merge with 5.5-tokudb tree. In particular:

* add TokuDB, together with the ft-index library
* cmake support, auto-detecting whether tokudb can be built
* fix packaging - tokudb-engine.rpm, deb
* remove PBXT
* add jemalloc
* the server is built with jemalloc by default even if TokuDB is not built
* documentation files in RPM are installed in the correct location
* support for optional deb packages (tokudb has specific build requirements)
* move plugins from mariadb-server deb to appropriate debs (server/test/libmariadbclient)
* correct mariadb-test.deb to be not architecture-independent
* fix out-of-tree builds to never modify in-tree files
* new handler::prepare_index_scan() method

100 files changed, 45843 insertions, 0 deletions

diff --git a/extra/jemalloc/COPYING b/extra/jemalloc/COPYING
new file mode 100644
index 00000000000..019e8132275
--- /dev/null
+++ b/extra/jemalloc/COPYING
@@ -0,0 +1,27 @@
+Unless otherwise specified, files in the jemalloc source distribution are
+subject to the following license:
+--------------------------------------------------------------------------------
+Copyright (C) 2002-2013 Jason Evans <jasone@canonware.com>.
+All rights reserved.
+Copyright (C) 2007-2012 Mozilla Foundation.  All rights reserved.
+Copyright (C) 2009-2013 Facebook, Inc.  All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+1. Redistributions of source code must retain the above copyright notice(s),
+   this list of conditions and the following disclaimer.
+2. Redistributions in binary form must reproduce the above copyright notice(s),
+   this list of conditions and the following disclaimer in the documentation
+   and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+--------------------------------------------------------------------------------
diff --git a/extra/jemalloc/ChangeLog b/extra/jemalloc/ChangeLog
new file mode 100644
index 00000000000..fc096d8f42f
--- /dev/null
+++ b/extra/jemalloc/ChangeLog
@@ -0,0 +1,425 @@
+Following are change highlights associated with official releases.  Important
+bug fixes are all mentioned, but internal enhancements are omitted here for
+brevity (even though they are more fun to write about).  Much more detail can be
+found in the git revision history:
+
+    http://www.canonware.com/cgi-bin/gitweb.cgi?p=jemalloc.git
+    git://canonware.com/jemalloc.git
+
+* 3.3.1 (March 6, 2013)
+
+  This version fixes bugs that are typically encountered only when utilizing
+  custom run-time options.
+
+  Bug fixes:
+  - Fix a locking order bug that could cause deadlock during fork if heap
+    profiling were enabled.
+  - Fix a chunk recycling bug that could cause the allocator to lose track of
+    whether a chunk was zeroed.   On FreeBSD, NetBSD, and OS X, it could cause
+    corruption if allocating via sbrk(2) (unlikely unless running with the
+    "dss:primary" option specified).  This was completely harmless on Linux
+    unless using mlockall(2) (and unlikely even then, unless the
+    --disable-munmap configure option or the "dss:primary" option was
+    specified).  This regression was introduced in 3.1.0 by the
+    mlockall(2)/madvise(2) interaction fix.
+  - Fix TLS-related memory corruption that could occur during thread exit if the
+    thread never allocated memory.  Only the quarantine and prof facilities were
+    susceptible.
+  - Fix two quarantine bugs:
+    + Internal reallocation of the quarantined object array leaked the old
+      array.
+    + Reallocation failure for internal reallocation of the quarantined object
+      array (very unlikely) resulted in memory corruption.
+  - Fix Valgrind integration to annotate all internally allocated memory in a
+    way that keeps Valgrind happy about internal data structure access.
+  - Fix building for s390 systems.
+
+* 3.3.0 (January 23, 2013)
+
+  This version includes a few minor performance improvements in addition to the
+  listed new features and bug fixes.
+
+  New features:
+  - Add clipping support to lg_chunk option processing.
+  - Add the --enable-ivsalloc option.
+  - Add the --without-export option.
+  - Add the --disable-zone-allocator option.
+
+  Bug fixes:
+  - Fix "arenas.extend" mallctl to output the number of arenas.
+  - Fix chunk_recycyle() to unconditionally inform Valgrind that returned memory
+    is undefined.
+  - Fix build break on FreeBSD related to alloca.h.
+
+* 3.2.0 (November 9, 2012)
+
+  In addition to a couple of bug fixes, this version modifies page run
+  allocation and dirty page purging algorithms in order to better control
+  page-level virtual memory fragmentation.
+
+  Incompatible changes:
+  - Change the "opt.lg_dirty_mult" default from 5 to 3 (32:1 to 8:1).
+
+  Bug fixes:
+  - Fix dss/mmap allocation precedence code to use recyclable mmap memory only
+    after primary dss allocation fails.
+  - Fix deadlock in the "arenas.purge" mallctl.  This regression was introduced
+    in 3.1.0 by the addition of the "arena.<i>.purge" mallctl.
+
+* 3.1.0 (October 16, 2012)
+
+  New features:
+  - Auto-detect whether running inside Valgrind, thus removing the need to
+    manually specify MALLOC_CONF=valgrind:true.
+  - Add the "arenas.extend" mallctl, which allows applications to create
+    manually managed arenas.
+  - Add the ALLOCM_ARENA() flag for {,r,d}allocm().
+  - Add the "opt.dss", "arena.<i>.dss", and "stats.arenas.<i>.dss" mallctls,
+    which provide control over dss/mmap precedence.
+  - Add the "arena.<i>.purge" mallctl, which obsoletes "arenas.purge".
+  - Define LG_QUANTUM for hppa.
+
+  Incompatible changes:
+  - Disable tcache by default if running inside Valgrind, in order to avoid
+    making unallocated objects appear reachable to Valgrind.
+  - Drop const from malloc_usable_size() argument on Linux.
+
+  Bug fixes:
+  - Fix heap profiling crash if sampled object is freed via realloc(p, 0).
+  - Remove const from __*_hook variable declarations, so that glibc can modify
+    them during process forking.
+  - Fix mlockall(2)/madvise(2) interaction.
+  - Fix fork(2)-related deadlocks.
+  - Fix error return value for "thread.tcache.enabled" mallctl.
+
+* 3.0.0 (May 11, 2012)
+
+  Although this version adds some major new features, the primary focus is on
+  internal code cleanup that facilitates maintainability and portability, most
+  of which is not reflected in the ChangeLog.  This is the first release to
+  incorporate substantial contributions from numerous other developers, and the
+  result is a more broadly useful allocator (see the git revision history for
+  contribution details).  Note that the license has been unified, thanks to
+  Facebook granting a license under the same terms as the other copyright
+  holders (see COPYING).
+
+  New features:
+  - Implement Valgrind support, redzones, and quarantine.
+  - Add support for additional platforms:
+    + FreeBSD
+    + Mac OS X Lion
+    + MinGW
+    + Windows (no support yet for replacing the system malloc)
+  - Add support for additional architectures:
+    + MIPS
+    + SH4
+    + Tilera
+  - Add support for cross compiling.
+  - Add nallocm(), which rounds a request size up to the nearest size class
+    without actually allocating.
+  - Implement aligned_alloc() (blame C11).
+  - Add the "thread.tcache.enabled" mallctl.
+  - Add the "opt.prof_final" mallctl.
+  - Update pprof (from gperftools 2.0).
+  - Add the --with-mangling option.
+  - Add the --disable-experimental option.
+  - Add the --disable-munmap option, and make it the default on Linux.
+  - Add the --enable-mremap option, which disables use of mremap(2) by default.
+
+  Incompatible changes:
+  - Enable stats by default.
+  - Enable fill by default.
+  - Disable lazy locking by default.
+  - Rename the "tcache.flush" mallctl to "thread.tcache.flush".
+  - Rename the "arenas.pagesize" mallctl to "arenas.page".
+  - Change the "opt.lg_prof_sample" default from 0 to 19 (1 B to 512 KiB).
+  - Change the "opt.prof_accum" default from true to false.
+
+  Removed features:
+  - Remove the swap feature, including the "config.swap", "swap.avail",
+    "swap.prezeroed", "swap.nfds", and "swap.fds" mallctls.
+  - Remove highruns statistics, including the
+    "stats.arenas.<i>.bins.<j>.highruns" and
+    "stats.arenas.<i>.lruns.<j>.highruns" mallctls.
+  - As part of small size class refactoring, remove the "opt.lg_[qc]space_max",
+    "arenas.cacheline", "arenas.subpage", "arenas.[tqcs]space_{min,max}", and
+    "arenas.[tqcs]bins" mallctls.
+  - Remove the "arenas.chunksize" mallctl.
+  - Remove the "opt.lg_prof_tcmax" option.
+  - Remove the "opt.lg_prof_bt_max" option.
+  - Remove the "opt.lg_tcache_gc_sweep" option.
+  - Remove the --disable-tiny option, including the "config.tiny" mallctl.
+  - Remove the --enable-dynamic-page-shift configure option.
+  - Remove the --enable-sysv configure option.
+
+  Bug fixes:
+  - Fix a statistics-related bug in the "thread.arena" mallctl that could cause
+    invalid statistics and crashes.
+  - Work around TLS deallocation via free() on Linux.  This bug could cause
+    write-after-free memory corruption.
+  - Fix a potential deadlock that could occur during interval- and
+    growth-triggered heap profile dumps.
+  - Fix large calloc() zeroing bugs due to dropping chunk map unzeroed flags.
+  - Fix chunk_alloc_dss() to stop claiming memory is zeroed.  This bug could
+    cause memory corruption and crashes with --enable-dss specified.
+  - Fix fork-related bugs that could cause deadlock in children between fork
+    and exec.
+  - Fix malloc_stats_print() to honor 'b' and 'l' in the opts parameter.
+  - Fix realloc(p, 0) to act like free(p).
+  - Do not enforce minimum alignment in memalign().
+  - Check for NULL pointer in malloc_usable_size().
+  - Fix an off-by-one heap profile statistics bug that could be observed in
+    interval- and growth-triggered heap profiles.
+  - Fix the "epoch" mallctl to update cached stats even if the passed in epoch
+    is 0.
+  - Fix bin->runcur management to fix a layout policy bug.  This bug did not
+    affect correctness.
+  - Fix a bug in choose_arena_hard() that potentially caused more arenas to be
+    initialized than necessary.
+  - Add missing "opt.lg_tcache_max" mallctl implementation.
+  - Use glibc allocator hooks to make mixed allocator usage less likely.
+  - Fix build issues for --disable-tcache.
+  - Don't mangle pthread_create() when --with-private-namespace is specified.
+
+* 2.2.5 (November 14, 2011)
+
+  Bug fixes:
+  - Fix huge_ralloc() race when using mremap(2).  This is a serious bug that
+    could cause memory corruption and/or crashes.
+  - Fix huge_ralloc() to maintain chunk statistics.
+  - Fix malloc_stats_print(..., "a") output.
+
+* 2.2.4 (November 5, 2011)
+
+  Bug fixes:
+  - Initialize arenas_tsd before using it.  This bug existed for 2.2.[0-3], as
+    well as for --disable-tls builds in earlier releases.
+  - Do not assume a 4 KiB page size in test/rallocm.c.
+
+* 2.2.3 (August 31, 2011)
+
+  This version fixes numerous bugs related to heap profiling.
+
+  Bug fixes:
+  - Fix a prof-related race condition.  This bug could cause memory corruption,
+    but only occurred in non-default configurations (prof_accum:false).
+  - Fix off-by-one backtracing issues (make sure that prof_alloc_prep() is
+    excluded from backtraces).
+  - Fix a prof-related bug in realloc() (only triggered by OOM errors).
+  - Fix prof-related bugs in allocm() and rallocm().
+  - Fix prof_tdata_cleanup() for --disable-tls builds.
+  - Fix a relative include path, to fix objdir builds.
+
+* 2.2.2 (July 30, 2011)
+
+  Bug fixes:
+  - Fix a build error for --disable-tcache.
+  - Fix assertions in arena_purge() (for real this time).
+  - Add the --with-private-namespace option.  This is a workaround for symbol
+    conflicts that can inadvertently arise when using static libraries.
+
+* 2.2.1 (March 30, 2011)
+
+  Bug fixes:
+  - Implement atomic operations for x86/x64.  This fixes compilation failures
+    for versions of gcc that are still in wide use.
+  - Fix an assertion in arena_purge().
+
+* 2.2.0 (March 22, 2011)
+
+  This version incorporates several improvements to algorithms and data
+  structures that tend to reduce fragmentation and increase speed.
+
+  New features:
+  - Add the "stats.cactive" mallctl.
+  - Update pprof (from google-perftools 1.7).
+  - Improve backtracing-related configuration logic, and add the
+    --disable-prof-libgcc option.
+
+  Bug fixes:
+  - Change default symbol visibility from "internal", to "hidden", which
+    decreases the overhead of library-internal function calls.
+  - Fix symbol visibility so that it is also set on OS X.
+  - Fix a build dependency regression caused by the introduction of the .pic.o
+    suffix for PIC object files.
+  - Add missing checks for mutex initialization failures.
+  - Don't use libgcc-based backtracing except on x64, where it is known to work.
+  - Fix deadlocks on OS X that were due to memory allocation in
+    pthread_mutex_lock().
+  - Heap profiling-specific fixes:
+    + Fix memory corruption due to integer overflow in small region index
+      computation, when using a small enough sample interval that profiling
+      context pointers are stored in small run headers.
+    + Fix a bootstrap ordering bug that only occurred with TLS disabled.
+    + Fix a rallocm() rsize bug.
+    + Fix error detection bugs for aligned memory allocation.
+
+* 2.1.3 (March 14, 2011)
+
+  Bug fixes:
+  - Fix a cpp logic regression (due to the "thread.{de,}allocatedp" mallctl fix
+    for OS X in 2.1.2).
+  - Fix a "thread.arena" mallctl bug.
+  - Fix a thread cache stats merging bug.
+
+* 2.1.2 (March 2, 2011)
+
+  Bug fixes:
+  - Fix "thread.{de,}allocatedp" mallctl for OS X.
+  - Add missing jemalloc.a to build system.
+
+* 2.1.1 (January 31, 2011)
+
+  Bug fixes:
+  - Fix aligned huge reallocation (affected allocm()).
+  - Fix the ALLOCM_LG_ALIGN macro definition.
+  - Fix a heap dumping deadlock.
+  - Fix a "thread.arena" mallctl bug.
+
+* 2.1.0 (December 3, 2010)
+
+  This version incorporates some optimizations that can't quite be considered
+  bug fixes.
+
+  New features:
+  - Use Linux's mremap(2) for huge object reallocation when possible.
+  - Avoid locking in mallctl*() when possible.
+  - Add the "thread.[de]allocatedp" mallctl's.
+  - Convert the manual page source from roff to DocBook, and generate both roff
+    and HTML manuals.
+
+  Bug fixes:
+  - Fix a crash due to incorrect bootstrap ordering.  This only impacted
+    --enable-debug --enable-dss configurations.
+  - Fix a minor statistics bug for mallctl("swap.avail", ...).
+
+* 2.0.1 (October 29, 2010)
+
+  Bug fixes:
+  - Fix a race condition in heap profiling that could cause undefined behavior
+    if "opt.prof_accum" were disabled.
+  - Add missing mutex unlocks for some OOM error paths in the heap profiling
+    code.
+  - Fix a compilation error for non-C99 builds.
+
+* 2.0.0 (October 24, 2010)
+
+  This version focuses on the experimental *allocm() API, and on improved
+  run-time configuration/introspection.  Nonetheless, numerous performance
+  improvements are also included.
+
+  New features:
+  - Implement the experimental {,r,s,d}allocm() API, which provides a superset
+    of the functionality available via malloc(), calloc(), posix_memalign(),
+    realloc(), malloc_usable_size(), and free().  These functions can be used to
+    allocate/reallocate aligned zeroed memory, ask for optional extra memory
+    during reallocation, prevent object movement during reallocation, etc.
+  - Replace JEMALLOC_OPTIONS/JEMALLOC_PROF_PREFIX with MALLOC_CONF, which is
+    more human-readable, and more flexible.  For example:
+      JEMALLOC_OPTIONS=AJP
+    is now:
+      MALLOC_CONF=abort:true,fill:true,stats_print:true
+  - Port to Apple OS X.  Sponsored by Mozilla.
+  - Make it possible for the application to control thread-->arena mappings via
+    the "thread.arena" mallctl.
+  - Add compile-time support for all TLS-related functionality via pthreads TSD.
+    This is mainly of interest for OS X, which does not support TLS, but has a
+    TSD implementation with similar performance.
+  - Override memalign() and valloc() if they are provided by the system.
+  - Add the "arenas.purge" mallctl, which can be used to synchronously purge all
+    dirty unused pages.
+  - Make cumulative heap profiling data optional, so that it is possible to
+    limit the amount of memory consumed by heap profiling data structures.
+  - Add per thread allocation counters that can be accessed via the
+    "thread.allocated" and "thread.deallocated" mallctls.
+
+  Incompatible changes:
+  - Remove JEMALLOC_OPTIONS and malloc_options (see MALLOC_CONF above).
+  - Increase default backtrace depth from 4 to 128 for heap profiling.
+  - Disable interval-based profile dumps by default.
+
+  Bug fixes:
+  - Remove bad assertions in fork handler functions.  These assertions could
+    cause aborts for some combinations of configure settings.
+  - Fix strerror_r() usage to deal with non-standard semantics in GNU libc.
+  - Fix leak context reporting.  This bug tended to cause the number of contexts
+    to be underreported (though the reported number of objects and bytes were
+    correct).
+  - Fix a realloc() bug for large in-place growing reallocation.  This bug could
+    cause memory corruption, but it was hard to trigger.
+  - Fix an allocation bug for small allocations that could be triggered if
+    multiple threads raced to create a new run of backing pages.
+  - Enhance the heap profiler to trigger samples based on usable size, rather
+    than request size.
+  - Fix a heap profiling bug due to sometimes losing track of requested object
+    size for sampled objects.
+
+* 1.0.3 (August 12, 2010)
+
+  Bug fixes:
+  - Fix the libunwind-based implementation of stack backtracing (used for heap
+    profiling).  This bug could cause zero-length backtraces to be reported.
+  - Add a missing mutex unlock in library initialization code.  If multiple
+    threads raced to initialize malloc, some of them could end up permanently
+    blocked.
+
+* 1.0.2 (May 11, 2010)
+
+  Bug fixes:
+  - Fix junk filling of large objects, which could cause memory corruption.
+  - Add MAP_NORESERVE support for chunk mapping, because otherwise virtual
+    memory limits could cause swap file configuration to fail.  Contributed by
+    Jordan DeLong.
+
+* 1.0.1 (April 14, 2010)
+
+  Bug fixes:
+  - Fix compilation when --enable-fill is specified.
+  - Fix threads-related profiling bugs that affected accuracy and caused memory
+    to be leaked during thread exit.
+  - Fix dirty page purging race conditions that could cause crashes.
+  - Fix crash in tcache flushing code during thread destruction.
+
+* 1.0.0 (April 11, 2010)
+
+  This release focuses on speed and run-time introspection.  Numerous
+  algorithmic improvements make this release substantially faster than its
+  predecessors.
+
+  New features:
+  - Implement autoconf-based configuration system.
+  - Add mallctl*(), for the purposes of introspection and run-time
+    configuration.
+  - Make it possible for the application to manually flush a thread's cache, via
+    the "tcache.flush" mallctl.
+  - Base maximum dirty page count on proportion of active memory.
+  - Compute various addtional run-time statistics, including per size class
+    statistics for large objects.
+  - Expose malloc_stats_print(), which can be called repeatedly by the
+    application.
+  - Simplify the malloc_message() signature to only take one string argument,
+    and incorporate an opaque data pointer argument for use by the application
+    in combination with malloc_stats_print().
+  - Add support for allocation backed by one or more swap files, and allow the
+    application to disable over-commit if swap files are in use.
+  - Implement allocation profiling and leak checking.
+
+  Removed features:
+  - Remove the dynamic arena rebalancing code, since thread-specific caching
+    reduces its utility.
+
+  Bug fixes:
+  - Modify chunk allocation to work when address space layout randomization
+    (ASLR) is in use.
+  - Fix thread cleanup bugs related to TLS destruction.
+  - Handle 0-size allocation requests in posix_memalign().
+  - Fix a chunk leak.  The leaked chunks were never touched, so this impacted
+    virtual memory usage, but not physical memory usage.
+
+* linux_2008082[78]a (August 27/28, 2008)
+
+  These snapshot releases are the simple result of incorporating Linux-specific
+  support into the FreeBSD malloc sources.
+
+--------------------------------------------------------------------------------
+vim:filetype=text:textwidth=80
diff --git a/extra/jemalloc/INSTALL b/extra/jemalloc/INSTALL
new file mode 100644
index 00000000000..6e371ce5095
--- /dev/null
+++ b/extra/jemalloc/INSTALL
@@ -0,0 +1,293 @@
+Building and installing jemalloc can be as simple as typing the following while
+in the root directory of the source tree:
+
+    ./configure
+    make
+    make install
+
+=== Advanced configuration =====================================================
+
+The 'configure' script supports numerous options that allow control of which
+functionality is enabled, where jemalloc is installed, etc.  Optionally, pass
+any of the following arguments (not a definitive list) to 'configure':
+
+--help
+    Print a definitive list of options.
+
+--prefix=<install-root-dir>
+    Set the base directory in which to install.  For example:
+
+        ./configure --prefix=/usr/local
+
+    will cause files to be installed into /usr/local/include, /usr/local/lib,
+    and /usr/local/man.
+
+--with-rpath=<colon-separated-rpath>
+    Embed one or more library paths, so that libjemalloc can find the libraries
+    it is linked to.  This works only on ELF-based systems.
+
+--with-mangling=<map>
+    Mangle public symbols specified in <map> which is a comma-separated list of
+    name:mangled pairs.
+
+    For example, to use ld's --wrap option as an alternative method for
+    overriding libc's malloc implementation, specify something like:
+
+      --with-mangling=malloc:__wrap_malloc,free:__wrap_free[...]
+
+    Note that mangling happens prior to application of the prefix specified by
+    --with-jemalloc-prefix, and mangled symbols are then ignored when applying
+    the prefix.
+
+--with-jemalloc-prefix=<prefix>
+    Prefix all public APIs with <prefix>.  For example, if <prefix> is
+    "prefix_", API changes like the following occur:
+
+      malloc()         --> prefix_malloc()
+      malloc_conf      --> prefix_malloc_conf
+      /etc/malloc.conf --> /etc/prefix_malloc.conf
+      MALLOC_CONF      --> PREFIX_MALLOC_CONF
+
+    This makes it possible to use jemalloc at the same time as the system
+    allocator, or even to use multiple copies of jemalloc simultaneously.
+
+    By default, the prefix is "", except on OS X, where it is "je_".  On OS X,
+    jemalloc overlays the default malloc zone, but makes no attempt to actually
+    replace the "malloc", "calloc", etc. symbols.
+
+--without-export
+    Don't export public APIs. This can be useful when building jemalloc as a
+    static library, or to avoid exporting public APIs when using the zone
+    allocator on OSX.
+
+--with-private-namespace=<prefix>
+    Prefix all library-private APIs with <prefix>.  For shared libraries,
+    symbol visibility mechanisms prevent these symbols from being exported, but
+    for static libraries, naming collisions are a real possibility.  By
+    default, the prefix is "" (empty string).
+
+--with-install-suffix=<suffix>
+    Append <suffix> to the base name of all installed files, such that multiple
+    versions of jemalloc can coexist in the same installation directory.  For
+    example, libjemalloc.so.0 becomes libjemalloc<suffix>.so.0.
+
+--enable-cc-silence
+    Enable code that silences non-useful compiler warnings.  This is helpful
+    when trying to tell serious warnings from those due to compiler
+    limitations, but it potentially incurs a performance penalty.
+
+--enable-debug
+    Enable assertions and validation code.  This incurs a substantial
+    performance hit, but is very useful during application development.
+    Implies --enable-ivsalloc.
+
+--enable-ivsalloc
+    Enable validation code, which verifies that pointers reside within
+    jemalloc-owned chunks before dereferencing them. This incurs a substantial
+    performance hit.
+
+--disable-stats
+    Disable statistics gathering functionality.  See the "opt.stats_print"
+    option documentation for usage details.
+
+--enable-prof
+    Enable heap profiling and leak detection functionality.  See the "opt.prof"
+    option documentation for usage details.  When enabled, there are several
+    approaches to backtracing, and the configure script chooses the first one
+    in the following list that appears to function correctly:
+
+    + libunwind      (requires --enable-prof-libunwind)
+    + libgcc         (unless --disable-prof-libgcc)
+    + gcc intrinsics (unless --disable-prof-gcc)
+
+--enable-prof-libunwind
+    Use the libunwind library (http://www.nongnu.org/libunwind/) for stack
+    backtracing.
+
+--disable-prof-libgcc
+    Disable the use of libgcc's backtracing functionality.
+
+--disable-prof-gcc
+    Disable the use of gcc intrinsics for backtracing.
+
+--with-static-libunwind=<libunwind.a>
+    Statically link against the specified libunwind.a rather than dynamically
+    linking with -lunwind.
+
+--disable-tcache
+    Disable thread-specific caches for small objects.  Objects are cached and
+    released in bulk, thus reducing the total number of mutex operations.  See
+    the "opt.tcache" option for usage details.
+
+--enable-mremap
+    Enable huge realloc() via mremap(2).  mremap() is disabled by default
+    because the flavor used is specific to Linux, which has a quirk in its
+    virtual memory allocation algorithm that causes semi-permanent VM map holes
+    under normal jemalloc operation.
+
+--disable-munmap
+    Disable virtual memory deallocation via munmap(2); instead keep track of
+    the virtual memory for later use.  munmap() is disabled by default (i.e.
+    --disable-munmap is implied) on Linux, which has a quirk in its virtual
+    memory allocation algorithm that causes semi-permanent VM map holes under
+    normal jemalloc operation.
+
+--enable-dss
+    Enable support for page allocation/deallocation via sbrk(2), in addition to
+    mmap(2).
+
+--disable-fill
+    Disable support for junk/zero filling of memory, quarantine, and redzones.
+    See the "opt.junk", "opt.zero", "opt.quarantine", and "opt.redzone" option
+    documentation for usage details.
+
+--disable-valgrind
+    Disable support for Valgrind.
+
+--disable-experimental
+    Disable support for the experimental API (*allocm()).
+
+--disable-zone-allocator
+    Disable zone allocator for Darwin. This means jemalloc won't be hooked as
+    the default allocator on OSX/iOS.
+
+--enable-utrace
+    Enable utrace(2)-based allocation tracing.  This feature is not broadly
+    portable (FreeBSD has it, but Linux and OS X do not).
+
+--enable-xmalloc
+    Enable support for optional immediate termination due to out-of-memory
+    errors, as is commonly implemented by "xmalloc" wrapper function for malloc.
+    See the "opt.xmalloc" option documentation for usage details.
+
+--enable-lazy-lock
+    Enable code that wraps pthread_create() to detect when an application
+    switches from single-threaded to multi-threaded mode, so that it can avoid
+    mutex locking/unlocking operations while in single-threaded mode.  In
+    practice, this feature usually has little impact on performance unless
+    thread-specific caching is disabled.
+
+--disable-tls
+    Disable thread-local storage (TLS), which allows for fast access to
+    thread-local variables via the __thread keyword.  If TLS is available,
+    jemalloc uses it for several purposes.
+
+--with-xslroot=<path>
+    Specify where to find DocBook XSL stylesheets when building the
+    documentation.
+
+The following environment variables (not a definitive list) impact configure's
+behavior:
+
+CFLAGS="?"
+    Pass these flags to the compiler.  You probably shouldn't define this unless
+    you know what you are doing.  (Use EXTRA_CFLAGS instead.)
+
+EXTRA_CFLAGS="?"
+    Append these flags to CFLAGS.  This makes it possible to add flags such as
+    -Werror, while allowing the configure script to determine what other flags
+    are appropriate for the specified configuration.
+
+    The configure script specifically checks whether an optimization flag (-O*)
+    is specified in EXTRA_CFLAGS, and refrains from specifying an optimization
+    level if it finds that one has already been specified.
+
+CPPFLAGS="?"
+    Pass these flags to the C preprocessor.  Note that CFLAGS is not passed to
+    'cpp' when 'configure' is looking for include files, so you must use
+    CPPFLAGS instead if you need to help 'configure' find header files.
+
+LD_LIBRARY_PATH="?"
+    'ld' uses this colon-separated list to find libraries.
+
+LDFLAGS="?"
+    Pass these flags when linking.
+
+PATH="?"
+    'configure' uses this to find programs.
+
+=== Advanced compilation =======================================================
+
+To build only parts of jemalloc, use the following targets:
+
+    build_lib_shared
+    build_lib_static
+    build_lib
+    build_doc_html
+    build_doc_man
+    build_doc
+
+To install only parts of jemalloc, use the following targets:
+
+    install_bin
+    install_include
+    install_lib_shared
+    install_lib_static
+    install_lib
+    install_doc_html
+    install_doc_man
+    install_doc
+
+To clean up build results to varying degrees, use the following make targets:
+
+    clean
+    distclean
+    relclean
+
+=== Advanced installation ======================================================
+
+Optionally, define make variables when invoking make, including (not
+exclusively):
+
+INCLUDEDIR="?"
+    Use this as the installation prefix for header files.
+
+LIBDIR="?"
+    Use this as the installation prefix for libraries.
+
+MANDIR="?"
+    Use this as the installation prefix for man pages.
+
+DESTDIR="?"
+    Prepend DESTDIR to INCLUDEDIR, LIBDIR, DATADIR, and MANDIR.  This is useful
+    when installing to a different path than was specified via --prefix.
+
+CC="?"
+    Use this to invoke the C compiler.
+
+CFLAGS="?"
+    Pass these flags to the compiler.
+
+CPPFLAGS="?"
+    Pass these flags to the C preprocessor.
+
+LDFLAGS="?"
+    Pass these flags when linking.
+
+PATH="?"
+    Use this to search for programs used during configuration and building.
+
+=== Development ================================================================
+
+If you intend to make non-trivial changes to jemalloc, use the 'autogen.sh'
+script rather than 'configure'.  This re-generates 'configure', enables
+configuration dependency rules, and enables re-generation of automatically
+generated source files.
+
+The build system supports using an object directory separate from the source
+tree.  For example, you can create an 'obj' directory, and from within that
+directory, issue configuration and build commands:
+
+    autoconf
+    mkdir obj
+    cd obj
+    ../configure --enable-autogen
+    make
+
+=== Documentation ==============================================================
+
+The manual page is generated in both html and roff formats.  Any web browser
+can be used to view the html manual.  The roff manual page can be formatted
+prior to installation via the following command:
+
+    nroff -man -t doc/jemalloc.3
diff --git a/extra/jemalloc/Makefile.in b/extra/jemalloc/Makefile.in
new file mode 100644
index 00000000000..74810472d11
--- /dev/null
+++ b/extra/jemalloc/Makefile.in
@@ -0,0 +1,324 @@
+# Clear out all vpaths, then set just one (default vpath) for the main build
+# directory.
+vpath
+vpath % .
+
+# Clear the default suffixes, so that built-in rules are not used.
+.SUFFIXES :
+
+SHELL := /bin/sh
+
+CC := @CC@
+
+# Configuration parameters.
+DESTDIR =
+BINDIR := $(DESTDIR)@BINDIR@
+INCLUDEDIR := $(DESTDIR)@INCLUDEDIR@
+LIBDIR := $(DESTDIR)@LIBDIR@
+DATADIR := $(DESTDIR)@DATADIR@
+MANDIR := $(DESTDIR)@MANDIR@
+srcroot := @srcroot@
+objroot := @objroot@
+abs_srcroot := @abs_srcroot@
+abs_objroot := @abs_objroot@
+
+# Build parameters.
+CPPFLAGS := @CPPFLAGS@ -I$(srcroot)include -I$(objroot)include
+CFLAGS := @CFLAGS@
+LDFLAGS := @LDFLAGS@
+EXTRA_LDFLAGS := @EXTRA_LDFLAGS@
+LIBS := @LIBS@
+RPATH_EXTRA := @RPATH_EXTRA@
+SO := @so@
+IMPORTLIB := @importlib@
+O := @o@
+A := @a@
+EXE := @exe@
+LIBPREFIX := @libprefix@
+REV := @rev@
+install_suffix := @install_suffix@
+ABI := @abi@
+XSLTPROC := @XSLTPROC@
+AUTOCONF := @AUTOCONF@
+_RPATH = @RPATH@
+RPATH = $(if $(1),$(call _RPATH,$(1)))
+cfghdrs_in := @cfghdrs_in@
+cfghdrs_out := @cfghdrs_out@
+cfgoutputs_in := @cfgoutputs_in@
+cfgoutputs_out := @cfgoutputs_out@
+enable_autogen := @enable_autogen@
+enable_experimental := @enable_experimental@
+enable_zone_allocator := @enable_zone_allocator@
+DSO_LDFLAGS = @DSO_LDFLAGS@
+SOREV = @SOREV@
+PIC_CFLAGS = @PIC_CFLAGS@
+CTARGET = @CTARGET@
+LDTARGET = @LDTARGET@
+MKLIB = @MKLIB@
+CC_MM = @CC_MM@
+
+ifeq (macho, $(ABI))
+TEST_LIBRARY_PATH := DYLD_FALLBACK_LIBRARY_PATH="$(objroot)lib"
+else
+ifeq (pecoff, $(ABI))
+TEST_LIBRARY_PATH := PATH="$(PATH):$(objroot)lib"
+else
+TEST_LIBRARY_PATH :=
+endif
+endif
+
+LIBJEMALLOC := $(LIBPREFIX)jemalloc$(install_suffix)
+
+# Lists of files.
+BINS := $(srcroot)bin/pprof $(objroot)bin/jemalloc.sh
+CHDRS := $(objroot)include/jemalloc/jemalloc$(install_suffix).h \
+	$(objroot)include/jemalloc/jemalloc_defs$(install_suffix).h
+CSRCS := $(srcroot)src/jemalloc.c $(srcroot)src/arena.c $(srcroot)src/atomic.c \
+	$(srcroot)src/base.c $(srcroot)src/bitmap.c $(srcroot)src/chunk.c \
+	$(srcroot)src/chunk_dss.c $(srcroot)src/chunk_mmap.c \
+	$(srcroot)src/ckh.c $(srcroot)src/ctl.c $(srcroot)src/extent.c \
+	$(srcroot)src/hash.c $(srcroot)src/huge.c $(srcroot)src/mb.c \
+	$(srcroot)src/mutex.c $(srcroot)src/prof.c $(srcroot)src/quarantine.c \
+	$(srcroot)src/rtree.c $(srcroot)src/stats.c $(srcroot)src/tcache.c \
+	$(srcroot)src/util.c $(srcroot)src/tsd.c
+ifeq ($(enable_zone_allocator), 1)
+CSRCS += $(srcroot)src/zone.c
+endif
+ifeq ($(IMPORTLIB),$(SO))
+STATIC_LIBS := $(objroot)lib/$(LIBJEMALLOC).$(A)
+endif
+ifdef PIC_CFLAGS
+STATIC_LIBS += $(objroot)lib/$(LIBJEMALLOC)_pic.$(A)
+else
+STATIC_LIBS += $(objroot)lib/$(LIBJEMALLOC)_s.$(A)
+endif
+DSOS := $(objroot)lib/$(LIBJEMALLOC).$(SOREV)
+ifneq ($(SOREV),$(SO))
+DSOS += $(objroot)lib/$(LIBJEMALLOC).$(SO)
+endif
+MAN3 := $(objroot)doc/jemalloc$(install_suffix).3
+DOCS_XML := $(objroot)doc/jemalloc$(install_suffix).xml
+DOCS_HTML := $(DOCS_XML:$(objroot)%.xml=$(srcroot)%.html)
+DOCS_MAN3 := $(DOCS_XML:$(objroot)%.xml=$(srcroot)%.3)
+DOCS := $(DOCS_HTML) $(DOCS_MAN3)
+CTESTS := $(srcroot)test/aligned_alloc.c $(srcroot)test/allocated.c \
+	$(srcroot)test/ALLOCM_ARENA.c $(srcroot)test/bitmap.c \
+	$(srcroot)test/mremap.c $(srcroot)test/posix_memalign.c \
+	$(srcroot)test/thread_arena.c $(srcroot)test/thread_tcache_enabled.c
+ifeq ($(enable_experimental), 1)
+CTESTS += $(srcroot)test/allocm.c $(srcroot)test/rallocm.c
+endif
+
+COBJS := $(CSRCS:$(srcroot)%.c=$(objroot)%.$(O))
+CPICOBJS := $(CSRCS:$(srcroot)%.c=$(objroot)%.pic.$(O))
+CTESTOBJS := $(CTESTS:$(srcroot)%.c=$(objroot)%.$(O))
+
+.PHONY: all dist build_doc_html build_doc_man build_doc
+.PHONY: install_bin install_include install_lib
+.PHONY: install_doc_html install_doc_man install_doc install
+.PHONY: tests check clean distclean relclean
+
+.SECONDARY : $(CTESTOBJS)
+
+# Default target.
+all: build
+
+dist: build_doc
+
+$(srcroot)doc/%.html : $(objroot)doc/%.xml $(srcroot)doc/stylesheet.xsl $(objroot)doc/html.xsl
+	$(XSLTPROC) -o $@ $(objroot)doc/html.xsl $<
+
+$(srcroot)doc/%.3 : $(objroot)doc/%.xml $(srcroot)doc/stylesheet.xsl $(objroot)doc/manpages.xsl
+	$(XSLTPROC) -o $@ $(objroot)doc/manpages.xsl $<
+
+build_doc_html: $(DOCS_HTML)
+build_doc_man: $(DOCS_MAN3)
+build_doc: $(DOCS)
+
+#
+# Include generated dependency files.
+#
+ifdef CC_MM
+-include $(COBJS:%.$(O)=%.d)
+-include $(CPICOBJS:%.$(O)=%.d)
+-include $(CTESTOBJS:%.$(O)=%.d)
+endif
+
+$(COBJS): $(objroot)src/%.$(O): $(srcroot)src/%.c
+$(CPICOBJS): $(objroot)src/%.pic.$(O): $(srcroot)src/%.c
+$(CPICOBJS): CFLAGS += $(PIC_CFLAGS)
+$(CTESTOBJS): $(objroot)test/%.$(O): $(srcroot)test/%.c
+$(CTESTOBJS): CPPFLAGS += -I$(objroot)test
+ifneq ($(IMPORTLIB),$(SO))
+$(COBJS): CPPFLAGS += -DDLLEXPORT
+endif
+
+ifndef CC_MM
+# Dependencies
+HEADER_DIRS = $(srcroot)include/jemalloc/internal \
+	$(objroot)include/jemalloc $(objroot)include/jemalloc/internal
+HEADERS = $(wildcard $(foreach dir,$(HEADER_DIRS),$(dir)/*.h))
+$(COBJS) $(CPICOBJS) $(CTESTOBJS): $(HEADERS)
+$(CTESTOBJS): $(objroot)test/jemalloc_test.h
+endif
+
+$(COBJS) $(CPICOBJS) $(CTESTOBJS): %.$(O):
+	@mkdir -p $(@D)
+	$(CC) $(CFLAGS) -c $(CPPFLAGS) $(CTARGET) $<
+ifdef CC_MM
+	@$(CC) -MM $(CPPFLAGS) -MT $@ -o $(@:%.$(O)=%.d) $<
+endif
+
+ifneq ($(SOREV),$(SO))
+%.$(SO) : %.$(SOREV)
+	@mkdir -p $(@D)
+	ln -sf $(<F) $@
+endif
+
+$(objroot)lib/$(LIBJEMALLOC).$(SOREV) : $(if $(PIC_CFLAGS),$(CPICOBJS),$(COBJS))
+	@mkdir -p $(@D)
+	$(CC) $(DSO_LDFLAGS) $(call RPATH,$(RPATH_EXTRA)) $(LDTARGET) $+ $(LDFLAGS) $(LIBS) $(EXTRA_LDFLAGS)
+
+$(objroot)lib/$(LIBJEMALLOC)_pic.$(A) : $(CPICOBJS)
+$(objroot)lib/$(LIBJEMALLOC).$(A) : $(COBJS)
+$(objroot)lib/$(LIBJEMALLOC)_s.$(A) : $(COBJS)
+
+$(STATIC_LIBS):
+	@mkdir -p $(@D)
+	$(MKLIB) $+
+
+$(objroot)test/bitmap$(EXE): $(objroot)src/bitmap.$(O)
+
+$(objroot)test/%$(EXE): $(objroot)test/%.$(O) $(objroot)src/util.$(O) $(DSOS)
+	@mkdir -p $(@D)
+	$(CC) $(LDTARGET) $(filter %.$(O),$^) $(call RPATH,$(objroot)lib) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB) $(filter -lpthread,$(LIBS)) $(EXTRA_LDFLAGS)
+
+build_lib_shared: $(DSOS)
+build_lib_static: $(STATIC_LIBS)
+build: build_lib_shared build_lib_static
+
+install_bin:
+	install -d $(BINDIR)
+	@for b in $(BINS); do \
+	echo "install -m 755 $$b $(BINDIR)"; \
+	install -m 755 $$b $(BINDIR); \
+done
+
+install_include:
+	install -d $(INCLUDEDIR)/jemalloc
+	@for h in $(CHDRS); do \
+	echo "install -m 644 $$h $(INCLUDEDIR)/jemalloc"; \
+	install -m 644 $$h $(INCLUDEDIR)/jemalloc; \
+done
+
+install_lib_shared: $(DSOS)
+	install -d $(LIBDIR)
+	install -m 755 $(objroot)lib/$(LIBJEMALLOC).$(SOREV) $(LIBDIR)
+ifneq ($(SOREV),$(SO))
+	ln -sf $(LIBJEMALLOC).$(SOREV) $(LIBDIR)/$(LIBJEMALLOC).$(SO)
+endif
+
+install_lib_static: $(STATIC_LIBS)
+	install -d $(LIBDIR)
+	@for l in $(STATIC_LIBS); do \
+	echo "install -m 755 $$l $(LIBDIR)"; \
+	install -m 755 $$l $(LIBDIR); \
+done
+
+install_lib: install_lib_shared install_lib_static
+
+install_doc_html:
+	install -d $(DATADIR)/doc/jemalloc$(install_suffix)
+	@for d in $(DOCS_HTML); do \
+	echo "install -m 644 $$d $(DATADIR)/doc/jemalloc$(install_suffix)"; \
+	install -m 644 $$d $(DATADIR)/doc/jemalloc$(install_suffix); \
+done
+
+install_doc_man:
+	install -d $(MANDIR)/man3
+	@for d in $(DOCS_MAN3); do \
+	echo "install -m 644 $$d $(MANDIR)/man3"; \
+	install -m 644 $$d $(MANDIR)/man3; \
+done
+
+install_doc: install_doc_html install_doc_man
+
+install: install_bin install_include install_lib install_doc
+
+tests: $(CTESTS:$(srcroot)%.c=$(objroot)%$(EXE))
+
+check: tests
+	@mkdir -p $(objroot)test
+	@$(SHELL) -c 'total=0; \
+		failures=0; \
+		echo "========================================="; \
+		for t in $(CTESTS:$(srcroot)%.c=$(objroot)%); do \
+			total=`expr $$total + 1`; \
+			/bin/echo -n "$${t} ... "; \
+			$(TEST_LIBRARY_PATH) $${t}$(EXE) $(abs_srcroot) \
+			  $(abs_objroot) > $(objroot)$${t}.out 2>&1; \
+			if test -e "$(srcroot)$${t}.exp"; then \
+				diff -w -u $(srcroot)$${t}.exp \
+				  $(objroot)$${t}.out >/dev/null 2>&1; \
+				fail=$$?; \
+				if test "$${fail}" -eq "1" ; then \
+					failures=`expr $${failures} + 1`; \
+					echo "*** FAIL ***"; \
+				else \
+					echo "pass"; \
+				fi; \
+			else \
+				echo "*** FAIL *** (.exp file is missing)"; \
+				failures=`expr $${failures} + 1`; \
+			fi; \
+		done; \
+		echo "========================================="; \
+		echo "Failures: $${failures}/$${total}"'
+
+clean:
+	rm -f $(COBJS)
+	rm -f $(CPICOBJS)
+	rm -f $(COBJS:%.$(O)=%.d)
+	rm -f $(CPICOBJS:%.$(O)=%.d)
+	rm -f $(CTESTOBJS:%.$(O)=%$(EXE))
+	rm -f $(CTESTOBJS)
+	rm -f $(CTESTOBJS:%.$(O)=%.d)
+	rm -f $(CTESTOBJS:%.$(O)=%.out)
+	rm -f $(DSOS) $(STATIC_LIBS)
+
+distclean: clean
+	rm -rf $(objroot)autom4te.cache
+	rm -f $(objroot)config.log
+	rm -f $(objroot)config.status
+	rm -f $(objroot)config.stamp
+	rm -f $(cfghdrs_out)
+	rm -f $(cfgoutputs_out)
+
+relclean: distclean
+	rm -f $(objroot)configure
+	rm -f $(srcroot)VERSION
+	rm -f $(DOCS_HTML)
+	rm -f $(DOCS_MAN3)
+
+#===============================================================================
+# Re-configuration rules.
+
+ifeq ($(enable_autogen), 1)
+$(srcroot)configure : $(srcroot)configure.ac
+	cd ./$(srcroot) && $(AUTOCONF)
+
+$(objroot)config.status : $(srcroot)configure
+	./$(objroot)config.status --recheck
+
+$(srcroot)config.stamp.in : $(srcroot)configure.ac
+	echo stamp > $(srcroot)config.stamp.in
+
+$(objroot)config.stamp : $(cfgoutputs_in) $(cfghdrs_in) $(srcroot)configure
+	./$(objroot)config.status
+	@touch $@
+
+# There must be some action in order for make to re-read Makefile when it is
+# out of date.
+$(cfgoutputs_out) $(cfghdrs_out) : $(objroot)config.stamp
+	@true
+endif
diff --git a/extra/jemalloc/README b/extra/jemalloc/README
new file mode 100644
index 00000000000..7661683bae7
--- /dev/null
+++ b/extra/jemalloc/README
@@ -0,0 +1,16 @@
+jemalloc is a general-purpose scalable concurrent malloc(3) implementation.
+This distribution is a "portable" implementation that currently targets
+FreeBSD, Linux, Apple OS X, and MinGW.  jemalloc is included as the default
+allocator in the FreeBSD and NetBSD operating systems, and it is used by the
+Mozilla Firefox web browser on Microsoft Windows-related platforms.  Depending
+on your needs, one of the other divergent versions may suit your needs better
+than this distribution.
+
+The COPYING file contains copyright and licensing information.
+
+The INSTALL file contains information on how to configure, build, and install
+jemalloc.
+
+The ChangeLog file contains a brief summary of changes for each release.
+
+URL: http://www.canonware.com/jemalloc/
diff --git a/extra/jemalloc/VERSION b/extra/jemalloc/VERSION
new file mode 100644
index 00000000000..900c82d1043
--- /dev/null
+++ b/extra/jemalloc/VERSION
@@ -0,0 +1 @@
+3.3.1-0-g9ef9d9e8c271cdf14f664b871a8f98c827714784
diff --git a/extra/jemalloc/autogen.sh b/extra/jemalloc/autogen.sh
new file mode 100755
index 00000000000..75f32da6873
--- /dev/null
+++ b/extra/jemalloc/autogen.sh
@@ -0,0 +1,17 @@
+#!/bin/sh
+
+for i in autoconf; do
+    echo "$i"
+    $i
+    if [ $? -ne 0 ]; then
+	echo "Error $? in $i"
+	exit 1
+    fi
+done
+
+echo "./configure --enable-autogen $@"
+./configure --enable-autogen $@
+if [ $? -ne 0 ]; then
+    echo "Error $? in ./configure"
+    exit 1
+fi
diff --git a/extra/jemalloc/bin/jemalloc.sh b/extra/jemalloc/bin/jemalloc.sh
new file mode 100644
index 00000000000..7c9f1b530da
--- /dev/null
+++ b/extra/jemalloc/bin/jemalloc.sh
@@ -0,0 +1,9 @@
+#!/bin/sh
+
+prefix=/usr/local
+exec_prefix=/usr/local
+libdir=${exec_prefix}/lib
+
+LD_PRELOAD=${libdir}/libjemalloc.so.1
+export LD_PRELOAD
+exec "$@"
diff --git a/extra/jemalloc/bin/jemalloc.sh.in b/extra/jemalloc/bin/jemalloc.sh.in
new file mode 100644
index 00000000000..cdf36737591
--- /dev/null
+++ b/extra/jemalloc/bin/jemalloc.sh.in
@@ -0,0 +1,9 @@
+#!/bin/sh
+
+prefix=@prefix@
+exec_prefix=@exec_prefix@
+libdir=@libdir@
+
+@LD_PRELOAD_VAR@=${libdir}/libjemalloc.@SOREV@
+export @LD_PRELOAD_VAR@
+exec "$@"
diff --git a/extra/jemalloc/bin/pprof b/extra/jemalloc/bin/pprof
new file mode 100755
index 00000000000..727eb43704f
--- /dev/null
+++ b/extra/jemalloc/bin/pprof
@@ -0,0 +1,5348 @@
+#! /usr/bin/env perl
+
+# Copyright (c) 1998-2007, Google Inc.
+# All rights reserved.
+# 
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+# 
+#     * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+#     * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+#     * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+# 
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# ---
+# Program for printing the profile generated by common/profiler.cc,
+# or by the heap profiler (common/debugallocation.cc)
+#
+# The profile contains a sequence of entries of the form:
+#       <count> <stack trace>
+# This program parses the profile, and generates user-readable
+# output.
+#
+# Examples:
+#
+# % tools/pprof "program" "profile"
+#   Enters "interactive" mode
+#
+# % tools/pprof --text "program" "profile"
+#   Generates one line per procedure
+#
+# % tools/pprof --gv "program" "profile"
+#   Generates annotated call-graph and displays via "gv"
+#
+# % tools/pprof --gv --focus=Mutex "program" "profile"
+#   Restrict to code paths that involve an entry that matches "Mutex"
+#
+# % tools/pprof --gv --focus=Mutex --ignore=string "program" "profile"
+#   Restrict to code paths that involve an entry that matches "Mutex"
+#   and does not match "string"
+#
+# % tools/pprof --list=IBF_CheckDocid "program" "profile"
+#   Generates disassembly listing of all routines with at least one
+#   sample that match the --list=<regexp> pattern.  The listing is
+#   annotated with the flat and cumulative sample counts at each line.
+#
+# % tools/pprof --disasm=IBF_CheckDocid "program" "profile"
+#   Generates disassembly listing of all routines with at least one
+#   sample that match the --disasm=<regexp> pattern.  The listing is
+#   annotated with the flat and cumulative sample counts at each PC value.
+#
+# TODO: Use color to indicate files?
+
+use strict;
+use warnings;
+use Getopt::Long;
+
+my $PPROF_VERSION = "2.0";
+
+# These are the object tools we use which can come from a
+# user-specified location using --tools, from the PPROF_TOOLS
+# environment variable, or from the environment.
+my %obj_tool_map = (
+  "objdump" => "objdump",
+  "nm" => "nm",
+  "addr2line" => "addr2line",
+  "c++filt" => "c++filt",
+  ## ConfigureObjTools may add architecture-specific entries:
+  #"nm_pdb" => "nm-pdb",       # for reading windows (PDB-format) executables
+  #"addr2line_pdb" => "addr2line-pdb",                                # ditto
+  #"otool" => "otool",         # equivalent of objdump on OS X
+);
+# NOTE: these are lists, so you can put in commandline flags if you want.
+my @DOT = ("dot");          # leave non-absolute, since it may be in /usr/local
+my @GV = ("gv");
+my @EVINCE = ("evince");    # could also be xpdf or perhaps acroread
+my @KCACHEGRIND = ("kcachegrind");
+my @PS2PDF = ("ps2pdf");
+# These are used for dynamic profiles
+my @URL_FETCHER = ("curl", "-s");
+
+# These are the web pages that servers need to support for dynamic profiles
+my $HEAP_PAGE = "/pprof/heap";
+my $PROFILE_PAGE = "/pprof/profile";   # must support cgi-param "?seconds=#"
+my $PMUPROFILE_PAGE = "/pprof/pmuprofile(?:\\?.*)?"; # must support cgi-param
+                                                # ?seconds=#&event=x&period=n
+my $GROWTH_PAGE = "/pprof/growth";
+my $CONTENTION_PAGE = "/pprof/contention";
+my $WALL_PAGE = "/pprof/wall(?:\\?.*)?";  # accepts options like namefilter
+my $FILTEREDPROFILE_PAGE = "/pprof/filteredprofile(?:\\?.*)?";
+my $CENSUSPROFILE_PAGE = "/pprof/censusprofile(?:\\?.*)?"; # must support cgi-param
+                                                       # "?seconds=#",
+                                                       # "?tags_regexp=#" and
+                                                       # "?type=#".
+my $SYMBOL_PAGE = "/pprof/symbol";     # must support symbol lookup via POST
+my $PROGRAM_NAME_PAGE = "/pprof/cmdline";
+
+# These are the web pages that can be named on the command line.
+# All the alternatives must begin with /.
+my $PROFILES = "($HEAP_PAGE|$PROFILE_PAGE|$PMUPROFILE_PAGE|" .
+               "$GROWTH_PAGE|$CONTENTION_PAGE|$WALL_PAGE|" .
+               "$FILTEREDPROFILE_PAGE|$CENSUSPROFILE_PAGE)";
+
+# default binary name
+my $UNKNOWN_BINARY = "(unknown)";
+
+# There is a pervasive dependency on the length (in hex characters,
+# i.e., nibbles) of an address, distinguishing between 32-bit and
+# 64-bit profiles.  To err on the safe size, default to 64-bit here:
+my $address_length = 16;
+
+my $dev_null = "/dev/null";
+if (! -e $dev_null && $^O =~ /MSWin/) {    # $^O is the OS perl was built for
+  $dev_null = "nul";
+}
+
+# A list of paths to search for shared object files
+my @prefix_list = ();
+
+# Special routine name that should not have any symbols.
+# Used as separator to parse "addr2line -i" output.
+my $sep_symbol = '_fini';
+my $sep_address = undef;
+
+##### Argument parsing #####
+
+sub usage_string {
+  return <<EOF;
+Usage:
+pprof [options] <program> <profiles>
+   <profiles> is a space separated list of profile names.
+pprof [options] <symbolized-profiles>
+   <symbolized-profiles> is a list of profile files where each file contains
+   the necessary symbol mappings  as well as profile data (likely generated
+   with --raw).
+pprof [options] <profile>
+   <profile> is a remote form.  Symbols are obtained from host:port$SYMBOL_PAGE
+
+   Each name can be:
+   /path/to/profile        - a path to a profile file
+   host:port[/<service>]   - a location of a service to get profile from
+
+   The /<service> can be $HEAP_PAGE, $PROFILE_PAGE, /pprof/pmuprofile,
+                         $GROWTH_PAGE, $CONTENTION_PAGE, /pprof/wall,
+                         $CENSUSPROFILE_PAGE, or /pprof/filteredprofile.
+   For instance:
+     pprof http://myserver.com:80$HEAP_PAGE
+   If /<service> is omitted, the service defaults to $PROFILE_PAGE (cpu profiling).
+pprof --symbols <program>
+   Maps addresses to symbol names.  In this mode, stdin should be a
+   list of library mappings, in the same format as is found in the heap-
+   and cpu-profile files (this loosely matches that of /proc/self/maps
+   on linux), followed by a list of hex addresses to map, one per line.
+
+   For more help with querying remote servers, including how to add the
+   necessary server-side support code, see this filename (or one like it):
+
+   /usr/doc/gperftools-$PPROF_VERSION/pprof_remote_servers.html
+
+Options:
+   --cum               Sort by cumulative data
+   --base=<base>       Subtract <base> from <profile> before display
+   --interactive       Run in interactive mode (interactive "help" gives help) [default]
+   --seconds=<n>       Length of time for dynamic profiles [default=30 secs]
+   --add_lib=<file>    Read additional symbols and line info from the given library
+   --lib_prefix=<dir>  Comma separated list of library path prefixes
+
+Reporting Granularity:
+   --addresses         Report at address level
+   --lines             Report at source line level
+   --functions         Report at function level [default]
+   --files             Report at source file level
+
+Output type:
+   --text              Generate text report
+   --callgrind         Generate callgrind format to stdout
+   --gv                Generate Postscript and display
+   --evince            Generate PDF and display
+   --web               Generate SVG and display
+   --list=<regexp>     Generate source listing of matching routines
+   --disasm=<regexp>   Generate disassembly of matching routines
+   --symbols           Print demangled symbol names found at given addresses
+   --dot               Generate DOT file to stdout
+   --ps                Generate Postcript to stdout
+   --pdf               Generate PDF to stdout
+   --svg               Generate SVG to stdout
+   --gif               Generate GIF to stdout
+   --raw               Generate symbolized pprof data (useful with remote fetch)
+
+Heap-Profile Options:
+   --inuse_space       Display in-use (mega)bytes [default]
+   --inuse_objects     Display in-use objects
+   --alloc_space       Display allocated (mega)bytes
+   --alloc_objects     Display allocated objects
+   --show_bytes        Display space in bytes
+   --drop_negative     Ignore negative differences
+
+Contention-profile options:
+   --total_delay       Display total delay at each region [default]
+   --contentions       Display number of delays at each region
+   --mean_delay        Display mean delay at each region
+
+Call-graph Options:
+   --nodecount=<n>     Show at most so many nodes [default=80]
+   --nodefraction=<f>  Hide nodes below <f>*total [default=.005]
+   --edgefraction=<f>  Hide edges below <f>*total [default=.001]
+   --maxdegree=<n>     Max incoming/outgoing edges per node [default=8]
+   --focus=<regexp>    Focus on nodes matching <regexp>
+   --ignore=<regexp>   Ignore nodes matching <regexp>
+   --scale=<n>         Set GV scaling [default=0]
+   --heapcheck         Make nodes with non-0 object counts
+                       (i.e. direct leak generators) more visible
+
+Miscellaneous:
+   --tools=<prefix or binary:fullpath>[,...]   \$PATH for object tool pathnames
+   --test              Run unit tests
+   --help              This message
+   --version           Version information
+
+Environment Variables:
+   PPROF_TMPDIR        Profiles directory. Defaults to \$HOME/pprof
+   PPROF_TOOLS         Prefix for object tools pathnames
+
+Examples:
+
+pprof /bin/ls ls.prof
+                       Enters "interactive" mode
+pprof --text /bin/ls ls.prof
+                       Outputs one line per procedure
+pprof --web /bin/ls ls.prof
+                       Displays annotated call-graph in web browser
+pprof --gv /bin/ls ls.prof
+                       Displays annotated call-graph via 'gv'
+pprof --gv --focus=Mutex /bin/ls ls.prof
+                       Restricts to code paths including a .*Mutex.* entry
+pprof --gv --focus=Mutex --ignore=string /bin/ls ls.prof
+                       Code paths including Mutex but not string
+pprof --list=getdir /bin/ls ls.prof
+                       (Per-line) annotated source listing for getdir()
+pprof --disasm=getdir /bin/ls ls.prof
+                       (Per-PC) annotated disassembly for getdir()
+
+pprof http://localhost:1234/
+                       Enters "interactive" mode
+pprof --text localhost:1234
+                       Outputs one line per procedure for localhost:1234
+pprof --raw localhost:1234 > ./local.raw
+pprof --text ./local.raw
+                       Fetches a remote profile for later analysis and then
+                       analyzes it in text mode.
+EOF
+}
+
+sub version_string {
+  return <<EOF
+pprof (part of gperftools $PPROF_VERSION)
+
+Copyright 1998-2007 Google Inc.
+
+This is BSD licensed software; see the source for copying conditions
+and license information.
+There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A
+PARTICULAR PURPOSE.
+EOF
+}
+
+sub usage {
+  my $msg = shift;
+  print STDERR "$msg\n\n";
+  print STDERR usage_string();
+  print STDERR "\nFATAL ERROR: $msg\n";    # just as a reminder
+  exit(1);
+}
+
+sub Init() {
+  # Setup tmp-file name and handler to clean it up.
+  # We do this in the very beginning so that we can use
+  # error() and cleanup() function anytime here after.
+  $main::tmpfile_sym = "/tmp/pprof$$.sym";
+  $main::tmpfile_ps = "/tmp/pprof$$";
+  $main::next_tmpfile = 0;
+  $SIG{'INT'} = \&sighandler;
+
+  # Cache from filename/linenumber to source code
+  $main::source_cache = ();
+
+  $main::opt_help = 0;
+  $main::opt_version = 0;
+
+  $main::opt_cum = 0;
+  $main::opt_base = '';
+  $main::opt_addresses = 0;
+  $main::opt_lines = 0;
+  $main::opt_functions = 0;
+  $main::opt_files = 0;
+  $main::opt_lib_prefix = "";
+
+  $main::opt_text = 0;
+  $main::opt_callgrind = 0;
+  $main::opt_list = "";
+  $main::opt_disasm = "";
+  $main::opt_symbols = 0;
+  $main::opt_gv = 0;
+  $main::opt_evince = 0;
+  $main::opt_web = 0;
+  $main::opt_dot = 0;
+  $main::opt_ps = 0;
+  $main::opt_pdf = 0;
+  $main::opt_gif = 0;
+  $main::opt_svg = 0;
+  $main::opt_raw = 0;
+
+  $main::opt_nodecount = 80;
+  $main::opt_nodefraction = 0.005;
+  $main::opt_edgefraction = 0.001;
+  $main::opt_maxdegree = 8;
+  $main::opt_focus = '';
+  $main::opt_ignore = '';
+  $main::opt_scale = 0;
+  $main::opt_heapcheck = 0;
+  $main::opt_seconds = 30;
+  $main::opt_lib = "";
+
+  $main::opt_inuse_space   = 0;
+  $main::opt_inuse_objects = 0;
+  $main::opt_alloc_space   = 0;
+  $main::opt_alloc_objects = 0;
+  $main::opt_show_bytes    = 0;
+  $main::opt_drop_negative = 0;
+  $main::opt_interactive   = 0;
+
+  $main::opt_total_delay = 0;
+  $main::opt_contentions = 0;
+  $main::opt_mean_delay = 0;
+
+  $main::opt_tools   = "";
+  $main::opt_debug   = 0;
+  $main::opt_test    = 0;
+
+  # These are undocumented flags used only by unittests.
+  $main::opt_test_stride = 0;
+
+  # Are we using $SYMBOL_PAGE?
+  $main::use_symbol_page = 0;
+
+  # Files returned by TempName.
+  %main::tempnames = ();
+
+  # Type of profile we are dealing with
+  # Supported types:
+  #     cpu
+  #     heap
+  #     growth
+  #     contention
+  $main::profile_type = '';     # Empty type means "unknown"
+
+  GetOptions("help!"          => \$main::opt_help,
+             "version!"       => \$main::opt_version,
+             "cum!"           => \$main::opt_cum,
+             "base=s"         => \$main::opt_base,
+             "seconds=i"      => \$main::opt_seconds,
+             "add_lib=s"      => \$main::opt_lib,
+             "lib_prefix=s"   => \$main::opt_lib_prefix,
+             "functions!"     => \$main::opt_functions,
+             "lines!"         => \$main::opt_lines,
+             "addresses!"     => \$main::opt_addresses,
+             "files!"         => \$main::opt_files,
+             "text!"          => \$main::opt_text,
+             "callgrind!"     => \$main::opt_callgrind,
+             "list=s"         => \$main::opt_list,
+             "disasm=s"       => \$main::opt_disasm,
+             "symbols!"       => \$main::opt_symbols,
+             "gv!"            => \$main::opt_gv,
+             "evince!"        => \$main::opt_evince,
+             "web!"           => \$main::opt_web,
+             "dot!"           => \$main::opt_dot,
+             "ps!"            => \$main::opt_ps,
+             "pdf!"           => \$main::opt_pdf,
+             "svg!"           => \$main::opt_svg,
+             "gif!"           => \$main::opt_gif,
+             "raw!"           => \$main::opt_raw,
+             "interactive!"   => \$main::opt_interactive,
+             "nodecount=i"    => \$main::opt_nodecount,
+             "nodefraction=f" => \$main::opt_nodefraction,
+             "edgefraction=f" => \$main::opt_edgefraction,
+             "maxdegree=i"    => \$main::opt_maxdegree,
+             "focus=s"        => \$main::opt_focus,
+             "ignore=s"       => \$main::opt_ignore,
+             "scale=i"        => \$main::opt_scale,
+             "heapcheck"      => \$main::opt_heapcheck,
+             "inuse_space!"   => \$main::opt_inuse_space,
+             "inuse_objects!" => \$main::opt_inuse_objects,
+             "alloc_space!"   => \$main::opt_alloc_space,
+             "alloc_objects!" => \$main::opt_alloc_objects,
+             "show_bytes!"    => \$main::opt_show_bytes,
+             "drop_negative!" => \$main::opt_drop_negative,
+             "total_delay!"   => \$main::opt_total_delay,
+             "contentions!"   => \$main::opt_contentions,
+             "mean_delay!"    => \$main::opt_mean_delay,
+             "tools=s"        => \$main::opt_tools,
+             "test!"          => \$main::opt_test,
+             "debug!"         => \$main::opt_debug,
+             # Undocumented flags used only by unittests:
+             "test_stride=i"  => \$main::opt_test_stride,
+      ) || usage("Invalid option(s)");
+
+  # Deal with the standard --help and --version
+  if ($main::opt_help) {
+    print usage_string();
+    exit(0);
+  }
+
+  if ($main::opt_version) {
+    print version_string();
+    exit(0);
+  }
+
+  # Disassembly/listing/symbols mode requires address-level info
+  if ($main::opt_disasm || $main::opt_list || $main::opt_symbols) {
+    $main::opt_functions = 0;
+    $main::opt_lines = 0;
+    $main::opt_addresses = 1;
+    $main::opt_files = 0;
+  }
+
+  # Check heap-profiling flags
+  if ($main::opt_inuse_space +
+      $main::opt_inuse_objects +
+      $main::opt_alloc_space +
+      $main::opt_alloc_objects > 1) {
+    usage("Specify at most on of --inuse/--alloc options");
+  }
+
+  # Check output granularities
+  my $grains =
+      $main::opt_functions +
+      $main::opt_lines +
+      $main::opt_addresses +
+      $main::opt_files +
+      0;
+  if ($grains > 1) {
+    usage("Only specify one output granularity option");
+  }
+  if ($grains == 0) {
+    $main::opt_functions = 1;
+  }
+
+  # Check output modes
+  my $modes =
+      $main::opt_text +
+      $main::opt_callgrind +
+      ($main::opt_list eq '' ? 0 : 1) +
+      ($main::opt_disasm eq '' ? 0 : 1) +
+      ($main::opt_symbols == 0 ? 0 : 1) +
+      $main::opt_gv +
+      $main::opt_evince +
+      $main::opt_web +
+      $main::opt_dot +
+      $main::opt_ps +
+      $main::opt_pdf +
+      $main::opt_svg +
+      $main::opt_gif +
+      $main::opt_raw +
+      $main::opt_interactive +
+      0;
+  if ($modes > 1) {
+    usage("Only specify one output mode");
+  }
+  if ($modes == 0) {
+    if (-t STDOUT) {  # If STDOUT is a tty, activate interactive mode
+      $main::opt_interactive = 1;
+    } else {
+      $main::opt_text = 1;
+    }
+  }
+
+  if ($main::opt_test) {
+    RunUnitTests();
+    # Should not return
+    exit(1);
+  }
+
+  # Binary name and profile arguments list
+  $main::prog = "";
+  @main::pfile_args = ();
+
+  # Remote profiling without a binary (using $SYMBOL_PAGE instead)
+  if (@ARGV > 0) {
+    if (IsProfileURL($ARGV[0])) {
+      $main::use_symbol_page = 1;
+    } elsif (IsSymbolizedProfileFile($ARGV[0])) {
+      $main::use_symbolized_profile = 1;
+      $main::prog = $UNKNOWN_BINARY;  # will be set later from the profile file
+    }
+  }
+
+  if ($main::use_symbol_page || $main::use_symbolized_profile) {
+    # We don't need a binary!
+    my %disabled = ('--lines' => $main::opt_lines,
+                    '--disasm' => $main::opt_disasm);
+    for my $option (keys %disabled) {
+      usage("$option cannot be used without a binary") if $disabled{$option};
+    }
+    # Set $main::prog later...
+    scalar(@ARGV) || usage("Did not specify profile file");
+  } elsif ($main::opt_symbols) {
+    # --symbols needs a binary-name (to run nm on, etc) but not profiles
+    $main::prog = shift(@ARGV) || usage("Did not specify program");
+  } else {
+    $main::prog = shift(@ARGV) || usage("Did not specify program");
+    scalar(@ARGV) || usage("Did not specify profile file");
+  }
+
+  # Parse profile file/location arguments
+  foreach my $farg (@ARGV) {
+    if ($farg =~ m/(.*)\@([0-9]+)(|\/.*)$/ ) {
+      my $machine = $1;
+      my $num_machines = $2;
+      my $path = $3;
+      for (my $i = 0; $i < $num_machines; $i++) {
+        unshift(@main::pfile_args, "$i.$machine$path");
+      }
+    } else {
+      unshift(@main::pfile_args, $farg);
+    }
+  }
+
+  if ($main::use_symbol_page) {
+    unless (IsProfileURL($main::pfile_args[0])) {
+      error("The first profile should be a remote form to use $SYMBOL_PAGE\n");
+    }
+    CheckSymbolPage();
+    $main::prog = FetchProgramName();
+  } elsif (!$main::use_symbolized_profile) {  # may not need objtools!
+    ConfigureObjTools($main::prog)
+  }
+
+  # Break the opt_lib_prefix into the prefix_list array
+  @prefix_list = split (',', $main::opt_lib_prefix);
+
+  # Remove trailing / from the prefixes, in the list to prevent
+  # searching things like /my/path//lib/mylib.so
+  foreach (@prefix_list) {
+    s|/+$||;
+  }
+}
+
+sub Main() {
+  Init();
+  $main::collected_profile = undef;
+  @main::profile_files = ();
+  $main::op_time = time();
+
+  # Printing symbols is special and requires a lot less info that most.
+  if ($main::opt_symbols) {
+    PrintSymbols(*STDIN);   # Get /proc/maps and symbols output from stdin
+    return;
+  }
+
+  # Fetch all profile data
+  FetchDynamicProfiles();
+
+  # this will hold symbols that we read from the profile files
+  my $symbol_map = {};
+
+  # Read one profile, pick the last item on the list
+  my $data = ReadProfile($main::prog, pop(@main::profile_files));
+  my $profile = $data->{profile};
+  my $pcs = $data->{pcs};
+  my $libs = $data->{libs};   # Info about main program and shared libraries
+  $symbol_map = MergeSymbols($symbol_map, $data->{symbols});
+
+  # Add additional profiles, if available.
+  if (scalar(@main::profile_files) > 0) {
+    foreach my $pname (@main::profile_files) {
+      my $data2 = ReadProfile($main::prog, $pname);
+      $profile = AddProfile($profile, $data2->{profile});
+      $pcs = AddPcs($pcs, $data2->{pcs});
+      $symbol_map = MergeSymbols($symbol_map, $data2->{symbols});
+    }
+  }
+
+  # Subtract base from profile, if specified
+  if ($main::opt_base ne '') {
+    my $base = ReadProfile($main::prog, $main::opt_base);
+    $profile = SubtractProfile($profile, $base->{profile});
+    $pcs = AddPcs($pcs, $base->{pcs});
+    $symbol_map = MergeSymbols($symbol_map, $base->{symbols});
+  }
+
+  # Get total data in profile
+  my $total = TotalProfile($profile);
+
+  # Collect symbols
+  my $symbols;
+  if ($main::use_symbolized_profile) {
+    $symbols = FetchSymbols($pcs, $symbol_map);
+  } elsif ($main::use_symbol_page) {
+    $symbols = FetchSymbols($pcs);
+  } else {
+    # TODO(csilvers): $libs uses the /proc/self/maps data from profile1,
+    # which may differ from the data from subsequent profiles, especially
+    # if they were run on different machines.  Use appropriate libs for
+    # each pc somehow.
+    $symbols = ExtractSymbols($libs, $pcs);
+  }
+
+  # Remove uniniteresting stack items
+  $profile = RemoveUninterestingFrames($symbols, $profile);
+
+  # Focus?
+  if ($main::opt_focus ne '') {
+    $profile = FocusProfile($symbols, $profile, $main::opt_focus);
+  }
+
+  # Ignore?
+  if ($main::opt_ignore ne '') {
+    $profile = IgnoreProfile($symbols, $profile, $main::opt_ignore);
+  }
+
+  my $calls = ExtractCalls($symbols, $profile);
+
+  # Reduce profiles to required output granularity, and also clean
+  # each stack trace so a given entry exists at most once.
+  my $reduced = ReduceProfile($symbols, $profile);
+
+  # Get derived profiles
+  my $flat = FlatProfile($reduced);
+  my $cumulative = CumulativeProfile($reduced);
+
+  # Print
+  if (!$main::opt_interactive) {
+    if ($main::opt_disasm) {
+      PrintDisassembly($libs, $flat, $cumulative, $main::opt_disasm);
+    } elsif ($main::opt_list) {
+      PrintListing($total, $libs, $flat, $cumulative, $main::opt_list, 0);
+    } elsif ($main::opt_text) {
+      # Make sure the output is empty when have nothing to report
+      # (only matters when --heapcheck is given but we must be
+      # compatible with old branches that did not pass --heapcheck always):
+      if ($total != 0) {
+        printf("Total: %s %s\n", Unparse($total), Units());
+      }
+      PrintText($symbols, $flat, $cumulative, -1);
+    } elsif ($main::opt_raw) {
+      PrintSymbolizedProfile($symbols, $profile, $main::prog);
+    } elsif ($main::opt_callgrind) {
+      PrintCallgrind($calls);
+    } else {
+      if (PrintDot($main::prog, $symbols, $profile, $flat, $cumulative, $total)) {
+        if ($main::opt_gv) {
+          RunGV(TempName($main::next_tmpfile, "ps"), "");
+        } elsif ($main::opt_evince) {
+          RunEvince(TempName($main::next_tmpfile, "pdf"), "");
+        } elsif ($main::opt_web) {
+          my $tmp = TempName($main::next_tmpfile, "svg");
+          RunWeb($tmp);
+          # The command we run might hand the file name off
+          # to an already running browser instance and then exit.
+          # Normally, we'd remove $tmp on exit (right now),
+          # but fork a child to remove $tmp a little later, so that the
+          # browser has time to load it first.
+          delete $main::tempnames{$tmp};
+          if (fork() == 0) {
+            sleep 5;
+            unlink($tmp);
+            exit(0);
+          }
+        }
+      } else {
+        cleanup();
+        exit(1);
+      }
+    }
+  } else {
+    InteractiveMode($profile, $symbols, $libs, $total);
+  }
+
+  cleanup();
+  exit(0);
+}
+
+##### Entry Point #####
+
+Main();
+
+# Temporary code to detect if we're running on a Goobuntu system.
+# These systems don't have the right stuff installed for the special
+# Readline libraries to work, so as a temporary workaround, we default
+# to using the normal stdio code, rather than the fancier readline-based
+# code
+sub ReadlineMightFail {
+  if (-e '/lib/libtermcap.so.2') {
+    return 0;  # libtermcap exists, so readline should be okay
+  } else {
+    return 1;
+  }
+}
+
+sub RunGV {
+  my $fname = shift;
+  my $bg = shift;       # "" or " &" if we should run in background
+  if (!system(ShellEscape(@GV, "--version") . " >$dev_null 2>&1")) {
+    # Options using double dash are supported by this gv version.
+    # Also, turn on noantialias to better handle bug in gv for
+    # postscript files with large dimensions.
+    # TODO: Maybe we should not pass the --noantialias flag
+    # if the gv version is known to work properly without the flag.
+    system(ShellEscape(@GV, "--scale=$main::opt_scale", "--noantialias", $fname)
+           . $bg);
+  } else {
+    # Old gv version - only supports options that use single dash.
+    print STDERR ShellEscape(@GV, "-scale", $main::opt_scale) . "\n";
+    system(ShellEscape(@GV, "-scale", "$main::opt_scale", $fname) . $bg);
+  }
+}
+
+sub RunEvince {
+  my $fname = shift;
+  my $bg = shift;       # "" or " &" if we should run in background
+  system(ShellEscape(@EVINCE, $fname) . $bg);
+}
+
+sub RunWeb {
+  my $fname = shift;
+  print STDERR "Loading web page file:///$fname\n";
+
+  if (`uname` =~ /Darwin/) {
+    # OS X: open will use standard preference for SVG files.
+    system("/usr/bin/open", $fname);
+    return;
+  }
+
+  # Some kind of Unix; try generic symlinks, then specific browsers.
+  # (Stop once we find one.)
+  # Works best if the browser is already running.
+  my @alt = (
+    "/etc/alternatives/gnome-www-browser",
+    "/etc/alternatives/x-www-browser",
+    "google-chrome",
+    "firefox",
+  );
+  foreach my $b (@alt) {
+    if (system($b, $fname) == 0) {
+      return;
+    }
+  }
+
+  print STDERR "Could not load web browser.\n";
+}
+
+sub RunKcachegrind {
+  my $fname = shift;
+  my $bg = shift;       # "" or " &" if we should run in background
+  print STDERR "Starting '@KCACHEGRIND " . $fname . $bg . "'\n";
+  system(ShellEscape(@KCACHEGRIND, $fname) . $bg);
+}
+
+
+##### Interactive helper routines #####
+
+sub InteractiveMode {
+  $| = 1;  # Make output unbuffered for interactive mode
+  my ($orig_profile, $symbols, $libs, $total) = @_;
+
+  print STDERR "Welcome to pprof!  For help, type 'help'.\n";
+
+  # Use ReadLine if it's installed and input comes from a console.
+  if ( -t STDIN &&
+       !ReadlineMightFail() &&
+       defined(eval {require Term::ReadLine}) ) {
+    my $term = new Term::ReadLine 'pprof';
+    while ( defined ($_ = $term->readline('(pprof) '))) {
+      $term->addhistory($_) if /\S/;
+      if (!InteractiveCommand($orig_profile, $symbols, $libs, $total, $_)) {
+        last;    # exit when we get an interactive command to quit
+      }
+    }
+  } else {       # don't have readline
+    while (1) {
+      print STDERR "(pprof) ";
+      $_ = <STDIN>;
+      last if ! defined $_ ;
+      s/\r//g;         # turn windows-looking lines into unix-looking lines
+
+      # Save some flags that might be reset by InteractiveCommand()
+      my $save_opt_lines = $main::opt_lines;
+
+      if (!InteractiveCommand($orig_profile, $symbols, $libs, $total, $_)) {
+        last;    # exit when we get an interactive command to quit
+      }
+
+      # Restore flags
+      $main::opt_lines = $save_opt_lines;
+    }
+  }
+}
+
+# Takes two args: orig profile, and command to run.
+# Returns 1 if we should keep going, or 0 if we were asked to quit
+sub InteractiveCommand {
+  my($orig_profile, $symbols, $libs, $total, $command) = @_;
+  $_ = $command;                # just to make future m//'s easier
+  if (!defined($_)) {
+    print STDERR "\n";
+    return 0;
+  }
+  if (m/^\s*quit/) {
+    return 0;
+  }
+  if (m/^\s*help/) {
+    InteractiveHelpMessage();
+    return 1;
+  }
+  # Clear all the mode options -- mode is controlled by "$command"
+  $main::opt_text = 0;
+  $main::opt_callgrind = 0;
+  $main::opt_disasm = 0;
+  $main::opt_list = 0;
+  $main::opt_gv = 0;
+  $main::opt_evince = 0;
+  $main::opt_cum = 0;
+
+  if (m/^\s*(text|top)(\d*)\s*(.*)/) {
+    $main::opt_text = 1;
+
+    my $line_limit = ($2 ne "") ? int($2) : 10;
+
+    my $routine;
+    my $ignore;
+    ($routine, $ignore) = ParseInteractiveArgs($3);
+
+    my $profile = ProcessProfile($total, $orig_profile, $symbols, "", $ignore);
+    my $reduced = ReduceProfile($symbols, $profile);
+
+    # Get derived profiles
+    my $flat = FlatProfile($reduced);
+    my $cumulative = CumulativeProfile($reduced);
+
+    PrintText($symbols, $flat, $cumulative, $line_limit);
+    return 1;
+  }
+  if (m/^\s*callgrind\s*([^ \n]*)/) {
+    $main::opt_callgrind = 1;
+
+    # Get derived profiles
+    my $calls = ExtractCalls($symbols, $orig_profile);
+    my $filename = $1;
+    if ( $1 eq '' ) {
+      $filename = TempName($main::next_tmpfile, "callgrind");
+    }
+    PrintCallgrind($calls, $filename);
+    if ( $1 eq '' ) {
+      RunKcachegrind($filename, " & ");
+      $main::next_tmpfile++;
+    }
+
+    return 1;
+  }
+  if (m/^\s*(web)?list\s*(.+)/) {
+    my $html = (defined($1) && ($1 eq "web"));
+    $main::opt_list = 1;
+
+    my $routine;
+    my $ignore;
+    ($routine, $ignore) = ParseInteractiveArgs($2);
+
+    my $profile = ProcessProfile($total, $orig_profile, $symbols, "", $ignore);
+    my $reduced = ReduceProfile($symbols, $profile);
+
+    # Get derived profiles
+    my $flat = FlatProfile($reduced);
+    my $cumulative = CumulativeProfile($reduced);
+
+    PrintListing($total, $libs, $flat, $cumulative, $routine, $html);
+    return 1;
+  }
+  if (m/^\s*disasm\s*(.+)/) {
+    $main::opt_disasm = 1;
+
+    my $routine;
+    my $ignore;
+    ($routine, $ignore) = ParseInteractiveArgs($1);
+
+    # Process current profile to account for various settings
+    my $profile = ProcessProfile($total, $orig_profile, $symbols, "", $ignore);
+    my $reduced = ReduceProfile($symbols, $profile);
+
+    # Get derived profiles
+    my $flat = FlatProfile($reduced);
+    my $cumulative = CumulativeProfile($reduced);
+
+    PrintDisassembly($libs, $flat, $cumulative, $routine);
+    return 1;
+  }
+  if (m/^\s*(gv|web|evince)\s*(.*)/) {
+    $main::opt_gv = 0;
+    $main::opt_evince = 0;
+    $main::opt_web = 0;
+    if ($1 eq "gv") {
+      $main::opt_gv = 1;
+    } elsif ($1 eq "evince") {
+      $main::opt_evince = 1;
+    } elsif ($1 eq "web") {
+      $main::opt_web = 1;
+    }
+
+    my $focus;
+    my $ignore;
+    ($focus, $ignore) = ParseInteractiveArgs($2);
+
+    # Process current profile to account for various settings
+    my $profile = ProcessProfile($total, $orig_profile, $symbols,
+                                 $focus, $ignore);
+    my $reduced = ReduceProfile($symbols, $profile);
+
+    # Get derived profiles
+    my $flat = FlatProfile($reduced);
+    my $cumulative = CumulativeProfile($reduced);
+
+    if (PrintDot($main::prog, $symbols, $profile, $flat, $cumulative, $total)) {
+      if ($main::opt_gv) {
+        RunGV(TempName($main::next_tmpfile, "ps"), " &");
+      } elsif ($main::opt_evince) {
+        RunEvince(TempName($main::next_tmpfile, "pdf"), " &");
+      } elsif ($main::opt_web) {
+        RunWeb(TempName($main::next_tmpfile, "svg"));
+      }
+      $main::next_tmpfile++;
+    }
+    return 1;
+  }
+  if (m/^\s*$/) {
+    return 1;
+  }
+  print STDERR "Unknown command: try 'help'.\n";
+  return 1;
+}
+
+
+sub ProcessProfile {
+  my $total_count = shift;
+  my $orig_profile = shift;
+  my $symbols = shift;
+  my $focus = shift;
+  my $ignore = shift;
+
+  # Process current profile to account for various settings
+  my $profile = $orig_profile;
+  printf("Total: %s %s\n", Unparse($total_count), Units());
+  if ($focus ne '') {
+    $profile = FocusProfile($symbols, $profile, $focus);
+    my $focus_count = TotalProfile($profile);
+    printf("After focusing on '%s': %s %s of %s (%0.1f%%)\n",
+           $focus,
+           Unparse($focus_count), Units(),
+           Unparse($total_count), ($focus_count*100.0) / $total_count);
+  }
+  if ($ignore ne '') {
+    $profile = IgnoreProfile($symbols, $profile, $ignore);
+    my $ignore_count = TotalProfile($profile);
+    printf("After ignoring '%s': %s %s of %s (%0.1f%%)\n",
+           $ignore,
+           Unparse($ignore_count), Units(),
+           Unparse($total_count),
+           ($ignore_count*100.0) / $total_count);
+  }
+
+  return $profile;
+}
+
+sub InteractiveHelpMessage {
+  print STDERR <<ENDOFHELP;
+Interactive pprof mode
+
+Commands:
+  gv
+  gv [focus] [-ignore1] [-ignore2]
+      Show graphical hierarchical display of current profile.  Without
+      any arguments, shows all samples in the profile.  With the optional
+      "focus" argument, restricts the samples shown to just those where
+      the "focus" regular expression matches a routine name on the stack
+      trace.
+
+  web
+  web [focus] [-ignore1] [-ignore2]
+      Like GV, but displays profile in your web browser instead of using
+      Ghostview. Works best if your web browser is already running.
+      To change the browser that gets used:
+      On Linux, set the /etc/alternatives/gnome-www-browser symlink.
+      On OS X, change the Finder association for SVG files.
+
+  list [routine_regexp] [-ignore1] [-ignore2]
+      Show source listing of routines whose names match "routine_regexp"
+
+  weblist [routine_regexp] [-ignore1] [-ignore2]
+     Displays a source listing of routines whose names match "routine_regexp"
+     in a web browser.  You can click on source lines to view the
+     corresponding disassembly.
+
+  top [--cum] [-ignore1] [-ignore2]
+  top20 [--cum] [-ignore1] [-ignore2]
+  top37 [--cum] [-ignore1] [-ignore2]
+      Show top lines ordered by flat profile count, or cumulative count
+      if --cum is specified.  If a number is present after 'top', the
+      top K routines will be shown (defaults to showing the top 10)
+
+  disasm [routine_regexp] [-ignore1] [-ignore2]
+      Show disassembly of routines whose names match "routine_regexp",
+      annotated with sample counts.
+
+  callgrind
+  callgrind [filename]
+      Generates callgrind file. If no filename is given, kcachegrind is called.
+
+  help - This listing
+  quit or ^D - End pprof
+
+For commands that accept optional -ignore tags, samples where any routine in
+the stack trace matches the regular expression in any of the -ignore
+parameters will be ignored.
+
+Further pprof details are available at this location (or one similar):
+
+ /usr/doc/gperftools-$PPROF_VERSION/cpu_profiler.html
+ /usr/doc/gperftools-$PPROF_VERSION/heap_profiler.html
+
+ENDOFHELP
+}
+sub ParseInteractiveArgs {
+  my $args = shift;
+  my $focus = "";
+  my $ignore = "";
+  my @x = split(/ +/, $args);
+  foreach $a (@x) {
+    if ($a =~ m/^(--|-)lines$/) {
+      $main::opt_lines = 1;
+    } elsif ($a =~ m/^(--|-)cum$/) {
+      $main::opt_cum = 1;
+    } elsif ($a =~ m/^-(.*)/) {
+      $ignore .= (($ignore ne "") ? "|" : "" ) . $1;
+    } else {
+      $focus .= (($focus ne "") ? "|" : "" ) . $a;
+    }
+  }
+  if ($ignore ne "") {
+    print STDERR "Ignoring samples in call stacks that match '$ignore'\n";
+  }
+  return ($focus, $ignore);
+}
+
+##### Output code #####
+
+sub TempName {
+  my $fnum = shift;
+  my $ext = shift;
+  my $file = "$main::tmpfile_ps.$fnum.$ext";
+  $main::tempnames{$file} = 1;
+  return $file;
+}
+
+# Print profile data in packed binary format (64-bit) to standard out
+sub PrintProfileData {
+  my $profile = shift;
+
+  # print header (64-bit style)
+  # (zero) (header-size) (version) (sample-period) (zero)
+  print pack('L*', 0, 0, 3, 0, 0, 0, 1, 0, 0, 0);
+
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    if ($#addrs >= 0) {
+      my $depth = $#addrs + 1;
+      # int(foo / 2**32) is the only reliable way to get rid of bottom
+      # 32 bits on both 32- and 64-bit systems.
+      print pack('L*', $count & 0xFFFFFFFF, int($count / 2**32));
+      print pack('L*', $depth & 0xFFFFFFFF, int($depth / 2**32));
+
+      foreach my $full_addr (@addrs) {
+        my $addr = $full_addr;
+        $addr =~ s/0x0*//;  # strip off leading 0x, zeroes
+        if (length($addr) > 16) {
+          print STDERR "Invalid address in profile: $full_addr\n";
+          next;
+        }
+        my $low_addr = substr($addr, -8);       # get last 8 hex chars
+        my $high_addr = substr($addr, -16, 8);  # get up to 8 more hex chars
+        print pack('L*', hex('0x' . $low_addr), hex('0x' . $high_addr));
+      }
+    }
+  }
+}
+
+# Print symbols and profile data
+sub PrintSymbolizedProfile {
+  my $symbols = shift;
+  my $profile = shift;
+  my $prog = shift;
+
+  $SYMBOL_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $symbol_marker = $&;
+
+  print '--- ', $symbol_marker, "\n";
+  if (defined($prog)) {
+    print 'binary=', $prog, "\n";
+  }
+  while (my ($pc, $name) = each(%{$symbols})) {
+    my $sep = ' ';
+    print '0x', $pc;
+    # We have a list of function names, which include the inlined
+    # calls.  They are separated (and terminated) by --, which is
+    # illegal in function names.
+    for (my $j = 2; $j <= $#{$name}; $j += 3) {
+      print $sep, $name->[$j];
+      $sep = '--';
+    }
+    print "\n";
+  }
+  print '---', "\n";
+
+  $PROFILE_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $profile_marker = $&;
+  print '--- ', $profile_marker, "\n";
+  if (defined($main::collected_profile)) {
+    # if used with remote fetch, simply dump the collected profile to output.
+    open(SRC, "<$main::collected_profile");
+    while (<SRC>) {
+      print $_;
+    }
+    close(SRC);
+  } else {
+    # dump a cpu-format profile to standard out
+    PrintProfileData($profile);
+  }
+}
+
+# Print text output
+sub PrintText {
+  my $symbols = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $line_limit = shift;
+
+  my $total = TotalProfile($flat);
+
+  # Which profile to sort by?
+  my $s = $main::opt_cum ? $cumulative : $flat;
+
+  my $running_sum = 0;
+  my $lines = 0;
+  foreach my $k (sort { GetEntry($s, $b) <=> GetEntry($s, $a) || $a cmp $b }
+                 keys(%{$cumulative})) {
+    my $f = GetEntry($flat, $k);
+    my $c = GetEntry($cumulative, $k);
+    $running_sum += $f;
+
+    my $sym = $k;
+    if (exists($symbols->{$k})) {
+      $sym = $symbols->{$k}->[0] . " " . $symbols->{$k}->[1];
+      if ($main::opt_addresses) {
+        $sym = $k . " " . $sym;
+      }
+    }
+
+    if ($f != 0 || $c != 0) {
+      printf("%8s %6s %6s %8s %6s %s\n",
+             Unparse($f),
+             Percent($f, $total),
+             Percent($running_sum, $total),
+             Unparse($c),
+             Percent($c, $total),
+             $sym);
+    }
+    $lines++;
+    last if ($line_limit >= 0 && $lines >= $line_limit);
+  }
+}
+
+# Callgrind format has a compression for repeated function and file
+# names.  You show the name the first time, and just use its number
+# subsequently.  This can cut down the file to about a third or a
+# quarter of its uncompressed size.  $key and $val are the key/value
+# pair that would normally be printed by callgrind; $map is a map from
+# value to number.
+sub CompressedCGName {
+  my($key, $val, $map) = @_;
+  my $idx = $map->{$val};
+  # For very short keys, providing an index hurts rather than helps.
+  if (length($val) <= 3) {
+    return "$key=$val\n";
+  } elsif (defined($idx)) {
+    return "$key=($idx)\n";
+  } else {
+    # scalar(keys $map) gives the number of items in the map.
+    $idx = scalar(keys(%{$map})) + 1;
+    $map->{$val} = $idx;
+    return "$key=($idx) $val\n";
+  }
+}
+
+# Print the call graph in a way that's suiteable for callgrind.
+sub PrintCallgrind {
+  my $calls = shift;
+  my $filename;
+  my %filename_to_index_map;
+  my %fnname_to_index_map;
+
+  if ($main::opt_interactive) {
+    $filename = shift;
+    print STDERR "Writing callgrind file to '$filename'.\n"
+  } else {
+    $filename = "&STDOUT";
+  }
+  open(CG, ">$filename");
+  printf CG ("events: Hits\n\n");
+  foreach my $call ( map { $_->[0] }
+                     sort { $a->[1] cmp $b ->[1] ||
+                            $a->[2] <=> $b->[2] }
+                     map { /([^:]+):(\d+):([^ ]+)( -> ([^:]+):(\d+):(.+))?/;
+                           [$_, $1, $2] }
+                     keys %$calls ) {
+    my $count = int($calls->{$call});
+    $call =~ /([^:]+):(\d+):([^ ]+)( -> ([^:]+):(\d+):(.+))?/;
+    my ( $caller_file, $caller_line, $caller_function,
+         $callee_file, $callee_line, $callee_function ) =
+       ( $1, $2, $3, $5, $6, $7 );
+
+    # TODO(csilvers): for better compression, collect all the
+    # caller/callee_files and functions first, before printing
+    # anything, and only compress those referenced more than once.
+    printf CG CompressedCGName("fl", $caller_file, \%filename_to_index_map);
+    printf CG CompressedCGName("fn", $caller_function, \%fnname_to_index_map);
+    if (defined $6) {
+      printf CG CompressedCGName("cfl", $callee_file, \%filename_to_index_map);
+      printf CG CompressedCGName("cfn", $callee_function, \%fnname_to_index_map);
+      printf CG ("calls=$count $callee_line\n");
+    }
+    printf CG ("$caller_line $count\n\n");
+  }
+}
+
+# Print disassembly for all all routines that match $main::opt_disasm
+sub PrintDisassembly {
+  my $libs = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $disasm_opts = shift;
+
+  my $total = TotalProfile($flat);
+
+  foreach my $lib (@{$libs}) {
+    my $symbol_table = GetProcedureBoundaries($lib->[0], $disasm_opts);
+    my $offset = AddressSub($lib->[1], $lib->[3]);
+    foreach my $routine (sort ByName keys(%{$symbol_table})) {
+      my $start_addr = $symbol_table->{$routine}->[0];
+      my $end_addr = $symbol_table->{$routine}->[1];
+      # See if there are any samples in this routine
+      my $length = hex(AddressSub($end_addr, $start_addr));
+      my $addr = AddressAdd($start_addr, $offset);
+      for (my $i = 0; $i < $length; $i++) {
+        if (defined($cumulative->{$addr})) {
+          PrintDisassembledFunction($lib->[0], $offset,
+                                    $routine, $flat, $cumulative,
+                                    $start_addr, $end_addr, $total);
+          last;
+        }
+        $addr = AddressInc($addr);
+      }
+    }
+  }
+}
+
+# Return reference to array of tuples of the form:
+#       [start_address, filename, linenumber, instruction, limit_address]
+# E.g.,
+#       ["0x806c43d", "/foo/bar.cc", 131, "ret", "0x806c440"]
+sub Disassemble {
+  my $prog = shift;
+  my $offset = shift;
+  my $start_addr = shift;
+  my $end_addr = shift;
+
+  my $objdump = $obj_tool_map{"objdump"};
+  my $cmd = ShellEscape($objdump, "-C", "-d", "-l", "--no-show-raw-insn",
+                        "--start-address=0x$start_addr",
+                        "--stop-address=0x$end_addr", $prog);
+  open(OBJDUMP, "$cmd |") || error("$cmd: $!\n");
+  my @result = ();
+  my $filename = "";
+  my $linenumber = -1;
+  my $last = ["", "", "", ""];
+  while (<OBJDUMP>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    chop;
+    if (m|\s*([^:\s]+):(\d+)\s*$|) {
+      # Location line of the form:
+      #   <filename>:<linenumber>
+      $filename = $1;
+      $linenumber = $2;
+    } elsif (m/^ +([0-9a-f]+):\s*(.*)/) {
+      # Disassembly line -- zero-extend address to full length
+      my $addr = HexExtend($1);
+      my $k = AddressAdd($addr, $offset);
+      $last->[4] = $k;   # Store ending address for previous instruction
+      $last = [$k, $filename, $linenumber, $2, $end_addr];
+      push(@result, $last);
+    }
+  }
+  close(OBJDUMP);
+  return @result;
+}
+
+# The input file should contain lines of the form /proc/maps-like
+# output (same format as expected from the profiles) or that looks
+# like hex addresses (like "0xDEADBEEF").  We will parse all
+# /proc/maps output, and for all the hex addresses, we will output
+# "short" symbol names, one per line, in the same order as the input.
+sub PrintSymbols {
+  my $maps_and_symbols_file = shift;
+
+  # ParseLibraries expects pcs to be in a set.  Fine by us...
+  my @pclist = ();   # pcs in sorted order
+  my $pcs = {};
+  my $map = "";
+  foreach my $line (<$maps_and_symbols_file>) {
+    $line =~ s/\r//g;    # turn windows-looking lines into unix-looking lines
+    if ($line =~ /\b(0x[0-9a-f]+)\b/i) {
+      push(@pclist, HexExtend($1));
+      $pcs->{$pclist[-1]} = 1;
+    } else {
+      $map .= $line;
+    }
+  }
+
+  my $libs = ParseLibraries($main::prog, $map, $pcs);
+  my $symbols = ExtractSymbols($libs, $pcs);
+
+  foreach my $pc (@pclist) {
+    # ->[0] is the shortname, ->[2] is the full name
+    print(($symbols->{$pc}->[0] || "??") . "\n");
+  }
+}
+
+
+# For sorting functions by name
+sub ByName {
+  return ShortFunctionName($a) cmp ShortFunctionName($b);
+}
+
+# Print source-listing for all all routines that match $list_opts
+sub PrintListing {
+  my $total = shift;
+  my $libs = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $list_opts = shift;
+  my $html = shift;
+
+  my $output = \*STDOUT;
+  my $fname = "";
+
+  if ($html) {
+    # Arrange to write the output to a temporary file
+    $fname = TempName($main::next_tmpfile, "html");
+    $main::next_tmpfile++;
+    if (!open(TEMP, ">$fname")) {
+      print STDERR "$fname: $!\n";
+      return;
+    }
+    $output = \*TEMP;
+    print $output HtmlListingHeader();
+    printf $output ("<div class=\"legend\">%s<br>Total: %s %s</div>\n",
+                    $main::prog, Unparse($total), Units());
+  }
+
+  my $listed = 0;
+  foreach my $lib (@{$libs}) {
+    my $symbol_table = GetProcedureBoundaries($lib->[0], $list_opts);
+    my $offset = AddressSub($lib->[1], $lib->[3]);
+    foreach my $routine (sort ByName keys(%{$symbol_table})) {
+      # Print if there are any samples in this routine
+      my $start_addr = $symbol_table->{$routine}->[0];
+      my $end_addr = $symbol_table->{$routine}->[1];
+      my $length = hex(AddressSub($end_addr, $start_addr));
+      my $addr = AddressAdd($start_addr, $offset);
+      for (my $i = 0; $i < $length; $i++) {
+        if (defined($cumulative->{$addr})) {
+          $listed += PrintSource(
+            $lib->[0], $offset,
+            $routine, $flat, $cumulative,
+            $start_addr, $end_addr,
+            $html,
+            $output);
+          last;
+        }
+        $addr = AddressInc($addr);
+      }
+    }
+  }
+
+  if ($html) {
+    if ($listed > 0) {
+      print $output HtmlListingFooter();
+      close($output);
+      RunWeb($fname);
+    } else {
+      close($output);
+      unlink($fname);
+    }
+  }
+}
+
+sub HtmlListingHeader {
+  return <<'EOF';
+<DOCTYPE html>
+<html>
+<head>
+<title>Pprof listing</title>
+<style type="text/css">
+body {
+  font-family: sans-serif;
+}
+h1 {
+  font-size: 1.5em;
+  margin-bottom: 4px;
+}
+.legend {
+  font-size: 1.25em;
+}
+.line {
+  color: #aaaaaa;
+}
+.nop {
+  color: #aaaaaa;
+}
+.unimportant {
+  color: #cccccc;
+}
+.disasmloc {
+  color: #000000;
+}
+.deadsrc {
+  cursor: pointer;
+}
+.deadsrc:hover {
+  background-color: #eeeeee;
+}
+.livesrc {
+  color: #0000ff;
+  cursor: pointer;
+}
+.livesrc:hover {
+  background-color: #eeeeee;
+}
+.asm {
+  color: #008800;
+  display: none;
+}
+</style>
+<script type="text/javascript">
+function pprof_toggle_asm(e) {
+  var target;
+  if (!e) e = window.event;
+  if (e.target) target = e.target;
+  else if (e.srcElement) target = e.srcElement;
+
+  if (target) {
+    var asm = target.nextSibling;
+    if (asm && asm.className == "asm") {
+      asm.style.display = (asm.style.display == "block" ? "" : "block");
+      e.preventDefault();
+      return false;
+    }
+  }
+}
+</script>
+</head>
+<body>
+EOF
+}
+
+sub HtmlListingFooter {
+  return <<'EOF';
+</body>
+</html>
+EOF
+}
+
+sub HtmlEscape {
+  my $text = shift;
+  $text =~ s/&/&amp;/g;
+  $text =~ s/</&lt;/g;
+  $text =~ s/>/&gt;/g;
+  return $text;
+}
+
+# Returns the indentation of the line, if it has any non-whitespace
+# characters.  Otherwise, returns -1.
+sub Indentation {
+  my $line = shift;
+  if (m/^(\s*)\S/) {
+    return length($1);
+  } else {
+    return -1;
+  }
+}
+
+# If the symbol table contains inlining info, Disassemble() may tag an
+# instruction with a location inside an inlined function.  But for
+# source listings, we prefer to use the location in the function we
+# are listing.  So use MapToSymbols() to fetch full location
+# information for each instruction and then pick out the first
+# location from a location list (location list contains callers before
+# callees in case of inlining).
+#
+# After this routine has run, each entry in $instructions contains:
+#   [0] start address
+#   [1] filename for function we are listing
+#   [2] line number for function we are listing
+#   [3] disassembly
+#   [4] limit address
+#   [5] most specific filename (may be different from [1] due to inlining)
+#   [6] most specific line number (may be different from [2] due to inlining)
+sub GetTopLevelLineNumbers {
+  my ($lib, $offset, $instructions) = @_;
+  my $pcs = [];
+  for (my $i = 0; $i <= $#{$instructions}; $i++) {
+    push(@{$pcs}, $instructions->[$i]->[0]);
+  }
+  my $symbols = {};
+  MapToSymbols($lib, $offset, $pcs, $symbols);
+  for (my $i = 0; $i <= $#{$instructions}; $i++) {
+    my $e = $instructions->[$i];
+    push(@{$e}, $e->[1]);
+    push(@{$e}, $e->[2]);
+    my $addr = $e->[0];
+    my $sym = $symbols->{$addr};
+    if (defined($sym)) {
+      if ($#{$sym} >= 2 && $sym->[1] =~ m/^(.*):(\d+)$/) {
+        $e->[1] = $1;  # File name
+        $e->[2] = $2;  # Line number
+      }
+    }
+  }
+}
+
+# Print source-listing for one routine
+sub PrintSource {
+  my $prog = shift;
+  my $offset = shift;
+  my $routine = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $start_addr = shift;
+  my $end_addr = shift;
+  my $html = shift;
+  my $output = shift;
+
+  # Disassemble all instructions (just to get line numbers)
+  my @instructions = Disassemble($prog, $offset, $start_addr, $end_addr);
+  GetTopLevelLineNumbers($prog, $offset, \@instructions);
+
+  # Hack 1: assume that the first source file encountered in the
+  # disassembly contains the routine
+  my $filename = undef;
+  for (my $i = 0; $i <= $#instructions; $i++) {
+    if ($instructions[$i]->[2] >= 0) {
+      $filename = $instructions[$i]->[1];
+      last;
+    }
+  }
+  if (!defined($filename)) {
+    print STDERR "no filename found in $routine\n";
+    return 0;
+  }
+
+  # Hack 2: assume that the largest line number from $filename is the
+  # end of the procedure.  This is typically safe since if P1 contains
+  # an inlined call to P2, then P2 usually occurs earlier in the
+  # source file.  If this does not work, we might have to compute a
+  # density profile or just print all regions we find.
+  my $lastline = 0;
+  for (my $i = 0; $i <= $#instructions; $i++) {
+    my $f = $instructions[$i]->[1];
+    my $l = $instructions[$i]->[2];
+    if (($f eq $filename) && ($l > $lastline)) {
+      $lastline = $l;
+    }
+  }
+
+  # Hack 3: assume the first source location from "filename" is the start of
+  # the source code.
+  my $firstline = 1;
+  for (my $i = 0; $i <= $#instructions; $i++) {
+    if ($instructions[$i]->[1] eq $filename) {
+      $firstline = $instructions[$i]->[2];
+      last;
+    }
+  }
+
+  # Hack 4: Extend last line forward until its indentation is less than
+  # the indentation we saw on $firstline
+  my $oldlastline = $lastline;
+  {
+    if (!open(FILE, "<$filename")) {
+      print STDERR "$filename: $!\n";
+      return 0;
+    }
+    my $l = 0;
+    my $first_indentation = -1;
+    while (<FILE>) {
+      s/\r//g;         # turn windows-looking lines into unix-looking lines
+      $l++;
+      my $indent = Indentation($_);
+      if ($l >= $firstline) {
+        if ($first_indentation < 0 && $indent >= 0) {
+          $first_indentation = $indent;
+          last if ($first_indentation == 0);
+        }
+      }
+      if ($l >= $lastline && $indent >= 0) {
+        if ($indent >= $first_indentation) {
+          $lastline = $l+1;
+        } else {
+          last;
+        }
+      }
+    }
+    close(FILE);
+  }
+
+  # Assign all samples to the range $firstline,$lastline,
+  # Hack 4: If an instruction does not occur in the range, its samples
+  # are moved to the next instruction that occurs in the range.
+  my $samples1 = {};        # Map from line number to flat count
+  my $samples2 = {};        # Map from line number to cumulative count
+  my $running1 = 0;         # Unassigned flat counts
+  my $running2 = 0;         # Unassigned cumulative counts
+  my $total1 = 0;           # Total flat counts
+  my $total2 = 0;           # Total cumulative counts
+  my %disasm = ();          # Map from line number to disassembly
+  my $running_disasm = "";  # Unassigned disassembly
+  my $skip_marker = "---\n";
+  if ($html) {
+    $skip_marker = "";
+    for (my $l = $firstline; $l <= $lastline; $l++) {
+      $disasm{$l} = "";
+    }
+  }
+  my $last_dis_filename = '';
+  my $last_dis_linenum = -1;
+  my $last_touched_line = -1;  # To detect gaps in disassembly for a line
+  foreach my $e (@instructions) {
+    # Add up counts for all address that fall inside this instruction
+    my $c1 = 0;
+    my $c2 = 0;
+    for (my $a = $e->[0]; $a lt $e->[4]; $a = AddressInc($a)) {
+      $c1 += GetEntry($flat, $a);
+      $c2 += GetEntry($cumulative, $a);
+    }
+
+    if ($html) {
+      my $dis = sprintf("      %6s %6s \t\t%8s: %s ",
+                        HtmlPrintNumber($c1),
+                        HtmlPrintNumber($c2),
+                        UnparseAddress($offset, $e->[0]),
+                        CleanDisassembly($e->[3]));
+      
+      # Append the most specific source line associated with this instruction
+      if (length($dis) < 80) { $dis .= (' ' x (80 - length($dis))) };
+      $dis = HtmlEscape($dis);
+      my $f = $e->[5];
+      my $l = $e->[6];
+      if ($f ne $last_dis_filename) {
+        $dis .= sprintf("<span class=disasmloc>%s:%d</span>", 
+                        HtmlEscape(CleanFileName($f)), $l);
+      } elsif ($l ne $last_dis_linenum) {
+        # De-emphasize the unchanged file name portion
+        $dis .= sprintf("<span class=unimportant>%s</span>" .
+                        "<span class=disasmloc>:%d</span>", 
+                        HtmlEscape(CleanFileName($f)), $l);
+      } else {
+        # De-emphasize the entire location
+        $dis .= sprintf("<span class=unimportant>%s:%d</span>", 
+                        HtmlEscape(CleanFileName($f)), $l);
+      }
+      $last_dis_filename = $f;
+      $last_dis_linenum = $l;
+      $running_disasm .= $dis;
+      $running_disasm .= "\n";
+    }
+
+    $running1 += $c1;
+    $running2 += $c2;
+    $total1 += $c1;
+    $total2 += $c2;
+    my $file = $e->[1];
+    my $line = $e->[2];
+    if (($file eq $filename) &&
+        ($line >= $firstline) &&
+        ($line <= $lastline)) {
+      # Assign all accumulated samples to this line
+      AddEntry($samples1, $line, $running1);
+      AddEntry($samples2, $line, $running2);
+      $running1 = 0;
+      $running2 = 0;
+      if ($html) {
+        if ($line != $last_touched_line && $disasm{$line} ne '') {
+          $disasm{$line} .= "\n";
+        }
+        $disasm{$line} .= $running_disasm;
+        $running_disasm = '';
+        $last_touched_line = $line;
+      }
+    }
+  }
+
+  # Assign any leftover samples to $lastline
+  AddEntry($samples1, $lastline, $running1);
+  AddEntry($samples2, $lastline, $running2);
+  if ($html) {
+    if ($lastline != $last_touched_line && $disasm{$lastline} ne '') {
+      $disasm{$lastline} .= "\n";
+    }
+    $disasm{$lastline} .= $running_disasm;
+  }
+
+  if ($html) {
+    printf $output (
+      "<h1>%s</h1>%s\n<pre onClick=\"pprof_toggle_asm()\">\n" .
+      "Total:%6s %6s (flat / cumulative %s)\n",
+      HtmlEscape(ShortFunctionName($routine)),
+      HtmlEscape(CleanFileName($filename)),
+      Unparse($total1),
+      Unparse($total2),
+      Units());
+  } else {
+    printf $output (
+      "ROUTINE ====================== %s in %s\n" .
+      "%6s %6s Total %s (flat / cumulative)\n",
+      ShortFunctionName($routine),
+      CleanFileName($filename),
+      Unparse($total1),
+      Unparse($total2),
+      Units());
+  }
+  if (!open(FILE, "<$filename")) {
+    print STDERR "$filename: $!\n";
+    return 0;
+  }
+  my $l = 0;
+  while (<FILE>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    $l++;
+    if ($l >= $firstline - 5 &&
+        (($l <= $oldlastline + 5) || ($l <= $lastline))) {
+      chop;
+      my $text = $_;
+      if ($l == $firstline) { print $output $skip_marker; }
+      my $n1 = GetEntry($samples1, $l);
+      my $n2 = GetEntry($samples2, $l);
+      if ($html) {
+        # Emit a span that has one of the following classes:
+        #    livesrc -- has samples
+        #    deadsrc -- has disassembly, but with no samples
+        #    nop     -- has no matching disasembly
+        # Also emit an optional span containing disassembly.
+        my $dis = $disasm{$l};
+        my $asm = "";
+        if (defined($dis) && $dis ne '') {
+          $asm = "<span class=\"asm\">" . $dis . "</span>";
+        }
+        my $source_class = (($n1 + $n2 > 0) 
+                            ? "livesrc" 
+                            : (($asm ne "") ? "deadsrc" : "nop"));
+        printf $output (
+          "<span class=\"line\">%5d</span> " .
+          "<span class=\"%s\">%6s %6s %s</span>%s\n",
+          $l, $source_class,
+          HtmlPrintNumber($n1),
+          HtmlPrintNumber($n2),
+          HtmlEscape($text),
+          $asm);
+      } else {
+        printf $output(
+          "%6s %6s %4d: %s\n",
+          UnparseAlt($n1),
+          UnparseAlt($n2),
+          $l,
+          $text);
+      }
+      if ($l == $lastline)  { print $output $skip_marker; }
+    };
+  }
+  close(FILE);
+  if ($html) {
+    print $output "</pre>\n";
+  }
+  return 1;
+}
+
+# Return the source line for the specified file/linenumber.
+# Returns undef if not found.
+sub SourceLine {
+  my $file = shift;
+  my $line = shift;
+
+  # Look in cache
+  if (!defined($main::source_cache{$file})) {
+    if (100 < scalar keys(%main::source_cache)) {
+      # Clear the cache when it gets too big
+      $main::source_cache = ();
+    }
+
+    # Read all lines from the file
+    if (!open(FILE, "<$file")) {
+      print STDERR "$file: $!\n";
+      $main::source_cache{$file} = [];  # Cache the negative result
+      return undef;
+    }
+    my $lines = [];
+    push(@{$lines}, "");        # So we can use 1-based line numbers as indices
+    while (<FILE>) {
+      push(@{$lines}, $_);
+    }
+    close(FILE);
+
+    # Save the lines in the cache
+    $main::source_cache{$file} = $lines;
+  }
+
+  my $lines = $main::source_cache{$file};
+  if (($line < 0) || ($line > $#{$lines})) {
+    return undef;
+  } else {
+    return $lines->[$line];
+  }
+}
+
+# Print disassembly for one routine with interspersed source if available
+sub PrintDisassembledFunction {
+  my $prog = shift;
+  my $offset = shift;
+  my $routine = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $start_addr = shift;
+  my $end_addr = shift;
+  my $total = shift;
+
+  # Disassemble all instructions
+  my @instructions = Disassemble($prog, $offset, $start_addr, $end_addr);
+
+  # Make array of counts per instruction
+  my @flat_count = ();
+  my @cum_count = ();
+  my $flat_total = 0;
+  my $cum_total = 0;
+  foreach my $e (@instructions) {
+    # Add up counts for all address that fall inside this instruction
+    my $c1 = 0;
+    my $c2 = 0;
+    for (my $a = $e->[0]; $a lt $e->[4]; $a = AddressInc($a)) {
+      $c1 += GetEntry($flat, $a);
+      $c2 += GetEntry($cumulative, $a);
+    }
+    push(@flat_count, $c1);
+    push(@cum_count, $c2);
+    $flat_total += $c1;
+    $cum_total += $c2;
+  }
+
+  # Print header with total counts
+  printf("ROUTINE ====================== %s\n" .
+         "%6s %6s %s (flat, cumulative) %.1f%% of total\n",
+         ShortFunctionName($routine),
+         Unparse($flat_total),
+         Unparse($cum_total),
+         Units(),
+         ($cum_total * 100.0) / $total);
+
+  # Process instructions in order
+  my $current_file = "";
+  for (my $i = 0; $i <= $#instructions; ) {
+    my $e = $instructions[$i];
+
+    # Print the new file name whenever we switch files
+    if ($e->[1] ne $current_file) {
+      $current_file = $e->[1];
+      my $fname = $current_file;
+      $fname =~ s|^\./||;   # Trim leading "./"
+
+      # Shorten long file names
+      if (length($fname) >= 58) {
+        $fname = "..." . substr($fname, -55);
+      }
+      printf("-------------------- %s\n", $fname);
+    }
+
+    # TODO: Compute range of lines to print together to deal with
+    # small reorderings.
+    my $first_line = $e->[2];
+    my $last_line = $first_line;
+    my %flat_sum = ();
+    my %cum_sum = ();
+    for (my $l = $first_line; $l <= $last_line; $l++) {
+      $flat_sum{$l} = 0;
+      $cum_sum{$l} = 0;
+    }
+
+    # Find run of instructions for this range of source lines
+    my $first_inst = $i;
+    while (($i <= $#instructions) &&
+           ($instructions[$i]->[2] >= $first_line) &&
+           ($instructions[$i]->[2] <= $last_line)) {
+      $e = $instructions[$i];
+      $flat_sum{$e->[2]} += $flat_count[$i];
+      $cum_sum{$e->[2]} += $cum_count[$i];
+      $i++;
+    }
+    my $last_inst = $i - 1;
+
+    # Print source lines
+    for (my $l = $first_line; $l <= $last_line; $l++) {
+      my $line = SourceLine($current_file, $l);
+      if (!defined($line)) {
+        $line = "?\n";
+        next;
+      } else {
+        $line =~ s/^\s+//;
+      }
+      printf("%6s %6s %5d: %s",
+             UnparseAlt($flat_sum{$l}),
+             UnparseAlt($cum_sum{$l}),
+             $l,
+             $line);
+    }
+
+    # Print disassembly
+    for (my $x = $first_inst; $x <= $last_inst; $x++) {
+      my $e = $instructions[$x];
+      printf("%6s %6s    %8s: %6s\n",
+             UnparseAlt($flat_count[$x]),
+             UnparseAlt($cum_count[$x]),
+             UnparseAddress($offset, $e->[0]),
+             CleanDisassembly($e->[3]));
+    }
+  }
+}
+
+# Print DOT graph
+sub PrintDot {
+  my $prog = shift;
+  my $symbols = shift;
+  my $raw = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $overall_total = shift;
+
+  # Get total
+  my $local_total = TotalProfile($flat);
+  my $nodelimit = int($main::opt_nodefraction * $local_total);
+  my $edgelimit = int($main::opt_edgefraction * $local_total);
+  my $nodecount = $main::opt_nodecount;
+
+  # Find nodes to include
+  my @list = (sort { abs(GetEntry($cumulative, $b)) <=>
+                     abs(GetEntry($cumulative, $a))
+                     || $a cmp $b }
+              keys(%{$cumulative}));
+  my $last = $nodecount - 1;
+  if ($last > $#list) {
+    $last = $#list;
+  }
+  while (($last >= 0) &&
+         (abs(GetEntry($cumulative, $list[$last])) <= $nodelimit)) {
+    $last--;
+  }
+  if ($last < 0) {
+    print STDERR "No nodes to print\n";
+    return 0;
+  }
+
+  if ($nodelimit > 0 || $edgelimit > 0) {
+    printf STDERR ("Dropping nodes with <= %s %s; edges with <= %s abs(%s)\n",
+                   Unparse($nodelimit), Units(),
+                   Unparse($edgelimit), Units());
+  }
+
+  # Open DOT output file
+  my $output;
+  my $escaped_dot = ShellEscape(@DOT);
+  my $escaped_ps2pdf = ShellEscape(@PS2PDF);
+  if ($main::opt_gv) {
+    my $escaped_outfile = ShellEscape(TempName($main::next_tmpfile, "ps"));
+    $output = "| $escaped_dot -Tps2 >$escaped_outfile";
+  } elsif ($main::opt_evince) {
+    my $escaped_outfile = ShellEscape(TempName($main::next_tmpfile, "pdf"));
+    $output = "| $escaped_dot -Tps2 | $escaped_ps2pdf - $escaped_outfile";
+  } elsif ($main::opt_ps) {
+    $output = "| $escaped_dot -Tps2";
+  } elsif ($main::opt_pdf) {
+    $output = "| $escaped_dot -Tps2 | $escaped_ps2pdf - -";
+  } elsif ($main::opt_web || $main::opt_svg) {
+    # We need to post-process the SVG, so write to a temporary file always.
+    my $escaped_outfile = ShellEscape(TempName($main::next_tmpfile, "svg"));
+    $output = "| $escaped_dot -Tsvg >$escaped_outfile";
+  } elsif ($main::opt_gif) {
+    $output = "| $escaped_dot -Tgif";
+  } else {
+    $output = ">&STDOUT";
+  }
+  open(DOT, $output) || error("$output: $!\n");
+
+  # Title
+  printf DOT ("digraph \"%s; %s %s\" {\n",
+              $prog,
+              Unparse($overall_total),
+              Units());
+  if ($main::opt_pdf) {
+    # The output is more printable if we set the page size for dot.
+    printf DOT ("size=\"8,11\"\n");
+  }
+  printf DOT ("node [width=0.375,height=0.25];\n");
+
+  # Print legend
+  printf DOT ("Legend [shape=box,fontsize=24,shape=plaintext," .
+              "label=\"%s\\l%s\\l%s\\l%s\\l%s\\l\"];\n",
+              $prog,
+              sprintf("Total %s: %s", Units(), Unparse($overall_total)),
+              sprintf("Focusing on: %s", Unparse($local_total)),
+              sprintf("Dropped nodes with <= %s abs(%s)",
+                      Unparse($nodelimit), Units()),
+              sprintf("Dropped edges with <= %s %s",
+                      Unparse($edgelimit), Units())
+              );
+
+  # Print nodes
+  my %node = ();
+  my $nextnode = 1;
+  foreach my $a (@list[0..$last]) {
+    # Pick font size
+    my $f = GetEntry($flat, $a);
+    my $c = GetEntry($cumulative, $a);
+
+    my $fs = 8;
+    if ($local_total > 0) {
+      $fs = 8 + (50.0 * sqrt(abs($f * 1.0 / $local_total)));
+    }
+
+    $node{$a} = $nextnode++;
+    my $sym = $a;
+    $sym =~ s/\s+/\\n/g;
+    $sym =~ s/::/\\n/g;
+
+    # Extra cumulative info to print for non-leaves
+    my $extra = "";
+    if ($f != $c) {
+      $extra = sprintf("\\rof %s (%s)",
+                       Unparse($c),
+                       Percent($c, $local_total));
+    }
+    my $style = "";
+    if ($main::opt_heapcheck) {
+      if ($f > 0) {
+        # make leak-causing nodes more visible (add a background)
+        $style = ",style=filled,fillcolor=gray"
+      } elsif ($f < 0) {
+        # make anti-leak-causing nodes (which almost never occur)
+        # stand out as well (triple border)
+        $style = ",peripheries=3"
+      }
+    }
+
+    printf DOT ("N%d [label=\"%s\\n%s (%s)%s\\r" .
+                "\",shape=box,fontsize=%.1f%s];\n",
+                $node{$a},
+                $sym,
+                Unparse($f),
+                Percent($f, $local_total),
+                $extra,
+                $fs,
+                $style,
+               );
+  }
+
+  # Get edges and counts per edge
+  my %edge = ();
+  my $n;
+  my $fullname_to_shortname_map = {};
+  FillFullnameToShortnameMap($symbols, $fullname_to_shortname_map);
+  foreach my $k (keys(%{$raw})) {
+    # TODO: omit low %age edges
+    $n = $raw->{$k};
+    my @translated = TranslateStack($symbols, $fullname_to_shortname_map, $k);
+    for (my $i = 1; $i <= $#translated; $i++) {
+      my $src = $translated[$i];
+      my $dst = $translated[$i-1];
+      #next if ($src eq $dst);  # Avoid self-edges?
+      if (exists($node{$src}) && exists($node{$dst})) {
+        my $edge_label = "$src\001$dst";
+        if (!exists($edge{$edge_label})) {
+          $edge{$edge_label} = 0;
+        }
+        $edge{$edge_label} += $n;
+      }
+    }
+  }
+
+  # Print edges (process in order of decreasing counts)
+  my %indegree = ();   # Number of incoming edges added per node so far
+  my %outdegree = ();  # Number of outgoing edges added per node so far
+  foreach my $e (sort { $edge{$b} <=> $edge{$a} } keys(%edge)) {
+    my @x = split(/\001/, $e);
+    $n = $edge{$e};
+
+    # Initialize degree of kept incoming and outgoing edges if necessary
+    my $src = $x[0];
+    my $dst = $x[1];
+    if (!exists($outdegree{$src})) { $outdegree{$src} = 0; }
+    if (!exists($indegree{$dst})) { $indegree{$dst} = 0; }
+
+    my $keep;
+    if ($indegree{$dst} == 0) {
+      # Keep edge if needed for reachability
+      $keep = 1;
+    } elsif (abs($n) <= $edgelimit) {
+      # Drop if we are below --edgefraction
+      $keep = 0;
+    } elsif ($outdegree{$src} >= $main::opt_maxdegree ||
+             $indegree{$dst} >= $main::opt_maxdegree) {
+      # Keep limited number of in/out edges per node
+      $keep = 0;
+    } else {
+      $keep = 1;
+    }
+
+    if ($keep) {
+      $outdegree{$src}++;
+      $indegree{$dst}++;
+
+      # Compute line width based on edge count
+      my $fraction = abs($local_total ? (3 * ($n / $local_total)) : 0);
+      if ($fraction > 1) { $fraction = 1; }
+      my $w = $fraction * 2;
+      if ($w < 1 && ($main::opt_web || $main::opt_svg)) {
+        # SVG output treats line widths < 1 poorly.
+        $w = 1;
+      }
+
+      # Dot sometimes segfaults if given edge weights that are too large, so
+      # we cap the weights at a large value
+      my $edgeweight = abs($n) ** 0.7;
+      if ($edgeweight > 100000) { $edgeweight = 100000; }
+      $edgeweight = int($edgeweight);
+
+      my $style = sprintf("setlinewidth(%f)", $w);
+      if ($x[1] =~ m/\(inline\)/) {
+        $style .= ",dashed";
+      }
+
+      # Use a slightly squashed function of the edge count as the weight
+      printf DOT ("N%s -> N%s [label=%s, weight=%d, style=\"%s\"];\n",
+                  $node{$x[0]},
+                  $node{$x[1]},
+                  Unparse($n),
+                  $edgeweight,
+                  $style);
+    }
+  }
+
+  print DOT ("}\n");
+  close(DOT);
+
+  if ($main::opt_web || $main::opt_svg) {
+    # Rewrite SVG to be more usable inside web browser.
+    RewriteSvg(TempName($main::next_tmpfile, "svg"));
+  }
+
+  return 1;
+}
+
+sub RewriteSvg {
+  my $svgfile = shift;
+
+  open(SVG, $svgfile) || die "open temp svg: $!";
+  my @svg = <SVG>;
+  close(SVG);
+  unlink $svgfile;
+  my $svg = join('', @svg);
+
+  # Dot's SVG output is
+  #
+  #    <svg width="___" height="___"
+  #     viewBox="___" xmlns=...>
+  #    <g id="graph0" transform="...">
+  #    ...
+  #    </g>
+  #    </svg>
+  #
+  # Change it to
+  #
+  #    <svg width="100%" height="100%"
+  #     xmlns=...>
+  #    $svg_javascript
+  #    <g id="viewport" transform="translate(0,0)">
+  #    <g id="graph0" transform="...">
+  #    ...
+  #    </g>
+  #    </g>
+  #    </svg>
+
+  # Fix width, height; drop viewBox.
+  $svg =~ s/(?s)<svg width="[^"]+" height="[^"]+"(.*?)viewBox="[^"]+"/<svg width="100%" height="100%"$1/;
+
+  # Insert script, viewport <g> above first <g>
+  my $svg_javascript = SvgJavascript();
+  my $viewport = "<g id=\"viewport\" transform=\"translate(0,0)\">\n";
+  $svg =~ s/<g id="graph\d"/$svg_javascript$viewport$&/;
+
+  # Insert final </g> above </svg>.
+  $svg =~ s/(.*)(<\/svg>)/$1<\/g>$2/;
+  $svg =~ s/<g id="graph\d"(.*?)/<g id="viewport"$1/;
+
+  if ($main::opt_svg) {
+    # --svg: write to standard output.
+    print $svg;
+  } else {
+    # Write back to temporary file.
+    open(SVG, ">$svgfile") || die "open $svgfile: $!";
+    print SVG $svg;
+    close(SVG);
+  }
+}
+
+sub SvgJavascript {
+  return <<'EOF';
+<script type="text/ecmascript"><![CDATA[
+// SVGPan
+// http://www.cyberz.org/blog/2009/12/08/svgpan-a-javascript-svg-panzoomdrag-library/
+// Local modification: if(true || ...) below to force panning, never moving.
+
+/**
+ *  SVGPan library 1.2
+ * ====================
+ *
+ * Given an unique existing element with id "viewport", including the
+ * the library into any SVG adds the following capabilities:
+ *
+ *  - Mouse panning
+ *  - Mouse zooming (using the wheel)
+ *  - Object dargging
+ *
+ * Known issues:
+ *
+ *  - Zooming (while panning) on Safari has still some issues
+ *
+ * Releases:
+ *
+ * 1.2, Sat Mar 20 08:42:50 GMT 2010, Zeng Xiaohui
+ *	Fixed a bug with browser mouse handler interaction
+ *
+ * 1.1, Wed Feb  3 17:39:33 GMT 2010, Zeng Xiaohui
+ *	Updated the zoom code to support the mouse wheel on Safari/Chrome
+ *
+ * 1.0, Andrea Leofreddi
+ *	First release
+ *
+ * This code is licensed under the following BSD license:
+ *
+ * Copyright 2009-2010 Andrea Leofreddi <a.leofreddi@itcharm.com>. All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without modification, are
+ * permitted provided that the following conditions are met:
+ *
+ *    1. Redistributions of source code must retain the above copyright notice, this list of
+ *       conditions and the following disclaimer.
+ *
+ *    2. Redistributions in binary form must reproduce the above copyright notice, this list
+ *       of conditions and the following disclaimer in the documentation and/or other materials
+ *       provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY Andrea Leofreddi ``AS IS'' AND ANY EXPRESS OR IMPLIED
+ * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
+ * FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL Andrea Leofreddi OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+ * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ * ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+ * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * The views and conclusions contained in the software and documentation are those of the
+ * authors and should not be interpreted as representing official policies, either expressed
+ * or implied, of Andrea Leofreddi.
+ */
+
+var root = document.documentElement;
+
+var state = 'none', stateTarget, stateOrigin, stateTf;
+
+setupHandlers(root);
+
+/**
+ * Register handlers
+ */
+function setupHandlers(root){
+	setAttributes(root, {
+		"onmouseup" : "add(evt)",
+		"onmousedown" : "handleMouseDown(evt)",
+		"onmousemove" : "handleMouseMove(evt)",
+		"onmouseup" : "handleMouseUp(evt)",
+		//"onmouseout" : "handleMouseUp(evt)", // Decomment this to stop the pan functionality when dragging out of the SVG element
+	});
+
+	if(navigator.userAgent.toLowerCase().indexOf('webkit') >= 0)
+		window.addEventListener('mousewheel', handleMouseWheel, false); // Chrome/Safari
+	else
+		window.addEventListener('DOMMouseScroll', handleMouseWheel, false); // Others
+
+	var g = svgDoc.getElementById("svg");
+	g.width = "100%";
+	g.height = "100%";
+}
+
+/**
+ * Instance an SVGPoint object with given event coordinates.
+ */
+function getEventPoint(evt) {
+	var p = root.createSVGPoint();
+
+	p.x = evt.clientX;
+	p.y = evt.clientY;
+
+	return p;
+}
+
+/**
+ * Sets the current transform matrix of an element.
+ */
+function setCTM(element, matrix) {
+	var s = "matrix(" + matrix.a + "," + matrix.b + "," + matrix.c + "," + matrix.d + "," + matrix.e + "," + matrix.f + ")";
+
+	element.setAttribute("transform", s);
+}
+
+/**
+ * Dumps a matrix to a string (useful for debug).
+ */
+function dumpMatrix(matrix) {
+	var s = "[ " + matrix.a + ", " + matrix.c + ", " + matrix.e + "\n  " + matrix.b + ", " + matrix.d + ", " + matrix.f + "\n  0, 0, 1 ]";
+
+	return s;
+}
+
+/**
+ * Sets attributes of an element.
+ */
+function setAttributes(element, attributes){
+	for (i in attributes)
+		element.setAttributeNS(null, i, attributes[i]);
+}
+
+/**
+ * Handle mouse move event.
+ */
+function handleMouseWheel(evt) {
+	if(evt.preventDefault)
+		evt.preventDefault();
+
+	evt.returnValue = false;
+
+	var svgDoc = evt.target.ownerDocument;
+
+	var delta;
+
+	if(evt.wheelDelta)
+		delta = evt.wheelDelta / 3600; // Chrome/Safari
+	else
+		delta = evt.detail / -90; // Mozilla
+
+	var z = 1 + delta; // Zoom factor: 0.9/1.1
+
+	var g = svgDoc.getElementById("viewport");
+
+	var p = getEventPoint(evt);
+
+	p = p.matrixTransform(g.getCTM().inverse());
+
+	// Compute new scale matrix in current mouse position
+	var k = root.createSVGMatrix().translate(p.x, p.y).scale(z).translate(-p.x, -p.y);
+
+        setCTM(g, g.getCTM().multiply(k));
+
+	stateTf = stateTf.multiply(k.inverse());
+}
+
+/**
+ * Handle mouse move event.
+ */
+function handleMouseMove(evt) {
+	if(evt.preventDefault)
+		evt.preventDefault();
+
+	evt.returnValue = false;
+
+	var svgDoc = evt.target.ownerDocument;
+
+	var g = svgDoc.getElementById("viewport");
+
+	if(state == 'pan') {
+		// Pan mode
+		var p = getEventPoint(evt).matrixTransform(stateTf);
+
+		setCTM(g, stateTf.inverse().translate(p.x - stateOrigin.x, p.y - stateOrigin.y));
+	} else if(state == 'move') {
+		// Move mode
+		var p = getEventPoint(evt).matrixTransform(g.getCTM().inverse());
+
+		setCTM(stateTarget, root.createSVGMatrix().translate(p.x - stateOrigin.x, p.y - stateOrigin.y).multiply(g.getCTM().inverse()).multiply(stateTarget.getCTM()));
+
+		stateOrigin = p;
+	}
+}
+
+/**
+ * Handle click event.
+ */
+function handleMouseDown(evt) {
+	if(evt.preventDefault)
+		evt.preventDefault();
+
+	evt.returnValue = false;
+
+	var svgDoc = evt.target.ownerDocument;
+
+	var g = svgDoc.getElementById("viewport");
+
+	if(true || evt.target.tagName == "svg") {
+		// Pan mode
+		state = 'pan';
+
+		stateTf = g.getCTM().inverse();
+
+		stateOrigin = getEventPoint(evt).matrixTransform(stateTf);
+	} else {
+		// Move mode
+		state = 'move';
+
+		stateTarget = evt.target;
+
+		stateTf = g.getCTM().inverse();
+
+		stateOrigin = getEventPoint(evt).matrixTransform(stateTf);
+	}
+}
+
+/**
+ * Handle mouse button release event.
+ */
+function handleMouseUp(evt) {
+	if(evt.preventDefault)
+		evt.preventDefault();
+
+	evt.returnValue = false;
+
+	var svgDoc = evt.target.ownerDocument;
+
+	if(state == 'pan' || state == 'move') {
+		// Quit pan mode
+		state = '';
+	}
+}
+
+]]></script>
+EOF
+}
+
+# Provides a map from fullname to shortname for cases where the
+# shortname is ambiguous.  The symlist has both the fullname and
+# shortname for all symbols, which is usually fine, but sometimes --
+# such as overloaded functions -- two different fullnames can map to
+# the same shortname.  In that case, we use the address of the
+# function to disambiguate the two.  This function fills in a map that
+# maps fullnames to modified shortnames in such cases.  If a fullname
+# is not present in the map, the 'normal' shortname provided by the
+# symlist is the appropriate one to use.
+sub FillFullnameToShortnameMap {
+  my $symbols = shift;
+  my $fullname_to_shortname_map = shift;
+  my $shortnames_seen_once = {};
+  my $shortnames_seen_more_than_once = {};
+
+  foreach my $symlist (values(%{$symbols})) {
+    # TODO(csilvers): deal with inlined symbols too.
+    my $shortname = $symlist->[0];
+    my $fullname = $symlist->[2];
+    if ($fullname !~ /<[0-9a-fA-F]+>$/) {  # fullname doesn't end in an address
+      next;       # the only collisions we care about are when addresses differ
+    }
+    if (defined($shortnames_seen_once->{$shortname}) &&
+        $shortnames_seen_once->{$shortname} ne $fullname) {
+      $shortnames_seen_more_than_once->{$shortname} = 1;
+    } else {
+      $shortnames_seen_once->{$shortname} = $fullname;
+    }
+  }
+
+  foreach my $symlist (values(%{$symbols})) {
+    my $shortname = $symlist->[0];
+    my $fullname = $symlist->[2];
+    # TODO(csilvers): take in a list of addresses we care about, and only
+    # store in the map if $symlist->[1] is in that list.  Saves space.
+    next if defined($fullname_to_shortname_map->{$fullname});
+    if (defined($shortnames_seen_more_than_once->{$shortname})) {
+      if ($fullname =~ /<0*([^>]*)>$/) {   # fullname has address at end of it
+        $fullname_to_shortname_map->{$fullname} = "$shortname\@$1";
+      }
+    }
+  }
+}
+
+# Return a small number that identifies the argument.
+# Multiple calls with the same argument will return the same number.
+# Calls with different arguments will return different numbers.
+sub ShortIdFor {
+  my $key = shift;
+  my $id = $main::uniqueid{$key};
+  if (!defined($id)) {
+    $id = keys(%main::uniqueid) + 1;
+    $main::uniqueid{$key} = $id;
+  }
+  return $id;
+}
+
+# Translate a stack of addresses into a stack of symbols
+sub TranslateStack {
+  my $symbols = shift;
+  my $fullname_to_shortname_map = shift;
+  my $k = shift;
+
+  my @addrs = split(/\n/, $k);
+  my @result = ();
+  for (my $i = 0; $i <= $#addrs; $i++) {
+    my $a = $addrs[$i];
+
+    # Skip large addresses since they sometimes show up as fake entries on RH9
+    if (length($a) > 8 && $a gt "7fffffffffffffff") {
+      next;
+    }
+
+    if ($main::opt_disasm || $main::opt_list) {
+      # We want just the address for the key
+      push(@result, $a);
+      next;
+    }
+
+    my $symlist = $symbols->{$a};
+    if (!defined($symlist)) {
+      $symlist = [$a, "", $a];
+    }
+
+    # We can have a sequence of symbols for a particular entry
+    # (more than one symbol in the case of inlining).  Callers
+    # come before callees in symlist, so walk backwards since
+    # the translated stack should contain callees before callers.
+    for (my $j = $#{$symlist}; $j >= 2; $j -= 3) {
+      my $func = $symlist->[$j-2];
+      my $fileline = $symlist->[$j-1];
+      my $fullfunc = $symlist->[$j];
+      if (defined($fullname_to_shortname_map->{$fullfunc})) {
+        $func = $fullname_to_shortname_map->{$fullfunc};
+      }
+      if ($j > 2) {
+        $func = "$func (inline)";
+      }
+
+      # Do not merge nodes corresponding to Callback::Run since that
+      # causes confusing cycles in dot display.  Instead, we synthesize
+      # a unique name for this frame per caller.
+      if ($func =~ m/Callback.*::Run$/) {
+        my $caller = ($i > 0) ? $addrs[$i-1] : 0;
+        $func = "Run#" . ShortIdFor($caller);
+      }
+
+      if ($main::opt_addresses) {
+        push(@result, "$a $func $fileline");
+      } elsif ($main::opt_lines) {
+        if ($func eq '??' && $fileline eq '??:0') {
+          push(@result, "$a");
+        } else {
+          push(@result, "$func $fileline");
+        }
+      } elsif ($main::opt_functions) {
+        if ($func eq '??') {
+          push(@result, "$a");
+        } else {
+          push(@result, $func);
+        }
+      } elsif ($main::opt_files) {
+        if ($fileline eq '??:0' || $fileline eq '') {
+          push(@result, "$a");
+        } else {
+          my $f = $fileline;
+          $f =~ s/:\d+$//;
+          push(@result, $f);
+        }
+      } else {
+        push(@result, $a);
+        last;  # Do not print inlined info
+      }
+    }
+  }
+
+  # print join(",", @addrs), " => ", join(",", @result), "\n";
+  return @result;
+}
+
+# Generate percent string for a number and a total
+sub Percent {
+  my $num = shift;
+  my $tot = shift;
+  if ($tot != 0) {
+    return sprintf("%.1f%%", $num * 100.0 / $tot);
+  } else {
+    return ($num == 0) ? "nan" : (($num > 0) ? "+inf" : "-inf");
+  }
+}
+
+# Generate pretty-printed form of number
+sub Unparse {
+  my $num = shift;
+  if ($main::profile_type eq 'heap' || $main::profile_type eq 'growth') {
+    if ($main::opt_inuse_objects || $main::opt_alloc_objects) {
+      return sprintf("%d", $num);
+    } else {
+      if ($main::opt_show_bytes) {
+        return sprintf("%d", $num);
+      } else {
+        return sprintf("%.1f", $num / 1048576.0);
+      }
+    }
+  } elsif ($main::profile_type eq 'contention' && !$main::opt_contentions) {
+    return sprintf("%.3f", $num / 1e9); # Convert nanoseconds to seconds
+  } else {
+    return sprintf("%d", $num);
+  }
+}
+
+# Alternate pretty-printed form: 0 maps to "."
+sub UnparseAlt {
+  my $num = shift;
+  if ($num == 0) {
+    return ".";
+  } else {
+    return Unparse($num);
+  }
+}
+
+# Alternate pretty-printed form: 0 maps to ""
+sub HtmlPrintNumber {
+  my $num = shift;
+  if ($num == 0) {
+    return "";
+  } else {
+    return Unparse($num);
+  }
+}
+
+# Return output units
+sub Units {
+  if ($main::profile_type eq 'heap' || $main::profile_type eq 'growth') {
+    if ($main::opt_inuse_objects || $main::opt_alloc_objects) {
+      return "objects";
+    } else {
+      if ($main::opt_show_bytes) {
+        return "B";
+      } else {
+        return "MB";
+      }
+    }
+  } elsif ($main::profile_type eq 'contention' && !$main::opt_contentions) {
+    return "seconds";
+  } else {
+    return "samples";
+  }
+}
+
+##### Profile manipulation code #####
+
+# Generate flattened profile:
+# If count is charged to stack [a,b,c,d], in generated profile,
+# it will be charged to [a]
+sub FlatProfile {
+  my $profile = shift;
+  my $result = {};
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    if ($#addrs >= 0) {
+      AddEntry($result, $addrs[0], $count);
+    }
+  }
+  return $result;
+}
+
+# Generate cumulative profile:
+# If count is charged to stack [a,b,c,d], in generated profile,
+# it will be charged to [a], [b], [c], [d]
+sub CumulativeProfile {
+  my $profile = shift;
+  my $result = {};
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    foreach my $a (@addrs) {
+      AddEntry($result, $a, $count);
+    }
+  }
+  return $result;
+}
+
+# If the second-youngest PC on the stack is always the same, returns
+# that pc.  Otherwise, returns undef.
+sub IsSecondPcAlwaysTheSame {
+  my $profile = shift;
+
+  my $second_pc = undef;
+  foreach my $k (keys(%{$profile})) {
+    my @addrs = split(/\n/, $k);
+    if ($#addrs < 1) {
+      return undef;
+    }
+    if (not defined $second_pc) {
+      $second_pc = $addrs[1];
+    } else {
+      if ($second_pc ne $addrs[1]) {
+        return undef;
+      }
+    }
+  }
+  return $second_pc;
+}
+
+sub ExtractSymbolLocation {
+  my $symbols = shift;
+  my $address = shift;
+  # 'addr2line' outputs "??:0" for unknown locations; we do the
+  # same to be consistent.
+  my $location = "??:0:unknown";
+  if (exists $symbols->{$address}) {
+    my $file = $symbols->{$address}->[1];
+    if ($file eq "?") {
+      $file = "??:0"
+    }
+    $location = $file . ":" . $symbols->{$address}->[0];
+  }
+  return $location;
+}
+
+# Extracts a graph of calls.
+sub ExtractCalls {
+  my $symbols = shift;
+  my $profile = shift;
+
+  my $calls = {};
+  while( my ($stack_trace, $count) = each %$profile ) {
+    my @address = split(/\n/, $stack_trace);
+    my $destination = ExtractSymbolLocation($symbols, $address[0]);
+    AddEntry($calls, $destination, $count);
+    for (my $i = 1; $i <= $#address; $i++) {
+      my $source = ExtractSymbolLocation($symbols, $address[$i]);
+      my $call = "$source -> $destination";
+      AddEntry($calls, $call, $count);
+      $destination = $source;
+    }
+  }
+
+  return $calls;
+}
+
+sub RemoveUninterestingFrames {
+  my $symbols = shift;
+  my $profile = shift;
+
+  # List of function names to skip
+  my %skip = ();
+  my $skip_regexp = 'NOMATCH';
+  if ($main::profile_type eq 'heap' || $main::profile_type eq 'growth') {
+    foreach my $name ('calloc',
+                      'cfree',
+                      'malloc',
+                      'free',
+                      'memalign',
+                      'posix_memalign',
+                      'pvalloc',
+                      'valloc',
+                      'realloc',
+                      'tc_calloc',
+                      'tc_cfree',
+                      'tc_malloc',
+                      'tc_free',
+                      'tc_memalign',
+                      'tc_posix_memalign',
+                      'tc_pvalloc',
+                      'tc_valloc',
+                      'tc_realloc',
+                      'tc_new',
+                      'tc_delete',
+                      'tc_newarray',
+                      'tc_deletearray',
+                      'tc_new_nothrow',
+                      'tc_newarray_nothrow',
+                      'do_malloc',
+                      '::do_malloc',   # new name -- got moved to an unnamed ns
+                      '::do_malloc_or_cpp_alloc',
+                      'DoSampledAllocation',
+                      'simple_alloc::allocate',
+                      '__malloc_alloc_template::allocate',
+                      '__builtin_delete',
+                      '__builtin_new',
+                      '__builtin_vec_delete',
+                      '__builtin_vec_new',
+                      'operator new',
+                      'operator new[]',
+                      # The entry to our memory-allocation routines on OS X
+                      'malloc_zone_malloc',
+                      'malloc_zone_calloc',
+                      'malloc_zone_valloc',
+                      'malloc_zone_realloc',
+                      'malloc_zone_memalign',
+                      'malloc_zone_free',
+                      # These mark the beginning/end of our custom sections
+                      '__start_google_malloc',
+                      '__stop_google_malloc',
+                      '__start_malloc_hook',
+                      '__stop_malloc_hook') {
+      $skip{$name} = 1;
+      $skip{"_" . $name} = 1;   # Mach (OS X) adds a _ prefix to everything
+    }
+    # TODO: Remove TCMalloc once everything has been
+    # moved into the tcmalloc:: namespace and we have flushed
+    # old code out of the system.
+    $skip_regexp = "TCMalloc|^tcmalloc::";
+  } elsif ($main::profile_type eq 'contention') {
+    foreach my $vname ('base::RecordLockProfileData',
+                       'base::SubmitMutexProfileData',
+                       'base::SubmitSpinLockProfileData',
+                       'Mutex::Unlock',
+                       'Mutex::UnlockSlow',
+                       'Mutex::ReaderUnlock',
+                       'MutexLock::~MutexLock',
+                       'SpinLock::Unlock',
+                       'SpinLock::SlowUnlock',
+                       'SpinLockHolder::~SpinLockHolder') {
+      $skip{$vname} = 1;
+    }
+  } elsif ($main::profile_type eq 'cpu') {
+    # Drop signal handlers used for CPU profile collection
+    # TODO(dpeng): this should not be necessary; it's taken
+    # care of by the general 2nd-pc mechanism below.
+    foreach my $name ('ProfileData::Add',           # historical
+                      'ProfileData::prof_handler',  # historical
+                      'CpuProfiler::prof_handler',
+                      '__FRAME_END__',
+                      '__pthread_sighandler',
+                      '__restore') {
+      $skip{$name} = 1;
+    }
+  } else {
+    # Nothing skipped for unknown types
+  }
+
+  if ($main::profile_type eq 'cpu') {
+    # If all the second-youngest program counters are the same,
+    # this STRONGLY suggests that it is an artifact of measurement,
+    # i.e., stack frames pushed by the CPU profiler signal handler.
+    # Hence, we delete them.
+    # (The topmost PC is read from the signal structure, not from
+    # the stack, so it does not get involved.)
+    while (my $second_pc = IsSecondPcAlwaysTheSame($profile)) {
+      my $result = {};
+      my $func = '';
+      if (exists($symbols->{$second_pc})) {
+        $second_pc = $symbols->{$second_pc}->[0];
+      }
+      print STDERR "Removing $second_pc from all stack traces.\n";
+      foreach my $k (keys(%{$profile})) {
+        my $count = $profile->{$k};
+        my @addrs = split(/\n/, $k);
+        splice @addrs, 1, 1;
+        my $reduced_path = join("\n", @addrs);
+        AddEntry($result, $reduced_path, $count);
+      }
+      $profile = $result;
+    }
+  }
+
+  my $result = {};
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    my @path = ();
+    foreach my $a (@addrs) {
+      if (exists($symbols->{$a})) {
+        my $func = $symbols->{$a}->[0];
+        if ($skip{$func} || ($func =~ m/$skip_regexp/)) {
+          next;
+        }
+      }
+      push(@path, $a);
+    }
+    my $reduced_path = join("\n", @path);
+    AddEntry($result, $reduced_path, $count);
+  }
+  return $result;
+}
+
+# Reduce profile to granularity given by user
+sub ReduceProfile {
+  my $symbols = shift;
+  my $profile = shift;
+  my $result = {};
+  my $fullname_to_shortname_map = {};
+  FillFullnameToShortnameMap($symbols, $fullname_to_shortname_map);
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @translated = TranslateStack($symbols, $fullname_to_shortname_map, $k);
+    my @path = ();
+    my %seen = ();
+    $seen{''} = 1;      # So that empty keys are skipped
+    foreach my $e (@translated) {
+      # To avoid double-counting due to recursion, skip a stack-trace
+      # entry if it has already been seen
+      if (!$seen{$e}) {
+        $seen{$e} = 1;
+        push(@path, $e);
+      }
+    }
+    my $reduced_path = join("\n", @path);
+    AddEntry($result, $reduced_path, $count);
+  }
+  return $result;
+}
+
+# Does the specified symbol array match the regexp?
+sub SymbolMatches {
+  my $sym = shift;
+  my $re = shift;
+  if (defined($sym)) {
+    for (my $i = 0; $i < $#{$sym}; $i += 3) {
+      if ($sym->[$i] =~ m/$re/ || $sym->[$i+1] =~ m/$re/) {
+        return 1;
+      }
+    }
+  }
+  return 0;
+}
+
+# Focus only on paths involving specified regexps
+sub FocusProfile {
+  my $symbols = shift;
+  my $profile = shift;
+  my $focus = shift;
+  my $result = {};
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    foreach my $a (@addrs) {
+      # Reply if it matches either the address/shortname/fileline
+      if (($a =~ m/$focus/) || SymbolMatches($symbols->{$a}, $focus)) {
+        AddEntry($result, $k, $count);
+        last;
+      }
+    }
+  }
+  return $result;
+}
+
+# Focus only on paths not involving specified regexps
+sub IgnoreProfile {
+  my $symbols = shift;
+  my $profile = shift;
+  my $ignore = shift;
+  my $result = {};
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    my $matched = 0;
+    foreach my $a (@addrs) {
+      # Reply if it matches either the address/shortname/fileline
+      if (($a =~ m/$ignore/) || SymbolMatches($symbols->{$a}, $ignore)) {
+        $matched = 1;
+        last;
+      }
+    }
+    if (!$matched) {
+      AddEntry($result, $k, $count);
+    }
+  }
+  return $result;
+}
+
+# Get total count in profile
+sub TotalProfile {
+  my $profile = shift;
+  my $result = 0;
+  foreach my $k (keys(%{$profile})) {
+    $result += $profile->{$k};
+  }
+  return $result;
+}
+
+# Add A to B
+sub AddProfile {
+  my $A = shift;
+  my $B = shift;
+
+  my $R = {};
+  # add all keys in A
+  foreach my $k (keys(%{$A})) {
+    my $v = $A->{$k};
+    AddEntry($R, $k, $v);
+  }
+  # add all keys in B
+  foreach my $k (keys(%{$B})) {
+    my $v = $B->{$k};
+    AddEntry($R, $k, $v);
+  }
+  return $R;
+}
+
+# Merges symbol maps
+sub MergeSymbols {
+  my $A = shift;
+  my $B = shift;
+
+  my $R = {};
+  foreach my $k (keys(%{$A})) {
+    $R->{$k} = $A->{$k};
+  }
+  if (defined($B)) {
+    foreach my $k (keys(%{$B})) {
+      $R->{$k} = $B->{$k};
+    }
+  }
+  return $R;
+}
+
+
+# Add A to B
+sub AddPcs {
+  my $A = shift;
+  my $B = shift;
+
+  my $R = {};
+  # add all keys in A
+  foreach my $k (keys(%{$A})) {
+    $R->{$k} = 1
+  }
+  # add all keys in B
+  foreach my $k (keys(%{$B})) {
+    $R->{$k} = 1
+  }
+  return $R;
+}
+
+# Subtract B from A
+sub SubtractProfile {
+  my $A = shift;
+  my $B = shift;
+
+  my $R = {};
+  foreach my $k (keys(%{$A})) {
+    my $v = $A->{$k} - GetEntry($B, $k);
+    if ($v < 0 && $main::opt_drop_negative) {
+      $v = 0;
+    }
+    AddEntry($R, $k, $v);
+  }
+  if (!$main::opt_drop_negative) {
+    # Take care of when subtracted profile has more entries
+    foreach my $k (keys(%{$B})) {
+      if (!exists($A->{$k})) {
+        AddEntry($R, $k, 0 - $B->{$k});
+      }
+    }
+  }
+  return $R;
+}
+
+# Get entry from profile; zero if not present
+sub GetEntry {
+  my $profile = shift;
+  my $k = shift;
+  if (exists($profile->{$k})) {
+    return $profile->{$k};
+  } else {
+    return 0;
+  }
+}
+
+# Add entry to specified profile
+sub AddEntry {
+  my $profile = shift;
+  my $k = shift;
+  my $n = shift;
+  if (!exists($profile->{$k})) {
+    $profile->{$k} = 0;
+  }
+  $profile->{$k} += $n;
+}
+
+# Add a stack of entries to specified profile, and add them to the $pcs
+# list.
+sub AddEntries {
+  my $profile = shift;
+  my $pcs = shift;
+  my $stack = shift;
+  my $count = shift;
+  my @k = ();
+
+  foreach my $e (split(/\s+/, $stack)) {
+    my $pc = HexExtend($e);
+    $pcs->{$pc} = 1;
+    push @k, $pc;
+  }
+  AddEntry($profile, (join "\n", @k), $count);
+}
+
+##### Code to profile a server dynamically #####
+
+sub CheckSymbolPage {
+  my $url = SymbolPageURL();
+  my $command = ShellEscape(@URL_FETCHER, $url);
+  open(SYMBOL, "$command |") or error($command);
+  my $line = <SYMBOL>;
+  $line =~ s/\r//g;         # turn windows-looking lines into unix-looking lines
+  close(SYMBOL);
+  unless (defined($line)) {
+    error("$url doesn't exist\n");
+  }
+
+  if ($line =~ /^num_symbols:\s+(\d+)$/) {
+    if ($1 == 0) {
+      error("Stripped binary. No symbols available.\n");
+    }
+  } else {
+    error("Failed to get the number of symbols from $url\n");
+  }
+}
+
+sub IsProfileURL {
+  my $profile_name = shift;
+  if (-f $profile_name) {
+    printf STDERR "Using local file $profile_name.\n";
+    return 0;
+  }
+  return 1;
+}
+
+sub ParseProfileURL {
+  my $profile_name = shift;
+
+  if (!defined($profile_name) || $profile_name eq "") {
+    return ();
+  }
+
+  # Split profile URL - matches all non-empty strings, so no test.
+  $profile_name =~ m,^(https?://)?([^/]+)(.*?)(/|$PROFILES)?$,;
+
+  my $proto = $1 || "http://";
+  my $hostport = $2;
+  my $prefix = $3;
+  my $profile = $4 || "/";
+
+  my $host = $hostport;
+  $host =~ s/:.*//;
+
+  my $baseurl = "$proto$hostport$prefix";
+  return ($host, $baseurl, $profile);
+}
+
+# We fetch symbols from the first profile argument.
+sub SymbolPageURL {
+  my ($host, $baseURL, $path) = ParseProfileURL($main::pfile_args[0]);
+  return "$baseURL$SYMBOL_PAGE";
+}
+
+sub FetchProgramName() {
+  my ($host, $baseURL, $path) = ParseProfileURL($main::pfile_args[0]);
+  my $url = "$baseURL$PROGRAM_NAME_PAGE";
+  my $command_line = ShellEscape(@URL_FETCHER, $url);
+  open(CMDLINE, "$command_line |") or error($command_line);
+  my $cmdline = <CMDLINE>;
+  $cmdline =~ s/\r//g;   # turn windows-looking lines into unix-looking lines
+  close(CMDLINE);
+  error("Failed to get program name from $url\n") unless defined($cmdline);
+  $cmdline =~ s/\x00.+//;  # Remove argv[1] and latters.
+  $cmdline =~ s!\n!!g;  # Remove LFs.
+  return $cmdline;
+}
+
+# Gee, curl's -L (--location) option isn't reliable at least
+# with its 7.12.3 version.  Curl will forget to post data if
+# there is a redirection.  This function is a workaround for
+# curl.  Redirection happens on borg hosts.
+sub ResolveRedirectionForCurl {
+  my $url = shift;
+  my $command_line = ShellEscape(@URL_FETCHER, "--head", $url);
+  open(CMDLINE, "$command_line |") or error($command_line);
+  while (<CMDLINE>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    if (/^Location: (.*)/) {
+      $url = $1;
+    }
+  }
+  close(CMDLINE);
+  return $url;
+}
+
+# Add a timeout flat to URL_FETCHER.  Returns a new list.
+sub AddFetchTimeout {
+  my $timeout = shift;
+  my @fetcher = shift;
+  if (defined($timeout)) {
+    if (join(" ", @fetcher) =~ m/\bcurl -s/) {
+      push(@fetcher, "--max-time", sprintf("%d", $timeout));
+    } elsif (join(" ", @fetcher) =~ m/\brpcget\b/) {
+      push(@fetcher, sprintf("--deadline=%d", $timeout));
+    }
+  }
+  return @fetcher;
+}
+
+# Reads a symbol map from the file handle name given as $1, returning
+# the resulting symbol map.  Also processes variables relating to symbols.
+# Currently, the only variable processed is 'binary=<value>' which updates
+# $main::prog to have the correct program name.
+sub ReadSymbols {
+  my $in = shift;
+  my $map = {};
+  while (<$in>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    # Removes all the leading zeroes from the symbols, see comment below.
+    if (m/^0x0*([0-9a-f]+)\s+(.+)/) {
+      $map->{$1} = $2;
+    } elsif (m/^---/) {
+      last;
+    } elsif (m/^([a-z][^=]*)=(.*)$/ ) {
+      my ($variable, $value) = ($1, $2);
+      for ($variable, $value) {
+        s/^\s+//;
+        s/\s+$//;
+      }
+      if ($variable eq "binary") {
+        if ($main::prog ne $UNKNOWN_BINARY && $main::prog ne $value) {
+          printf STDERR ("Warning: Mismatched binary name '%s', using '%s'.\n",
+                         $main::prog, $value);
+        }
+        $main::prog = $value;
+      } else {
+        printf STDERR ("Ignoring unknown variable in symbols list: " .
+            "'%s' = '%s'\n", $variable, $value);
+      }
+    }
+  }
+  return $map;
+}
+
+# Fetches and processes symbols to prepare them for use in the profile output
+# code.  If the optional 'symbol_map' arg is not given, fetches symbols from
+# $SYMBOL_PAGE for all PC values found in profile.  Otherwise, the raw symbols
+# are assumed to have already been fetched into 'symbol_map' and are simply
+# extracted and processed.
+sub FetchSymbols {
+  my $pcset = shift;
+  my $symbol_map = shift;
+
+  my %seen = ();
+  my @pcs = grep { !$seen{$_}++ } keys(%$pcset);  # uniq
+
+  if (!defined($symbol_map)) {
+    my $post_data = join("+", sort((map {"0x" . "$_"} @pcs)));
+
+    open(POSTFILE, ">$main::tmpfile_sym");
+    print POSTFILE $post_data;
+    close(POSTFILE);
+
+    my $url = SymbolPageURL();
+
+    my $command_line;
+    if (join(" ", @URL_FETCHER) =~ m/\bcurl -s/) {
+      $url = ResolveRedirectionForCurl($url);
+      $command_line = ShellEscape(@URL_FETCHER, "-d", "\@$main::tmpfile_sym",
+                                  $url);
+    } else {
+      $command_line = (ShellEscape(@URL_FETCHER, "--post", $url)
+                       . " < " . ShellEscape($main::tmpfile_sym));
+    }
+    # We use c++filt in case $SYMBOL_PAGE gives us mangled symbols.
+    my $escaped_cppfilt = ShellEscape($obj_tool_map{"c++filt"});
+    open(SYMBOL, "$command_line | $escaped_cppfilt |") or error($command_line);
+    $symbol_map = ReadSymbols(*SYMBOL{IO});
+    close(SYMBOL);
+  }
+
+  my $symbols = {};
+  foreach my $pc (@pcs) {
+    my $fullname;
+    # For 64 bits binaries, symbols are extracted with 8 leading zeroes.
+    # Then /symbol reads the long symbols in as uint64, and outputs
+    # the result with a "0x%08llx" format which get rid of the zeroes.
+    # By removing all the leading zeroes in both $pc and the symbols from
+    # /symbol, the symbols match and are retrievable from the map.
+    my $shortpc = $pc;
+    $shortpc =~ s/^0*//;
+    # Each line may have a list of names, which includes the function
+    # and also other functions it has inlined.  They are separated (in
+    # PrintSymbolizedProfile), by --, which is illegal in function names.
+    my $fullnames;
+    if (defined($symbol_map->{$shortpc})) {
+      $fullnames = $symbol_map->{$shortpc};
+    } else {
+      $fullnames = "0x" . $pc;  # Just use addresses
+    }
+    my $sym = [];
+    $symbols->{$pc} = $sym;
+    foreach my $fullname (split("--", $fullnames)) {
+      my $name = ShortFunctionName($fullname);
+      push(@{$sym}, $name, "?", $fullname);
+    }
+  }
+  return $symbols;
+}
+
+sub BaseName {
+  my $file_name = shift;
+  $file_name =~ s!^.*/!!;  # Remove directory name
+  return $file_name;
+}
+
+sub MakeProfileBaseName {
+  my ($binary_name, $profile_name) = @_;
+  my ($host, $baseURL, $path) = ParseProfileURL($profile_name);
+  my $binary_shortname = BaseName($binary_name);
+  return sprintf("%s.%s.%s",
+                 $binary_shortname, $main::op_time, $host);
+}
+
+sub FetchDynamicProfile {
+  my $binary_name = shift;
+  my $profile_name = shift;
+  my $fetch_name_only = shift;
+  my $encourage_patience = shift;
+
+  if (!IsProfileURL($profile_name)) {
+    return $profile_name;
+  } else {
+    my ($host, $baseURL, $path) = ParseProfileURL($profile_name);
+    if ($path eq "" || $path eq "/") {
+      # Missing type specifier defaults to cpu-profile
+      $path = $PROFILE_PAGE;
+    }
+
+    my $profile_file = MakeProfileBaseName($binary_name, $profile_name);
+
+    my $url = "$baseURL$path";
+    my $fetch_timeout = undef;
+    if ($path =~ m/$PROFILE_PAGE|$PMUPROFILE_PAGE/) {
+      if ($path =~ m/[?]/) {
+        $url .= "&";
+      } else {
+        $url .= "?";
+      }
+      $url .= sprintf("seconds=%d", $main::opt_seconds);
+      $fetch_timeout = $main::opt_seconds * 1.01 + 60;
+    } else {
+      # For non-CPU profiles, we add a type-extension to
+      # the target profile file name.
+      my $suffix = $path;
+      $suffix =~ s,/,.,g;
+      $profile_file .= $suffix;
+    }
+
+    my $profile_dir = $ENV{"PPROF_TMPDIR"} || ($ENV{HOME} . "/pprof");
+    if (! -d $profile_dir) {
+      mkdir($profile_dir)
+          || die("Unable to create profile directory $profile_dir: $!\n");
+    }
+    my $tmp_profile = "$profile_dir/.tmp.$profile_file";
+    my $real_profile = "$profile_dir/$profile_file";
+
+    if ($fetch_name_only > 0) {
+      return $real_profile;
+    }
+
+    my @fetcher = AddFetchTimeout($fetch_timeout, @URL_FETCHER);
+    my $cmd = ShellEscape(@fetcher, $url) . " > " . ShellEscape($tmp_profile);
+    if ($path =~ m/$PROFILE_PAGE|$PMUPROFILE_PAGE|$CENSUSPROFILE_PAGE/){
+      print STDERR "Gathering CPU profile from $url for $main::opt_seconds seconds to\n  ${real_profile}\n";
+      if ($encourage_patience) {
+        print STDERR "Be patient...\n";
+      }
+    } else {
+      print STDERR "Fetching $path profile from $url to\n  ${real_profile}\n";
+    }
+
+    (system($cmd) == 0) || error("Failed to get profile: $cmd: $!\n");
+    (system("mv", $tmp_profile, $real_profile) == 0) || error("Unable to rename profile\n");
+    print STDERR "Wrote profile to $real_profile\n";
+    $main::collected_profile = $real_profile;
+    return $main::collected_profile;
+  }
+}
+
+# Collect profiles in parallel
+sub FetchDynamicProfiles {
+  my $items = scalar(@main::pfile_args);
+  my $levels = log($items) / log(2);
+
+  if ($items == 1) {
+    $main::profile_files[0] = FetchDynamicProfile($main::prog, $main::pfile_args[0], 0, 1);
+  } else {
+    # math rounding issues
+    if ((2 ** $levels) < $items) {
+     $levels++;
+    }
+    my $count = scalar(@main::pfile_args);
+    for (my $i = 0; $i < $count; $i++) {
+      $main::profile_files[$i] = FetchDynamicProfile($main::prog, $main::pfile_args[$i], 1, 0);
+    }
+    print STDERR "Fetching $count profiles, Be patient...\n";
+    FetchDynamicProfilesRecurse($levels, 0, 0);
+    $main::collected_profile = join(" \\\n    ", @main::profile_files);
+  }
+}
+
+# Recursively fork a process to get enough processes
+# collecting profiles
+sub FetchDynamicProfilesRecurse {
+  my $maxlevel = shift;
+  my $level = shift;
+  my $position = shift;
+
+  if (my $pid = fork()) {
+    $position = 0 | ($position << 1);
+    TryCollectProfile($maxlevel, $level, $position);
+    wait;
+  } else {
+    $position = 1 | ($position << 1);
+    TryCollectProfile($maxlevel, $level, $position);
+    cleanup();
+    exit(0);
+  }
+}
+
+# Collect a single profile
+sub TryCollectProfile {
+  my $maxlevel = shift;
+  my $level = shift;
+  my $position = shift;
+
+  if ($level >= ($maxlevel - 1)) {
+    if ($position < scalar(@main::pfile_args)) {
+      FetchDynamicProfile($main::prog, $main::pfile_args[$position], 0, 0);
+    }
+  } else {
+    FetchDynamicProfilesRecurse($maxlevel, $level+1, $position);
+  }
+}
+
+##### Parsing code #####
+
+# Provide a small streaming-read module to handle very large
+# cpu-profile files.  Stream in chunks along a sliding window.
+# Provides an interface to get one 'slot', correctly handling
+# endian-ness differences.  A slot is one 32-bit or 64-bit word
+# (depending on the input profile).  We tell endianness and bit-size
+# for the profile by looking at the first 8 bytes: in cpu profiles,
+# the second slot is always 3 (we'll accept anything that's not 0).
+BEGIN {
+  package CpuProfileStream;
+
+  sub new {
+    my ($class, $file, $fname) = @_;
+    my $self = { file        => $file,
+                 base        => 0,
+                 stride      => 512 * 1024,   # must be a multiple of bitsize/8
+                 slots       => [],
+                 unpack_code => "",           # N for big-endian, V for little
+                 perl_is_64bit => 1,          # matters if profile is 64-bit
+    };
+    bless $self, $class;
+    # Let unittests adjust the stride
+    if ($main::opt_test_stride > 0) {
+      $self->{stride} = $main::opt_test_stride;
+    }
+    # Read the first two slots to figure out bitsize and endianness.
+    my $slots = $self->{slots};
+    my $str;
+    read($self->{file}, $str, 8);
+    # Set the global $address_length based on what we see here.
+    # 8 is 32-bit (8 hexadecimal chars); 16 is 64-bit (16 hexadecimal chars).
+    $address_length = ($str eq (chr(0)x8)) ? 16 : 8;
+    if ($address_length == 8) {
+      if (substr($str, 6, 2) eq chr(0)x2) {
+        $self->{unpack_code} = 'V';  # Little-endian.
+      } elsif (substr($str, 4, 2) eq chr(0)x2) {
+        $self->{unpack_code} = 'N';  # Big-endian
+      } else {
+        ::error("$fname: header size >= 2**16\n");
+      }
+      @$slots = unpack($self->{unpack_code} . "*", $str);
+    } else {
+      # If we're a 64-bit profile, check if we're a 64-bit-capable
+      # perl.  Otherwise, each slot will be represented as a float
+      # instead of an int64, losing precision and making all the
+      # 64-bit addresses wrong.  We won't complain yet, but will
+      # later if we ever see a value that doesn't fit in 32 bits.
+      my $has_q = 0;
+      eval { $has_q = pack("Q", "1") ? 1 : 1; };
+      if (!$has_q) {
+        $self->{perl_is_64bit} = 0;
+      }
+      read($self->{file}, $str, 8);
+      if (substr($str, 4, 4) eq chr(0)x4) {
+        # We'd love to use 'Q', but it's a) not universal, b) not endian-proof.
+        $self->{unpack_code} = 'V';  # Little-endian.
+      } elsif (substr($str, 0, 4) eq chr(0)x4) {
+        $self->{unpack_code} = 'N';  # Big-endian
+      } else {
+        ::error("$fname: header size >= 2**32\n");
+      }
+      my @pair = unpack($self->{unpack_code} . "*", $str);
+      # Since we know one of the pair is 0, it's fine to just add them.
+      @$slots = (0, $pair[0] + $pair[1]);
+    }
+    return $self;
+  }
+
+  # Load more data when we access slots->get(X) which is not yet in memory.
+  sub overflow {
+    my ($self) = @_;
+    my $slots = $self->{slots};
+    $self->{base} += $#$slots + 1;   # skip over data we're replacing
+    my $str;
+    read($self->{file}, $str, $self->{stride});
+    if ($address_length == 8) {      # the 32-bit case
+      # This is the easy case: unpack provides 32-bit unpacking primitives.
+      @$slots = unpack($self->{unpack_code} . "*", $str);
+    } else {
+      # We need to unpack 32 bits at a time and combine.
+      my @b32_values = unpack($self->{unpack_code} . "*", $str);
+      my @b64_values = ();
+      for (my $i = 0; $i < $#b32_values; $i += 2) {
+        # TODO(csilvers): if this is a 32-bit perl, the math below
+        #    could end up in a too-large int, which perl will promote
+        #    to a double, losing necessary precision.  Deal with that.
+        #    Right now, we just die.
+        my ($lo, $hi) = ($b32_values[$i], $b32_values[$i+1]);
+        if ($self->{unpack_code} eq 'N') {    # big-endian
+          ($lo, $hi) = ($hi, $lo);
+        }
+        my $value = $lo + $hi * (2**32);
+        if (!$self->{perl_is_64bit} &&   # check value is exactly represented
+            (($value % (2**32)) != $lo || int($value / (2**32)) != $hi)) {
+          ::error("Need a 64-bit perl to process this 64-bit profile.\n");
+        }
+        push(@b64_values, $value);
+      }
+      @$slots = @b64_values;
+    }
+  }
+
+  # Access the i-th long in the file (logically), or -1 at EOF.
+  sub get {
+    my ($self, $idx) = @_;
+    my $slots = $self->{slots};
+    while ($#$slots >= 0) {
+      if ($idx < $self->{base}) {
+        # The only time we expect a reference to $slots[$i - something]
+        # after referencing $slots[$i] is reading the very first header.
+        # Since $stride > |header|, that shouldn't cause any lookback
+        # errors.  And everything after the header is sequential.
+        print STDERR "Unexpected look-back reading CPU profile";
+        return -1;   # shrug, don't know what better to return
+      } elsif ($idx > $self->{base} + $#$slots) {
+        $self->overflow();
+      } else {
+        return $slots->[$idx - $self->{base}];
+      }
+    }
+    # If we get here, $slots is [], which means we've reached EOF
+    return -1;  # unique since slots is supposed to hold unsigned numbers
+  }
+}
+
+# Reads the top, 'header' section of a profile, and returns the last
+# line of the header, commonly called a 'header line'.  The header
+# section of a profile consists of zero or more 'command' lines that
+# are instructions to pprof, which pprof executes when reading the
+# header.  All 'command' lines start with a %.  After the command
+# lines is the 'header line', which is a profile-specific line that
+# indicates what type of profile it is, and perhaps other global
+# information about the profile.  For instance, here's a header line
+# for a heap profile:
+#   heap profile:     53:    38236 [  5525:  1284029] @ heapprofile
+# For historical reasons, the CPU profile does not contain a text-
+# readable header line.  If the profile looks like a CPU profile,
+# this function returns "".  If no header line could be found, this
+# function returns undef.
+#
+# The following commands are recognized:
+#   %warn -- emit the rest of this line to stderr, prefixed by 'WARNING:'
+#
+# The input file should be in binmode.
+sub ReadProfileHeader {
+  local *PROFILE = shift;
+  my $firstchar = "";
+  my $line = "";
+  read(PROFILE, $firstchar, 1);
+  seek(PROFILE, -1, 1);                    # unread the firstchar
+  if ($firstchar !~ /[[:print:]]/) {       # is not a text character
+    return "";
+  }
+  while (defined($line = <PROFILE>)) {
+    $line =~ s/\r//g;   # turn windows-looking lines into unix-looking lines
+    if ($line =~ /^%warn\s+(.*)/) {        # 'warn' command
+      # Note this matches both '%warn blah\n' and '%warn\n'.
+      print STDERR "WARNING: $1\n";        # print the rest of the line
+    } elsif ($line =~ /^%/) {
+      print STDERR "Ignoring unknown command from profile header: $line";
+    } else {
+      # End of commands, must be the header line.
+      return $line;
+    }
+  }
+  return undef;     # got to EOF without seeing a header line
+}
+
+sub IsSymbolizedProfileFile {
+  my $file_name = shift;
+  if (!(-e $file_name) || !(-r $file_name)) {
+    return 0;
+  }
+  # Check if the file contains a symbol-section marker.
+  open(TFILE, "<$file_name");
+  binmode TFILE;
+  my $firstline = ReadProfileHeader(*TFILE);
+  close(TFILE);
+  if (!$firstline) {
+    return 0;
+  }
+  $SYMBOL_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $symbol_marker = $&;
+  return $firstline =~ /^--- *$symbol_marker/;
+}
+
+# Parse profile generated by common/profiler.cc and return a reference
+# to a map:
+#      $result->{version}     Version number of profile file
+#      $result->{period}      Sampling period (in microseconds)
+#      $result->{profile}     Profile object
+#      $result->{map}         Memory map info from profile
+#      $result->{pcs}         Hash of all PC values seen, key is hex address
+sub ReadProfile {
+  my $prog = shift;
+  my $fname = shift;
+  my $result;            # return value
+
+  $CONTENTION_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $contention_marker = $&;
+  $GROWTH_PAGE  =~ m,[^/]+$,;    # matches everything after the last slash
+  my $growth_marker = $&;
+  $SYMBOL_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $symbol_marker = $&;
+  $PROFILE_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $profile_marker = $&;
+
+  # Look at first line to see if it is a heap or a CPU profile.
+  # CPU profile may start with no header at all, and just binary data
+  # (starting with \0\0\0\0) -- in that case, don't try to read the
+  # whole firstline, since it may be gigabytes(!) of data.
+  open(PROFILE, "<$fname") || error("$fname: $!\n");
+  binmode PROFILE;      # New perls do UTF-8 processing
+  my $header = ReadProfileHeader(*PROFILE);
+  if (!defined($header)) {   # means "at EOF"
+    error("Profile is empty.\n");
+  }
+
+  my $symbols;
+  if ($header =~ m/^--- *$symbol_marker/o) {
+    # Verify that the user asked for a symbolized profile
+    if (!$main::use_symbolized_profile) {
+      # we have both a binary and symbolized profiles, abort
+      error("FATAL ERROR: Symbolized profile\n   $fname\ncannot be used with " .
+            "a binary arg. Try again without passing\n   $prog\n");
+    }
+    # Read the symbol section of the symbolized profile file.
+    $symbols = ReadSymbols(*PROFILE{IO});
+    # Read the next line to get the header for the remaining profile.
+    $header = ReadProfileHeader(*PROFILE) || "";
+  }
+
+  $main::profile_type = '';
+  if ($header =~ m/^heap profile:.*$growth_marker/o) {
+    $main::profile_type = 'growth';
+    $result =  ReadHeapProfile($prog, *PROFILE, $header);
+  } elsif ($header =~ m/^heap profile:/) {
+    $main::profile_type = 'heap';
+    $result =  ReadHeapProfile($prog, *PROFILE, $header);
+  } elsif ($header =~ m/^--- *$contention_marker/o) {
+    $main::profile_type = 'contention';
+    $result = ReadSynchProfile($prog, *PROFILE);
+  } elsif ($header =~ m/^--- *Stacks:/) {
+    print STDERR
+      "Old format contention profile: mistakenly reports " .
+      "condition variable signals as lock contentions.\n";
+    $main::profile_type = 'contention';
+    $result = ReadSynchProfile($prog, *PROFILE);
+  } elsif ($header =~ m/^--- *$profile_marker/) {
+    # the binary cpu profile data starts immediately after this line
+    $main::profile_type = 'cpu';
+    $result = ReadCPUProfile($prog, $fname, *PROFILE);
+  } else {
+    if (defined($symbols)) {
+      # a symbolized profile contains a format we don't recognize, bail out
+      error("$fname: Cannot recognize profile section after symbols.\n");
+    }
+    # no ascii header present -- must be a CPU profile
+    $main::profile_type = 'cpu';
+    $result = ReadCPUProfile($prog, $fname, *PROFILE);
+  }
+
+  close(PROFILE);
+
+  # if we got symbols along with the profile, return those as well
+  if (defined($symbols)) {
+    $result->{symbols} = $symbols;
+  }
+
+  return $result;
+}
+
+# Subtract one from caller pc so we map back to call instr.
+# However, don't do this if we're reading a symbolized profile
+# file, in which case the subtract-one was done when the file
+# was written.
+#
+# We apply the same logic to all readers, though ReadCPUProfile uses an
+# independent implementation.
+sub FixCallerAddresses {
+  my $stack = shift;
+  if ($main::use_symbolized_profile) {
+    return $stack;
+  } else {
+    $stack =~ /(\s)/;
+    my $delimiter = $1;
+    my @addrs = split(' ', $stack);
+    my @fixedaddrs;
+    $#fixedaddrs = $#addrs;
+    if ($#addrs >= 0) {
+      $fixedaddrs[0] = $addrs[0];
+    }
+    for (my $i = 1; $i <= $#addrs; $i++) {
+      $fixedaddrs[$i] = AddressSub($addrs[$i], "0x1");
+    }
+    return join $delimiter, @fixedaddrs;
+  }
+}
+
+# CPU profile reader
+sub ReadCPUProfile {
+  my $prog = shift;
+  my $fname = shift;       # just used for logging
+  local *PROFILE = shift;
+  my $version;
+  my $period;
+  my $i;
+  my $profile = {};
+  my $pcs = {};
+
+  # Parse string into array of slots.
+  my $slots = CpuProfileStream->new(*PROFILE, $fname);
+
+  # Read header.  The current header version is a 5-element structure
+  # containing:
+  #   0: header count (always 0)
+  #   1: header "words" (after this one: 3)
+  #   2: format version (0)
+  #   3: sampling period (usec)
+  #   4: unused padding (always 0)
+  if ($slots->get(0) != 0 ) {
+    error("$fname: not a profile file, or old format profile file\n");
+  }
+  $i = 2 + $slots->get(1);
+  $version = $slots->get(2);
+  $period = $slots->get(3);
+  # Do some sanity checking on these header values.
+  if ($version > (2**32) || $period > (2**32) || $i > (2**32) || $i < 5) {
+    error("$fname: not a profile file, or corrupted profile file\n");
+  }
+
+  # Parse profile
+  while ($slots->get($i) != -1) {
+    my $n = $slots->get($i++);
+    my $d = $slots->get($i++);
+    if ($d > (2**16)) {  # TODO(csilvers): what's a reasonable max-stack-depth?
+      my $addr = sprintf("0%o", $i * ($address_length == 8 ? 4 : 8));
+      print STDERR "At index $i (address $addr):\n";
+      error("$fname: stack trace depth >= 2**32\n");
+    }
+    if ($slots->get($i) == 0) {
+      # End of profile data marker
+      $i += $d;
+      last;
+    }
+
+    # Make key out of the stack entries
+    my @k = ();
+    for (my $j = 0; $j < $d; $j++) {
+      my $pc = $slots->get($i+$j);
+      # Subtract one from caller pc so we map back to call instr.
+      # However, don't do this if we're reading a symbolized profile
+      # file, in which case the subtract-one was done when the file
+      # was written.
+      if ($j > 0 && !$main::use_symbolized_profile) {
+        $pc--;
+      }
+      $pc = sprintf("%0*x", $address_length, $pc);
+      $pcs->{$pc} = 1;
+      push @k, $pc;
+    }
+
+    AddEntry($profile, (join "\n", @k), $n);
+    $i += $d;
+  }
+
+  # Parse map
+  my $map = '';
+  seek(PROFILE, $i * 4, 0);
+  read(PROFILE, $map, (stat PROFILE)[7]);
+
+  my $r = {};
+  $r->{version} = $version;
+  $r->{period} = $period;
+  $r->{profile} = $profile;
+  $r->{libs} = ParseLibraries($prog, $map, $pcs);
+  $r->{pcs} = $pcs;
+
+  return $r;
+}
+
+sub ReadHeapProfile {
+  my $prog = shift;
+  local *PROFILE = shift;
+  my $header = shift;
+
+  my $index = 1;
+  if ($main::opt_inuse_space) {
+    $index = 1;
+  } elsif ($main::opt_inuse_objects) {
+    $index = 0;
+  } elsif ($main::opt_alloc_space) {
+    $index = 3;
+  } elsif ($main::opt_alloc_objects) {
+    $index = 2;
+  }
+
+  # Find the type of this profile.  The header line looks like:
+  #    heap profile:   1246:  8800744 [  1246:  8800744] @ <heap-url>/266053
+  # There are two pairs <count: size>, the first inuse objects/space, and the
+  # second allocated objects/space.  This is followed optionally by a profile
+  # type, and if that is present, optionally by a sampling frequency.
+  # For remote heap profiles (v1):
+  # The interpretation of the sampling frequency is that the profiler, for
+  # each sample, calculates a uniformly distributed random integer less than
+  # the given value, and records the next sample after that many bytes have
+  # been allocated.  Therefore, the expected sample interval is half of the
+  # given frequency.  By default, if not specified, the expected sample
+  # interval is 128KB.  Only remote-heap-page profiles are adjusted for
+  # sample size.
+  # For remote heap profiles (v2):
+  # The sampling frequency is the rate of a Poisson process. This means that
+  # the probability of sampling an allocation of size X with sampling rate Y
+  # is 1 - exp(-X/Y)
+  # For version 2, a typical header line might look like this:
+  # heap profile:   1922: 127792360 [  1922: 127792360] @ <heap-url>_v2/524288
+  # the trailing number (524288) is the sampling rate. (Version 1 showed
+  # double the 'rate' here)
+  my $sampling_algorithm = 0;
+  my $sample_adjustment = 0;
+  chomp($header);
+  my $type = "unknown";
+  if ($header =~ m"^heap profile:\s*(\d+):\s+(\d+)\s+\[\s*(\d+):\s+(\d+)\](\s*@\s*([^/]*)(/(\d+))?)?") {
+    if (defined($6) && ($6 ne '')) {
+      $type = $6;
+      my $sample_period = $8;
+      # $type is "heapprofile" for profiles generated by the
+      # heap-profiler, and either "heap" or "heap_v2" for profiles
+      # generated by sampling directly within tcmalloc.  It can also
+      # be "growth" for heap-growth profiles.  The first is typically
+      # found for profiles generated locally, and the others for
+      # remote profiles.
+      if (($type eq "heapprofile") || ($type !~ /heap/) ) {
+        # No need to adjust for the sampling rate with heap-profiler-derived data
+        $sampling_algorithm = 0;
+      } elsif ($type =~ /_v2/) {
+        $sampling_algorithm = 2;     # version 2 sampling
+        if (defined($sample_period) && ($sample_period ne '')) {
+          $sample_adjustment = int($sample_period);
+        }
+      } else {
+        $sampling_algorithm = 1;     # version 1 sampling
+        if (defined($sample_period) && ($sample_period ne '')) {
+          $sample_adjustment = int($sample_period)/2;
+        }
+      }
+    } else {
+      # We detect whether or not this is a remote-heap profile by checking
+      # that the total-allocated stats ($n2,$s2) are exactly the
+      # same as the in-use stats ($n1,$s1).  It is remotely conceivable
+      # that a non-remote-heap profile may pass this check, but it is hard
+      # to imagine how that could happen.
+      # In this case it's so old it's guaranteed to be remote-heap version 1.
+      my ($n1, $s1, $n2, $s2) = ($1, $2, $3, $4);
+      if (($n1 == $n2) && ($s1 == $s2)) {
+        # This is likely to be a remote-heap based sample profile
+        $sampling_algorithm = 1;
+      }
+    }
+  }
+
+  if ($sampling_algorithm > 0) {
+    # For remote-heap generated profiles, adjust the counts and sizes to
+    # account for the sample rate (we sample once every 128KB by default).
+    if ($sample_adjustment == 0) {
+      # Turn on profile adjustment.
+      $sample_adjustment = 128*1024;
+      print STDERR "Adjusting heap profiles for 1-in-128KB sampling rate\n";
+    } else {
+      printf STDERR ("Adjusting heap profiles for 1-in-%d sampling rate\n",
+                     $sample_adjustment);
+    }
+    if ($sampling_algorithm > 1) {
+      # We don't bother printing anything for the original version (version 1)
+      printf STDERR "Heap version $sampling_algorithm\n";
+    }
+  }
+
+  my $profile = {};
+  my $pcs = {};
+  my $map = "";
+
+  while (<PROFILE>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    if (/^MAPPED_LIBRARIES:/) {
+      # Read the /proc/self/maps data
+      while (<PROFILE>) {
+        s/\r//g;         # turn windows-looking lines into unix-looking lines
+        $map .= $_;
+      }
+      last;
+    }
+
+    if (/^--- Memory map:/) {
+      # Read /proc/self/maps data as formatted by DumpAddressMap()
+      my $buildvar = "";
+      while (<PROFILE>) {
+        s/\r//g;         # turn windows-looking lines into unix-looking lines
+        # Parse "build=<dir>" specification if supplied
+        if (m/^\s*build=(.*)\n/) {
+          $buildvar = $1;
+        }
+
+        # Expand "$build" variable if available
+        $_ =~ s/\$build\b/$buildvar/g;
+
+        $map .= $_;
+      }
+      last;
+    }
+
+    # Read entry of the form:
+    #  <count1>: <bytes1> [<count2>: <bytes2>] @ a1 a2 a3 ... an
+    s/^\s*//;
+    s/\s*$//;
+    if (m/^\s*(\d+):\s+(\d+)\s+\[\s*(\d+):\s+(\d+)\]\s+@\s+(.*)$/) {
+      my $stack = $5;
+      my ($n1, $s1, $n2, $s2) = ($1, $2, $3, $4);
+
+      if ($sample_adjustment) {
+        if ($sampling_algorithm == 2) {
+          # Remote-heap version 2
+          # The sampling frequency is the rate of a Poisson process.
+          # This means that the probability of sampling an allocation of
+          # size X with sampling rate Y is 1 - exp(-X/Y)
+          if ($n1 != 0) {
+            my $ratio = (($s1*1.0)/$n1)/($sample_adjustment);
+            my $scale_factor = 1/(1 - exp(-$ratio));
+            $n1 *= $scale_factor;
+            $s1 *= $scale_factor;
+          }
+          if ($n2 != 0) {
+            my $ratio = (($s2*1.0)/$n2)/($sample_adjustment);
+            my $scale_factor = 1/(1 - exp(-$ratio));
+            $n2 *= $scale_factor;
+            $s2 *= $scale_factor;
+          }
+        } else {
+          # Remote-heap version 1
+          my $ratio;
+          $ratio = (($s1*1.0)/$n1)/($sample_adjustment);
+          if ($ratio < 1) {
+            $n1 /= $ratio;
+            $s1 /= $ratio;
+          }
+          $ratio = (($s2*1.0)/$n2)/($sample_adjustment);
+          if ($ratio < 1) {
+            $n2 /= $ratio;
+            $s2 /= $ratio;
+          }
+        }
+      }
+
+      my @counts = ($n1, $s1, $n2, $s2);
+      AddEntries($profile, $pcs, FixCallerAddresses($stack), $counts[$index]);
+    }
+  }
+
+  my $r = {};
+  $r->{version} = "heap";
+  $r->{period} = 1;
+  $r->{profile} = $profile;
+  $r->{libs} = ParseLibraries($prog, $map, $pcs);
+  $r->{pcs} = $pcs;
+  return $r;
+}
+
+sub ReadSynchProfile {
+  my $prog = shift;
+  local *PROFILE = shift;
+  my $header = shift;
+
+  my $map = '';
+  my $profile = {};
+  my $pcs = {};
+  my $sampling_period = 1;
+  my $cyclespernanosec = 2.8;   # Default assumption for old binaries
+  my $seen_clockrate = 0;
+  my $line;
+
+  my $index = 0;
+  if ($main::opt_total_delay) {
+    $index = 0;
+  } elsif ($main::opt_contentions) {
+    $index = 1;
+  } elsif ($main::opt_mean_delay) {
+    $index = 2;
+  }
+
+  while ( $line = <PROFILE> ) {
+    $line =~ s/\r//g;      # turn windows-looking lines into unix-looking lines
+    if ( $line =~ /^\s*(\d+)\s+(\d+) \@\s*(.*?)\s*$/ ) {
+      my ($cycles, $count, $stack) = ($1, $2, $3);
+
+      # Convert cycles to nanoseconds
+      $cycles /= $cyclespernanosec;
+
+      # Adjust for sampling done by application
+      $cycles *= $sampling_period;
+      $count *= $sampling_period;
+
+      my @values = ($cycles, $count, $cycles / $count);
+      AddEntries($profile, $pcs, FixCallerAddresses($stack), $values[$index]);
+
+    } elsif ( $line =~ /^(slow release).*thread \d+  \@\s*(.*?)\s*$/ ||
+              $line =~ /^\s*(\d+) \@\s*(.*?)\s*$/ ) {
+      my ($cycles, $stack) = ($1, $2);
+      if ($cycles !~ /^\d+$/) {
+        next;
+      }
+
+      # Convert cycles to nanoseconds
+      $cycles /= $cyclespernanosec;
+
+      # Adjust for sampling done by application
+      $cycles *= $sampling_period;
+
+      AddEntries($profile, $pcs, FixCallerAddresses($stack), $cycles);
+
+    } elsif ( $line =~ m/^([a-z][^=]*)=(.*)$/ ) {
+      my ($variable, $value) = ($1,$2);
+      for ($variable, $value) {
+        s/^\s+//;
+        s/\s+$//;
+      }
+      if ($variable eq "cycles/second") {
+        $cyclespernanosec = $value / 1e9;
+        $seen_clockrate = 1;
+      } elsif ($variable eq "sampling period") {
+        $sampling_period = $value;
+      } elsif ($variable eq "ms since reset") {
+        # Currently nothing is done with this value in pprof
+        # So we just silently ignore it for now
+      } elsif ($variable eq "discarded samples") {
+        # Currently nothing is done with this value in pprof
+        # So we just silently ignore it for now
+      } else {
+        printf STDERR ("Ignoring unnknown variable in /contention output: " .
+                       "'%s' = '%s'\n",$variable,$value);
+      }
+    } else {
+      # Memory map entry
+      $map .= $line;
+    }
+  }
+
+  if (!$seen_clockrate) {
+    printf STDERR ("No cycles/second entry in profile; Guessing %.1f GHz\n",
+                   $cyclespernanosec);
+  }
+
+  my $r = {};
+  $r->{version} = 0;
+  $r->{period} = $sampling_period;
+  $r->{profile} = $profile;
+  $r->{libs} = ParseLibraries($prog, $map, $pcs);
+  $r->{pcs} = $pcs;
+  return $r;
+}
+
+# Given a hex value in the form "0x1abcd" or "1abcd", return either
+# "0001abcd" or "000000000001abcd", depending on the current (global)
+# address length.
+sub HexExtend {
+  my $addr = shift;
+
+  $addr =~ s/^(0x)?0*//;
+  my $zeros_needed = $address_length - length($addr);
+  if ($zeros_needed < 0) {
+    printf STDERR "Warning: address $addr is longer than address length $address_length\n";
+    return $addr;
+  }
+  return ("0" x $zeros_needed) . $addr;
+}
+
+##### Symbol extraction #####
+
+# Aggressively search the lib_prefix values for the given library
+# If all else fails, just return the name of the library unmodified.
+# If the lib_prefix is "/my/path,/other/path" and $file is "/lib/dir/mylib.so"
+# it will search the following locations in this order, until it finds a file:
+#   /my/path/lib/dir/mylib.so
+#   /other/path/lib/dir/mylib.so
+#   /my/path/dir/mylib.so
+#   /other/path/dir/mylib.so
+#   /my/path/mylib.so
+#   /other/path/mylib.so
+#   /lib/dir/mylib.so              (returned as last resort)
+sub FindLibrary {
+  my $file = shift;
+  my $suffix = $file;
+
+  # Search for the library as described above
+  do {
+    foreach my $prefix (@prefix_list) {
+      my $fullpath = $prefix . $suffix;
+      if (-e $fullpath) {
+        return $fullpath;
+      }
+    }
+  } while ($suffix =~ s|^/[^/]+/|/|);
+  return $file;
+}
+
+# Return path to library with debugging symbols.
+# For libc libraries, the copy in /usr/lib/debug contains debugging symbols
+sub DebuggingLibrary {
+  my $file = shift;
+  if ($file =~ m|^/| && -f "/usr/lib/debug$file") {
+    return "/usr/lib/debug$file";
+  }
+  return undef;
+}
+
+# Parse text section header of a library using objdump
+sub ParseTextSectionHeaderFromObjdump {
+  my $lib = shift;
+
+  my $size = undef;
+  my $vma;
+  my $file_offset;
+  # Get objdump output from the library file to figure out how to
+  # map between mapped addresses and addresses in the library.
+  my $cmd = ShellEscape($obj_tool_map{"objdump"}, "-h", $lib);
+  open(OBJDUMP, "$cmd |") || error("$cmd: $!\n");
+  while (<OBJDUMP>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    # Idx Name          Size      VMA       LMA       File off  Algn
+    #  10 .text         00104b2c  420156f0  420156f0  000156f0  2**4
+    # For 64-bit objects, VMA and LMA will be 16 hex digits, size and file
+    # offset may still be 8.  But AddressSub below will still handle that.
+    my @x = split;
+    if (($#x >= 6) && ($x[1] eq '.text')) {
+      $size = $x[2];
+      $vma = $x[3];
+      $file_offset = $x[5];
+      last;
+    }
+  }
+  close(OBJDUMP);
+
+  if (!defined($size)) {
+    return undef;
+  }
+
+  my $r = {};
+  $r->{size} = $size;
+  $r->{vma} = $vma;
+  $r->{file_offset} = $file_offset;
+
+  return $r;
+}
+
+# Parse text section header of a library using otool (on OS X)
+sub ParseTextSectionHeaderFromOtool {
+  my $lib = shift;
+
+  my $size = undef;
+  my $vma = undef;
+  my $file_offset = undef;
+  # Get otool output from the library file to figure out how to
+  # map between mapped addresses and addresses in the library.
+  my $command = ShellEscape($obj_tool_map{"otool"}, "-l", $lib);
+  open(OTOOL, "$command |") || error("$command: $!\n");
+  my $cmd = "";
+  my $sectname = "";
+  my $segname = "";
+  foreach my $line (<OTOOL>) {
+    $line =~ s/\r//g;      # turn windows-looking lines into unix-looking lines
+    # Load command <#>
+    #       cmd LC_SEGMENT
+    # [...]
+    # Section
+    #   sectname __text
+    #    segname __TEXT
+    #       addr 0x000009f8
+    #       size 0x00018b9e
+    #     offset 2552
+    #      align 2^2 (4)
+    # We will need to strip off the leading 0x from the hex addresses,
+    # and convert the offset into hex.
+    if ($line =~ /Load command/) {
+      $cmd = "";
+      $sectname = "";
+      $segname = "";
+    } elsif ($line =~ /Section/) {
+      $sectname = "";
+      $segname = "";
+    } elsif ($line =~ /cmd (\w+)/) {
+      $cmd = $1;
+    } elsif ($line =~ /sectname (\w+)/) {
+      $sectname = $1;
+    } elsif ($line =~ /segname (\w+)/) {
+      $segname = $1;
+    } elsif (!(($cmd eq "LC_SEGMENT" || $cmd eq "LC_SEGMENT_64") &&
+               $sectname eq "__text" &&
+               $segname eq "__TEXT")) {
+      next;
+    } elsif ($line =~ /\baddr 0x([0-9a-fA-F]+)/) {
+      $vma = $1;
+    } elsif ($line =~ /\bsize 0x([0-9a-fA-F]+)/) {
+      $size = $1;
+    } elsif ($line =~ /\boffset ([0-9]+)/) {
+      $file_offset = sprintf("%016x", $1);
+    }
+    if (defined($vma) && defined($size) && defined($file_offset)) {
+      last;
+    }
+  }
+  close(OTOOL);
+
+  if (!defined($vma) || !defined($size) || !defined($file_offset)) {
+     return undef;
+  }
+
+  my $r = {};
+  $r->{size} = $size;
+  $r->{vma} = $vma;
+  $r->{file_offset} = $file_offset;
+
+  return $r;
+}
+
+sub ParseTextSectionHeader {
+  # obj_tool_map("otool") is only defined if we're in a Mach-O environment
+  if (defined($obj_tool_map{"otool"})) {
+    my $r = ParseTextSectionHeaderFromOtool(@_);
+    if (defined($r)){
+      return $r;
+    }
+  }
+  # If otool doesn't work, or we don't have it, fall back to objdump
+  return ParseTextSectionHeaderFromObjdump(@_);
+}
+
+# Split /proc/pid/maps dump into a list of libraries
+sub ParseLibraries {
+  return if $main::use_symbol_page;  # We don't need libraries info.
+  my $prog = shift;
+  my $map = shift;
+  my $pcs = shift;
+
+  my $result = [];
+  my $h = "[a-f0-9]+";
+  my $zero_offset = HexExtend("0");
+
+  my $buildvar = "";
+  foreach my $l (split("\n", $map)) {
+    if ($l =~ m/^\s*build=(.*)$/) {
+      $buildvar = $1;
+    }
+
+    my $start;
+    my $finish;
+    my $offset;
+    my $lib;
+    if ($l =~ /^($h)-($h)\s+..x.\s+($h)\s+\S+:\S+\s+\d+\s+(\S+\.(so|dll|dylib|bundle)((\.\d+)+\w*(\.\d+){0,3})?)$/i) {
+      # Full line from /proc/self/maps.  Example:
+      #   40000000-40015000 r-xp 00000000 03:01 12845071   /lib/ld-2.3.2.so
+      $start = HexExtend($1);
+      $finish = HexExtend($2);
+      $offset = HexExtend($3);
+      $lib = $4;
+      $lib =~ s|\\|/|g;     # turn windows-style paths into unix-style paths
+    } elsif ($l =~ /^\s*($h)-($h):\s*(\S+\.so(\.\d+)*)/) {
+      # Cooked line from DumpAddressMap.  Example:
+      #   40000000-40015000: /lib/ld-2.3.2.so
+      $start = HexExtend($1);
+      $finish = HexExtend($2);
+      $offset = $zero_offset;
+      $lib = $3;
+    } else {
+      next;
+    }
+
+    # Expand "$build" variable if available
+    $lib =~ s/\$build\b/$buildvar/g;
+
+    $lib = FindLibrary($lib);
+
+    # Check for pre-relocated libraries, which use pre-relocated symbol tables
+    # and thus require adjusting the offset that we'll use to translate
+    # VM addresses into symbol table addresses.
+    # Only do this if we're not going to fetch the symbol table from a
+    # debugging copy of the library.
+    if (!DebuggingLibrary($lib)) {
+      my $text = ParseTextSectionHeader($lib);
+      if (defined($text)) {
+         my $vma_offset = AddressSub($text->{vma}, $text->{file_offset});
+         $offset = AddressAdd($offset, $vma_offset);
+      }
+    }
+
+    push(@{$result}, [$lib, $start, $finish, $offset]);
+  }
+
+  # Append special entry for additional library (not relocated)
+  if ($main::opt_lib ne "") {
+    my $text = ParseTextSectionHeader($main::opt_lib);
+    if (defined($text)) {
+       my $start = $text->{vma};
+       my $finish = AddressAdd($start, $text->{size});
+
+       push(@{$result}, [$main::opt_lib, $start, $finish, $start]);
+    }
+  }
+
+  # Append special entry for the main program.  This covers
+  # 0..max_pc_value_seen, so that we assume pc values not found in one
+  # of the library ranges will be treated as coming from the main
+  # program binary.
+  my $min_pc = HexExtend("0");
+  my $max_pc = $min_pc;          # find the maximal PC value in any sample
+  foreach my $pc (keys(%{$pcs})) {
+    if (HexExtend($pc) gt $max_pc) { $max_pc = HexExtend($pc); }
+  }
+  push(@{$result}, [$prog, $min_pc, $max_pc, $zero_offset]);
+
+  return $result;
+}
+
+# Add two hex addresses of length $address_length.
+# Run pprof --test for unit test if this is changed.
+sub AddressAdd {
+  my $addr1 = shift;
+  my $addr2 = shift;
+  my $sum;
+
+  if ($address_length == 8) {
+    # Perl doesn't cope with wraparound arithmetic, so do it explicitly:
+    $sum = (hex($addr1)+hex($addr2)) % (0x10000000 * 16);
+    return sprintf("%08x", $sum);
+
+  } else {
+    # Do the addition in 7-nibble chunks to trivialize carry handling.
+
+    if ($main::opt_debug and $main::opt_test) {
+      print STDERR "AddressAdd $addr1 + $addr2 = ";
+    }
+
+    my $a1 = substr($addr1,-7);
+    $addr1 = substr($addr1,0,-7);
+    my $a2 = substr($addr2,-7);
+    $addr2 = substr($addr2,0,-7);
+    $sum = hex($a1) + hex($a2);
+    my $c = 0;
+    if ($sum > 0xfffffff) {
+      $c = 1;
+      $sum -= 0x10000000;
+    }
+    my $r = sprintf("%07x", $sum);
+
+    $a1 = substr($addr1,-7);
+    $addr1 = substr($addr1,0,-7);
+    $a2 = substr($addr2,-7);
+    $addr2 = substr($addr2,0,-7);
+    $sum = hex($a1) + hex($a2) + $c;
+    $c = 0;
+    if ($sum > 0xfffffff) {
+      $c = 1;
+      $sum -= 0x10000000;
+    }
+    $r = sprintf("%07x", $sum) . $r;
+
+    $sum = hex($addr1) + hex($addr2) + $c;
+    if ($sum > 0xff) { $sum -= 0x100; }
+    $r = sprintf("%02x", $sum) . $r;
+
+    if ($main::opt_debug and $main::opt_test) { print STDERR "$r\n"; }
+
+    return $r;
+  }
+}
+
+
+# Subtract two hex addresses of length $address_length.
+# Run pprof --test for unit test if this is changed.
+sub AddressSub {
+  my $addr1 = shift;
+  my $addr2 = shift;
+  my $diff;
+
+  if ($address_length == 8) {
+    # Perl doesn't cope with wraparound arithmetic, so do it explicitly:
+    $diff = (hex($addr1)-hex($addr2)) % (0x10000000 * 16);
+    return sprintf("%08x", $diff);
+
+  } else {
+    # Do the addition in 7-nibble chunks to trivialize borrow handling.
+    # if ($main::opt_debug) { print STDERR "AddressSub $addr1 - $addr2 = "; }
+
+    my $a1 = hex(substr($addr1,-7));
+    $addr1 = substr($addr1,0,-7);
+    my $a2 = hex(substr($addr2,-7));
+    $addr2 = substr($addr2,0,-7);
+    my $b = 0;
+    if ($a2 > $a1) {
+      $b = 1;
+      $a1 += 0x10000000;
+    }
+    $diff = $a1 - $a2;
+    my $r = sprintf("%07x", $diff);
+
+    $a1 = hex(substr($addr1,-7));
+    $addr1 = substr($addr1,0,-7);
+    $a2 = hex(substr($addr2,-7)) + $b;
+    $addr2 = substr($addr2,0,-7);
+    $b = 0;
+    if ($a2 > $a1) {
+      $b = 1;
+      $a1 += 0x10000000;
+    }
+    $diff = $a1 - $a2;
+    $r = sprintf("%07x", $diff) . $r;
+
+    $a1 = hex($addr1);
+    $a2 = hex($addr2) + $b;
+    if ($a2 > $a1) { $a1 += 0x100; }
+    $diff = $a1 - $a2;
+    $r = sprintf("%02x", $diff) . $r;
+
+    # if ($main::opt_debug) { print STDERR "$r\n"; }
+
+    return $r;
+  }
+}
+
+# Increment a hex addresses of length $address_length.
+# Run pprof --test for unit test if this is changed.
+sub AddressInc {
+  my $addr = shift;
+  my $sum;
+
+  if ($address_length == 8) {
+    # Perl doesn't cope with wraparound arithmetic, so do it explicitly:
+    $sum = (hex($addr)+1) % (0x10000000 * 16);
+    return sprintf("%08x", $sum);
+
+  } else {
+    # Do the addition in 7-nibble chunks to trivialize carry handling.
+    # We are always doing this to step through the addresses in a function,
+    # and will almost never overflow the first chunk, so we check for this
+    # case and exit early.
+
+    # if ($main::opt_debug) { print STDERR "AddressInc $addr1 = "; }
+
+    my $a1 = substr($addr,-7);
+    $addr = substr($addr,0,-7);
+    $sum = hex($a1) + 1;
+    my $r = sprintf("%07x", $sum);
+    if ($sum <= 0xfffffff) {
+      $r = $addr . $r;
+      # if ($main::opt_debug) { print STDERR "$r\n"; }
+      return HexExtend($r);
+    } else {
+      $r = "0000000";
+    }
+
+    $a1 = substr($addr,-7);
+    $addr = substr($addr,0,-7);
+    $sum = hex($a1) + 1;
+    $r = sprintf("%07x", $sum) . $r;
+    if ($sum <= 0xfffffff) {
+      $r = $addr . $r;
+      # if ($main::opt_debug) { print STDERR "$r\n"; }
+      return HexExtend($r);
+    } else {
+      $r = "00000000000000";
+    }
+
+    $sum = hex($addr) + 1;
+    if ($sum > 0xff) { $sum -= 0x100; }
+    $r = sprintf("%02x", $sum) . $r;
+
+    # if ($main::opt_debug) { print STDERR "$r\n"; }
+    return $r;
+  }
+}
+
+# Extract symbols for all PC values found in profile
+sub ExtractSymbols {
+  my $libs = shift;
+  my $pcset = shift;
+
+  my $symbols = {};
+
+  # Map each PC value to the containing library.  To make this faster,
+  # we sort libraries by their starting pc value (highest first), and
+  # advance through the libraries as we advance the pc.  Sometimes the
+  # addresses of libraries may overlap with the addresses of the main
+  # binary, so to make sure the libraries 'win', we iterate over the
+  # libraries in reverse order (which assumes the binary doesn't start
+  # in the middle of a library, which seems a fair assumption).
+  my @pcs = (sort { $a cmp $b } keys(%{$pcset}));  # pcset is 0-extended strings
+  foreach my $lib (sort {$b->[1] cmp $a->[1]} @{$libs}) {
+    my $libname = $lib->[0];
+    my $start = $lib->[1];
+    my $finish = $lib->[2];
+    my $offset = $lib->[3];
+
+    # Get list of pcs that belong in this library.
+    my $contained = [];
+    my ($start_pc_index, $finish_pc_index);
+    # Find smallest finish_pc_index such that $finish < $pc[$finish_pc_index].
+    for ($finish_pc_index = $#pcs + 1; $finish_pc_index > 0;
+         $finish_pc_index--) {
+      last if $pcs[$finish_pc_index - 1] le $finish;
+    }
+    # Find smallest start_pc_index such that $start <= $pc[$start_pc_index].
+    for ($start_pc_index = $finish_pc_index; $start_pc_index > 0;
+         $start_pc_index--) {
+      last if $pcs[$start_pc_index - 1] lt $start;
+    }
+    # This keeps PC values higher than $pc[$finish_pc_index] in @pcs,
+    # in case there are overlaps in libraries and the main binary.
+    @{$contained} = splice(@pcs, $start_pc_index,
+                           $finish_pc_index - $start_pc_index);
+    # Map to symbols
+    MapToSymbols($libname, AddressSub($start, $offset), $contained, $symbols);
+  }
+
+  return $symbols;
+}
+
+# Map list of PC values to symbols for a given image
+sub MapToSymbols {
+  my $image = shift;
+  my $offset = shift;
+  my $pclist = shift;
+  my $symbols = shift;
+
+  my $debug = 0;
+
+  # Ignore empty binaries
+  if ($#{$pclist} < 0) { return; }
+
+  # Figure out the addr2line command to use
+  my $addr2line = $obj_tool_map{"addr2line"};
+  my $cmd = ShellEscape($addr2line, "-f", "-C", "-e", $image);
+  if (exists $obj_tool_map{"addr2line_pdb"}) {
+    $addr2line = $obj_tool_map{"addr2line_pdb"};
+    $cmd = ShellEscape($addr2line, "--demangle", "-f", "-C", "-e", $image);
+  }
+
+  # If "addr2line" isn't installed on the system at all, just use
+  # nm to get what info we can (function names, but not line numbers).
+  if (system(ShellEscape($addr2line, "--help") . " >$dev_null 2>&1") != 0) {
+    MapSymbolsWithNM($image, $offset, $pclist, $symbols);
+    return;
+  }
+
+  # "addr2line -i" can produce a variable number of lines per input
+  # address, with no separator that allows us to tell when data for
+  # the next address starts.  So we find the address for a special
+  # symbol (_fini) and interleave this address between all real
+  # addresses passed to addr2line.  The name of this special symbol
+  # can then be used as a separator.
+  $sep_address = undef;  # May be filled in by MapSymbolsWithNM()
+  my $nm_symbols = {};
+  MapSymbolsWithNM($image, $offset, $pclist, $nm_symbols);
+  if (defined($sep_address)) {
+    # Only add " -i" to addr2line if the binary supports it.
+    # addr2line --help returns 0, but not if it sees an unknown flag first.
+    if (system("$cmd -i --help >$dev_null 2>&1") == 0) {
+      $cmd .= " -i";
+    } else {
+      $sep_address = undef;   # no need for sep_address if we don't support -i
+    }
+  }
+
+  # Make file with all PC values with intervening 'sep_address' so
+  # that we can reliably detect the end of inlined function list
+  open(ADDRESSES, ">$main::tmpfile_sym") || error("$main::tmpfile_sym: $!\n");
+  if ($debug) { print("---- $image ---\n"); }
+  for (my $i = 0; $i <= $#{$pclist}; $i++) {
+    # addr2line always reads hex addresses, and does not need '0x' prefix.
+    if ($debug) { printf STDERR ("%s\n", $pclist->[$i]); }
+    printf ADDRESSES ("%s\n", AddressSub($pclist->[$i], $offset));
+    if (defined($sep_address)) {
+      printf ADDRESSES ("%s\n", $sep_address);
+    }
+  }
+  close(ADDRESSES);
+  if ($debug) {
+    print("----\n");
+    system("cat", $main::tmpfile_sym);
+    print("----\n");
+    system("$cmd < " . ShellEscape($main::tmpfile_sym));
+    print("----\n");
+  }
+
+  open(SYMBOLS, "$cmd <" . ShellEscape($main::tmpfile_sym) . " |")
+      || error("$cmd: $!\n");
+  my $count = 0;   # Index in pclist
+  while (<SYMBOLS>) {
+    # Read fullfunction and filelineinfo from next pair of lines
+    s/\r?\n$//g;
+    my $fullfunction = $_;
+    $_ = <SYMBOLS>;
+    s/\r?\n$//g;
+    my $filelinenum = $_;
+
+    if (defined($sep_address) && $fullfunction eq $sep_symbol) {
+      # Terminating marker for data for this address
+      $count++;
+      next;
+    }
+
+    $filelinenum =~ s|\\|/|g; # turn windows-style paths into unix-style paths
+
+    my $pcstr = $pclist->[$count];
+    my $function = ShortFunctionName($fullfunction);
+    my $nms = $nm_symbols->{$pcstr};
+    if (defined($nms)) {
+      if ($fullfunction eq '??') {
+        # nm found a symbol for us.
+        $function = $nms->[0];
+        $fullfunction = $nms->[2];
+      } else {
+	# MapSymbolsWithNM tags each routine with its starting address,
+	# useful in case the image has multiple occurrences of this
+	# routine.  (It uses a syntax that resembles template paramters,
+	# that are automatically stripped out by ShortFunctionName().)
+	# addr2line does not provide the same information.  So we check
+	# if nm disambiguated our symbol, and if so take the annotated
+	# (nm) version of the routine-name.  TODO(csilvers): this won't
+	# catch overloaded, inlined symbols, which nm doesn't see.
+	# Better would be to do a check similar to nm's, in this fn.
+	if ($nms->[2] =~ m/^\Q$function\E/) {  # sanity check it's the right fn
+	  $function = $nms->[0];
+	  $fullfunction = $nms->[2];
+	}
+      }
+    }
+    
+    # Prepend to accumulated symbols for pcstr
+    # (so that caller comes before callee)
+    my $sym = $symbols->{$pcstr};
+    if (!defined($sym)) {
+      $sym = [];
+      $symbols->{$pcstr} = $sym;
+    }
+    unshift(@{$sym}, $function, $filelinenum, $fullfunction);
+    if ($debug) { printf STDERR ("%s => [%s]\n", $pcstr, join(" ", @{$sym})); }
+    if (!defined($sep_address)) {
+      # Inlining is off, so this entry ends immediately
+      $count++;
+    }
+  }
+  close(SYMBOLS);
+}
+
+# Use nm to map the list of referenced PCs to symbols.  Return true iff we
+# are able to read procedure information via nm.
+sub MapSymbolsWithNM {
+  my $image = shift;
+  my $offset = shift;
+  my $pclist = shift;
+  my $symbols = shift;
+
+  # Get nm output sorted by increasing address
+  my $symbol_table = GetProcedureBoundaries($image, ".");
+  if (!%{$symbol_table}) {
+    return 0;
+  }
+  # Start addresses are already the right length (8 or 16 hex digits).
+  my @names = sort { $symbol_table->{$a}->[0] cmp $symbol_table->{$b}->[0] }
+    keys(%{$symbol_table});
+
+  if ($#names < 0) {
+    # No symbols: just use addresses
+    foreach my $pc (@{$pclist}) {
+      my $pcstr = "0x" . $pc;
+      $symbols->{$pc} = [$pcstr, "?", $pcstr];
+    }
+    return 0;
+  }
+
+  # Sort addresses so we can do a join against nm output
+  my $index = 0;
+  my $fullname = $names[0];
+  my $name = ShortFunctionName($fullname);
+  foreach my $pc (sort { $a cmp $b } @{$pclist}) {
+    # Adjust for mapped offset
+    my $mpc = AddressSub($pc, $offset);
+    while (($index < $#names) && ($mpc ge $symbol_table->{$fullname}->[1])){
+      $index++;
+      $fullname = $names[$index];
+      $name = ShortFunctionName($fullname);
+    }
+    if ($mpc lt $symbol_table->{$fullname}->[1]) {
+      $symbols->{$pc} = [$name, "?", $fullname];
+    } else {
+      my $pcstr = "0x" . $pc;
+      $symbols->{$pc} = [$pcstr, "?", $pcstr];
+    }
+  }
+  return 1;
+}
+
+sub ShortFunctionName {
+  my $function = shift;
+  while ($function =~ s/\([^()]*\)(\s*const)?//g) { }   # Argument types
+  while ($function =~ s/<[^<>]*>//g)  { }    # Remove template arguments
+  $function =~ s/^.*\s+(\w+::)/$1/;          # Remove leading type
+  return $function;
+}
+
+# Trim overly long symbols found in disassembler output
+sub CleanDisassembly {
+  my $d = shift;
+  while ($d =~ s/\([^()%]*\)(\s*const)?//g) { } # Argument types, not (%rax)
+  while ($d =~ s/(\w+)<[^<>]*>/$1/g)  { }       # Remove template arguments
+  return $d;
+}
+
+# Clean file name for display
+sub CleanFileName {
+  my ($f) = @_;
+  $f =~ s|^/proc/self/cwd/||;
+  $f =~ s|^\./||;
+  return $f;
+}
+
+# Make address relative to section and clean up for display
+sub UnparseAddress {
+  my ($offset, $address) = @_;
+  $address = AddressSub($address, $offset);
+  $address =~ s/^0x//;
+  $address =~ s/^0*//;
+  return $address;
+}
+
+##### Miscellaneous #####
+
+# Find the right versions of the above object tools to use.  The
+# argument is the program file being analyzed, and should be an ELF
+# 32-bit or ELF 64-bit executable file.  The location of the tools
+# is determined by considering the following options in this order:
+#   1) --tools option, if set
+#   2) PPROF_TOOLS environment variable, if set
+#   3) the environment
+sub ConfigureObjTools {
+  my $prog_file = shift;
+
+  # Check for the existence of $prog_file because /usr/bin/file does not
+  # predictably return error status in prod.
+  (-e $prog_file)  || error("$prog_file does not exist.\n");
+
+  my $file_type = undef;
+  if (-e "/usr/bin/file") {
+    # Follow symlinks (at least for systems where "file" supports that).
+    my $escaped_prog_file = ShellEscape($prog_file);
+    $file_type = `/usr/bin/file -L $escaped_prog_file 2>$dev_null ||
+                  /usr/bin/file $escaped_prog_file`;
+  } elsif ($^O == "MSWin32") {
+    $file_type = "MS Windows";
+  } else {
+    print STDERR "WARNING: Can't determine the file type of $prog_file";
+  }
+
+  if ($file_type =~ /64-bit/) {
+    # Change $address_length to 16 if the program file is ELF 64-bit.
+    # We can't detect this from many (most?) heap or lock contention
+    # profiles, since the actual addresses referenced are generally in low
+    # memory even for 64-bit programs.
+    $address_length = 16;
+  }
+
+  if ($file_type =~ /MS Windows/) {
+    # For windows, we provide a version of nm and addr2line as part of
+    # the opensource release, which is capable of parsing
+    # Windows-style PDB executables.  It should live in the path, or
+    # in the same directory as pprof.
+    $obj_tool_map{"nm_pdb"} = "nm-pdb";
+    $obj_tool_map{"addr2line_pdb"} = "addr2line-pdb";
+  }
+
+  if ($file_type =~ /Mach-O/) {
+    # OS X uses otool to examine Mach-O files, rather than objdump.
+    $obj_tool_map{"otool"} = "otool";
+    $obj_tool_map{"addr2line"} = "false";  # no addr2line
+    $obj_tool_map{"objdump"} = "false";  # no objdump
+  }
+
+  # Go fill in %obj_tool_map with the pathnames to use:
+  foreach my $tool (keys %obj_tool_map) {
+    $obj_tool_map{$tool} = ConfigureTool($obj_tool_map{$tool});
+  }
+}
+
+# Returns the path of a caller-specified object tool.  If --tools or
+# PPROF_TOOLS are specified, then returns the full path to the tool
+# with that prefix.  Otherwise, returns the path unmodified (which
+# means we will look for it on PATH).
+sub ConfigureTool {
+  my $tool = shift;
+  my $path;
+
+  # --tools (or $PPROF_TOOLS) is a comma separated list, where each
+  # item is either a) a pathname prefix, or b) a map of the form
+  # <tool>:<path>.  First we look for an entry of type (b) for our
+  # tool.  If one is found, we use it.  Otherwise, we consider all the
+  # pathname prefixes in turn, until one yields an existing file.  If
+  # none does, we use a default path.
+  my $tools = $main::opt_tools || $ENV{"PPROF_TOOLS"} || "";
+  if ($tools =~ m/(,|^)\Q$tool\E:([^,]*)/) {
+    $path = $2;
+    # TODO(csilvers): sanity-check that $path exists?  Hard if it's relative.
+  } elsif ($tools ne '') {
+    foreach my $prefix (split(',', $tools)) {
+      next if ($prefix =~ /:/);    # ignore "tool:fullpath" entries in the list
+      if (-x $prefix . $tool) {
+        $path = $prefix . $tool;
+        last;
+      }
+    }
+    if (!$path) {
+      error("No '$tool' found with prefix specified by " .
+            "--tools (or \$PPROF_TOOLS) '$tools'\n");
+    }
+  } else {
+    # ... otherwise use the version that exists in the same directory as
+    # pprof.  If there's nothing there, use $PATH.
+    $0 =~ m,[^/]*$,;     # this is everything after the last slash
+    my $dirname = $`;    # this is everything up to and including the last slash
+    if (-x "$dirname$tool") {
+      $path = "$dirname$tool";
+    } else { 
+      $path = $tool;
+    }
+  }
+  if ($main::opt_debug) { print STDERR "Using '$path' for '$tool'.\n"; }
+  return $path;
+}
+
+sub ShellEscape {
+  my @escaped_words = ();
+  foreach my $word (@_) {
+    my $escaped_word = $word;
+    if ($word =~ m![^a-zA-Z0-9/.,_=-]!) {  # check for anything not in whitelist
+      $escaped_word =~ s/'/'\\''/;
+      $escaped_word = "'$escaped_word'";
+    }
+    push(@escaped_words, $escaped_word);
+  }
+  return join(" ", @escaped_words);
+}
+
+sub cleanup {
+  unlink($main::tmpfile_sym);
+  unlink(keys %main::tempnames);
+
+  # We leave any collected profiles in $HOME/pprof in case the user wants
+  # to look at them later.  We print a message informing them of this.
+  if ((scalar(@main::profile_files) > 0) &&
+      defined($main::collected_profile)) {
+    if (scalar(@main::profile_files) == 1) {
+      print STDERR "Dynamically gathered profile is in $main::collected_profile\n";
+    }
+    print STDERR "If you want to investigate this profile further, you can do:\n";
+    print STDERR "\n";
+    print STDERR "  pprof \\\n";
+    print STDERR "    $main::prog \\\n";
+    print STDERR "    $main::collected_profile\n";
+    print STDERR "\n";
+  }
+}
+
+sub sighandler {
+  cleanup();
+  exit(1);
+}
+
+sub error {
+  my $msg = shift;
+  print STDERR $msg;
+  cleanup();
+  exit(1);
+}
+
+
+# Run $nm_command and get all the resulting procedure boundaries whose
+# names match "$regexp" and returns them in a hashtable mapping from
+# procedure name to a two-element vector of [start address, end address]
+sub GetProcedureBoundariesViaNm {
+  my $escaped_nm_command = shift;    # shell-escaped
+  my $regexp = shift;
+
+  my $symbol_table = {};
+  open(NM, "$escaped_nm_command |") || error("$escaped_nm_command: $!\n");
+  my $last_start = "0";
+  my $routine = "";
+  while (<NM>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    if (m/^\s*([0-9a-f]+) (.) (..*)/) {
+      my $start_val = $1;
+      my $type = $2;
+      my $this_routine = $3;
+
+      # It's possible for two symbols to share the same address, if
+      # one is a zero-length variable (like __start_google_malloc) or
+      # one symbol is a weak alias to another (like __libc_malloc).
+      # In such cases, we want to ignore all values except for the
+      # actual symbol, which in nm-speak has type "T".  The logic
+      # below does this, though it's a bit tricky: what happens when
+      # we have a series of lines with the same address, is the first
+      # one gets queued up to be processed.  However, it won't
+      # *actually* be processed until later, when we read a line with
+      # a different address.  That means that as long as we're reading
+      # lines with the same address, we have a chance to replace that
+      # item in the queue, which we do whenever we see a 'T' entry --
+      # that is, a line with type 'T'.  If we never see a 'T' entry,
+      # we'll just go ahead and process the first entry (which never
+      # got touched in the queue), and ignore the others.
+      if ($start_val eq $last_start && $type =~ /t/i) {
+        # We are the 'T' symbol at this address, replace previous symbol.
+        $routine = $this_routine;
+        next;
+      } elsif ($start_val eq $last_start) {
+        # We're not the 'T' symbol at this address, so ignore us.
+        next;
+      }
+
+      if ($this_routine eq $sep_symbol) {
+        $sep_address = HexExtend($start_val);
+      }
+
+      # Tag this routine with the starting address in case the image
+      # has multiple occurrences of this routine.  We use a syntax
+      # that resembles template paramters that are automatically
+      # stripped out by ShortFunctionName()
+      $this_routine .= "<$start_val>";
+
+      if (defined($routine) && $routine =~ m/$regexp/) {
+        $symbol_table->{$routine} = [HexExtend($last_start),
+                                     HexExtend($start_val)];
+      }
+      $last_start = $start_val;
+      $routine = $this_routine;
+    } elsif (m/^Loaded image name: (.+)/) {
+      # The win32 nm workalike emits information about the binary it is using.
+      if ($main::opt_debug) { print STDERR "Using Image $1\n"; }
+    } elsif (m/^PDB file name: (.+)/) {
+      # The win32 nm workalike emits information about the pdb it is using.
+      if ($main::opt_debug) { print STDERR "Using PDB $1\n"; }
+    }
+  }
+  close(NM);
+  # Handle the last line in the nm output.  Unfortunately, we don't know
+  # how big this last symbol is, because we don't know how big the file
+  # is.  For now, we just give it a size of 0.
+  # TODO(csilvers): do better here.
+  if (defined($routine) && $routine =~ m/$regexp/) {
+    $symbol_table->{$routine} = [HexExtend($last_start),
+                                 HexExtend($last_start)];
+  }
+  return $symbol_table;
+}
+
+# Gets the procedure boundaries for all routines in "$image" whose names
+# match "$regexp" and returns them in a hashtable mapping from procedure
+# name to a two-element vector of [start address, end address].
+# Will return an empty map if nm is not installed or not working properly.
+sub GetProcedureBoundaries {
+  my $image = shift;
+  my $regexp = shift;
+
+  # If $image doesn't start with /, then put ./ in front of it.  This works
+  # around an obnoxious bug in our probing of nm -f behavior.
+  # "nm -f $image" is supposed to fail on GNU nm, but if:
+  #
+  # a. $image starts with [BbSsPp] (for example, bin/foo/bar), AND
+  # b. you have a.out in your current directory (a not uncommon occurence)
+  #
+  # then "nm -f $image" succeeds because -f only looks at the first letter of
+  # the argument, which looks valid because it's [BbSsPp], and then since
+  # there's no image provided, it looks for a.out and finds it.
+  #
+  # This regex makes sure that $image starts with . or /, forcing the -f
+  # parsing to fail since . and / are not valid formats.
+  $image =~ s#^[^/]#./$&#;
+
+  # For libc libraries, the copy in /usr/lib/debug contains debugging symbols
+  my $debugging = DebuggingLibrary($image);
+  if ($debugging) {
+    $image = $debugging;
+  }
+
+  my $nm = $obj_tool_map{"nm"};
+  my $cppfilt = $obj_tool_map{"c++filt"};
+
+  # nm can fail for two reasons: 1) $image isn't a debug library; 2) nm
+  # binary doesn't support --demangle.  In addition, for OS X we need
+  # to use the -f flag to get 'flat' nm output (otherwise we don't sort
+  # properly and get incorrect results).  Unfortunately, GNU nm uses -f
+  # in an incompatible way.  So first we test whether our nm supports
+  # --demangle and -f.
+  my $demangle_flag = "";
+  my $cppfilt_flag = "";
+  my $to_devnull = ">$dev_null 2>&1";
+  if (system(ShellEscape($nm, "--demangle", "image") . $to_devnull) == 0) {
+    # In this mode, we do "nm --demangle <foo>"
+    $demangle_flag = "--demangle";
+    $cppfilt_flag = "";
+  } elsif (system(ShellEscape($cppfilt, $image) . $to_devnull) == 0) {
+    # In this mode, we do "nm <foo> | c++filt"
+    $cppfilt_flag = " | " . ShellEscape($cppfilt);
+  };
+  my $flatten_flag = "";
+  if (system(ShellEscape($nm, "-f", $image) . $to_devnull) == 0) {
+    $flatten_flag = "-f";
+  }
+
+  # Finally, in the case $imagie isn't a debug library, we try again with
+  # -D to at least get *exported* symbols.  If we can't use --demangle,
+  # we use c++filt instead, if it exists on this system.
+  my @nm_commands = (ShellEscape($nm, "-n", $flatten_flag, $demangle_flag,
+                                 $image) . " 2>$dev_null $cppfilt_flag",
+                     ShellEscape($nm, "-D", "-n", $flatten_flag, $demangle_flag,
+                                 $image) . " 2>$dev_null $cppfilt_flag",
+                     # 6nm is for Go binaries
+                     ShellEscape("6nm", "$image") . " 2>$dev_null | sort",
+                     );
+
+  # If the executable is an MS Windows PDB-format executable, we'll
+  # have set up obj_tool_map("nm_pdb").  In this case, we actually
+  # want to use both unix nm and windows-specific nm_pdb, since
+  # PDB-format executables can apparently include dwarf .o files.
+  if (exists $obj_tool_map{"nm_pdb"}) {
+    push(@nm_commands,
+         ShellEscape($obj_tool_map{"nm_pdb"}, "--demangle", $image)
+         . " 2>$dev_null");
+  }
+
+  foreach my $nm_command (@nm_commands) {
+    my $symbol_table = GetProcedureBoundariesViaNm($nm_command, $regexp);
+    return $symbol_table if (%{$symbol_table});
+  }
+  my $symbol_table = {};
+  return $symbol_table;
+}
+
+
+# The test vectors for AddressAdd/Sub/Inc are 8-16-nibble hex strings.
+# To make them more readable, we add underscores at interesting places.
+# This routine removes the underscores, producing the canonical representation
+# used by pprof to represent addresses, particularly in the tested routines.
+sub CanonicalHex {
+  my $arg = shift;
+  return join '', (split '_',$arg);
+}
+
+
+# Unit test for AddressAdd:
+sub AddressAddUnitTest {
+  my $test_data_8 = shift;
+  my $test_data_16 = shift;
+  my $error_count = 0;
+  my $fail_count = 0;
+  my $pass_count = 0;
+  # print STDERR "AddressAddUnitTest: ", 1+$#{$test_data_8}, " tests\n";
+
+  # First a few 8-nibble addresses.  Note that this implementation uses
+  # plain old arithmetic, so a quick sanity check along with verifying what
+  # happens to overflow (we want it to wrap):
+  $address_length = 8;
+  foreach my $row (@{$test_data_8}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressAdd ($row->[0], $row->[1]);
+    if ($sum ne $row->[2]) {
+      printf STDERR "ERROR: %s != %s + %s = %s\n", $sum,
+             $row->[0], $row->[1], $row->[2];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressAdd 32-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count = $fail_count;
+  $fail_count = 0;
+  $pass_count = 0;
+
+  # Now 16-nibble addresses.
+  $address_length = 16;
+  foreach my $row (@{$test_data_16}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressAdd (CanonicalHex($row->[0]), CanonicalHex($row->[1]));
+    my $expected = join '', (split '_',$row->[2]);
+    if ($sum ne CanonicalHex($row->[2])) {
+      printf STDERR "ERROR: %s != %s + %s = %s\n", $sum,
+             $row->[0], $row->[1], $row->[2];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressAdd 64-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count += $fail_count;
+
+  return $error_count;
+}
+
+
+# Unit test for AddressSub:
+sub AddressSubUnitTest {
+  my $test_data_8 = shift;
+  my $test_data_16 = shift;
+  my $error_count = 0;
+  my $fail_count = 0;
+  my $pass_count = 0;
+  # print STDERR "AddressSubUnitTest: ", 1+$#{$test_data_8}, " tests\n";
+
+  # First a few 8-nibble addresses.  Note that this implementation uses
+  # plain old arithmetic, so a quick sanity check along with verifying what
+  # happens to overflow (we want it to wrap):
+  $address_length = 8;
+  foreach my $row (@{$test_data_8}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressSub ($row->[0], $row->[1]);
+    if ($sum ne $row->[3]) {
+      printf STDERR "ERROR: %s != %s - %s = %s\n", $sum,
+             $row->[0], $row->[1], $row->[3];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressSub 32-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count = $fail_count;
+  $fail_count = 0;
+  $pass_count = 0;
+
+  # Now 16-nibble addresses.
+  $address_length = 16;
+  foreach my $row (@{$test_data_16}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressSub (CanonicalHex($row->[0]), CanonicalHex($row->[1]));
+    if ($sum ne CanonicalHex($row->[3])) {
+      printf STDERR "ERROR: %s != %s - %s = %s\n", $sum,
+             $row->[0], $row->[1], $row->[3];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressSub 64-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count += $fail_count;
+
+  return $error_count;
+}
+
+
+# Unit test for AddressInc:
+sub AddressIncUnitTest {
+  my $test_data_8 = shift;
+  my $test_data_16 = shift;
+  my $error_count = 0;
+  my $fail_count = 0;
+  my $pass_count = 0;
+  # print STDERR "AddressIncUnitTest: ", 1+$#{$test_data_8}, " tests\n";
+
+  # First a few 8-nibble addresses.  Note that this implementation uses
+  # plain old arithmetic, so a quick sanity check along with verifying what
+  # happens to overflow (we want it to wrap):
+  $address_length = 8;
+  foreach my $row (@{$test_data_8}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressInc ($row->[0]);
+    if ($sum ne $row->[4]) {
+      printf STDERR "ERROR: %s != %s + 1 = %s\n", $sum,
+             $row->[0], $row->[4];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressInc 32-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count = $fail_count;
+  $fail_count = 0;
+  $pass_count = 0;
+
+  # Now 16-nibble addresses.
+  $address_length = 16;
+  foreach my $row (@{$test_data_16}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressInc (CanonicalHex($row->[0]));
+    if ($sum ne CanonicalHex($row->[4])) {
+      printf STDERR "ERROR: %s != %s + 1 = %s\n", $sum,
+             $row->[0], $row->[4];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressInc 64-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count += $fail_count;
+
+  return $error_count;
+}
+
+
+# Driver for unit tests.
+# Currently just the address add/subtract/increment routines for 64-bit.
+sub RunUnitTests {
+  my $error_count = 0;
+
+  # This is a list of tuples [a, b, a+b, a-b, a+1]
+  my $unit_test_data_8 = [
+    [qw(aaaaaaaa 50505050 fafafafa 5a5a5a5a aaaaaaab)],
+    [qw(50505050 aaaaaaaa fafafafa a5a5a5a6 50505051)],
+    [qw(ffffffff aaaaaaaa aaaaaaa9 55555555 00000000)],
+    [qw(00000001 ffffffff 00000000 00000002 00000002)],
+    [qw(00000001 fffffff0 fffffff1 00000011 00000002)],
+  ];
+  my $unit_test_data_16 = [
+    # The implementation handles data in 7-nibble chunks, so those are the
+    # interesting boundaries.
+    [qw(aaaaaaaa 50505050
+        00_000000f_afafafa 00_0000005_a5a5a5a 00_000000a_aaaaaab)],
+    [qw(50505050 aaaaaaaa
+        00_000000f_afafafa ff_ffffffa_5a5a5a6 00_0000005_0505051)],
+    [qw(ffffffff aaaaaaaa
+        00_000001a_aaaaaa9 00_0000005_5555555 00_0000010_0000000)],
+    [qw(00000001 ffffffff
+        00_0000010_0000000 ff_ffffff0_0000002 00_0000000_0000002)],
+    [qw(00000001 fffffff0
+        00_000000f_ffffff1 ff_ffffff0_0000011 00_0000000_0000002)],
+
+    [qw(00_a00000a_aaaaaaa 50505050
+        00_a00000f_afafafa 00_a000005_a5a5a5a 00_a00000a_aaaaaab)],
+    [qw(0f_fff0005_0505050 aaaaaaaa
+        0f_fff000f_afafafa 0f_ffefffa_5a5a5a6 0f_fff0005_0505051)],
+    [qw(00_000000f_fffffff 01_800000a_aaaaaaa
+        01_800001a_aaaaaa9 fe_8000005_5555555 00_0000010_0000000)],
+    [qw(00_0000000_0000001 ff_fffffff_fffffff
+        00_0000000_0000000 00_0000000_0000002 00_0000000_0000002)],
+    [qw(00_0000000_0000001 ff_fffffff_ffffff0
+        ff_fffffff_ffffff1 00_0000000_0000011 00_0000000_0000002)],
+  ];
+
+  $error_count += AddressAddUnitTest($unit_test_data_8, $unit_test_data_16);
+  $error_count += AddressSubUnitTest($unit_test_data_8, $unit_test_data_16);
+  $error_count += AddressIncUnitTest($unit_test_data_8, $unit_test_data_16);
+  if ($error_count > 0) {
+    print STDERR $error_count, " errors: FAILED\n";
+  } else {
+    print STDERR "PASS\n";
+  }
+  exit ($error_count);
+}
diff --git a/extra/jemalloc/config.guess b/extra/jemalloc/config.guess
new file mode 100755
index 00000000000..d622a44e551
--- /dev/null
+++ b/extra/jemalloc/config.guess
@@ -0,0 +1,1530 @@
+#! /bin/sh
+# Attempt to guess a canonical system name.
+#   Copyright (C) 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
+#   2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010,
+#   2011, 2012 Free Software Foundation, Inc.
+
+timestamp='2012-02-10'
+
+# This file is free software; you can redistribute it and/or modify it
+# under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful, but
+# WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+# General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, see <http://www.gnu.org/licenses/>.
+#
+# As a special exception to the GNU General Public License, if you
+# distribute this file as part of a program that contains a
+# configuration script generated by Autoconf, you may include it under
+# the same distribution terms that you use for the rest of that program.
+
+
+# Originally written by Per Bothner.  Please send patches (context
+# diff format) to <config-patches@gnu.org> and include a ChangeLog
+# entry.
+#
+# This script attempts to guess a canonical system name similar to
+# config.sub.  If it succeeds, it prints the system name on stdout, and
+# exits with 0.  Otherwise, it exits with 1.
+#
+# You can get the latest version of this script from:
+# http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.guess;hb=HEAD
+
+me=`echo "$0" | sed -e 's,.*/,,'`
+
+usage="\
+Usage: $0 [OPTION]
+
+Output the configuration name of the system \`$me' is run on.
+
+Operation modes:
+  -h, --help         print this help, then exit
+  -t, --time-stamp   print date of last modification, then exit
+  -v, --version      print version number, then exit
+
+Report bugs and patches to <config-patches@gnu.org>."
+
+version="\
+GNU config.guess ($timestamp)
+
+Originally written by Per Bothner.
+Copyright (C) 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
+2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012
+Free Software Foundation, Inc.
+
+This is free software; see the source for copying conditions.  There is NO
+warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE."
+
+help="
+Try \`$me --help' for more information."
+
+# Parse command line
+while test $# -gt 0 ; do
+  case $1 in
+    --time-stamp | --time* | -t )
+       echo "$timestamp" ; exit ;;
+    --version | -v )
+       echo "$version" ; exit ;;
+    --help | --h* | -h )
+       echo "$usage"; exit ;;
+    -- )     # Stop option processing
+       shift; break ;;
+    - )	# Use stdin as input.
+       break ;;
+    -* )
+       echo "$me: invalid option $1$help" >&2
+       exit 1 ;;
+    * )
+       break ;;
+  esac
+done
+
+if test $# != 0; then
+  echo "$me: too many arguments$help" >&2
+  exit 1
+fi
+
+trap 'exit 1' 1 2 15
+
+# CC_FOR_BUILD -- compiler used by this script. Note that the use of a
+# compiler to aid in system detection is discouraged as it requires
+# temporary files to be created and, as you can see below, it is a
+# headache to deal with in a portable fashion.
+
+# Historically, `CC_FOR_BUILD' used to be named `HOST_CC'. We still
+# use `HOST_CC' if defined, but it is deprecated.
+
+# Portable tmp directory creation inspired by the Autoconf team.
+
+set_cc_for_build='
+trap "exitcode=\$?; (rm -f \$tmpfiles 2>/dev/null; rmdir \$tmp 2>/dev/null) && exit \$exitcode" 0 ;
+trap "rm -f \$tmpfiles 2>/dev/null; rmdir \$tmp 2>/dev/null; exit 1" 1 2 13 15 ;
+: ${TMPDIR=/tmp} ;
+ { tmp=`(umask 077 && mktemp -d "$TMPDIR/cgXXXXXX") 2>/dev/null` && test -n "$tmp" && test -d "$tmp" ; } ||
+ { test -n "$RANDOM" && tmp=$TMPDIR/cg$$-$RANDOM && (umask 077 && mkdir $tmp) ; } ||
+ { tmp=$TMPDIR/cg-$$ && (umask 077 && mkdir $tmp) && echo "Warning: creating insecure temp directory" >&2 ; } ||
+ { echo "$me: cannot create a temporary directory in $TMPDIR" >&2 ; exit 1 ; } ;
+dummy=$tmp/dummy ;
+tmpfiles="$dummy.c $dummy.o $dummy.rel $dummy" ;
+case $CC_FOR_BUILD,$HOST_CC,$CC in
+ ,,)    echo "int x;" > $dummy.c ;
+	for c in cc gcc c89 c99 ; do
+	  if ($c -c -o $dummy.o $dummy.c) >/dev/null 2>&1 ; then
+	     CC_FOR_BUILD="$c"; break ;
+	  fi ;
+	done ;
+	if test x"$CC_FOR_BUILD" = x ; then
+	  CC_FOR_BUILD=no_compiler_found ;
+	fi
+	;;
+ ,,*)   CC_FOR_BUILD=$CC ;;
+ ,*,*)  CC_FOR_BUILD=$HOST_CC ;;
+esac ; set_cc_for_build= ;'
+
+# This is needed to find uname on a Pyramid OSx when run in the BSD universe.
+# (ghazi@noc.rutgers.edu 1994-08-24)
+if (test -f /.attbin/uname) >/dev/null 2>&1 ; then
+	PATH=$PATH:/.attbin ; export PATH
+fi
+
+UNAME_MACHINE=`(uname -m) 2>/dev/null` || UNAME_MACHINE=unknown
+UNAME_RELEASE=`(uname -r) 2>/dev/null` || UNAME_RELEASE=unknown
+UNAME_SYSTEM=`(uname -s) 2>/dev/null`  || UNAME_SYSTEM=unknown
+UNAME_VERSION=`(uname -v) 2>/dev/null` || UNAME_VERSION=unknown
+
+# Note: order is significant - the case branches are not exclusive.
+
+case "${UNAME_MACHINE}:${UNAME_SYSTEM}:${UNAME_RELEASE}:${UNAME_VERSION}" in
+    *:NetBSD:*:*)
+	# NetBSD (nbsd) targets should (where applicable) match one or
+	# more of the tuples: *-*-netbsdelf*, *-*-netbsdaout*,
+	# *-*-netbsdecoff* and *-*-netbsd*.  For targets that recently
+	# switched to ELF, *-*-netbsd* would select the old
+	# object file format.  This provides both forward
+	# compatibility and a consistent mechanism for selecting the
+	# object file format.
+	#
+	# Note: NetBSD doesn't particularly care about the vendor
+	# portion of the name.  We always set it to "unknown".
+	sysctl="sysctl -n hw.machine_arch"
+	UNAME_MACHINE_ARCH=`(/sbin/$sysctl 2>/dev/null || \
+	    /usr/sbin/$sysctl 2>/dev/null || echo unknown)`
+	case "${UNAME_MACHINE_ARCH}" in
+	    armeb) machine=armeb-unknown ;;
+	    arm*) machine=arm-unknown ;;
+	    sh3el) machine=shl-unknown ;;
+	    sh3eb) machine=sh-unknown ;;
+	    sh5el) machine=sh5le-unknown ;;
+	    *) machine=${UNAME_MACHINE_ARCH}-unknown ;;
+	esac
+	# The Operating System including object format, if it has switched
+	# to ELF recently, or will in the future.
+	case "${UNAME_MACHINE_ARCH}" in
+	    arm*|i386|m68k|ns32k|sh3*|sparc|vax)
+		eval $set_cc_for_build
+		if echo __ELF__ | $CC_FOR_BUILD -E - 2>/dev/null \
+			| grep -q __ELF__
+		then
+		    # Once all utilities can be ECOFF (netbsdecoff) or a.out (netbsdaout).
+		    # Return netbsd for either.  FIX?
+		    os=netbsd
+		else
+		    os=netbsdelf
+		fi
+		;;
+	    *)
+		os=netbsd
+		;;
+	esac
+	# The OS release
+	# Debian GNU/NetBSD machines have a different userland, and
+	# thus, need a distinct triplet. However, they do not need
+	# kernel version information, so it can be replaced with a
+	# suitable tag, in the style of linux-gnu.
+	case "${UNAME_VERSION}" in
+	    Debian*)
+		release='-gnu'
+		;;
+	    *)
+		release=`echo ${UNAME_RELEASE}|sed -e 's/[-_].*/\./'`
+		;;
+	esac
+	# Since CPU_TYPE-MANUFACTURER-KERNEL-OPERATING_SYSTEM:
+	# contains redundant information, the shorter form:
+	# CPU_TYPE-MANUFACTURER-OPERATING_SYSTEM is used.
+	echo "${machine}-${os}${release}"
+	exit ;;
+    *:OpenBSD:*:*)
+	UNAME_MACHINE_ARCH=`arch | sed 's/OpenBSD.//'`
+	echo ${UNAME_MACHINE_ARCH}-unknown-openbsd${UNAME_RELEASE}
+	exit ;;
+    *:ekkoBSD:*:*)
+	echo ${UNAME_MACHINE}-unknown-ekkobsd${UNAME_RELEASE}
+	exit ;;
+    *:SolidBSD:*:*)
+	echo ${UNAME_MACHINE}-unknown-solidbsd${UNAME_RELEASE}
+	exit ;;
+    macppc:MirBSD:*:*)
+	echo powerpc-unknown-mirbsd${UNAME_RELEASE}
+	exit ;;
+    *:MirBSD:*:*)
+	echo ${UNAME_MACHINE}-unknown-mirbsd${UNAME_RELEASE}
+	exit ;;
+    alpha:OSF1:*:*)
+	case $UNAME_RELEASE in
+	*4.0)
+		UNAME_RELEASE=`/usr/sbin/sizer -v | awk '{print $3}'`
+		;;
+	*5.*)
+		UNAME_RELEASE=`/usr/sbin/sizer -v | awk '{print $4}'`
+		;;
+	esac
+	# According to Compaq, /usr/sbin/psrinfo has been available on
+	# OSF/1 and Tru64 systems produced since 1995.  I hope that
+	# covers most systems running today.  This code pipes the CPU
+	# types through head -n 1, so we only detect the type of CPU 0.
+	ALPHA_CPU_TYPE=`/usr/sbin/psrinfo -v | sed -n -e 's/^  The alpha \(.*\) processor.*$/\1/p' | head -n 1`
+	case "$ALPHA_CPU_TYPE" in
+	    "EV4 (21064)")
+		UNAME_MACHINE="alpha" ;;
+	    "EV4.5 (21064)")
+		UNAME_MACHINE="alpha" ;;
+	    "LCA4 (21066/21068)")
+		UNAME_MACHINE="alpha" ;;
+	    "EV5 (21164)")
+		UNAME_MACHINE="alphaev5" ;;
+	    "EV5.6 (21164A)")
+		UNAME_MACHINE="alphaev56" ;;
+	    "EV5.6 (21164PC)")
+		UNAME_MACHINE="alphapca56" ;;
+	    "EV5.7 (21164PC)")
+		UNAME_MACHINE="alphapca57" ;;
+	    "EV6 (21264)")
+		UNAME_MACHINE="alphaev6" ;;
+	    "EV6.7 (21264A)")
+		UNAME_MACHINE="alphaev67" ;;
+	    "EV6.8CB (21264C)")
+		UNAME_MACHINE="alphaev68" ;;
+	    "EV6.8AL (21264B)")
+		UNAME_MACHINE="alphaev68" ;;
+	    "EV6.8CX (21264D)")
+		UNAME_MACHINE="alphaev68" ;;
+	    "EV6.9A (21264/EV69A)")
+		UNAME_MACHINE="alphaev69" ;;
+	    "EV7 (21364)")
+		UNAME_MACHINE="alphaev7" ;;
+	    "EV7.9 (21364A)")
+		UNAME_MACHINE="alphaev79" ;;
+	esac
+	# A Pn.n version is a patched version.
+	# A Vn.n version is a released version.
+	# A Tn.n version is a released field test version.
+	# A Xn.n version is an unreleased experimental baselevel.
+	# 1.2 uses "1.2" for uname -r.
+	echo ${UNAME_MACHINE}-dec-osf`echo ${UNAME_RELEASE} | sed -e 's/^[PVTX]//' | tr 'ABCDEFGHIJKLMNOPQRSTUVWXYZ' 'abcdefghijklmnopqrstuvwxyz'`
+	# Reset EXIT trap before exiting to avoid spurious non-zero exit code.
+	exitcode=$?
+	trap '' 0
+	exit $exitcode ;;
+    Alpha\ *:Windows_NT*:*)
+	# How do we know it's Interix rather than the generic POSIX subsystem?
+	# Should we change UNAME_MACHINE based on the output of uname instead
+	# of the specific Alpha model?
+	echo alpha-pc-interix
+	exit ;;
+    21064:Windows_NT:50:3)
+	echo alpha-dec-winnt3.5
+	exit ;;
+    Amiga*:UNIX_System_V:4.0:*)
+	echo m68k-unknown-sysv4
+	exit ;;
+    *:[Aa]miga[Oo][Ss]:*:*)
+	echo ${UNAME_MACHINE}-unknown-amigaos
+	exit ;;
+    *:[Mm]orph[Oo][Ss]:*:*)
+	echo ${UNAME_MACHINE}-unknown-morphos
+	exit ;;
+    *:OS/390:*:*)
+	echo i370-ibm-openedition
+	exit ;;
+    *:z/VM:*:*)
+	echo s390-ibm-zvmoe
+	exit ;;
+    *:OS400:*:*)
+	echo powerpc-ibm-os400
+	exit ;;
+    arm:RISC*:1.[012]*:*|arm:riscix:1.[012]*:*)
+	echo arm-acorn-riscix${UNAME_RELEASE}
+	exit ;;
+    arm:riscos:*:*|arm:RISCOS:*:*)
+	echo arm-unknown-riscos
+	exit ;;
+    SR2?01:HI-UX/MPP:*:* | SR8000:HI-UX/MPP:*:*)
+	echo hppa1.1-hitachi-hiuxmpp
+	exit ;;
+    Pyramid*:OSx*:*:* | MIS*:OSx*:*:* | MIS*:SMP_DC-OSx*:*:*)
+	# akee@wpdis03.wpafb.af.mil (Earle F. Ake) contributed MIS and NILE.
+	if test "`(/bin/universe) 2>/dev/null`" = att ; then
+		echo pyramid-pyramid-sysv3
+	else
+		echo pyramid-pyramid-bsd
+	fi
+	exit ;;
+    NILE*:*:*:dcosx)
+	echo pyramid-pyramid-svr4
+	exit ;;
+    DRS?6000:unix:4.0:6*)
+	echo sparc-icl-nx6
+	exit ;;
+    DRS?6000:UNIX_SV:4.2*:7* | DRS?6000:isis:4.2*:7*)
+	case `/usr/bin/uname -p` in
+	    sparc) echo sparc-icl-nx7; exit ;;
+	esac ;;
+    s390x:SunOS:*:*)
+	echo ${UNAME_MACHINE}-ibm-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    sun4H:SunOS:5.*:*)
+	echo sparc-hal-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    sun4*:SunOS:5.*:* | tadpole*:SunOS:5.*:*)
+	echo sparc-sun-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    i86pc:AuroraUX:5.*:* | i86xen:AuroraUX:5.*:*)
+	echo i386-pc-auroraux${UNAME_RELEASE}
+	exit ;;
+    i86pc:SunOS:5.*:* | i86xen:SunOS:5.*:*)
+	eval $set_cc_for_build
+	SUN_ARCH="i386"
+	# If there is a compiler, see if it is configured for 64-bit objects.
+	# Note that the Sun cc does not turn __LP64__ into 1 like gcc does.
+	# This test works for both compilers.
+	if [ "$CC_FOR_BUILD" != 'no_compiler_found' ]; then
+	    if (echo '#ifdef __amd64'; echo IS_64BIT_ARCH; echo '#endif') | \
+		(CCOPTS= $CC_FOR_BUILD -E - 2>/dev/null) | \
+		grep IS_64BIT_ARCH >/dev/null
+	    then
+		SUN_ARCH="x86_64"
+	    fi
+	fi
+	echo ${SUN_ARCH}-pc-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    sun4*:SunOS:6*:*)
+	# According to config.sub, this is the proper way to canonicalize
+	# SunOS6.  Hard to guess exactly what SunOS6 will be like, but
+	# it's likely to be more like Solaris than SunOS4.
+	echo sparc-sun-solaris3`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    sun4*:SunOS:*:*)
+	case "`/usr/bin/arch -k`" in
+	    Series*|S4*)
+		UNAME_RELEASE=`uname -v`
+		;;
+	esac
+	# Japanese Language versions have a version number like `4.1.3-JL'.
+	echo sparc-sun-sunos`echo ${UNAME_RELEASE}|sed -e 's/-/_/'`
+	exit ;;
+    sun3*:SunOS:*:*)
+	echo m68k-sun-sunos${UNAME_RELEASE}
+	exit ;;
+    sun*:*:4.2BSD:*)
+	UNAME_RELEASE=`(sed 1q /etc/motd | awk '{print substr($5,1,3)}') 2>/dev/null`
+	test "x${UNAME_RELEASE}" = "x" && UNAME_RELEASE=3
+	case "`/bin/arch`" in
+	    sun3)
+		echo m68k-sun-sunos${UNAME_RELEASE}
+		;;
+	    sun4)
+		echo sparc-sun-sunos${UNAME_RELEASE}
+		;;
+	esac
+	exit ;;
+    aushp:SunOS:*:*)
+	echo sparc-auspex-sunos${UNAME_RELEASE}
+	exit ;;
+    # The situation for MiNT is a little confusing.  The machine name
+    # can be virtually everything (everything which is not
+    # "atarist" or "atariste" at least should have a processor
+    # > m68000).  The system name ranges from "MiNT" over "FreeMiNT"
+    # to the lowercase version "mint" (or "freemint").  Finally
+    # the system name "TOS" denotes a system which is actually not
+    # MiNT.  But MiNT is downward compatible to TOS, so this should
+    # be no problem.
+    atarist[e]:*MiNT:*:* | atarist[e]:*mint:*:* | atarist[e]:*TOS:*:*)
+	echo m68k-atari-mint${UNAME_RELEASE}
+	exit ;;
+    atari*:*MiNT:*:* | atari*:*mint:*:* | atarist[e]:*TOS:*:*)
+	echo m68k-atari-mint${UNAME_RELEASE}
+	exit ;;
+    *falcon*:*MiNT:*:* | *falcon*:*mint:*:* | *falcon*:*TOS:*:*)
+	echo m68k-atari-mint${UNAME_RELEASE}
+	exit ;;
+    milan*:*MiNT:*:* | milan*:*mint:*:* | *milan*:*TOS:*:*)
+	echo m68k-milan-mint${UNAME_RELEASE}
+	exit ;;
+    hades*:*MiNT:*:* | hades*:*mint:*:* | *hades*:*TOS:*:*)
+	echo m68k-hades-mint${UNAME_RELEASE}
+	exit ;;
+    *:*MiNT:*:* | *:*mint:*:* | *:*TOS:*:*)
+	echo m68k-unknown-mint${UNAME_RELEASE}
+	exit ;;
+    m68k:machten:*:*)
+	echo m68k-apple-machten${UNAME_RELEASE}
+	exit ;;
+    powerpc:machten:*:*)
+	echo powerpc-apple-machten${UNAME_RELEASE}
+	exit ;;
+    RISC*:Mach:*:*)
+	echo mips-dec-mach_bsd4.3
+	exit ;;
+    RISC*:ULTRIX:*:*)
+	echo mips-dec-ultrix${UNAME_RELEASE}
+	exit ;;
+    VAX*:ULTRIX*:*:*)
+	echo vax-dec-ultrix${UNAME_RELEASE}
+	exit ;;
+    2020:CLIX:*:* | 2430:CLIX:*:*)
+	echo clipper-intergraph-clix${UNAME_RELEASE}
+	exit ;;
+    mips:*:*:UMIPS | mips:*:*:RISCos)
+	eval $set_cc_for_build
+	sed 's/^	//' << EOF >$dummy.c
+#ifdef __cplusplus
+#include <stdio.h>  /* for printf() prototype */
+	int main (int argc, char *argv[]) {
+#else
+	int main (argc, argv) int argc; char *argv[]; {
+#endif
+	#if defined (host_mips) && defined (MIPSEB)
+	#if defined (SYSTYPE_SYSV)
+	  printf ("mips-mips-riscos%ssysv\n", argv[1]); exit (0);
+	#endif
+	#if defined (SYSTYPE_SVR4)
+	  printf ("mips-mips-riscos%ssvr4\n", argv[1]); exit (0);
+	#endif
+	#if defined (SYSTYPE_BSD43) || defined(SYSTYPE_BSD)
+	  printf ("mips-mips-riscos%sbsd\n", argv[1]); exit (0);
+	#endif
+	#endif
+	  exit (-1);
+	}
+EOF
+	$CC_FOR_BUILD -o $dummy $dummy.c &&
+	  dummyarg=`echo "${UNAME_RELEASE}" | sed -n 's/\([0-9]*\).*/\1/p'` &&
+	  SYSTEM_NAME=`$dummy $dummyarg` &&
+	    { echo "$SYSTEM_NAME"; exit; }
+	echo mips-mips-riscos${UNAME_RELEASE}
+	exit ;;
+    Motorola:PowerMAX_OS:*:*)
+	echo powerpc-motorola-powermax
+	exit ;;
+    Motorola:*:4.3:PL8-*)
+	echo powerpc-harris-powermax
+	exit ;;
+    Night_Hawk:*:*:PowerMAX_OS | Synergy:PowerMAX_OS:*:*)
+	echo powerpc-harris-powermax
+	exit ;;
+    Night_Hawk:Power_UNIX:*:*)
+	echo powerpc-harris-powerunix
+	exit ;;
+    m88k:CX/UX:7*:*)
+	echo m88k-harris-cxux7
+	exit ;;
+    m88k:*:4*:R4*)
+	echo m88k-motorola-sysv4
+	exit ;;
+    m88k:*:3*:R3*)
+	echo m88k-motorola-sysv3
+	exit ;;
+    AViiON:dgux:*:*)
+	# DG/UX returns AViiON for all architectures
+	UNAME_PROCESSOR=`/usr/bin/uname -p`
+	if [ $UNAME_PROCESSOR = mc88100 ] || [ $UNAME_PROCESSOR = mc88110 ]
+	then
+	    if [ ${TARGET_BINARY_INTERFACE}x = m88kdguxelfx ] || \
+	       [ ${TARGET_BINARY_INTERFACE}x = x ]
+	    then
+		echo m88k-dg-dgux${UNAME_RELEASE}
+	    else
+		echo m88k-dg-dguxbcs${UNAME_RELEASE}
+	    fi
+	else
+	    echo i586-dg-dgux${UNAME_RELEASE}
+	fi
+	exit ;;
+    M88*:DolphinOS:*:*)	# DolphinOS (SVR3)
+	echo m88k-dolphin-sysv3
+	exit ;;
+    M88*:*:R3*:*)
+	# Delta 88k system running SVR3
+	echo m88k-motorola-sysv3
+	exit ;;
+    XD88*:*:*:*) # Tektronix XD88 system running UTekV (SVR3)
+	echo m88k-tektronix-sysv3
+	exit ;;
+    Tek43[0-9][0-9]:UTek:*:*) # Tektronix 4300 system running UTek (BSD)
+	echo m68k-tektronix-bsd
+	exit ;;
+    *:IRIX*:*:*)
+	echo mips-sgi-irix`echo ${UNAME_RELEASE}|sed -e 's/-/_/g'`
+	exit ;;
+    ????????:AIX?:[12].1:2)   # AIX 2.2.1 or AIX 2.1.1 is RT/PC AIX.
+	echo romp-ibm-aix     # uname -m gives an 8 hex-code CPU id
+	exit ;;               # Note that: echo "'`uname -s`'" gives 'AIX '
+    i*86:AIX:*:*)
+	echo i386-ibm-aix
+	exit ;;
+    ia64:AIX:*:*)
+	if [ -x /usr/bin/oslevel ] ; then
+		IBM_REV=`/usr/bin/oslevel`
+	else
+		IBM_REV=${UNAME_VERSION}.${UNAME_RELEASE}
+	fi
+	echo ${UNAME_MACHINE}-ibm-aix${IBM_REV}
+	exit ;;
+    *:AIX:2:3)
+	if grep bos325 /usr/include/stdio.h >/dev/null 2>&1; then
+		eval $set_cc_for_build
+		sed 's/^		//' << EOF >$dummy.c
+		#include <sys/systemcfg.h>
+
+		main()
+			{
+			if (!__power_pc())
+				exit(1);
+			puts("powerpc-ibm-aix3.2.5");
+			exit(0);
+			}
+EOF
+		if $CC_FOR_BUILD -o $dummy $dummy.c && SYSTEM_NAME=`$dummy`
+		then
+			echo "$SYSTEM_NAME"
+		else
+			echo rs6000-ibm-aix3.2.5
+		fi
+	elif grep bos324 /usr/include/stdio.h >/dev/null 2>&1; then
+		echo rs6000-ibm-aix3.2.4
+	else
+		echo rs6000-ibm-aix3.2
+	fi
+	exit ;;
+    *:AIX:*:[4567])
+	IBM_CPU_ID=`/usr/sbin/lsdev -C -c processor -S available | sed 1q | awk '{ print $1 }'`
+	if /usr/sbin/lsattr -El ${IBM_CPU_ID} | grep ' POWER' >/dev/null 2>&1; then
+		IBM_ARCH=rs6000
+	else
+		IBM_ARCH=powerpc
+	fi
+	if [ -x /usr/bin/oslevel ] ; then
+		IBM_REV=`/usr/bin/oslevel`
+	else
+		IBM_REV=${UNAME_VERSION}.${UNAME_RELEASE}
+	fi
+	echo ${IBM_ARCH}-ibm-aix${IBM_REV}
+	exit ;;
+    *:AIX:*:*)
+	echo rs6000-ibm-aix
+	exit ;;
+    ibmrt:4.4BSD:*|romp-ibm:BSD:*)
+	echo romp-ibm-bsd4.4
+	exit ;;
+    ibmrt:*BSD:*|romp-ibm:BSD:*)            # covers RT/PC BSD and
+	echo romp-ibm-bsd${UNAME_RELEASE}   # 4.3 with uname added to
+	exit ;;                             # report: romp-ibm BSD 4.3
+    *:BOSX:*:*)
+	echo rs6000-bull-bosx
+	exit ;;
+    DPX/2?00:B.O.S.:*:*)
+	echo m68k-bull-sysv3
+	exit ;;
+    9000/[34]??:4.3bsd:1.*:*)
+	echo m68k-hp-bsd
+	exit ;;
+    hp300:4.4BSD:*:* | 9000/[34]??:4.3bsd:2.*:*)
+	echo m68k-hp-bsd4.4
+	exit ;;
+    9000/[34678]??:HP-UX:*:*)
+	HPUX_REV=`echo ${UNAME_RELEASE}|sed -e 's/[^.]*.[0B]*//'`
+	case "${UNAME_MACHINE}" in
+	    9000/31? )            HP_ARCH=m68000 ;;
+	    9000/[34]?? )         HP_ARCH=m68k ;;
+	    9000/[678][0-9][0-9])
+		if [ -x /usr/bin/getconf ]; then
+		    sc_cpu_version=`/usr/bin/getconf SC_CPU_VERSION 2>/dev/null`
+		    sc_kernel_bits=`/usr/bin/getconf SC_KERNEL_BITS 2>/dev/null`
+		    case "${sc_cpu_version}" in
+		      523) HP_ARCH="hppa1.0" ;; # CPU_PA_RISC1_0
+		      528) HP_ARCH="hppa1.1" ;; # CPU_PA_RISC1_1
+		      532)                      # CPU_PA_RISC2_0
+			case "${sc_kernel_bits}" in
+			  32) HP_ARCH="hppa2.0n" ;;
+			  64) HP_ARCH="hppa2.0w" ;;
+			  '') HP_ARCH="hppa2.0" ;;   # HP-UX 10.20
+			esac ;;
+		    esac
+		fi
+		if [ "${HP_ARCH}" = "" ]; then
+		    eval $set_cc_for_build
+		    sed 's/^		//' << EOF >$dummy.c
+
+		#define _HPUX_SOURCE
+		#include <stdlib.h>
+		#include <unistd.h>
+
+		int main ()
+		{
+		#if defined(_SC_KERNEL_BITS)
+		    long bits = sysconf(_SC_KERNEL_BITS);
+		#endif
+		    long cpu  = sysconf (_SC_CPU_VERSION);
+
+		    switch (cpu)
+			{
+			case CPU_PA_RISC1_0: puts ("hppa1.0"); break;
+			case CPU_PA_RISC1_1: puts ("hppa1.1"); break;
+			case CPU_PA_RISC2_0:
+		#if defined(_SC_KERNEL_BITS)
+			    switch (bits)
+				{
+				case 64: puts ("hppa2.0w"); break;
+				case 32: puts ("hppa2.0n"); break;
+				default: puts ("hppa2.0"); break;
+				} break;
+		#else  /* !defined(_SC_KERNEL_BITS) */
+			    puts ("hppa2.0"); break;
+		#endif
+			default: puts ("hppa1.0"); break;
+			}
+		    exit (0);
+		}
+EOF
+		    (CCOPTS= $CC_FOR_BUILD -o $dummy $dummy.c 2>/dev/null) && HP_ARCH=`$dummy`
+		    test -z "$HP_ARCH" && HP_ARCH=hppa
+		fi ;;
+	esac
+	if [ ${HP_ARCH} = "hppa2.0w" ]
+	then
+	    eval $set_cc_for_build
+
+	    # hppa2.0w-hp-hpux* has a 64-bit kernel and a compiler generating
+	    # 32-bit code.  hppa64-hp-hpux* has the same kernel and a compiler
+	    # generating 64-bit code.  GNU and HP use different nomenclature:
+	    #
+	    # $ CC_FOR_BUILD=cc ./config.guess
+	    # => hppa2.0w-hp-hpux11.23
+	    # $ CC_FOR_BUILD="cc +DA2.0w" ./config.guess
+	    # => hppa64-hp-hpux11.23
+
+	    if echo __LP64__ | (CCOPTS= $CC_FOR_BUILD -E - 2>/dev/null) |
+		grep -q __LP64__
+	    then
+		HP_ARCH="hppa2.0w"
+	    else
+		HP_ARCH="hppa64"
+	    fi
+	fi
+	echo ${HP_ARCH}-hp-hpux${HPUX_REV}
+	exit ;;
+    ia64:HP-UX:*:*)
+	HPUX_REV=`echo ${UNAME_RELEASE}|sed -e 's/[^.]*.[0B]*//'`
+	echo ia64-hp-hpux${HPUX_REV}
+	exit ;;
+    3050*:HI-UX:*:*)
+	eval $set_cc_for_build
+	sed 's/^	//' << EOF >$dummy.c
+	#include <unistd.h>
+	int
+	main ()
+	{
+	  long cpu = sysconf (_SC_CPU_VERSION);
+	  /* The order matters, because CPU_IS_HP_MC68K erroneously returns
+	     true for CPU_PA_RISC1_0.  CPU_IS_PA_RISC returns correct
+	     results, however.  */
+	  if (CPU_IS_PA_RISC (cpu))
+	    {
+	      switch (cpu)
+		{
+		  case CPU_PA_RISC1_0: puts ("hppa1.0-hitachi-hiuxwe2"); break;
+		  case CPU_PA_RISC1_1: puts ("hppa1.1-hitachi-hiuxwe2"); break;
+		  case CPU_PA_RISC2_0: puts ("hppa2.0-hitachi-hiuxwe2"); break;
+		  default: puts ("hppa-hitachi-hiuxwe2"); break;
+		}
+	    }
+	  else if (CPU_IS_HP_MC68K (cpu))
+	    puts ("m68k-hitachi-hiuxwe2");
+	  else puts ("unknown-hitachi-hiuxwe2");
+	  exit (0);
+	}
+EOF
+	$CC_FOR_BUILD -o $dummy $dummy.c && SYSTEM_NAME=`$dummy` &&
+		{ echo "$SYSTEM_NAME"; exit; }
+	echo unknown-hitachi-hiuxwe2
+	exit ;;
+    9000/7??:4.3bsd:*:* | 9000/8?[79]:4.3bsd:*:* )
+	echo hppa1.1-hp-bsd
+	exit ;;
+    9000/8??:4.3bsd:*:*)
+	echo hppa1.0-hp-bsd
+	exit ;;
+    *9??*:MPE/iX:*:* | *3000*:MPE/iX:*:*)
+	echo hppa1.0-hp-mpeix
+	exit ;;
+    hp7??:OSF1:*:* | hp8?[79]:OSF1:*:* )
+	echo hppa1.1-hp-osf
+	exit ;;
+    hp8??:OSF1:*:*)
+	echo hppa1.0-hp-osf
+	exit ;;
+    i*86:OSF1:*:*)
+	if [ -x /usr/sbin/sysversion ] ; then
+	    echo ${UNAME_MACHINE}-unknown-osf1mk
+	else
+	    echo ${UNAME_MACHINE}-unknown-osf1
+	fi
+	exit ;;
+    parisc*:Lites*:*:*)
+	echo hppa1.1-hp-lites
+	exit ;;
+    C1*:ConvexOS:*:* | convex:ConvexOS:C1*:*)
+	echo c1-convex-bsd
+	exit ;;
+    C2*:ConvexOS:*:* | convex:ConvexOS:C2*:*)
+	if getsysinfo -f scalar_acc
+	then echo c32-convex-bsd
+	else echo c2-convex-bsd
+	fi
+	exit ;;
+    C34*:ConvexOS:*:* | convex:ConvexOS:C34*:*)
+	echo c34-convex-bsd
+	exit ;;
+    C38*:ConvexOS:*:* | convex:ConvexOS:C38*:*)
+	echo c38-convex-bsd
+	exit ;;
+    C4*:ConvexOS:*:* | convex:ConvexOS:C4*:*)
+	echo c4-convex-bsd
+	exit ;;
+    CRAY*Y-MP:*:*:*)
+	echo ymp-cray-unicos${UNAME_RELEASE} | sed -e 's/\.[^.]*$/.X/'
+	exit ;;
+    CRAY*[A-Z]90:*:*:*)
+	echo ${UNAME_MACHINE}-cray-unicos${UNAME_RELEASE} \
+	| sed -e 's/CRAY.*\([A-Z]90\)/\1/' \
+	      -e y/ABCDEFGHIJKLMNOPQRSTUVWXYZ/abcdefghijklmnopqrstuvwxyz/ \
+	      -e 's/\.[^.]*$/.X/'
+	exit ;;
+    CRAY*TS:*:*:*)
+	echo t90-cray-unicos${UNAME_RELEASE} | sed -e 's/\.[^.]*$/.X/'
+	exit ;;
+    CRAY*T3E:*:*:*)
+	echo alphaev5-cray-unicosmk${UNAME_RELEASE} | sed -e 's/\.[^.]*$/.X/'
+	exit ;;
+    CRAY*SV1:*:*:*)
+	echo sv1-cray-unicos${UNAME_RELEASE} | sed -e 's/\.[^.]*$/.X/'
+	exit ;;
+    *:UNICOS/mp:*:*)
+	echo craynv-cray-unicosmp${UNAME_RELEASE} | sed -e 's/\.[^.]*$/.X/'
+	exit ;;
+    F30[01]:UNIX_System_V:*:* | F700:UNIX_System_V:*:*)
+	FUJITSU_PROC=`uname -m | tr 'ABCDEFGHIJKLMNOPQRSTUVWXYZ' 'abcdefghijklmnopqrstuvwxyz'`
+	FUJITSU_SYS=`uname -p | tr 'ABCDEFGHIJKLMNOPQRSTUVWXYZ' 'abcdefghijklmnopqrstuvwxyz' | sed -e 's/\///'`
+	FUJITSU_REL=`echo ${UNAME_RELEASE} | sed -e 's/ /_/'`
+	echo "${FUJITSU_PROC}-fujitsu-${FUJITSU_SYS}${FUJITSU_REL}"
+	exit ;;
+    5000:UNIX_System_V:4.*:*)
+	FUJITSU_SYS=`uname -p | tr 'ABCDEFGHIJKLMNOPQRSTUVWXYZ' 'abcdefghijklmnopqrstuvwxyz' | sed -e 's/\///'`
+	FUJITSU_REL=`echo ${UNAME_RELEASE} | tr 'ABCDEFGHIJKLMNOPQRSTUVWXYZ' 'abcdefghijklmnopqrstuvwxyz' | sed -e 's/ /_/'`
+	echo "sparc-fujitsu-${FUJITSU_SYS}${FUJITSU_REL}"
+	exit ;;
+    i*86:BSD/386:*:* | i*86:BSD/OS:*:* | *:Ascend\ Embedded/OS:*:*)
+	echo ${UNAME_MACHINE}-pc-bsdi${UNAME_RELEASE}
+	exit ;;
+    sparc*:BSD/OS:*:*)
+	echo sparc-unknown-bsdi${UNAME_RELEASE}
+	exit ;;
+    *:BSD/OS:*:*)
+	echo ${UNAME_MACHINE}-unknown-bsdi${UNAME_RELEASE}
+	exit ;;
+    *:FreeBSD:*:*)
+	UNAME_PROCESSOR=`/usr/bin/uname -p`
+	case ${UNAME_PROCESSOR} in
+	    amd64)
+		echo x86_64-unknown-freebsd`echo ${UNAME_RELEASE}|sed -e 's/[-(].*//'` ;;
+	    *)
+		echo ${UNAME_PROCESSOR}-unknown-freebsd`echo ${UNAME_RELEASE}|sed -e 's/[-(].*//'` ;;
+	esac
+	exit ;;
+    i*:CYGWIN*:*)
+	echo ${UNAME_MACHINE}-pc-cygwin
+	exit ;;
+    *:MINGW*:*)
+	echo ${UNAME_MACHINE}-pc-mingw32
+	exit ;;
+    i*:MSYS*:*)
+	echo ${UNAME_MACHINE}-pc-msys
+	exit ;;
+    i*:windows32*:*)
+	# uname -m includes "-pc" on this system.
+	echo ${UNAME_MACHINE}-mingw32
+	exit ;;
+    i*:PW*:*)
+	echo ${UNAME_MACHINE}-pc-pw32
+	exit ;;
+    *:Interix*:*)
+	case ${UNAME_MACHINE} in
+	    x86)
+		echo i586-pc-interix${UNAME_RELEASE}
+		exit ;;
+	    authenticamd | genuineintel | EM64T)
+		echo x86_64-unknown-interix${UNAME_RELEASE}
+		exit ;;
+	    IA64)
+		echo ia64-unknown-interix${UNAME_RELEASE}
+		exit ;;
+	esac ;;
+    [345]86:Windows_95:* | [345]86:Windows_98:* | [345]86:Windows_NT:*)
+	echo i${UNAME_MACHINE}-pc-mks
+	exit ;;
+    8664:Windows_NT:*)
+	echo x86_64-pc-mks
+	exit ;;
+    i*:Windows_NT*:* | Pentium*:Windows_NT*:*)
+	# How do we know it's Interix rather than the generic POSIX subsystem?
+	# It also conflicts with pre-2.0 versions of AT&T UWIN. Should we
+	# UNAME_MACHINE based on the output of uname instead of i386?
+	echo i586-pc-interix
+	exit ;;
+    i*:UWIN*:*)
+	echo ${UNAME_MACHINE}-pc-uwin
+	exit ;;
+    amd64:CYGWIN*:*:* | x86_64:CYGWIN*:*:*)
+	echo x86_64-unknown-cygwin
+	exit ;;
+    p*:CYGWIN*:*)
+	echo powerpcle-unknown-cygwin
+	exit ;;
+    prep*:SunOS:5.*:*)
+	echo powerpcle-unknown-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    *:GNU:*:*)
+	# the GNU system
+	echo `echo ${UNAME_MACHINE}|sed -e 's,[-/].*$,,'`-unknown-gnu`echo ${UNAME_RELEASE}|sed -e 's,/.*$,,'`
+	exit ;;
+    *:GNU/*:*:*)
+	# other systems with GNU libc and userland
+	echo ${UNAME_MACHINE}-unknown-`echo ${UNAME_SYSTEM} | sed 's,^[^/]*/,,' | tr '[A-Z]' '[a-z]'``echo ${UNAME_RELEASE}|sed -e 's/[-(].*//'`-gnu
+	exit ;;
+    i*86:Minix:*:*)
+	echo ${UNAME_MACHINE}-pc-minix
+	exit ;;
+    aarch64:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-gnu
+	exit ;;
+    aarch64_be:Linux:*:*)
+	UNAME_MACHINE=aarch64_be
+	echo ${UNAME_MACHINE}-unknown-linux-gnu
+	exit ;;
+    alpha:Linux:*:*)
+	case `sed -n '/^cpu model/s/^.*: \(.*\)/\1/p' < /proc/cpuinfo` in
+	  EV5)   UNAME_MACHINE=alphaev5 ;;
+	  EV56)  UNAME_MACHINE=alphaev56 ;;
+	  PCA56) UNAME_MACHINE=alphapca56 ;;
+	  PCA57) UNAME_MACHINE=alphapca56 ;;
+	  EV6)   UNAME_MACHINE=alphaev6 ;;
+	  EV67)  UNAME_MACHINE=alphaev67 ;;
+	  EV68*) UNAME_MACHINE=alphaev68 ;;
+	esac
+	objdump --private-headers /bin/sh | grep -q ld.so.1
+	if test "$?" = 0 ; then LIBC="libc1" ; else LIBC="" ; fi
+	echo ${UNAME_MACHINE}-unknown-linux-gnu${LIBC}
+	exit ;;
+    arm*:Linux:*:*)
+	eval $set_cc_for_build
+	if echo __ARM_EABI__ | $CC_FOR_BUILD -E - 2>/dev/null \
+	    | grep -q __ARM_EABI__
+	then
+	    echo ${UNAME_MACHINE}-unknown-linux-gnu
+	else
+	    if echo __ARM_PCS_VFP | $CC_FOR_BUILD -E - 2>/dev/null \
+		| grep -q __ARM_PCS_VFP
+	    then
+		echo ${UNAME_MACHINE}-unknown-linux-gnueabi
+	    else
+		echo ${UNAME_MACHINE}-unknown-linux-gnueabihf
+	    fi
+	fi
+	exit ;;
+    avr32*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-gnu
+	exit ;;
+    cris:Linux:*:*)
+	echo ${UNAME_MACHINE}-axis-linux-gnu
+	exit ;;
+    crisv32:Linux:*:*)
+	echo ${UNAME_MACHINE}-axis-linux-gnu
+	exit ;;
+    frv:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-gnu
+	exit ;;
+    hexagon:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-gnu
+	exit ;;
+    i*86:Linux:*:*)
+	LIBC=gnu
+	eval $set_cc_for_build
+	sed 's/^	//' << EOF >$dummy.c
+	#ifdef __dietlibc__
+	LIBC=dietlibc
+	#endif
+EOF
+	eval `$CC_FOR_BUILD -E $dummy.c 2>/dev/null | grep '^LIBC'`
+	echo "${UNAME_MACHINE}-pc-linux-${LIBC}"
+	exit ;;
+    ia64:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-gnu
+	exit ;;
+    m32r*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-gnu
+	exit ;;
+    m68*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-gnu
+	exit ;;
+    mips:Linux:*:* | mips64:Linux:*:*)
+	eval $set_cc_for_build
+	sed 's/^	//' << EOF >$dummy.c
+	#undef CPU
+	#undef ${UNAME_MACHINE}
+	#undef ${UNAME_MACHINE}el
+	#if defined(__MIPSEL__) || defined(__MIPSEL) || defined(_MIPSEL) || defined(MIPSEL)
+	CPU=${UNAME_MACHINE}el
+	#else
+	#if defined(__MIPSEB__) || defined(__MIPSEB) || defined(_MIPSEB) || defined(MIPSEB)
+	CPU=${UNAME_MACHINE}
+	#else
+	CPU=
+	#endif
+	#endif
+EOF
+	eval `$CC_FOR_BUILD -E $dummy.c 2>/dev/null | grep '^CPU'`
+	test x"${CPU}" != x && { echo "${CPU}-unknown-linux-gnu"; exit; }
+	;;
+    or32:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-gnu
+	exit ;;
+    padre:Linux:*:*)
+	echo sparc-unknown-linux-gnu
+	exit ;;
+    parisc64:Linux:*:* | hppa64:Linux:*:*)
+	echo hppa64-unknown-linux-gnu
+	exit ;;
+    parisc:Linux:*:* | hppa:Linux:*:*)
+	# Look for CPU level
+	case `grep '^cpu[^a-z]*:' /proc/cpuinfo 2>/dev/null | cut -d' ' -f2` in
+	  PA7*) echo hppa1.1-unknown-linux-gnu ;;
+	  PA8*) echo hppa2.0-unknown-linux-gnu ;;
+	  *)    echo hppa-unknown-linux-gnu ;;
+	esac
+	exit ;;
+    ppc64:Linux:*:*)
+	echo powerpc64-unknown-linux-gnu
+	exit ;;
+    ppc:Linux:*:*)
+	echo powerpc-unknown-linux-gnu
+	exit ;;
+    s390:Linux:*:* | s390x:Linux:*:*)
+	echo ${UNAME_MACHINE}-ibm-linux
+	exit ;;
+    sh64*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-gnu
+	exit ;;
+    sh*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-gnu
+	exit ;;
+    sparc:Linux:*:* | sparc64:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-gnu
+	exit ;;
+    tile*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-gnu
+	exit ;;
+    vax:Linux:*:*)
+	echo ${UNAME_MACHINE}-dec-linux-gnu
+	exit ;;
+    x86_64:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-gnu
+	exit ;;
+    xtensa*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-gnu
+	exit ;;
+    i*86:DYNIX/ptx:4*:*)
+	# ptx 4.0 does uname -s correctly, with DYNIX/ptx in there.
+	# earlier versions are messed up and put the nodename in both
+	# sysname and nodename.
+	echo i386-sequent-sysv4
+	exit ;;
+    i*86:UNIX_SV:4.2MP:2.*)
+	# Unixware is an offshoot of SVR4, but it has its own version
+	# number series starting with 2...
+	# I am not positive that other SVR4 systems won't match this,
+	# I just have to hope.  -- rms.
+	# Use sysv4.2uw... so that sysv4* matches it.
+	echo ${UNAME_MACHINE}-pc-sysv4.2uw${UNAME_VERSION}
+	exit ;;
+    i*86:OS/2:*:*)
+	# If we were able to find `uname', then EMX Unix compatibility
+	# is probably installed.
+	echo ${UNAME_MACHINE}-pc-os2-emx
+	exit ;;
+    i*86:XTS-300:*:STOP)
+	echo ${UNAME_MACHINE}-unknown-stop
+	exit ;;
+    i*86:atheos:*:*)
+	echo ${UNAME_MACHINE}-unknown-atheos
+	exit ;;
+    i*86:syllable:*:*)
+	echo ${UNAME_MACHINE}-pc-syllable
+	exit ;;
+    i*86:LynxOS:2.*:* | i*86:LynxOS:3.[01]*:* | i*86:LynxOS:4.[02]*:*)
+	echo i386-unknown-lynxos${UNAME_RELEASE}
+	exit ;;
+    i*86:*DOS:*:*)
+	echo ${UNAME_MACHINE}-pc-msdosdjgpp
+	exit ;;
+    i*86:*:4.*:* | i*86:SYSTEM_V:4.*:*)
+	UNAME_REL=`echo ${UNAME_RELEASE} | sed 's/\/MP$//'`
+	if grep Novell /usr/include/link.h >/dev/null 2>/dev/null; then
+		echo ${UNAME_MACHINE}-univel-sysv${UNAME_REL}
+	else
+		echo ${UNAME_MACHINE}-pc-sysv${UNAME_REL}
+	fi
+	exit ;;
+    i*86:*:5:[678]*)
+	# UnixWare 7.x, OpenUNIX and OpenServer 6.
+	case `/bin/uname -X | grep "^Machine"` in
+	    *486*)	     UNAME_MACHINE=i486 ;;
+	    *Pentium)	     UNAME_MACHINE=i586 ;;
+	    *Pent*|*Celeron) UNAME_MACHINE=i686 ;;
+	esac
+	echo ${UNAME_MACHINE}-unknown-sysv${UNAME_RELEASE}${UNAME_SYSTEM}${UNAME_VERSION}
+	exit ;;
+    i*86:*:3.2:*)
+	if test -f /usr/options/cb.name; then
+		UNAME_REL=`sed -n 's/.*Version //p' </usr/options/cb.name`
+		echo ${UNAME_MACHINE}-pc-isc$UNAME_REL
+	elif /bin/uname -X 2>/dev/null >/dev/null ; then
+		UNAME_REL=`(/bin/uname -X|grep Release|sed -e 's/.*= //')`
+		(/bin/uname -X|grep i80486 >/dev/null) && UNAME_MACHINE=i486
+		(/bin/uname -X|grep '^Machine.*Pentium' >/dev/null) \
+			&& UNAME_MACHINE=i586
+		(/bin/uname -X|grep '^Machine.*Pent *II' >/dev/null) \
+			&& UNAME_MACHINE=i686
+		(/bin/uname -X|grep '^Machine.*Pentium Pro' >/dev/null) \
+			&& UNAME_MACHINE=i686
+		echo ${UNAME_MACHINE}-pc-sco$UNAME_REL
+	else
+		echo ${UNAME_MACHINE}-pc-sysv32
+	fi
+	exit ;;
+    pc:*:*:*)
+	# Left here for compatibility:
+	# uname -m prints for DJGPP always 'pc', but it prints nothing about
+	# the processor, so we play safe by assuming i586.
+	# Note: whatever this is, it MUST be the same as what config.sub
+	# prints for the "djgpp" host, or else GDB configury will decide that
+	# this is a cross-build.
+	echo i586-pc-msdosdjgpp
+	exit ;;
+    Intel:Mach:3*:*)
+	echo i386-pc-mach3
+	exit ;;
+    paragon:*:*:*)
+	echo i860-intel-osf1
+	exit ;;
+    i860:*:4.*:*) # i860-SVR4
+	if grep Stardent /usr/include/sys/uadmin.h >/dev/null 2>&1 ; then
+	  echo i860-stardent-sysv${UNAME_RELEASE} # Stardent Vistra i860-SVR4
+	else # Add other i860-SVR4 vendors below as they are discovered.
+	  echo i860-unknown-sysv${UNAME_RELEASE}  # Unknown i860-SVR4
+	fi
+	exit ;;
+    mini*:CTIX:SYS*5:*)
+	# "miniframe"
+	echo m68010-convergent-sysv
+	exit ;;
+    mc68k:UNIX:SYSTEM5:3.51m)
+	echo m68k-convergent-sysv
+	exit ;;
+    M680?0:D-NIX:5.3:*)
+	echo m68k-diab-dnix
+	exit ;;
+    M68*:*:R3V[5678]*:*)
+	test -r /sysV68 && { echo 'm68k-motorola-sysv'; exit; } ;;
+    3[345]??:*:4.0:3.0 | 3[34]??A:*:4.0:3.0 | 3[34]??,*:*:4.0:3.0 | 3[34]??/*:*:4.0:3.0 | 4400:*:4.0:3.0 | 4850:*:4.0:3.0 | SKA40:*:4.0:3.0 | SDS2:*:4.0:3.0 | SHG2:*:4.0:3.0 | S7501*:*:4.0:3.0)
+	OS_REL=''
+	test -r /etc/.relid \
+	&& OS_REL=.`sed -n 's/[^ ]* [^ ]* \([0-9][0-9]\).*/\1/p' < /etc/.relid`
+	/bin/uname -p 2>/dev/null | grep 86 >/dev/null \
+	  && { echo i486-ncr-sysv4.3${OS_REL}; exit; }
+	/bin/uname -p 2>/dev/null | /bin/grep entium >/dev/null \
+	  && { echo i586-ncr-sysv4.3${OS_REL}; exit; } ;;
+    3[34]??:*:4.0:* | 3[34]??,*:*:4.0:*)
+	/bin/uname -p 2>/dev/null | grep 86 >/dev/null \
+	  && { echo i486-ncr-sysv4; exit; } ;;
+    NCR*:*:4.2:* | MPRAS*:*:4.2:*)
+	OS_REL='.3'
+	test -r /etc/.relid \
+	    && OS_REL=.`sed -n 's/[^ ]* [^ ]* \([0-9][0-9]\).*/\1/p' < /etc/.relid`
+	/bin/uname -p 2>/dev/null | grep 86 >/dev/null \
+	    && { echo i486-ncr-sysv4.3${OS_REL}; exit; }
+	/bin/uname -p 2>/dev/null | /bin/grep entium >/dev/null \
+	    && { echo i586-ncr-sysv4.3${OS_REL}; exit; }
+	/bin/uname -p 2>/dev/null | /bin/grep pteron >/dev/null \
+	    && { echo i586-ncr-sysv4.3${OS_REL}; exit; } ;;
+    m68*:LynxOS:2.*:* | m68*:LynxOS:3.0*:*)
+	echo m68k-unknown-lynxos${UNAME_RELEASE}
+	exit ;;
+    mc68030:UNIX_System_V:4.*:*)
+	echo m68k-atari-sysv4
+	exit ;;
+    TSUNAMI:LynxOS:2.*:*)
+	echo sparc-unknown-lynxos${UNAME_RELEASE}
+	exit ;;
+    rs6000:LynxOS:2.*:*)
+	echo rs6000-unknown-lynxos${UNAME_RELEASE}
+	exit ;;
+    PowerPC:LynxOS:2.*:* | PowerPC:LynxOS:3.[01]*:* | PowerPC:LynxOS:4.[02]*:*)
+	echo powerpc-unknown-lynxos${UNAME_RELEASE}
+	exit ;;
+    SM[BE]S:UNIX_SV:*:*)
+	echo mips-dde-sysv${UNAME_RELEASE}
+	exit ;;
+    RM*:ReliantUNIX-*:*:*)
+	echo mips-sni-sysv4
+	exit ;;
+    RM*:SINIX-*:*:*)
+	echo mips-sni-sysv4
+	exit ;;
+    *:SINIX-*:*:*)
+	if uname -p 2>/dev/null >/dev/null ; then
+		UNAME_MACHINE=`(uname -p) 2>/dev/null`
+		echo ${UNAME_MACHINE}-sni-sysv4
+	else
+		echo ns32k-sni-sysv
+	fi
+	exit ;;
+    PENTIUM:*:4.0*:*)	# Unisys `ClearPath HMP IX 4000' SVR4/MP effort
+			# says <Richard.M.Bartel@ccMail.Census.GOV>
+	echo i586-unisys-sysv4
+	exit ;;
+    *:UNIX_System_V:4*:FTX*)
+	# From Gerald Hewes <hewes@openmarket.com>.
+	# How about differentiating between stratus architectures? -djm
+	echo hppa1.1-stratus-sysv4
+	exit ;;
+    *:*:*:FTX*)
+	# From seanf@swdc.stratus.com.
+	echo i860-stratus-sysv4
+	exit ;;
+    i*86:VOS:*:*)
+	# From Paul.Green@stratus.com.
+	echo ${UNAME_MACHINE}-stratus-vos
+	exit ;;
+    *:VOS:*:*)
+	# From Paul.Green@stratus.com.
+	echo hppa1.1-stratus-vos
+	exit ;;
+    mc68*:A/UX:*:*)
+	echo m68k-apple-aux${UNAME_RELEASE}
+	exit ;;
+    news*:NEWS-OS:6*:*)
+	echo mips-sony-newsos6
+	exit ;;
+    R[34]000:*System_V*:*:* | R4000:UNIX_SYSV:*:* | R*000:UNIX_SV:*:*)
+	if [ -d /usr/nec ]; then
+		echo mips-nec-sysv${UNAME_RELEASE}
+	else
+		echo mips-unknown-sysv${UNAME_RELEASE}
+	fi
+	exit ;;
+    BeBox:BeOS:*:*)	# BeOS running on hardware made by Be, PPC only.
+	echo powerpc-be-beos
+	exit ;;
+    BeMac:BeOS:*:*)	# BeOS running on Mac or Mac clone, PPC only.
+	echo powerpc-apple-beos
+	exit ;;
+    BePC:BeOS:*:*)	# BeOS running on Intel PC compatible.
+	echo i586-pc-beos
+	exit ;;
+    BePC:Haiku:*:*)	# Haiku running on Intel PC compatible.
+	echo i586-pc-haiku
+	exit ;;
+    SX-4:SUPER-UX:*:*)
+	echo sx4-nec-superux${UNAME_RELEASE}
+	exit ;;
+    SX-5:SUPER-UX:*:*)
+	echo sx5-nec-superux${UNAME_RELEASE}
+	exit ;;
+    SX-6:SUPER-UX:*:*)
+	echo sx6-nec-superux${UNAME_RELEASE}
+	exit ;;
+    SX-7:SUPER-UX:*:*)
+	echo sx7-nec-superux${UNAME_RELEASE}
+	exit ;;
+    SX-8:SUPER-UX:*:*)
+	echo sx8-nec-superux${UNAME_RELEASE}
+	exit ;;
+    SX-8R:SUPER-UX:*:*)
+	echo sx8r-nec-superux${UNAME_RELEASE}
+	exit ;;
+    Power*:Rhapsody:*:*)
+	echo powerpc-apple-rhapsody${UNAME_RELEASE}
+	exit ;;
+    *:Rhapsody:*:*)
+	echo ${UNAME_MACHINE}-apple-rhapsody${UNAME_RELEASE}
+	exit ;;
+    *:Darwin:*:*)
+	UNAME_PROCESSOR=`uname -p` || UNAME_PROCESSOR=unknown
+	case $UNAME_PROCESSOR in
+	    i386)
+		eval $set_cc_for_build
+		if [ "$CC_FOR_BUILD" != 'no_compiler_found' ]; then
+		  if (echo '#ifdef __LP64__'; echo IS_64BIT_ARCH; echo '#endif') | \
+		      (CCOPTS= $CC_FOR_BUILD -E - 2>/dev/null) | \
+		      grep IS_64BIT_ARCH >/dev/null
+		  then
+		      UNAME_PROCESSOR="x86_64"
+		  fi
+		fi ;;
+	    unknown) UNAME_PROCESSOR=powerpc ;;
+	esac
+	echo ${UNAME_PROCESSOR}-apple-darwin${UNAME_RELEASE}
+	exit ;;
+    *:procnto*:*:* | *:QNX:[0123456789]*:*)
+	UNAME_PROCESSOR=`uname -p`
+	if test "$UNAME_PROCESSOR" = "x86"; then
+		UNAME_PROCESSOR=i386
+		UNAME_MACHINE=pc
+	fi
+	echo ${UNAME_PROCESSOR}-${UNAME_MACHINE}-nto-qnx${UNAME_RELEASE}
+	exit ;;
+    *:QNX:*:4*)
+	echo i386-pc-qnx
+	exit ;;
+    NEO-?:NONSTOP_KERNEL:*:*)
+	echo neo-tandem-nsk${UNAME_RELEASE}
+	exit ;;
+    NSE-?:NONSTOP_KERNEL:*:*)
+	echo nse-tandem-nsk${UNAME_RELEASE}
+	exit ;;
+    NSR-?:NONSTOP_KERNEL:*:*)
+	echo nsr-tandem-nsk${UNAME_RELEASE}
+	exit ;;
+    *:NonStop-UX:*:*)
+	echo mips-compaq-nonstopux
+	exit ;;
+    BS2000:POSIX*:*:*)
+	echo bs2000-siemens-sysv
+	exit ;;
+    DS/*:UNIX_System_V:*:*)
+	echo ${UNAME_MACHINE}-${UNAME_SYSTEM}-${UNAME_RELEASE}
+	exit ;;
+    *:Plan9:*:*)
+	# "uname -m" is not consistent, so use $cputype instead. 386
+	# is converted to i386 for consistency with other x86
+	# operating systems.
+	if test "$cputype" = "386"; then
+	    UNAME_MACHINE=i386
+	else
+	    UNAME_MACHINE="$cputype"
+	fi
+	echo ${UNAME_MACHINE}-unknown-plan9
+	exit ;;
+    *:TOPS-10:*:*)
+	echo pdp10-unknown-tops10
+	exit ;;
+    *:TENEX:*:*)
+	echo pdp10-unknown-tenex
+	exit ;;
+    KS10:TOPS-20:*:* | KL10:TOPS-20:*:* | TYPE4:TOPS-20:*:*)
+	echo pdp10-dec-tops20
+	exit ;;
+    XKL-1:TOPS-20:*:* | TYPE5:TOPS-20:*:*)
+	echo pdp10-xkl-tops20
+	exit ;;
+    *:TOPS-20:*:*)
+	echo pdp10-unknown-tops20
+	exit ;;
+    *:ITS:*:*)
+	echo pdp10-unknown-its
+	exit ;;
+    SEI:*:*:SEIUX)
+	echo mips-sei-seiux${UNAME_RELEASE}
+	exit ;;
+    *:DragonFly:*:*)
+	echo ${UNAME_MACHINE}-unknown-dragonfly`echo ${UNAME_RELEASE}|sed -e 's/[-(].*//'`
+	exit ;;
+    *:*VMS:*:*)
+	UNAME_MACHINE=`(uname -p) 2>/dev/null`
+	case "${UNAME_MACHINE}" in
+	    A*) echo alpha-dec-vms ; exit ;;
+	    I*) echo ia64-dec-vms ; exit ;;
+	    V*) echo vax-dec-vms ; exit ;;
+	esac ;;
+    *:XENIX:*:SysV)
+	echo i386-pc-xenix
+	exit ;;
+    i*86:skyos:*:*)
+	echo ${UNAME_MACHINE}-pc-skyos`echo ${UNAME_RELEASE}` | sed -e 's/ .*$//'
+	exit ;;
+    i*86:rdos:*:*)
+	echo ${UNAME_MACHINE}-pc-rdos
+	exit ;;
+    i*86:AROS:*:*)
+	echo ${UNAME_MACHINE}-pc-aros
+	exit ;;
+    x86_64:VMkernel:*:*)
+	echo ${UNAME_MACHINE}-unknown-esx
+	exit ;;
+esac
+
+#echo '(No uname command or uname output not recognized.)' 1>&2
+#echo "${UNAME_MACHINE}:${UNAME_SYSTEM}:${UNAME_RELEASE}:${UNAME_VERSION}" 1>&2
+
+eval $set_cc_for_build
+cat >$dummy.c <<EOF
+#ifdef _SEQUENT_
+# include <sys/types.h>
+# include <sys/utsname.h>
+#endif
+main ()
+{
+#if defined (sony)
+#if defined (MIPSEB)
+  /* BFD wants "bsd" instead of "newsos".  Perhaps BFD should be changed,
+     I don't know....  */
+  printf ("mips-sony-bsd\n"); exit (0);
+#else
+#include <sys/param.h>
+  printf ("m68k-sony-newsos%s\n",
+#ifdef NEWSOS4
+	"4"
+#else
+	""
+#endif
+	); exit (0);
+#endif
+#endif
+
+#if defined (__arm) && defined (__acorn) && defined (__unix)
+  printf ("arm-acorn-riscix\n"); exit (0);
+#endif
+
+#if defined (hp300) && !defined (hpux)
+  printf ("m68k-hp-bsd\n"); exit (0);
+#endif
+
+#if defined (NeXT)
+#if !defined (__ARCHITECTURE__)
+#define __ARCHITECTURE__ "m68k"
+#endif
+  int version;
+  version=`(hostinfo | sed -n 's/.*NeXT Mach \([0-9]*\).*/\1/p') 2>/dev/null`;
+  if (version < 4)
+    printf ("%s-next-nextstep%d\n", __ARCHITECTURE__, version);
+  else
+    printf ("%s-next-openstep%d\n", __ARCHITECTURE__, version);
+  exit (0);
+#endif
+
+#if defined (MULTIMAX) || defined (n16)
+#if defined (UMAXV)
+  printf ("ns32k-encore-sysv\n"); exit (0);
+#else
+#if defined (CMU)
+  printf ("ns32k-encore-mach\n"); exit (0);
+#else
+  printf ("ns32k-encore-bsd\n"); exit (0);
+#endif
+#endif
+#endif
+
+#if defined (__386BSD__)
+  printf ("i386-pc-bsd\n"); exit (0);
+#endif
+
+#if defined (sequent)
+#if defined (i386)
+  printf ("i386-sequent-dynix\n"); exit (0);
+#endif
+#if defined (ns32000)
+  printf ("ns32k-sequent-dynix\n"); exit (0);
+#endif
+#endif
+
+#if defined (_SEQUENT_)
+    struct utsname un;
+
+    uname(&un);
+
+    if (strncmp(un.version, "V2", 2) == 0) {
+	printf ("i386-sequent-ptx2\n"); exit (0);
+    }
+    if (strncmp(un.version, "V1", 2) == 0) { /* XXX is V1 correct? */
+	printf ("i386-sequent-ptx1\n"); exit (0);
+    }
+    printf ("i386-sequent-ptx\n"); exit (0);
+
+#endif
+
+#if defined (vax)
+# if !defined (ultrix)
+#  include <sys/param.h>
+#  if defined (BSD)
+#   if BSD == 43
+      printf ("vax-dec-bsd4.3\n"); exit (0);
+#   else
+#    if BSD == 199006
+      printf ("vax-dec-bsd4.3reno\n"); exit (0);
+#    else
+      printf ("vax-dec-bsd\n"); exit (0);
+#    endif
+#   endif
+#  else
+    printf ("vax-dec-bsd\n"); exit (0);
+#  endif
+# else
+    printf ("vax-dec-ultrix\n"); exit (0);
+# endif
+#endif
+
+#if defined (alliant) && defined (i860)
+  printf ("i860-alliant-bsd\n"); exit (0);
+#endif
+
+  exit (1);
+}
+EOF
+
+$CC_FOR_BUILD -o $dummy $dummy.c 2>/dev/null && SYSTEM_NAME=`$dummy` &&
+	{ echo "$SYSTEM_NAME"; exit; }
+
+# Apollos put the system type in the environment.
+
+test -d /usr/apollo && { echo ${ISP}-apollo-${SYSTYPE}; exit; }
+
+# Convex versions that predate uname can use getsysinfo(1)
+
+if [ -x /usr/convex/getsysinfo ]
+then
+    case `getsysinfo -f cpu_type` in
+    c1*)
+	echo c1-convex-bsd
+	exit ;;
+    c2*)
+	if getsysinfo -f scalar_acc
+	then echo c32-convex-bsd
+	else echo c2-convex-bsd
+	fi
+	exit ;;
+    c34*)
+	echo c34-convex-bsd
+	exit ;;
+    c38*)
+	echo c38-convex-bsd
+	exit ;;
+    c4*)
+	echo c4-convex-bsd
+	exit ;;
+    esac
+fi
+
+cat >&2 <<EOF
+$0: unable to guess system type
+
+This script, last modified $timestamp, has failed to recognize
+the operating system you are using. It is advised that you
+download the most up to date version of the config scripts from
+
+  http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.guess;hb=HEAD
+and
+  http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.sub;hb=HEAD
+
+If the version you run ($0) is already up to date, please
+send the following data and any information you think might be
+pertinent to <config-patches@gnu.org> in order to provide the needed
+information to handle your system.
+
+config.guess timestamp = $timestamp
+
+uname -m = `(uname -m) 2>/dev/null || echo unknown`
+uname -r = `(uname -r) 2>/dev/null || echo unknown`
+uname -s = `(uname -s) 2>/dev/null || echo unknown`
+uname -v = `(uname -v) 2>/dev/null || echo unknown`
+
+/usr/bin/uname -p = `(/usr/bin/uname -p) 2>/dev/null`
+/bin/uname -X     = `(/bin/uname -X) 2>/dev/null`
+
+hostinfo               = `(hostinfo) 2>/dev/null`
+/bin/universe          = `(/bin/universe) 2>/dev/null`
+/usr/bin/arch -k       = `(/usr/bin/arch -k) 2>/dev/null`
+/bin/arch              = `(/bin/arch) 2>/dev/null`
+/usr/bin/oslevel       = `(/usr/bin/oslevel) 2>/dev/null`
+/usr/convex/getsysinfo = `(/usr/convex/getsysinfo) 2>/dev/null`
+
+UNAME_MACHINE = ${UNAME_MACHINE}
+UNAME_RELEASE = ${UNAME_RELEASE}
+UNAME_SYSTEM  = ${UNAME_SYSTEM}
+UNAME_VERSION = ${UNAME_VERSION}
+EOF
+
+exit 1
+
+# Local variables:
+# eval: (add-hook 'write-file-hooks 'time-stamp)
+# time-stamp-start: "timestamp='"
+# time-stamp-format: "%:y-%02m-%02d"
+# time-stamp-end: "'"
+# End:
diff --git a/extra/jemalloc/config.stamp.in b/extra/jemalloc/config.stamp.in
new file mode 100644
index 00000000000..e69de29bb2d
--- /dev/null
+++ b/extra/jemalloc/config.stamp.in
diff --git a/extra/jemalloc/config.sub b/extra/jemalloc/config.sub
new file mode 100755
index 00000000000..c894da45500
--- /dev/null
+++ b/extra/jemalloc/config.sub
@@ -0,0 +1,1773 @@
+#! /bin/sh
+# Configuration validation subroutine script.
+#   Copyright (C) 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
+#   2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010,
+#   2011, 2012 Free Software Foundation, Inc.
+
+timestamp='2012-02-10'
+
+# This file is (in principle) common to ALL GNU software.
+# The presence of a machine in this file suggests that SOME GNU software
+# can handle that machine.  It does not imply ALL GNU software can.
+#
+# This file is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, see <http://www.gnu.org/licenses/>.
+#
+# As a special exception to the GNU General Public License, if you
+# distribute this file as part of a program that contains a
+# configuration script generated by Autoconf, you may include it under
+# the same distribution terms that you use for the rest of that program.
+
+
+# Please send patches to <config-patches@gnu.org>.  Submit a context
+# diff and a properly formatted GNU ChangeLog entry.
+#
+# Configuration subroutine to validate and canonicalize a configuration type.
+# Supply the specified configuration type as an argument.
+# If it is invalid, we print an error message on stderr and exit with code 1.
+# Otherwise, we print the canonical config type on stdout and succeed.
+
+# You can get the latest version of this script from:
+# http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.sub;hb=HEAD
+
+# This file is supposed to be the same for all GNU packages
+# and recognize all the CPU types, system types and aliases
+# that are meaningful with *any* GNU software.
+# Each package is responsible for reporting which valid configurations
+# it does not support.  The user should be able to distinguish
+# a failure to support a valid configuration from a meaningless
+# configuration.
+
+# The goal of this file is to map all the various variations of a given
+# machine specification into a single specification in the form:
+#	CPU_TYPE-MANUFACTURER-OPERATING_SYSTEM
+# or in some cases, the newer four-part form:
+#	CPU_TYPE-MANUFACTURER-KERNEL-OPERATING_SYSTEM
+# It is wrong to echo any other type of specification.
+
+me=`echo "$0" | sed -e 's,.*/,,'`
+
+usage="\
+Usage: $0 [OPTION] CPU-MFR-OPSYS
+       $0 [OPTION] ALIAS
+
+Canonicalize a configuration name.
+
+Operation modes:
+  -h, --help         print this help, then exit
+  -t, --time-stamp   print date of last modification, then exit
+  -v, --version      print version number, then exit
+
+Report bugs and patches to <config-patches@gnu.org>."
+
+version="\
+GNU config.sub ($timestamp)
+
+Copyright (C) 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
+2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012
+Free Software Foundation, Inc.
+
+This is free software; see the source for copying conditions.  There is NO
+warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE."
+
+help="
+Try \`$me --help' for more information."
+
+# Parse command line
+while test $# -gt 0 ; do
+  case $1 in
+    --time-stamp | --time* | -t )
+       echo "$timestamp" ; exit ;;
+    --version | -v )
+       echo "$version" ; exit ;;
+    --help | --h* | -h )
+       echo "$usage"; exit ;;
+    -- )     # Stop option processing
+       shift; break ;;
+    - )	# Use stdin as input.
+       break ;;
+    -* )
+       echo "$me: invalid option $1$help"
+       exit 1 ;;
+
+    *local*)
+       # First pass through any local machine types.
+       echo $1
+       exit ;;
+
+    * )
+       break ;;
+  esac
+done
+
+case $# in
+ 0) echo "$me: missing argument$help" >&2
+    exit 1;;
+ 1) ;;
+ *) echo "$me: too many arguments$help" >&2
+    exit 1;;
+esac
+
+# Separate what the user gave into CPU-COMPANY and OS or KERNEL-OS (if any).
+# Here we must recognize all the valid KERNEL-OS combinations.
+maybe_os=`echo $1 | sed 's/^\(.*\)-\([^-]*-[^-]*\)$/\2/'`
+case $maybe_os in
+  nto-qnx* | linux-gnu* | linux-android* | linux-dietlibc | linux-newlib* | \
+  linux-uclibc* | uclinux-uclibc* | uclinux-gnu* | kfreebsd*-gnu* | \
+  knetbsd*-gnu* | netbsd*-gnu* | \
+  kopensolaris*-gnu* | \
+  storm-chaos* | os2-emx* | rtmk-nova*)
+    os=-$maybe_os
+    basic_machine=`echo $1 | sed 's/^\(.*\)-\([^-]*-[^-]*\)$/\1/'`
+    ;;
+  android-linux)
+    os=-linux-android
+    basic_machine=`echo $1 | sed 's/^\(.*\)-\([^-]*-[^-]*\)$/\1/'`-unknown
+    ;;
+  *)
+    basic_machine=`echo $1 | sed 's/-[^-]*$//'`
+    if [ $basic_machine != $1 ]
+    then os=`echo $1 | sed 's/.*-/-/'`
+    else os=; fi
+    ;;
+esac
+
+### Let's recognize common machines as not being operating systems so
+### that things like config.sub decstation-3100 work.  We also
+### recognize some manufacturers as not being operating systems, so we
+### can provide default operating systems below.
+case $os in
+	-sun*os*)
+		# Prevent following clause from handling this invalid input.
+		;;
+	-dec* | -mips* | -sequent* | -encore* | -pc532* | -sgi* | -sony* | \
+	-att* | -7300* | -3300* | -delta* | -motorola* | -sun[234]* | \
+	-unicom* | -ibm* | -next | -hp | -isi* | -apollo | -altos* | \
+	-convergent* | -ncr* | -news | -32* | -3600* | -3100* | -hitachi* |\
+	-c[123]* | -convex* | -sun | -crds | -omron* | -dg | -ultra | -tti* | \
+	-harris | -dolphin | -highlevel | -gould | -cbm | -ns | -masscomp | \
+	-apple | -axis | -knuth | -cray | -microblaze)
+		os=
+		basic_machine=$1
+		;;
+	-bluegene*)
+		os=-cnk
+		;;
+	-sim | -cisco | -oki | -wec | -winbond)
+		os=
+		basic_machine=$1
+		;;
+	-scout)
+		;;
+	-wrs)
+		os=-vxworks
+		basic_machine=$1
+		;;
+	-chorusos*)
+		os=-chorusos
+		basic_machine=$1
+		;;
+	-chorusrdb)
+		os=-chorusrdb
+		basic_machine=$1
+		;;
+	-hiux*)
+		os=-hiuxwe2
+		;;
+	-sco6)
+		os=-sco5v6
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco5)
+		os=-sco3.2v5
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco4)
+		os=-sco3.2v4
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco3.2.[4-9]*)
+		os=`echo $os | sed -e 's/sco3.2./sco3.2v/'`
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco3.2v[4-9]*)
+		# Don't forget version if it is 3.2v4 or newer.
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco5v6*)
+		# Don't forget version if it is 3.2v4 or newer.
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco*)
+		os=-sco3.2v2
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-udk*)
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-isc)
+		os=-isc2.2
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-clix*)
+		basic_machine=clipper-intergraph
+		;;
+	-isc*)
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-lynx*)
+		os=-lynxos
+		;;
+	-ptx*)
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-sequent/'`
+		;;
+	-windowsnt*)
+		os=`echo $os | sed -e 's/windowsnt/winnt/'`
+		;;
+	-psos*)
+		os=-psos
+		;;
+	-mint | -mint[0-9]*)
+		basic_machine=m68k-atari
+		os=-mint
+		;;
+esac
+
+# Decode aliases for certain CPU-COMPANY combinations.
+case $basic_machine in
+	# Recognize the basic CPU types without company name.
+	# Some are omitted here because they have special meanings below.
+	1750a | 580 \
+	| a29k \
+	| aarch64 | aarch64_be \
+	| alpha | alphaev[4-8] | alphaev56 | alphaev6[78] | alphapca5[67] \
+	| alpha64 | alpha64ev[4-8] | alpha64ev56 | alpha64ev6[78] | alpha64pca5[67] \
+	| am33_2.0 \
+	| arc | arm | arm[bl]e | arme[lb] | armv[2345] | armv[345][lb] | avr | avr32 \
+        | be32 | be64 \
+	| bfin \
+	| c4x | clipper \
+	| d10v | d30v | dlx | dsp16xx \
+	| epiphany \
+	| fido | fr30 | frv \
+	| h8300 | h8500 | hppa | hppa1.[01] | hppa2.0 | hppa2.0[nw] | hppa64 \
+	| hexagon \
+	| i370 | i860 | i960 | ia64 \
+	| ip2k | iq2000 \
+	| le32 | le64 \
+	| lm32 \
+	| m32c | m32r | m32rle | m68000 | m68k | m88k \
+	| maxq | mb | microblaze | mcore | mep | metag \
+	| mips | mipsbe | mipseb | mipsel | mipsle \
+	| mips16 \
+	| mips64 | mips64el \
+	| mips64octeon | mips64octeonel \
+	| mips64orion | mips64orionel \
+	| mips64r5900 | mips64r5900el \
+	| mips64vr | mips64vrel \
+	| mips64vr4100 | mips64vr4100el \
+	| mips64vr4300 | mips64vr4300el \
+	| mips64vr5000 | mips64vr5000el \
+	| mips64vr5900 | mips64vr5900el \
+	| mipsisa32 | mipsisa32el \
+	| mipsisa32r2 | mipsisa32r2el \
+	| mipsisa64 | mipsisa64el \
+	| mipsisa64r2 | mipsisa64r2el \
+	| mipsisa64sb1 | mipsisa64sb1el \
+	| mipsisa64sr71k | mipsisa64sr71kel \
+	| mipstx39 | mipstx39el \
+	| mn10200 | mn10300 \
+	| moxie \
+	| mt \
+	| msp430 \
+	| nds32 | nds32le | nds32be \
+	| nios | nios2 \
+	| ns16k | ns32k \
+	| open8 \
+	| or32 \
+	| pdp10 | pdp11 | pj | pjl \
+	| powerpc | powerpc64 | powerpc64le | powerpcle \
+	| pyramid \
+	| rl78 | rx \
+	| score \
+	| sh | sh[1234] | sh[24]a | sh[24]aeb | sh[23]e | sh[34]eb | sheb | shbe | shle | sh[1234]le | sh3ele \
+	| sh64 | sh64le \
+	| sparc | sparc64 | sparc64b | sparc64v | sparc86x | sparclet | sparclite \
+	| sparcv8 | sparcv9 | sparcv9b | sparcv9v \
+	| spu \
+	| tahoe | tic4x | tic54x | tic55x | tic6x | tic80 | tron \
+	| ubicom32 \
+	| v850 | v850e | v850e1 | v850e2 | v850es | v850e2v3 \
+	| we32k \
+	| x86 | xc16x | xstormy16 | xtensa \
+	| z8k | z80)
+		basic_machine=$basic_machine-unknown
+		;;
+	c54x)
+		basic_machine=tic54x-unknown
+		;;
+	c55x)
+		basic_machine=tic55x-unknown
+		;;
+	c6x)
+		basic_machine=tic6x-unknown
+		;;
+	m6811 | m68hc11 | m6812 | m68hc12 | m68hcs12x | picochip)
+		basic_machine=$basic_machine-unknown
+		os=-none
+		;;
+	m88110 | m680[12346]0 | m683?2 | m68360 | m5200 | v70 | w65 | z8k)
+		;;
+	ms1)
+		basic_machine=mt-unknown
+		;;
+
+	strongarm | thumb | xscale)
+		basic_machine=arm-unknown
+		;;
+	xgate)
+		basic_machine=$basic_machine-unknown
+		os=-none
+		;;
+	xscaleeb)
+		basic_machine=armeb-unknown
+		;;
+
+	xscaleel)
+		basic_machine=armel-unknown
+		;;
+
+	# We use `pc' rather than `unknown'
+	# because (1) that's what they normally are, and
+	# (2) the word "unknown" tends to confuse beginning users.
+	i*86 | x86_64)
+	  basic_machine=$basic_machine-pc
+	  ;;
+	# Object if more than one company name word.
+	*-*-*)
+		echo Invalid configuration \`$1\': machine \`$basic_machine\' not recognized 1>&2
+		exit 1
+		;;
+	# Recognize the basic CPU types with company name.
+	580-* \
+	| a29k-* \
+	| aarch64-* | aarch64_be-* \
+	| alpha-* | alphaev[4-8]-* | alphaev56-* | alphaev6[78]-* \
+	| alpha64-* | alpha64ev[4-8]-* | alpha64ev56-* | alpha64ev6[78]-* \
+	| alphapca5[67]-* | alpha64pca5[67]-* | arc-* \
+	| arm-*  | armbe-* | armle-* | armeb-* | armv*-* \
+	| avr-* | avr32-* \
+	| be32-* | be64-* \
+	| bfin-* | bs2000-* \
+	| c[123]* | c30-* | [cjt]90-* | c4x-* \
+	| clipper-* | craynv-* | cydra-* \
+	| d10v-* | d30v-* | dlx-* \
+	| elxsi-* \
+	| f30[01]-* | f700-* | fido-* | fr30-* | frv-* | fx80-* \
+	| h8300-* | h8500-* \
+	| hppa-* | hppa1.[01]-* | hppa2.0-* | hppa2.0[nw]-* | hppa64-* \
+	| hexagon-* \
+	| i*86-* | i860-* | i960-* | ia64-* \
+	| ip2k-* | iq2000-* \
+	| le32-* | le64-* \
+	| lm32-* \
+	| m32c-* | m32r-* | m32rle-* \
+	| m68000-* | m680[012346]0-* | m68360-* | m683?2-* | m68k-* \
+	| m88110-* | m88k-* | maxq-* | mcore-* | metag-* | microblaze-* \
+	| mips-* | mipsbe-* | mipseb-* | mipsel-* | mipsle-* \
+	| mips16-* \
+	| mips64-* | mips64el-* \
+	| mips64octeon-* | mips64octeonel-* \
+	| mips64orion-* | mips64orionel-* \
+	| mips64r5900-* | mips64r5900el-* \
+	| mips64vr-* | mips64vrel-* \
+	| mips64vr4100-* | mips64vr4100el-* \
+	| mips64vr4300-* | mips64vr4300el-* \
+	| mips64vr5000-* | mips64vr5000el-* \
+	| mips64vr5900-* | mips64vr5900el-* \
+	| mipsisa32-* | mipsisa32el-* \
+	| mipsisa32r2-* | mipsisa32r2el-* \
+	| mipsisa64-* | mipsisa64el-* \
+	| mipsisa64r2-* | mipsisa64r2el-* \
+	| mipsisa64sb1-* | mipsisa64sb1el-* \
+	| mipsisa64sr71k-* | mipsisa64sr71kel-* \
+	| mipstx39-* | mipstx39el-* \
+	| mmix-* \
+	| mt-* \
+	| msp430-* \
+	| nds32-* | nds32le-* | nds32be-* \
+	| nios-* | nios2-* \
+	| none-* | np1-* | ns16k-* | ns32k-* \
+	| open8-* \
+	| orion-* \
+	| pdp10-* | pdp11-* | pj-* | pjl-* | pn-* | power-* \
+	| powerpc-* | powerpc64-* | powerpc64le-* | powerpcle-* \
+	| pyramid-* \
+	| rl78-* | romp-* | rs6000-* | rx-* \
+	| sh-* | sh[1234]-* | sh[24]a-* | sh[24]aeb-* | sh[23]e-* | sh[34]eb-* | sheb-* | shbe-* \
+	| shle-* | sh[1234]le-* | sh3ele-* | sh64-* | sh64le-* \
+	| sparc-* | sparc64-* | sparc64b-* | sparc64v-* | sparc86x-* | sparclet-* \
+	| sparclite-* \
+	| sparcv8-* | sparcv9-* | sparcv9b-* | sparcv9v-* | sv1-* | sx?-* \
+	| tahoe-* \
+	| tic30-* | tic4x-* | tic54x-* | tic55x-* | tic6x-* | tic80-* \
+	| tile*-* \
+	| tron-* \
+	| ubicom32-* \
+	| v850-* | v850e-* | v850e1-* | v850es-* | v850e2-* | v850e2v3-* \
+	| vax-* \
+	| we32k-* \
+	| x86-* | x86_64-* | xc16x-* | xps100-* \
+	| xstormy16-* | xtensa*-* \
+	| ymp-* \
+	| z8k-* | z80-*)
+		;;
+	# Recognize the basic CPU types without company name, with glob match.
+	xtensa*)
+		basic_machine=$basic_machine-unknown
+		;;
+	# Recognize the various machine names and aliases which stand
+	# for a CPU type and a company and sometimes even an OS.
+	386bsd)
+		basic_machine=i386-unknown
+		os=-bsd
+		;;
+	3b1 | 7300 | 7300-att | att-7300 | pc7300 | safari | unixpc)
+		basic_machine=m68000-att
+		;;
+	3b*)
+		basic_machine=we32k-att
+		;;
+	a29khif)
+		basic_machine=a29k-amd
+		os=-udi
+		;;
+	abacus)
+		basic_machine=abacus-unknown
+		;;
+	adobe68k)
+		basic_machine=m68010-adobe
+		os=-scout
+		;;
+	alliant | fx80)
+		basic_machine=fx80-alliant
+		;;
+	altos | altos3068)
+		basic_machine=m68k-altos
+		;;
+	am29k)
+		basic_machine=a29k-none
+		os=-bsd
+		;;
+	amd64)
+		basic_machine=x86_64-pc
+		;;
+	amd64-*)
+		basic_machine=x86_64-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	amdahl)
+		basic_machine=580-amdahl
+		os=-sysv
+		;;
+	amiga | amiga-*)
+		basic_machine=m68k-unknown
+		;;
+	amigaos | amigados)
+		basic_machine=m68k-unknown
+		os=-amigaos
+		;;
+	amigaunix | amix)
+		basic_machine=m68k-unknown
+		os=-sysv4
+		;;
+	apollo68)
+		basic_machine=m68k-apollo
+		os=-sysv
+		;;
+	apollo68bsd)
+		basic_machine=m68k-apollo
+		os=-bsd
+		;;
+	aros)
+		basic_machine=i386-pc
+		os=-aros
+		;;
+	aux)
+		basic_machine=m68k-apple
+		os=-aux
+		;;
+	balance)
+		basic_machine=ns32k-sequent
+		os=-dynix
+		;;
+	blackfin)
+		basic_machine=bfin-unknown
+		os=-linux
+		;;
+	blackfin-*)
+		basic_machine=bfin-`echo $basic_machine | sed 's/^[^-]*-//'`
+		os=-linux
+		;;
+	bluegene*)
+		basic_machine=powerpc-ibm
+		os=-cnk
+		;;
+	c54x-*)
+		basic_machine=tic54x-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	c55x-*)
+		basic_machine=tic55x-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	c6x-*)
+		basic_machine=tic6x-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	c90)
+		basic_machine=c90-cray
+		os=-unicos
+		;;
+	cegcc)
+		basic_machine=arm-unknown
+		os=-cegcc
+		;;
+	convex-c1)
+		basic_machine=c1-convex
+		os=-bsd
+		;;
+	convex-c2)
+		basic_machine=c2-convex
+		os=-bsd
+		;;
+	convex-c32)
+		basic_machine=c32-convex
+		os=-bsd
+		;;
+	convex-c34)
+		basic_machine=c34-convex
+		os=-bsd
+		;;
+	convex-c38)
+		basic_machine=c38-convex
+		os=-bsd
+		;;
+	cray | j90)
+		basic_machine=j90-cray
+		os=-unicos
+		;;
+	craynv)
+		basic_machine=craynv-cray
+		os=-unicosmp
+		;;
+	cr16 | cr16-*)
+		basic_machine=cr16-unknown
+		os=-elf
+		;;
+	crds | unos)
+		basic_machine=m68k-crds
+		;;
+	crisv32 | crisv32-* | etraxfs*)
+		basic_machine=crisv32-axis
+		;;
+	cris | cris-* | etrax*)
+		basic_machine=cris-axis
+		;;
+	crx)
+		basic_machine=crx-unknown
+		os=-elf
+		;;
+	da30 | da30-*)
+		basic_machine=m68k-da30
+		;;
+	decstation | decstation-3100 | pmax | pmax-* | pmin | dec3100 | decstatn)
+		basic_machine=mips-dec
+		;;
+	decsystem10* | dec10*)
+		basic_machine=pdp10-dec
+		os=-tops10
+		;;
+	decsystem20* | dec20*)
+		basic_machine=pdp10-dec
+		os=-tops20
+		;;
+	delta | 3300 | motorola-3300 | motorola-delta \
+	      | 3300-motorola | delta-motorola)
+		basic_machine=m68k-motorola
+		;;
+	delta88)
+		basic_machine=m88k-motorola
+		os=-sysv3
+		;;
+	dicos)
+		basic_machine=i686-pc
+		os=-dicos
+		;;
+	djgpp)
+		basic_machine=i586-pc
+		os=-msdosdjgpp
+		;;
+	dpx20 | dpx20-*)
+		basic_machine=rs6000-bull
+		os=-bosx
+		;;
+	dpx2* | dpx2*-bull)
+		basic_machine=m68k-bull
+		os=-sysv3
+		;;
+	ebmon29k)
+		basic_machine=a29k-amd
+		os=-ebmon
+		;;
+	elxsi)
+		basic_machine=elxsi-elxsi
+		os=-bsd
+		;;
+	encore | umax | mmax)
+		basic_machine=ns32k-encore
+		;;
+	es1800 | OSE68k | ose68k | ose | OSE)
+		basic_machine=m68k-ericsson
+		os=-ose
+		;;
+	fx2800)
+		basic_machine=i860-alliant
+		;;
+	genix)
+		basic_machine=ns32k-ns
+		;;
+	gmicro)
+		basic_machine=tron-gmicro
+		os=-sysv
+		;;
+	go32)
+		basic_machine=i386-pc
+		os=-go32
+		;;
+	h3050r* | hiux*)
+		basic_machine=hppa1.1-hitachi
+		os=-hiuxwe2
+		;;
+	h8300hms)
+		basic_machine=h8300-hitachi
+		os=-hms
+		;;
+	h8300xray)
+		basic_machine=h8300-hitachi
+		os=-xray
+		;;
+	h8500hms)
+		basic_machine=h8500-hitachi
+		os=-hms
+		;;
+	harris)
+		basic_machine=m88k-harris
+		os=-sysv3
+		;;
+	hp300-*)
+		basic_machine=m68k-hp
+		;;
+	hp300bsd)
+		basic_machine=m68k-hp
+		os=-bsd
+		;;
+	hp300hpux)
+		basic_machine=m68k-hp
+		os=-hpux
+		;;
+	hp3k9[0-9][0-9] | hp9[0-9][0-9])
+		basic_machine=hppa1.0-hp
+		;;
+	hp9k2[0-9][0-9] | hp9k31[0-9])
+		basic_machine=m68000-hp
+		;;
+	hp9k3[2-9][0-9])
+		basic_machine=m68k-hp
+		;;
+	hp9k6[0-9][0-9] | hp6[0-9][0-9])
+		basic_machine=hppa1.0-hp
+		;;
+	hp9k7[0-79][0-9] | hp7[0-79][0-9])
+		basic_machine=hppa1.1-hp
+		;;
+	hp9k78[0-9] | hp78[0-9])
+		# FIXME: really hppa2.0-hp
+		basic_machine=hppa1.1-hp
+		;;
+	hp9k8[67]1 | hp8[67]1 | hp9k80[24] | hp80[24] | hp9k8[78]9 | hp8[78]9 | hp9k893 | hp893)
+		# FIXME: really hppa2.0-hp
+		basic_machine=hppa1.1-hp
+		;;
+	hp9k8[0-9][13679] | hp8[0-9][13679])
+		basic_machine=hppa1.1-hp
+		;;
+	hp9k8[0-9][0-9] | hp8[0-9][0-9])
+		basic_machine=hppa1.0-hp
+		;;
+	hppa-next)
+		os=-nextstep3
+		;;
+	hppaosf)
+		basic_machine=hppa1.1-hp
+		os=-osf
+		;;
+	hppro)
+		basic_machine=hppa1.1-hp
+		os=-proelf
+		;;
+	i370-ibm* | ibm*)
+		basic_machine=i370-ibm
+		;;
+	i*86v32)
+		basic_machine=`echo $1 | sed -e 's/86.*/86-pc/'`
+		os=-sysv32
+		;;
+	i*86v4*)
+		basic_machine=`echo $1 | sed -e 's/86.*/86-pc/'`
+		os=-sysv4
+		;;
+	i*86v)
+		basic_machine=`echo $1 | sed -e 's/86.*/86-pc/'`
+		os=-sysv
+		;;
+	i*86sol2)
+		basic_machine=`echo $1 | sed -e 's/86.*/86-pc/'`
+		os=-solaris2
+		;;
+	i386mach)
+		basic_machine=i386-mach
+		os=-mach
+		;;
+	i386-vsta | vsta)
+		basic_machine=i386-unknown
+		os=-vsta
+		;;
+	iris | iris4d)
+		basic_machine=mips-sgi
+		case $os in
+		    -irix*)
+			;;
+		    *)
+			os=-irix4
+			;;
+		esac
+		;;
+	isi68 | isi)
+		basic_machine=m68k-isi
+		os=-sysv
+		;;
+	m68knommu)
+		basic_machine=m68k-unknown
+		os=-linux
+		;;
+	m68knommu-*)
+		basic_machine=m68k-`echo $basic_machine | sed 's/^[^-]*-//'`
+		os=-linux
+		;;
+	m88k-omron*)
+		basic_machine=m88k-omron
+		;;
+	magnum | m3230)
+		basic_machine=mips-mips
+		os=-sysv
+		;;
+	merlin)
+		basic_machine=ns32k-utek
+		os=-sysv
+		;;
+	microblaze)
+		basic_machine=microblaze-xilinx
+		;;
+	mingw32)
+		basic_machine=i386-pc
+		os=-mingw32
+		;;
+	mingw32ce)
+		basic_machine=arm-unknown
+		os=-mingw32ce
+		;;
+	miniframe)
+		basic_machine=m68000-convergent
+		;;
+	*mint | -mint[0-9]* | *MiNT | *MiNT[0-9]*)
+		basic_machine=m68k-atari
+		os=-mint
+		;;
+	mips3*-*)
+		basic_machine=`echo $basic_machine | sed -e 's/mips3/mips64/'`
+		;;
+	mips3*)
+		basic_machine=`echo $basic_machine | sed -e 's/mips3/mips64/'`-unknown
+		;;
+	monitor)
+		basic_machine=m68k-rom68k
+		os=-coff
+		;;
+	morphos)
+		basic_machine=powerpc-unknown
+		os=-morphos
+		;;
+	msdos)
+		basic_machine=i386-pc
+		os=-msdos
+		;;
+	ms1-*)
+		basic_machine=`echo $basic_machine | sed -e 's/ms1-/mt-/'`
+		;;
+	msys)
+		basic_machine=i386-pc
+		os=-msys
+		;;
+	mvs)
+		basic_machine=i370-ibm
+		os=-mvs
+		;;
+	nacl)
+		basic_machine=le32-unknown
+		os=-nacl
+		;;
+	ncr3000)
+		basic_machine=i486-ncr
+		os=-sysv4
+		;;
+	netbsd386)
+		basic_machine=i386-unknown
+		os=-netbsd
+		;;
+	netwinder)
+		basic_machine=armv4l-rebel
+		os=-linux
+		;;
+	news | news700 | news800 | news900)
+		basic_machine=m68k-sony
+		os=-newsos
+		;;
+	news1000)
+		basic_machine=m68030-sony
+		os=-newsos
+		;;
+	news-3600 | risc-news)
+		basic_machine=mips-sony
+		os=-newsos
+		;;
+	necv70)
+		basic_machine=v70-nec
+		os=-sysv
+		;;
+	next | m*-next )
+		basic_machine=m68k-next
+		case $os in
+		    -nextstep* )
+			;;
+		    -ns2*)
+		      os=-nextstep2
+			;;
+		    *)
+		      os=-nextstep3
+			;;
+		esac
+		;;
+	nh3000)
+		basic_machine=m68k-harris
+		os=-cxux
+		;;
+	nh[45]000)
+		basic_machine=m88k-harris
+		os=-cxux
+		;;
+	nindy960)
+		basic_machine=i960-intel
+		os=-nindy
+		;;
+	mon960)
+		basic_machine=i960-intel
+		os=-mon960
+		;;
+	nonstopux)
+		basic_machine=mips-compaq
+		os=-nonstopux
+		;;
+	np1)
+		basic_machine=np1-gould
+		;;
+	neo-tandem)
+		basic_machine=neo-tandem
+		;;
+	nse-tandem)
+		basic_machine=nse-tandem
+		;;
+	nsr-tandem)
+		basic_machine=nsr-tandem
+		;;
+	op50n-* | op60c-*)
+		basic_machine=hppa1.1-oki
+		os=-proelf
+		;;
+	openrisc | openrisc-*)
+		basic_machine=or32-unknown
+		;;
+	os400)
+		basic_machine=powerpc-ibm
+		os=-os400
+		;;
+	OSE68000 | ose68000)
+		basic_machine=m68000-ericsson
+		os=-ose
+		;;
+	os68k)
+		basic_machine=m68k-none
+		os=-os68k
+		;;
+	pa-hitachi)
+		basic_machine=hppa1.1-hitachi
+		os=-hiuxwe2
+		;;
+	paragon)
+		basic_machine=i860-intel
+		os=-osf
+		;;
+	parisc)
+		basic_machine=hppa-unknown
+		os=-linux
+		;;
+	parisc-*)
+		basic_machine=hppa-`echo $basic_machine | sed 's/^[^-]*-//'`
+		os=-linux
+		;;
+	pbd)
+		basic_machine=sparc-tti
+		;;
+	pbb)
+		basic_machine=m68k-tti
+		;;
+	pc532 | pc532-*)
+		basic_machine=ns32k-pc532
+		;;
+	pc98)
+		basic_machine=i386-pc
+		;;
+	pc98-*)
+		basic_machine=i386-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	pentium | p5 | k5 | k6 | nexgen | viac3)
+		basic_machine=i586-pc
+		;;
+	pentiumpro | p6 | 6x86 | athlon | athlon_*)
+		basic_machine=i686-pc
+		;;
+	pentiumii | pentium2 | pentiumiii | pentium3)
+		basic_machine=i686-pc
+		;;
+	pentium4)
+		basic_machine=i786-pc
+		;;
+	pentium-* | p5-* | k5-* | k6-* | nexgen-* | viac3-*)
+		basic_machine=i586-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	pentiumpro-* | p6-* | 6x86-* | athlon-*)
+		basic_machine=i686-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	pentiumii-* | pentium2-* | pentiumiii-* | pentium3-*)
+		basic_machine=i686-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	pentium4-*)
+		basic_machine=i786-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	pn)
+		basic_machine=pn-gould
+		;;
+	power)	basic_machine=power-ibm
+		;;
+	ppc | ppcbe)	basic_machine=powerpc-unknown
+		;;
+	ppc-* | ppcbe-*)
+		basic_machine=powerpc-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	ppcle | powerpclittle | ppc-le | powerpc-little)
+		basic_machine=powerpcle-unknown
+		;;
+	ppcle-* | powerpclittle-*)
+		basic_machine=powerpcle-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	ppc64)	basic_machine=powerpc64-unknown
+		;;
+	ppc64-*) basic_machine=powerpc64-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	ppc64le | powerpc64little | ppc64-le | powerpc64-little)
+		basic_machine=powerpc64le-unknown
+		;;
+	ppc64le-* | powerpc64little-*)
+		basic_machine=powerpc64le-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	ps2)
+		basic_machine=i386-ibm
+		;;
+	pw32)
+		basic_machine=i586-unknown
+		os=-pw32
+		;;
+	rdos)
+		basic_machine=i386-pc
+		os=-rdos
+		;;
+	rom68k)
+		basic_machine=m68k-rom68k
+		os=-coff
+		;;
+	rm[46]00)
+		basic_machine=mips-siemens
+		;;
+	rtpc | rtpc-*)
+		basic_machine=romp-ibm
+		;;
+	s390 | s390-*)
+		basic_machine=s390-ibm
+		;;
+	s390x | s390x-*)
+		basic_machine=s390x-ibm
+		;;
+	sa29200)
+		basic_machine=a29k-amd
+		os=-udi
+		;;
+	sb1)
+		basic_machine=mipsisa64sb1-unknown
+		;;
+	sb1el)
+		basic_machine=mipsisa64sb1el-unknown
+		;;
+	sde)
+		basic_machine=mipsisa32-sde
+		os=-elf
+		;;
+	sei)
+		basic_machine=mips-sei
+		os=-seiux
+		;;
+	sequent)
+		basic_machine=i386-sequent
+		;;
+	sh)
+		basic_machine=sh-hitachi
+		os=-hms
+		;;
+	sh5el)
+		basic_machine=sh5le-unknown
+		;;
+	sh64)
+		basic_machine=sh64-unknown
+		;;
+	sparclite-wrs | simso-wrs)
+		basic_machine=sparclite-wrs
+		os=-vxworks
+		;;
+	sps7)
+		basic_machine=m68k-bull
+		os=-sysv2
+		;;
+	spur)
+		basic_machine=spur-unknown
+		;;
+	st2000)
+		basic_machine=m68k-tandem
+		;;
+	stratus)
+		basic_machine=i860-stratus
+		os=-sysv4
+		;;
+	strongarm-* | thumb-*)
+		basic_machine=arm-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	sun2)
+		basic_machine=m68000-sun
+		;;
+	sun2os3)
+		basic_machine=m68000-sun
+		os=-sunos3
+		;;
+	sun2os4)
+		basic_machine=m68000-sun
+		os=-sunos4
+		;;
+	sun3os3)
+		basic_machine=m68k-sun
+		os=-sunos3
+		;;
+	sun3os4)
+		basic_machine=m68k-sun
+		os=-sunos4
+		;;
+	sun4os3)
+		basic_machine=sparc-sun
+		os=-sunos3
+		;;
+	sun4os4)
+		basic_machine=sparc-sun
+		os=-sunos4
+		;;
+	sun4sol2)
+		basic_machine=sparc-sun
+		os=-solaris2
+		;;
+	sun3 | sun3-*)
+		basic_machine=m68k-sun
+		;;
+	sun4)
+		basic_machine=sparc-sun
+		;;
+	sun386 | sun386i | roadrunner)
+		basic_machine=i386-sun
+		;;
+	sv1)
+		basic_machine=sv1-cray
+		os=-unicos
+		;;
+	symmetry)
+		basic_machine=i386-sequent
+		os=-dynix
+		;;
+	t3e)
+		basic_machine=alphaev5-cray
+		os=-unicos
+		;;
+	t90)
+		basic_machine=t90-cray
+		os=-unicos
+		;;
+	tile*)
+		basic_machine=$basic_machine-unknown
+		os=-linux-gnu
+		;;
+	tx39)
+		basic_machine=mipstx39-unknown
+		;;
+	tx39el)
+		basic_machine=mipstx39el-unknown
+		;;
+	toad1)
+		basic_machine=pdp10-xkl
+		os=-tops20
+		;;
+	tower | tower-32)
+		basic_machine=m68k-ncr
+		;;
+	tpf)
+		basic_machine=s390x-ibm
+		os=-tpf
+		;;
+	udi29k)
+		basic_machine=a29k-amd
+		os=-udi
+		;;
+	ultra3)
+		basic_machine=a29k-nyu
+		os=-sym1
+		;;
+	v810 | necv810)
+		basic_machine=v810-nec
+		os=-none
+		;;
+	vaxv)
+		basic_machine=vax-dec
+		os=-sysv
+		;;
+	vms)
+		basic_machine=vax-dec
+		os=-vms
+		;;
+	vpp*|vx|vx-*)
+		basic_machine=f301-fujitsu
+		;;
+	vxworks960)
+		basic_machine=i960-wrs
+		os=-vxworks
+		;;
+	vxworks68)
+		basic_machine=m68k-wrs
+		os=-vxworks
+		;;
+	vxworks29k)
+		basic_machine=a29k-wrs
+		os=-vxworks
+		;;
+	w65*)
+		basic_machine=w65-wdc
+		os=-none
+		;;
+	w89k-*)
+		basic_machine=hppa1.1-winbond
+		os=-proelf
+		;;
+	xbox)
+		basic_machine=i686-pc
+		os=-mingw32
+		;;
+	xps | xps100)
+		basic_machine=xps100-honeywell
+		;;
+	xscale-* | xscalee[bl]-*)
+		basic_machine=`echo $basic_machine | sed 's/^xscale/arm/'`
+		;;
+	ymp)
+		basic_machine=ymp-cray
+		os=-unicos
+		;;
+	z8k-*-coff)
+		basic_machine=z8k-unknown
+		os=-sim
+		;;
+	z80-*-coff)
+		basic_machine=z80-unknown
+		os=-sim
+		;;
+	none)
+		basic_machine=none-none
+		os=-none
+		;;
+
+# Here we handle the default manufacturer of certain CPU types.  It is in
+# some cases the only manufacturer, in others, it is the most popular.
+	w89k)
+		basic_machine=hppa1.1-winbond
+		;;
+	op50n)
+		basic_machine=hppa1.1-oki
+		;;
+	op60c)
+		basic_machine=hppa1.1-oki
+		;;
+	romp)
+		basic_machine=romp-ibm
+		;;
+	mmix)
+		basic_machine=mmix-knuth
+		;;
+	rs6000)
+		basic_machine=rs6000-ibm
+		;;
+	vax)
+		basic_machine=vax-dec
+		;;
+	pdp10)
+		# there are many clones, so DEC is not a safe bet
+		basic_machine=pdp10-unknown
+		;;
+	pdp11)
+		basic_machine=pdp11-dec
+		;;
+	we32k)
+		basic_machine=we32k-att
+		;;
+	sh[1234] | sh[24]a | sh[24]aeb | sh[34]eb | sh[1234]le | sh[23]ele)
+		basic_machine=sh-unknown
+		;;
+	sparc | sparcv8 | sparcv9 | sparcv9b | sparcv9v)
+		basic_machine=sparc-sun
+		;;
+	cydra)
+		basic_machine=cydra-cydrome
+		;;
+	orion)
+		basic_machine=orion-highlevel
+		;;
+	orion105)
+		basic_machine=clipper-highlevel
+		;;
+	mac | mpw | mac-mpw)
+		basic_machine=m68k-apple
+		;;
+	pmac | pmac-mpw)
+		basic_machine=powerpc-apple
+		;;
+	*-unknown)
+		# Make sure to match an already-canonicalized machine name.
+		;;
+	*)
+		echo Invalid configuration \`$1\': machine \`$basic_machine\' not recognized 1>&2
+		exit 1
+		;;
+esac
+
+# Here we canonicalize certain aliases for manufacturers.
+case $basic_machine in
+	*-digital*)
+		basic_machine=`echo $basic_machine | sed 's/digital.*/dec/'`
+		;;
+	*-commodore*)
+		basic_machine=`echo $basic_machine | sed 's/commodore.*/cbm/'`
+		;;
+	*)
+		;;
+esac
+
+# Decode manufacturer-specific aliases for certain operating systems.
+
+if [ x"$os" != x"" ]
+then
+case $os in
+	# First match some system type aliases
+	# that might get confused with valid system types.
+	# -solaris* is a basic system type, with this one exception.
+	-auroraux)
+		os=-auroraux
+		;;
+	-solaris1 | -solaris1.*)
+		os=`echo $os | sed -e 's|solaris1|sunos4|'`
+		;;
+	-solaris)
+		os=-solaris2
+		;;
+	-svr4*)
+		os=-sysv4
+		;;
+	-unixware*)
+		os=-sysv4.2uw
+		;;
+	-gnu/linux*)
+		os=`echo $os | sed -e 's|gnu/linux|linux-gnu|'`
+		;;
+	# First accept the basic system types.
+	# The portable systems comes first.
+	# Each alternative MUST END IN A *, to match a version number.
+	# -sysv* is not here because it comes later, after sysvr4.
+	-gnu* | -bsd* | -mach* | -minix* | -genix* | -ultrix* | -irix* \
+	      | -*vms* | -sco* | -esix* | -isc* | -aix* | -cnk* | -sunos | -sunos[34]*\
+	      | -hpux* | -unos* | -osf* | -luna* | -dgux* | -auroraux* | -solaris* \
+	      | -sym* | -kopensolaris* \
+	      | -amigaos* | -amigados* | -msdos* | -newsos* | -unicos* | -aof* \
+	      | -aos* | -aros* \
+	      | -nindy* | -vxsim* | -vxworks* | -ebmon* | -hms* | -mvs* \
+	      | -clix* | -riscos* | -uniplus* | -iris* | -rtu* | -xenix* \
+	      | -hiux* | -386bsd* | -knetbsd* | -mirbsd* | -netbsd* \
+	      | -openbsd* | -solidbsd* \
+	      | -ekkobsd* | -kfreebsd* | -freebsd* | -riscix* | -lynxos* \
+	      | -bosx* | -nextstep* | -cxux* | -aout* | -elf* | -oabi* \
+	      | -ptx* | -coff* | -ecoff* | -winnt* | -domain* | -vsta* \
+	      | -udi* | -eabi* | -lites* | -ieee* | -go32* | -aux* \
+	      | -chorusos* | -chorusrdb* | -cegcc* \
+	      | -cygwin* | -msys* | -pe* | -psos* | -moss* | -proelf* | -rtems* \
+	      | -mingw32* | -linux-gnu* | -linux-android* \
+	      | -linux-newlib* | -linux-uclibc* \
+	      | -uxpv* | -beos* | -mpeix* | -udk* \
+	      | -interix* | -uwin* | -mks* | -rhapsody* | -darwin* | -opened* \
+	      | -openstep* | -oskit* | -conix* | -pw32* | -nonstopux* \
+	      | -storm-chaos* | -tops10* | -tenex* | -tops20* | -its* \
+	      | -os2* | -vos* | -palmos* | -uclinux* | -nucleus* \
+	      | -morphos* | -superux* | -rtmk* | -rtmk-nova* | -windiss* \
+	      | -powermax* | -dnix* | -nx6 | -nx7 | -sei* | -dragonfly* \
+	      | -skyos* | -haiku* | -rdos* | -toppers* | -drops* | -es*)
+	# Remember, each alternative MUST END IN *, to match a version number.
+		;;
+	-qnx*)
+		case $basic_machine in
+		    x86-* | i*86-*)
+			;;
+		    *)
+			os=-nto$os
+			;;
+		esac
+		;;
+	-nto-qnx*)
+		;;
+	-nto*)
+		os=`echo $os | sed -e 's|nto|nto-qnx|'`
+		;;
+	-sim | -es1800* | -hms* | -xray | -os68k* | -none* | -v88r* \
+	      | -windows* | -osx | -abug | -netware* | -os9* | -beos* | -haiku* \
+	      | -macos* | -mpw* | -magic* | -mmixware* | -mon960* | -lnews*)
+		;;
+	-mac*)
+		os=`echo $os | sed -e 's|mac|macos|'`
+		;;
+	-linux-dietlibc)
+		os=-linux-dietlibc
+		;;
+	-linux*)
+		os=`echo $os | sed -e 's|linux|linux-gnu|'`
+		;;
+	-sunos5*)
+		os=`echo $os | sed -e 's|sunos5|solaris2|'`
+		;;
+	-sunos6*)
+		os=`echo $os | sed -e 's|sunos6|solaris3|'`
+		;;
+	-opened*)
+		os=-openedition
+		;;
+	-os400*)
+		os=-os400
+		;;
+	-wince*)
+		os=-wince
+		;;
+	-osfrose*)
+		os=-osfrose
+		;;
+	-osf*)
+		os=-osf
+		;;
+	-utek*)
+		os=-bsd
+		;;
+	-dynix*)
+		os=-bsd
+		;;
+	-acis*)
+		os=-aos
+		;;
+	-atheos*)
+		os=-atheos
+		;;
+	-syllable*)
+		os=-syllable
+		;;
+	-386bsd)
+		os=-bsd
+		;;
+	-ctix* | -uts*)
+		os=-sysv
+		;;
+	-nova*)
+		os=-rtmk-nova
+		;;
+	-ns2 )
+		os=-nextstep2
+		;;
+	-nsk*)
+		os=-nsk
+		;;
+	# Preserve the version number of sinix5.
+	-sinix5.*)
+		os=`echo $os | sed -e 's|sinix|sysv|'`
+		;;
+	-sinix*)
+		os=-sysv4
+		;;
+	-tpf*)
+		os=-tpf
+		;;
+	-triton*)
+		os=-sysv3
+		;;
+	-oss*)
+		os=-sysv3
+		;;
+	-svr4)
+		os=-sysv4
+		;;
+	-svr3)
+		os=-sysv3
+		;;
+	-sysvr4)
+		os=-sysv4
+		;;
+	# This must come after -sysvr4.
+	-sysv*)
+		;;
+	-ose*)
+		os=-ose
+		;;
+	-es1800*)
+		os=-ose
+		;;
+	-xenix)
+		os=-xenix
+		;;
+	-*mint | -mint[0-9]* | -*MiNT | -MiNT[0-9]*)
+		os=-mint
+		;;
+	-aros*)
+		os=-aros
+		;;
+	-kaos*)
+		os=-kaos
+		;;
+	-zvmoe)
+		os=-zvmoe
+		;;
+	-dicos*)
+		os=-dicos
+		;;
+	-nacl*)
+		;;
+	-none)
+		;;
+	*)
+		# Get rid of the `-' at the beginning of $os.
+		os=`echo $os | sed 's/[^-]*-//'`
+		echo Invalid configuration \`$1\': system \`$os\' not recognized 1>&2
+		exit 1
+		;;
+esac
+else
+
+# Here we handle the default operating systems that come with various machines.
+# The value should be what the vendor currently ships out the door with their
+# machine or put another way, the most popular os provided with the machine.
+
+# Note that if you're going to try to match "-MANUFACTURER" here (say,
+# "-sun"), then you have to tell the case statement up towards the top
+# that MANUFACTURER isn't an operating system.  Otherwise, code above
+# will signal an error saying that MANUFACTURER isn't an operating
+# system, and we'll never get to this point.
+
+case $basic_machine in
+	score-*)
+		os=-elf
+		;;
+	spu-*)
+		os=-elf
+		;;
+	*-acorn)
+		os=-riscix1.2
+		;;
+	arm*-rebel)
+		os=-linux
+		;;
+	arm*-semi)
+		os=-aout
+		;;
+	c4x-* | tic4x-*)
+		os=-coff
+		;;
+	tic54x-*)
+		os=-coff
+		;;
+	tic55x-*)
+		os=-coff
+		;;
+	tic6x-*)
+		os=-coff
+		;;
+	# This must come before the *-dec entry.
+	pdp10-*)
+		os=-tops20
+		;;
+	pdp11-*)
+		os=-none
+		;;
+	*-dec | vax-*)
+		os=-ultrix4.2
+		;;
+	m68*-apollo)
+		os=-domain
+		;;
+	i386-sun)
+		os=-sunos4.0.2
+		;;
+	m68000-sun)
+		os=-sunos3
+		;;
+	m68*-cisco)
+		os=-aout
+		;;
+	mep-*)
+		os=-elf
+		;;
+	mips*-cisco)
+		os=-elf
+		;;
+	mips*-*)
+		os=-elf
+		;;
+	or32-*)
+		os=-coff
+		;;
+	*-tti)	# must be before sparc entry or we get the wrong os.
+		os=-sysv3
+		;;
+	sparc-* | *-sun)
+		os=-sunos4.1.1
+		;;
+	*-be)
+		os=-beos
+		;;
+	*-haiku)
+		os=-haiku
+		;;
+	*-ibm)
+		os=-aix
+		;;
+	*-knuth)
+		os=-mmixware
+		;;
+	*-wec)
+		os=-proelf
+		;;
+	*-winbond)
+		os=-proelf
+		;;
+	*-oki)
+		os=-proelf
+		;;
+	*-hp)
+		os=-hpux
+		;;
+	*-hitachi)
+		os=-hiux
+		;;
+	i860-* | *-att | *-ncr | *-altos | *-motorola | *-convergent)
+		os=-sysv
+		;;
+	*-cbm)
+		os=-amigaos
+		;;
+	*-dg)
+		os=-dgux
+		;;
+	*-dolphin)
+		os=-sysv3
+		;;
+	m68k-ccur)
+		os=-rtu
+		;;
+	m88k-omron*)
+		os=-luna
+		;;
+	*-next )
+		os=-nextstep
+		;;
+	*-sequent)
+		os=-ptx
+		;;
+	*-crds)
+		os=-unos
+		;;
+	*-ns)
+		os=-genix
+		;;
+	i370-*)
+		os=-mvs
+		;;
+	*-next)
+		os=-nextstep3
+		;;
+	*-gould)
+		os=-sysv
+		;;
+	*-highlevel)
+		os=-bsd
+		;;
+	*-encore)
+		os=-bsd
+		;;
+	*-sgi)
+		os=-irix
+		;;
+	*-siemens)
+		os=-sysv4
+		;;
+	*-masscomp)
+		os=-rtu
+		;;
+	f30[01]-fujitsu | f700-fujitsu)
+		os=-uxpv
+		;;
+	*-rom68k)
+		os=-coff
+		;;
+	*-*bug)
+		os=-coff
+		;;
+	*-apple)
+		os=-macos
+		;;
+	*-atari*)
+		os=-mint
+		;;
+	*)
+		os=-none
+		;;
+esac
+fi
+
+# Here we handle the case where we know the os, and the CPU type, but not the
+# manufacturer.  We pick the logical manufacturer.
+vendor=unknown
+case $basic_machine in
+	*-unknown)
+		case $os in
+			-riscix*)
+				vendor=acorn
+				;;
+			-sunos*)
+				vendor=sun
+				;;
+			-cnk*|-aix*)
+				vendor=ibm
+				;;
+			-beos*)
+				vendor=be
+				;;
+			-hpux*)
+				vendor=hp
+				;;
+			-mpeix*)
+				vendor=hp
+				;;
+			-hiux*)
+				vendor=hitachi
+				;;
+			-unos*)
+				vendor=crds
+				;;
+			-dgux*)
+				vendor=dg
+				;;
+			-luna*)
+				vendor=omron
+				;;
+			-genix*)
+				vendor=ns
+				;;
+			-mvs* | -opened*)
+				vendor=ibm
+				;;
+			-os400*)
+				vendor=ibm
+				;;
+			-ptx*)
+				vendor=sequent
+				;;
+			-tpf*)
+				vendor=ibm
+				;;
+			-vxsim* | -vxworks* | -windiss*)
+				vendor=wrs
+				;;
+			-aux*)
+				vendor=apple
+				;;
+			-hms*)
+				vendor=hitachi
+				;;
+			-mpw* | -macos*)
+				vendor=apple
+				;;
+			-*mint | -mint[0-9]* | -*MiNT | -MiNT[0-9]*)
+				vendor=atari
+				;;
+			-vos*)
+				vendor=stratus
+				;;
+		esac
+		basic_machine=`echo $basic_machine | sed "s/unknown/$vendor/"`
+		;;
+esac
+
+echo $basic_machine$os
+exit
+
+# Local variables:
+# eval: (add-hook 'write-file-hooks 'time-stamp)
+# time-stamp-start: "timestamp='"
+# time-stamp-format: "%:y-%02m-%02d"
+# time-stamp-end: "'"
+# End:
diff --git a/extra/jemalloc/configure b/extra/jemalloc/configure
new file mode 100755
index 00000000000..c112a47d45d
--- /dev/null
+++ b/extra/jemalloc/configure
@@ -0,0 +1,8339 @@
+#! /bin/sh
+# Guess values for system-dependent variables and create Makefiles.
+# Generated by GNU Autoconf 2.68.
+#
+#
+# Copyright (C) 1992, 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2001,
+# 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010 Free Software
+# Foundation, Inc.
+#
+#
+# This configure script is free software; the Free Software Foundation
+# gives unlimited permission to copy, distribute and modify it.
+## -------------------- ##
+## M4sh Initialization. ##
+## -------------------- ##
+
+# Be more Bourne compatible
+DUALCASE=1; export DUALCASE # for MKS sh
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '${1+"$@"}'='"$@"'
+  setopt NO_GLOB_SUBST
+else
+  case `(set -o) 2>/dev/null` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
+esac
+fi
+
+
+as_nl='
+'
+export as_nl
+# Printing a long string crashes Solaris 7 /usr/bin/printf.
+as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
+as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
+as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
+# Prefer a ksh shell builtin over an external printf program on Solaris,
+# but without wasting forks for bash or zsh.
+if test -z "$BASH_VERSION$ZSH_VERSION" \
+    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
+  as_echo='print -r --'
+  as_echo_n='print -rn --'
+elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+  as_echo='printf %s\n'
+  as_echo_n='printf %s'
+else
+  if test "X`(/usr/ucb/echo -n -n $as_echo) 2>/dev/null`" = "X-n $as_echo"; then
+    as_echo_body='eval /usr/ucb/echo -n "$1$as_nl"'
+    as_echo_n='/usr/ucb/echo -n'
+  else
+    as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
+    as_echo_n_body='eval
+      arg=$1;
+      case $arg in #(
+      *"$as_nl"*)
+	expr "X$arg" : "X\\(.*\\)$as_nl";
+	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
+      esac;
+      expr "X$arg" : "X\\(.*\\)" | tr -d "$as_nl"
+    '
+    export as_echo_n_body
+    as_echo_n='sh -c $as_echo_n_body as_echo'
+  fi
+  export as_echo_body
+  as_echo='sh -c $as_echo_body as_echo'
+fi
+
+# The user is always right.
+if test "${PATH_SEPARATOR+set}" != set; then
+  PATH_SEPARATOR=:
+  (PATH='/bin;/bin'; FPATH=$PATH; sh -c :) >/dev/null 2>&1 && {
+    (PATH='/bin:/bin'; FPATH=$PATH; sh -c :) >/dev/null 2>&1 ||
+      PATH_SEPARATOR=';'
+  }
+fi
+
+
+# IFS
+# We need space, tab and new line, in precisely that order.  Quoting is
+# there to prevent editors from complaining about space-tab.
+# (If _AS_PATH_WALK were called with IFS unset, it would disable word
+# splitting by setting IFS to empty value.)
+IFS=" ""	$as_nl"
+
+# Find who we are.  Look in the path if we contain no directory separator.
+as_myself=
+case $0 in #((
+  *[\\/]* ) as_myself=$0 ;;
+  *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+  done
+IFS=$as_save_IFS
+
+     ;;
+esac
+# We did not find ourselves, most probably we were run as `sh COMMAND'
+# in which case we are not to be found in the path.
+if test "x$as_myself" = x; then
+  as_myself=$0
+fi
+if test ! -f "$as_myself"; then
+  $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
+  exit 1
+fi
+
+# Unset variables that we do not need and which cause bugs (e.g. in
+# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
+# suppresses any "Segmentation fault" message there.  '((' could
+# trigger a bug in pdksh 5.2.14.
+for as_var in BASH_ENV ENV MAIL MAILPATH
+do eval test x\${$as_var+set} = xset \
+  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
+done
+PS1='$ '
+PS2='> '
+PS4='+ '
+
+# NLS nuisances.
+LC_ALL=C
+export LC_ALL
+LANGUAGE=C
+export LANGUAGE
+
+# CDPATH.
+(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
+
+if test "x$CONFIG_SHELL" = x; then
+  as_bourne_compatible="if test -n \"\${ZSH_VERSION+set}\" && (emulate sh) >/dev/null 2>&1; then :
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on \${1+\"\$@\"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '\${1+\"\$@\"}'='\"\$@\"'
+  setopt NO_GLOB_SUBST
+else
+  case \`(set -o) 2>/dev/null\` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
+esac
+fi
+"
+  as_required="as_fn_return () { (exit \$1); }
+as_fn_success () { as_fn_return 0; }
+as_fn_failure () { as_fn_return 1; }
+as_fn_ret_success () { return 0; }
+as_fn_ret_failure () { return 1; }
+
+exitcode=0
+as_fn_success || { exitcode=1; echo as_fn_success failed.; }
+as_fn_failure && { exitcode=1; echo as_fn_failure succeeded.; }
+as_fn_ret_success || { exitcode=1; echo as_fn_ret_success failed.; }
+as_fn_ret_failure && { exitcode=1; echo as_fn_ret_failure succeeded.; }
+if ( set x; as_fn_ret_success y && test x = \"\$1\" ); then :
+
+else
+  exitcode=1; echo positional parameters were not saved.
+fi
+test x\$exitcode = x0 || exit 1"
+  as_suggested="  as_lineno_1=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_1a=\$LINENO
+  as_lineno_2=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_2a=\$LINENO
+  eval 'test \"x\$as_lineno_1'\$as_run'\" != \"x\$as_lineno_2'\$as_run'\" &&
+  test \"x\`expr \$as_lineno_1'\$as_run' + 1\`\" = \"x\$as_lineno_2'\$as_run'\"' || exit 1
+test \$(( 1 + 1 )) = 2 || exit 1"
+  if (eval "$as_required") 2>/dev/null; then :
+  as_have_required=yes
+else
+  as_have_required=no
+fi
+  if test x$as_have_required = xyes && (eval "$as_suggested") 2>/dev/null; then :
+
+else
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+as_found=false
+for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+  as_found=:
+  case $as_dir in #(
+	 /*)
+	   for as_base in sh bash ksh sh5; do
+	     # Try only shells that exist, to save several forks.
+	     as_shell=$as_dir/$as_base
+	     if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
+		    { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$as_shell"; } 2>/dev/null; then :
+  CONFIG_SHELL=$as_shell as_have_required=yes
+		   if { $as_echo "$as_bourne_compatible""$as_suggested" | as_run=a "$as_shell"; } 2>/dev/null; then :
+  break 2
+fi
+fi
+	   done;;
+       esac
+  as_found=false
+done
+$as_found || { if { test -f "$SHELL" || test -f "$SHELL.exe"; } &&
+	      { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$SHELL"; } 2>/dev/null; then :
+  CONFIG_SHELL=$SHELL as_have_required=yes
+fi; }
+IFS=$as_save_IFS
+
+
+      if test "x$CONFIG_SHELL" != x; then :
+  # We cannot yet assume a decent shell, so we have to provide a
+	# neutralization value for shells without unset; and this also
+	# works around shells that cannot unset nonexistent variables.
+	# Preserve -v and -x to the replacement shell.
+	BASH_ENV=/dev/null
+	ENV=/dev/null
+	(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
+	export CONFIG_SHELL
+	case $- in # ((((
+	  *v*x* | *x*v* ) as_opts=-vx ;;
+	  *v* ) as_opts=-v ;;
+	  *x* ) as_opts=-x ;;
+	  * ) as_opts= ;;
+	esac
+	exec "$CONFIG_SHELL" $as_opts "$as_myself" ${1+"$@"}
+fi
+
+    if test x$as_have_required = xno; then :
+  $as_echo "$0: This script requires a shell more modern than all"
+  $as_echo "$0: the shells that I found on your system."
+  if test x${ZSH_VERSION+set} = xset ; then
+    $as_echo "$0: In particular, zsh $ZSH_VERSION has bugs and should"
+    $as_echo "$0: be upgraded to zsh 4.3.4 or later."
+  else
+    $as_echo "$0: Please tell bug-autoconf@gnu.org about your system,
+$0: including any error possibly output before this
+$0: message. Then install a modern shell, or manually run
+$0: the script under such a shell if you do have one."
+  fi
+  exit 1
+fi
+fi
+fi
+SHELL=${CONFIG_SHELL-/bin/sh}
+export SHELL
+# Unset more variables known to interfere with behavior of common tools.
+CLICOLOR_FORCE= GREP_OPTIONS=
+unset CLICOLOR_FORCE GREP_OPTIONS
+
+## --------------------- ##
+## M4sh Shell Functions. ##
+## --------------------- ##
+# as_fn_unset VAR
+# ---------------
+# Portably unset VAR.
+as_fn_unset ()
+{
+  { eval $1=; unset $1;}
+}
+as_unset=as_fn_unset
+
+# as_fn_set_status STATUS
+# -----------------------
+# Set $? to STATUS, without forking.
+as_fn_set_status ()
+{
+  return $1
+} # as_fn_set_status
+
+# as_fn_exit STATUS
+# -----------------
+# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
+as_fn_exit ()
+{
+  set +e
+  as_fn_set_status $1
+  exit $1
+} # as_fn_exit
+
+# as_fn_mkdir_p
+# -------------
+# Create "$as_dir" as a directory, including parents if necessary.
+as_fn_mkdir_p ()
+{
+
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || eval $as_mkdir_p || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
+
+
+} # as_fn_mkdir_p
+# as_fn_append VAR VALUE
+# ----------------------
+# Append the text in VALUE to the end of the definition contained in VAR. Take
+# advantage of any shell optimizations that allow amortized linear growth over
+# repeated appends, instead of the typical quadratic growth present in naive
+# implementations.
+if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
+  eval 'as_fn_append ()
+  {
+    eval $1+=\$2
+  }'
+else
+  as_fn_append ()
+  {
+    eval $1=\$$1\$2
+  }
+fi # as_fn_append
+
+# as_fn_arith ARG...
+# ------------------
+# Perform arithmetic evaluation on the ARGs, and store the result in the
+# global $as_val. Take advantage of shells that can avoid forks. The arguments
+# must be portable across $(()) and expr.
+if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
+  eval 'as_fn_arith ()
+  {
+    as_val=$(( $* ))
+  }'
+else
+  as_fn_arith ()
+  {
+    as_val=`expr "$@" || test $? -eq 1`
+  }
+fi # as_fn_arith
+
+
+# as_fn_error STATUS ERROR [LINENO LOG_FD]
+# ----------------------------------------
+# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
+# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
+# script with STATUS, using 1 if that was 0.
+as_fn_error ()
+{
+  as_status=$1; test $as_status -eq 0 && as_status=1
+  if test "$4"; then
+    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
+  fi
+  $as_echo "$as_me: error: $2" >&2
+  as_fn_exit $as_status
+} # as_fn_error
+
+if expr a : '\(a\)' >/dev/null 2>&1 &&
+   test "X`expr 00001 : '.*\(...\)'`" = X001; then
+  as_expr=expr
+else
+  as_expr=false
+fi
+
+if (basename -- /) >/dev/null 2>&1 && test "X`basename -- / 2>&1`" = "X/"; then
+  as_basename=basename
+else
+  as_basename=false
+fi
+
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
+
+as_me=`$as_basename -- "$0" ||
+$as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
+	 X"$0" : 'X\(//\)$' \| \
+	 X"$0" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X/"$0" |
+    sed '/^.*\/\([^/][^/]*\)\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\/\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\/\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
+
+
+  as_lineno_1=$LINENO as_lineno_1a=$LINENO
+  as_lineno_2=$LINENO as_lineno_2a=$LINENO
+  eval 'test "x$as_lineno_1'$as_run'" != "x$as_lineno_2'$as_run'" &&
+  test "x`expr $as_lineno_1'$as_run' + 1`" = "x$as_lineno_2'$as_run'"' || {
+  # Blame Lee E. McMahon (1931-1989) for sed's syntax.  :-)
+  sed -n '
+    p
+    /[$]LINENO/=
+  ' <$as_myself |
+    sed '
+      s/[$]LINENO.*/&-/
+      t lineno
+      b
+      :lineno
+      N
+      :loop
+      s/[$]LINENO\([^'$as_cr_alnum'_].*\n\)\(.*\)/\2\1\2/
+      t loop
+      s/-\n.*//
+    ' >$as_me.lineno &&
+  chmod +x "$as_me.lineno" ||
+    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2; as_fn_exit 1; }
+
+  # Don't try to exec as it changes $[0], causing all sort of problems
+  # (the dirname of $[0] is not the place where we might find the
+  # original and so on.  Autoconf is especially sensitive to this).
+  . "./$as_me.lineno"
+  # Exit status is that of the last command.
+  exit
+}
+
+ECHO_C= ECHO_N= ECHO_T=
+case `echo -n x` in #(((((
+-n*)
+  case `echo 'xy\c'` in
+  *c*) ECHO_T='	';;	# ECHO_T is single tab character.
+  xy)  ECHO_C='\c';;
+  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
+       ECHO_T='	';;
+  esac;;
+*)
+  ECHO_N='-n';;
+esac
+
+rm -f conf$$ conf$$.exe conf$$.file
+if test -d conf$$.dir; then
+  rm -f conf$$.dir/conf$$.file
+else
+  rm -f conf$$.dir
+  mkdir conf$$.dir 2>/dev/null
+fi
+if (echo >conf$$.file) 2>/dev/null; then
+  if ln -s conf$$.file conf$$ 2>/dev/null; then
+    as_ln_s='ln -s'
+    # ... but there are two gotchas:
+    # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
+    # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
+    # In both cases, we have to default to `cp -p'.
+    ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
+      as_ln_s='cp -p'
+  elif ln conf$$.file conf$$ 2>/dev/null; then
+    as_ln_s=ln
+  else
+    as_ln_s='cp -p'
+  fi
+else
+  as_ln_s='cp -p'
+fi
+rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
+rmdir conf$$.dir 2>/dev/null
+
+if mkdir -p . 2>/dev/null; then
+  as_mkdir_p='mkdir -p "$as_dir"'
+else
+  test -d ./-p && rmdir ./-p
+  as_mkdir_p=false
+fi
+
+if test -x / >/dev/null 2>&1; then
+  as_test_x='test -x'
+else
+  if ls -dL / >/dev/null 2>&1; then
+    as_ls_L_option=L
+  else
+    as_ls_L_option=
+  fi
+  as_test_x='
+    eval sh -c '\''
+      if test -d "$1"; then
+	test -d "$1/.";
+      else
+	case $1 in #(
+	-*)set "./$1";;
+	esac;
+	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in #((
+	???[sx]*):;;*)false;;esac;fi
+    '\'' sh
+  '
+fi
+as_executable_p=$as_test_x
+
+# Sed expression to map a string onto a valid CPP name.
+as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
+
+# Sed expression to map a string onto a valid variable name.
+as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
+
+
+test -n "$DJDIR" || exec 7<&0 </dev/null
+exec 6>&1
+
+# Name of the host.
+# hostname on some systems (SVR3.2, old GNU/Linux) returns a bogus exit status,
+# so uname gets run too.
+ac_hostname=`(hostname || uname -n) 2>/dev/null | sed 1q`
+
+#
+# Initializations.
+#
+ac_default_prefix=/usr/local
+ac_clean_files=
+ac_config_libobj_dir=.
+LIBOBJS=
+cross_compiling=no
+subdirs=
+MFLAGS=
+MAKEFLAGS=
+
+# Identity of this package.
+PACKAGE_NAME=
+PACKAGE_TARNAME=
+PACKAGE_VERSION=
+PACKAGE_STRING=
+PACKAGE_BUGREPORT=
+PACKAGE_URL=
+
+ac_unique_file="Makefile.in"
+# Factoring default headers for most tests.
+ac_includes_default="\
+#include <stdio.h>
+#ifdef HAVE_SYS_TYPES_H
+# include <sys/types.h>
+#endif
+#ifdef HAVE_SYS_STAT_H
+# include <sys/stat.h>
+#endif
+#ifdef STDC_HEADERS
+# include <stdlib.h>
+# include <stddef.h>
+#else
+# ifdef HAVE_STDLIB_H
+#  include <stdlib.h>
+# endif
+#endif
+#ifdef HAVE_STRING_H
+# if !defined STDC_HEADERS && defined HAVE_MEMORY_H
+#  include <memory.h>
+# endif
+# include <string.h>
+#endif
+#ifdef HAVE_STRINGS_H
+# include <strings.h>
+#endif
+#ifdef HAVE_INTTYPES_H
+# include <inttypes.h>
+#endif
+#ifdef HAVE_STDINT_H
+# include <stdint.h>
+#endif
+#ifdef HAVE_UNISTD_H
+# include <unistd.h>
+#endif"
+
+ac_subst_vars='LTLIBOBJS
+LIBOBJS
+cfgoutputs_out
+cfgoutputs_in
+cfghdrs_out
+cfghdrs_in
+enable_zone_allocator
+enable_tls
+enable_lazy_lock
+jemalloc_version_gid
+jemalloc_version_nrev
+jemalloc_version_bugfix
+jemalloc_version_minor
+jemalloc_version_major
+jemalloc_version
+enable_xmalloc
+enable_valgrind
+enable_utrace
+enable_fill
+enable_dss
+enable_munmap
+enable_mremap
+enable_tcache
+enable_prof
+enable_stats
+enable_debug
+install_suffix
+enable_experimental
+AUTOCONF
+LD
+AR
+RANLIB
+INSTALL_DATA
+INSTALL_SCRIPT
+INSTALL_PROGRAM
+enable_autogen
+RPATH_EXTRA
+CC_MM
+MKLIB
+LDTARGET
+CTARGET
+PIC_CFLAGS
+SOREV
+EXTRA_LDFLAGS
+DSO_LDFLAGS
+libprefix
+exe
+a
+o
+importlib
+so
+LD_PRELOAD_VAR
+RPATH
+abi
+host_os
+host_vendor
+host_cpu
+host
+build_os
+build_vendor
+build_cpu
+build
+EGREP
+GREP
+CPP
+OBJEXT
+EXEEXT
+ac_ct_CC
+CPPFLAGS
+LDFLAGS
+CFLAGS
+CC
+XSLROOT
+XSLTPROC
+MANDIR
+DATADIR
+LIBDIR
+INCLUDEDIR
+BINDIR
+PREFIX
+abs_objroot
+objroot
+abs_srcroot
+srcroot
+rev
+target_alias
+host_alias
+build_alias
+LIBS
+ECHO_T
+ECHO_N
+ECHO_C
+DEFS
+mandir
+localedir
+libdir
+psdir
+pdfdir
+dvidir
+htmldir
+infodir
+docdir
+oldincludedir
+includedir
+localstatedir
+sharedstatedir
+sysconfdir
+datadir
+datarootdir
+libexecdir
+sbindir
+bindir
+program_transform_name
+prefix
+exec_prefix
+PACKAGE_URL
+PACKAGE_BUGREPORT
+PACKAGE_STRING
+PACKAGE_VERSION
+PACKAGE_TARNAME
+PACKAGE_NAME
+PATH_SEPARATOR
+SHELL'
+ac_subst_files=''
+ac_user_opts='
+enable_option_checking
+with_xslroot
+with_rpath
+enable_autogen
+enable_experimental
+with_mangling
+with_jemalloc_prefix
+with_export
+with_private_namespace
+with_install_suffix
+enable_cc_silence
+enable_debug
+enable_ivsalloc
+enable_stats
+enable_prof
+enable_prof_libunwind
+with_static_libunwind
+enable_prof_libgcc
+enable_prof_gcc
+enable_tcache
+enable_mremap
+enable_munmap
+enable_dss
+enable_fill
+enable_utrace
+enable_valgrind
+enable_xmalloc
+enable_lazy_lock
+enable_tls
+enable_zone_allocator
+'
+      ac_precious_vars='build_alias
+host_alias
+target_alias
+CC
+CFLAGS
+LDFLAGS
+LIBS
+CPPFLAGS
+CPP'
+
+
+# Initialize some variables set by options.
+ac_init_help=
+ac_init_version=false
+ac_unrecognized_opts=
+ac_unrecognized_sep=
+# The variables have the same names as the options, with
+# dashes changed to underlines.
+cache_file=/dev/null
+exec_prefix=NONE
+no_create=
+no_recursion=
+prefix=NONE
+program_prefix=NONE
+program_suffix=NONE
+program_transform_name=s,x,x,
+silent=
+site=
+srcdir=
+verbose=
+x_includes=NONE
+x_libraries=NONE
+
+# Installation directory options.
+# These are left unexpanded so users can "make install exec_prefix=/foo"
+# and all the variables that are supposed to be based on exec_prefix
+# by default will actually change.
+# Use braces instead of parens because sh, perl, etc. also accept them.
+# (The list follows the same order as the GNU Coding Standards.)
+bindir='${exec_prefix}/bin'
+sbindir='${exec_prefix}/sbin'
+libexecdir='${exec_prefix}/libexec'
+datarootdir='${prefix}/share'
+datadir='${datarootdir}'
+sysconfdir='${prefix}/etc'
+sharedstatedir='${prefix}/com'
+localstatedir='${prefix}/var'
+includedir='${prefix}/include'
+oldincludedir='/usr/include'
+docdir='${datarootdir}/doc/${PACKAGE}'
+infodir='${datarootdir}/info'
+htmldir='${docdir}'
+dvidir='${docdir}'
+pdfdir='${docdir}'
+psdir='${docdir}'
+libdir='${exec_prefix}/lib'
+localedir='${datarootdir}/locale'
+mandir='${datarootdir}/man'
+
+ac_prev=
+ac_dashdash=
+for ac_option
+do
+  # If the previous option needs an argument, assign it.
+  if test -n "$ac_prev"; then
+    eval $ac_prev=\$ac_option
+    ac_prev=
+    continue
+  fi
+
+  case $ac_option in
+  *=?*) ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
+  *=)   ac_optarg= ;;
+  *)    ac_optarg=yes ;;
+  esac
+
+  # Accept the important Cygnus configure options, so we can diagnose typos.
+
+  case $ac_dashdash$ac_option in
+  --)
+    ac_dashdash=yes ;;
+
+  -bindir | --bindir | --bindi | --bind | --bin | --bi)
+    ac_prev=bindir ;;
+  -bindir=* | --bindir=* | --bindi=* | --bind=* | --bin=* | --bi=*)
+    bindir=$ac_optarg ;;
+
+  -build | --build | --buil | --bui | --bu)
+    ac_prev=build_alias ;;
+  -build=* | --build=* | --buil=* | --bui=* | --bu=*)
+    build_alias=$ac_optarg ;;
+
+  -cache-file | --cache-file | --cache-fil | --cache-fi \
+  | --cache-f | --cache- | --cache | --cach | --cac | --ca | --c)
+    ac_prev=cache_file ;;
+  -cache-file=* | --cache-file=* | --cache-fil=* | --cache-fi=* \
+  | --cache-f=* | --cache-=* | --cache=* | --cach=* | --cac=* | --ca=* | --c=*)
+    cache_file=$ac_optarg ;;
+
+  --config-cache | -C)
+    cache_file=config.cache ;;
+
+  -datadir | --datadir | --datadi | --datad)
+    ac_prev=datadir ;;
+  -datadir=* | --datadir=* | --datadi=* | --datad=*)
+    datadir=$ac_optarg ;;
+
+  -datarootdir | --datarootdir | --datarootdi | --datarootd | --dataroot \
+  | --dataroo | --dataro | --datar)
+    ac_prev=datarootdir ;;
+  -datarootdir=* | --datarootdir=* | --datarootdi=* | --datarootd=* \
+  | --dataroot=* | --dataroo=* | --dataro=* | --datar=*)
+    datarootdir=$ac_optarg ;;
+
+  -disable-* | --disable-*)
+    ac_useropt=`expr "x$ac_option" : 'x-*disable-\(.*\)'`
+    # Reject names that are not valid shell variable names.
+    expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
+      as_fn_error $? "invalid feature name: $ac_useropt"
+    ac_useropt_orig=$ac_useropt
+    ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
+    case $ac_user_opts in
+      *"
+"enable_$ac_useropt"
+"*) ;;
+      *) ac_unrecognized_opts="$ac_unrecognized_opts$ac_unrecognized_sep--disable-$ac_useropt_orig"
+	 ac_unrecognized_sep=', ';;
+    esac
+    eval enable_$ac_useropt=no ;;
+
+  -docdir | --docdir | --docdi | --doc | --do)
+    ac_prev=docdir ;;
+  -docdir=* | --docdir=* | --docdi=* | --doc=* | --do=*)
+    docdir=$ac_optarg ;;
+
+  -dvidir | --dvidir | --dvidi | --dvid | --dvi | --dv)
+    ac_prev=dvidir ;;
+  -dvidir=* | --dvidir=* | --dvidi=* | --dvid=* | --dvi=* | --dv=*)
+    dvidir=$ac_optarg ;;
+
+  -enable-* | --enable-*)
+    ac_useropt=`expr "x$ac_option" : 'x-*enable-\([^=]*\)'`
+    # Reject names that are not valid shell variable names.
+    expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
+      as_fn_error $? "invalid feature name: $ac_useropt"
+    ac_useropt_orig=$ac_useropt
+    ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
+    case $ac_user_opts in
+      *"
+"enable_$ac_useropt"
+"*) ;;
+      *) ac_unrecognized_opts="$ac_unrecognized_opts$ac_unrecognized_sep--enable-$ac_useropt_orig"
+	 ac_unrecognized_sep=', ';;
+    esac
+    eval enable_$ac_useropt=\$ac_optarg ;;
+
+  -exec-prefix | --exec_prefix | --exec-prefix | --exec-prefi \
+  | --exec-pref | --exec-pre | --exec-pr | --exec-p | --exec- \
+  | --exec | --exe | --ex)
+    ac_prev=exec_prefix ;;
+  -exec-prefix=* | --exec_prefix=* | --exec-prefix=* | --exec-prefi=* \
+  | --exec-pref=* | --exec-pre=* | --exec-pr=* | --exec-p=* | --exec-=* \
+  | --exec=* | --exe=* | --ex=*)
+    exec_prefix=$ac_optarg ;;
+
+  -gas | --gas | --ga | --g)
+    # Obsolete; use --with-gas.
+    with_gas=yes ;;
+
+  -help | --help | --hel | --he | -h)
+    ac_init_help=long ;;
+  -help=r* | --help=r* | --hel=r* | --he=r* | -hr*)
+    ac_init_help=recursive ;;
+  -help=s* | --help=s* | --hel=s* | --he=s* | -hs*)
+    ac_init_help=short ;;
+
+  -host | --host | --hos | --ho)
+    ac_prev=host_alias ;;
+  -host=* | --host=* | --hos=* | --ho=*)
+    host_alias=$ac_optarg ;;
+
+  -htmldir | --htmldir | --htmldi | --htmld | --html | --htm | --ht)
+    ac_prev=htmldir ;;
+  -htmldir=* | --htmldir=* | --htmldi=* | --htmld=* | --html=* | --htm=* \
+  | --ht=*)
+    htmldir=$ac_optarg ;;
+
+  -includedir | --includedir | --includedi | --included | --include \
+  | --includ | --inclu | --incl | --inc)
+    ac_prev=includedir ;;
+  -includedir=* | --includedir=* | --includedi=* | --included=* | --include=* \
+  | --includ=* | --inclu=* | --incl=* | --inc=*)
+    includedir=$ac_optarg ;;
+
+  -infodir | --infodir | --infodi | --infod | --info | --inf)
+    ac_prev=infodir ;;
+  -infodir=* | --infodir=* | --infodi=* | --infod=* | --info=* | --inf=*)
+    infodir=$ac_optarg ;;
+
+  -libdir | --libdir | --libdi | --libd)
+    ac_prev=libdir ;;
+  -libdir=* | --libdir=* | --libdi=* | --libd=*)
+    libdir=$ac_optarg ;;
+
+  -libexecdir | --libexecdir | --libexecdi | --libexecd | --libexec \
+  | --libexe | --libex | --libe)
+    ac_prev=libexecdir ;;
+  -libexecdir=* | --libexecdir=* | --libexecdi=* | --libexecd=* | --libexec=* \
+  | --libexe=* | --libex=* | --libe=*)
+    libexecdir=$ac_optarg ;;
+
+  -localedir | --localedir | --localedi | --localed | --locale)
+    ac_prev=localedir ;;
+  -localedir=* | --localedir=* | --localedi=* | --localed=* | --locale=*)
+    localedir=$ac_optarg ;;
+
+  -localstatedir | --localstatedir | --localstatedi | --localstated \
+  | --localstate | --localstat | --localsta | --localst | --locals)
+    ac_prev=localstatedir ;;
+  -localstatedir=* | --localstatedir=* | --localstatedi=* | --localstated=* \
+  | --localstate=* | --localstat=* | --localsta=* | --localst=* | --locals=*)
+    localstatedir=$ac_optarg ;;
+
+  -mandir | --mandir | --mandi | --mand | --man | --ma | --m)
+    ac_prev=mandir ;;
+  -mandir=* | --mandir=* | --mandi=* | --mand=* | --man=* | --ma=* | --m=*)
+    mandir=$ac_optarg ;;
+
+  -nfp | --nfp | --nf)
+    # Obsolete; use --without-fp.
+    with_fp=no ;;
+
+  -no-create | --no-create | --no-creat | --no-crea | --no-cre \
+  | --no-cr | --no-c | -n)
+    no_create=yes ;;
+
+  -no-recursion | --no-recursion | --no-recursio | --no-recursi \
+  | --no-recurs | --no-recur | --no-recu | --no-rec | --no-re | --no-r)
+    no_recursion=yes ;;
+
+  -oldincludedir | --oldincludedir | --oldincludedi | --oldincluded \
+  | --oldinclude | --oldinclud | --oldinclu | --oldincl | --oldinc \
+  | --oldin | --oldi | --old | --ol | --o)
+    ac_prev=oldincludedir ;;
+  -oldincludedir=* | --oldincludedir=* | --oldincludedi=* | --oldincluded=* \
+  | --oldinclude=* | --oldinclud=* | --oldinclu=* | --oldincl=* | --oldinc=* \
+  | --oldin=* | --oldi=* | --old=* | --ol=* | --o=*)
+    oldincludedir=$ac_optarg ;;
+
+  -prefix | --prefix | --prefi | --pref | --pre | --pr | --p)
+    ac_prev=prefix ;;
+  -prefix=* | --prefix=* | --prefi=* | --pref=* | --pre=* | --pr=* | --p=*)
+    prefix=$ac_optarg ;;
+
+  -program-prefix | --program-prefix | --program-prefi | --program-pref \
+  | --program-pre | --program-pr | --program-p)
+    ac_prev=program_prefix ;;
+  -program-prefix=* | --program-prefix=* | --program-prefi=* \
+  | --program-pref=* | --program-pre=* | --program-pr=* | --program-p=*)
+    program_prefix=$ac_optarg ;;
+
+  -program-suffix | --program-suffix | --program-suffi | --program-suff \
+  | --program-suf | --program-su | --program-s)
+    ac_prev=program_suffix ;;
+  -program-suffix=* | --program-suffix=* | --program-suffi=* \
+  | --program-suff=* | --program-suf=* | --program-su=* | --program-s=*)
+    program_suffix=$ac_optarg ;;
+
+  -program-transform-name | --program-transform-name \
+  | --program-transform-nam | --program-transform-na \
+  | --program-transform-n | --program-transform- \
+  | --program-transform | --program-transfor \
+  | --program-transfo | --program-transf \
+  | --program-trans | --program-tran \
+  | --progr-tra | --program-tr | --program-t)
+    ac_prev=program_transform_name ;;
+  -program-transform-name=* | --program-transform-name=* \
+  | --program-transform-nam=* | --program-transform-na=* \
+  | --program-transform-n=* | --program-transform-=* \
+  | --program-transform=* | --program-transfor=* \
+  | --program-transfo=* | --program-transf=* \
+  | --program-trans=* | --program-tran=* \
+  | --progr-tra=* | --program-tr=* | --program-t=*)
+    program_transform_name=$ac_optarg ;;
+
+  -pdfdir | --pdfdir | --pdfdi | --pdfd | --pdf | --pd)
+    ac_prev=pdfdir ;;
+  -pdfdir=* | --pdfdir=* | --pdfdi=* | --pdfd=* | --pdf=* | --pd=*)
+    pdfdir=$ac_optarg ;;
+
+  -psdir | --psdir | --psdi | --psd | --ps)
+    ac_prev=psdir ;;
+  -psdir=* | --psdir=* | --psdi=* | --psd=* | --ps=*)
+    psdir=$ac_optarg ;;
+
+  -q | -quiet | --quiet | --quie | --qui | --qu | --q \
+  | -silent | --silent | --silen | --sile | --sil)
+    silent=yes ;;
+
+  -sbindir | --sbindir | --sbindi | --sbind | --sbin | --sbi | --sb)
+    ac_prev=sbindir ;;
+  -sbindir=* | --sbindir=* | --sbindi=* | --sbind=* | --sbin=* \
+  | --sbi=* | --sb=*)
+    sbindir=$ac_optarg ;;
+
+  -sharedstatedir | --sharedstatedir | --sharedstatedi \
+  | --sharedstated | --sharedstate | --sharedstat | --sharedsta \
+  | --sharedst | --shareds | --shared | --share | --shar \
+  | --sha | --sh)
+    ac_prev=sharedstatedir ;;
+  -sharedstatedir=* | --sharedstatedir=* | --sharedstatedi=* \
+  | --sharedstated=* | --sharedstate=* | --sharedstat=* | --sharedsta=* \
+  | --sharedst=* | --shareds=* | --shared=* | --share=* | --shar=* \
+  | --sha=* | --sh=*)
+    sharedstatedir=$ac_optarg ;;
+
+  -site | --site | --sit)
+    ac_prev=site ;;
+  -site=* | --site=* | --sit=*)
+    site=$ac_optarg ;;
+
+  -srcdir | --srcdir | --srcdi | --srcd | --src | --sr)
+    ac_prev=srcdir ;;
+  -srcdir=* | --srcdir=* | --srcdi=* | --srcd=* | --src=* | --sr=*)
+    srcdir=$ac_optarg ;;
+
+  -sysconfdir | --sysconfdir | --sysconfdi | --sysconfd | --sysconf \
+  | --syscon | --sysco | --sysc | --sys | --sy)
+    ac_prev=sysconfdir ;;
+  -sysconfdir=* | --sysconfdir=* | --sysconfdi=* | --sysconfd=* | --sysconf=* \
+  | --syscon=* | --sysco=* | --sysc=* | --sys=* | --sy=*)
+    sysconfdir=$ac_optarg ;;
+
+  -target | --target | --targe | --targ | --tar | --ta | --t)
+    ac_prev=target_alias ;;
+  -target=* | --target=* | --targe=* | --targ=* | --tar=* | --ta=* | --t=*)
+    target_alias=$ac_optarg ;;
+
+  -v | -verbose | --verbose | --verbos | --verbo | --verb)
+    verbose=yes ;;
+
+  -version | --version | --versio | --versi | --vers | -V)
+    ac_init_version=: ;;
+
+  -with-* | --with-*)
+    ac_useropt=`expr "x$ac_option" : 'x-*with-\([^=]*\)'`
+    # Reject names that are not valid shell variable names.
+    expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
+      as_fn_error $? "invalid package name: $ac_useropt"
+    ac_useropt_orig=$ac_useropt
+    ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
+    case $ac_user_opts in
+      *"
+"with_$ac_useropt"
+"*) ;;
+      *) ac_unrecognized_opts="$ac_unrecognized_opts$ac_unrecognized_sep--with-$ac_useropt_orig"
+	 ac_unrecognized_sep=', ';;
+    esac
+    eval with_$ac_useropt=\$ac_optarg ;;
+
+  -without-* | --without-*)
+    ac_useropt=`expr "x$ac_option" : 'x-*without-\(.*\)'`
+    # Reject names that are not valid shell variable names.
+    expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
+      as_fn_error $? "invalid package name: $ac_useropt"
+    ac_useropt_orig=$ac_useropt
+    ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
+    case $ac_user_opts in
+      *"
+"with_$ac_useropt"
+"*) ;;
+      *) ac_unrecognized_opts="$ac_unrecognized_opts$ac_unrecognized_sep--without-$ac_useropt_orig"
+	 ac_unrecognized_sep=', ';;
+    esac
+    eval with_$ac_useropt=no ;;
+
+  --x)
+    # Obsolete; use --with-x.
+    with_x=yes ;;
+
+  -x-includes | --x-includes | --x-include | --x-includ | --x-inclu \
+  | --x-incl | --x-inc | --x-in | --x-i)
+    ac_prev=x_includes ;;
+  -x-includes=* | --x-includes=* | --x-include=* | --x-includ=* | --x-inclu=* \
+  | --x-incl=* | --x-inc=* | --x-in=* | --x-i=*)
+    x_includes=$ac_optarg ;;
+
+  -x-libraries | --x-libraries | --x-librarie | --x-librari \
+  | --x-librar | --x-libra | --x-libr | --x-lib | --x-li | --x-l)
+    ac_prev=x_libraries ;;
+  -x-libraries=* | --x-libraries=* | --x-librarie=* | --x-librari=* \
+  | --x-librar=* | --x-libra=* | --x-libr=* | --x-lib=* | --x-li=* | --x-l=*)
+    x_libraries=$ac_optarg ;;
+
+  -*) as_fn_error $? "unrecognized option: \`$ac_option'
+Try \`$0 --help' for more information"
+    ;;
+
+  *=*)
+    ac_envvar=`expr "x$ac_option" : 'x\([^=]*\)='`
+    # Reject names that are not valid shell variable names.
+    case $ac_envvar in #(
+      '' | [0-9]* | *[!_$as_cr_alnum]* )
+      as_fn_error $? "invalid variable name: \`$ac_envvar'" ;;
+    esac
+    eval $ac_envvar=\$ac_optarg
+    export $ac_envvar ;;
+
+  *)
+    # FIXME: should be removed in autoconf 3.0.
+    $as_echo "$as_me: WARNING: you should use --build, --host, --target" >&2
+    expr "x$ac_option" : ".*[^-._$as_cr_alnum]" >/dev/null &&
+      $as_echo "$as_me: WARNING: invalid host type: $ac_option" >&2
+    : "${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}"
+    ;;
+
+  esac
+done
+
+if test -n "$ac_prev"; then
+  ac_option=--`echo $ac_prev | sed 's/_/-/g'`
+  as_fn_error $? "missing argument to $ac_option"
+fi
+
+if test -n "$ac_unrecognized_opts"; then
+  case $enable_option_checking in
+    no) ;;
+    fatal) as_fn_error $? "unrecognized options: $ac_unrecognized_opts" ;;
+    *)     $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2 ;;
+  esac
+fi
+
+# Check all directory arguments for consistency.
+for ac_var in	exec_prefix prefix bindir sbindir libexecdir datarootdir \
+		datadir sysconfdir sharedstatedir localstatedir includedir \
+		oldincludedir docdir infodir htmldir dvidir pdfdir psdir \
+		libdir localedir mandir
+do
+  eval ac_val=\$$ac_var
+  # Remove trailing slashes.
+  case $ac_val in
+    */ )
+      ac_val=`expr "X$ac_val" : 'X\(.*[^/]\)' \| "X$ac_val" : 'X\(.*\)'`
+      eval $ac_var=\$ac_val;;
+  esac
+  # Be sure to have absolute directory names.
+  case $ac_val in
+    [\\/$]* | ?:[\\/]* )  continue;;
+    NONE | '' ) case $ac_var in *prefix ) continue;; esac;;
+  esac
+  as_fn_error $? "expected an absolute directory name for --$ac_var: $ac_val"
+done
+
+# There might be people who depend on the old broken behavior: `$host'
+# used to hold the argument of --host etc.
+# FIXME: To remove some day.
+build=$build_alias
+host=$host_alias
+target=$target_alias
+
+# FIXME: To remove some day.
+if test "x$host_alias" != x; then
+  if test "x$build_alias" = x; then
+    cross_compiling=maybe
+    $as_echo "$as_me: WARNING: if you wanted to set the --build type, don't use --host.
+    If a cross compiler is detected then cross compile mode will be used" >&2
+  elif test "x$build_alias" != "x$host_alias"; then
+    cross_compiling=yes
+  fi
+fi
+
+ac_tool_prefix=
+test -n "$host_alias" && ac_tool_prefix=$host_alias-
+
+test "$silent" = yes && exec 6>/dev/null
+
+
+ac_pwd=`pwd` && test -n "$ac_pwd" &&
+ac_ls_di=`ls -di .` &&
+ac_pwd_ls_di=`cd "$ac_pwd" && ls -di .` ||
+  as_fn_error $? "working directory cannot be determined"
+test "X$ac_ls_di" = "X$ac_pwd_ls_di" ||
+  as_fn_error $? "pwd does not report name of working directory"
+
+
+# Find the source files, if location was not specified.
+if test -z "$srcdir"; then
+  ac_srcdir_defaulted=yes
+  # Try the directory containing this script, then the parent directory.
+  ac_confdir=`$as_dirname -- "$as_myself" ||
+$as_expr X"$as_myself" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_myself" : 'X\(//\)[^/]' \| \
+	 X"$as_myself" : 'X\(//\)$' \| \
+	 X"$as_myself" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_myself" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+  srcdir=$ac_confdir
+  if test ! -r "$srcdir/$ac_unique_file"; then
+    srcdir=..
+  fi
+else
+  ac_srcdir_defaulted=no
+fi
+if test ! -r "$srcdir/$ac_unique_file"; then
+  test "$ac_srcdir_defaulted" = yes && srcdir="$ac_confdir or .."
+  as_fn_error $? "cannot find sources ($ac_unique_file) in $srcdir"
+fi
+ac_msg="sources are in $srcdir, but \`cd $srcdir' does not work"
+ac_abs_confdir=`(
+	cd "$srcdir" && test -r "./$ac_unique_file" || as_fn_error $? "$ac_msg"
+	pwd)`
+# When building in place, set srcdir=.
+if test "$ac_abs_confdir" = "$ac_pwd"; then
+  srcdir=.
+fi
+# Remove unnecessary trailing slashes from srcdir.
+# Double slashes in file names in object file debugging info
+# mess up M-x gdb in Emacs.
+case $srcdir in
+*/) srcdir=`expr "X$srcdir" : 'X\(.*[^/]\)' \| "X$srcdir" : 'X\(.*\)'`;;
+esac
+for ac_var in $ac_precious_vars; do
+  eval ac_env_${ac_var}_set=\${${ac_var}+set}
+  eval ac_env_${ac_var}_value=\$${ac_var}
+  eval ac_cv_env_${ac_var}_set=\${${ac_var}+set}
+  eval ac_cv_env_${ac_var}_value=\$${ac_var}
+done
+
+#
+# Report the --help message.
+#
+if test "$ac_init_help" = "long"; then
+  # Omit some internal or obsolete options to make the list less imposing.
+  # This message is too long to be a string in the A/UX 3.1 sh.
+  cat <<_ACEOF
+\`configure' configures this package to adapt to many kinds of systems.
+
+Usage: $0 [OPTION]... [VAR=VALUE]...
+
+To assign environment variables (e.g., CC, CFLAGS...), specify them as
+VAR=VALUE.  See below for descriptions of some of the useful variables.
+
+Defaults for the options are specified in brackets.
+
+Configuration:
+  -h, --help              display this help and exit
+      --help=short        display options specific to this package
+      --help=recursive    display the short help of all the included packages
+  -V, --version           display version information and exit
+  -q, --quiet, --silent   do not print \`checking ...' messages
+      --cache-file=FILE   cache test results in FILE [disabled]
+  -C, --config-cache      alias for \`--cache-file=config.cache'
+  -n, --no-create         do not create output files
+      --srcdir=DIR        find the sources in DIR [configure dir or \`..']
+
+Installation directories:
+  --prefix=PREFIX         install architecture-independent files in PREFIX
+                          [$ac_default_prefix]
+  --exec-prefix=EPREFIX   install architecture-dependent files in EPREFIX
+                          [PREFIX]
+
+By default, \`make install' will install all the files in
+\`$ac_default_prefix/bin', \`$ac_default_prefix/lib' etc.  You can specify
+an installation prefix other than \`$ac_default_prefix' using \`--prefix',
+for instance \`--prefix=\$HOME'.
+
+For better control, use the options below.
+
+Fine tuning of the installation directories:
+  --bindir=DIR            user executables [EPREFIX/bin]
+  --sbindir=DIR           system admin executables [EPREFIX/sbin]
+  --libexecdir=DIR        program executables [EPREFIX/libexec]
+  --sysconfdir=DIR        read-only single-machine data [PREFIX/etc]
+  --sharedstatedir=DIR    modifiable architecture-independent data [PREFIX/com]
+  --localstatedir=DIR     modifiable single-machine data [PREFIX/var]
+  --libdir=DIR            object code libraries [EPREFIX/lib]
+  --includedir=DIR        C header files [PREFIX/include]
+  --oldincludedir=DIR     C header files for non-gcc [/usr/include]
+  --datarootdir=DIR       read-only arch.-independent data root [PREFIX/share]
+  --datadir=DIR           read-only architecture-independent data [DATAROOTDIR]
+  --infodir=DIR           info documentation [DATAROOTDIR/info]
+  --localedir=DIR         locale-dependent data [DATAROOTDIR/locale]
+  --mandir=DIR            man documentation [DATAROOTDIR/man]
+  --docdir=DIR            documentation root [DATAROOTDIR/doc/PACKAGE]
+  --htmldir=DIR           html documentation [DOCDIR]
+  --dvidir=DIR            dvi documentation [DOCDIR]
+  --pdfdir=DIR            pdf documentation [DOCDIR]
+  --psdir=DIR             ps documentation [DOCDIR]
+_ACEOF
+
+  cat <<\_ACEOF
+
+System types:
+  --build=BUILD     configure for building on BUILD [guessed]
+  --host=HOST       cross-compile to build programs to run on HOST [BUILD]
+_ACEOF
+fi
+
+if test -n "$ac_init_help"; then
+
+  cat <<\_ACEOF
+
+Optional Features:
+  --disable-option-checking  ignore unrecognized --enable/--with options
+  --disable-FEATURE       do not include FEATURE (same as --enable-FEATURE=no)
+  --enable-FEATURE[=ARG]  include FEATURE [ARG=yes]
+  --enable-autogen        Automatically regenerate configure output
+  --disable-experimental  Disable support for the experimental API
+  --enable-cc-silence     Silence irrelevant compiler warnings
+  --enable-debug          Build debugging code (implies --enable-ivsalloc)
+  --enable-ivsalloc       Validate pointers passed through the public API
+  --disable-stats         Disable statistics calculation/reporting
+  --enable-prof           Enable allocation profiling
+  --enable-prof-libunwind Use libunwind for backtracing
+  --disable-prof-libgcc   Do not use libgcc for backtracing
+  --disable-prof-gcc      Do not use gcc intrinsics for backtracing
+  --disable-tcache        Disable per thread caches
+  --enable-mremap         Enable mremap(2) for huge realloc()
+  --disable-munmap        Disable VM deallocation via munmap(2)
+  --enable-dss            Enable allocation from DSS
+  --disable-fill          Disable support for junk/zero filling, quarantine,
+                          and redzones
+  --enable-utrace         Enable utrace(2)-based tracing
+  --disable-valgrind      Disable support for Valgrind
+  --enable-xmalloc        Support xmalloc option
+  --enable-lazy-lock      Enable lazy locking (only lock when multi-threaded)
+  --disable-tls           Disable thread-local storage (__thread keyword)
+  --disable-zone-allocator
+                          Disable zone allocator for Darwin
+
+Optional Packages:
+  --with-PACKAGE[=ARG]    use PACKAGE [ARG=yes]
+  --without-PACKAGE       do not use PACKAGE (same as --with-PACKAGE=no)
+  --with-xslroot=<path>   XSL stylesheet root path
+  --with-rpath=<rpath>    Colon-separated rpath (ELF systems only)
+  --with-mangling=<map>   Mangle symbols in <map>
+  --with-jemalloc-prefix=<prefix>
+                          Prefix to prepend to all public APIs
+  --without-export        disable exporting jemalloc public APIs
+  --with-private-namespace=<prefix>
+                          Prefix to prepend to all library-private APIs
+  --with-install-suffix=<suffix>
+                          Suffix to append to all installed files
+  --with-static-libunwind=<libunwind.a>
+                          Path to static libunwind library; use rather than
+                          dynamically linking
+
+Some influential environment variables:
+  CC          C compiler command
+  CFLAGS      C compiler flags
+  LDFLAGS     linker flags, e.g. -L<lib dir> if you have libraries in a
+              nonstandard directory <lib dir>
+  LIBS        libraries to pass to the linker, e.g. -l<library>
+  CPPFLAGS    (Objective) C/C++ preprocessor flags, e.g. -I<include dir> if
+              you have headers in a nonstandard directory <include dir>
+  CPP         C preprocessor
+
+Use these variables to override the choices made by `configure' or to help
+it to find libraries and programs with nonstandard names/locations.
+
+Report bugs to the package provider.
+_ACEOF
+ac_status=$?
+fi
+
+if test "$ac_init_help" = "recursive"; then
+  # If there are subdirs, report their specific --help.
+  for ac_dir in : $ac_subdirs_all; do test "x$ac_dir" = x: && continue
+    test -d "$ac_dir" ||
+      { cd "$srcdir" && ac_pwd=`pwd` && srcdir=. && test -d "$ac_dir"; } ||
+      continue
+    ac_builddir=.
+
+case "$ac_dir" in
+.) ac_dir_suffix= ac_top_builddir_sub=. ac_top_build_prefix= ;;
+*)
+  ac_dir_suffix=/`$as_echo "$ac_dir" | sed 's|^\.[\\/]||'`
+  # A ".." for each directory in $ac_dir_suffix.
+  ac_top_builddir_sub=`$as_echo "$ac_dir_suffix" | sed 's|/[^\\/]*|/..|g;s|/||'`
+  case $ac_top_builddir_sub in
+  "") ac_top_builddir_sub=. ac_top_build_prefix= ;;
+  *)  ac_top_build_prefix=$ac_top_builddir_sub/ ;;
+  esac ;;
+esac
+ac_abs_top_builddir=$ac_pwd
+ac_abs_builddir=$ac_pwd$ac_dir_suffix
+# for backward compatibility:
+ac_top_builddir=$ac_top_build_prefix
+
+case $srcdir in
+  .)  # We are building in place.
+    ac_srcdir=.
+    ac_top_srcdir=$ac_top_builddir_sub
+    ac_abs_top_srcdir=$ac_pwd ;;
+  [\\/]* | ?:[\\/]* )  # Absolute name.
+    ac_srcdir=$srcdir$ac_dir_suffix;
+    ac_top_srcdir=$srcdir
+    ac_abs_top_srcdir=$srcdir ;;
+  *) # Relative name.
+    ac_srcdir=$ac_top_build_prefix$srcdir$ac_dir_suffix
+    ac_top_srcdir=$ac_top_build_prefix$srcdir
+    ac_abs_top_srcdir=$ac_pwd/$srcdir ;;
+esac
+ac_abs_srcdir=$ac_abs_top_srcdir$ac_dir_suffix
+
+    cd "$ac_dir" || { ac_status=$?; continue; }
+    # Check for guested configure.
+    if test -f "$ac_srcdir/configure.gnu"; then
+      echo &&
+      $SHELL "$ac_srcdir/configure.gnu" --help=recursive
+    elif test -f "$ac_srcdir/configure"; then
+      echo &&
+      $SHELL "$ac_srcdir/configure" --help=recursive
+    else
+      $as_echo "$as_me: WARNING: no configuration information is in $ac_dir" >&2
+    fi || ac_status=$?
+    cd "$ac_pwd" || { ac_status=$?; break; }
+  done
+fi
+
+test -n "$ac_init_help" && exit $ac_status
+if $ac_init_version; then
+  cat <<\_ACEOF
+configure
+generated by GNU Autoconf 2.68
+
+Copyright (C) 2010 Free Software Foundation, Inc.
+This configure script is free software; the Free Software Foundation
+gives unlimited permission to copy, distribute and modify it.
+_ACEOF
+  exit
+fi
+
+## ------------------------ ##
+## Autoconf initialization. ##
+## ------------------------ ##
+
+# ac_fn_c_try_compile LINENO
+# --------------------------
+# Try to compile conftest.$ac_ext, and return whether this succeeded.
+ac_fn_c_try_compile ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  rm -f conftest.$ac_objext
+  if { { ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compile") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_retval=1
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_compile
+
+# ac_fn_c_try_cpp LINENO
+# ----------------------
+# Try to preprocess conftest.$ac_ext, and return whether this succeeded.
+ac_fn_c_try_cpp ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if { { ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } > conftest.i && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+    ac_retval=1
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_cpp
+
+# ac_fn_c_try_run LINENO
+# ----------------------
+# Try to link conftest.$ac_ext, and return whether this succeeded. Assumes
+# that executables *can* be run.
+ac_fn_c_try_run ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && { ac_try='./conftest$ac_exeext'
+  { { case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: program exited with status $ac_status" >&5
+       $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+       ac_retval=$ac_status
+fi
+  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_run
+
+# ac_fn_c_compute_int LINENO EXPR VAR INCLUDES
+# --------------------------------------------
+# Tries to find the compile-time value of EXPR in a program that includes
+# INCLUDES, setting VAR accordingly. Returns whether the value could be
+# computed
+ac_fn_c_compute_int ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if test "$cross_compiling" = yes; then
+    # Depending upon the size, compute the lo and hi bounds.
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+static int test_array [1 - 2 * !(($2) >= 0)];
+test_array [0] = 0
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_lo=0 ac_mid=0
+  while :; do
+    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+static int test_array [1 - 2 * !(($2) <= $ac_mid)];
+test_array [0] = 0
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_hi=$ac_mid; break
+else
+  as_fn_arith $ac_mid + 1 && ac_lo=$as_val
+			if test $ac_lo -le $ac_mid; then
+			  ac_lo= ac_hi=
+			  break
+			fi
+			as_fn_arith 2 '*' $ac_mid + 1 && ac_mid=$as_val
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+  done
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+static int test_array [1 - 2 * !(($2) < 0)];
+test_array [0] = 0
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_hi=-1 ac_mid=-1
+  while :; do
+    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+static int test_array [1 - 2 * !(($2) >= $ac_mid)];
+test_array [0] = 0
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_lo=$ac_mid; break
+else
+  as_fn_arith '(' $ac_mid ')' - 1 && ac_hi=$as_val
+			if test $ac_mid -le $ac_hi; then
+			  ac_lo= ac_hi=
+			  break
+			fi
+			as_fn_arith 2 '*' $ac_mid && ac_mid=$as_val
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+  done
+else
+  ac_lo= ac_hi=
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+# Binary search between lo and hi bounds.
+while test "x$ac_lo" != "x$ac_hi"; do
+  as_fn_arith '(' $ac_hi - $ac_lo ')' / 2 + $ac_lo && ac_mid=$as_val
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+static int test_array [1 - 2 * !(($2) <= $ac_mid)];
+test_array [0] = 0
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_hi=$ac_mid
+else
+  as_fn_arith '(' $ac_mid ')' + 1 && ac_lo=$as_val
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+done
+case $ac_lo in #((
+?*) eval "$3=\$ac_lo"; ac_retval=0 ;;
+'') ac_retval=1 ;;
+esac
+  else
+    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+static long int longval () { return $2; }
+static unsigned long int ulongval () { return $2; }
+#include <stdio.h>
+#include <stdlib.h>
+int
+main ()
+{
+
+  FILE *f = fopen ("conftest.val", "w");
+  if (! f)
+    return 1;
+  if (($2) < 0)
+    {
+      long int i = longval ();
+      if (i != ($2))
+	return 1;
+      fprintf (f, "%ld", i);
+    }
+  else
+    {
+      unsigned long int i = ulongval ();
+      if (i != ($2))
+	return 1;
+      fprintf (f, "%lu", i);
+    }
+  /* Do not output a trailing newline, as this causes \r\n confusion
+     on some platforms.  */
+  return ferror (f) || fclose (f) != 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_run "$LINENO"; then :
+  echo >>conftest.val; read $3 <conftest.val; ac_retval=0
+else
+  ac_retval=1
+fi
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
+  conftest.$ac_objext conftest.beam conftest.$ac_ext
+rm -f conftest.val
+
+  fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_compute_int
+
+# ac_fn_c_check_header_compile LINENO HEADER VAR INCLUDES
+# -------------------------------------------------------
+# Tests whether HEADER exists and can be compiled using the include files in
+# INCLUDES, setting the cache variable VAR accordingly.
+ac_fn_c_check_header_compile ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+#include <$2>
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  eval "$3=yes"
+else
+  eval "$3=no"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_header_compile
+
+# ac_fn_c_try_link LINENO
+# -----------------------
+# Try to link conftest.$ac_ext, and return whether this succeeded.
+ac_fn_c_try_link ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  rm -f conftest.$ac_objext conftest$ac_exeext
+  if { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 $as_test_x conftest$ac_exeext
+       }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_retval=1
+fi
+  # Delete the IPA/IPO (Inter Procedural Analysis/Optimization) information
+  # created by the PGI compiler (conftest_ipa8_conftest.oo), as it would
+  # interfere with the next link command; also delete a directory that is
+  # left behind by Apple's compiler.  We do this before executing the actions.
+  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_link
+
+# ac_fn_c_check_func LINENO FUNC VAR
+# ----------------------------------
+# Tests whether FUNC exists, setting the cache variable VAR accordingly
+ac_fn_c_check_func ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+/* Define $2 to an innocuous variant, in case <limits.h> declares $2.
+   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
+#define $2 innocuous_$2
+
+/* System header to define __stub macros and hopefully few prototypes,
+    which can conflict with char $2 (); below.
+    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
+    <limits.h> exists even on freestanding compilers.  */
+
+#ifdef __STDC__
+# include <limits.h>
+#else
+# include <assert.h>
+#endif
+
+#undef $2
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char $2 ();
+/* The GNU C library defines this for functions which it implements
+    to always fail with ENOSYS.  Some functions are actually named
+    something starting with __ and the normal name is an alias.  */
+#if defined __stub_$2 || defined __stub___$2
+choke me
+#endif
+
+int
+main ()
+{
+return $2 ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  eval "$3=yes"
+else
+  eval "$3=no"
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_func
+
+# ac_fn_c_check_header_mongrel LINENO HEADER VAR INCLUDES
+# -------------------------------------------------------
+# Tests whether HEADER exists, giving a warning if it cannot be compiled using
+# the include files in INCLUDES and setting the cache variable VAR
+# accordingly.
+ac_fn_c_check_header_mongrel ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if eval \${$3+:} false; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 usability" >&5
+$as_echo_n "checking $2 usability... " >&6; }
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+#include <$2>
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_header_compiler=yes
+else
+  ac_header_compiler=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 presence" >&5
+$as_echo_n "checking $2 presence... " >&6; }
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <$2>
+_ACEOF
+if ac_fn_c_try_cpp "$LINENO"; then :
+  ac_header_preproc=yes
+else
+  ac_header_preproc=no
+fi
+rm -f conftest.err conftest.i conftest.$ac_ext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in #((
+  yes:no: )
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $2: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $2:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $2: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
+    ;;
+esac
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=\$ac_header_compiler"
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_header_mongrel
+
+# ac_fn_c_check_type LINENO TYPE VAR INCLUDES
+# -------------------------------------------
+# Tests whether TYPE exists after having included INCLUDES, setting cache
+# variable VAR accordingly.
+ac_fn_c_check_type ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=no"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+if (sizeof ($2))
+	 return 0;
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+if (sizeof (($2)))
+	    return 0;
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+
+else
+  eval "$3=yes"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_type
+cat >config.log <<_ACEOF
+This file contains any messages produced by compilers while
+running configure, to aid debugging if configure makes a mistake.
+
+It was created by $as_me, which was
+generated by GNU Autoconf 2.68.  Invocation command line was
+
+  $ $0 $@
+
+_ACEOF
+exec 5>>config.log
+{
+cat <<_ASUNAME
+## --------- ##
+## Platform. ##
+## --------- ##
+
+hostname = `(hostname || uname -n) 2>/dev/null | sed 1q`
+uname -m = `(uname -m) 2>/dev/null || echo unknown`
+uname -r = `(uname -r) 2>/dev/null || echo unknown`
+uname -s = `(uname -s) 2>/dev/null || echo unknown`
+uname -v = `(uname -v) 2>/dev/null || echo unknown`
+
+/usr/bin/uname -p = `(/usr/bin/uname -p) 2>/dev/null || echo unknown`
+/bin/uname -X     = `(/bin/uname -X) 2>/dev/null     || echo unknown`
+
+/bin/arch              = `(/bin/arch) 2>/dev/null              || echo unknown`
+/usr/bin/arch -k       = `(/usr/bin/arch -k) 2>/dev/null       || echo unknown`
+/usr/convex/getsysinfo = `(/usr/convex/getsysinfo) 2>/dev/null || echo unknown`
+/usr/bin/hostinfo      = `(/usr/bin/hostinfo) 2>/dev/null      || echo unknown`
+/bin/machine           = `(/bin/machine) 2>/dev/null           || echo unknown`
+/usr/bin/oslevel       = `(/usr/bin/oslevel) 2>/dev/null       || echo unknown`
+/bin/universe          = `(/bin/universe) 2>/dev/null          || echo unknown`
+
+_ASUNAME
+
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    $as_echo "PATH: $as_dir"
+  done
+IFS=$as_save_IFS
+
+} >&5
+
+cat >&5 <<_ACEOF
+
+
+## ----------- ##
+## Core tests. ##
+## ----------- ##
+
+_ACEOF
+
+
+# Keep a trace of the command line.
+# Strip out --no-create and --no-recursion so they do not pile up.
+# Strip out --silent because we don't want to record it for future runs.
+# Also quote any args containing shell meta-characters.
+# Make two passes to allow for proper duplicate-argument suppression.
+ac_configure_args=
+ac_configure_args0=
+ac_configure_args1=
+ac_must_keep_next=false
+for ac_pass in 1 2
+do
+  for ac_arg
+  do
+    case $ac_arg in
+    -no-create | --no-c* | -n | -no-recursion | --no-r*) continue ;;
+    -q | -quiet | --quiet | --quie | --qui | --qu | --q \
+    | -silent | --silent | --silen | --sile | --sil)
+      continue ;;
+    *\'*)
+      ac_arg=`$as_echo "$ac_arg" | sed "s/'/'\\\\\\\\''/g"` ;;
+    esac
+    case $ac_pass in
+    1) as_fn_append ac_configure_args0 " '$ac_arg'" ;;
+    2)
+      as_fn_append ac_configure_args1 " '$ac_arg'"
+      if test $ac_must_keep_next = true; then
+	ac_must_keep_next=false # Got value, back to normal.
+      else
+	case $ac_arg in
+	  *=* | --config-cache | -C | -disable-* | --disable-* \
+	  | -enable-* | --enable-* | -gas | --g* | -nfp | --nf* \
+	  | -q | -quiet | --q* | -silent | --sil* | -v | -verb* \
+	  | -with-* | --with-* | -without-* | --without-* | --x)
+	    case "$ac_configure_args0 " in
+	      "$ac_configure_args1"*" '$ac_arg' "* ) continue ;;
+	    esac
+	    ;;
+	  -* ) ac_must_keep_next=true ;;
+	esac
+      fi
+      as_fn_append ac_configure_args " '$ac_arg'"
+      ;;
+    esac
+  done
+done
+{ ac_configure_args0=; unset ac_configure_args0;}
+{ ac_configure_args1=; unset ac_configure_args1;}
+
+# When interrupted or exit'd, cleanup temporary files, and complete
+# config.log.  We remove comments because anyway the quotes in there
+# would cause problems or look ugly.
+# WARNING: Use '\'' to represent an apostrophe within the trap.
+# WARNING: Do not start the trap code with a newline, due to a FreeBSD 4.0 bug.
+trap 'exit_status=$?
+  # Save into config.log some information that might help in debugging.
+  {
+    echo
+
+    $as_echo "## ---------------- ##
+## Cache variables. ##
+## ---------------- ##"
+    echo
+    # The following way of writing the cache mishandles newlines in values,
+(
+  for ac_var in `(set) 2>&1 | sed -n '\''s/^\([a-zA-Z_][a-zA-Z0-9_]*\)=.*/\1/p'\''`; do
+    eval ac_val=\$$ac_var
+    case $ac_val in #(
+    *${as_nl}*)
+      case $ac_var in #(
+      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
+$as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
+      esac
+      case $ac_var in #(
+      _ | IFS | as_nl) ;; #(
+      BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
+      *) { eval $ac_var=; unset $ac_var;} ;;
+      esac ;;
+    esac
+  done
+  (set) 2>&1 |
+    case $as_nl`(ac_space='\'' '\''; set) 2>&1` in #(
+    *${as_nl}ac_space=\ *)
+      sed -n \
+	"s/'\''/'\''\\\\'\'''\''/g;
+	  s/^\\([_$as_cr_alnum]*_cv_[_$as_cr_alnum]*\\)=\\(.*\\)/\\1='\''\\2'\''/p"
+      ;; #(
+    *)
+      sed -n "/^[_$as_cr_alnum]*_cv_[_$as_cr_alnum]*=/p"
+      ;;
+    esac |
+    sort
+)
+    echo
+
+    $as_echo "## ----------------- ##
+## Output variables. ##
+## ----------------- ##"
+    echo
+    for ac_var in $ac_subst_vars
+    do
+      eval ac_val=\$$ac_var
+      case $ac_val in
+      *\'\''*) ac_val=`$as_echo "$ac_val" | sed "s/'\''/'\''\\\\\\\\'\'''\''/g"`;;
+      esac
+      $as_echo "$ac_var='\''$ac_val'\''"
+    done | sort
+    echo
+
+    if test -n "$ac_subst_files"; then
+      $as_echo "## ------------------- ##
+## File substitutions. ##
+## ------------------- ##"
+      echo
+      for ac_var in $ac_subst_files
+      do
+	eval ac_val=\$$ac_var
+	case $ac_val in
+	*\'\''*) ac_val=`$as_echo "$ac_val" | sed "s/'\''/'\''\\\\\\\\'\'''\''/g"`;;
+	esac
+	$as_echo "$ac_var='\''$ac_val'\''"
+      done | sort
+      echo
+    fi
+
+    if test -s confdefs.h; then
+      $as_echo "## ----------- ##
+## confdefs.h. ##
+## ----------- ##"
+      echo
+      cat confdefs.h
+      echo
+    fi
+    test "$ac_signal" != 0 &&
+      $as_echo "$as_me: caught signal $ac_signal"
+    $as_echo "$as_me: exit $exit_status"
+  } >&5
+  rm -f core *.core core.conftest.* &&
+    rm -f -r conftest* confdefs* conf$$* $ac_clean_files &&
+    exit $exit_status
+' 0
+for ac_signal in 1 2 13 15; do
+  trap 'ac_signal='$ac_signal'; as_fn_exit 1' $ac_signal
+done
+ac_signal=0
+
+# confdefs.h avoids OS command line length limits that DEFS can exceed.
+rm -f -r conftest* confdefs.h
+
+$as_echo "/* confdefs.h */" > confdefs.h
+
+# Predefined preprocessor variables.
+
+cat >>confdefs.h <<_ACEOF
+#define PACKAGE_NAME "$PACKAGE_NAME"
+_ACEOF
+
+cat >>confdefs.h <<_ACEOF
+#define PACKAGE_TARNAME "$PACKAGE_TARNAME"
+_ACEOF
+
+cat >>confdefs.h <<_ACEOF
+#define PACKAGE_VERSION "$PACKAGE_VERSION"
+_ACEOF
+
+cat >>confdefs.h <<_ACEOF
+#define PACKAGE_STRING "$PACKAGE_STRING"
+_ACEOF
+
+cat >>confdefs.h <<_ACEOF
+#define PACKAGE_BUGREPORT "$PACKAGE_BUGREPORT"
+_ACEOF
+
+cat >>confdefs.h <<_ACEOF
+#define PACKAGE_URL "$PACKAGE_URL"
+_ACEOF
+
+
+# Let the site file select an alternate cache file if it wants to.
+# Prefer an explicitly selected file to automatically selected ones.
+ac_site_file1=NONE
+ac_site_file2=NONE
+if test -n "$CONFIG_SITE"; then
+  # We do not want a PATH search for config.site.
+  case $CONFIG_SITE in #((
+    -*)  ac_site_file1=./$CONFIG_SITE;;
+    */*) ac_site_file1=$CONFIG_SITE;;
+    *)   ac_site_file1=./$CONFIG_SITE;;
+  esac
+elif test "x$prefix" != xNONE; then
+  ac_site_file1=$prefix/share/config.site
+  ac_site_file2=$prefix/etc/config.site
+else
+  ac_site_file1=$ac_default_prefix/share/config.site
+  ac_site_file2=$ac_default_prefix/etc/config.site
+fi
+for ac_site_file in "$ac_site_file1" "$ac_site_file2"
+do
+  test "x$ac_site_file" = xNONE && continue
+  if test /dev/null != "$ac_site_file" && test -r "$ac_site_file"; then
+    { $as_echo "$as_me:${as_lineno-$LINENO}: loading site script $ac_site_file" >&5
+$as_echo "$as_me: loading site script $ac_site_file" >&6;}
+    sed 's/^/| /' "$ac_site_file" >&5
+    . "$ac_site_file" \
+      || { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "failed to load site script $ac_site_file
+See \`config.log' for more details" "$LINENO" 5; }
+  fi
+done
+
+if test -r "$cache_file"; then
+  # Some versions of bash will fail to source /dev/null (special files
+  # actually), so we avoid doing that.  DJGPP emulates it as a regular file.
+  if test /dev/null != "$cache_file" && test -f "$cache_file"; then
+    { $as_echo "$as_me:${as_lineno-$LINENO}: loading cache $cache_file" >&5
+$as_echo "$as_me: loading cache $cache_file" >&6;}
+    case $cache_file in
+      [\\/]* | ?:[\\/]* ) . "$cache_file";;
+      *)                      . "./$cache_file";;
+    esac
+  fi
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: creating cache $cache_file" >&5
+$as_echo "$as_me: creating cache $cache_file" >&6;}
+  >$cache_file
+fi
+
+# Check that the precious variables saved in the cache have kept the same
+# value.
+ac_cache_corrupted=false
+for ac_var in $ac_precious_vars; do
+  eval ac_old_set=\$ac_cv_env_${ac_var}_set
+  eval ac_new_set=\$ac_env_${ac_var}_set
+  eval ac_old_val=\$ac_cv_env_${ac_var}_value
+  eval ac_new_val=\$ac_env_${ac_var}_value
+  case $ac_old_set,$ac_new_set in
+    set,)
+      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
+$as_echo "$as_me: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&2;}
+      ac_cache_corrupted=: ;;
+    ,set)
+      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was not set in the previous run" >&5
+$as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
+      ac_cache_corrupted=: ;;
+    ,);;
+    *)
+      if test "x$ac_old_val" != "x$ac_new_val"; then
+	# differences in whitespace do not lead to failure.
+	ac_old_val_w=`echo x $ac_old_val`
+	ac_new_val_w=`echo x $ac_new_val`
+	if test "$ac_old_val_w" != "$ac_new_val_w"; then
+	  { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' has changed since the previous run:" >&5
+$as_echo "$as_me: error: \`$ac_var' has changed since the previous run:" >&2;}
+	  ac_cache_corrupted=:
+	else
+	  { $as_echo "$as_me:${as_lineno-$LINENO}: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
+$as_echo "$as_me: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&2;}
+	  eval $ac_var=\$ac_old_val
+	fi
+	{ $as_echo "$as_me:${as_lineno-$LINENO}:   former value:  \`$ac_old_val'" >&5
+$as_echo "$as_me:   former value:  \`$ac_old_val'" >&2;}
+	{ $as_echo "$as_me:${as_lineno-$LINENO}:   current value: \`$ac_new_val'" >&5
+$as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
+      fi;;
+  esac
+  # Pass precious variables to config.status.
+  if test "$ac_new_set" = set; then
+    case $ac_new_val in
+    *\'*) ac_arg=$ac_var=`$as_echo "$ac_new_val" | sed "s/'/'\\\\\\\\''/g"` ;;
+    *) ac_arg=$ac_var=$ac_new_val ;;
+    esac
+    case " $ac_configure_args " in
+      *" '$ac_arg' "*) ;; # Avoid dups.  Use of quotes ensures accuracy.
+      *) as_fn_append ac_configure_args " '$ac_arg'" ;;
+    esac
+  fi
+done
+if $ac_cache_corrupted; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+  { $as_echo "$as_me:${as_lineno-$LINENO}: error: changes in the environment can compromise the build" >&5
+$as_echo "$as_me: error: changes in the environment can compromise the build" >&2;}
+  as_fn_error $? "run \`make distclean' and/or \`rm $cache_file' and start over" "$LINENO" 5
+fi
+## -------------------- ##
+## Main body of script. ##
+## -------------------- ##
+
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+
+
+
+
+
+
+
+
+
+rev=1
+
+
+srcroot=$srcdir
+if test "x${srcroot}" = "x." ; then
+  srcroot=""
+else
+  srcroot="${srcroot}/"
+fi
+
+abs_srcroot="`cd \"${srcdir}\"; pwd`/"
+
+
+objroot=""
+
+abs_objroot="`pwd`/"
+
+
+if test "x$prefix" = "xNONE" ; then
+  prefix="/usr/local"
+fi
+if test "x$exec_prefix" = "xNONE" ; then
+  exec_prefix=$prefix
+fi
+PREFIX=$prefix
+
+BINDIR=`eval echo $bindir`
+BINDIR=`eval echo $BINDIR`
+
+INCLUDEDIR=`eval echo $includedir`
+INCLUDEDIR=`eval echo $INCLUDEDIR`
+
+LIBDIR=`eval echo $libdir`
+LIBDIR=`eval echo $LIBDIR`
+
+DATADIR=`eval echo $datadir`
+DATADIR=`eval echo $DATADIR`
+
+MANDIR=`eval echo $mandir`
+MANDIR=`eval echo $MANDIR`
+
+
+# Extract the first word of "xsltproc", so it can be a program name with args.
+set dummy xsltproc; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_path_XSLTPROC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  case $XSLTPROC in
+  [\\/]* | ?:[\\/]*)
+  ac_cv_path_XSLTPROC="$XSLTPROC" # Let the user override the test with a path.
+  ;;
+  *)
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    ac_cv_path_XSLTPROC="$as_dir/$ac_word$ac_exec_ext"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+  test -z "$ac_cv_path_XSLTPROC" && ac_cv_path_XSLTPROC="false"
+  ;;
+esac
+fi
+XSLTPROC=$ac_cv_path_XSLTPROC
+if test -n "$XSLTPROC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $XSLTPROC" >&5
+$as_echo "$XSLTPROC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+if test -d "/usr/share/xml/docbook/stylesheet/docbook-xsl" ; then
+  DEFAULT_XSLROOT="/usr/share/xml/docbook/stylesheet/docbook-xsl"
+elif test -d "/usr/share/sgml/docbook/xsl-stylesheets" ; then
+  DEFAULT_XSLROOT="/usr/share/sgml/docbook/xsl-stylesheets"
+else
+    DEFAULT_XSLROOT=""
+fi
+
+# Check whether --with-xslroot was given.
+if test "${with_xslroot+set}" = set; then :
+  withval=$with_xslroot;
+if test "x$with_xslroot" = "xno" ; then
+  XSLROOT="${DEFAULT_XSLROOT}"
+else
+  XSLROOT="${with_xslroot}"
+fi
+
+else
+  XSLROOT="${DEFAULT_XSLROOT}"
+
+fi
+
+
+
+CFLAGS=$CFLAGS
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+if test -n "$ac_tool_prefix"; then
+  # Extract the first word of "${ac_tool_prefix}gcc", so it can be a program name with args.
+set dummy ${ac_tool_prefix}gcc; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$CC"; then
+  ac_cv_prog_CC="$CC" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    ac_cv_prog_CC="${ac_tool_prefix}gcc"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+CC=$ac_cv_prog_CC
+if test -n "$CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+$as_echo "$CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+fi
+if test -z "$ac_cv_prog_CC"; then
+  ac_ct_CC=$CC
+  # Extract the first word of "gcc", so it can be a program name with args.
+set dummy gcc; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_ac_ct_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$ac_ct_CC"; then
+  ac_cv_prog_ac_ct_CC="$ac_ct_CC" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    ac_cv_prog_ac_ct_CC="gcc"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+ac_ct_CC=$ac_cv_prog_ac_ct_CC
+if test -n "$ac_ct_CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
+$as_echo "$ac_ct_CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+  if test "x$ac_ct_CC" = x; then
+    CC=""
+  else
+    case $cross_compiling:$ac_tool_warned in
+yes:)
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+$as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
+ac_tool_warned=yes ;;
+esac
+    CC=$ac_ct_CC
+  fi
+else
+  CC="$ac_cv_prog_CC"
+fi
+
+if test -z "$CC"; then
+          if test -n "$ac_tool_prefix"; then
+    # Extract the first word of "${ac_tool_prefix}cc", so it can be a program name with args.
+set dummy ${ac_tool_prefix}cc; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$CC"; then
+  ac_cv_prog_CC="$CC" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    ac_cv_prog_CC="${ac_tool_prefix}cc"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+CC=$ac_cv_prog_CC
+if test -n "$CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+$as_echo "$CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+  fi
+fi
+if test -z "$CC"; then
+  # Extract the first word of "cc", so it can be a program name with args.
+set dummy cc; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$CC"; then
+  ac_cv_prog_CC="$CC" # Let the user override the test.
+else
+  ac_prog_rejected=no
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    if test "$as_dir/$ac_word$ac_exec_ext" = "/usr/ucb/cc"; then
+       ac_prog_rejected=yes
+       continue
+     fi
+    ac_cv_prog_CC="cc"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+if test $ac_prog_rejected = yes; then
+  # We found a bogon in the path, so make sure we never use it.
+  set dummy $ac_cv_prog_CC
+  shift
+  if test $# != 0; then
+    # We chose a different compiler from the bogus one.
+    # However, it has the same basename, so the bogon will be chosen
+    # first if we set CC to just the basename; use the full file name.
+    shift
+    ac_cv_prog_CC="$as_dir/$ac_word${1+' '}$@"
+  fi
+fi
+fi
+fi
+CC=$ac_cv_prog_CC
+if test -n "$CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+$as_echo "$CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+fi
+if test -z "$CC"; then
+  if test -n "$ac_tool_prefix"; then
+  for ac_prog in cl.exe
+  do
+    # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
+set dummy $ac_tool_prefix$ac_prog; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$CC"; then
+  ac_cv_prog_CC="$CC" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    ac_cv_prog_CC="$ac_tool_prefix$ac_prog"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+CC=$ac_cv_prog_CC
+if test -n "$CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+$as_echo "$CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+    test -n "$CC" && break
+  done
+fi
+if test -z "$CC"; then
+  ac_ct_CC=$CC
+  for ac_prog in cl.exe
+do
+  # Extract the first word of "$ac_prog", so it can be a program name with args.
+set dummy $ac_prog; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_ac_ct_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$ac_ct_CC"; then
+  ac_cv_prog_ac_ct_CC="$ac_ct_CC" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    ac_cv_prog_ac_ct_CC="$ac_prog"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+ac_ct_CC=$ac_cv_prog_ac_ct_CC
+if test -n "$ac_ct_CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
+$as_echo "$ac_ct_CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+  test -n "$ac_ct_CC" && break
+done
+
+  if test "x$ac_ct_CC" = x; then
+    CC=""
+  else
+    case $cross_compiling:$ac_tool_warned in
+yes:)
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+$as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
+ac_tool_warned=yes ;;
+esac
+    CC=$ac_ct_CC
+  fi
+fi
+
+fi
+
+
+test -z "$CC" && { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "no acceptable C compiler found in \$PATH
+See \`config.log' for more details" "$LINENO" 5; }
+
+# Provide some information about the compiler.
+$as_echo "$as_me:${as_lineno-$LINENO}: checking for C compiler version" >&5
+set X $ac_compile
+ac_compiler=$2
+for ac_option in --version -v -V -qversion; do
+  { { ac_try="$ac_compiler $ac_option >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    sed '10a\
+... rest of stderr output deleted ...
+         10q' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+  fi
+  rm -f conftest.er1 conftest.err
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }
+done
+
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+ac_clean_files_save=$ac_clean_files
+ac_clean_files="$ac_clean_files a.out a.out.dSYM a.exe b.out"
+# Try to create an executable without -o first, disregard a.out.
+# It will help us diagnose broken compilers, and finding out an intuition
+# of exeext.
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the C compiler works" >&5
+$as_echo_n "checking whether the C compiler works... " >&6; }
+ac_link_default=`$as_echo "$ac_link" | sed 's/ -o *conftest[^ ]*//'`
+
+# The possible output files:
+ac_files="a.out conftest.exe conftest a.exe a_out.exe b.out conftest.*"
+
+ac_rmfiles=
+for ac_file in $ac_files
+do
+  case $ac_file in
+    *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | *.map | *.inf | *.dSYM | *.o | *.obj ) ;;
+    * ) ac_rmfiles="$ac_rmfiles $ac_file";;
+  esac
+done
+rm -f $ac_rmfiles
+
+if { { ac_try="$ac_link_default"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link_default") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
+  # Autoconf-2.13 could set the ac_cv_exeext variable to `no'.
+# So ignore a value of `no', otherwise this would lead to `EXEEXT = no'
+# in a Makefile.  We should not override ac_cv_exeext if it was cached,
+# so that the user can short-circuit this test for compilers unknown to
+# Autoconf.
+for ac_file in $ac_files ''
+do
+  test -f "$ac_file" || continue
+  case $ac_file in
+    *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | *.map | *.inf | *.dSYM | *.o | *.obj )
+	;;
+    [ab].out )
+	# We found the default executable, but exeext='' is most
+	# certainly right.
+	break;;
+    *.* )
+	if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
+	then :; else
+	   ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
+	fi
+	# We set ac_cv_exeext here because the later test for it is not
+	# safe: cross compilers may not add the suffix if given an `-o'
+	# argument, so we may need to know it at that point already.
+	# Even if this section looks crufty: it has the advantage of
+	# actually working.
+	break;;
+    * )
+	break;;
+  esac
+done
+test "$ac_cv_exeext" = no && ac_cv_exeext=
+
+else
+  ac_file=''
+fi
+if test -z "$ac_file"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+$as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error 77 "C compiler cannot create executables
+See \`config.log' for more details" "$LINENO" 5; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for C compiler default output file name" >&5
+$as_echo_n "checking for C compiler default output file name... " >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_file" >&5
+$as_echo "$ac_file" >&6; }
+ac_exeext=$ac_cv_exeext
+
+rm -f -r a.out a.out.dSYM a.exe conftest$ac_cv_exeext b.out
+ac_clean_files=$ac_clean_files_save
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of executables" >&5
+$as_echo_n "checking for suffix of executables... " >&6; }
+if { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
+  # If both `conftest.exe' and `conftest' are `present' (well, observable)
+# catch `conftest.exe'.  For instance with Cygwin, `ls conftest' will
+# work properly (i.e., refer to `conftest.exe'), while it won't with
+# `rm'.
+for ac_file in conftest.exe conftest conftest.*; do
+  test -f "$ac_file" || continue
+  case $ac_file in
+    *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | *.map | *.inf | *.dSYM | *.o | *.obj ) ;;
+    *.* ) ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
+	  break;;
+    * ) break;;
+  esac
+done
+else
+  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "cannot compute suffix of executables: cannot compile and link
+See \`config.log' for more details" "$LINENO" 5; }
+fi
+rm -f conftest conftest$ac_cv_exeext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_exeext" >&5
+$as_echo "$ac_cv_exeext" >&6; }
+
+rm -f conftest.$ac_ext
+EXEEXT=$ac_cv_exeext
+ac_exeext=$EXEEXT
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdio.h>
+int
+main ()
+{
+FILE *f = fopen ("conftest.out", "w");
+ return ferror (f) || fclose (f) != 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+ac_clean_files="$ac_clean_files conftest.out"
+# Check that the compiler produces executables we can run.  If not, either
+# the compiler is broken, or we cross compile.
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are cross compiling" >&5
+$as_echo_n "checking whether we are cross compiling... " >&6; }
+if test "$cross_compiling" != yes; then
+  { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }
+  if { ac_try='./conftest$ac_cv_exeext'
+  { { case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; }; then
+    cross_compiling=no
+  else
+    if test "$cross_compiling" = maybe; then
+	cross_compiling=yes
+    else
+	{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "cannot run C compiled programs.
+If you meant to cross compile, use \`--host'.
+See \`config.log' for more details" "$LINENO" 5; }
+    fi
+  fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $cross_compiling" >&5
+$as_echo "$cross_compiling" >&6; }
+
+rm -f conftest.$ac_ext conftest$ac_cv_exeext conftest.out
+ac_clean_files=$ac_clean_files_save
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of object files" >&5
+$as_echo_n "checking for suffix of object files... " >&6; }
+if ${ac_cv_objext+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.o conftest.obj
+if { { ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compile") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
+  for ac_file in conftest.o conftest.obj conftest.*; do
+  test -f "$ac_file" || continue;
+  case $ac_file in
+    *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | *.map | *.inf | *.dSYM ) ;;
+    *) ac_cv_objext=`expr "$ac_file" : '.*\.\(.*\)'`
+       break;;
+  esac
+done
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "cannot compute suffix of object files: cannot compile
+See \`config.log' for more details" "$LINENO" 5; }
+fi
+rm -f conftest.$ac_cv_objext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_objext" >&5
+$as_echo "$ac_cv_objext" >&6; }
+OBJEXT=$ac_cv_objext
+ac_objext=$OBJEXT
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C compiler" >&5
+$as_echo_n "checking whether we are using the GNU C compiler... " >&6; }
+if ${ac_cv_c_compiler_gnu+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+#ifndef __GNUC__
+       choke me
+#endif
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_compiler_gnu=yes
+else
+  ac_compiler_gnu=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+ac_cv_c_compiler_gnu=$ac_compiler_gnu
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_c_compiler_gnu" >&5
+$as_echo "$ac_cv_c_compiler_gnu" >&6; }
+if test $ac_compiler_gnu = yes; then
+  GCC=yes
+else
+  GCC=
+fi
+ac_test_CFLAGS=${CFLAGS+set}
+ac_save_CFLAGS=$CFLAGS
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC accepts -g" >&5
+$as_echo_n "checking whether $CC accepts -g... " >&6; }
+if ${ac_cv_prog_cc_g+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_save_c_werror_flag=$ac_c_werror_flag
+   ac_c_werror_flag=yes
+   ac_cv_prog_cc_g=no
+   CFLAGS="-g"
+   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_cv_prog_cc_g=yes
+else
+  CFLAGS=""
+      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+
+else
+  ac_c_werror_flag=$ac_save_c_werror_flag
+	 CFLAGS="-g"
+	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_cv_prog_cc_g=yes
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+   ac_c_werror_flag=$ac_save_c_werror_flag
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_g" >&5
+$as_echo "$ac_cv_prog_cc_g" >&6; }
+if test "$ac_test_CFLAGS" = set; then
+  CFLAGS=$ac_save_CFLAGS
+elif test $ac_cv_prog_cc_g = yes; then
+  if test "$GCC" = yes; then
+    CFLAGS="-g -O2"
+  else
+    CFLAGS="-g"
+  fi
+else
+  if test "$GCC" = yes; then
+    CFLAGS="-O2"
+  else
+    CFLAGS=
+  fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to accept ISO C89" >&5
+$as_echo_n "checking for $CC option to accept ISO C89... " >&6; }
+if ${ac_cv_prog_cc_c89+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_cv_prog_cc_c89=no
+ac_save_CC=$CC
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdarg.h>
+#include <stdio.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+/* Most of the following tests are stolen from RCS 5.7's src/conf.sh.  */
+struct buf { int x; };
+FILE * (*rcsopen) (struct buf *, struct stat *, int);
+static char *e (p, i)
+     char **p;
+     int i;
+{
+  return p[i];
+}
+static char *f (char * (*g) (char **, int), char **p, ...)
+{
+  char *s;
+  va_list v;
+  va_start (v,p);
+  s = g (p, va_arg (v,int));
+  va_end (v);
+  return s;
+}
+
+/* OSF 4.0 Compaq cc is some sort of almost-ANSI by default.  It has
+   function prototypes and stuff, but not '\xHH' hex character constants.
+   These don't provoke an error unfortunately, instead are silently treated
+   as 'x'.  The following induces an error, until -std is added to get
+   proper ANSI mode.  Curiously '\x00'!='x' always comes out true, for an
+   array size at least.  It's necessary to write '\x00'==0 to get something
+   that's true only with -std.  */
+int osf4_cc_array ['\x00' == 0 ? 1 : -1];
+
+/* IBM C 6 for AIX is almost-ANSI by default, but it replaces macro parameters
+   inside strings and character constants.  */
+#define FOO(x) 'x'
+int xlc6_cc_array[FOO(a) == 'x' ? 1 : -1];
+
+int test (int i, double x);
+struct s1 {int (*f) (int a);};
+struct s2 {int (*f) (double a);};
+int pairnames (int, char **, FILE *(*)(struct buf *, struct stat *, int), int, int);
+int argc;
+char **argv;
+int
+main ()
+{
+return f (e, argv, 0) != argv[0]  ||  f (e, argv, 1) != argv[1];
+  ;
+  return 0;
+}
+_ACEOF
+for ac_arg in '' -qlanglvl=extc89 -qlanglvl=ansi -std \
+	-Ae "-Aa -D_HPUX_SOURCE" "-Xc -D__EXTENSIONS__"
+do
+  CC="$ac_save_CC $ac_arg"
+  if ac_fn_c_try_compile "$LINENO"; then :
+  ac_cv_prog_cc_c89=$ac_arg
+fi
+rm -f core conftest.err conftest.$ac_objext
+  test "x$ac_cv_prog_cc_c89" != "xno" && break
+done
+rm -f conftest.$ac_ext
+CC=$ac_save_CC
+
+fi
+# AC_CACHE_VAL
+case "x$ac_cv_prog_cc_c89" in
+  x)
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: none needed" >&5
+$as_echo "none needed" >&6; } ;;
+  xno)
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: unsupported" >&5
+$as_echo "unsupported" >&6; } ;;
+  *)
+    CC="$CC $ac_cv_prog_cc_c89"
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_c89" >&5
+$as_echo "$ac_cv_prog_cc_c89" >&6; } ;;
+esac
+if test "x$ac_cv_prog_cc_c89" != xno; then :
+
+fi
+
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+
+if test "x$GCC" != "xyes" ; then
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether compiler is MSVC" >&5
+$as_echo_n "checking whether compiler is MSVC... " >&6; }
+if ${je_cv_msvc+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+#ifndef _MSC_VER
+  int fail-1;
+#endif
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  je_cv_msvc=yes
+else
+  je_cv_msvc=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $je_cv_msvc" >&5
+$as_echo "$je_cv_msvc" >&6; }
+fi
+
+if test "x$CFLAGS" = "x" ; then
+  no_CFLAGS="yes"
+  if test "x$GCC" = "xyes" ; then
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether compiler supports -std=gnu99" >&5
+$as_echo_n "checking whether compiler supports -std=gnu99... " >&6; }
+TCFLAGS="${CFLAGS}"
+if test "x${CFLAGS}" = "x" ; then
+  CFLAGS="-std=gnu99"
+else
+  CFLAGS="${CFLAGS} -std=gnu99"
+fi
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+
+int
+main ()
+{
+
+    return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+              CFLAGS="${TCFLAGS}"
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether compiler supports -Wall" >&5
+$as_echo_n "checking whether compiler supports -Wall... " >&6; }
+TCFLAGS="${CFLAGS}"
+if test "x${CFLAGS}" = "x" ; then
+  CFLAGS="-Wall"
+else
+  CFLAGS="${CFLAGS} -Wall"
+fi
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+
+int
+main ()
+{
+
+    return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+              CFLAGS="${TCFLAGS}"
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether compiler supports -pipe" >&5
+$as_echo_n "checking whether compiler supports -pipe... " >&6; }
+TCFLAGS="${CFLAGS}"
+if test "x${CFLAGS}" = "x" ; then
+  CFLAGS="-pipe"
+else
+  CFLAGS="${CFLAGS} -pipe"
+fi
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+
+int
+main ()
+{
+
+    return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+              CFLAGS="${TCFLAGS}"
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether compiler supports -g3" >&5
+$as_echo_n "checking whether compiler supports -g3... " >&6; }
+TCFLAGS="${CFLAGS}"
+if test "x${CFLAGS}" = "x" ; then
+  CFLAGS="-g3"
+else
+  CFLAGS="${CFLAGS} -g3"
+fi
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+
+int
+main ()
+{
+
+    return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+              CFLAGS="${TCFLAGS}"
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+  elif test "x$je_cv_msvc" = "xyes" ; then
+    CC="$CC -nologo"
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether compiler supports -Zi" >&5
+$as_echo_n "checking whether compiler supports -Zi... " >&6; }
+TCFLAGS="${CFLAGS}"
+if test "x${CFLAGS}" = "x" ; then
+  CFLAGS="-Zi"
+else
+  CFLAGS="${CFLAGS} -Zi"
+fi
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+
+int
+main ()
+{
+
+    return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+              CFLAGS="${TCFLAGS}"
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether compiler supports -MT" >&5
+$as_echo_n "checking whether compiler supports -MT... " >&6; }
+TCFLAGS="${CFLAGS}"
+if test "x${CFLAGS}" = "x" ; then
+  CFLAGS="-MT"
+else
+  CFLAGS="${CFLAGS} -MT"
+fi
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+
+int
+main ()
+{
+
+    return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+              CFLAGS="${TCFLAGS}"
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether compiler supports -W3" >&5
+$as_echo_n "checking whether compiler supports -W3... " >&6; }
+TCFLAGS="${CFLAGS}"
+if test "x${CFLAGS}" = "x" ; then
+  CFLAGS="-W3"
+else
+  CFLAGS="${CFLAGS} -W3"
+fi
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+
+int
+main ()
+{
+
+    return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+              CFLAGS="${TCFLAGS}"
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+    CPPFLAGS="$CPPFLAGS -I${srcroot}/include/msvc_compat"
+  fi
+fi
+if test "x$EXTRA_CFLAGS" != "x" ; then
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether compiler supports $EXTRA_CFLAGS" >&5
+$as_echo_n "checking whether compiler supports $EXTRA_CFLAGS... " >&6; }
+TCFLAGS="${CFLAGS}"
+if test "x${CFLAGS}" = "x" ; then
+  CFLAGS="$EXTRA_CFLAGS"
+else
+  CFLAGS="${CFLAGS} $EXTRA_CFLAGS"
+fi
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+
+int
+main ()
+{
+
+    return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+              CFLAGS="${TCFLAGS}"
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+fi
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking how to run the C preprocessor" >&5
+$as_echo_n "checking how to run the C preprocessor... " >&6; }
+# On Suns, sometimes $CPP names a directory.
+if test -n "$CPP" && test -d "$CPP"; then
+  CPP=
+fi
+if test -z "$CPP"; then
+  if ${ac_cv_prog_CPP+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+      # Double quotes because CPP needs to be expanded
+    for CPP in "$CC -E" "$CC -E -traditional-cpp" "/lib/cpp"
+    do
+      ac_preproc_ok=false
+for ac_c_preproc_warn_flag in '' yes
+do
+  # Use a header file that comes with gcc, so configuring glibc
+  # with a fresh cross-compiler works.
+  # Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
+  # <limits.h> exists even on freestanding compilers.
+  # On the NeXT, cc -E runs the code through the compiler's parser,
+  # not just through cpp. "Syntax error" is here to catch this case.
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#ifdef __STDC__
+# include <limits.h>
+#else
+# include <assert.h>
+#endif
+		     Syntax error
+_ACEOF
+if ac_fn_c_try_cpp "$LINENO"; then :
+
+else
+  # Broken: fails on valid input.
+continue
+fi
+rm -f conftest.err conftest.i conftest.$ac_ext
+
+  # OK, works on sane cases.  Now check whether nonexistent headers
+  # can be detected and how.
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <ac_nonexistent.h>
+_ACEOF
+if ac_fn_c_try_cpp "$LINENO"; then :
+  # Broken: success on invalid input.
+continue
+else
+  # Passes both tests.
+ac_preproc_ok=:
+break
+fi
+rm -f conftest.err conftest.i conftest.$ac_ext
+
+done
+# Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
+rm -f conftest.i conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then :
+  break
+fi
+
+    done
+    ac_cv_prog_CPP=$CPP
+
+fi
+  CPP=$ac_cv_prog_CPP
+else
+  ac_cv_prog_CPP=$CPP
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $CPP" >&5
+$as_echo "$CPP" >&6; }
+ac_preproc_ok=false
+for ac_c_preproc_warn_flag in '' yes
+do
+  # Use a header file that comes with gcc, so configuring glibc
+  # with a fresh cross-compiler works.
+  # Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
+  # <limits.h> exists even on freestanding compilers.
+  # On the NeXT, cc -E runs the code through the compiler's parser,
+  # not just through cpp. "Syntax error" is here to catch this case.
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#ifdef __STDC__
+# include <limits.h>
+#else
+# include <assert.h>
+#endif
+		     Syntax error
+_ACEOF
+if ac_fn_c_try_cpp "$LINENO"; then :
+
+else
+  # Broken: fails on valid input.
+continue
+fi
+rm -f conftest.err conftest.i conftest.$ac_ext
+
+  # OK, works on sane cases.  Now check whether nonexistent headers
+  # can be detected and how.
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <ac_nonexistent.h>
+_ACEOF
+if ac_fn_c_try_cpp "$LINENO"; then :
+  # Broken: success on invalid input.
+continue
+else
+  # Passes both tests.
+ac_preproc_ok=:
+break
+fi
+rm -f conftest.err conftest.i conftest.$ac_ext
+
+done
+# Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
+rm -f conftest.i conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then :
+
+else
+  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "C preprocessor \"$CPP\" fails sanity check
+See \`config.log' for more details" "$LINENO" 5; }
+fi
+
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for grep that handles long lines and -e" >&5
+$as_echo_n "checking for grep that handles long lines and -e... " >&6; }
+if ${ac_cv_path_GREP+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -z "$GREP"; then
+  ac_path_GREP_found=false
+  # Loop through the user's path and test for each of PROGNAME-LIST
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_prog in grep ggrep; do
+    for ac_exec_ext in '' $ac_executable_extensions; do
+      ac_path_GREP="$as_dir/$ac_prog$ac_exec_ext"
+      { test -f "$ac_path_GREP" && $as_test_x "$ac_path_GREP"; } || continue
+# Check for GNU ac_path_GREP and select it if it is found.
+  # Check for GNU $ac_path_GREP
+case `"$ac_path_GREP" --version 2>&1` in
+*GNU*)
+  ac_cv_path_GREP="$ac_path_GREP" ac_path_GREP_found=:;;
+*)
+  ac_count=0
+  $as_echo_n 0123456789 >"conftest.in"
+  while :
+  do
+    cat "conftest.in" "conftest.in" >"conftest.tmp"
+    mv "conftest.tmp" "conftest.in"
+    cp "conftest.in" "conftest.nl"
+    $as_echo 'GREP' >> "conftest.nl"
+    "$ac_path_GREP" -e 'GREP$' -e '-(cannot match)-' < "conftest.nl" >"conftest.out" 2>/dev/null || break
+    diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
+    as_fn_arith $ac_count + 1 && ac_count=$as_val
+    if test $ac_count -gt ${ac_path_GREP_max-0}; then
+      # Best one so far, save it but keep looking for a better one
+      ac_cv_path_GREP="$ac_path_GREP"
+      ac_path_GREP_max=$ac_count
+    fi
+    # 10*(2^10) chars as input seems more than enough
+    test $ac_count -gt 10 && break
+  done
+  rm -f conftest.in conftest.tmp conftest.nl conftest.out;;
+esac
+
+      $ac_path_GREP_found && break 3
+    done
+  done
+  done
+IFS=$as_save_IFS
+  if test -z "$ac_cv_path_GREP"; then
+    as_fn_error $? "no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
+  fi
+else
+  ac_cv_path_GREP=$GREP
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_GREP" >&5
+$as_echo "$ac_cv_path_GREP" >&6; }
+ GREP="$ac_cv_path_GREP"
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for egrep" >&5
+$as_echo_n "checking for egrep... " >&6; }
+if ${ac_cv_path_EGREP+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if echo a | $GREP -E '(a|b)' >/dev/null 2>&1
+   then ac_cv_path_EGREP="$GREP -E"
+   else
+     if test -z "$EGREP"; then
+  ac_path_EGREP_found=false
+  # Loop through the user's path and test for each of PROGNAME-LIST
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_prog in egrep; do
+    for ac_exec_ext in '' $ac_executable_extensions; do
+      ac_path_EGREP="$as_dir/$ac_prog$ac_exec_ext"
+      { test -f "$ac_path_EGREP" && $as_test_x "$ac_path_EGREP"; } || continue
+# Check for GNU ac_path_EGREP and select it if it is found.
+  # Check for GNU $ac_path_EGREP
+case `"$ac_path_EGREP" --version 2>&1` in
+*GNU*)
+  ac_cv_path_EGREP="$ac_path_EGREP" ac_path_EGREP_found=:;;
+*)
+  ac_count=0
+  $as_echo_n 0123456789 >"conftest.in"
+  while :
+  do
+    cat "conftest.in" "conftest.in" >"conftest.tmp"
+    mv "conftest.tmp" "conftest.in"
+    cp "conftest.in" "conftest.nl"
+    $as_echo 'EGREP' >> "conftest.nl"
+    "$ac_path_EGREP" 'EGREP$' < "conftest.nl" >"conftest.out" 2>/dev/null || break
+    diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
+    as_fn_arith $ac_count + 1 && ac_count=$as_val
+    if test $ac_count -gt ${ac_path_EGREP_max-0}; then
+      # Best one so far, save it but keep looking for a better one
+      ac_cv_path_EGREP="$ac_path_EGREP"
+      ac_path_EGREP_max=$ac_count
+    fi
+    # 10*(2^10) chars as input seems more than enough
+    test $ac_count -gt 10 && break
+  done
+  rm -f conftest.in conftest.tmp conftest.nl conftest.out;;
+esac
+
+      $ac_path_EGREP_found && break 3
+    done
+  done
+  done
+IFS=$as_save_IFS
+  if test -z "$ac_cv_path_EGREP"; then
+    as_fn_error $? "no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
+  fi
+else
+  ac_cv_path_EGREP=$EGREP
+fi
+
+   fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_EGREP" >&5
+$as_echo "$ac_cv_path_EGREP" >&6; }
+ EGREP="$ac_cv_path_EGREP"
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for ANSI C header files" >&5
+$as_echo_n "checking for ANSI C header files... " >&6; }
+if ${ac_cv_header_stdc+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdlib.h>
+#include <stdarg.h>
+#include <string.h>
+#include <float.h>
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_cv_header_stdc=yes
+else
+  ac_cv_header_stdc=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+if test $ac_cv_header_stdc = yes; then
+  # SunOS 4.x string.h does not declare mem*, contrary to ANSI.
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <string.h>
+
+_ACEOF
+if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
+  $EGREP "memchr" >/dev/null 2>&1; then :
+
+else
+  ac_cv_header_stdc=no
+fi
+rm -f conftest*
+
+fi
+
+if test $ac_cv_header_stdc = yes; then
+  # ISC 2.0.2 stdlib.h does not declare free, contrary to ANSI.
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdlib.h>
+
+_ACEOF
+if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
+  $EGREP "free" >/dev/null 2>&1; then :
+
+else
+  ac_cv_header_stdc=no
+fi
+rm -f conftest*
+
+fi
+
+if test $ac_cv_header_stdc = yes; then
+  # /bin/cc in Irix-4.0.5 gets non-ANSI ctype macros unless using -ansi.
+  if test "$cross_compiling" = yes; then :
+  :
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <ctype.h>
+#include <stdlib.h>
+#if ((' ' & 0x0FF) == 0x020)
+# define ISLOWER(c) ('a' <= (c) && (c) <= 'z')
+# define TOUPPER(c) (ISLOWER(c) ? 'A' + ((c) - 'a') : (c))
+#else
+# define ISLOWER(c) \
+		   (('a' <= (c) && (c) <= 'i') \
+		     || ('j' <= (c) && (c) <= 'r') \
+		     || ('s' <= (c) && (c) <= 'z'))
+# define TOUPPER(c) (ISLOWER(c) ? ((c) | 0x40) : (c))
+#endif
+
+#define XOR(e, f) (((e) && !(f)) || (!(e) && (f)))
+int
+main ()
+{
+  int i;
+  for (i = 0; i < 256; i++)
+    if (XOR (islower (i), ISLOWER (i))
+	|| toupper (i) != TOUPPER (i))
+      return 2;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_run "$LINENO"; then :
+
+else
+  ac_cv_header_stdc=no
+fi
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
+  conftest.$ac_objext conftest.beam conftest.$ac_ext
+fi
+
+fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_header_stdc" >&5
+$as_echo "$ac_cv_header_stdc" >&6; }
+if test $ac_cv_header_stdc = yes; then
+
+$as_echo "#define STDC_HEADERS 1" >>confdefs.h
+
+fi
+
+# On IRIX 5.3, sys/types and inttypes.h are conflicting.
+for ac_header in sys/types.h sys/stat.h stdlib.h string.h memory.h strings.h \
+		  inttypes.h stdint.h unistd.h
+do :
+  as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+ac_fn_c_check_header_compile "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default
+"
+if eval test \"x\$"$as_ac_Header"\" = x"yes"; then :
+  cat >>confdefs.h <<_ACEOF
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+_ACEOF
+
+fi
+
+done
+
+
+# The cast to long int works around a bug in the HP C Compiler
+# version HP92453-01 B.11.11.23709.GP, which incorrectly rejects
+# declarations like `int a3[[(sizeof (unsigned char)) >= 0]];'.
+# This bug is HP SR number 8606223364.
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking size of void *" >&5
+$as_echo_n "checking size of void *... " >&6; }
+if ${ac_cv_sizeof_void_p+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if ac_fn_c_compute_int "$LINENO" "(long int) (sizeof (void *))" "ac_cv_sizeof_void_p"        "$ac_includes_default"; then :
+
+else
+  if test "$ac_cv_type_void_p" = yes; then
+     { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error 77 "cannot compute sizeof (void *)
+See \`config.log' for more details" "$LINENO" 5; }
+   else
+     ac_cv_sizeof_void_p=0
+   fi
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_sizeof_void_p" >&5
+$as_echo "$ac_cv_sizeof_void_p" >&6; }
+
+
+
+cat >>confdefs.h <<_ACEOF
+#define SIZEOF_VOID_P $ac_cv_sizeof_void_p
+_ACEOF
+
+
+if test "x${ac_cv_sizeof_void_p}" = "x8" ; then
+  LG_SIZEOF_PTR=3
+elif test "x${ac_cv_sizeof_void_p}" = "x4" ; then
+  LG_SIZEOF_PTR=2
+else
+  as_fn_error $? "Unsupported pointer size: ${ac_cv_sizeof_void_p}" "$LINENO" 5
+fi
+cat >>confdefs.h <<_ACEOF
+#define LG_SIZEOF_PTR $LG_SIZEOF_PTR
+_ACEOF
+
+
+# The cast to long int works around a bug in the HP C Compiler
+# version HP92453-01 B.11.11.23709.GP, which incorrectly rejects
+# declarations like `int a3[[(sizeof (unsigned char)) >= 0]];'.
+# This bug is HP SR number 8606223364.
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking size of int" >&5
+$as_echo_n "checking size of int... " >&6; }
+if ${ac_cv_sizeof_int+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if ac_fn_c_compute_int "$LINENO" "(long int) (sizeof (int))" "ac_cv_sizeof_int"        "$ac_includes_default"; then :
+
+else
+  if test "$ac_cv_type_int" = yes; then
+     { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error 77 "cannot compute sizeof (int)
+See \`config.log' for more details" "$LINENO" 5; }
+   else
+     ac_cv_sizeof_int=0
+   fi
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_sizeof_int" >&5
+$as_echo "$ac_cv_sizeof_int" >&6; }
+
+
+
+cat >>confdefs.h <<_ACEOF
+#define SIZEOF_INT $ac_cv_sizeof_int
+_ACEOF
+
+
+if test "x${ac_cv_sizeof_int}" = "x8" ; then
+  LG_SIZEOF_INT=3
+elif test "x${ac_cv_sizeof_int}" = "x4" ; then
+  LG_SIZEOF_INT=2
+else
+  as_fn_error $? "Unsupported int size: ${ac_cv_sizeof_int}" "$LINENO" 5
+fi
+cat >>confdefs.h <<_ACEOF
+#define LG_SIZEOF_INT $LG_SIZEOF_INT
+_ACEOF
+
+
+# The cast to long int works around a bug in the HP C Compiler
+# version HP92453-01 B.11.11.23709.GP, which incorrectly rejects
+# declarations like `int a3[[(sizeof (unsigned char)) >= 0]];'.
+# This bug is HP SR number 8606223364.
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking size of long" >&5
+$as_echo_n "checking size of long... " >&6; }
+if ${ac_cv_sizeof_long+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if ac_fn_c_compute_int "$LINENO" "(long int) (sizeof (long))" "ac_cv_sizeof_long"        "$ac_includes_default"; then :
+
+else
+  if test "$ac_cv_type_long" = yes; then
+     { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error 77 "cannot compute sizeof (long)
+See \`config.log' for more details" "$LINENO" 5; }
+   else
+     ac_cv_sizeof_long=0
+   fi
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_sizeof_long" >&5
+$as_echo "$ac_cv_sizeof_long" >&6; }
+
+
+
+cat >>confdefs.h <<_ACEOF
+#define SIZEOF_LONG $ac_cv_sizeof_long
+_ACEOF
+
+
+if test "x${ac_cv_sizeof_long}" = "x8" ; then
+  LG_SIZEOF_LONG=3
+elif test "x${ac_cv_sizeof_long}" = "x4" ; then
+  LG_SIZEOF_LONG=2
+else
+  as_fn_error $? "Unsupported long size: ${ac_cv_sizeof_long}" "$LINENO" 5
+fi
+cat >>confdefs.h <<_ACEOF
+#define LG_SIZEOF_LONG $LG_SIZEOF_LONG
+_ACEOF
+
+
+# The cast to long int works around a bug in the HP C Compiler
+# version HP92453-01 B.11.11.23709.GP, which incorrectly rejects
+# declarations like `int a3[[(sizeof (unsigned char)) >= 0]];'.
+# This bug is HP SR number 8606223364.
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking size of intmax_t" >&5
+$as_echo_n "checking size of intmax_t... " >&6; }
+if ${ac_cv_sizeof_intmax_t+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if ac_fn_c_compute_int "$LINENO" "(long int) (sizeof (intmax_t))" "ac_cv_sizeof_intmax_t"        "$ac_includes_default"; then :
+
+else
+  if test "$ac_cv_type_intmax_t" = yes; then
+     { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error 77 "cannot compute sizeof (intmax_t)
+See \`config.log' for more details" "$LINENO" 5; }
+   else
+     ac_cv_sizeof_intmax_t=0
+   fi
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_sizeof_intmax_t" >&5
+$as_echo "$ac_cv_sizeof_intmax_t" >&6; }
+
+
+
+cat >>confdefs.h <<_ACEOF
+#define SIZEOF_INTMAX_T $ac_cv_sizeof_intmax_t
+_ACEOF
+
+
+if test "x${ac_cv_sizeof_intmax_t}" = "x16" ; then
+  LG_SIZEOF_INTMAX_T=4
+elif test "x${ac_cv_sizeof_intmax_t}" = "x8" ; then
+  LG_SIZEOF_INTMAX_T=3
+elif test "x${ac_cv_sizeof_intmax_t}" = "x4" ; then
+  LG_SIZEOF_INTMAX_T=2
+else
+  as_fn_error $? "Unsupported intmax_t size: ${ac_cv_sizeof_intmax_t}" "$LINENO" 5
+fi
+cat >>confdefs.h <<_ACEOF
+#define LG_SIZEOF_INTMAX_T $LG_SIZEOF_INTMAX_T
+_ACEOF
+
+
+ac_aux_dir=
+for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
+  if test -f "$ac_dir/install-sh"; then
+    ac_aux_dir=$ac_dir
+    ac_install_sh="$ac_aux_dir/install-sh -c"
+    break
+  elif test -f "$ac_dir/install.sh"; then
+    ac_aux_dir=$ac_dir
+    ac_install_sh="$ac_aux_dir/install.sh -c"
+    break
+  elif test -f "$ac_dir/shtool"; then
+    ac_aux_dir=$ac_dir
+    ac_install_sh="$ac_aux_dir/shtool install -c"
+    break
+  fi
+done
+if test -z "$ac_aux_dir"; then
+  as_fn_error $? "cannot find install-sh, install.sh, or shtool in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" "$LINENO" 5
+fi
+
+# These three variables are undocumented and unsupported,
+# and are intended to be withdrawn in a future Autoconf release.
+# They can cause serious problems if a builder's source tree is in a directory
+# whose full name contains unusual characters.
+ac_config_guess="$SHELL $ac_aux_dir/config.guess"  # Please don't use this var.
+ac_config_sub="$SHELL $ac_aux_dir/config.sub"  # Please don't use this var.
+ac_configure="$SHELL $ac_aux_dir/configure"  # Please don't use this var.
+
+
+# Make sure we can run config.sub.
+$SHELL "$ac_aux_dir/config.sub" sun4 >/dev/null 2>&1 ||
+  as_fn_error $? "cannot run $SHELL $ac_aux_dir/config.sub" "$LINENO" 5
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking build system type" >&5
+$as_echo_n "checking build system type... " >&6; }
+if ${ac_cv_build+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_build_alias=$build_alias
+test "x$ac_build_alias" = x &&
+  ac_build_alias=`$SHELL "$ac_aux_dir/config.guess"`
+test "x$ac_build_alias" = x &&
+  as_fn_error $? "cannot guess build type; you must specify one" "$LINENO" 5
+ac_cv_build=`$SHELL "$ac_aux_dir/config.sub" $ac_build_alias` ||
+  as_fn_error $? "$SHELL $ac_aux_dir/config.sub $ac_build_alias failed" "$LINENO" 5
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_build" >&5
+$as_echo "$ac_cv_build" >&6; }
+case $ac_cv_build in
+*-*-*) ;;
+*) as_fn_error $? "invalid value of canonical build" "$LINENO" 5;;
+esac
+build=$ac_cv_build
+ac_save_IFS=$IFS; IFS='-'
+set x $ac_cv_build
+shift
+build_cpu=$1
+build_vendor=$2
+shift; shift
+# Remember, the first character of IFS is used to create $*,
+# except with old shells:
+build_os=$*
+IFS=$ac_save_IFS
+case $build_os in *\ *) build_os=`echo "$build_os" | sed 's/ /-/g'`;; esac
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking host system type" >&5
+$as_echo_n "checking host system type... " >&6; }
+if ${ac_cv_host+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test "x$host_alias" = x; then
+  ac_cv_host=$ac_cv_build
+else
+  ac_cv_host=`$SHELL "$ac_aux_dir/config.sub" $host_alias` ||
+    as_fn_error $? "$SHELL $ac_aux_dir/config.sub $host_alias failed" "$LINENO" 5
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_host" >&5
+$as_echo "$ac_cv_host" >&6; }
+case $ac_cv_host in
+*-*-*) ;;
+*) as_fn_error $? "invalid value of canonical host" "$LINENO" 5;;
+esac
+host=$ac_cv_host
+ac_save_IFS=$IFS; IFS='-'
+set x $ac_cv_host
+shift
+host_cpu=$1
+host_vendor=$2
+shift; shift
+# Remember, the first character of IFS is used to create $*,
+# except with old shells:
+host_os=$*
+IFS=$ac_save_IFS
+case $host_os in *\ *) host_os=`echo "$host_os" | sed 's/ /-/g'`;; esac
+
+
+CPU_SPINWAIT=""
+case "${host_cpu}" in
+  i[345]86)
+	;;
+  i686)
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether __asm__ is compilable" >&5
+$as_echo_n "checking whether __asm__ is compilable... " >&6; }
+if ${je_cv_asm+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+__asm__ volatile("pause"); return 0;
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  je_cv_asm=yes
+else
+  je_cv_asm=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $je_cv_asm" >&5
+$as_echo "$je_cv_asm" >&6; }
+
+	if test "x${je_cv_asm}" = "xyes" ; then
+	    CPU_SPINWAIT='__asm__ volatile("pause")'
+	fi
+	;;
+  x86_64)
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether __asm__ syntax is compilable" >&5
+$as_echo_n "checking whether __asm__ syntax is compilable... " >&6; }
+if ${je_cv_asm+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+__asm__ volatile("pause"); return 0;
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  je_cv_asm=yes
+else
+  je_cv_asm=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $je_cv_asm" >&5
+$as_echo "$je_cv_asm" >&6; }
+
+	if test "x${je_cv_asm}" = "xyes" ; then
+	    CPU_SPINWAIT='__asm__ volatile("pause")'
+	fi
+	;;
+  *)
+	;;
+esac
+cat >>confdefs.h <<_ACEOF
+#define CPU_SPINWAIT $CPU_SPINWAIT
+_ACEOF
+
+
+LD_PRELOAD_VAR="LD_PRELOAD"
+so="so"
+importlib="${so}"
+o="$ac_objext"
+a="a"
+exe="$ac_exeext"
+libprefix="lib"
+DSO_LDFLAGS='-shared -Wl,-soname,$(@F)'
+RPATH='-Wl,-rpath,$(1)'
+SOREV="${so}.${rev}"
+PIC_CFLAGS='-fPIC -DPIC'
+CTARGET='-o $@'
+LDTARGET='-o $@'
+EXTRA_LDFLAGS=
+MKLIB='ar crus $@'
+CC_MM=1
+
+default_munmap="1"
+JEMALLOC_USABLE_SIZE_CONST="const"
+case "${host}" in
+  *-*-darwin*)
+	CFLAGS="$CFLAGS"
+	abi="macho"
+	$as_echo "#define JEMALLOC_PURGE_MADVISE_FREE  " >>confdefs.h
+
+	RPATH=""
+	LD_PRELOAD_VAR="DYLD_INSERT_LIBRARIES"
+	so="dylib"
+	importlib="${so}"
+	force_tls="0"
+	DSO_LDFLAGS='-shared -Wl,-dylib_install_name,$(@F)'
+	SOREV="${rev}.${so}"
+	;;
+  *-*-freebsd*)
+	CFLAGS="$CFLAGS"
+	abi="elf"
+	$as_echo "#define JEMALLOC_PURGE_MADVISE_FREE  " >>confdefs.h
+
+	force_lazy_lock="1"
+	;;
+  *-*-linux*)
+	CFLAGS="$CFLAGS"
+	CPPFLAGS="$CPPFLAGS -D_GNU_SOURCE"
+	abi="elf"
+	$as_echo "#define JEMALLOC_HAS_ALLOCA_H 1" >>confdefs.h
+
+	$as_echo "#define JEMALLOC_PURGE_MADVISE_DONTNEED  " >>confdefs.h
+
+	$as_echo "#define JEMALLOC_THREADED_INIT  " >>confdefs.h
+
+	JEMALLOC_USABLE_SIZE_CONST=""
+	default_munmap="0"
+	;;
+  *-*-netbsd*)
+	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking ABI" >&5
+$as_echo_n "checking ABI... " >&6; }
+        cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#ifdef __ELF__
+/* ELF */
+#else
+#error aout
+#endif
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  CFLAGS="$CFLAGS"; abi="elf"
+else
+  abi="aout"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $abi" >&5
+$as_echo "$abi" >&6; }
+	$as_echo "#define JEMALLOC_PURGE_MADVISE_FREE  " >>confdefs.h
+
+	;;
+  *-*-solaris2*)
+	CFLAGS="$CFLAGS"
+	abi="elf"
+	RPATH='-Wl,-R,$(1)'
+		CPPFLAGS="$CPPFLAGS -D_POSIX_PTHREAD_SEMANTICS"
+	LIBS="$LIBS -lposix4 -lsocket -lnsl"
+	;;
+  *-ibm-aix*)
+	if "$LG_SIZEOF_PTR" = "8"; then
+	  	  LD_PRELOAD_VAR="LDR_PRELOAD64"
+	else
+	  	  LD_PRELOAD_VAR="LDR_PRELOAD"
+	fi
+	abi="xcoff"
+	;;
+  *-*-mingw*)
+	abi="pecoff"
+	force_tls="0"
+	RPATH=""
+	so="dll"
+	if test "x$je_cv_msvc" = "xyes" ; then
+	  importlib="lib"
+	  DSO_LDFLAGS="-LD"
+	  EXTRA_LDFLAGS="-link -DEBUG"
+	  CTARGET='-Fo$@'
+	  LDTARGET='-Fe$@'
+	  MKLIB='lib -nologo -out:$@'
+	  CC_MM=
+        else
+	  importlib="${so}"
+	  DSO_LDFLAGS="-shared"
+	fi
+	a="lib"
+	libprefix=""
+	SOREV="${so}"
+	PIC_CFLAGS=""
+	;;
+  *)
+	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: Unsupported operating system: ${host}" >&5
+$as_echo "Unsupported operating system: ${host}" >&6; }
+	abi="elf"
+	;;
+esac
+cat >>confdefs.h <<_ACEOF
+#define JEMALLOC_USABLE_SIZE_CONST $JEMALLOC_USABLE_SIZE_CONST
+_ACEOF
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+if test "x$abi" != "xpecoff"; then
+    LIBS="$LIBS -lm"
+fi
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether __attribute__ syntax is compilable" >&5
+$as_echo_n "checking whether __attribute__ syntax is compilable... " >&6; }
+if ${je_cv_attribute+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+static __attribute__((unused)) void foo(void){}
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  je_cv_attribute=yes
+else
+  je_cv_attribute=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $je_cv_attribute" >&5
+$as_echo "$je_cv_attribute" >&6; }
+
+if test "x${je_cv_attribute}" = "xyes" ; then
+  $as_echo "#define JEMALLOC_HAVE_ATTR  " >>confdefs.h
+
+  if test "x${GCC}" = "xyes" -a "x${abi}" = "xelf"; then
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether compiler supports -fvisibility=hidden" >&5
+$as_echo_n "checking whether compiler supports -fvisibility=hidden... " >&6; }
+TCFLAGS="${CFLAGS}"
+if test "x${CFLAGS}" = "x" ; then
+  CFLAGS="-fvisibility=hidden"
+else
+  CFLAGS="${CFLAGS} -fvisibility=hidden"
+fi
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+
+int
+main ()
+{
+
+    return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+              CFLAGS="${TCFLAGS}"
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+  fi
+fi
+SAVED_CFLAGS="${CFLAGS}"
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether compiler supports -Werror" >&5
+$as_echo_n "checking whether compiler supports -Werror... " >&6; }
+TCFLAGS="${CFLAGS}"
+if test "x${CFLAGS}" = "x" ; then
+  CFLAGS="-Werror"
+else
+  CFLAGS="${CFLAGS} -Werror"
+fi
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+
+int
+main ()
+{
+
+    return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+              CFLAGS="${TCFLAGS}"
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether tls_model attribute is compilable" >&5
+$as_echo_n "checking whether tls_model attribute is compilable... " >&6; }
+if ${je_cv_tls_model+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+static __thread int
+               __attribute__((tls_model("initial-exec"))) foo;
+               foo = 0;
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  je_cv_tls_model=yes
+else
+  je_cv_tls_model=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $je_cv_tls_model" >&5
+$as_echo "$je_cv_tls_model" >&6; }
+
+CFLAGS="${SAVED_CFLAGS}"
+if test "x${je_cv_tls_model}" = "xyes" ; then
+  $as_echo "#define JEMALLOC_TLS_MODEL __attribute__((tls_model(\"initial-exec\")))" >>confdefs.h
+
+else
+  $as_echo "#define JEMALLOC_TLS_MODEL  " >>confdefs.h
+
+fi
+
+
+# Check whether --with-rpath was given.
+if test "${with_rpath+set}" = set; then :
+  withval=$with_rpath; if test "x$with_rpath" = "xno" ; then
+  RPATH_EXTRA=
+else
+  RPATH_EXTRA="`echo $with_rpath | tr \":\" \" \"`"
+fi
+else
+  RPATH_EXTRA=
+
+fi
+
+
+
+# Check whether --enable-autogen was given.
+if test "${enable_autogen+set}" = set; then :
+  enableval=$enable_autogen; if test "x$enable_autogen" = "xno" ; then
+  enable_autogen="0"
+else
+  enable_autogen="1"
+fi
+
+else
+  enable_autogen="0"
+
+fi
+
+
+
+# Find a good install program.  We prefer a C program (faster),
+# so one script is as good as another.  But avoid the broken or
+# incompatible versions:
+# SysV /etc/install, /usr/sbin/install
+# SunOS /usr/etc/install
+# IRIX /sbin/install
+# AIX /bin/install
+# AmigaOS /C/install, which installs bootblocks on floppy discs
+# AIX 4 /usr/bin/installbsd, which doesn't work without a -g flag
+# AFS /usr/afsws/bin/install, which mishandles nonexistent args
+# SVR4 /usr/ucb/install, which tries to use the nonexistent group "staff"
+# OS/2's system install, which has a completely different semantic
+# ./install, which can be erroneously created by make from ./install.sh.
+# Reject install programs that cannot install multiple files.
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a BSD-compatible install" >&5
+$as_echo_n "checking for a BSD-compatible install... " >&6; }
+if test -z "$INSTALL"; then
+if ${ac_cv_path_install+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    # Account for people who put trailing slashes in PATH elements.
+case $as_dir/ in #((
+  ./ | .// | /[cC]/* | \
+  /etc/* | /usr/sbin/* | /usr/etc/* | /sbin/* | /usr/afsws/bin/* | \
+  ?:[\\/]os2[\\/]install[\\/]* | ?:[\\/]OS2[\\/]INSTALL[\\/]* | \
+  /usr/ucb/* ) ;;
+  *)
+    # OSF1 and SCO ODT 3.0 have their own names for install.
+    # Don't use installbsd from OSF since it installs stuff as root
+    # by default.
+    for ac_prog in ginstall scoinst install; do
+      for ac_exec_ext in '' $ac_executable_extensions; do
+	if { test -f "$as_dir/$ac_prog$ac_exec_ext" && $as_test_x "$as_dir/$ac_prog$ac_exec_ext"; }; then
+	  if test $ac_prog = install &&
+	    grep dspmsg "$as_dir/$ac_prog$ac_exec_ext" >/dev/null 2>&1; then
+	    # AIX install.  It has an incompatible calling convention.
+	    :
+	  elif test $ac_prog = install &&
+	    grep pwplus "$as_dir/$ac_prog$ac_exec_ext" >/dev/null 2>&1; then
+	    # program-specific install script used by HP pwplus--don't use.
+	    :
+	  else
+	    rm -rf conftest.one conftest.two conftest.dir
+	    echo one > conftest.one
+	    echo two > conftest.two
+	    mkdir conftest.dir
+	    if "$as_dir/$ac_prog$ac_exec_ext" -c conftest.one conftest.two "`pwd`/conftest.dir" &&
+	      test -s conftest.one && test -s conftest.two &&
+	      test -s conftest.dir/conftest.one &&
+	      test -s conftest.dir/conftest.two
+	    then
+	      ac_cv_path_install="$as_dir/$ac_prog$ac_exec_ext -c"
+	      break 3
+	    fi
+	  fi
+	fi
+      done
+    done
+    ;;
+esac
+
+  done
+IFS=$as_save_IFS
+
+rm -rf conftest.one conftest.two conftest.dir
+
+fi
+  if test "${ac_cv_path_install+set}" = set; then
+    INSTALL=$ac_cv_path_install
+  else
+    # As a last resort, use the slow shell script.  Don't cache a
+    # value for INSTALL within a source directory, because that will
+    # break other packages using the cache if that directory is
+    # removed, or if the value is a relative name.
+    INSTALL=$ac_install_sh
+  fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $INSTALL" >&5
+$as_echo "$INSTALL" >&6; }
+
+# Use test -z because SunOS4 sh mishandles braces in ${var-val}.
+# It thinks the first close brace ends the variable substitution.
+test -z "$INSTALL_PROGRAM" && INSTALL_PROGRAM='${INSTALL}'
+
+test -z "$INSTALL_SCRIPT" && INSTALL_SCRIPT='${INSTALL}'
+
+test -z "$INSTALL_DATA" && INSTALL_DATA='${INSTALL} -m 644'
+
+if test -n "$ac_tool_prefix"; then
+  # Extract the first word of "${ac_tool_prefix}ranlib", so it can be a program name with args.
+set dummy ${ac_tool_prefix}ranlib; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_RANLIB+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$RANLIB"; then
+  ac_cv_prog_RANLIB="$RANLIB" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    ac_cv_prog_RANLIB="${ac_tool_prefix}ranlib"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+RANLIB=$ac_cv_prog_RANLIB
+if test -n "$RANLIB"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $RANLIB" >&5
+$as_echo "$RANLIB" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+fi
+if test -z "$ac_cv_prog_RANLIB"; then
+  ac_ct_RANLIB=$RANLIB
+  # Extract the first word of "ranlib", so it can be a program name with args.
+set dummy ranlib; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_ac_ct_RANLIB+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$ac_ct_RANLIB"; then
+  ac_cv_prog_ac_ct_RANLIB="$ac_ct_RANLIB" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    ac_cv_prog_ac_ct_RANLIB="ranlib"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+ac_ct_RANLIB=$ac_cv_prog_ac_ct_RANLIB
+if test -n "$ac_ct_RANLIB"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_RANLIB" >&5
+$as_echo "$ac_ct_RANLIB" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+  if test "x$ac_ct_RANLIB" = x; then
+    RANLIB=":"
+  else
+    case $cross_compiling:$ac_tool_warned in
+yes:)
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+$as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
+ac_tool_warned=yes ;;
+esac
+    RANLIB=$ac_ct_RANLIB
+  fi
+else
+  RANLIB="$ac_cv_prog_RANLIB"
+fi
+
+# Extract the first word of "ar", so it can be a program name with args.
+set dummy ar; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_path_AR+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  case $AR in
+  [\\/]* | ?:[\\/]*)
+  ac_cv_path_AR="$AR" # Let the user override the test with a path.
+  ;;
+  *)
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    ac_cv_path_AR="$as_dir/$ac_word$ac_exec_ext"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+  test -z "$ac_cv_path_AR" && ac_cv_path_AR="false"
+  ;;
+esac
+fi
+AR=$ac_cv_path_AR
+if test -n "$AR"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $AR" >&5
+$as_echo "$AR" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+# Extract the first word of "ld", so it can be a program name with args.
+set dummy ld; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_path_LD+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  case $LD in
+  [\\/]* | ?:[\\/]*)
+  ac_cv_path_LD="$LD" # Let the user override the test with a path.
+  ;;
+  *)
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    ac_cv_path_LD="$as_dir/$ac_word$ac_exec_ext"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+  test -z "$ac_cv_path_LD" && ac_cv_path_LD="false"
+  ;;
+esac
+fi
+LD=$ac_cv_path_LD
+if test -n "$LD"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $LD" >&5
+$as_echo "$LD" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+# Extract the first word of "autoconf", so it can be a program name with args.
+set dummy autoconf; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_path_AUTOCONF+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  case $AUTOCONF in
+  [\\/]* | ?:[\\/]*)
+  ac_cv_path_AUTOCONF="$AUTOCONF" # Let the user override the test with a path.
+  ;;
+  *)
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    ac_cv_path_AUTOCONF="$as_dir/$ac_word$ac_exec_ext"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+  test -z "$ac_cv_path_AUTOCONF" && ac_cv_path_AUTOCONF="false"
+  ;;
+esac
+fi
+AUTOCONF=$ac_cv_path_AUTOCONF
+if test -n "$AUTOCONF"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $AUTOCONF" >&5
+$as_echo "$AUTOCONF" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+
+public_syms="malloc_conf malloc_message malloc calloc posix_memalign aligned_alloc realloc free malloc_usable_size malloc_stats_print mallctl mallctlnametomib mallctlbymib"
+
+ac_fn_c_check_func "$LINENO" "memalign" "ac_cv_func_memalign"
+if test "x$ac_cv_func_memalign" = xyes; then :
+  $as_echo "#define JEMALLOC_OVERRIDE_MEMALIGN  " >>confdefs.h
+
+	       public_syms="${public_syms} memalign"
+fi
+
+ac_fn_c_check_func "$LINENO" "valloc" "ac_cv_func_valloc"
+if test "x$ac_cv_func_valloc" = xyes; then :
+  $as_echo "#define JEMALLOC_OVERRIDE_VALLOC  " >>confdefs.h
+
+	       public_syms="${public_syms} valloc"
+fi
+
+
+# Check whether --enable-experimental was given.
+if test "${enable_experimental+set}" = set; then :
+  enableval=$enable_experimental; if test "x$enable_experimental" = "xno" ; then
+  enable_experimental="0"
+else
+  enable_experimental="1"
+fi
+
+else
+  enable_experimental="1"
+
+fi
+
+if test "x$enable_experimental" = "x1" ; then
+  $as_echo "#define JEMALLOC_EXPERIMENTAL  " >>confdefs.h
+
+  public_syms="${public_syms} allocm dallocm nallocm rallocm sallocm"
+fi
+
+
+
+# Check whether --with-mangling was given.
+if test "${with_mangling+set}" = set; then :
+  withval=$with_mangling; mangling_map="$with_mangling"
+else
+  mangling_map=""
+fi
+
+for nm in `echo ${mangling_map} |tr ',' ' '` ; do
+  k="`echo ${nm} |tr ':' ' ' |awk '{print $1}'`"
+  n="je_${k}"
+  m=`echo ${nm} |tr ':' ' ' |awk '{print $2}'`
+  cat >>confdefs.h <<_ACEOF
+#define ${n} ${m}
+_ACEOF
+
+    public_syms=`for sym in ${public_syms}; do echo "${sym}"; done |grep -v "^${k}\$" |tr '\n' ' '`
+done
+
+
+# Check whether --with-jemalloc_prefix was given.
+if test "${with_jemalloc_prefix+set}" = set; then :
+  withval=$with_jemalloc_prefix; JEMALLOC_PREFIX="$with_jemalloc_prefix"
+else
+  if test "x$abi" != "xmacho" -a "x$abi" != "xpecoff"; then
+  JEMALLOC_PREFIX=""
+else
+  JEMALLOC_PREFIX="je_"
+fi
+
+fi
+
+if test "x$JEMALLOC_PREFIX" != "x" ; then
+  JEMALLOC_CPREFIX=`echo ${JEMALLOC_PREFIX} | tr "a-z" "A-Z"`
+  cat >>confdefs.h <<_ACEOF
+#define JEMALLOC_PREFIX "$JEMALLOC_PREFIX"
+_ACEOF
+
+  cat >>confdefs.h <<_ACEOF
+#define JEMALLOC_CPREFIX "$JEMALLOC_CPREFIX"
+_ACEOF
+
+fi
+for stem in ${public_syms}; do
+  n="je_${stem}"
+  m="${JEMALLOC_PREFIX}${stem}"
+  cat >>confdefs.h <<_ACEOF
+#define ${n} ${m}
+_ACEOF
+
+done
+
+
+# Check whether --with-export was given.
+if test "${with_export+set}" = set; then :
+  withval=$with_export; if test "x$with_export" = "xno"; then
+  $as_echo "#define JEMALLOC_EXPORT /**/" >>confdefs.h
+
+fi
+
+fi
+
+
+
+# Check whether --with-private_namespace was given.
+if test "${with_private_namespace+set}" = set; then :
+  withval=$with_private_namespace; JEMALLOC_PRIVATE_NAMESPACE="$with_private_namespace"
+else
+  JEMALLOC_PRIVATE_NAMESPACE=""
+
+fi
+
+cat >>confdefs.h <<_ACEOF
+#define JEMALLOC_PRIVATE_NAMESPACE "$JEMALLOC_PRIVATE_NAMESPACE"
+_ACEOF
+
+if test "x$JEMALLOC_PRIVATE_NAMESPACE" != "x" ; then
+  cat >>confdefs.h <<_ACEOF
+#define JEMALLOC_N(string_that_no_one_should_want_to_use_as_a_jemalloc_private_namespace_prefix) ${JEMALLOC_PRIVATE_NAMESPACE}##string_that_no_one_should_want_to_use_as_a_jemalloc_private_namespace_prefix
+_ACEOF
+
+else
+  cat >>confdefs.h <<_ACEOF
+#define JEMALLOC_N(string_that_no_one_should_want_to_use_as_a_jemalloc_private_namespace_prefix) string_that_no_one_should_want_to_use_as_a_jemalloc_private_namespace_prefix
+_ACEOF
+
+fi
+
+
+# Check whether --with-install_suffix was given.
+if test "${with_install_suffix+set}" = set; then :
+  withval=$with_install_suffix; INSTALL_SUFFIX="$with_install_suffix"
+else
+  INSTALL_SUFFIX=
+
+fi
+
+install_suffix="$INSTALL_SUFFIX"
+
+
+cfgoutputs_in="${srcroot}Makefile.in"
+cfgoutputs_in="${cfgoutputs_in} ${srcroot}doc/html.xsl.in"
+cfgoutputs_in="${cfgoutputs_in} ${srcroot}doc/manpages.xsl.in"
+cfgoutputs_in="${cfgoutputs_in} ${srcroot}doc/jemalloc.xml.in"
+cfgoutputs_in="${cfgoutputs_in} ${srcroot}include/jemalloc/jemalloc.h.in"
+cfgoutputs_in="${cfgoutputs_in} ${srcroot}include/jemalloc/internal/jemalloc_internal.h.in"
+cfgoutputs_in="${cfgoutputs_in} ${srcroot}test/jemalloc_test.h.in"
+
+cfgoutputs_out="Makefile"
+cfgoutputs_out="${cfgoutputs_out} doc/html.xsl"
+cfgoutputs_out="${cfgoutputs_out} doc/manpages.xsl"
+cfgoutputs_out="${cfgoutputs_out} doc/jemalloc${install_suffix}.xml"
+cfgoutputs_out="${cfgoutputs_out} include/jemalloc/jemalloc${install_suffix}.h"
+cfgoutputs_out="${cfgoutputs_out} include/jemalloc/internal/jemalloc_internal.h"
+cfgoutputs_out="${cfgoutputs_out} test/jemalloc_test.h"
+
+cfgoutputs_tup="Makefile"
+cfgoutputs_tup="${cfgoutputs_tup} doc/html.xsl:doc/html.xsl.in"
+cfgoutputs_tup="${cfgoutputs_tup} doc/manpages.xsl:doc/manpages.xsl.in"
+cfgoutputs_tup="${cfgoutputs_tup} doc/jemalloc${install_suffix}.xml:doc/jemalloc.xml.in"
+cfgoutputs_tup="${cfgoutputs_tup} include/jemalloc/jemalloc${install_suffix}.h:include/jemalloc/jemalloc.h.in"
+cfgoutputs_tup="${cfgoutputs_tup} include/jemalloc/internal/jemalloc_internal.h"
+cfgoutputs_tup="${cfgoutputs_tup} test/jemalloc_test.h:test/jemalloc_test.h.in"
+
+cfghdrs_in="${srcroot}include/jemalloc/jemalloc_defs.h.in"
+cfghdrs_in="${cfghdrs_in} ${srcroot}include/jemalloc/internal/size_classes.sh"
+
+cfghdrs_out="include/jemalloc/jemalloc_defs${install_suffix}.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/internal/size_classes.h"
+
+cfghdrs_tup="include/jemalloc/jemalloc_defs${install_suffix}.h:include/jemalloc/jemalloc_defs.h.in"
+
+# Check whether --enable-cc-silence was given.
+if test "${enable_cc_silence+set}" = set; then :
+  enableval=$enable_cc_silence; if test "x$enable_cc_silence" = "xno" ; then
+  enable_cc_silence="0"
+else
+  enable_cc_silence="1"
+fi
+
+else
+  enable_cc_silence="0"
+
+fi
+
+if test "x$enable_cc_silence" = "x1" ; then
+  $as_echo "#define JEMALLOC_CC_SILENCE  " >>confdefs.h
+
+fi
+
+# Check whether --enable-debug was given.
+if test "${enable_debug+set}" = set; then :
+  enableval=$enable_debug; if test "x$enable_debug" = "xno" ; then
+  enable_debug="0"
+else
+  enable_debug="1"
+fi
+
+else
+  enable_debug="0"
+
+fi
+
+if test "x$enable_debug" = "x1" ; then
+  $as_echo "#define JEMALLOC_DEBUG  " >>confdefs.h
+
+  enable_ivsalloc="1"
+fi
+
+
+# Check whether --enable-ivsalloc was given.
+if test "${enable_ivsalloc+set}" = set; then :
+  enableval=$enable_ivsalloc; if test "x$enable_ivsalloc" = "xno" ; then
+  enable_ivsalloc="0"
+else
+  enable_ivsalloc="1"
+fi
+
+else
+  enable_ivsalloc="0"
+
+fi
+
+if test "x$enable_ivsalloc" = "x1" ; then
+  $as_echo "#define JEMALLOC_IVSALLOC  " >>confdefs.h
+
+fi
+
+if test "x$enable_debug" = "x0" -a "x$no_CFLAGS" = "xyes" ; then
+    optimize="no"
+  echo "$EXTRA_CFLAGS" | grep "\-O" >/dev/null || optimize="yes"
+  if test "x${optimize}" = "xyes" ; then
+    if test "x$GCC" = "xyes" ; then
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether compiler supports -O3" >&5
+$as_echo_n "checking whether compiler supports -O3... " >&6; }
+TCFLAGS="${CFLAGS}"
+if test "x${CFLAGS}" = "x" ; then
+  CFLAGS="-O3"
+else
+  CFLAGS="${CFLAGS} -O3"
+fi
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+
+int
+main ()
+{
+
+    return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+              CFLAGS="${TCFLAGS}"
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether compiler supports -funroll-loops" >&5
+$as_echo_n "checking whether compiler supports -funroll-loops... " >&6; }
+TCFLAGS="${CFLAGS}"
+if test "x${CFLAGS}" = "x" ; then
+  CFLAGS="-funroll-loops"
+else
+  CFLAGS="${CFLAGS} -funroll-loops"
+fi
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+
+int
+main ()
+{
+
+    return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+              CFLAGS="${TCFLAGS}"
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+    elif test "x$je_cv_msvc" = "xyes" ; then
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether compiler supports -O2" >&5
+$as_echo_n "checking whether compiler supports -O2... " >&6; }
+TCFLAGS="${CFLAGS}"
+if test "x${CFLAGS}" = "x" ; then
+  CFLAGS="-O2"
+else
+  CFLAGS="${CFLAGS} -O2"
+fi
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+
+int
+main ()
+{
+
+    return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+              CFLAGS="${TCFLAGS}"
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+    else
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether compiler supports -O" >&5
+$as_echo_n "checking whether compiler supports -O... " >&6; }
+TCFLAGS="${CFLAGS}"
+if test "x${CFLAGS}" = "x" ; then
+  CFLAGS="-O"
+else
+  CFLAGS="${CFLAGS} -O"
+fi
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+
+int
+main ()
+{
+
+    return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+              CFLAGS="${TCFLAGS}"
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+    fi
+  fi
+fi
+
+# Check whether --enable-stats was given.
+if test "${enable_stats+set}" = set; then :
+  enableval=$enable_stats; if test "x$enable_stats" = "xno" ; then
+  enable_stats="0"
+else
+  enable_stats="1"
+fi
+
+else
+  enable_stats="1"
+
+fi
+
+if test "x$enable_stats" = "x1" ; then
+  $as_echo "#define JEMALLOC_STATS  " >>confdefs.h
+
+fi
+
+
+# Check whether --enable-prof was given.
+if test "${enable_prof+set}" = set; then :
+  enableval=$enable_prof; if test "x$enable_prof" = "xno" ; then
+  enable_prof="0"
+else
+  enable_prof="1"
+fi
+
+else
+  enable_prof="0"
+
+fi
+
+if test "x$enable_prof" = "x1" ; then
+  backtrace_method=""
+else
+  backtrace_method="N/A"
+fi
+
+# Check whether --enable-prof-libunwind was given.
+if test "${enable_prof_libunwind+set}" = set; then :
+  enableval=$enable_prof_libunwind; if test "x$enable_prof_libunwind" = "xno" ; then
+  enable_prof_libunwind="0"
+else
+  enable_prof_libunwind="1"
+fi
+
+else
+  enable_prof_libunwind="0"
+
+fi
+
+
+# Check whether --with-static_libunwind was given.
+if test "${with_static_libunwind+set}" = set; then :
+  withval=$with_static_libunwind; if test "x$with_static_libunwind" = "xno" ; then
+  LUNWIND="-lunwind"
+else
+  if test ! -f "$with_static_libunwind" ; then
+    as_fn_error $? "Static libunwind not found: $with_static_libunwind" "$LINENO" 5
+  fi
+  LUNWIND="$with_static_libunwind"
+fi
+else
+  LUNWIND="-lunwind"
+
+fi
+
+if test "x$backtrace_method" = "x" -a "x$enable_prof_libunwind" = "x1" ; then
+  for ac_header in libunwind.h
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "libunwind.h" "ac_cv_header_libunwind_h" "$ac_includes_default"
+if test "x$ac_cv_header_libunwind_h" = xyes; then :
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_LIBUNWIND_H 1
+_ACEOF
+
+else
+  enable_prof_libunwind="0"
+fi
+
+done
+
+  if test "x$LUNWIND" = "x-lunwind" ; then
+    { $as_echo "$as_me:${as_lineno-$LINENO}: checking for backtrace in -lunwind" >&5
+$as_echo_n "checking for backtrace in -lunwind... " >&6; }
+if ${ac_cv_lib_unwind_backtrace+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_check_lib_save_LIBS=$LIBS
+LIBS="-lunwind  $LIBS"
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char backtrace ();
+int
+main ()
+{
+return backtrace ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  ac_cv_lib_unwind_backtrace=yes
+else
+  ac_cv_lib_unwind_backtrace=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+LIBS=$ac_check_lib_save_LIBS
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_unwind_backtrace" >&5
+$as_echo "$ac_cv_lib_unwind_backtrace" >&6; }
+if test "x$ac_cv_lib_unwind_backtrace" = xyes; then :
+  LIBS="$LIBS $LUNWIND"
+else
+  enable_prof_libunwind="0"
+fi
+
+  else
+    LIBS="$LIBS $LUNWIND"
+  fi
+  if test "x${enable_prof_libunwind}" = "x1" ; then
+    backtrace_method="libunwind"
+    $as_echo "#define JEMALLOC_PROF_LIBUNWIND  " >>confdefs.h
+
+  fi
+fi
+
+# Check whether --enable-prof-libgcc was given.
+if test "${enable_prof_libgcc+set}" = set; then :
+  enableval=$enable_prof_libgcc; if test "x$enable_prof_libgcc" = "xno" ; then
+  enable_prof_libgcc="0"
+else
+  enable_prof_libgcc="1"
+fi
+
+else
+  enable_prof_libgcc="1"
+
+fi
+
+if test "x$backtrace_method" = "x" -a "x$enable_prof_libgcc" = "x1" \
+     -a "x$GCC" = "xyes" ; then
+  for ac_header in unwind.h
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "unwind.h" "ac_cv_header_unwind_h" "$ac_includes_default"
+if test "x$ac_cv_header_unwind_h" = xyes; then :
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_UNWIND_H 1
+_ACEOF
+
+else
+  enable_prof_libgcc="0"
+fi
+
+done
+
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for _Unwind_Backtrace in -lgcc" >&5
+$as_echo_n "checking for _Unwind_Backtrace in -lgcc... " >&6; }
+if ${ac_cv_lib_gcc__Unwind_Backtrace+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_check_lib_save_LIBS=$LIBS
+LIBS="-lgcc  $LIBS"
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char _Unwind_Backtrace ();
+int
+main ()
+{
+return _Unwind_Backtrace ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  ac_cv_lib_gcc__Unwind_Backtrace=yes
+else
+  ac_cv_lib_gcc__Unwind_Backtrace=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+LIBS=$ac_check_lib_save_LIBS
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_gcc__Unwind_Backtrace" >&5
+$as_echo "$ac_cv_lib_gcc__Unwind_Backtrace" >&6; }
+if test "x$ac_cv_lib_gcc__Unwind_Backtrace" = xyes; then :
+  LIBS="$LIBS -lgcc"
+else
+  enable_prof_libgcc="0"
+fi
+
+      { $as_echo "$as_me:${as_lineno-$LINENO}: checking libgcc-based backtracing reliability on ${host_cpu}" >&5
+$as_echo_n "checking libgcc-based backtracing reliability on ${host_cpu}... " >&6; }
+  case "${host_cpu}" in
+    i[3456]86)
+      { $as_echo "$as_me:${as_lineno-$LINENO}: result: unreliable" >&5
+$as_echo "unreliable" >&6; }
+      enable_prof_libgcc="0";
+      ;;
+    x86_64)
+      { $as_echo "$as_me:${as_lineno-$LINENO}: result: reliable" >&5
+$as_echo "reliable" >&6; }
+      ;;
+    *)
+      { $as_echo "$as_me:${as_lineno-$LINENO}: result: unreliable" >&5
+$as_echo "unreliable" >&6; }
+      enable_prof_libgcc="0";
+      ;;
+  esac
+  if test "x${enable_prof_libgcc}" = "x1" ; then
+    backtrace_method="libgcc"
+    $as_echo "#define JEMALLOC_PROF_LIBGCC  " >>confdefs.h
+
+  fi
+else
+  enable_prof_libgcc="0"
+fi
+
+# Check whether --enable-prof-gcc was given.
+if test "${enable_prof_gcc+set}" = set; then :
+  enableval=$enable_prof_gcc; if test "x$enable_prof_gcc" = "xno" ; then
+  enable_prof_gcc="0"
+else
+  enable_prof_gcc="1"
+fi
+
+else
+  enable_prof_gcc="1"
+
+fi
+
+if test "x$backtrace_method" = "x" -a "x$enable_prof_gcc" = "x1" \
+     -a "x$GCC" = "xyes" ; then
+  backtrace_method="gcc intrinsics"
+  $as_echo "#define JEMALLOC_PROF_GCC  " >>confdefs.h
+
+else
+  enable_prof_gcc="0"
+fi
+
+if test "x$backtrace_method" = "x" ; then
+  backtrace_method="none (disabling profiling)"
+  enable_prof="0"
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking configured backtracing method" >&5
+$as_echo_n "checking configured backtracing method... " >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $backtrace_method" >&5
+$as_echo "$backtrace_method" >&6; }
+if test "x$enable_prof" = "x1" ; then
+  if test "x${force_tls}" = "x0" ; then
+    as_fn_error $? "Heap profiling requires TLS" "$LINENO" 5;
+  fi
+  force_tls="1"
+  $as_echo "#define JEMALLOC_PROF  " >>confdefs.h
+
+fi
+
+
+# Check whether --enable-tcache was given.
+if test "${enable_tcache+set}" = set; then :
+  enableval=$enable_tcache; if test "x$enable_tcache" = "xno" ; then
+  enable_tcache="0"
+else
+  enable_tcache="1"
+fi
+
+else
+  enable_tcache="1"
+
+fi
+
+if test "x$enable_tcache" = "x1" ; then
+  $as_echo "#define JEMALLOC_TCACHE  " >>confdefs.h
+
+fi
+
+
+# Check whether --enable-mremap was given.
+if test "${enable_mremap+set}" = set; then :
+  enableval=$enable_mremap; if test "x$enable_mremap" = "xno" ; then
+  enable_mremap="0"
+else
+  enable_mremap="1"
+fi
+
+else
+  enable_mremap="0"
+
+fi
+
+if test "x$enable_mremap" = "x1" ; then
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether mremap(...MREMAP_FIXED...) is compilable" >&5
+$as_echo_n "checking whether mremap(...MREMAP_FIXED...) is compilable... " >&6; }
+if ${je_cv_mremap_fixed+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+#define _GNU_SOURCE
+#include <sys/mman.h>
+
+int
+main ()
+{
+
+void *p = mremap((void *)0, 0, 0, MREMAP_MAYMOVE|MREMAP_FIXED, (void *)0);
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  je_cv_mremap_fixed=yes
+else
+  je_cv_mremap_fixed=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $je_cv_mremap_fixed" >&5
+$as_echo "$je_cv_mremap_fixed" >&6; }
+
+  if test "x${je_cv_mremap_fixed}" = "xno" ; then
+    enable_mremap="0"
+  fi
+fi
+if test "x$enable_mremap" = "x1" ; then
+  $as_echo "#define JEMALLOC_MREMAP  " >>confdefs.h
+
+fi
+
+
+# Check whether --enable-munmap was given.
+if test "${enable_munmap+set}" = set; then :
+  enableval=$enable_munmap; if test "x$enable_munmap" = "xno" ; then
+  enable_munmap="0"
+else
+  enable_munmap="1"
+fi
+
+else
+  enable_munmap="${default_munmap}"
+
+fi
+
+if test "x$enable_munmap" = "x1" ; then
+  $as_echo "#define JEMALLOC_MUNMAP  " >>confdefs.h
+
+fi
+
+
+# Check whether --enable-dss was given.
+if test "${enable_dss+set}" = set; then :
+  enableval=$enable_dss; if test "x$enable_dss" = "xno" ; then
+  enable_dss="0"
+else
+  enable_dss="1"
+fi
+
+else
+  enable_dss="0"
+
+fi
+
+ac_fn_c_check_func "$LINENO" "sbrk" "ac_cv_func_sbrk"
+if test "x$ac_cv_func_sbrk" = xyes; then :
+  have_sbrk="1"
+else
+  have_sbrk="0"
+fi
+
+if test "x$have_sbrk" = "x1" ; then
+  $as_echo "#define JEMALLOC_HAVE_SBRK  " >>confdefs.h
+
+else
+  enable_dss="0"
+fi
+
+if test "x$enable_dss" = "x1" ; then
+  $as_echo "#define JEMALLOC_DSS  " >>confdefs.h
+
+fi
+
+
+# Check whether --enable-fill was given.
+if test "${enable_fill+set}" = set; then :
+  enableval=$enable_fill; if test "x$enable_fill" = "xno" ; then
+  enable_fill="0"
+else
+  enable_fill="1"
+fi
+
+else
+  enable_fill="1"
+
+fi
+
+if test "x$enable_fill" = "x1" ; then
+  $as_echo "#define JEMALLOC_FILL  " >>confdefs.h
+
+fi
+
+
+# Check whether --enable-utrace was given.
+if test "${enable_utrace+set}" = set; then :
+  enableval=$enable_utrace; if test "x$enable_utrace" = "xno" ; then
+  enable_utrace="0"
+else
+  enable_utrace="1"
+fi
+
+else
+  enable_utrace="0"
+
+fi
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether utrace(2) is compilable" >&5
+$as_echo_n "checking whether utrace(2) is compilable... " >&6; }
+if ${je_cv_utrace+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+#include <sys/types.h>
+#include <sys/param.h>
+#include <sys/time.h>
+#include <sys/uio.h>
+#include <sys/ktrace.h>
+
+int
+main ()
+{
+
+	utrace((void *)0, 0);
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  je_cv_utrace=yes
+else
+  je_cv_utrace=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $je_cv_utrace" >&5
+$as_echo "$je_cv_utrace" >&6; }
+
+if test "x${je_cv_utrace}" = "xno" ; then
+  enable_utrace="0"
+fi
+if test "x$enable_utrace" = "x1" ; then
+  $as_echo "#define JEMALLOC_UTRACE  " >>confdefs.h
+
+fi
+
+
+# Check whether --enable-valgrind was given.
+if test "${enable_valgrind+set}" = set; then :
+  enableval=$enable_valgrind; if test "x$enable_valgrind" = "xno" ; then
+  enable_valgrind="0"
+else
+  enable_valgrind="1"
+fi
+
+else
+  enable_valgrind="1"
+
+fi
+
+if test "x$enable_valgrind" = "x1" ; then
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether valgrind is compilable" >&5
+$as_echo_n "checking whether valgrind is compilable... " >&6; }
+if ${je_cv_valgrind+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+#include <valgrind/valgrind.h>
+#include <valgrind/memcheck.h>
+
+#if !defined(VALGRIND_RESIZEINPLACE_BLOCK)
+#  error "Incompatible Valgrind version"
+#endif
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  je_cv_valgrind=yes
+else
+  je_cv_valgrind=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $je_cv_valgrind" >&5
+$as_echo "$je_cv_valgrind" >&6; }
+
+  if test "x${je_cv_valgrind}" = "xno" ; then
+    enable_valgrind="0"
+  fi
+  if test "x$enable_valgrind" = "x1" ; then
+    $as_echo "#define JEMALLOC_VALGRIND  " >>confdefs.h
+
+  fi
+fi
+
+
+# Check whether --enable-xmalloc was given.
+if test "${enable_xmalloc+set}" = set; then :
+  enableval=$enable_xmalloc; if test "x$enable_xmalloc" = "xno" ; then
+  enable_xmalloc="0"
+else
+  enable_xmalloc="1"
+fi
+
+else
+  enable_xmalloc="0"
+
+fi
+
+if test "x$enable_xmalloc" = "x1" ; then
+  $as_echo "#define JEMALLOC_XMALLOC  " >>confdefs.h
+
+fi
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking STATIC_PAGE_SHIFT" >&5
+$as_echo_n "checking STATIC_PAGE_SHIFT... " >&6; }
+if ${je_cv_static_page_shift+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test "$cross_compiling" = yes; then :
+  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "cannot run test program while cross compiling
+See \`config.log' for more details" "$LINENO" 5; }
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+#include <strings.h>
+#ifdef _WIN32
+#include <windows.h>
+#else
+#include <unistd.h>
+#endif
+#include <stdio.h>
+
+int
+main ()
+{
+
+    int result;
+    FILE *f;
+
+#ifdef _WIN32
+    SYSTEM_INFO si;
+    GetSystemInfo(&si);
+    result = si.dwPageSize;
+#else
+    result = sysconf(_SC_PAGESIZE);
+#endif
+    if (result == -1) {
+	return 1;
+    }
+    result = ffsl(result) - 1;
+
+    f = fopen("conftest.out", "w");
+    if (f == NULL) {
+	return 1;
+    }
+    fprintf(f, "%d\n", result);
+    fclose(f);
+
+    return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_run "$LINENO"; then :
+  je_cv_static_page_shift=`cat conftest.out`
+else
+  je_cv_static_page_shift=undefined
+fi
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
+  conftest.$ac_objext conftest.beam conftest.$ac_ext
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $je_cv_static_page_shift" >&5
+$as_echo "$je_cv_static_page_shift" >&6; }
+
+if test "x$je_cv_static_page_shift" != "xundefined"; then
+   cat >>confdefs.h <<_ACEOF
+#define STATIC_PAGE_SHIFT $je_cv_static_page_shift
+_ACEOF
+
+else
+   as_fn_error $? "cannot determine value for STATIC_PAGE_SHIFT" "$LINENO" 5
+fi
+
+
+if test -d "${srcroot}.git" ; then
+  git describe --long --abbrev=40 > ${srcroot}VERSION
+fi
+jemalloc_version=`cat ${srcroot}VERSION`
+jemalloc_version_major=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print $1}'`
+jemalloc_version_minor=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print $2}'`
+jemalloc_version_bugfix=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print $3}'`
+jemalloc_version_nrev=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print $4}'`
+jemalloc_version_gid=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print $5}'`
+
+
+
+
+
+
+
+
+if test "x$abi" != "xpecoff" ; then
+  for ac_header in pthread.h
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "pthread.h" "ac_cv_header_pthread_h" "$ac_includes_default"
+if test "x$ac_cv_header_pthread_h" = xyes; then :
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_PTHREAD_H 1
+_ACEOF
+
+else
+  as_fn_error $? "pthread.h is missing" "$LINENO" 5
+fi
+
+done
+
+      { $as_echo "$as_me:${as_lineno-$LINENO}: checking for pthread_create in -lpthread" >&5
+$as_echo_n "checking for pthread_create in -lpthread... " >&6; }
+if ${ac_cv_lib_pthread_pthread_create+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_check_lib_save_LIBS=$LIBS
+LIBS="-lpthread  $LIBS"
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char pthread_create ();
+int
+main ()
+{
+return pthread_create ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  ac_cv_lib_pthread_pthread_create=yes
+else
+  ac_cv_lib_pthread_pthread_create=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+LIBS=$ac_check_lib_save_LIBS
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_pthread_pthread_create" >&5
+$as_echo "$ac_cv_lib_pthread_pthread_create" >&6; }
+if test "x$ac_cv_lib_pthread_pthread_create" = xyes; then :
+  LIBS="$LIBS -lpthread"
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for library containing pthread_create" >&5
+$as_echo_n "checking for library containing pthread_create... " >&6; }
+if ${ac_cv_search_pthread_create+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_func_search_save_LIBS=$LIBS
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char pthread_create ();
+int
+main ()
+{
+return pthread_create ();
+  ;
+  return 0;
+}
+_ACEOF
+for ac_lib in '' ; do
+  if test -z "$ac_lib"; then
+    ac_res="none required"
+  else
+    ac_res=-l$ac_lib
+    LIBS="-l$ac_lib  $ac_func_search_save_LIBS"
+  fi
+  if ac_fn_c_try_link "$LINENO"; then :
+  ac_cv_search_pthread_create=$ac_res
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext
+  if ${ac_cv_search_pthread_create+:} false; then :
+  break
+fi
+done
+if ${ac_cv_search_pthread_create+:} false; then :
+
+else
+  ac_cv_search_pthread_create=no
+fi
+rm conftest.$ac_ext
+LIBS=$ac_func_search_save_LIBS
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_search_pthread_create" >&5
+$as_echo "$ac_cv_search_pthread_create" >&6; }
+ac_res=$ac_cv_search_pthread_create
+if test "$ac_res" != no; then :
+  test "$ac_res" = "none required" || LIBS="$ac_res $LIBS"
+
+else
+  as_fn_error $? "libpthread is missing" "$LINENO" 5
+fi
+
+fi
+
+fi
+
+CPPFLAGS="$CPPFLAGS -D_REENTRANT"
+
+ac_fn_c_check_func "$LINENO" "_malloc_thread_cleanup" "ac_cv_func__malloc_thread_cleanup"
+if test "x$ac_cv_func__malloc_thread_cleanup" = xyes; then :
+  have__malloc_thread_cleanup="1"
+else
+  have__malloc_thread_cleanup="0"
+
+fi
+
+if test "x$have__malloc_thread_cleanup" = "x1" ; then
+  $as_echo "#define JEMALLOC_MALLOC_THREAD_CLEANUP  " >>confdefs.h
+
+  force_tls="1"
+fi
+
+ac_fn_c_check_func "$LINENO" "_pthread_mutex_init_calloc_cb" "ac_cv_func__pthread_mutex_init_calloc_cb"
+if test "x$ac_cv_func__pthread_mutex_init_calloc_cb" = xyes; then :
+  have__pthread_mutex_init_calloc_cb="1"
+else
+  have__pthread_mutex_init_calloc_cb="0"
+
+fi
+
+if test "x$have__pthread_mutex_init_calloc_cb" = "x1" ; then
+  $as_echo "#define JEMALLOC_MUTEX_INIT_CB 1" >>confdefs.h
+
+fi
+
+# Check whether --enable-lazy_lock was given.
+if test "${enable_lazy_lock+set}" = set; then :
+  enableval=$enable_lazy_lock; if test "x$enable_lazy_lock" = "xno" ; then
+  enable_lazy_lock="0"
+else
+  enable_lazy_lock="1"
+fi
+
+else
+  enable_lazy_lock="0"
+
+fi
+
+if test "x$enable_lazy_lock" = "x0" -a "x${force_lazy_lock}" = "x1" ; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: Forcing lazy-lock to avoid allocator/threading bootstrap issues" >&5
+$as_echo "Forcing lazy-lock to avoid allocator/threading bootstrap issues" >&6; }
+  enable_lazy_lock="1"
+fi
+if test "x$enable_lazy_lock" = "x1" ; then
+  if test "x$abi" != "xpecoff" ; then
+    for ac_header in dlfcn.h
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "dlfcn.h" "ac_cv_header_dlfcn_h" "$ac_includes_default"
+if test "x$ac_cv_header_dlfcn_h" = xyes; then :
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_DLFCN_H 1
+_ACEOF
+
+else
+  as_fn_error $? "dlfcn.h is missing" "$LINENO" 5
+fi
+
+done
+
+    ac_fn_c_check_func "$LINENO" "dlsym" "ac_cv_func_dlsym"
+if test "x$ac_cv_func_dlsym" = xyes; then :
+
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for dlsym in -ldl" >&5
+$as_echo_n "checking for dlsym in -ldl... " >&6; }
+if ${ac_cv_lib_dl_dlsym+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_check_lib_save_LIBS=$LIBS
+LIBS="-ldl  $LIBS"
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char dlsym ();
+int
+main ()
+{
+return dlsym ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  ac_cv_lib_dl_dlsym=yes
+else
+  ac_cv_lib_dl_dlsym=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+LIBS=$ac_check_lib_save_LIBS
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_dl_dlsym" >&5
+$as_echo "$ac_cv_lib_dl_dlsym" >&6; }
+if test "x$ac_cv_lib_dl_dlsym" = xyes; then :
+  LIBS="$LIBS -ldl"
+else
+  as_fn_error $? "libdl is missing" "$LINENO" 5
+fi
+
+
+fi
+
+  fi
+  $as_echo "#define JEMALLOC_LAZY_LOCK  " >>confdefs.h
+
+fi
+
+
+# Check whether --enable-tls was given.
+if test "${enable_tls+set}" = set; then :
+  enableval=$enable_tls; if test "x$enable_tls" = "xno" ; then
+  enable_tls="0"
+else
+  enable_tls="1"
+fi
+
+else
+  enable_tls="1"
+
+fi
+
+if test "x${enable_tls}" = "x0" -a "x${force_tls}" = "x1" ; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: Forcing TLS to avoid allocator/threading bootstrap issues" >&5
+$as_echo "Forcing TLS to avoid allocator/threading bootstrap issues" >&6; }
+  enable_tls="1"
+fi
+if test "x${enable_tls}" = "x1" -a "x${force_tls}" = "x0" ; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: Forcing no TLS to avoid allocator/threading bootstrap issues" >&5
+$as_echo "Forcing no TLS to avoid allocator/threading bootstrap issues" >&6; }
+  enable_tls="0"
+fi
+if test "x${enable_tls}" = "x1" ; then
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for TLS" >&5
+$as_echo_n "checking for TLS... " >&6; }
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+    __thread int x;
+
+int
+main ()
+{
+
+    x = 42;
+
+    return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+              enable_tls="0"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+
+if test "x${enable_tls}" = "x1" ; then
+  cat >>confdefs.h <<_ACEOF
+#define JEMALLOC_TLS
+_ACEOF
+
+elif test "x${force_tls}" = "x1" ; then
+  as_fn_error $? "Failed to configure TLS, which is mandatory for correct function" "$LINENO" 5
+fi
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether a program using ffsl is compilable" >&5
+$as_echo_n "checking whether a program using ffsl is compilable... " >&6; }
+if ${je_cv_function_ffsl+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+#include <stdio.h>
+#include <strings.h>
+#include <string.h>
+
+int
+main ()
+{
+
+	{
+		int rv = ffsl(0x08);
+		printf("%d\n", rv);
+	}
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  je_cv_function_ffsl=yes
+else
+  je_cv_function_ffsl=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $je_cv_function_ffsl" >&5
+$as_echo "$je_cv_function_ffsl" >&6; }
+
+if test "x${je_cv_function_ffsl}" != "xyes" ; then
+   as_fn_error $? "Cannot build without ffsl(3)" "$LINENO" 5
+fi
+
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether atomic(9) is compilable" >&5
+$as_echo_n "checking whether atomic(9) is compilable... " >&6; }
+if ${je_cv_atomic9+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+#include <sys/types.h>
+#include <machine/atomic.h>
+#include <inttypes.h>
+
+int
+main ()
+{
+
+	{
+		uint32_t x32 = 0;
+		volatile uint32_t *x32p = &x32;
+		atomic_fetchadd_32(x32p, 1);
+	}
+	{
+		unsigned long xlong = 0;
+		volatile unsigned long *xlongp = &xlong;
+		atomic_fetchadd_long(xlongp, 1);
+	}
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  je_cv_atomic9=yes
+else
+  je_cv_atomic9=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $je_cv_atomic9" >&5
+$as_echo "$je_cv_atomic9" >&6; }
+
+if test "x${je_cv_atomic9}" = "xyes" ; then
+  $as_echo "#define JEMALLOC_ATOMIC9 1" >>confdefs.h
+
+fi
+
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether Darwin OSAtomic*() is compilable" >&5
+$as_echo_n "checking whether Darwin OSAtomic*() is compilable... " >&6; }
+if ${je_cv_osatomic+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+#include <libkern/OSAtomic.h>
+#include <inttypes.h>
+
+int
+main ()
+{
+
+	{
+		int32_t x32 = 0;
+		volatile int32_t *x32p = &x32;
+		OSAtomicAdd32(1, x32p);
+	}
+	{
+		int64_t x64 = 0;
+		volatile int64_t *x64p = &x64;
+		OSAtomicAdd64(1, x64p);
+	}
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  je_cv_osatomic=yes
+else
+  je_cv_osatomic=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $je_cv_osatomic" >&5
+$as_echo "$je_cv_osatomic" >&6; }
+
+if test "x${je_cv_osatomic}" = "xyes" ; then
+  $as_echo "#define JEMALLOC_OSATOMIC  " >>confdefs.h
+
+fi
+
+
+
+
+if test "x${je_cv_atomic9}" != "xyes" -a "x${je_cv_osatomic}" != "xyes" ; then
+
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether to force 32-bit __sync_{add,sub}_and_fetch()" >&5
+$as_echo_n "checking whether to force 32-bit __sync_{add,sub}_and_fetch()... " >&6; }
+if ${je_cv_sync_compare_and_swap_4+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+                                                 #include <stdint.h>
+
+int
+main ()
+{
+
+                                                 #ifndef __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4
+                                                 {
+                                                    uint32_t x32 = 0;
+                                                    __sync_add_and_fetch(&x32, 42);
+                                                    __sync_sub_and_fetch(&x32, 1);
+                                                 }
+                                                 #else
+                                                 #error __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4 is defined, no need to force
+                                                 #endif
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  je_cv_sync_compare_and_swap_4=yes
+else
+  je_cv_sync_compare_and_swap_4=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $je_cv_sync_compare_and_swap_4" >&5
+$as_echo "$je_cv_sync_compare_and_swap_4" >&6; }
+
+  if test "x${je_cv_sync_compare_and_swap_4}" = "xyes" ; then
+    $as_echo "#define JE_FORCE_SYNC_COMPARE_AND_SWAP_4  " >>confdefs.h
+
+  fi
+
+
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether to force 64-bit __sync_{add,sub}_and_fetch()" >&5
+$as_echo_n "checking whether to force 64-bit __sync_{add,sub}_and_fetch()... " >&6; }
+if ${je_cv_sync_compare_and_swap_8+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+                                                 #include <stdint.h>
+
+int
+main ()
+{
+
+                                                 #ifndef __GCC_HAVE_SYNC_COMPARE_AND_SWAP_8
+                                                 {
+                                                    uint64_t x64 = 0;
+                                                    __sync_add_and_fetch(&x64, 42);
+                                                    __sync_sub_and_fetch(&x64, 1);
+                                                 }
+                                                 #else
+                                                 #error __GCC_HAVE_SYNC_COMPARE_AND_SWAP_8 is defined, no need to force
+                                                 #endif
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  je_cv_sync_compare_and_swap_8=yes
+else
+  je_cv_sync_compare_and_swap_8=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $je_cv_sync_compare_and_swap_8" >&5
+$as_echo "$je_cv_sync_compare_and_swap_8" >&6; }
+
+  if test "x${je_cv_sync_compare_and_swap_8}" = "xyes" ; then
+    $as_echo "#define JE_FORCE_SYNC_COMPARE_AND_SWAP_8  " >>confdefs.h
+
+  fi
+
+fi
+
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether Darwin OSSpin*() is compilable" >&5
+$as_echo_n "checking whether Darwin OSSpin*() is compilable... " >&6; }
+if ${je_cv_osspin+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+#include <libkern/OSAtomic.h>
+#include <inttypes.h>
+
+int
+main ()
+{
+
+	OSSpinLock lock = 0;
+	OSSpinLockLock(&lock);
+	OSSpinLockUnlock(&lock);
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  je_cv_osspin=yes
+else
+  je_cv_osspin=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $je_cv_osspin" >&5
+$as_echo "$je_cv_osspin" >&6; }
+
+if test "x${je_cv_osspin}" = "xyes" ; then
+  $as_echo "#define JEMALLOC_OSSPIN  " >>confdefs.h
+
+fi
+
+
+# Check whether --enable-zone-allocator was given.
+if test "${enable_zone_allocator+set}" = set; then :
+  enableval=$enable_zone_allocator; if test "x$enable_zone_allocator" = "xno" ; then
+  enable_zone_allocator="0"
+else
+  enable_zone_allocator="1"
+fi
+
+else
+  if test "x${abi}" = "xmacho"; then
+  enable_zone_allocator="1"
+fi
+
+
+fi
+
+
+
+if test "x${enable_zone_allocator}" = "x1" ; then
+  if test "x${abi}" != "xmacho"; then
+    as_fn_error $? "--enable-zone-allocator is only supported on Darwin" "$LINENO" 5
+  fi
+  $as_echo "#define JEMALLOC_IVSALLOC  " >>confdefs.h
+
+  $as_echo "#define JEMALLOC_ZONE  " >>confdefs.h
+
+
+        { $as_echo "$as_me:${as_lineno-$LINENO}: checking malloc zone version" >&5
+$as_echo_n "checking malloc zone version... " >&6; }
+
+
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <malloc/malloc.h>
+int
+main ()
+{
+static foo[sizeof(malloc_zone_t) == sizeof(void *) * 14 ? 1 : -1]
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  JEMALLOC_ZONE_VERSION=3
+else
+
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <malloc/malloc.h>
+int
+main ()
+{
+static foo[sizeof(malloc_zone_t) == sizeof(void *) * 15 ? 1 : -1]
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  JEMALLOC_ZONE_VERSION=5
+else
+
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <malloc/malloc.h>
+int
+main ()
+{
+static foo[sizeof(malloc_zone_t) == sizeof(void *) * 16 ? 1 : -1]
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+
+    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <malloc/malloc.h>
+int
+main ()
+{
+static foo[sizeof(malloc_introspection_t) == sizeof(void *) * 9 ? 1 : -1]
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  JEMALLOC_ZONE_VERSION=6
+else
+
+    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <malloc/malloc.h>
+int
+main ()
+{
+static foo[sizeof(malloc_introspection_t) == sizeof(void *) * 13 ? 1 : -1]
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  JEMALLOC_ZONE_VERSION=7
+else
+  JEMALLOC_ZONE_VERSION=
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+else
+
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <malloc/malloc.h>
+int
+main ()
+{
+static foo[sizeof(malloc_zone_t) == sizeof(void *) * 17 ? 1 : -1]
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  JEMALLOC_ZONE_VERSION=8
+else
+
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <malloc/malloc.h>
+int
+main ()
+{
+static foo[sizeof(malloc_zone_t) > sizeof(void *) * 17 ? 1 : -1]
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  JEMALLOC_ZONE_VERSION=9
+else
+  JEMALLOC_ZONE_VERSION=
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+  if test "x${JEMALLOC_ZONE_VERSION}" = "x"; then
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: unsupported" >&5
+$as_echo "unsupported" >&6; }
+    as_fn_error $? "Unsupported malloc zone version" "$LINENO" 5
+  fi
+  if test "${JEMALLOC_ZONE_VERSION}" = 9; then
+    JEMALLOC_ZONE_VERSION=8
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: > 8" >&5
+$as_echo "> 8" >&6; }
+  else
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $JEMALLOC_ZONE_VERSION" >&5
+$as_echo "$JEMALLOC_ZONE_VERSION" >&6; }
+  fi
+  cat >>confdefs.h <<_ACEOF
+#define JEMALLOC_ZONE_VERSION $JEMALLOC_ZONE_VERSION
+_ACEOF
+
+fi
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for stdbool.h that conforms to C99" >&5
+$as_echo_n "checking for stdbool.h that conforms to C99... " >&6; }
+if ${ac_cv_header_stdbool_h+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+#include <stdbool.h>
+#ifndef bool
+ "error: bool is not defined"
+#endif
+#ifndef false
+ "error: false is not defined"
+#endif
+#if false
+ "error: false is not 0"
+#endif
+#ifndef true
+ "error: true is not defined"
+#endif
+#if true != 1
+ "error: true is not 1"
+#endif
+#ifndef __bool_true_false_are_defined
+ "error: __bool_true_false_are_defined is not defined"
+#endif
+
+	struct s { _Bool s: 1; _Bool t; } s;
+
+	char a[true == 1 ? 1 : -1];
+	char b[false == 0 ? 1 : -1];
+	char c[__bool_true_false_are_defined == 1 ? 1 : -1];
+	char d[(bool) 0.5 == true ? 1 : -1];
+	/* See body of main program for 'e'.  */
+	char f[(_Bool) 0.0 == false ? 1 : -1];
+	char g[true];
+	char h[sizeof (_Bool)];
+	char i[sizeof s.t];
+	enum { j = false, k = true, l = false * true, m = true * 256 };
+	/* The following fails for
+	   HP aC++/ANSI C B3910B A.05.55 [Dec 04 2003]. */
+	_Bool n[m];
+	char o[sizeof n == m * sizeof n[0] ? 1 : -1];
+	char p[-1 - (_Bool) 0 < 0 && -1 - (bool) 0 < 0 ? 1 : -1];
+	/* Catch a bug in an HP-UX C compiler.  See
+	   http://gcc.gnu.org/ml/gcc-patches/2003-12/msg02303.html
+	   http://lists.gnu.org/archive/html/bug-coreutils/2005-11/msg00161.html
+	 */
+	_Bool q = true;
+	_Bool *pq = &q;
+
+int
+main ()
+{
+
+	bool e = &s;
+	*pq |= q;
+	*pq |= ! q;
+	/* Refer to every declared value, to avoid compiler optimizations.  */
+	return (!a + !b + !c + !d + !e + !f + !g + !h + !i + !!j + !k + !!l
+		+ !m + !n + !o + !p + !q + !pq);
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_cv_header_stdbool_h=yes
+else
+  ac_cv_header_stdbool_h=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_header_stdbool_h" >&5
+$as_echo "$ac_cv_header_stdbool_h" >&6; }
+ac_fn_c_check_type "$LINENO" "_Bool" "ac_cv_type__Bool" "$ac_includes_default"
+if test "x$ac_cv_type__Bool" = xyes; then :
+
+cat >>confdefs.h <<_ACEOF
+#define HAVE__BOOL 1
+_ACEOF
+
+
+fi
+
+if test $ac_cv_header_stdbool_h = yes; then
+
+$as_echo "#define HAVE_STDBOOL_H 1" >>confdefs.h
+
+fi
+
+
+ac_config_commands="$ac_config_commands include/jemalloc/internal/size_classes.h"
+
+
+
+
+ac_config_headers="$ac_config_headers $cfghdrs_tup"
+
+
+ac_config_files="$ac_config_files $cfgoutputs_tup config.stamp bin/jemalloc.sh"
+
+
+
+cat >confcache <<\_ACEOF
+# This file is a shell script that caches the results of configure
+# tests run on this system so they can be shared between configure
+# scripts and configure runs, see configure's option --config-cache.
+# It is not useful on other systems.  If it contains results you don't
+# want to keep, you may remove or edit it.
+#
+# config.status only pays attention to the cache file if you give it
+# the --recheck option to rerun configure.
+#
+# `ac_cv_env_foo' variables (set or unset) will be overridden when
+# loading this file, other *unset* `ac_cv_foo' will be assigned the
+# following values.
+
+_ACEOF
+
+# The following way of writing the cache mishandles newlines in values,
+# but we know of no workaround that is simple, portable, and efficient.
+# So, we kill variables containing newlines.
+# Ultrix sh set writes to stderr and can't be redirected directly,
+# and sets the high bit in the cache file unless we assign to the vars.
+(
+  for ac_var in `(set) 2>&1 | sed -n 's/^\([a-zA-Z_][a-zA-Z0-9_]*\)=.*/\1/p'`; do
+    eval ac_val=\$$ac_var
+    case $ac_val in #(
+    *${as_nl}*)
+      case $ac_var in #(
+      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
+$as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
+      esac
+      case $ac_var in #(
+      _ | IFS | as_nl) ;; #(
+      BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
+      *) { eval $ac_var=; unset $ac_var;} ;;
+      esac ;;
+    esac
+  done
+
+  (set) 2>&1 |
+    case $as_nl`(ac_space=' '; set) 2>&1` in #(
+    *${as_nl}ac_space=\ *)
+      # `set' does not quote correctly, so add quotes: double-quote
+      # substitution turns \\\\ into \\, and sed turns \\ into \.
+      sed -n \
+	"s/'/'\\\\''/g;
+	  s/^\\([_$as_cr_alnum]*_cv_[_$as_cr_alnum]*\\)=\\(.*\\)/\\1='\\2'/p"
+      ;; #(
+    *)
+      # `set' quotes correctly as required by POSIX, so do not add quotes.
+      sed -n "/^[_$as_cr_alnum]*_cv_[_$as_cr_alnum]*=/p"
+      ;;
+    esac |
+    sort
+) |
+  sed '
+     /^ac_cv_env_/b end
+     t clear
+     :clear
+     s/^\([^=]*\)=\(.*[{}].*\)$/test "${\1+set}" = set || &/
+     t end
+     s/^\([^=]*\)=\(.*\)$/\1=${\1=\2}/
+     :end' >>confcache
+if diff "$cache_file" confcache >/dev/null 2>&1; then :; else
+  if test -w "$cache_file"; then
+    if test "x$cache_file" != "x/dev/null"; then
+      { $as_echo "$as_me:${as_lineno-$LINENO}: updating cache $cache_file" >&5
+$as_echo "$as_me: updating cache $cache_file" >&6;}
+      if test ! -f "$cache_file" || test -h "$cache_file"; then
+	cat confcache >"$cache_file"
+      else
+        case $cache_file in #(
+        */* | ?:*)
+	  mv -f confcache "$cache_file"$$ &&
+	  mv -f "$cache_file"$$ "$cache_file" ;; #(
+        *)
+	  mv -f confcache "$cache_file" ;;
+	esac
+      fi
+    fi
+  else
+    { $as_echo "$as_me:${as_lineno-$LINENO}: not updating unwritable cache $cache_file" >&5
+$as_echo "$as_me: not updating unwritable cache $cache_file" >&6;}
+  fi
+fi
+rm -f confcache
+
+test "x$prefix" = xNONE && prefix=$ac_default_prefix
+# Let make expand exec_prefix.
+test "x$exec_prefix" = xNONE && exec_prefix='${prefix}'
+
+DEFS=-DHAVE_CONFIG_H
+
+ac_libobjs=
+ac_ltlibobjs=
+U=
+for ac_i in : $LIBOBJS; do test "x$ac_i" = x: && continue
+  # 1. Remove the extension, and $U if already installed.
+  ac_script='s/\$U\././;s/\.o$//;s/\.obj$//'
+  ac_i=`$as_echo "$ac_i" | sed "$ac_script"`
+  # 2. Prepend LIBOBJDIR.  When used with automake>=1.10 LIBOBJDIR
+  #    will be set to the directory where LIBOBJS objects are built.
+  as_fn_append ac_libobjs " \${LIBOBJDIR}$ac_i\$U.$ac_objext"
+  as_fn_append ac_ltlibobjs " \${LIBOBJDIR}$ac_i"'$U.lo'
+done
+LIBOBJS=$ac_libobjs
+
+LTLIBOBJS=$ac_ltlibobjs
+
+
+
+: "${CONFIG_STATUS=./config.status}"
+ac_write_fail=0
+ac_clean_files_save=$ac_clean_files
+ac_clean_files="$ac_clean_files $CONFIG_STATUS"
+{ $as_echo "$as_me:${as_lineno-$LINENO}: creating $CONFIG_STATUS" >&5
+$as_echo "$as_me: creating $CONFIG_STATUS" >&6;}
+as_write_fail=0
+cat >$CONFIG_STATUS <<_ASEOF || as_write_fail=1
+#! $SHELL
+# Generated by $as_me.
+# Run this file to recreate the current configuration.
+# Compiler output produced by configure, useful for debugging
+# configure, is in config.log if it exists.
+
+debug=false
+ac_cs_recheck=false
+ac_cs_silent=false
+
+SHELL=\${CONFIG_SHELL-$SHELL}
+export SHELL
+_ASEOF
+cat >>$CONFIG_STATUS <<\_ASEOF || as_write_fail=1
+## -------------------- ##
+## M4sh Initialization. ##
+## -------------------- ##
+
+# Be more Bourne compatible
+DUALCASE=1; export DUALCASE # for MKS sh
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '${1+"$@"}'='"$@"'
+  setopt NO_GLOB_SUBST
+else
+  case `(set -o) 2>/dev/null` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
+esac
+fi
+
+
+as_nl='
+'
+export as_nl
+# Printing a long string crashes Solaris 7 /usr/bin/printf.
+as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
+as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
+as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
+# Prefer a ksh shell builtin over an external printf program on Solaris,
+# but without wasting forks for bash or zsh.
+if test -z "$BASH_VERSION$ZSH_VERSION" \
+    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
+  as_echo='print -r --'
+  as_echo_n='print -rn --'
+elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+  as_echo='printf %s\n'
+  as_echo_n='printf %s'
+else
+  if test "X`(/usr/ucb/echo -n -n $as_echo) 2>/dev/null`" = "X-n $as_echo"; then
+    as_echo_body='eval /usr/ucb/echo -n "$1$as_nl"'
+    as_echo_n='/usr/ucb/echo -n'
+  else
+    as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
+    as_echo_n_body='eval
+      arg=$1;
+      case $arg in #(
+      *"$as_nl"*)
+	expr "X$arg" : "X\\(.*\\)$as_nl";
+	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
+      esac;
+      expr "X$arg" : "X\\(.*\\)" | tr -d "$as_nl"
+    '
+    export as_echo_n_body
+    as_echo_n='sh -c $as_echo_n_body as_echo'
+  fi
+  export as_echo_body
+  as_echo='sh -c $as_echo_body as_echo'
+fi
+
+# The user is always right.
+if test "${PATH_SEPARATOR+set}" != set; then
+  PATH_SEPARATOR=:
+  (PATH='/bin;/bin'; FPATH=$PATH; sh -c :) >/dev/null 2>&1 && {
+    (PATH='/bin:/bin'; FPATH=$PATH; sh -c :) >/dev/null 2>&1 ||
+      PATH_SEPARATOR=';'
+  }
+fi
+
+
+# IFS
+# We need space, tab and new line, in precisely that order.  Quoting is
+# there to prevent editors from complaining about space-tab.
+# (If _AS_PATH_WALK were called with IFS unset, it would disable word
+# splitting by setting IFS to empty value.)
+IFS=" ""	$as_nl"
+
+# Find who we are.  Look in the path if we contain no directory separator.
+as_myself=
+case $0 in #((
+  *[\\/]* ) as_myself=$0 ;;
+  *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+  done
+IFS=$as_save_IFS
+
+     ;;
+esac
+# We did not find ourselves, most probably we were run as `sh COMMAND'
+# in which case we are not to be found in the path.
+if test "x$as_myself" = x; then
+  as_myself=$0
+fi
+if test ! -f "$as_myself"; then
+  $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
+  exit 1
+fi
+
+# Unset variables that we do not need and which cause bugs (e.g. in
+# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
+# suppresses any "Segmentation fault" message there.  '((' could
+# trigger a bug in pdksh 5.2.14.
+for as_var in BASH_ENV ENV MAIL MAILPATH
+do eval test x\${$as_var+set} = xset \
+  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
+done
+PS1='$ '
+PS2='> '
+PS4='+ '
+
+# NLS nuisances.
+LC_ALL=C
+export LC_ALL
+LANGUAGE=C
+export LANGUAGE
+
+# CDPATH.
+(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
+
+
+# as_fn_error STATUS ERROR [LINENO LOG_FD]
+# ----------------------------------------
+# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
+# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
+# script with STATUS, using 1 if that was 0.
+as_fn_error ()
+{
+  as_status=$1; test $as_status -eq 0 && as_status=1
+  if test "$4"; then
+    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
+  fi
+  $as_echo "$as_me: error: $2" >&2
+  as_fn_exit $as_status
+} # as_fn_error
+
+
+# as_fn_set_status STATUS
+# -----------------------
+# Set $? to STATUS, without forking.
+as_fn_set_status ()
+{
+  return $1
+} # as_fn_set_status
+
+# as_fn_exit STATUS
+# -----------------
+# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
+as_fn_exit ()
+{
+  set +e
+  as_fn_set_status $1
+  exit $1
+} # as_fn_exit
+
+# as_fn_unset VAR
+# ---------------
+# Portably unset VAR.
+as_fn_unset ()
+{
+  { eval $1=; unset $1;}
+}
+as_unset=as_fn_unset
+# as_fn_append VAR VALUE
+# ----------------------
+# Append the text in VALUE to the end of the definition contained in VAR. Take
+# advantage of any shell optimizations that allow amortized linear growth over
+# repeated appends, instead of the typical quadratic growth present in naive
+# implementations.
+if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
+  eval 'as_fn_append ()
+  {
+    eval $1+=\$2
+  }'
+else
+  as_fn_append ()
+  {
+    eval $1=\$$1\$2
+  }
+fi # as_fn_append
+
+# as_fn_arith ARG...
+# ------------------
+# Perform arithmetic evaluation on the ARGs, and store the result in the
+# global $as_val. Take advantage of shells that can avoid forks. The arguments
+# must be portable across $(()) and expr.
+if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
+  eval 'as_fn_arith ()
+  {
+    as_val=$(( $* ))
+  }'
+else
+  as_fn_arith ()
+  {
+    as_val=`expr "$@" || test $? -eq 1`
+  }
+fi # as_fn_arith
+
+
+if expr a : '\(a\)' >/dev/null 2>&1 &&
+   test "X`expr 00001 : '.*\(...\)'`" = X001; then
+  as_expr=expr
+else
+  as_expr=false
+fi
+
+if (basename -- /) >/dev/null 2>&1 && test "X`basename -- / 2>&1`" = "X/"; then
+  as_basename=basename
+else
+  as_basename=false
+fi
+
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
+
+as_me=`$as_basename -- "$0" ||
+$as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
+	 X"$0" : 'X\(//\)$' \| \
+	 X"$0" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X/"$0" |
+    sed '/^.*\/\([^/][^/]*\)\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\/\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\/\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
+
+ECHO_C= ECHO_N= ECHO_T=
+case `echo -n x` in #(((((
+-n*)
+  case `echo 'xy\c'` in
+  *c*) ECHO_T='	';;	# ECHO_T is single tab character.
+  xy)  ECHO_C='\c';;
+  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
+       ECHO_T='	';;
+  esac;;
+*)
+  ECHO_N='-n';;
+esac
+
+rm -f conf$$ conf$$.exe conf$$.file
+if test -d conf$$.dir; then
+  rm -f conf$$.dir/conf$$.file
+else
+  rm -f conf$$.dir
+  mkdir conf$$.dir 2>/dev/null
+fi
+if (echo >conf$$.file) 2>/dev/null; then
+  if ln -s conf$$.file conf$$ 2>/dev/null; then
+    as_ln_s='ln -s'
+    # ... but there are two gotchas:
+    # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
+    # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
+    # In both cases, we have to default to `cp -p'.
+    ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
+      as_ln_s='cp -p'
+  elif ln conf$$.file conf$$ 2>/dev/null; then
+    as_ln_s=ln
+  else
+    as_ln_s='cp -p'
+  fi
+else
+  as_ln_s='cp -p'
+fi
+rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
+rmdir conf$$.dir 2>/dev/null
+
+
+# as_fn_mkdir_p
+# -------------
+# Create "$as_dir" as a directory, including parents if necessary.
+as_fn_mkdir_p ()
+{
+
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || eval $as_mkdir_p || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
+
+
+} # as_fn_mkdir_p
+if mkdir -p . 2>/dev/null; then
+  as_mkdir_p='mkdir -p "$as_dir"'
+else
+  test -d ./-p && rmdir ./-p
+  as_mkdir_p=false
+fi
+
+if test -x / >/dev/null 2>&1; then
+  as_test_x='test -x'
+else
+  if ls -dL / >/dev/null 2>&1; then
+    as_ls_L_option=L
+  else
+    as_ls_L_option=
+  fi
+  as_test_x='
+    eval sh -c '\''
+      if test -d "$1"; then
+	test -d "$1/.";
+      else
+	case $1 in #(
+	-*)set "./$1";;
+	esac;
+	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in #((
+	???[sx]*):;;*)false;;esac;fi
+    '\'' sh
+  '
+fi
+as_executable_p=$as_test_x
+
+# Sed expression to map a string onto a valid CPP name.
+as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
+
+# Sed expression to map a string onto a valid variable name.
+as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
+
+
+exec 6>&1
+## ----------------------------------- ##
+## Main body of $CONFIG_STATUS script. ##
+## ----------------------------------- ##
+_ASEOF
+test $as_write_fail = 0 && chmod +x $CONFIG_STATUS || ac_write_fail=1
+
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+# Save the log message, to keep $0 and so on meaningful, and to
+# report actual input values of CONFIG_FILES etc. instead of their
+# values after options handling.
+ac_log="
+This file was extended by $as_me, which was
+generated by GNU Autoconf 2.68.  Invocation command line was
+
+  CONFIG_FILES    = $CONFIG_FILES
+  CONFIG_HEADERS  = $CONFIG_HEADERS
+  CONFIG_LINKS    = $CONFIG_LINKS
+  CONFIG_COMMANDS = $CONFIG_COMMANDS
+  $ $0 $@
+
+on `(hostname || uname -n) 2>/dev/null | sed 1q`
+"
+
+_ACEOF
+
+case $ac_config_files in *"
+"*) set x $ac_config_files; shift; ac_config_files=$*;;
+esac
+
+case $ac_config_headers in *"
+"*) set x $ac_config_headers; shift; ac_config_headers=$*;;
+esac
+
+
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+# Files that config.status was made for.
+config_files="$ac_config_files"
+config_headers="$ac_config_headers"
+config_commands="$ac_config_commands"
+
+_ACEOF
+
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+ac_cs_usage="\
+\`$as_me' instantiates files and other configuration actions
+from templates according to the current configuration.  Unless the files
+and actions are specified as TAGs, all are instantiated by default.
+
+Usage: $0 [OPTION]... [TAG]...
+
+  -h, --help       print this help, then exit
+  -V, --version    print version number and configuration settings, then exit
+      --config     print configuration, then exit
+  -q, --quiet, --silent
+                   do not print progress messages
+  -d, --debug      don't remove temporary files
+      --recheck    update $as_me by reconfiguring in the same conditions
+      --file=FILE[:TEMPLATE]
+                   instantiate the configuration file FILE
+      --header=FILE[:TEMPLATE]
+                   instantiate the configuration header FILE
+
+Configuration files:
+$config_files
+
+Configuration headers:
+$config_headers
+
+Configuration commands:
+$config_commands
+
+Report bugs to the package provider."
+
+_ACEOF
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+ac_cs_config="`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`"
+ac_cs_version="\\
+config.status
+configured by $0, generated by GNU Autoconf 2.68,
+  with options \\"\$ac_cs_config\\"
+
+Copyright (C) 2010 Free Software Foundation, Inc.
+This config.status script is free software; the Free Software Foundation
+gives unlimited permission to copy, distribute and modify it."
+
+ac_pwd='$ac_pwd'
+srcdir='$srcdir'
+INSTALL='$INSTALL'
+test -n "\$AWK" || AWK=awk
+_ACEOF
+
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+# The default lists apply if the user does not specify any file.
+ac_need_defaults=:
+while test $# != 0
+do
+  case $1 in
+  --*=?*)
+    ac_option=`expr "X$1" : 'X\([^=]*\)='`
+    ac_optarg=`expr "X$1" : 'X[^=]*=\(.*\)'`
+    ac_shift=:
+    ;;
+  --*=)
+    ac_option=`expr "X$1" : 'X\([^=]*\)='`
+    ac_optarg=
+    ac_shift=:
+    ;;
+  *)
+    ac_option=$1
+    ac_optarg=$2
+    ac_shift=shift
+    ;;
+  esac
+
+  case $ac_option in
+  # Handling of the options.
+  -recheck | --recheck | --rechec | --reche | --rech | --rec | --re | --r)
+    ac_cs_recheck=: ;;
+  --version | --versio | --versi | --vers | --ver | --ve | --v | -V )
+    $as_echo "$ac_cs_version"; exit ;;
+  --config | --confi | --conf | --con | --co | --c )
+    $as_echo "$ac_cs_config"; exit ;;
+  --debug | --debu | --deb | --de | --d | -d )
+    debug=: ;;
+  --file | --fil | --fi | --f )
+    $ac_shift
+    case $ac_optarg in
+    *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
+    '') as_fn_error $? "missing file argument" ;;
+    esac
+    as_fn_append CONFIG_FILES " '$ac_optarg'"
+    ac_need_defaults=false;;
+  --header | --heade | --head | --hea )
+    $ac_shift
+    case $ac_optarg in
+    *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
+    esac
+    as_fn_append CONFIG_HEADERS " '$ac_optarg'"
+    ac_need_defaults=false;;
+  --he | --h)
+    # Conflict between --help and --header
+    as_fn_error $? "ambiguous option: \`$1'
+Try \`$0 --help' for more information.";;
+  --help | --hel | -h )
+    $as_echo "$ac_cs_usage"; exit ;;
+  -q | -quiet | --quiet | --quie | --qui | --qu | --q \
+  | -silent | --silent | --silen | --sile | --sil | --si | --s)
+    ac_cs_silent=: ;;
+
+  # This is an error.
+  -*) as_fn_error $? "unrecognized option: \`$1'
+Try \`$0 --help' for more information." ;;
+
+  *) as_fn_append ac_config_targets " $1"
+     ac_need_defaults=false ;;
+
+  esac
+  shift
+done
+
+ac_configure_extra_args=
+
+if $ac_cs_silent; then
+  exec 6>/dev/null
+  ac_configure_extra_args="$ac_configure_extra_args --silent"
+fi
+
+_ACEOF
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+if \$ac_cs_recheck; then
+  set X '$SHELL' '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
+  shift
+  \$as_echo "running CONFIG_SHELL=$SHELL \$*" >&6
+  CONFIG_SHELL='$SHELL'
+  export CONFIG_SHELL
+  exec "\$@"
+fi
+
+_ACEOF
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+exec 5>>config.log
+{
+  echo
+  sed 'h;s/./-/g;s/^.../## /;s/...$/ ##/;p;x;p;x' <<_ASBOX
+## Running $as_me. ##
+_ASBOX
+  $as_echo "$ac_log"
+} >&5
+
+_ACEOF
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+_ACEOF
+
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+
+# Handling of arguments.
+for ac_config_target in $ac_config_targets
+do
+  case $ac_config_target in
+    "include/jemalloc/internal/size_classes.h") CONFIG_COMMANDS="$CONFIG_COMMANDS include/jemalloc/internal/size_classes.h" ;;
+    "$cfghdrs_tup") CONFIG_HEADERS="$CONFIG_HEADERS $cfghdrs_tup" ;;
+    "$cfgoutputs_tup") CONFIG_FILES="$CONFIG_FILES $cfgoutputs_tup" ;;
+    "config.stamp") CONFIG_FILES="$CONFIG_FILES config.stamp" ;;
+    "bin/jemalloc.sh") CONFIG_FILES="$CONFIG_FILES bin/jemalloc.sh" ;;
+
+  *) as_fn_error $? "invalid argument: \`$ac_config_target'" "$LINENO" 5;;
+  esac
+done
+
+
+# If the user did not use the arguments to specify the items to instantiate,
+# then the envvar interface is used.  Set only those that are not.
+# We use the long form for the default assignment because of an extremely
+# bizarre bug on SunOS 4.1.3.
+if $ac_need_defaults; then
+  test "${CONFIG_FILES+set}" = set || CONFIG_FILES=$config_files
+  test "${CONFIG_HEADERS+set}" = set || CONFIG_HEADERS=$config_headers
+  test "${CONFIG_COMMANDS+set}" = set || CONFIG_COMMANDS=$config_commands
+fi
+
+# Have a temporary directory for convenience.  Make it in the build tree
+# simply because there is no reason against having it here, and in addition,
+# creating and moving files from /tmp can sometimes cause problems.
+# Hook for its removal unless debugging.
+# Note that there is a small window in which the directory will not be cleaned:
+# after its creation but before its name has been assigned to `$tmp'.
+$debug ||
+{
+  tmp= ac_tmp=
+  trap 'exit_status=$?
+  : "${ac_tmp:=$tmp}"
+  { test ! -d "$ac_tmp" || rm -fr "$ac_tmp"; } && exit $exit_status
+' 0
+  trap 'as_fn_exit 1' 1 2 13 15
+}
+# Create a (secure) tmp directory for tmp files.
+
+{
+  tmp=`(umask 077 && mktemp -d "./confXXXXXX") 2>/dev/null` &&
+  test -d "$tmp"
+}  ||
+{
+  tmp=./conf$$-$RANDOM
+  (umask 077 && mkdir "$tmp")
+} || as_fn_error $? "cannot create a temporary directory in ." "$LINENO" 5
+ac_tmp=$tmp
+
+# Set up the scripts for CONFIG_FILES section.
+# No need to generate them if there are no CONFIG_FILES.
+# This happens for instance with `./config.status config.h'.
+if test -n "$CONFIG_FILES"; then
+
+
+ac_cr=`echo X | tr X '\015'`
+# On cygwin, bash can eat \r inside `` if the user requested igncr.
+# But we know of no other shell where ac_cr would be empty at this
+# point, so we can use a bashism as a fallback.
+if test "x$ac_cr" = x; then
+  eval ac_cr=\$\'\\r\'
+fi
+ac_cs_awk_cr=`$AWK 'BEGIN { print "a\rb" }' </dev/null 2>/dev/null`
+if test "$ac_cs_awk_cr" = "a${ac_cr}b"; then
+  ac_cs_awk_cr='\\r'
+else
+  ac_cs_awk_cr=$ac_cr
+fi
+
+echo 'BEGIN {' >"$ac_tmp/subs1.awk" &&
+_ACEOF
+
+
+{
+  echo "cat >conf$$subs.awk <<_ACEOF" &&
+  echo "$ac_subst_vars" | sed 's/.*/&!$&$ac_delim/' &&
+  echo "_ACEOF"
+} >conf$$subs.sh ||
+  as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+ac_delim_num=`echo "$ac_subst_vars" | grep -c '^'`
+ac_delim='%!_!# '
+for ac_last_try in false false false false false :; do
+  . ./conf$$subs.sh ||
+    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+
+  ac_delim_n=`sed -n "s/.*$ac_delim\$/X/p" conf$$subs.awk | grep -c X`
+  if test $ac_delim_n = $ac_delim_num; then
+    break
+  elif $ac_last_try; then
+    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+  else
+    ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
+  fi
+done
+rm -f conf$$subs.sh
+
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+cat >>"\$ac_tmp/subs1.awk" <<\\_ACAWK &&
+_ACEOF
+sed -n '
+h
+s/^/S["/; s/!.*/"]=/
+p
+g
+s/^[^!]*!//
+:repl
+t repl
+s/'"$ac_delim"'$//
+t delim
+:nl
+h
+s/\(.\{148\}\)..*/\1/
+t more1
+s/["\\]/\\&/g; s/^/"/; s/$/\\n"\\/
+p
+n
+b repl
+:more1
+s/["\\]/\\&/g; s/^/"/; s/$/"\\/
+p
+g
+s/.\{148\}//
+t nl
+:delim
+h
+s/\(.\{148\}\)..*/\1/
+t more2
+s/["\\]/\\&/g; s/^/"/; s/$/"/
+p
+b
+:more2
+s/["\\]/\\&/g; s/^/"/; s/$/"\\/
+p
+g
+s/.\{148\}//
+t delim
+' <conf$$subs.awk | sed '
+/^[^""]/{
+  N
+  s/\n//
+}
+' >>$CONFIG_STATUS || ac_write_fail=1
+rm -f conf$$subs.awk
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+_ACAWK
+cat >>"\$ac_tmp/subs1.awk" <<_ACAWK &&
+  for (key in S) S_is_set[key] = 1
+  FS = ""
+
+}
+{
+  line = $ 0
+  nfields = split(line, field, "@")
+  substed = 0
+  len = length(field[1])
+  for (i = 2; i < nfields; i++) {
+    key = field[i]
+    keylen = length(key)
+    if (S_is_set[key]) {
+      value = S[key]
+      line = substr(line, 1, len) "" value "" substr(line, len + keylen + 3)
+      len += length(value) + length(field[++i])
+      substed = 1
+    } else
+      len += 1 + keylen
+  }
+
+  print line
+}
+
+_ACAWK
+_ACEOF
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+if sed "s/$ac_cr//" < /dev/null > /dev/null 2>&1; then
+  sed "s/$ac_cr\$//; s/$ac_cr/$ac_cs_awk_cr/g"
+else
+  cat
+fi < "$ac_tmp/subs1.awk" > "$ac_tmp/subs.awk" \
+  || as_fn_error $? "could not setup config files machinery" "$LINENO" 5
+_ACEOF
+
+# VPATH may cause trouble with some makes, so we remove sole $(srcdir),
+# ${srcdir} and @srcdir@ entries from VPATH if srcdir is ".", strip leading and
+# trailing colons and then remove the whole line if VPATH becomes empty
+# (actually we leave an empty line to preserve line numbers).
+if test "x$srcdir" = x.; then
+  ac_vpsub='/^[	 ]*VPATH[	 ]*=[	 ]*/{
+h
+s///
+s/^/:/
+s/[	 ]*$/:/
+s/:\$(srcdir):/:/g
+s/:\${srcdir}:/:/g
+s/:@srcdir@:/:/g
+s/^:*//
+s/:*$//
+x
+s/\(=[	 ]*\).*/\1/
+G
+s/\n//
+s/^[^=]*=[	 ]*$//
+}'
+fi
+
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+fi # test -n "$CONFIG_FILES"
+
+# Set up the scripts for CONFIG_HEADERS section.
+# No need to generate them if there are no CONFIG_HEADERS.
+# This happens for instance with `./config.status Makefile'.
+if test -n "$CONFIG_HEADERS"; then
+cat >"$ac_tmp/defines.awk" <<\_ACAWK ||
+BEGIN {
+_ACEOF
+
+# Transform confdefs.h into an awk script `defines.awk', embedded as
+# here-document in config.status, that substitutes the proper values into
+# config.h.in to produce config.h.
+
+# Create a delimiter string that does not exist in confdefs.h, to ease
+# handling of long lines.
+ac_delim='%!_!# '
+for ac_last_try in false false :; do
+  ac_tt=`sed -n "/$ac_delim/p" confdefs.h`
+  if test -z "$ac_tt"; then
+    break
+  elif $ac_last_try; then
+    as_fn_error $? "could not make $CONFIG_HEADERS" "$LINENO" 5
+  else
+    ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
+  fi
+done
+
+# For the awk script, D is an array of macro values keyed by name,
+# likewise P contains macro parameters if any.  Preserve backslash
+# newline sequences.
+
+ac_word_re=[_$as_cr_Letters][_$as_cr_alnum]*
+sed -n '
+s/.\{148\}/&'"$ac_delim"'/g
+t rset
+:rset
+s/^[	 ]*#[	 ]*define[	 ][	 ]*/ /
+t def
+d
+:def
+s/\\$//
+t bsnl
+s/["\\]/\\&/g
+s/^ \('"$ac_word_re"'\)\(([^()]*)\)[	 ]*\(.*\)/P["\1"]="\2"\
+D["\1"]=" \3"/p
+s/^ \('"$ac_word_re"'\)[	 ]*\(.*\)/D["\1"]=" \2"/p
+d
+:bsnl
+s/["\\]/\\&/g
+s/^ \('"$ac_word_re"'\)\(([^()]*)\)[	 ]*\(.*\)/P["\1"]="\2"\
+D["\1"]=" \3\\\\\\n"\\/p
+t cont
+s/^ \('"$ac_word_re"'\)[	 ]*\(.*\)/D["\1"]=" \2\\\\\\n"\\/p
+t cont
+d
+:cont
+n
+s/.\{148\}/&'"$ac_delim"'/g
+t clear
+:clear
+s/\\$//
+t bsnlc
+s/["\\]/\\&/g; s/^/"/; s/$/"/p
+d
+:bsnlc
+s/["\\]/\\&/g; s/^/"/; s/$/\\\\\\n"\\/p
+b cont
+' <confdefs.h | sed '
+s/'"$ac_delim"'/"\\\
+"/g' >>$CONFIG_STATUS || ac_write_fail=1
+
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+  for (key in D) D_is_set[key] = 1
+  FS = ""
+}
+/^[\t ]*#[\t ]*(define|undef)[\t ]+$ac_word_re([\t (]|\$)/ {
+  line = \$ 0
+  split(line, arg, " ")
+  if (arg[1] == "#") {
+    defundef = arg[2]
+    mac1 = arg[3]
+  } else {
+    defundef = substr(arg[1], 2)
+    mac1 = arg[2]
+  }
+  split(mac1, mac2, "(") #)
+  macro = mac2[1]
+  prefix = substr(line, 1, index(line, defundef) - 1)
+  if (D_is_set[macro]) {
+    # Preserve the white space surrounding the "#".
+    print prefix "define", macro P[macro] D[macro]
+    next
+  } else {
+    # Replace #undef with comments.  This is necessary, for example,
+    # in the case of _POSIX_SOURCE, which is predefined and required
+    # on some systems where configure will not decide to define it.
+    if (defundef == "undef") {
+      print "/*", prefix defundef, macro, "*/"
+      next
+    }
+  }
+}
+{ print }
+_ACAWK
+_ACEOF
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+  as_fn_error $? "could not setup config headers machinery" "$LINENO" 5
+fi # test -n "$CONFIG_HEADERS"
+
+
+eval set X "  :F $CONFIG_FILES  :H $CONFIG_HEADERS    :C $CONFIG_COMMANDS"
+shift
+for ac_tag
+do
+  case $ac_tag in
+  :[FHLC]) ac_mode=$ac_tag; continue;;
+  esac
+  case $ac_mode$ac_tag in
+  :[FHL]*:*);;
+  :L* | :C*:*) as_fn_error $? "invalid tag \`$ac_tag'" "$LINENO" 5;;
+  :[FH]-) ac_tag=-:-;;
+  :[FH]*) ac_tag=$ac_tag:$ac_tag.in;;
+  esac
+  ac_save_IFS=$IFS
+  IFS=:
+  set x $ac_tag
+  IFS=$ac_save_IFS
+  shift
+  ac_file=$1
+  shift
+
+  case $ac_mode in
+  :L) ac_source=$1;;
+  :[FH])
+    ac_file_inputs=
+    for ac_f
+    do
+      case $ac_f in
+      -) ac_f="$ac_tmp/stdin";;
+      *) # Look for the file first in the build tree, then in the source tree
+	 # (if the path is not absolute).  The absolute path cannot be DOS-style,
+	 # because $ac_f cannot contain `:'.
+	 test -f "$ac_f" ||
+	   case $ac_f in
+	   [\\/$]*) false;;
+	   *) test -f "$srcdir/$ac_f" && ac_f="$srcdir/$ac_f";;
+	   esac ||
+	   as_fn_error 1 "cannot find input file: \`$ac_f'" "$LINENO" 5;;
+      esac
+      case $ac_f in *\'*) ac_f=`$as_echo "$ac_f" | sed "s/'/'\\\\\\\\''/g"`;; esac
+      as_fn_append ac_file_inputs " '$ac_f'"
+    done
+
+    # Let's still pretend it is `configure' which instantiates (i.e., don't
+    # use $as_me), people would be surprised to read:
+    #    /* config.h.  Generated by config.status.  */
+    configure_input='Generated from '`
+	  $as_echo "$*" | sed 's|^[^:]*/||;s|:[^:]*/|, |g'
+	`' by configure.'
+    if test x"$ac_file" != x-; then
+      configure_input="$ac_file.  $configure_input"
+      { $as_echo "$as_me:${as_lineno-$LINENO}: creating $ac_file" >&5
+$as_echo "$as_me: creating $ac_file" >&6;}
+    fi
+    # Neutralize special characters interpreted by sed in replacement strings.
+    case $configure_input in #(
+    *\&* | *\|* | *\\* )
+       ac_sed_conf_input=`$as_echo "$configure_input" |
+       sed 's/[\\\\&|]/\\\\&/g'`;; #(
+    *) ac_sed_conf_input=$configure_input;;
+    esac
+
+    case $ac_tag in
+    *:-:* | *:-) cat >"$ac_tmp/stdin" \
+      || as_fn_error $? "could not create $ac_file" "$LINENO" 5 ;;
+    esac
+    ;;
+  esac
+
+  ac_dir=`$as_dirname -- "$ac_file" ||
+$as_expr X"$ac_file" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$ac_file" : 'X\(//\)[^/]' \| \
+	 X"$ac_file" : 'X\(//\)$' \| \
+	 X"$ac_file" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$ac_file" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+  as_dir="$ac_dir"; as_fn_mkdir_p
+  ac_builddir=.
+
+case "$ac_dir" in
+.) ac_dir_suffix= ac_top_builddir_sub=. ac_top_build_prefix= ;;
+*)
+  ac_dir_suffix=/`$as_echo "$ac_dir" | sed 's|^\.[\\/]||'`
+  # A ".." for each directory in $ac_dir_suffix.
+  ac_top_builddir_sub=`$as_echo "$ac_dir_suffix" | sed 's|/[^\\/]*|/..|g;s|/||'`
+  case $ac_top_builddir_sub in
+  "") ac_top_builddir_sub=. ac_top_build_prefix= ;;
+  *)  ac_top_build_prefix=$ac_top_builddir_sub/ ;;
+  esac ;;
+esac
+ac_abs_top_builddir=$ac_pwd
+ac_abs_builddir=$ac_pwd$ac_dir_suffix
+# for backward compatibility:
+ac_top_builddir=$ac_top_build_prefix
+
+case $srcdir in
+  .)  # We are building in place.
+    ac_srcdir=.
+    ac_top_srcdir=$ac_top_builddir_sub
+    ac_abs_top_srcdir=$ac_pwd ;;
+  [\\/]* | ?:[\\/]* )  # Absolute name.
+    ac_srcdir=$srcdir$ac_dir_suffix;
+    ac_top_srcdir=$srcdir
+    ac_abs_top_srcdir=$srcdir ;;
+  *) # Relative name.
+    ac_srcdir=$ac_top_build_prefix$srcdir$ac_dir_suffix
+    ac_top_srcdir=$ac_top_build_prefix$srcdir
+    ac_abs_top_srcdir=$ac_pwd/$srcdir ;;
+esac
+ac_abs_srcdir=$ac_abs_top_srcdir$ac_dir_suffix
+
+
+  case $ac_mode in
+  :F)
+  #
+  # CONFIG_FILE
+  #
+
+  case $INSTALL in
+  [\\/$]* | ?:[\\/]* ) ac_INSTALL=$INSTALL ;;
+  *) ac_INSTALL=$ac_top_build_prefix$INSTALL ;;
+  esac
+_ACEOF
+
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+# If the template does not know about datarootdir, expand it.
+# FIXME: This hack should be removed a few years after 2.60.
+ac_datarootdir_hack=; ac_datarootdir_seen=
+ac_sed_dataroot='
+/datarootdir/ {
+  p
+  q
+}
+/@datadir@/p
+/@docdir@/p
+/@infodir@/p
+/@localedir@/p
+/@mandir@/p'
+case `eval "sed -n \"\$ac_sed_dataroot\" $ac_file_inputs"` in
+*datarootdir*) ac_datarootdir_seen=yes;;
+*@datadir@*|*@docdir@*|*@infodir@*|*@localedir@*|*@mandir@*)
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
+$as_echo "$as_me: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&2;}
+_ACEOF
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+  ac_datarootdir_hack='
+  s&@datadir@&$datadir&g
+  s&@docdir@&$docdir&g
+  s&@infodir@&$infodir&g
+  s&@localedir@&$localedir&g
+  s&@mandir@&$mandir&g
+  s&\\\${datarootdir}&$datarootdir&g' ;;
+esac
+_ACEOF
+
+# Neutralize VPATH when `$srcdir' = `.'.
+# Shell code in configure.ac might set extrasub.
+# FIXME: do we really want to maintain this feature?
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+ac_sed_extra="$ac_vpsub
+$extrasub
+_ACEOF
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+:t
+/@[a-zA-Z_][a-zA-Z_0-9]*@/!b
+s|@configure_input@|$ac_sed_conf_input|;t t
+s&@top_builddir@&$ac_top_builddir_sub&;t t
+s&@top_build_prefix@&$ac_top_build_prefix&;t t
+s&@srcdir@&$ac_srcdir&;t t
+s&@abs_srcdir@&$ac_abs_srcdir&;t t
+s&@top_srcdir@&$ac_top_srcdir&;t t
+s&@abs_top_srcdir@&$ac_abs_top_srcdir&;t t
+s&@builddir@&$ac_builddir&;t t
+s&@abs_builddir@&$ac_abs_builddir&;t t
+s&@abs_top_builddir@&$ac_abs_top_builddir&;t t
+s&@INSTALL@&$ac_INSTALL&;t t
+$ac_datarootdir_hack
+"
+eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$ac_tmp/subs.awk" \
+  >$ac_tmp/out || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+
+test -z "$ac_datarootdir_hack$ac_datarootdir_seen" &&
+  { ac_out=`sed -n '/\${datarootdir}/p' "$ac_tmp/out"`; test -n "$ac_out"; } &&
+  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' \
+      "$ac_tmp/out"`; test -z "$ac_out"; } &&
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file contains a reference to the variable \`datarootdir'
+which seems to be undefined.  Please make sure it is defined" >&5
+$as_echo "$as_me: WARNING: $ac_file contains a reference to the variable \`datarootdir'
+which seems to be undefined.  Please make sure it is defined" >&2;}
+
+  rm -f "$ac_tmp/stdin"
+  case $ac_file in
+  -) cat "$ac_tmp/out" && rm -f "$ac_tmp/out";;
+  *) rm -f "$ac_file" && mv "$ac_tmp/out" "$ac_file";;
+  esac \
+  || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+ ;;
+  :H)
+  #
+  # CONFIG_HEADER
+  #
+  if test x"$ac_file" != x-; then
+    {
+      $as_echo "/* $configure_input  */" \
+      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs"
+    } >"$ac_tmp/config.h" \
+      || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+    if diff "$ac_file" "$ac_tmp/config.h" >/dev/null 2>&1; then
+      { $as_echo "$as_me:${as_lineno-$LINENO}: $ac_file is unchanged" >&5
+$as_echo "$as_me: $ac_file is unchanged" >&6;}
+    else
+      rm -f "$ac_file"
+      mv "$ac_tmp/config.h" "$ac_file" \
+	|| as_fn_error $? "could not create $ac_file" "$LINENO" 5
+    fi
+  else
+    $as_echo "/* $configure_input  */" \
+      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs" \
+      || as_fn_error $? "could not create -" "$LINENO" 5
+  fi
+ ;;
+
+  :C)  { $as_echo "$as_me:${as_lineno-$LINENO}: executing $ac_file commands" >&5
+$as_echo "$as_me: executing $ac_file commands" >&6;}
+ ;;
+  esac
+
+
+  case $ac_file$ac_mode in
+    "include/jemalloc/internal/size_classes.h":C)
+  mkdir -p "include/jemalloc/internal"
+  "${srcdir}/include/jemalloc/internal/size_classes.sh" > "${objroot}include/jemalloc/internal/size_classes.h"
+ ;;
+
+  esac
+done # for ac_tag
+
+
+as_fn_exit 0
+_ACEOF
+ac_clean_files=$ac_clean_files_save
+
+test $ac_write_fail = 0 ||
+  as_fn_error $? "write failure creating $CONFIG_STATUS" "$LINENO" 5
+
+
+# configure is writing to config.log, and then calls config.status.
+# config.status does its own redirection, appending to config.log.
+# Unfortunately, on DOS this fails, as config.log is still kept open
+# by configure, so config.status won't be able to write to it; its
+# output is simply discarded.  So we exec the FD to /dev/null,
+# effectively closing config.log, so it can be properly (re)opened and
+# appended to by config.status.  When coming back to configure, we
+# need to make the FD available again.
+if test "$no_create" != yes; then
+  ac_cs_success=:
+  ac_config_status_args=
+  test "$silent" = yes &&
+    ac_config_status_args="$ac_config_status_args --quiet"
+  exec 5>/dev/null
+  $SHELL $CONFIG_STATUS $ac_config_status_args || ac_cs_success=false
+  exec 5>>config.log
+  # Use ||, not &&, to avoid exiting from the if with $? = 1, which
+  # would make configure fail if this is the last instruction.
+  $ac_cs_success || as_fn_exit 1
+fi
+if test -n "$ac_unrecognized_opts" && test "$enable_option_checking" != no; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
+$as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2;}
+fi
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: ===============================================================================" >&5
+$as_echo "===============================================================================" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: jemalloc version   : ${jemalloc_version}" >&5
+$as_echo "jemalloc version   : ${jemalloc_version}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: library revision   : ${rev}" >&5
+$as_echo "library revision   : ${rev}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: " >&5
+$as_echo "" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: CC                 : ${CC}" >&5
+$as_echo "CC                 : ${CC}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: CPPFLAGS           : ${CPPFLAGS}" >&5
+$as_echo "CPPFLAGS           : ${CPPFLAGS}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: CFLAGS             : ${CFLAGS}" >&5
+$as_echo "CFLAGS             : ${CFLAGS}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: LDFLAGS            : ${LDFLAGS}" >&5
+$as_echo "LDFLAGS            : ${LDFLAGS}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: LIBS               : ${LIBS}" >&5
+$as_echo "LIBS               : ${LIBS}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: RPATH_EXTRA        : ${RPATH_EXTRA}" >&5
+$as_echo "RPATH_EXTRA        : ${RPATH_EXTRA}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: " >&5
+$as_echo "" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: XSLTPROC           : ${XSLTPROC}" >&5
+$as_echo "XSLTPROC           : ${XSLTPROC}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: XSLROOT            : ${XSLROOT}" >&5
+$as_echo "XSLROOT            : ${XSLROOT}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: " >&5
+$as_echo "" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: PREFIX             : ${PREFIX}" >&5
+$as_echo "PREFIX             : ${PREFIX}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: BINDIR             : ${BINDIR}" >&5
+$as_echo "BINDIR             : ${BINDIR}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: INCLUDEDIR         : ${INCLUDEDIR}" >&5
+$as_echo "INCLUDEDIR         : ${INCLUDEDIR}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: LIBDIR             : ${LIBDIR}" >&5
+$as_echo "LIBDIR             : ${LIBDIR}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: DATADIR            : ${DATADIR}" >&5
+$as_echo "DATADIR            : ${DATADIR}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: MANDIR             : ${MANDIR}" >&5
+$as_echo "MANDIR             : ${MANDIR}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: " >&5
+$as_echo "" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: srcroot            : ${srcroot}" >&5
+$as_echo "srcroot            : ${srcroot}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: abs_srcroot        : ${abs_srcroot}" >&5
+$as_echo "abs_srcroot        : ${abs_srcroot}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: objroot            : ${objroot}" >&5
+$as_echo "objroot            : ${objroot}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: abs_objroot        : ${abs_objroot}" >&5
+$as_echo "abs_objroot        : ${abs_objroot}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: " >&5
+$as_echo "" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: JEMALLOC_PREFIX    : ${JEMALLOC_PREFIX}" >&5
+$as_echo "JEMALLOC_PREFIX    : ${JEMALLOC_PREFIX}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: JEMALLOC_PRIVATE_NAMESPACE" >&5
+$as_echo "JEMALLOC_PRIVATE_NAMESPACE" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result:                    : ${JEMALLOC_PRIVATE_NAMESPACE}" >&5
+$as_echo "                   : ${JEMALLOC_PRIVATE_NAMESPACE}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: install_suffix     : ${install_suffix}" >&5
+$as_echo "install_suffix     : ${install_suffix}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: autogen            : ${enable_autogen}" >&5
+$as_echo "autogen            : ${enable_autogen}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: experimental       : ${enable_experimental}" >&5
+$as_echo "experimental       : ${enable_experimental}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: cc-silence         : ${enable_cc_silence}" >&5
+$as_echo "cc-silence         : ${enable_cc_silence}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: debug              : ${enable_debug}" >&5
+$as_echo "debug              : ${enable_debug}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: stats              : ${enable_stats}" >&5
+$as_echo "stats              : ${enable_stats}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: prof               : ${enable_prof}" >&5
+$as_echo "prof               : ${enable_prof}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: prof-libunwind     : ${enable_prof_libunwind}" >&5
+$as_echo "prof-libunwind     : ${enable_prof_libunwind}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: prof-libgcc        : ${enable_prof_libgcc}" >&5
+$as_echo "prof-libgcc        : ${enable_prof_libgcc}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: prof-gcc           : ${enable_prof_gcc}" >&5
+$as_echo "prof-gcc           : ${enable_prof_gcc}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: tcache             : ${enable_tcache}" >&5
+$as_echo "tcache             : ${enable_tcache}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: fill               : ${enable_fill}" >&5
+$as_echo "fill               : ${enable_fill}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: utrace             : ${enable_utrace}" >&5
+$as_echo "utrace             : ${enable_utrace}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: valgrind           : ${enable_valgrind}" >&5
+$as_echo "valgrind           : ${enable_valgrind}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: xmalloc            : ${enable_xmalloc}" >&5
+$as_echo "xmalloc            : ${enable_xmalloc}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: mremap             : ${enable_mremap}" >&5
+$as_echo "mremap             : ${enable_mremap}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: munmap             : ${enable_munmap}" >&5
+$as_echo "munmap             : ${enable_munmap}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: dss                : ${enable_dss}" >&5
+$as_echo "dss                : ${enable_dss}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: lazy_lock          : ${enable_lazy_lock}" >&5
+$as_echo "lazy_lock          : ${enable_lazy_lock}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: tls                : ${enable_tls}" >&5
+$as_echo "tls                : ${enable_tls}" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: ===============================================================================" >&5
+$as_echo "===============================================================================" >&6; }
diff --git a/extra/jemalloc/configure.ac b/extra/jemalloc/configure.ac
new file mode 100644
index 00000000000..c270662b6d4
--- /dev/null
+++ b/extra/jemalloc/configure.ac
@@ -0,0 +1,1333 @@
+dnl Process this file with autoconf to produce a configure script.
+AC_INIT([Makefile.in])
+
+dnl ============================================================================
+dnl Custom macro definitions.
+
+dnl JE_CFLAGS_APPEND(cflag)
+AC_DEFUN([JE_CFLAGS_APPEND],
+[
+AC_MSG_CHECKING([whether compiler supports $1])
+TCFLAGS="${CFLAGS}"
+if test "x${CFLAGS}" = "x" ; then
+  CFLAGS="$1"
+else
+  CFLAGS="${CFLAGS} $1"
+fi
+AC_COMPILE_IFELSE([AC_LANG_PROGRAM(
+[[
+]], [[
+    return 0;
+]])],
+              AC_MSG_RESULT([yes]),
+              AC_MSG_RESULT([no])
+              [CFLAGS="${TCFLAGS}"]
+)
+])
+
+dnl JE_COMPILABLE(label, hcode, mcode, rvar)
+dnl 
+dnl Use AC_LINK_IFELSE() rather than AC_COMPILE_IFELSE() so that linker errors
+dnl cause failure.
+AC_DEFUN([JE_COMPILABLE],
+[
+AC_CACHE_CHECK([whether $1 is compilable],
+               [$4],
+               [AC_LINK_IFELSE([AC_LANG_PROGRAM([$2],
+                                                [$3])],
+                               [$4=yes],
+                               [$4=no])])
+])
+
+dnl ============================================================================
+
+dnl Library revision.
+rev=1
+AC_SUBST([rev])
+
+srcroot=$srcdir
+if test "x${srcroot}" = "x." ; then
+  srcroot=""
+else
+  srcroot="${srcroot}/"
+fi
+AC_SUBST([srcroot])
+abs_srcroot="`cd \"${srcdir}\"; pwd`/"
+AC_SUBST([abs_srcroot])
+
+objroot=""
+AC_SUBST([objroot])
+abs_objroot="`pwd`/"
+AC_SUBST([abs_objroot])
+
+dnl Munge install path variables.
+if test "x$prefix" = "xNONE" ; then
+  prefix="/usr/local"
+fi
+if test "x$exec_prefix" = "xNONE" ; then
+  exec_prefix=$prefix
+fi
+PREFIX=$prefix
+AC_SUBST([PREFIX])
+BINDIR=`eval echo $bindir`
+BINDIR=`eval echo $BINDIR`
+AC_SUBST([BINDIR])
+INCLUDEDIR=`eval echo $includedir`
+INCLUDEDIR=`eval echo $INCLUDEDIR`
+AC_SUBST([INCLUDEDIR])
+LIBDIR=`eval echo $libdir`
+LIBDIR=`eval echo $LIBDIR`
+AC_SUBST([LIBDIR])
+DATADIR=`eval echo $datadir`
+DATADIR=`eval echo $DATADIR`
+AC_SUBST([DATADIR])
+MANDIR=`eval echo $mandir`
+MANDIR=`eval echo $MANDIR`
+AC_SUBST([MANDIR])
+
+dnl Support for building documentation.
+AC_PATH_PROG([XSLTPROC], [xsltproc], [false], [$PATH])
+if test -d "/usr/share/xml/docbook/stylesheet/docbook-xsl" ; then
+  DEFAULT_XSLROOT="/usr/share/xml/docbook/stylesheet/docbook-xsl"
+elif test -d "/usr/share/sgml/docbook/xsl-stylesheets" ; then
+  DEFAULT_XSLROOT="/usr/share/sgml/docbook/xsl-stylesheets"
+else
+  dnl Documentation building will fail if this default gets used.
+  DEFAULT_XSLROOT=""
+fi
+AC_ARG_WITH([xslroot],
+  [AS_HELP_STRING([--with-xslroot=<path>], [XSL stylesheet root path])], [
+if test "x$with_xslroot" = "xno" ; then
+  XSLROOT="${DEFAULT_XSLROOT}"
+else
+  XSLROOT="${with_xslroot}"
+fi
+],
+  XSLROOT="${DEFAULT_XSLROOT}"
+)
+AC_SUBST([XSLROOT])
+
+dnl If CFLAGS isn't defined, set CFLAGS to something reasonable.  Otherwise,
+dnl just prevent autoconf from molesting CFLAGS.
+CFLAGS=$CFLAGS
+AC_PROG_CC
+if test "x$GCC" != "xyes" ; then
+  AC_CACHE_CHECK([whether compiler is MSVC],
+                 [je_cv_msvc],
+                 [AC_COMPILE_IFELSE([AC_LANG_PROGRAM([],
+                                                     [
+#ifndef _MSC_VER
+  int fail[-1];
+#endif
+])],
+                               [je_cv_msvc=yes],
+                               [je_cv_msvc=no])])
+fi
+
+if test "x$CFLAGS" = "x" ; then
+  no_CFLAGS="yes"
+  if test "x$GCC" = "xyes" ; then
+    JE_CFLAGS_APPEND([-std=gnu99])
+    JE_CFLAGS_APPEND([-Wall])
+    JE_CFLAGS_APPEND([-pipe])
+    JE_CFLAGS_APPEND([-g3])
+  elif test "x$je_cv_msvc" = "xyes" ; then
+    CC="$CC -nologo"
+    JE_CFLAGS_APPEND([-Zi])
+    JE_CFLAGS_APPEND([-MT])
+    JE_CFLAGS_APPEND([-W3])
+    CPPFLAGS="$CPPFLAGS -I${srcroot}/include/msvc_compat"
+  fi
+fi
+dnl Append EXTRA_CFLAGS to CFLAGS, if defined.
+if test "x$EXTRA_CFLAGS" != "x" ; then
+  JE_CFLAGS_APPEND([$EXTRA_CFLAGS])
+fi
+AC_PROG_CPP
+
+AC_CHECK_SIZEOF([void *])
+if test "x${ac_cv_sizeof_void_p}" = "x8" ; then
+  LG_SIZEOF_PTR=3
+elif test "x${ac_cv_sizeof_void_p}" = "x4" ; then
+  LG_SIZEOF_PTR=2
+else
+  AC_MSG_ERROR([Unsupported pointer size: ${ac_cv_sizeof_void_p}])
+fi
+AC_DEFINE_UNQUOTED([LG_SIZEOF_PTR], [$LG_SIZEOF_PTR])
+
+AC_CHECK_SIZEOF([int])
+if test "x${ac_cv_sizeof_int}" = "x8" ; then
+  LG_SIZEOF_INT=3
+elif test "x${ac_cv_sizeof_int}" = "x4" ; then
+  LG_SIZEOF_INT=2
+else
+  AC_MSG_ERROR([Unsupported int size: ${ac_cv_sizeof_int}])
+fi
+AC_DEFINE_UNQUOTED([LG_SIZEOF_INT], [$LG_SIZEOF_INT])
+
+AC_CHECK_SIZEOF([long])
+if test "x${ac_cv_sizeof_long}" = "x8" ; then
+  LG_SIZEOF_LONG=3
+elif test "x${ac_cv_sizeof_long}" = "x4" ; then
+  LG_SIZEOF_LONG=2
+else
+  AC_MSG_ERROR([Unsupported long size: ${ac_cv_sizeof_long}])
+fi
+AC_DEFINE_UNQUOTED([LG_SIZEOF_LONG], [$LG_SIZEOF_LONG])
+
+AC_CHECK_SIZEOF([intmax_t])
+if test "x${ac_cv_sizeof_intmax_t}" = "x16" ; then
+  LG_SIZEOF_INTMAX_T=4
+elif test "x${ac_cv_sizeof_intmax_t}" = "x8" ; then
+  LG_SIZEOF_INTMAX_T=3
+elif test "x${ac_cv_sizeof_intmax_t}" = "x4" ; then
+  LG_SIZEOF_INTMAX_T=2
+else
+  AC_MSG_ERROR([Unsupported intmax_t size: ${ac_cv_sizeof_intmax_t}])
+fi
+AC_DEFINE_UNQUOTED([LG_SIZEOF_INTMAX_T], [$LG_SIZEOF_INTMAX_T])
+
+AC_CANONICAL_HOST
+dnl CPU-specific settings.
+CPU_SPINWAIT=""
+case "${host_cpu}" in
+  i[[345]]86)
+	;;
+  i686)
+	JE_COMPILABLE([__asm__], [], [[__asm__ volatile("pause"); return 0;]],
+	              [je_cv_asm])
+	if test "x${je_cv_asm}" = "xyes" ; then
+	    CPU_SPINWAIT='__asm__ volatile("pause")'
+	fi
+	;;
+  x86_64)
+	JE_COMPILABLE([__asm__ syntax], [],
+	              [[__asm__ volatile("pause"); return 0;]], [je_cv_asm])
+	if test "x${je_cv_asm}" = "xyes" ; then
+	    CPU_SPINWAIT='__asm__ volatile("pause")'
+	fi
+	;;
+  *)
+	;;
+esac
+AC_DEFINE_UNQUOTED([CPU_SPINWAIT], [$CPU_SPINWAIT])
+
+LD_PRELOAD_VAR="LD_PRELOAD"
+so="so"
+importlib="${so}"
+o="$ac_objext"
+a="a"
+exe="$ac_exeext"
+libprefix="lib"
+DSO_LDFLAGS='-shared -Wl,-soname,$(@F)'
+RPATH='-Wl,-rpath,$(1)'
+SOREV="${so}.${rev}"
+PIC_CFLAGS='-fPIC -DPIC'
+CTARGET='-o $@'
+LDTARGET='-o $@'
+EXTRA_LDFLAGS=
+MKLIB='ar crus $@'
+CC_MM=1
+
+dnl Platform-specific settings.  abi and RPATH can probably be determined
+dnl programmatically, but doing so is error-prone, which makes it generally
+dnl not worth the trouble.
+dnl 
+dnl Define cpp macros in CPPFLAGS, rather than doing AC_DEFINE(macro), since the
+dnl definitions need to be seen before any headers are included, which is a pain
+dnl to make happen otherwise.
+default_munmap="1"
+JEMALLOC_USABLE_SIZE_CONST="const"
+case "${host}" in
+  *-*-darwin*)
+	CFLAGS="$CFLAGS"
+	abi="macho"
+	AC_DEFINE([JEMALLOC_PURGE_MADVISE_FREE], [ ])
+	RPATH=""
+	LD_PRELOAD_VAR="DYLD_INSERT_LIBRARIES"
+	so="dylib"
+	importlib="${so}"
+	force_tls="0"
+	DSO_LDFLAGS='-shared -Wl,-dylib_install_name,$(@F)'
+	SOREV="${rev}.${so}"
+	;;
+  *-*-freebsd*)
+	CFLAGS="$CFLAGS"
+	abi="elf"
+	AC_DEFINE([JEMALLOC_PURGE_MADVISE_FREE], [ ])
+	force_lazy_lock="1"
+	;;
+  *-*-linux*)
+	CFLAGS="$CFLAGS"
+	CPPFLAGS="$CPPFLAGS -D_GNU_SOURCE"
+	abi="elf"
+	AC_DEFINE([JEMALLOC_HAS_ALLOCA_H])
+	AC_DEFINE([JEMALLOC_PURGE_MADVISE_DONTNEED], [ ])
+	AC_DEFINE([JEMALLOC_THREADED_INIT], [ ])
+	JEMALLOC_USABLE_SIZE_CONST=""
+	default_munmap="0"
+	;;
+  *-*-netbsd*)
+	AC_MSG_CHECKING([ABI])
+        AC_COMPILE_IFELSE([AC_LANG_PROGRAM(
+[[#ifdef __ELF__
+/* ELF */
+#else
+#error aout
+#endif
+]])],
+                          [CFLAGS="$CFLAGS"; abi="elf"],
+                          [abi="aout"])
+	AC_MSG_RESULT([$abi])
+	AC_DEFINE([JEMALLOC_PURGE_MADVISE_FREE], [ ])
+	;;
+  *-*-solaris2*)
+	CFLAGS="$CFLAGS"
+	abi="elf"
+	RPATH='-Wl,-R,$(1)'
+	dnl Solaris needs this for sigwait().
+	CPPFLAGS="$CPPFLAGS -D_POSIX_PTHREAD_SEMANTICS"
+	LIBS="$LIBS -lposix4 -lsocket -lnsl"
+	;;
+  *-ibm-aix*)
+	if "$LG_SIZEOF_PTR" = "8"; then
+	  dnl 64bit AIX
+	  LD_PRELOAD_VAR="LDR_PRELOAD64"
+	else
+	  dnl 32bit AIX
+	  LD_PRELOAD_VAR="LDR_PRELOAD"
+	fi
+	abi="xcoff"
+	;;
+  *-*-mingw*)
+	abi="pecoff"
+	force_tls="0"
+	RPATH=""
+	so="dll"
+	if test "x$je_cv_msvc" = "xyes" ; then
+	  importlib="lib"
+	  DSO_LDFLAGS="-LD"
+	  EXTRA_LDFLAGS="-link -DEBUG"
+	  CTARGET='-Fo$@'
+	  LDTARGET='-Fe$@'
+	  MKLIB='lib -nologo -out:$@'
+	  CC_MM=
+        else
+	  importlib="${so}"
+	  DSO_LDFLAGS="-shared"
+	fi
+	a="lib"
+	libprefix=""
+	SOREV="${so}"
+	PIC_CFLAGS=""
+	;;
+  *)
+	AC_MSG_RESULT([Unsupported operating system: ${host}])
+	abi="elf"
+	;;
+esac
+AC_DEFINE_UNQUOTED([JEMALLOC_USABLE_SIZE_CONST], [$JEMALLOC_USABLE_SIZE_CONST])
+AC_SUBST([abi])
+AC_SUBST([RPATH])
+AC_SUBST([LD_PRELOAD_VAR])
+AC_SUBST([so])
+AC_SUBST([importlib])
+AC_SUBST([o])
+AC_SUBST([a])
+AC_SUBST([exe])
+AC_SUBST([libprefix])
+AC_SUBST([DSO_LDFLAGS])
+AC_SUBST([EXTRA_LDFLAGS])
+AC_SUBST([SOREV])
+AC_SUBST([PIC_CFLAGS])
+AC_SUBST([CTARGET])
+AC_SUBST([LDTARGET])
+AC_SUBST([MKLIB])
+AC_SUBST([CC_MM])
+
+if test "x$abi" != "xpecoff"; then
+  dnl Heap profiling uses the log(3) function.
+  LIBS="$LIBS -lm"
+fi
+
+JE_COMPILABLE([__attribute__ syntax],
+              [static __attribute__((unused)) void foo(void){}],
+              [],
+              [je_cv_attribute])
+if test "x${je_cv_attribute}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_ATTR], [ ])
+  if test "x${GCC}" = "xyes" -a "x${abi}" = "xelf"; then
+    JE_CFLAGS_APPEND([-fvisibility=hidden])
+  fi
+fi
+dnl Check for tls_model attribute support (clang 3.0 still lacks support).
+SAVED_CFLAGS="${CFLAGS}"
+JE_CFLAGS_APPEND([-Werror])
+JE_COMPILABLE([tls_model attribute], [],
+              [static __thread int
+               __attribute__((tls_model("initial-exec"))) foo;
+               foo = 0;],
+              [je_cv_tls_model])
+CFLAGS="${SAVED_CFLAGS}"
+if test "x${je_cv_tls_model}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_TLS_MODEL],
+            [__attribute__((tls_model("initial-exec")))])
+else
+  AC_DEFINE([JEMALLOC_TLS_MODEL], [ ])
+fi
+
+dnl Support optional additions to rpath.
+AC_ARG_WITH([rpath],
+  [AS_HELP_STRING([--with-rpath=<rpath>], [Colon-separated rpath (ELF systems only)])],
+if test "x$with_rpath" = "xno" ; then
+  RPATH_EXTRA=
+else
+  RPATH_EXTRA="`echo $with_rpath | tr \":\" \" \"`"
+fi,
+  RPATH_EXTRA=
+)
+AC_SUBST([RPATH_EXTRA])
+
+dnl Disable rules that do automatic regeneration of configure output by default.
+AC_ARG_ENABLE([autogen],
+  [AS_HELP_STRING([--enable-autogen], [Automatically regenerate configure output])],
+if test "x$enable_autogen" = "xno" ; then
+  enable_autogen="0"
+else
+  enable_autogen="1"
+fi
+,
+enable_autogen="0"
+)
+AC_SUBST([enable_autogen])
+
+AC_PROG_INSTALL
+AC_PROG_RANLIB
+AC_PATH_PROG([AR], [ar], [false], [$PATH])
+AC_PATH_PROG([LD], [ld], [false], [$PATH])
+AC_PATH_PROG([AUTOCONF], [autoconf], [false], [$PATH])
+
+public_syms="malloc_conf malloc_message malloc calloc posix_memalign aligned_alloc realloc free malloc_usable_size malloc_stats_print mallctl mallctlnametomib mallctlbymib"
+
+dnl Check for allocator-related functions that should be wrapped.
+AC_CHECK_FUNC([memalign],
+	      [AC_DEFINE([JEMALLOC_OVERRIDE_MEMALIGN], [ ])
+	       public_syms="${public_syms} memalign"])
+AC_CHECK_FUNC([valloc],
+	      [AC_DEFINE([JEMALLOC_OVERRIDE_VALLOC], [ ])
+	       public_syms="${public_syms} valloc"])
+
+dnl Support the experimental API by default.
+AC_ARG_ENABLE([experimental],
+  [AS_HELP_STRING([--disable-experimental],
+   [Disable support for the experimental API])],
+[if test "x$enable_experimental" = "xno" ; then
+  enable_experimental="0"
+else
+  enable_experimental="1"
+fi
+],
+[enable_experimental="1"]
+)
+if test "x$enable_experimental" = "x1" ; then
+  AC_DEFINE([JEMALLOC_EXPERIMENTAL], [ ])
+  public_syms="${public_syms} allocm dallocm nallocm rallocm sallocm"
+fi
+AC_SUBST([enable_experimental])
+
+dnl Perform no name mangling by default.
+AC_ARG_WITH([mangling],
+  [AS_HELP_STRING([--with-mangling=<map>], [Mangle symbols in <map>])],
+  [mangling_map="$with_mangling"], [mangling_map=""])
+for nm in `echo ${mangling_map} |tr ',' ' '` ; do
+  k="`echo ${nm} |tr ':' ' ' |awk '{print $1}'`"
+  n="je_${k}"
+  m=`echo ${nm} |tr ':' ' ' |awk '{print $2}'`
+  AC_DEFINE_UNQUOTED([${n}], [${m}])
+  dnl Remove key from public_syms so that it isn't redefined later.
+  public_syms=`for sym in ${public_syms}; do echo "${sym}"; done |grep -v "^${k}\$" |tr '\n' ' '`
+done
+
+dnl Do not prefix public APIs by default.
+AC_ARG_WITH([jemalloc_prefix],
+  [AS_HELP_STRING([--with-jemalloc-prefix=<prefix>], [Prefix to prepend to all public APIs])],
+  [JEMALLOC_PREFIX="$with_jemalloc_prefix"],
+  [if test "x$abi" != "xmacho" -a "x$abi" != "xpecoff"; then
+  JEMALLOC_PREFIX=""
+else
+  JEMALLOC_PREFIX="je_"
+fi]
+)
+if test "x$JEMALLOC_PREFIX" != "x" ; then
+  JEMALLOC_CPREFIX=`echo ${JEMALLOC_PREFIX} | tr "a-z" "A-Z"`
+  AC_DEFINE_UNQUOTED([JEMALLOC_PREFIX], ["$JEMALLOC_PREFIX"])
+  AC_DEFINE_UNQUOTED([JEMALLOC_CPREFIX], ["$JEMALLOC_CPREFIX"])
+fi
+dnl Generate macros to rename public symbols.  All public symbols are prefixed
+dnl with je_ in the source code, so these macro definitions are needed even if
+dnl --with-jemalloc-prefix wasn't specified.
+for stem in ${public_syms}; do
+  n="je_${stem}"
+  m="${JEMALLOC_PREFIX}${stem}"
+  AC_DEFINE_UNQUOTED([${n}], [${m}])
+done
+
+AC_ARG_WITH([export],
+  [AS_HELP_STRING([--without-export], [disable exporting jemalloc public APIs])],
+  [if test "x$with_export" = "xno"; then
+  AC_DEFINE([JEMALLOC_EXPORT],[])
+fi]
+)
+
+dnl Do not mangle library-private APIs by default.
+AC_ARG_WITH([private_namespace],
+  [AS_HELP_STRING([--with-private-namespace=<prefix>], [Prefix to prepend to all library-private APIs])],
+  [JEMALLOC_PRIVATE_NAMESPACE="$with_private_namespace"],
+  [JEMALLOC_PRIVATE_NAMESPACE=""]
+)
+AC_DEFINE_UNQUOTED([JEMALLOC_PRIVATE_NAMESPACE], ["$JEMALLOC_PRIVATE_NAMESPACE"])
+if test "x$JEMALLOC_PRIVATE_NAMESPACE" != "x" ; then
+  AC_DEFINE_UNQUOTED([JEMALLOC_N(string_that_no_one_should_want_to_use_as_a_jemalloc_private_namespace_prefix)], [${JEMALLOC_PRIVATE_NAMESPACE}##string_that_no_one_should_want_to_use_as_a_jemalloc_private_namespace_prefix])
+else
+  AC_DEFINE_UNQUOTED([JEMALLOC_N(string_that_no_one_should_want_to_use_as_a_jemalloc_private_namespace_prefix)], [string_that_no_one_should_want_to_use_as_a_jemalloc_private_namespace_prefix])
+fi
+
+dnl Do not add suffix to installed files by default.
+AC_ARG_WITH([install_suffix],
+  [AS_HELP_STRING([--with-install-suffix=<suffix>], [Suffix to append to all installed files])],
+  [INSTALL_SUFFIX="$with_install_suffix"],
+  [INSTALL_SUFFIX=]
+)
+install_suffix="$INSTALL_SUFFIX"
+AC_SUBST([install_suffix])
+
+cfgoutputs_in="${srcroot}Makefile.in"
+cfgoutputs_in="${cfgoutputs_in} ${srcroot}doc/html.xsl.in"
+cfgoutputs_in="${cfgoutputs_in} ${srcroot}doc/manpages.xsl.in"
+cfgoutputs_in="${cfgoutputs_in} ${srcroot}doc/jemalloc.xml.in"
+cfgoutputs_in="${cfgoutputs_in} ${srcroot}include/jemalloc/jemalloc.h.in"
+cfgoutputs_in="${cfgoutputs_in} ${srcroot}include/jemalloc/internal/jemalloc_internal.h.in"
+cfgoutputs_in="${cfgoutputs_in} ${srcroot}test/jemalloc_test.h.in"
+
+cfgoutputs_out="Makefile"
+cfgoutputs_out="${cfgoutputs_out} doc/html.xsl"
+cfgoutputs_out="${cfgoutputs_out} doc/manpages.xsl"
+cfgoutputs_out="${cfgoutputs_out} doc/jemalloc${install_suffix}.xml"
+cfgoutputs_out="${cfgoutputs_out} include/jemalloc/jemalloc${install_suffix}.h"
+cfgoutputs_out="${cfgoutputs_out} include/jemalloc/internal/jemalloc_internal.h"
+cfgoutputs_out="${cfgoutputs_out} test/jemalloc_test.h"
+
+cfgoutputs_tup="Makefile"
+cfgoutputs_tup="${cfgoutputs_tup} doc/html.xsl:doc/html.xsl.in"
+cfgoutputs_tup="${cfgoutputs_tup} doc/manpages.xsl:doc/manpages.xsl.in"
+cfgoutputs_tup="${cfgoutputs_tup} doc/jemalloc${install_suffix}.xml:doc/jemalloc.xml.in"
+cfgoutputs_tup="${cfgoutputs_tup} include/jemalloc/jemalloc${install_suffix}.h:include/jemalloc/jemalloc.h.in"
+cfgoutputs_tup="${cfgoutputs_tup} include/jemalloc/internal/jemalloc_internal.h"
+cfgoutputs_tup="${cfgoutputs_tup} test/jemalloc_test.h:test/jemalloc_test.h.in"
+
+cfghdrs_in="${srcroot}include/jemalloc/jemalloc_defs.h.in"
+cfghdrs_in="${cfghdrs_in} ${srcroot}include/jemalloc/internal/size_classes.sh"
+
+cfghdrs_out="include/jemalloc/jemalloc_defs${install_suffix}.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/internal/size_classes.h"
+
+cfghdrs_tup="include/jemalloc/jemalloc_defs${install_suffix}.h:include/jemalloc/jemalloc_defs.h.in"
+
+dnl Do not silence irrelevant compiler warnings by default, since enabling this
+dnl option incurs a performance penalty.
+AC_ARG_ENABLE([cc-silence],
+  [AS_HELP_STRING([--enable-cc-silence],
+                  [Silence irrelevant compiler warnings])],
+[if test "x$enable_cc_silence" = "xno" ; then
+  enable_cc_silence="0"
+else
+  enable_cc_silence="1"
+fi
+],
+[enable_cc_silence="0"]
+)
+if test "x$enable_cc_silence" = "x1" ; then
+  AC_DEFINE([JEMALLOC_CC_SILENCE], [ ])
+fi
+
+dnl Do not compile with debugging by default.
+AC_ARG_ENABLE([debug],
+  [AS_HELP_STRING([--enable-debug], [Build debugging code (implies --enable-ivsalloc)])],
+[if test "x$enable_debug" = "xno" ; then
+  enable_debug="0"
+else
+  enable_debug="1"
+fi
+],
+[enable_debug="0"]
+)
+if test "x$enable_debug" = "x1" ; then
+  AC_DEFINE([JEMALLOC_DEBUG], [ ])
+  enable_ivsalloc="1"
+fi
+AC_SUBST([enable_debug])
+
+dnl Do not validate pointers by default.
+AC_ARG_ENABLE([ivsalloc],
+  [AS_HELP_STRING([--enable-ivsalloc], [Validate pointers passed through the public API])],
+[if test "x$enable_ivsalloc" = "xno" ; then
+  enable_ivsalloc="0"
+else
+  enable_ivsalloc="1"
+fi
+],
+[enable_ivsalloc="0"]
+)
+if test "x$enable_ivsalloc" = "x1" ; then
+  AC_DEFINE([JEMALLOC_IVSALLOC], [ ])
+fi
+
+dnl Only optimize if not debugging.
+if test "x$enable_debug" = "x0" -a "x$no_CFLAGS" = "xyes" ; then
+  dnl Make sure that an optimization flag was not specified in EXTRA_CFLAGS.
+  optimize="no"
+  echo "$EXTRA_CFLAGS" | grep "\-O" >/dev/null || optimize="yes"
+  if test "x${optimize}" = "xyes" ; then
+    if test "x$GCC" = "xyes" ; then
+      JE_CFLAGS_APPEND([-O3])
+      JE_CFLAGS_APPEND([-funroll-loops])
+    elif test "x$je_cv_msvc" = "xyes" ; then
+      JE_CFLAGS_APPEND([-O2])
+    else
+      JE_CFLAGS_APPEND([-O])
+    fi
+  fi
+fi
+
+dnl Enable statistics calculation by default.
+AC_ARG_ENABLE([stats],
+  [AS_HELP_STRING([--disable-stats],
+                  [Disable statistics calculation/reporting])],
+[if test "x$enable_stats" = "xno" ; then
+  enable_stats="0"
+else
+  enable_stats="1"
+fi
+],
+[enable_stats="1"]
+)
+if test "x$enable_stats" = "x1" ; then
+  AC_DEFINE([JEMALLOC_STATS], [ ])
+fi
+AC_SUBST([enable_stats])
+
+dnl Do not enable profiling by default.
+AC_ARG_ENABLE([prof],
+  [AS_HELP_STRING([--enable-prof], [Enable allocation profiling])],
+[if test "x$enable_prof" = "xno" ; then
+  enable_prof="0"
+else
+  enable_prof="1"
+fi
+],
+[enable_prof="0"]
+)
+if test "x$enable_prof" = "x1" ; then
+  backtrace_method=""
+else
+  backtrace_method="N/A"
+fi
+
+AC_ARG_ENABLE([prof-libunwind],
+  [AS_HELP_STRING([--enable-prof-libunwind], [Use libunwind for backtracing])],
+[if test "x$enable_prof_libunwind" = "xno" ; then
+  enable_prof_libunwind="0"
+else
+  enable_prof_libunwind="1"
+fi
+],
+[enable_prof_libunwind="0"]
+)
+AC_ARG_WITH([static_libunwind],
+  [AS_HELP_STRING([--with-static-libunwind=<libunwind.a>],
+  [Path to static libunwind library; use rather than dynamically linking])],
+if test "x$with_static_libunwind" = "xno" ; then
+  LUNWIND="-lunwind"
+else
+  if test ! -f "$with_static_libunwind" ; then
+    AC_MSG_ERROR([Static libunwind not found: $with_static_libunwind])
+  fi
+  LUNWIND="$with_static_libunwind"
+fi,
+  LUNWIND="-lunwind"
+)
+if test "x$backtrace_method" = "x" -a "x$enable_prof_libunwind" = "x1" ; then
+  AC_CHECK_HEADERS([libunwind.h], , [enable_prof_libunwind="0"])
+  if test "x$LUNWIND" = "x-lunwind" ; then
+    AC_CHECK_LIB([unwind], [backtrace], [LIBS="$LIBS $LUNWIND"],
+                 [enable_prof_libunwind="0"])
+  else
+    LIBS="$LIBS $LUNWIND"
+  fi
+  if test "x${enable_prof_libunwind}" = "x1" ; then
+    backtrace_method="libunwind"
+    AC_DEFINE([JEMALLOC_PROF_LIBUNWIND], [ ])
+  fi
+fi
+
+AC_ARG_ENABLE([prof-libgcc],
+  [AS_HELP_STRING([--disable-prof-libgcc],
+  [Do not use libgcc for backtracing])],
+[if test "x$enable_prof_libgcc" = "xno" ; then
+  enable_prof_libgcc="0"
+else
+  enable_prof_libgcc="1"
+fi
+],
+[enable_prof_libgcc="1"]
+)
+if test "x$backtrace_method" = "x" -a "x$enable_prof_libgcc" = "x1" \
+     -a "x$GCC" = "xyes" ; then
+  AC_CHECK_HEADERS([unwind.h], , [enable_prof_libgcc="0"])
+  AC_CHECK_LIB([gcc], [_Unwind_Backtrace], [LIBS="$LIBS -lgcc"], [enable_prof_libgcc="0"])
+  dnl The following is conservative, in that it only has entries for CPUs on
+  dnl which jemalloc has been tested.
+  AC_MSG_CHECKING([libgcc-based backtracing reliability on ${host_cpu}])
+  case "${host_cpu}" in
+    i[[3456]]86)
+      AC_MSG_RESULT([unreliable])
+      enable_prof_libgcc="0";
+      ;;
+    x86_64)
+      AC_MSG_RESULT([reliable])
+      ;;
+    *)
+      AC_MSG_RESULT([unreliable])
+      enable_prof_libgcc="0";
+      ;;
+  esac
+  if test "x${enable_prof_libgcc}" = "x1" ; then
+    backtrace_method="libgcc"
+    AC_DEFINE([JEMALLOC_PROF_LIBGCC], [ ])
+  fi
+else
+  enable_prof_libgcc="0"
+fi
+
+AC_ARG_ENABLE([prof-gcc],
+  [AS_HELP_STRING([--disable-prof-gcc],
+  [Do not use gcc intrinsics for backtracing])],
+[if test "x$enable_prof_gcc" = "xno" ; then
+  enable_prof_gcc="0"
+else
+  enable_prof_gcc="1"
+fi
+],
+[enable_prof_gcc="1"]
+)
+if test "x$backtrace_method" = "x" -a "x$enable_prof_gcc" = "x1" \
+     -a "x$GCC" = "xyes" ; then
+  backtrace_method="gcc intrinsics"
+  AC_DEFINE([JEMALLOC_PROF_GCC], [ ])
+else
+  enable_prof_gcc="0"
+fi
+
+if test "x$backtrace_method" = "x" ; then
+  backtrace_method="none (disabling profiling)"
+  enable_prof="0"
+fi
+AC_MSG_CHECKING([configured backtracing method])
+AC_MSG_RESULT([$backtrace_method])
+if test "x$enable_prof" = "x1" ; then
+  if test "x${force_tls}" = "x0" ; then
+    AC_MSG_ERROR([Heap profiling requires TLS]);
+  fi
+  force_tls="1"
+  AC_DEFINE([JEMALLOC_PROF], [ ])
+fi
+AC_SUBST([enable_prof])
+
+dnl Enable thread-specific caching by default.
+AC_ARG_ENABLE([tcache],
+  [AS_HELP_STRING([--disable-tcache], [Disable per thread caches])],
+[if test "x$enable_tcache" = "xno" ; then
+  enable_tcache="0"
+else
+  enable_tcache="1"
+fi
+],
+[enable_tcache="1"]
+)
+if test "x$enable_tcache" = "x1" ; then
+  AC_DEFINE([JEMALLOC_TCACHE], [ ])
+fi
+AC_SUBST([enable_tcache])
+
+dnl Disable mremap() for huge realloc() by default.
+AC_ARG_ENABLE([mremap],
+  [AS_HELP_STRING([--enable-mremap], [Enable mremap(2) for huge realloc()])],
+[if test "x$enable_mremap" = "xno" ; then
+  enable_mremap="0"
+else
+  enable_mremap="1"
+fi
+],
+[enable_mremap="0"]
+)
+if test "x$enable_mremap" = "x1" ; then
+  JE_COMPILABLE([mremap(...MREMAP_FIXED...)], [
+#define _GNU_SOURCE
+#include <sys/mman.h>
+], [
+void *p = mremap((void *)0, 0, 0, MREMAP_MAYMOVE|MREMAP_FIXED, (void *)0);
+], [je_cv_mremap_fixed])
+  if test "x${je_cv_mremap_fixed}" = "xno" ; then
+    enable_mremap="0"
+  fi
+fi
+if test "x$enable_mremap" = "x1" ; then
+  AC_DEFINE([JEMALLOC_MREMAP], [ ])
+fi
+AC_SUBST([enable_mremap])
+
+dnl Enable VM deallocation via munmap() by default.
+AC_ARG_ENABLE([munmap],
+  [AS_HELP_STRING([--disable-munmap], [Disable VM deallocation via munmap(2)])],
+[if test "x$enable_munmap" = "xno" ; then
+  enable_munmap="0"
+else
+  enable_munmap="1"
+fi
+],
+[enable_munmap="${default_munmap}"]
+)
+if test "x$enable_munmap" = "x1" ; then
+  AC_DEFINE([JEMALLOC_MUNMAP], [ ])
+fi
+AC_SUBST([enable_munmap])
+
+dnl Do not enable allocation from DSS by default.
+AC_ARG_ENABLE([dss],
+  [AS_HELP_STRING([--enable-dss], [Enable allocation from DSS])],
+[if test "x$enable_dss" = "xno" ; then
+  enable_dss="0"
+else
+  enable_dss="1"
+fi
+],
+[enable_dss="0"]
+)
+dnl Check whether the BSD/SUSv1 sbrk() exists.  If not, disable DSS support.
+AC_CHECK_FUNC([sbrk], [have_sbrk="1"], [have_sbrk="0"])
+if test "x$have_sbrk" = "x1" ; then
+  AC_DEFINE([JEMALLOC_HAVE_SBRK], [ ])
+else
+  enable_dss="0"
+fi
+
+if test "x$enable_dss" = "x1" ; then
+  AC_DEFINE([JEMALLOC_DSS], [ ])
+fi
+AC_SUBST([enable_dss])
+
+dnl Support the junk/zero filling option by default.
+AC_ARG_ENABLE([fill],
+  [AS_HELP_STRING([--disable-fill],
+                  [Disable support for junk/zero filling, quarantine, and redzones])],
+[if test "x$enable_fill" = "xno" ; then
+  enable_fill="0"
+else
+  enable_fill="1"
+fi
+],
+[enable_fill="1"]
+)
+if test "x$enable_fill" = "x1" ; then
+  AC_DEFINE([JEMALLOC_FILL], [ ])
+fi
+AC_SUBST([enable_fill])
+
+dnl Disable utrace(2)-based tracing by default.
+AC_ARG_ENABLE([utrace],
+  [AS_HELP_STRING([--enable-utrace], [Enable utrace(2)-based tracing])],
+[if test "x$enable_utrace" = "xno" ; then
+  enable_utrace="0"
+else
+  enable_utrace="1"
+fi
+],
+[enable_utrace="0"]
+)
+JE_COMPILABLE([utrace(2)], [
+#include <sys/types.h>
+#include <sys/param.h>
+#include <sys/time.h>
+#include <sys/uio.h>
+#include <sys/ktrace.h>
+], [
+	utrace((void *)0, 0);
+], [je_cv_utrace])
+if test "x${je_cv_utrace}" = "xno" ; then
+  enable_utrace="0"
+fi
+if test "x$enable_utrace" = "x1" ; then
+  AC_DEFINE([JEMALLOC_UTRACE], [ ])
+fi
+AC_SUBST([enable_utrace])
+
+dnl Support Valgrind by default.
+AC_ARG_ENABLE([valgrind],
+  [AS_HELP_STRING([--disable-valgrind], [Disable support for Valgrind])],
+[if test "x$enable_valgrind" = "xno" ; then
+  enable_valgrind="0"
+else
+  enable_valgrind="1"
+fi
+],
+[enable_valgrind="1"]
+)
+if test "x$enable_valgrind" = "x1" ; then
+  JE_COMPILABLE([valgrind], [
+#include <valgrind/valgrind.h>
+#include <valgrind/memcheck.h>
+
+#if !defined(VALGRIND_RESIZEINPLACE_BLOCK)
+#  error "Incompatible Valgrind version"
+#endif
+], [], [je_cv_valgrind])
+  if test "x${je_cv_valgrind}" = "xno" ; then
+    enable_valgrind="0"
+  fi
+  if test "x$enable_valgrind" = "x1" ; then
+    AC_DEFINE([JEMALLOC_VALGRIND], [ ])
+  fi
+fi
+AC_SUBST([enable_valgrind])
+
+dnl Do not support the xmalloc option by default.
+AC_ARG_ENABLE([xmalloc],
+  [AS_HELP_STRING([--enable-xmalloc], [Support xmalloc option])],
+[if test "x$enable_xmalloc" = "xno" ; then
+  enable_xmalloc="0"
+else
+  enable_xmalloc="1"
+fi
+],
+[enable_xmalloc="0"]
+)
+if test "x$enable_xmalloc" = "x1" ; then
+  AC_DEFINE([JEMALLOC_XMALLOC], [ ])
+fi
+AC_SUBST([enable_xmalloc])
+
+AC_CACHE_CHECK([STATIC_PAGE_SHIFT],
+               [je_cv_static_page_shift],
+               AC_RUN_IFELSE([AC_LANG_PROGRAM(
+[[
+#include <strings.h>
+#ifdef _WIN32
+#include <windows.h>
+#else
+#include <unistd.h>
+#endif
+#include <stdio.h>
+]],
+[[
+    int result;
+    FILE *f;
+
+#ifdef _WIN32
+    SYSTEM_INFO si;
+    GetSystemInfo(&si);
+    result = si.dwPageSize;
+#else
+    result = sysconf(_SC_PAGESIZE);
+#endif
+    if (result == -1) {
+	return 1;
+    }
+    result = ffsl(result) - 1;
+
+    f = fopen("conftest.out", "w");
+    if (f == NULL) {
+	return 1;
+    }
+    fprintf(f, "%d\n", result);
+    fclose(f);
+
+    return 0;
+]])],
+                             [je_cv_static_page_shift=`cat conftest.out`],
+                             [je_cv_static_page_shift=undefined]))
+
+if test "x$je_cv_static_page_shift" != "xundefined"; then
+   AC_DEFINE_UNQUOTED([STATIC_PAGE_SHIFT], [$je_cv_static_page_shift])
+else
+   AC_MSG_ERROR([cannot determine value for STATIC_PAGE_SHIFT])
+fi
+
+dnl ============================================================================
+dnl jemalloc configuration.
+dnl 
+
+dnl Set VERSION if source directory has an embedded git repository.
+if test -d "${srcroot}.git" ; then
+  git describe --long --abbrev=40 > ${srcroot}VERSION
+fi
+jemalloc_version=`cat ${srcroot}VERSION`
+jemalloc_version_major=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print [$]1}'`
+jemalloc_version_minor=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print [$]2}'`
+jemalloc_version_bugfix=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print [$]3}'`
+jemalloc_version_nrev=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print [$]4}'`
+jemalloc_version_gid=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print [$]5}'`
+AC_SUBST([jemalloc_version])
+AC_SUBST([jemalloc_version_major])
+AC_SUBST([jemalloc_version_minor])
+AC_SUBST([jemalloc_version_bugfix])
+AC_SUBST([jemalloc_version_nrev])
+AC_SUBST([jemalloc_version_gid])
+
+dnl ============================================================================
+dnl Configure pthreads.
+
+if test "x$abi" != "xpecoff" ; then
+  AC_CHECK_HEADERS([pthread.h], , [AC_MSG_ERROR([pthread.h is missing])])
+  dnl Some systems may embed pthreads functionality in libc; check for libpthread
+  dnl first, but try libc too before failing.
+  AC_CHECK_LIB([pthread], [pthread_create], [LIBS="$LIBS -lpthread"],
+               [AC_SEARCH_LIBS([pthread_create], , ,
+                               AC_MSG_ERROR([libpthread is missing]))])
+fi
+
+CPPFLAGS="$CPPFLAGS -D_REENTRANT"
+
+dnl Check whether the BSD-specific _malloc_thread_cleanup() exists.  If so, use
+dnl it rather than pthreads TSD cleanup functions to support cleanup during
+dnl thread exit, in order to avoid pthreads library recursion during
+dnl bootstrapping.
+AC_CHECK_FUNC([_malloc_thread_cleanup],
+              [have__malloc_thread_cleanup="1"],
+              [have__malloc_thread_cleanup="0"]
+             )
+if test "x$have__malloc_thread_cleanup" = "x1" ; then
+  AC_DEFINE([JEMALLOC_MALLOC_THREAD_CLEANUP], [ ])
+  force_tls="1"
+fi
+
+dnl Check whether the BSD-specific _pthread_mutex_init_calloc_cb() exists.  If
+dnl so, mutex initialization causes allocation, and we need to implement this
+dnl callback function in order to prevent recursive allocation.
+AC_CHECK_FUNC([_pthread_mutex_init_calloc_cb],
+              [have__pthread_mutex_init_calloc_cb="1"],
+              [have__pthread_mutex_init_calloc_cb="0"]
+             )
+if test "x$have__pthread_mutex_init_calloc_cb" = "x1" ; then
+  AC_DEFINE([JEMALLOC_MUTEX_INIT_CB])
+fi
+
+dnl Disable lazy locking by default.
+AC_ARG_ENABLE([lazy_lock],
+  [AS_HELP_STRING([--enable-lazy-lock],
+  [Enable lazy locking (only lock when multi-threaded)])],
+[if test "x$enable_lazy_lock" = "xno" ; then
+  enable_lazy_lock="0"
+else
+  enable_lazy_lock="1"
+fi
+],
+[enable_lazy_lock="0"]
+)
+if test "x$enable_lazy_lock" = "x0" -a "x${force_lazy_lock}" = "x1" ; then
+  AC_MSG_RESULT([Forcing lazy-lock to avoid allocator/threading bootstrap issues])
+  enable_lazy_lock="1"
+fi
+if test "x$enable_lazy_lock" = "x1" ; then
+  if test "x$abi" != "xpecoff" ; then
+    AC_CHECK_HEADERS([dlfcn.h], , [AC_MSG_ERROR([dlfcn.h is missing])])
+    AC_CHECK_FUNC([dlsym], [],
+      [AC_CHECK_LIB([dl], [dlsym], [LIBS="$LIBS -ldl"],
+                    [AC_MSG_ERROR([libdl is missing])])
+      ])
+  fi
+  AC_DEFINE([JEMALLOC_LAZY_LOCK], [ ])
+fi
+AC_SUBST([enable_lazy_lock])
+
+AC_ARG_ENABLE([tls],
+  [AS_HELP_STRING([--disable-tls], [Disable thread-local storage (__thread keyword)])],
+if test "x$enable_tls" = "xno" ; then
+  enable_tls="0"
+else
+  enable_tls="1"
+fi
+,
+enable_tls="1"
+)
+if test "x${enable_tls}" = "x0" -a "x${force_tls}" = "x1" ; then
+  AC_MSG_RESULT([Forcing TLS to avoid allocator/threading bootstrap issues])
+  enable_tls="1"
+fi
+if test "x${enable_tls}" = "x1" -a "x${force_tls}" = "x0" ; then
+  AC_MSG_RESULT([Forcing no TLS to avoid allocator/threading bootstrap issues])
+  enable_tls="0"
+fi
+if test "x${enable_tls}" = "x1" ; then
+AC_MSG_CHECKING([for TLS])
+AC_COMPILE_IFELSE([AC_LANG_PROGRAM(
+[[
+    __thread int x;
+]], [[
+    x = 42;
+
+    return 0;
+]])],
+              AC_MSG_RESULT([yes]),
+              AC_MSG_RESULT([no])
+              enable_tls="0")
+fi
+AC_SUBST([enable_tls])
+if test "x${enable_tls}" = "x1" ; then
+  AC_DEFINE_UNQUOTED([JEMALLOC_TLS], [ ])
+elif test "x${force_tls}" = "x1" ; then
+  AC_MSG_ERROR([Failed to configure TLS, which is mandatory for correct function])
+fi
+
+dnl ============================================================================
+dnl Check for ffsl(3), and fail if not found.  This function exists on all
+dnl platforms that jemalloc currently has a chance of functioning on without
+dnl modification.
+JE_COMPILABLE([a program using ffsl], [
+#include <stdio.h>
+#include <strings.h>
+#include <string.h>
+], [
+	{
+		int rv = ffsl(0x08);
+		printf("%d\n", rv);
+	}
+], [je_cv_function_ffsl])
+if test "x${je_cv_function_ffsl}" != "xyes" ; then
+   AC_MSG_ERROR([Cannot build without ffsl(3)])
+fi
+
+dnl ============================================================================
+dnl Check for atomic(9) operations as provided on FreeBSD.
+
+JE_COMPILABLE([atomic(9)], [
+#include <sys/types.h>
+#include <machine/atomic.h>
+#include <inttypes.h>
+], [
+	{
+		uint32_t x32 = 0;
+		volatile uint32_t *x32p = &x32;
+		atomic_fetchadd_32(x32p, 1);
+	}
+	{
+		unsigned long xlong = 0;
+		volatile unsigned long *xlongp = &xlong;
+		atomic_fetchadd_long(xlongp, 1);
+	}
+], [je_cv_atomic9])
+if test "x${je_cv_atomic9}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_ATOMIC9])
+fi
+
+dnl ============================================================================
+dnl Check for atomic(3) operations as provided on Darwin.
+
+JE_COMPILABLE([Darwin OSAtomic*()], [
+#include <libkern/OSAtomic.h>
+#include <inttypes.h>
+], [
+	{
+		int32_t x32 = 0;
+		volatile int32_t *x32p = &x32;
+		OSAtomicAdd32(1, x32p);
+	}
+	{
+		int64_t x64 = 0;
+		volatile int64_t *x64p = &x64;
+		OSAtomicAdd64(1, x64p);
+	}
+], [je_cv_osatomic])
+if test "x${je_cv_osatomic}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_OSATOMIC], [ ])
+fi
+
+dnl ============================================================================
+dnl Check whether __sync_{add,sub}_and_fetch() are available despite
+dnl __GCC_HAVE_SYNC_COMPARE_AND_SWAP_n macros being undefined.
+
+AC_DEFUN([JE_SYNC_COMPARE_AND_SWAP_CHECK],[
+  AC_CACHE_CHECK([whether to force $1-bit __sync_{add,sub}_and_fetch()],
+               [je_cv_sync_compare_and_swap_$2],
+               [AC_LINK_IFELSE([AC_LANG_PROGRAM([
+                                                 #include <stdint.h>
+                                                ],
+                                                [
+                                                 #ifndef __GCC_HAVE_SYNC_COMPARE_AND_SWAP_$2
+                                                 {
+                                                    uint$1_t x$1 = 0;
+                                                    __sync_add_and_fetch(&x$1, 42);
+                                                    __sync_sub_and_fetch(&x$1, 1);
+                                                 }
+                                                 #else
+                                                 #error __GCC_HAVE_SYNC_COMPARE_AND_SWAP_$2 is defined, no need to force
+                                                 #endif
+                                                ])],
+                               [je_cv_sync_compare_and_swap_$2=yes],
+                               [je_cv_sync_compare_and_swap_$2=no])])
+
+  if test "x${je_cv_sync_compare_and_swap_$2}" = "xyes" ; then
+    AC_DEFINE([JE_FORCE_SYNC_COMPARE_AND_SWAP_$2], [ ])
+  fi
+])
+
+if test "x${je_cv_atomic9}" != "xyes" -a "x${je_cv_osatomic}" != "xyes" ; then
+  JE_SYNC_COMPARE_AND_SWAP_CHECK(32, 4)
+  JE_SYNC_COMPARE_AND_SWAP_CHECK(64, 8)
+fi
+
+dnl ============================================================================
+dnl Check for spinlock(3) operations as provided on Darwin.
+
+JE_COMPILABLE([Darwin OSSpin*()], [
+#include <libkern/OSAtomic.h>
+#include <inttypes.h>
+], [
+	OSSpinLock lock = 0;
+	OSSpinLockLock(&lock);
+	OSSpinLockUnlock(&lock);
+], [je_cv_osspin])
+if test "x${je_cv_osspin}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_OSSPIN], [ ])
+fi
+
+dnl ============================================================================
+dnl Darwin-related configuration.
+
+AC_ARG_ENABLE([zone-allocator],
+  [AS_HELP_STRING([--disable-zone-allocator],
+                  [Disable zone allocator for Darwin])],
+[if test "x$enable_zone_allocator" = "xno" ; then
+  enable_zone_allocator="0"
+else
+  enable_zone_allocator="1"
+fi
+],
+[if test "x${abi}" = "xmacho"; then
+  enable_zone_allocator="1"
+fi
+]
+)
+AC_SUBST([enable_zone_allocator])
+
+if test "x${enable_zone_allocator}" = "x1" ; then
+  if test "x${abi}" != "xmacho"; then
+    AC_MSG_ERROR([--enable-zone-allocator is only supported on Darwin])
+  fi
+  AC_DEFINE([JEMALLOC_IVSALLOC], [ ])
+  AC_DEFINE([JEMALLOC_ZONE], [ ])
+
+  dnl The szone version jumped from 3 to 6 between the OS X 10.5.x and 10.6
+  dnl releases.  malloc_zone_t and malloc_introspection_t have new fields in
+  dnl 10.6, which is the only source-level indication of the change.
+  AC_MSG_CHECKING([malloc zone version])
+  AC_DEFUN([JE_ZONE_PROGRAM],
+    [AC_LANG_PROGRAM(
+      [#include <malloc/malloc.h>],
+      [static foo[[sizeof($1) $2 sizeof(void *) * $3 ? 1 : -1]]]
+    )])
+
+  AC_COMPILE_IFELSE([JE_ZONE_PROGRAM(malloc_zone_t,==,14)],[JEMALLOC_ZONE_VERSION=3],[
+  AC_COMPILE_IFELSE([JE_ZONE_PROGRAM(malloc_zone_t,==,15)],[JEMALLOC_ZONE_VERSION=5],[
+  AC_COMPILE_IFELSE([JE_ZONE_PROGRAM(malloc_zone_t,==,16)],[
+    AC_COMPILE_IFELSE([JE_ZONE_PROGRAM(malloc_introspection_t,==,9)],[JEMALLOC_ZONE_VERSION=6],[
+    AC_COMPILE_IFELSE([JE_ZONE_PROGRAM(malloc_introspection_t,==,13)],[JEMALLOC_ZONE_VERSION=7],[JEMALLOC_ZONE_VERSION=]
+  )])],[
+  AC_COMPILE_IFELSE([JE_ZONE_PROGRAM(malloc_zone_t,==,17)],[JEMALLOC_ZONE_VERSION=8],[
+  AC_COMPILE_IFELSE([JE_ZONE_PROGRAM(malloc_zone_t,>,17)],[JEMALLOC_ZONE_VERSION=9],[JEMALLOC_ZONE_VERSION=]
+  )])])])])
+  if test "x${JEMALLOC_ZONE_VERSION}" = "x"; then
+    AC_MSG_RESULT([unsupported])
+    AC_MSG_ERROR([Unsupported malloc zone version])
+  fi
+  if test "${JEMALLOC_ZONE_VERSION}" = 9; then
+    JEMALLOC_ZONE_VERSION=8
+    AC_MSG_RESULT([> 8])
+  else
+    AC_MSG_RESULT([$JEMALLOC_ZONE_VERSION])
+  fi
+  AC_DEFINE_UNQUOTED(JEMALLOC_ZONE_VERSION, [$JEMALLOC_ZONE_VERSION])
+fi
+
+dnl ============================================================================
+dnl Check for typedefs, structures, and compiler characteristics.
+AC_HEADER_STDBOOL
+
+AC_CONFIG_COMMANDS([include/jemalloc/internal/size_classes.h], [
+  mkdir -p "include/jemalloc/internal"
+  "${srcdir}/include/jemalloc/internal/size_classes.sh" > "${objroot}include/jemalloc/internal/size_classes.h"
+])
+
+dnl Process .in files.
+AC_SUBST([cfghdrs_in])
+AC_SUBST([cfghdrs_out])
+AC_CONFIG_HEADERS([$cfghdrs_tup])
+
+dnl ============================================================================
+dnl Generate outputs.
+AC_CONFIG_FILES([$cfgoutputs_tup config.stamp bin/jemalloc.sh])
+AC_SUBST([cfgoutputs_in])
+AC_SUBST([cfgoutputs_out])
+AC_OUTPUT
+
+dnl ============================================================================
+dnl Print out the results of configuration.
+AC_MSG_RESULT([===============================================================================])
+AC_MSG_RESULT([jemalloc version   : ${jemalloc_version}])
+AC_MSG_RESULT([library revision   : ${rev}])
+AC_MSG_RESULT([])
+AC_MSG_RESULT([CC                 : ${CC}])
+AC_MSG_RESULT([CPPFLAGS           : ${CPPFLAGS}])
+AC_MSG_RESULT([CFLAGS             : ${CFLAGS}])
+AC_MSG_RESULT([LDFLAGS            : ${LDFLAGS}])
+AC_MSG_RESULT([LIBS               : ${LIBS}])
+AC_MSG_RESULT([RPATH_EXTRA        : ${RPATH_EXTRA}])
+AC_MSG_RESULT([])
+AC_MSG_RESULT([XSLTPROC           : ${XSLTPROC}])
+AC_MSG_RESULT([XSLROOT            : ${XSLROOT}])
+AC_MSG_RESULT([])
+AC_MSG_RESULT([PREFIX             : ${PREFIX}])
+AC_MSG_RESULT([BINDIR             : ${BINDIR}])
+AC_MSG_RESULT([INCLUDEDIR         : ${INCLUDEDIR}])
+AC_MSG_RESULT([LIBDIR             : ${LIBDIR}])
+AC_MSG_RESULT([DATADIR            : ${DATADIR}])
+AC_MSG_RESULT([MANDIR             : ${MANDIR}])
+AC_MSG_RESULT([])
+AC_MSG_RESULT([srcroot            : ${srcroot}])
+AC_MSG_RESULT([abs_srcroot        : ${abs_srcroot}])
+AC_MSG_RESULT([objroot            : ${objroot}])
+AC_MSG_RESULT([abs_objroot        : ${abs_objroot}])
+AC_MSG_RESULT([])
+AC_MSG_RESULT([JEMALLOC_PREFIX    : ${JEMALLOC_PREFIX}])
+AC_MSG_RESULT([JEMALLOC_PRIVATE_NAMESPACE])
+AC_MSG_RESULT([                   : ${JEMALLOC_PRIVATE_NAMESPACE}])
+AC_MSG_RESULT([install_suffix     : ${install_suffix}])
+AC_MSG_RESULT([autogen            : ${enable_autogen}])
+AC_MSG_RESULT([experimental       : ${enable_experimental}])
+AC_MSG_RESULT([cc-silence         : ${enable_cc_silence}])
+AC_MSG_RESULT([debug              : ${enable_debug}])
+AC_MSG_RESULT([stats              : ${enable_stats}])
+AC_MSG_RESULT([prof               : ${enable_prof}])
+AC_MSG_RESULT([prof-libunwind     : ${enable_prof_libunwind}])
+AC_MSG_RESULT([prof-libgcc        : ${enable_prof_libgcc}])
+AC_MSG_RESULT([prof-gcc           : ${enable_prof_gcc}])
+AC_MSG_RESULT([tcache             : ${enable_tcache}])
+AC_MSG_RESULT([fill               : ${enable_fill}])
+AC_MSG_RESULT([utrace             : ${enable_utrace}])
+AC_MSG_RESULT([valgrind           : ${enable_valgrind}])
+AC_MSG_RESULT([xmalloc            : ${enable_xmalloc}])
+AC_MSG_RESULT([mremap             : ${enable_mremap}])
+AC_MSG_RESULT([munmap             : ${enable_munmap}])
+AC_MSG_RESULT([dss                : ${enable_dss}])
+AC_MSG_RESULT([lazy_lock          : ${enable_lazy_lock}])
+AC_MSG_RESULT([tls                : ${enable_tls}])
+AC_MSG_RESULT([===============================================================================])
diff --git a/extra/jemalloc/doc/html.xsl.in b/extra/jemalloc/doc/html.xsl.in
new file mode 100644
index 00000000000..a91d9746f62
--- /dev/null
+++ b/extra/jemalloc/doc/html.xsl.in
@@ -0,0 +1,4 @@
+<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
+  <xsl:import href="@XSLROOT@/html/docbook.xsl"/>
+  <xsl:import href="@abs_srcroot@doc/stylesheet.xsl"/>
+</xsl:stylesheet>
diff --git a/extra/jemalloc/doc/jemalloc.3 b/extra/jemalloc/doc/jemalloc.3
new file mode 100644
index 00000000000..1462e2c2b34
--- /dev/null
+++ b/extra/jemalloc/doc/jemalloc.3
@@ -0,0 +1,1482 @@
+'\" t
+.\"     Title: JEMALLOC
+.\"    Author: Jason Evans
+.\" Generator: DocBook XSL Stylesheets v1.76.1 <http://docbook.sf.net/>
+.\"      Date: 03/06/2013
+.\"    Manual: User Manual
+.\"    Source: jemalloc 3.3.1-0-g9ef9d9e8c271cdf14f664b871a8f98c827714784
+.\"  Language: English
+.\"
+.TH "JEMALLOC" "3" "03/06/2013" "jemalloc 3.3.1-0-g9ef9d9e8c271" "User Manual"
+.\" -----------------------------------------------------------------
+.\" * Define some portability stuff
+.\" -----------------------------------------------------------------
+.\" ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+.\" http://bugs.debian.org/507673
+.\" http://lists.gnu.org/archive/html/groff/2009-02/msg00013.html
+.\" ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+.ie \n(.g .ds Aq \(aq
+.el       .ds Aq '
+.\" -----------------------------------------------------------------
+.\" * set default formatting
+.\" -----------------------------------------------------------------
+.\" disable hyphenation
+.nh
+.\" disable justification (adjust text to left margin only)
+.ad l
+.\" -----------------------------------------------------------------
+.\" * MAIN CONTENT STARTS HERE *
+.\" -----------------------------------------------------------------
+.SH "NAME"
+jemalloc \- general purpose memory allocation functions
+.SH "LIBRARY"
+.PP
+This manual describes jemalloc 3\&.3\&.1\-0\-g9ef9d9e8c271cdf14f664b871a8f98c827714784\&. More information can be found at the
+\m[blue]\fBjemalloc website\fR\m[]\&\s-2\u[1]\d\s+2\&.
+.SH "SYNOPSIS"
+.sp
+.ft B
+.nf
+#include <stdlib\&.h>
+#include <jemalloc/jemalloc\&.h>
+.fi
+.ft
+.SS "Standard API"
+.HP \w'void\ *malloc('u
+.BI "void *malloc(size_t\ " "size" ");"
+.HP \w'void\ *calloc('u
+.BI "void *calloc(size_t\ " "number" ", size_t\ " "size" ");"
+.HP \w'int\ posix_memalign('u
+.BI "int posix_memalign(void\ **" "ptr" ", size_t\ " "alignment" ", size_t\ " "size" ");"
+.HP \w'void\ *aligned_alloc('u
+.BI "void *aligned_alloc(size_t\ " "alignment" ", size_t\ " "size" ");"
+.HP \w'void\ *realloc('u
+.BI "void *realloc(void\ *" "ptr" ", size_t\ " "size" ");"
+.HP \w'void\ free('u
+.BI "void free(void\ *" "ptr" ");"
+.SS "Non\-standard API"
+.HP \w'size_t\ malloc_usable_size('u
+.BI "size_t malloc_usable_size(const\ void\ *" "ptr" ");"
+.HP \w'void\ malloc_stats_print('u
+.BI "void malloc_stats_print(void\ " "(*write_cb)" "\ (void\ *,\ const\ char\ *), void\ *" "cbopaque" ", const\ char\ *" "opts" ");"
+.HP \w'int\ mallctl('u
+.BI "int mallctl(const\ char\ *" "name" ", void\ *" "oldp" ", size_t\ *" "oldlenp" ", void\ *" "newp" ", size_t\ " "newlen" ");"
+.HP \w'int\ mallctlnametomib('u
+.BI "int mallctlnametomib(const\ char\ *" "name" ", size_t\ *" "mibp" ", size_t\ *" "miblenp" ");"
+.HP \w'int\ mallctlbymib('u
+.BI "int mallctlbymib(const\ size_t\ *" "mib" ", size_t\ " "miblen" ", void\ *" "oldp" ", size_t\ *" "oldlenp" ", void\ *" "newp" ", size_t\ " "newlen" ");"
+.HP \w'void\ (*malloc_message)('u
+.BI "void (*malloc_message)(void\ *" "cbopaque" ", const\ char\ *" "s" ");"
+.PP
+const char *\fImalloc_conf\fR;
+.SS "Experimental API"
+.HP \w'int\ allocm('u
+.BI "int allocm(void\ **" "ptr" ", size_t\ *" "rsize" ", size_t\ " "size" ", int\ " "flags" ");"
+.HP \w'int\ rallocm('u
+.BI "int rallocm(void\ **" "ptr" ", size_t\ *" "rsize" ", size_t\ " "size" ", size_t\ " "extra" ", int\ " "flags" ");"
+.HP \w'int\ sallocm('u
+.BI "int sallocm(const\ void\ *" "ptr" ", size_t\ *" "rsize" ", int\ " "flags" ");"
+.HP \w'int\ dallocm('u
+.BI "int dallocm(void\ *" "ptr" ", int\ " "flags" ");"
+.HP \w'int\ nallocm('u
+.BI "int nallocm(size_t\ *" "rsize" ", size_t\ " "size" ", int\ " "flags" ");"
+.SH "DESCRIPTION"
+.SS "Standard API"
+.PP
+The
+\fBmalloc\fR\fB\fR
+function allocates
+\fIsize\fR
+bytes of uninitialized memory\&. The allocated space is suitably aligned (after possible pointer coercion) for storage of any type of object\&.
+.PP
+The
+\fBcalloc\fR\fB\fR
+function allocates space for
+\fInumber\fR
+objects, each
+\fIsize\fR
+bytes in length\&. The result is identical to calling
+\fBmalloc\fR\fB\fR
+with an argument of
+\fInumber\fR
+*
+\fIsize\fR, with the exception that the allocated memory is explicitly initialized to zero bytes\&.
+.PP
+The
+\fBposix_memalign\fR\fB\fR
+function allocates
+\fIsize\fR
+bytes of memory such that the allocation\*(Aqs base address is an even multiple of
+\fIalignment\fR, and returns the allocation in the value pointed to by
+\fIptr\fR\&. The requested
+\fIalignment\fR
+must be a power of 2 at least as large as
+sizeof(\fBvoid *\fR)\&.
+.PP
+The
+\fBaligned_alloc\fR\fB\fR
+function allocates
+\fIsize\fR
+bytes of memory such that the allocation\*(Aqs base address is an even multiple of
+\fIalignment\fR\&. The requested
+\fIalignment\fR
+must be a power of 2\&. Behavior is undefined if
+\fIsize\fR
+is not an integral multiple of
+\fIalignment\fR\&.
+.PP
+The
+\fBrealloc\fR\fB\fR
+function changes the size of the previously allocated memory referenced by
+\fIptr\fR
+to
+\fIsize\fR
+bytes\&. The contents of the memory are unchanged up to the lesser of the new and old sizes\&. If the new size is larger, the contents of the newly allocated portion of the memory are undefined\&. Upon success, the memory referenced by
+\fIptr\fR
+is freed and a pointer to the newly allocated memory is returned\&. Note that
+\fBrealloc\fR\fB\fR
+may move the memory allocation, resulting in a different return value than
+\fIptr\fR\&. If
+\fIptr\fR
+is
+\fBNULL\fR, the
+\fBrealloc\fR\fB\fR
+function behaves identically to
+\fBmalloc\fR\fB\fR
+for the specified size\&.
+.PP
+The
+\fBfree\fR\fB\fR
+function causes the allocated memory referenced by
+\fIptr\fR
+to be made available for future allocations\&. If
+\fIptr\fR
+is
+\fBNULL\fR, no action occurs\&.
+.SS "Non\-standard API"
+.PP
+The
+\fBmalloc_usable_size\fR\fB\fR
+function returns the usable size of the allocation pointed to by
+\fIptr\fR\&. The return value may be larger than the size that was requested during allocation\&. The
+\fBmalloc_usable_size\fR\fB\fR
+function is not a mechanism for in\-place
+\fBrealloc\fR\fB\fR; rather it is provided solely as a tool for introspection purposes\&. Any discrepancy between the requested allocation size and the size reported by
+\fBmalloc_usable_size\fR\fB\fR
+should not be depended on, since such behavior is entirely implementation\-dependent\&.
+.PP
+The
+\fBmalloc_stats_print\fR\fB\fR
+function writes human\-readable summary statistics via the
+\fIwrite_cb\fR
+callback function pointer and
+\fIcbopaque\fR
+data passed to
+\fIwrite_cb\fR, or
+\fBmalloc_message\fR\fB\fR
+if
+\fIwrite_cb\fR
+is
+\fBNULL\fR\&. This function can be called repeatedly\&. General information that never changes during execution can be omitted by specifying "g" as a character within the
+\fIopts\fR
+string\&. Note that
+\fBmalloc_message\fR\fB\fR
+uses the
+\fBmallctl*\fR\fB\fR
+functions internally, so inconsistent statistics can be reported if multiple threads use these functions simultaneously\&. If
+\fB\-\-enable\-stats\fR
+is specified during configuration, \(lqm\(rq and \(lqa\(rq can be specified to omit merged arena and per arena statistics, respectively; \(lqb\(rq and \(lql\(rq can be specified to omit per size class statistics for bins and large objects, respectively\&. Unrecognized characters are silently ignored\&. Note that thread caching may prevent some statistics from being completely up to date, since extra locking would be required to merge counters that track thread cache operations\&.
+.PP
+The
+\fBmallctl\fR\fB\fR
+function provides a general interface for introspecting the memory allocator, as well as setting modifiable parameters and triggering actions\&. The period\-separated
+\fIname\fR
+argument specifies a location in a tree\-structured namespace; see the
+MALLCTL NAMESPACE
+section for documentation on the tree contents\&. To read a value, pass a pointer via
+\fIoldp\fR
+to adequate space to contain the value, and a pointer to its length via
+\fIoldlenp\fR; otherwise pass
+\fBNULL\fR
+and
+\fBNULL\fR\&. Similarly, to write a value, pass a pointer to the value via
+\fInewp\fR, and its length via
+\fInewlen\fR; otherwise pass
+\fBNULL\fR
+and
+\fB0\fR\&.
+.PP
+The
+\fBmallctlnametomib\fR\fB\fR
+function provides a way to avoid repeated name lookups for applications that repeatedly query the same portion of the namespace, by translating a name to a \(lqManagement Information Base\(rq (MIB) that can be passed repeatedly to
+\fBmallctlbymib\fR\fB\fR\&. Upon successful return from
+\fBmallctlnametomib\fR\fB\fR,
+\fImibp\fR
+contains an array of
+\fI*miblenp\fR
+integers, where
+\fI*miblenp\fR
+is the lesser of the number of components in
+\fIname\fR
+and the input value of
+\fI*miblenp\fR\&. Thus it is possible to pass a
+\fI*miblenp\fR
+that is smaller than the number of period\-separated name components, which results in a partial MIB that can be used as the basis for constructing a complete MIB\&. For name components that are integers (e\&.g\&. the 2 in
+"arenas\&.bin\&.2\&.size"), the corresponding MIB component will always be that integer\&. Therefore, it is legitimate to construct code like the following:
+.sp
+.if n \{\
+.RS 4
+.\}
+.nf
+unsigned nbins, i;
+
+int mib[4];
+size_t len, miblen;
+
+len = sizeof(nbins);
+mallctl("arenas\&.nbins", &nbins, &len, NULL, 0);
+
+miblen = 4;
+mallnametomib("arenas\&.bin\&.0\&.size", mib, &miblen);
+for (i = 0; i < nbins; i++) {
+	size_t bin_size;
+
+	mib[2] = i;
+	len = sizeof(bin_size);
+	mallctlbymib(mib, miblen, &bin_size, &len, NULL, 0);
+	/* Do something with bin_size\&.\&.\&. */
+}
+.fi
+.if n \{\
+.RE
+.\}
+.SS "Experimental API"
+.PP
+The experimental API is subject to change or removal without regard for backward compatibility\&. If
+\fB\-\-disable\-experimental\fR
+is specified during configuration, the experimental API is omitted\&.
+.PP
+The
+\fBallocm\fR\fB\fR,
+\fBrallocm\fR\fB\fR,
+\fBsallocm\fR\fB\fR,
+\fBdallocm\fR\fB\fR, and
+\fBnallocm\fR\fB\fR
+functions all have a
+\fIflags\fR
+argument that can be used to specify options\&. The functions only check the options that are contextually relevant\&. Use bitwise or (|) operations to specify one or more of the following:
+.PP
+\fBALLOCM_LG_ALIGN(\fR\fB\fIla\fR\fR\fB) \fR
+.RS 4
+Align the memory allocation to start at an address that is a multiple of
+(1 << \fIla\fR)\&. This macro does not validate that
+\fIla\fR
+is within the valid range\&.
+.RE
+.PP
+\fBALLOCM_ALIGN(\fR\fB\fIa\fR\fR\fB) \fR
+.RS 4
+Align the memory allocation to start at an address that is a multiple of
+\fIa\fR, where
+\fIa\fR
+is a power of two\&. This macro does not validate that
+\fIa\fR
+is a power of 2\&.
+.RE
+.PP
+\fBALLOCM_ZERO\fR
+.RS 4
+Initialize newly allocated memory to contain zero bytes\&. In the growing reallocation case, the real size prior to reallocation defines the boundary between untouched bytes and those that are initialized to contain zero bytes\&. If this option is absent, newly allocated memory is uninitialized\&.
+.RE
+.PP
+\fBALLOCM_NO_MOVE\fR
+.RS 4
+For reallocation, fail rather than moving the object\&. This constraint can apply to both growth and shrinkage\&.
+.RE
+.PP
+\fBALLOCM_ARENA(\fR\fB\fIa\fR\fR\fB) \fR
+.RS 4
+Use the arena specified by the index
+\fIa\fR\&. This macro does not validate that
+\fIa\fR
+specifies an arena in the valid range\&.
+.RE
+.PP
+The
+\fBallocm\fR\fB\fR
+function allocates at least
+\fIsize\fR
+bytes of memory, sets
+\fI*ptr\fR
+to the base address of the allocation, and sets
+\fI*rsize\fR
+to the real size of the allocation if
+\fIrsize\fR
+is not
+\fBNULL\fR\&. Behavior is undefined if
+\fIsize\fR
+is
+\fB0\fR\&.
+.PP
+The
+\fBrallocm\fR\fB\fR
+function resizes the allocation at
+\fI*ptr\fR
+to be at least
+\fIsize\fR
+bytes, sets
+\fI*ptr\fR
+to the base address of the allocation if it moved, and sets
+\fI*rsize\fR
+to the real size of the allocation if
+\fIrsize\fR
+is not
+\fBNULL\fR\&. If
+\fIextra\fR
+is non\-zero, an attempt is made to resize the allocation to be at least
+\fIsize\fR + \fIextra\fR)
+bytes, though inability to allocate the extra byte(s) will not by itself result in failure\&. Behavior is undefined if
+\fIsize\fR
+is
+\fB0\fR, or if
+(\fIsize\fR + \fIextra\fR > \fBSIZE_T_MAX\fR)\&.
+.PP
+The
+\fBsallocm\fR\fB\fR
+function sets
+\fI*rsize\fR
+to the real size of the allocation\&.
+.PP
+The
+\fBdallocm\fR\fB\fR
+function causes the memory referenced by
+\fIptr\fR
+to be made available for future allocations\&.
+.PP
+The
+\fBnallocm\fR\fB\fR
+function allocates no memory, but it performs the same size computation as the
+\fBallocm\fR\fB\fR
+function, and if
+\fIrsize\fR
+is not
+\fBNULL\fR
+it sets
+\fI*rsize\fR
+to the real size of the allocation that would result from the equivalent
+\fBallocm\fR\fB\fR
+function call\&. Behavior is undefined if
+\fIsize\fR
+is
+\fB0\fR\&.
+.SH "TUNING"
+.PP
+Once, when the first call is made to one of the memory allocation routines, the allocator initializes its internals based in part on various options that can be specified at compile\- or run\-time\&.
+.PP
+The string pointed to by the global variable
+\fImalloc_conf\fR, the \(lqname\(rq of the file referenced by the symbolic link named
+/etc/malloc\&.conf, and the value of the environment variable
+\fBMALLOC_CONF\fR, will be interpreted, in that order, from left to right as options\&.
+.PP
+An options string is a comma\-separated list of option:value pairs\&. There is one key corresponding to each
+"opt\&.*"
+mallctl (see the
+MALLCTL NAMESPACE
+section for options documentation)\&. For example,
+abort:true,narenas:1
+sets the
+"opt\&.abort"
+and
+"opt\&.narenas"
+options\&. Some options have boolean values (true/false), others have integer values (base 8, 10, or 16, depending on prefix), and yet others have raw string values\&.
+.SH "IMPLEMENTATION NOTES"
+.PP
+Traditionally, allocators have used
+\fBsbrk\fR(2)
+to obtain memory, which is suboptimal for several reasons, including race conditions, increased fragmentation, and artificial limitations on maximum usable memory\&. If
+\fB\-\-enable\-dss\fR
+is specified during configuration, this allocator uses both
+\fBmmap\fR(2)
+and
+\fBsbrk\fR(2), in that order of preference; otherwise only
+\fBmmap\fR(2)
+is used\&.
+.PP
+This allocator uses multiple arenas in order to reduce lock contention for threaded programs on multi\-processor systems\&. This works well with regard to threading scalability, but incurs some costs\&. There is a small fixed per\-arena overhead, and additionally, arenas manage memory completely independently of each other, which means a small fixed increase in overall memory fragmentation\&. These overheads are not generally an issue, given the number of arenas normally used\&. Note that using substantially more arenas than the default is not likely to improve performance, mainly due to reduced cache performance\&. However, it may make sense to reduce the number of arenas if an application does not make much use of the allocation functions\&.
+.PP
+In addition to multiple arenas, unless
+\fB\-\-disable\-tcache\fR
+is specified during configuration, this allocator supports thread\-specific caching for small and large objects, in order to make it possible to completely avoid synchronization for most allocation requests\&. Such caching allows very fast allocation in the common case, but it increases memory usage and fragmentation, since a bounded number of objects can remain allocated in each thread cache\&.
+.PP
+Memory is conceptually broken into equal\-sized chunks, where the chunk size is a power of two that is greater than the page size\&. Chunks are always aligned to multiples of the chunk size\&. This alignment makes it possible to find metadata for user objects very quickly\&.
+.PP
+User objects are broken into three categories according to size: small, large, and huge\&. Small objects are smaller than one page\&. Large objects are smaller than the chunk size\&. Huge objects are a multiple of the chunk size\&. Small and large objects are managed by arenas; huge objects are managed separately in a single data structure that is shared by all threads\&. Huge objects are used by applications infrequently enough that this single data structure is not a scalability issue\&.
+.PP
+Each chunk that is managed by an arena tracks its contents as runs of contiguous pages (unused, backing a set of small objects, or backing one large object)\&. The combination of chunk alignment and chunk page maps makes it possible to determine all metadata regarding small and large allocations in constant time\&.
+.PP
+Small objects are managed in groups by page runs\&. Each run maintains a frontier and free list to track which regions are in use\&. Allocation requests that are no more than half the quantum (8 or 16, depending on architecture) are rounded up to the nearest power of two that is at least
+sizeof(\fBdouble\fR)\&. All other small object size classes are multiples of the quantum, spaced such that internal fragmentation is limited to approximately 25% for all but the smallest size classes\&. Allocation requests that are larger than the maximum small size class, but small enough to fit in an arena\-managed chunk (see the
+"opt\&.lg_chunk"
+option), are rounded up to the nearest run size\&. Allocation requests that are too large to fit in an arena\-managed chunk are rounded up to the nearest multiple of the chunk size\&.
+.PP
+Allocations are packed tightly together, which can be an issue for multi\-threaded applications\&. If you need to assure that allocations do not suffer from cacheline sharing, round your allocation requests up to the nearest multiple of the cacheline size, or specify cacheline alignment when allocating\&.
+.PP
+Assuming 4 MiB chunks, 4 KiB pages, and a 16\-byte quantum on a 64\-bit system, the size classes in each category are as shown in
+Table 1\&.
+.sp
+.it 1 an-trap
+.nr an-no-space-flag 1
+.nr an-break-flag 1
+.br
+.B Table\ \&1.\ \&Size classes
+.TS
+allbox tab(:);
+lB rB lB.
+T{
+Category
+T}:T{
+Spacing
+T}:T{
+Size
+T}
+.T&
+l r l
+^ r l
+^ r l
+^ r l
+^ r l
+^ r l
+^ r l
+l r l
+l r l.
+T{
+Small
+T}:T{
+lg
+T}:T{
+[8]
+T}
+:T{
+16
+T}:T{
+[16, 32, 48, \&.\&.\&., 128]
+T}
+:T{
+32
+T}:T{
+[160, 192, 224, 256]
+T}
+:T{
+64
+T}:T{
+[320, 384, 448, 512]
+T}
+:T{
+128
+T}:T{
+[640, 768, 896, 1024]
+T}
+:T{
+256
+T}:T{
+[1280, 1536, 1792, 2048]
+T}
+:T{
+512
+T}:T{
+[2560, 3072, 3584]
+T}
+T{
+Large
+T}:T{
+4 KiB
+T}:T{
+[4 KiB, 8 KiB, 12 KiB, \&.\&.\&., 4072 KiB]
+T}
+T{
+Huge
+T}:T{
+4 MiB
+T}:T{
+[4 MiB, 8 MiB, 12 MiB, \&.\&.\&.]
+T}
+.TE
+.sp 1
+.SH "MALLCTL NAMESPACE"
+.PP
+The following names are defined in the namespace accessible via the
+\fBmallctl*\fR\fB\fR
+functions\&. Value types are specified in parentheses, their readable/writable statuses are encoded as
+rw,
+r\-,
+\-w, or
+\-\-, and required build configuration flags follow, if any\&. A name element encoded as
+<i>
+or
+<j>
+indicates an integer component, where the integer varies from 0 to some upper value that must be determined via introspection\&. In the case of
+"stats\&.arenas\&.<i>\&.*",
+<i>
+equal to
+"arenas\&.narenas"
+can be used to access the summation of statistics from all arenas\&. Take special note of the
+"epoch"
+mallctl, which controls refreshing of cached dynamic statistics\&.
+.PP
+"version" (\fBconst char *\fR) r\-
+.RS 4
+Return the jemalloc version string\&.
+.RE
+.PP
+"epoch" (\fBuint64_t\fR) rw
+.RS 4
+If a value is passed in, refresh the data from which the
+\fBmallctl*\fR\fB\fR
+functions report values, and increment the epoch\&. Return the current epoch\&. This is useful for detecting whether another thread caused a refresh\&.
+.RE
+.PP
+"config\&.debug" (\fBbool\fR) r\-
+.RS 4
+\fB\-\-enable\-debug\fR
+was specified during build configuration\&.
+.RE
+.PP
+"config\&.dss" (\fBbool\fR) r\-
+.RS 4
+\fB\-\-enable\-dss\fR
+was specified during build configuration\&.
+.RE
+.PP
+"config\&.fill" (\fBbool\fR) r\-
+.RS 4
+\fB\-\-enable\-fill\fR
+was specified during build configuration\&.
+.RE
+.PP
+"config\&.lazy_lock" (\fBbool\fR) r\-
+.RS 4
+\fB\-\-enable\-lazy\-lock\fR
+was specified during build configuration\&.
+.RE
+.PP
+"config\&.mremap" (\fBbool\fR) r\-
+.RS 4
+\fB\-\-enable\-mremap\fR
+was specified during build configuration\&.
+.RE
+.PP
+"config\&.munmap" (\fBbool\fR) r\-
+.RS 4
+\fB\-\-enable\-munmap\fR
+was specified during build configuration\&.
+.RE
+.PP
+"config\&.prof" (\fBbool\fR) r\-
+.RS 4
+\fB\-\-enable\-prof\fR
+was specified during build configuration\&.
+.RE
+.PP
+"config\&.prof_libgcc" (\fBbool\fR) r\-
+.RS 4
+\fB\-\-disable\-prof\-libgcc\fR
+was not specified during build configuration\&.
+.RE
+.PP
+"config\&.prof_libunwind" (\fBbool\fR) r\-
+.RS 4
+\fB\-\-enable\-prof\-libunwind\fR
+was specified during build configuration\&.
+.RE
+.PP
+"config\&.stats" (\fBbool\fR) r\-
+.RS 4
+\fB\-\-enable\-stats\fR
+was specified during build configuration\&.
+.RE
+.PP
+"config\&.tcache" (\fBbool\fR) r\-
+.RS 4
+\fB\-\-disable\-tcache\fR
+was not specified during build configuration\&.
+.RE
+.PP
+"config\&.tls" (\fBbool\fR) r\-
+.RS 4
+\fB\-\-disable\-tls\fR
+was not specified during build configuration\&.
+.RE
+.PP
+"config\&.utrace" (\fBbool\fR) r\-
+.RS 4
+\fB\-\-enable\-utrace\fR
+was specified during build configuration\&.
+.RE
+.PP
+"config\&.valgrind" (\fBbool\fR) r\-
+.RS 4
+\fB\-\-enable\-valgrind\fR
+was specified during build configuration\&.
+.RE
+.PP
+"config\&.xmalloc" (\fBbool\fR) r\-
+.RS 4
+\fB\-\-enable\-xmalloc\fR
+was specified during build configuration\&.
+.RE
+.PP
+"opt\&.abort" (\fBbool\fR) r\-
+.RS 4
+Abort\-on\-warning enabled/disabled\&. If true, most warnings are fatal\&. The process will call
+\fBabort\fR(3)
+in these cases\&. This option is disabled by default unless
+\fB\-\-enable\-debug\fR
+is specified during configuration, in which case it is enabled by default\&.
+.RE
+.PP
+"opt\&.lg_chunk" (\fBsize_t\fR) r\-
+.RS 4
+Virtual memory chunk size (log base 2)\&. If a chunk size outside the supported size range is specified, the size is silently clipped to the minimum/maximum supported size\&. The default chunk size is 4 MiB (2^22)\&.
+.RE
+.PP
+"opt\&.dss" (\fBconst char *\fR) r\-
+.RS 4
+dss (\fBsbrk\fR(2)) allocation precedence as related to
+\fBmmap\fR(2)
+allocation\&. The following settings are supported: \(lqdisabled\(rq, \(lqprimary\(rq, and \(lqsecondary\(rq (default)\&.
+.RE
+.PP
+"opt\&.narenas" (\fBsize_t\fR) r\-
+.RS 4
+Maximum number of arenas to use for automatic multiplexing of threads and arenas\&. The default is four times the number of CPUs, or one if there is a single CPU\&.
+.RE
+.PP
+"opt\&.lg_dirty_mult" (\fBssize_t\fR) r\-
+.RS 4
+Per\-arena minimum ratio (log base 2) of active to dirty pages\&. Some dirty unused pages may be allowed to accumulate, within the limit set by the ratio (or one chunk worth of dirty pages, whichever is greater), before informing the kernel about some of those pages via
+\fBmadvise\fR(2)
+or a similar system call\&. This provides the kernel with sufficient information to recycle dirty pages if physical memory becomes scarce and the pages remain unused\&. The default minimum ratio is 8:1 (2^3:1); an option value of \-1 will disable dirty page purging\&.
+.RE
+.PP
+"opt\&.stats_print" (\fBbool\fR) r\-
+.RS 4
+Enable/disable statistics printing at exit\&. If enabled, the
+\fBmalloc_stats_print\fR\fB\fR
+function is called at program exit via an
+\fBatexit\fR(3)
+function\&. If
+\fB\-\-enable\-stats\fR
+is specified during configuration, this has the potential to cause deadlock for a multi\-threaded process that exits while one or more threads are executing in the memory allocation functions\&. Therefore, this option should only be used with care; it is primarily intended as a performance tuning aid during application development\&. This option is disabled by default\&.
+.RE
+.PP
+"opt\&.junk" (\fBbool\fR) r\- [\fB\-\-enable\-fill\fR]
+.RS 4
+Junk filling enabled/disabled\&. If enabled, each byte of uninitialized allocated memory will be initialized to
+0xa5\&. All deallocated memory will be initialized to
+0x5a\&. This is intended for debugging and will impact performance negatively\&. This option is disabled by default unless
+\fB\-\-enable\-debug\fR
+is specified during configuration, in which case it is enabled by default unless running inside
+\m[blue]\fBValgrind\fR\m[]\&\s-2\u[2]\d\s+2\&.
+.RE
+.PP
+"opt\&.quarantine" (\fBsize_t\fR) r\- [\fB\-\-enable\-fill\fR]
+.RS 4
+Per thread quarantine size in bytes\&. If non\-zero, each thread maintains a FIFO object quarantine that stores up to the specified number of bytes of memory\&. The quarantined memory is not freed until it is released from quarantine, though it is immediately junk\-filled if the
+"opt\&.junk"
+option is enabled\&. This feature is of particular use in combination with
+\m[blue]\fBValgrind\fR\m[]\&\s-2\u[2]\d\s+2, which can detect attempts to access quarantined objects\&. This is intended for debugging and will impact performance negatively\&. The default quarantine size is 0 unless running inside Valgrind, in which case the default is 16 MiB\&.
+.RE
+.PP
+"opt\&.redzone" (\fBbool\fR) r\- [\fB\-\-enable\-fill\fR]
+.RS 4
+Redzones enabled/disabled\&. If enabled, small allocations have redzones before and after them\&. Furthermore, if the
+"opt\&.junk"
+option is enabled, the redzones are checked for corruption during deallocation\&. However, the primary intended purpose of this feature is to be used in combination with
+\m[blue]\fBValgrind\fR\m[]\&\s-2\u[2]\d\s+2, which needs redzones in order to do effective buffer overflow/underflow detection\&. This option is intended for debugging and will impact performance negatively\&. This option is disabled by default unless running inside Valgrind\&.
+.RE
+.PP
+"opt\&.zero" (\fBbool\fR) r\- [\fB\-\-enable\-fill\fR]
+.RS 4
+Zero filling enabled/disabled\&. If enabled, each byte of uninitialized allocated memory will be initialized to 0\&. Note that this initialization only happens once for each byte, so
+\fBrealloc\fR\fB\fR
+and
+\fBrallocm\fR\fB\fR
+calls do not zero memory that was previously allocated\&. This is intended for debugging and will impact performance negatively\&. This option is disabled by default\&.
+.RE
+.PP
+"opt\&.utrace" (\fBbool\fR) r\- [\fB\-\-enable\-utrace\fR]
+.RS 4
+Allocation tracing based on
+\fButrace\fR(2)
+enabled/disabled\&. This option is disabled by default\&.
+.RE
+.PP
+"opt\&.valgrind" (\fBbool\fR) r\- [\fB\-\-enable\-valgrind\fR]
+.RS 4
+\m[blue]\fBValgrind\fR\m[]\&\s-2\u[2]\d\s+2
+support enabled/disabled\&. This option is vestigal because jemalloc auto\-detects whether it is running inside Valgrind\&. This option is disabled by default, unless running inside Valgrind\&.
+.RE
+.PP
+"opt\&.xmalloc" (\fBbool\fR) r\- [\fB\-\-enable\-xmalloc\fR]
+.RS 4
+Abort\-on\-out\-of\-memory enabled/disabled\&. If enabled, rather than returning failure for any allocation function, display a diagnostic message on
+\fBSTDERR_FILENO\fR
+and cause the program to drop core (using
+\fBabort\fR(3))\&. If an application is designed to depend on this behavior, set the option at compile time by including the following in the source code:
+.sp
+.if n \{\
+.RS 4
+.\}
+.nf
+malloc_conf = "xmalloc:true";
+.fi
+.if n \{\
+.RE
+.\}
+.sp
+This option is disabled by default\&.
+.RE
+.PP
+"opt\&.tcache" (\fBbool\fR) r\- [\fB\-\-enable\-tcache\fR]
+.RS 4
+Thread\-specific caching enabled/disabled\&. When there are multiple threads, each thread uses a thread\-specific cache for objects up to a certain size\&. Thread\-specific caching allows many allocations to be satisfied without performing any thread synchronization, at the cost of increased memory use\&. See the
+"opt\&.lg_tcache_max"
+option for related tuning information\&. This option is enabled by default unless running inside
+\m[blue]\fBValgrind\fR\m[]\&\s-2\u[2]\d\s+2\&.
+.RE
+.PP
+"opt\&.lg_tcache_max" (\fBsize_t\fR) r\- [\fB\-\-enable\-tcache\fR]
+.RS 4
+Maximum size class (log base 2) to cache in the thread\-specific cache\&. At a minimum, all small size classes are cached, and at a maximum all large size classes are cached\&. The default maximum is 32 KiB (2^15)\&.
+.RE
+.PP
+"opt\&.prof" (\fBbool\fR) r\- [\fB\-\-enable\-prof\fR]
+.RS 4
+Memory profiling enabled/disabled\&. If enabled, profile memory allocation activity\&. See the
+"opt\&.prof_active"
+option for on\-the\-fly activation/deactivation\&. See the
+"opt\&.lg_prof_sample"
+option for probabilistic sampling control\&. See the
+"opt\&.prof_accum"
+option for control of cumulative sample reporting\&. See the
+"opt\&.lg_prof_interval"
+option for information on interval\-triggered profile dumping, the
+"opt\&.prof_gdump"
+option for information on high\-water\-triggered profile dumping, and the
+"opt\&.prof_final"
+option for final profile dumping\&. Profile output is compatible with the included
+\fBpprof\fR
+Perl script, which originates from the
+\m[blue]\fBgperftools package\fR\m[]\&\s-2\u[3]\d\s+2\&.
+.RE
+.PP
+"opt\&.prof_prefix" (\fBconst char *\fR) r\- [\fB\-\-enable\-prof\fR]
+.RS 4
+Filename prefix for profile dumps\&. If the prefix is set to the empty string, no automatic dumps will occur; this is primarily useful for disabling the automatic final heap dump (which also disables leak reporting, if enabled)\&. The default prefix is
+jeprof\&.
+.RE
+.PP
+"opt\&.prof_active" (\fBbool\fR) r\- [\fB\-\-enable\-prof\fR]
+.RS 4
+Profiling activated/deactivated\&. This is a secondary control mechanism that makes it possible to start the application with profiling enabled (see the
+"opt\&.prof"
+option) but inactive, then toggle profiling at any time during program execution with the
+"prof\&.active"
+mallctl\&. This option is enabled by default\&.
+.RE
+.PP
+"opt\&.lg_prof_sample" (\fBssize_t\fR) r\- [\fB\-\-enable\-prof\fR]
+.RS 4
+Average interval (log base 2) between allocation samples, as measured in bytes of allocation activity\&. Increasing the sampling interval decreases profile fidelity, but also decreases the computational overhead\&. The default sample interval is 512 KiB (2^19 B)\&.
+.RE
+.PP
+"opt\&.prof_accum" (\fBbool\fR) r\- [\fB\-\-enable\-prof\fR]
+.RS 4
+Reporting of cumulative object/byte counts in profile dumps enabled/disabled\&. If this option is enabled, every unique backtrace must be stored for the duration of execution\&. Depending on the application, this can impose a large memory overhead, and the cumulative counts are not always of interest\&. This option is disabled by default\&.
+.RE
+.PP
+"opt\&.lg_prof_interval" (\fBssize_t\fR) r\- [\fB\-\-enable\-prof\fR]
+.RS 4
+Average interval (log base 2) between memory profile dumps, as measured in bytes of allocation activity\&. The actual interval between dumps may be sporadic because decentralized allocation counters are used to avoid synchronization bottlenecks\&. Profiles are dumped to files named according to the pattern
+<prefix>\&.<pid>\&.<seq>\&.i<iseq>\&.heap, where
+<prefix>
+is controlled by the
+"opt\&.prof_prefix"
+option\&. By default, interval\-triggered profile dumping is disabled (encoded as \-1)\&.
+.RE
+.PP
+"opt\&.prof_gdump" (\fBbool\fR) r\- [\fB\-\-enable\-prof\fR]
+.RS 4
+Trigger a memory profile dump every time the total virtual memory exceeds the previous maximum\&. Profiles are dumped to files named according to the pattern
+<prefix>\&.<pid>\&.<seq>\&.u<useq>\&.heap, where
+<prefix>
+is controlled by the
+"opt\&.prof_prefix"
+option\&. This option is disabled by default\&.
+.RE
+.PP
+"opt\&.prof_final" (\fBbool\fR) r\- [\fB\-\-enable\-prof\fR]
+.RS 4
+Use an
+\fBatexit\fR(3)
+function to dump final memory usage to a file named according to the pattern
+<prefix>\&.<pid>\&.<seq>\&.f\&.heap, where
+<prefix>
+is controlled by the
+"opt\&.prof_prefix"
+option\&. This option is enabled by default\&.
+.RE
+.PP
+"opt\&.prof_leak" (\fBbool\fR) r\- [\fB\-\-enable\-prof\fR]
+.RS 4
+Leak reporting enabled/disabled\&. If enabled, use an
+\fBatexit\fR(3)
+function to report memory leaks detected by allocation sampling\&. See the
+"opt\&.prof"
+option for information on analyzing heap profile output\&. This option is disabled by default\&.
+.RE
+.PP
+"thread\&.arena" (\fBunsigned\fR) rw
+.RS 4
+Get or set the arena associated with the calling thread\&. If the specified arena was not initialized beforehand (see the
+"arenas\&.initialized"
+mallctl), it will be automatically initialized as a side effect of calling this interface\&.
+.RE
+.PP
+"thread\&.allocated" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Get the total number of bytes ever allocated by the calling thread\&. This counter has the potential to wrap around; it is up to the application to appropriately interpret the counter in such cases\&.
+.RE
+.PP
+"thread\&.allocatedp" (\fBuint64_t *\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Get a pointer to the the value that is returned by the
+"thread\&.allocated"
+mallctl\&. This is useful for avoiding the overhead of repeated
+\fBmallctl*\fR\fB\fR
+calls\&.
+.RE
+.PP
+"thread\&.deallocated" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Get the total number of bytes ever deallocated by the calling thread\&. This counter has the potential to wrap around; it is up to the application to appropriately interpret the counter in such cases\&.
+.RE
+.PP
+"thread\&.deallocatedp" (\fBuint64_t *\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Get a pointer to the the value that is returned by the
+"thread\&.deallocated"
+mallctl\&. This is useful for avoiding the overhead of repeated
+\fBmallctl*\fR\fB\fR
+calls\&.
+.RE
+.PP
+"thread\&.tcache\&.enabled" (\fBbool\fR) rw [\fB\-\-enable\-tcache\fR]
+.RS 4
+Enable/disable calling thread\*(Aqs tcache\&. The tcache is implicitly flushed as a side effect of becoming disabled (see
+"thread\&.tcache\&.flush")\&.
+.RE
+.PP
+"thread\&.tcache\&.flush" (\fBvoid\fR) \-\- [\fB\-\-enable\-tcache\fR]
+.RS 4
+Flush calling thread\*(Aqs tcache\&. This interface releases all cached objects and internal data structures associated with the calling thread\*(Aqs thread\-specific cache\&. Ordinarily, this interface need not be called, since automatic periodic incremental garbage collection occurs, and the thread cache is automatically discarded when a thread exits\&. However, garbage collection is triggered by allocation activity, so it is possible for a thread that stops allocating/deallocating to retain its cache indefinitely, in which case the developer may find manual flushing useful\&.
+.RE
+.PP
+"arena\&.<i>\&.purge" (\fBunsigned\fR) \-\-
+.RS 4
+Purge unused dirty pages for arena <i>, or for all arenas if <i> equals
+"arenas\&.narenas"\&.
+.RE
+.PP
+"arena\&.<i>\&.dss" (\fBconst char *\fR) rw
+.RS 4
+Set the precedence of dss allocation as related to mmap allocation for arena <i>, or for all arenas if <i> equals
+"arenas\&.narenas"\&. See
+"opt\&.dss"
+for supported settings\&.
+.RE
+.PP
+"arenas\&.narenas" (\fBunsigned\fR) r\-
+.RS 4
+Current limit on number of arenas\&.
+.RE
+.PP
+"arenas\&.initialized" (\fBbool *\fR) r\-
+.RS 4
+An array of
+"arenas\&.narenas"
+booleans\&. Each boolean indicates whether the corresponding arena is initialized\&.
+.RE
+.PP
+"arenas\&.quantum" (\fBsize_t\fR) r\-
+.RS 4
+Quantum size\&.
+.RE
+.PP
+"arenas\&.page" (\fBsize_t\fR) r\-
+.RS 4
+Page size\&.
+.RE
+.PP
+"arenas\&.tcache_max" (\fBsize_t\fR) r\- [\fB\-\-enable\-tcache\fR]
+.RS 4
+Maximum thread\-cached size class\&.
+.RE
+.PP
+"arenas\&.nbins" (\fBunsigned\fR) r\-
+.RS 4
+Number of bin size classes\&.
+.RE
+.PP
+"arenas\&.nhbins" (\fBunsigned\fR) r\- [\fB\-\-enable\-tcache\fR]
+.RS 4
+Total number of thread cache bin size classes\&.
+.RE
+.PP
+"arenas\&.bin\&.<i>\&.size" (\fBsize_t\fR) r\-
+.RS 4
+Maximum size supported by size class\&.
+.RE
+.PP
+"arenas\&.bin\&.<i>\&.nregs" (\fBuint32_t\fR) r\-
+.RS 4
+Number of regions per page run\&.
+.RE
+.PP
+"arenas\&.bin\&.<i>\&.run_size" (\fBsize_t\fR) r\-
+.RS 4
+Number of bytes per page run\&.
+.RE
+.PP
+"arenas\&.nlruns" (\fBsize_t\fR) r\-
+.RS 4
+Total number of large size classes\&.
+.RE
+.PP
+"arenas\&.lrun\&.<i>\&.size" (\fBsize_t\fR) r\-
+.RS 4
+Maximum size supported by this large size class\&.
+.RE
+.PP
+"arenas\&.purge" (\fBunsigned\fR) \-w
+.RS 4
+Purge unused dirty pages for the specified arena, or for all arenas if none is specified\&.
+.RE
+.PP
+"arenas\&.extend" (\fBunsigned\fR) r\-
+.RS 4
+Extend the array of arenas by appending a new arena, and returning the new arena index\&.
+.RE
+.PP
+"prof\&.active" (\fBbool\fR) rw [\fB\-\-enable\-prof\fR]
+.RS 4
+Control whether sampling is currently active\&. See the
+"opt\&.prof_active"
+option for additional information\&.
+.RE
+.PP
+"prof\&.dump" (\fBconst char *\fR) \-w [\fB\-\-enable\-prof\fR]
+.RS 4
+Dump a memory profile to the specified file, or if NULL is specified, to a file according to the pattern
+<prefix>\&.<pid>\&.<seq>\&.m<mseq>\&.heap, where
+<prefix>
+is controlled by the
+"opt\&.prof_prefix"
+option\&.
+.RE
+.PP
+"prof\&.interval" (\fBuint64_t\fR) r\- [\fB\-\-enable\-prof\fR]
+.RS 4
+Average number of bytes allocated between inverval\-based profile dumps\&. See the
+"opt\&.lg_prof_interval"
+option for additional information\&.
+.RE
+.PP
+"stats\&.cactive" (\fBsize_t *\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Pointer to a counter that contains an approximate count of the current number of bytes in active pages\&. The estimate may be high, but never low, because each arena rounds up to the nearest multiple of the chunk size when computing its contribution to the counter\&. Note that the
+"epoch"
+mallctl has no bearing on this counter\&. Furthermore, counter consistency is maintained via atomic operations, so it is necessary to use an atomic operation in order to guarantee a consistent read when dereferencing the pointer\&.
+.RE
+.PP
+"stats\&.allocated" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Total number of bytes allocated by the application\&.
+.RE
+.PP
+"stats\&.active" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Total number of bytes in active pages allocated by the application\&. This is a multiple of the page size, and greater than or equal to
+"stats\&.allocated"\&. This does not include
+"stats\&.arenas\&.<i>\&.pdirty"
+and pages entirely devoted to allocator metadata\&.
+.RE
+.PP
+"stats\&.mapped" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Total number of bytes in chunks mapped on behalf of the application\&. This is a multiple of the chunk size, and is at least as large as
+"stats\&.active"\&. This does not include inactive chunks\&.
+.RE
+.PP
+"stats\&.chunks\&.current" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Total number of chunks actively mapped on behalf of the application\&. This does not include inactive chunks\&.
+.RE
+.PP
+"stats\&.chunks\&.total" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Cumulative number of chunks allocated\&.
+.RE
+.PP
+"stats\&.chunks\&.high" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Maximum number of active chunks at any time thus far\&.
+.RE
+.PP
+"stats\&.huge\&.allocated" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Number of bytes currently allocated by huge objects\&.
+.RE
+.PP
+"stats\&.huge\&.nmalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Cumulative number of huge allocation requests\&.
+.RE
+.PP
+"stats\&.huge\&.ndalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Cumulative number of huge deallocation requests\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.dss" (\fBconst char *\fR) r\-
+.RS 4
+dss (\fBsbrk\fR(2)) allocation precedence as related to
+\fBmmap\fR(2)
+allocation\&. See
+"opt\&.dss"
+for details\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.nthreads" (\fBunsigned\fR) r\-
+.RS 4
+Number of threads currently assigned to arena\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.pactive" (\fBsize_t\fR) r\-
+.RS 4
+Number of pages in active runs\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.pdirty" (\fBsize_t\fR) r\-
+.RS 4
+Number of pages within unused runs that are potentially dirty, and for which
+\fBmadvise\fR\fB\fI\&.\&.\&.\fR\fR\fB \fR\fB\fI\fBMADV_DONTNEED\fR\fR\fR
+or similar has not been called\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.mapped" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Number of mapped bytes\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.npurge" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Number of dirty page purge sweeps performed\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.nmadvise" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Number of
+\fBmadvise\fR\fB\fI\&.\&.\&.\fR\fR\fB \fR\fB\fI\fBMADV_DONTNEED\fR\fR\fR
+or similar calls made to purge dirty pages\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.npurged" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Number of pages purged\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.small\&.allocated" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Number of bytes currently allocated by small objects\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.small\&.nmalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Cumulative number of allocation requests served by small bins\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.small\&.ndalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Cumulative number of small objects returned to bins\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.small\&.nrequests" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Cumulative number of small allocation requests\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.large\&.allocated" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Number of bytes currently allocated by large objects\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.large\&.nmalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Cumulative number of large allocation requests served directly by the arena\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.large\&.ndalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Cumulative number of large deallocation requests served directly by the arena\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.large\&.nrequests" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Cumulative number of large allocation requests\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.bins\&.<j>\&.allocated" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Current number of bytes allocated by bin\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.bins\&.<j>\&.nmalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Cumulative number of allocations served by bin\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.bins\&.<j>\&.ndalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Cumulative number of allocations returned to bin\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.bins\&.<j>\&.nrequests" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Cumulative number of allocation requests\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.bins\&.<j>\&.nfills" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR \fB\-\-enable\-tcache\fR]
+.RS 4
+Cumulative number of tcache fills\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.bins\&.<j>\&.nflushes" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR \fB\-\-enable\-tcache\fR]
+.RS 4
+Cumulative number of tcache flushes\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.bins\&.<j>\&.nruns" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Cumulative number of runs created\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.bins\&.<j>\&.nreruns" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Cumulative number of times the current run from which to allocate changed\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.bins\&.<j>\&.curruns" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Current number of runs\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.lruns\&.<j>\&.nmalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Cumulative number of allocation requests for this size class served directly by the arena\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.lruns\&.<j>\&.ndalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Cumulative number of deallocation requests for this size class served directly by the arena\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.lruns\&.<j>\&.nrequests" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Cumulative number of allocation requests for this size class\&.
+.RE
+.PP
+"stats\&.arenas\&.<i>\&.lruns\&.<j>\&.curruns" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR]
+.RS 4
+Current number of runs for this size class\&.
+.RE
+.SH "DEBUGGING MALLOC PROBLEMS"
+.PP
+When debugging, it is a good idea to configure/build jemalloc with the
+\fB\-\-enable\-debug\fR
+and
+\fB\-\-enable\-fill\fR
+options, and recompile the program with suitable options and symbols for debugger support\&. When so configured, jemalloc incorporates a wide variety of run\-time assertions that catch application errors such as double\-free, write\-after\-free, etc\&.
+.PP
+Programs often accidentally depend on \(lquninitialized\(rq memory actually being filled with zero bytes\&. Junk filling (see the
+"opt\&.junk"
+option) tends to expose such bugs in the form of obviously incorrect results and/or coredumps\&. Conversely, zero filling (see the
+"opt\&.zero"
+option) eliminates the symptoms of such bugs\&. Between these two options, it is usually possible to quickly detect, diagnose, and eliminate such bugs\&.
+.PP
+This implementation does not provide much detail about the problems it detects, because the performance impact for storing such information would be prohibitive\&. However, jemalloc does integrate with the most excellent
+\m[blue]\fBValgrind\fR\m[]\&\s-2\u[2]\d\s+2
+tool if the
+\fB\-\-enable\-valgrind\fR
+configuration option is enabled\&.
+.SH "DIAGNOSTIC MESSAGES"
+.PP
+If any of the memory allocation/deallocation functions detect an error or warning condition, a message will be printed to file descriptor
+\fBSTDERR_FILENO\fR\&. Errors will result in the process dumping core\&. If the
+"opt\&.abort"
+option is set, most warnings are treated as errors\&.
+.PP
+The
+\fImalloc_message\fR
+variable allows the programmer to override the function which emits the text strings forming the errors and warnings if for some reason the
+\fBSTDERR_FILENO\fR
+file descriptor is not suitable for this\&.
+\fBmalloc_message\fR\fB\fR
+takes the
+\fIcbopaque\fR
+pointer argument that is
+\fBNULL\fR
+unless overridden by the arguments in a call to
+\fBmalloc_stats_print\fR\fB\fR, followed by a string pointer\&. Please note that doing anything which tries to allocate memory in this function is likely to result in a crash or deadlock\&.
+.PP
+All messages are prefixed by \(lq<jemalloc>:\(rq\&.
+.SH "RETURN VALUES"
+.SS "Standard API"
+.PP
+The
+\fBmalloc\fR\fB\fR
+and
+\fBcalloc\fR\fB\fR
+functions return a pointer to the allocated memory if successful; otherwise a
+\fBNULL\fR
+pointer is returned and
+\fIerrno\fR
+is set to
+ENOMEM\&.
+.PP
+The
+\fBposix_memalign\fR\fB\fR
+function returns the value 0 if successful; otherwise it returns an error value\&. The
+\fBposix_memalign\fR\fB\fR
+function will fail if:
+.PP
+EINVAL
+.RS 4
+The
+\fIalignment\fR
+parameter is not a power of 2 at least as large as
+sizeof(\fBvoid *\fR)\&.
+.RE
+.PP
+ENOMEM
+.RS 4
+Memory allocation error\&.
+.RE
+.PP
+The
+\fBaligned_alloc\fR\fB\fR
+function returns a pointer to the allocated memory if successful; otherwise a
+\fBNULL\fR
+pointer is returned and
+\fIerrno\fR
+is set\&. The
+\fBaligned_alloc\fR\fB\fR
+function will fail if:
+.PP
+EINVAL
+.RS 4
+The
+\fIalignment\fR
+parameter is not a power of 2\&.
+.RE
+.PP
+ENOMEM
+.RS 4
+Memory allocation error\&.
+.RE
+.PP
+The
+\fBrealloc\fR\fB\fR
+function returns a pointer, possibly identical to
+\fIptr\fR, to the allocated memory if successful; otherwise a
+\fBNULL\fR
+pointer is returned, and
+\fIerrno\fR
+is set to
+ENOMEM
+if the error was the result of an allocation failure\&. The
+\fBrealloc\fR\fB\fR
+function always leaves the original buffer intact when an error occurs\&.
+.PP
+The
+\fBfree\fR\fB\fR
+function returns no value\&.
+.SS "Non\-standard API"
+.PP
+The
+\fBmalloc_usable_size\fR\fB\fR
+function returns the usable size of the allocation pointed to by
+\fIptr\fR\&.
+.PP
+The
+\fBmallctl\fR\fB\fR,
+\fBmallctlnametomib\fR\fB\fR, and
+\fBmallctlbymib\fR\fB\fR
+functions return 0 on success; otherwise they return an error value\&. The functions will fail if:
+.PP
+EINVAL
+.RS 4
+\fInewp\fR
+is not
+\fBNULL\fR, and
+\fInewlen\fR
+is too large or too small\&. Alternatively,
+\fI*oldlenp\fR
+is too large or too small; in this case as much data as possible are read despite the error\&.
+.RE
+.PP
+ENOMEM
+.RS 4
+\fI*oldlenp\fR
+is too short to hold the requested value\&.
+.RE
+.PP
+ENOENT
+.RS 4
+\fIname\fR
+or
+\fImib\fR
+specifies an unknown/invalid value\&.
+.RE
+.PP
+EPERM
+.RS 4
+Attempt to read or write void value, or attempt to write read\-only value\&.
+.RE
+.PP
+EAGAIN
+.RS 4
+A memory allocation failure occurred\&.
+.RE
+.PP
+EFAULT
+.RS 4
+An interface with side effects failed in some way not directly related to
+\fBmallctl*\fR\fB\fR
+read/write processing\&.
+.RE
+.SS "Experimental API"
+.PP
+The
+\fBallocm\fR\fB\fR,
+\fBrallocm\fR\fB\fR,
+\fBsallocm\fR\fB\fR,
+\fBdallocm\fR\fB\fR, and
+\fBnallocm\fR\fB\fR
+functions return
+\fBALLOCM_SUCCESS\fR
+on success; otherwise they return an error value\&. The
+\fBallocm\fR\fB\fR,
+\fBrallocm\fR\fB\fR, and
+\fBnallocm\fR\fB\fR
+functions will fail if:
+.PP
+ALLOCM_ERR_OOM
+.RS 4
+Out of memory\&. Insufficient contiguous memory was available to service the allocation request\&. The
+\fBallocm\fR\fB\fR
+function additionally sets
+\fI*ptr\fR
+to
+\fBNULL\fR, whereas the
+\fBrallocm\fR\fB\fR
+function leaves
+\fB*ptr\fR
+unmodified\&.
+.RE
+The
+\fBrallocm\fR\fB\fR
+function will also fail if:
+.PP
+ALLOCM_ERR_NOT_MOVED
+.RS 4
+\fBALLOCM_NO_MOVE\fR
+was specified, but the reallocation request could not be serviced without moving the object\&.
+.RE
+.SH "ENVIRONMENT"
+.PP
+The following environment variable affects the execution of the allocation functions:
+.PP
+\fBMALLOC_CONF\fR
+.RS 4
+If the environment variable
+\fBMALLOC_CONF\fR
+is set, the characters it contains will be interpreted as options\&.
+.RE
+.SH "EXAMPLES"
+.PP
+To dump core whenever a problem occurs:
+.sp
+.if n \{\
+.RS 4
+.\}
+.nf
+ln \-s \*(Aqabort:true\*(Aq /etc/malloc\&.conf
+.fi
+.if n \{\
+.RE
+.\}
+.PP
+To specify in the source a chunk size that is 16 MiB:
+.sp
+.if n \{\
+.RS 4
+.\}
+.nf
+malloc_conf = "lg_chunk:24";
+.fi
+.if n \{\
+.RE
+.\}
+.SH "SEE ALSO"
+.PP
+\fBmadvise\fR(2),
+\fBmmap\fR(2),
+\fBsbrk\fR(2),
+\fButrace\fR(2),
+\fBalloca\fR(3),
+\fBatexit\fR(3),
+\fBgetpagesize\fR(3)
+.SH "STANDARDS"
+.PP
+The
+\fBmalloc\fR\fB\fR,
+\fBcalloc\fR\fB\fR,
+\fBrealloc\fR\fB\fR, and
+\fBfree\fR\fB\fR
+functions conform to ISO/IEC 9899:1990 (\(lqISO C90\(rq)\&.
+.PP
+The
+\fBposix_memalign\fR\fB\fR
+function conforms to IEEE Std 1003\&.1\-2001 (\(lqPOSIX\&.1\(rq)\&.
+.SH "AUTHOR"
+.PP
+\fBJason Evans\fR
+.RS 4
+.RE
+.SH "NOTES"
+.IP " 1." 4
+jemalloc website
+.RS 4
+\%http://www.canonware.com/jemalloc/
+.RE
+.IP " 2." 4
+Valgrind
+.RS 4
+\%http://valgrind.org/
+.RE
+.IP " 3." 4
+gperftools package
+.RS 4
+\%http://code.google.com/p/gperftools/
+.RE
diff --git a/extra/jemalloc/doc/jemalloc.html b/extra/jemalloc/doc/jemalloc.html
new file mode 100644
index 00000000000..3e0fe26d1bd
--- /dev/null
+++ b/extra/jemalloc/doc/jemalloc.html
@@ -0,0 +1,1417 @@
+<html><head><meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1"><title>JEMALLOC</title><meta name="generator" content="DocBook XSL Stylesheets V1.76.1"></head><body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF"><div class="refentry" title="JEMALLOC"><a name="id286949159"></a><div class="titlepage"></div><div class="refnamediv"><h2>Name</h2><p>jemalloc &#8212; general purpose memory allocation functions</p></div><div class="refsect1" title="LIBRARY"><a name="library"></a><h2>LIBRARY</h2><p>This manual describes jemalloc 3.3.1-0-g9ef9d9e8c271cdf14f664b871a8f98c827714784.  More information
+    can be found at the <a class="ulink" href="http://www.canonware.com/jemalloc/" target="_top">jemalloc website</a>.</p></div><div class="refsynopsisdiv" title="SYNOPSIS"><h2>SYNOPSIS</h2><div class="funcsynopsis"><pre class="funcsynopsisinfo">#include &lt;<code class="filename">stdlib.h</code>&gt;
+#include &lt;<code class="filename">jemalloc/jemalloc.h</code>&gt;</pre><div class="refsect2" title="Standard API"><a name="id286901505"></a><h3>Standard API</h3><table border="0" summary="Function synopsis" cellspacing="0" cellpadding="0" class="funcprototype-table"><tr><td><code class="funcdef">void *<b class="fsfunc">malloc</b>(</code></td><td>size_t <var class="pdparam">size</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer"> </div><table border="0" summary="Function synopsis" cellspacing="0" cellpadding="0" class="funcprototype-table"><tr><td><code class="funcdef">void *<b class="fsfunc">calloc</b>(</code></td><td>size_t <var class="pdparam">number</var>, </td></tr><tr><td> </td><td>size_t <var class="pdparam">size</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer"> </div><table border="0" summary="Function synopsis" cellspacing="0" cellpadding="0" class="funcprototype-table"><tr><td><code class="funcdef">int <b class="fsfunc">posix_memalign</b>(</code></td><td>void **<var class="pdparam">ptr</var>, </td></tr><tr><td> </td><td>size_t <var class="pdparam">alignment</var>, </td></tr><tr><td> </td><td>size_t <var class="pdparam">size</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer"> </div><table border="0" summary="Function synopsis" cellspacing="0" cellpadding="0" class="funcprototype-table"><tr><td><code class="funcdef">void *<b class="fsfunc">aligned_alloc</b>(</code></td><td>size_t <var class="pdparam">alignment</var>, </td></tr><tr><td> </td><td>size_t <var class="pdparam">size</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer"> </div><table border="0" summary="Function synopsis" cellspacing="0" cellpadding="0" class="funcprototype-table"><tr><td><code class="funcdef">void *<b class="fsfunc">realloc</b>(</code></td><td>void *<var class="pdparam">ptr</var>, </td></tr><tr><td> </td><td>size_t <var class="pdparam">size</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer"> </div><table border="0" summary="Function synopsis" cellspacing="0" cellpadding="0" class="funcprototype-table"><tr><td><code class="funcdef">void <b class="fsfunc">free</b>(</code></td><td>void *<var class="pdparam">ptr</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer"> </div></div><div class="refsect2" title="Non-standard API"><a name="id286900549"></a><h3>Non-standard API</h3><table border="0" summary="Function synopsis" cellspacing="0" cellpadding="0" class="funcprototype-table"><tr><td><code class="funcdef">size_t <b class="fsfunc">malloc_usable_size</b>(</code></td><td>const void *<var class="pdparam">ptr</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer"> </div><table border="0" summary="Function synopsis" cellspacing="0" cellpadding="0" class="funcprototype-table"><tr><td><code class="funcdef">void <b class="fsfunc">malloc_stats_print</b>(</code></td><td>void <var class="pdparam">(*write_cb)</var>
+            <code>(</code>void *, const char *<code>)</code>
+          , </td></tr><tr><td> </td><td>void *<var class="pdparam">cbopaque</var>, </td></tr><tr><td> </td><td>const char *<var class="pdparam">opts</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer"> </div><table border="0" summary="Function synopsis" cellspacing="0" cellpadding="0" class="funcprototype-table"><tr><td><code class="funcdef">int <b class="fsfunc">mallctl</b>(</code></td><td>const char *<var class="pdparam">name</var>, </td></tr><tr><td> </td><td>void *<var class="pdparam">oldp</var>, </td></tr><tr><td> </td><td>size_t *<var class="pdparam">oldlenp</var>, </td></tr><tr><td> </td><td>void *<var class="pdparam">newp</var>, </td></tr><tr><td> </td><td>size_t <var class="pdparam">newlen</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer"> </div><table border="0" summary="Function synopsis" cellspacing="0" cellpadding="0" class="funcprototype-table"><tr><td><code class="funcdef">int <b class="fsfunc">mallctlnametomib</b>(</code></td><td>const char *<var class="pdparam">name</var>, </td></tr><tr><td> </td><td>size_t *<var class="pdparam">mibp</var>, </td></tr><tr><td> </td><td>size_t *<var class="pdparam">miblenp</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer"> </div><table border="0" summary="Function synopsis" cellspacing="0" cellpadding="0" class="funcprototype-table"><tr><td><code class="funcdef">int <b class="fsfunc">mallctlbymib</b>(</code></td><td>const size_t *<var class="pdparam">mib</var>, </td></tr><tr><td> </td><td>size_t <var class="pdparam">miblen</var>, </td></tr><tr><td> </td><td>void *<var class="pdparam">oldp</var>, </td></tr><tr><td> </td><td>size_t *<var class="pdparam">oldlenp</var>, </td></tr><tr><td> </td><td>void *<var class="pdparam">newp</var>, </td></tr><tr><td> </td><td>size_t <var class="pdparam">newlen</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer"> </div><table border="0" summary="Function synopsis" cellspacing="0" cellpadding="0" class="funcprototype-table"><tr><td><code class="funcdef">void <b class="fsfunc">(*malloc_message)</b>(</code></td><td>void *<var class="pdparam">cbopaque</var>, </td></tr><tr><td> </td><td>const char *<var class="pdparam">s</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer"> </div><p><span class="type">const char *</span><code class="varname">malloc_conf</code>;</p></div><div class="refsect2" title="Experimental API"><a name="id286900756"></a><h3>Experimental API</h3><table border="0" summary="Function synopsis" cellspacing="0" cellpadding="0" class="funcprototype-table"><tr><td><code class="funcdef">int <b class="fsfunc">allocm</b>(</code></td><td>void **<var class="pdparam">ptr</var>, </td></tr><tr><td> </td><td>size_t *<var class="pdparam">rsize</var>, </td></tr><tr><td> </td><td>size_t <var class="pdparam">size</var>, </td></tr><tr><td> </td><td>int <var class="pdparam">flags</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer"> </div><table border="0" summary="Function synopsis" cellspacing="0" cellpadding="0" class="funcprototype-table"><tr><td><code class="funcdef">int <b class="fsfunc">rallocm</b>(</code></td><td>void **<var class="pdparam">ptr</var>, </td></tr><tr><td> </td><td>size_t *<var class="pdparam">rsize</var>, </td></tr><tr><td> </td><td>size_t <var class="pdparam">size</var>, </td></tr><tr><td> </td><td>size_t <var class="pdparam">extra</var>, </td></tr><tr><td> </td><td>int <var class="pdparam">flags</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer"> </div><table border="0" summary="Function synopsis" cellspacing="0" cellpadding="0" class="funcprototype-table"><tr><td><code class="funcdef">int <b class="fsfunc">sallocm</b>(</code></td><td>const void *<var class="pdparam">ptr</var>, </td></tr><tr><td> </td><td>size_t *<var class="pdparam">rsize</var>, </td></tr><tr><td> </td><td>int <var class="pdparam">flags</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer"> </div><table border="0" summary="Function synopsis" cellspacing="0" cellpadding="0" class="funcprototype-table"><tr><td><code class="funcdef">int <b class="fsfunc">dallocm</b>(</code></td><td>void *<var class="pdparam">ptr</var>, </td></tr><tr><td> </td><td>int <var class="pdparam">flags</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer"> </div><table border="0" summary="Function synopsis" cellspacing="0" cellpadding="0" class="funcprototype-table"><tr><td><code class="funcdef">int <b class="fsfunc">nallocm</b>(</code></td><td>size_t *<var class="pdparam">rsize</var>, </td></tr><tr><td> </td><td>size_t <var class="pdparam">size</var>, </td></tr><tr><td> </td><td>int <var class="pdparam">flags</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer"> </div></div></div></div><div class="refsect1" title="DESCRIPTION"><a name="description"></a><h2>DESCRIPTION</h2><div class="refsect2" title="Standard API"><a name="id286949297"></a><h3>Standard API</h3><p>The <code class="function">malloc</code>(<em class="parameter"><code></code></em>) function allocates
+      <em class="parameter"><code>size</code></em> bytes of uninitialized memory.  The allocated
+      space is suitably aligned (after possible pointer coercion) for storage
+      of any type of object.</p><p>The <code class="function">calloc</code>(<em class="parameter"><code></code></em>) function allocates
+      space for <em class="parameter"><code>number</code></em> objects, each
+      <em class="parameter"><code>size</code></em> bytes in length.  The result is identical to
+      calling <code class="function">malloc</code>(<em class="parameter"><code></code></em>) with an argument of
+      <em class="parameter"><code>number</code></em> * <em class="parameter"><code>size</code></em>, with the
+      exception that the allocated memory is explicitly initialized to zero
+      bytes.</p><p>The <code class="function">posix_memalign</code>(<em class="parameter"><code></code></em>) function
+      allocates <em class="parameter"><code>size</code></em> bytes of memory such that the
+      allocation's base address is an even multiple of
+      <em class="parameter"><code>alignment</code></em>, and returns the allocation in the value
+      pointed to by <em class="parameter"><code>ptr</code></em>.  The requested
+      <em class="parameter"><code>alignment</code></em> must be a power of 2 at least as large
+      as <code class="code">sizeof(<span class="type">void *</span>)</code>.</p><p>The <code class="function">aligned_alloc</code>(<em class="parameter"><code></code></em>) function
+      allocates <em class="parameter"><code>size</code></em> bytes of memory such that the
+      allocation's base address is an even multiple of
+      <em class="parameter"><code>alignment</code></em>.  The requested
+      <em class="parameter"><code>alignment</code></em> must be a power of 2.  Behavior is
+      undefined if <em class="parameter"><code>size</code></em> is not an integral multiple of
+      <em class="parameter"><code>alignment</code></em>.</p><p>The <code class="function">realloc</code>(<em class="parameter"><code></code></em>) function changes the
+      size of the previously allocated memory referenced by
+      <em class="parameter"><code>ptr</code></em> to <em class="parameter"><code>size</code></em> bytes.  The
+      contents of the memory are unchanged up to the lesser of the new and old
+      sizes.  If the new size is larger, the contents of the newly allocated
+      portion of the memory are undefined.  Upon success, the memory referenced
+      by <em class="parameter"><code>ptr</code></em> is freed and a pointer to the newly
+      allocated memory is returned.  Note that
+      <code class="function">realloc</code>(<em class="parameter"><code></code></em>) may move the memory allocation,
+      resulting in a different return value than <em class="parameter"><code>ptr</code></em>.
+      If <em class="parameter"><code>ptr</code></em> is <code class="constant">NULL</code>, the
+      <code class="function">realloc</code>(<em class="parameter"><code></code></em>) function behaves identically to
+      <code class="function">malloc</code>(<em class="parameter"><code></code></em>) for the specified size.</p><p>The <code class="function">free</code>(<em class="parameter"><code></code></em>) function causes the
+      allocated memory referenced by <em class="parameter"><code>ptr</code></em> to be made
+      available for future allocations.  If <em class="parameter"><code>ptr</code></em> is
+      <code class="constant">NULL</code>, no action occurs.</p></div><div class="refsect2" title="Non-standard API"><a name="id286949561"></a><h3>Non-standard API</h3><p>The <code class="function">malloc_usable_size</code>(<em class="parameter"><code></code></em>) function
+      returns the usable size of the allocation pointed to by
+      <em class="parameter"><code>ptr</code></em>.  The return value may be larger than the size
+      that was requested during allocation.  The
+      <code class="function">malloc_usable_size</code>(<em class="parameter"><code></code></em>) function is not a
+      mechanism for in-place <code class="function">realloc</code>(<em class="parameter"><code></code></em>); rather
+      it is provided solely as a tool for introspection purposes.  Any
+      discrepancy between the requested allocation size and the size reported
+      by <code class="function">malloc_usable_size</code>(<em class="parameter"><code></code></em>) should not be
+      depended on, since such behavior is entirely implementation-dependent.
+      </p><p>The <code class="function">malloc_stats_print</code>(<em class="parameter"><code></code></em>) function
+      writes human-readable summary statistics via the
+      <em class="parameter"><code>write_cb</code></em> callback function pointer and
+      <em class="parameter"><code>cbopaque</code></em> data passed to
+      <em class="parameter"><code>write_cb</code></em>, or
+      <code class="function">malloc_message</code>(<em class="parameter"><code></code></em>) if
+      <em class="parameter"><code>write_cb</code></em> is <code class="constant">NULL</code>.  This
+      function can be called repeatedly.  General information that never
+      changes during execution can be omitted by specifying "g" as a character
+      within the <em class="parameter"><code>opts</code></em> string.  Note that
+      <code class="function">malloc_message</code>(<em class="parameter"><code></code></em>) uses the
+      <code class="function">mallctl*</code>(<em class="parameter"><code></code></em>) functions internally, so
+      inconsistent statistics can be reported if multiple threads use these
+      functions simultaneously.  If <code class="option">--enable-stats</code> is
+      specified during configuration, &#8220;m&#8221; and &#8220;a&#8221; can
+      be specified to omit merged arena and per arena statistics, respectively;
+      &#8220;b&#8221; and &#8220;l&#8221; can be specified to omit per size
+      class statistics for bins and large objects, respectively.  Unrecognized
+      characters are silently ignored.  Note that thread caching may prevent
+      some statistics from being completely up to date, since extra locking
+      would be required to merge counters that track thread cache operations.
+      </p><p>The <code class="function">mallctl</code>(<em class="parameter"><code></code></em>) function provides a
+      general interface for introspecting the memory allocator, as well as
+      setting modifiable parameters and triggering actions.  The
+      period-separated <em class="parameter"><code>name</code></em> argument specifies a
+      location in a tree-structured namespace; see the <a class="xref" href="#mallctl_namespace" title="MALLCTL NAMESPACE">MALLCTL NAMESPACE</a> section for
+      documentation on the tree contents.  To read a value, pass a pointer via
+      <em class="parameter"><code>oldp</code></em> to adequate space to contain the value, and a
+      pointer to its length via <em class="parameter"><code>oldlenp</code></em>; otherwise pass
+      <code class="constant">NULL</code> and <code class="constant">NULL</code>.  Similarly, to
+      write a value, pass a pointer to the value via
+      <em class="parameter"><code>newp</code></em>, and its length via
+      <em class="parameter"><code>newlen</code></em>; otherwise pass <code class="constant">NULL</code>
+      and <code class="constant">0</code>.</p><p>The <code class="function">mallctlnametomib</code>(<em class="parameter"><code></code></em>) function
+      provides a way to avoid repeated name lookups for applications that
+      repeatedly query the same portion of the namespace, by translating a name
+      to a &#8220;Management Information Base&#8221; (MIB) that can be passed
+      repeatedly to <code class="function">mallctlbymib</code>(<em class="parameter"><code></code></em>).  Upon
+      successful return from <code class="function">mallctlnametomib</code>(<em class="parameter"><code></code></em>),
+      <em class="parameter"><code>mibp</code></em> contains an array of
+      <em class="parameter"><code>*miblenp</code></em> integers, where
+      <em class="parameter"><code>*miblenp</code></em> is the lesser of the number of components
+      in <em class="parameter"><code>name</code></em> and the input value of
+      <em class="parameter"><code>*miblenp</code></em>.  Thus it is possible to pass a
+      <em class="parameter"><code>*miblenp</code></em> that is smaller than the number of
+      period-separated name components, which results in a partial MIB that can
+      be used as the basis for constructing a complete MIB.  For name
+      components that are integers (e.g. the 2 in
+      <a class="link" href="#arenas.bin.i.size">
+    "<code class="mallctl">arenas.bin.2.size</code>"
+  </a>),
+      the corresponding MIB component will always be that integer.  Therefore,
+      it is legitimate to construct code like the following: </p><pre class="programlisting">
+unsigned nbins, i;
+
+int mib[4];
+size_t len, miblen;
+
+len = sizeof(nbins);
+mallctl("arenas.nbins", &amp;nbins, &amp;len, NULL, 0);
+
+miblen = 4;
+mallnametomib("arenas.bin.0.size", mib, &amp;miblen);
+for (i = 0; i &lt; nbins; i++) {
+	size_t bin_size;
+
+	mib[2] = i;
+	len = sizeof(bin_size);
+	mallctlbymib(mib, miblen, &amp;bin_size, &amp;len, NULL, 0);
+	/* Do something with bin_size... */
+}</pre></div><div class="refsect2" title="Experimental API"><a name="id286949870"></a><h3>Experimental API</h3><p>The experimental API is subject to change or removal without regard
+      for backward compatibility.  If <code class="option">--disable-experimental</code>
+      is specified during configuration, the experimental API is
+      omitted.</p><p>The <code class="function">allocm</code>(<em class="parameter"><code></code></em>),
+      <code class="function">rallocm</code>(<em class="parameter"><code></code></em>),
+      <code class="function">sallocm</code>(<em class="parameter"><code></code></em>),
+      <code class="function">dallocm</code>(<em class="parameter"><code></code></em>), and
+      <code class="function">nallocm</code>(<em class="parameter"><code></code></em>) functions all have a
+      <em class="parameter"><code>flags</code></em> argument that can be used to specify
+      options.  The functions only check the options that are contextually
+      relevant.  Use bitwise or (<code class="code">|</code>) operations to
+      specify one or more of the following:
+        </p><div class="variablelist"><dl><dt><span class="term"><code class="constant">ALLOCM_LG_ALIGN(<em class="parameter"><code>la</code></em>)
+            </code></span></dt><dd><p>Align the memory allocation to start at an address
+            that is a multiple of <code class="code">(1 &lt;&lt;
+            <em class="parameter"><code>la</code></em>)</code>.  This macro does not validate
+            that <em class="parameter"><code>la</code></em> is within the valid
+            range.</p></dd><dt><span class="term"><code class="constant">ALLOCM_ALIGN(<em class="parameter"><code>a</code></em>)
+            </code></span></dt><dd><p>Align the memory allocation to start at an address
+            that is a multiple of <em class="parameter"><code>a</code></em>, where
+            <em class="parameter"><code>a</code></em> is a power of two.  This macro does not
+            validate that <em class="parameter"><code>a</code></em> is a power of 2.
+            </p></dd><dt><span class="term"><code class="constant">ALLOCM_ZERO</code></span></dt><dd><p>Initialize newly allocated memory to contain zero
+            bytes.  In the growing reallocation case, the real size prior to
+            reallocation defines the boundary between untouched bytes and those
+            that are initialized to contain zero bytes.  If this option is
+            absent, newly allocated memory is uninitialized.</p></dd><dt><span class="term"><code class="constant">ALLOCM_NO_MOVE</code></span></dt><dd><p>For reallocation, fail rather than moving the
+            object.  This constraint can apply to both growth and
+            shrinkage.</p></dd><dt><span class="term"><code class="constant">ALLOCM_ARENA(<em class="parameter"><code>a</code></em>)
+            </code></span></dt><dd><p>Use the arena specified by the index
+            <em class="parameter"><code>a</code></em>.  This macro does not validate that
+            <em class="parameter"><code>a</code></em> specifies an arena in the valid
+            range.</p></dd></dl></div><p>
+      </p><p>The <code class="function">allocm</code>(<em class="parameter"><code></code></em>) function allocates at
+      least <em class="parameter"><code>size</code></em> bytes of memory, sets
+      <em class="parameter"><code>*ptr</code></em> to the base address of the allocation, and
+      sets <em class="parameter"><code>*rsize</code></em> to the real size of the allocation if
+      <em class="parameter"><code>rsize</code></em> is not <code class="constant">NULL</code>.  Behavior
+      is undefined if <em class="parameter"><code>size</code></em> is
+      <code class="constant">0</code>.</p><p>The <code class="function">rallocm</code>(<em class="parameter"><code></code></em>) function resizes the
+      allocation at <em class="parameter"><code>*ptr</code></em> to be at least
+      <em class="parameter"><code>size</code></em> bytes, sets <em class="parameter"><code>*ptr</code></em> to
+      the base address of the allocation if it moved, and sets
+      <em class="parameter"><code>*rsize</code></em> to the real size of the allocation if
+      <em class="parameter"><code>rsize</code></em> is not <code class="constant">NULL</code>.  If
+      <em class="parameter"><code>extra</code></em> is non-zero, an attempt is made to resize
+      the allocation to be at least <code class="code"><em class="parameter"><code>size</code></em> +
+      <em class="parameter"><code>extra</code></em>)</code> bytes, though inability to allocate
+      the extra byte(s) will not by itself result in failure.  Behavior is
+      undefined if <em class="parameter"><code>size</code></em> is <code class="constant">0</code>, or if
+      <code class="code">(<em class="parameter"><code>size</code></em> +
+      <em class="parameter"><code>extra</code></em> &gt;
+      <code class="constant">SIZE_T_MAX</code>)</code>.</p><p>The <code class="function">sallocm</code>(<em class="parameter"><code></code></em>) function sets
+      <em class="parameter"><code>*rsize</code></em> to the real size of the allocation.</p><p>The <code class="function">dallocm</code>(<em class="parameter"><code></code></em>) function causes the
+      memory referenced by <em class="parameter"><code>ptr</code></em> to be made available for
+      future allocations.</p><p>The <code class="function">nallocm</code>(<em class="parameter"><code></code></em>) function allocates no
+      memory, but it performs the same size computation as the
+      <code class="function">allocm</code>(<em class="parameter"><code></code></em>) function, and if
+      <em class="parameter"><code>rsize</code></em> is not <code class="constant">NULL</code> it sets
+      <em class="parameter"><code>*rsize</code></em> to the real size of the allocation that
+      would result from the equivalent <code class="function">allocm</code>(<em class="parameter"><code></code></em>)
+      function call.  Behavior is undefined if
+      <em class="parameter"><code>size</code></em> is <code class="constant">0</code>.</p></div></div><div class="refsect1" title="TUNING"><a name="tuning"></a><h2>TUNING</h2><p>Once, when the first call is made to one of the memory allocation
+    routines, the allocator initializes its internals based in part on various
+    options that can be specified at compile- or run-time.</p><p>The string pointed to by the global variable
+    <code class="varname">malloc_conf</code>, the &#8220;name&#8221; of the file
+    referenced by the symbolic link named <code class="filename">/etc/malloc.conf</code>, and the value of the
+    environment variable <code class="envar">MALLOC_CONF</code>, will be interpreted, in
+    that order, from left to right as options.</p><p>An options string is a comma-separated list of option:value pairs.
+    There is one key corresponding to each <a class="link" href="#opt.abort">
+    "<code class="mallctl">opt.*</code>"
+  </a> mallctl (see the <a class="xref" href="#mallctl_namespace" title="MALLCTL NAMESPACE">MALLCTL NAMESPACE</a> section for options
+    documentation).  For example, <code class="literal">abort:true,narenas:1</code> sets
+    the <a class="link" href="#opt.abort">
+    "<code class="mallctl">opt.abort</code>"
+  </a> and <a class="link" href="#opt.narenas">
+    "<code class="mallctl">opt.narenas</code>"
+  </a> options.  Some
+    options have boolean values (true/false), others have integer values (base
+    8, 10, or 16, depending on prefix), and yet others have raw string
+    values.</p></div><div class="refsect1" title="IMPLEMENTATION NOTES"><a name="implementation_notes"></a><h2>IMPLEMENTATION NOTES</h2><p>Traditionally, allocators have used
+    <span class="citerefentry"><span class="refentrytitle">sbrk</span>(2)</span> to obtain memory, which is
+    suboptimal for several reasons, including race conditions, increased
+    fragmentation, and artificial limitations on maximum usable memory.  If
+    <code class="option">--enable-dss</code> is specified during configuration, this
+    allocator uses both <span class="citerefentry"><span class="refentrytitle">mmap</span>(2)</span> and
+    <span class="citerefentry"><span class="refentrytitle">sbrk</span>(2)</span>, in that order of preference;
+    otherwise only <span class="citerefentry"><span class="refentrytitle">mmap</span>(2)</span> is used.</p><p>This allocator uses multiple arenas in order to reduce lock
+    contention for threaded programs on multi-processor systems.  This works
+    well with regard to threading scalability, but incurs some costs.  There is
+    a small fixed per-arena overhead, and additionally, arenas manage memory
+    completely independently of each other, which means a small fixed increase
+    in overall memory fragmentation.  These overheads are not generally an
+    issue, given the number of arenas normally used.  Note that using
+    substantially more arenas than the default is not likely to improve
+    performance, mainly due to reduced cache performance.  However, it may make
+    sense to reduce the number of arenas if an application does not make much
+    use of the allocation functions.</p><p>In addition to multiple arenas, unless
+    <code class="option">--disable-tcache</code> is specified during configuration, this
+    allocator supports thread-specific caching for small and large objects, in
+    order to make it possible to completely avoid synchronization for most
+    allocation requests.  Such caching allows very fast allocation in the
+    common case, but it increases memory usage and fragmentation, since a
+    bounded number of objects can remain allocated in each thread cache.</p><p>Memory is conceptually broken into equal-sized chunks, where the
+    chunk size is a power of two that is greater than the page size.  Chunks
+    are always aligned to multiples of the chunk size.  This alignment makes it
+    possible to find metadata for user objects very quickly.</p><p>User objects are broken into three categories according to size:
+    small, large, and huge.  Small objects are smaller than one page.  Large
+    objects are smaller than the chunk size.  Huge objects are a multiple of
+    the chunk size.  Small and large objects are managed by arenas; huge
+    objects are managed separately in a single data structure that is shared by
+    all threads.  Huge objects are used by applications infrequently enough
+    that this single data structure is not a scalability issue.</p><p>Each chunk that is managed by an arena tracks its contents as runs of
+    contiguous pages (unused, backing a set of small objects, or backing one
+    large object).  The combination of chunk alignment and chunk page maps
+    makes it possible to determine all metadata regarding small and large
+    allocations in constant time.</p><p>Small objects are managed in groups by page runs.  Each run maintains
+    a frontier and free list to track which regions are in use.  Allocation
+    requests that are no more than half the quantum (8 or 16, depending on
+    architecture) are rounded up to the nearest power of two that is at least
+    <code class="code">sizeof(<span class="type">double</span>)</code>.  All other small
+    object size classes are multiples of the quantum, spaced such that internal
+    fragmentation is limited to approximately 25% for all but the smallest size
+    classes.  Allocation requests that are larger than the maximum small size
+    class, but small enough to fit in an arena-managed chunk (see the <a class="link" href="#opt.lg_chunk">
+    "<code class="mallctl">opt.lg_chunk</code>"
+  </a> option), are
+    rounded up to the nearest run size.  Allocation requests that are too large
+    to fit in an arena-managed chunk are rounded up to the nearest multiple of
+    the chunk size.</p><p>Allocations are packed tightly together, which can be an issue for
+    multi-threaded applications.  If you need to assure that allocations do not
+    suffer from cacheline sharing, round your allocation requests up to the
+    nearest multiple of the cacheline size, or specify cacheline alignment when
+    allocating.</p><p>Assuming 4 MiB chunks, 4 KiB pages, and a 16-byte quantum on a 64-bit
+    system, the size classes in each category are as shown in <a class="xref" href="#size_classes" title="Table 1. Size classes">Table 1</a>.</p><div class="table"><a name="size_classes"></a><p class="title"><b>Table 1. Size classes</b></p><div class="table-contents"><table summary="Size classes" border="1"><colgroup><col align="left" class="c1"><col align="right" class="c2"><col align="left" class="c3"></colgroup><thead><tr><th align="left">Category</th><th align="right">Spacing</th><th align="left">Size</th></tr></thead><tbody><tr><td rowspan="7" align="left">Small</td><td align="right">lg</td><td align="left">[8]</td></tr><tr><td align="right">16</td><td align="left">[16, 32, 48, ..., 128]</td></tr><tr><td align="right">32</td><td align="left">[160, 192, 224, 256]</td></tr><tr><td align="right">64</td><td align="left">[320, 384, 448, 512]</td></tr><tr><td align="right">128</td><td align="left">[640, 768, 896, 1024]</td></tr><tr><td align="right">256</td><td align="left">[1280, 1536, 1792, 2048]</td></tr><tr><td align="right">512</td><td align="left">[2560, 3072, 3584]</td></tr><tr><td align="left">Large</td><td align="right">4 KiB</td><td align="left">[4 KiB, 8 KiB, 12 KiB, ..., 4072 KiB]</td></tr><tr><td align="left">Huge</td><td align="right">4 MiB</td><td align="left">[4 MiB, 8 MiB, 12 MiB, ...]</td></tr></tbody></table></div></div><br class="table-break"></div><div class="refsect1" title="MALLCTL NAMESPACE"><a name="mallctl_namespace"></a><h2>MALLCTL NAMESPACE</h2><p>The following names are defined in the namespace accessible via the
+    <code class="function">mallctl*</code>(<em class="parameter"><code></code></em>) functions.  Value types are
+    specified in parentheses, their readable/writable statuses are encoded as
+    <code class="literal">rw</code>, <code class="literal">r-</code>, <code class="literal">-w</code>, or
+    <code class="literal">--</code>, and required build configuration flags follow, if
+    any.  A name element encoded as <code class="literal">&lt;i&gt;</code> or
+    <code class="literal">&lt;j&gt;</code> indicates an integer component, where the
+    integer varies from 0 to some upper value that must be determined via
+    introspection.  In the case of 
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.*</code>"
+  ,
+    <code class="literal">&lt;i&gt;</code> equal to <a class="link" href="#arenas.narenas">
+    "<code class="mallctl">arenas.narenas</code>"
+  </a> can be
+    used to access the summation of statistics from all arenas.  Take special
+    note of the <a class="link" href="#epoch">
+    "<code class="mallctl">epoch</code>"
+  </a> mallctl,
+    which controls refreshing of cached dynamic statistics.</p><div class="variablelist"><dl><dt><span class="term">
+          
+    "<code class="mallctl">version</code>"
+  
+          (<span class="type">const char *</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Return the jemalloc version string.</p></dd><dt><a name="epoch"></a><span class="term">
+          
+    "<code class="mallctl">epoch</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">rw</code>
+        </span></dt><dd><p>If a value is passed in, refresh the data from which
+        the <code class="function">mallctl*</code>(<em class="parameter"><code></code></em>) functions report values,
+        and increment the epoch.  Return the current epoch.  This is useful for
+        detecting whether another thread caused a refresh.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">config.debug</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p><code class="option">--enable-debug</code> was specified during
+        build configuration.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">config.dss</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p><code class="option">--enable-dss</code> was specified during
+        build configuration.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">config.fill</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p><code class="option">--enable-fill</code> was specified during
+        build configuration.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">config.lazy_lock</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p><code class="option">--enable-lazy-lock</code> was specified
+        during build configuration.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">config.mremap</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p><code class="option">--enable-mremap</code> was specified during
+        build configuration.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">config.munmap</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p><code class="option">--enable-munmap</code> was specified during
+        build configuration.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">config.prof</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p><code class="option">--enable-prof</code> was specified during
+        build configuration.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">config.prof_libgcc</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p><code class="option">--disable-prof-libgcc</code> was not
+        specified during build configuration.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">config.prof_libunwind</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p><code class="option">--enable-prof-libunwind</code> was specified
+        during build configuration.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">config.stats</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p><code class="option">--enable-stats</code> was specified during
+        build configuration.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">config.tcache</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p><code class="option">--disable-tcache</code> was not specified
+        during build configuration.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">config.tls</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p><code class="option">--disable-tls</code> was not specified during
+        build configuration.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">config.utrace</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p><code class="option">--enable-utrace</code> was specified during
+        build configuration.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">config.valgrind</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p><code class="option">--enable-valgrind</code> was specified during
+        build configuration.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">config.xmalloc</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p><code class="option">--enable-xmalloc</code> was specified during
+        build configuration.</p></dd><dt><a name="opt.abort"></a><span class="term">
+          
+    "<code class="mallctl">opt.abort</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Abort-on-warning enabled/disabled.  If true, most
+        warnings are fatal.  The process will call
+        <span class="citerefentry"><span class="refentrytitle">abort</span>(3)</span> in these cases.  This option is
+        disabled by default unless <code class="option">--enable-debug</code> is
+        specified during configuration, in which case it is enabled by default.
+        </p></dd><dt><a name="opt.lg_chunk"></a><span class="term">
+          
+    "<code class="mallctl">opt.lg_chunk</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Virtual memory chunk size (log base 2).  If a chunk
+        size outside the supported size range is specified, the size is
+        silently clipped to the minimum/maximum supported size.  The default
+        chunk size is 4 MiB (2^22).
+        </p></dd><dt><a name="opt.dss"></a><span class="term">
+          
+    "<code class="mallctl">opt.dss</code>"
+  
+          (<span class="type">const char *</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>dss (<span class="citerefentry"><span class="refentrytitle">sbrk</span>(2)</span>) allocation precedence as
+        related to <span class="citerefentry"><span class="refentrytitle">mmap</span>(2)</span> allocation.  The following
+        settings are supported: &#8220;disabled&#8221;, &#8220;primary&#8221;,
+        and &#8220;secondary&#8221; (default).</p></dd><dt><a name="opt.narenas"></a><span class="term">
+          
+    "<code class="mallctl">opt.narenas</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Maximum number of arenas to use for automatic
+        multiplexing of threads and arenas.  The default is four times the
+        number of CPUs, or one if there is a single CPU.</p></dd><dt><a name="opt.lg_dirty_mult"></a><span class="term">
+          
+    "<code class="mallctl">opt.lg_dirty_mult</code>"
+  
+          (<span class="type">ssize_t</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Per-arena minimum ratio (log base 2) of active to dirty
+        pages.  Some dirty unused pages may be allowed to accumulate, within
+        the limit set by the ratio (or one chunk worth of dirty pages,
+        whichever is greater), before informing the kernel about some of those
+        pages via <span class="citerefentry"><span class="refentrytitle">madvise</span>(2)</span> or a similar system call.  This
+        provides the kernel with sufficient information to recycle dirty pages
+        if physical memory becomes scarce and the pages remain unused.  The
+        default minimum ratio is 8:1 (2^3:1); an option value of -1 will
+        disable dirty page purging.</p></dd><dt><a name="opt.stats_print"></a><span class="term">
+          
+    "<code class="mallctl">opt.stats_print</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Enable/disable statistics printing at exit.  If
+        enabled, the <code class="function">malloc_stats_print</code>(<em class="parameter"><code></code></em>)
+        function is called at program exit via an
+        <span class="citerefentry"><span class="refentrytitle">atexit</span>(3)</span> function.  If
+        <code class="option">--enable-stats</code> is specified during configuration, this
+        has the potential to cause deadlock for a multi-threaded process that
+        exits while one or more threads are executing in the memory allocation
+        functions.  Therefore, this option should only be used with care; it is
+        primarily intended as a performance tuning aid during application
+        development.  This option is disabled by default.</p></dd><dt><a name="opt.junk"></a><span class="term">
+          
+    "<code class="mallctl">opt.junk</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-fill</code>]
+        </span></dt><dd><p>Junk filling enabled/disabled.  If enabled, each byte
+        of uninitialized allocated memory will be initialized to
+        <code class="literal">0xa5</code>.  All deallocated memory will be initialized to
+        <code class="literal">0x5a</code>.  This is intended for debugging and will
+        impact performance negatively.  This option is disabled by default
+        unless <code class="option">--enable-debug</code> is specified during
+        configuration, in which case it is enabled by default unless running
+        inside <a class="ulink" href="http://valgrind.org/" target="_top">Valgrind</a>.</p></dd><dt><a name="opt.quarantine"></a><span class="term">
+          
+    "<code class="mallctl">opt.quarantine</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-fill</code>]
+        </span></dt><dd><p>Per thread quarantine size in bytes.  If non-zero, each
+        thread maintains a FIFO object quarantine that stores up to the
+        specified number of bytes of memory.  The quarantined memory is not
+        freed until it is released from quarantine, though it is immediately
+        junk-filled if the <a class="link" href="#opt.junk">
+    "<code class="mallctl">opt.junk</code>"
+  </a> option is
+        enabled.  This feature is of particular use in combination with <a class="ulink" href="http://valgrind.org/" target="_top">Valgrind</a>, which can detect attempts
+        to access quarantined objects.  This is intended for debugging and will
+        impact performance negatively.  The default quarantine size is 0 unless
+        running inside Valgrind, in which case the default is 16
+        MiB.</p></dd><dt><a name="opt.redzone"></a><span class="term">
+          
+    "<code class="mallctl">opt.redzone</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-fill</code>]
+        </span></dt><dd><p>Redzones enabled/disabled.  If enabled, small
+        allocations have redzones before and after them.  Furthermore, if the
+        <a class="link" href="#opt.junk">
+    "<code class="mallctl">opt.junk</code>"
+  </a> option is
+        enabled, the redzones are checked for corruption during deallocation.
+        However, the primary intended purpose of this feature is to be used in
+        combination with <a class="ulink" href="http://valgrind.org/" target="_top">Valgrind</a>,
+        which needs redzones in order to do effective buffer overflow/underflow
+        detection.  This option is intended for debugging and will impact
+        performance negatively.  This option is disabled by
+        default unless running inside Valgrind.</p></dd><dt><a name="opt.zero"></a><span class="term">
+          
+    "<code class="mallctl">opt.zero</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-fill</code>]
+        </span></dt><dd><p>Zero filling enabled/disabled.  If enabled, each byte
+        of uninitialized allocated memory will be initialized to 0.  Note that
+        this initialization only happens once for each byte, so
+        <code class="function">realloc</code>(<em class="parameter"><code></code></em>) and
+        <code class="function">rallocm</code>(<em class="parameter"><code></code></em>) calls do not zero memory that
+        was previously allocated.  This is intended for debugging and will
+        impact performance negatively.  This option is disabled by default.
+        </p></dd><dt><a name="opt.utrace"></a><span class="term">
+          
+    "<code class="mallctl">opt.utrace</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-utrace</code>]
+        </span></dt><dd><p>Allocation tracing based on
+        <span class="citerefentry"><span class="refentrytitle">utrace</span>(2)</span> enabled/disabled.  This option
+        is disabled by default.</p></dd><dt><a name="opt.valgrind"></a><span class="term">
+          
+    "<code class="mallctl">opt.valgrind</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-valgrind</code>]
+        </span></dt><dd><p><a class="ulink" href="http://valgrind.org/" target="_top">Valgrind</a>
+        support enabled/disabled.  This option is vestigal because jemalloc
+        auto-detects whether it is running inside Valgrind.  This option is
+        disabled by default, unless running inside Valgrind.</p></dd><dt><a name="opt.xmalloc"></a><span class="term">
+          
+    "<code class="mallctl">opt.xmalloc</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-xmalloc</code>]
+        </span></dt><dd><p>Abort-on-out-of-memory enabled/disabled.  If enabled,
+        rather than returning failure for any allocation function, display a
+        diagnostic message on <code class="constant">STDERR_FILENO</code> and cause the
+        program to drop core (using
+        <span class="citerefentry"><span class="refentrytitle">abort</span>(3)</span>).  If an application is
+        designed to depend on this behavior, set the option at compile time by
+        including the following in the source code:
+        </p><pre class="programlisting">
+malloc_conf = "xmalloc:true";</pre><p>
+        This option is disabled by default.</p></dd><dt><a name="opt.tcache"></a><span class="term">
+          
+    "<code class="mallctl">opt.tcache</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-tcache</code>]
+        </span></dt><dd><p>Thread-specific caching enabled/disabled.  When there
+        are multiple threads, each thread uses a thread-specific cache for
+        objects up to a certain size.  Thread-specific caching allows many
+        allocations to be satisfied without performing any thread
+        synchronization, at the cost of increased memory use.  See the
+        <a class="link" href="#opt.lg_tcache_max">
+    "<code class="mallctl">opt.lg_tcache_max</code>"
+  </a>
+        option for related tuning information.  This option is enabled by
+        default unless running inside <a class="ulink" href="http://valgrind.org/" target="_top">Valgrind</a>.</p></dd><dt><a name="opt.lg_tcache_max"></a><span class="term">
+          
+    "<code class="mallctl">opt.lg_tcache_max</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-tcache</code>]
+        </span></dt><dd><p>Maximum size class (log base 2) to cache in the
+        thread-specific cache.  At a minimum, all small size classes are
+        cached, and at a maximum all large size classes are cached.  The
+        default maximum is 32 KiB (2^15).</p></dd><dt><a name="opt.prof"></a><span class="term">
+          
+    "<code class="mallctl">opt.prof</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-prof</code>]
+        </span></dt><dd><p>Memory profiling enabled/disabled.  If enabled, profile
+        memory allocation activity.  See the <a class="link" href="#opt.prof_active">
+    "<code class="mallctl">opt.prof_active</code>"
+  </a>
+        option for on-the-fly activation/deactivation.  See the <a class="link" href="#opt.lg_prof_sample">
+    "<code class="mallctl">opt.lg_prof_sample</code>"
+  </a>
+        option for probabilistic sampling control.  See the <a class="link" href="#opt.prof_accum">
+    "<code class="mallctl">opt.prof_accum</code>"
+  </a>
+        option for control of cumulative sample reporting.  See the <a class="link" href="#opt.lg_prof_interval">
+    "<code class="mallctl">opt.lg_prof_interval</code>"
+  </a>
+        option for information on interval-triggered profile dumping, the <a class="link" href="#opt.prof_gdump">
+    "<code class="mallctl">opt.prof_gdump</code>"
+  </a>
+        option for information on high-water-triggered profile dumping, and the
+        <a class="link" href="#opt.prof_final">
+    "<code class="mallctl">opt.prof_final</code>"
+  </a>
+        option for final profile dumping.  Profile output is compatible with
+        the included <span class="command"><strong>pprof</strong></span> Perl script, which originates
+        from the <a class="ulink" href="http://code.google.com/p/gperftools/" target="_top">gperftools
+        package</a>.</p></dd><dt><a name="opt.prof_prefix"></a><span class="term">
+          
+    "<code class="mallctl">opt.prof_prefix</code>"
+  
+          (<span class="type">const char *</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-prof</code>]
+        </span></dt><dd><p>Filename prefix for profile dumps.  If the prefix is
+        set to the empty string, no automatic dumps will occur; this is
+        primarily useful for disabling the automatic final heap dump (which
+        also disables leak reporting, if enabled).  The default prefix is
+        <code class="filename">jeprof</code>.</p></dd><dt><a name="opt.prof_active"></a><span class="term">
+          
+    "<code class="mallctl">opt.prof_active</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-prof</code>]
+        </span></dt><dd><p>Profiling activated/deactivated.  This is a secondary
+        control mechanism that makes it possible to start the application with
+        profiling enabled (see the <a class="link" href="#opt.prof">
+    "<code class="mallctl">opt.prof</code>"
+  </a> option) but
+        inactive, then toggle profiling at any time during program execution
+        with the <a class="link" href="#prof.active">
+    "<code class="mallctl">prof.active</code>"
+  </a> mallctl.
+        This option is enabled by default.</p></dd><dt><a name="opt.lg_prof_sample"></a><span class="term">
+          
+    "<code class="mallctl">opt.lg_prof_sample</code>"
+  
+          (<span class="type">ssize_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-prof</code>]
+        </span></dt><dd><p>Average interval (log base 2) between allocation
+        samples, as measured in bytes of allocation activity.  Increasing the
+        sampling interval decreases profile fidelity, but also decreases the
+        computational overhead.  The default sample interval is 512 KiB (2^19
+        B).</p></dd><dt><a name="opt.prof_accum"></a><span class="term">
+          
+    "<code class="mallctl">opt.prof_accum</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-prof</code>]
+        </span></dt><dd><p>Reporting of cumulative object/byte counts in profile
+        dumps enabled/disabled.  If this option is enabled, every unique
+        backtrace must be stored for the duration of execution.  Depending on
+        the application, this can impose a large memory overhead, and the
+        cumulative counts are not always of interest.  This option is disabled
+        by default.</p></dd><dt><a name="opt.lg_prof_interval"></a><span class="term">
+          
+    "<code class="mallctl">opt.lg_prof_interval</code>"
+  
+          (<span class="type">ssize_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-prof</code>]
+        </span></dt><dd><p>Average interval (log base 2) between memory profile
+        dumps, as measured in bytes of allocation activity.  The actual
+        interval between dumps may be sporadic because decentralized allocation
+        counters are used to avoid synchronization bottlenecks.  Profiles are
+        dumped to files named according to the pattern
+        <code class="filename">&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.i&lt;iseq&gt;.heap</code>,
+        where <code class="literal">&lt;prefix&gt;</code> is controlled by the
+        <a class="link" href="#opt.prof_prefix">
+    "<code class="mallctl">opt.prof_prefix</code>"
+  </a>
+        option.  By default, interval-triggered profile dumping is disabled
+        (encoded as -1).
+        </p></dd><dt><a name="opt.prof_gdump"></a><span class="term">
+          
+    "<code class="mallctl">opt.prof_gdump</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-prof</code>]
+        </span></dt><dd><p>Trigger a memory profile dump every time the total
+        virtual memory exceeds the previous maximum.  Profiles are dumped to
+        files named according to the pattern
+        <code class="filename">&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.u&lt;useq&gt;.heap</code>,
+        where <code class="literal">&lt;prefix&gt;</code> is controlled by the <a class="link" href="#opt.prof_prefix">
+    "<code class="mallctl">opt.prof_prefix</code>"
+  </a>
+        option.  This option is disabled by default.</p></dd><dt><a name="opt.prof_final"></a><span class="term">
+          
+    "<code class="mallctl">opt.prof_final</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-prof</code>]
+        </span></dt><dd><p>Use an
+        <span class="citerefentry"><span class="refentrytitle">atexit</span>(3)</span> function to dump final memory
+        usage to a file named according to the pattern
+        <code class="filename">&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.f.heap</code>,
+        where <code class="literal">&lt;prefix&gt;</code> is controlled by the <a class="link" href="#opt.prof_prefix">
+    "<code class="mallctl">opt.prof_prefix</code>"
+  </a>
+        option.  This option is enabled by default.</p></dd><dt><a name="opt.prof_leak"></a><span class="term">
+          
+    "<code class="mallctl">opt.prof_leak</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-prof</code>]
+        </span></dt><dd><p>Leak reporting enabled/disabled.  If enabled, use an
+        <span class="citerefentry"><span class="refentrytitle">atexit</span>(3)</span> function to report memory leaks
+        detected by allocation sampling.  See the
+        <a class="link" href="#opt.prof">
+    "<code class="mallctl">opt.prof</code>"
+  </a> option for
+        information on analyzing heap profile output.  This option is disabled
+        by default.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">thread.arena</code>"
+  
+          (<span class="type">unsigned</span>)
+          <code class="literal">rw</code>
+        </span></dt><dd><p>Get or set the arena associated with the calling
+        thread.  If the specified arena was not initialized beforehand (see the
+        <a class="link" href="#arenas.initialized">
+    "<code class="mallctl">arenas.initialized</code>"
+  </a>
+        mallctl), it will be automatically initialized as a side effect of
+        calling this interface.</p></dd><dt><a name="thread.allocated"></a><span class="term">
+          
+    "<code class="mallctl">thread.allocated</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Get the total number of bytes ever allocated by the
+        calling thread.  This counter has the potential to wrap around; it is
+        up to the application to appropriately interpret the counter in such
+        cases.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">thread.allocatedp</code>"
+  
+          (<span class="type">uint64_t *</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Get a pointer to the the value that is returned by the
+        <a class="link" href="#thread.allocated">
+    "<code class="mallctl">thread.allocated</code>"
+  </a>
+        mallctl.  This is useful for avoiding the overhead of repeated
+        <code class="function">mallctl*</code>(<em class="parameter"><code></code></em>) calls.</p></dd><dt><a name="thread.deallocated"></a><span class="term">
+          
+    "<code class="mallctl">thread.deallocated</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Get the total number of bytes ever deallocated by the
+        calling thread.  This counter has the potential to wrap around; it is
+        up to the application to appropriately interpret the counter in such
+        cases.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">thread.deallocatedp</code>"
+  
+          (<span class="type">uint64_t *</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Get a pointer to the the value that is returned by the
+        <a class="link" href="#thread.deallocated">
+    "<code class="mallctl">thread.deallocated</code>"
+  </a>
+        mallctl.  This is useful for avoiding the overhead of repeated
+        <code class="function">mallctl*</code>(<em class="parameter"><code></code></em>) calls.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">thread.tcache.enabled</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">rw</code>
+          [<code class="option">--enable-tcache</code>]
+        </span></dt><dd><p>Enable/disable calling thread's tcache.  The tcache is
+        implicitly flushed as a side effect of becoming
+        disabled (see 
+    "<code class="mallctl">thread.tcache.flush</code>"
+  ).
+        </p></dd><dt><span class="term">
+          
+    "<code class="mallctl">thread.tcache.flush</code>"
+  
+          (<span class="type">void</span>)
+          <code class="literal">--</code>
+          [<code class="option">--enable-tcache</code>]
+        </span></dt><dd><p>Flush calling thread's tcache.  This interface releases
+        all cached objects and internal data structures associated with the
+        calling thread's thread-specific cache.  Ordinarily, this interface
+        need not be called, since automatic periodic incremental garbage
+        collection occurs, and the thread cache is automatically discarded when
+        a thread exits.  However, garbage collection is triggered by allocation
+        activity, so it is possible for a thread that stops
+        allocating/deallocating to retain its cache indefinitely, in which case
+        the developer may find manual flushing useful.</p></dd><dt><a name="arena.i.purge"></a><span class="term">
+          
+    "<code class="mallctl">arena.&lt;i&gt;.purge</code>"
+  
+          (<span class="type">unsigned</span>)
+          <code class="literal">--</code>
+        </span></dt><dd><p>Purge unused dirty pages for arena &lt;i&gt;, or for
+        all arenas if &lt;i&gt; equals <a class="link" href="#arenas.narenas">
+    "<code class="mallctl">arenas.narenas</code>"
+  </a>.
+        </p></dd><dt><a name="arena.i.dss"></a><span class="term">
+          
+    "<code class="mallctl">arena.&lt;i&gt;.dss</code>"
+  
+          (<span class="type">const char *</span>)
+          <code class="literal">rw</code>
+        </span></dt><dd><p>Set the precedence of dss allocation as related to mmap
+        allocation for arena &lt;i&gt;, or for all arenas if &lt;i&gt; equals
+        <a class="link" href="#arenas.narenas">
+    "<code class="mallctl">arenas.narenas</code>"
+  </a>.  See
+        <a class="link" href="#opt.dss">
+    "<code class="mallctl">opt.dss</code>"
+  </a> for supported
+        settings.
+        </p></dd><dt><a name="arenas.narenas"></a><span class="term">
+          
+    "<code class="mallctl">arenas.narenas</code>"
+  
+          (<span class="type">unsigned</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Current limit on number of arenas.</p></dd><dt><a name="arenas.initialized"></a><span class="term">
+          
+    "<code class="mallctl">arenas.initialized</code>"
+  
+          (<span class="type">bool *</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>An array of <a class="link" href="#arenas.narenas">
+    "<code class="mallctl">arenas.narenas</code>"
+  </a>
+        booleans.  Each boolean indicates whether the corresponding arena is
+        initialized.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">arenas.quantum</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Quantum size.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">arenas.page</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Page size.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">arenas.tcache_max</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-tcache</code>]
+        </span></dt><dd><p>Maximum thread-cached size class.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">arenas.nbins</code>"
+  
+          (<span class="type">unsigned</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Number of bin size classes.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">arenas.nhbins</code>"
+  
+          (<span class="type">unsigned</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-tcache</code>]
+        </span></dt><dd><p>Total number of thread cache bin size
+        classes.</p></dd><dt><a name="arenas.bin.i.size"></a><span class="term">
+          
+    "<code class="mallctl">arenas.bin.&lt;i&gt;.size</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Maximum size supported by size class.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">arenas.bin.&lt;i&gt;.nregs</code>"
+  
+          (<span class="type">uint32_t</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Number of regions per page run.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">arenas.bin.&lt;i&gt;.run_size</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Number of bytes per page run.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">arenas.nlruns</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Total number of large size classes.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">arenas.lrun.&lt;i&gt;.size</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Maximum size supported by this large size
+        class.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">arenas.purge</code>"
+  
+          (<span class="type">unsigned</span>)
+          <code class="literal">-w</code>
+        </span></dt><dd><p>Purge unused dirty pages for the specified arena, or
+        for all arenas if none is specified.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">arenas.extend</code>"
+  
+          (<span class="type">unsigned</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Extend the array of arenas by appending a new arena,
+        and returning the new arena index.</p></dd><dt><a name="prof.active"></a><span class="term">
+          
+    "<code class="mallctl">prof.active</code>"
+  
+          (<span class="type">bool</span>)
+          <code class="literal">rw</code>
+          [<code class="option">--enable-prof</code>]
+        </span></dt><dd><p>Control whether sampling is currently active.  See the
+        <a class="link" href="#opt.prof_active">
+    "<code class="mallctl">opt.prof_active</code>"
+  </a>
+        option for additional information.
+        </p></dd><dt><span class="term">
+          
+    "<code class="mallctl">prof.dump</code>"
+  
+          (<span class="type">const char *</span>)
+          <code class="literal">-w</code>
+          [<code class="option">--enable-prof</code>]
+        </span></dt><dd><p>Dump a memory profile to the specified file, or if NULL
+        is specified, to a file according to the pattern
+        <code class="filename">&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.m&lt;mseq&gt;.heap</code>,
+        where <code class="literal">&lt;prefix&gt;</code> is controlled by the
+        <a class="link" href="#opt.prof_prefix">
+    "<code class="mallctl">opt.prof_prefix</code>"
+  </a>
+        option.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">prof.interval</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-prof</code>]
+        </span></dt><dd><p>Average number of bytes allocated between
+        inverval-based profile dumps.  See the
+        <a class="link" href="#opt.lg_prof_interval">
+    "<code class="mallctl">opt.lg_prof_interval</code>"
+  </a>
+        option for additional information.</p></dd><dt><a name="stats.cactive"></a><span class="term">
+          
+    "<code class="mallctl">stats.cactive</code>"
+  
+          (<span class="type">size_t *</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Pointer to a counter that contains an approximate count
+        of the current number of bytes in active pages.  The estimate may be
+        high, but never low, because each arena rounds up to the nearest
+        multiple of the chunk size when computing its contribution to the
+        counter.  Note that the <a class="link" href="#epoch">
+    "<code class="mallctl">epoch</code>"
+  </a> mallctl has no bearing
+        on this counter.  Furthermore, counter consistency is maintained via
+        atomic operations, so it is necessary to use an atomic operation in
+        order to guarantee a consistent read when dereferencing the pointer.
+        </p></dd><dt><a name="stats.allocated"></a><span class="term">
+          
+    "<code class="mallctl">stats.allocated</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Total number of bytes allocated by the
+        application.</p></dd><dt><a name="stats.active"></a><span class="term">
+          
+    "<code class="mallctl">stats.active</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Total number of bytes in active pages allocated by the
+        application.  This is a multiple of the page size, and greater than or
+        equal to <a class="link" href="#stats.allocated">
+    "<code class="mallctl">stats.allocated</code>"
+  </a>.
+        This does not include <a class="link" href="#stats.arenas.i.pdirty">
+        
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.pdirty</code>"
+  </a> and pages
+        entirely devoted to allocator metadata.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.mapped</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Total number of bytes in chunks mapped on behalf of the
+        application.  This is a multiple of the chunk size, and is at least as
+        large as <a class="link" href="#stats.active">
+    "<code class="mallctl">stats.active</code>"
+  </a>.  This
+        does not include inactive chunks.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.chunks.current</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Total number of chunks actively mapped on behalf of the
+        application.  This does not include inactive chunks.
+        </p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.chunks.total</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Cumulative number of chunks allocated.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.chunks.high</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Maximum number of active chunks at any time thus far.
+        </p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.huge.allocated</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Number of bytes currently allocated by huge objects.
+        </p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.huge.nmalloc</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Cumulative number of huge allocation requests.
+        </p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.huge.ndalloc</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Cumulative number of huge deallocation requests.
+        </p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.dss</code>"
+  
+          (<span class="type">const char *</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>dss (<span class="citerefentry"><span class="refentrytitle">sbrk</span>(2)</span>) allocation precedence as
+        related to <span class="citerefentry"><span class="refentrytitle">mmap</span>(2)</span> allocation.  See <a class="link" href="#opt.dss">
+    "<code class="mallctl">opt.dss</code>"
+  </a> for details.
+        </p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.nthreads</code>"
+  
+          (<span class="type">unsigned</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Number of threads currently assigned to
+        arena.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.pactive</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Number of pages in active runs.</p></dd><dt><a name="stats.arenas.i.pdirty"></a><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.pdirty</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+        </span></dt><dd><p>Number of pages within unused runs that are potentially
+        dirty, and for which <code class="function">madvise</code>(<em class="parameter"><code>...</code></em>, 
+        <em class="parameter"><code><code class="constant">MADV_DONTNEED</code></code></em>) or
+        similar has not been called.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.mapped</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Number of mapped bytes.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.npurge</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Number of dirty page purge sweeps performed.
+        </p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.nmadvise</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Number of <code class="function">madvise</code>(<em class="parameter"><code>...</code></em>, 
+        <em class="parameter"><code><code class="constant">MADV_DONTNEED</code></code></em>) or
+        similar calls made to purge dirty pages.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.npurged</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Number of pages purged.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.small.allocated</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Number of bytes currently allocated by small objects.
+        </p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.small.nmalloc</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Cumulative number of allocation requests served by
+        small bins.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.small.ndalloc</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Cumulative number of small objects returned to bins.
+        </p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.small.nrequests</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Cumulative number of small allocation requests.
+        </p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.large.allocated</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Number of bytes currently allocated by large objects.
+        </p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.large.nmalloc</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Cumulative number of large allocation requests served
+        directly by the arena.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.large.ndalloc</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Cumulative number of large deallocation requests served
+        directly by the arena.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.large.nrequests</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Cumulative number of large allocation requests.
+        </p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.allocated</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Current number of bytes allocated by
+        bin.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nmalloc</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Cumulative number of allocations served by bin.
+        </p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.ndalloc</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Cumulative number of allocations returned to bin.
+        </p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nrequests</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Cumulative number of allocation
+        requests.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nfills</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code> <code class="option">--enable-tcache</code>]
+        </span></dt><dd><p>Cumulative number of tcache fills.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nflushes</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code> <code class="option">--enable-tcache</code>]
+        </span></dt><dd><p>Cumulative number of tcache flushes.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nruns</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Cumulative number of runs created.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nreruns</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Cumulative number of times the current run from which
+        to allocate changed.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.curruns</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Current number of runs.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.lruns.&lt;j&gt;.nmalloc</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Cumulative number of allocation requests for this size
+        class served directly by the arena.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.lruns.&lt;j&gt;.ndalloc</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Cumulative number of deallocation requests for this
+        size class served directly by the arena.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.lruns.&lt;j&gt;.nrequests</code>"
+  
+          (<span class="type">uint64_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Cumulative number of allocation requests for this size
+        class.</p></dd><dt><span class="term">
+          
+    "<code class="mallctl">stats.arenas.&lt;i&gt;.lruns.&lt;j&gt;.curruns</code>"
+  
+          (<span class="type">size_t</span>)
+          <code class="literal">r-</code>
+          [<code class="option">--enable-stats</code>]
+        </span></dt><dd><p>Current number of runs for this size class.
+        </p></dd></dl></div></div><div class="refsect1" title="DEBUGGING MALLOC PROBLEMS"><a name="debugging_malloc_problems"></a><h2>DEBUGGING MALLOC PROBLEMS</h2><p>When debugging, it is a good idea to configure/build jemalloc with
+    the <code class="option">--enable-debug</code> and <code class="option">--enable-fill</code>
+    options, and recompile the program with suitable options and symbols for
+    debugger support.  When so configured, jemalloc incorporates a wide variety
+    of run-time assertions that catch application errors such as double-free,
+    write-after-free, etc.</p><p>Programs often accidentally depend on &#8220;uninitialized&#8221;
+    memory actually being filled with zero bytes.  Junk filling
+    (see the <a class="link" href="#opt.junk">
+    "<code class="mallctl">opt.junk</code>"
+  </a>
+    option) tends to expose such bugs in the form of obviously incorrect
+    results and/or coredumps.  Conversely, zero
+    filling (see the <a class="link" href="#opt.zero">
+    "<code class="mallctl">opt.zero</code>"
+  </a> option) eliminates
+    the symptoms of such bugs.  Between these two options, it is usually
+    possible to quickly detect, diagnose, and eliminate such bugs.</p><p>This implementation does not provide much detail about the problems
+    it detects, because the performance impact for storing such information
+    would be prohibitive.  However, jemalloc does integrate with the most
+    excellent <a class="ulink" href="http://valgrind.org/" target="_top">Valgrind</a> tool if the
+    <code class="option">--enable-valgrind</code> configuration option is enabled.</p></div><div class="refsect1" title="DIAGNOSTIC MESSAGES"><a name="diagnostic_messages"></a><h2>DIAGNOSTIC MESSAGES</h2><p>If any of the memory allocation/deallocation functions detect an
+    error or warning condition, a message will be printed to file descriptor
+    <code class="constant">STDERR_FILENO</code>.  Errors will result in the process
+    dumping core.  If the <a class="link" href="#opt.abort">
+    "<code class="mallctl">opt.abort</code>"
+  </a> option is set, most
+    warnings are treated as errors.</p><p>The <code class="varname">malloc_message</code> variable allows the programmer
+    to override the function which emits the text strings forming the errors
+    and warnings if for some reason the <code class="constant">STDERR_FILENO</code> file
+    descriptor is not suitable for this.
+    <code class="function">malloc_message</code>(<em class="parameter"><code></code></em>) takes the
+    <em class="parameter"><code>cbopaque</code></em> pointer argument that is
+    <code class="constant">NULL</code> unless overridden by the arguments in a call to
+    <code class="function">malloc_stats_print</code>(<em class="parameter"><code></code></em>), followed by a string
+    pointer.  Please note that doing anything which tries to allocate memory in
+    this function is likely to result in a crash or deadlock.</p><p>All messages are prefixed by
+    &#8220;<code class="computeroutput">&lt;jemalloc&gt;: </code>&#8221;.</p></div><div class="refsect1" title="RETURN VALUES"><a name="return_values"></a><h2>RETURN VALUES</h2><div class="refsect2" title="Standard API"><a name="id286954473"></a><h3>Standard API</h3><p>The <code class="function">malloc</code>(<em class="parameter"><code></code></em>) and
+      <code class="function">calloc</code>(<em class="parameter"><code></code></em>) functions return a pointer to the
+      allocated memory if successful; otherwise a <code class="constant">NULL</code>
+      pointer is returned and <code class="varname">errno</code> is set to
+      <span class="errorname">ENOMEM</span>.</p><p>The <code class="function">posix_memalign</code>(<em class="parameter"><code></code></em>) function
+      returns the value 0 if successful; otherwise it returns an error value.
+      The <code class="function">posix_memalign</code>(<em class="parameter"><code></code></em>) function will fail
+      if:
+        </p><div class="variablelist"><dl><dt><span class="term"><span class="errorname">EINVAL</span></span></dt><dd><p>The <em class="parameter"><code>alignment</code></em> parameter is
+            not a power of 2 at least as large as
+            <code class="code">sizeof(<span class="type">void *</span>)</code>.
+            </p></dd><dt><span class="term"><span class="errorname">ENOMEM</span></span></dt><dd><p>Memory allocation error.</p></dd></dl></div><p>
+      </p><p>The <code class="function">aligned_alloc</code>(<em class="parameter"><code></code></em>) function returns
+      a pointer to the allocated memory if successful; otherwise a
+      <code class="constant">NULL</code> pointer is returned and
+      <code class="varname">errno</code> is set.  The
+      <code class="function">aligned_alloc</code>(<em class="parameter"><code></code></em>) function will fail if:
+        </p><div class="variablelist"><dl><dt><span class="term"><span class="errorname">EINVAL</span></span></dt><dd><p>The <em class="parameter"><code>alignment</code></em> parameter is
+            not a power of 2.
+            </p></dd><dt><span class="term"><span class="errorname">ENOMEM</span></span></dt><dd><p>Memory allocation error.</p></dd></dl></div><p>
+      </p><p>The <code class="function">realloc</code>(<em class="parameter"><code></code></em>) function returns a
+      pointer, possibly identical to <em class="parameter"><code>ptr</code></em>, to the
+      allocated memory if successful; otherwise a <code class="constant">NULL</code>
+      pointer is returned, and <code class="varname">errno</code> is set to
+      <span class="errorname">ENOMEM</span> if the error was the result of an
+      allocation failure.  The <code class="function">realloc</code>(<em class="parameter"><code></code></em>)
+      function always leaves the original buffer intact when an error occurs.
+      </p><p>The <code class="function">free</code>(<em class="parameter"><code></code></em>) function returns no
+      value.</p></div><div class="refsect2" title="Non-standard API"><a name="id286954690"></a><h3>Non-standard API</h3><p>The <code class="function">malloc_usable_size</code>(<em class="parameter"><code></code></em>) function
+      returns the usable size of the allocation pointed to by
+      <em class="parameter"><code>ptr</code></em>.  </p><p>The <code class="function">mallctl</code>(<em class="parameter"><code></code></em>),
+      <code class="function">mallctlnametomib</code>(<em class="parameter"><code></code></em>), and
+      <code class="function">mallctlbymib</code>(<em class="parameter"><code></code></em>) functions return 0 on
+      success; otherwise they return an error value.  The functions will fail
+      if:
+        </p><div class="variablelist"><dl><dt><span class="term"><span class="errorname">EINVAL</span></span></dt><dd><p><em class="parameter"><code>newp</code></em> is not
+            <code class="constant">NULL</code>, and <em class="parameter"><code>newlen</code></em> is too
+            large or too small.  Alternatively, <em class="parameter"><code>*oldlenp</code></em>
+            is too large or too small; in this case as much data as possible
+            are read despite the error.</p></dd><dt><span class="term"><span class="errorname">ENOMEM</span></span></dt><dd><p><em class="parameter"><code>*oldlenp</code></em> is too short to
+            hold the requested value.</p></dd><dt><span class="term"><span class="errorname">ENOENT</span></span></dt><dd><p><em class="parameter"><code>name</code></em> or
+            <em class="parameter"><code>mib</code></em> specifies an unknown/invalid
+            value.</p></dd><dt><span class="term"><span class="errorname">EPERM</span></span></dt><dd><p>Attempt to read or write void value, or attempt to
+            write read-only value.</p></dd><dt><span class="term"><span class="errorname">EAGAIN</span></span></dt><dd><p>A memory allocation failure
+            occurred.</p></dd><dt><span class="term"><span class="errorname">EFAULT</span></span></dt><dd><p>An interface with side effects failed in some way
+            not directly related to <code class="function">mallctl*</code>(<em class="parameter"><code></code></em>)
+            read/write processing.</p></dd></dl></div><p>
+      </p></div><div class="refsect2" title="Experimental API"><a name="id286954842"></a><h3>Experimental API</h3><p>The <code class="function">allocm</code>(<em class="parameter"><code></code></em>),
+      <code class="function">rallocm</code>(<em class="parameter"><code></code></em>),
+      <code class="function">sallocm</code>(<em class="parameter"><code></code></em>),
+      <code class="function">dallocm</code>(<em class="parameter"><code></code></em>), and
+      <code class="function">nallocm</code>(<em class="parameter"><code></code></em>) functions return
+      <code class="constant">ALLOCM_SUCCESS</code> on success; otherwise they return an
+      error value.  The <code class="function">allocm</code>(<em class="parameter"><code></code></em>),
+      <code class="function">rallocm</code>(<em class="parameter"><code></code></em>), and
+      <code class="function">nallocm</code>(<em class="parameter"><code></code></em>) functions will fail if:
+        </p><div class="variablelist"><dl><dt><span class="term"><span class="errorname">ALLOCM_ERR_OOM</span></span></dt><dd><p>Out of memory.  Insufficient contiguous memory was
+            available to service the allocation request.  The
+            <code class="function">allocm</code>(<em class="parameter"><code></code></em>) function additionally sets
+            <em class="parameter"><code>*ptr</code></em> to <code class="constant">NULL</code>, whereas
+            the <code class="function">rallocm</code>(<em class="parameter"><code></code></em>) function leaves
+            <code class="constant">*ptr</code> unmodified.</p></dd></dl></div><p>
+      The <code class="function">rallocm</code>(<em class="parameter"><code></code></em>) function will also
+      fail if:
+        </p><div class="variablelist"><dl><dt><span class="term"><span class="errorname">ALLOCM_ERR_NOT_MOVED</span></span></dt><dd><p><code class="constant">ALLOCM_NO_MOVE</code> was specified,
+            but the reallocation request could not be serviced without moving
+            the object.</p></dd></dl></div><p>
+      </p></div></div><div class="refsect1" title="ENVIRONMENT"><a name="environment"></a><h2>ENVIRONMENT</h2><p>The following environment variable affects the execution of the
+    allocation functions:
+      </p><div class="variablelist"><dl><dt><span class="term"><code class="envar">MALLOC_CONF</code></span></dt><dd><p>If the environment variable
+          <code class="envar">MALLOC_CONF</code> is set, the characters it contains
+          will be interpreted as options.</p></dd></dl></div><p>
+    </p></div><div class="refsect1" title="EXAMPLES"><a name="examples"></a><h2>EXAMPLES</h2><p>To dump core whenever a problem occurs:
+      </p><pre class="screen">ln -s 'abort:true' /etc/malloc.conf</pre><p>
+    </p><p>To specify in the source a chunk size that is 16 MiB:
+      </p><pre class="programlisting">
+malloc_conf = "lg_chunk:24";</pre></div><div class="refsect1" title="SEE ALSO"><a name="see_also"></a><h2>SEE ALSO</h2><p><span class="citerefentry"><span class="refentrytitle">madvise</span>(2)</span>,
+    <span class="citerefentry"><span class="refentrytitle">mmap</span>(2)</span>,
+    <span class="citerefentry"><span class="refentrytitle">sbrk</span>(2)</span>,
+    <span class="citerefentry"><span class="refentrytitle">utrace</span>(2)</span>,
+    <span class="citerefentry"><span class="refentrytitle">alloca</span>(3)</span>,
+    <span class="citerefentry"><span class="refentrytitle">atexit</span>(3)</span>,
+    <span class="citerefentry"><span class="refentrytitle">getpagesize</span>(3)</span></p></div><div class="refsect1" title="STANDARDS"><a name="standards"></a><h2>STANDARDS</h2><p>The <code class="function">malloc</code>(<em class="parameter"><code></code></em>),
+    <code class="function">calloc</code>(<em class="parameter"><code></code></em>),
+    <code class="function">realloc</code>(<em class="parameter"><code></code></em>), and
+    <code class="function">free</code>(<em class="parameter"><code></code></em>) functions conform to ISO/IEC
+    9899:1990 (&#8220;ISO C90&#8221;).</p><p>The <code class="function">posix_memalign</code>(<em class="parameter"><code></code></em>) function conforms
+    to IEEE Std 1003.1-2001 (&#8220;POSIX.1&#8221;).</p></div></div></body></html>
diff --git a/extra/jemalloc/doc/jemalloc.xml.in b/extra/jemalloc/doc/jemalloc.xml.in
new file mode 100644
index 00000000000..09305801bab
--- /dev/null
+++ b/extra/jemalloc/doc/jemalloc.xml.in
@@ -0,0 +1,2176 @@
+<?xml version='1.0' encoding='UTF-8'?>
+<?xml-stylesheet type="text/xsl"
+        href="http://docbook.sourceforge.net/release/xsl/current/manpages/docbook.xsl"?>
+<!DOCTYPE refentry PUBLIC "-//OASIS//DTD DocBook XML V4.4//EN"
+        "http://www.oasis-open.org/docbook/xml/4.4/docbookx.dtd" [
+]>
+
+<refentry>
+  <refentryinfo>
+    <title>User Manual</title>
+    <productname>jemalloc</productname>
+    <releaseinfo role="version">@jemalloc_version@</releaseinfo>
+    <authorgroup>
+      <author>
+        <firstname>Jason</firstname>
+        <surname>Evans</surname>
+        <personblurb>Author</personblurb>
+      </author>
+    </authorgroup>
+  </refentryinfo>
+  <refmeta>
+    <refentrytitle>JEMALLOC</refentrytitle>
+    <manvolnum>3</manvolnum>
+  </refmeta>
+  <refnamediv>
+    <refdescriptor>jemalloc</refdescriptor>
+    <refname>jemalloc</refname>
+    <!-- Each refname causes a man page file to be created.  Only if this were
+         the system malloc(3) implementation would these files be appropriate.
+    <refname>malloc</refname>
+    <refname>calloc</refname>
+    <refname>posix_memalign</refname>
+    <refname>aligned_alloc</refname>
+    <refname>realloc</refname>
+    <refname>free</refname>
+    <refname>malloc_usable_size</refname>
+    <refname>malloc_stats_print</refname>
+    <refname>mallctl</refname>
+    <refname>mallctlnametomib</refname>
+    <refname>mallctlbymib</refname>
+    <refname>allocm</refname>
+    <refname>rallocm</refname>
+    <refname>sallocm</refname>
+    <refname>dallocm</refname>
+    <refname>nallocm</refname>
+    -->
+    <refpurpose>general purpose memory allocation functions</refpurpose>
+  </refnamediv>
+  <refsect1 id="library">
+    <title>LIBRARY</title>
+    <para>This manual describes jemalloc @jemalloc_version@.  More information
+    can be found at the <ulink
+    url="http://www.canonware.com/jemalloc/">jemalloc website</ulink>.</para>
+  </refsect1>
+  <refsynopsisdiv>
+    <title>SYNOPSIS</title>
+    <funcsynopsis>
+      <funcsynopsisinfo>#include &lt;<filename class="headerfile">stdlib.h</filename>&gt;
+#include &lt;<filename class="headerfile">jemalloc/jemalloc.h</filename>&gt;</funcsynopsisinfo>
+      <refsect2>
+        <title>Standard API</title>
+        <funcprototype>
+          <funcdef>void *<function>malloc</function></funcdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void *<function>calloc</function></funcdef>
+          <paramdef>size_t <parameter>number</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>int <function>posix_memalign</function></funcdef>
+          <paramdef>void **<parameter>ptr</parameter></paramdef>
+          <paramdef>size_t <parameter>alignment</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void *<function>aligned_alloc</function></funcdef>
+          <paramdef>size_t <parameter>alignment</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void *<function>realloc</function></funcdef>
+          <paramdef>void *<parameter>ptr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void <function>free</function></funcdef>
+          <paramdef>void *<parameter>ptr</parameter></paramdef>
+        </funcprototype>
+      </refsect2>
+      <refsect2>
+        <title>Non-standard API</title>
+        <funcprototype>
+          <funcdef>size_t <function>malloc_usable_size</function></funcdef>
+          <paramdef>const void *<parameter>ptr</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void <function>malloc_stats_print</function></funcdef>
+          <paramdef>void <parameter>(*write_cb)</parameter>
+            <funcparams>void *, const char *</funcparams>
+          </paramdef>
+          <paramdef>void *<parameter>cbopaque</parameter></paramdef>
+          <paramdef>const char *<parameter>opts</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>int <function>mallctl</function></funcdef>
+          <paramdef>const char *<parameter>name</parameter></paramdef>
+          <paramdef>void *<parameter>oldp</parameter></paramdef>
+          <paramdef>size_t *<parameter>oldlenp</parameter></paramdef>
+          <paramdef>void *<parameter>newp</parameter></paramdef>
+          <paramdef>size_t <parameter>newlen</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>int <function>mallctlnametomib</function></funcdef>
+          <paramdef>const char *<parameter>name</parameter></paramdef>
+          <paramdef>size_t *<parameter>mibp</parameter></paramdef>
+          <paramdef>size_t *<parameter>miblenp</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>int <function>mallctlbymib</function></funcdef>
+          <paramdef>const size_t *<parameter>mib</parameter></paramdef>
+          <paramdef>size_t <parameter>miblen</parameter></paramdef>
+          <paramdef>void *<parameter>oldp</parameter></paramdef>
+          <paramdef>size_t *<parameter>oldlenp</parameter></paramdef>
+          <paramdef>void *<parameter>newp</parameter></paramdef>
+          <paramdef>size_t <parameter>newlen</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void <function>(*malloc_message)</function></funcdef>
+          <paramdef>void *<parameter>cbopaque</parameter></paramdef>
+          <paramdef>const char *<parameter>s</parameter></paramdef>
+        </funcprototype>
+        <para><type>const char *</type><varname>malloc_conf</varname>;</para>
+      </refsect2>
+      <refsect2>
+      <title>Experimental API</title>
+        <funcprototype>
+          <funcdef>int <function>allocm</function></funcdef>
+          <paramdef>void **<parameter>ptr</parameter></paramdef>
+          <paramdef>size_t *<parameter>rsize</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>int <function>rallocm</function></funcdef>
+          <paramdef>void **<parameter>ptr</parameter></paramdef>
+          <paramdef>size_t *<parameter>rsize</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>size_t <parameter>extra</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>int <function>sallocm</function></funcdef>
+          <paramdef>const void *<parameter>ptr</parameter></paramdef>
+          <paramdef>size_t *<parameter>rsize</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>int <function>dallocm</function></funcdef>
+          <paramdef>void *<parameter>ptr</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>int <function>nallocm</function></funcdef>
+          <paramdef>size_t *<parameter>rsize</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+      </refsect2>
+    </funcsynopsis>
+  </refsynopsisdiv>
+  <refsect1 id="description">
+    <title>DESCRIPTION</title>
+    <refsect2>
+      <title>Standard API</title>
+
+      <para>The <function>malloc<parameter/></function> function allocates
+      <parameter>size</parameter> bytes of uninitialized memory.  The allocated
+      space is suitably aligned (after possible pointer coercion) for storage
+      of any type of object.</para>
+
+      <para>The <function>calloc<parameter/></function> function allocates
+      space for <parameter>number</parameter> objects, each
+      <parameter>size</parameter> bytes in length.  The result is identical to
+      calling <function>malloc<parameter/></function> with an argument of
+      <parameter>number</parameter> * <parameter>size</parameter>, with the
+      exception that the allocated memory is explicitly initialized to zero
+      bytes.</para>
+
+      <para>The <function>posix_memalign<parameter/></function> function
+      allocates <parameter>size</parameter> bytes of memory such that the
+      allocation's base address is an even multiple of
+      <parameter>alignment</parameter>, and returns the allocation in the value
+      pointed to by <parameter>ptr</parameter>.  The requested
+      <parameter>alignment</parameter> must be a power of 2 at least as large
+      as <code language="C">sizeof(<type>void *</type>)</code>.</para>
+
+      <para>The <function>aligned_alloc<parameter/></function> function
+      allocates <parameter>size</parameter> bytes of memory such that the
+      allocation's base address is an even multiple of
+      <parameter>alignment</parameter>.  The requested
+      <parameter>alignment</parameter> must be a power of 2.  Behavior is
+      undefined if <parameter>size</parameter> is not an integral multiple of
+      <parameter>alignment</parameter>.</para>
+
+      <para>The <function>realloc<parameter/></function> function changes the
+      size of the previously allocated memory referenced by
+      <parameter>ptr</parameter> to <parameter>size</parameter> bytes.  The
+      contents of the memory are unchanged up to the lesser of the new and old
+      sizes.  If the new size is larger, the contents of the newly allocated
+      portion of the memory are undefined.  Upon success, the memory referenced
+      by <parameter>ptr</parameter> is freed and a pointer to the newly
+      allocated memory is returned.  Note that
+      <function>realloc<parameter/></function> may move the memory allocation,
+      resulting in a different return value than <parameter>ptr</parameter>.
+      If <parameter>ptr</parameter> is <constant>NULL</constant>, the
+      <function>realloc<parameter/></function> function behaves identically to
+      <function>malloc<parameter/></function> for the specified size.</para>
+
+      <para>The <function>free<parameter/></function> function causes the
+      allocated memory referenced by <parameter>ptr</parameter> to be made
+      available for future allocations.  If <parameter>ptr</parameter> is
+      <constant>NULL</constant>, no action occurs.</para>
+    </refsect2>
+    <refsect2>
+      <title>Non-standard API</title>
+
+      <para>The <function>malloc_usable_size<parameter/></function> function
+      returns the usable size of the allocation pointed to by
+      <parameter>ptr</parameter>.  The return value may be larger than the size
+      that was requested during allocation.  The
+      <function>malloc_usable_size<parameter/></function> function is not a
+      mechanism for in-place <function>realloc<parameter/></function>; rather
+      it is provided solely as a tool for introspection purposes.  Any
+      discrepancy between the requested allocation size and the size reported
+      by <function>malloc_usable_size<parameter/></function> should not be
+      depended on, since such behavior is entirely implementation-dependent.
+      </para>
+
+      <para>The <function>malloc_stats_print<parameter/></function> function
+      writes human-readable summary statistics via the
+      <parameter>write_cb</parameter> callback function pointer and
+      <parameter>cbopaque</parameter> data passed to
+      <parameter>write_cb</parameter>, or
+      <function>malloc_message<parameter/></function> if
+      <parameter>write_cb</parameter> is <constant>NULL</constant>.  This
+      function can be called repeatedly.  General information that never
+      changes during execution can be omitted by specifying "g" as a character
+      within the <parameter>opts</parameter> string.  Note that
+      <function>malloc_message<parameter/></function> uses the
+      <function>mallctl*<parameter/></function> functions internally, so
+      inconsistent statistics can be reported if multiple threads use these
+      functions simultaneously.  If <option>--enable-stats</option> is
+      specified during configuration, &ldquo;m&rdquo; and &ldquo;a&rdquo; can
+      be specified to omit merged arena and per arena statistics, respectively;
+      &ldquo;b&rdquo; and &ldquo;l&rdquo; can be specified to omit per size
+      class statistics for bins and large objects, respectively.  Unrecognized
+      characters are silently ignored.  Note that thread caching may prevent
+      some statistics from being completely up to date, since extra locking
+      would be required to merge counters that track thread cache operations.
+      </para>
+
+      <para>The <function>mallctl<parameter/></function> function provides a
+      general interface for introspecting the memory allocator, as well as
+      setting modifiable parameters and triggering actions.  The
+      period-separated <parameter>name</parameter> argument specifies a
+      location in a tree-structured namespace; see the <xref
+      linkend="mallctl_namespace" xrefstyle="template:%t"/> section for
+      documentation on the tree contents.  To read a value, pass a pointer via
+      <parameter>oldp</parameter> to adequate space to contain the value, and a
+      pointer to its length via <parameter>oldlenp</parameter>; otherwise pass
+      <constant>NULL</constant> and <constant>NULL</constant>.  Similarly, to
+      write a value, pass a pointer to the value via
+      <parameter>newp</parameter>, and its length via
+      <parameter>newlen</parameter>; otherwise pass <constant>NULL</constant>
+      and <constant>0</constant>.</para>
+
+      <para>The <function>mallctlnametomib<parameter/></function> function
+      provides a way to avoid repeated name lookups for applications that
+      repeatedly query the same portion of the namespace, by translating a name
+      to a &ldquo;Management Information Base&rdquo; (MIB) that can be passed
+      repeatedly to <function>mallctlbymib<parameter/></function>.  Upon
+      successful return from <function>mallctlnametomib<parameter/></function>,
+      <parameter>mibp</parameter> contains an array of
+      <parameter>*miblenp</parameter> integers, where
+      <parameter>*miblenp</parameter> is the lesser of the number of components
+      in <parameter>name</parameter> and the input value of
+      <parameter>*miblenp</parameter>.  Thus it is possible to pass a
+      <parameter>*miblenp</parameter> that is smaller than the number of
+      period-separated name components, which results in a partial MIB that can
+      be used as the basis for constructing a complete MIB.  For name
+      components that are integers (e.g. the 2 in
+      <link
+      linkend="arenas.bin.i.size"><mallctl>arenas.bin.2.size</mallctl></link>),
+      the corresponding MIB component will always be that integer.  Therefore,
+      it is legitimate to construct code like the following: <programlisting
+      language="C"><![CDATA[
+unsigned nbins, i;
+
+int mib[4];
+size_t len, miblen;
+
+len = sizeof(nbins);
+mallctl("arenas.nbins", &nbins, &len, NULL, 0);
+
+miblen = 4;
+mallnametomib("arenas.bin.0.size", mib, &miblen);
+for (i = 0; i < nbins; i++) {
+	size_t bin_size;
+
+	mib[2] = i;
+	len = sizeof(bin_size);
+	mallctlbymib(mib, miblen, &bin_size, &len, NULL, 0);
+	/* Do something with bin_size... */
+}]]></programlisting></para>
+    </refsect2>
+    <refsect2>
+      <title>Experimental API</title>
+      <para>The experimental API is subject to change or removal without regard
+      for backward compatibility.  If <option>--disable-experimental</option>
+      is specified during configuration, the experimental API is
+      omitted.</para>
+
+      <para>The <function>allocm<parameter/></function>,
+      <function>rallocm<parameter/></function>,
+      <function>sallocm<parameter/></function>,
+      <function>dallocm<parameter/></function>, and
+      <function>nallocm<parameter/></function> functions all have a
+      <parameter>flags</parameter> argument that can be used to specify
+      options.  The functions only check the options that are contextually
+      relevant.  Use bitwise or (<code language="C">|</code>) operations to
+      specify one or more of the following:
+        <variablelist>
+          <varlistentry>
+            <term><constant>ALLOCM_LG_ALIGN(<parameter>la</parameter>)
+            </constant></term>
+
+            <listitem><para>Align the memory allocation to start at an address
+            that is a multiple of <code language="C">(1 &lt;&lt;
+            <parameter>la</parameter>)</code>.  This macro does not validate
+            that <parameter>la</parameter> is within the valid
+            range.</para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><constant>ALLOCM_ALIGN(<parameter>a</parameter>)
+            </constant></term>
+
+            <listitem><para>Align the memory allocation to start at an address
+            that is a multiple of <parameter>a</parameter>, where
+            <parameter>a</parameter> is a power of two.  This macro does not
+            validate that <parameter>a</parameter> is a power of 2.
+            </para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><constant>ALLOCM_ZERO</constant></term>
+
+            <listitem><para>Initialize newly allocated memory to contain zero
+            bytes.  In the growing reallocation case, the real size prior to
+            reallocation defines the boundary between untouched bytes and those
+            that are initialized to contain zero bytes.  If this option is
+            absent, newly allocated memory is uninitialized.</para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><constant>ALLOCM_NO_MOVE</constant></term>
+
+            <listitem><para>For reallocation, fail rather than moving the
+            object.  This constraint can apply to both growth and
+            shrinkage.</para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><constant>ALLOCM_ARENA(<parameter>a</parameter>)
+            </constant></term>
+
+            <listitem><para>Use the arena specified by the index
+            <parameter>a</parameter>.  This macro does not validate that
+            <parameter>a</parameter> specifies an arena in the valid
+            range.</para></listitem>
+          </varlistentry>
+        </variablelist>
+      </para>
+
+      <para>The <function>allocm<parameter/></function> function allocates at
+      least <parameter>size</parameter> bytes of memory, sets
+      <parameter>*ptr</parameter> to the base address of the allocation, and
+      sets <parameter>*rsize</parameter> to the real size of the allocation if
+      <parameter>rsize</parameter> is not <constant>NULL</constant>.  Behavior
+      is undefined if <parameter>size</parameter> is
+      <constant>0</constant>.</para>
+
+      <para>The <function>rallocm<parameter/></function> function resizes the
+      allocation at <parameter>*ptr</parameter> to be at least
+      <parameter>size</parameter> bytes, sets <parameter>*ptr</parameter> to
+      the base address of the allocation if it moved, and sets
+      <parameter>*rsize</parameter> to the real size of the allocation if
+      <parameter>rsize</parameter> is not <constant>NULL</constant>.  If
+      <parameter>extra</parameter> is non-zero, an attempt is made to resize
+      the allocation to be at least <code
+      language="C"><parameter>size</parameter> +
+      <parameter>extra</parameter>)</code> bytes, though inability to allocate
+      the extra byte(s) will not by itself result in failure.  Behavior is
+      undefined if <parameter>size</parameter> is <constant>0</constant>, or if
+      <code language="C">(<parameter>size</parameter> +
+      <parameter>extra</parameter> &gt;
+      <constant>SIZE_T_MAX</constant>)</code>.</para>
+
+      <para>The <function>sallocm<parameter/></function> function sets
+      <parameter>*rsize</parameter> to the real size of the allocation.</para>
+
+      <para>The <function>dallocm<parameter/></function> function causes the
+      memory referenced by <parameter>ptr</parameter> to be made available for
+      future allocations.</para>
+
+      <para>The <function>nallocm<parameter/></function> function allocates no
+      memory, but it performs the same size computation as the
+      <function>allocm<parameter/></function> function, and if
+      <parameter>rsize</parameter> is not <constant>NULL</constant> it sets
+      <parameter>*rsize</parameter> to the real size of the allocation that
+      would result from the equivalent <function>allocm<parameter/></function>
+      function call.  Behavior is undefined if
+      <parameter>size</parameter> is <constant>0</constant>.</para>
+    </refsect2>
+  </refsect1>
+  <refsect1 id="tuning">
+    <title>TUNING</title>
+    <para>Once, when the first call is made to one of the memory allocation
+    routines, the allocator initializes its internals based in part on various
+    options that can be specified at compile- or run-time.</para>
+
+    <para>The string pointed to by the global variable
+    <varname>malloc_conf</varname>, the &ldquo;name&rdquo; of the file
+    referenced by the symbolic link named <filename
+    class="symlink">/etc/malloc.conf</filename>, and the value of the
+    environment variable <envar>MALLOC_CONF</envar>, will be interpreted, in
+    that order, from left to right as options.</para>
+
+    <para>An options string is a comma-separated list of option:value pairs.
+    There is one key corresponding to each <link
+    linkend="opt.abort"><mallctl>opt.*</mallctl></link> mallctl (see the <xref
+    linkend="mallctl_namespace" xrefstyle="template:%t"/> section for options
+    documentation).  For example, <literal>abort:true,narenas:1</literal> sets
+    the <link linkend="opt.abort"><mallctl>opt.abort</mallctl></link> and <link
+    linkend="opt.narenas"><mallctl>opt.narenas</mallctl></link> options.  Some
+    options have boolean values (true/false), others have integer values (base
+    8, 10, or 16, depending on prefix), and yet others have raw string
+    values.</para>
+  </refsect1>
+  <refsect1 id="implementation_notes">
+    <title>IMPLEMENTATION NOTES</title>
+    <para>Traditionally, allocators have used
+    <citerefentry><refentrytitle>sbrk</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry> to obtain memory, which is
+    suboptimal for several reasons, including race conditions, increased
+    fragmentation, and artificial limitations on maximum usable memory.  If
+    <option>--enable-dss</option> is specified during configuration, this
+    allocator uses both <citerefentry><refentrytitle>mmap</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry> and
+    <citerefentry><refentrytitle>sbrk</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry>, in that order of preference;
+    otherwise only <citerefentry><refentrytitle>mmap</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry> is used.</para>
+
+    <para>This allocator uses multiple arenas in order to reduce lock
+    contention for threaded programs on multi-processor systems.  This works
+    well with regard to threading scalability, but incurs some costs.  There is
+    a small fixed per-arena overhead, and additionally, arenas manage memory
+    completely independently of each other, which means a small fixed increase
+    in overall memory fragmentation.  These overheads are not generally an
+    issue, given the number of arenas normally used.  Note that using
+    substantially more arenas than the default is not likely to improve
+    performance, mainly due to reduced cache performance.  However, it may make
+    sense to reduce the number of arenas if an application does not make much
+    use of the allocation functions.</para>
+
+    <para>In addition to multiple arenas, unless
+    <option>--disable-tcache</option> is specified during configuration, this
+    allocator supports thread-specific caching for small and large objects, in
+    order to make it possible to completely avoid synchronization for most
+    allocation requests.  Such caching allows very fast allocation in the
+    common case, but it increases memory usage and fragmentation, since a
+    bounded number of objects can remain allocated in each thread cache.</para>
+
+    <para>Memory is conceptually broken into equal-sized chunks, where the
+    chunk size is a power of two that is greater than the page size.  Chunks
+    are always aligned to multiples of the chunk size.  This alignment makes it
+    possible to find metadata for user objects very quickly.</para>
+
+    <para>User objects are broken into three categories according to size:
+    small, large, and huge.  Small objects are smaller than one page.  Large
+    objects are smaller than the chunk size.  Huge objects are a multiple of
+    the chunk size.  Small and large objects are managed by arenas; huge
+    objects are managed separately in a single data structure that is shared by
+    all threads.  Huge objects are used by applications infrequently enough
+    that this single data structure is not a scalability issue.</para>
+
+    <para>Each chunk that is managed by an arena tracks its contents as runs of
+    contiguous pages (unused, backing a set of small objects, or backing one
+    large object).  The combination of chunk alignment and chunk page maps
+    makes it possible to determine all metadata regarding small and large
+    allocations in constant time.</para>
+
+    <para>Small objects are managed in groups by page runs.  Each run maintains
+    a frontier and free list to track which regions are in use.  Allocation
+    requests that are no more than half the quantum (8 or 16, depending on
+    architecture) are rounded up to the nearest power of two that is at least
+    <code language="C">sizeof(<type>double</type>)</code>.  All other small
+    object size classes are multiples of the quantum, spaced such that internal
+    fragmentation is limited to approximately 25% for all but the smallest size
+    classes.  Allocation requests that are larger than the maximum small size
+    class, but small enough to fit in an arena-managed chunk (see the <link
+    linkend="opt.lg_chunk"><mallctl>opt.lg_chunk</mallctl></link> option), are
+    rounded up to the nearest run size.  Allocation requests that are too large
+    to fit in an arena-managed chunk are rounded up to the nearest multiple of
+    the chunk size.</para>
+
+    <para>Allocations are packed tightly together, which can be an issue for
+    multi-threaded applications.  If you need to assure that allocations do not
+    suffer from cacheline sharing, round your allocation requests up to the
+    nearest multiple of the cacheline size, or specify cacheline alignment when
+    allocating.</para>
+
+    <para>Assuming 4 MiB chunks, 4 KiB pages, and a 16-byte quantum on a 64-bit
+    system, the size classes in each category are as shown in <xref
+    linkend="size_classes" xrefstyle="template:Table %n"/>.</para>
+
+    <table xml:id="size_classes" frame="all">
+      <title>Size classes</title>
+      <tgroup cols="3" colsep="1" rowsep="1">
+      <colspec colname="c1" align="left"/>
+      <colspec colname="c2" align="right"/>
+      <colspec colname="c3" align="left"/>
+      <thead>
+        <row>
+          <entry>Category</entry>
+          <entry>Spacing</entry>
+          <entry>Size</entry>
+        </row>
+      </thead>
+      <tbody>
+        <row>
+          <entry morerows="6">Small</entry>
+          <entry>lg</entry>
+          <entry>[8]</entry>
+        </row>
+        <row>
+          <entry>16</entry>
+          <entry>[16, 32, 48, ..., 128]</entry>
+        </row>
+        <row>
+          <entry>32</entry>
+          <entry>[160, 192, 224, 256]</entry>
+        </row>
+        <row>
+          <entry>64</entry>
+          <entry>[320, 384, 448, 512]</entry>
+        </row>
+        <row>
+          <entry>128</entry>
+          <entry>[640, 768, 896, 1024]</entry>
+        </row>
+        <row>
+          <entry>256</entry>
+          <entry>[1280, 1536, 1792, 2048]</entry>
+        </row>
+        <row>
+          <entry>512</entry>
+          <entry>[2560, 3072, 3584]</entry>
+        </row>
+        <row>
+          <entry>Large</entry>
+          <entry>4 KiB</entry>
+          <entry>[4 KiB, 8 KiB, 12 KiB, ..., 4072 KiB]</entry>
+        </row>
+        <row>
+          <entry>Huge</entry>
+          <entry>4 MiB</entry>
+          <entry>[4 MiB, 8 MiB, 12 MiB, ...]</entry>
+        </row>
+      </tbody>
+      </tgroup>
+    </table>
+  </refsect1>
+  <refsect1 id="mallctl_namespace">
+    <title>MALLCTL NAMESPACE</title>
+    <para>The following names are defined in the namespace accessible via the
+    <function>mallctl*<parameter/></function> functions.  Value types are
+    specified in parentheses, their readable/writable statuses are encoded as
+    <literal>rw</literal>, <literal>r-</literal>, <literal>-w</literal>, or
+    <literal>--</literal>, and required build configuration flags follow, if
+    any.  A name element encoded as <literal>&lt;i&gt;</literal> or
+    <literal>&lt;j&gt;</literal> indicates an integer component, where the
+    integer varies from 0 to some upper value that must be determined via
+    introspection.  In the case of <mallctl>stats.arenas.&lt;i&gt;.*</mallctl>,
+    <literal>&lt;i&gt;</literal> equal to <link
+    linkend="arenas.narenas"><mallctl>arenas.narenas</mallctl></link> can be
+    used to access the summation of statistics from all arenas.  Take special
+    note of the <link linkend="epoch"><mallctl>epoch</mallctl></link> mallctl,
+    which controls refreshing of cached dynamic statistics.</para>
+
+    <variablelist>
+      <varlistentry>
+        <term>
+          <mallctl>version</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Return the jemalloc version string.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="epoch">
+        <term>
+          <mallctl>epoch</mallctl>
+          (<type>uint64_t</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>If a value is passed in, refresh the data from which
+        the <function>mallctl*<parameter/></function> functions report values,
+        and increment the epoch.  Return the current epoch.  This is useful for
+        detecting whether another thread caused a refresh.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>config.debug</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-debug</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>config.dss</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-dss</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>config.fill</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-fill</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>config.lazy_lock</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-lazy-lock</option> was specified
+        during build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>config.mremap</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-mremap</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>config.munmap</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-munmap</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>config.prof</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-prof</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>config.prof_libgcc</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--disable-prof-libgcc</option> was not
+        specified during build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>config.prof_libunwind</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-prof-libunwind</option> was specified
+        during build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>config.stats</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-stats</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>config.tcache</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--disable-tcache</option> was not specified
+        during build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>config.tls</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--disable-tls</option> was not specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>config.utrace</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-utrace</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>config.valgrind</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-valgrind</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>config.xmalloc</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-xmalloc</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.abort">
+        <term>
+          <mallctl>opt.abort</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Abort-on-warning enabled/disabled.  If true, most
+        warnings are fatal.  The process will call
+        <citerefentry><refentrytitle>abort</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry> in these cases.  This option is
+        disabled by default unless <option>--enable-debug</option> is
+        specified during configuration, in which case it is enabled by default.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.lg_chunk">
+        <term>
+          <mallctl>opt.lg_chunk</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Virtual memory chunk size (log base 2).  If a chunk
+        size outside the supported size range is specified, the size is
+        silently clipped to the minimum/maximum supported size.  The default
+        chunk size is 4 MiB (2^22).
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.dss">
+        <term>
+          <mallctl>opt.dss</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>dss (<citerefentry><refentrytitle>sbrk</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry>) allocation precedence as
+        related to <citerefentry><refentrytitle>mmap</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> allocation.  The following
+        settings are supported: &ldquo;disabled&rdquo;, &ldquo;primary&rdquo;,
+        and &ldquo;secondary&rdquo; (default).</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.narenas">
+        <term>
+          <mallctl>opt.narenas</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Maximum number of arenas to use for automatic
+        multiplexing of threads and arenas.  The default is four times the
+        number of CPUs, or one if there is a single CPU.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.lg_dirty_mult">
+        <term>
+          <mallctl>opt.lg_dirty_mult</mallctl>
+          (<type>ssize_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Per-arena minimum ratio (log base 2) of active to dirty
+        pages.  Some dirty unused pages may be allowed to accumulate, within
+        the limit set by the ratio (or one chunk worth of dirty pages,
+        whichever is greater), before informing the kernel about some of those
+        pages via <citerefentry><refentrytitle>madvise</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> or a similar system call.  This
+        provides the kernel with sufficient information to recycle dirty pages
+        if physical memory becomes scarce and the pages remain unused.  The
+        default minimum ratio is 8:1 (2^3:1); an option value of -1 will
+        disable dirty page purging.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.stats_print">
+        <term>
+          <mallctl>opt.stats_print</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Enable/disable statistics printing at exit.  If
+        enabled, the <function>malloc_stats_print<parameter/></function>
+        function is called at program exit via an
+        <citerefentry><refentrytitle>atexit</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry> function.  If
+        <option>--enable-stats</option> is specified during configuration, this
+        has the potential to cause deadlock for a multi-threaded process that
+        exits while one or more threads are executing in the memory allocation
+        functions.  Therefore, this option should only be used with care; it is
+        primarily intended as a performance tuning aid during application
+        development.  This option is disabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.junk">
+        <term>
+          <mallctl>opt.junk</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-fill</option>]
+        </term>
+        <listitem><para>Junk filling enabled/disabled.  If enabled, each byte
+        of uninitialized allocated memory will be initialized to
+        <literal>0xa5</literal>.  All deallocated memory will be initialized to
+        <literal>0x5a</literal>.  This is intended for debugging and will
+        impact performance negatively.  This option is disabled by default
+        unless <option>--enable-debug</option> is specified during
+        configuration, in which case it is enabled by default unless running
+        inside <ulink
+        url="http://valgrind.org/">Valgrind</ulink>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.quarantine">
+        <term>
+          <mallctl>opt.quarantine</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-fill</option>]
+        </term>
+        <listitem><para>Per thread quarantine size in bytes.  If non-zero, each
+        thread maintains a FIFO object quarantine that stores up to the
+        specified number of bytes of memory.  The quarantined memory is not
+        freed until it is released from quarantine, though it is immediately
+        junk-filled if the <link
+        linkend="opt.junk"><mallctl>opt.junk</mallctl></link> option is
+        enabled.  This feature is of particular use in combination with <ulink
+        url="http://valgrind.org/">Valgrind</ulink>, which can detect attempts
+        to access quarantined objects.  This is intended for debugging and will
+        impact performance negatively.  The default quarantine size is 0 unless
+        running inside Valgrind, in which case the default is 16
+        MiB.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.redzone">
+        <term>
+          <mallctl>opt.redzone</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-fill</option>]
+        </term>
+        <listitem><para>Redzones enabled/disabled.  If enabled, small
+        allocations have redzones before and after them.  Furthermore, if the
+        <link linkend="opt.junk"><mallctl>opt.junk</mallctl></link> option is
+        enabled, the redzones are checked for corruption during deallocation.
+        However, the primary intended purpose of this feature is to be used in
+        combination with <ulink url="http://valgrind.org/">Valgrind</ulink>,
+        which needs redzones in order to do effective buffer overflow/underflow
+        detection.  This option is intended for debugging and will impact
+        performance negatively.  This option is disabled by
+        default unless running inside Valgrind.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.zero">
+        <term>
+          <mallctl>opt.zero</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-fill</option>]
+        </term>
+        <listitem><para>Zero filling enabled/disabled.  If enabled, each byte
+        of uninitialized allocated memory will be initialized to 0.  Note that
+        this initialization only happens once for each byte, so
+        <function>realloc<parameter/></function> and
+        <function>rallocm<parameter/></function> calls do not zero memory that
+        was previously allocated.  This is intended for debugging and will
+        impact performance negatively.  This option is disabled by default.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.utrace">
+        <term>
+          <mallctl>opt.utrace</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-utrace</option>]
+        </term>
+        <listitem><para>Allocation tracing based on
+        <citerefentry><refentrytitle>utrace</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> enabled/disabled.  This option
+        is disabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.valgrind">
+        <term>
+          <mallctl>opt.valgrind</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-valgrind</option>]
+        </term>
+        <listitem><para><ulink url="http://valgrind.org/">Valgrind</ulink>
+        support enabled/disabled.  This option is vestigal because jemalloc
+        auto-detects whether it is running inside Valgrind.  This option is
+        disabled by default, unless running inside Valgrind.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.xmalloc">
+        <term>
+          <mallctl>opt.xmalloc</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-xmalloc</option>]
+        </term>
+        <listitem><para>Abort-on-out-of-memory enabled/disabled.  If enabled,
+        rather than returning failure for any allocation function, display a
+        diagnostic message on <constant>STDERR_FILENO</constant> and cause the
+        program to drop core (using
+        <citerefentry><refentrytitle>abort</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry>).  If an application is
+        designed to depend on this behavior, set the option at compile time by
+        including the following in the source code:
+        <programlisting language="C"><![CDATA[
+malloc_conf = "xmalloc:true";]]></programlisting>
+        This option is disabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.tcache">
+        <term>
+          <mallctl>opt.tcache</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-tcache</option>]
+        </term>
+        <listitem><para>Thread-specific caching enabled/disabled.  When there
+        are multiple threads, each thread uses a thread-specific cache for
+        objects up to a certain size.  Thread-specific caching allows many
+        allocations to be satisfied without performing any thread
+        synchronization, at the cost of increased memory use.  See the
+        <link
+        linkend="opt.lg_tcache_max"><mallctl>opt.lg_tcache_max</mallctl></link>
+        option for related tuning information.  This option is enabled by
+        default unless running inside <ulink
+        url="http://valgrind.org/">Valgrind</ulink>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.lg_tcache_max">
+        <term>
+          <mallctl>opt.lg_tcache_max</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-tcache</option>]
+        </term>
+        <listitem><para>Maximum size class (log base 2) to cache in the
+        thread-specific cache.  At a minimum, all small size classes are
+        cached, and at a maximum all large size classes are cached.  The
+        default maximum is 32 KiB (2^15).</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof">
+        <term>
+          <mallctl>opt.prof</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Memory profiling enabled/disabled.  If enabled, profile
+        memory allocation activity.  See the <link
+        linkend="opt.prof_active"><mallctl>opt.prof_active</mallctl></link>
+        option for on-the-fly activation/deactivation.  See the <link
+        linkend="opt.lg_prof_sample"><mallctl>opt.lg_prof_sample</mallctl></link>
+        option for probabilistic sampling control.  See the <link
+        linkend="opt.prof_accum"><mallctl>opt.prof_accum</mallctl></link>
+        option for control of cumulative sample reporting.  See the <link
+        linkend="opt.lg_prof_interval"><mallctl>opt.lg_prof_interval</mallctl></link>
+        option for information on interval-triggered profile dumping, the <link
+        linkend="opt.prof_gdump"><mallctl>opt.prof_gdump</mallctl></link>
+        option for information on high-water-triggered profile dumping, and the
+        <link linkend="opt.prof_final"><mallctl>opt.prof_final</mallctl></link>
+        option for final profile dumping.  Profile output is compatible with
+        the included <command>pprof</command> Perl script, which originates
+        from the <ulink url="http://code.google.com/p/gperftools/">gperftools
+        package</ulink>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_prefix">
+        <term>
+          <mallctl>opt.prof_prefix</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Filename prefix for profile dumps.  If the prefix is
+        set to the empty string, no automatic dumps will occur; this is
+        primarily useful for disabling the automatic final heap dump (which
+        also disables leak reporting, if enabled).  The default prefix is
+        <filename>jeprof</filename>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_active">
+        <term>
+          <mallctl>opt.prof_active</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Profiling activated/deactivated.  This is a secondary
+        control mechanism that makes it possible to start the application with
+        profiling enabled (see the <link
+        linkend="opt.prof"><mallctl>opt.prof</mallctl></link> option) but
+        inactive, then toggle profiling at any time during program execution
+        with the <link
+        linkend="prof.active"><mallctl>prof.active</mallctl></link> mallctl.
+        This option is enabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.lg_prof_sample">
+        <term>
+          <mallctl>opt.lg_prof_sample</mallctl>
+          (<type>ssize_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Average interval (log base 2) between allocation
+        samples, as measured in bytes of allocation activity.  Increasing the
+        sampling interval decreases profile fidelity, but also decreases the
+        computational overhead.  The default sample interval is 512 KiB (2^19
+        B).</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_accum">
+        <term>
+          <mallctl>opt.prof_accum</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Reporting of cumulative object/byte counts in profile
+        dumps enabled/disabled.  If this option is enabled, every unique
+        backtrace must be stored for the duration of execution.  Depending on
+        the application, this can impose a large memory overhead, and the
+        cumulative counts are not always of interest.  This option is disabled
+        by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.lg_prof_interval">
+        <term>
+          <mallctl>opt.lg_prof_interval</mallctl>
+          (<type>ssize_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Average interval (log base 2) between memory profile
+        dumps, as measured in bytes of allocation activity.  The actual
+        interval between dumps may be sporadic because decentralized allocation
+        counters are used to avoid synchronization bottlenecks.  Profiles are
+        dumped to files named according to the pattern
+        <filename>&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.i&lt;iseq&gt;.heap</filename>,
+        where <literal>&lt;prefix&gt;</literal> is controlled by the
+        <link
+        linkend="opt.prof_prefix"><mallctl>opt.prof_prefix</mallctl></link>
+        option.  By default, interval-triggered profile dumping is disabled
+        (encoded as -1).
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_gdump">
+        <term>
+          <mallctl>opt.prof_gdump</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Trigger a memory profile dump every time the total
+        virtual memory exceeds the previous maximum.  Profiles are dumped to
+        files named according to the pattern
+        <filename>&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.u&lt;useq&gt;.heap</filename>,
+        where <literal>&lt;prefix&gt;</literal> is controlled by the <link
+        linkend="opt.prof_prefix"><mallctl>opt.prof_prefix</mallctl></link>
+        option.  This option is disabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_final">
+        <term>
+          <mallctl>opt.prof_final</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Use an
+        <citerefentry><refentrytitle>atexit</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry> function to dump final memory
+        usage to a file named according to the pattern
+        <filename>&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.f.heap</filename>,
+        where <literal>&lt;prefix&gt;</literal> is controlled by the <link
+        linkend="opt.prof_prefix"><mallctl>opt.prof_prefix</mallctl></link>
+        option.  This option is enabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_leak">
+        <term>
+          <mallctl>opt.prof_leak</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Leak reporting enabled/disabled.  If enabled, use an
+        <citerefentry><refentrytitle>atexit</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry> function to report memory leaks
+        detected by allocation sampling.  See the
+        <link linkend="opt.prof"><mallctl>opt.prof</mallctl></link> option for
+        information on analyzing heap profile output.  This option is disabled
+        by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>thread.arena</mallctl>
+          (<type>unsigned</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Get or set the arena associated with the calling
+        thread.  If the specified arena was not initialized beforehand (see the
+        <link
+        linkend="arenas.initialized"><mallctl>arenas.initialized</mallctl></link>
+        mallctl), it will be automatically initialized as a side effect of
+        calling this interface.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.allocated">
+        <term>
+          <mallctl>thread.allocated</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Get the total number of bytes ever allocated by the
+        calling thread.  This counter has the potential to wrap around; it is
+        up to the application to appropriately interpret the counter in such
+        cases.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>thread.allocatedp</mallctl>
+          (<type>uint64_t *</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Get a pointer to the the value that is returned by the
+        <link
+        linkend="thread.allocated"><mallctl>thread.allocated</mallctl></link>
+        mallctl.  This is useful for avoiding the overhead of repeated
+        <function>mallctl*<parameter/></function> calls.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.deallocated">
+        <term>
+          <mallctl>thread.deallocated</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Get the total number of bytes ever deallocated by the
+        calling thread.  This counter has the potential to wrap around; it is
+        up to the application to appropriately interpret the counter in such
+        cases.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>thread.deallocatedp</mallctl>
+          (<type>uint64_t *</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Get a pointer to the the value that is returned by the
+        <link
+        linkend="thread.deallocated"><mallctl>thread.deallocated</mallctl></link>
+        mallctl.  This is useful for avoiding the overhead of repeated
+        <function>mallctl*<parameter/></function> calls.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>thread.tcache.enabled</mallctl>
+          (<type>bool</type>)
+          <literal>rw</literal>
+          [<option>--enable-tcache</option>]
+        </term>
+        <listitem><para>Enable/disable calling thread's tcache.  The tcache is
+        implicitly flushed as a side effect of becoming
+        disabled (see <link
+        lenkend="thread.tcache.flush"><mallctl>thread.tcache.flush</mallctl></link>).
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>thread.tcache.flush</mallctl>
+          (<type>void</type>)
+          <literal>--</literal>
+          [<option>--enable-tcache</option>]
+        </term>
+        <listitem><para>Flush calling thread's tcache.  This interface releases
+        all cached objects and internal data structures associated with the
+        calling thread's thread-specific cache.  Ordinarily, this interface
+        need not be called, since automatic periodic incremental garbage
+        collection occurs, and the thread cache is automatically discarded when
+        a thread exits.  However, garbage collection is triggered by allocation
+        activity, so it is possible for a thread that stops
+        allocating/deallocating to retain its cache indefinitely, in which case
+        the developer may find manual flushing useful.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.purge">
+        <term>
+          <mallctl>arena.&lt;i&gt;.purge</mallctl>
+          (<type>unsigned</type>)
+          <literal>--</literal>
+        </term>
+        <listitem><para>Purge unused dirty pages for arena &lt;i&gt;, or for
+        all arenas if &lt;i&gt; equals <link
+        linkend="arenas.narenas"><mallctl>arenas.narenas</mallctl></link>.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.dss">
+        <term>
+          <mallctl>arena.&lt;i&gt;.dss</mallctl>
+          (<type>const char *</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Set the precedence of dss allocation as related to mmap
+        allocation for arena &lt;i&gt;, or for all arenas if &lt;i&gt; equals
+        <link
+        linkend="arenas.narenas"><mallctl>arenas.narenas</mallctl></link>.  See
+        <link linkend="opt.dss"><mallctl>opt.dss</mallctl></link> for supported
+        settings.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.narenas">
+        <term>
+          <mallctl>arenas.narenas</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Current limit on number of arenas.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.initialized">
+        <term>
+          <mallctl>arenas.initialized</mallctl>
+          (<type>bool *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>An array of <link
+        linkend="arenas.narenas"><mallctl>arenas.narenas</mallctl></link>
+        booleans.  Each boolean indicates whether the corresponding arena is
+        initialized.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>arenas.quantum</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Quantum size.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>arenas.page</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Page size.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>arenas.tcache_max</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-tcache</option>]
+        </term>
+        <listitem><para>Maximum thread-cached size class.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>arenas.nbins</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of bin size classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>arenas.nhbins</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+          [<option>--enable-tcache</option>]
+        </term>
+        <listitem><para>Total number of thread cache bin size
+        classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.bin.i.size">
+        <term>
+          <mallctl>arenas.bin.&lt;i&gt;.size</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Maximum size supported by size class.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>arenas.bin.&lt;i&gt;.nregs</mallctl>
+          (<type>uint32_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of regions per page run.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>arenas.bin.&lt;i&gt;.run_size</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of bytes per page run.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>arenas.nlruns</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Total number of large size classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>arenas.lrun.&lt;i&gt;.size</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Maximum size supported by this large size
+        class.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>arenas.purge</mallctl>
+          (<type>unsigned</type>)
+          <literal>-w</literal>
+        </term>
+        <listitem><para>Purge unused dirty pages for the specified arena, or
+        for all arenas if none is specified.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>arenas.extend</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Extend the array of arenas by appending a new arena,
+        and returning the new arena index.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="prof.active">
+        <term>
+          <mallctl>prof.active</mallctl>
+          (<type>bool</type>)
+          <literal>rw</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Control whether sampling is currently active.  See the
+        <link
+        linkend="opt.prof_active"><mallctl>opt.prof_active</mallctl></link>
+        option for additional information.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>prof.dump</mallctl>
+          (<type>const char *</type>)
+          <literal>-w</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Dump a memory profile to the specified file, or if NULL
+        is specified, to a file according to the pattern
+        <filename>&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.m&lt;mseq&gt;.heap</filename>,
+        where <literal>&lt;prefix&gt;</literal> is controlled by the
+        <link
+        linkend="opt.prof_prefix"><mallctl>opt.prof_prefix</mallctl></link>
+        option.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>prof.interval</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Average number of bytes allocated between
+        inverval-based profile dumps.  See the
+        <link
+        linkend="opt.lg_prof_interval"><mallctl>opt.lg_prof_interval</mallctl></link>
+        option for additional information.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.cactive">
+        <term>
+          <mallctl>stats.cactive</mallctl>
+          (<type>size_t *</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Pointer to a counter that contains an approximate count
+        of the current number of bytes in active pages.  The estimate may be
+        high, but never low, because each arena rounds up to the nearest
+        multiple of the chunk size when computing its contribution to the
+        counter.  Note that the <link
+        linkend="epoch"><mallctl>epoch</mallctl></link> mallctl has no bearing
+        on this counter.  Furthermore, counter consistency is maintained via
+        atomic operations, so it is necessary to use an atomic operation in
+        order to guarantee a consistent read when dereferencing the pointer.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.allocated">
+        <term>
+          <mallctl>stats.allocated</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Total number of bytes allocated by the
+        application.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.active">
+        <term>
+          <mallctl>stats.active</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Total number of bytes in active pages allocated by the
+        application.  This is a multiple of the page size, and greater than or
+        equal to <link
+        linkend="stats.allocated"><mallctl>stats.allocated</mallctl></link>.
+        This does not include <link linkend="stats.arenas.i.pdirty">
+        <mallctl>stats.arenas.&lt;i&gt;.pdirty</mallctl></link> and pages
+        entirely devoted to allocator metadata.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.mapped</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Total number of bytes in chunks mapped on behalf of the
+        application.  This is a multiple of the chunk size, and is at least as
+        large as <link
+        linkend="stats.active"><mallctl>stats.active</mallctl></link>.  This
+        does not include inactive chunks.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.chunks.current</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Total number of chunks actively mapped on behalf of the
+        application.  This does not include inactive chunks.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.chunks.total</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of chunks allocated.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.chunks.high</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Maximum number of active chunks at any time thus far.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.huge.allocated</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of bytes currently allocated by huge objects.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.huge.nmalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of huge allocation requests.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.huge.ndalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of huge deallocation requests.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.dss</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>dss (<citerefentry><refentrytitle>sbrk</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry>) allocation precedence as
+        related to <citerefentry><refentrytitle>mmap</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> allocation.  See <link
+        linkend="opt.dss"><mallctl>opt.dss</mallctl></link> for details.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.nthreads</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of threads currently assigned to
+        arena.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.pactive</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of pages in active runs.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.pdirty">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.pdirty</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of pages within unused runs that are potentially
+        dirty, and for which <function>madvise<parameter>...</parameter>
+        <parameter><constant>MADV_DONTNEED</constant></parameter></function> or
+        similar has not been called.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mapped</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of mapped bytes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.npurge</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of dirty page purge sweeps performed.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.nmadvise</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of <function>madvise<parameter>...</parameter>
+        <parameter><constant>MADV_DONTNEED</constant></parameter></function> or
+        similar calls made to purge dirty pages.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.npurged</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of pages purged.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.small.allocated</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of bytes currently allocated by small objects.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.small.nmalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of allocation requests served by
+        small bins.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.small.ndalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of small objects returned to bins.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.small.nrequests</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of small allocation requests.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.large.allocated</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of bytes currently allocated by large objects.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.large.nmalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of large allocation requests served
+        directly by the arena.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.large.ndalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of large deallocation requests served
+        directly by the arena.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.large.nrequests</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of large allocation requests.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.allocated</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Current number of bytes allocated by
+        bin.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nmalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of allocations served by bin.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.ndalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of allocations returned to bin.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nrequests</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of allocation
+        requests.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nfills</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option> <option>--enable-tcache</option>]
+        </term>
+        <listitem><para>Cumulative number of tcache fills.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nflushes</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option> <option>--enable-tcache</option>]
+        </term>
+        <listitem><para>Cumulative number of tcache flushes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nruns</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of runs created.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nreruns</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times the current run from which
+        to allocate changed.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.curruns</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Current number of runs.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.lruns.&lt;j&gt;.nmalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of allocation requests for this size
+        class served directly by the arena.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.lruns.&lt;j&gt;.ndalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of deallocation requests for this
+        size class served directly by the arena.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.lruns.&lt;j&gt;.nrequests</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of allocation requests for this size
+        class.</para></listitem>
+      </varlistentry>
+
+      <varlistentry>
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.lruns.&lt;j&gt;.curruns</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Current number of runs for this size class.
+        </para></listitem>
+      </varlistentry>
+    </variablelist>
+  </refsect1>
+  <refsect1 id="debugging_malloc_problems">
+    <title>DEBUGGING MALLOC PROBLEMS</title>
+    <para>When debugging, it is a good idea to configure/build jemalloc with
+    the <option>--enable-debug</option> and <option>--enable-fill</option>
+    options, and recompile the program with suitable options and symbols for
+    debugger support.  When so configured, jemalloc incorporates a wide variety
+    of run-time assertions that catch application errors such as double-free,
+    write-after-free, etc.</para>
+
+    <para>Programs often accidentally depend on &ldquo;uninitialized&rdquo;
+    memory actually being filled with zero bytes.  Junk filling
+    (see the <link linkend="opt.junk"><mallctl>opt.junk</mallctl></link>
+    option) tends to expose such bugs in the form of obviously incorrect
+    results and/or coredumps.  Conversely, zero
+    filling (see the <link
+    linkend="opt.zero"><mallctl>opt.zero</mallctl></link> option) eliminates
+    the symptoms of such bugs.  Between these two options, it is usually
+    possible to quickly detect, diagnose, and eliminate such bugs.</para>
+
+    <para>This implementation does not provide much detail about the problems
+    it detects, because the performance impact for storing such information
+    would be prohibitive.  However, jemalloc does integrate with the most
+    excellent <ulink url="http://valgrind.org/">Valgrind</ulink> tool if the
+    <option>--enable-valgrind</option> configuration option is enabled.</para>
+  </refsect1>
+  <refsect1 id="diagnostic_messages">
+    <title>DIAGNOSTIC MESSAGES</title>
+    <para>If any of the memory allocation/deallocation functions detect an
+    error or warning condition, a message will be printed to file descriptor
+    <constant>STDERR_FILENO</constant>.  Errors will result in the process
+    dumping core.  If the <link
+    linkend="opt.abort"><mallctl>opt.abort</mallctl></link> option is set, most
+    warnings are treated as errors.</para>
+
+    <para>The <varname>malloc_message</varname> variable allows the programmer
+    to override the function which emits the text strings forming the errors
+    and warnings if for some reason the <constant>STDERR_FILENO</constant> file
+    descriptor is not suitable for this.
+    <function>malloc_message<parameter/></function> takes the
+    <parameter>cbopaque</parameter> pointer argument that is
+    <constant>NULL</constant> unless overridden by the arguments in a call to
+    <function>malloc_stats_print<parameter/></function>, followed by a string
+    pointer.  Please note that doing anything which tries to allocate memory in
+    this function is likely to result in a crash or deadlock.</para>
+
+    <para>All messages are prefixed by
+    &ldquo;<computeroutput>&lt;jemalloc&gt;: </computeroutput>&rdquo;.</para>
+  </refsect1>
+  <refsect1 id="return_values">
+    <title>RETURN VALUES</title>
+    <refsect2>
+      <title>Standard API</title>
+      <para>The <function>malloc<parameter/></function> and
+      <function>calloc<parameter/></function> functions return a pointer to the
+      allocated memory if successful; otherwise a <constant>NULL</constant>
+      pointer is returned and <varname>errno</varname> is set to
+      <errorname>ENOMEM</errorname>.</para>
+
+      <para>The <function>posix_memalign<parameter/></function> function
+      returns the value 0 if successful; otherwise it returns an error value.
+      The <function>posix_memalign<parameter/></function> function will fail
+      if:
+        <variablelist>
+          <varlistentry>
+            <term><errorname>EINVAL</errorname></term>
+
+            <listitem><para>The <parameter>alignment</parameter> parameter is
+            not a power of 2 at least as large as
+            <code language="C">sizeof(<type>void *</type>)</code>.
+            </para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>ENOMEM</errorname></term>
+
+            <listitem><para>Memory allocation error.</para></listitem>
+          </varlistentry>
+        </variablelist>
+      </para>
+
+      <para>The <function>aligned_alloc<parameter/></function> function returns
+      a pointer to the allocated memory if successful; otherwise a
+      <constant>NULL</constant> pointer is returned and
+      <varname>errno</varname> is set.  The
+      <function>aligned_alloc<parameter/></function> function will fail if:
+        <variablelist>
+          <varlistentry>
+            <term><errorname>EINVAL</errorname></term>
+
+            <listitem><para>The <parameter>alignment</parameter> parameter is
+            not a power of 2.
+            </para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>ENOMEM</errorname></term>
+
+            <listitem><para>Memory allocation error.</para></listitem>
+          </varlistentry>
+        </variablelist>
+      </para>
+
+      <para>The <function>realloc<parameter/></function> function returns a
+      pointer, possibly identical to <parameter>ptr</parameter>, to the
+      allocated memory if successful; otherwise a <constant>NULL</constant>
+      pointer is returned, and <varname>errno</varname> is set to
+      <errorname>ENOMEM</errorname> if the error was the result of an
+      allocation failure.  The <function>realloc<parameter/></function>
+      function always leaves the original buffer intact when an error occurs.
+      </para>
+
+      <para>The <function>free<parameter/></function> function returns no
+      value.</para>
+    </refsect2>
+    <refsect2>
+      <title>Non-standard API</title>
+      <para>The <function>malloc_usable_size<parameter/></function> function
+      returns the usable size of the allocation pointed to by
+      <parameter>ptr</parameter>.  </para>
+
+      <para>The <function>mallctl<parameter/></function>,
+      <function>mallctlnametomib<parameter/></function>, and
+      <function>mallctlbymib<parameter/></function> functions return 0 on
+      success; otherwise they return an error value.  The functions will fail
+      if:
+        <variablelist>
+          <varlistentry>
+            <term><errorname>EINVAL</errorname></term>
+
+            <listitem><para><parameter>newp</parameter> is not
+            <constant>NULL</constant>, and <parameter>newlen</parameter> is too
+            large or too small.  Alternatively, <parameter>*oldlenp</parameter>
+            is too large or too small; in this case as much data as possible
+            are read despite the error.</para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>ENOMEM</errorname></term>
+
+            <listitem><para><parameter>*oldlenp</parameter> is too short to
+            hold the requested value.</para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>ENOENT</errorname></term>
+
+            <listitem><para><parameter>name</parameter> or
+            <parameter>mib</parameter> specifies an unknown/invalid
+            value.</para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>EPERM</errorname></term>
+
+            <listitem><para>Attempt to read or write void value, or attempt to
+            write read-only value.</para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>EAGAIN</errorname></term>
+
+            <listitem><para>A memory allocation failure
+            occurred.</para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>EFAULT</errorname></term>
+
+            <listitem><para>An interface with side effects failed in some way
+            not directly related to <function>mallctl*<parameter/></function>
+            read/write processing.</para></listitem>
+          </varlistentry>
+        </variablelist>
+      </para>
+    </refsect2>
+    <refsect2>
+      <title>Experimental API</title>
+      <para>The <function>allocm<parameter/></function>,
+      <function>rallocm<parameter/></function>,
+      <function>sallocm<parameter/></function>,
+      <function>dallocm<parameter/></function>, and
+      <function>nallocm<parameter/></function> functions return
+      <constant>ALLOCM_SUCCESS</constant> on success; otherwise they return an
+      error value.  The <function>allocm<parameter/></function>,
+      <function>rallocm<parameter/></function>, and
+      <function>nallocm<parameter/></function> functions will fail if:
+        <variablelist>
+          <varlistentry>
+            <term><errorname>ALLOCM_ERR_OOM</errorname></term>
+
+            <listitem><para>Out of memory.  Insufficient contiguous memory was
+            available to service the allocation request.  The
+            <function>allocm<parameter/></function> function additionally sets
+            <parameter>*ptr</parameter> to <constant>NULL</constant>, whereas
+            the <function>rallocm<parameter/></function> function leaves
+            <constant>*ptr</constant> unmodified.</para></listitem>
+          </varlistentry>
+        </variablelist>
+      The <function>rallocm<parameter/></function> function will also
+      fail if:
+        <variablelist>
+          <varlistentry>
+            <term><errorname>ALLOCM_ERR_NOT_MOVED</errorname></term>
+
+            <listitem><para><constant>ALLOCM_NO_MOVE</constant> was specified,
+            but the reallocation request could not be serviced without moving
+            the object.</para></listitem>
+          </varlistentry>
+        </variablelist>
+      </para>
+    </refsect2>
+  </refsect1>
+  <refsect1 id="environment">
+    <title>ENVIRONMENT</title>
+    <para>The following environment variable affects the execution of the
+    allocation functions:
+      <variablelist>
+        <varlistentry>
+          <term><envar>MALLOC_CONF</envar></term>
+
+          <listitem><para>If the environment variable
+          <envar>MALLOC_CONF</envar> is set, the characters it contains
+          will be interpreted as options.</para></listitem>
+        </varlistentry>
+      </variablelist>
+    </para>
+  </refsect1>
+  <refsect1 id="examples">
+    <title>EXAMPLES</title>
+    <para>To dump core whenever a problem occurs:
+      <screen>ln -s 'abort:true' /etc/malloc.conf</screen>
+    </para>
+    <para>To specify in the source a chunk size that is 16 MiB:
+      <programlisting language="C"><![CDATA[
+malloc_conf = "lg_chunk:24";]]></programlisting></para>
+  </refsect1>
+  <refsect1 id="see_also">
+    <title>SEE ALSO</title>
+    <para><citerefentry><refentrytitle>madvise</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>mmap</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>sbrk</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>utrace</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>alloca</refentrytitle>
+    <manvolnum>3</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>atexit</refentrytitle>
+    <manvolnum>3</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>getpagesize</refentrytitle>
+    <manvolnum>3</manvolnum></citerefentry></para>
+  </refsect1>
+  <refsect1 id="standards">
+    <title>STANDARDS</title>
+    <para>The <function>malloc<parameter/></function>,
+    <function>calloc<parameter/></function>,
+    <function>realloc<parameter/></function>, and
+    <function>free<parameter/></function> functions conform to ISO/IEC
+    9899:1990 (&ldquo;ISO C90&rdquo;).</para>
+
+    <para>The <function>posix_memalign<parameter/></function> function conforms
+    to IEEE Std 1003.1-2001 (&ldquo;POSIX.1&rdquo;).</para>
+  </refsect1>
+</refentry>
diff --git a/extra/jemalloc/doc/manpages.xsl.in b/extra/jemalloc/doc/manpages.xsl.in
new file mode 100644
index 00000000000..88b2626b958
--- /dev/null
+++ b/extra/jemalloc/doc/manpages.xsl.in
@@ -0,0 +1,4 @@
+<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
+  <xsl:import href="@XSLROOT@/manpages/docbook.xsl"/>
+  <xsl:import href="@abs_srcroot@doc/stylesheet.xsl"/>
+</xsl:stylesheet>
diff --git a/extra/jemalloc/doc/stylesheet.xsl b/extra/jemalloc/doc/stylesheet.xsl
new file mode 100644
index 00000000000..4e334a86f87
--- /dev/null
+++ b/extra/jemalloc/doc/stylesheet.xsl
@@ -0,0 +1,7 @@
+<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
+  <xsl:param name="funcsynopsis.style">ansi</xsl:param>
+  <xsl:param name="function.parens" select="1"/>
+  <xsl:template match="mallctl">
+    "<xsl:call-template name="inline.monoseq"/>"
+  </xsl:template>
+</xsl:stylesheet>
diff --git a/extra/jemalloc/include/jemalloc/internal/arena.h b/extra/jemalloc/include/jemalloc/internal/arena.h
new file mode 100644
index 00000000000..f2c18f43543
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/arena.h
@@ -0,0 +1,1022 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+/*
+ * RUN_MAX_OVRHD indicates maximum desired run header overhead.  Runs are sized
+ * as small as possible such that this setting is still honored, without
+ * violating other constraints.  The goal is to make runs as small as possible
+ * without exceeding a per run external fragmentation threshold.
+ *
+ * We use binary fixed point math for overhead computations, where the binary
+ * point is implicitly RUN_BFP bits to the left.
+ *
+ * Note that it is possible to set RUN_MAX_OVRHD low enough that it cannot be
+ * honored for some/all object sizes, since when heap profiling is enabled
+ * there is one pointer of header overhead per object (plus a constant).  This
+ * constraint is relaxed (ignored) for runs that are so small that the
+ * per-region overhead is greater than:
+ *
+ *   (RUN_MAX_OVRHD / (reg_interval << (3+RUN_BFP))
+ */
+#define	RUN_BFP			12
+/*                                    \/   Implicit binary fixed point. */
+#define	RUN_MAX_OVRHD		0x0000003dU
+#define	RUN_MAX_OVRHD_RELAX	0x00001800U
+
+/* Maximum number of regions in one run. */
+#define	LG_RUN_MAXREGS		11
+#define	RUN_MAXREGS		(1U << LG_RUN_MAXREGS)
+
+/*
+ * Minimum redzone size.  Redzones may be larger than this if necessary to
+ * preserve region alignment.
+ */
+#define	REDZONE_MINSIZE		16
+
+/*
+ * The minimum ratio of active:dirty pages per arena is computed as:
+ *
+ *   (nactive >> opt_lg_dirty_mult) >= ndirty
+ *
+ * So, supposing that opt_lg_dirty_mult is 3, there can be no less than 8 times
+ * as many active pages as dirty pages.
+ */
+#define	LG_DIRTY_MULT_DEFAULT	3
+
+typedef struct arena_chunk_map_s arena_chunk_map_t;
+typedef struct arena_chunk_s arena_chunk_t;
+typedef struct arena_run_s arena_run_t;
+typedef struct arena_bin_info_s arena_bin_info_t;
+typedef struct arena_bin_s arena_bin_t;
+typedef struct arena_s arena_t;
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+/* Each element of the chunk map corresponds to one page within the chunk. */
+struct arena_chunk_map_s {
+#ifndef JEMALLOC_PROF
+	/*
+	 * Overlay prof_ctx in order to allow it to be referenced by dead code.
+	 * Such antics aren't warranted for per arena data structures, but
+	 * chunk map overhead accounts for a percentage of memory, rather than
+	 * being just a fixed cost.
+	 */
+	union {
+#endif
+	union {
+		/*
+		 * Linkage for run trees.  There are two disjoint uses:
+		 *
+		 * 1) arena_t's runs_avail tree.
+		 * 2) arena_run_t conceptually uses this linkage for in-use
+		 *    non-full runs, rather than directly embedding linkage.
+		 */
+		rb_node(arena_chunk_map_t)	rb_link;
+		/*
+		 * List of runs currently in purgatory.  arena_chunk_purge()
+		 * temporarily allocates runs that contain dirty pages while
+		 * purging, so that other threads cannot use the runs while the
+		 * purging thread is operating without the arena lock held.
+		 */
+		ql_elm(arena_chunk_map_t)	ql_link;
+	}				u;
+
+	/* Profile counters, used for large object runs. */
+	prof_ctx_t			*prof_ctx;
+#ifndef JEMALLOC_PROF
+	}; /* union { ... }; */
+#endif
+
+	/*
+	 * Run address (or size) and various flags are stored together.  The bit
+	 * layout looks like (assuming 32-bit system):
+	 *
+	 *   ???????? ???????? ????nnnn nnnndula
+	 *
+	 * ? : Unallocated: Run address for first/last pages, unset for internal
+	 *                  pages.
+	 *     Small: Run page offset.
+	 *     Large: Run size for first page, unset for trailing pages.
+	 * n : binind for small size class, BININD_INVALID for large size class.
+	 * d : dirty?
+	 * u : unzeroed?
+	 * l : large?
+	 * a : allocated?
+	 *
+	 * Following are example bit patterns for the three types of runs.
+	 *
+	 * p : run page offset
+	 * s : run size
+	 * n : binind for size class; large objects set these to BININD_INVALID
+	 *     except for promoted allocations (see prof_promote)
+	 * x : don't care
+	 * - : 0
+	 * + : 1
+	 * [DULA] : bit set
+	 * [dula] : bit unset
+	 *
+	 *   Unallocated (clean):
+	 *     ssssssss ssssssss ssss++++ ++++du-a
+	 *     xxxxxxxx xxxxxxxx xxxxxxxx xxxx-Uxx
+	 *     ssssssss ssssssss ssss++++ ++++dU-a
+	 *
+	 *   Unallocated (dirty):
+	 *     ssssssss ssssssss ssss++++ ++++D--a
+	 *     xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx
+	 *     ssssssss ssssssss ssss++++ ++++D--a
+	 *
+	 *   Small:
+	 *     pppppppp pppppppp ppppnnnn nnnnd--A
+	 *     pppppppp pppppppp ppppnnnn nnnn---A
+	 *     pppppppp pppppppp ppppnnnn nnnnd--A
+	 *
+	 *   Large:
+	 *     ssssssss ssssssss ssss++++ ++++D-LA
+	 *     xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx
+	 *     -------- -------- ----++++ ++++D-LA
+	 *
+	 *   Large (sampled, size <= PAGE):
+	 *     ssssssss ssssssss ssssnnnn nnnnD-LA
+	 *
+	 *   Large (not sampled, size == PAGE):
+	 *     ssssssss ssssssss ssss++++ ++++D-LA
+	 */
+	size_t				bits;
+#define	CHUNK_MAP_BININD_SHIFT	4
+#define	BININD_INVALID		((size_t)0xffU)
+/*     CHUNK_MAP_BININD_MASK == (BININD_INVALID << CHUNK_MAP_BININD_SHIFT) */
+#define	CHUNK_MAP_BININD_MASK	((size_t)0xff0U)
+#define	CHUNK_MAP_BININD_INVALID CHUNK_MAP_BININD_MASK
+#define	CHUNK_MAP_FLAGS_MASK	((size_t)0xcU)
+#define	CHUNK_MAP_DIRTY		((size_t)0x8U)
+#define	CHUNK_MAP_UNZEROED	((size_t)0x4U)
+#define	CHUNK_MAP_LARGE		((size_t)0x2U)
+#define	CHUNK_MAP_ALLOCATED	((size_t)0x1U)
+#define	CHUNK_MAP_KEY		CHUNK_MAP_ALLOCATED
+};
+typedef rb_tree(arena_chunk_map_t) arena_avail_tree_t;
+typedef rb_tree(arena_chunk_map_t) arena_run_tree_t;
+
+/* Arena chunk header. */
+struct arena_chunk_s {
+	/* Arena that owns the chunk. */
+	arena_t			*arena;
+
+	/* Linkage for tree of arena chunks that contain dirty runs. */
+	rb_node(arena_chunk_t)	dirty_link;
+
+	/* Number of dirty pages. */
+	size_t			ndirty;
+
+	/* Number of available runs. */
+	size_t			nruns_avail;
+
+	/*
+	 * Number of available run adjacencies.  Clean and dirty available runs
+	 * are not coalesced, which causes virtual memory fragmentation.  The
+	 * ratio of (nruns_avail-nruns_adjac):nruns_adjac is used for tracking
+	 * this fragmentation.
+	 * */
+	size_t			nruns_adjac;
+
+	/*
+	 * Map of pages within chunk that keeps track of free/large/small.  The
+	 * first map_bias entries are omitted, since the chunk header does not
+	 * need to be tracked in the map.  This omission saves a header page
+	 * for common chunk sizes (e.g. 4 MiB).
+	 */
+	arena_chunk_map_t	map[1]; /* Dynamically sized. */
+};
+typedef rb_tree(arena_chunk_t) arena_chunk_tree_t;
+
+struct arena_run_s {
+	/* Bin this run is associated with. */
+	arena_bin_t	*bin;
+
+	/* Index of next region that has never been allocated, or nregs. */
+	uint32_t	nextind;
+
+	/* Number of free regions in run. */
+	unsigned	nfree;
+};
+
+/*
+ * Read-only information associated with each element of arena_t's bins array
+ * is stored separately, partly to reduce memory usage (only one copy, rather
+ * than one per arena), but mainly to avoid false cacheline sharing.
+ *
+ * Each run has the following layout:
+ *
+ *               /--------------------\
+ *               | arena_run_t header |
+ *               | ...                |
+ * bitmap_offset | bitmap             |
+ *               | ...                |
+ *   ctx0_offset | ctx map            |
+ *               | ...                |
+ *               |--------------------|
+ *               | redzone            |
+ *   reg0_offset | region 0           |
+ *               | redzone            |
+ *               |--------------------| \
+ *               | redzone            | |
+ *               | region 1           |  > reg_interval
+ *               | redzone            | /
+ *               |--------------------|
+ *               | ...                |
+ *               | ...                |
+ *               | ...                |
+ *               |--------------------|
+ *               | redzone            |
+ *               | region nregs-1     |
+ *               | redzone            |
+ *               |--------------------|
+ *               | alignment pad?     |
+ *               \--------------------/
+ *
+ * reg_interval has at least the same minimum alignment as reg_size; this
+ * preserves the alignment constraint that sa2u() depends on.  Alignment pad is
+ * either 0 or redzone_size; it is present only if needed to align reg0_offset.
+ */
+struct arena_bin_info_s {
+	/* Size of regions in a run for this bin's size class. */
+	size_t		reg_size;
+
+	/* Redzone size. */
+	size_t		redzone_size;
+
+	/* Interval between regions (reg_size + (redzone_size << 1)). */
+	size_t		reg_interval;
+
+	/* Total size of a run for this bin's size class. */
+	size_t		run_size;
+
+	/* Total number of regions in a run for this bin's size class. */
+	uint32_t	nregs;
+
+	/*
+	 * Offset of first bitmap_t element in a run header for this bin's size
+	 * class.
+	 */
+	uint32_t	bitmap_offset;
+
+	/*
+	 * Metadata used to manipulate bitmaps for runs associated with this
+	 * bin.
+	 */
+	bitmap_info_t	bitmap_info;
+
+	/*
+	 * Offset of first (prof_ctx_t *) in a run header for this bin's size
+	 * class, or 0 if (config_prof == false || opt_prof == false).
+	 */
+	uint32_t	ctx0_offset;
+
+	/* Offset of first region in a run for this bin's size class. */
+	uint32_t	reg0_offset;
+};
+
+struct arena_bin_s {
+	/*
+	 * All operations on runcur, runs, and stats require that lock be
+	 * locked.  Run allocation/deallocation are protected by the arena lock,
+	 * which may be acquired while holding one or more bin locks, but not
+	 * vise versa.
+	 */
+	malloc_mutex_t	lock;
+
+	/*
+	 * Current run being used to service allocations of this bin's size
+	 * class.
+	 */
+	arena_run_t	*runcur;
+
+	/*
+	 * Tree of non-full runs.  This tree is used when looking for an
+	 * existing run when runcur is no longer usable.  We choose the
+	 * non-full run that is lowest in memory; this policy tends to keep
+	 * objects packed well, and it can also help reduce the number of
+	 * almost-empty chunks.
+	 */
+	arena_run_tree_t runs;
+
+	/* Bin statistics. */
+	malloc_bin_stats_t stats;
+};
+
+struct arena_s {
+	/* This arena's index within the arenas array. */
+	unsigned		ind;
+
+	/*
+	 * Number of threads currently assigned to this arena.  This field is
+	 * protected by arenas_lock.
+	 */
+	unsigned		nthreads;
+
+	/*
+	 * There are three classes of arena operations from a locking
+	 * perspective:
+	 * 1) Thread asssignment (modifies nthreads) is protected by
+	 *    arenas_lock.
+	 * 2) Bin-related operations are protected by bin locks.
+	 * 3) Chunk- and run-related operations are protected by this mutex.
+	 */
+	malloc_mutex_t		lock;
+
+	arena_stats_t		stats;
+	/*
+	 * List of tcaches for extant threads associated with this arena.
+	 * Stats from these are merged incrementally, and at exit.
+	 */
+	ql_head(tcache_t)	tcache_ql;
+
+	uint64_t		prof_accumbytes;
+
+	dss_prec_t		dss_prec;
+
+	/* Tree of dirty-page-containing chunks this arena manages. */
+	arena_chunk_tree_t	chunks_dirty;
+
+	/*
+	 * In order to avoid rapid chunk allocation/deallocation when an arena
+	 * oscillates right on the cusp of needing a new chunk, cache the most
+	 * recently freed chunk.  The spare is left in the arena's chunk trees
+	 * until it is deleted.
+	 *
+	 * There is one spare chunk per arena, rather than one spare total, in
+	 * order to avoid interactions between multiple threads that could make
+	 * a single spare inadequate.
+	 */
+	arena_chunk_t		*spare;
+
+	/* Number of pages in active runs. */
+	size_t			nactive;
+
+	/*
+	 * Current count of pages within unused runs that are potentially
+	 * dirty, and for which madvise(... MADV_DONTNEED) has not been called.
+	 * By tracking this, we can institute a limit on how much dirty unused
+	 * memory is mapped for each arena.
+	 */
+	size_t			ndirty;
+
+	/*
+	 * Approximate number of pages being purged.  It is possible for
+	 * multiple threads to purge dirty pages concurrently, and they use
+	 * npurgatory to indicate the total number of pages all threads are
+	 * attempting to purge.
+	 */
+	size_t			npurgatory;
+
+	/*
+	 * Size/address-ordered trees of this arena's available runs.  The trees
+	 * are used for first-best-fit run allocation.
+	 */
+	arena_avail_tree_t	runs_avail;
+
+	/* bins is used to store trees of free regions. */
+	arena_bin_t		bins[NBINS];
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+extern ssize_t	opt_lg_dirty_mult;
+/*
+ * small_size2bin is a compact lookup table that rounds request sizes up to
+ * size classes.  In order to reduce cache footprint, the table is compressed,
+ * and all accesses are via the SMALL_SIZE2BIN macro.
+ */
+extern uint8_t const	small_size2bin[];
+#define	SMALL_SIZE2BIN(s)	(small_size2bin[(s-1) >> LG_TINY_MIN])
+
+extern arena_bin_info_t	arena_bin_info[NBINS];
+
+/* Number of large size classes. */
+#define			nlclasses (chunk_npages - map_bias)
+
+void	arena_purge_all(arena_t *arena);
+void	arena_tcache_fill_small(arena_t *arena, tcache_bin_t *tbin,
+    size_t binind, uint64_t prof_accumbytes);
+void	arena_alloc_junk_small(void *ptr, arena_bin_info_t *bin_info,
+    bool zero);
+void	arena_dalloc_junk_small(void *ptr, arena_bin_info_t *bin_info);
+void	*arena_malloc_small(arena_t *arena, size_t size, bool zero);
+void	*arena_malloc_large(arena_t *arena, size_t size, bool zero);
+void	*arena_palloc(arena_t *arena, size_t size, size_t alignment, bool zero);
+void	arena_prof_promoted(const void *ptr, size_t size);
+void	arena_dalloc_bin_locked(arena_t *arena, arena_chunk_t *chunk, void *ptr,
+    arena_chunk_map_t *mapelm);
+void	arena_dalloc_bin(arena_t *arena, arena_chunk_t *chunk, void *ptr,
+    size_t pageind, arena_chunk_map_t *mapelm);
+void	arena_dalloc_small(arena_t *arena, arena_chunk_t *chunk, void *ptr,
+    size_t pageind);
+void	arena_dalloc_large_locked(arena_t *arena, arena_chunk_t *chunk,
+    void *ptr);
+void	arena_dalloc_large(arena_t *arena, arena_chunk_t *chunk, void *ptr);
+void	*arena_ralloc_no_move(void *ptr, size_t oldsize, size_t size,
+    size_t extra, bool zero);
+void	*arena_ralloc(arena_t *arena, void *ptr, size_t oldsize, size_t size,
+    size_t extra, size_t alignment, bool zero, bool try_tcache_alloc,
+    bool try_tcache_dalloc);
+dss_prec_t	arena_dss_prec_get(arena_t *arena);
+void	arena_dss_prec_set(arena_t *arena, dss_prec_t dss_prec);
+void	arena_stats_merge(arena_t *arena, const char **dss, size_t *nactive,
+    size_t *ndirty, arena_stats_t *astats, malloc_bin_stats_t *bstats,
+    malloc_large_stats_t *lstats);
+bool	arena_new(arena_t *arena, unsigned ind);
+void	arena_boot(void);
+void	arena_prefork(arena_t *arena);
+void	arena_postfork_parent(arena_t *arena);
+void	arena_postfork_child(arena_t *arena);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+arena_chunk_map_t	*arena_mapp_get(arena_chunk_t *chunk, size_t pageind);
+size_t	*arena_mapbitsp_get(arena_chunk_t *chunk, size_t pageind);
+size_t	arena_mapbits_get(arena_chunk_t *chunk, size_t pageind);
+size_t	arena_mapbits_unallocated_size_get(arena_chunk_t *chunk,
+    size_t pageind);
+size_t	arena_mapbits_large_size_get(arena_chunk_t *chunk, size_t pageind);
+size_t	arena_mapbits_small_runind_get(arena_chunk_t *chunk, size_t pageind);
+size_t	arena_mapbits_binind_get(arena_chunk_t *chunk, size_t pageind);
+size_t	arena_mapbits_dirty_get(arena_chunk_t *chunk, size_t pageind);
+size_t	arena_mapbits_unzeroed_get(arena_chunk_t *chunk, size_t pageind);
+size_t	arena_mapbits_large_get(arena_chunk_t *chunk, size_t pageind);
+size_t	arena_mapbits_allocated_get(arena_chunk_t *chunk, size_t pageind);
+void	arena_mapbits_unallocated_set(arena_chunk_t *chunk, size_t pageind,
+    size_t size, size_t flags);
+void	arena_mapbits_unallocated_size_set(arena_chunk_t *chunk, size_t pageind,
+    size_t size);
+void	arena_mapbits_large_set(arena_chunk_t *chunk, size_t pageind,
+    size_t size, size_t flags);
+void	arena_mapbits_large_binind_set(arena_chunk_t *chunk, size_t pageind,
+    size_t binind);
+void	arena_mapbits_small_set(arena_chunk_t *chunk, size_t pageind,
+    size_t runind, size_t binind, size_t flags);
+void	arena_mapbits_unzeroed_set(arena_chunk_t *chunk, size_t pageind,
+    size_t unzeroed);
+bool	arena_prof_accum_impl(arena_t *arena, uint64_t accumbytes);
+bool	arena_prof_accum_locked(arena_t *arena, uint64_t accumbytes);
+bool	arena_prof_accum(arena_t *arena, uint64_t accumbytes);
+size_t	arena_ptr_small_binind_get(const void *ptr, size_t mapbits);
+size_t	arena_bin_index(arena_t *arena, arena_bin_t *bin);
+unsigned	arena_run_regind(arena_run_t *run, arena_bin_info_t *bin_info,
+    const void *ptr);
+prof_ctx_t	*arena_prof_ctx_get(const void *ptr);
+void	arena_prof_ctx_set(const void *ptr, prof_ctx_t *ctx);
+void	*arena_malloc(arena_t *arena, size_t size, bool zero, bool try_tcache);
+size_t	arena_salloc(const void *ptr, bool demote);
+void	arena_dalloc(arena_t *arena, arena_chunk_t *chunk, void *ptr,
+    bool try_tcache);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_ARENA_C_))
+#  ifdef JEMALLOC_ARENA_INLINE_A
+JEMALLOC_ALWAYS_INLINE arena_chunk_map_t *
+arena_mapp_get(arena_chunk_t *chunk, size_t pageind)
+{
+
+	assert(pageind >= map_bias);
+	assert(pageind < chunk_npages);
+
+	return (&chunk->map[pageind-map_bias]);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t *
+arena_mapbitsp_get(arena_chunk_t *chunk, size_t pageind)
+{
+
+	return (&arena_mapp_get(chunk, pageind)->bits);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+arena_mapbits_get(arena_chunk_t *chunk, size_t pageind)
+{
+
+	return (*arena_mapbitsp_get(chunk, pageind));
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+arena_mapbits_unallocated_size_get(arena_chunk_t *chunk, size_t pageind)
+{
+	size_t mapbits;
+
+	mapbits = arena_mapbits_get(chunk, pageind);
+	assert((mapbits & (CHUNK_MAP_LARGE|CHUNK_MAP_ALLOCATED)) == 0);
+	return (mapbits & ~PAGE_MASK);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+arena_mapbits_large_size_get(arena_chunk_t *chunk, size_t pageind)
+{
+	size_t mapbits;
+
+	mapbits = arena_mapbits_get(chunk, pageind);
+	assert((mapbits & (CHUNK_MAP_LARGE|CHUNK_MAP_ALLOCATED)) ==
+	    (CHUNK_MAP_LARGE|CHUNK_MAP_ALLOCATED));
+	return (mapbits & ~PAGE_MASK);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+arena_mapbits_small_runind_get(arena_chunk_t *chunk, size_t pageind)
+{
+	size_t mapbits;
+
+	mapbits = arena_mapbits_get(chunk, pageind);
+	assert((mapbits & (CHUNK_MAP_LARGE|CHUNK_MAP_ALLOCATED)) ==
+	    CHUNK_MAP_ALLOCATED);
+	return (mapbits >> LG_PAGE);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+arena_mapbits_binind_get(arena_chunk_t *chunk, size_t pageind)
+{
+	size_t mapbits;
+	size_t binind;
+
+	mapbits = arena_mapbits_get(chunk, pageind);
+	binind = (mapbits & CHUNK_MAP_BININD_MASK) >> CHUNK_MAP_BININD_SHIFT;
+	assert(binind < NBINS || binind == BININD_INVALID);
+	return (binind);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+arena_mapbits_dirty_get(arena_chunk_t *chunk, size_t pageind)
+{
+	size_t mapbits;
+
+	mapbits = arena_mapbits_get(chunk, pageind);
+	return (mapbits & CHUNK_MAP_DIRTY);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+arena_mapbits_unzeroed_get(arena_chunk_t *chunk, size_t pageind)
+{
+	size_t mapbits;
+
+	mapbits = arena_mapbits_get(chunk, pageind);
+	return (mapbits & CHUNK_MAP_UNZEROED);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+arena_mapbits_large_get(arena_chunk_t *chunk, size_t pageind)
+{
+	size_t mapbits;
+
+	mapbits = arena_mapbits_get(chunk, pageind);
+	return (mapbits & CHUNK_MAP_LARGE);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+arena_mapbits_allocated_get(arena_chunk_t *chunk, size_t pageind)
+{
+	size_t mapbits;
+
+	mapbits = arena_mapbits_get(chunk, pageind);
+	return (mapbits & CHUNK_MAP_ALLOCATED);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_mapbits_unallocated_set(arena_chunk_t *chunk, size_t pageind, size_t size,
+    size_t flags)
+{
+	size_t *mapbitsp;
+
+	mapbitsp = arena_mapbitsp_get(chunk, pageind);
+	assert((size & PAGE_MASK) == 0);
+	assert((flags & ~CHUNK_MAP_FLAGS_MASK) == 0);
+	assert((flags & (CHUNK_MAP_DIRTY|CHUNK_MAP_UNZEROED)) == flags);
+	*mapbitsp = size | CHUNK_MAP_BININD_INVALID | flags;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_mapbits_unallocated_size_set(arena_chunk_t *chunk, size_t pageind,
+    size_t size)
+{
+	size_t *mapbitsp;
+
+	mapbitsp = arena_mapbitsp_get(chunk, pageind);
+	assert((size & PAGE_MASK) == 0);
+	assert((*mapbitsp & (CHUNK_MAP_LARGE|CHUNK_MAP_ALLOCATED)) == 0);
+	*mapbitsp = size | (*mapbitsp & PAGE_MASK);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_mapbits_large_set(arena_chunk_t *chunk, size_t pageind, size_t size,
+    size_t flags)
+{
+	size_t *mapbitsp;
+	size_t unzeroed;
+
+	mapbitsp = arena_mapbitsp_get(chunk, pageind);
+	assert((size & PAGE_MASK) == 0);
+	assert((flags & CHUNK_MAP_DIRTY) == flags);
+	unzeroed = *mapbitsp & CHUNK_MAP_UNZEROED; /* Preserve unzeroed. */
+	*mapbitsp = size | CHUNK_MAP_BININD_INVALID | flags | unzeroed |
+	    CHUNK_MAP_LARGE | CHUNK_MAP_ALLOCATED;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_mapbits_large_binind_set(arena_chunk_t *chunk, size_t pageind,
+    size_t binind)
+{
+	size_t *mapbitsp;
+
+	assert(binind <= BININD_INVALID);
+	mapbitsp = arena_mapbitsp_get(chunk, pageind);
+	assert(arena_mapbits_large_size_get(chunk, pageind) == PAGE);
+	*mapbitsp = (*mapbitsp & ~CHUNK_MAP_BININD_MASK) | (binind <<
+	    CHUNK_MAP_BININD_SHIFT);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_mapbits_small_set(arena_chunk_t *chunk, size_t pageind, size_t runind,
+    size_t binind, size_t flags)
+{
+	size_t *mapbitsp;
+	size_t unzeroed;
+
+	assert(binind < BININD_INVALID);
+	mapbitsp = arena_mapbitsp_get(chunk, pageind);
+	assert(pageind - runind >= map_bias);
+	assert((flags & CHUNK_MAP_DIRTY) == flags);
+	unzeroed = *mapbitsp & CHUNK_MAP_UNZEROED; /* Preserve unzeroed. */
+	*mapbitsp = (runind << LG_PAGE) | (binind << CHUNK_MAP_BININD_SHIFT) |
+	    flags | unzeroed | CHUNK_MAP_ALLOCATED;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_mapbits_unzeroed_set(arena_chunk_t *chunk, size_t pageind,
+    size_t unzeroed)
+{
+	size_t *mapbitsp;
+
+	mapbitsp = arena_mapbitsp_get(chunk, pageind);
+	*mapbitsp = (*mapbitsp & ~CHUNK_MAP_UNZEROED) | unzeroed;
+}
+
+JEMALLOC_INLINE bool
+arena_prof_accum_impl(arena_t *arena, uint64_t accumbytes)
+{
+
+	cassert(config_prof);
+	assert(prof_interval != 0);
+
+	arena->prof_accumbytes += accumbytes;
+	if (arena->prof_accumbytes >= prof_interval) {
+		arena->prof_accumbytes -= prof_interval;
+		return (true);
+	}
+	return (false);
+}
+
+JEMALLOC_INLINE bool
+arena_prof_accum_locked(arena_t *arena, uint64_t accumbytes)
+{
+
+	cassert(config_prof);
+
+	if (prof_interval == 0)
+		return (false);
+	return (arena_prof_accum_impl(arena, accumbytes));
+}
+
+JEMALLOC_INLINE bool
+arena_prof_accum(arena_t *arena, uint64_t accumbytes)
+{
+
+	cassert(config_prof);
+
+	if (prof_interval == 0)
+		return (false);
+
+	{
+		bool ret;
+
+		malloc_mutex_lock(&arena->lock);
+		ret = arena_prof_accum_impl(arena, accumbytes);
+		malloc_mutex_unlock(&arena->lock);
+		return (ret);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+arena_ptr_small_binind_get(const void *ptr, size_t mapbits)
+{
+	size_t binind;
+
+	binind = (mapbits & CHUNK_MAP_BININD_MASK) >> CHUNK_MAP_BININD_SHIFT;
+
+	if (config_debug) {
+		arena_chunk_t *chunk;
+		arena_t *arena;
+		size_t pageind;
+		size_t actual_mapbits;
+		arena_run_t *run;
+		arena_bin_t *bin;
+		size_t actual_binind;
+		arena_bin_info_t *bin_info;
+
+		assert(binind != BININD_INVALID);
+		assert(binind < NBINS);
+		chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+		arena = chunk->arena;
+		pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+		actual_mapbits = arena_mapbits_get(chunk, pageind);
+		assert(mapbits == actual_mapbits);
+		assert(arena_mapbits_large_get(chunk, pageind) == 0);
+		assert(arena_mapbits_allocated_get(chunk, pageind) != 0);
+		run = (arena_run_t *)((uintptr_t)chunk + (uintptr_t)((pageind -
+		    (actual_mapbits >> LG_PAGE)) << LG_PAGE));
+		bin = run->bin;
+		actual_binind = bin - arena->bins;
+		assert(binind == actual_binind);
+		bin_info = &arena_bin_info[actual_binind];
+		assert(((uintptr_t)ptr - ((uintptr_t)run +
+		    (uintptr_t)bin_info->reg0_offset)) % bin_info->reg_interval
+		    == 0);
+	}
+
+	return (binind);
+}
+#  endif /* JEMALLOC_ARENA_INLINE_A */
+
+#  ifdef JEMALLOC_ARENA_INLINE_B
+JEMALLOC_INLINE size_t
+arena_bin_index(arena_t *arena, arena_bin_t *bin)
+{
+	size_t binind = bin - arena->bins;
+	assert(binind < NBINS);
+	return (binind);
+}
+
+JEMALLOC_INLINE unsigned
+arena_run_regind(arena_run_t *run, arena_bin_info_t *bin_info, const void *ptr)
+{
+	unsigned shift, diff, regind;
+	size_t interval;
+
+	/*
+	 * Freeing a pointer lower than region zero can cause assertion
+	 * failure.
+	 */
+	assert((uintptr_t)ptr >= (uintptr_t)run +
+	    (uintptr_t)bin_info->reg0_offset);
+
+	/*
+	 * Avoid doing division with a variable divisor if possible.  Using
+	 * actual division here can reduce allocator throughput by over 20%!
+	 */
+	diff = (unsigned)((uintptr_t)ptr - (uintptr_t)run -
+	    bin_info->reg0_offset);
+
+	/* Rescale (factor powers of 2 out of the numerator and denominator). */
+	interval = bin_info->reg_interval;
+	shift = ffs(interval) - 1;
+	diff >>= shift;
+	interval >>= shift;
+
+	if (interval == 1) {
+		/* The divisor was a power of 2. */
+		regind = diff;
+	} else {
+		/*
+		 * To divide by a number D that is not a power of two we
+		 * multiply by (2^21 / D) and then right shift by 21 positions.
+		 *
+		 *   X / D
+		 *
+		 * becomes
+		 *
+		 *   (X * interval_invs[D - 3]) >> SIZE_INV_SHIFT
+		 *
+		 * We can omit the first three elements, because we never
+		 * divide by 0, and 1 and 2 are both powers of two, which are
+		 * handled above.
+		 */
+#define	SIZE_INV_SHIFT	((sizeof(unsigned) << 3) - LG_RUN_MAXREGS)
+#define	SIZE_INV(s)	(((1U << SIZE_INV_SHIFT) / (s)) + 1)
+		static const unsigned interval_invs[] = {
+		    SIZE_INV(3),
+		    SIZE_INV(4), SIZE_INV(5), SIZE_INV(6), SIZE_INV(7),
+		    SIZE_INV(8), SIZE_INV(9), SIZE_INV(10), SIZE_INV(11),
+		    SIZE_INV(12), SIZE_INV(13), SIZE_INV(14), SIZE_INV(15),
+		    SIZE_INV(16), SIZE_INV(17), SIZE_INV(18), SIZE_INV(19),
+		    SIZE_INV(20), SIZE_INV(21), SIZE_INV(22), SIZE_INV(23),
+		    SIZE_INV(24), SIZE_INV(25), SIZE_INV(26), SIZE_INV(27),
+		    SIZE_INV(28), SIZE_INV(29), SIZE_INV(30), SIZE_INV(31)
+		};
+
+		if (interval <= ((sizeof(interval_invs) / sizeof(unsigned)) +
+		    2)) {
+			regind = (diff * interval_invs[interval - 3]) >>
+			    SIZE_INV_SHIFT;
+		} else
+			regind = diff / interval;
+#undef SIZE_INV
+#undef SIZE_INV_SHIFT
+	}
+	assert(diff == regind * interval);
+	assert(regind < bin_info->nregs);
+
+	return (regind);
+}
+
+JEMALLOC_INLINE prof_ctx_t *
+arena_prof_ctx_get(const void *ptr)
+{
+	prof_ctx_t *ret;
+	arena_chunk_t *chunk;
+	size_t pageind, mapbits;
+
+	cassert(config_prof);
+	assert(ptr != NULL);
+	assert(CHUNK_ADDR2BASE(ptr) != ptr);
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+	pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+	mapbits = arena_mapbits_get(chunk, pageind);
+	assert((mapbits & CHUNK_MAP_ALLOCATED) != 0);
+	if ((mapbits & CHUNK_MAP_LARGE) == 0) {
+		if (prof_promote)
+			ret = (prof_ctx_t *)(uintptr_t)1U;
+		else {
+			arena_run_t *run = (arena_run_t *)((uintptr_t)chunk +
+			    (uintptr_t)((pageind - (mapbits >> LG_PAGE)) <<
+			    LG_PAGE));
+			size_t binind = arena_ptr_small_binind_get(ptr,
+			    mapbits);
+			arena_bin_info_t *bin_info = &arena_bin_info[binind];
+			unsigned regind;
+
+			regind = arena_run_regind(run, bin_info, ptr);
+			ret = *(prof_ctx_t **)((uintptr_t)run +
+			    bin_info->ctx0_offset + (regind *
+			    sizeof(prof_ctx_t *)));
+		}
+	} else
+		ret = arena_mapp_get(chunk, pageind)->prof_ctx;
+
+	return (ret);
+}
+
+JEMALLOC_INLINE void
+arena_prof_ctx_set(const void *ptr, prof_ctx_t *ctx)
+{
+	arena_chunk_t *chunk;
+	size_t pageind, mapbits;
+
+	cassert(config_prof);
+	assert(ptr != NULL);
+	assert(CHUNK_ADDR2BASE(ptr) != ptr);
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+	pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+	mapbits = arena_mapbits_get(chunk, pageind);
+	assert((mapbits & CHUNK_MAP_ALLOCATED) != 0);
+	if ((mapbits & CHUNK_MAP_LARGE) == 0) {
+		if (prof_promote == false) {
+			arena_run_t *run = (arena_run_t *)((uintptr_t)chunk +
+			    (uintptr_t)((pageind - (mapbits >> LG_PAGE)) <<
+			    LG_PAGE));
+			size_t binind;
+			arena_bin_info_t *bin_info;
+			unsigned regind;
+
+			binind = arena_ptr_small_binind_get(ptr, mapbits);
+			bin_info = &arena_bin_info[binind];
+			regind = arena_run_regind(run, bin_info, ptr);
+
+			*((prof_ctx_t **)((uintptr_t)run + bin_info->ctx0_offset
+			    + (regind * sizeof(prof_ctx_t *)))) = ctx;
+		} else
+			assert((uintptr_t)ctx == (uintptr_t)1U);
+	} else
+		arena_mapp_get(chunk, pageind)->prof_ctx = ctx;
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+arena_malloc(arena_t *arena, size_t size, bool zero, bool try_tcache)
+{
+	tcache_t *tcache;
+
+	assert(size != 0);
+	assert(size <= arena_maxclass);
+
+	if (size <= SMALL_MAXCLASS) {
+		if (try_tcache && (tcache = tcache_get(true)) != NULL)
+			return (tcache_alloc_small(tcache, size, zero));
+		else {
+			return (arena_malloc_small(choose_arena(arena), size,
+			    zero));
+		}
+	} else {
+		/*
+		 * Initialize tcache after checking size in order to avoid
+		 * infinite recursion during tcache initialization.
+		 */
+		if (try_tcache && size <= tcache_maxclass && (tcache =
+		    tcache_get(true)) != NULL)
+			return (tcache_alloc_large(tcache, size, zero));
+		else {
+			return (arena_malloc_large(choose_arena(arena), size,
+			    zero));
+		}
+	}
+}
+
+/* Return the size of the allocation pointed to by ptr. */
+JEMALLOC_ALWAYS_INLINE size_t
+arena_salloc(const void *ptr, bool demote)
+{
+	size_t ret;
+	arena_chunk_t *chunk;
+	size_t pageind, binind;
+
+	assert(ptr != NULL);
+	assert(CHUNK_ADDR2BASE(ptr) != ptr);
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+	pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+	assert(arena_mapbits_allocated_get(chunk, pageind) != 0);
+	binind = arena_mapbits_binind_get(chunk, pageind);
+	if (binind == BININD_INVALID || (config_prof && demote == false &&
+	    prof_promote && arena_mapbits_large_get(chunk, pageind) != 0)) {
+		/*
+		 * Large allocation.  In the common case (demote == true), and
+		 * as this is an inline function, most callers will only end up
+		 * looking at binind to determine that ptr is a small
+		 * allocation.
+		 */
+		assert(((uintptr_t)ptr & PAGE_MASK) == 0);
+		ret = arena_mapbits_large_size_get(chunk, pageind);
+		assert(ret != 0);
+		assert(pageind + (ret>>LG_PAGE) <= chunk_npages);
+		assert(ret == PAGE || arena_mapbits_large_size_get(chunk,
+		    pageind+(ret>>LG_PAGE)-1) == 0);
+		assert(binind == arena_mapbits_binind_get(chunk,
+		    pageind+(ret>>LG_PAGE)-1));
+		assert(arena_mapbits_dirty_get(chunk, pageind) ==
+		    arena_mapbits_dirty_get(chunk, pageind+(ret>>LG_PAGE)-1));
+	} else {
+		/*
+		 * Small allocation (possibly promoted to a large object due to
+		 * prof_promote).
+		 */
+		assert(arena_mapbits_large_get(chunk, pageind) != 0 ||
+		    arena_ptr_small_binind_get(ptr, arena_mapbits_get(chunk,
+		    pageind)) == binind);
+		ret = arena_bin_info[binind].reg_size;
+	}
+
+	return (ret);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_dalloc(arena_t *arena, arena_chunk_t *chunk, void *ptr, bool try_tcache)
+{
+	size_t pageind, mapbits;
+	tcache_t *tcache;
+
+	assert(arena != NULL);
+	assert(chunk->arena == arena);
+	assert(ptr != NULL);
+	assert(CHUNK_ADDR2BASE(ptr) != ptr);
+
+	pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+	mapbits = arena_mapbits_get(chunk, pageind);
+	assert(arena_mapbits_allocated_get(chunk, pageind) != 0);
+	if ((mapbits & CHUNK_MAP_LARGE) == 0) {
+		/* Small allocation. */
+		if (try_tcache && (tcache = tcache_get(false)) != NULL) {
+			size_t binind;
+
+			binind = arena_ptr_small_binind_get(ptr, mapbits);
+			tcache_dalloc_small(tcache, ptr, binind);
+		} else
+			arena_dalloc_small(arena, chunk, ptr, pageind);
+	} else {
+		size_t size = arena_mapbits_large_size_get(chunk, pageind);
+
+		assert(((uintptr_t)ptr & PAGE_MASK) == 0);
+
+		if (try_tcache && size <= tcache_maxclass && (tcache =
+		    tcache_get(false)) != NULL) {
+			tcache_dalloc_large(tcache, ptr, size);
+		} else
+			arena_dalloc_large(arena, chunk, ptr);
+	}
+}
+#  endif /* JEMALLOC_ARENA_INLINE_B */
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/internal/atomic.h b/extra/jemalloc/include/jemalloc/internal/atomic.h
new file mode 100644
index 00000000000..11a7b47fe0f
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/atomic.h
@@ -0,0 +1,304 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+#define	atomic_read_uint64(p)	atomic_add_uint64(p, 0)
+#define	atomic_read_uint32(p)	atomic_add_uint32(p, 0)
+#define	atomic_read_z(p)	atomic_add_z(p, 0)
+#define	atomic_read_u(p)	atomic_add_u(p, 0)
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+uint64_t	atomic_add_uint64(uint64_t *p, uint64_t x);
+uint64_t	atomic_sub_uint64(uint64_t *p, uint64_t x);
+uint32_t	atomic_add_uint32(uint32_t *p, uint32_t x);
+uint32_t	atomic_sub_uint32(uint32_t *p, uint32_t x);
+size_t	atomic_add_z(size_t *p, size_t x);
+size_t	atomic_sub_z(size_t *p, size_t x);
+unsigned	atomic_add_u(unsigned *p, unsigned x);
+unsigned	atomic_sub_u(unsigned *p, unsigned x);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_ATOMIC_C_))
+/******************************************************************************/
+/* 64-bit operations. */
+#if (LG_SIZEOF_PTR == 3 || LG_SIZEOF_INT == 3)
+#  ifdef __GCC_HAVE_SYNC_COMPARE_AND_SWAP_8
+JEMALLOC_INLINE uint64_t
+atomic_add_uint64(uint64_t *p, uint64_t x)
+{
+
+	return (__sync_add_and_fetch(p, x));
+}
+
+JEMALLOC_INLINE uint64_t
+atomic_sub_uint64(uint64_t *p, uint64_t x)
+{
+
+	return (__sync_sub_and_fetch(p, x));
+}
+#elif (defined(_MSC_VER))
+JEMALLOC_INLINE uint64_t
+atomic_add_uint64(uint64_t *p, uint64_t x)
+{
+
+	return (InterlockedExchangeAdd64(p, x));
+}
+
+JEMALLOC_INLINE uint64_t
+atomic_sub_uint64(uint64_t *p, uint64_t x)
+{
+
+	return (InterlockedExchangeAdd64(p, -((int64_t)x)));
+}
+#elif (defined(JEMALLOC_OSATOMIC))
+JEMALLOC_INLINE uint64_t
+atomic_add_uint64(uint64_t *p, uint64_t x)
+{
+
+	return (OSAtomicAdd64((int64_t)x, (int64_t *)p));
+}
+
+JEMALLOC_INLINE uint64_t
+atomic_sub_uint64(uint64_t *p, uint64_t x)
+{
+
+	return (OSAtomicAdd64(-((int64_t)x), (int64_t *)p));
+}
+#  elif (defined(__amd64__) || defined(__x86_64__))
+JEMALLOC_INLINE uint64_t
+atomic_add_uint64(uint64_t *p, uint64_t x)
+{
+
+	asm volatile (
+	    "lock; xaddq %0, %1;"
+	    : "+r" (x), "=m" (*p) /* Outputs. */
+	    : "m" (*p) /* Inputs. */
+	    );
+
+	return (x);
+}
+
+JEMALLOC_INLINE uint64_t
+atomic_sub_uint64(uint64_t *p, uint64_t x)
+{
+
+	x = (uint64_t)(-(int64_t)x);
+	asm volatile (
+	    "lock; xaddq %0, %1;"
+	    : "+r" (x), "=m" (*p) /* Outputs. */
+	    : "m" (*p) /* Inputs. */
+	    );
+
+	return (x);
+}
+#  elif (defined(JEMALLOC_ATOMIC9))
+JEMALLOC_INLINE uint64_t
+atomic_add_uint64(uint64_t *p, uint64_t x)
+{
+
+	/*
+	 * atomic_fetchadd_64() doesn't exist, but we only ever use this
+	 * function on LP64 systems, so atomic_fetchadd_long() will do.
+	 */
+	assert(sizeof(uint64_t) == sizeof(unsigned long));
+
+	return (atomic_fetchadd_long(p, (unsigned long)x) + x);
+}
+
+JEMALLOC_INLINE uint64_t
+atomic_sub_uint64(uint64_t *p, uint64_t x)
+{
+
+	assert(sizeof(uint64_t) == sizeof(unsigned long));
+
+	return (atomic_fetchadd_long(p, (unsigned long)(-(long)x)) - x);
+}
+#  elif (defined(JE_FORCE_SYNC_COMPARE_AND_SWAP_8))
+JEMALLOC_INLINE uint64_t
+atomic_add_uint64(uint64_t *p, uint64_t x)
+{
+
+	return (__sync_add_and_fetch(p, x));
+}
+
+JEMALLOC_INLINE uint64_t
+atomic_sub_uint64(uint64_t *p, uint64_t x)
+{
+
+	return (__sync_sub_and_fetch(p, x));
+}
+#  else
+#    error "Missing implementation for 64-bit atomic operations"
+#  endif
+#endif
+
+/******************************************************************************/
+/* 32-bit operations. */
+#ifdef __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4
+JEMALLOC_INLINE uint32_t
+atomic_add_uint32(uint32_t *p, uint32_t x)
+{
+
+	return (__sync_add_and_fetch(p, x));
+}
+
+JEMALLOC_INLINE uint32_t
+atomic_sub_uint32(uint32_t *p, uint32_t x)
+{
+
+	return (__sync_sub_and_fetch(p, x));
+}
+#elif (defined(_MSC_VER))
+JEMALLOC_INLINE uint32_t
+atomic_add_uint32(uint32_t *p, uint32_t x)
+{
+
+	return (InterlockedExchangeAdd(p, x));
+}
+
+JEMALLOC_INLINE uint32_t
+atomic_sub_uint32(uint32_t *p, uint32_t x)
+{
+
+	return (InterlockedExchangeAdd(p, -((int32_t)x)));
+}
+#elif (defined(JEMALLOC_OSATOMIC))
+JEMALLOC_INLINE uint32_t
+atomic_add_uint32(uint32_t *p, uint32_t x)
+{
+
+	return (OSAtomicAdd32((int32_t)x, (int32_t *)p));
+}
+
+JEMALLOC_INLINE uint32_t
+atomic_sub_uint32(uint32_t *p, uint32_t x)
+{
+
+	return (OSAtomicAdd32(-((int32_t)x), (int32_t *)p));
+}
+#elif (defined(__i386__) || defined(__amd64__) || defined(__x86_64__))
+JEMALLOC_INLINE uint32_t
+atomic_add_uint32(uint32_t *p, uint32_t x)
+{
+
+	asm volatile (
+	    "lock; xaddl %0, %1;"
+	    : "+r" (x), "=m" (*p) /* Outputs. */
+	    : "m" (*p) /* Inputs. */
+	    );
+
+	return (x);
+}
+
+JEMALLOC_INLINE uint32_t
+atomic_sub_uint32(uint32_t *p, uint32_t x)
+{
+
+	x = (uint32_t)(-(int32_t)x);
+	asm volatile (
+	    "lock; xaddl %0, %1;"
+	    : "+r" (x), "=m" (*p) /* Outputs. */
+	    : "m" (*p) /* Inputs. */
+	    );
+
+	return (x);
+}
+#elif (defined(JEMALLOC_ATOMIC9))
+JEMALLOC_INLINE uint32_t
+atomic_add_uint32(uint32_t *p, uint32_t x)
+{
+
+	return (atomic_fetchadd_32(p, x) + x);
+}
+
+JEMALLOC_INLINE uint32_t
+atomic_sub_uint32(uint32_t *p, uint32_t x)
+{
+
+	return (atomic_fetchadd_32(p, (uint32_t)(-(int32_t)x)) - x);
+}
+#elif (defined(JE_FORCE_SYNC_COMPARE_AND_SWAP_4))
+JEMALLOC_INLINE uint32_t
+atomic_add_uint32(uint32_t *p, uint32_t x)
+{
+
+	return (__sync_add_and_fetch(p, x));
+}
+
+JEMALLOC_INLINE uint32_t
+atomic_sub_uint32(uint32_t *p, uint32_t x)
+{
+
+	return (__sync_sub_and_fetch(p, x));
+}
+#else
+#  error "Missing implementation for 32-bit atomic operations"
+#endif
+
+/******************************************************************************/
+/* size_t operations. */
+JEMALLOC_INLINE size_t
+atomic_add_z(size_t *p, size_t x)
+{
+
+#if (LG_SIZEOF_PTR == 3)
+	return ((size_t)atomic_add_uint64((uint64_t *)p, (uint64_t)x));
+#elif (LG_SIZEOF_PTR == 2)
+	return ((size_t)atomic_add_uint32((uint32_t *)p, (uint32_t)x));
+#endif
+}
+
+JEMALLOC_INLINE size_t
+atomic_sub_z(size_t *p, size_t x)
+{
+
+#if (LG_SIZEOF_PTR == 3)
+	return ((size_t)atomic_add_uint64((uint64_t *)p,
+	    (uint64_t)-((int64_t)x)));
+#elif (LG_SIZEOF_PTR == 2)
+	return ((size_t)atomic_add_uint32((uint32_t *)p,
+	    (uint32_t)-((int32_t)x)));
+#endif
+}
+
+/******************************************************************************/
+/* unsigned operations. */
+JEMALLOC_INLINE unsigned
+atomic_add_u(unsigned *p, unsigned x)
+{
+
+#if (LG_SIZEOF_INT == 3)
+	return ((unsigned)atomic_add_uint64((uint64_t *)p, (uint64_t)x));
+#elif (LG_SIZEOF_INT == 2)
+	return ((unsigned)atomic_add_uint32((uint32_t *)p, (uint32_t)x));
+#endif
+}
+
+JEMALLOC_INLINE unsigned
+atomic_sub_u(unsigned *p, unsigned x)
+{
+
+#if (LG_SIZEOF_INT == 3)
+	return ((unsigned)atomic_add_uint64((uint64_t *)p,
+	    (uint64_t)-((int64_t)x)));
+#elif (LG_SIZEOF_INT == 2)
+	return ((unsigned)atomic_add_uint32((uint32_t *)p,
+	    (uint32_t)-((int32_t)x)));
+#endif
+}
+/******************************************************************************/
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/internal/base.h b/extra/jemalloc/include/jemalloc/internal/base.h
new file mode 100644
index 00000000000..9cf75ffb0b3
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/base.h
@@ -0,0 +1,26 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+void	*base_alloc(size_t size);
+void	*base_calloc(size_t number, size_t size);
+extent_node_t *base_node_alloc(void);
+void	base_node_dealloc(extent_node_t *node);
+bool	base_boot(void);
+void	base_prefork(void);
+void	base_postfork_parent(void);
+void	base_postfork_child(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/internal/bitmap.h b/extra/jemalloc/include/jemalloc/internal/bitmap.h
new file mode 100644
index 00000000000..605ebac58c1
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/bitmap.h
@@ -0,0 +1,184 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+/* Maximum bitmap bit count is 2^LG_BITMAP_MAXBITS. */
+#define	LG_BITMAP_MAXBITS	LG_RUN_MAXREGS
+
+typedef struct bitmap_level_s bitmap_level_t;
+typedef struct bitmap_info_s bitmap_info_t;
+typedef unsigned long bitmap_t;
+#define	LG_SIZEOF_BITMAP	LG_SIZEOF_LONG
+
+/* Number of bits per group. */
+#define	LG_BITMAP_GROUP_NBITS		(LG_SIZEOF_BITMAP + 3)
+#define	BITMAP_GROUP_NBITS		(ZU(1) << LG_BITMAP_GROUP_NBITS)
+#define	BITMAP_GROUP_NBITS_MASK		(BITMAP_GROUP_NBITS-1)
+
+/* Maximum number of levels possible. */
+#define	BITMAP_MAX_LEVELS						\
+    (LG_BITMAP_MAXBITS / LG_SIZEOF_BITMAP)				\
+    + !!(LG_BITMAP_MAXBITS % LG_SIZEOF_BITMAP)
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+struct bitmap_level_s {
+	/* Offset of this level's groups within the array of groups. */
+	size_t group_offset;
+};
+
+struct bitmap_info_s {
+	/* Logical number of bits in bitmap (stored at bottom level). */
+	size_t nbits;
+
+	/* Number of levels necessary for nbits. */
+	unsigned nlevels;
+
+	/*
+	 * Only the first (nlevels+1) elements are used, and levels are ordered
+	 * bottom to top (e.g. the bottom level is stored in levels[0]).
+	 */
+	bitmap_level_t levels[BITMAP_MAX_LEVELS+1];
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+void	bitmap_info_init(bitmap_info_t *binfo, size_t nbits);
+size_t	bitmap_info_ngroups(const bitmap_info_t *binfo);
+size_t	bitmap_size(size_t nbits);
+void	bitmap_init(bitmap_t *bitmap, const bitmap_info_t *binfo);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+bool	bitmap_full(bitmap_t *bitmap, const bitmap_info_t *binfo);
+bool	bitmap_get(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit);
+void	bitmap_set(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit);
+size_t	bitmap_sfu(bitmap_t *bitmap, const bitmap_info_t *binfo);
+void	bitmap_unset(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_BITMAP_C_))
+JEMALLOC_INLINE bool
+bitmap_full(bitmap_t *bitmap, const bitmap_info_t *binfo)
+{
+	unsigned rgoff = binfo->levels[binfo->nlevels].group_offset - 1;
+	bitmap_t rg = bitmap[rgoff];
+	/* The bitmap is full iff the root group is 0. */
+	return (rg == 0);
+}
+
+JEMALLOC_INLINE bool
+bitmap_get(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit)
+{
+	size_t goff;
+	bitmap_t g;
+
+	assert(bit < binfo->nbits);
+	goff = bit >> LG_BITMAP_GROUP_NBITS;
+	g = bitmap[goff];
+	return (!(g & (1LU << (bit & BITMAP_GROUP_NBITS_MASK))));
+}
+
+JEMALLOC_INLINE void
+bitmap_set(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit)
+{
+	size_t goff;
+	bitmap_t *gp;
+	bitmap_t g;
+
+	assert(bit < binfo->nbits);
+	assert(bitmap_get(bitmap, binfo, bit) == false);
+	goff = bit >> LG_BITMAP_GROUP_NBITS;
+	gp = &bitmap[goff];
+	g = *gp;
+	assert(g & (1LU << (bit & BITMAP_GROUP_NBITS_MASK)));
+	g ^= 1LU << (bit & BITMAP_GROUP_NBITS_MASK);
+	*gp = g;
+	assert(bitmap_get(bitmap, binfo, bit));
+	/* Propagate group state transitions up the tree. */
+	if (g == 0) {
+		unsigned i;
+		for (i = 1; i < binfo->nlevels; i++) {
+			bit = goff;
+			goff = bit >> LG_BITMAP_GROUP_NBITS;
+			gp = &bitmap[binfo->levels[i].group_offset + goff];
+			g = *gp;
+			assert(g & (1LU << (bit & BITMAP_GROUP_NBITS_MASK)));
+			g ^= 1LU << (bit & BITMAP_GROUP_NBITS_MASK);
+			*gp = g;
+			if (g != 0)
+				break;
+		}
+	}
+}
+
+/* sfu: set first unset. */
+JEMALLOC_INLINE size_t
+bitmap_sfu(bitmap_t *bitmap, const bitmap_info_t *binfo)
+{
+	size_t bit;
+	bitmap_t g;
+	unsigned i;
+
+	assert(bitmap_full(bitmap, binfo) == false);
+
+	i = binfo->nlevels - 1;
+	g = bitmap[binfo->levels[i].group_offset];
+	bit = ffsl(g) - 1;
+	while (i > 0) {
+		i--;
+		g = bitmap[binfo->levels[i].group_offset + bit];
+		bit = (bit << LG_BITMAP_GROUP_NBITS) + (ffsl(g) - 1);
+	}
+
+	bitmap_set(bitmap, binfo, bit);
+	return (bit);
+}
+
+JEMALLOC_INLINE void
+bitmap_unset(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit)
+{
+	size_t goff;
+	bitmap_t *gp;
+	bitmap_t g;
+	bool propagate;
+
+	assert(bit < binfo->nbits);
+	assert(bitmap_get(bitmap, binfo, bit));
+	goff = bit >> LG_BITMAP_GROUP_NBITS;
+	gp = &bitmap[goff];
+	g = *gp;
+	propagate = (g == 0);
+	assert((g & (1LU << (bit & BITMAP_GROUP_NBITS_MASK))) == 0);
+	g ^= 1LU << (bit & BITMAP_GROUP_NBITS_MASK);
+	*gp = g;
+	assert(bitmap_get(bitmap, binfo, bit) == false);
+	/* Propagate group state transitions up the tree. */
+	if (propagate) {
+		unsigned i;
+		for (i = 1; i < binfo->nlevels; i++) {
+			bit = goff;
+			goff = bit >> LG_BITMAP_GROUP_NBITS;
+			gp = &bitmap[binfo->levels[i].group_offset + goff];
+			g = *gp;
+			propagate = (g == 0);
+			assert((g & (1LU << (bit & BITMAP_GROUP_NBITS_MASK)))
+			    == 0);
+			g ^= 1LU << (bit & BITMAP_GROUP_NBITS_MASK);
+			*gp = g;
+			if (propagate == false)
+				break;
+		}
+	}
+}
+
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/internal/chunk.h b/extra/jemalloc/include/jemalloc/internal/chunk.h
new file mode 100644
index 00000000000..87d8700dac8
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/chunk.h
@@ -0,0 +1,63 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+/*
+ * Size and alignment of memory chunks that are allocated by the OS's virtual
+ * memory system.
+ */
+#define	LG_CHUNK_DEFAULT	22
+
+/* Return the chunk address for allocation address a. */
+#define	CHUNK_ADDR2BASE(a)						\
+	((void *)((uintptr_t)(a) & ~chunksize_mask))
+
+/* Return the chunk offset of address a. */
+#define	CHUNK_ADDR2OFFSET(a)						\
+	((size_t)((uintptr_t)(a) & chunksize_mask))
+
+/* Return the smallest chunk multiple that is >= s. */
+#define	CHUNK_CEILING(s)						\
+	(((s) + chunksize_mask) & ~chunksize_mask)
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+extern size_t		opt_lg_chunk;
+extern const char	*opt_dss;
+
+/* Protects stats_chunks; currently not used for any other purpose. */
+extern malloc_mutex_t	chunks_mtx;
+/* Chunk statistics. */
+extern chunk_stats_t	stats_chunks;
+
+extern rtree_t		*chunks_rtree;
+
+extern size_t		chunksize;
+extern size_t		chunksize_mask; /* (chunksize - 1). */
+extern size_t		chunk_npages;
+extern size_t		map_bias; /* Number of arena chunk header pages. */
+extern size_t		arena_maxclass; /* Max size class for arenas. */
+
+void	*chunk_alloc(size_t size, size_t alignment, bool base, bool *zero,
+    dss_prec_t dss_prec);
+void	chunk_unmap(void *chunk, size_t size);
+void	chunk_dealloc(void *chunk, size_t size, bool unmap);
+bool	chunk_boot(void);
+void	chunk_prefork(void);
+void	chunk_postfork_parent(void);
+void	chunk_postfork_child(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
+
+#include "jemalloc/internal/chunk_dss.h"
+#include "jemalloc/internal/chunk_mmap.h"
diff --git a/extra/jemalloc/include/jemalloc/internal/chunk_dss.h b/extra/jemalloc/include/jemalloc/internal/chunk_dss.h
new file mode 100644
index 00000000000..6585f071bbe
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/chunk_dss.h
@@ -0,0 +1,38 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+typedef enum {
+	dss_prec_disabled  = 0,
+	dss_prec_primary   = 1,
+	dss_prec_secondary = 2,
+
+	dss_prec_limit     = 3
+} dss_prec_t ;
+#define	DSS_PREC_DEFAULT	dss_prec_secondary
+#define	DSS_DEFAULT		"secondary"
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+extern const char *dss_prec_names[];
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+dss_prec_t	chunk_dss_prec_get(void);
+bool	chunk_dss_prec_set(dss_prec_t dss_prec);
+void	*chunk_alloc_dss(size_t size, size_t alignment, bool *zero);
+bool	chunk_in_dss(void *chunk);
+bool	chunk_dss_boot(void);
+void	chunk_dss_prefork(void);
+void	chunk_dss_postfork_parent(void);
+void	chunk_dss_postfork_child(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/internal/chunk_mmap.h b/extra/jemalloc/include/jemalloc/internal/chunk_mmap.h
new file mode 100644
index 00000000000..f24abac7538
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/chunk_mmap.h
@@ -0,0 +1,22 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+bool	pages_purge(void *addr, size_t length);
+
+void	*chunk_alloc_mmap(size_t size, size_t alignment, bool *zero);
+bool	chunk_dealloc_mmap(void *chunk, size_t size);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/internal/ckh.h b/extra/jemalloc/include/jemalloc/internal/ckh.h
new file mode 100644
index 00000000000..50c39ed9581
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/ckh.h
@@ -0,0 +1,88 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+typedef struct ckh_s ckh_t;
+typedef struct ckhc_s ckhc_t;
+
+/* Typedefs to allow easy function pointer passing. */
+typedef void ckh_hash_t (const void *, size_t[2]);
+typedef bool ckh_keycomp_t (const void *, const void *);
+
+/* Maintain counters used to get an idea of performance. */
+/* #define	CKH_COUNT */
+/* Print counter values in ckh_delete() (requires CKH_COUNT). */
+/* #define	CKH_VERBOSE */
+
+/*
+ * There are 2^LG_CKH_BUCKET_CELLS cells in each hash table bucket.  Try to fit
+ * one bucket per L1 cache line.
+ */
+#define LG_CKH_BUCKET_CELLS (LG_CACHELINE - LG_SIZEOF_PTR - 1)
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+/* Hash table cell. */
+struct ckhc_s {
+	const void	*key;
+	const void	*data;
+};
+
+struct ckh_s {
+#ifdef CKH_COUNT
+	/* Counters used to get an idea of performance. */
+	uint64_t	ngrows;
+	uint64_t	nshrinks;
+	uint64_t	nshrinkfails;
+	uint64_t	ninserts;
+	uint64_t	nrelocs;
+#endif
+
+	/* Used for pseudo-random number generation. */
+#define	CKH_A		1103515241
+#define	CKH_C		12347
+	uint32_t	prng_state;
+
+	/* Total number of items. */
+	size_t		count;
+
+	/*
+	 * Minimum and current number of hash table buckets.  There are
+	 * 2^LG_CKH_BUCKET_CELLS cells per bucket.
+	 */
+	unsigned	lg_minbuckets;
+	unsigned	lg_curbuckets;
+
+	/* Hash and comparison functions. */
+	ckh_hash_t	*hash;
+	ckh_keycomp_t	*keycomp;
+
+	/* Hash table with 2^lg_curbuckets buckets. */
+	ckhc_t		*tab;
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+bool	ckh_new(ckh_t *ckh, size_t minitems, ckh_hash_t *hash,
+    ckh_keycomp_t *keycomp);
+void	ckh_delete(ckh_t *ckh);
+size_t	ckh_count(ckh_t *ckh);
+bool	ckh_iter(ckh_t *ckh, size_t *tabind, void **key, void **data);
+bool	ckh_insert(ckh_t *ckh, const void *key, const void *data);
+bool	ckh_remove(ckh_t *ckh, const void *searchkey, void **key,
+    void **data);
+bool	ckh_search(ckh_t *ckh, const void *seachkey, void **key, void **data);
+void	ckh_string_hash(const void *key, size_t r_hash[2]);
+bool	ckh_string_keycomp(const void *k1, const void *k2);
+void	ckh_pointer_hash(const void *key, size_t r_hash[2]);
+bool	ckh_pointer_keycomp(const void *k1, const void *k2);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/internal/ctl.h b/extra/jemalloc/include/jemalloc/internal/ctl.h
new file mode 100644
index 00000000000..0ffecc5f2a2
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/ctl.h
@@ -0,0 +1,117 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+typedef struct ctl_node_s ctl_node_t;
+typedef struct ctl_named_node_s ctl_named_node_t;
+typedef struct ctl_indexed_node_s ctl_indexed_node_t;
+typedef struct ctl_arena_stats_s ctl_arena_stats_t;
+typedef struct ctl_stats_s ctl_stats_t;
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+struct ctl_node_s {
+	bool			named;
+};
+
+struct ctl_named_node_s {
+	struct ctl_node_s	node;
+	const char		*name;
+	/* If (nchildren == 0), this is a terminal node. */
+	unsigned		nchildren;
+	const			ctl_node_t *children;
+	int			(*ctl)(const size_t *, size_t, void *, size_t *,
+	    void *, size_t);
+};
+
+struct ctl_indexed_node_s {
+	struct ctl_node_s	node;
+	const ctl_named_node_t	*(*index)(const size_t *, size_t, size_t);
+};
+
+struct ctl_arena_stats_s {
+	bool			initialized;
+	unsigned		nthreads;
+	const char		*dss;
+	size_t			pactive;
+	size_t			pdirty;
+	arena_stats_t		astats;
+
+	/* Aggregate stats for small size classes, based on bin stats. */
+	size_t			allocated_small;
+	uint64_t		nmalloc_small;
+	uint64_t		ndalloc_small;
+	uint64_t		nrequests_small;
+
+	malloc_bin_stats_t	bstats[NBINS];
+	malloc_large_stats_t	*lstats;	/* nlclasses elements. */
+};
+
+struct ctl_stats_s {
+	size_t			allocated;
+	size_t			active;
+	size_t			mapped;
+	struct {
+		size_t		current;	/* stats_chunks.curchunks */
+		uint64_t	total;		/* stats_chunks.nchunks */
+		size_t		high;		/* stats_chunks.highchunks */
+	} chunks;
+	struct {
+		size_t		allocated;	/* huge_allocated */
+		uint64_t	nmalloc;	/* huge_nmalloc */
+		uint64_t	ndalloc;	/* huge_ndalloc */
+	} huge;
+	unsigned		narenas;
+	ctl_arena_stats_t	*arenas;	/* (narenas + 1) elements. */
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+int	ctl_byname(const char *name, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen);
+int	ctl_nametomib(const char *name, size_t *mibp, size_t *miblenp);
+
+int	ctl_bymib(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen);
+bool	ctl_boot(void);
+void	ctl_prefork(void);
+void	ctl_postfork_parent(void);
+void	ctl_postfork_child(void);
+
+#define	xmallctl(name, oldp, oldlenp, newp, newlen) do {		\
+	if (je_mallctl(name, oldp, oldlenp, newp, newlen)		\
+	    != 0) {							\
+		malloc_printf(						\
+		    "<jemalloc>: Failure in xmallctl(\"%s\", ...)\n",	\
+		    name);						\
+		abort();						\
+	}								\
+} while (0)
+
+#define	xmallctlnametomib(name, mibp, miblenp) do {			\
+	if (je_mallctlnametomib(name, mibp, miblenp) != 0) {		\
+		malloc_printf("<jemalloc>: Failure in "			\
+		    "xmallctlnametomib(\"%s\", ...)\n", name);		\
+		abort();						\
+	}								\
+} while (0)
+
+#define	xmallctlbymib(mib, miblen, oldp, oldlenp, newp, newlen) do {	\
+	if (je_mallctlbymib(mib, miblen, oldp, oldlenp, newp,		\
+	    newlen) != 0) {						\
+		malloc_write(						\
+		    "<jemalloc>: Failure in xmallctlbymib()\n");	\
+		abort();						\
+	}								\
+} while (0)
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
+
diff --git a/extra/jemalloc/include/jemalloc/internal/extent.h b/extra/jemalloc/include/jemalloc/internal/extent.h
new file mode 100644
index 00000000000..ba95ca816bd
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/extent.h
@@ -0,0 +1,46 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+typedef struct extent_node_s extent_node_t;
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+/* Tree of extents. */
+struct extent_node_s {
+	/* Linkage for the size/address-ordered tree. */
+	rb_node(extent_node_t)	link_szad;
+
+	/* Linkage for the address-ordered tree. */
+	rb_node(extent_node_t)	link_ad;
+
+	/* Profile counters, used for huge objects. */
+	prof_ctx_t		*prof_ctx;
+
+	/* Pointer to the extent that this tree node is responsible for. */
+	void			*addr;
+
+	/* Total region size. */
+	size_t			size;
+
+	/* True if zero-filled; used by chunk recycling code. */
+	bool			zeroed;
+};
+typedef rb_tree(extent_node_t) extent_tree_t;
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+rb_proto(, extent_tree_szad_, extent_tree_t, extent_node_t)
+
+rb_proto(, extent_tree_ad_, extent_tree_t, extent_node_t)
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
+
diff --git a/extra/jemalloc/include/jemalloc/internal/hash.h b/extra/jemalloc/include/jemalloc/internal/hash.h
new file mode 100644
index 00000000000..56ecc793b36
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/hash.h
@@ -0,0 +1,331 @@
+/*
+ * The following hash function is based on MurmurHash3, placed into the public
+ * domain by Austin Appleby.  See http://code.google.com/p/smhasher/ for
+ * details.
+ */
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+void	hash(const void *key, size_t len, const uint32_t seed,
+    size_t r_hash[2]);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_HASH_C_))
+/******************************************************************************/
+/* Internal implementation. */
+JEMALLOC_INLINE uint32_t
+hash_rotl_32(uint32_t x, int8_t r)
+{
+
+	return (x << r) | (x >> (32 - r));
+}
+
+JEMALLOC_INLINE uint64_t
+hash_rotl_64(uint64_t x, int8_t r)
+{
+	return (x << r) | (x >> (64 - r));
+}
+
+JEMALLOC_INLINE uint32_t
+hash_get_block_32(const uint32_t *p, int i)
+{
+
+	return p[i];
+}
+
+JEMALLOC_INLINE uint64_t
+hash_get_block_64(const uint64_t *p, int i)
+{
+
+	return p[i];
+}
+
+JEMALLOC_INLINE uint32_t
+hash_fmix_32(uint32_t h)
+{
+
+	h ^= h >> 16;
+	h *= 0x85ebca6b;
+	h ^= h >> 13;
+	h *= 0xc2b2ae35;
+	h ^= h >> 16;
+
+	return h;
+}
+
+JEMALLOC_INLINE uint64_t
+hash_fmix_64(uint64_t k)
+{
+
+	k ^= k >> 33;
+	k *= QU(0xff51afd7ed558ccdLLU);
+	k ^= k >> 33;
+	k *= QU(0xc4ceb9fe1a85ec53LLU);
+	k ^= k >> 33;
+
+	return k;
+}
+
+JEMALLOC_INLINE uint32_t
+hash_x86_32(const void *key, int len, uint32_t seed)
+{
+	const uint8_t *data = (const uint8_t *) key;
+	const int nblocks = len / 4;
+
+	uint32_t h1 = seed;
+
+	const uint32_t c1 = 0xcc9e2d51;
+	const uint32_t c2 = 0x1b873593;
+
+	/* body */
+	{
+		const uint32_t *blocks = (const uint32_t *) (data + nblocks*4);
+		int i;
+
+		for (i = -nblocks; i; i++) {
+			uint32_t k1 = hash_get_block_32(blocks, i);
+
+			k1 *= c1;
+			k1 = hash_rotl_32(k1, 15);
+			k1 *= c2;
+
+			h1 ^= k1;
+			h1 = hash_rotl_32(h1, 13);
+			h1 = h1*5 + 0xe6546b64;
+		}
+	}
+
+	/* tail */
+	{
+		const uint8_t *tail = (const uint8_t *) (data + nblocks*4);
+
+		uint32_t k1 = 0;
+
+		switch (len & 3) {
+		case 3: k1 ^= tail[2] << 16;
+		case 2: k1 ^= tail[1] << 8;
+		case 1: k1 ^= tail[0]; k1 *= c1; k1 = hash_rotl_32(k1, 15);
+			k1 *= c2; h1 ^= k1;
+		}
+	}
+
+	/* finalization */
+	h1 ^= len;
+
+	h1 = hash_fmix_32(h1);
+
+	return h1;
+}
+
+UNUSED JEMALLOC_INLINE void
+hash_x86_128(const void *key, const int len, uint32_t seed,
+  uint64_t r_out[2])
+{
+	const uint8_t * data = (const uint8_t *) key;
+	const int nblocks = len / 16;
+
+	uint32_t h1 = seed;
+	uint32_t h2 = seed;
+	uint32_t h3 = seed;
+	uint32_t h4 = seed;
+
+	const uint32_t c1 = 0x239b961b;
+	const uint32_t c2 = 0xab0e9789;
+	const uint32_t c3 = 0x38b34ae5;
+	const uint32_t c4 = 0xa1e38b93;
+
+	/* body */
+	{
+		const uint32_t *blocks = (const uint32_t *) (data + nblocks*16);
+		int i;
+
+		for (i = -nblocks; i; i++) {
+			uint32_t k1 = hash_get_block_32(blocks, i*4 + 0);
+			uint32_t k2 = hash_get_block_32(blocks, i*4 + 1);
+			uint32_t k3 = hash_get_block_32(blocks, i*4 + 2);
+			uint32_t k4 = hash_get_block_32(blocks, i*4 + 3);
+
+			k1 *= c1; k1 = hash_rotl_32(k1, 15); k1 *= c2; h1 ^= k1;
+
+			h1 = hash_rotl_32(h1, 19); h1 += h2;
+			h1 = h1*5 + 0x561ccd1b;
+
+			k2 *= c2; k2 = hash_rotl_32(k2, 16); k2 *= c3; h2 ^= k2;
+
+			h2 = hash_rotl_32(h2, 17); h2 += h3;
+			h2 = h2*5 + 0x0bcaa747;
+
+			k3 *= c3; k3 = hash_rotl_32(k3, 17); k3 *= c4; h3 ^= k3;
+
+			h3 = hash_rotl_32(h3, 15); h3 += h4;
+			h3 = h3*5 + 0x96cd1c35;
+
+			k4 *= c4; k4 = hash_rotl_32(k4, 18); k4 *= c1; h4 ^= k4;
+
+			h4 = hash_rotl_32(h4, 13); h4 += h1;
+			h4 = h4*5 + 0x32ac3b17;
+		}
+	}
+
+	/* tail */
+	{
+		const uint8_t *tail = (const uint8_t *) (data + nblocks*16);
+		uint32_t k1 = 0;
+		uint32_t k2 = 0;
+		uint32_t k3 = 0;
+		uint32_t k4 = 0;
+
+		switch (len & 15) {
+		case 15: k4 ^= tail[14] << 16;
+		case 14: k4 ^= tail[13] << 8;
+		case 13: k4 ^= tail[12] << 0;
+			k4 *= c4; k4 = hash_rotl_32(k4, 18); k4 *= c1; h4 ^= k4;
+
+		case 12: k3 ^= tail[11] << 24;
+		case 11: k3 ^= tail[10] << 16;
+		case 10: k3 ^= tail[ 9] << 8;
+		case  9: k3 ^= tail[ 8] << 0;
+		     k3 *= c3; k3 = hash_rotl_32(k3, 17); k3 *= c4; h3 ^= k3;
+
+		case  8: k2 ^= tail[ 7] << 24;
+		case  7: k2 ^= tail[ 6] << 16;
+		case  6: k2 ^= tail[ 5] << 8;
+		case  5: k2 ^= tail[ 4] << 0;
+			k2 *= c2; k2 = hash_rotl_32(k2, 16); k2 *= c3; h2 ^= k2;
+
+		case  4: k1 ^= tail[ 3] << 24;
+		case  3: k1 ^= tail[ 2] << 16;
+		case  2: k1 ^= tail[ 1] << 8;
+		case  1: k1 ^= tail[ 0] << 0;
+			k1 *= c1; k1 = hash_rotl_32(k1, 15); k1 *= c2; h1 ^= k1;
+		}
+	}
+
+	/* finalization */
+	h1 ^= len; h2 ^= len; h3 ^= len; h4 ^= len;
+
+	h1 += h2; h1 += h3; h1 += h4;
+	h2 += h1; h3 += h1; h4 += h1;
+
+	h1 = hash_fmix_32(h1);
+	h2 = hash_fmix_32(h2);
+	h3 = hash_fmix_32(h3);
+	h4 = hash_fmix_32(h4);
+
+	h1 += h2; h1 += h3; h1 += h4;
+	h2 += h1; h3 += h1; h4 += h1;
+
+	r_out[0] = (((uint64_t) h2) << 32) | h1;
+	r_out[1] = (((uint64_t) h4) << 32) | h3;
+}
+
+UNUSED JEMALLOC_INLINE void
+hash_x64_128(const void *key, const int len, const uint32_t seed,
+  uint64_t r_out[2])
+{
+	const uint8_t *data = (const uint8_t *) key;
+	const int nblocks = len / 16;
+
+	uint64_t h1 = seed;
+	uint64_t h2 = seed;
+
+	const uint64_t c1 = QU(0x87c37b91114253d5LLU);
+	const uint64_t c2 = QU(0x4cf5ad432745937fLLU);
+
+	/* body */
+	{
+		const uint64_t *blocks = (const uint64_t *) (data);
+		int i;
+
+		for (i = 0; i < nblocks; i++) {
+			uint64_t k1 = hash_get_block_64(blocks, i*2 + 0);
+			uint64_t k2 = hash_get_block_64(blocks, i*2 + 1);
+
+			k1 *= c1; k1 = hash_rotl_64(k1, 31); k1 *= c2; h1 ^= k1;
+
+			h1 = hash_rotl_64(h1, 27); h1 += h2;
+			h1 = h1*5 + 0x52dce729;
+
+			k2 *= c2; k2 = hash_rotl_64(k2, 33); k2 *= c1; h2 ^= k2;
+
+			h2 = hash_rotl_64(h2, 31); h2 += h1;
+			h2 = h2*5 + 0x38495ab5;
+		}
+	}
+
+	/* tail */
+	{
+		const uint8_t *tail = (const uint8_t*)(data + nblocks*16);
+		uint64_t k1 = 0;
+		uint64_t k2 = 0;
+
+		switch (len & 15) {
+		case 15: k2 ^= ((uint64_t)(tail[14])) << 48;
+		case 14: k2 ^= ((uint64_t)(tail[13])) << 40;
+		case 13: k2 ^= ((uint64_t)(tail[12])) << 32;
+		case 12: k2 ^= ((uint64_t)(tail[11])) << 24;
+		case 11: k2 ^= ((uint64_t)(tail[10])) << 16;
+		case 10: k2 ^= ((uint64_t)(tail[ 9])) << 8;
+		case  9: k2 ^= ((uint64_t)(tail[ 8])) << 0;
+			k2 *= c2; k2 = hash_rotl_64(k2, 33); k2 *= c1; h2 ^= k2;
+
+		case  8: k1 ^= ((uint64_t)(tail[ 7])) << 56;
+		case  7: k1 ^= ((uint64_t)(tail[ 6])) << 48;
+		case  6: k1 ^= ((uint64_t)(tail[ 5])) << 40;
+		case  5: k1 ^= ((uint64_t)(tail[ 4])) << 32;
+		case  4: k1 ^= ((uint64_t)(tail[ 3])) << 24;
+		case  3: k1 ^= ((uint64_t)(tail[ 2])) << 16;
+		case  2: k1 ^= ((uint64_t)(tail[ 1])) << 8;
+		case  1: k1 ^= ((uint64_t)(tail[ 0])) << 0;
+			k1 *= c1; k1 = hash_rotl_64(k1, 31); k1 *= c2; h1 ^= k1;
+		}
+	}
+
+	/* finalization */
+	h1 ^= len; h2 ^= len;
+
+	h1 += h2;
+	h2 += h1;
+
+	h1 = hash_fmix_64(h1);
+	h2 = hash_fmix_64(h2);
+
+	h1 += h2;
+	h2 += h1;
+
+	r_out[0] = h1;
+	r_out[1] = h2;
+}
+
+
+/******************************************************************************/
+/* API. */
+JEMALLOC_INLINE void
+hash(const void *key, size_t len, const uint32_t seed, size_t r_hash[2])
+{
+#if (LG_SIZEOF_PTR == 3)
+	hash_x64_128(key, len, seed, (uint64_t *)r_hash);
+#else
+	uint64_t hashes[2];
+	hash_x86_128(key, len, seed, hashes);
+	r_hash[0] = (size_t)hashes[0];
+	r_hash[1] = (size_t)hashes[1];
+#endif
+}
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/internal/huge.h b/extra/jemalloc/include/jemalloc/internal/huge.h
new file mode 100644
index 00000000000..d987d370767
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/huge.h
@@ -0,0 +1,40 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+/* Huge allocation statistics. */
+extern uint64_t		huge_nmalloc;
+extern uint64_t		huge_ndalloc;
+extern size_t		huge_allocated;
+
+/* Protects chunk-related data structures. */
+extern malloc_mutex_t	huge_mtx;
+
+void	*huge_malloc(size_t size, bool zero);
+void	*huge_palloc(size_t size, size_t alignment, bool zero);
+void	*huge_ralloc_no_move(void *ptr, size_t oldsize, size_t size,
+    size_t extra);
+void	*huge_ralloc(void *ptr, size_t oldsize, size_t size, size_t extra,
+    size_t alignment, bool zero, bool try_tcache_dalloc);
+void	huge_dalloc(void *ptr, bool unmap);
+size_t	huge_salloc(const void *ptr);
+prof_ctx_t	*huge_prof_ctx_get(const void *ptr);
+void	huge_prof_ctx_set(const void *ptr, prof_ctx_t *ctx);
+bool	huge_boot(void);
+void	huge_prefork(void);
+void	huge_postfork_parent(void);
+void	huge_postfork_child(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/internal/jemalloc_internal.h.in b/extra/jemalloc/include/jemalloc/internal/jemalloc_internal.h.in
new file mode 100644
index 00000000000..50d84cabf69
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/jemalloc_internal.h.in
@@ -0,0 +1,1026 @@
+#ifndef JEMALLOC_INTERNAL_H
+#define JEMALLOC_INTERNAL_H
+#include <math.h>
+#ifdef _WIN32
+#  include <windows.h>
+#  define ENOENT ERROR_PATH_NOT_FOUND
+#  define EINVAL ERROR_BAD_ARGUMENTS
+#  define EAGAIN ERROR_OUTOFMEMORY
+#  define EPERM  ERROR_WRITE_FAULT
+#  define EFAULT ERROR_INVALID_ADDRESS
+#  define ENOMEM ERROR_NOT_ENOUGH_MEMORY
+#  undef ERANGE
+#  define ERANGE ERROR_INVALID_DATA
+#else
+#  include <sys/param.h>
+#  include <sys/mman.h>
+#  include <sys/syscall.h>
+#  if !defined(SYS_write) && defined(__NR_write)
+#    define SYS_write __NR_write
+#  endif
+#  include <sys/uio.h>
+#  include <pthread.h>
+#  include <errno.h>
+#endif
+#include <sys/types.h>
+
+#include <limits.h>
+#ifndef SIZE_T_MAX
+#  define SIZE_T_MAX	SIZE_MAX
+#endif
+#include <stdarg.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <stddef.h>
+#ifndef offsetof
+#  define offsetof(type, member)	((size_t)&(((type *)NULL)->member))
+#endif
+#include <inttypes.h>
+#include <string.h>
+#include <strings.h>
+#include <ctype.h>
+#ifdef _MSC_VER
+#  include <io.h>
+typedef intptr_t ssize_t;
+#  define PATH_MAX 1024
+#  define STDERR_FILENO 2
+#  define __func__ __FUNCTION__
+/* Disable warnings about deprecated system functions */
+#  pragma warning(disable: 4996)
+#else
+#  include <unistd.h>
+#endif
+#include <fcntl.h>
+
+#define	JEMALLOC_NO_DEMANGLE
+#include "../jemalloc@install_suffix@.h"
+
+#ifdef JEMALLOC_UTRACE
+#include <sys/ktrace.h>
+#endif
+
+#ifdef JEMALLOC_VALGRIND
+#include <valgrind/valgrind.h>
+#include <valgrind/memcheck.h>
+#endif
+
+#include "jemalloc/internal/private_namespace.h"
+
+#ifdef JEMALLOC_CC_SILENCE
+#define	UNUSED JEMALLOC_ATTR(unused)
+#else
+#define	UNUSED
+#endif
+
+static const bool config_debug =
+#ifdef JEMALLOC_DEBUG
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_dss =
+#ifdef JEMALLOC_DSS
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_fill =
+#ifdef JEMALLOC_FILL
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_lazy_lock =
+#ifdef JEMALLOC_LAZY_LOCK
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_prof =
+#ifdef JEMALLOC_PROF
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_prof_libgcc =
+#ifdef JEMALLOC_PROF_LIBGCC
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_prof_libunwind =
+#ifdef JEMALLOC_PROF_LIBUNWIND
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_mremap =
+#ifdef JEMALLOC_MREMAP
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_munmap =
+#ifdef JEMALLOC_MUNMAP
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_stats =
+#ifdef JEMALLOC_STATS
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_tcache =
+#ifdef JEMALLOC_TCACHE
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_tls =
+#ifdef JEMALLOC_TLS
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_utrace =
+#ifdef JEMALLOC_UTRACE
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_valgrind =
+#ifdef JEMALLOC_VALGRIND
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_xmalloc =
+#ifdef JEMALLOC_XMALLOC
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_ivsalloc =
+#ifdef JEMALLOC_IVSALLOC
+    true
+#else
+    false
+#endif
+    ;
+
+#ifdef JEMALLOC_ATOMIC9
+#include <machine/atomic.h>
+#endif
+
+#if (defined(JEMALLOC_OSATOMIC) || defined(JEMALLOC_OSSPIN))
+#include <libkern/OSAtomic.h>
+#endif
+
+#ifdef JEMALLOC_ZONE
+#include <mach/mach_error.h>
+#include <mach/mach_init.h>
+#include <mach/vm_map.h>
+#include <malloc/malloc.h>
+#endif
+
+#define	RB_COMPACT
+#include "jemalloc/internal/rb.h"
+#include "jemalloc/internal/qr.h"
+#include "jemalloc/internal/ql.h"
+
+/*
+ * jemalloc can conceptually be broken into components (arena, tcache, etc.),
+ * but there are circular dependencies that cannot be broken without
+ * substantial performance degradation.  In order to reduce the effect on
+ * visual code flow, read the header files in multiple passes, with one of the
+ * following cpp variables defined during each pass:
+ *
+ *   JEMALLOC_H_TYPES   : Preprocessor-defined constants and psuedo-opaque data
+ *                        types.
+ *   JEMALLOC_H_STRUCTS : Data structures.
+ *   JEMALLOC_H_EXTERNS : Extern data declarations and function prototypes.
+ *   JEMALLOC_H_INLINES : Inline functions.
+ */
+/******************************************************************************/
+#define JEMALLOC_H_TYPES
+
+#define	ALLOCM_LG_ALIGN_MASK	((int)0x3f)
+
+#define	ZU(z)	((size_t)z)
+#define	QU(q)	((uint64_t)q)
+
+#ifndef __DECONST
+#  define	__DECONST(type, var)	((type)(uintptr_t)(const void *)(var))
+#endif
+
+#ifdef JEMALLOC_DEBUG
+   /* Disable inlining to make debugging easier. */
+#  define JEMALLOC_ALWAYS_INLINE
+#  define JEMALLOC_INLINE
+#  define inline
+#else
+#  define JEMALLOC_ENABLE_INLINE
+#  ifdef JEMALLOC_HAVE_ATTR
+#    define JEMALLOC_ALWAYS_INLINE \
+	 static inline JEMALLOC_ATTR(unused) JEMALLOC_ATTR(always_inline)
+#  else
+#    define JEMALLOC_ALWAYS_INLINE static inline
+#  endif
+#  define JEMALLOC_INLINE static inline
+#  ifdef _MSC_VER
+#    define inline _inline
+#  endif
+#endif
+
+/* Smallest size class to support. */
+#define	LG_TINY_MIN		3
+#define	TINY_MIN		(1U << LG_TINY_MIN)
+
+/*
+ * Minimum alignment of allocations is 2^LG_QUANTUM bytes (ignoring tiny size
+ * classes).
+ */
+#ifndef LG_QUANTUM
+#  if (defined(__i386__) || defined(_M_IX86))
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __ia64__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __alpha__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __sparc64__
+#    define LG_QUANTUM		4
+#  endif
+#  if (defined(__amd64__) || defined(__x86_64__) || defined(_M_X64))
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __arm__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __hppa__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __mips__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __powerpc__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __s390__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __SH4__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __tile__
+#    define LG_QUANTUM		4
+#  endif
+#  ifndef LG_QUANTUM
+#    error "No LG_QUANTUM definition for architecture; specify via CPPFLAGS"
+#  endif
+#endif
+
+#define	QUANTUM			((size_t)(1U << LG_QUANTUM))
+#define	QUANTUM_MASK		(QUANTUM - 1)
+
+/* Return the smallest quantum multiple that is >= a. */
+#define	QUANTUM_CEILING(a)						\
+	(((a) + QUANTUM_MASK) & ~QUANTUM_MASK)
+
+#define	LONG			((size_t)(1U << LG_SIZEOF_LONG))
+#define	LONG_MASK		(LONG - 1)
+
+/* Return the smallest long multiple that is >= a. */
+#define	LONG_CEILING(a)							\
+	(((a) + LONG_MASK) & ~LONG_MASK)
+
+#define	SIZEOF_PTR		(1U << LG_SIZEOF_PTR)
+#define	PTR_MASK		(SIZEOF_PTR - 1)
+
+/* Return the smallest (void *) multiple that is >= a. */
+#define	PTR_CEILING(a)							\
+	(((a) + PTR_MASK) & ~PTR_MASK)
+
+/*
+ * Maximum size of L1 cache line.  This is used to avoid cache line aliasing.
+ * In addition, this controls the spacing of cacheline-spaced size classes.
+ *
+ * CACHELINE cannot be based on LG_CACHELINE because __declspec(align()) can
+ * only handle raw constants.
+ */
+#define	LG_CACHELINE		6
+#define	CACHELINE		64
+#define	CACHELINE_MASK		(CACHELINE - 1)
+
+/* Return the smallest cacheline multiple that is >= s. */
+#define	CACHELINE_CEILING(s)						\
+	(((s) + CACHELINE_MASK) & ~CACHELINE_MASK)
+
+/* Page size.  STATIC_PAGE_SHIFT is determined by the configure script. */
+#ifdef PAGE_MASK
+#  undef PAGE_MASK
+#endif
+#define	LG_PAGE		STATIC_PAGE_SHIFT
+#define	PAGE		((size_t)(1U << STATIC_PAGE_SHIFT))
+#define	PAGE_MASK	((size_t)(PAGE - 1))
+
+/* Return the smallest pagesize multiple that is >= s. */
+#define	PAGE_CEILING(s)							\
+	(((s) + PAGE_MASK) & ~PAGE_MASK)
+
+/* Return the nearest aligned address at or below a. */
+#define	ALIGNMENT_ADDR2BASE(a, alignment)				\
+	((void *)((uintptr_t)(a) & (-(alignment))))
+
+/* Return the offset between a and the nearest aligned address at or below a. */
+#define	ALIGNMENT_ADDR2OFFSET(a, alignment)				\
+	((size_t)((uintptr_t)(a) & (alignment - 1)))
+
+/* Return the smallest alignment multiple that is >= s. */
+#define	ALIGNMENT_CEILING(s, alignment)					\
+	(((s) + (alignment - 1)) & (-(alignment)))
+
+/* Declare a variable length array */
+#if __STDC_VERSION__ < 199901L
+#  ifdef _MSC_VER
+#    include <malloc.h>
+#    define alloca _alloca
+#  else
+#    ifdef JEMALLOC_HAS_ALLOCA_H
+#      include <alloca.h>
+#    else
+#      include <stdlib.h>
+#    endif
+#  endif
+#  define VARIABLE_ARRAY(type, name, count) \
+	type *name = alloca(sizeof(type) * count)
+#else
+#  define VARIABLE_ARRAY(type, name, count) type name[count]
+#endif
+
+#ifdef JEMALLOC_VALGRIND
+/*
+ * The JEMALLOC_VALGRIND_*() macros must be macros rather than functions
+ * so that when Valgrind reports errors, there are no extra stack frames
+ * in the backtraces.
+ *
+ * The size that is reported to valgrind must be consistent through a chain of
+ * malloc..realloc..realloc calls.  Request size isn't recorded anywhere in
+ * jemalloc, so it is critical that all callers of these macros provide usize
+ * rather than request size.  As a result, buffer overflow detection is
+ * technically weakened for the standard API, though it is generally accepted
+ * practice to consider any extra bytes reported by malloc_usable_size() as
+ * usable space.
+ */
+#define	JEMALLOC_VALGRIND_MALLOC(cond, ptr, usize, zero) do {		\
+	if (config_valgrind && opt_valgrind && cond)			\
+		VALGRIND_MALLOCLIKE_BLOCK(ptr, usize, p2rz(ptr), zero);	\
+} while (0)
+#define	JEMALLOC_VALGRIND_REALLOC(ptr, usize, old_ptr, old_usize,	\
+    old_rzsize, zero)  do {						\
+	if (config_valgrind && opt_valgrind) {				\
+		size_t rzsize = p2rz(ptr);				\
+									\
+		if (ptr == old_ptr) {					\
+			VALGRIND_RESIZEINPLACE_BLOCK(ptr, old_usize,	\
+			    usize, rzsize);				\
+			if (zero && old_usize < usize) {		\
+				VALGRIND_MAKE_MEM_DEFINED(		\
+				    (void *)((uintptr_t)ptr +		\
+				    old_usize), usize - old_usize);	\
+			}						\
+		} else {						\
+			if (old_ptr != NULL) {				\
+				VALGRIND_FREELIKE_BLOCK(old_ptr,	\
+				    old_rzsize);			\
+			}						\
+			if (ptr != NULL) {				\
+				size_t copy_size = (old_usize < usize)	\
+				    ?  old_usize : usize;		\
+				size_t tail_size = usize - copy_size;	\
+				VALGRIND_MALLOCLIKE_BLOCK(ptr, usize,	\
+				    rzsize, false);			\
+				if (copy_size > 0) {			\
+					VALGRIND_MAKE_MEM_DEFINED(ptr,	\
+					    copy_size);			\
+				}					\
+				if (zero && tail_size > 0) {		\
+					VALGRIND_MAKE_MEM_DEFINED(	\
+					    (void *)((uintptr_t)ptr +	\
+					    copy_size), tail_size);	\
+				}					\
+			}						\
+		}							\
+	}								\
+} while (0)
+#define	JEMALLOC_VALGRIND_FREE(ptr, rzsize) do {			\
+	if (config_valgrind && opt_valgrind)				\
+		VALGRIND_FREELIKE_BLOCK(ptr, rzsize);			\
+} while (0)
+#else
+#define	RUNNING_ON_VALGRIND	((unsigned)0)
+#define	VALGRIND_MALLOCLIKE_BLOCK(addr, sizeB, rzB, is_zeroed) \
+    do {} while (0)
+#define	VALGRIND_RESIZEINPLACE_BLOCK(addr, oldSizeB, newSizeB, rzB) \
+    do {} while (0)
+#define	VALGRIND_FREELIKE_BLOCK(addr, rzB) do {} while (0)
+#define	VALGRIND_MAKE_MEM_NOACCESS(_qzz_addr, _qzz_len) do {} while (0)
+#define	VALGRIND_MAKE_MEM_UNDEFINED(_qzz_addr, _qzz_len) do {} while (0)
+#define	VALGRIND_MAKE_MEM_DEFINED(_qzz_addr, _qzz_len) do {} while (0)
+#define	JEMALLOC_VALGRIND_MALLOC(cond, ptr, usize, zero) do {} while (0)
+#define	JEMALLOC_VALGRIND_REALLOC(ptr, usize, old_ptr, old_usize,	\
+    old_rzsize, zero) do {} while (0)
+#define	JEMALLOC_VALGRIND_FREE(ptr, rzsize) do {} while (0)
+#endif
+
+#include "jemalloc/internal/util.h"
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/prng.h"
+#include "jemalloc/internal/ckh.h"
+#include "jemalloc/internal/size_classes.h"
+#include "jemalloc/internal/stats.h"
+#include "jemalloc/internal/ctl.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/tsd.h"
+#include "jemalloc/internal/mb.h"
+#include "jemalloc/internal/extent.h"
+#include "jemalloc/internal/arena.h"
+#include "jemalloc/internal/bitmap.h"
+#include "jemalloc/internal/base.h"
+#include "jemalloc/internal/chunk.h"
+#include "jemalloc/internal/huge.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/tcache.h"
+#include "jemalloc/internal/hash.h"
+#include "jemalloc/internal/quarantine.h"
+#include "jemalloc/internal/prof.h"
+
+#undef JEMALLOC_H_TYPES
+/******************************************************************************/
+#define JEMALLOC_H_STRUCTS
+
+#include "jemalloc/internal/util.h"
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/prng.h"
+#include "jemalloc/internal/ckh.h"
+#include "jemalloc/internal/size_classes.h"
+#include "jemalloc/internal/stats.h"
+#include "jemalloc/internal/ctl.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/tsd.h"
+#include "jemalloc/internal/mb.h"
+#include "jemalloc/internal/bitmap.h"
+#include "jemalloc/internal/extent.h"
+#include "jemalloc/internal/arena.h"
+#include "jemalloc/internal/base.h"
+#include "jemalloc/internal/chunk.h"
+#include "jemalloc/internal/huge.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/tcache.h"
+#include "jemalloc/internal/hash.h"
+#include "jemalloc/internal/quarantine.h"
+#include "jemalloc/internal/prof.h"
+
+typedef struct {
+	uint64_t	allocated;
+	uint64_t	deallocated;
+} thread_allocated_t;
+/*
+ * The JEMALLOC_CONCAT() wrapper is necessary to pass {0, 0} via a cpp macro
+ * argument.
+ */
+#define	THREAD_ALLOCATED_INITIALIZER	JEMALLOC_CONCAT({0, 0})
+
+#undef JEMALLOC_H_STRUCTS
+/******************************************************************************/
+#define JEMALLOC_H_EXTERNS
+
+extern bool	opt_abort;
+extern bool	opt_junk;
+extern size_t	opt_quarantine;
+extern bool	opt_redzone;
+extern bool	opt_utrace;
+extern bool	opt_valgrind;
+extern bool	opt_xmalloc;
+extern bool	opt_zero;
+extern size_t	opt_narenas;
+
+/* Number of CPUs. */
+extern unsigned		ncpus;
+
+/* Protects arenas initialization (arenas, arenas_total). */
+extern malloc_mutex_t	arenas_lock;
+/*
+ * Arenas that are used to service external requests.  Not all elements of the
+ * arenas array are necessarily used; arenas are created lazily as needed.
+ *
+ * arenas[0..narenas_auto) are used for automatic multiplexing of threads and
+ * arenas.  arenas[narenas_auto..narenas_total) are only used if the application
+ * takes some action to create them and allocate from them.
+ */
+extern arena_t		**arenas;
+extern unsigned		narenas_total;
+extern unsigned		narenas_auto; /* Read-only after initialization. */
+
+arena_t	*arenas_extend(unsigned ind);
+void	arenas_cleanup(void *arg);
+arena_t	*choose_arena_hard(void);
+void	jemalloc_prefork(void);
+void	jemalloc_postfork_parent(void);
+void	jemalloc_postfork_child(void);
+
+#include "jemalloc/internal/util.h"
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/prng.h"
+#include "jemalloc/internal/ckh.h"
+#include "jemalloc/internal/size_classes.h"
+#include "jemalloc/internal/stats.h"
+#include "jemalloc/internal/ctl.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/tsd.h"
+#include "jemalloc/internal/mb.h"
+#include "jemalloc/internal/bitmap.h"
+#include "jemalloc/internal/extent.h"
+#include "jemalloc/internal/arena.h"
+#include "jemalloc/internal/base.h"
+#include "jemalloc/internal/chunk.h"
+#include "jemalloc/internal/huge.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/tcache.h"
+#include "jemalloc/internal/hash.h"
+#include "jemalloc/internal/quarantine.h"
+#include "jemalloc/internal/prof.h"
+
+#undef JEMALLOC_H_EXTERNS
+/******************************************************************************/
+#define JEMALLOC_H_INLINES
+
+#include "jemalloc/internal/util.h"
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/prng.h"
+#include "jemalloc/internal/ckh.h"
+#include "jemalloc/internal/size_classes.h"
+#include "jemalloc/internal/stats.h"
+#include "jemalloc/internal/ctl.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/tsd.h"
+#include "jemalloc/internal/mb.h"
+#include "jemalloc/internal/extent.h"
+#include "jemalloc/internal/base.h"
+#include "jemalloc/internal/chunk.h"
+#include "jemalloc/internal/huge.h"
+
+#ifndef JEMALLOC_ENABLE_INLINE
+malloc_tsd_protos(JEMALLOC_ATTR(unused), arenas, arena_t *)
+
+size_t	s2u(size_t size);
+size_t	sa2u(size_t size, size_t alignment);
+unsigned	narenas_total_get(void);
+arena_t	*choose_arena(arena_t *arena);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_C_))
+/*
+ * Map of pthread_self() --> arenas[???], used for selecting an arena to use
+ * for allocations.
+ */
+malloc_tsd_externs(arenas, arena_t *)
+malloc_tsd_funcs(JEMALLOC_ALWAYS_INLINE, arenas, arena_t *, NULL,
+    arenas_cleanup)
+
+/*
+ * Compute usable size that would result from allocating an object with the
+ * specified size.
+ */
+JEMALLOC_ALWAYS_INLINE size_t
+s2u(size_t size)
+{
+
+	if (size <= SMALL_MAXCLASS)
+		return (arena_bin_info[SMALL_SIZE2BIN(size)].reg_size);
+	if (size <= arena_maxclass)
+		return (PAGE_CEILING(size));
+	return (CHUNK_CEILING(size));
+}
+
+/*
+ * Compute usable size that would result from allocating an object with the
+ * specified size and alignment.
+ */
+JEMALLOC_ALWAYS_INLINE size_t
+sa2u(size_t size, size_t alignment)
+{
+	size_t usize;
+
+	assert(alignment != 0 && ((alignment - 1) & alignment) == 0);
+
+	/*
+	 * Round size up to the nearest multiple of alignment.
+	 *
+	 * This done, we can take advantage of the fact that for each small
+	 * size class, every object is aligned at the smallest power of two
+	 * that is non-zero in the base two representation of the size.  For
+	 * example:
+	 *
+	 *   Size |   Base 2 | Minimum alignment
+	 *   -----+----------+------------------
+	 *     96 |  1100000 |  32
+	 *    144 | 10100000 |  32
+	 *    192 | 11000000 |  64
+	 */
+	usize = ALIGNMENT_CEILING(size, alignment);
+	/*
+	 * (usize < size) protects against the combination of maximal
+	 * alignment and size greater than maximal alignment.
+	 */
+	if (usize < size) {
+		/* size_t overflow. */
+		return (0);
+	}
+
+	if (usize <= arena_maxclass && alignment <= PAGE) {
+		if (usize <= SMALL_MAXCLASS)
+			return (arena_bin_info[SMALL_SIZE2BIN(usize)].reg_size);
+		return (PAGE_CEILING(usize));
+	} else {
+		size_t run_size;
+
+		/*
+		 * We can't achieve subpage alignment, so round up alignment
+		 * permanently; it makes later calculations simpler.
+		 */
+		alignment = PAGE_CEILING(alignment);
+		usize = PAGE_CEILING(size);
+		/*
+		 * (usize < size) protects against very large sizes within
+		 * PAGE of SIZE_T_MAX.
+		 *
+		 * (usize + alignment < usize) protects against the
+		 * combination of maximal alignment and usize large enough
+		 * to cause overflow.  This is similar to the first overflow
+		 * check above, but it needs to be repeated due to the new
+		 * usize value, which may now be *equal* to maximal
+		 * alignment, whereas before we only detected overflow if the
+		 * original size was *greater* than maximal alignment.
+		 */
+		if (usize < size || usize + alignment < usize) {
+			/* size_t overflow. */
+			return (0);
+		}
+
+		/*
+		 * Calculate the size of the over-size run that arena_palloc()
+		 * would need to allocate in order to guarantee the alignment.
+		 * If the run wouldn't fit within a chunk, round up to a huge
+		 * allocation size.
+		 */
+		run_size = usize + alignment - PAGE;
+		if (run_size <= arena_maxclass)
+			return (PAGE_CEILING(usize));
+		return (CHUNK_CEILING(usize));
+	}
+}
+
+JEMALLOC_INLINE unsigned
+narenas_total_get(void)
+{
+	unsigned narenas;
+
+	malloc_mutex_lock(&arenas_lock);
+	narenas = narenas_total;
+	malloc_mutex_unlock(&arenas_lock);
+
+	return (narenas);
+}
+
+/* Choose an arena based on a per-thread value. */
+JEMALLOC_INLINE arena_t *
+choose_arena(arena_t *arena)
+{
+	arena_t *ret;
+
+	if (arena != NULL)
+		return (arena);
+
+	if ((ret = *arenas_tsd_get()) == NULL) {
+		ret = choose_arena_hard();
+		assert(ret != NULL);
+	}
+
+	return (ret);
+}
+#endif
+
+#include "jemalloc/internal/bitmap.h"
+#include "jemalloc/internal/rtree.h"
+/*
+ * Include arena.h twice in order to resolve circular dependencies with
+ * tcache.h.
+ */
+#define	JEMALLOC_ARENA_INLINE_A
+#include "jemalloc/internal/arena.h"
+#undef JEMALLOC_ARENA_INLINE_A
+#include "jemalloc/internal/tcache.h"
+#define	JEMALLOC_ARENA_INLINE_B
+#include "jemalloc/internal/arena.h"
+#undef JEMALLOC_ARENA_INLINE_B
+#include "jemalloc/internal/hash.h"
+#include "jemalloc/internal/quarantine.h"
+
+#ifndef JEMALLOC_ENABLE_INLINE
+void	*imallocx(size_t size, bool try_tcache, arena_t *arena);
+void	*imalloc(size_t size);
+void	*icallocx(size_t size, bool try_tcache, arena_t *arena);
+void	*icalloc(size_t size);
+void	*ipallocx(size_t usize, size_t alignment, bool zero, bool try_tcache,
+    arena_t *arena);
+void	*ipalloc(size_t usize, size_t alignment, bool zero);
+size_t	isalloc(const void *ptr, bool demote);
+size_t	ivsalloc(const void *ptr, bool demote);
+size_t	u2rz(size_t usize);
+size_t	p2rz(const void *ptr);
+void	idallocx(void *ptr, bool try_tcache);
+void	idalloc(void *ptr);
+void	iqallocx(void *ptr, bool try_tcache);
+void	iqalloc(void *ptr);
+void	*irallocx(void *ptr, size_t size, size_t extra, size_t alignment,
+    bool zero, bool no_move, bool try_tcache_alloc, bool try_tcache_dalloc,
+    arena_t *arena);
+void	*iralloc(void *ptr, size_t size, size_t extra, size_t alignment,
+    bool zero, bool no_move);
+malloc_tsd_protos(JEMALLOC_ATTR(unused), thread_allocated, thread_allocated_t)
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_C_))
+JEMALLOC_ALWAYS_INLINE void *
+imallocx(size_t size, bool try_tcache, arena_t *arena)
+{
+
+	assert(size != 0);
+
+	if (size <= arena_maxclass)
+		return (arena_malloc(arena, size, false, try_tcache));
+	else
+		return (huge_malloc(size, false));
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+imalloc(size_t size)
+{
+
+	return (imallocx(size, true, NULL));
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+icallocx(size_t size, bool try_tcache, arena_t *arena)
+{
+
+	if (size <= arena_maxclass)
+		return (arena_malloc(arena, size, true, try_tcache));
+	else
+		return (huge_malloc(size, true));
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+icalloc(size_t size)
+{
+
+	return (icallocx(size, true, NULL));
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+ipallocx(size_t usize, size_t alignment, bool zero, bool try_tcache,
+    arena_t *arena)
+{
+	void *ret;
+
+	assert(usize != 0);
+	assert(usize == sa2u(usize, alignment));
+
+	if (usize <= arena_maxclass && alignment <= PAGE)
+		ret = arena_malloc(arena, usize, zero, try_tcache);
+	else {
+		if (usize <= arena_maxclass) {
+			ret = arena_palloc(choose_arena(arena), usize,
+			    alignment, zero);
+		} else if (alignment <= chunksize)
+			ret = huge_malloc(usize, zero);
+		else
+			ret = huge_palloc(usize, alignment, zero);
+	}
+
+	assert(ALIGNMENT_ADDR2BASE(ret, alignment) == ret);
+	return (ret);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+ipalloc(size_t usize, size_t alignment, bool zero)
+{
+
+	return (ipallocx(usize, alignment, zero, true, NULL));
+}
+
+/*
+ * Typical usage:
+ *   void *ptr = [...]
+ *   size_t sz = isalloc(ptr, config_prof);
+ */
+JEMALLOC_ALWAYS_INLINE size_t
+isalloc(const void *ptr, bool demote)
+{
+	size_t ret;
+	arena_chunk_t *chunk;
+
+	assert(ptr != NULL);
+	/* Demotion only makes sense if config_prof is true. */
+	assert(config_prof || demote == false);
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+	if (chunk != ptr)
+		ret = arena_salloc(ptr, demote);
+	else
+		ret = huge_salloc(ptr);
+
+	return (ret);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+ivsalloc(const void *ptr, bool demote)
+{
+
+	/* Return 0 if ptr is not within a chunk managed by jemalloc. */
+	if (rtree_get(chunks_rtree, (uintptr_t)CHUNK_ADDR2BASE(ptr)) == NULL)
+		return (0);
+
+	return (isalloc(ptr, demote));
+}
+
+JEMALLOC_INLINE size_t
+u2rz(size_t usize)
+{
+	size_t ret;
+
+	if (usize <= SMALL_MAXCLASS) {
+		size_t binind = SMALL_SIZE2BIN(usize);
+		ret = arena_bin_info[binind].redzone_size;
+	} else
+		ret = 0;
+
+	return (ret);
+}
+
+JEMALLOC_INLINE size_t
+p2rz(const void *ptr)
+{
+	size_t usize = isalloc(ptr, false);
+
+	return (u2rz(usize));
+}
+
+JEMALLOC_ALWAYS_INLINE void
+idallocx(void *ptr, bool try_tcache)
+{
+	arena_chunk_t *chunk;
+
+	assert(ptr != NULL);
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+	if (chunk != ptr)
+		arena_dalloc(chunk->arena, chunk, ptr, try_tcache);
+	else
+		huge_dalloc(ptr, true);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+idalloc(void *ptr)
+{
+
+	idallocx(ptr, true);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+iqallocx(void *ptr, bool try_tcache)
+{
+
+	if (config_fill && opt_quarantine)
+		quarantine(ptr);
+	else
+		idallocx(ptr, try_tcache);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+iqalloc(void *ptr)
+{
+
+	iqallocx(ptr, true);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+irallocx(void *ptr, size_t size, size_t extra, size_t alignment, bool zero,
+    bool no_move, bool try_tcache_alloc, bool try_tcache_dalloc, arena_t *arena)
+{
+	void *ret;
+	size_t oldsize;
+
+	assert(ptr != NULL);
+	assert(size != 0);
+
+	oldsize = isalloc(ptr, config_prof);
+
+	if (alignment != 0 && ((uintptr_t)ptr & ((uintptr_t)alignment-1))
+	    != 0) {
+		size_t usize, copysize;
+
+		/*
+		 * Existing object alignment is inadequate; allocate new space
+		 * and copy.
+		 */
+		if (no_move)
+			return (NULL);
+		usize = sa2u(size + extra, alignment);
+		if (usize == 0)
+			return (NULL);
+		ret = ipallocx(usize, alignment, zero, try_tcache_alloc, arena);
+		if (ret == NULL) {
+			if (extra == 0)
+				return (NULL);
+			/* Try again, without extra this time. */
+			usize = sa2u(size, alignment);
+			if (usize == 0)
+				return (NULL);
+			ret = ipallocx(usize, alignment, zero, try_tcache_alloc,
+			    arena);
+			if (ret == NULL)
+				return (NULL);
+		}
+		/*
+		 * Copy at most size bytes (not size+extra), since the caller
+		 * has no expectation that the extra bytes will be reliably
+		 * preserved.
+		 */
+		copysize = (size < oldsize) ? size : oldsize;
+		memcpy(ret, ptr, copysize);
+		iqallocx(ptr, try_tcache_dalloc);
+		return (ret);
+	}
+
+	if (no_move) {
+		if (size <= arena_maxclass) {
+			return (arena_ralloc_no_move(ptr, oldsize, size,
+			    extra, zero));
+		} else {
+			return (huge_ralloc_no_move(ptr, oldsize, size,
+			    extra));
+		}
+	} else {
+		if (size + extra <= arena_maxclass) {
+			return (arena_ralloc(arena, ptr, oldsize, size, extra,
+			    alignment, zero, try_tcache_alloc,
+			    try_tcache_dalloc));
+		} else {
+			return (huge_ralloc(ptr, oldsize, size, extra,
+			    alignment, zero, try_tcache_dalloc));
+		}
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+iralloc(void *ptr, size_t size, size_t extra, size_t alignment, bool zero,
+    bool no_move)
+{
+
+	return (irallocx(ptr, size, extra, alignment, zero, no_move, true, true,
+	    NULL));
+}
+
+malloc_tsd_externs(thread_allocated, thread_allocated_t)
+malloc_tsd_funcs(JEMALLOC_ALWAYS_INLINE, thread_allocated, thread_allocated_t,
+    THREAD_ALLOCATED_INITIALIZER, malloc_tsd_no_cleanup)
+#endif
+
+#include "jemalloc/internal/prof.h"
+
+#undef JEMALLOC_H_INLINES
+/******************************************************************************/
+#endif /* JEMALLOC_INTERNAL_H */
diff --git a/extra/jemalloc/include/jemalloc/internal/mb.h b/extra/jemalloc/include/jemalloc/internal/mb.h
new file mode 100644
index 00000000000..3cfa7872942
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/mb.h
@@ -0,0 +1,115 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+void	mb_write(void);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_MB_C_))
+#ifdef __i386__
+/*
+ * According to the Intel Architecture Software Developer's Manual, current
+ * processors execute instructions in order from the perspective of other
+ * processors in a multiprocessor system, but 1) Intel reserves the right to
+ * change that, and 2) the compiler's optimizer could re-order instructions if
+ * there weren't some form of barrier.  Therefore, even if running on an
+ * architecture that does not need memory barriers (everything through at least
+ * i686), an "optimizer barrier" is necessary.
+ */
+JEMALLOC_INLINE void
+mb_write(void)
+{
+
+#  if 0
+	/* This is a true memory barrier. */
+	asm volatile ("pusha;"
+	    "xor  %%eax,%%eax;"
+	    "cpuid;"
+	    "popa;"
+	    : /* Outputs. */
+	    : /* Inputs. */
+	    : "memory" /* Clobbers. */
+	    );
+#else
+	/*
+	 * This is hopefully enough to keep the compiler from reordering
+	 * instructions around this one.
+	 */
+	asm volatile ("nop;"
+	    : /* Outputs. */
+	    : /* Inputs. */
+	    : "memory" /* Clobbers. */
+	    );
+#endif
+}
+#elif (defined(__amd64__) || defined(__x86_64__))
+JEMALLOC_INLINE void
+mb_write(void)
+{
+
+	asm volatile ("sfence"
+	    : /* Outputs. */
+	    : /* Inputs. */
+	    : "memory" /* Clobbers. */
+	    );
+}
+#elif defined(__powerpc__)
+JEMALLOC_INLINE void
+mb_write(void)
+{
+
+	asm volatile ("eieio"
+	    : /* Outputs. */
+	    : /* Inputs. */
+	    : "memory" /* Clobbers. */
+	    );
+}
+#elif defined(__sparc64__)
+JEMALLOC_INLINE void
+mb_write(void)
+{
+
+	asm volatile ("membar #StoreStore"
+	    : /* Outputs. */
+	    : /* Inputs. */
+	    : "memory" /* Clobbers. */
+	    );
+}
+#elif defined(__tile__)
+JEMALLOC_INLINE void
+mb_write(void)
+{
+
+	__sync_synchronize();
+}
+#else
+/*
+ * This is much slower than a simple memory barrier, but the semantics of mutex
+ * unlock make this work.
+ */
+JEMALLOC_INLINE void
+mb_write(void)
+{
+	malloc_mutex_t mtx;
+
+	malloc_mutex_init(&mtx);
+	malloc_mutex_lock(&mtx);
+	malloc_mutex_unlock(&mtx);
+}
+#endif
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/internal/mutex.h b/extra/jemalloc/include/jemalloc/internal/mutex.h
new file mode 100644
index 00000000000..de44e1435ad
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/mutex.h
@@ -0,0 +1,99 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+typedef struct malloc_mutex_s malloc_mutex_t;
+
+#ifdef _WIN32
+#  define MALLOC_MUTEX_INITIALIZER
+#elif (defined(JEMALLOC_OSSPIN))
+#  define MALLOC_MUTEX_INITIALIZER {0}
+#elif (defined(JEMALLOC_MUTEX_INIT_CB))
+#  define MALLOC_MUTEX_INITIALIZER {PTHREAD_MUTEX_INITIALIZER, NULL}
+#else
+#  if (defined(PTHREAD_MUTEX_ADAPTIVE_NP) &&				\
+       defined(PTHREAD_ADAPTIVE_MUTEX_INITIALIZER_NP))
+#    define MALLOC_MUTEX_TYPE PTHREAD_MUTEX_ADAPTIVE_NP
+#    define MALLOC_MUTEX_INITIALIZER {PTHREAD_ADAPTIVE_MUTEX_INITIALIZER_NP}
+#  else
+#    define MALLOC_MUTEX_TYPE PTHREAD_MUTEX_DEFAULT
+#    define MALLOC_MUTEX_INITIALIZER {PTHREAD_MUTEX_INITIALIZER}
+#  endif
+#endif
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+struct malloc_mutex_s {
+#ifdef _WIN32
+	CRITICAL_SECTION	lock;
+#elif (defined(JEMALLOC_OSSPIN))
+	OSSpinLock		lock;
+#elif (defined(JEMALLOC_MUTEX_INIT_CB))
+	pthread_mutex_t		lock;
+	malloc_mutex_t		*postponed_next;
+#else
+	pthread_mutex_t		lock;
+#endif
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+#ifdef JEMALLOC_LAZY_LOCK
+extern bool isthreaded;
+#else
+#  undef isthreaded /* Undo private_namespace.h definition. */
+#  define isthreaded true
+#endif
+
+bool	malloc_mutex_init(malloc_mutex_t *mutex);
+void	malloc_mutex_prefork(malloc_mutex_t *mutex);
+void	malloc_mutex_postfork_parent(malloc_mutex_t *mutex);
+void	malloc_mutex_postfork_child(malloc_mutex_t *mutex);
+bool	mutex_boot(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+void	malloc_mutex_lock(malloc_mutex_t *mutex);
+void	malloc_mutex_unlock(malloc_mutex_t *mutex);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_MUTEX_C_))
+JEMALLOC_INLINE void
+malloc_mutex_lock(malloc_mutex_t *mutex)
+{
+
+	if (isthreaded) {
+#ifdef _WIN32
+		EnterCriticalSection(&mutex->lock);
+#elif (defined(JEMALLOC_OSSPIN))
+		OSSpinLockLock(&mutex->lock);
+#else
+		pthread_mutex_lock(&mutex->lock);
+#endif
+	}
+}
+
+JEMALLOC_INLINE void
+malloc_mutex_unlock(malloc_mutex_t *mutex)
+{
+
+	if (isthreaded) {
+#ifdef _WIN32
+		LeaveCriticalSection(&mutex->lock);
+#elif (defined(JEMALLOC_OSSPIN))
+		OSSpinLockUnlock(&mutex->lock);
+#else
+		pthread_mutex_unlock(&mutex->lock);
+#endif
+	}
+}
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/internal/private_namespace.h b/extra/jemalloc/include/jemalloc/internal/private_namespace.h
new file mode 100644
index 00000000000..65de3163fd3
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/private_namespace.h
@@ -0,0 +1,390 @@
+#define	a0calloc JEMALLOC_N(a0calloc)
+#define	a0free JEMALLOC_N(a0free)
+#define	a0malloc JEMALLOC_N(a0malloc)
+#define	arena_alloc_junk_small JEMALLOC_N(arena_alloc_junk_small)
+#define	arena_bin_index JEMALLOC_N(arena_bin_index)
+#define	arena_bin_info JEMALLOC_N(arena_bin_info)
+#define	arena_boot JEMALLOC_N(arena_boot)
+#define	arena_dalloc JEMALLOC_N(arena_dalloc)
+#define	arena_dalloc_bin JEMALLOC_N(arena_dalloc_bin)
+#define	arena_dalloc_bin_locked JEMALLOC_N(arena_dalloc_bin_locked)
+#define	arena_dalloc_junk_small JEMALLOC_N(arena_dalloc_junk_small)
+#define	arena_dalloc_large JEMALLOC_N(arena_dalloc_large)
+#define	arena_dalloc_large_locked JEMALLOC_N(arena_dalloc_large_locked)
+#define	arena_dalloc_small JEMALLOC_N(arena_dalloc_small)
+#define	arena_dss_prec_get JEMALLOC_N(arena_dss_prec_get)
+#define	arena_dss_prec_set JEMALLOC_N(arena_dss_prec_set)
+#define	arena_malloc JEMALLOC_N(arena_malloc)
+#define	arena_malloc_large JEMALLOC_N(arena_malloc_large)
+#define	arena_malloc_small JEMALLOC_N(arena_malloc_small)
+#define	arena_mapbits_allocated_get JEMALLOC_N(arena_mapbits_allocated_get)
+#define	arena_mapbits_binind_get JEMALLOC_N(arena_mapbits_binind_get)
+#define	arena_mapbits_dirty_get JEMALLOC_N(arena_mapbits_dirty_get)
+#define	arena_mapbits_get JEMALLOC_N(arena_mapbits_get)
+#define	arena_mapbits_large_binind_set JEMALLOC_N(arena_mapbits_large_binind_set)
+#define	arena_mapbits_large_get JEMALLOC_N(arena_mapbits_large_get)
+#define	arena_mapbits_large_set JEMALLOC_N(arena_mapbits_large_set)
+#define	arena_mapbits_large_size_get JEMALLOC_N(arena_mapbits_large_size_get)
+#define	arena_mapbits_small_runind_get JEMALLOC_N(arena_mapbits_small_runind_get)
+#define	arena_mapbits_small_set JEMALLOC_N(arena_mapbits_small_set)
+#define	arena_mapbits_unallocated_set JEMALLOC_N(arena_mapbits_unallocated_set)
+#define	arena_mapbits_unallocated_size_get JEMALLOC_N(arena_mapbits_unallocated_size_get)
+#define	arena_mapbits_unallocated_size_set JEMALLOC_N(arena_mapbits_unallocated_size_set)
+#define	arena_mapbits_unzeroed_get JEMALLOC_N(arena_mapbits_unzeroed_get)
+#define	arena_mapbits_unzeroed_set JEMALLOC_N(arena_mapbits_unzeroed_set)
+#define	arena_mapbitsp_get JEMALLOC_N(arena_mapbitsp_get)
+#define	arena_mapp_get JEMALLOC_N(arena_mapp_get)
+#define	arena_maxclass JEMALLOC_N(arena_maxclass)
+#define	arena_new JEMALLOC_N(arena_new)
+#define	arena_palloc JEMALLOC_N(arena_palloc)
+#define	arena_postfork_child JEMALLOC_N(arena_postfork_child)
+#define	arena_postfork_parent JEMALLOC_N(arena_postfork_parent)
+#define	arena_prefork JEMALLOC_N(arena_prefork)
+#define	arena_prof_accum JEMALLOC_N(arena_prof_accum)
+#define	arena_prof_accum_impl JEMALLOC_N(arena_prof_accum_impl)
+#define	arena_prof_accum_locked JEMALLOC_N(arena_prof_accum_locked)
+#define	arena_prof_ctx_get JEMALLOC_N(arena_prof_ctx_get)
+#define	arena_prof_ctx_set JEMALLOC_N(arena_prof_ctx_set)
+#define	arena_prof_promoted JEMALLOC_N(arena_prof_promoted)
+#define	arena_ptr_small_binind_get JEMALLOC_N(arena_ptr_small_binind_get)
+#define	arena_purge_all JEMALLOC_N(arena_purge_all)
+#define	arena_ralloc JEMALLOC_N(arena_ralloc)
+#define	arena_ralloc_no_move JEMALLOC_N(arena_ralloc_no_move)
+#define	arena_run_regind JEMALLOC_N(arena_run_regind)
+#define	arena_salloc JEMALLOC_N(arena_salloc)
+#define	arena_stats_merge JEMALLOC_N(arena_stats_merge)
+#define	arena_tcache_fill_small JEMALLOC_N(arena_tcache_fill_small)
+#define	arenas JEMALLOC_N(arenas)
+#define	arenas_booted JEMALLOC_N(arenas_booted)
+#define	arenas_cleanup JEMALLOC_N(arenas_cleanup)
+#define	arenas_extend JEMALLOC_N(arenas_extend)
+#define	arenas_initialized JEMALLOC_N(arenas_initialized)
+#define	arenas_lock JEMALLOC_N(arenas_lock)
+#define	arenas_tls JEMALLOC_N(arenas_tls)
+#define	arenas_tsd JEMALLOC_N(arenas_tsd)
+#define	arenas_tsd_boot JEMALLOC_N(arenas_tsd_boot)
+#define	arenas_tsd_cleanup_wrapper JEMALLOC_N(arenas_tsd_cleanup_wrapper)
+#define	arenas_tsd_get JEMALLOC_N(arenas_tsd_get)
+#define	arenas_tsd_get_wrapper JEMALLOC_N(arenas_tsd_get_wrapper)
+#define	arenas_tsd_set JEMALLOC_N(arenas_tsd_set)
+#define	atomic_add_u JEMALLOC_N(atomic_add_u)
+#define	atomic_add_uint32 JEMALLOC_N(atomic_add_uint32)
+#define	atomic_add_uint64 JEMALLOC_N(atomic_add_uint64)
+#define	atomic_add_z JEMALLOC_N(atomic_add_z)
+#define	atomic_sub_u JEMALLOC_N(atomic_sub_u)
+#define	atomic_sub_uint32 JEMALLOC_N(atomic_sub_uint32)
+#define	atomic_sub_uint64 JEMALLOC_N(atomic_sub_uint64)
+#define	atomic_sub_z JEMALLOC_N(atomic_sub_z)
+#define	base_alloc JEMALLOC_N(base_alloc)
+#define	base_boot JEMALLOC_N(base_boot)
+#define	base_calloc JEMALLOC_N(base_calloc)
+#define	base_node_alloc JEMALLOC_N(base_node_alloc)
+#define	base_node_dealloc JEMALLOC_N(base_node_dealloc)
+#define	base_postfork_child JEMALLOC_N(base_postfork_child)
+#define	base_postfork_parent JEMALLOC_N(base_postfork_parent)
+#define	base_prefork JEMALLOC_N(base_prefork)
+#define	bitmap_full JEMALLOC_N(bitmap_full)
+#define	bitmap_get JEMALLOC_N(bitmap_get)
+#define	bitmap_info_init JEMALLOC_N(bitmap_info_init)
+#define	bitmap_info_ngroups JEMALLOC_N(bitmap_info_ngroups)
+#define	bitmap_init JEMALLOC_N(bitmap_init)
+#define	bitmap_set JEMALLOC_N(bitmap_set)
+#define	bitmap_sfu JEMALLOC_N(bitmap_sfu)
+#define	bitmap_size JEMALLOC_N(bitmap_size)
+#define	bitmap_unset JEMALLOC_N(bitmap_unset)
+#define	bt_init JEMALLOC_N(bt_init)
+#define	buferror JEMALLOC_N(buferror)
+#define	choose_arena JEMALLOC_N(choose_arena)
+#define	choose_arena_hard JEMALLOC_N(choose_arena_hard)
+#define	chunk_alloc JEMALLOC_N(chunk_alloc)
+#define	chunk_alloc_dss JEMALLOC_N(chunk_alloc_dss)
+#define	chunk_alloc_mmap JEMALLOC_N(chunk_alloc_mmap)
+#define	chunk_boot JEMALLOC_N(chunk_boot)
+#define	chunk_dealloc JEMALLOC_N(chunk_dealloc)
+#define	chunk_dealloc_mmap JEMALLOC_N(chunk_dealloc_mmap)
+#define	chunk_dss_boot JEMALLOC_N(chunk_dss_boot)
+#define	chunk_dss_postfork_child JEMALLOC_N(chunk_dss_postfork_child)
+#define	chunk_dss_postfork_parent JEMALLOC_N(chunk_dss_postfork_parent)
+#define	chunk_dss_prec_get JEMALLOC_N(chunk_dss_prec_get)
+#define	chunk_dss_prec_set JEMALLOC_N(chunk_dss_prec_set)
+#define	chunk_dss_prefork JEMALLOC_N(chunk_dss_prefork)
+#define	chunk_in_dss JEMALLOC_N(chunk_in_dss)
+#define	chunk_npages JEMALLOC_N(chunk_npages)
+#define	chunk_postfork_child JEMALLOC_N(chunk_postfork_child)
+#define	chunk_postfork_parent JEMALLOC_N(chunk_postfork_parent)
+#define	chunk_prefork JEMALLOC_N(chunk_prefork)
+#define	chunk_unmap JEMALLOC_N(chunk_unmap)
+#define	chunks_mtx JEMALLOC_N(chunks_mtx)
+#define	chunks_rtree JEMALLOC_N(chunks_rtree)
+#define	chunksize JEMALLOC_N(chunksize)
+#define	chunksize_mask JEMALLOC_N(chunksize_mask)
+#define	ckh_bucket_search JEMALLOC_N(ckh_bucket_search)
+#define	ckh_count JEMALLOC_N(ckh_count)
+#define	ckh_delete JEMALLOC_N(ckh_delete)
+#define	ckh_evict_reloc_insert JEMALLOC_N(ckh_evict_reloc_insert)
+#define	ckh_insert JEMALLOC_N(ckh_insert)
+#define	ckh_isearch JEMALLOC_N(ckh_isearch)
+#define	ckh_iter JEMALLOC_N(ckh_iter)
+#define	ckh_new JEMALLOC_N(ckh_new)
+#define	ckh_pointer_hash JEMALLOC_N(ckh_pointer_hash)
+#define	ckh_pointer_keycomp JEMALLOC_N(ckh_pointer_keycomp)
+#define	ckh_rebuild JEMALLOC_N(ckh_rebuild)
+#define	ckh_remove JEMALLOC_N(ckh_remove)
+#define	ckh_search JEMALLOC_N(ckh_search)
+#define	ckh_string_hash JEMALLOC_N(ckh_string_hash)
+#define	ckh_string_keycomp JEMALLOC_N(ckh_string_keycomp)
+#define	ckh_try_bucket_insert JEMALLOC_N(ckh_try_bucket_insert)
+#define	ckh_try_insert JEMALLOC_N(ckh_try_insert)
+#define	ctl_boot JEMALLOC_N(ctl_boot)
+#define	ctl_bymib JEMALLOC_N(ctl_bymib)
+#define	ctl_byname JEMALLOC_N(ctl_byname)
+#define	ctl_nametomib JEMALLOC_N(ctl_nametomib)
+#define	ctl_postfork_child JEMALLOC_N(ctl_postfork_child)
+#define	ctl_postfork_parent JEMALLOC_N(ctl_postfork_parent)
+#define	ctl_prefork JEMALLOC_N(ctl_prefork)
+#define	dss_prec_names JEMALLOC_N(dss_prec_names)
+#define	extent_tree_ad_first JEMALLOC_N(extent_tree_ad_first)
+#define	extent_tree_ad_insert JEMALLOC_N(extent_tree_ad_insert)
+#define	extent_tree_ad_iter JEMALLOC_N(extent_tree_ad_iter)
+#define	extent_tree_ad_iter_recurse JEMALLOC_N(extent_tree_ad_iter_recurse)
+#define	extent_tree_ad_iter_start JEMALLOC_N(extent_tree_ad_iter_start)
+#define	extent_tree_ad_last JEMALLOC_N(extent_tree_ad_last)
+#define	extent_tree_ad_new JEMALLOC_N(extent_tree_ad_new)
+#define	extent_tree_ad_next JEMALLOC_N(extent_tree_ad_next)
+#define	extent_tree_ad_nsearch JEMALLOC_N(extent_tree_ad_nsearch)
+#define	extent_tree_ad_prev JEMALLOC_N(extent_tree_ad_prev)
+#define	extent_tree_ad_psearch JEMALLOC_N(extent_tree_ad_psearch)
+#define	extent_tree_ad_remove JEMALLOC_N(extent_tree_ad_remove)
+#define	extent_tree_ad_reverse_iter JEMALLOC_N(extent_tree_ad_reverse_iter)
+#define	extent_tree_ad_reverse_iter_recurse JEMALLOC_N(extent_tree_ad_reverse_iter_recurse)
+#define	extent_tree_ad_reverse_iter_start JEMALLOC_N(extent_tree_ad_reverse_iter_start)
+#define	extent_tree_ad_search JEMALLOC_N(extent_tree_ad_search)
+#define	extent_tree_szad_first JEMALLOC_N(extent_tree_szad_first)
+#define	extent_tree_szad_insert JEMALLOC_N(extent_tree_szad_insert)
+#define	extent_tree_szad_iter JEMALLOC_N(extent_tree_szad_iter)
+#define	extent_tree_szad_iter_recurse JEMALLOC_N(extent_tree_szad_iter_recurse)
+#define	extent_tree_szad_iter_start JEMALLOC_N(extent_tree_szad_iter_start)
+#define	extent_tree_szad_last JEMALLOC_N(extent_tree_szad_last)
+#define	extent_tree_szad_new JEMALLOC_N(extent_tree_szad_new)
+#define	extent_tree_szad_next JEMALLOC_N(extent_tree_szad_next)
+#define	extent_tree_szad_nsearch JEMALLOC_N(extent_tree_szad_nsearch)
+#define	extent_tree_szad_prev JEMALLOC_N(extent_tree_szad_prev)
+#define	extent_tree_szad_psearch JEMALLOC_N(extent_tree_szad_psearch)
+#define	extent_tree_szad_remove JEMALLOC_N(extent_tree_szad_remove)
+#define	extent_tree_szad_reverse_iter JEMALLOC_N(extent_tree_szad_reverse_iter)
+#define	extent_tree_szad_reverse_iter_recurse JEMALLOC_N(extent_tree_szad_reverse_iter_recurse)
+#define	extent_tree_szad_reverse_iter_start JEMALLOC_N(extent_tree_szad_reverse_iter_start)
+#define	extent_tree_szad_search JEMALLOC_N(extent_tree_szad_search)
+#define	get_errno JEMALLOC_N(get_errno)
+#define	hash JEMALLOC_N(hash)
+#define	hash_fmix_32 JEMALLOC_N(hash_fmix_32)
+#define	hash_fmix_64 JEMALLOC_N(hash_fmix_64)
+#define	hash_get_block_32 JEMALLOC_N(hash_get_block_32)
+#define	hash_get_block_64 JEMALLOC_N(hash_get_block_64)
+#define	hash_rotl_32 JEMALLOC_N(hash_rotl_32)
+#define	hash_rotl_64 JEMALLOC_N(hash_rotl_64)
+#define	hash_x64_128 JEMALLOC_N(hash_x64_128)
+#define	hash_x86_128 JEMALLOC_N(hash_x86_128)
+#define	hash_x86_32 JEMALLOC_N(hash_x86_32)
+#define	huge_allocated JEMALLOC_N(huge_allocated)
+#define	huge_boot JEMALLOC_N(huge_boot)
+#define	huge_dalloc JEMALLOC_N(huge_dalloc)
+#define	huge_malloc JEMALLOC_N(huge_malloc)
+#define	huge_mtx JEMALLOC_N(huge_mtx)
+#define	huge_ndalloc JEMALLOC_N(huge_ndalloc)
+#define	huge_nmalloc JEMALLOC_N(huge_nmalloc)
+#define	huge_palloc JEMALLOC_N(huge_palloc)
+#define	huge_postfork_child JEMALLOC_N(huge_postfork_child)
+#define	huge_postfork_parent JEMALLOC_N(huge_postfork_parent)
+#define	huge_prefork JEMALLOC_N(huge_prefork)
+#define	huge_prof_ctx_get JEMALLOC_N(huge_prof_ctx_get)
+#define	huge_prof_ctx_set JEMALLOC_N(huge_prof_ctx_set)
+#define	huge_ralloc JEMALLOC_N(huge_ralloc)
+#define	huge_ralloc_no_move JEMALLOC_N(huge_ralloc_no_move)
+#define	huge_salloc JEMALLOC_N(huge_salloc)
+#define	iallocm JEMALLOC_N(iallocm)
+#define	icalloc JEMALLOC_N(icalloc)
+#define	icallocx JEMALLOC_N(icallocx)
+#define	idalloc JEMALLOC_N(idalloc)
+#define	idallocx JEMALLOC_N(idallocx)
+#define	imalloc JEMALLOC_N(imalloc)
+#define	imallocx JEMALLOC_N(imallocx)
+#define	ipalloc JEMALLOC_N(ipalloc)
+#define	ipallocx JEMALLOC_N(ipallocx)
+#define	iqalloc JEMALLOC_N(iqalloc)
+#define	iqallocx JEMALLOC_N(iqallocx)
+#define	iralloc JEMALLOC_N(iralloc)
+#define	irallocx JEMALLOC_N(irallocx)
+#define	isalloc JEMALLOC_N(isalloc)
+#define	isthreaded JEMALLOC_N(isthreaded)
+#define	ivsalloc JEMALLOC_N(ivsalloc)
+#define	jemalloc_postfork_child JEMALLOC_N(jemalloc_postfork_child)
+#define	jemalloc_postfork_parent JEMALLOC_N(jemalloc_postfork_parent)
+#define	jemalloc_prefork JEMALLOC_N(jemalloc_prefork)
+#define	malloc_cprintf JEMALLOC_N(malloc_cprintf)
+#define	malloc_mutex_init JEMALLOC_N(malloc_mutex_init)
+#define	malloc_mutex_lock JEMALLOC_N(malloc_mutex_lock)
+#define	malloc_mutex_postfork_child JEMALLOC_N(malloc_mutex_postfork_child)
+#define	malloc_mutex_postfork_parent JEMALLOC_N(malloc_mutex_postfork_parent)
+#define	malloc_mutex_prefork JEMALLOC_N(malloc_mutex_prefork)
+#define	malloc_mutex_unlock JEMALLOC_N(malloc_mutex_unlock)
+#define	malloc_printf JEMALLOC_N(malloc_printf)
+#define	malloc_snprintf JEMALLOC_N(malloc_snprintf)
+#define	malloc_strtoumax JEMALLOC_N(malloc_strtoumax)
+#define	malloc_tsd_boot JEMALLOC_N(malloc_tsd_boot)
+#define	malloc_tsd_cleanup_register JEMALLOC_N(malloc_tsd_cleanup_register)
+#define	malloc_tsd_dalloc JEMALLOC_N(malloc_tsd_dalloc)
+#define	malloc_tsd_malloc JEMALLOC_N(malloc_tsd_malloc)
+#define	malloc_tsd_no_cleanup JEMALLOC_N(malloc_tsd_no_cleanup)
+#define	malloc_vcprintf JEMALLOC_N(malloc_vcprintf)
+#define	malloc_vsnprintf JEMALLOC_N(malloc_vsnprintf)
+#define	malloc_write JEMALLOC_N(malloc_write)
+#define	map_bias JEMALLOC_N(map_bias)
+#define	mb_write JEMALLOC_N(mb_write)
+#define	mutex_boot JEMALLOC_N(mutex_boot)
+#define	narenas_auto JEMALLOC_N(narenas_auto)
+#define	narenas_total JEMALLOC_N(narenas_total)
+#define	narenas_total_get JEMALLOC_N(narenas_total_get)
+#define	ncpus JEMALLOC_N(ncpus)
+#define	nhbins JEMALLOC_N(nhbins)
+#define	opt_abort JEMALLOC_N(opt_abort)
+#define	opt_junk JEMALLOC_N(opt_junk)
+#define	opt_lg_chunk JEMALLOC_N(opt_lg_chunk)
+#define	opt_lg_dirty_mult JEMALLOC_N(opt_lg_dirty_mult)
+#define	opt_lg_prof_interval JEMALLOC_N(opt_lg_prof_interval)
+#define	opt_lg_prof_sample JEMALLOC_N(opt_lg_prof_sample)
+#define	opt_lg_tcache_max JEMALLOC_N(opt_lg_tcache_max)
+#define	opt_narenas JEMALLOC_N(opt_narenas)
+#define	opt_prof JEMALLOC_N(opt_prof)
+#define	opt_prof_accum JEMALLOC_N(opt_prof_accum)
+#define	opt_prof_active JEMALLOC_N(opt_prof_active)
+#define	opt_prof_final JEMALLOC_N(opt_prof_final)
+#define	opt_prof_gdump JEMALLOC_N(opt_prof_gdump)
+#define	opt_prof_leak JEMALLOC_N(opt_prof_leak)
+#define	opt_prof_prefix JEMALLOC_N(opt_prof_prefix)
+#define	opt_quarantine JEMALLOC_N(opt_quarantine)
+#define	opt_redzone JEMALLOC_N(opt_redzone)
+#define	opt_stats_print JEMALLOC_N(opt_stats_print)
+#define	opt_tcache JEMALLOC_N(opt_tcache)
+#define	opt_utrace JEMALLOC_N(opt_utrace)
+#define	opt_valgrind JEMALLOC_N(opt_valgrind)
+#define	opt_xmalloc JEMALLOC_N(opt_xmalloc)
+#define	opt_zero JEMALLOC_N(opt_zero)
+#define	p2rz JEMALLOC_N(p2rz)
+#define	pages_purge JEMALLOC_N(pages_purge)
+#define	pow2_ceil JEMALLOC_N(pow2_ceil)
+#define	prof_backtrace JEMALLOC_N(prof_backtrace)
+#define	prof_boot0 JEMALLOC_N(prof_boot0)
+#define	prof_boot1 JEMALLOC_N(prof_boot1)
+#define	prof_boot2 JEMALLOC_N(prof_boot2)
+#define	prof_ctx_get JEMALLOC_N(prof_ctx_get)
+#define	prof_ctx_set JEMALLOC_N(prof_ctx_set)
+#define	prof_free JEMALLOC_N(prof_free)
+#define	prof_gdump JEMALLOC_N(prof_gdump)
+#define	prof_idump JEMALLOC_N(prof_idump)
+#define	prof_interval JEMALLOC_N(prof_interval)
+#define	prof_lookup JEMALLOC_N(prof_lookup)
+#define	prof_malloc JEMALLOC_N(prof_malloc)
+#define	prof_mdump JEMALLOC_N(prof_mdump)
+#define	prof_postfork_child JEMALLOC_N(prof_postfork_child)
+#define	prof_postfork_parent JEMALLOC_N(prof_postfork_parent)
+#define	prof_prefork JEMALLOC_N(prof_prefork)
+#define	prof_promote JEMALLOC_N(prof_promote)
+#define	prof_realloc JEMALLOC_N(prof_realloc)
+#define	prof_sample_accum_update JEMALLOC_N(prof_sample_accum_update)
+#define	prof_sample_threshold_update JEMALLOC_N(prof_sample_threshold_update)
+#define	prof_tdata_booted JEMALLOC_N(prof_tdata_booted)
+#define	prof_tdata_cleanup JEMALLOC_N(prof_tdata_cleanup)
+#define	prof_tdata_get JEMALLOC_N(prof_tdata_get)
+#define	prof_tdata_init JEMALLOC_N(prof_tdata_init)
+#define	prof_tdata_initialized JEMALLOC_N(prof_tdata_initialized)
+#define	prof_tdata_tls JEMALLOC_N(prof_tdata_tls)
+#define	prof_tdata_tsd JEMALLOC_N(prof_tdata_tsd)
+#define	prof_tdata_tsd_boot JEMALLOC_N(prof_tdata_tsd_boot)
+#define	prof_tdata_tsd_cleanup_wrapper JEMALLOC_N(prof_tdata_tsd_cleanup_wrapper)
+#define	prof_tdata_tsd_get JEMALLOC_N(prof_tdata_tsd_get)
+#define	prof_tdata_tsd_get_wrapper JEMALLOC_N(prof_tdata_tsd_get_wrapper)
+#define	prof_tdata_tsd_set JEMALLOC_N(prof_tdata_tsd_set)
+#define	quarantine JEMALLOC_N(quarantine)
+#define	quarantine_alloc_hook JEMALLOC_N(quarantine_alloc_hook)
+#define	quarantine_boot JEMALLOC_N(quarantine_boot)
+#define	quarantine_booted JEMALLOC_N(quarantine_booted)
+#define	quarantine_cleanup JEMALLOC_N(quarantine_cleanup)
+#define	quarantine_init JEMALLOC_N(quarantine_init)
+#define	quarantine_tls JEMALLOC_N(quarantine_tls)
+#define	quarantine_tsd JEMALLOC_N(quarantine_tsd)
+#define	quarantine_tsd_boot JEMALLOC_N(quarantine_tsd_boot)
+#define	quarantine_tsd_cleanup_wrapper JEMALLOC_N(quarantine_tsd_cleanup_wrapper)
+#define	quarantine_tsd_get JEMALLOC_N(quarantine_tsd_get)
+#define	quarantine_tsd_get_wrapper JEMALLOC_N(quarantine_tsd_get_wrapper)
+#define	quarantine_tsd_set JEMALLOC_N(quarantine_tsd_set)
+#define	register_zone JEMALLOC_N(register_zone)
+#define	rtree_get JEMALLOC_N(rtree_get)
+#define	rtree_get_locked JEMALLOC_N(rtree_get_locked)
+#define	rtree_new JEMALLOC_N(rtree_new)
+#define	rtree_postfork_child JEMALLOC_N(rtree_postfork_child)
+#define	rtree_postfork_parent JEMALLOC_N(rtree_postfork_parent)
+#define	rtree_prefork JEMALLOC_N(rtree_prefork)
+#define	rtree_set JEMALLOC_N(rtree_set)
+#define	s2u JEMALLOC_N(s2u)
+#define	sa2u JEMALLOC_N(sa2u)
+#define	set_errno JEMALLOC_N(set_errno)
+#define	stats_cactive JEMALLOC_N(stats_cactive)
+#define	stats_cactive_add JEMALLOC_N(stats_cactive_add)
+#define	stats_cactive_get JEMALLOC_N(stats_cactive_get)
+#define	stats_cactive_sub JEMALLOC_N(stats_cactive_sub)
+#define	stats_chunks JEMALLOC_N(stats_chunks)
+#define	stats_print JEMALLOC_N(stats_print)
+#define	tcache_alloc_easy JEMALLOC_N(tcache_alloc_easy)
+#define	tcache_alloc_large JEMALLOC_N(tcache_alloc_large)
+#define	tcache_alloc_small JEMALLOC_N(tcache_alloc_small)
+#define	tcache_alloc_small_hard JEMALLOC_N(tcache_alloc_small_hard)
+#define	tcache_arena_associate JEMALLOC_N(tcache_arena_associate)
+#define	tcache_arena_dissociate JEMALLOC_N(tcache_arena_dissociate)
+#define	tcache_bin_flush_large JEMALLOC_N(tcache_bin_flush_large)
+#define	tcache_bin_flush_small JEMALLOC_N(tcache_bin_flush_small)
+#define	tcache_bin_info JEMALLOC_N(tcache_bin_info)
+#define	tcache_boot0 JEMALLOC_N(tcache_boot0)
+#define	tcache_boot1 JEMALLOC_N(tcache_boot1)
+#define	tcache_booted JEMALLOC_N(tcache_booted)
+#define	tcache_create JEMALLOC_N(tcache_create)
+#define	tcache_dalloc_large JEMALLOC_N(tcache_dalloc_large)
+#define	tcache_dalloc_small JEMALLOC_N(tcache_dalloc_small)
+#define	tcache_destroy JEMALLOC_N(tcache_destroy)
+#define	tcache_enabled_booted JEMALLOC_N(tcache_enabled_booted)
+#define	tcache_enabled_get JEMALLOC_N(tcache_enabled_get)
+#define	tcache_enabled_initialized JEMALLOC_N(tcache_enabled_initialized)
+#define	tcache_enabled_set JEMALLOC_N(tcache_enabled_set)
+#define	tcache_enabled_tls JEMALLOC_N(tcache_enabled_tls)
+#define	tcache_enabled_tsd JEMALLOC_N(tcache_enabled_tsd)
+#define	tcache_enabled_tsd_boot JEMALLOC_N(tcache_enabled_tsd_boot)
+#define	tcache_enabled_tsd_cleanup_wrapper JEMALLOC_N(tcache_enabled_tsd_cleanup_wrapper)
+#define	tcache_enabled_tsd_get JEMALLOC_N(tcache_enabled_tsd_get)
+#define	tcache_enabled_tsd_get_wrapper JEMALLOC_N(tcache_enabled_tsd_get_wrapper)
+#define	tcache_enabled_tsd_set JEMALLOC_N(tcache_enabled_tsd_set)
+#define	tcache_event JEMALLOC_N(tcache_event)
+#define	tcache_event_hard JEMALLOC_N(tcache_event_hard)
+#define	tcache_flush JEMALLOC_N(tcache_flush)
+#define	tcache_get JEMALLOC_N(tcache_get)
+#define	tcache_initialized JEMALLOC_N(tcache_initialized)
+#define	tcache_maxclass JEMALLOC_N(tcache_maxclass)
+#define	tcache_salloc JEMALLOC_N(tcache_salloc)
+#define	tcache_stats_merge JEMALLOC_N(tcache_stats_merge)
+#define	tcache_thread_cleanup JEMALLOC_N(tcache_thread_cleanup)
+#define	tcache_tls JEMALLOC_N(tcache_tls)
+#define	tcache_tsd JEMALLOC_N(tcache_tsd)
+#define	tcache_tsd_boot JEMALLOC_N(tcache_tsd_boot)
+#define	tcache_tsd_cleanup_wrapper JEMALLOC_N(tcache_tsd_cleanup_wrapper)
+#define	tcache_tsd_get JEMALLOC_N(tcache_tsd_get)
+#define	tcache_tsd_get_wrapper JEMALLOC_N(tcache_tsd_get_wrapper)
+#define	tcache_tsd_set JEMALLOC_N(tcache_tsd_set)
+#define	thread_allocated_booted JEMALLOC_N(thread_allocated_booted)
+#define	thread_allocated_initialized JEMALLOC_N(thread_allocated_initialized)
+#define	thread_allocated_tls JEMALLOC_N(thread_allocated_tls)
+#define	thread_allocated_tsd JEMALLOC_N(thread_allocated_tsd)
+#define	thread_allocated_tsd_boot JEMALLOC_N(thread_allocated_tsd_boot)
+#define	thread_allocated_tsd_cleanup_wrapper JEMALLOC_N(thread_allocated_tsd_cleanup_wrapper)
+#define	thread_allocated_tsd_get JEMALLOC_N(thread_allocated_tsd_get)
+#define	thread_allocated_tsd_get_wrapper JEMALLOC_N(thread_allocated_tsd_get_wrapper)
+#define	thread_allocated_tsd_set JEMALLOC_N(thread_allocated_tsd_set)
+#define	u2rz JEMALLOC_N(u2rz)
diff --git a/extra/jemalloc/include/jemalloc/internal/prng.h b/extra/jemalloc/include/jemalloc/internal/prng.h
new file mode 100644
index 00000000000..83a5462b4dd
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/prng.h
@@ -0,0 +1,60 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+/*
+ * Simple linear congruential pseudo-random number generator:
+ *
+ *   prng(y) = (a*x + c) % m
+ *
+ * where the following constants ensure maximal period:
+ *
+ *   a == Odd number (relatively prime to 2^n), and (a-1) is a multiple of 4.
+ *   c == Odd number (relatively prime to 2^n).
+ *   m == 2^32
+ *
+ * See Knuth's TAOCP 3rd Ed., Vol. 2, pg. 17 for details on these constraints.
+ *
+ * This choice of m has the disadvantage that the quality of the bits is
+ * proportional to bit position.  For example. the lowest bit has a cycle of 2,
+ * the next has a cycle of 4, etc.  For this reason, we prefer to use the upper
+ * bits.
+ *
+ * Macro parameters:
+ *   uint32_t r          : Result.
+ *   unsigned lg_range   : (0..32], number of least significant bits to return.
+ *   uint32_t state      : Seed value.
+ *   const uint32_t a, c : See above discussion.
+ */
+#define prng32(r, lg_range, state, a, c) do {				\
+	assert(lg_range > 0);						\
+	assert(lg_range <= 32);						\
+									\
+	r = (state * (a)) + (c);					\
+	state = r;							\
+	r >>= (32 - lg_range);						\
+} while (false)
+
+/* Same as prng32(), but 64 bits of pseudo-randomness, using uint64_t. */
+#define prng64(r, lg_range, state, a, c) do {				\
+	assert(lg_range > 0);						\
+	assert(lg_range <= 64);						\
+									\
+	r = (state * (a)) + (c);					\
+	state = r;							\
+	r >>= (64 - lg_range);						\
+} while (false)
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/internal/prof.h b/extra/jemalloc/include/jemalloc/internal/prof.h
new file mode 100644
index 00000000000..119a5b1bcb7
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/prof.h
@@ -0,0 +1,579 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+typedef struct prof_bt_s prof_bt_t;
+typedef struct prof_cnt_s prof_cnt_t;
+typedef struct prof_thr_cnt_s prof_thr_cnt_t;
+typedef struct prof_ctx_s prof_ctx_t;
+typedef struct prof_tdata_s prof_tdata_t;
+
+/* Option defaults. */
+#define	PROF_PREFIX_DEFAULT		"jeprof"
+#define	LG_PROF_SAMPLE_DEFAULT		19
+#define	LG_PROF_INTERVAL_DEFAULT	-1
+
+/*
+ * Hard limit on stack backtrace depth.  The version of prof_backtrace() that
+ * is based on __builtin_return_address() necessarily has a hard-coded number
+ * of backtrace frame handlers, and should be kept in sync with this setting.
+ */
+#define	PROF_BT_MAX			128
+
+/* Maximum number of backtraces to store in each per thread LRU cache. */
+#define	PROF_TCMAX			1024
+
+/* Initial hash table size. */
+#define	PROF_CKH_MINITEMS		64
+
+/* Size of memory buffer to use when writing dump files. */
+#define	PROF_DUMP_BUFSIZE		65536
+
+/* Size of stack-allocated buffer used by prof_printf(). */
+#define	PROF_PRINTF_BUFSIZE		128
+
+/*
+ * Number of mutexes shared among all ctx's.  No space is allocated for these
+ * unless profiling is enabled, so it's okay to over-provision.
+ */
+#define	PROF_NCTX_LOCKS			1024
+
+/*
+ * prof_tdata pointers close to NULL are used to encode state information that
+ * is used for cleaning up during thread shutdown.
+ */
+#define	PROF_TDATA_STATE_REINCARNATED	((prof_tdata_t *)(uintptr_t)1)
+#define	PROF_TDATA_STATE_PURGATORY	((prof_tdata_t *)(uintptr_t)2)
+#define	PROF_TDATA_STATE_MAX		PROF_TDATA_STATE_PURGATORY
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+struct prof_bt_s {
+	/* Backtrace, stored as len program counters. */
+	void		**vec;
+	unsigned	len;
+};
+
+#ifdef JEMALLOC_PROF_LIBGCC
+/* Data structure passed to libgcc _Unwind_Backtrace() callback functions. */
+typedef struct {
+	prof_bt_t	*bt;
+	unsigned	nignore;
+	unsigned	max;
+} prof_unwind_data_t;
+#endif
+
+struct prof_cnt_s {
+	/*
+	 * Profiling counters.  An allocation/deallocation pair can operate on
+	 * different prof_thr_cnt_t objects that are linked into the same
+	 * prof_ctx_t cnts_ql, so it is possible for the cur* counters to go
+	 * negative.  In principle it is possible for the *bytes counters to
+	 * overflow/underflow, but a general solution would require something
+	 * like 128-bit counters; this implementation doesn't bother to solve
+	 * that problem.
+	 */
+	int64_t		curobjs;
+	int64_t		curbytes;
+	uint64_t	accumobjs;
+	uint64_t	accumbytes;
+};
+
+struct prof_thr_cnt_s {
+	/* Linkage into prof_ctx_t's cnts_ql. */
+	ql_elm(prof_thr_cnt_t)	cnts_link;
+
+	/* Linkage into thread's LRU. */
+	ql_elm(prof_thr_cnt_t)	lru_link;
+
+	/*
+	 * Associated context.  If a thread frees an object that it did not
+	 * allocate, it is possible that the context is not cached in the
+	 * thread's hash table, in which case it must be able to look up the
+	 * context, insert a new prof_thr_cnt_t into the thread's hash table,
+	 * and link it into the prof_ctx_t's cnts_ql.
+	 */
+	prof_ctx_t		*ctx;
+
+	/*
+	 * Threads use memory barriers to update the counters.  Since there is
+	 * only ever one writer, the only challenge is for the reader to get a
+	 * consistent read of the counters.
+	 *
+	 * The writer uses this series of operations:
+	 *
+	 * 1) Increment epoch to an odd number.
+	 * 2) Update counters.
+	 * 3) Increment epoch to an even number.
+	 *
+	 * The reader must assure 1) that the epoch is even while it reads the
+	 * counters, and 2) that the epoch doesn't change between the time it
+	 * starts and finishes reading the counters.
+	 */
+	unsigned		epoch;
+
+	/* Profiling counters. */
+	prof_cnt_t		cnts;
+};
+
+struct prof_ctx_s {
+	/* Associated backtrace. */
+	prof_bt_t		*bt;
+
+	/* Protects nlimbo, cnt_merged, and cnts_ql. */
+	malloc_mutex_t		*lock;
+
+	/*
+	 * Number of threads that currently cause this ctx to be in a state of
+	 * limbo due to one of:
+	 *   - Initializing per thread counters associated with this ctx.
+	 *   - Preparing to destroy this ctx.
+	 * nlimbo must be 1 (single destroyer) in order to safely destroy the
+	 * ctx.
+	 */
+	unsigned		nlimbo;
+
+	/* Temporary storage for summation during dump. */
+	prof_cnt_t		cnt_summed;
+
+	/* When threads exit, they merge their stats into cnt_merged. */
+	prof_cnt_t		cnt_merged;
+
+	/*
+	 * List of profile counters, one for each thread that has allocated in
+	 * this context.
+	 */
+	ql_head(prof_thr_cnt_t)	cnts_ql;
+};
+
+struct prof_tdata_s {
+	/*
+	 * Hash of (prof_bt_t *)-->(prof_thr_cnt_t *).  Each thread keeps a
+	 * cache of backtraces, with associated thread-specific prof_thr_cnt_t
+	 * objects.  Other threads may read the prof_thr_cnt_t contents, but no
+	 * others will ever write them.
+	 *
+	 * Upon thread exit, the thread must merge all the prof_thr_cnt_t
+	 * counter data into the associated prof_ctx_t objects, and unlink/free
+	 * the prof_thr_cnt_t objects.
+	 */
+	ckh_t			bt2cnt;
+
+	/* LRU for contents of bt2cnt. */
+	ql_head(prof_thr_cnt_t)	lru_ql;
+
+	/* Backtrace vector, used for calls to prof_backtrace(). */
+	void			**vec;
+
+	/* Sampling state. */
+	uint64_t		prng_state;
+	uint64_t		threshold;
+	uint64_t		accum;
+
+	/* State used to avoid dumping while operating on prof internals. */
+	bool			enq;
+	bool			enq_idump;
+	bool			enq_gdump;
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+extern bool	opt_prof;
+/*
+ * Even if opt_prof is true, sampling can be temporarily disabled by setting
+ * opt_prof_active to false.  No locking is used when updating opt_prof_active,
+ * so there are no guarantees regarding how long it will take for all threads
+ * to notice state changes.
+ */
+extern bool	opt_prof_active;
+extern size_t	opt_lg_prof_sample;   /* Mean bytes between samples. */
+extern ssize_t	opt_lg_prof_interval; /* lg(prof_interval). */
+extern bool	opt_prof_gdump;       /* High-water memory dumping. */
+extern bool	opt_prof_final;       /* Final profile dumping. */
+extern bool	opt_prof_leak;        /* Dump leak summary at exit. */
+extern bool	opt_prof_accum;       /* Report cumulative bytes. */
+extern char	opt_prof_prefix[PATH_MAX + 1];
+
+/*
+ * Profile dump interval, measured in bytes allocated.  Each arena triggers a
+ * profile dump when it reaches this threshold.  The effect is that the
+ * interval between profile dumps averages prof_interval, though the actual
+ * interval between dumps will tend to be sporadic, and the interval will be a
+ * maximum of approximately (prof_interval * narenas).
+ */
+extern uint64_t	prof_interval;
+
+/*
+ * If true, promote small sampled objects to large objects, since small run
+ * headers do not have embedded profile context pointers.
+ */
+extern bool	prof_promote;
+
+void	bt_init(prof_bt_t *bt, void **vec);
+void	prof_backtrace(prof_bt_t *bt, unsigned nignore);
+prof_thr_cnt_t	*prof_lookup(prof_bt_t *bt);
+void	prof_idump(void);
+bool	prof_mdump(const char *filename);
+void	prof_gdump(void);
+prof_tdata_t	*prof_tdata_init(void);
+void	prof_tdata_cleanup(void *arg);
+void	prof_boot0(void);
+void	prof_boot1(void);
+bool	prof_boot2(void);
+void	prof_prefork(void);
+void	prof_postfork_parent(void);
+void	prof_postfork_child(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#define	PROF_ALLOC_PREP(nignore, size, ret) do {			\
+	prof_tdata_t *prof_tdata;					\
+	prof_bt_t bt;							\
+									\
+	assert(size == s2u(size));					\
+									\
+	prof_tdata = prof_tdata_get(true);				\
+	if ((uintptr_t)prof_tdata <= (uintptr_t)PROF_TDATA_STATE_MAX) {	\
+		if (prof_tdata != NULL)					\
+			ret = (prof_thr_cnt_t *)(uintptr_t)1U;		\
+		else							\
+			ret = NULL;					\
+		break;							\
+	}								\
+									\
+	if (opt_prof_active == false) {					\
+		/* Sampling is currently inactive, so avoid sampling. */\
+		ret = (prof_thr_cnt_t *)(uintptr_t)1U;			\
+	} else if (opt_lg_prof_sample == 0) {				\
+		/* Don't bother with sampling logic, since sampling   */\
+		/* interval is 1.                                     */\
+		bt_init(&bt, prof_tdata->vec);				\
+		prof_backtrace(&bt, nignore);				\
+		ret = prof_lookup(&bt);					\
+	} else {							\
+		if (prof_tdata->threshold == 0) {			\
+			/* Initialize.  Seed the prng differently for */\
+			/* each thread.                               */\
+			prof_tdata->prng_state =			\
+			    (uint64_t)(uintptr_t)&size;			\
+			prof_sample_threshold_update(prof_tdata);	\
+		}							\
+									\
+		/* Determine whether to capture a backtrace based on  */\
+		/* whether size is enough for prof_accum to reach     */\
+		/* prof_tdata->threshold.  However, delay updating    */\
+		/* these variables until prof_{m,re}alloc(), because  */\
+		/* we don't know for sure that the allocation will    */\
+		/* succeed.                                           */\
+		/*                                                    */\
+		/* Use subtraction rather than addition to avoid      */\
+		/* potential integer overflow.                        */\
+		if (size >= prof_tdata->threshold -			\
+		    prof_tdata->accum) {				\
+			bt_init(&bt, prof_tdata->vec);			\
+			prof_backtrace(&bt, nignore);			\
+			ret = prof_lookup(&bt);				\
+		} else							\
+			ret = (prof_thr_cnt_t *)(uintptr_t)1U;		\
+	}								\
+} while (0)
+
+#ifndef JEMALLOC_ENABLE_INLINE
+malloc_tsd_protos(JEMALLOC_ATTR(unused), prof_tdata, prof_tdata_t *)
+
+prof_tdata_t	*prof_tdata_get(bool create);
+void	prof_sample_threshold_update(prof_tdata_t *prof_tdata);
+prof_ctx_t	*prof_ctx_get(const void *ptr);
+void	prof_ctx_set(const void *ptr, prof_ctx_t *ctx);
+bool	prof_sample_accum_update(size_t size);
+void	prof_malloc(const void *ptr, size_t size, prof_thr_cnt_t *cnt);
+void	prof_realloc(const void *ptr, size_t size, prof_thr_cnt_t *cnt,
+    size_t old_size, prof_ctx_t *old_ctx);
+void	prof_free(const void *ptr, size_t size);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_PROF_C_))
+/* Thread-specific backtrace cache, used to reduce bt2ctx contention. */
+malloc_tsd_externs(prof_tdata, prof_tdata_t *)
+malloc_tsd_funcs(JEMALLOC_INLINE, prof_tdata, prof_tdata_t *, NULL,
+    prof_tdata_cleanup)
+
+JEMALLOC_INLINE prof_tdata_t *
+prof_tdata_get(bool create)
+{
+	prof_tdata_t *prof_tdata;
+
+	cassert(config_prof);
+
+	prof_tdata = *prof_tdata_tsd_get();
+	if (create && prof_tdata == NULL)
+		prof_tdata = prof_tdata_init();
+
+	return (prof_tdata);
+}
+
+JEMALLOC_INLINE void
+prof_sample_threshold_update(prof_tdata_t *prof_tdata)
+{
+	uint64_t r;
+	double u;
+
+	cassert(config_prof);
+
+	/*
+	 * Compute sample threshold as a geometrically distributed random
+	 * variable with mean (2^opt_lg_prof_sample).
+	 *
+	 *                         __        __
+	 *                         |  log(u)  |                     1
+	 * prof_tdata->threshold = | -------- |, where p = -------------------
+	 *                         | log(1-p) |             opt_lg_prof_sample
+	 *                                                 2
+	 *
+	 * For more information on the math, see:
+	 *
+	 *   Non-Uniform Random Variate Generation
+	 *   Luc Devroye
+	 *   Springer-Verlag, New York, 1986
+	 *   pp 500
+	 *   (http://cg.scs.carleton.ca/~luc/rnbookindex.html)
+	 */
+	prng64(r, 53, prof_tdata->prng_state,
+	    UINT64_C(6364136223846793005), UINT64_C(1442695040888963407));
+	u = (double)r * (1.0/9007199254740992.0L);
+	prof_tdata->threshold = (uint64_t)(log(u) /
+	    log(1.0 - (1.0 / (double)((uint64_t)1U << opt_lg_prof_sample))))
+	    + (uint64_t)1U;
+}
+
+JEMALLOC_INLINE prof_ctx_t *
+prof_ctx_get(const void *ptr)
+{
+	prof_ctx_t *ret;
+	arena_chunk_t *chunk;
+
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+	if (chunk != ptr) {
+		/* Region. */
+		ret = arena_prof_ctx_get(ptr);
+	} else
+		ret = huge_prof_ctx_get(ptr);
+
+	return (ret);
+}
+
+JEMALLOC_INLINE void
+prof_ctx_set(const void *ptr, prof_ctx_t *ctx)
+{
+	arena_chunk_t *chunk;
+
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+	if (chunk != ptr) {
+		/* Region. */
+		arena_prof_ctx_set(ptr, ctx);
+	} else
+		huge_prof_ctx_set(ptr, ctx);
+}
+
+JEMALLOC_INLINE bool
+prof_sample_accum_update(size_t size)
+{
+	prof_tdata_t *prof_tdata;
+
+	cassert(config_prof);
+	/* Sampling logic is unnecessary if the interval is 1. */
+	assert(opt_lg_prof_sample != 0);
+
+	prof_tdata = prof_tdata_get(false);
+	if ((uintptr_t)prof_tdata <= (uintptr_t)PROF_TDATA_STATE_MAX)
+		return (true);
+
+	/* Take care to avoid integer overflow. */
+	if (size >= prof_tdata->threshold - prof_tdata->accum) {
+		prof_tdata->accum -= (prof_tdata->threshold - size);
+		/* Compute new sample threshold. */
+		prof_sample_threshold_update(prof_tdata);
+		while (prof_tdata->accum >= prof_tdata->threshold) {
+			prof_tdata->accum -= prof_tdata->threshold;
+			prof_sample_threshold_update(prof_tdata);
+		}
+		return (false);
+	} else {
+		prof_tdata->accum += size;
+		return (true);
+	}
+}
+
+JEMALLOC_INLINE void
+prof_malloc(const void *ptr, size_t size, prof_thr_cnt_t *cnt)
+{
+
+	cassert(config_prof);
+	assert(ptr != NULL);
+	assert(size == isalloc(ptr, true));
+
+	if (opt_lg_prof_sample != 0) {
+		if (prof_sample_accum_update(size)) {
+			/*
+			 * Don't sample.  For malloc()-like allocation, it is
+			 * always possible to tell in advance how large an
+			 * object's usable size will be, so there should never
+			 * be a difference between the size passed to
+			 * PROF_ALLOC_PREP() and prof_malloc().
+			 */
+			assert((uintptr_t)cnt == (uintptr_t)1U);
+		}
+	}
+
+	if ((uintptr_t)cnt > (uintptr_t)1U) {
+		prof_ctx_set(ptr, cnt->ctx);
+
+		cnt->epoch++;
+		/*********/
+		mb_write();
+		/*********/
+		cnt->cnts.curobjs++;
+		cnt->cnts.curbytes += size;
+		if (opt_prof_accum) {
+			cnt->cnts.accumobjs++;
+			cnt->cnts.accumbytes += size;
+		}
+		/*********/
+		mb_write();
+		/*********/
+		cnt->epoch++;
+		/*********/
+		mb_write();
+		/*********/
+	} else
+		prof_ctx_set(ptr, (prof_ctx_t *)(uintptr_t)1U);
+}
+
+JEMALLOC_INLINE void
+prof_realloc(const void *ptr, size_t size, prof_thr_cnt_t *cnt,
+    size_t old_size, prof_ctx_t *old_ctx)
+{
+	prof_thr_cnt_t *told_cnt;
+
+	cassert(config_prof);
+	assert(ptr != NULL || (uintptr_t)cnt <= (uintptr_t)1U);
+
+	if (ptr != NULL) {
+		assert(size == isalloc(ptr, true));
+		if (opt_lg_prof_sample != 0) {
+			if (prof_sample_accum_update(size)) {
+				/*
+				 * Don't sample.  The size passed to
+				 * PROF_ALLOC_PREP() was larger than what
+				 * actually got allocated, so a backtrace was
+				 * captured for this allocation, even though
+				 * its actual size was insufficient to cross
+				 * the sample threshold.
+				 */
+				cnt = (prof_thr_cnt_t *)(uintptr_t)1U;
+			}
+		}
+	}
+
+	if ((uintptr_t)old_ctx > (uintptr_t)1U) {
+		told_cnt = prof_lookup(old_ctx->bt);
+		if (told_cnt == NULL) {
+			/*
+			 * It's too late to propagate OOM for this realloc(),
+			 * so operate directly on old_cnt->ctx->cnt_merged.
+			 */
+			malloc_mutex_lock(old_ctx->lock);
+			old_ctx->cnt_merged.curobjs--;
+			old_ctx->cnt_merged.curbytes -= old_size;
+			malloc_mutex_unlock(old_ctx->lock);
+			told_cnt = (prof_thr_cnt_t *)(uintptr_t)1U;
+		}
+	} else
+		told_cnt = (prof_thr_cnt_t *)(uintptr_t)1U;
+
+	if ((uintptr_t)told_cnt > (uintptr_t)1U)
+		told_cnt->epoch++;
+	if ((uintptr_t)cnt > (uintptr_t)1U) {
+		prof_ctx_set(ptr, cnt->ctx);
+		cnt->epoch++;
+	} else if (ptr != NULL)
+		prof_ctx_set(ptr, (prof_ctx_t *)(uintptr_t)1U);
+	/*********/
+	mb_write();
+	/*********/
+	if ((uintptr_t)told_cnt > (uintptr_t)1U) {
+		told_cnt->cnts.curobjs--;
+		told_cnt->cnts.curbytes -= old_size;
+	}
+	if ((uintptr_t)cnt > (uintptr_t)1U) {
+		cnt->cnts.curobjs++;
+		cnt->cnts.curbytes += size;
+		if (opt_prof_accum) {
+			cnt->cnts.accumobjs++;
+			cnt->cnts.accumbytes += size;
+		}
+	}
+	/*********/
+	mb_write();
+	/*********/
+	if ((uintptr_t)told_cnt > (uintptr_t)1U)
+		told_cnt->epoch++;
+	if ((uintptr_t)cnt > (uintptr_t)1U)
+		cnt->epoch++;
+	/*********/
+	mb_write(); /* Not strictly necessary. */
+}
+
+JEMALLOC_INLINE void
+prof_free(const void *ptr, size_t size)
+{
+	prof_ctx_t *ctx = prof_ctx_get(ptr);
+
+	cassert(config_prof);
+
+	if ((uintptr_t)ctx > (uintptr_t)1) {
+		prof_thr_cnt_t *tcnt;
+		assert(size == isalloc(ptr, true));
+		tcnt = prof_lookup(ctx->bt);
+
+		if (tcnt != NULL) {
+			tcnt->epoch++;
+			/*********/
+			mb_write();
+			/*********/
+			tcnt->cnts.curobjs--;
+			tcnt->cnts.curbytes -= size;
+			/*********/
+			mb_write();
+			/*********/
+			tcnt->epoch++;
+			/*********/
+			mb_write();
+			/*********/
+		} else {
+			/*
+			 * OOM during free() cannot be propagated, so operate
+			 * directly on cnt->ctx->cnt_merged.
+			 */
+			malloc_mutex_lock(ctx->lock);
+			ctx->cnt_merged.curobjs--;
+			ctx->cnt_merged.curbytes -= size;
+			malloc_mutex_unlock(ctx->lock);
+		}
+	}
+}
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/internal/ql.h b/extra/jemalloc/include/jemalloc/internal/ql.h
new file mode 100644
index 00000000000..a9ed2393f0c
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/ql.h
@@ -0,0 +1,83 @@
+/*
+ * List definitions.
+ */
+#define ql_head(a_type)							\
+struct {								\
+	a_type *qlh_first;						\
+}
+
+#define ql_head_initializer(a_head) {NULL}
+
+#define ql_elm(a_type)	qr(a_type)
+
+/* List functions. */
+#define ql_new(a_head) do {						\
+	(a_head)->qlh_first = NULL;					\
+} while (0)
+
+#define ql_elm_new(a_elm, a_field) qr_new((a_elm), a_field)
+
+#define ql_first(a_head) ((a_head)->qlh_first)
+
+#define ql_last(a_head, a_field)					\
+	((ql_first(a_head) != NULL)					\
+	    ? qr_prev(ql_first(a_head), a_field) : NULL)
+
+#define ql_next(a_head, a_elm, a_field)					\
+	((ql_last(a_head, a_field) != (a_elm))				\
+	    ? qr_next((a_elm), a_field)	: NULL)
+
+#define ql_prev(a_head, a_elm, a_field)					\
+	((ql_first(a_head) != (a_elm)) ? qr_prev((a_elm), a_field)	\
+				       : NULL)
+
+#define ql_before_insert(a_head, a_qlelm, a_elm, a_field) do {		\
+	qr_before_insert((a_qlelm), (a_elm), a_field);			\
+	if (ql_first(a_head) == (a_qlelm)) {				\
+		ql_first(a_head) = (a_elm);				\
+	}								\
+} while (0)
+
+#define ql_after_insert(a_qlelm, a_elm, a_field)			\
+	qr_after_insert((a_qlelm), (a_elm), a_field)
+
+#define ql_head_insert(a_head, a_elm, a_field) do {			\
+	if (ql_first(a_head) != NULL) {					\
+		qr_before_insert(ql_first(a_head), (a_elm), a_field);	\
+	}								\
+	ql_first(a_head) = (a_elm);					\
+} while (0)
+
+#define ql_tail_insert(a_head, a_elm, a_field) do {			\
+	if (ql_first(a_head) != NULL) {					\
+		qr_before_insert(ql_first(a_head), (a_elm), a_field);	\
+	}								\
+	ql_first(a_head) = qr_next((a_elm), a_field);			\
+} while (0)
+
+#define ql_remove(a_head, a_elm, a_field) do {				\
+	if (ql_first(a_head) == (a_elm)) {				\
+		ql_first(a_head) = qr_next(ql_first(a_head), a_field);	\
+	}								\
+	if (ql_first(a_head) != (a_elm)) {				\
+		qr_remove((a_elm), a_field);				\
+	} else {							\
+		ql_first(a_head) = NULL;				\
+	}								\
+} while (0)
+
+#define ql_head_remove(a_head, a_type, a_field) do {			\
+	a_type *t = ql_first(a_head);					\
+	ql_remove((a_head), t, a_field);				\
+} while (0)
+
+#define ql_tail_remove(a_head, a_type, a_field) do {			\
+	a_type *t = ql_last(a_head, a_field);				\
+	ql_remove((a_head), t, a_field);				\
+} while (0)
+
+#define ql_foreach(a_var, a_head, a_field)				\
+	qr_foreach((a_var), ql_first(a_head), a_field)
+
+#define ql_reverse_foreach(a_var, a_head, a_field)			\
+	qr_reverse_foreach((a_var), ql_first(a_head), a_field)
diff --git a/extra/jemalloc/include/jemalloc/internal/qr.h b/extra/jemalloc/include/jemalloc/internal/qr.h
new file mode 100644
index 00000000000..fe22352fedd
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/qr.h
@@ -0,0 +1,67 @@
+/* Ring definitions. */
+#define qr(a_type)							\
+struct {								\
+	a_type	*qre_next;						\
+	a_type	*qre_prev;						\
+}
+
+/* Ring functions. */
+#define qr_new(a_qr, a_field) do {					\
+	(a_qr)->a_field.qre_next = (a_qr);				\
+	(a_qr)->a_field.qre_prev = (a_qr);				\
+} while (0)
+
+#define qr_next(a_qr, a_field) ((a_qr)->a_field.qre_next)
+
+#define qr_prev(a_qr, a_field) ((a_qr)->a_field.qre_prev)
+
+#define qr_before_insert(a_qrelm, a_qr, a_field) do {			\
+	(a_qr)->a_field.qre_prev = (a_qrelm)->a_field.qre_prev;		\
+	(a_qr)->a_field.qre_next = (a_qrelm);				\
+	(a_qr)->a_field.qre_prev->a_field.qre_next = (a_qr);		\
+	(a_qrelm)->a_field.qre_prev = (a_qr);				\
+} while (0)
+
+#define qr_after_insert(a_qrelm, a_qr, a_field)				\
+    do									\
+    {									\
+	(a_qr)->a_field.qre_next = (a_qrelm)->a_field.qre_next;		\
+	(a_qr)->a_field.qre_prev = (a_qrelm);				\
+	(a_qr)->a_field.qre_next->a_field.qre_prev = (a_qr);		\
+	(a_qrelm)->a_field.qre_next = (a_qr);				\
+    } while (0)
+
+#define qr_meld(a_qr_a, a_qr_b, a_field) do {				\
+	void *t;							\
+	(a_qr_a)->a_field.qre_prev->a_field.qre_next = (a_qr_b);	\
+	(a_qr_b)->a_field.qre_prev->a_field.qre_next = (a_qr_a);	\
+	t = (a_qr_a)->a_field.qre_prev;					\
+	(a_qr_a)->a_field.qre_prev = (a_qr_b)->a_field.qre_prev;	\
+	(a_qr_b)->a_field.qre_prev = t;					\
+} while (0)
+
+/* qr_meld() and qr_split() are functionally equivalent, so there's no need to
+ * have two copies of the code. */
+#define qr_split(a_qr_a, a_qr_b, a_field)				\
+	qr_meld((a_qr_a), (a_qr_b), a_field)
+
+#define qr_remove(a_qr, a_field) do {					\
+	(a_qr)->a_field.qre_prev->a_field.qre_next			\
+	    = (a_qr)->a_field.qre_next;					\
+	(a_qr)->a_field.qre_next->a_field.qre_prev			\
+	    = (a_qr)->a_field.qre_prev;					\
+	(a_qr)->a_field.qre_next = (a_qr);				\
+	(a_qr)->a_field.qre_prev = (a_qr);				\
+} while (0)
+
+#define qr_foreach(var, a_qr, a_field)					\
+	for ((var) = (a_qr);						\
+	    (var) != NULL;						\
+	    (var) = (((var)->a_field.qre_next != (a_qr))		\
+	    ? (var)->a_field.qre_next : NULL))
+
+#define qr_reverse_foreach(var, a_qr, a_field)				\
+	for ((var) = ((a_qr) != NULL) ? qr_prev(a_qr, a_field) : NULL;	\
+	    (var) != NULL;						\
+	    (var) = (((var) != (a_qr))					\
+	    ? (var)->a_field.qre_prev : NULL))
diff --git a/extra/jemalloc/include/jemalloc/internal/quarantine.h b/extra/jemalloc/include/jemalloc/internal/quarantine.h
new file mode 100644
index 00000000000..16f677f73da
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/quarantine.h
@@ -0,0 +1,67 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+typedef struct quarantine_obj_s quarantine_obj_t;
+typedef struct quarantine_s quarantine_t;
+
+/* Default per thread quarantine size if valgrind is enabled. */
+#define	JEMALLOC_VALGRIND_QUARANTINE_DEFAULT	(ZU(1) << 24)
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+struct quarantine_obj_s {
+	void	*ptr;
+	size_t	usize;
+};
+
+struct quarantine_s {
+	size_t			curbytes;
+	size_t			curobjs;
+	size_t			first;
+#define	LG_MAXOBJS_INIT 10
+	size_t			lg_maxobjs;
+	quarantine_obj_t	objs[1]; /* Dynamically sized ring buffer. */
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+quarantine_t	*quarantine_init(size_t lg_maxobjs);
+void	quarantine(void *ptr);
+void	quarantine_cleanup(void *arg);
+bool	quarantine_boot(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+malloc_tsd_protos(JEMALLOC_ATTR(unused), quarantine, quarantine_t *)
+
+void	quarantine_alloc_hook(void);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_QUARANTINE_C_))
+malloc_tsd_externs(quarantine, quarantine_t *)
+malloc_tsd_funcs(JEMALLOC_ALWAYS_INLINE, quarantine, quarantine_t *, NULL,
+    quarantine_cleanup)
+
+JEMALLOC_ALWAYS_INLINE void
+quarantine_alloc_hook(void)
+{
+	quarantine_t *quarantine;
+
+	assert(config_fill && opt_quarantine);
+
+	quarantine = *quarantine_tsd_get();
+	if (quarantine == NULL)
+		quarantine_init(LG_MAXOBJS_INIT);
+}
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
+
diff --git a/extra/jemalloc/include/jemalloc/internal/rb.h b/extra/jemalloc/include/jemalloc/internal/rb.h
new file mode 100644
index 00000000000..7b675f09051
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/rb.h
@@ -0,0 +1,973 @@
+/*-
+ *******************************************************************************
+ *
+ * cpp macro implementation of left-leaning 2-3 red-black trees.  Parent
+ * pointers are not used, and color bits are stored in the least significant
+ * bit of right-child pointers (if RB_COMPACT is defined), thus making node
+ * linkage as compact as is possible for red-black trees.
+ *
+ * Usage:
+ *
+ *   #include <stdint.h>
+ *   #include <stdbool.h>
+ *   #define NDEBUG // (Optional, see assert(3).)
+ *   #include <assert.h>
+ *   #define RB_COMPACT // (Optional, embed color bits in right-child pointers.)
+ *   #include <rb.h>
+ *   ...
+ *
+ *******************************************************************************
+ */
+
+#ifndef RB_H_
+#define	RB_H_
+
+#if 0
+__FBSDID("$FreeBSD: head/lib/libc/stdlib/rb.h 204493 2010-02-28 22:57:13Z jasone $");
+#endif
+
+#ifdef RB_COMPACT
+/* Node structure. */
+#define	rb_node(a_type)							\
+struct {								\
+    a_type *rbn_left;							\
+    a_type *rbn_right_red;						\
+}
+#else
+#define	rb_node(a_type)							\
+struct {								\
+    a_type *rbn_left;							\
+    a_type *rbn_right;							\
+    bool rbn_red;							\
+}
+#endif
+
+/* Root structure. */
+#define	rb_tree(a_type)							\
+struct {								\
+    a_type *rbt_root;							\
+    a_type rbt_nil;							\
+}
+
+/* Left accessors. */
+#define	rbtn_left_get(a_type, a_field, a_node)				\
+    ((a_node)->a_field.rbn_left)
+#define	rbtn_left_set(a_type, a_field, a_node, a_left) do {		\
+    (a_node)->a_field.rbn_left = a_left;				\
+} while (0)
+
+#ifdef RB_COMPACT
+/* Right accessors. */
+#define	rbtn_right_get(a_type, a_field, a_node)				\
+    ((a_type *) (((intptr_t) (a_node)->a_field.rbn_right_red)		\
+      & ((ssize_t)-2)))
+#define	rbtn_right_set(a_type, a_field, a_node, a_right) do {		\
+    (a_node)->a_field.rbn_right_red = (a_type *) (((uintptr_t) a_right)	\
+      | (((uintptr_t) (a_node)->a_field.rbn_right_red) & ((size_t)1)));	\
+} while (0)
+
+/* Color accessors. */
+#define	rbtn_red_get(a_type, a_field, a_node)				\
+    ((bool) (((uintptr_t) (a_node)->a_field.rbn_right_red)		\
+      & ((size_t)1)))
+#define	rbtn_color_set(a_type, a_field, a_node, a_red) do {		\
+    (a_node)->a_field.rbn_right_red = (a_type *) ((((intptr_t)		\
+      (a_node)->a_field.rbn_right_red) & ((ssize_t)-2))			\
+      | ((ssize_t)a_red));						\
+} while (0)
+#define	rbtn_red_set(a_type, a_field, a_node) do {			\
+    (a_node)->a_field.rbn_right_red = (a_type *) (((uintptr_t)		\
+      (a_node)->a_field.rbn_right_red) | ((size_t)1));			\
+} while (0)
+#define	rbtn_black_set(a_type, a_field, a_node) do {			\
+    (a_node)->a_field.rbn_right_red = (a_type *) (((intptr_t)		\
+      (a_node)->a_field.rbn_right_red) & ((ssize_t)-2));		\
+} while (0)
+#else
+/* Right accessors. */
+#define	rbtn_right_get(a_type, a_field, a_node)				\
+    ((a_node)->a_field.rbn_right)
+#define	rbtn_right_set(a_type, a_field, a_node, a_right) do {		\
+    (a_node)->a_field.rbn_right = a_right;				\
+} while (0)
+
+/* Color accessors. */
+#define	rbtn_red_get(a_type, a_field, a_node)				\
+    ((a_node)->a_field.rbn_red)
+#define	rbtn_color_set(a_type, a_field, a_node, a_red) do {		\
+    (a_node)->a_field.rbn_red = (a_red);				\
+} while (0)
+#define	rbtn_red_set(a_type, a_field, a_node) do {			\
+    (a_node)->a_field.rbn_red = true;					\
+} while (0)
+#define	rbtn_black_set(a_type, a_field, a_node) do {			\
+    (a_node)->a_field.rbn_red = false;					\
+} while (0)
+#endif
+
+/* Node initializer. */
+#define	rbt_node_new(a_type, a_field, a_rbt, a_node) do {		\
+    rbtn_left_set(a_type, a_field, (a_node), &(a_rbt)->rbt_nil);	\
+    rbtn_right_set(a_type, a_field, (a_node), &(a_rbt)->rbt_nil);	\
+    rbtn_red_set(a_type, a_field, (a_node));				\
+} while (0)
+
+/* Tree initializer. */
+#define	rb_new(a_type, a_field, a_rbt) do {				\
+    (a_rbt)->rbt_root = &(a_rbt)->rbt_nil;				\
+    rbt_node_new(a_type, a_field, a_rbt, &(a_rbt)->rbt_nil);		\
+    rbtn_black_set(a_type, a_field, &(a_rbt)->rbt_nil);			\
+} while (0)
+
+/* Internal utility macros. */
+#define	rbtn_first(a_type, a_field, a_rbt, a_root, r_node) do {		\
+    (r_node) = (a_root);						\
+    if ((r_node) != &(a_rbt)->rbt_nil) {				\
+	for (;								\
+	  rbtn_left_get(a_type, a_field, (r_node)) != &(a_rbt)->rbt_nil;\
+	  (r_node) = rbtn_left_get(a_type, a_field, (r_node))) {	\
+	}								\
+    }									\
+} while (0)
+
+#define	rbtn_last(a_type, a_field, a_rbt, a_root, r_node) do {		\
+    (r_node) = (a_root);						\
+    if ((r_node) != &(a_rbt)->rbt_nil) {				\
+	for (; rbtn_right_get(a_type, a_field, (r_node)) !=		\
+	  &(a_rbt)->rbt_nil; (r_node) = rbtn_right_get(a_type, a_field,	\
+	  (r_node))) {							\
+	}								\
+    }									\
+} while (0)
+
+#define	rbtn_rotate_left(a_type, a_field, a_node, r_node) do {		\
+    (r_node) = rbtn_right_get(a_type, a_field, (a_node));		\
+    rbtn_right_set(a_type, a_field, (a_node),				\
+      rbtn_left_get(a_type, a_field, (r_node)));			\
+    rbtn_left_set(a_type, a_field, (r_node), (a_node));			\
+} while (0)
+
+#define	rbtn_rotate_right(a_type, a_field, a_node, r_node) do {		\
+    (r_node) = rbtn_left_get(a_type, a_field, (a_node));		\
+    rbtn_left_set(a_type, a_field, (a_node),				\
+      rbtn_right_get(a_type, a_field, (r_node)));			\
+    rbtn_right_set(a_type, a_field, (r_node), (a_node));		\
+} while (0)
+
+/*
+ * The rb_proto() macro generates function prototypes that correspond to the
+ * functions generated by an equivalently parameterized call to rb_gen().
+ */
+
+#define	rb_proto(a_attr, a_prefix, a_rbt_type, a_type)			\
+a_attr void								\
+a_prefix##new(a_rbt_type *rbtree);					\
+a_attr a_type *								\
+a_prefix##first(a_rbt_type *rbtree);					\
+a_attr a_type *								\
+a_prefix##last(a_rbt_type *rbtree);					\
+a_attr a_type *								\
+a_prefix##next(a_rbt_type *rbtree, a_type *node);			\
+a_attr a_type *								\
+a_prefix##prev(a_rbt_type *rbtree, a_type *node);			\
+a_attr a_type *								\
+a_prefix##search(a_rbt_type *rbtree, a_type *key);			\
+a_attr a_type *								\
+a_prefix##nsearch(a_rbt_type *rbtree, a_type *key);			\
+a_attr a_type *								\
+a_prefix##psearch(a_rbt_type *rbtree, a_type *key);			\
+a_attr void								\
+a_prefix##insert(a_rbt_type *rbtree, a_type *node);			\
+a_attr void								\
+a_prefix##remove(a_rbt_type *rbtree, a_type *node);			\
+a_attr a_type *								\
+a_prefix##iter(a_rbt_type *rbtree, a_type *start, a_type *(*cb)(	\
+  a_rbt_type *, a_type *, void *), void *arg);				\
+a_attr a_type *								\
+a_prefix##reverse_iter(a_rbt_type *rbtree, a_type *start,		\
+  a_type *(*cb)(a_rbt_type *, a_type *, void *), void *arg);
+
+/*
+ * The rb_gen() macro generates a type-specific red-black tree implementation,
+ * based on the above cpp macros.
+ *
+ * Arguments:
+ *
+ *   a_attr    : Function attribute for generated functions (ex: static).
+ *   a_prefix  : Prefix for generated functions (ex: ex_).
+ *   a_rb_type : Type for red-black tree data structure (ex: ex_t).
+ *   a_type    : Type for red-black tree node data structure (ex: ex_node_t).
+ *   a_field   : Name of red-black tree node linkage (ex: ex_link).
+ *   a_cmp     : Node comparison function name, with the following prototype:
+ *                 int (a_cmp *)(a_type *a_node, a_type *a_other);
+ *                                       ^^^^^^
+ *                                    or a_key
+ *               Interpretation of comparision function return values:
+ *                 -1 : a_node <  a_other
+ *                  0 : a_node == a_other
+ *                  1 : a_node >  a_other
+ *               In all cases, the a_node or a_key macro argument is the first
+ *               argument to the comparison function, which makes it possible
+ *               to write comparison functions that treat the first argument
+ *               specially.
+ *
+ * Assuming the following setup:
+ *
+ *   typedef struct ex_node_s ex_node_t;
+ *   struct ex_node_s {
+ *       rb_node(ex_node_t) ex_link;
+ *   };
+ *   typedef rb_tree(ex_node_t) ex_t;
+ *   rb_gen(static, ex_, ex_t, ex_node_t, ex_link, ex_cmp)
+ *
+ * The following API is generated:
+ *
+ *   static void
+ *   ex_new(ex_t *tree);
+ *       Description: Initialize a red-black tree structure.
+ *       Args:
+ *         tree: Pointer to an uninitialized red-black tree object.
+ *
+ *   static ex_node_t *
+ *   ex_first(ex_t *tree);
+ *   static ex_node_t *
+ *   ex_last(ex_t *tree);
+ *       Description: Get the first/last node in tree.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *       Ret: First/last node in tree, or NULL if tree is empty.
+ *
+ *   static ex_node_t *
+ *   ex_next(ex_t *tree, ex_node_t *node);
+ *   static ex_node_t *
+ *   ex_prev(ex_t *tree, ex_node_t *node);
+ *       Description: Get node's successor/predecessor.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *         node: A node in tree.
+ *       Ret: node's successor/predecessor in tree, or NULL if node is
+ *            last/first.
+ *
+ *   static ex_node_t *
+ *   ex_search(ex_t *tree, ex_node_t *key);
+ *       Description: Search for node that matches key.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *         key : Search key.
+ *       Ret: Node in tree that matches key, or NULL if no match.
+ *
+ *   static ex_node_t *
+ *   ex_nsearch(ex_t *tree, ex_node_t *key);
+ *   static ex_node_t *
+ *   ex_psearch(ex_t *tree, ex_node_t *key);
+ *       Description: Search for node that matches key.  If no match is found,
+ *                    return what would be key's successor/predecessor, were
+ *                    key in tree.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *         key : Search key.
+ *       Ret: Node in tree that matches key, or if no match, hypothetical node's
+ *            successor/predecessor (NULL if no successor/predecessor).
+ *
+ *   static void
+ *   ex_insert(ex_t *tree, ex_node_t *node);
+ *       Description: Insert node into tree.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *         node: Node to be inserted into tree.
+ *
+ *   static void
+ *   ex_remove(ex_t *tree, ex_node_t *node);
+ *       Description: Remove node from tree.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *         node: Node in tree to be removed.
+ *
+ *   static ex_node_t *
+ *   ex_iter(ex_t *tree, ex_node_t *start, ex_node_t *(*cb)(ex_t *,
+ *     ex_node_t *, void *), void *arg);
+ *   static ex_node_t *
+ *   ex_reverse_iter(ex_t *tree, ex_node_t *start, ex_node *(*cb)(ex_t *,
+ *     ex_node_t *, void *), void *arg);
+ *       Description: Iterate forward/backward over tree, starting at node.  If
+ *                    tree is modified, iteration must be immediately
+ *                    terminated by the callback function that causes the
+ *                    modification.
+ *       Args:
+ *         tree : Pointer to an initialized red-black tree object.
+ *         start: Node at which to start iteration, or NULL to start at
+ *                first/last node.
+ *         cb   : Callback function, which is called for each node during
+ *                iteration.  Under normal circumstances the callback function
+ *                should return NULL, which causes iteration to continue.  If a
+ *                callback function returns non-NULL, iteration is immediately
+ *                terminated and the non-NULL return value is returned by the
+ *                iterator.  This is useful for re-starting iteration after
+ *                modifying tree.
+ *         arg  : Opaque pointer passed to cb().
+ *       Ret: NULL if iteration completed, or the non-NULL callback return value
+ *            that caused termination of the iteration.
+ */
+#define	rb_gen(a_attr, a_prefix, a_rbt_type, a_type, a_field, a_cmp)	\
+a_attr void								\
+a_prefix##new(a_rbt_type *rbtree) {					\
+    rb_new(a_type, a_field, rbtree);					\
+}									\
+a_attr a_type *								\
+a_prefix##first(a_rbt_type *rbtree) {					\
+    a_type *ret;							\
+    rbtn_first(a_type, a_field, rbtree, rbtree->rbt_root, ret);		\
+    if (ret == &rbtree->rbt_nil) {					\
+	ret = NULL;							\
+    }									\
+    return (ret);							\
+}									\
+a_attr a_type *								\
+a_prefix##last(a_rbt_type *rbtree) {					\
+    a_type *ret;							\
+    rbtn_last(a_type, a_field, rbtree, rbtree->rbt_root, ret);		\
+    if (ret == &rbtree->rbt_nil) {					\
+	ret = NULL;							\
+    }									\
+    return (ret);							\
+}									\
+a_attr a_type *								\
+a_prefix##next(a_rbt_type *rbtree, a_type *node) {			\
+    a_type *ret;							\
+    if (rbtn_right_get(a_type, a_field, node) != &rbtree->rbt_nil) {	\
+	rbtn_first(a_type, a_field, rbtree, rbtn_right_get(a_type,	\
+	  a_field, node), ret);						\
+    } else {								\
+	a_type *tnode = rbtree->rbt_root;				\
+	assert(tnode != &rbtree->rbt_nil);				\
+	ret = &rbtree->rbt_nil;						\
+	while (true) {							\
+	    int cmp = (a_cmp)(node, tnode);				\
+	    if (cmp < 0) {						\
+		ret = tnode;						\
+		tnode = rbtn_left_get(a_type, a_field, tnode);		\
+	    } else if (cmp > 0) {					\
+		tnode = rbtn_right_get(a_type, a_field, tnode);		\
+	    } else {							\
+		break;							\
+	    }								\
+	    assert(tnode != &rbtree->rbt_nil);				\
+	}								\
+    }									\
+    if (ret == &rbtree->rbt_nil) {					\
+	ret = (NULL);							\
+    }									\
+    return (ret);							\
+}									\
+a_attr a_type *								\
+a_prefix##prev(a_rbt_type *rbtree, a_type *node) {			\
+    a_type *ret;							\
+    if (rbtn_left_get(a_type, a_field, node) != &rbtree->rbt_nil) {	\
+	rbtn_last(a_type, a_field, rbtree, rbtn_left_get(a_type,	\
+	  a_field, node), ret);						\
+    } else {								\
+	a_type *tnode = rbtree->rbt_root;				\
+	assert(tnode != &rbtree->rbt_nil);				\
+	ret = &rbtree->rbt_nil;						\
+	while (true) {							\
+	    int cmp = (a_cmp)(node, tnode);				\
+	    if (cmp < 0) {						\
+		tnode = rbtn_left_get(a_type, a_field, tnode);		\
+	    } else if (cmp > 0) {					\
+		ret = tnode;						\
+		tnode = rbtn_right_get(a_type, a_field, tnode);		\
+	    } else {							\
+		break;							\
+	    }								\
+	    assert(tnode != &rbtree->rbt_nil);				\
+	}								\
+    }									\
+    if (ret == &rbtree->rbt_nil) {					\
+	ret = (NULL);							\
+    }									\
+    return (ret);							\
+}									\
+a_attr a_type *								\
+a_prefix##search(a_rbt_type *rbtree, a_type *key) {			\
+    a_type *ret;							\
+    int cmp;								\
+    ret = rbtree->rbt_root;						\
+    while (ret != &rbtree->rbt_nil					\
+      && (cmp = (a_cmp)(key, ret)) != 0) {				\
+	if (cmp < 0) {							\
+	    ret = rbtn_left_get(a_type, a_field, ret);			\
+	} else {							\
+	    ret = rbtn_right_get(a_type, a_field, ret);			\
+	}								\
+    }									\
+    if (ret == &rbtree->rbt_nil) {					\
+	ret = (NULL);							\
+    }									\
+    return (ret);							\
+}									\
+a_attr a_type *								\
+a_prefix##nsearch(a_rbt_type *rbtree, a_type *key) {			\
+    a_type *ret;							\
+    a_type *tnode = rbtree->rbt_root;					\
+    ret = &rbtree->rbt_nil;						\
+    while (tnode != &rbtree->rbt_nil) {					\
+	int cmp = (a_cmp)(key, tnode);					\
+	if (cmp < 0) {							\
+	    ret = tnode;						\
+	    tnode = rbtn_left_get(a_type, a_field, tnode);		\
+	} else if (cmp > 0) {						\
+	    tnode = rbtn_right_get(a_type, a_field, tnode);		\
+	} else {							\
+	    ret = tnode;						\
+	    break;							\
+	}								\
+    }									\
+    if (ret == &rbtree->rbt_nil) {					\
+	ret = (NULL);							\
+    }									\
+    return (ret);							\
+}									\
+a_attr a_type *								\
+a_prefix##psearch(a_rbt_type *rbtree, a_type *key) {			\
+    a_type *ret;							\
+    a_type *tnode = rbtree->rbt_root;					\
+    ret = &rbtree->rbt_nil;						\
+    while (tnode != &rbtree->rbt_nil) {					\
+	int cmp = (a_cmp)(key, tnode);					\
+	if (cmp < 0) {							\
+	    tnode = rbtn_left_get(a_type, a_field, tnode);		\
+	} else if (cmp > 0) {						\
+	    ret = tnode;						\
+	    tnode = rbtn_right_get(a_type, a_field, tnode);		\
+	} else {							\
+	    ret = tnode;						\
+	    break;							\
+	}								\
+    }									\
+    if (ret == &rbtree->rbt_nil) {					\
+	ret = (NULL);							\
+    }									\
+    return (ret);							\
+}									\
+a_attr void								\
+a_prefix##insert(a_rbt_type *rbtree, a_type *node) {			\
+    struct {								\
+	a_type *node;							\
+	int cmp;							\
+    } path[sizeof(void *) << 4], *pathp;				\
+    rbt_node_new(a_type, a_field, rbtree, node);			\
+    /* Wind. */								\
+    path->node = rbtree->rbt_root;					\
+    for (pathp = path; pathp->node != &rbtree->rbt_nil; pathp++) {	\
+	int cmp = pathp->cmp = a_cmp(node, pathp->node);		\
+	assert(cmp != 0);						\
+	if (cmp < 0) {							\
+	    pathp[1].node = rbtn_left_get(a_type, a_field,		\
+	      pathp->node);						\
+	} else {							\
+	    pathp[1].node = rbtn_right_get(a_type, a_field,		\
+	      pathp->node);						\
+	}								\
+    }									\
+    pathp->node = node;							\
+    /* Unwind. */							\
+    for (pathp--; (uintptr_t)pathp >= (uintptr_t)path; pathp--) {	\
+	a_type *cnode = pathp->node;					\
+	if (pathp->cmp < 0) {						\
+	    a_type *left = pathp[1].node;				\
+	    rbtn_left_set(a_type, a_field, cnode, left);		\
+	    if (rbtn_red_get(a_type, a_field, left)) {			\
+		a_type *leftleft = rbtn_left_get(a_type, a_field, left);\
+		if (rbtn_red_get(a_type, a_field, leftleft)) {		\
+		    /* Fix up 4-node. */				\
+		    a_type *tnode;					\
+		    rbtn_black_set(a_type, a_field, leftleft);		\
+		    rbtn_rotate_right(a_type, a_field, cnode, tnode);	\
+		    cnode = tnode;					\
+		}							\
+	    } else {							\
+		return;							\
+	    }								\
+	} else {							\
+	    a_type *right = pathp[1].node;				\
+	    rbtn_right_set(a_type, a_field, cnode, right);		\
+	    if (rbtn_red_get(a_type, a_field, right)) {			\
+		a_type *left = rbtn_left_get(a_type, a_field, cnode);	\
+		if (rbtn_red_get(a_type, a_field, left)) {		\
+		    /* Split 4-node. */					\
+		    rbtn_black_set(a_type, a_field, left);		\
+		    rbtn_black_set(a_type, a_field, right);		\
+		    rbtn_red_set(a_type, a_field, cnode);		\
+		} else {						\
+		    /* Lean left. */					\
+		    a_type *tnode;					\
+		    bool tred = rbtn_red_get(a_type, a_field, cnode);	\
+		    rbtn_rotate_left(a_type, a_field, cnode, tnode);	\
+		    rbtn_color_set(a_type, a_field, tnode, tred);	\
+		    rbtn_red_set(a_type, a_field, cnode);		\
+		    cnode = tnode;					\
+		}							\
+	    } else {							\
+		return;							\
+	    }								\
+	}								\
+	pathp->node = cnode;						\
+    }									\
+    /* Set root, and make it black. */					\
+    rbtree->rbt_root = path->node;					\
+    rbtn_black_set(a_type, a_field, rbtree->rbt_root);			\
+}									\
+a_attr void								\
+a_prefix##remove(a_rbt_type *rbtree, a_type *node) {			\
+    struct {								\
+	a_type *node;							\
+	int cmp;							\
+    } *pathp, *nodep, path[sizeof(void *) << 4];			\
+    /* Wind. */								\
+    nodep = NULL; /* Silence compiler warning. */			\
+    path->node = rbtree->rbt_root;					\
+    for (pathp = path; pathp->node != &rbtree->rbt_nil; pathp++) {	\
+	int cmp = pathp->cmp = a_cmp(node, pathp->node);		\
+	if (cmp < 0) {							\
+	    pathp[1].node = rbtn_left_get(a_type, a_field,		\
+	      pathp->node);						\
+	} else {							\
+	    pathp[1].node = rbtn_right_get(a_type, a_field,		\
+	      pathp->node);						\
+	    if (cmp == 0) {						\
+	        /* Find node's successor, in preparation for swap. */	\
+		pathp->cmp = 1;						\
+		nodep = pathp;						\
+		for (pathp++; pathp->node != &rbtree->rbt_nil;		\
+		  pathp++) {						\
+		    pathp->cmp = -1;					\
+		    pathp[1].node = rbtn_left_get(a_type, a_field,	\
+		      pathp->node);					\
+		}							\
+		break;							\
+	    }								\
+	}								\
+    }									\
+    assert(nodep->node == node);					\
+    pathp--;								\
+    if (pathp->node != node) {						\
+	/* Swap node with its successor. */				\
+	bool tred = rbtn_red_get(a_type, a_field, pathp->node);		\
+	rbtn_color_set(a_type, a_field, pathp->node,			\
+	  rbtn_red_get(a_type, a_field, node));				\
+	rbtn_left_set(a_type, a_field, pathp->node,			\
+	  rbtn_left_get(a_type, a_field, node));			\
+	/* If node's successor is its right child, the following code */\
+	/* will do the wrong thing for the right child pointer.       */\
+	/* However, it doesn't matter, because the pointer will be    */\
+	/* properly set when the successor is pruned.                 */\
+	rbtn_right_set(a_type, a_field, pathp->node,			\
+	  rbtn_right_get(a_type, a_field, node));			\
+	rbtn_color_set(a_type, a_field, node, tred);			\
+	/* The pruned leaf node's child pointers are never accessed   */\
+	/* again, so don't bother setting them to nil.                */\
+	nodep->node = pathp->node;					\
+	pathp->node = node;						\
+	if (nodep == path) {						\
+	    rbtree->rbt_root = nodep->node;				\
+	} else {							\
+	    if (nodep[-1].cmp < 0) {					\
+		rbtn_left_set(a_type, a_field, nodep[-1].node,		\
+		  nodep->node);						\
+	    } else {							\
+		rbtn_right_set(a_type, a_field, nodep[-1].node,		\
+		  nodep->node);						\
+	    }								\
+	}								\
+    } else {								\
+	a_type *left = rbtn_left_get(a_type, a_field, node);		\
+	if (left != &rbtree->rbt_nil) {					\
+	    /* node has no successor, but it has a left child.        */\
+	    /* Splice node out, without losing the left child.        */\
+	    assert(rbtn_red_get(a_type, a_field, node) == false);	\
+	    assert(rbtn_red_get(a_type, a_field, left));		\
+	    rbtn_black_set(a_type, a_field, left);			\
+	    if (pathp == path) {					\
+		rbtree->rbt_root = left;				\
+	    } else {							\
+		if (pathp[-1].cmp < 0) {				\
+		    rbtn_left_set(a_type, a_field, pathp[-1].node,	\
+		      left);						\
+		} else {						\
+		    rbtn_right_set(a_type, a_field, pathp[-1].node,	\
+		      left);						\
+		}							\
+	    }								\
+	    return;							\
+	} else if (pathp == path) {					\
+	    /* The tree only contained one node. */			\
+	    rbtree->rbt_root = &rbtree->rbt_nil;			\
+	    return;							\
+	}								\
+    }									\
+    if (rbtn_red_get(a_type, a_field, pathp->node)) {			\
+	/* Prune red node, which requires no fixup. */			\
+	assert(pathp[-1].cmp < 0);					\
+	rbtn_left_set(a_type, a_field, pathp[-1].node,			\
+	  &rbtree->rbt_nil);						\
+	return;								\
+    }									\
+    /* The node to be pruned is black, so unwind until balance is     */\
+    /* restored.                                                      */\
+    pathp->node = &rbtree->rbt_nil;					\
+    for (pathp--; (uintptr_t)pathp >= (uintptr_t)path; pathp--) {	\
+	assert(pathp->cmp != 0);					\
+	if (pathp->cmp < 0) {						\
+	    rbtn_left_set(a_type, a_field, pathp->node,			\
+	      pathp[1].node);						\
+	    assert(rbtn_red_get(a_type, a_field, pathp[1].node)		\
+	      == false);						\
+	    if (rbtn_red_get(a_type, a_field, pathp->node)) {		\
+		a_type *right = rbtn_right_get(a_type, a_field,		\
+		  pathp->node);						\
+		a_type *rightleft = rbtn_left_get(a_type, a_field,	\
+		  right);						\
+		a_type *tnode;						\
+		if (rbtn_red_get(a_type, a_field, rightleft)) {		\
+		    /* In the following diagrams, ||, //, and \\      */\
+		    /* indicate the path to the removed node.         */\
+		    /*                                                */\
+		    /*      ||                                        */\
+		    /*    pathp(r)                                    */\
+		    /*  //        \                                   */\
+		    /* (b)        (b)                                 */\
+		    /*           /                                    */\
+		    /*          (r)                                   */\
+		    /*                                                */\
+		    rbtn_black_set(a_type, a_field, pathp->node);	\
+		    rbtn_rotate_right(a_type, a_field, right, tnode);	\
+		    rbtn_right_set(a_type, a_field, pathp->node, tnode);\
+		    rbtn_rotate_left(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		} else {						\
+		    /*      ||                                        */\
+		    /*    pathp(r)                                    */\
+		    /*  //        \                                   */\
+		    /* (b)        (b)                                 */\
+		    /*           /                                    */\
+		    /*          (b)                                   */\
+		    /*                                                */\
+		    rbtn_rotate_left(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		}							\
+		/* Balance restored, but rotation modified subtree    */\
+		/* root.                                              */\
+		assert((uintptr_t)pathp > (uintptr_t)path);		\
+		if (pathp[-1].cmp < 0) {				\
+		    rbtn_left_set(a_type, a_field, pathp[-1].node,	\
+		      tnode);						\
+		} else {						\
+		    rbtn_right_set(a_type, a_field, pathp[-1].node,	\
+		      tnode);						\
+		}							\
+		return;							\
+	    } else {							\
+		a_type *right = rbtn_right_get(a_type, a_field,		\
+		  pathp->node);						\
+		a_type *rightleft = rbtn_left_get(a_type, a_field,	\
+		  right);						\
+		if (rbtn_red_get(a_type, a_field, rightleft)) {		\
+		    /*      ||                                        */\
+		    /*    pathp(b)                                    */\
+		    /*  //        \                                   */\
+		    /* (b)        (b)                                 */\
+		    /*           /                                    */\
+		    /*          (r)                                   */\
+		    a_type *tnode;					\
+		    rbtn_black_set(a_type, a_field, rightleft);		\
+		    rbtn_rotate_right(a_type, a_field, right, tnode);	\
+		    rbtn_right_set(a_type, a_field, pathp->node, tnode);\
+		    rbtn_rotate_left(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    /* Balance restored, but rotation modified        */\
+		    /* subree root, which may actually be the tree    */\
+		    /* root.                                          */\
+		    if (pathp == path) {				\
+			/* Set root. */					\
+			rbtree->rbt_root = tnode;			\
+		    } else {						\
+			if (pathp[-1].cmp < 0) {			\
+			    rbtn_left_set(a_type, a_field,		\
+			      pathp[-1].node, tnode);			\
+			} else {					\
+			    rbtn_right_set(a_type, a_field,		\
+			      pathp[-1].node, tnode);			\
+			}						\
+		    }							\
+		    return;						\
+		} else {						\
+		    /*      ||                                        */\
+		    /*    pathp(b)                                    */\
+		    /*  //        \                                   */\
+		    /* (b)        (b)                                 */\
+		    /*           /                                    */\
+		    /*          (b)                                   */\
+		    a_type *tnode;					\
+		    rbtn_red_set(a_type, a_field, pathp->node);		\
+		    rbtn_rotate_left(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    pathp->node = tnode;				\
+		}							\
+	    }								\
+	} else {							\
+	    a_type *left;						\
+	    rbtn_right_set(a_type, a_field, pathp->node,		\
+	      pathp[1].node);						\
+	    left = rbtn_left_get(a_type, a_field, pathp->node);		\
+	    if (rbtn_red_get(a_type, a_field, left)) {			\
+		a_type *tnode;						\
+		a_type *leftright = rbtn_right_get(a_type, a_field,	\
+		  left);						\
+		a_type *leftrightleft = rbtn_left_get(a_type, a_field,	\
+		  leftright);						\
+		if (rbtn_red_get(a_type, a_field, leftrightleft)) {	\
+		    /*      ||                                        */\
+		    /*    pathp(b)                                    */\
+		    /*   /        \\                                  */\
+		    /* (r)        (b)                                 */\
+		    /*   \                                            */\
+		    /*   (b)                                          */\
+		    /*   /                                            */\
+		    /* (r)                                            */\
+		    a_type *unode;					\
+		    rbtn_black_set(a_type, a_field, leftrightleft);	\
+		    rbtn_rotate_right(a_type, a_field, pathp->node,	\
+		      unode);						\
+		    rbtn_rotate_right(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    rbtn_right_set(a_type, a_field, unode, tnode);	\
+		    rbtn_rotate_left(a_type, a_field, unode, tnode);	\
+		} else {						\
+		    /*      ||                                        */\
+		    /*    pathp(b)                                    */\
+		    /*   /        \\                                  */\
+		    /* (r)        (b)                                 */\
+		    /*   \                                            */\
+		    /*   (b)                                          */\
+		    /*   /                                            */\
+		    /* (b)                                            */\
+		    assert(leftright != &rbtree->rbt_nil);		\
+		    rbtn_red_set(a_type, a_field, leftright);		\
+		    rbtn_rotate_right(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    rbtn_black_set(a_type, a_field, tnode);		\
+		}							\
+		/* Balance restored, but rotation modified subtree    */\
+		/* root, which may actually be the tree root.         */\
+		if (pathp == path) {					\
+		    /* Set root. */					\
+		    rbtree->rbt_root = tnode;				\
+		} else {						\
+		    if (pathp[-1].cmp < 0) {				\
+			rbtn_left_set(a_type, a_field, pathp[-1].node,	\
+			  tnode);					\
+		    } else {						\
+			rbtn_right_set(a_type, a_field, pathp[-1].node,	\
+			  tnode);					\
+		    }							\
+		}							\
+		return;							\
+	    } else if (rbtn_red_get(a_type, a_field, pathp->node)) {	\
+		a_type *leftleft = rbtn_left_get(a_type, a_field, left);\
+		if (rbtn_red_get(a_type, a_field, leftleft)) {		\
+		    /*        ||                                      */\
+		    /*      pathp(r)                                  */\
+		    /*     /        \\                                */\
+		    /*   (b)        (b)                               */\
+		    /*   /                                            */\
+		    /* (r)                                            */\
+		    a_type *tnode;					\
+		    rbtn_black_set(a_type, a_field, pathp->node);	\
+		    rbtn_red_set(a_type, a_field, left);		\
+		    rbtn_black_set(a_type, a_field, leftleft);		\
+		    rbtn_rotate_right(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    /* Balance restored, but rotation modified        */\
+		    /* subtree root.                                  */\
+		    assert((uintptr_t)pathp > (uintptr_t)path);		\
+		    if (pathp[-1].cmp < 0) {				\
+			rbtn_left_set(a_type, a_field, pathp[-1].node,	\
+			  tnode);					\
+		    } else {						\
+			rbtn_right_set(a_type, a_field, pathp[-1].node,	\
+			  tnode);					\
+		    }							\
+		    return;						\
+		} else {						\
+		    /*        ||                                      */\
+		    /*      pathp(r)                                  */\
+		    /*     /        \\                                */\
+		    /*   (b)        (b)                               */\
+		    /*   /                                            */\
+		    /* (b)                                            */\
+		    rbtn_red_set(a_type, a_field, left);		\
+		    rbtn_black_set(a_type, a_field, pathp->node);	\
+		    /* Balance restored. */				\
+		    return;						\
+		}							\
+	    } else {							\
+		a_type *leftleft = rbtn_left_get(a_type, a_field, left);\
+		if (rbtn_red_get(a_type, a_field, leftleft)) {		\
+		    /*               ||                               */\
+		    /*             pathp(b)                           */\
+		    /*            /        \\                         */\
+		    /*          (b)        (b)                        */\
+		    /*          /                                     */\
+		    /*        (r)                                     */\
+		    a_type *tnode;					\
+		    rbtn_black_set(a_type, a_field, leftleft);		\
+		    rbtn_rotate_right(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    /* Balance restored, but rotation modified        */\
+		    /* subtree root, which may actually be the tree   */\
+		    /* root.                                          */\
+		    if (pathp == path) {				\
+			/* Set root. */					\
+			rbtree->rbt_root = tnode;			\
+		    } else {						\
+			if (pathp[-1].cmp < 0) {			\
+			    rbtn_left_set(a_type, a_field,		\
+			      pathp[-1].node, tnode);			\
+			} else {					\
+			    rbtn_right_set(a_type, a_field,		\
+			      pathp[-1].node, tnode);			\
+			}						\
+		    }							\
+		    return;						\
+		} else {						\
+		    /*               ||                               */\
+		    /*             pathp(b)                           */\
+		    /*            /        \\                         */\
+		    /*          (b)        (b)                        */\
+		    /*          /                                     */\
+		    /*        (b)                                     */\
+		    rbtn_red_set(a_type, a_field, left);		\
+		}							\
+	    }								\
+	}								\
+    }									\
+    /* Set root. */							\
+    rbtree->rbt_root = path->node;					\
+    assert(rbtn_red_get(a_type, a_field, rbtree->rbt_root) == false);	\
+}									\
+a_attr a_type *								\
+a_prefix##iter_recurse(a_rbt_type *rbtree, a_type *node,		\
+  a_type *(*cb)(a_rbt_type *, a_type *, void *), void *arg) {		\
+    if (node == &rbtree->rbt_nil) {					\
+	return (&rbtree->rbt_nil);					\
+    } else {								\
+	a_type *ret;							\
+	if ((ret = a_prefix##iter_recurse(rbtree, rbtn_left_get(a_type,	\
+	  a_field, node), cb, arg)) != &rbtree->rbt_nil			\
+	  || (ret = cb(rbtree, node, arg)) != NULL) {			\
+	    return (ret);						\
+	}								\
+	return (a_prefix##iter_recurse(rbtree, rbtn_right_get(a_type,	\
+	  a_field, node), cb, arg));					\
+    }									\
+}									\
+a_attr a_type *								\
+a_prefix##iter_start(a_rbt_type *rbtree, a_type *start, a_type *node,	\
+  a_type *(*cb)(a_rbt_type *, a_type *, void *), void *arg) {		\
+    int cmp = a_cmp(start, node);					\
+    if (cmp < 0) {							\
+	a_type *ret;							\
+	if ((ret = a_prefix##iter_start(rbtree, start,			\
+	  rbtn_left_get(a_type, a_field, node), cb, arg)) !=		\
+	  &rbtree->rbt_nil || (ret = cb(rbtree, node, arg)) != NULL) {	\
+	    return (ret);						\
+	}								\
+	return (a_prefix##iter_recurse(rbtree, rbtn_right_get(a_type,	\
+	  a_field, node), cb, arg));					\
+    } else if (cmp > 0) {						\
+	return (a_prefix##iter_start(rbtree, start,			\
+	  rbtn_right_get(a_type, a_field, node), cb, arg));		\
+    } else {								\
+	a_type *ret;							\
+	if ((ret = cb(rbtree, node, arg)) != NULL) {			\
+	    return (ret);						\
+	}								\
+	return (a_prefix##iter_recurse(rbtree, rbtn_right_get(a_type,	\
+	  a_field, node), cb, arg));					\
+    }									\
+}									\
+a_attr a_type *								\
+a_prefix##iter(a_rbt_type *rbtree, a_type *start, a_type *(*cb)(	\
+  a_rbt_type *, a_type *, void *), void *arg) {				\
+    a_type *ret;							\
+    if (start != NULL) {						\
+	ret = a_prefix##iter_start(rbtree, start, rbtree->rbt_root,	\
+	  cb, arg);							\
+    } else {								\
+	ret = a_prefix##iter_recurse(rbtree, rbtree->rbt_root, cb, arg);\
+    }									\
+    if (ret == &rbtree->rbt_nil) {					\
+	ret = NULL;							\
+    }									\
+    return (ret);							\
+}									\
+a_attr a_type *								\
+a_prefix##reverse_iter_recurse(a_rbt_type *rbtree, a_type *node,	\
+  a_type *(*cb)(a_rbt_type *, a_type *, void *), void *arg) {		\
+    if (node == &rbtree->rbt_nil) {					\
+	return (&rbtree->rbt_nil);					\
+    } else {								\
+	a_type *ret;							\
+	if ((ret = a_prefix##reverse_iter_recurse(rbtree,		\
+	  rbtn_right_get(a_type, a_field, node), cb, arg)) !=		\
+	  &rbtree->rbt_nil || (ret = cb(rbtree, node, arg)) != NULL) {	\
+	    return (ret);						\
+	}								\
+	return (a_prefix##reverse_iter_recurse(rbtree,			\
+	  rbtn_left_get(a_type, a_field, node), cb, arg));		\
+    }									\
+}									\
+a_attr a_type *								\
+a_prefix##reverse_iter_start(a_rbt_type *rbtree, a_type *start,		\
+  a_type *node, a_type *(*cb)(a_rbt_type *, a_type *, void *),		\
+  void *arg) {								\
+    int cmp = a_cmp(start, node);					\
+    if (cmp > 0) {							\
+	a_type *ret;							\
+	if ((ret = a_prefix##reverse_iter_start(rbtree, start,		\
+	  rbtn_right_get(a_type, a_field, node), cb, arg)) !=		\
+	  &rbtree->rbt_nil || (ret = cb(rbtree, node, arg)) != NULL) {	\
+	    return (ret);						\
+	}								\
+	return (a_prefix##reverse_iter_recurse(rbtree,			\
+	  rbtn_left_get(a_type, a_field, node), cb, arg));		\
+    } else if (cmp < 0) {						\
+	return (a_prefix##reverse_iter_start(rbtree, start,		\
+	  rbtn_left_get(a_type, a_field, node), cb, arg));		\
+    } else {								\
+	a_type *ret;							\
+	if ((ret = cb(rbtree, node, arg)) != NULL) {			\
+	    return (ret);						\
+	}								\
+	return (a_prefix##reverse_iter_recurse(rbtree,			\
+	  rbtn_left_get(a_type, a_field, node), cb, arg));		\
+    }									\
+}									\
+a_attr a_type *								\
+a_prefix##reverse_iter(a_rbt_type *rbtree, a_type *start,		\
+  a_type *(*cb)(a_rbt_type *, a_type *, void *), void *arg) {		\
+    a_type *ret;							\
+    if (start != NULL) {						\
+	ret = a_prefix##reverse_iter_start(rbtree, start,		\
+	  rbtree->rbt_root, cb, arg);					\
+    } else {								\
+	ret = a_prefix##reverse_iter_recurse(rbtree, rbtree->rbt_root,	\
+	  cb, arg);							\
+    }									\
+    if (ret == &rbtree->rbt_nil) {					\
+	ret = NULL;							\
+    }									\
+    return (ret);							\
+}
+
+#endif /* RB_H_ */
diff --git a/extra/jemalloc/include/jemalloc/internal/rtree.h b/extra/jemalloc/include/jemalloc/internal/rtree.h
new file mode 100644
index 00000000000..9bd98548cfe
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/rtree.h
@@ -0,0 +1,164 @@
+/*
+ * This radix tree implementation is tailored to the singular purpose of
+ * tracking which chunks are currently owned by jemalloc.  This functionality
+ * is mandatory for OS X, where jemalloc must be able to respond to object
+ * ownership queries.
+ *
+ *******************************************************************************
+ */
+#ifdef JEMALLOC_H_TYPES
+
+typedef struct rtree_s rtree_t;
+
+/*
+ * Size of each radix tree node (must be a power of 2).  This impacts tree
+ * depth.
+ */
+#if (LG_SIZEOF_PTR == 2)
+#  define RTREE_NODESIZE (1U << 14)
+#else
+#  define RTREE_NODESIZE CACHELINE
+#endif
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+struct rtree_s {
+	malloc_mutex_t	mutex;
+	void		**root;
+	unsigned	height;
+	unsigned	level2bits[1]; /* Dynamically sized. */
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+rtree_t	*rtree_new(unsigned bits);
+void	rtree_prefork(rtree_t *rtree);
+void	rtree_postfork_parent(rtree_t *rtree);
+void	rtree_postfork_child(rtree_t *rtree);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+#ifndef JEMALLOC_DEBUG
+void	*rtree_get_locked(rtree_t *rtree, uintptr_t key);
+#endif
+void	*rtree_get(rtree_t *rtree, uintptr_t key);
+bool	rtree_set(rtree_t *rtree, uintptr_t key, void *val);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_RTREE_C_))
+#define	RTREE_GET_GENERATE(f)						\
+/* The least significant bits of the key are ignored. */		\
+JEMALLOC_INLINE void *							\
+f(rtree_t *rtree, uintptr_t key)					\
+{									\
+	void *ret;							\
+	uintptr_t subkey;						\
+	unsigned i, lshift, height, bits;				\
+	void **node, **child;						\
+									\
+	RTREE_LOCK(&rtree->mutex);					\
+	for (i = lshift = 0, height = rtree->height, node = rtree->root;\
+	    i < height - 1;						\
+	    i++, lshift += bits, node = child) {			\
+		bits = rtree->level2bits[i];				\
+		subkey = (key << lshift) >> ((ZU(1) << (LG_SIZEOF_PTR + \
+		    3)) - bits);					\
+		child = (void**)node[subkey];				\
+		if (child == NULL) {					\
+			RTREE_UNLOCK(&rtree->mutex);			\
+			return (NULL);					\
+		}							\
+	}								\
+									\
+	/*								\
+	 * node is a leaf, so it contains values rather than node	\
+	 * pointers.							\
+	 */								\
+	bits = rtree->level2bits[i];					\
+	subkey = (key << lshift) >> ((ZU(1) << (LG_SIZEOF_PTR+3)) -	\
+	    bits);							\
+	ret = node[subkey];						\
+	RTREE_UNLOCK(&rtree->mutex);					\
+									\
+	RTREE_GET_VALIDATE						\
+	return (ret);							\
+}
+
+#ifdef JEMALLOC_DEBUG
+#  define RTREE_LOCK(l)		malloc_mutex_lock(l)
+#  define RTREE_UNLOCK(l)	malloc_mutex_unlock(l)
+#  define RTREE_GET_VALIDATE
+RTREE_GET_GENERATE(rtree_get_locked)
+#  undef RTREE_LOCK
+#  undef RTREE_UNLOCK
+#  undef RTREE_GET_VALIDATE
+#endif
+
+#define	RTREE_LOCK(l)
+#define	RTREE_UNLOCK(l)
+#ifdef JEMALLOC_DEBUG
+   /*
+    * Suppose that it were possible for a jemalloc-allocated chunk to be
+    * munmap()ped, followed by a different allocator in another thread re-using
+    * overlapping virtual memory, all without invalidating the cached rtree
+    * value.  The result would be a false positive (the rtree would claim that
+    * jemalloc owns memory that it had actually discarded).  This scenario
+    * seems impossible, but the following assertion is a prudent sanity check.
+    */
+#  define RTREE_GET_VALIDATE						\
+	assert(rtree_get_locked(rtree, key) == ret);
+#else
+#  define RTREE_GET_VALIDATE
+#endif
+RTREE_GET_GENERATE(rtree_get)
+#undef RTREE_LOCK
+#undef RTREE_UNLOCK
+#undef RTREE_GET_VALIDATE
+
+JEMALLOC_INLINE bool
+rtree_set(rtree_t *rtree, uintptr_t key, void *val)
+{
+	uintptr_t subkey;
+	unsigned i, lshift, height, bits;
+	void **node, **child;
+
+	malloc_mutex_lock(&rtree->mutex);
+	for (i = lshift = 0, height = rtree->height, node = rtree->root;
+	    i < height - 1;
+	    i++, lshift += bits, node = child) {
+		bits = rtree->level2bits[i];
+		subkey = (key << lshift) >> ((ZU(1) << (LG_SIZEOF_PTR+3)) -
+		    bits);
+		child = (void**)node[subkey];
+		if (child == NULL) {
+			child = (void**)base_alloc(sizeof(void *) <<
+			    rtree->level2bits[i+1]);
+			if (child == NULL) {
+				malloc_mutex_unlock(&rtree->mutex);
+				return (true);
+			}
+			memset(child, 0, sizeof(void *) <<
+			    rtree->level2bits[i+1]);
+			node[subkey] = child;
+		}
+	}
+
+	/* node is a leaf, so it contains values rather than node pointers. */
+	bits = rtree->level2bits[i];
+	subkey = (key << lshift) >> ((ZU(1) << (LG_SIZEOF_PTR+3)) - bits);
+	node[subkey] = val;
+	malloc_mutex_unlock(&rtree->mutex);
+
+	return (false);
+}
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/internal/size_classes.sh b/extra/jemalloc/include/jemalloc/internal/size_classes.sh
new file mode 100755
index 00000000000..29c80c1fb8d
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/size_classes.sh
@@ -0,0 +1,122 @@
+#!/bin/sh
+
+# The following limits are chosen such that they cover all supported platforms.
+
+# Range of quanta.
+lg_qmin=3
+lg_qmax=4
+
+# The range of tiny size classes is [2^lg_tmin..2^(lg_q-1)].
+lg_tmin=3
+
+# Range of page sizes.
+lg_pmin=12
+lg_pmax=16
+
+pow2() {
+  e=$1
+  pow2_result=1
+  while [ ${e} -gt 0 ] ; do
+    pow2_result=$((${pow2_result} + ${pow2_result}))
+    e=$((${e} - 1))
+  done
+}
+
+cat <<EOF
+/* This file was automatically generated by size_classes.sh. */
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+EOF
+
+lg_q=${lg_qmin}
+while [ ${lg_q} -le ${lg_qmax} ] ; do
+  lg_t=${lg_tmin}
+  while [ ${lg_t} -le ${lg_q} ] ; do
+    lg_p=${lg_pmin}
+    while [ ${lg_p} -le ${lg_pmax} ] ; do
+      echo "#if (LG_TINY_MIN == ${lg_t} && LG_QUANTUM == ${lg_q} && LG_PAGE == ${lg_p})"
+      echo "#define	SIZE_CLASSES_DEFINED"
+      pow2 ${lg_q}; q=${pow2_result}
+      pow2 ${lg_t}; t=${pow2_result}
+      pow2 ${lg_p}; p=${pow2_result}
+      bin=0
+      psz=0
+      sz=${t}
+      delta=$((${sz} - ${psz}))
+      echo "/*  SIZE_CLASS(bin,	delta,	sz) */"
+      echo "#define	SIZE_CLASSES							\\"
+
+      # Tiny size classes.
+      while [ ${sz} -lt ${q} ] ; do
+        echo "    SIZE_CLASS(${bin},	${delta},	${sz})					\\"
+        bin=$((${bin} + 1))
+        psz=${sz}
+        sz=$((${sz} + ${sz}))
+        delta=$((${sz} - ${psz}))
+      done
+      # Quantum-multiple size classes.  For each doubling of sz, as many as 4
+      # size classes exist.  Their spacing is the greater of:
+      # - q
+      # - sz/4, where sz is a power of 2
+      while [ ${sz} -lt ${p} ] ; do
+        if [ ${sz} -ge $((${q} * 4)) ] ; then
+          i=$((${sz} / 4))
+        else
+          i=${q}
+        fi
+        next_2pow=$((${sz} * 2))
+        while [ ${sz} -lt $next_2pow ] ; do
+          echo "    SIZE_CLASS(${bin},	${delta},	${sz})					\\"
+          bin=$((${bin} + 1))
+          psz=${sz}
+          sz=$((${sz} + ${i}))
+          delta=$((${sz} - ${psz}))
+        done
+      done
+      echo
+      echo "#define	NBINS		${bin}"
+      echo "#define	SMALL_MAXCLASS	${psz}"
+      echo "#endif"
+      echo
+      lg_p=$((${lg_p} + 1))
+    done
+    lg_t=$((${lg_t} + 1))
+  done
+  lg_q=$((${lg_q} + 1))
+done
+
+cat <<EOF
+#ifndef SIZE_CLASSES_DEFINED
+#  error "No size class definitions match configuration"
+#endif
+#undef SIZE_CLASSES_DEFINED
+/*
+ * The small_size2bin lookup table uses uint8_t to encode each bin index, so we
+ * cannot support more than 256 small size classes.  Further constrain NBINS to
+ * 255 to support prof_promote, since all small size classes, plus a "not
+ * small" size class must be stored in 8 bits of arena_chunk_map_t's bits
+ * field.
+ */
+#if (NBINS > 255)
+#  error "Too many small size classes"
+#endif
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
+EOF
diff --git a/extra/jemalloc/include/jemalloc/internal/stats.h b/extra/jemalloc/include/jemalloc/internal/stats.h
new file mode 100644
index 00000000000..27f68e3681c
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/stats.h
@@ -0,0 +1,173 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+typedef struct tcache_bin_stats_s tcache_bin_stats_t;
+typedef struct malloc_bin_stats_s malloc_bin_stats_t;
+typedef struct malloc_large_stats_s malloc_large_stats_t;
+typedef struct arena_stats_s arena_stats_t;
+typedef struct chunk_stats_s chunk_stats_t;
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+struct tcache_bin_stats_s {
+	/*
+	 * Number of allocation requests that corresponded to the size of this
+	 * bin.
+	 */
+	uint64_t	nrequests;
+};
+
+struct malloc_bin_stats_s {
+	/*
+	 * Current number of bytes allocated, including objects currently
+	 * cached by tcache.
+	 */
+	size_t		allocated;
+
+	/*
+	 * Total number of allocation/deallocation requests served directly by
+	 * the bin.  Note that tcache may allocate an object, then recycle it
+	 * many times, resulting many increments to nrequests, but only one
+	 * each to nmalloc and ndalloc.
+	 */
+	uint64_t	nmalloc;
+	uint64_t	ndalloc;
+
+	/*
+	 * Number of allocation requests that correspond to the size of this
+	 * bin.  This includes requests served by tcache, though tcache only
+	 * periodically merges into this counter.
+	 */
+	uint64_t	nrequests;
+
+	/* Number of tcache fills from this bin. */
+	uint64_t	nfills;
+
+	/* Number of tcache flushes to this bin. */
+	uint64_t	nflushes;
+
+	/* Total number of runs created for this bin's size class. */
+	uint64_t	nruns;
+
+	/*
+	 * Total number of runs reused by extracting them from the runs tree for
+	 * this bin's size class.
+	 */
+	uint64_t	reruns;
+
+	/* Current number of runs in this bin. */
+	size_t		curruns;
+};
+
+struct malloc_large_stats_s {
+	/*
+	 * Total number of allocation/deallocation requests served directly by
+	 * the arena.  Note that tcache may allocate an object, then recycle it
+	 * many times, resulting many increments to nrequests, but only one
+	 * each to nmalloc and ndalloc.
+	 */
+	uint64_t	nmalloc;
+	uint64_t	ndalloc;
+
+	/*
+	 * Number of allocation requests that correspond to this size class.
+	 * This includes requests served by tcache, though tcache only
+	 * periodically merges into this counter.
+	 */
+	uint64_t	nrequests;
+
+	/* Current number of runs of this size class. */
+	size_t		curruns;
+};
+
+struct arena_stats_s {
+	/* Number of bytes currently mapped. */
+	size_t		mapped;
+
+	/*
+	 * Total number of purge sweeps, total number of madvise calls made,
+	 * and total pages purged in order to keep dirty unused memory under
+	 * control.
+	 */
+	uint64_t	npurge;
+	uint64_t	nmadvise;
+	uint64_t	purged;
+
+	/* Per-size-category statistics. */
+	size_t		allocated_large;
+	uint64_t	nmalloc_large;
+	uint64_t	ndalloc_large;
+	uint64_t	nrequests_large;
+
+	/*
+	 * One element for each possible size class, including sizes that
+	 * overlap with bin size classes.  This is necessary because ipalloc()
+	 * sometimes has to use such large objects in order to assure proper
+	 * alignment.
+	 */
+	malloc_large_stats_t	*lstats;
+};
+
+struct chunk_stats_s {
+	/* Number of chunks that were allocated. */
+	uint64_t	nchunks;
+
+	/* High-water mark for number of chunks allocated. */
+	size_t		highchunks;
+
+	/*
+	 * Current number of chunks allocated.  This value isn't maintained for
+	 * any other purpose, so keep track of it in order to be able to set
+	 * highchunks.
+	 */
+	size_t		curchunks;
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+extern bool	opt_stats_print;
+
+extern size_t	stats_cactive;
+
+void	stats_print(void (*write)(void *, const char *), void *cbopaque,
+    const char *opts);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+size_t	stats_cactive_get(void);
+void	stats_cactive_add(size_t size);
+void	stats_cactive_sub(size_t size);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_STATS_C_))
+JEMALLOC_INLINE size_t
+stats_cactive_get(void)
+{
+
+	return (atomic_read_z(&stats_cactive));
+}
+
+JEMALLOC_INLINE void
+stats_cactive_add(size_t size)
+{
+
+	atomic_add_z(&stats_cactive, size);
+}
+
+JEMALLOC_INLINE void
+stats_cactive_sub(size_t size)
+{
+
+	atomic_sub_z(&stats_cactive, size);
+}
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/internal/tcache.h b/extra/jemalloc/include/jemalloc/internal/tcache.h
new file mode 100644
index 00000000000..ba36204ff21
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/tcache.h
@@ -0,0 +1,442 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+typedef struct tcache_bin_info_s tcache_bin_info_t;
+typedef struct tcache_bin_s tcache_bin_t;
+typedef struct tcache_s tcache_t;
+
+/*
+ * tcache pointers close to NULL are used to encode state information that is
+ * used for two purposes: preventing thread caching on a per thread basis and
+ * cleaning up during thread shutdown.
+ */
+#define	TCACHE_STATE_DISABLED		((tcache_t *)(uintptr_t)1)
+#define	TCACHE_STATE_REINCARNATED	((tcache_t *)(uintptr_t)2)
+#define	TCACHE_STATE_PURGATORY		((tcache_t *)(uintptr_t)3)
+#define	TCACHE_STATE_MAX		TCACHE_STATE_PURGATORY
+
+/*
+ * Absolute maximum number of cache slots for each small bin in the thread
+ * cache.  This is an additional constraint beyond that imposed as: twice the
+ * number of regions per run for this size class.
+ *
+ * This constant must be an even number.
+ */
+#define	TCACHE_NSLOTS_SMALL_MAX		200
+
+/* Number of cache slots for large size classes. */
+#define	TCACHE_NSLOTS_LARGE		20
+
+/* (1U << opt_lg_tcache_max) is used to compute tcache_maxclass. */
+#define	LG_TCACHE_MAXCLASS_DEFAULT	15
+
+/*
+ * TCACHE_GC_SWEEP is the approximate number of allocation events between
+ * full GC sweeps.  Integer rounding may cause the actual number to be
+ * slightly higher, since GC is performed incrementally.
+ */
+#define	TCACHE_GC_SWEEP			8192
+
+/* Number of tcache allocation/deallocation events between incremental GCs. */
+#define	TCACHE_GC_INCR							\
+    ((TCACHE_GC_SWEEP / NBINS) + ((TCACHE_GC_SWEEP / NBINS == 0) ? 0 : 1))
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+typedef enum {
+	tcache_enabled_false   = 0, /* Enable cast to/from bool. */
+	tcache_enabled_true    = 1,
+	tcache_enabled_default = 2
+} tcache_enabled_t;
+
+/*
+ * Read-only information associated with each element of tcache_t's tbins array
+ * is stored separately, mainly to reduce memory usage.
+ */
+struct tcache_bin_info_s {
+	unsigned	ncached_max;	/* Upper limit on ncached. */
+};
+
+struct tcache_bin_s {
+	tcache_bin_stats_t tstats;
+	int		low_water;	/* Min # cached since last GC. */
+	unsigned	lg_fill_div;	/* Fill (ncached_max >> lg_fill_div). */
+	unsigned	ncached;	/* # of cached objects. */
+	void		**avail;	/* Stack of available objects. */
+};
+
+struct tcache_s {
+	ql_elm(tcache_t) link;		/* Used for aggregating stats. */
+	uint64_t	prof_accumbytes;/* Cleared after arena_prof_accum() */
+	arena_t		*arena;		/* This thread's arena. */
+	unsigned	ev_cnt;		/* Event count since incremental GC. */
+	unsigned	next_gc_bin;	/* Next bin to GC. */
+	tcache_bin_t	tbins[1];	/* Dynamically sized. */
+	/*
+	 * The pointer stacks associated with tbins follow as a contiguous
+	 * array.  During tcache initialization, the avail pointer in each
+	 * element of tbins is initialized to point to the proper offset within
+	 * this array.
+	 */
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+extern bool	opt_tcache;
+extern ssize_t	opt_lg_tcache_max;
+
+extern tcache_bin_info_t	*tcache_bin_info;
+
+/*
+ * Number of tcache bins.  There are NBINS small-object bins, plus 0 or more
+ * large-object bins.
+ */
+extern size_t			nhbins;
+
+/* Maximum cached size class. */
+extern size_t			tcache_maxclass;
+
+size_t	tcache_salloc(const void *ptr);
+void	tcache_event_hard(tcache_t *tcache);
+void	*tcache_alloc_small_hard(tcache_t *tcache, tcache_bin_t *tbin,
+    size_t binind);
+void	tcache_bin_flush_small(tcache_bin_t *tbin, size_t binind, unsigned rem,
+    tcache_t *tcache);
+void	tcache_bin_flush_large(tcache_bin_t *tbin, size_t binind, unsigned rem,
+    tcache_t *tcache);
+void	tcache_arena_associate(tcache_t *tcache, arena_t *arena);
+void	tcache_arena_dissociate(tcache_t *tcache);
+tcache_t *tcache_create(arena_t *arena);
+void	tcache_destroy(tcache_t *tcache);
+void	tcache_thread_cleanup(void *arg);
+void	tcache_stats_merge(tcache_t *tcache, arena_t *arena);
+bool	tcache_boot0(void);
+bool	tcache_boot1(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+malloc_tsd_protos(JEMALLOC_ATTR(unused), tcache, tcache_t *)
+malloc_tsd_protos(JEMALLOC_ATTR(unused), tcache_enabled, tcache_enabled_t)
+
+void	tcache_event(tcache_t *tcache);
+void	tcache_flush(void);
+bool	tcache_enabled_get(void);
+tcache_t *tcache_get(bool create);
+void	tcache_enabled_set(bool enabled);
+void	*tcache_alloc_easy(tcache_bin_t *tbin);
+void	*tcache_alloc_small(tcache_t *tcache, size_t size, bool zero);
+void	*tcache_alloc_large(tcache_t *tcache, size_t size, bool zero);
+void	tcache_dalloc_small(tcache_t *tcache, void *ptr, size_t binind);
+void	tcache_dalloc_large(tcache_t *tcache, void *ptr, size_t size);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_TCACHE_C_))
+/* Map of thread-specific caches. */
+malloc_tsd_externs(tcache, tcache_t *)
+malloc_tsd_funcs(JEMALLOC_ALWAYS_INLINE, tcache, tcache_t *, NULL,
+    tcache_thread_cleanup)
+/* Per thread flag that allows thread caches to be disabled. */
+malloc_tsd_externs(tcache_enabled, tcache_enabled_t)
+malloc_tsd_funcs(JEMALLOC_ALWAYS_INLINE, tcache_enabled, tcache_enabled_t,
+    tcache_enabled_default, malloc_tsd_no_cleanup)
+
+JEMALLOC_INLINE void
+tcache_flush(void)
+{
+	tcache_t *tcache;
+
+	cassert(config_tcache);
+
+	tcache = *tcache_tsd_get();
+	if ((uintptr_t)tcache <= (uintptr_t)TCACHE_STATE_MAX)
+		return;
+	tcache_destroy(tcache);
+	tcache = NULL;
+	tcache_tsd_set(&tcache);
+}
+
+JEMALLOC_INLINE bool
+tcache_enabled_get(void)
+{
+	tcache_enabled_t tcache_enabled;
+
+	cassert(config_tcache);
+
+	tcache_enabled = *tcache_enabled_tsd_get();
+	if (tcache_enabled == tcache_enabled_default) {
+		tcache_enabled = (tcache_enabled_t)opt_tcache;
+		tcache_enabled_tsd_set(&tcache_enabled);
+	}
+
+	return ((bool)tcache_enabled);
+}
+
+JEMALLOC_INLINE void
+tcache_enabled_set(bool enabled)
+{
+	tcache_enabled_t tcache_enabled;
+	tcache_t *tcache;
+
+	cassert(config_tcache);
+
+	tcache_enabled = (tcache_enabled_t)enabled;
+	tcache_enabled_tsd_set(&tcache_enabled);
+	tcache = *tcache_tsd_get();
+	if (enabled) {
+		if (tcache == TCACHE_STATE_DISABLED) {
+			tcache = NULL;
+			tcache_tsd_set(&tcache);
+		}
+	} else /* disabled */ {
+		if (tcache > TCACHE_STATE_MAX) {
+			tcache_destroy(tcache);
+			tcache = NULL;
+		}
+		if (tcache == NULL) {
+			tcache = TCACHE_STATE_DISABLED;
+			tcache_tsd_set(&tcache);
+		}
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE tcache_t *
+tcache_get(bool create)
+{
+	tcache_t *tcache;
+
+	if (config_tcache == false)
+		return (NULL);
+	if (config_lazy_lock && isthreaded == false)
+		return (NULL);
+
+	tcache = *tcache_tsd_get();
+	if ((uintptr_t)tcache <= (uintptr_t)TCACHE_STATE_MAX) {
+		if (tcache == TCACHE_STATE_DISABLED)
+			return (NULL);
+		if (tcache == NULL) {
+			if (create == false) {
+				/*
+				 * Creating a tcache here would cause
+				 * allocation as a side effect of free().
+				 * Ordinarily that would be okay since
+				 * tcache_create() failure is a soft failure
+				 * that doesn't propagate.  However, if TLS
+				 * data are freed via free() as in glibc,
+				 * subtle corruption could result from setting
+				 * a TLS variable after its backing memory is
+				 * freed.
+				 */
+				return (NULL);
+			}
+			if (tcache_enabled_get() == false) {
+				tcache_enabled_set(false); /* Memoize. */
+				return (NULL);
+			}
+			return (tcache_create(choose_arena(NULL)));
+		}
+		if (tcache == TCACHE_STATE_PURGATORY) {
+			/*
+			 * Make a note that an allocator function was called
+			 * after tcache_thread_cleanup() was called.
+			 */
+			tcache = TCACHE_STATE_REINCARNATED;
+			tcache_tsd_set(&tcache);
+			return (NULL);
+		}
+		if (tcache == TCACHE_STATE_REINCARNATED)
+			return (NULL);
+		not_reached();
+	}
+
+	return (tcache);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tcache_event(tcache_t *tcache)
+{
+
+	if (TCACHE_GC_INCR == 0)
+		return;
+
+	tcache->ev_cnt++;
+	assert(tcache->ev_cnt <= TCACHE_GC_INCR);
+	if (tcache->ev_cnt == TCACHE_GC_INCR)
+		tcache_event_hard(tcache);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+tcache_alloc_easy(tcache_bin_t *tbin)
+{
+	void *ret;
+
+	if (tbin->ncached == 0) {
+		tbin->low_water = -1;
+		return (NULL);
+	}
+	tbin->ncached--;
+	if ((int)tbin->ncached < tbin->low_water)
+		tbin->low_water = tbin->ncached;
+	ret = tbin->avail[tbin->ncached];
+	return (ret);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+tcache_alloc_small(tcache_t *tcache, size_t size, bool zero)
+{
+	void *ret;
+	size_t binind;
+	tcache_bin_t *tbin;
+
+	binind = SMALL_SIZE2BIN(size);
+	assert(binind < NBINS);
+	tbin = &tcache->tbins[binind];
+	ret = tcache_alloc_easy(tbin);
+	if (ret == NULL) {
+		ret = tcache_alloc_small_hard(tcache, tbin, binind);
+		if (ret == NULL)
+			return (NULL);
+	}
+	assert(tcache_salloc(ret) == arena_bin_info[binind].reg_size);
+
+	if (zero == false) {
+		if (config_fill) {
+			if (opt_junk) {
+				arena_alloc_junk_small(ret,
+				    &arena_bin_info[binind], false);
+			} else if (opt_zero)
+				memset(ret, 0, size);
+		}
+	} else {
+		if (config_fill && opt_junk) {
+			arena_alloc_junk_small(ret, &arena_bin_info[binind],
+			    true);
+		}
+		VALGRIND_MAKE_MEM_UNDEFINED(ret, size);
+		memset(ret, 0, size);
+	}
+	VALGRIND_MAKE_MEM_UNDEFINED(ret, size);
+
+	if (config_stats)
+		tbin->tstats.nrequests++;
+	if (config_prof)
+		tcache->prof_accumbytes += arena_bin_info[binind].reg_size;
+	tcache_event(tcache);
+	return (ret);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+tcache_alloc_large(tcache_t *tcache, size_t size, bool zero)
+{
+	void *ret;
+	size_t binind;
+	tcache_bin_t *tbin;
+
+	size = PAGE_CEILING(size);
+	assert(size <= tcache_maxclass);
+	binind = NBINS + (size >> LG_PAGE) - 1;
+	assert(binind < nhbins);
+	tbin = &tcache->tbins[binind];
+	ret = tcache_alloc_easy(tbin);
+	if (ret == NULL) {
+		/*
+		 * Only allocate one large object at a time, because it's quite
+		 * expensive to create one and not use it.
+		 */
+		ret = arena_malloc_large(tcache->arena, size, zero);
+		if (ret == NULL)
+			return (NULL);
+	} else {
+		if (config_prof && prof_promote && size == PAGE) {
+			arena_chunk_t *chunk =
+			    (arena_chunk_t *)CHUNK_ADDR2BASE(ret);
+			size_t pageind = (((uintptr_t)ret - (uintptr_t)chunk) >>
+			    LG_PAGE);
+			arena_mapbits_large_binind_set(chunk, pageind,
+			    BININD_INVALID);
+		}
+		if (zero == false) {
+			if (config_fill) {
+				if (opt_junk)
+					memset(ret, 0xa5, size);
+				else if (opt_zero)
+					memset(ret, 0, size);
+			}
+		} else {
+			VALGRIND_MAKE_MEM_UNDEFINED(ret, size);
+			memset(ret, 0, size);
+		}
+		VALGRIND_MAKE_MEM_UNDEFINED(ret, size);
+
+		if (config_stats)
+			tbin->tstats.nrequests++;
+		if (config_prof)
+			tcache->prof_accumbytes += size;
+	}
+
+	tcache_event(tcache);
+	return (ret);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tcache_dalloc_small(tcache_t *tcache, void *ptr, size_t binind)
+{
+	tcache_bin_t *tbin;
+	tcache_bin_info_t *tbin_info;
+
+	assert(tcache_salloc(ptr) <= SMALL_MAXCLASS);
+
+	if (config_fill && opt_junk)
+		arena_dalloc_junk_small(ptr, &arena_bin_info[binind]);
+
+	tbin = &tcache->tbins[binind];
+	tbin_info = &tcache_bin_info[binind];
+	if (tbin->ncached == tbin_info->ncached_max) {
+		tcache_bin_flush_small(tbin, binind, (tbin_info->ncached_max >>
+		    1), tcache);
+	}
+	assert(tbin->ncached < tbin_info->ncached_max);
+	tbin->avail[tbin->ncached] = ptr;
+	tbin->ncached++;
+
+	tcache_event(tcache);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tcache_dalloc_large(tcache_t *tcache, void *ptr, size_t size)
+{
+	size_t binind;
+	tcache_bin_t *tbin;
+	tcache_bin_info_t *tbin_info;
+
+	assert((size & PAGE_MASK) == 0);
+	assert(tcache_salloc(ptr) > SMALL_MAXCLASS);
+	assert(tcache_salloc(ptr) <= tcache_maxclass);
+
+	binind = NBINS + (size >> LG_PAGE) - 1;
+
+	if (config_fill && opt_junk)
+		memset(ptr, 0x5a, size);
+
+	tbin = &tcache->tbins[binind];
+	tbin_info = &tcache_bin_info[binind];
+	if (tbin->ncached == tbin_info->ncached_max) {
+		tcache_bin_flush_large(tbin, binind, (tbin_info->ncached_max >>
+		    1), tcache);
+	}
+	assert(tbin->ncached < tbin_info->ncached_max);
+	tbin->avail[tbin->ncached] = ptr;
+	tbin->ncached++;
+
+	tcache_event(tcache);
+}
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/internal/tsd.h b/extra/jemalloc/include/jemalloc/internal/tsd.h
new file mode 100644
index 00000000000..0037cf35e70
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/tsd.h
@@ -0,0 +1,397 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+/* Maximum number of malloc_tsd users with cleanup functions. */
+#define	MALLOC_TSD_CLEANUPS_MAX	8
+
+typedef bool (*malloc_tsd_cleanup_t)(void);
+
+/*
+ * TLS/TSD-agnostic macro-based implementation of thread-specific data.  There
+ * are four macros that support (at least) three use cases: file-private,
+ * library-private, and library-private inlined.  Following is an example
+ * library-private tsd variable:
+ *
+ * In example.h:
+ *   typedef struct {
+ *           int x;
+ *           int y;
+ *   } example_t;
+ *   #define EX_INITIALIZER JEMALLOC_CONCAT({0, 0})
+ *   malloc_tsd_protos(, example, example_t *)
+ *   malloc_tsd_externs(example, example_t *)
+ * In example.c:
+ *   malloc_tsd_data(, example, example_t *, EX_INITIALIZER)
+ *   malloc_tsd_funcs(, example, example_t *, EX_INITIALIZER,
+ *       example_tsd_cleanup)
+ *
+ * The result is a set of generated functions, e.g.:
+ *
+ *   bool example_tsd_boot(void) {...}
+ *   example_t **example_tsd_get() {...}
+ *   void example_tsd_set(example_t **val) {...}
+ *
+ * Note that all of the functions deal in terms of (a_type *) rather than
+ * (a_type)  so that it is possible to support non-pointer types (unlike
+ * pthreads TSD).  example_tsd_cleanup() is passed an (a_type *) pointer that is
+ * cast to (void *).  This means that the cleanup function needs to cast *and*
+ * dereference the function argument, e.g.:
+ *
+ *   void
+ *   example_tsd_cleanup(void *arg)
+ *   {
+ *           example_t *example = *(example_t **)arg;
+ *
+ *           [...]
+ *           if ([want the cleanup function to be called again]) {
+ *                   example_tsd_set(&example);
+ *           }
+ *   }
+ *
+ * If example_tsd_set() is called within example_tsd_cleanup(), it will be
+ * called again.  This is similar to how pthreads TSD destruction works, except
+ * that pthreads only calls the cleanup function again if the value was set to
+ * non-NULL.
+ */
+
+/* malloc_tsd_protos(). */
+#define	malloc_tsd_protos(a_attr, a_name, a_type)			\
+a_attr bool								\
+a_name##_tsd_boot(void);						\
+a_attr a_type *								\
+a_name##_tsd_get(void);							\
+a_attr void								\
+a_name##_tsd_set(a_type *val);
+
+/* malloc_tsd_externs(). */
+#ifdef JEMALLOC_MALLOC_THREAD_CLEANUP
+#define	malloc_tsd_externs(a_name, a_type)				\
+extern __thread a_type	a_name##_tls;					\
+extern __thread bool	a_name##_initialized;				\
+extern bool		a_name##_booted;
+#elif (defined(JEMALLOC_TLS))
+#define	malloc_tsd_externs(a_name, a_type)				\
+extern __thread a_type	a_name##_tls;					\
+extern pthread_key_t	a_name##_tsd;					\
+extern bool		a_name##_booted;
+#elif (defined(_WIN32))
+#define malloc_tsd_externs(a_name, a_type)				\
+extern DWORD		a_name##_tsd;					\
+extern bool		a_name##_booted;
+#else
+#define	malloc_tsd_externs(a_name, a_type)				\
+extern pthread_key_t	a_name##_tsd;					\
+extern bool		a_name##_booted;
+#endif
+
+/* malloc_tsd_data(). */
+#ifdef JEMALLOC_MALLOC_THREAD_CLEANUP
+#define	malloc_tsd_data(a_attr, a_name, a_type, a_initializer)		\
+a_attr __thread a_type JEMALLOC_TLS_MODEL				\
+    a_name##_tls = a_initializer;					\
+a_attr __thread bool JEMALLOC_TLS_MODEL					\
+    a_name##_initialized = false;					\
+a_attr bool		a_name##_booted = false;
+#elif (defined(JEMALLOC_TLS))
+#define	malloc_tsd_data(a_attr, a_name, a_type, a_initializer)		\
+a_attr __thread a_type JEMALLOC_TLS_MODEL				\
+    a_name##_tls = a_initializer;					\
+a_attr pthread_key_t	a_name##_tsd;					\
+a_attr bool		a_name##_booted = false;
+#elif (defined(_WIN32))
+#define	malloc_tsd_data(a_attr, a_name, a_type, a_initializer)		\
+a_attr DWORD		a_name##_tsd;					\
+a_attr bool		a_name##_booted = false;
+#else
+#define	malloc_tsd_data(a_attr, a_name, a_type, a_initializer)		\
+a_attr pthread_key_t	a_name##_tsd;					\
+a_attr bool		a_name##_booted = false;
+#endif
+
+/* malloc_tsd_funcs(). */
+#ifdef JEMALLOC_MALLOC_THREAD_CLEANUP
+#define	malloc_tsd_funcs(a_attr, a_name, a_type, a_initializer,		\
+    a_cleanup)								\
+/* Initialization/cleanup. */						\
+a_attr bool								\
+a_name##_tsd_cleanup_wrapper(void)					\
+{									\
+									\
+	if (a_name##_initialized) {					\
+		a_name##_initialized = false;				\
+		a_cleanup(&a_name##_tls);				\
+	}								\
+	return (a_name##_initialized);					\
+}									\
+a_attr bool								\
+a_name##_tsd_boot(void)							\
+{									\
+									\
+	if (a_cleanup != malloc_tsd_no_cleanup) {			\
+		malloc_tsd_cleanup_register(				\
+		    &a_name##_tsd_cleanup_wrapper);			\
+	}								\
+	a_name##_booted = true;						\
+	return (false);							\
+}									\
+/* Get/set. */								\
+a_attr a_type *								\
+a_name##_tsd_get(void)							\
+{									\
+									\
+	assert(a_name##_booted);					\
+	return (&a_name##_tls);						\
+}									\
+a_attr void								\
+a_name##_tsd_set(a_type *val)						\
+{									\
+									\
+	assert(a_name##_booted);					\
+	a_name##_tls = (*val);						\
+	if (a_cleanup != malloc_tsd_no_cleanup)				\
+		a_name##_initialized = true;				\
+}
+#elif (defined(JEMALLOC_TLS))
+#define	malloc_tsd_funcs(a_attr, a_name, a_type, a_initializer,		\
+    a_cleanup)								\
+/* Initialization/cleanup. */						\
+a_attr bool								\
+a_name##_tsd_boot(void)							\
+{									\
+									\
+	if (a_cleanup != malloc_tsd_no_cleanup) {			\
+		if (pthread_key_create(&a_name##_tsd, a_cleanup) != 0)	\
+			return (true);					\
+	}								\
+	a_name##_booted = true;						\
+	return (false);							\
+}									\
+/* Get/set. */								\
+a_attr a_type *								\
+a_name##_tsd_get(void)							\
+{									\
+									\
+	assert(a_name##_booted);					\
+	return (&a_name##_tls);						\
+}									\
+a_attr void								\
+a_name##_tsd_set(a_type *val)						\
+{									\
+									\
+	assert(a_name##_booted);					\
+	a_name##_tls = (*val);						\
+	if (a_cleanup != malloc_tsd_no_cleanup) {			\
+		if (pthread_setspecific(a_name##_tsd,			\
+		    (void *)(&a_name##_tls))) {				\
+			malloc_write("<jemalloc>: Error"		\
+			    " setting TSD for "#a_name"\n");		\
+			if (opt_abort)					\
+				abort();				\
+		}							\
+	}								\
+}
+#elif (defined(_WIN32))
+#define	malloc_tsd_funcs(a_attr, a_name, a_type, a_initializer,		\
+    a_cleanup)								\
+/* Data structure. */							\
+typedef struct {							\
+	bool	initialized;						\
+	a_type	val;							\
+} a_name##_tsd_wrapper_t;						\
+/* Initialization/cleanup. */						\
+a_attr bool								\
+a_name##_tsd_cleanup_wrapper(void)					\
+{									\
+	a_name##_tsd_wrapper_t *wrapper;				\
+									\
+	wrapper = (a_name##_tsd_wrapper_t *) TlsGetValue(a_name##_tsd);	\
+	if (wrapper == NULL)						\
+		return (false);						\
+	if (a_cleanup != malloc_tsd_no_cleanup &&			\
+	    wrapper->initialized) {					\
+		a_type val = wrapper->val;				\
+		a_type tsd_static_data = a_initializer;			\
+		wrapper->initialized = false;				\
+		wrapper->val = tsd_static_data;				\
+		a_cleanup(&val);					\
+		if (wrapper->initialized) {				\
+			/* Trigger another cleanup round. */		\
+			return (true);					\
+		}							\
+	}								\
+	malloc_tsd_dalloc(wrapper);					\
+	return (false);							\
+}									\
+a_attr bool								\
+a_name##_tsd_boot(void)							\
+{									\
+									\
+	a_name##_tsd = TlsAlloc();					\
+	if (a_name##_tsd == TLS_OUT_OF_INDEXES)				\
+		return (true);						\
+	if (a_cleanup != malloc_tsd_no_cleanup) {			\
+		malloc_tsd_cleanup_register(				\
+		    &a_name##_tsd_cleanup_wrapper);			\
+	}								\
+	a_name##_booted = true;						\
+	return (false);							\
+}									\
+/* Get/set. */								\
+a_attr a_name##_tsd_wrapper_t *						\
+a_name##_tsd_get_wrapper(void)						\
+{									\
+	a_name##_tsd_wrapper_t *wrapper = (a_name##_tsd_wrapper_t *)	\
+	    TlsGetValue(a_name##_tsd);					\
+									\
+	if (wrapper == NULL) {						\
+		wrapper = (a_name##_tsd_wrapper_t *)			\
+		    malloc_tsd_malloc(sizeof(a_name##_tsd_wrapper_t));	\
+		if (wrapper == NULL) {					\
+			malloc_write("<jemalloc>: Error allocating"	\
+			    " TSD for "#a_name"\n");			\
+			abort();					\
+		} else {						\
+			static a_type tsd_static_data = a_initializer;	\
+			wrapper->initialized = false;			\
+			wrapper->val = tsd_static_data;			\
+		}							\
+		if (!TlsSetValue(a_name##_tsd, (void *)wrapper)) {	\
+			malloc_write("<jemalloc>: Error setting"	\
+			    " TSD for "#a_name"\n");			\
+			abort();					\
+		}							\
+	}								\
+	return (wrapper);						\
+}									\
+a_attr a_type *								\
+a_name##_tsd_get(void)							\
+{									\
+	a_name##_tsd_wrapper_t *wrapper;				\
+									\
+	assert(a_name##_booted);					\
+	wrapper = a_name##_tsd_get_wrapper();				\
+	return (&wrapper->val);						\
+}									\
+a_attr void								\
+a_name##_tsd_set(a_type *val)						\
+{									\
+	a_name##_tsd_wrapper_t *wrapper;				\
+									\
+	assert(a_name##_booted);					\
+	wrapper = a_name##_tsd_get_wrapper();				\
+	wrapper->val = *(val);						\
+	if (a_cleanup != malloc_tsd_no_cleanup)				\
+		wrapper->initialized = true;				\
+}
+#else
+#define	malloc_tsd_funcs(a_attr, a_name, a_type, a_initializer,		\
+    a_cleanup)								\
+/* Data structure. */							\
+typedef struct {							\
+	bool	initialized;						\
+	a_type	val;							\
+} a_name##_tsd_wrapper_t;						\
+/* Initialization/cleanup. */						\
+a_attr void								\
+a_name##_tsd_cleanup_wrapper(void *arg)					\
+{									\
+	a_name##_tsd_wrapper_t *wrapper = (a_name##_tsd_wrapper_t *)arg;\
+									\
+	if (a_cleanup != malloc_tsd_no_cleanup &&			\
+	    wrapper->initialized) {					\
+		wrapper->initialized = false;				\
+		a_cleanup(&wrapper->val);				\
+		if (wrapper->initialized) {				\
+			/* Trigger another cleanup round. */		\
+			if (pthread_setspecific(a_name##_tsd,		\
+			    (void *)wrapper)) {				\
+				malloc_write("<jemalloc>: Error"	\
+				    " setting TSD for "#a_name"\n");	\
+				if (opt_abort)				\
+					abort();			\
+			}						\
+			return;						\
+		}							\
+	}								\
+	malloc_tsd_dalloc(wrapper);					\
+}									\
+a_attr bool								\
+a_name##_tsd_boot(void)							\
+{									\
+									\
+	if (pthread_key_create(&a_name##_tsd,				\
+	    a_name##_tsd_cleanup_wrapper) != 0)				\
+		return (true);						\
+	a_name##_booted = true;						\
+	return (false);							\
+}									\
+/* Get/set. */								\
+a_attr a_name##_tsd_wrapper_t *						\
+a_name##_tsd_get_wrapper(void)						\
+{									\
+	a_name##_tsd_wrapper_t *wrapper = (a_name##_tsd_wrapper_t *)	\
+	    pthread_getspecific(a_name##_tsd);				\
+									\
+	if (wrapper == NULL) {						\
+		wrapper = (a_name##_tsd_wrapper_t *)			\
+		    malloc_tsd_malloc(sizeof(a_name##_tsd_wrapper_t));	\
+		if (wrapper == NULL) {					\
+			malloc_write("<jemalloc>: Error allocating"	\
+			    " TSD for "#a_name"\n");			\
+			abort();					\
+		} else {						\
+			static a_type tsd_static_data = a_initializer;	\
+			wrapper->initialized = false;			\
+			wrapper->val = tsd_static_data;			\
+		}							\
+		if (pthread_setspecific(a_name##_tsd,			\
+		    (void *)wrapper)) {					\
+			malloc_write("<jemalloc>: Error setting"	\
+			    " TSD for "#a_name"\n");			\
+			abort();					\
+		}							\
+	}								\
+	return (wrapper);						\
+}									\
+a_attr a_type *								\
+a_name##_tsd_get(void)							\
+{									\
+	a_name##_tsd_wrapper_t *wrapper;				\
+									\
+	assert(a_name##_booted);					\
+	wrapper = a_name##_tsd_get_wrapper();				\
+	return (&wrapper->val);						\
+}									\
+a_attr void								\
+a_name##_tsd_set(a_type *val)						\
+{									\
+	a_name##_tsd_wrapper_t *wrapper;				\
+									\
+	assert(a_name##_booted);					\
+	wrapper = a_name##_tsd_get_wrapper();				\
+	wrapper->val = *(val);						\
+	if (a_cleanup != malloc_tsd_no_cleanup)				\
+		wrapper->initialized = true;				\
+}
+#endif
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+void	*malloc_tsd_malloc(size_t size);
+void	malloc_tsd_dalloc(void *wrapper);
+void	malloc_tsd_no_cleanup(void *);
+void	malloc_tsd_cleanup_register(bool (*f)(void));
+void	malloc_tsd_boot(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/internal/util.h b/extra/jemalloc/include/jemalloc/internal/util.h
new file mode 100644
index 00000000000..8479693631a
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/internal/util.h
@@ -0,0 +1,160 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+/* Size of stack-allocated buffer passed to buferror(). */
+#define	BUFERROR_BUF		64
+
+/*
+ * Size of stack-allocated buffer used by malloc_{,v,vc}printf().  This must be
+ * large enough for all possible uses within jemalloc.
+ */
+#define	MALLOC_PRINTF_BUFSIZE	4096
+
+/*
+ * Wrap a cpp argument that contains commas such that it isn't broken up into
+ * multiple arguments.
+ */
+#define JEMALLOC_CONCAT(...) __VA_ARGS__
+
+/*
+ * Silence compiler warnings due to uninitialized values.  This is used
+ * wherever the compiler fails to recognize that the variable is never used
+ * uninitialized.
+ */
+#ifdef JEMALLOC_CC_SILENCE
+#  define JEMALLOC_CC_SILENCE_INIT(v) = v
+#else
+#  define JEMALLOC_CC_SILENCE_INIT(v)
+#endif
+
+/*
+ * Define a custom assert() in order to reduce the chances of deadlock during
+ * assertion failure.
+ */
+#ifndef assert
+#define	assert(e) do {							\
+	if (config_debug && !(e)) {					\
+		malloc_printf(						\
+		    "<jemalloc>: %s:%d: Failed assertion: \"%s\"\n",	\
+		    __FILE__, __LINE__, #e);				\
+		abort();						\
+	}								\
+} while (0)
+#endif
+
+/* Use to assert a particular configuration, e.g., cassert(config_debug). */
+#define	cassert(c) do {							\
+	if ((c) == false)						\
+		assert(false);						\
+} while (0)
+
+#ifndef not_reached
+#define	not_reached() do {						\
+	if (config_debug) {						\
+		malloc_printf(						\
+		    "<jemalloc>: %s:%d: Unreachable code reached\n",	\
+		    __FILE__, __LINE__);				\
+		abort();						\
+	}								\
+} while (0)
+#endif
+
+#ifndef not_implemented
+#define	not_implemented() do {						\
+	if (config_debug) {						\
+		malloc_printf("<jemalloc>: %s:%d: Not implemented\n",	\
+		    __FILE__, __LINE__);				\
+		abort();						\
+	}								\
+} while (0)
+#endif
+
+#define	assert_not_implemented(e) do {					\
+	if (config_debug && !(e))					\
+		not_implemented();					\
+} while (0)
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+int	buferror(char *buf, size_t buflen);
+uintmax_t	malloc_strtoumax(const char *nptr, char **endptr, int base);
+void	malloc_write(const char *s);
+
+/*
+ * malloc_vsnprintf() supports a subset of snprintf(3) that avoids floating
+ * point math.
+ */
+int	malloc_vsnprintf(char *str, size_t size, const char *format,
+    va_list ap);
+int	malloc_snprintf(char *str, size_t size, const char *format, ...)
+    JEMALLOC_ATTR(format(printf, 3, 4));
+void	malloc_vcprintf(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *format, va_list ap);
+void malloc_cprintf(void (*write)(void *, const char *), void *cbopaque,
+    const char *format, ...) JEMALLOC_ATTR(format(printf, 3, 4));
+void	malloc_printf(const char *format, ...)
+    JEMALLOC_ATTR(format(printf, 1, 2));
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+size_t	pow2_ceil(size_t x);
+void	malloc_write(const char *s);
+void	set_errno(int errnum);
+int	get_errno(void);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_UTIL_C_))
+/* Compute the smallest power of 2 that is >= x. */
+JEMALLOC_INLINE size_t
+pow2_ceil(size_t x)
+{
+
+	x--;
+	x |= x >> 1;
+	x |= x >> 2;
+	x |= x >> 4;
+	x |= x >> 8;
+	x |= x >> 16;
+#if (LG_SIZEOF_PTR == 3)
+	x |= x >> 32;
+#endif
+	x++;
+	return (x);
+}
+
+/* Sets error code */
+JEMALLOC_INLINE void
+set_errno(int errnum)
+{
+
+#ifdef _WIN32
+	SetLastError(errnum);
+#else
+	errno = errnum;
+#endif
+}
+
+/* Get last error code */
+JEMALLOC_INLINE int
+get_errno(void)
+{
+
+#ifdef _WIN32
+	return (GetLastError());
+#else
+	return (errno);
+#endif
+}
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
diff --git a/extra/jemalloc/include/jemalloc/jemalloc.h.in b/extra/jemalloc/include/jemalloc/jemalloc.h.in
new file mode 100644
index 00000000000..31b1304a20a
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/jemalloc.h.in
@@ -0,0 +1,157 @@
+#ifndef JEMALLOC_H_
+#define	JEMALLOC_H_
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <limits.h>
+#include <strings.h>
+
+#define	JEMALLOC_VERSION "@jemalloc_version@"
+#define	JEMALLOC_VERSION_MAJOR @jemalloc_version_major@
+#define	JEMALLOC_VERSION_MINOR @jemalloc_version_minor@
+#define	JEMALLOC_VERSION_BUGFIX @jemalloc_version_bugfix@
+#define	JEMALLOC_VERSION_NREV @jemalloc_version_nrev@
+#define	JEMALLOC_VERSION_GID "@jemalloc_version_gid@"
+
+#include "jemalloc_defs@install_suffix@.h"
+
+#ifdef JEMALLOC_EXPERIMENTAL
+#define	ALLOCM_LG_ALIGN(la)	(la)
+#if LG_SIZEOF_PTR == 2
+#define	ALLOCM_ALIGN(a)	(ffs(a)-1)
+#else
+#define	ALLOCM_ALIGN(a)	((a < (size_t)INT_MAX) ? ffs(a)-1 : ffs(a>>32)+31)
+#endif
+#define	ALLOCM_ZERO	((int)0x40)
+#define	ALLOCM_NO_MOVE	((int)0x80)
+/* Bias arena index bits so that 0 encodes "ALLOCM_ARENA() unspecified". */
+#define	ALLOCM_ARENA(a)	((int)(((a)+1) << 8))
+
+#define	ALLOCM_SUCCESS		0
+#define	ALLOCM_ERR_OOM		1
+#define	ALLOCM_ERR_NOT_MOVED	2
+#endif
+
+/*
+ * The je_ prefix on the following public symbol declarations is an artifact of
+ * namespace management, and should be omitted in application code unless
+ * JEMALLOC_NO_DEMANGLE is defined (see below).
+ */
+extern JEMALLOC_EXPORT const char	*je_malloc_conf;
+extern JEMALLOC_EXPORT void		(*je_malloc_message)(void *cbopaque,
+    const char *s);
+
+JEMALLOC_EXPORT void	*je_malloc(size_t size) JEMALLOC_ATTR(malloc);
+JEMALLOC_EXPORT void	*je_calloc(size_t num, size_t size)
+    JEMALLOC_ATTR(malloc);
+JEMALLOC_EXPORT int	je_posix_memalign(void **memptr, size_t alignment,
+    size_t size) JEMALLOC_ATTR(nonnull(1));
+JEMALLOC_EXPORT void	*je_aligned_alloc(size_t alignment, size_t size)
+    JEMALLOC_ATTR(malloc);
+JEMALLOC_EXPORT void	*je_realloc(void *ptr, size_t size);
+JEMALLOC_EXPORT void	je_free(void *ptr);
+
+#ifdef JEMALLOC_OVERRIDE_MEMALIGN
+JEMALLOC_EXPORT void *	je_memalign(size_t alignment, size_t size)
+    JEMALLOC_ATTR(malloc);
+#endif
+
+#ifdef JEMALLOC_OVERRIDE_VALLOC
+JEMALLOC_EXPORT void *	je_valloc(size_t size) JEMALLOC_ATTR(malloc);
+#endif
+
+JEMALLOC_EXPORT size_t	je_malloc_usable_size(
+    JEMALLOC_USABLE_SIZE_CONST void *ptr);
+JEMALLOC_EXPORT void	je_malloc_stats_print(void (*write_cb)(void *,
+    const char *), void *je_cbopaque, const char *opts);
+JEMALLOC_EXPORT int	je_mallctl(const char *name, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen);
+JEMALLOC_EXPORT int	je_mallctlnametomib(const char *name, size_t *mibp,
+    size_t *miblenp);
+JEMALLOC_EXPORT int	je_mallctlbymib(const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen);
+
+#ifdef JEMALLOC_EXPERIMENTAL
+JEMALLOC_EXPORT int	je_allocm(void **ptr, size_t *rsize, size_t size,
+    int flags) JEMALLOC_ATTR(nonnull(1));
+JEMALLOC_EXPORT int	je_rallocm(void **ptr, size_t *rsize, size_t size,
+    size_t extra, int flags) JEMALLOC_ATTR(nonnull(1));
+JEMALLOC_EXPORT int	je_sallocm(const void *ptr, size_t *rsize, int flags)
+    JEMALLOC_ATTR(nonnull(1));
+JEMALLOC_EXPORT int	je_dallocm(void *ptr, int flags)
+    JEMALLOC_ATTR(nonnull(1));
+JEMALLOC_EXPORT int	je_nallocm(size_t *rsize, size_t size, int flags);
+#endif
+
+/*
+ * By default application code must explicitly refer to mangled symbol names,
+ * so that it is possible to use jemalloc in conjunction with another allocator
+ * in the same application.  Define JEMALLOC_MANGLE in order to cause automatic
+ * name mangling that matches the API prefixing that happened as a result of
+ * --with-mangling and/or --with-jemalloc-prefix configuration settings.
+ */
+#ifdef JEMALLOC_MANGLE
+#ifndef JEMALLOC_NO_DEMANGLE
+#define	JEMALLOC_NO_DEMANGLE
+#endif
+#define	malloc_conf je_malloc_conf
+#define	malloc_message je_malloc_message
+#define	malloc je_malloc
+#define	calloc je_calloc
+#define	posix_memalign je_posix_memalign
+#define	aligned_alloc je_aligned_alloc
+#define	realloc je_realloc
+#define	free je_free
+#define	malloc_usable_size je_malloc_usable_size
+#define	malloc_stats_print je_malloc_stats_print
+#define	mallctl je_mallctl
+#define	mallctlnametomib je_mallctlnametomib
+#define	mallctlbymib je_mallctlbymib
+#define	memalign je_memalign
+#define	valloc je_valloc
+#ifdef JEMALLOC_EXPERIMENTAL
+#define	allocm je_allocm
+#define	rallocm je_rallocm
+#define	sallocm je_sallocm
+#define	dallocm je_dallocm
+#define	nallocm je_nallocm
+#endif
+#endif
+
+/*
+ * The je_* macros can be used as stable alternative names for the public
+ * jemalloc API if JEMALLOC_NO_DEMANGLE is defined.  This is primarily meant
+ * for use in jemalloc itself, but it can be used by application code to
+ * provide isolation from the name mangling specified via --with-mangling
+ * and/or --with-jemalloc-prefix.
+ */
+#ifndef JEMALLOC_NO_DEMANGLE
+#undef je_malloc_conf
+#undef je_malloc_message
+#undef je_malloc
+#undef je_calloc
+#undef je_posix_memalign
+#undef je_aligned_alloc
+#undef je_realloc
+#undef je_free
+#undef je_malloc_usable_size
+#undef je_malloc_stats_print
+#undef je_mallctl
+#undef je_mallctlnametomib
+#undef je_mallctlbymib
+#undef je_memalign
+#undef je_valloc
+#ifdef JEMALLOC_EXPERIMENTAL
+#undef je_allocm
+#undef je_rallocm
+#undef je_sallocm
+#undef je_dallocm
+#undef je_nallocm
+#endif
+#endif
+
+#ifdef __cplusplus
+};
+#endif
+#endif /* JEMALLOC_H_ */
diff --git a/extra/jemalloc/include/jemalloc/jemalloc_defs.h.in b/extra/jemalloc/include/jemalloc/jemalloc_defs.h.in
new file mode 100644
index 00000000000..3fcf93ce5d2
--- /dev/null
+++ b/extra/jemalloc/include/jemalloc/jemalloc_defs.h.in
@@ -0,0 +1,267 @@
+/*
+ * If JEMALLOC_PREFIX is defined via --with-jemalloc-prefix, it will cause all
+ * public APIs to be prefixed.  This makes it possible, with some care, to use
+ * multiple allocators simultaneously.
+ */
+#undef JEMALLOC_PREFIX
+#undef JEMALLOC_CPREFIX
+
+/*
+ * Name mangling for public symbols is controlled by --with-mangling and
+ * --with-jemalloc-prefix.  With default settings the je_ prefix is stripped by
+ * these macro definitions.
+ */
+#undef je_malloc_conf
+#undef je_malloc_message
+#undef je_malloc
+#undef je_calloc
+#undef je_posix_memalign
+#undef je_aligned_alloc
+#undef je_realloc
+#undef je_free
+#undef je_malloc_usable_size
+#undef je_malloc_stats_print
+#undef je_mallctl
+#undef je_mallctlnametomib
+#undef je_mallctlbymib
+#undef je_memalign
+#undef je_valloc
+#undef je_allocm
+#undef je_rallocm
+#undef je_sallocm
+#undef je_dallocm
+#undef je_nallocm
+
+/*
+ * JEMALLOC_PRIVATE_NAMESPACE is used as a prefix for all library-private APIs.
+ * For shared libraries, symbol visibility mechanisms prevent these symbols
+ * from being exported, but for static libraries, naming collisions are a real
+ * possibility.
+ */
+#undef JEMALLOC_PRIVATE_NAMESPACE
+#undef JEMALLOC_N
+
+/*
+ * Hyper-threaded CPUs may need a special instruction inside spin loops in
+ * order to yield to another virtual CPU.
+ */
+#undef CPU_SPINWAIT
+
+/* Defined if the equivalent of FreeBSD's atomic(9) functions are available. */
+#undef JEMALLOC_ATOMIC9
+
+/*
+ * Defined if OSAtomic*() functions are available, as provided by Darwin, and
+ * documented in the atomic(3) manual page.
+ */
+#undef JEMALLOC_OSATOMIC
+
+/*
+ * Defined if __sync_add_and_fetch(uint32_t *, uint32_t) and
+ * __sync_sub_and_fetch(uint32_t *, uint32_t) are available, despite
+ * __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4 not being defined (which means the
+ * functions are defined in libgcc instead of being inlines)
+ */
+#undef JE_FORCE_SYNC_COMPARE_AND_SWAP_4
+
+/*
+ * Defined if __sync_add_and_fetch(uint64_t *, uint64_t) and
+ * __sync_sub_and_fetch(uint64_t *, uint64_t) are available, despite
+ * __GCC_HAVE_SYNC_COMPARE_AND_SWAP_8 not being defined (which means the
+ * functions are defined in libgcc instead of being inlines)
+ */
+#undef JE_FORCE_SYNC_COMPARE_AND_SWAP_8
+
+/*
+ * Defined if OSSpin*() functions are available, as provided by Darwin, and
+ * documented in the spinlock(3) manual page.
+ */
+#undef JEMALLOC_OSSPIN
+
+/*
+ * Defined if _malloc_thread_cleanup() exists.  At least in the case of
+ * FreeBSD, pthread_key_create() allocates, which if used during malloc
+ * bootstrapping will cause recursion into the pthreads library.  Therefore, if
+ * _malloc_thread_cleanup() exists, use it as the basis for thread cleanup in
+ * malloc_tsd.
+ */
+#undef JEMALLOC_MALLOC_THREAD_CLEANUP
+
+/*
+ * Defined if threaded initialization is known to be safe on this platform.
+ * Among other things, it must be possible to initialize a mutex without
+ * triggering allocation in order for threaded allocation to be safe.
+ */
+#undef JEMALLOC_THREADED_INIT
+
+/*
+ * Defined if the pthreads implementation defines
+ * _pthread_mutex_init_calloc_cb(), in which case the function is used in order
+ * to avoid recursive allocation during mutex initialization.
+ */
+#undef JEMALLOC_MUTEX_INIT_CB
+
+/* Defined if __attribute__((...)) syntax is supported. */
+#undef JEMALLOC_HAVE_ATTR
+#ifdef JEMALLOC_HAVE_ATTR
+#  define JEMALLOC_ATTR(s) __attribute__((s))
+#  define JEMALLOC_EXPORT JEMALLOC_ATTR(visibility("default"))
+#  define JEMALLOC_ALIGNED(s) JEMALLOC_ATTR(aligned(s))
+#  define JEMALLOC_SECTION(s) JEMALLOC_ATTR(section(s))
+#  define JEMALLOC_NOINLINE JEMALLOC_ATTR(noinline)
+#elif _MSC_VER
+#  define JEMALLOC_ATTR(s)
+#  ifdef DLLEXPORT
+#    define JEMALLOC_EXPORT __declspec(dllexport)
+#  else
+#    define JEMALLOC_EXPORT __declspec(dllimport)
+#  endif
+#  define JEMALLOC_ALIGNED(s) __declspec(align(s))
+#  define JEMALLOC_SECTION(s) __declspec(allocate(s))
+#  define JEMALLOC_NOINLINE __declspec(noinline)
+#else
+#  define JEMALLOC_ATTR(s)
+#  define JEMALLOC_EXPORT
+#  define JEMALLOC_ALIGNED(s)
+#  define JEMALLOC_SECTION(s)
+#  define JEMALLOC_NOINLINE
+#endif
+
+/* Defined if sbrk() is supported. */
+#undef JEMALLOC_HAVE_SBRK
+
+/* Non-empty if the tls_model attribute is supported. */
+#undef JEMALLOC_TLS_MODEL
+
+/* JEMALLOC_CC_SILENCE enables code that silences unuseful compiler warnings. */
+#undef JEMALLOC_CC_SILENCE
+
+/*
+ * JEMALLOC_DEBUG enables assertions and other sanity checks, and disables
+ * inline functions.
+ */
+#undef JEMALLOC_DEBUG
+
+/* JEMALLOC_STATS enables statistics calculation. */
+#undef JEMALLOC_STATS
+
+/* JEMALLOC_PROF enables allocation profiling. */
+#undef JEMALLOC_PROF
+
+/* Use libunwind for profile backtracing if defined. */
+#undef JEMALLOC_PROF_LIBUNWIND
+
+/* Use libgcc for profile backtracing if defined. */
+#undef JEMALLOC_PROF_LIBGCC
+
+/* Use gcc intrinsics for profile backtracing if defined. */
+#undef JEMALLOC_PROF_GCC
+
+/*
+ * JEMALLOC_TCACHE enables a thread-specific caching layer for small objects.
+ * This makes it possible to allocate/deallocate objects without any locking
+ * when the cache is in the steady state.
+ */
+#undef JEMALLOC_TCACHE
+
+/*
+ * JEMALLOC_DSS enables use of sbrk(2) to allocate chunks from the data storage
+ * segment (DSS).
+ */
+#undef JEMALLOC_DSS
+
+/* Support memory filling (junk/zero/quarantine/redzone). */
+#undef JEMALLOC_FILL
+
+/* Support the experimental API. */
+#undef JEMALLOC_EXPERIMENTAL
+
+/* Support utrace(2)-based tracing. */
+#undef JEMALLOC_UTRACE
+
+/* Support Valgrind. */
+#undef JEMALLOC_VALGRIND
+
+/* Support optional abort() on OOM. */
+#undef JEMALLOC_XMALLOC
+
+/* Support lazy locking (avoid locking unless a second thread is launched). */
+#undef JEMALLOC_LAZY_LOCK
+
+/* One page is 2^STATIC_PAGE_SHIFT bytes. */
+#undef STATIC_PAGE_SHIFT
+
+/*
+ * If defined, use munmap() to unmap freed chunks, rather than storing them for
+ * later reuse.  This is disabled by default on Linux because common sequences
+ * of mmap()/munmap() calls will cause virtual memory map holes.
+ */
+#undef JEMALLOC_MUNMAP
+
+/*
+ * If defined, use mremap(...MREMAP_FIXED...) for huge realloc().  This is
+ * disabled by default because it is Linux-specific and it will cause virtual
+ * memory map holes, much like munmap(2) does.
+ */
+#undef JEMALLOC_MREMAP
+
+/* TLS is used to map arenas and magazine caches to threads. */
+#undef JEMALLOC_TLS
+
+/*
+ * JEMALLOC_IVSALLOC enables ivsalloc(), which verifies that pointers reside
+ * within jemalloc-owned chunks before dereferencing them.
+ */
+#undef JEMALLOC_IVSALLOC
+
+/*
+ * Define overrides for non-standard allocator-related functions if they
+ * are present on the system.
+ */
+#undef JEMALLOC_OVERRIDE_MEMALIGN
+#undef JEMALLOC_OVERRIDE_VALLOC
+
+/*
+ * At least Linux omits the "const" in:
+ *
+ *   size_t malloc_usable_size(const void *ptr);
+ *
+ * Match the operating system's prototype.
+ */
+#undef JEMALLOC_USABLE_SIZE_CONST
+
+/*
+ * Darwin (OS X) uses zones to work around Mach-O symbol override shortcomings.
+ */
+#undef JEMALLOC_ZONE
+#undef JEMALLOC_ZONE_VERSION
+
+/*
+ * Methods for purging unused pages differ between operating systems.
+ *
+ *   madvise(..., MADV_DONTNEED) : On Linux, this immediately discards pages,
+ *                                 such that new pages will be demand-zeroed if
+ *                                 the address region is later touched.
+ *   madvise(..., MADV_FREE) : On FreeBSD and Darwin, this marks pages as being
+ *                             unused, such that they will be discarded rather
+ *                             than swapped out.
+ */
+#undef JEMALLOC_PURGE_MADVISE_DONTNEED
+#undef JEMALLOC_PURGE_MADVISE_FREE
+
+/*
+ * Define if operating system has alloca.h header.
+ */
+#undef JEMALLOC_HAS_ALLOCA_H
+
+/* sizeof(void *) == 2^LG_SIZEOF_PTR. */
+#undef LG_SIZEOF_PTR
+
+/* sizeof(int) == 2^LG_SIZEOF_INT. */
+#undef LG_SIZEOF_INT
+
+/* sizeof(long) == 2^LG_SIZEOF_LONG. */
+#undef LG_SIZEOF_LONG
+
+/* sizeof(intmax_t) == 2^LG_SIZEOF_INTMAX_T. */
+#undef LG_SIZEOF_INTMAX_T
diff --git a/extra/jemalloc/include/msvc_compat/inttypes.h b/extra/jemalloc/include/msvc_compat/inttypes.h
new file mode 100644
index 00000000000..a4e6b75cb91
--- /dev/null
+++ b/extra/jemalloc/include/msvc_compat/inttypes.h
@@ -0,0 +1,313 @@
+// ISO C9x  compliant inttypes.h for Microsoft Visual Studio
+// Based on ISO/IEC 9899:TC2 Committee draft (May 6, 2005) WG14/N1124 
+// 
+//  Copyright (c) 2006 Alexander Chemeris
+// 
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are met:
+// 
+//   1. Redistributions of source code must retain the above copyright notice,
+//      this list of conditions and the following disclaimer.
+// 
+//   2. Redistributions in binary form must reproduce the above copyright
+//      notice, this list of conditions and the following disclaimer in the
+//      documentation and/or other materials provided with the distribution.
+// 
+//   3. The name of the author may be used to endorse or promote products
+//      derived from this software without specific prior written permission.
+// 
+// THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED
+// WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+// MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
+// EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+// OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, 
+// WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+// OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+// ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+// 
+///////////////////////////////////////////////////////////////////////////////
+
+#ifndef _MSC_VER // [
+#error "Use this header only with Microsoft Visual C++ compilers!"
+#endif // _MSC_VER ]
+
+#ifndef _MSC_INTTYPES_H_ // [
+#define _MSC_INTTYPES_H_
+
+#if _MSC_VER > 1000
+#pragma once
+#endif
+
+#include "stdint.h"
+
+// 7.8 Format conversion of integer types
+
+typedef struct {
+   intmax_t quot;
+   intmax_t rem;
+} imaxdiv_t;
+
+// 7.8.1 Macros for format specifiers
+
+#if !defined(__cplusplus) || defined(__STDC_FORMAT_MACROS) // [   See footnote 185 at page 198
+
+#ifdef _WIN64
+#  define __PRI64_PREFIX        "l"
+#  define __PRIPTR_PREFIX       "l"
+#else
+#  define __PRI64_PREFIX        "ll"
+#  define __PRIPTR_PREFIX
+#endif
+
+// The fprintf macros for signed integers are:
+#define PRId8       "d"
+#define PRIi8       "i"
+#define PRIdLEAST8  "d"
+#define PRIiLEAST8  "i"
+#define PRIdFAST8   "d"
+#define PRIiFAST8   "i"
+
+#define PRId16       "hd"
+#define PRIi16       "hi"
+#define PRIdLEAST16  "hd"
+#define PRIiLEAST16  "hi"
+#define PRIdFAST16   "hd"
+#define PRIiFAST16   "hi"
+
+#define PRId32       "d"
+#define PRIi32       "i"
+#define PRIdLEAST32  "d"
+#define PRIiLEAST32  "i"
+#define PRIdFAST32   "d"
+#define PRIiFAST32   "i"
+
+#define PRId64       __PRI64_PREFIX "d"
+#define PRIi64       __PRI64_PREFIX "i"
+#define PRIdLEAST64  __PRI64_PREFIX "d"
+#define PRIiLEAST64  __PRI64_PREFIX "i"
+#define PRIdFAST64   __PRI64_PREFIX "d"
+#define PRIiFAST64   __PRI64_PREFIX "i"
+
+#define PRIdMAX     __PRI64_PREFIX "d"
+#define PRIiMAX     __PRI64_PREFIX "i"
+
+#define PRIdPTR     __PRIPTR_PREFIX "d"
+#define PRIiPTR     __PRIPTR_PREFIX "i"
+
+// The fprintf macros for unsigned integers are:
+#define PRIo8       "o"
+#define PRIu8       "u"
+#define PRIx8       "x"
+#define PRIX8       "X"
+#define PRIoLEAST8  "o"
+#define PRIuLEAST8  "u"
+#define PRIxLEAST8  "x"
+#define PRIXLEAST8  "X"
+#define PRIoFAST8   "o"
+#define PRIuFAST8   "u"
+#define PRIxFAST8   "x"
+#define PRIXFAST8   "X"
+
+#define PRIo16       "ho"
+#define PRIu16       "hu"
+#define PRIx16       "hx"
+#define PRIX16       "hX"
+#define PRIoLEAST16  "ho"
+#define PRIuLEAST16  "hu"
+#define PRIxLEAST16  "hx"
+#define PRIXLEAST16  "hX"
+#define PRIoFAST16   "ho"
+#define PRIuFAST16   "hu"
+#define PRIxFAST16   "hx"
+#define PRIXFAST16   "hX"
+
+#define PRIo32       "o"
+#define PRIu32       "u"
+#define PRIx32       "x"
+#define PRIX32       "X"
+#define PRIoLEAST32  "o"
+#define PRIuLEAST32  "u"
+#define PRIxLEAST32  "x"
+#define PRIXLEAST32  "X"
+#define PRIoFAST32   "o"
+#define PRIuFAST32   "u"
+#define PRIxFAST32   "x"
+#define PRIXFAST32   "X"
+
+#define PRIo64       __PRI64_PREFIX "o"
+#define PRIu64       __PRI64_PREFIX "u"
+#define PRIx64       __PRI64_PREFIX "x"
+#define PRIX64       __PRI64_PREFIX "X"
+#define PRIoLEAST64  __PRI64_PREFIX "o"
+#define PRIuLEAST64  __PRI64_PREFIX "u"
+#define PRIxLEAST64  __PRI64_PREFIX "x"
+#define PRIXLEAST64  __PRI64_PREFIX "X"
+#define PRIoFAST64   __PRI64_PREFIX "o"
+#define PRIuFAST64   __PRI64_PREFIX "u"
+#define PRIxFAST64   __PRI64_PREFIX "x"
+#define PRIXFAST64   __PRI64_PREFIX "X"
+
+#define PRIoMAX     __PRI64_PREFIX "o"
+#define PRIuMAX     __PRI64_PREFIX "u"
+#define PRIxMAX     __PRI64_PREFIX "x"
+#define PRIXMAX     __PRI64_PREFIX "X"
+
+#define PRIoPTR     __PRIPTR_PREFIX "o"
+#define PRIuPTR     __PRIPTR_PREFIX "u"
+#define PRIxPTR     __PRIPTR_PREFIX "x"
+#define PRIXPTR     __PRIPTR_PREFIX "X"
+
+// The fscanf macros for signed integers are:
+#define SCNd8       "d"
+#define SCNi8       "i"
+#define SCNdLEAST8  "d"
+#define SCNiLEAST8  "i"
+#define SCNdFAST8   "d"
+#define SCNiFAST8   "i"
+
+#define SCNd16       "hd"
+#define SCNi16       "hi"
+#define SCNdLEAST16  "hd"
+#define SCNiLEAST16  "hi"
+#define SCNdFAST16   "hd"
+#define SCNiFAST16   "hi"
+
+#define SCNd32       "ld"
+#define SCNi32       "li"
+#define SCNdLEAST32  "ld"
+#define SCNiLEAST32  "li"
+#define SCNdFAST32   "ld"
+#define SCNiFAST32   "li"
+
+#define SCNd64       "I64d"
+#define SCNi64       "I64i"
+#define SCNdLEAST64  "I64d"
+#define SCNiLEAST64  "I64i"
+#define SCNdFAST64   "I64d"
+#define SCNiFAST64   "I64i"
+
+#define SCNdMAX     "I64d"
+#define SCNiMAX     "I64i"
+
+#ifdef _WIN64 // [
+#  define SCNdPTR     "I64d"
+#  define SCNiPTR     "I64i"
+#else  // _WIN64 ][
+#  define SCNdPTR     "ld"
+#  define SCNiPTR     "li"
+#endif  // _WIN64 ]
+
+// The fscanf macros for unsigned integers are:
+#define SCNo8       "o"
+#define SCNu8       "u"
+#define SCNx8       "x"
+#define SCNX8       "X"
+#define SCNoLEAST8  "o"
+#define SCNuLEAST8  "u"
+#define SCNxLEAST8  "x"
+#define SCNXLEAST8  "X"
+#define SCNoFAST8   "o"
+#define SCNuFAST8   "u"
+#define SCNxFAST8   "x"
+#define SCNXFAST8   "X"
+
+#define SCNo16       "ho"
+#define SCNu16       "hu"
+#define SCNx16       "hx"
+#define SCNX16       "hX"
+#define SCNoLEAST16  "ho"
+#define SCNuLEAST16  "hu"
+#define SCNxLEAST16  "hx"
+#define SCNXLEAST16  "hX"
+#define SCNoFAST16   "ho"
+#define SCNuFAST16   "hu"
+#define SCNxFAST16   "hx"
+#define SCNXFAST16   "hX"
+
+#define SCNo32       "lo"
+#define SCNu32       "lu"
+#define SCNx32       "lx"
+#define SCNX32       "lX"
+#define SCNoLEAST32  "lo"
+#define SCNuLEAST32  "lu"
+#define SCNxLEAST32  "lx"
+#define SCNXLEAST32  "lX"
+#define SCNoFAST32   "lo"
+#define SCNuFAST32   "lu"
+#define SCNxFAST32   "lx"
+#define SCNXFAST32   "lX"
+
+#define SCNo64       "I64o"
+#define SCNu64       "I64u"
+#define SCNx64       "I64x"
+#define SCNX64       "I64X"
+#define SCNoLEAST64  "I64o"
+#define SCNuLEAST64  "I64u"
+#define SCNxLEAST64  "I64x"
+#define SCNXLEAST64  "I64X"
+#define SCNoFAST64   "I64o"
+#define SCNuFAST64   "I64u"
+#define SCNxFAST64   "I64x"
+#define SCNXFAST64   "I64X"
+
+#define SCNoMAX     "I64o"
+#define SCNuMAX     "I64u"
+#define SCNxMAX     "I64x"
+#define SCNXMAX     "I64X"
+
+#ifdef _WIN64 // [
+#  define SCNoPTR     "I64o"
+#  define SCNuPTR     "I64u"
+#  define SCNxPTR     "I64x"
+#  define SCNXPTR     "I64X"
+#else  // _WIN64 ][
+#  define SCNoPTR     "lo"
+#  define SCNuPTR     "lu"
+#  define SCNxPTR     "lx"
+#  define SCNXPTR     "lX"
+#endif  // _WIN64 ]
+
+#endif // __STDC_FORMAT_MACROS ]
+
+// 7.8.2 Functions for greatest-width integer types
+
+// 7.8.2.1 The imaxabs function
+#define imaxabs _abs64
+
+// 7.8.2.2 The imaxdiv function
+
+// This is modified version of div() function from Microsoft's div.c found
+// in %MSVC.NET%\crt\src\div.c
+#ifdef STATIC_IMAXDIV // [
+static
+#else // STATIC_IMAXDIV ][
+_inline
+#endif // STATIC_IMAXDIV ]
+imaxdiv_t __cdecl imaxdiv(intmax_t numer, intmax_t denom)
+{
+   imaxdiv_t result;
+
+   result.quot = numer / denom;
+   result.rem = numer % denom;
+
+   if (numer < 0 && result.rem > 0) {
+      // did division wrong; must fix up
+      ++result.quot;
+      result.rem -= denom;
+   }
+
+   return result;
+}
+
+// 7.8.2.3 The strtoimax and strtoumax functions
+#define strtoimax _strtoi64
+#define strtoumax _strtoui64
+
+// 7.8.2.4 The wcstoimax and wcstoumax functions
+#define wcstoimax _wcstoi64
+#define wcstoumax _wcstoui64
+
+
+#endif // _MSC_INTTYPES_H_ ]
diff --git a/extra/jemalloc/include/msvc_compat/stdbool.h b/extra/jemalloc/include/msvc_compat/stdbool.h
new file mode 100644
index 00000000000..da9ee8b809b
--- /dev/null
+++ b/extra/jemalloc/include/msvc_compat/stdbool.h
@@ -0,0 +1,16 @@
+#ifndef stdbool_h
+#define stdbool_h
+
+#include <wtypes.h>
+
+/* MSVC doesn't define _Bool or bool in C, but does have BOOL */
+/* Note this doesn't pass autoconf's test because (bool) 0.5 != true */
+typedef BOOL _Bool;
+
+#define bool _Bool
+#define true 1
+#define false 0
+
+#define __bool_true_false_are_defined 1
+
+#endif /* stdbool_h */
diff --git a/extra/jemalloc/include/msvc_compat/stdint.h b/extra/jemalloc/include/msvc_compat/stdint.h
new file mode 100644
index 00000000000..d02608a5972
--- /dev/null
+++ b/extra/jemalloc/include/msvc_compat/stdint.h
@@ -0,0 +1,247 @@
+// ISO C9x  compliant stdint.h for Microsoft Visual Studio
+// Based on ISO/IEC 9899:TC2 Committee draft (May 6, 2005) WG14/N1124 
+// 
+//  Copyright (c) 2006-2008 Alexander Chemeris
+// 
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are met:
+// 
+//   1. Redistributions of source code must retain the above copyright notice,
+//      this list of conditions and the following disclaimer.
+// 
+//   2. Redistributions in binary form must reproduce the above copyright
+//      notice, this list of conditions and the following disclaimer in the
+//      documentation and/or other materials provided with the distribution.
+// 
+//   3. The name of the author may be used to endorse or promote products
+//      derived from this software without specific prior written permission.
+// 
+// THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED
+// WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+// MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
+// EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+// OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, 
+// WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+// OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+// ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+// 
+///////////////////////////////////////////////////////////////////////////////
+
+#ifndef _MSC_VER // [
+#error "Use this header only with Microsoft Visual C++ compilers!"
+#endif // _MSC_VER ]
+
+#ifndef _MSC_STDINT_H_ // [
+#define _MSC_STDINT_H_
+
+#if _MSC_VER > 1000
+#pragma once
+#endif
+
+#include <limits.h>
+
+// For Visual Studio 6 in C++ mode and for many Visual Studio versions when
+// compiling for ARM we should wrap <wchar.h> include with 'extern "C++" {}'
+// or compiler give many errors like this:
+//   error C2733: second C linkage of overloaded function 'wmemchr' not allowed
+#ifdef __cplusplus
+extern "C" {
+#endif
+#  include <wchar.h>
+#ifdef __cplusplus
+}
+#endif
+
+// Define _W64 macros to mark types changing their size, like intptr_t.
+#ifndef _W64
+#  if !defined(__midl) && (defined(_X86_) || defined(_M_IX86)) && _MSC_VER >= 1300
+#     define _W64 __w64
+#  else
+#     define _W64
+#  endif
+#endif
+
+
+// 7.18.1 Integer types
+
+// 7.18.1.1 Exact-width integer types
+
+// Visual Studio 6 and Embedded Visual C++ 4 doesn't
+// realize that, e.g. char has the same size as __int8
+// so we give up on __intX for them.
+#if (_MSC_VER < 1300)
+   typedef signed char       int8_t;
+   typedef signed short      int16_t;
+   typedef signed int        int32_t;
+   typedef unsigned char     uint8_t;
+   typedef unsigned short    uint16_t;
+   typedef unsigned int      uint32_t;
+#else
+   typedef signed __int8     int8_t;
+   typedef signed __int16    int16_t;
+   typedef signed __int32    int32_t;
+   typedef unsigned __int8   uint8_t;
+   typedef unsigned __int16  uint16_t;
+   typedef unsigned __int32  uint32_t;
+#endif
+typedef signed __int64       int64_t;
+typedef unsigned __int64     uint64_t;
+
+
+// 7.18.1.2 Minimum-width integer types
+typedef int8_t    int_least8_t;
+typedef int16_t   int_least16_t;
+typedef int32_t   int_least32_t;
+typedef int64_t   int_least64_t;
+typedef uint8_t   uint_least8_t;
+typedef uint16_t  uint_least16_t;
+typedef uint32_t  uint_least32_t;
+typedef uint64_t  uint_least64_t;
+
+// 7.18.1.3 Fastest minimum-width integer types
+typedef int8_t    int_fast8_t;
+typedef int16_t   int_fast16_t;
+typedef int32_t   int_fast32_t;
+typedef int64_t   int_fast64_t;
+typedef uint8_t   uint_fast8_t;
+typedef uint16_t  uint_fast16_t;
+typedef uint32_t  uint_fast32_t;
+typedef uint64_t  uint_fast64_t;
+
+// 7.18.1.4 Integer types capable of holding object pointers
+#ifdef _WIN64 // [
+   typedef signed __int64    intptr_t;
+   typedef unsigned __int64  uintptr_t;
+#else // _WIN64 ][
+   typedef _W64 signed int   intptr_t;
+   typedef _W64 unsigned int uintptr_t;
+#endif // _WIN64 ]
+
+// 7.18.1.5 Greatest-width integer types
+typedef int64_t   intmax_t;
+typedef uint64_t  uintmax_t;
+
+
+// 7.18.2 Limits of specified-width integer types
+
+#if !defined(__cplusplus) || defined(__STDC_LIMIT_MACROS) // [   See footnote 220 at page 257 and footnote 221 at page 259
+
+// 7.18.2.1 Limits of exact-width integer types
+#define INT8_MIN     ((int8_t)_I8_MIN)
+#define INT8_MAX     _I8_MAX
+#define INT16_MIN    ((int16_t)_I16_MIN)
+#define INT16_MAX    _I16_MAX
+#define INT32_MIN    ((int32_t)_I32_MIN)
+#define INT32_MAX    _I32_MAX
+#define INT64_MIN    ((int64_t)_I64_MIN)
+#define INT64_MAX    _I64_MAX
+#define UINT8_MAX    _UI8_MAX
+#define UINT16_MAX   _UI16_MAX
+#define UINT32_MAX   _UI32_MAX
+#define UINT64_MAX   _UI64_MAX
+
+// 7.18.2.2 Limits of minimum-width integer types
+#define INT_LEAST8_MIN    INT8_MIN
+#define INT_LEAST8_MAX    INT8_MAX
+#define INT_LEAST16_MIN   INT16_MIN
+#define INT_LEAST16_MAX   INT16_MAX
+#define INT_LEAST32_MIN   INT32_MIN
+#define INT_LEAST32_MAX   INT32_MAX
+#define INT_LEAST64_MIN   INT64_MIN
+#define INT_LEAST64_MAX   INT64_MAX
+#define UINT_LEAST8_MAX   UINT8_MAX
+#define UINT_LEAST16_MAX  UINT16_MAX
+#define UINT_LEAST32_MAX  UINT32_MAX
+#define UINT_LEAST64_MAX  UINT64_MAX
+
+// 7.18.2.3 Limits of fastest minimum-width integer types
+#define INT_FAST8_MIN    INT8_MIN
+#define INT_FAST8_MAX    INT8_MAX
+#define INT_FAST16_MIN   INT16_MIN
+#define INT_FAST16_MAX   INT16_MAX
+#define INT_FAST32_MIN   INT32_MIN
+#define INT_FAST32_MAX   INT32_MAX
+#define INT_FAST64_MIN   INT64_MIN
+#define INT_FAST64_MAX   INT64_MAX
+#define UINT_FAST8_MAX   UINT8_MAX
+#define UINT_FAST16_MAX  UINT16_MAX
+#define UINT_FAST32_MAX  UINT32_MAX
+#define UINT_FAST64_MAX  UINT64_MAX
+
+// 7.18.2.4 Limits of integer types capable of holding object pointers
+#ifdef _WIN64 // [
+#  define INTPTR_MIN   INT64_MIN
+#  define INTPTR_MAX   INT64_MAX
+#  define UINTPTR_MAX  UINT64_MAX
+#else // _WIN64 ][
+#  define INTPTR_MIN   INT32_MIN
+#  define INTPTR_MAX   INT32_MAX
+#  define UINTPTR_MAX  UINT32_MAX
+#endif // _WIN64 ]
+
+// 7.18.2.5 Limits of greatest-width integer types
+#define INTMAX_MIN   INT64_MIN
+#define INTMAX_MAX   INT64_MAX
+#define UINTMAX_MAX  UINT64_MAX
+
+// 7.18.3 Limits of other integer types
+
+#ifdef _WIN64 // [
+#  define PTRDIFF_MIN  _I64_MIN
+#  define PTRDIFF_MAX  _I64_MAX
+#else  // _WIN64 ][
+#  define PTRDIFF_MIN  _I32_MIN
+#  define PTRDIFF_MAX  _I32_MAX
+#endif  // _WIN64 ]
+
+#define SIG_ATOMIC_MIN  INT_MIN
+#define SIG_ATOMIC_MAX  INT_MAX
+
+#ifndef SIZE_MAX // [
+#  ifdef _WIN64 // [
+#     define SIZE_MAX  _UI64_MAX
+#  else // _WIN64 ][
+#     define SIZE_MAX  _UI32_MAX
+#  endif // _WIN64 ]
+#endif // SIZE_MAX ]
+
+// WCHAR_MIN and WCHAR_MAX are also defined in <wchar.h>
+#ifndef WCHAR_MIN // [
+#  define WCHAR_MIN  0
+#endif  // WCHAR_MIN ]
+#ifndef WCHAR_MAX // [
+#  define WCHAR_MAX  _UI16_MAX
+#endif  // WCHAR_MAX ]
+
+#define WINT_MIN  0
+#define WINT_MAX  _UI16_MAX
+
+#endif // __STDC_LIMIT_MACROS ]
+
+
+// 7.18.4 Limits of other integer types
+
+#if !defined(__cplusplus) || defined(__STDC_CONSTANT_MACROS) // [   See footnote 224 at page 260
+
+// 7.18.4.1 Macros for minimum-width integer constants
+
+#define INT8_C(val)  val##i8
+#define INT16_C(val) val##i16
+#define INT32_C(val) val##i32
+#define INT64_C(val) val##i64
+
+#define UINT8_C(val)  val##ui8
+#define UINT16_C(val) val##ui16
+#define UINT32_C(val) val##ui32
+#define UINT64_C(val) val##ui64
+
+// 7.18.4.2 Macros for greatest-width integer constants
+#define INTMAX_C   INT64_C
+#define UINTMAX_C  UINT64_C
+
+#endif // __STDC_CONSTANT_MACROS ]
+
+
+#endif // _MSC_STDINT_H_ ]
diff --git a/extra/jemalloc/include/msvc_compat/strings.h b/extra/jemalloc/include/msvc_compat/strings.h
new file mode 100644
index 00000000000..c84975b6b8e
--- /dev/null
+++ b/extra/jemalloc/include/msvc_compat/strings.h
@@ -0,0 +1,23 @@
+#ifndef strings_h
+#define strings_h
+
+/* MSVC doesn't define ffs/ffsl. This dummy strings.h header is provided
+ * for both */
+#include <intrin.h>
+#pragma intrinsic(_BitScanForward)
+static __forceinline int ffsl(long x)
+{
+	unsigned long i;
+
+	if (_BitScanForward(&i, x))
+		return (i + 1);
+	return (0);
+}
+
+static __forceinline int ffs(int x)
+{
+
+	return (ffsl(x));
+}
+
+#endif
diff --git a/extra/jemalloc/install-sh b/extra/jemalloc/install-sh
new file mode 100755
index 00000000000..ebc66913e94
--- /dev/null
+++ b/extra/jemalloc/install-sh
@@ -0,0 +1,250 @@
+#! /bin/sh
+#
+# install - install a program, script, or datafile
+# This comes from X11R5 (mit/util/scripts/install.sh).
+#
+# Copyright 1991 by the Massachusetts Institute of Technology
+#
+# Permission to use, copy, modify, distribute, and sell this software and its
+# documentation for any purpose is hereby granted without fee, provided that
+# the above copyright notice appear in all copies and that both that
+# copyright notice and this permission notice appear in supporting
+# documentation, and that the name of M.I.T. not be used in advertising or
+# publicity pertaining to distribution of the software without specific,
+# written prior permission.  M.I.T. makes no representations about the
+# suitability of this software for any purpose.  It is provided "as is"
+# without express or implied warranty.
+#
+# Calling this script install-sh is preferred over install.sh, to prevent
+# `make' implicit rules from creating a file called install from it
+# when there is no Makefile.
+#
+# This script is compatible with the BSD install script, but was written
+# from scratch.  It can only install one file at a time, a restriction
+# shared with many OS's install programs.
+
+
+# set DOITPROG to echo to test this script
+
+# Don't use :- since 4.3BSD and earlier shells don't like it.
+doit="${DOITPROG-}"
+
+
+# put in absolute paths if you don't have them in your path; or use env. vars.
+
+mvprog="${MVPROG-mv}"
+cpprog="${CPPROG-cp}"
+chmodprog="${CHMODPROG-chmod}"
+chownprog="${CHOWNPROG-chown}"
+chgrpprog="${CHGRPPROG-chgrp}"
+stripprog="${STRIPPROG-strip}"
+rmprog="${RMPROG-rm}"
+mkdirprog="${MKDIRPROG-mkdir}"
+
+transformbasename=""
+transform_arg=""
+instcmd="$mvprog"
+chmodcmd="$chmodprog 0755"
+chowncmd=""
+chgrpcmd=""
+stripcmd=""
+rmcmd="$rmprog -f"
+mvcmd="$mvprog"
+src=""
+dst=""
+dir_arg=""
+
+while [ x"$1" != x ]; do
+    case $1 in
+	-c) instcmd="$cpprog"
+	    shift
+	    continue;;
+
+	-d) dir_arg=true
+	    shift
+	    continue;;
+
+	-m) chmodcmd="$chmodprog $2"
+	    shift
+	    shift
+	    continue;;
+
+	-o) chowncmd="$chownprog $2"
+	    shift
+	    shift
+	    continue;;
+
+	-g) chgrpcmd="$chgrpprog $2"
+	    shift
+	    shift
+	    continue;;
+
+	-s) stripcmd="$stripprog"
+	    shift
+	    continue;;
+
+	-t=*) transformarg=`echo $1 | sed 's/-t=//'`
+	    shift
+	    continue;;
+
+	-b=*) transformbasename=`echo $1 | sed 's/-b=//'`
+	    shift
+	    continue;;
+
+	*)  if [ x"$src" = x ]
+	    then
+		src=$1
+	    else
+		# this colon is to work around a 386BSD /bin/sh bug
+		:
+		dst=$1
+	    fi
+	    shift
+	    continue;;
+    esac
+done
+
+if [ x"$src" = x ]
+then
+	echo "install:	no input file specified"
+	exit 1
+else
+	true
+fi
+
+if [ x"$dir_arg" != x ]; then
+	dst=$src
+	src=""
+	
+	if [ -d $dst ]; then
+		instcmd=:
+	else
+		instcmd=mkdir
+	fi
+else
+
+# Waiting for this to be detected by the "$instcmd $src $dsttmp" command
+# might cause directories to be created, which would be especially bad 
+# if $src (and thus $dsttmp) contains '*'.
+
+	if [ -f $src -o -d $src ]
+	then
+		true
+	else
+		echo "install:  $src does not exist"
+		exit 1
+	fi
+	
+	if [ x"$dst" = x ]
+	then
+		echo "install:	no destination specified"
+		exit 1
+	else
+		true
+	fi
+
+# If destination is a directory, append the input filename; if your system
+# does not like double slashes in filenames, you may need to add some logic
+
+	if [ -d $dst ]
+	then
+		dst="$dst"/`basename $src`
+	else
+		true
+	fi
+fi
+
+## this sed command emulates the dirname command
+dstdir=`echo $dst | sed -e 's,[^/]*$,,;s,/$,,;s,^$,.,'`
+
+# Make sure that the destination directory exists.
+#  this part is taken from Noah Friedman's mkinstalldirs script
+
+# Skip lots of stat calls in the usual case.
+if [ ! -d "$dstdir" ]; then
+defaultIFS='	
+'
+IFS="${IFS-${defaultIFS}}"
+
+oIFS="${IFS}"
+# Some sh's can't handle IFS=/ for some reason.
+IFS='%'
+set - `echo ${dstdir} | sed -e 's@/@%@g' -e 's@^%@/@'`
+IFS="${oIFS}"
+
+pathcomp=''
+
+while [ $# -ne 0 ] ; do
+	pathcomp="${pathcomp}${1}"
+	shift
+
+	if [ ! -d "${pathcomp}" ] ;
+        then
+		$mkdirprog "${pathcomp}"
+	else
+		true
+	fi
+
+	pathcomp="${pathcomp}/"
+done
+fi
+
+if [ x"$dir_arg" != x ]
+then
+	$doit $instcmd $dst &&
+
+	if [ x"$chowncmd" != x ]; then $doit $chowncmd $dst; else true ; fi &&
+	if [ x"$chgrpcmd" != x ]; then $doit $chgrpcmd $dst; else true ; fi &&
+	if [ x"$stripcmd" != x ]; then $doit $stripcmd $dst; else true ; fi &&
+	if [ x"$chmodcmd" != x ]; then $doit $chmodcmd $dst; else true ; fi
+else
+
+# If we're going to rename the final executable, determine the name now.
+
+	if [ x"$transformarg" = x ] 
+	then
+		dstfile=`basename $dst`
+	else
+		dstfile=`basename $dst $transformbasename | 
+			sed $transformarg`$transformbasename
+	fi
+
+# don't allow the sed command to completely eliminate the filename
+
+	if [ x"$dstfile" = x ] 
+	then
+		dstfile=`basename $dst`
+	else
+		true
+	fi
+
+# Make a temp file name in the proper directory.
+
+	dsttmp=$dstdir/#inst.$$#
+
+# Move or copy the file name to the temp name
+
+	$doit $instcmd $src $dsttmp &&
+
+	trap "rm -f ${dsttmp}" 0 &&
+
+# and set any options; do chmod last to preserve setuid bits
+
+# If any of these fail, we abort the whole thing.  If we want to
+# ignore errors from any of these, just make sure not to ignore
+# errors from the above "$doit $instcmd $src $dsttmp" command.
+
+	if [ x"$chowncmd" != x ]; then $doit $chowncmd $dsttmp; else true;fi &&
+	if [ x"$chgrpcmd" != x ]; then $doit $chgrpcmd $dsttmp; else true;fi &&
+	if [ x"$stripcmd" != x ]; then $doit $stripcmd $dsttmp; else true;fi &&
+	if [ x"$chmodcmd" != x ]; then $doit $chmodcmd $dsttmp; else true;fi &&
+
+# Now rename the file to the real destination.
+
+	$doit $rmcmd -f $dstdir/$dstfile &&
+	$doit $mvcmd $dsttmp $dstdir/$dstfile 
+
+fi &&
+
+
+exit 0
diff --git a/extra/jemalloc/src/arena.c b/extra/jemalloc/src/arena.c
new file mode 100644
index 00000000000..05a787f89d9
--- /dev/null
+++ b/extra/jemalloc/src/arena.c
@@ -0,0 +1,2365 @@
+#define	JEMALLOC_ARENA_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Data. */
+
+ssize_t		opt_lg_dirty_mult = LG_DIRTY_MULT_DEFAULT;
+arena_bin_info_t	arena_bin_info[NBINS];
+
+JEMALLOC_ALIGNED(CACHELINE)
+const uint8_t	small_size2bin[] = {
+#define	S2B_8(i)	i,
+#define	S2B_16(i)	S2B_8(i) S2B_8(i)
+#define	S2B_32(i)	S2B_16(i) S2B_16(i)
+#define	S2B_64(i)	S2B_32(i) S2B_32(i)
+#define	S2B_128(i)	S2B_64(i) S2B_64(i)
+#define	S2B_256(i)	S2B_128(i) S2B_128(i)
+#define	S2B_512(i)	S2B_256(i) S2B_256(i)
+#define	S2B_1024(i)	S2B_512(i) S2B_512(i)
+#define	S2B_2048(i)	S2B_1024(i) S2B_1024(i)
+#define	S2B_4096(i)	S2B_2048(i) S2B_2048(i)
+#define	S2B_8192(i)	S2B_4096(i) S2B_4096(i)
+#define	SIZE_CLASS(bin, delta, size)					\
+	S2B_##delta(bin)
+	SIZE_CLASSES
+#undef S2B_8
+#undef S2B_16
+#undef S2B_32
+#undef S2B_64
+#undef S2B_128
+#undef S2B_256
+#undef S2B_512
+#undef S2B_1024
+#undef S2B_2048
+#undef S2B_4096
+#undef S2B_8192
+#undef SIZE_CLASS
+};
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static void	arena_avail_insert(arena_t *arena, arena_chunk_t *chunk,
+    size_t pageind, size_t npages, bool maybe_adjac_pred,
+    bool maybe_adjac_succ);
+static void	arena_avail_remove(arena_t *arena, arena_chunk_t *chunk,
+    size_t pageind, size_t npages, bool maybe_adjac_pred,
+    bool maybe_adjac_succ);
+static void	arena_run_split(arena_t *arena, arena_run_t *run, size_t size,
+    bool large, size_t binind, bool zero);
+static arena_chunk_t *arena_chunk_alloc(arena_t *arena);
+static void	arena_chunk_dealloc(arena_t *arena, arena_chunk_t *chunk);
+static arena_run_t	*arena_run_alloc_helper(arena_t *arena, size_t size,
+    bool large, size_t binind, bool zero);
+static arena_run_t *arena_run_alloc(arena_t *arena, size_t size, bool large,
+    size_t binind, bool zero);
+static arena_chunk_t	*chunks_dirty_iter_cb(arena_chunk_tree_t *tree,
+    arena_chunk_t *chunk, void *arg);
+static void	arena_purge(arena_t *arena, bool all);
+static void	arena_run_dalloc(arena_t *arena, arena_run_t *run, bool dirty,
+    bool cleaned);
+static void	arena_run_trim_head(arena_t *arena, arena_chunk_t *chunk,
+    arena_run_t *run, size_t oldsize, size_t newsize);
+static void	arena_run_trim_tail(arena_t *arena, arena_chunk_t *chunk,
+    arena_run_t *run, size_t oldsize, size_t newsize, bool dirty);
+static arena_run_t	*arena_bin_runs_first(arena_bin_t *bin);
+static void	arena_bin_runs_insert(arena_bin_t *bin, arena_run_t *run);
+static void	arena_bin_runs_remove(arena_bin_t *bin, arena_run_t *run);
+static arena_run_t *arena_bin_nonfull_run_tryget(arena_bin_t *bin);
+static arena_run_t *arena_bin_nonfull_run_get(arena_t *arena, arena_bin_t *bin);
+static void	*arena_bin_malloc_hard(arena_t *arena, arena_bin_t *bin);
+static void	arena_dissociate_bin_run(arena_chunk_t *chunk, arena_run_t *run,
+    arena_bin_t *bin);
+static void	arena_dalloc_bin_run(arena_t *arena, arena_chunk_t *chunk,
+    arena_run_t *run, arena_bin_t *bin);
+static void	arena_bin_lower_run(arena_t *arena, arena_chunk_t *chunk,
+    arena_run_t *run, arena_bin_t *bin);
+static void	arena_ralloc_large_shrink(arena_t *arena, arena_chunk_t *chunk,
+    void *ptr, size_t oldsize, size_t size);
+static bool	arena_ralloc_large_grow(arena_t *arena, arena_chunk_t *chunk,
+    void *ptr, size_t oldsize, size_t size, size_t extra, bool zero);
+static bool	arena_ralloc_large(void *ptr, size_t oldsize, size_t size,
+    size_t extra, bool zero);
+static size_t	bin_info_run_size_calc(arena_bin_info_t *bin_info,
+    size_t min_run_size);
+static void	bin_info_init(void);
+
+/******************************************************************************/
+
+static inline int
+arena_run_comp(arena_chunk_map_t *a, arena_chunk_map_t *b)
+{
+	uintptr_t a_mapelm = (uintptr_t)a;
+	uintptr_t b_mapelm = (uintptr_t)b;
+
+	assert(a != NULL);
+	assert(b != NULL);
+
+	return ((a_mapelm > b_mapelm) - (a_mapelm < b_mapelm));
+}
+
+/* Generate red-black tree functions. */
+rb_gen(static UNUSED, arena_run_tree_, arena_run_tree_t, arena_chunk_map_t,
+    u.rb_link, arena_run_comp)
+
+static inline int
+arena_avail_comp(arena_chunk_map_t *a, arena_chunk_map_t *b)
+{
+	int ret;
+	size_t a_size = a->bits & ~PAGE_MASK;
+	size_t b_size = b->bits & ~PAGE_MASK;
+
+	ret = (a_size > b_size) - (a_size < b_size);
+	if (ret == 0) {
+		uintptr_t a_mapelm, b_mapelm;
+
+		if ((a->bits & CHUNK_MAP_KEY) != CHUNK_MAP_KEY)
+			a_mapelm = (uintptr_t)a;
+		else {
+			/*
+			 * Treat keys as though they are lower than anything
+			 * else.
+			 */
+			a_mapelm = 0;
+		}
+		b_mapelm = (uintptr_t)b;
+
+		ret = (a_mapelm > b_mapelm) - (a_mapelm < b_mapelm);
+	}
+
+	return (ret);
+}
+
+/* Generate red-black tree functions. */
+rb_gen(static UNUSED, arena_avail_tree_, arena_avail_tree_t, arena_chunk_map_t,
+    u.rb_link, arena_avail_comp)
+
+static inline int
+arena_chunk_dirty_comp(arena_chunk_t *a, arena_chunk_t *b)
+{
+
+	assert(a != NULL);
+	assert(b != NULL);
+
+	/*
+	 * Short-circuit for self comparison.  The following comparison code
+	 * would come to the same result, but at the cost of executing the slow
+	 * path.
+	 */
+	if (a == b)
+		return (0);
+
+	/*
+	 * Order such that chunks with higher fragmentation are "less than"
+	 * those with lower fragmentation -- purging order is from "least" to
+	 * "greatest".  Fragmentation is measured as:
+	 *
+	 *     mean current avail run size
+	 *   --------------------------------
+	 *   mean defragmented avail run size
+	 *
+	 *            navail
+	 *         -----------
+	 *         nruns_avail           nruns_avail-nruns_adjac
+	 * = ========================= = -----------------------
+	 *            navail                  nruns_avail
+	 *    -----------------------
+	 *    nruns_avail-nruns_adjac
+	 *
+	 * The following code multiplies away the denominator prior to
+	 * comparison, in order to avoid division.
+	 *
+	 */
+	{
+		size_t a_val = (a->nruns_avail - a->nruns_adjac) *
+		    b->nruns_avail;
+		size_t b_val = (b->nruns_avail - b->nruns_adjac) *
+		    a->nruns_avail;
+
+		if (a_val < b_val)
+			return (1);
+		if (a_val > b_val)
+			return (-1);
+	}
+	/*
+	 * Break ties by chunk address.  For fragmented chunks, report lower
+	 * addresses as "lower", so that fragmentation reduction happens first
+	 * at lower addresses.  However, use the opposite ordering for
+	 * unfragmented chunks, in order to increase the chances of
+	 * re-allocating dirty runs.
+	 */
+	{
+		uintptr_t a_chunk = (uintptr_t)a;
+		uintptr_t b_chunk = (uintptr_t)b;
+		int ret = ((a_chunk > b_chunk) - (a_chunk < b_chunk));
+		if (a->nruns_adjac == 0) {
+			assert(b->nruns_adjac == 0);
+			ret = -ret;
+		}
+		return (ret);
+	}
+}
+
+/* Generate red-black tree functions. */
+rb_gen(static UNUSED, arena_chunk_dirty_, arena_chunk_tree_t, arena_chunk_t,
+    dirty_link, arena_chunk_dirty_comp)
+
+static inline bool
+arena_avail_adjac_pred(arena_chunk_t *chunk, size_t pageind)
+{
+	bool ret;
+
+	if (pageind-1 < map_bias)
+		ret = false;
+	else {
+		ret = (arena_mapbits_allocated_get(chunk, pageind-1) == 0);
+		assert(ret == false || arena_mapbits_dirty_get(chunk,
+		    pageind-1) != arena_mapbits_dirty_get(chunk, pageind));
+	}
+	return (ret);
+}
+
+static inline bool
+arena_avail_adjac_succ(arena_chunk_t *chunk, size_t pageind, size_t npages)
+{
+	bool ret;
+
+	if (pageind+npages == chunk_npages)
+		ret = false;
+	else {
+		assert(pageind+npages < chunk_npages);
+		ret = (arena_mapbits_allocated_get(chunk, pageind+npages) == 0);
+		assert(ret == false || arena_mapbits_dirty_get(chunk, pageind)
+		    != arena_mapbits_dirty_get(chunk, pageind+npages));
+	}
+	return (ret);
+}
+
+static inline bool
+arena_avail_adjac(arena_chunk_t *chunk, size_t pageind, size_t npages)
+{
+
+	return (arena_avail_adjac_pred(chunk, pageind) ||
+	    arena_avail_adjac_succ(chunk, pageind, npages));
+}
+
+static void
+arena_avail_insert(arena_t *arena, arena_chunk_t *chunk, size_t pageind,
+    size_t npages, bool maybe_adjac_pred, bool maybe_adjac_succ)
+{
+
+	assert(npages == (arena_mapbits_unallocated_size_get(chunk, pageind) >>
+	    LG_PAGE));
+
+	/*
+	 * chunks_dirty is keyed by nruns_{avail,adjac}, so the chunk must be
+	 * removed and reinserted even if the run to be inserted is clean.
+	 */
+	if (chunk->ndirty != 0)
+		arena_chunk_dirty_remove(&arena->chunks_dirty, chunk);
+
+	if (maybe_adjac_pred && arena_avail_adjac_pred(chunk, pageind))
+		chunk->nruns_adjac++;
+	if (maybe_adjac_succ && arena_avail_adjac_succ(chunk, pageind, npages))
+		chunk->nruns_adjac++;
+	chunk->nruns_avail++;
+	assert(chunk->nruns_avail > chunk->nruns_adjac);
+
+	if (arena_mapbits_dirty_get(chunk, pageind) != 0) {
+		arena->ndirty += npages;
+		chunk->ndirty += npages;
+	}
+	if (chunk->ndirty != 0)
+		arena_chunk_dirty_insert(&arena->chunks_dirty, chunk);
+
+	arena_avail_tree_insert(&arena->runs_avail, arena_mapp_get(chunk,
+	    pageind));
+}
+
+static void
+arena_avail_remove(arena_t *arena, arena_chunk_t *chunk, size_t pageind,
+    size_t npages, bool maybe_adjac_pred, bool maybe_adjac_succ)
+{
+
+	assert(npages == (arena_mapbits_unallocated_size_get(chunk, pageind) >>
+	    LG_PAGE));
+
+	/*
+	 * chunks_dirty is keyed by nruns_{avail,adjac}, so the chunk must be
+	 * removed and reinserted even if the run to be removed is clean.
+	 */
+	if (chunk->ndirty != 0)
+		arena_chunk_dirty_remove(&arena->chunks_dirty, chunk);
+
+	if (maybe_adjac_pred && arena_avail_adjac_pred(chunk, pageind))
+		chunk->nruns_adjac--;
+	if (maybe_adjac_succ && arena_avail_adjac_succ(chunk, pageind, npages))
+		chunk->nruns_adjac--;
+	chunk->nruns_avail--;
+	assert(chunk->nruns_avail > chunk->nruns_adjac || (chunk->nruns_avail
+	    == 0 && chunk->nruns_adjac == 0));
+
+	if (arena_mapbits_dirty_get(chunk, pageind) != 0) {
+		arena->ndirty -= npages;
+		chunk->ndirty -= npages;
+	}
+	if (chunk->ndirty != 0)
+		arena_chunk_dirty_insert(&arena->chunks_dirty, chunk);
+
+	arena_avail_tree_remove(&arena->runs_avail, arena_mapp_get(chunk,
+	    pageind));
+}
+
+static inline void *
+arena_run_reg_alloc(arena_run_t *run, arena_bin_info_t *bin_info)
+{
+	void *ret;
+	unsigned regind;
+	bitmap_t *bitmap = (bitmap_t *)((uintptr_t)run +
+	    (uintptr_t)bin_info->bitmap_offset);
+
+	assert(run->nfree > 0);
+	assert(bitmap_full(bitmap, &bin_info->bitmap_info) == false);
+
+	regind = bitmap_sfu(bitmap, &bin_info->bitmap_info);
+	ret = (void *)((uintptr_t)run + (uintptr_t)bin_info->reg0_offset +
+	    (uintptr_t)(bin_info->reg_interval * regind));
+	run->nfree--;
+	if (regind == run->nextind)
+		run->nextind++;
+	assert(regind < run->nextind);
+	return (ret);
+}
+
+static inline void
+arena_run_reg_dalloc(arena_run_t *run, void *ptr)
+{
+	arena_chunk_t *chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run);
+	size_t pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+	size_t mapbits = arena_mapbits_get(chunk, pageind);
+	size_t binind = arena_ptr_small_binind_get(ptr, mapbits);
+	arena_bin_info_t *bin_info = &arena_bin_info[binind];
+	unsigned regind = arena_run_regind(run, bin_info, ptr);
+	bitmap_t *bitmap = (bitmap_t *)((uintptr_t)run +
+	    (uintptr_t)bin_info->bitmap_offset);
+
+	assert(run->nfree < bin_info->nregs);
+	/* Freeing an interior pointer can cause assertion failure. */
+	assert(((uintptr_t)ptr - ((uintptr_t)run +
+	    (uintptr_t)bin_info->reg0_offset)) %
+	    (uintptr_t)bin_info->reg_interval == 0);
+	assert((uintptr_t)ptr >= (uintptr_t)run +
+	    (uintptr_t)bin_info->reg0_offset);
+	/* Freeing an unallocated pointer can cause assertion failure. */
+	assert(bitmap_get(bitmap, &bin_info->bitmap_info, regind));
+
+	bitmap_unset(bitmap, &bin_info->bitmap_info, regind);
+	run->nfree++;
+}
+
+static inline void
+arena_run_zero(arena_chunk_t *chunk, size_t run_ind, size_t npages)
+{
+
+	VALGRIND_MAKE_MEM_UNDEFINED((void *)((uintptr_t)chunk + (run_ind <<
+	    LG_PAGE)), (npages << LG_PAGE));
+	memset((void *)((uintptr_t)chunk + (run_ind << LG_PAGE)), 0,
+	    (npages << LG_PAGE));
+}
+
+static inline void
+arena_run_page_validate_zeroed(arena_chunk_t *chunk, size_t run_ind)
+{
+	size_t i;
+	UNUSED size_t *p = (size_t *)((uintptr_t)chunk + (run_ind << LG_PAGE));
+
+	VALGRIND_MAKE_MEM_DEFINED((void *)((uintptr_t)chunk + (run_ind <<
+	    LG_PAGE)), PAGE);
+	for (i = 0; i < PAGE / sizeof(size_t); i++)
+		assert(p[i] == 0);
+}
+
+static void
+arena_run_split(arena_t *arena, arena_run_t *run, size_t size, bool large,
+    size_t binind, bool zero)
+{
+	arena_chunk_t *chunk;
+	size_t run_ind, total_pages, need_pages, rem_pages, i;
+	size_t flag_dirty;
+
+	assert((large && binind == BININD_INVALID) || (large == false && binind
+	    != BININD_INVALID));
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run);
+	run_ind = (unsigned)(((uintptr_t)run - (uintptr_t)chunk) >> LG_PAGE);
+	flag_dirty = arena_mapbits_dirty_get(chunk, run_ind);
+	total_pages = arena_mapbits_unallocated_size_get(chunk, run_ind) >>
+	    LG_PAGE;
+	assert(arena_mapbits_dirty_get(chunk, run_ind+total_pages-1) ==
+	    flag_dirty);
+	need_pages = (size >> LG_PAGE);
+	assert(need_pages > 0);
+	assert(need_pages <= total_pages);
+	rem_pages = total_pages - need_pages;
+
+	arena_avail_remove(arena, chunk, run_ind, total_pages, true, true);
+	if (config_stats) {
+		/*
+		 * Update stats_cactive if nactive is crossing a chunk
+		 * multiple.
+		 */
+		size_t cactive_diff = CHUNK_CEILING((arena->nactive +
+		    need_pages) << LG_PAGE) - CHUNK_CEILING(arena->nactive <<
+		    LG_PAGE);
+		if (cactive_diff != 0)
+			stats_cactive_add(cactive_diff);
+	}
+	arena->nactive += need_pages;
+
+	/* Keep track of trailing unused pages for later use. */
+	if (rem_pages > 0) {
+		if (flag_dirty != 0) {
+			arena_mapbits_unallocated_set(chunk, run_ind+need_pages,
+			    (rem_pages << LG_PAGE), CHUNK_MAP_DIRTY);
+			arena_mapbits_unallocated_set(chunk,
+			    run_ind+total_pages-1, (rem_pages << LG_PAGE),
+			    CHUNK_MAP_DIRTY);
+		} else {
+			arena_mapbits_unallocated_set(chunk, run_ind+need_pages,
+			    (rem_pages << LG_PAGE),
+			    arena_mapbits_unzeroed_get(chunk,
+			    run_ind+need_pages));
+			arena_mapbits_unallocated_set(chunk,
+			    run_ind+total_pages-1, (rem_pages << LG_PAGE),
+			    arena_mapbits_unzeroed_get(chunk,
+			    run_ind+total_pages-1));
+		}
+		arena_avail_insert(arena, chunk, run_ind+need_pages, rem_pages,
+		    false, true);
+	}
+
+	/*
+	 * Update the page map separately for large vs. small runs, since it is
+	 * possible to avoid iteration for large mallocs.
+	 */
+	if (large) {
+		if (zero) {
+			if (flag_dirty == 0) {
+				/*
+				 * The run is clean, so some pages may be
+				 * zeroed (i.e. never before touched).
+				 */
+				for (i = 0; i < need_pages; i++) {
+					if (arena_mapbits_unzeroed_get(chunk,
+					    run_ind+i) != 0) {
+						arena_run_zero(chunk, run_ind+i,
+						    1);
+					} else if (config_debug) {
+						arena_run_page_validate_zeroed(
+						    chunk, run_ind+i);
+					}
+				}
+			} else {
+				/*
+				 * The run is dirty, so all pages must be
+				 * zeroed.
+				 */
+				arena_run_zero(chunk, run_ind, need_pages);
+			}
+		}
+
+		/*
+		 * Set the last element first, in case the run only contains one
+		 * page (i.e. both statements set the same element).
+		 */
+		arena_mapbits_large_set(chunk, run_ind+need_pages-1, 0,
+		    flag_dirty);
+		arena_mapbits_large_set(chunk, run_ind, size, flag_dirty);
+	} else {
+		assert(zero == false);
+		/*
+		 * Propagate the dirty and unzeroed flags to the allocated
+		 * small run, so that arena_dalloc_bin_run() has the ability to
+		 * conditionally trim clean pages.
+		 */
+		arena_mapbits_small_set(chunk, run_ind, 0, binind, flag_dirty);
+		/*
+		 * The first page will always be dirtied during small run
+		 * initialization, so a validation failure here would not
+		 * actually cause an observable failure.
+		 */
+		if (config_debug && flag_dirty == 0 &&
+		    arena_mapbits_unzeroed_get(chunk, run_ind) == 0)
+			arena_run_page_validate_zeroed(chunk, run_ind);
+		for (i = 1; i < need_pages - 1; i++) {
+			arena_mapbits_small_set(chunk, run_ind+i, i, binind, 0);
+			if (config_debug && flag_dirty == 0 &&
+			    arena_mapbits_unzeroed_get(chunk, run_ind+i) == 0) {
+				arena_run_page_validate_zeroed(chunk,
+				    run_ind+i);
+			}
+		}
+		arena_mapbits_small_set(chunk, run_ind+need_pages-1,
+		    need_pages-1, binind, flag_dirty);
+		if (config_debug && flag_dirty == 0 &&
+		    arena_mapbits_unzeroed_get(chunk, run_ind+need_pages-1) ==
+		    0) {
+			arena_run_page_validate_zeroed(chunk,
+			    run_ind+need_pages-1);
+		}
+	}
+	VALGRIND_MAKE_MEM_UNDEFINED((void *)((uintptr_t)chunk + (run_ind <<
+	    LG_PAGE)), (need_pages << LG_PAGE));
+}
+
+static arena_chunk_t *
+arena_chunk_alloc(arena_t *arena)
+{
+	arena_chunk_t *chunk;
+	size_t i;
+
+	if (arena->spare != NULL) {
+		chunk = arena->spare;
+		arena->spare = NULL;
+
+		assert(arena_mapbits_allocated_get(chunk, map_bias) == 0);
+		assert(arena_mapbits_allocated_get(chunk, chunk_npages-1) == 0);
+		assert(arena_mapbits_unallocated_size_get(chunk, map_bias) ==
+		    arena_maxclass);
+		assert(arena_mapbits_unallocated_size_get(chunk,
+		    chunk_npages-1) == arena_maxclass);
+		assert(arena_mapbits_dirty_get(chunk, map_bias) ==
+		    arena_mapbits_dirty_get(chunk, chunk_npages-1));
+	} else {
+		bool zero;
+		size_t unzeroed;
+
+		zero = false;
+		malloc_mutex_unlock(&arena->lock);
+		chunk = (arena_chunk_t *)chunk_alloc(chunksize, chunksize,
+		    false, &zero, arena->dss_prec);
+		malloc_mutex_lock(&arena->lock);
+		if (chunk == NULL)
+			return (NULL);
+		if (config_stats)
+			arena->stats.mapped += chunksize;
+
+		chunk->arena = arena;
+
+		/*
+		 * Claim that no pages are in use, since the header is merely
+		 * overhead.
+		 */
+		chunk->ndirty = 0;
+
+		chunk->nruns_avail = 0;
+		chunk->nruns_adjac = 0;
+
+		/*
+		 * Initialize the map to contain one maximal free untouched run.
+		 * Mark the pages as zeroed iff chunk_alloc() returned a zeroed
+		 * chunk.
+		 */
+		unzeroed = zero ? 0 : CHUNK_MAP_UNZEROED;
+		arena_mapbits_unallocated_set(chunk, map_bias, arena_maxclass,
+		    unzeroed);
+		/*
+		 * There is no need to initialize the internal page map entries
+		 * unless the chunk is not zeroed.
+		 */
+		if (zero == false) {
+			for (i = map_bias+1; i < chunk_npages-1; i++)
+				arena_mapbits_unzeroed_set(chunk, i, unzeroed);
+		} else if (config_debug) {
+			VALGRIND_MAKE_MEM_DEFINED(
+			    (void *)arena_mapp_get(chunk, map_bias+1),
+			    (void *)((uintptr_t)
+			    arena_mapp_get(chunk, chunk_npages-1)
+			    - (uintptr_t)arena_mapp_get(chunk, map_bias+1)));
+			for (i = map_bias+1; i < chunk_npages-1; i++) {
+				assert(arena_mapbits_unzeroed_get(chunk, i) ==
+				    unzeroed);
+			}
+		}
+		arena_mapbits_unallocated_set(chunk, chunk_npages-1,
+		    arena_maxclass, unzeroed);
+	}
+
+	/* Insert the run into the runs_avail tree. */
+	arena_avail_insert(arena, chunk, map_bias, chunk_npages-map_bias,
+	    false, false);
+
+	return (chunk);
+}
+
+static void
+arena_chunk_dealloc(arena_t *arena, arena_chunk_t *chunk)
+{
+	assert(arena_mapbits_allocated_get(chunk, map_bias) == 0);
+	assert(arena_mapbits_allocated_get(chunk, chunk_npages-1) == 0);
+	assert(arena_mapbits_unallocated_size_get(chunk, map_bias) ==
+	    arena_maxclass);
+	assert(arena_mapbits_unallocated_size_get(chunk, chunk_npages-1) ==
+	    arena_maxclass);
+	assert(arena_mapbits_dirty_get(chunk, map_bias) ==
+	    arena_mapbits_dirty_get(chunk, chunk_npages-1));
+
+	/*
+	 * Remove run from the runs_avail tree, so that the arena does not use
+	 * it.
+	 */
+	arena_avail_remove(arena, chunk, map_bias, chunk_npages-map_bias,
+	    false, false);
+
+	if (arena->spare != NULL) {
+		arena_chunk_t *spare = arena->spare;
+
+		arena->spare = chunk;
+		malloc_mutex_unlock(&arena->lock);
+		chunk_dealloc((void *)spare, chunksize, true);
+		malloc_mutex_lock(&arena->lock);
+		if (config_stats)
+			arena->stats.mapped -= chunksize;
+	} else
+		arena->spare = chunk;
+}
+
+static arena_run_t *
+arena_run_alloc_helper(arena_t *arena, size_t size, bool large, size_t binind,
+    bool zero)
+{
+	arena_run_t *run;
+	arena_chunk_map_t *mapelm, key;
+
+	key.bits = size | CHUNK_MAP_KEY;
+	mapelm = arena_avail_tree_nsearch(&arena->runs_avail, &key);
+	if (mapelm != NULL) {
+		arena_chunk_t *run_chunk = CHUNK_ADDR2BASE(mapelm);
+		size_t pageind = (((uintptr_t)mapelm -
+		    (uintptr_t)run_chunk->map) / sizeof(arena_chunk_map_t))
+		    + map_bias;
+
+		run = (arena_run_t *)((uintptr_t)run_chunk + (pageind <<
+		    LG_PAGE));
+		arena_run_split(arena, run, size, large, binind, zero);
+		return (run);
+	}
+
+	return (NULL);
+}
+
+static arena_run_t *
+arena_run_alloc(arena_t *arena, size_t size, bool large, size_t binind,
+    bool zero)
+{
+	arena_chunk_t *chunk;
+	arena_run_t *run;
+
+	assert(size <= arena_maxclass);
+	assert((size & PAGE_MASK) == 0);
+	assert((large && binind == BININD_INVALID) || (large == false && binind
+	    != BININD_INVALID));
+
+	/* Search the arena's chunks for the lowest best fit. */
+	run = arena_run_alloc_helper(arena, size, large, binind, zero);
+	if (run != NULL)
+		return (run);
+
+	/*
+	 * No usable runs.  Create a new chunk from which to allocate the run.
+	 */
+	chunk = arena_chunk_alloc(arena);
+	if (chunk != NULL) {
+		run = (arena_run_t *)((uintptr_t)chunk + (map_bias << LG_PAGE));
+		arena_run_split(arena, run, size, large, binind, zero);
+		return (run);
+	}
+
+	/*
+	 * arena_chunk_alloc() failed, but another thread may have made
+	 * sufficient memory available while this one dropped arena->lock in
+	 * arena_chunk_alloc(), so search one more time.
+	 */
+	return (arena_run_alloc_helper(arena, size, large, binind, zero));
+}
+
+static inline void
+arena_maybe_purge(arena_t *arena)
+{
+	size_t npurgeable, threshold;
+
+	/* Don't purge if the option is disabled. */
+	if (opt_lg_dirty_mult < 0)
+		return;
+	/* Don't purge if all dirty pages are already being purged. */
+	if (arena->ndirty <= arena->npurgatory)
+		return;
+	npurgeable = arena->ndirty - arena->npurgatory;
+	threshold = (arena->nactive >> opt_lg_dirty_mult);
+	/*
+	 * Don't purge unless the number of purgeable pages exceeds the
+	 * threshold.
+	 */
+	if (npurgeable <= threshold)
+		return;
+
+	arena_purge(arena, false);
+}
+
+static inline size_t
+arena_chunk_purge(arena_t *arena, arena_chunk_t *chunk, bool all)
+{
+	size_t npurged;
+	ql_head(arena_chunk_map_t) mapelms;
+	arena_chunk_map_t *mapelm;
+	size_t pageind, npages;
+	size_t nmadvise;
+
+	ql_new(&mapelms);
+
+	/*
+	 * If chunk is the spare, temporarily re-allocate it, 1) so that its
+	 * run is reinserted into runs_avail, and 2) so that it cannot be
+	 * completely discarded by another thread while arena->lock is dropped
+	 * by this thread.  Note that the arena_run_dalloc() call will
+	 * implicitly deallocate the chunk, so no explicit action is required
+	 * in this function to deallocate the chunk.
+	 *
+	 * Note that once a chunk contains dirty pages, it cannot again contain
+	 * a single run unless 1) it is a dirty run, or 2) this function purges
+	 * dirty pages and causes the transition to a single clean run.  Thus
+	 * (chunk == arena->spare) is possible, but it is not possible for
+	 * this function to be called on the spare unless it contains a dirty
+	 * run.
+	 */
+	if (chunk == arena->spare) {
+		assert(arena_mapbits_dirty_get(chunk, map_bias) != 0);
+		assert(arena_mapbits_dirty_get(chunk, chunk_npages-1) != 0);
+
+		arena_chunk_alloc(arena);
+	}
+
+	if (config_stats)
+		arena->stats.purged += chunk->ndirty;
+
+	/*
+	 * Operate on all dirty runs if there is no clean/dirty run
+	 * fragmentation.
+	 */
+	if (chunk->nruns_adjac == 0)
+		all = true;
+
+	/*
+	 * Temporarily allocate free dirty runs within chunk.  If all is false,
+	 * only operate on dirty runs that are fragments; otherwise operate on
+	 * all dirty runs.
+	 */
+	for (pageind = map_bias; pageind < chunk_npages; pageind += npages) {
+		mapelm = arena_mapp_get(chunk, pageind);
+		if (arena_mapbits_allocated_get(chunk, pageind) == 0) {
+			size_t run_size =
+			    arena_mapbits_unallocated_size_get(chunk, pageind);
+
+			npages = run_size >> LG_PAGE;
+			assert(pageind + npages <= chunk_npages);
+			assert(arena_mapbits_dirty_get(chunk, pageind) ==
+			    arena_mapbits_dirty_get(chunk, pageind+npages-1));
+
+			if (arena_mapbits_dirty_get(chunk, pageind) != 0 &&
+			    (all || arena_avail_adjac(chunk, pageind,
+			    npages))) {
+				arena_run_t *run = (arena_run_t *)((uintptr_t)
+				    chunk + (uintptr_t)(pageind << LG_PAGE));
+
+				arena_run_split(arena, run, run_size, true,
+				    BININD_INVALID, false);
+				/* Append to list for later processing. */
+				ql_elm_new(mapelm, u.ql_link);
+				ql_tail_insert(&mapelms, mapelm, u.ql_link);
+			}
+		} else {
+			/* Skip run. */
+			if (arena_mapbits_large_get(chunk, pageind) != 0) {
+				npages = arena_mapbits_large_size_get(chunk,
+				    pageind) >> LG_PAGE;
+			} else {
+				size_t binind;
+				arena_bin_info_t *bin_info;
+				arena_run_t *run = (arena_run_t *)((uintptr_t)
+				    chunk + (uintptr_t)(pageind << LG_PAGE));
+
+				assert(arena_mapbits_small_runind_get(chunk,
+				    pageind) == 0);
+				binind = arena_bin_index(arena, run->bin);
+				bin_info = &arena_bin_info[binind];
+				npages = bin_info->run_size >> LG_PAGE;
+			}
+		}
+	}
+	assert(pageind == chunk_npages);
+	assert(chunk->ndirty == 0 || all == false);
+	assert(chunk->nruns_adjac == 0);
+
+	malloc_mutex_unlock(&arena->lock);
+	if (config_stats)
+		nmadvise = 0;
+	npurged = 0;
+	ql_foreach(mapelm, &mapelms, u.ql_link) {
+		bool unzeroed;
+		size_t flag_unzeroed, i;
+
+		pageind = (((uintptr_t)mapelm - (uintptr_t)chunk->map) /
+		    sizeof(arena_chunk_map_t)) + map_bias;
+		npages = arena_mapbits_large_size_get(chunk, pageind) >>
+		    LG_PAGE;
+		assert(pageind + npages <= chunk_npages);
+		unzeroed = pages_purge((void *)((uintptr_t)chunk + (pageind <<
+		    LG_PAGE)), (npages << LG_PAGE));
+		flag_unzeroed = unzeroed ? CHUNK_MAP_UNZEROED : 0;
+		/*
+		 * Set the unzeroed flag for all pages, now that pages_purge()
+		 * has returned whether the pages were zeroed as a side effect
+		 * of purging.  This chunk map modification is safe even though
+		 * the arena mutex isn't currently owned by this thread,
+		 * because the run is marked as allocated, thus protecting it
+		 * from being modified by any other thread.  As long as these
+		 * writes don't perturb the first and last elements'
+		 * CHUNK_MAP_ALLOCATED bits, behavior is well defined.
+		 */
+		for (i = 0; i < npages; i++) {
+			arena_mapbits_unzeroed_set(chunk, pageind+i,
+			    flag_unzeroed);
+		}
+		npurged += npages;
+		if (config_stats)
+			nmadvise++;
+	}
+	malloc_mutex_lock(&arena->lock);
+	if (config_stats)
+		arena->stats.nmadvise += nmadvise;
+
+	/* Deallocate runs. */
+	for (mapelm = ql_first(&mapelms); mapelm != NULL;
+	    mapelm = ql_first(&mapelms)) {
+		arena_run_t *run;
+
+		pageind = (((uintptr_t)mapelm - (uintptr_t)chunk->map) /
+		    sizeof(arena_chunk_map_t)) + map_bias;
+		run = (arena_run_t *)((uintptr_t)chunk + (uintptr_t)(pageind <<
+		    LG_PAGE));
+		ql_remove(&mapelms, mapelm, u.ql_link);
+		arena_run_dalloc(arena, run, false, true);
+	}
+
+	return (npurged);
+}
+
+static arena_chunk_t *
+chunks_dirty_iter_cb(arena_chunk_tree_t *tree, arena_chunk_t *chunk, void *arg)
+{
+       size_t *ndirty = (size_t *)arg;
+
+       assert(chunk->ndirty != 0);
+       *ndirty += chunk->ndirty;
+       return (NULL);
+}
+
+static void
+arena_purge(arena_t *arena, bool all)
+{
+	arena_chunk_t *chunk;
+	size_t npurgatory;
+	if (config_debug) {
+		size_t ndirty = 0;
+
+		arena_chunk_dirty_iter(&arena->chunks_dirty, NULL,
+		    chunks_dirty_iter_cb, (void *)&ndirty);
+		assert(ndirty == arena->ndirty);
+	}
+	assert(arena->ndirty > arena->npurgatory || all);
+	assert((arena->nactive >> opt_lg_dirty_mult) < (arena->ndirty -
+	    arena->npurgatory) || all);
+
+	if (config_stats)
+		arena->stats.npurge++;
+
+	/*
+	 * Compute the minimum number of pages that this thread should try to
+	 * purge, and add the result to arena->npurgatory.  This will keep
+	 * multiple threads from racing to reduce ndirty below the threshold.
+	 */
+	{
+		size_t npurgeable = arena->ndirty - arena->npurgatory;
+
+		if (all == false) {
+			size_t threshold = (arena->nactive >>
+			    opt_lg_dirty_mult);
+
+			npurgatory = npurgeable - threshold;
+		} else
+			npurgatory = npurgeable;
+	}
+	arena->npurgatory += npurgatory;
+
+	while (npurgatory > 0) {
+		size_t npurgeable, npurged, nunpurged;
+
+		/* Get next chunk with dirty pages. */
+		chunk = arena_chunk_dirty_first(&arena->chunks_dirty);
+		if (chunk == NULL) {
+			/*
+			 * This thread was unable to purge as many pages as
+			 * originally intended, due to races with other threads
+			 * that either did some of the purging work, or re-used
+			 * dirty pages.
+			 */
+			arena->npurgatory -= npurgatory;
+			return;
+		}
+		npurgeable = chunk->ndirty;
+		assert(npurgeable != 0);
+
+		if (npurgeable > npurgatory && chunk->nruns_adjac == 0) {
+			/*
+			 * This thread will purge all the dirty pages in chunk,
+			 * so set npurgatory to reflect this thread's intent to
+			 * purge the pages.  This tends to reduce the chances
+			 * of the following scenario:
+			 *
+			 * 1) This thread sets arena->npurgatory such that
+			 *    (arena->ndirty - arena->npurgatory) is at the
+			 *    threshold.
+			 * 2) This thread drops arena->lock.
+			 * 3) Another thread causes one or more pages to be
+			 *    dirtied, and immediately determines that it must
+			 *    purge dirty pages.
+			 *
+			 * If this scenario *does* play out, that's okay,
+			 * because all of the purging work being done really
+			 * needs to happen.
+			 */
+			arena->npurgatory += npurgeable - npurgatory;
+			npurgatory = npurgeable;
+		}
+
+		/*
+		 * Keep track of how many pages are purgeable, versus how many
+		 * actually get purged, and adjust counters accordingly.
+		 */
+		arena->npurgatory -= npurgeable;
+		npurgatory -= npurgeable;
+		npurged = arena_chunk_purge(arena, chunk, all);
+		nunpurged = npurgeable - npurged;
+		arena->npurgatory += nunpurged;
+		npurgatory += nunpurged;
+	}
+}
+
+void
+arena_purge_all(arena_t *arena)
+{
+
+	malloc_mutex_lock(&arena->lock);
+	arena_purge(arena, true);
+	malloc_mutex_unlock(&arena->lock);
+}
+
+static void
+arena_run_dalloc(arena_t *arena, arena_run_t *run, bool dirty, bool cleaned)
+{
+	arena_chunk_t *chunk;
+	size_t size, run_ind, run_pages, flag_dirty;
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run);
+	run_ind = (size_t)(((uintptr_t)run - (uintptr_t)chunk) >> LG_PAGE);
+	assert(run_ind >= map_bias);
+	assert(run_ind < chunk_npages);
+	if (arena_mapbits_large_get(chunk, run_ind) != 0) {
+		size = arena_mapbits_large_size_get(chunk, run_ind);
+		assert(size == PAGE ||
+		    arena_mapbits_large_size_get(chunk,
+		    run_ind+(size>>LG_PAGE)-1) == 0);
+	} else {
+		size_t binind = arena_bin_index(arena, run->bin);
+		arena_bin_info_t *bin_info = &arena_bin_info[binind];
+		size = bin_info->run_size;
+	}
+	run_pages = (size >> LG_PAGE);
+	if (config_stats) {
+		/*
+		 * Update stats_cactive if nactive is crossing a chunk
+		 * multiple.
+		 */
+		size_t cactive_diff = CHUNK_CEILING(arena->nactive << LG_PAGE) -
+		    CHUNK_CEILING((arena->nactive - run_pages) << LG_PAGE);
+		if (cactive_diff != 0)
+			stats_cactive_sub(cactive_diff);
+	}
+	arena->nactive -= run_pages;
+
+	/*
+	 * The run is dirty if the caller claims to have dirtied it, as well as
+	 * if it was already dirty before being allocated and the caller
+	 * doesn't claim to have cleaned it.
+	 */
+	assert(arena_mapbits_dirty_get(chunk, run_ind) ==
+	    arena_mapbits_dirty_get(chunk, run_ind+run_pages-1));
+	if (cleaned == false && arena_mapbits_dirty_get(chunk, run_ind) != 0)
+		dirty = true;
+	flag_dirty = dirty ? CHUNK_MAP_DIRTY : 0;
+
+	/* Mark pages as unallocated in the chunk map. */
+	if (dirty) {
+		arena_mapbits_unallocated_set(chunk, run_ind, size,
+		    CHUNK_MAP_DIRTY);
+		arena_mapbits_unallocated_set(chunk, run_ind+run_pages-1, size,
+		    CHUNK_MAP_DIRTY);
+	} else {
+		arena_mapbits_unallocated_set(chunk, run_ind, size,
+		    arena_mapbits_unzeroed_get(chunk, run_ind));
+		arena_mapbits_unallocated_set(chunk, run_ind+run_pages-1, size,
+		    arena_mapbits_unzeroed_get(chunk, run_ind+run_pages-1));
+	}
+
+	/* Try to coalesce forward. */
+	if (run_ind + run_pages < chunk_npages &&
+	    arena_mapbits_allocated_get(chunk, run_ind+run_pages) == 0 &&
+	    arena_mapbits_dirty_get(chunk, run_ind+run_pages) == flag_dirty) {
+		size_t nrun_size = arena_mapbits_unallocated_size_get(chunk,
+		    run_ind+run_pages);
+		size_t nrun_pages = nrun_size >> LG_PAGE;
+
+		/*
+		 * Remove successor from runs_avail; the coalesced run is
+		 * inserted later.
+		 */
+		assert(arena_mapbits_unallocated_size_get(chunk,
+		    run_ind+run_pages+nrun_pages-1) == nrun_size);
+		assert(arena_mapbits_dirty_get(chunk,
+		    run_ind+run_pages+nrun_pages-1) == flag_dirty);
+		arena_avail_remove(arena, chunk, run_ind+run_pages, nrun_pages,
+		    false, true);
+
+		size += nrun_size;
+		run_pages += nrun_pages;
+
+		arena_mapbits_unallocated_size_set(chunk, run_ind, size);
+		arena_mapbits_unallocated_size_set(chunk, run_ind+run_pages-1,
+		    size);
+	}
+
+	/* Try to coalesce backward. */
+	if (run_ind > map_bias && arena_mapbits_allocated_get(chunk, run_ind-1)
+	    == 0 && arena_mapbits_dirty_get(chunk, run_ind-1) == flag_dirty) {
+		size_t prun_size = arena_mapbits_unallocated_size_get(chunk,
+		    run_ind-1);
+		size_t prun_pages = prun_size >> LG_PAGE;
+
+		run_ind -= prun_pages;
+
+		/*
+		 * Remove predecessor from runs_avail; the coalesced run is
+		 * inserted later.
+		 */
+		assert(arena_mapbits_unallocated_size_get(chunk, run_ind) ==
+		    prun_size);
+		assert(arena_mapbits_dirty_get(chunk, run_ind) == flag_dirty);
+		arena_avail_remove(arena, chunk, run_ind, prun_pages, true,
+		    false);
+
+		size += prun_size;
+		run_pages += prun_pages;
+
+		arena_mapbits_unallocated_size_set(chunk, run_ind, size);
+		arena_mapbits_unallocated_size_set(chunk, run_ind+run_pages-1,
+		    size);
+	}
+
+	/* Insert into runs_avail, now that coalescing is complete. */
+	assert(arena_mapbits_unallocated_size_get(chunk, run_ind) ==
+	    arena_mapbits_unallocated_size_get(chunk, run_ind+run_pages-1));
+	assert(arena_mapbits_dirty_get(chunk, run_ind) ==
+	    arena_mapbits_dirty_get(chunk, run_ind+run_pages-1));
+	arena_avail_insert(arena, chunk, run_ind, run_pages, true, true);
+
+	/* Deallocate chunk if it is now completely unused. */
+	if (size == arena_maxclass) {
+		assert(run_ind == map_bias);
+		assert(run_pages == (arena_maxclass >> LG_PAGE));
+		arena_chunk_dealloc(arena, chunk);
+	}
+
+	/*
+	 * It is okay to do dirty page processing here even if the chunk was
+	 * deallocated above, since in that case it is the spare.  Waiting
+	 * until after possible chunk deallocation to do dirty processing
+	 * allows for an old spare to be fully deallocated, thus decreasing the
+	 * chances of spuriously crossing the dirty page purging threshold.
+	 */
+	if (dirty)
+		arena_maybe_purge(arena);
+}
+
+static void
+arena_run_trim_head(arena_t *arena, arena_chunk_t *chunk, arena_run_t *run,
+    size_t oldsize, size_t newsize)
+{
+	size_t pageind = ((uintptr_t)run - (uintptr_t)chunk) >> LG_PAGE;
+	size_t head_npages = (oldsize - newsize) >> LG_PAGE;
+	size_t flag_dirty = arena_mapbits_dirty_get(chunk, pageind);
+
+	assert(oldsize > newsize);
+
+	/*
+	 * Update the chunk map so that arena_run_dalloc() can treat the
+	 * leading run as separately allocated.  Set the last element of each
+	 * run first, in case of single-page runs.
+	 */
+	assert(arena_mapbits_large_size_get(chunk, pageind) == oldsize);
+	arena_mapbits_large_set(chunk, pageind+head_npages-1, 0, flag_dirty);
+	arena_mapbits_large_set(chunk, pageind, oldsize-newsize, flag_dirty);
+
+	if (config_debug) {
+		UNUSED size_t tail_npages = newsize >> LG_PAGE;
+		assert(arena_mapbits_large_size_get(chunk,
+		    pageind+head_npages+tail_npages-1) == 0);
+		assert(arena_mapbits_dirty_get(chunk,
+		    pageind+head_npages+tail_npages-1) == flag_dirty);
+	}
+	arena_mapbits_large_set(chunk, pageind+head_npages, newsize,
+	    flag_dirty);
+
+	arena_run_dalloc(arena, run, false, false);
+}
+
+static void
+arena_run_trim_tail(arena_t *arena, arena_chunk_t *chunk, arena_run_t *run,
+    size_t oldsize, size_t newsize, bool dirty)
+{
+	size_t pageind = ((uintptr_t)run - (uintptr_t)chunk) >> LG_PAGE;
+	size_t head_npages = newsize >> LG_PAGE;
+	size_t flag_dirty = arena_mapbits_dirty_get(chunk, pageind);
+
+	assert(oldsize > newsize);
+
+	/*
+	 * Update the chunk map so that arena_run_dalloc() can treat the
+	 * trailing run as separately allocated.  Set the last element of each
+	 * run first, in case of single-page runs.
+	 */
+	assert(arena_mapbits_large_size_get(chunk, pageind) == oldsize);
+	arena_mapbits_large_set(chunk, pageind+head_npages-1, 0, flag_dirty);
+	arena_mapbits_large_set(chunk, pageind, newsize, flag_dirty);
+
+	if (config_debug) {
+		UNUSED size_t tail_npages = (oldsize - newsize) >> LG_PAGE;
+		assert(arena_mapbits_large_size_get(chunk,
+		    pageind+head_npages+tail_npages-1) == 0);
+		assert(arena_mapbits_dirty_get(chunk,
+		    pageind+head_npages+tail_npages-1) == flag_dirty);
+	}
+	arena_mapbits_large_set(chunk, pageind+head_npages, oldsize-newsize,
+	    flag_dirty);
+
+	arena_run_dalloc(arena, (arena_run_t *)((uintptr_t)run + newsize),
+	    dirty, false);
+}
+
+static arena_run_t *
+arena_bin_runs_first(arena_bin_t *bin)
+{
+	arena_chunk_map_t *mapelm = arena_run_tree_first(&bin->runs);
+	if (mapelm != NULL) {
+		arena_chunk_t *chunk;
+		size_t pageind;
+		arena_run_t *run;
+
+		chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(mapelm);
+		pageind = ((((uintptr_t)mapelm - (uintptr_t)chunk->map) /
+		    sizeof(arena_chunk_map_t))) + map_bias;
+		run = (arena_run_t *)((uintptr_t)chunk + (uintptr_t)((pageind -
+		    arena_mapbits_small_runind_get(chunk, pageind)) <<
+		    LG_PAGE));
+		return (run);
+	}
+
+	return (NULL);
+}
+
+static void
+arena_bin_runs_insert(arena_bin_t *bin, arena_run_t *run)
+{
+	arena_chunk_t *chunk = CHUNK_ADDR2BASE(run);
+	size_t pageind = ((uintptr_t)run - (uintptr_t)chunk) >> LG_PAGE;
+	arena_chunk_map_t *mapelm = arena_mapp_get(chunk, pageind);
+
+	assert(arena_run_tree_search(&bin->runs, mapelm) == NULL);
+
+	arena_run_tree_insert(&bin->runs, mapelm);
+}
+
+static void
+arena_bin_runs_remove(arena_bin_t *bin, arena_run_t *run)
+{
+	arena_chunk_t *chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run);
+	size_t pageind = ((uintptr_t)run - (uintptr_t)chunk) >> LG_PAGE;
+	arena_chunk_map_t *mapelm = arena_mapp_get(chunk, pageind);
+
+	assert(arena_run_tree_search(&bin->runs, mapelm) != NULL);
+
+	arena_run_tree_remove(&bin->runs, mapelm);
+}
+
+static arena_run_t *
+arena_bin_nonfull_run_tryget(arena_bin_t *bin)
+{
+	arena_run_t *run = arena_bin_runs_first(bin);
+	if (run != NULL) {
+		arena_bin_runs_remove(bin, run);
+		if (config_stats)
+			bin->stats.reruns++;
+	}
+	return (run);
+}
+
+static arena_run_t *
+arena_bin_nonfull_run_get(arena_t *arena, arena_bin_t *bin)
+{
+	arena_run_t *run;
+	size_t binind;
+	arena_bin_info_t *bin_info;
+
+	/* Look for a usable run. */
+	run = arena_bin_nonfull_run_tryget(bin);
+	if (run != NULL)
+		return (run);
+	/* No existing runs have any space available. */
+
+	binind = arena_bin_index(arena, bin);
+	bin_info = &arena_bin_info[binind];
+
+	/* Allocate a new run. */
+	malloc_mutex_unlock(&bin->lock);
+	/******************************/
+	malloc_mutex_lock(&arena->lock);
+	run = arena_run_alloc(arena, bin_info->run_size, false, binind, false);
+	if (run != NULL) {
+		bitmap_t *bitmap = (bitmap_t *)((uintptr_t)run +
+		    (uintptr_t)bin_info->bitmap_offset);
+
+		/* Initialize run internals. */
+		run->bin = bin;
+		run->nextind = 0;
+		run->nfree = bin_info->nregs;
+		bitmap_init(bitmap, &bin_info->bitmap_info);
+	}
+	malloc_mutex_unlock(&arena->lock);
+	/********************************/
+	malloc_mutex_lock(&bin->lock);
+	if (run != NULL) {
+		if (config_stats) {
+			bin->stats.nruns++;
+			bin->stats.curruns++;
+		}
+		return (run);
+	}
+
+	/*
+	 * arena_run_alloc() failed, but another thread may have made
+	 * sufficient memory available while this one dropped bin->lock above,
+	 * so search one more time.
+	 */
+	run = arena_bin_nonfull_run_tryget(bin);
+	if (run != NULL)
+		return (run);
+
+	return (NULL);
+}
+
+/* Re-fill bin->runcur, then call arena_run_reg_alloc(). */
+static void *
+arena_bin_malloc_hard(arena_t *arena, arena_bin_t *bin)
+{
+	void *ret;
+	size_t binind;
+	arena_bin_info_t *bin_info;
+	arena_run_t *run;
+
+	binind = arena_bin_index(arena, bin);
+	bin_info = &arena_bin_info[binind];
+	bin->runcur = NULL;
+	run = arena_bin_nonfull_run_get(arena, bin);
+	if (bin->runcur != NULL && bin->runcur->nfree > 0) {
+		/*
+		 * Another thread updated runcur while this one ran without the
+		 * bin lock in arena_bin_nonfull_run_get().
+		 */
+		assert(bin->runcur->nfree > 0);
+		ret = arena_run_reg_alloc(bin->runcur, bin_info);
+		if (run != NULL) {
+			arena_chunk_t *chunk;
+
+			/*
+			 * arena_run_alloc() may have allocated run, or it may
+			 * have pulled run from the bin's run tree.  Therefore
+			 * it is unsafe to make any assumptions about how run
+			 * has previously been used, and arena_bin_lower_run()
+			 * must be called, as if a region were just deallocated
+			 * from the run.
+			 */
+			chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run);
+			if (run->nfree == bin_info->nregs)
+				arena_dalloc_bin_run(arena, chunk, run, bin);
+			else
+				arena_bin_lower_run(arena, chunk, run, bin);
+		}
+		return (ret);
+	}
+
+	if (run == NULL)
+		return (NULL);
+
+	bin->runcur = run;
+
+	assert(bin->runcur->nfree > 0);
+
+	return (arena_run_reg_alloc(bin->runcur, bin_info));
+}
+
+void
+arena_tcache_fill_small(arena_t *arena, tcache_bin_t *tbin, size_t binind,
+    uint64_t prof_accumbytes)
+{
+	unsigned i, nfill;
+	arena_bin_t *bin;
+	arena_run_t *run;
+	void *ptr;
+
+	assert(tbin->ncached == 0);
+
+	if (config_prof && arena_prof_accum(arena, prof_accumbytes))
+		prof_idump();
+	bin = &arena->bins[binind];
+	malloc_mutex_lock(&bin->lock);
+	for (i = 0, nfill = (tcache_bin_info[binind].ncached_max >>
+	    tbin->lg_fill_div); i < nfill; i++) {
+		if ((run = bin->runcur) != NULL && run->nfree > 0)
+			ptr = arena_run_reg_alloc(run, &arena_bin_info[binind]);
+		else
+			ptr = arena_bin_malloc_hard(arena, bin);
+		if (ptr == NULL)
+			break;
+		if (config_fill && opt_junk) {
+			arena_alloc_junk_small(ptr, &arena_bin_info[binind],
+			    true);
+		}
+		/* Insert such that low regions get used first. */
+		tbin->avail[nfill - 1 - i] = ptr;
+	}
+	if (config_stats) {
+		bin->stats.allocated += i * arena_bin_info[binind].reg_size;
+		bin->stats.nmalloc += i;
+		bin->stats.nrequests += tbin->tstats.nrequests;
+		bin->stats.nfills++;
+		tbin->tstats.nrequests = 0;
+	}
+	malloc_mutex_unlock(&bin->lock);
+	tbin->ncached = i;
+}
+
+void
+arena_alloc_junk_small(void *ptr, arena_bin_info_t *bin_info, bool zero)
+{
+
+	if (zero) {
+		size_t redzone_size = bin_info->redzone_size;
+		memset((void *)((uintptr_t)ptr - redzone_size), 0xa5,
+		    redzone_size);
+		memset((void *)((uintptr_t)ptr + bin_info->reg_size), 0xa5,
+		    redzone_size);
+	} else {
+		memset((void *)((uintptr_t)ptr - bin_info->redzone_size), 0xa5,
+		    bin_info->reg_interval);
+	}
+}
+
+void
+arena_dalloc_junk_small(void *ptr, arena_bin_info_t *bin_info)
+{
+	size_t size = bin_info->reg_size;
+	size_t redzone_size = bin_info->redzone_size;
+	size_t i;
+	bool error = false;
+
+	for (i = 1; i <= redzone_size; i++) {
+		unsigned byte;
+		if ((byte = *(uint8_t *)((uintptr_t)ptr - i)) != 0xa5) {
+			error = true;
+			malloc_printf("<jemalloc>: Corrupt redzone "
+			    "%zu byte%s before %p (size %zu), byte=%#x\n", i,
+			    (i == 1) ? "" : "s", ptr, size, byte);
+		}
+	}
+	for (i = 0; i < redzone_size; i++) {
+		unsigned byte;
+		if ((byte = *(uint8_t *)((uintptr_t)ptr + size + i)) != 0xa5) {
+			error = true;
+			malloc_printf("<jemalloc>: Corrupt redzone "
+			    "%zu byte%s after end of %p (size %zu), byte=%#x\n",
+			    i, (i == 1) ? "" : "s", ptr, size, byte);
+		}
+	}
+	if (opt_abort && error)
+		abort();
+
+	memset((void *)((uintptr_t)ptr - redzone_size), 0x5a,
+	    bin_info->reg_interval);
+}
+
+void *
+arena_malloc_small(arena_t *arena, size_t size, bool zero)
+{
+	void *ret;
+	arena_bin_t *bin;
+	arena_run_t *run;
+	size_t binind;
+
+	binind = SMALL_SIZE2BIN(size);
+	assert(binind < NBINS);
+	bin = &arena->bins[binind];
+	size = arena_bin_info[binind].reg_size;
+
+	malloc_mutex_lock(&bin->lock);
+	if ((run = bin->runcur) != NULL && run->nfree > 0)
+		ret = arena_run_reg_alloc(run, &arena_bin_info[binind]);
+	else
+		ret = arena_bin_malloc_hard(arena, bin);
+
+	if (ret == NULL) {
+		malloc_mutex_unlock(&bin->lock);
+		return (NULL);
+	}
+
+	if (config_stats) {
+		bin->stats.allocated += size;
+		bin->stats.nmalloc++;
+		bin->stats.nrequests++;
+	}
+	malloc_mutex_unlock(&bin->lock);
+	if (config_prof && isthreaded == false && arena_prof_accum(arena, size))
+		prof_idump();
+
+	if (zero == false) {
+		if (config_fill) {
+			if (opt_junk) {
+				arena_alloc_junk_small(ret,
+				    &arena_bin_info[binind], false);
+			} else if (opt_zero)
+				memset(ret, 0, size);
+		}
+	} else {
+		if (config_fill && opt_junk) {
+			arena_alloc_junk_small(ret, &arena_bin_info[binind],
+			    true);
+		}
+		VALGRIND_MAKE_MEM_UNDEFINED(ret, size);
+		memset(ret, 0, size);
+	}
+	VALGRIND_MAKE_MEM_UNDEFINED(ret, size);
+
+	return (ret);
+}
+
+void *
+arena_malloc_large(arena_t *arena, size_t size, bool zero)
+{
+	void *ret;
+	UNUSED bool idump;
+
+	/* Large allocation. */
+	size = PAGE_CEILING(size);
+	malloc_mutex_lock(&arena->lock);
+	ret = (void *)arena_run_alloc(arena, size, true, BININD_INVALID, zero);
+	if (ret == NULL) {
+		malloc_mutex_unlock(&arena->lock);
+		return (NULL);
+	}
+	if (config_stats) {
+		arena->stats.nmalloc_large++;
+		arena->stats.nrequests_large++;
+		arena->stats.allocated_large += size;
+		arena->stats.lstats[(size >> LG_PAGE) - 1].nmalloc++;
+		arena->stats.lstats[(size >> LG_PAGE) - 1].nrequests++;
+		arena->stats.lstats[(size >> LG_PAGE) - 1].curruns++;
+	}
+	if (config_prof)
+		idump = arena_prof_accum_locked(arena, size);
+	malloc_mutex_unlock(&arena->lock);
+	if (config_prof && idump)
+		prof_idump();
+
+	if (zero == false) {
+		if (config_fill) {
+			if (opt_junk)
+				memset(ret, 0xa5, size);
+			else if (opt_zero)
+				memset(ret, 0, size);
+		}
+	}
+
+	return (ret);
+}
+
+/* Only handles large allocations that require more than page alignment. */
+void *
+arena_palloc(arena_t *arena, size_t size, size_t alignment, bool zero)
+{
+	void *ret;
+	size_t alloc_size, leadsize, trailsize;
+	arena_run_t *run;
+	arena_chunk_t *chunk;
+
+	assert((size & PAGE_MASK) == 0);
+
+	alignment = PAGE_CEILING(alignment);
+	alloc_size = size + alignment - PAGE;
+
+	malloc_mutex_lock(&arena->lock);
+	run = arena_run_alloc(arena, alloc_size, true, BININD_INVALID, zero);
+	if (run == NULL) {
+		malloc_mutex_unlock(&arena->lock);
+		return (NULL);
+	}
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run);
+
+	leadsize = ALIGNMENT_CEILING((uintptr_t)run, alignment) -
+	    (uintptr_t)run;
+	assert(alloc_size >= leadsize + size);
+	trailsize = alloc_size - leadsize - size;
+	ret = (void *)((uintptr_t)run + leadsize);
+	if (leadsize != 0) {
+		arena_run_trim_head(arena, chunk, run, alloc_size, alloc_size -
+		    leadsize);
+	}
+	if (trailsize != 0) {
+		arena_run_trim_tail(arena, chunk, ret, size + trailsize, size,
+		    false);
+	}
+
+	if (config_stats) {
+		arena->stats.nmalloc_large++;
+		arena->stats.nrequests_large++;
+		arena->stats.allocated_large += size;
+		arena->stats.lstats[(size >> LG_PAGE) - 1].nmalloc++;
+		arena->stats.lstats[(size >> LG_PAGE) - 1].nrequests++;
+		arena->stats.lstats[(size >> LG_PAGE) - 1].curruns++;
+	}
+	malloc_mutex_unlock(&arena->lock);
+
+	if (config_fill && zero == false) {
+		if (opt_junk)
+			memset(ret, 0xa5, size);
+		else if (opt_zero)
+			memset(ret, 0, size);
+	}
+	return (ret);
+}
+
+void
+arena_prof_promoted(const void *ptr, size_t size)
+{
+	arena_chunk_t *chunk;
+	size_t pageind, binind;
+
+	cassert(config_prof);
+	assert(ptr != NULL);
+	assert(CHUNK_ADDR2BASE(ptr) != ptr);
+	assert(isalloc(ptr, false) == PAGE);
+	assert(isalloc(ptr, true) == PAGE);
+	assert(size <= SMALL_MAXCLASS);
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+	pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+	binind = SMALL_SIZE2BIN(size);
+	assert(binind < NBINS);
+	arena_mapbits_large_binind_set(chunk, pageind, binind);
+
+	assert(isalloc(ptr, false) == PAGE);
+	assert(isalloc(ptr, true) == size);
+}
+
+static void
+arena_dissociate_bin_run(arena_chunk_t *chunk, arena_run_t *run,
+    arena_bin_t *bin)
+{
+
+	/* Dissociate run from bin. */
+	if (run == bin->runcur)
+		bin->runcur = NULL;
+	else {
+		size_t binind = arena_bin_index(chunk->arena, bin);
+		arena_bin_info_t *bin_info = &arena_bin_info[binind];
+
+		if (bin_info->nregs != 1) {
+			/*
+			 * This block's conditional is necessary because if the
+			 * run only contains one region, then it never gets
+			 * inserted into the non-full runs tree.
+			 */
+			arena_bin_runs_remove(bin, run);
+		}
+	}
+}
+
+static void
+arena_dalloc_bin_run(arena_t *arena, arena_chunk_t *chunk, arena_run_t *run,
+    arena_bin_t *bin)
+{
+	size_t binind;
+	arena_bin_info_t *bin_info;
+	size_t npages, run_ind, past;
+
+	assert(run != bin->runcur);
+	assert(arena_run_tree_search(&bin->runs,
+	    arena_mapp_get(chunk, ((uintptr_t)run-(uintptr_t)chunk)>>LG_PAGE))
+	    == NULL);
+
+	binind = arena_bin_index(chunk->arena, run->bin);
+	bin_info = &arena_bin_info[binind];
+
+	malloc_mutex_unlock(&bin->lock);
+	/******************************/
+	npages = bin_info->run_size >> LG_PAGE;
+	run_ind = (size_t)(((uintptr_t)run - (uintptr_t)chunk) >> LG_PAGE);
+	past = (size_t)(PAGE_CEILING((uintptr_t)run +
+	    (uintptr_t)bin_info->reg0_offset + (uintptr_t)(run->nextind *
+	    bin_info->reg_interval - bin_info->redzone_size) -
+	    (uintptr_t)chunk) >> LG_PAGE);
+	malloc_mutex_lock(&arena->lock);
+
+	/*
+	 * If the run was originally clean, and some pages were never touched,
+	 * trim the clean pages before deallocating the dirty portion of the
+	 * run.
+	 */
+	assert(arena_mapbits_dirty_get(chunk, run_ind) ==
+	    arena_mapbits_dirty_get(chunk, run_ind+npages-1));
+	if (arena_mapbits_dirty_get(chunk, run_ind) == 0 && past - run_ind <
+	    npages) {
+		/* Trim clean pages.  Convert to large run beforehand. */
+		assert(npages > 0);
+		arena_mapbits_large_set(chunk, run_ind, bin_info->run_size, 0);
+		arena_mapbits_large_set(chunk, run_ind+npages-1, 0, 0);
+		arena_run_trim_tail(arena, chunk, run, (npages << LG_PAGE),
+		    ((past - run_ind) << LG_PAGE), false);
+		/* npages = past - run_ind; */
+	}
+	arena_run_dalloc(arena, run, true, false);
+	malloc_mutex_unlock(&arena->lock);
+	/****************************/
+	malloc_mutex_lock(&bin->lock);
+	if (config_stats)
+		bin->stats.curruns--;
+}
+
+static void
+arena_bin_lower_run(arena_t *arena, arena_chunk_t *chunk, arena_run_t *run,
+    arena_bin_t *bin)
+{
+
+	/*
+	 * Make sure that if bin->runcur is non-NULL, it refers to the lowest
+	 * non-full run.  It is okay to NULL runcur out rather than proactively
+	 * keeping it pointing at the lowest non-full run.
+	 */
+	if ((uintptr_t)run < (uintptr_t)bin->runcur) {
+		/* Switch runcur. */
+		if (bin->runcur->nfree > 0)
+			arena_bin_runs_insert(bin, bin->runcur);
+		bin->runcur = run;
+		if (config_stats)
+			bin->stats.reruns++;
+	} else
+		arena_bin_runs_insert(bin, run);
+}
+
+void
+arena_dalloc_bin_locked(arena_t *arena, arena_chunk_t *chunk, void *ptr,
+    arena_chunk_map_t *mapelm)
+{
+	size_t pageind;
+	arena_run_t *run;
+	arena_bin_t *bin;
+	arena_bin_info_t *bin_info;
+	size_t size, binind;
+
+	pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+	run = (arena_run_t *)((uintptr_t)chunk + (uintptr_t)((pageind -
+	    arena_mapbits_small_runind_get(chunk, pageind)) << LG_PAGE));
+	bin = run->bin;
+	binind = arena_ptr_small_binind_get(ptr, mapelm->bits);
+	bin_info = &arena_bin_info[binind];
+	if (config_fill || config_stats)
+		size = bin_info->reg_size;
+
+	if (config_fill && opt_junk)
+		arena_dalloc_junk_small(ptr, bin_info);
+
+	arena_run_reg_dalloc(run, ptr);
+	if (run->nfree == bin_info->nregs) {
+		arena_dissociate_bin_run(chunk, run, bin);
+		arena_dalloc_bin_run(arena, chunk, run, bin);
+	} else if (run->nfree == 1 && run != bin->runcur)
+		arena_bin_lower_run(arena, chunk, run, bin);
+
+	if (config_stats) {
+		bin->stats.allocated -= size;
+		bin->stats.ndalloc++;
+	}
+}
+
+void
+arena_dalloc_bin(arena_t *arena, arena_chunk_t *chunk, void *ptr,
+    size_t pageind, arena_chunk_map_t *mapelm)
+{
+	arena_run_t *run;
+	arena_bin_t *bin;
+
+	run = (arena_run_t *)((uintptr_t)chunk + (uintptr_t)((pageind -
+	    arena_mapbits_small_runind_get(chunk, pageind)) << LG_PAGE));
+	bin = run->bin;
+	malloc_mutex_lock(&bin->lock);
+	arena_dalloc_bin_locked(arena, chunk, ptr, mapelm);
+	malloc_mutex_unlock(&bin->lock);
+}
+
+void
+arena_dalloc_small(arena_t *arena, arena_chunk_t *chunk, void *ptr,
+    size_t pageind)
+{
+	arena_chunk_map_t *mapelm;
+
+	if (config_debug) {
+		/* arena_ptr_small_binind_get() does extra sanity checking. */
+		assert(arena_ptr_small_binind_get(ptr, arena_mapbits_get(chunk,
+		    pageind)) != BININD_INVALID);
+	}
+	mapelm = arena_mapp_get(chunk, pageind);
+	arena_dalloc_bin(arena, chunk, ptr, pageind, mapelm);
+}
+
+void
+arena_dalloc_large_locked(arena_t *arena, arena_chunk_t *chunk, void *ptr)
+{
+
+	if (config_fill || config_stats) {
+		size_t pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+		size_t size = arena_mapbits_large_size_get(chunk, pageind);
+
+		if (config_fill && config_stats && opt_junk)
+			memset(ptr, 0x5a, size);
+		if (config_stats) {
+			arena->stats.ndalloc_large++;
+			arena->stats.allocated_large -= size;
+			arena->stats.lstats[(size >> LG_PAGE) - 1].ndalloc++;
+			arena->stats.lstats[(size >> LG_PAGE) - 1].curruns--;
+		}
+	}
+
+	arena_run_dalloc(arena, (arena_run_t *)ptr, true, false);
+}
+
+void
+arena_dalloc_large(arena_t *arena, arena_chunk_t *chunk, void *ptr)
+{
+
+	malloc_mutex_lock(&arena->lock);
+	arena_dalloc_large_locked(arena, chunk, ptr);
+	malloc_mutex_unlock(&arena->lock);
+}
+
+static void
+arena_ralloc_large_shrink(arena_t *arena, arena_chunk_t *chunk, void *ptr,
+    size_t oldsize, size_t size)
+{
+
+	assert(size < oldsize);
+
+	/*
+	 * Shrink the run, and make trailing pages available for other
+	 * allocations.
+	 */
+	malloc_mutex_lock(&arena->lock);
+	arena_run_trim_tail(arena, chunk, (arena_run_t *)ptr, oldsize, size,
+	    true);
+	if (config_stats) {
+		arena->stats.ndalloc_large++;
+		arena->stats.allocated_large -= oldsize;
+		arena->stats.lstats[(oldsize >> LG_PAGE) - 1].ndalloc++;
+		arena->stats.lstats[(oldsize >> LG_PAGE) - 1].curruns--;
+
+		arena->stats.nmalloc_large++;
+		arena->stats.nrequests_large++;
+		arena->stats.allocated_large += size;
+		arena->stats.lstats[(size >> LG_PAGE) - 1].nmalloc++;
+		arena->stats.lstats[(size >> LG_PAGE) - 1].nrequests++;
+		arena->stats.lstats[(size >> LG_PAGE) - 1].curruns++;
+	}
+	malloc_mutex_unlock(&arena->lock);
+}
+
+static bool
+arena_ralloc_large_grow(arena_t *arena, arena_chunk_t *chunk, void *ptr,
+    size_t oldsize, size_t size, size_t extra, bool zero)
+{
+	size_t pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+	size_t npages = oldsize >> LG_PAGE;
+	size_t followsize;
+
+	assert(oldsize == arena_mapbits_large_size_get(chunk, pageind));
+
+	/* Try to extend the run. */
+	assert(size + extra > oldsize);
+	malloc_mutex_lock(&arena->lock);
+	if (pageind + npages < chunk_npages &&
+	    arena_mapbits_allocated_get(chunk, pageind+npages) == 0 &&
+	    (followsize = arena_mapbits_unallocated_size_get(chunk,
+	    pageind+npages)) >= size - oldsize) {
+		/*
+		 * The next run is available and sufficiently large.  Split the
+		 * following run, then merge the first part with the existing
+		 * allocation.
+		 */
+		size_t flag_dirty;
+		size_t splitsize = (oldsize + followsize <= size + extra)
+		    ? followsize : size + extra - oldsize;
+		arena_run_split(arena, (arena_run_t *)((uintptr_t)chunk +
+		    ((pageind+npages) << LG_PAGE)), splitsize, true,
+		    BININD_INVALID, zero);
+
+		size = oldsize + splitsize;
+		npages = size >> LG_PAGE;
+
+		/*
+		 * Mark the extended run as dirty if either portion of the run
+		 * was dirty before allocation.  This is rather pedantic,
+		 * because there's not actually any sequence of events that
+		 * could cause the resulting run to be passed to
+		 * arena_run_dalloc() with the dirty argument set to false
+		 * (which is when dirty flag consistency would really matter).
+		 */
+		flag_dirty = arena_mapbits_dirty_get(chunk, pageind) |
+		    arena_mapbits_dirty_get(chunk, pageind+npages-1);
+		arena_mapbits_large_set(chunk, pageind, size, flag_dirty);
+		arena_mapbits_large_set(chunk, pageind+npages-1, 0, flag_dirty);
+
+		if (config_stats) {
+			arena->stats.ndalloc_large++;
+			arena->stats.allocated_large -= oldsize;
+			arena->stats.lstats[(oldsize >> LG_PAGE) - 1].ndalloc++;
+			arena->stats.lstats[(oldsize >> LG_PAGE) - 1].curruns--;
+
+			arena->stats.nmalloc_large++;
+			arena->stats.nrequests_large++;
+			arena->stats.allocated_large += size;
+			arena->stats.lstats[(size >> LG_PAGE) - 1].nmalloc++;
+			arena->stats.lstats[(size >> LG_PAGE) - 1].nrequests++;
+			arena->stats.lstats[(size >> LG_PAGE) - 1].curruns++;
+		}
+		malloc_mutex_unlock(&arena->lock);
+		return (false);
+	}
+	malloc_mutex_unlock(&arena->lock);
+
+	return (true);
+}
+
+/*
+ * Try to resize a large allocation, in order to avoid copying.  This will
+ * always fail if growing an object, and the following run is already in use.
+ */
+static bool
+arena_ralloc_large(void *ptr, size_t oldsize, size_t size, size_t extra,
+    bool zero)
+{
+	size_t psize;
+
+	psize = PAGE_CEILING(size + extra);
+	if (psize == oldsize) {
+		/* Same size class. */
+		if (config_fill && opt_junk && size < oldsize) {
+			memset((void *)((uintptr_t)ptr + size), 0x5a, oldsize -
+			    size);
+		}
+		return (false);
+	} else {
+		arena_chunk_t *chunk;
+		arena_t *arena;
+
+		chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+		arena = chunk->arena;
+
+		if (psize < oldsize) {
+			/* Fill before shrinking in order avoid a race. */
+			if (config_fill && opt_junk) {
+				memset((void *)((uintptr_t)ptr + size), 0x5a,
+				    oldsize - size);
+			}
+			arena_ralloc_large_shrink(arena, chunk, ptr, oldsize,
+			    psize);
+			return (false);
+		} else {
+			bool ret = arena_ralloc_large_grow(arena, chunk, ptr,
+			    oldsize, PAGE_CEILING(size),
+			    psize - PAGE_CEILING(size), zero);
+			if (config_fill && ret == false && zero == false &&
+			    opt_zero) {
+				memset((void *)((uintptr_t)ptr + oldsize), 0,
+				    size - oldsize);
+			}
+			return (ret);
+		}
+	}
+}
+
+void *
+arena_ralloc_no_move(void *ptr, size_t oldsize, size_t size, size_t extra,
+    bool zero)
+{
+
+	/*
+	 * Avoid moving the allocation if the size class can be left the same.
+	 */
+	if (oldsize <= arena_maxclass) {
+		if (oldsize <= SMALL_MAXCLASS) {
+			assert(arena_bin_info[SMALL_SIZE2BIN(oldsize)].reg_size
+			    == oldsize);
+			if ((size + extra <= SMALL_MAXCLASS &&
+			    SMALL_SIZE2BIN(size + extra) ==
+			    SMALL_SIZE2BIN(oldsize)) || (size <= oldsize &&
+			    size + extra >= oldsize)) {
+				if (config_fill && opt_junk && size < oldsize) {
+					memset((void *)((uintptr_t)ptr + size),
+					    0x5a, oldsize - size);
+				}
+				return (ptr);
+			}
+		} else {
+			assert(size <= arena_maxclass);
+			if (size + extra > SMALL_MAXCLASS) {
+				if (arena_ralloc_large(ptr, oldsize, size,
+				    extra, zero) == false)
+					return (ptr);
+			}
+		}
+	}
+
+	/* Reallocation would require a move. */
+	return (NULL);
+}
+
+void *
+arena_ralloc(arena_t *arena, void *ptr, size_t oldsize, size_t size,
+    size_t extra, size_t alignment, bool zero, bool try_tcache_alloc,
+    bool try_tcache_dalloc)
+{
+	void *ret;
+	size_t copysize;
+
+	/* Try to avoid moving the allocation. */
+	ret = arena_ralloc_no_move(ptr, oldsize, size, extra, zero);
+	if (ret != NULL)
+		return (ret);
+
+	/*
+	 * size and oldsize are different enough that we need to move the
+	 * object.  In that case, fall back to allocating new space and
+	 * copying.
+	 */
+	if (alignment != 0) {
+		size_t usize = sa2u(size + extra, alignment);
+		if (usize == 0)
+			return (NULL);
+		ret = ipallocx(usize, alignment, zero, try_tcache_alloc, arena);
+	} else
+		ret = arena_malloc(arena, size + extra, zero, try_tcache_alloc);
+
+	if (ret == NULL) {
+		if (extra == 0)
+			return (NULL);
+		/* Try again, this time without extra. */
+		if (alignment != 0) {
+			size_t usize = sa2u(size, alignment);
+			if (usize == 0)
+				return (NULL);
+			ret = ipallocx(usize, alignment, zero, try_tcache_alloc,
+			    arena);
+		} else
+			ret = arena_malloc(arena, size, zero, try_tcache_alloc);
+
+		if (ret == NULL)
+			return (NULL);
+	}
+
+	/* Junk/zero-filling were already done by ipalloc()/arena_malloc(). */
+
+	/*
+	 * Copy at most size bytes (not size+extra), since the caller has no
+	 * expectation that the extra bytes will be reliably preserved.
+	 */
+	copysize = (size < oldsize) ? size : oldsize;
+	VALGRIND_MAKE_MEM_UNDEFINED(ret, copysize);
+	memcpy(ret, ptr, copysize);
+	iqallocx(ptr, try_tcache_dalloc);
+	return (ret);
+}
+
+dss_prec_t
+arena_dss_prec_get(arena_t *arena)
+{
+	dss_prec_t ret;
+
+	malloc_mutex_lock(&arena->lock);
+	ret = arena->dss_prec;
+	malloc_mutex_unlock(&arena->lock);
+	return (ret);
+}
+
+void
+arena_dss_prec_set(arena_t *arena, dss_prec_t dss_prec)
+{
+
+	malloc_mutex_lock(&arena->lock);
+	arena->dss_prec = dss_prec;
+	malloc_mutex_unlock(&arena->lock);
+}
+
+void
+arena_stats_merge(arena_t *arena, const char **dss, size_t *nactive,
+    size_t *ndirty, arena_stats_t *astats, malloc_bin_stats_t *bstats,
+    malloc_large_stats_t *lstats)
+{
+	unsigned i;
+
+	malloc_mutex_lock(&arena->lock);
+	*dss = dss_prec_names[arena->dss_prec];
+	*nactive += arena->nactive;
+	*ndirty += arena->ndirty;
+
+	astats->mapped += arena->stats.mapped;
+	astats->npurge += arena->stats.npurge;
+	astats->nmadvise += arena->stats.nmadvise;
+	astats->purged += arena->stats.purged;
+	astats->allocated_large += arena->stats.allocated_large;
+	astats->nmalloc_large += arena->stats.nmalloc_large;
+	astats->ndalloc_large += arena->stats.ndalloc_large;
+	astats->nrequests_large += arena->stats.nrequests_large;
+
+	for (i = 0; i < nlclasses; i++) {
+		lstats[i].nmalloc += arena->stats.lstats[i].nmalloc;
+		lstats[i].ndalloc += arena->stats.lstats[i].ndalloc;
+		lstats[i].nrequests += arena->stats.lstats[i].nrequests;
+		lstats[i].curruns += arena->stats.lstats[i].curruns;
+	}
+	malloc_mutex_unlock(&arena->lock);
+
+	for (i = 0; i < NBINS; i++) {
+		arena_bin_t *bin = &arena->bins[i];
+
+		malloc_mutex_lock(&bin->lock);
+		bstats[i].allocated += bin->stats.allocated;
+		bstats[i].nmalloc += bin->stats.nmalloc;
+		bstats[i].ndalloc += bin->stats.ndalloc;
+		bstats[i].nrequests += bin->stats.nrequests;
+		if (config_tcache) {
+			bstats[i].nfills += bin->stats.nfills;
+			bstats[i].nflushes += bin->stats.nflushes;
+		}
+		bstats[i].nruns += bin->stats.nruns;
+		bstats[i].reruns += bin->stats.reruns;
+		bstats[i].curruns += bin->stats.curruns;
+		malloc_mutex_unlock(&bin->lock);
+	}
+}
+
+bool
+arena_new(arena_t *arena, unsigned ind)
+{
+	unsigned i;
+	arena_bin_t *bin;
+
+	arena->ind = ind;
+	arena->nthreads = 0;
+
+	if (malloc_mutex_init(&arena->lock))
+		return (true);
+
+	if (config_stats) {
+		memset(&arena->stats, 0, sizeof(arena_stats_t));
+		arena->stats.lstats =
+		    (malloc_large_stats_t *)base_alloc(nlclasses *
+		    sizeof(malloc_large_stats_t));
+		if (arena->stats.lstats == NULL)
+			return (true);
+		memset(arena->stats.lstats, 0, nlclasses *
+		    sizeof(malloc_large_stats_t));
+		if (config_tcache)
+			ql_new(&arena->tcache_ql);
+	}
+
+	if (config_prof)
+		arena->prof_accumbytes = 0;
+
+	arena->dss_prec = chunk_dss_prec_get();
+
+	/* Initialize chunks. */
+	arena_chunk_dirty_new(&arena->chunks_dirty);
+	arena->spare = NULL;
+
+	arena->nactive = 0;
+	arena->ndirty = 0;
+	arena->npurgatory = 0;
+
+	arena_avail_tree_new(&arena->runs_avail);
+
+	/* Initialize bins. */
+	for (i = 0; i < NBINS; i++) {
+		bin = &arena->bins[i];
+		if (malloc_mutex_init(&bin->lock))
+			return (true);
+		bin->runcur = NULL;
+		arena_run_tree_new(&bin->runs);
+		if (config_stats)
+			memset(&bin->stats, 0, sizeof(malloc_bin_stats_t));
+	}
+
+	return (false);
+}
+
+/*
+ * Calculate bin_info->run_size such that it meets the following constraints:
+ *
+ *   *) bin_info->run_size >= min_run_size
+ *   *) bin_info->run_size <= arena_maxclass
+ *   *) run header overhead <= RUN_MAX_OVRHD (or header overhead relaxed).
+ *   *) bin_info->nregs <= RUN_MAXREGS
+ *
+ * bin_info->nregs, bin_info->bitmap_offset, and bin_info->reg0_offset are also
+ * calculated here, since these settings are all interdependent.
+ */
+static size_t
+bin_info_run_size_calc(arena_bin_info_t *bin_info, size_t min_run_size)
+{
+	size_t pad_size;
+	size_t try_run_size, good_run_size;
+	uint32_t try_nregs, good_nregs;
+	uint32_t try_hdr_size, good_hdr_size;
+	uint32_t try_bitmap_offset, good_bitmap_offset;
+	uint32_t try_ctx0_offset, good_ctx0_offset;
+	uint32_t try_redzone0_offset, good_redzone0_offset;
+
+	assert(min_run_size >= PAGE);
+	assert(min_run_size <= arena_maxclass);
+
+	/*
+	 * Determine redzone size based on minimum alignment and minimum
+	 * redzone size.  Add padding to the end of the run if it is needed to
+	 * align the regions.  The padding allows each redzone to be half the
+	 * minimum alignment; without the padding, each redzone would have to
+	 * be twice as large in order to maintain alignment.
+	 */
+	if (config_fill && opt_redzone) {
+		size_t align_min = ZU(1) << (ffs(bin_info->reg_size) - 1);
+		if (align_min <= REDZONE_MINSIZE) {
+			bin_info->redzone_size = REDZONE_MINSIZE;
+			pad_size = 0;
+		} else {
+			bin_info->redzone_size = align_min >> 1;
+			pad_size = bin_info->redzone_size;
+		}
+	} else {
+		bin_info->redzone_size = 0;
+		pad_size = 0;
+	}
+	bin_info->reg_interval = bin_info->reg_size +
+	    (bin_info->redzone_size << 1);
+
+	/*
+	 * Calculate known-valid settings before entering the run_size
+	 * expansion loop, so that the first part of the loop always copies
+	 * valid settings.
+	 *
+	 * The do..while loop iteratively reduces the number of regions until
+	 * the run header and the regions no longer overlap.  A closed formula
+	 * would be quite messy, since there is an interdependency between the
+	 * header's mask length and the number of regions.
+	 */
+	try_run_size = min_run_size;
+	try_nregs = ((try_run_size - sizeof(arena_run_t)) /
+	    bin_info->reg_interval)
+	    + 1; /* Counter-act try_nregs-- in loop. */
+	if (try_nregs > RUN_MAXREGS) {
+		try_nregs = RUN_MAXREGS
+		    + 1; /* Counter-act try_nregs-- in loop. */
+	}
+	do {
+		try_nregs--;
+		try_hdr_size = sizeof(arena_run_t);
+		/* Pad to a long boundary. */
+		try_hdr_size = LONG_CEILING(try_hdr_size);
+		try_bitmap_offset = try_hdr_size;
+		/* Add space for bitmap. */
+		try_hdr_size += bitmap_size(try_nregs);
+		if (config_prof && opt_prof && prof_promote == false) {
+			/* Pad to a quantum boundary. */
+			try_hdr_size = QUANTUM_CEILING(try_hdr_size);
+			try_ctx0_offset = try_hdr_size;
+			/* Add space for one (prof_ctx_t *) per region. */
+			try_hdr_size += try_nregs * sizeof(prof_ctx_t *);
+		} else
+			try_ctx0_offset = 0;
+		try_redzone0_offset = try_run_size - (try_nregs *
+		    bin_info->reg_interval) - pad_size;
+	} while (try_hdr_size > try_redzone0_offset);
+
+	/* run_size expansion loop. */
+	do {
+		/*
+		 * Copy valid settings before trying more aggressive settings.
+		 */
+		good_run_size = try_run_size;
+		good_nregs = try_nregs;
+		good_hdr_size = try_hdr_size;
+		good_bitmap_offset = try_bitmap_offset;
+		good_ctx0_offset = try_ctx0_offset;
+		good_redzone0_offset = try_redzone0_offset;
+
+		/* Try more aggressive settings. */
+		try_run_size += PAGE;
+		try_nregs = ((try_run_size - sizeof(arena_run_t) - pad_size) /
+		    bin_info->reg_interval)
+		    + 1; /* Counter-act try_nregs-- in loop. */
+		if (try_nregs > RUN_MAXREGS) {
+			try_nregs = RUN_MAXREGS
+			    + 1; /* Counter-act try_nregs-- in loop. */
+		}
+		do {
+			try_nregs--;
+			try_hdr_size = sizeof(arena_run_t);
+			/* Pad to a long boundary. */
+			try_hdr_size = LONG_CEILING(try_hdr_size);
+			try_bitmap_offset = try_hdr_size;
+			/* Add space for bitmap. */
+			try_hdr_size += bitmap_size(try_nregs);
+			if (config_prof && opt_prof && prof_promote == false) {
+				/* Pad to a quantum boundary. */
+				try_hdr_size = QUANTUM_CEILING(try_hdr_size);
+				try_ctx0_offset = try_hdr_size;
+				/*
+				 * Add space for one (prof_ctx_t *) per region.
+				 */
+				try_hdr_size += try_nregs *
+				    sizeof(prof_ctx_t *);
+			}
+			try_redzone0_offset = try_run_size - (try_nregs *
+			    bin_info->reg_interval) - pad_size;
+		} while (try_hdr_size > try_redzone0_offset);
+	} while (try_run_size <= arena_maxclass
+	    && try_run_size <= arena_maxclass
+	    && RUN_MAX_OVRHD * (bin_info->reg_interval << 3) >
+	    RUN_MAX_OVRHD_RELAX
+	    && (try_redzone0_offset << RUN_BFP) > RUN_MAX_OVRHD * try_run_size
+	    && try_nregs < RUN_MAXREGS);
+
+	assert(good_hdr_size <= good_redzone0_offset);
+
+	/* Copy final settings. */
+	bin_info->run_size = good_run_size;
+	bin_info->nregs = good_nregs;
+	bin_info->bitmap_offset = good_bitmap_offset;
+	bin_info->ctx0_offset = good_ctx0_offset;
+	bin_info->reg0_offset = good_redzone0_offset + bin_info->redzone_size;
+
+	assert(bin_info->reg0_offset - bin_info->redzone_size + (bin_info->nregs
+	    * bin_info->reg_interval) + pad_size == bin_info->run_size);
+
+	return (good_run_size);
+}
+
+static void
+bin_info_init(void)
+{
+	arena_bin_info_t *bin_info;
+	size_t prev_run_size = PAGE;
+
+#define	SIZE_CLASS(bin, delta, size)					\
+	bin_info = &arena_bin_info[bin];				\
+	bin_info->reg_size = size;					\
+	prev_run_size = bin_info_run_size_calc(bin_info, prev_run_size);\
+	bitmap_info_init(&bin_info->bitmap_info, bin_info->nregs);
+	SIZE_CLASSES
+#undef SIZE_CLASS
+}
+
+void
+arena_boot(void)
+{
+	size_t header_size;
+	unsigned i;
+
+	/*
+	 * Compute the header size such that it is large enough to contain the
+	 * page map.  The page map is biased to omit entries for the header
+	 * itself, so some iteration is necessary to compute the map bias.
+	 *
+	 * 1) Compute safe header_size and map_bias values that include enough
+	 *    space for an unbiased page map.
+	 * 2) Refine map_bias based on (1) to omit the header pages in the page
+	 *    map.  The resulting map_bias may be one too small.
+	 * 3) Refine map_bias based on (2).  The result will be >= the result
+	 *    from (2), and will always be correct.
+	 */
+	map_bias = 0;
+	for (i = 0; i < 3; i++) {
+		header_size = offsetof(arena_chunk_t, map) +
+		    (sizeof(arena_chunk_map_t) * (chunk_npages-map_bias));
+		map_bias = (header_size >> LG_PAGE) + ((header_size & PAGE_MASK)
+		    != 0);
+	}
+	assert(map_bias > 0);
+
+	arena_maxclass = chunksize - (map_bias << LG_PAGE);
+
+	bin_info_init();
+}
+
+void
+arena_prefork(arena_t *arena)
+{
+	unsigned i;
+
+	malloc_mutex_prefork(&arena->lock);
+	for (i = 0; i < NBINS; i++)
+		malloc_mutex_prefork(&arena->bins[i].lock);
+}
+
+void
+arena_postfork_parent(arena_t *arena)
+{
+	unsigned i;
+
+	for (i = 0; i < NBINS; i++)
+		malloc_mutex_postfork_parent(&arena->bins[i].lock);
+	malloc_mutex_postfork_parent(&arena->lock);
+}
+
+void
+arena_postfork_child(arena_t *arena)
+{
+	unsigned i;
+
+	for (i = 0; i < NBINS; i++)
+		malloc_mutex_postfork_child(&arena->bins[i].lock);
+	malloc_mutex_postfork_child(&arena->lock);
+}
diff --git a/extra/jemalloc/src/atomic.c b/extra/jemalloc/src/atomic.c
new file mode 100644
index 00000000000..77ee313113b
--- /dev/null
+++ b/extra/jemalloc/src/atomic.c
@@ -0,0 +1,2 @@
+#define	JEMALLOC_ATOMIC_C_
+#include "jemalloc/internal/jemalloc_internal.h"
diff --git a/extra/jemalloc/src/base.c b/extra/jemalloc/src/base.c
new file mode 100644
index 00000000000..4e62e8fa918
--- /dev/null
+++ b/extra/jemalloc/src/base.c
@@ -0,0 +1,142 @@
+#define	JEMALLOC_BASE_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Data. */
+
+static malloc_mutex_t	base_mtx;
+
+/*
+ * Current pages that are being used for internal memory allocations.  These
+ * pages are carved up in cacheline-size quanta, so that there is no chance of
+ * false cache line sharing.
+ */
+static void		*base_pages;
+static void		*base_next_addr;
+static void		*base_past_addr; /* Addr immediately past base_pages. */
+static extent_node_t	*base_nodes;
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static bool	base_pages_alloc(size_t minsize);
+
+/******************************************************************************/
+
+static bool
+base_pages_alloc(size_t minsize)
+{
+	size_t csize;
+	bool zero;
+
+	assert(minsize != 0);
+	csize = CHUNK_CEILING(minsize);
+	zero = false;
+	base_pages = chunk_alloc(csize, chunksize, true, &zero,
+	    chunk_dss_prec_get());
+	if (base_pages == NULL)
+		return (true);
+	base_next_addr = base_pages;
+	base_past_addr = (void *)((uintptr_t)base_pages + csize);
+
+	return (false);
+}
+
+void *
+base_alloc(size_t size)
+{
+	void *ret;
+	size_t csize;
+
+	/* Round size up to nearest multiple of the cacheline size. */
+	csize = CACHELINE_CEILING(size);
+
+	malloc_mutex_lock(&base_mtx);
+	/* Make sure there's enough space for the allocation. */
+	if ((uintptr_t)base_next_addr + csize > (uintptr_t)base_past_addr) {
+		if (base_pages_alloc(csize)) {
+			malloc_mutex_unlock(&base_mtx);
+			return (NULL);
+		}
+	}
+	/* Allocate. */
+	ret = base_next_addr;
+	base_next_addr = (void *)((uintptr_t)base_next_addr + csize);
+	malloc_mutex_unlock(&base_mtx);
+	VALGRIND_MAKE_MEM_UNDEFINED(ret, csize);
+
+	return (ret);
+}
+
+void *
+base_calloc(size_t number, size_t size)
+{
+	void *ret = base_alloc(number * size);
+
+	if (ret != NULL)
+		memset(ret, 0, number * size);
+
+	return (ret);
+}
+
+extent_node_t *
+base_node_alloc(void)
+{
+	extent_node_t *ret;
+
+	malloc_mutex_lock(&base_mtx);
+	if (base_nodes != NULL) {
+		ret = base_nodes;
+		base_nodes = *(extent_node_t **)ret;
+		malloc_mutex_unlock(&base_mtx);
+		VALGRIND_MAKE_MEM_UNDEFINED(ret, sizeof(extent_node_t));
+	} else {
+		malloc_mutex_unlock(&base_mtx);
+		ret = (extent_node_t *)base_alloc(sizeof(extent_node_t));
+	}
+
+	return (ret);
+}
+
+void
+base_node_dealloc(extent_node_t *node)
+{
+
+	VALGRIND_MAKE_MEM_UNDEFINED(node, sizeof(extent_node_t));
+	malloc_mutex_lock(&base_mtx);
+	*(extent_node_t **)node = base_nodes;
+	base_nodes = node;
+	malloc_mutex_unlock(&base_mtx);
+}
+
+bool
+base_boot(void)
+{
+
+	base_nodes = NULL;
+	if (malloc_mutex_init(&base_mtx))
+		return (true);
+
+	return (false);
+}
+
+void
+base_prefork(void)
+{
+
+	malloc_mutex_prefork(&base_mtx);
+}
+
+void
+base_postfork_parent(void)
+{
+
+	malloc_mutex_postfork_parent(&base_mtx);
+}
+
+void
+base_postfork_child(void)
+{
+
+	malloc_mutex_postfork_child(&base_mtx);
+}
diff --git a/extra/jemalloc/src/bitmap.c b/extra/jemalloc/src/bitmap.c
new file mode 100644
index 00000000000..b47e2629093
--- /dev/null
+++ b/extra/jemalloc/src/bitmap.c
@@ -0,0 +1,90 @@
+#define JEMALLOC_BITMAP_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static size_t	bits2groups(size_t nbits);
+
+/******************************************************************************/
+
+static size_t
+bits2groups(size_t nbits)
+{
+
+	return ((nbits >> LG_BITMAP_GROUP_NBITS) +
+	    !!(nbits & BITMAP_GROUP_NBITS_MASK));
+}
+
+void
+bitmap_info_init(bitmap_info_t *binfo, size_t nbits)
+{
+	unsigned i;
+	size_t group_count;
+
+	assert(nbits > 0);
+	assert(nbits <= (ZU(1) << LG_BITMAP_MAXBITS));
+
+	/*
+	 * Compute the number of groups necessary to store nbits bits, and
+	 * progressively work upward through the levels until reaching a level
+	 * that requires only one group.
+	 */
+	binfo->levels[0].group_offset = 0;
+	group_count = bits2groups(nbits);
+	for (i = 1; group_count > 1; i++) {
+		assert(i < BITMAP_MAX_LEVELS);
+		binfo->levels[i].group_offset = binfo->levels[i-1].group_offset
+		    + group_count;
+		group_count = bits2groups(group_count);
+	}
+	binfo->levels[i].group_offset = binfo->levels[i-1].group_offset
+	    + group_count;
+	binfo->nlevels = i;
+	binfo->nbits = nbits;
+}
+
+size_t
+bitmap_info_ngroups(const bitmap_info_t *binfo)
+{
+
+	return (binfo->levels[binfo->nlevels].group_offset << LG_SIZEOF_BITMAP);
+}
+
+size_t
+bitmap_size(size_t nbits)
+{
+	bitmap_info_t binfo;
+
+	bitmap_info_init(&binfo, nbits);
+	return (bitmap_info_ngroups(&binfo));
+}
+
+void
+bitmap_init(bitmap_t *bitmap, const bitmap_info_t *binfo)
+{
+	size_t extra;
+	unsigned i;
+
+	/*
+	 * Bits are actually inverted with regard to the external bitmap
+	 * interface, so the bitmap starts out with all 1 bits, except for
+	 * trailing unused bits (if any).  Note that each group uses bit 0 to
+	 * correspond to the first logical bit in the group, so extra bits
+	 * are the most significant bits of the last group.
+	 */
+	memset(bitmap, 0xffU, binfo->levels[binfo->nlevels].group_offset <<
+	    LG_SIZEOF_BITMAP);
+	extra = (BITMAP_GROUP_NBITS - (binfo->nbits & BITMAP_GROUP_NBITS_MASK))
+	    & BITMAP_GROUP_NBITS_MASK;
+	if (extra != 0)
+		bitmap[binfo->levels[1].group_offset - 1] >>= extra;
+	for (i = 1; i < binfo->nlevels; i++) {
+		size_t group_count = binfo->levels[i].group_offset -
+		    binfo->levels[i-1].group_offset;
+		extra = (BITMAP_GROUP_NBITS - (group_count &
+		    BITMAP_GROUP_NBITS_MASK)) & BITMAP_GROUP_NBITS_MASK;
+		if (extra != 0)
+			bitmap[binfo->levels[i+1].group_offset - 1] >>= extra;
+	}
+}
diff --git a/extra/jemalloc/src/chunk.c b/extra/jemalloc/src/chunk.c
new file mode 100644
index 00000000000..044f76be96c
--- /dev/null
+++ b/extra/jemalloc/src/chunk.c
@@ -0,0 +1,385 @@
+#define	JEMALLOC_CHUNK_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Data. */
+
+const char	*opt_dss = DSS_DEFAULT;
+size_t		opt_lg_chunk = LG_CHUNK_DEFAULT;
+
+malloc_mutex_t	chunks_mtx;
+chunk_stats_t	stats_chunks;
+
+/*
+ * Trees of chunks that were previously allocated (trees differ only in node
+ * ordering).  These are used when allocating chunks, in an attempt to re-use
+ * address space.  Depending on function, different tree orderings are needed,
+ * which is why there are two trees with the same contents.
+ */
+static extent_tree_t	chunks_szad_mmap;
+static extent_tree_t	chunks_ad_mmap;
+static extent_tree_t	chunks_szad_dss;
+static extent_tree_t	chunks_ad_dss;
+
+rtree_t		*chunks_rtree;
+
+/* Various chunk-related settings. */
+size_t		chunksize;
+size_t		chunksize_mask; /* (chunksize - 1). */
+size_t		chunk_npages;
+size_t		map_bias;
+size_t		arena_maxclass; /* Max size class for arenas. */
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static void	*chunk_recycle(extent_tree_t *chunks_szad,
+    extent_tree_t *chunks_ad, size_t size, size_t alignment, bool base,
+    bool *zero);
+static void	chunk_record(extent_tree_t *chunks_szad,
+    extent_tree_t *chunks_ad, void *chunk, size_t size);
+
+/******************************************************************************/
+
+static void *
+chunk_recycle(extent_tree_t *chunks_szad, extent_tree_t *chunks_ad, size_t size,
+    size_t alignment, bool base, bool *zero)
+{
+	void *ret;
+	extent_node_t *node;
+	extent_node_t key;
+	size_t alloc_size, leadsize, trailsize;
+	bool zeroed;
+
+	if (base) {
+		/*
+		 * This function may need to call base_node_{,de}alloc(), but
+		 * the current chunk allocation request is on behalf of the
+		 * base allocator.  Avoid deadlock (and if that weren't an
+		 * issue, potential for infinite recursion) by returning NULL.
+		 */
+		return (NULL);
+	}
+
+	alloc_size = size + alignment - chunksize;
+	/* Beware size_t wrap-around. */
+	if (alloc_size < size)
+		return (NULL);
+	key.addr = NULL;
+	key.size = alloc_size;
+	malloc_mutex_lock(&chunks_mtx);
+	node = extent_tree_szad_nsearch(chunks_szad, &key);
+	if (node == NULL) {
+		malloc_mutex_unlock(&chunks_mtx);
+		return (NULL);
+	}
+	leadsize = ALIGNMENT_CEILING((uintptr_t)node->addr, alignment) -
+	    (uintptr_t)node->addr;
+	assert(node->size >= leadsize + size);
+	trailsize = node->size - leadsize - size;
+	ret = (void *)((uintptr_t)node->addr + leadsize);
+	zeroed = node->zeroed;
+	if (zeroed)
+	    *zero = true;
+	/* Remove node from the tree. */
+	extent_tree_szad_remove(chunks_szad, node);
+	extent_tree_ad_remove(chunks_ad, node);
+	if (leadsize != 0) {
+		/* Insert the leading space as a smaller chunk. */
+		node->size = leadsize;
+		extent_tree_szad_insert(chunks_szad, node);
+		extent_tree_ad_insert(chunks_ad, node);
+		node = NULL;
+	}
+	if (trailsize != 0) {
+		/* Insert the trailing space as a smaller chunk. */
+		if (node == NULL) {
+			/*
+			 * An additional node is required, but
+			 * base_node_alloc() can cause a new base chunk to be
+			 * allocated.  Drop chunks_mtx in order to avoid
+			 * deadlock, and if node allocation fails, deallocate
+			 * the result before returning an error.
+			 */
+			malloc_mutex_unlock(&chunks_mtx);
+			node = base_node_alloc();
+			if (node == NULL) {
+				chunk_dealloc(ret, size, true);
+				return (NULL);
+			}
+			malloc_mutex_lock(&chunks_mtx);
+		}
+		node->addr = (void *)((uintptr_t)(ret) + size);
+		node->size = trailsize;
+		node->zeroed = zeroed;
+		extent_tree_szad_insert(chunks_szad, node);
+		extent_tree_ad_insert(chunks_ad, node);
+		node = NULL;
+	}
+	malloc_mutex_unlock(&chunks_mtx);
+
+	if (node != NULL)
+		base_node_dealloc(node);
+	if (*zero) {
+		if (zeroed == false)
+			memset(ret, 0, size);
+		else if (config_debug) {
+			size_t i;
+			size_t *p = (size_t *)(uintptr_t)ret;
+
+			VALGRIND_MAKE_MEM_DEFINED(ret, size);
+			for (i = 0; i < size / sizeof(size_t); i++)
+				assert(p[i] == 0);
+		}
+	}
+	return (ret);
+}
+
+/*
+ * If the caller specifies (*zero == false), it is still possible to receive
+ * zeroed memory, in which case *zero is toggled to true.  arena_chunk_alloc()
+ * takes advantage of this to avoid demanding zeroed chunks, but taking
+ * advantage of them if they are returned.
+ */
+void *
+chunk_alloc(size_t size, size_t alignment, bool base, bool *zero,
+    dss_prec_t dss_prec)
+{
+	void *ret;
+
+	assert(size != 0);
+	assert((size & chunksize_mask) == 0);
+	assert(alignment != 0);
+	assert((alignment & chunksize_mask) == 0);
+
+	/* "primary" dss. */
+	if (config_dss && dss_prec == dss_prec_primary) {
+		if ((ret = chunk_recycle(&chunks_szad_dss, &chunks_ad_dss, size,
+		    alignment, base, zero)) != NULL)
+			goto label_return;
+		if ((ret = chunk_alloc_dss(size, alignment, zero)) != NULL)
+			goto label_return;
+	}
+	/* mmap. */
+	if ((ret = chunk_recycle(&chunks_szad_mmap, &chunks_ad_mmap, size,
+	    alignment, base, zero)) != NULL)
+		goto label_return;
+	if ((ret = chunk_alloc_mmap(size, alignment, zero)) != NULL)
+		goto label_return;
+	/* "secondary" dss. */
+	if (config_dss && dss_prec == dss_prec_secondary) {
+		if ((ret = chunk_recycle(&chunks_szad_dss, &chunks_ad_dss, size,
+		    alignment, base, zero)) != NULL)
+			goto label_return;
+		if ((ret = chunk_alloc_dss(size, alignment, zero)) != NULL)
+			goto label_return;
+	}
+
+	/* All strategies for allocation failed. */
+	ret = NULL;
+label_return:
+	if (ret != NULL) {
+		if (config_ivsalloc && base == false) {
+			if (rtree_set(chunks_rtree, (uintptr_t)ret, ret)) {
+				chunk_dealloc(ret, size, true);
+				return (NULL);
+			}
+		}
+		if (config_stats || config_prof) {
+			bool gdump;
+			malloc_mutex_lock(&chunks_mtx);
+			if (config_stats)
+				stats_chunks.nchunks += (size / chunksize);
+			stats_chunks.curchunks += (size / chunksize);
+			if (stats_chunks.curchunks > stats_chunks.highchunks) {
+				stats_chunks.highchunks =
+				    stats_chunks.curchunks;
+				if (config_prof)
+					gdump = true;
+			} else if (config_prof)
+				gdump = false;
+			malloc_mutex_unlock(&chunks_mtx);
+			if (config_prof && opt_prof && opt_prof_gdump && gdump)
+				prof_gdump();
+		}
+		if (config_valgrind)
+			VALGRIND_MAKE_MEM_UNDEFINED(ret, size);
+	}
+	assert(CHUNK_ADDR2BASE(ret) == ret);
+	return (ret);
+}
+
+static void
+chunk_record(extent_tree_t *chunks_szad, extent_tree_t *chunks_ad, void *chunk,
+    size_t size)
+{
+	bool unzeroed;
+	extent_node_t *xnode, *node, *prev, key;
+
+	unzeroed = pages_purge(chunk, size);
+	VALGRIND_MAKE_MEM_NOACCESS(chunk, size);
+
+	/*
+	 * Allocate a node before acquiring chunks_mtx even though it might not
+	 * be needed, because base_node_alloc() may cause a new base chunk to
+	 * be allocated, which could cause deadlock if chunks_mtx were already
+	 * held.
+	 */
+	xnode = base_node_alloc();
+
+	malloc_mutex_lock(&chunks_mtx);
+	key.addr = (void *)((uintptr_t)chunk + size);
+	node = extent_tree_ad_nsearch(chunks_ad, &key);
+	/* Try to coalesce forward. */
+	if (node != NULL && node->addr == key.addr) {
+		/*
+		 * Coalesce chunk with the following address range.  This does
+		 * not change the position within chunks_ad, so only
+		 * remove/insert from/into chunks_szad.
+		 */
+		extent_tree_szad_remove(chunks_szad, node);
+		node->addr = chunk;
+		node->size += size;
+		node->zeroed = (node->zeroed && (unzeroed == false));
+		extent_tree_szad_insert(chunks_szad, node);
+		if (xnode != NULL)
+			base_node_dealloc(xnode);
+	} else {
+		/* Coalescing forward failed, so insert a new node. */
+		if (xnode == NULL) {
+			/*
+			 * base_node_alloc() failed, which is an exceedingly
+			 * unlikely failure.  Leak chunk; its pages have
+			 * already been purged, so this is only a virtual
+			 * memory leak.
+			 */
+			malloc_mutex_unlock(&chunks_mtx);
+			return;
+		}
+		node = xnode;
+		node->addr = chunk;
+		node->size = size;
+		node->zeroed = (unzeroed == false);
+		extent_tree_ad_insert(chunks_ad, node);
+		extent_tree_szad_insert(chunks_szad, node);
+	}
+
+	/* Try to coalesce backward. */
+	prev = extent_tree_ad_prev(chunks_ad, node);
+	if (prev != NULL && (void *)((uintptr_t)prev->addr + prev->size) ==
+	    chunk) {
+		/*
+		 * Coalesce chunk with the previous address range.  This does
+		 * not change the position within chunks_ad, so only
+		 * remove/insert node from/into chunks_szad.
+		 */
+		extent_tree_szad_remove(chunks_szad, prev);
+		extent_tree_ad_remove(chunks_ad, prev);
+
+		extent_tree_szad_remove(chunks_szad, node);
+		node->addr = prev->addr;
+		node->size += prev->size;
+		node->zeroed = (node->zeroed && prev->zeroed);
+		extent_tree_szad_insert(chunks_szad, node);
+
+		base_node_dealloc(prev);
+	}
+	malloc_mutex_unlock(&chunks_mtx);
+}
+
+void
+chunk_unmap(void *chunk, size_t size)
+{
+	assert(chunk != NULL);
+	assert(CHUNK_ADDR2BASE(chunk) == chunk);
+	assert(size != 0);
+	assert((size & chunksize_mask) == 0);
+
+	if (config_dss && chunk_in_dss(chunk))
+		chunk_record(&chunks_szad_dss, &chunks_ad_dss, chunk, size);
+	else if (chunk_dealloc_mmap(chunk, size))
+		chunk_record(&chunks_szad_mmap, &chunks_ad_mmap, chunk, size);
+}
+
+void
+chunk_dealloc(void *chunk, size_t size, bool unmap)
+{
+
+	assert(chunk != NULL);
+	assert(CHUNK_ADDR2BASE(chunk) == chunk);
+	assert(size != 0);
+	assert((size & chunksize_mask) == 0);
+
+	if (config_ivsalloc)
+		rtree_set(chunks_rtree, (uintptr_t)chunk, NULL);
+	if (config_stats || config_prof) {
+		malloc_mutex_lock(&chunks_mtx);
+		assert(stats_chunks.curchunks >= (size / chunksize));
+		stats_chunks.curchunks -= (size / chunksize);
+		malloc_mutex_unlock(&chunks_mtx);
+	}
+
+	if (unmap)
+		chunk_unmap(chunk, size);
+}
+
+bool
+chunk_boot(void)
+{
+
+	/* Set variables according to the value of opt_lg_chunk. */
+	chunksize = (ZU(1) << opt_lg_chunk);
+	assert(chunksize >= PAGE);
+	chunksize_mask = chunksize - 1;
+	chunk_npages = (chunksize >> LG_PAGE);
+
+	if (config_stats || config_prof) {
+		if (malloc_mutex_init(&chunks_mtx))
+			return (true);
+		memset(&stats_chunks, 0, sizeof(chunk_stats_t));
+	}
+	if (config_dss && chunk_dss_boot())
+		return (true);
+	extent_tree_szad_new(&chunks_szad_mmap);
+	extent_tree_ad_new(&chunks_ad_mmap);
+	extent_tree_szad_new(&chunks_szad_dss);
+	extent_tree_ad_new(&chunks_ad_dss);
+	if (config_ivsalloc) {
+		chunks_rtree = rtree_new((ZU(1) << (LG_SIZEOF_PTR+3)) -
+		    opt_lg_chunk);
+		if (chunks_rtree == NULL)
+			return (true);
+	}
+
+	return (false);
+}
+
+void
+chunk_prefork(void)
+{
+
+	malloc_mutex_lock(&chunks_mtx);
+	if (config_ivsalloc)
+		rtree_prefork(chunks_rtree);
+	chunk_dss_prefork();
+}
+
+void
+chunk_postfork_parent(void)
+{
+
+	chunk_dss_postfork_parent();
+	if (config_ivsalloc)
+		rtree_postfork_parent(chunks_rtree);
+	malloc_mutex_postfork_parent(&chunks_mtx);
+}
+
+void
+chunk_postfork_child(void)
+{
+
+	chunk_dss_postfork_child();
+	if (config_ivsalloc)
+		rtree_postfork_child(chunks_rtree);
+	malloc_mutex_postfork_child(&chunks_mtx);
+}
diff --git a/extra/jemalloc/src/chunk_dss.c b/extra/jemalloc/src/chunk_dss.c
new file mode 100644
index 00000000000..24781cc52dc
--- /dev/null
+++ b/extra/jemalloc/src/chunk_dss.c
@@ -0,0 +1,197 @@
+#define	JEMALLOC_CHUNK_DSS_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+/******************************************************************************/
+/* Data. */
+
+const char	*dss_prec_names[] = {
+	"disabled",
+	"primary",
+	"secondary",
+	"N/A"
+};
+
+/* Current dss precedence default, used when creating new arenas. */
+static dss_prec_t	dss_prec_default = DSS_PREC_DEFAULT;
+
+/*
+ * Protects sbrk() calls.  This avoids malloc races among threads, though it
+ * does not protect against races with threads that call sbrk() directly.
+ */
+static malloc_mutex_t	dss_mtx;
+
+/* Base address of the DSS. */
+static void		*dss_base;
+/* Current end of the DSS, or ((void *)-1) if the DSS is exhausted. */
+static void		*dss_prev;
+/* Current upper limit on DSS addresses. */
+static void		*dss_max;
+
+/******************************************************************************/
+
+#ifndef JEMALLOC_HAVE_SBRK
+static void *
+sbrk(intptr_t increment)
+{
+
+	not_implemented();
+
+	return (NULL);
+}
+#endif
+
+dss_prec_t
+chunk_dss_prec_get(void)
+{
+	dss_prec_t ret;
+
+	if (config_dss == false)
+		return (dss_prec_disabled);
+	malloc_mutex_lock(&dss_mtx);
+	ret = dss_prec_default;
+	malloc_mutex_unlock(&dss_mtx);
+	return (ret);
+}
+
+bool
+chunk_dss_prec_set(dss_prec_t dss_prec)
+{
+
+	if (config_dss == false)
+		return (true);
+	malloc_mutex_lock(&dss_mtx);
+	dss_prec_default = dss_prec;
+	malloc_mutex_unlock(&dss_mtx);
+	return (false);
+}
+
+void *
+chunk_alloc_dss(size_t size, size_t alignment, bool *zero)
+{
+	void *ret;
+
+	cassert(config_dss);
+	assert(size > 0 && (size & chunksize_mask) == 0);
+	assert(alignment > 0 && (alignment & chunksize_mask) == 0);
+
+	/*
+	 * sbrk() uses a signed increment argument, so take care not to
+	 * interpret a huge allocation request as a negative increment.
+	 */
+	if ((intptr_t)size < 0)
+		return (NULL);
+
+	malloc_mutex_lock(&dss_mtx);
+	if (dss_prev != (void *)-1) {
+		size_t gap_size, cpad_size;
+		void *cpad, *dss_next;
+		intptr_t incr;
+
+		/*
+		 * The loop is necessary to recover from races with other
+		 * threads that are using the DSS for something other than
+		 * malloc.
+		 */
+		do {
+			/* Get the current end of the DSS. */
+			dss_max = sbrk(0);
+			/*
+			 * Calculate how much padding is necessary to
+			 * chunk-align the end of the DSS.
+			 */
+			gap_size = (chunksize - CHUNK_ADDR2OFFSET(dss_max)) &
+			    chunksize_mask;
+			/*
+			 * Compute how much chunk-aligned pad space (if any) is
+			 * necessary to satisfy alignment.  This space can be
+			 * recycled for later use.
+			 */
+			cpad = (void *)((uintptr_t)dss_max + gap_size);
+			ret = (void *)ALIGNMENT_CEILING((uintptr_t)dss_max,
+			    alignment);
+			cpad_size = (uintptr_t)ret - (uintptr_t)cpad;
+			dss_next = (void *)((uintptr_t)ret + size);
+			if ((uintptr_t)ret < (uintptr_t)dss_max ||
+			    (uintptr_t)dss_next < (uintptr_t)dss_max) {
+				/* Wrap-around. */
+				malloc_mutex_unlock(&dss_mtx);
+				return (NULL);
+			}
+			incr = gap_size + cpad_size + size;
+			dss_prev = sbrk(incr);
+			if (dss_prev == dss_max) {
+				/* Success. */
+				dss_max = dss_next;
+				malloc_mutex_unlock(&dss_mtx);
+				if (cpad_size != 0)
+					chunk_unmap(cpad, cpad_size);
+				if (*zero) {
+					VALGRIND_MAKE_MEM_UNDEFINED(ret, size);
+					memset(ret, 0, size);
+				}
+				return (ret);
+			}
+		} while (dss_prev != (void *)-1);
+	}
+	malloc_mutex_unlock(&dss_mtx);
+
+	return (NULL);
+}
+
+bool
+chunk_in_dss(void *chunk)
+{
+	bool ret;
+
+	cassert(config_dss);
+
+	malloc_mutex_lock(&dss_mtx);
+	if ((uintptr_t)chunk >= (uintptr_t)dss_base
+	    && (uintptr_t)chunk < (uintptr_t)dss_max)
+		ret = true;
+	else
+		ret = false;
+	malloc_mutex_unlock(&dss_mtx);
+
+	return (ret);
+}
+
+bool
+chunk_dss_boot(void)
+{
+
+	cassert(config_dss);
+
+	if (malloc_mutex_init(&dss_mtx))
+		return (true);
+	dss_base = sbrk(0);
+	dss_prev = dss_base;
+	dss_max = dss_base;
+
+	return (false);
+}
+
+void
+chunk_dss_prefork(void)
+{
+
+	if (config_dss)
+		malloc_mutex_prefork(&dss_mtx);
+}
+
+void
+chunk_dss_postfork_parent(void)
+{
+
+	if (config_dss)
+		malloc_mutex_postfork_parent(&dss_mtx);
+}
+
+void
+chunk_dss_postfork_child(void)
+{
+
+	if (config_dss)
+		malloc_mutex_postfork_child(&dss_mtx);
+}
+
+/******************************************************************************/
diff --git a/extra/jemalloc/src/chunk_mmap.c b/extra/jemalloc/src/chunk_mmap.c
new file mode 100644
index 00000000000..8a42e75915f
--- /dev/null
+++ b/extra/jemalloc/src/chunk_mmap.c
@@ -0,0 +1,210 @@
+#define	JEMALLOC_CHUNK_MMAP_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static void	*pages_map(void *addr, size_t size);
+static void	pages_unmap(void *addr, size_t size);
+static void	*chunk_alloc_mmap_slow(size_t size, size_t alignment,
+    bool *zero);
+
+/******************************************************************************/
+
+static void *
+pages_map(void *addr, size_t size)
+{
+	void *ret;
+
+	assert(size != 0);
+
+#ifdef _WIN32
+	/*
+	 * If VirtualAlloc can't allocate at the given address when one is
+	 * given, it fails and returns NULL.
+	 */
+	ret = VirtualAlloc(addr, size, MEM_COMMIT | MEM_RESERVE,
+	    PAGE_READWRITE);
+#else
+	/*
+	 * We don't use MAP_FIXED here, because it can cause the *replacement*
+	 * of existing mappings, and we only want to create new mappings.
+	 */
+	ret = mmap(addr, size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANON,
+	    -1, 0);
+	assert(ret != NULL);
+
+	if (ret == MAP_FAILED)
+		ret = NULL;
+	else if (addr != NULL && ret != addr) {
+		/*
+		 * We succeeded in mapping memory, but not in the right place.
+		 */
+		if (munmap(ret, size) == -1) {
+			char buf[BUFERROR_BUF];
+
+			buferror(buf, sizeof(buf));
+			malloc_printf("<jemalloc: Error in munmap(): %s\n",
+			    buf);
+			if (opt_abort)
+				abort();
+		}
+		ret = NULL;
+	}
+#endif
+	assert(ret == NULL || (addr == NULL && ret != addr)
+	    || (addr != NULL && ret == addr));
+	return (ret);
+}
+
+static void
+pages_unmap(void *addr, size_t size)
+{
+
+#ifdef _WIN32
+	if (VirtualFree(addr, 0, MEM_RELEASE) == 0)
+#else
+	if (munmap(addr, size) == -1)
+#endif
+	{
+		char buf[BUFERROR_BUF];
+
+		buferror(buf, sizeof(buf));
+		malloc_printf("<jemalloc>: Error in "
+#ifdef _WIN32
+		              "VirtualFree"
+#else
+		              "munmap"
+#endif
+		              "(): %s\n", buf);
+		if (opt_abort)
+			abort();
+	}
+}
+
+static void *
+pages_trim(void *addr, size_t alloc_size, size_t leadsize, size_t size)
+{
+	void *ret = (void *)((uintptr_t)addr + leadsize);
+
+	assert(alloc_size >= leadsize + size);
+#ifdef _WIN32
+	{
+		void *new_addr;
+
+		pages_unmap(addr, alloc_size);
+		new_addr = pages_map(ret, size);
+		if (new_addr == ret)
+			return (ret);
+		if (new_addr)
+			pages_unmap(new_addr, size);
+		return (NULL);
+	}
+#else
+	{
+		size_t trailsize = alloc_size - leadsize - size;
+
+		if (leadsize != 0)
+			pages_unmap(addr, leadsize);
+		if (trailsize != 0)
+			pages_unmap((void *)((uintptr_t)ret + size), trailsize);
+		return (ret);
+	}
+#endif
+}
+
+bool
+pages_purge(void *addr, size_t length)
+{
+	bool unzeroed;
+
+#ifdef _WIN32
+	VirtualAlloc(addr, length, MEM_RESET, PAGE_READWRITE);
+	unzeroed = true;
+#else
+#  ifdef JEMALLOC_PURGE_MADVISE_DONTNEED
+#    define JEMALLOC_MADV_PURGE MADV_DONTNEED
+#    define JEMALLOC_MADV_ZEROS true
+#  elif defined(JEMALLOC_PURGE_MADVISE_FREE)
+#    define JEMALLOC_MADV_PURGE MADV_FREE
+#    define JEMALLOC_MADV_ZEROS false
+#  else
+#    error "No method defined for purging unused dirty pages."
+#  endif
+	int err = madvise(addr, length, JEMALLOC_MADV_PURGE);
+	unzeroed = (JEMALLOC_MADV_ZEROS == false || err != 0);
+#  undef JEMALLOC_MADV_PURGE
+#  undef JEMALLOC_MADV_ZEROS
+#endif
+	return (unzeroed);
+}
+
+static void *
+chunk_alloc_mmap_slow(size_t size, size_t alignment, bool *zero)
+{
+	void *ret, *pages;
+	size_t alloc_size, leadsize;
+
+	alloc_size = size + alignment - PAGE;
+	/* Beware size_t wrap-around. */
+	if (alloc_size < size)
+		return (NULL);
+	do {
+		pages = pages_map(NULL, alloc_size);
+		if (pages == NULL)
+			return (NULL);
+		leadsize = ALIGNMENT_CEILING((uintptr_t)pages, alignment) -
+		    (uintptr_t)pages;
+		ret = pages_trim(pages, alloc_size, leadsize, size);
+	} while (ret == NULL);
+
+	assert(ret != NULL);
+	*zero = true;
+	return (ret);
+}
+
+void *
+chunk_alloc_mmap(size_t size, size_t alignment, bool *zero)
+{
+	void *ret;
+	size_t offset;
+
+	/*
+	 * Ideally, there would be a way to specify alignment to mmap() (like
+	 * NetBSD has), but in the absence of such a feature, we have to work
+	 * hard to efficiently create aligned mappings.  The reliable, but
+	 * slow method is to create a mapping that is over-sized, then trim the
+	 * excess.  However, that always results in one or two calls to
+	 * pages_unmap().
+	 *
+	 * Optimistically try mapping precisely the right amount before falling
+	 * back to the slow method, with the expectation that the optimistic
+	 * approach works most of the time.
+	 */
+
+	assert(alignment != 0);
+	assert((alignment & chunksize_mask) == 0);
+
+	ret = pages_map(NULL, size);
+	if (ret == NULL)
+		return (NULL);
+	offset = ALIGNMENT_ADDR2OFFSET(ret, alignment);
+	if (offset != 0) {
+		pages_unmap(ret, size);
+		return (chunk_alloc_mmap_slow(size, alignment, zero));
+	}
+
+	assert(ret != NULL);
+	*zero = true;
+	return (ret);
+}
+
+bool
+chunk_dealloc_mmap(void *chunk, size_t size)
+{
+
+	if (config_munmap)
+		pages_unmap(chunk, size);
+
+	return (config_munmap == false);
+}
diff --git a/extra/jemalloc/src/ckh.c b/extra/jemalloc/src/ckh.c
new file mode 100644
index 00000000000..2f38348bb85
--- /dev/null
+++ b/extra/jemalloc/src/ckh.c
@@ -0,0 +1,563 @@
+/*
+ *******************************************************************************
+ * Implementation of (2^1+,2) cuckoo hashing, where 2^1+ indicates that each
+ * hash bucket contains 2^n cells, for n >= 1, and 2 indicates that two hash
+ * functions are employed.  The original cuckoo hashing algorithm was described
+ * in:
+ *
+ *   Pagh, R., F.F. Rodler (2004) Cuckoo Hashing.  Journal of Algorithms
+ *     51(2):122-144.
+ *
+ * Generalization of cuckoo hashing was discussed in:
+ *
+ *   Erlingsson, U., M. Manasse, F. McSherry (2006) A cool and practical
+ *     alternative to traditional hash tables.  In Proceedings of the 7th
+ *     Workshop on Distributed Data and Structures (WDAS'06), Santa Clara, CA,
+ *     January 2006.
+ *
+ * This implementation uses precisely two hash functions because that is the
+ * fewest that can work, and supporting multiple hashes is an implementation
+ * burden.  Here is a reproduction of Figure 1 from Erlingsson et al. (2006)
+ * that shows approximate expected maximum load factors for various
+ * configurations:
+ *
+ *           |         #cells/bucket         |
+ *   #hashes |   1   |   2   |   4   |   8   |
+ *   --------+-------+-------+-------+-------+
+ *         1 | 0.006 | 0.006 | 0.03  | 0.12  |
+ *         2 | 0.49  | 0.86  |>0.93< |>0.96< |
+ *         3 | 0.91  | 0.97  | 0.98  | 0.999 |
+ *         4 | 0.97  | 0.99  | 0.999 |       |
+ *
+ * The number of cells per bucket is chosen such that a bucket fits in one cache
+ * line.  So, on 32- and 64-bit systems, we use (8,2) and (4,2) cuckoo hashing,
+ * respectively.
+ *
+ ******************************************************************************/
+#define	JEMALLOC_CKH_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static bool	ckh_grow(ckh_t *ckh);
+static void	ckh_shrink(ckh_t *ckh);
+
+/******************************************************************************/
+
+/*
+ * Search bucket for key and return the cell number if found; SIZE_T_MAX
+ * otherwise.
+ */
+JEMALLOC_INLINE size_t
+ckh_bucket_search(ckh_t *ckh, size_t bucket, const void *key)
+{
+	ckhc_t *cell;
+	unsigned i;
+
+	for (i = 0; i < (ZU(1) << LG_CKH_BUCKET_CELLS); i++) {
+		cell = &ckh->tab[(bucket << LG_CKH_BUCKET_CELLS) + i];
+		if (cell->key != NULL && ckh->keycomp(key, cell->key))
+			return ((bucket << LG_CKH_BUCKET_CELLS) + i);
+	}
+
+	return (SIZE_T_MAX);
+}
+
+/*
+ * Search table for key and return cell number if found; SIZE_T_MAX otherwise.
+ */
+JEMALLOC_INLINE size_t
+ckh_isearch(ckh_t *ckh, const void *key)
+{
+	size_t hashes[2], bucket, cell;
+
+	assert(ckh != NULL);
+
+	ckh->hash(key, hashes);
+
+	/* Search primary bucket. */
+	bucket = hashes[0] & ((ZU(1) << ckh->lg_curbuckets) - 1);
+	cell = ckh_bucket_search(ckh, bucket, key);
+	if (cell != SIZE_T_MAX)
+		return (cell);
+
+	/* Search secondary bucket. */
+	bucket = hashes[1] & ((ZU(1) << ckh->lg_curbuckets) - 1);
+	cell = ckh_bucket_search(ckh, bucket, key);
+	return (cell);
+}
+
+JEMALLOC_INLINE bool
+ckh_try_bucket_insert(ckh_t *ckh, size_t bucket, const void *key,
+    const void *data)
+{
+	ckhc_t *cell;
+	unsigned offset, i;
+
+	/*
+	 * Cycle through the cells in the bucket, starting at a random position.
+	 * The randomness avoids worst-case search overhead as buckets fill up.
+	 */
+	prng32(offset, LG_CKH_BUCKET_CELLS, ckh->prng_state, CKH_A, CKH_C);
+	for (i = 0; i < (ZU(1) << LG_CKH_BUCKET_CELLS); i++) {
+		cell = &ckh->tab[(bucket << LG_CKH_BUCKET_CELLS) +
+		    ((i + offset) & ((ZU(1) << LG_CKH_BUCKET_CELLS) - 1))];
+		if (cell->key == NULL) {
+			cell->key = key;
+			cell->data = data;
+			ckh->count++;
+			return (false);
+		}
+	}
+
+	return (true);
+}
+
+/*
+ * No space is available in bucket.  Randomly evict an item, then try to find an
+ * alternate location for that item.  Iteratively repeat this
+ * eviction/relocation procedure until either success or detection of an
+ * eviction/relocation bucket cycle.
+ */
+JEMALLOC_INLINE bool
+ckh_evict_reloc_insert(ckh_t *ckh, size_t argbucket, void const **argkey,
+    void const **argdata)
+{
+	const void *key, *data, *tkey, *tdata;
+	ckhc_t *cell;
+	size_t hashes[2], bucket, tbucket;
+	unsigned i;
+
+	bucket = argbucket;
+	key = *argkey;
+	data = *argdata;
+	while (true) {
+		/*
+		 * Choose a random item within the bucket to evict.  This is
+		 * critical to correct function, because without (eventually)
+		 * evicting all items within a bucket during iteration, it
+		 * would be possible to get stuck in an infinite loop if there
+		 * were an item for which both hashes indicated the same
+		 * bucket.
+		 */
+		prng32(i, LG_CKH_BUCKET_CELLS, ckh->prng_state, CKH_A, CKH_C);
+		cell = &ckh->tab[(bucket << LG_CKH_BUCKET_CELLS) + i];
+		assert(cell->key != NULL);
+
+		/* Swap cell->{key,data} and {key,data} (evict). */
+		tkey = cell->key; tdata = cell->data;
+		cell->key = key; cell->data = data;
+		key = tkey; data = tdata;
+
+#ifdef CKH_COUNT
+		ckh->nrelocs++;
+#endif
+
+		/* Find the alternate bucket for the evicted item. */
+		ckh->hash(key, hashes);
+		tbucket = hashes[1] & ((ZU(1) << ckh->lg_curbuckets) - 1);
+		if (tbucket == bucket) {
+			tbucket = hashes[0] & ((ZU(1) << ckh->lg_curbuckets)
+			    - 1);
+			/*
+			 * It may be that (tbucket == bucket) still, if the
+			 * item's hashes both indicate this bucket.  However,
+			 * we are guaranteed to eventually escape this bucket
+			 * during iteration, assuming pseudo-random item
+			 * selection (true randomness would make infinite
+			 * looping a remote possibility).  The reason we can
+			 * never get trapped forever is that there are two
+			 * cases:
+			 *
+			 * 1) This bucket == argbucket, so we will quickly
+			 *    detect an eviction cycle and terminate.
+			 * 2) An item was evicted to this bucket from another,
+			 *    which means that at least one item in this bucket
+			 *    has hashes that indicate distinct buckets.
+			 */
+		}
+		/* Check for a cycle. */
+		if (tbucket == argbucket) {
+			*argkey = key;
+			*argdata = data;
+			return (true);
+		}
+
+		bucket = tbucket;
+		if (ckh_try_bucket_insert(ckh, bucket, key, data) == false)
+			return (false);
+	}
+}
+
+JEMALLOC_INLINE bool
+ckh_try_insert(ckh_t *ckh, void const**argkey, void const**argdata)
+{
+	size_t hashes[2], bucket;
+	const void *key = *argkey;
+	const void *data = *argdata;
+
+	ckh->hash(key, hashes);
+
+	/* Try to insert in primary bucket. */
+	bucket = hashes[0] & ((ZU(1) << ckh->lg_curbuckets) - 1);
+	if (ckh_try_bucket_insert(ckh, bucket, key, data) == false)
+		return (false);
+
+	/* Try to insert in secondary bucket. */
+	bucket = hashes[1] & ((ZU(1) << ckh->lg_curbuckets) - 1);
+	if (ckh_try_bucket_insert(ckh, bucket, key, data) == false)
+		return (false);
+
+	/*
+	 * Try to find a place for this item via iterative eviction/relocation.
+	 */
+	return (ckh_evict_reloc_insert(ckh, bucket, argkey, argdata));
+}
+
+/*
+ * Try to rebuild the hash table from scratch by inserting all items from the
+ * old table into the new.
+ */
+JEMALLOC_INLINE bool
+ckh_rebuild(ckh_t *ckh, ckhc_t *aTab)
+{
+	size_t count, i, nins;
+	const void *key, *data;
+
+	count = ckh->count;
+	ckh->count = 0;
+	for (i = nins = 0; nins < count; i++) {
+		if (aTab[i].key != NULL) {
+			key = aTab[i].key;
+			data = aTab[i].data;
+			if (ckh_try_insert(ckh, &key, &data)) {
+				ckh->count = count;
+				return (true);
+			}
+			nins++;
+		}
+	}
+
+	return (false);
+}
+
+static bool
+ckh_grow(ckh_t *ckh)
+{
+	bool ret;
+	ckhc_t *tab, *ttab;
+	size_t lg_curcells;
+	unsigned lg_prevbuckets;
+
+#ifdef CKH_COUNT
+	ckh->ngrows++;
+#endif
+
+	/*
+	 * It is possible (though unlikely, given well behaved hashes) that the
+	 * table will have to be doubled more than once in order to create a
+	 * usable table.
+	 */
+	lg_prevbuckets = ckh->lg_curbuckets;
+	lg_curcells = ckh->lg_curbuckets + LG_CKH_BUCKET_CELLS;
+	while (true) {
+		size_t usize;
+
+		lg_curcells++;
+		usize = sa2u(sizeof(ckhc_t) << lg_curcells, CACHELINE);
+		if (usize == 0) {
+			ret = true;
+			goto label_return;
+		}
+		tab = (ckhc_t *)ipalloc(usize, CACHELINE, true);
+		if (tab == NULL) {
+			ret = true;
+			goto label_return;
+		}
+		/* Swap in new table. */
+		ttab = ckh->tab;
+		ckh->tab = tab;
+		tab = ttab;
+		ckh->lg_curbuckets = lg_curcells - LG_CKH_BUCKET_CELLS;
+
+		if (ckh_rebuild(ckh, tab) == false) {
+			idalloc(tab);
+			break;
+		}
+
+		/* Rebuilding failed, so back out partially rebuilt table. */
+		idalloc(ckh->tab);
+		ckh->tab = tab;
+		ckh->lg_curbuckets = lg_prevbuckets;
+	}
+
+	ret = false;
+label_return:
+	return (ret);
+}
+
+static void
+ckh_shrink(ckh_t *ckh)
+{
+	ckhc_t *tab, *ttab;
+	size_t lg_curcells, usize;
+	unsigned lg_prevbuckets;
+
+	/*
+	 * It is possible (though unlikely, given well behaved hashes) that the
+	 * table rebuild will fail.
+	 */
+	lg_prevbuckets = ckh->lg_curbuckets;
+	lg_curcells = ckh->lg_curbuckets + LG_CKH_BUCKET_CELLS - 1;
+	usize = sa2u(sizeof(ckhc_t) << lg_curcells, CACHELINE);
+	if (usize == 0)
+		return;
+	tab = (ckhc_t *)ipalloc(usize, CACHELINE, true);
+	if (tab == NULL) {
+		/*
+		 * An OOM error isn't worth propagating, since it doesn't
+		 * prevent this or future operations from proceeding.
+		 */
+		return;
+	}
+	/* Swap in new table. */
+	ttab = ckh->tab;
+	ckh->tab = tab;
+	tab = ttab;
+	ckh->lg_curbuckets = lg_curcells - LG_CKH_BUCKET_CELLS;
+
+	if (ckh_rebuild(ckh, tab) == false) {
+		idalloc(tab);
+#ifdef CKH_COUNT
+		ckh->nshrinks++;
+#endif
+		return;
+	}
+
+	/* Rebuilding failed, so back out partially rebuilt table. */
+	idalloc(ckh->tab);
+	ckh->tab = tab;
+	ckh->lg_curbuckets = lg_prevbuckets;
+#ifdef CKH_COUNT
+	ckh->nshrinkfails++;
+#endif
+}
+
+bool
+ckh_new(ckh_t *ckh, size_t minitems, ckh_hash_t *hash, ckh_keycomp_t *keycomp)
+{
+	bool ret;
+	size_t mincells, usize;
+	unsigned lg_mincells;
+
+	assert(minitems > 0);
+	assert(hash != NULL);
+	assert(keycomp != NULL);
+
+#ifdef CKH_COUNT
+	ckh->ngrows = 0;
+	ckh->nshrinks = 0;
+	ckh->nshrinkfails = 0;
+	ckh->ninserts = 0;
+	ckh->nrelocs = 0;
+#endif
+	ckh->prng_state = 42; /* Value doesn't really matter. */
+	ckh->count = 0;
+
+	/*
+	 * Find the minimum power of 2 that is large enough to fit aBaseCount
+	 * entries.  We are using (2+,2) cuckoo hashing, which has an expected
+	 * maximum load factor of at least ~0.86, so 0.75 is a conservative load
+	 * factor that will typically allow 2^aLgMinItems to fit without ever
+	 * growing the table.
+	 */
+	assert(LG_CKH_BUCKET_CELLS > 0);
+	mincells = ((minitems + (3 - (minitems % 3))) / 3) << 2;
+	for (lg_mincells = LG_CKH_BUCKET_CELLS;
+	    (ZU(1) << lg_mincells) < mincells;
+	    lg_mincells++)
+		; /* Do nothing. */
+	ckh->lg_minbuckets = lg_mincells - LG_CKH_BUCKET_CELLS;
+	ckh->lg_curbuckets = lg_mincells - LG_CKH_BUCKET_CELLS;
+	ckh->hash = hash;
+	ckh->keycomp = keycomp;
+
+	usize = sa2u(sizeof(ckhc_t) << lg_mincells, CACHELINE);
+	if (usize == 0) {
+		ret = true;
+		goto label_return;
+	}
+	ckh->tab = (ckhc_t *)ipalloc(usize, CACHELINE, true);
+	if (ckh->tab == NULL) {
+		ret = true;
+		goto label_return;
+	}
+
+	ret = false;
+label_return:
+	return (ret);
+}
+
+void
+ckh_delete(ckh_t *ckh)
+{
+
+	assert(ckh != NULL);
+
+#ifdef CKH_VERBOSE
+	malloc_printf(
+	    "%s(%p): ngrows: %"PRIu64", nshrinks: %"PRIu64","
+	    " nshrinkfails: %"PRIu64", ninserts: %"PRIu64","
+	    " nrelocs: %"PRIu64"\n", __func__, ckh,
+	    (unsigned long long)ckh->ngrows,
+	    (unsigned long long)ckh->nshrinks,
+	    (unsigned long long)ckh->nshrinkfails,
+	    (unsigned long long)ckh->ninserts,
+	    (unsigned long long)ckh->nrelocs);
+#endif
+
+	idalloc(ckh->tab);
+	if (config_debug)
+		memset(ckh, 0x5a, sizeof(ckh_t));
+}
+
+size_t
+ckh_count(ckh_t *ckh)
+{
+
+	assert(ckh != NULL);
+
+	return (ckh->count);
+}
+
+bool
+ckh_iter(ckh_t *ckh, size_t *tabind, void **key, void **data)
+{
+	size_t i, ncells;
+
+	for (i = *tabind, ncells = (ZU(1) << (ckh->lg_curbuckets +
+	    LG_CKH_BUCKET_CELLS)); i < ncells; i++) {
+		if (ckh->tab[i].key != NULL) {
+			if (key != NULL)
+				*key = (void *)ckh->tab[i].key;
+			if (data != NULL)
+				*data = (void *)ckh->tab[i].data;
+			*tabind = i + 1;
+			return (false);
+		}
+	}
+
+	return (true);
+}
+
+bool
+ckh_insert(ckh_t *ckh, const void *key, const void *data)
+{
+	bool ret;
+
+	assert(ckh != NULL);
+	assert(ckh_search(ckh, key, NULL, NULL));
+
+#ifdef CKH_COUNT
+	ckh->ninserts++;
+#endif
+
+	while (ckh_try_insert(ckh, &key, &data)) {
+		if (ckh_grow(ckh)) {
+			ret = true;
+			goto label_return;
+		}
+	}
+
+	ret = false;
+label_return:
+	return (ret);
+}
+
+bool
+ckh_remove(ckh_t *ckh, const void *searchkey, void **key, void **data)
+{
+	size_t cell;
+
+	assert(ckh != NULL);
+
+	cell = ckh_isearch(ckh, searchkey);
+	if (cell != SIZE_T_MAX) {
+		if (key != NULL)
+			*key = (void *)ckh->tab[cell].key;
+		if (data != NULL)
+			*data = (void *)ckh->tab[cell].data;
+		ckh->tab[cell].key = NULL;
+		ckh->tab[cell].data = NULL; /* Not necessary. */
+
+		ckh->count--;
+		/* Try to halve the table if it is less than 1/4 full. */
+		if (ckh->count < (ZU(1) << (ckh->lg_curbuckets
+		    + LG_CKH_BUCKET_CELLS - 2)) && ckh->lg_curbuckets
+		    > ckh->lg_minbuckets) {
+			/* Ignore error due to OOM. */
+			ckh_shrink(ckh);
+		}
+
+		return (false);
+	}
+
+	return (true);
+}
+
+bool
+ckh_search(ckh_t *ckh, const void *searchkey, void **key, void **data)
+{
+	size_t cell;
+
+	assert(ckh != NULL);
+
+	cell = ckh_isearch(ckh, searchkey);
+	if (cell != SIZE_T_MAX) {
+		if (key != NULL)
+			*key = (void *)ckh->tab[cell].key;
+		if (data != NULL)
+			*data = (void *)ckh->tab[cell].data;
+		return (false);
+	}
+
+	return (true);
+}
+
+void
+ckh_string_hash(const void *key, size_t r_hash[2])
+{
+
+	hash(key, strlen((const char *)key), 0x94122f33U, r_hash);
+}
+
+bool
+ckh_string_keycomp(const void *k1, const void *k2)
+{
+
+    assert(k1 != NULL);
+    assert(k2 != NULL);
+
+    return (strcmp((char *)k1, (char *)k2) ? false : true);
+}
+
+void
+ckh_pointer_hash(const void *key, size_t r_hash[2])
+{
+	union {
+		const void	*v;
+		size_t		i;
+	} u;
+
+	assert(sizeof(u.v) == sizeof(u.i));
+	u.v = key;
+	hash(&u.i, sizeof(u.i), 0xd983396eU, r_hash);
+}
+
+bool
+ckh_pointer_keycomp(const void *k1, const void *k2)
+{
+
+	return ((k1 == k2) ? true : false);
+}
diff --git a/extra/jemalloc/src/ctl.c b/extra/jemalloc/src/ctl.c
new file mode 100644
index 00000000000..f2ef4e60611
--- /dev/null
+++ b/extra/jemalloc/src/ctl.c
@@ -0,0 +1,1673 @@
+#define	JEMALLOC_CTL_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Data. */
+
+/*
+ * ctl_mtx protects the following:
+ * - ctl_stats.*
+ * - opt_prof_active
+ */
+static malloc_mutex_t	ctl_mtx;
+static bool		ctl_initialized;
+static uint64_t		ctl_epoch;
+static ctl_stats_t	ctl_stats;
+
+/******************************************************************************/
+/* Helpers for named and indexed nodes. */
+
+static inline const ctl_named_node_t *
+ctl_named_node(const ctl_node_t *node)
+{
+
+	return ((node->named) ? (const ctl_named_node_t *)node : NULL);
+}
+
+static inline const ctl_named_node_t *
+ctl_named_children(const ctl_named_node_t *node, int index)
+{
+	const ctl_named_node_t *children = ctl_named_node(node->children);
+
+	return (children ? &children[index] : NULL);
+}
+
+static inline const ctl_indexed_node_t *
+ctl_indexed_node(const ctl_node_t *node)
+{
+
+	return ((node->named == false) ? (const ctl_indexed_node_t *)node :
+	    NULL);
+}
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+#define	CTL_PROTO(n)							\
+static int	n##_ctl(const size_t *mib, size_t miblen, void *oldp,	\
+    size_t *oldlenp, void *newp, size_t newlen);
+
+#define	INDEX_PROTO(n)							\
+static const ctl_named_node_t	*n##_index(const size_t *mib,		\
+    size_t miblen, size_t i);
+
+static bool	ctl_arena_init(ctl_arena_stats_t *astats);
+static void	ctl_arena_clear(ctl_arena_stats_t *astats);
+static void	ctl_arena_stats_amerge(ctl_arena_stats_t *cstats,
+    arena_t *arena);
+static void	ctl_arena_stats_smerge(ctl_arena_stats_t *sstats,
+    ctl_arena_stats_t *astats);
+static void	ctl_arena_refresh(arena_t *arena, unsigned i);
+static bool	ctl_grow(void);
+static void	ctl_refresh(void);
+static bool	ctl_init(void);
+static int	ctl_lookup(const char *name, ctl_node_t const **nodesp,
+    size_t *mibp, size_t *depthp);
+
+CTL_PROTO(version)
+CTL_PROTO(epoch)
+CTL_PROTO(thread_tcache_enabled)
+CTL_PROTO(thread_tcache_flush)
+CTL_PROTO(thread_arena)
+CTL_PROTO(thread_allocated)
+CTL_PROTO(thread_allocatedp)
+CTL_PROTO(thread_deallocated)
+CTL_PROTO(thread_deallocatedp)
+CTL_PROTO(config_debug)
+CTL_PROTO(config_dss)
+CTL_PROTO(config_fill)
+CTL_PROTO(config_lazy_lock)
+CTL_PROTO(config_mremap)
+CTL_PROTO(config_munmap)
+CTL_PROTO(config_prof)
+CTL_PROTO(config_prof_libgcc)
+CTL_PROTO(config_prof_libunwind)
+CTL_PROTO(config_stats)
+CTL_PROTO(config_tcache)
+CTL_PROTO(config_tls)
+CTL_PROTO(config_utrace)
+CTL_PROTO(config_valgrind)
+CTL_PROTO(config_xmalloc)
+CTL_PROTO(opt_abort)
+CTL_PROTO(opt_dss)
+CTL_PROTO(opt_lg_chunk)
+CTL_PROTO(opt_narenas)
+CTL_PROTO(opt_lg_dirty_mult)
+CTL_PROTO(opt_stats_print)
+CTL_PROTO(opt_junk)
+CTL_PROTO(opt_zero)
+CTL_PROTO(opt_quarantine)
+CTL_PROTO(opt_redzone)
+CTL_PROTO(opt_utrace)
+CTL_PROTO(opt_valgrind)
+CTL_PROTO(opt_xmalloc)
+CTL_PROTO(opt_tcache)
+CTL_PROTO(opt_lg_tcache_max)
+CTL_PROTO(opt_prof)
+CTL_PROTO(opt_prof_prefix)
+CTL_PROTO(opt_prof_active)
+CTL_PROTO(opt_lg_prof_sample)
+CTL_PROTO(opt_lg_prof_interval)
+CTL_PROTO(opt_prof_gdump)
+CTL_PROTO(opt_prof_final)
+CTL_PROTO(opt_prof_leak)
+CTL_PROTO(opt_prof_accum)
+CTL_PROTO(arena_i_purge)
+static void	arena_purge(unsigned arena_ind);
+CTL_PROTO(arena_i_dss)
+INDEX_PROTO(arena_i)
+CTL_PROTO(arenas_bin_i_size)
+CTL_PROTO(arenas_bin_i_nregs)
+CTL_PROTO(arenas_bin_i_run_size)
+INDEX_PROTO(arenas_bin_i)
+CTL_PROTO(arenas_lrun_i_size)
+INDEX_PROTO(arenas_lrun_i)
+CTL_PROTO(arenas_narenas)
+CTL_PROTO(arenas_initialized)
+CTL_PROTO(arenas_quantum)
+CTL_PROTO(arenas_page)
+CTL_PROTO(arenas_tcache_max)
+CTL_PROTO(arenas_nbins)
+CTL_PROTO(arenas_nhbins)
+CTL_PROTO(arenas_nlruns)
+CTL_PROTO(arenas_purge)
+CTL_PROTO(arenas_extend)
+CTL_PROTO(prof_active)
+CTL_PROTO(prof_dump)
+CTL_PROTO(prof_interval)
+CTL_PROTO(stats_chunks_current)
+CTL_PROTO(stats_chunks_total)
+CTL_PROTO(stats_chunks_high)
+CTL_PROTO(stats_huge_allocated)
+CTL_PROTO(stats_huge_nmalloc)
+CTL_PROTO(stats_huge_ndalloc)
+CTL_PROTO(stats_arenas_i_small_allocated)
+CTL_PROTO(stats_arenas_i_small_nmalloc)
+CTL_PROTO(stats_arenas_i_small_ndalloc)
+CTL_PROTO(stats_arenas_i_small_nrequests)
+CTL_PROTO(stats_arenas_i_large_allocated)
+CTL_PROTO(stats_arenas_i_large_nmalloc)
+CTL_PROTO(stats_arenas_i_large_ndalloc)
+CTL_PROTO(stats_arenas_i_large_nrequests)
+CTL_PROTO(stats_arenas_i_bins_j_allocated)
+CTL_PROTO(stats_arenas_i_bins_j_nmalloc)
+CTL_PROTO(stats_arenas_i_bins_j_ndalloc)
+CTL_PROTO(stats_arenas_i_bins_j_nrequests)
+CTL_PROTO(stats_arenas_i_bins_j_nfills)
+CTL_PROTO(stats_arenas_i_bins_j_nflushes)
+CTL_PROTO(stats_arenas_i_bins_j_nruns)
+CTL_PROTO(stats_arenas_i_bins_j_nreruns)
+CTL_PROTO(stats_arenas_i_bins_j_curruns)
+INDEX_PROTO(stats_arenas_i_bins_j)
+CTL_PROTO(stats_arenas_i_lruns_j_nmalloc)
+CTL_PROTO(stats_arenas_i_lruns_j_ndalloc)
+CTL_PROTO(stats_arenas_i_lruns_j_nrequests)
+CTL_PROTO(stats_arenas_i_lruns_j_curruns)
+INDEX_PROTO(stats_arenas_i_lruns_j)
+CTL_PROTO(stats_arenas_i_nthreads)
+CTL_PROTO(stats_arenas_i_dss)
+CTL_PROTO(stats_arenas_i_pactive)
+CTL_PROTO(stats_arenas_i_pdirty)
+CTL_PROTO(stats_arenas_i_mapped)
+CTL_PROTO(stats_arenas_i_npurge)
+CTL_PROTO(stats_arenas_i_nmadvise)
+CTL_PROTO(stats_arenas_i_purged)
+INDEX_PROTO(stats_arenas_i)
+CTL_PROTO(stats_cactive)
+CTL_PROTO(stats_allocated)
+CTL_PROTO(stats_active)
+CTL_PROTO(stats_mapped)
+
+/******************************************************************************/
+/* mallctl tree. */
+
+/* Maximum tree depth. */
+#define	CTL_MAX_DEPTH	6
+
+#define	NAME(n)	{true},	n
+#define	CHILD(t, c)							\
+	sizeof(c##_node) / sizeof(ctl_##t##_node_t),			\
+	(ctl_node_t *)c##_node,						\
+	NULL
+#define	CTL(c)	0, NULL, c##_ctl
+
+/*
+ * Only handles internal indexed nodes, since there are currently no external
+ * ones.
+ */
+#define	INDEX(i)	{false},	i##_index
+
+static const ctl_named_node_t	tcache_node[] = {
+	{NAME("enabled"),	CTL(thread_tcache_enabled)},
+	{NAME("flush"),		CTL(thread_tcache_flush)}
+};
+
+static const ctl_named_node_t	thread_node[] = {
+	{NAME("arena"),		CTL(thread_arena)},
+	{NAME("allocated"),	CTL(thread_allocated)},
+	{NAME("allocatedp"),	CTL(thread_allocatedp)},
+	{NAME("deallocated"),	CTL(thread_deallocated)},
+	{NAME("deallocatedp"),	CTL(thread_deallocatedp)},
+	{NAME("tcache"),	CHILD(named, tcache)}
+};
+
+static const ctl_named_node_t	config_node[] = {
+	{NAME("debug"),			CTL(config_debug)},
+	{NAME("dss"),			CTL(config_dss)},
+	{NAME("fill"),			CTL(config_fill)},
+	{NAME("lazy_lock"),		CTL(config_lazy_lock)},
+	{NAME("mremap"),		CTL(config_mremap)},
+	{NAME("munmap"),		CTL(config_munmap)},
+	{NAME("prof"),			CTL(config_prof)},
+	{NAME("prof_libgcc"),		CTL(config_prof_libgcc)},
+	{NAME("prof_libunwind"),	CTL(config_prof_libunwind)},
+	{NAME("stats"),			CTL(config_stats)},
+	{NAME("tcache"),		CTL(config_tcache)},
+	{NAME("tls"),			CTL(config_tls)},
+	{NAME("utrace"),		CTL(config_utrace)},
+	{NAME("valgrind"),		CTL(config_valgrind)},
+	{NAME("xmalloc"),		CTL(config_xmalloc)}
+};
+
+static const ctl_named_node_t opt_node[] = {
+	{NAME("abort"),			CTL(opt_abort)},
+	{NAME("dss"),			CTL(opt_dss)},
+	{NAME("lg_chunk"),		CTL(opt_lg_chunk)},
+	{NAME("narenas"),		CTL(opt_narenas)},
+	{NAME("lg_dirty_mult"),		CTL(opt_lg_dirty_mult)},
+	{NAME("stats_print"),		CTL(opt_stats_print)},
+	{NAME("junk"),			CTL(opt_junk)},
+	{NAME("zero"),			CTL(opt_zero)},
+	{NAME("quarantine"),		CTL(opt_quarantine)},
+	{NAME("redzone"),		CTL(opt_redzone)},
+	{NAME("utrace"),		CTL(opt_utrace)},
+	{NAME("valgrind"),		CTL(opt_valgrind)},
+	{NAME("xmalloc"),		CTL(opt_xmalloc)},
+	{NAME("tcache"),		CTL(opt_tcache)},
+	{NAME("lg_tcache_max"),		CTL(opt_lg_tcache_max)},
+	{NAME("prof"),			CTL(opt_prof)},
+	{NAME("prof_prefix"),		CTL(opt_prof_prefix)},
+	{NAME("prof_active"),		CTL(opt_prof_active)},
+	{NAME("lg_prof_sample"),	CTL(opt_lg_prof_sample)},
+	{NAME("lg_prof_interval"),	CTL(opt_lg_prof_interval)},
+	{NAME("prof_gdump"),		CTL(opt_prof_gdump)},
+	{NAME("prof_final"),		CTL(opt_prof_final)},
+	{NAME("prof_leak"),		CTL(opt_prof_leak)},
+	{NAME("prof_accum"),		CTL(opt_prof_accum)}
+};
+
+static const ctl_named_node_t arena_i_node[] = {
+	{NAME("purge"),			CTL(arena_i_purge)},
+	{NAME("dss"),			CTL(arena_i_dss)}
+};
+static const ctl_named_node_t super_arena_i_node[] = {
+	{NAME(""),			CHILD(named, arena_i)}
+};
+
+static const ctl_indexed_node_t arena_node[] = {
+	{INDEX(arena_i)}
+};
+
+static const ctl_named_node_t arenas_bin_i_node[] = {
+	{NAME("size"),			CTL(arenas_bin_i_size)},
+	{NAME("nregs"),			CTL(arenas_bin_i_nregs)},
+	{NAME("run_size"),		CTL(arenas_bin_i_run_size)}
+};
+static const ctl_named_node_t super_arenas_bin_i_node[] = {
+	{NAME(""),			CHILD(named, arenas_bin_i)}
+};
+
+static const ctl_indexed_node_t arenas_bin_node[] = {
+	{INDEX(arenas_bin_i)}
+};
+
+static const ctl_named_node_t arenas_lrun_i_node[] = {
+	{NAME("size"),			CTL(arenas_lrun_i_size)}
+};
+static const ctl_named_node_t super_arenas_lrun_i_node[] = {
+	{NAME(""),			CHILD(named, arenas_lrun_i)}
+};
+
+static const ctl_indexed_node_t arenas_lrun_node[] = {
+	{INDEX(arenas_lrun_i)}
+};
+
+static const ctl_named_node_t arenas_node[] = {
+	{NAME("narenas"),		CTL(arenas_narenas)},
+	{NAME("initialized"),		CTL(arenas_initialized)},
+	{NAME("quantum"),		CTL(arenas_quantum)},
+	{NAME("page"),			CTL(arenas_page)},
+	{NAME("tcache_max"),		CTL(arenas_tcache_max)},
+	{NAME("nbins"),			CTL(arenas_nbins)},
+	{NAME("nhbins"),		CTL(arenas_nhbins)},
+	{NAME("bin"),			CHILD(indexed, arenas_bin)},
+	{NAME("nlruns"),		CTL(arenas_nlruns)},
+	{NAME("lrun"),			CHILD(indexed, arenas_lrun)},
+	{NAME("purge"),			CTL(arenas_purge)},
+	{NAME("extend"),		CTL(arenas_extend)}
+};
+
+static const ctl_named_node_t	prof_node[] = {
+	{NAME("active"),	CTL(prof_active)},
+	{NAME("dump"),		CTL(prof_dump)},
+	{NAME("interval"),	CTL(prof_interval)}
+};
+
+static const ctl_named_node_t stats_chunks_node[] = {
+	{NAME("current"),		CTL(stats_chunks_current)},
+	{NAME("total"),			CTL(stats_chunks_total)},
+	{NAME("high"),			CTL(stats_chunks_high)}
+};
+
+static const ctl_named_node_t stats_huge_node[] = {
+	{NAME("allocated"),		CTL(stats_huge_allocated)},
+	{NAME("nmalloc"),		CTL(stats_huge_nmalloc)},
+	{NAME("ndalloc"),		CTL(stats_huge_ndalloc)}
+};
+
+static const ctl_named_node_t stats_arenas_i_small_node[] = {
+	{NAME("allocated"),		CTL(stats_arenas_i_small_allocated)},
+	{NAME("nmalloc"),		CTL(stats_arenas_i_small_nmalloc)},
+	{NAME("ndalloc"),		CTL(stats_arenas_i_small_ndalloc)},
+	{NAME("nrequests"),		CTL(stats_arenas_i_small_nrequests)}
+};
+
+static const ctl_named_node_t stats_arenas_i_large_node[] = {
+	{NAME("allocated"),		CTL(stats_arenas_i_large_allocated)},
+	{NAME("nmalloc"),		CTL(stats_arenas_i_large_nmalloc)},
+	{NAME("ndalloc"),		CTL(stats_arenas_i_large_ndalloc)},
+	{NAME("nrequests"),		CTL(stats_arenas_i_large_nrequests)}
+};
+
+static const ctl_named_node_t stats_arenas_i_bins_j_node[] = {
+	{NAME("allocated"),		CTL(stats_arenas_i_bins_j_allocated)},
+	{NAME("nmalloc"),		CTL(stats_arenas_i_bins_j_nmalloc)},
+	{NAME("ndalloc"),		CTL(stats_arenas_i_bins_j_ndalloc)},
+	{NAME("nrequests"),		CTL(stats_arenas_i_bins_j_nrequests)},
+	{NAME("nfills"),		CTL(stats_arenas_i_bins_j_nfills)},
+	{NAME("nflushes"),		CTL(stats_arenas_i_bins_j_nflushes)},
+	{NAME("nruns"),			CTL(stats_arenas_i_bins_j_nruns)},
+	{NAME("nreruns"),		CTL(stats_arenas_i_bins_j_nreruns)},
+	{NAME("curruns"),		CTL(stats_arenas_i_bins_j_curruns)}
+};
+static const ctl_named_node_t super_stats_arenas_i_bins_j_node[] = {
+	{NAME(""),			CHILD(named, stats_arenas_i_bins_j)}
+};
+
+static const ctl_indexed_node_t stats_arenas_i_bins_node[] = {
+	{INDEX(stats_arenas_i_bins_j)}
+};
+
+static const ctl_named_node_t stats_arenas_i_lruns_j_node[] = {
+	{NAME("nmalloc"),		CTL(stats_arenas_i_lruns_j_nmalloc)},
+	{NAME("ndalloc"),		CTL(stats_arenas_i_lruns_j_ndalloc)},
+	{NAME("nrequests"),		CTL(stats_arenas_i_lruns_j_nrequests)},
+	{NAME("curruns"),		CTL(stats_arenas_i_lruns_j_curruns)}
+};
+static const ctl_named_node_t super_stats_arenas_i_lruns_j_node[] = {
+	{NAME(""),			CHILD(named, stats_arenas_i_lruns_j)}
+};
+
+static const ctl_indexed_node_t stats_arenas_i_lruns_node[] = {
+	{INDEX(stats_arenas_i_lruns_j)}
+};
+
+static const ctl_named_node_t stats_arenas_i_node[] = {
+	{NAME("nthreads"),		CTL(stats_arenas_i_nthreads)},
+	{NAME("dss"),			CTL(stats_arenas_i_dss)},
+	{NAME("pactive"),		CTL(stats_arenas_i_pactive)},
+	{NAME("pdirty"),		CTL(stats_arenas_i_pdirty)},
+	{NAME("mapped"),		CTL(stats_arenas_i_mapped)},
+	{NAME("npurge"),		CTL(stats_arenas_i_npurge)},
+	{NAME("nmadvise"),		CTL(stats_arenas_i_nmadvise)},
+	{NAME("purged"),		CTL(stats_arenas_i_purged)},
+	{NAME("small"),			CHILD(named, stats_arenas_i_small)},
+	{NAME("large"),			CHILD(named, stats_arenas_i_large)},
+	{NAME("bins"),			CHILD(indexed, stats_arenas_i_bins)},
+	{NAME("lruns"),			CHILD(indexed, stats_arenas_i_lruns)}
+};
+static const ctl_named_node_t super_stats_arenas_i_node[] = {
+	{NAME(""),			CHILD(named, stats_arenas_i)}
+};
+
+static const ctl_indexed_node_t stats_arenas_node[] = {
+	{INDEX(stats_arenas_i)}
+};
+
+static const ctl_named_node_t stats_node[] = {
+	{NAME("cactive"),		CTL(stats_cactive)},
+	{NAME("allocated"),		CTL(stats_allocated)},
+	{NAME("active"),		CTL(stats_active)},
+	{NAME("mapped"),		CTL(stats_mapped)},
+	{NAME("chunks"),		CHILD(named, stats_chunks)},
+	{NAME("huge"),			CHILD(named, stats_huge)},
+	{NAME("arenas"),		CHILD(indexed, stats_arenas)}
+};
+
+static const ctl_named_node_t	root_node[] = {
+	{NAME("version"),	CTL(version)},
+	{NAME("epoch"),		CTL(epoch)},
+	{NAME("thread"),	CHILD(named, thread)},
+	{NAME("config"),	CHILD(named, config)},
+	{NAME("opt"),		CHILD(named, opt)},
+	{NAME("arena"),		CHILD(indexed, arena)},
+	{NAME("arenas"),	CHILD(named, arenas)},
+	{NAME("prof"),		CHILD(named, prof)},
+	{NAME("stats"),		CHILD(named, stats)}
+};
+static const ctl_named_node_t super_root_node[] = {
+	{NAME(""),		CHILD(named, root)}
+};
+
+#undef NAME
+#undef CHILD
+#undef CTL
+#undef INDEX
+
+/******************************************************************************/
+
+static bool
+ctl_arena_init(ctl_arena_stats_t *astats)
+{
+
+	if (astats->lstats == NULL) {
+		astats->lstats = (malloc_large_stats_t *)base_alloc(nlclasses *
+		    sizeof(malloc_large_stats_t));
+		if (astats->lstats == NULL)
+			return (true);
+	}
+
+	return (false);
+}
+
+static void
+ctl_arena_clear(ctl_arena_stats_t *astats)
+{
+
+	astats->dss = dss_prec_names[dss_prec_limit];
+	astats->pactive = 0;
+	astats->pdirty = 0;
+	if (config_stats) {
+		memset(&astats->astats, 0, sizeof(arena_stats_t));
+		astats->allocated_small = 0;
+		astats->nmalloc_small = 0;
+		astats->ndalloc_small = 0;
+		astats->nrequests_small = 0;
+		memset(astats->bstats, 0, NBINS * sizeof(malloc_bin_stats_t));
+		memset(astats->lstats, 0, nlclasses *
+		    sizeof(malloc_large_stats_t));
+	}
+}
+
+static void
+ctl_arena_stats_amerge(ctl_arena_stats_t *cstats, arena_t *arena)
+{
+	unsigned i;
+
+	arena_stats_merge(arena, &cstats->dss, &cstats->pactive,
+	    &cstats->pdirty, &cstats->astats, cstats->bstats, cstats->lstats);
+
+	for (i = 0; i < NBINS; i++) {
+		cstats->allocated_small += cstats->bstats[i].allocated;
+		cstats->nmalloc_small += cstats->bstats[i].nmalloc;
+		cstats->ndalloc_small += cstats->bstats[i].ndalloc;
+		cstats->nrequests_small += cstats->bstats[i].nrequests;
+	}
+}
+
+static void
+ctl_arena_stats_smerge(ctl_arena_stats_t *sstats, ctl_arena_stats_t *astats)
+{
+	unsigned i;
+
+	sstats->pactive += astats->pactive;
+	sstats->pdirty += astats->pdirty;
+
+	sstats->astats.mapped += astats->astats.mapped;
+	sstats->astats.npurge += astats->astats.npurge;
+	sstats->astats.nmadvise += astats->astats.nmadvise;
+	sstats->astats.purged += astats->astats.purged;
+
+	sstats->allocated_small += astats->allocated_small;
+	sstats->nmalloc_small += astats->nmalloc_small;
+	sstats->ndalloc_small += astats->ndalloc_small;
+	sstats->nrequests_small += astats->nrequests_small;
+
+	sstats->astats.allocated_large += astats->astats.allocated_large;
+	sstats->astats.nmalloc_large += astats->astats.nmalloc_large;
+	sstats->astats.ndalloc_large += astats->astats.ndalloc_large;
+	sstats->astats.nrequests_large += astats->astats.nrequests_large;
+
+	for (i = 0; i < nlclasses; i++) {
+		sstats->lstats[i].nmalloc += astats->lstats[i].nmalloc;
+		sstats->lstats[i].ndalloc += astats->lstats[i].ndalloc;
+		sstats->lstats[i].nrequests += astats->lstats[i].nrequests;
+		sstats->lstats[i].curruns += astats->lstats[i].curruns;
+	}
+
+	for (i = 0; i < NBINS; i++) {
+		sstats->bstats[i].allocated += astats->bstats[i].allocated;
+		sstats->bstats[i].nmalloc += astats->bstats[i].nmalloc;
+		sstats->bstats[i].ndalloc += astats->bstats[i].ndalloc;
+		sstats->bstats[i].nrequests += astats->bstats[i].nrequests;
+		if (config_tcache) {
+			sstats->bstats[i].nfills += astats->bstats[i].nfills;
+			sstats->bstats[i].nflushes +=
+			    astats->bstats[i].nflushes;
+		}
+		sstats->bstats[i].nruns += astats->bstats[i].nruns;
+		sstats->bstats[i].reruns += astats->bstats[i].reruns;
+		sstats->bstats[i].curruns += astats->bstats[i].curruns;
+	}
+}
+
+static void
+ctl_arena_refresh(arena_t *arena, unsigned i)
+{
+	ctl_arena_stats_t *astats = &ctl_stats.arenas[i];
+	ctl_arena_stats_t *sstats = &ctl_stats.arenas[ctl_stats.narenas];
+
+	ctl_arena_clear(astats);
+
+	sstats->nthreads += astats->nthreads;
+	if (config_stats) {
+		ctl_arena_stats_amerge(astats, arena);
+		/* Merge into sum stats as well. */
+		ctl_arena_stats_smerge(sstats, astats);
+	} else {
+		astats->pactive += arena->nactive;
+		astats->pdirty += arena->ndirty;
+		/* Merge into sum stats as well. */
+		sstats->pactive += arena->nactive;
+		sstats->pdirty += arena->ndirty;
+	}
+}
+
+static bool
+ctl_grow(void)
+{
+	size_t astats_size;
+	ctl_arena_stats_t *astats;
+	arena_t **tarenas;
+
+	/* Extend arena stats and arenas arrays. */
+	astats_size = (ctl_stats.narenas + 2) * sizeof(ctl_arena_stats_t);
+	if (ctl_stats.narenas == narenas_auto) {
+		/* ctl_stats.arenas and arenas came from base_alloc(). */
+		astats = (ctl_arena_stats_t *)imalloc(astats_size);
+		if (astats == NULL)
+			return (true);
+		memcpy(astats, ctl_stats.arenas, (ctl_stats.narenas + 1) *
+		    sizeof(ctl_arena_stats_t));
+
+		tarenas = (arena_t **)imalloc((ctl_stats.narenas + 1) *
+		    sizeof(arena_t *));
+		if (tarenas == NULL) {
+			idalloc(astats);
+			return (true);
+		}
+		memcpy(tarenas, arenas, ctl_stats.narenas * sizeof(arena_t *));
+	} else {
+		astats = (ctl_arena_stats_t *)iralloc(ctl_stats.arenas,
+		    astats_size, 0, 0, false, false);
+		if (astats == NULL)
+			return (true);
+
+		tarenas = (arena_t **)iralloc(arenas, (ctl_stats.narenas + 1) *
+		    sizeof(arena_t *), 0, 0, false, false);
+		if (tarenas == NULL)
+			return (true);
+	}
+	/* Initialize the new astats and arenas elements. */
+	memset(&astats[ctl_stats.narenas + 1], 0, sizeof(ctl_arena_stats_t));
+	if (ctl_arena_init(&astats[ctl_stats.narenas + 1]))
+		return (true);
+	tarenas[ctl_stats.narenas] = NULL;
+	/* Swap merged stats to their new location. */
+	{
+		ctl_arena_stats_t tstats;
+		memcpy(&tstats, &astats[ctl_stats.narenas],
+		    sizeof(ctl_arena_stats_t));
+		memcpy(&astats[ctl_stats.narenas],
+		    &astats[ctl_stats.narenas + 1], sizeof(ctl_arena_stats_t));
+		memcpy(&astats[ctl_stats.narenas + 1], &tstats,
+		    sizeof(ctl_arena_stats_t));
+	}
+	ctl_stats.arenas = astats;
+	ctl_stats.narenas++;
+	malloc_mutex_lock(&arenas_lock);
+	arenas = tarenas;
+	narenas_total++;
+	arenas_extend(narenas_total - 1);
+	malloc_mutex_unlock(&arenas_lock);
+
+	return (false);
+}
+
+static void
+ctl_refresh(void)
+{
+	unsigned i;
+	VARIABLE_ARRAY(arena_t *, tarenas, ctl_stats.narenas);
+
+	if (config_stats) {
+		malloc_mutex_lock(&chunks_mtx);
+		ctl_stats.chunks.current = stats_chunks.curchunks;
+		ctl_stats.chunks.total = stats_chunks.nchunks;
+		ctl_stats.chunks.high = stats_chunks.highchunks;
+		malloc_mutex_unlock(&chunks_mtx);
+
+		malloc_mutex_lock(&huge_mtx);
+		ctl_stats.huge.allocated = huge_allocated;
+		ctl_stats.huge.nmalloc = huge_nmalloc;
+		ctl_stats.huge.ndalloc = huge_ndalloc;
+		malloc_mutex_unlock(&huge_mtx);
+	}
+
+	/*
+	 * Clear sum stats, since they will be merged into by
+	 * ctl_arena_refresh().
+	 */
+	ctl_stats.arenas[ctl_stats.narenas].nthreads = 0;
+	ctl_arena_clear(&ctl_stats.arenas[ctl_stats.narenas]);
+
+	malloc_mutex_lock(&arenas_lock);
+	memcpy(tarenas, arenas, sizeof(arena_t *) * ctl_stats.narenas);
+	for (i = 0; i < ctl_stats.narenas; i++) {
+		if (arenas[i] != NULL)
+			ctl_stats.arenas[i].nthreads = arenas[i]->nthreads;
+		else
+			ctl_stats.arenas[i].nthreads = 0;
+	}
+	malloc_mutex_unlock(&arenas_lock);
+	for (i = 0; i < ctl_stats.narenas; i++) {
+		bool initialized = (tarenas[i] != NULL);
+
+		ctl_stats.arenas[i].initialized = initialized;
+		if (initialized)
+			ctl_arena_refresh(tarenas[i], i);
+	}
+
+	if (config_stats) {
+		ctl_stats.allocated =
+		    ctl_stats.arenas[ctl_stats.narenas].allocated_small
+		    + ctl_stats.arenas[ctl_stats.narenas].astats.allocated_large
+		    + ctl_stats.huge.allocated;
+		ctl_stats.active =
+		    (ctl_stats.arenas[ctl_stats.narenas].pactive << LG_PAGE)
+		    + ctl_stats.huge.allocated;
+		ctl_stats.mapped = (ctl_stats.chunks.current << opt_lg_chunk);
+	}
+
+	ctl_epoch++;
+}
+
+static bool
+ctl_init(void)
+{
+	bool ret;
+
+	malloc_mutex_lock(&ctl_mtx);
+	if (ctl_initialized == false) {
+		/*
+		 * Allocate space for one extra arena stats element, which
+		 * contains summed stats across all arenas.
+		 */
+		assert(narenas_auto == narenas_total_get());
+		ctl_stats.narenas = narenas_auto;
+		ctl_stats.arenas = (ctl_arena_stats_t *)base_alloc(
+		    (ctl_stats.narenas + 1) * sizeof(ctl_arena_stats_t));
+		if (ctl_stats.arenas == NULL) {
+			ret = true;
+			goto label_return;
+		}
+		memset(ctl_stats.arenas, 0, (ctl_stats.narenas + 1) *
+		    sizeof(ctl_arena_stats_t));
+
+		/*
+		 * Initialize all stats structures, regardless of whether they
+		 * ever get used.  Lazy initialization would allow errors to
+		 * cause inconsistent state to be viewable by the application.
+		 */
+		if (config_stats) {
+			unsigned i;
+			for (i = 0; i <= ctl_stats.narenas; i++) {
+				if (ctl_arena_init(&ctl_stats.arenas[i])) {
+					ret = true;
+					goto label_return;
+				}
+			}
+		}
+		ctl_stats.arenas[ctl_stats.narenas].initialized = true;
+
+		ctl_epoch = 0;
+		ctl_refresh();
+		ctl_initialized = true;
+	}
+
+	ret = false;
+label_return:
+	malloc_mutex_unlock(&ctl_mtx);
+	return (ret);
+}
+
+static int
+ctl_lookup(const char *name, ctl_node_t const **nodesp, size_t *mibp,
+    size_t *depthp)
+{
+	int ret;
+	const char *elm, *tdot, *dot;
+	size_t elen, i, j;
+	const ctl_named_node_t *node;
+
+	elm = name;
+	/* Equivalent to strchrnul(). */
+	dot = ((tdot = strchr(elm, '.')) != NULL) ? tdot : strchr(elm, '\0');
+	elen = (size_t)((uintptr_t)dot - (uintptr_t)elm);
+	if (elen == 0) {
+		ret = ENOENT;
+		goto label_return;
+	}
+	node = super_root_node;
+	for (i = 0; i < *depthp; i++) {
+		assert(node);
+		assert(node->nchildren > 0);
+		if (ctl_named_node(node->children) != NULL) {
+			const ctl_named_node_t *pnode = node;
+
+			/* Children are named. */
+			for (j = 0; j < node->nchildren; j++) {
+				const ctl_named_node_t *child =
+				    ctl_named_children(node, j);
+				if (strlen(child->name) == elen &&
+				    strncmp(elm, child->name, elen) == 0) {
+					node = child;
+					if (nodesp != NULL)
+						nodesp[i] =
+						    (const ctl_node_t *)node;
+					mibp[i] = j;
+					break;
+				}
+			}
+			if (node == pnode) {
+				ret = ENOENT;
+				goto label_return;
+			}
+		} else {
+			uintmax_t index;
+			const ctl_indexed_node_t *inode;
+
+			/* Children are indexed. */
+			index = malloc_strtoumax(elm, NULL, 10);
+			if (index == UINTMAX_MAX || index > SIZE_T_MAX) {
+				ret = ENOENT;
+				goto label_return;
+			}
+
+			inode = ctl_indexed_node(node->children);
+			node = inode->index(mibp, *depthp, (size_t)index);
+			if (node == NULL) {
+				ret = ENOENT;
+				goto label_return;
+			}
+
+			if (nodesp != NULL)
+				nodesp[i] = (const ctl_node_t *)node;
+			mibp[i] = (size_t)index;
+		}
+
+		if (node->ctl != NULL) {
+			/* Terminal node. */
+			if (*dot != '\0') {
+				/*
+				 * The name contains more elements than are
+				 * in this path through the tree.
+				 */
+				ret = ENOENT;
+				goto label_return;
+			}
+			/* Complete lookup successful. */
+			*depthp = i + 1;
+			break;
+		}
+
+		/* Update elm. */
+		if (*dot == '\0') {
+			/* No more elements. */
+			ret = ENOENT;
+			goto label_return;
+		}
+		elm = &dot[1];
+		dot = ((tdot = strchr(elm, '.')) != NULL) ? tdot :
+		    strchr(elm, '\0');
+		elen = (size_t)((uintptr_t)dot - (uintptr_t)elm);
+	}
+
+	ret = 0;
+label_return:
+	return (ret);
+}
+
+int
+ctl_byname(const char *name, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen)
+{
+	int ret;
+	size_t depth;
+	ctl_node_t const *nodes[CTL_MAX_DEPTH];
+	size_t mib[CTL_MAX_DEPTH];
+	const ctl_named_node_t *node;
+
+	if (ctl_initialized == false && ctl_init()) {
+		ret = EAGAIN;
+		goto label_return;
+	}
+
+	depth = CTL_MAX_DEPTH;
+	ret = ctl_lookup(name, nodes, mib, &depth);
+	if (ret != 0)
+		goto label_return;
+
+	node = ctl_named_node(nodes[depth-1]);
+	if (node != NULL && node->ctl)
+		ret = node->ctl(mib, depth, oldp, oldlenp, newp, newlen);
+	else {
+		/* The name refers to a partial path through the ctl tree. */
+		ret = ENOENT;
+	}
+
+label_return:
+	return(ret);
+}
+
+int
+ctl_nametomib(const char *name, size_t *mibp, size_t *miblenp)
+{
+	int ret;
+
+	if (ctl_initialized == false && ctl_init()) {
+		ret = EAGAIN;
+		goto label_return;
+	}
+
+	ret = ctl_lookup(name, NULL, mibp, miblenp);
+label_return:
+	return(ret);
+}
+
+int
+ctl_bymib(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen)
+{
+	int ret;
+	const ctl_named_node_t *node;
+	size_t i;
+
+	if (ctl_initialized == false && ctl_init()) {
+		ret = EAGAIN;
+		goto label_return;
+	}
+
+	/* Iterate down the tree. */
+	node = super_root_node;
+	for (i = 0; i < miblen; i++) {
+		assert(node);
+		assert(node->nchildren > 0);
+		if (ctl_named_node(node->children) != NULL) {
+			/* Children are named. */
+			if (node->nchildren <= mib[i]) {
+				ret = ENOENT;
+				goto label_return;
+			}
+			node = ctl_named_children(node, mib[i]);
+		} else {
+			const ctl_indexed_node_t *inode;
+
+			/* Indexed element. */
+			inode = ctl_indexed_node(node->children);
+			node = inode->index(mib, miblen, mib[i]);
+			if (node == NULL) {
+				ret = ENOENT;
+				goto label_return;
+			}
+		}
+	}
+
+	/* Call the ctl function. */
+	if (node && node->ctl)
+		ret = node->ctl(mib, miblen, oldp, oldlenp, newp, newlen);
+	else {
+		/* Partial MIB. */
+		ret = ENOENT;
+	}
+
+label_return:
+	return(ret);
+}
+
+bool
+ctl_boot(void)
+{
+
+	if (malloc_mutex_init(&ctl_mtx))
+		return (true);
+
+	ctl_initialized = false;
+
+	return (false);
+}
+
+void
+ctl_prefork(void)
+{
+
+	malloc_mutex_lock(&ctl_mtx);
+}
+
+void
+ctl_postfork_parent(void)
+{
+
+	malloc_mutex_postfork_parent(&ctl_mtx);
+}
+
+void
+ctl_postfork_child(void)
+{
+
+	malloc_mutex_postfork_child(&ctl_mtx);
+}
+
+/******************************************************************************/
+/* *_ctl() functions. */
+
+#define	READONLY()	do {						\
+	if (newp != NULL || newlen != 0) {				\
+		ret = EPERM;						\
+		goto label_return;					\
+	}								\
+} while (0)
+
+#define	WRITEONLY()	do {						\
+	if (oldp != NULL || oldlenp != NULL) {				\
+		ret = EPERM;						\
+		goto label_return;					\
+	}								\
+} while (0)
+
+#define	READ(v, t)	do {						\
+	if (oldp != NULL && oldlenp != NULL) {				\
+		if (*oldlenp != sizeof(t)) {				\
+			size_t	copylen = (sizeof(t) <= *oldlenp)	\
+			    ? sizeof(t) : *oldlenp;			\
+			memcpy(oldp, (void *)&(v), copylen);		\
+			ret = EINVAL;					\
+			goto label_return;				\
+		} else							\
+			*(t *)oldp = (v);				\
+	}								\
+} while (0)
+
+#define	WRITE(v, t)	do {						\
+	if (newp != NULL) {						\
+		if (newlen != sizeof(t)) {				\
+			ret = EINVAL;					\
+			goto label_return;				\
+		}							\
+		(v) = *(t *)newp;					\
+	}								\
+} while (0)
+
+/*
+ * There's a lot of code duplication in the following macros due to limitations
+ * in how nested cpp macros are expanded.
+ */
+#define	CTL_RO_CLGEN(c, l, n, v, t)					\
+static int								\
+n##_ctl(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,	\
+    void *newp, size_t newlen)						\
+{									\
+	int ret;							\
+	t oldval;							\
+									\
+	if ((c) == false)						\
+		return (ENOENT);					\
+	if (l)								\
+		malloc_mutex_lock(&ctl_mtx);				\
+	READONLY();							\
+	oldval = (v);							\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	if (l)								\
+		malloc_mutex_unlock(&ctl_mtx);				\
+	return (ret);							\
+}
+
+#define	CTL_RO_CGEN(c, n, v, t)						\
+static int								\
+n##_ctl(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,	\
+    void *newp, size_t newlen)						\
+{									\
+	int ret;							\
+	t oldval;							\
+									\
+	if ((c) == false)						\
+		return (ENOENT);					\
+	malloc_mutex_lock(&ctl_mtx);					\
+	READONLY();							\
+	oldval = (v);							\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	malloc_mutex_unlock(&ctl_mtx);					\
+	return (ret);							\
+}
+
+#define	CTL_RO_GEN(n, v, t)						\
+static int								\
+n##_ctl(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,	\
+    void *newp, size_t newlen)						\
+{									\
+	int ret;							\
+	t oldval;							\
+									\
+	malloc_mutex_lock(&ctl_mtx);					\
+	READONLY();							\
+	oldval = (v);							\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	malloc_mutex_unlock(&ctl_mtx);					\
+	return (ret);							\
+}
+
+/*
+ * ctl_mtx is not acquired, under the assumption that no pertinent data will
+ * mutate during the call.
+ */
+#define	CTL_RO_NL_CGEN(c, n, v, t)					\
+static int								\
+n##_ctl(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,	\
+    void *newp, size_t newlen)						\
+{									\
+	int ret;							\
+	t oldval;							\
+									\
+	if ((c) == false)						\
+		return (ENOENT);					\
+	READONLY();							\
+	oldval = (v);							\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	return (ret);							\
+}
+
+#define	CTL_RO_NL_GEN(n, v, t)						\
+static int								\
+n##_ctl(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,	\
+    void *newp, size_t newlen)						\
+{									\
+	int ret;							\
+	t oldval;							\
+									\
+	READONLY();							\
+	oldval = (v);							\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	return (ret);							\
+}
+
+#define	CTL_RO_BOOL_CONFIG_GEN(n)					\
+static int								\
+n##_ctl(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,	\
+    void *newp, size_t newlen)						\
+{									\
+	int ret;							\
+	bool oldval;							\
+									\
+	READONLY();							\
+	oldval = n;							\
+	READ(oldval, bool);						\
+									\
+	ret = 0;							\
+label_return:								\
+	return (ret);							\
+}
+
+CTL_RO_NL_GEN(version, JEMALLOC_VERSION, const char *)
+
+static int
+epoch_ctl(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen)
+{
+	int ret;
+	uint64_t newval;
+
+	malloc_mutex_lock(&ctl_mtx);
+	WRITE(newval, uint64_t);
+	if (newp != NULL)
+		ctl_refresh();
+	READ(ctl_epoch, uint64_t);
+
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(&ctl_mtx);
+	return (ret);
+}
+
+static int
+thread_tcache_enabled_ctl(const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen)
+{
+	int ret;
+	bool oldval;
+
+	if (config_tcache == false)
+		return (ENOENT);
+
+	oldval = tcache_enabled_get();
+	if (newp != NULL) {
+		if (newlen != sizeof(bool)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		tcache_enabled_set(*(bool *)newp);
+	}
+	READ(oldval, bool);
+
+	ret = 0;
+label_return:
+	return (ret);
+}
+
+static int
+thread_tcache_flush_ctl(const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen)
+{
+	int ret;
+
+	if (config_tcache == false)
+		return (ENOENT);
+
+	READONLY();
+	WRITEONLY();
+
+	tcache_flush();
+
+	ret = 0;
+label_return:
+	return (ret);
+}
+
+static int
+thread_arena_ctl(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen)
+{
+	int ret;
+	unsigned newind, oldind;
+
+	malloc_mutex_lock(&ctl_mtx);
+	newind = oldind = choose_arena(NULL)->ind;
+	WRITE(newind, unsigned);
+	READ(oldind, unsigned);
+	if (newind != oldind) {
+		arena_t *arena;
+
+		if (newind >= ctl_stats.narenas) {
+			/* New arena index is out of range. */
+			ret = EFAULT;
+			goto label_return;
+		}
+
+		/* Initialize arena if necessary. */
+		malloc_mutex_lock(&arenas_lock);
+		if ((arena = arenas[newind]) == NULL && (arena =
+		    arenas_extend(newind)) == NULL) {
+			malloc_mutex_unlock(&arenas_lock);
+			ret = EAGAIN;
+			goto label_return;
+		}
+		assert(arena == arenas[newind]);
+		arenas[oldind]->nthreads--;
+		arenas[newind]->nthreads++;
+		malloc_mutex_unlock(&arenas_lock);
+
+		/* Set new arena association. */
+		if (config_tcache) {
+			tcache_t *tcache;
+			if ((uintptr_t)(tcache = *tcache_tsd_get()) >
+			    (uintptr_t)TCACHE_STATE_MAX) {
+				tcache_arena_dissociate(tcache);
+				tcache_arena_associate(tcache, arena);
+			}
+		}
+		arenas_tsd_set(&arena);
+	}
+
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(&ctl_mtx);
+	return (ret);
+}
+
+CTL_RO_NL_CGEN(config_stats, thread_allocated,
+    thread_allocated_tsd_get()->allocated, uint64_t)
+CTL_RO_NL_CGEN(config_stats, thread_allocatedp,
+    &thread_allocated_tsd_get()->allocated, uint64_t *)
+CTL_RO_NL_CGEN(config_stats, thread_deallocated,
+    thread_allocated_tsd_get()->deallocated, uint64_t)
+CTL_RO_NL_CGEN(config_stats, thread_deallocatedp,
+    &thread_allocated_tsd_get()->deallocated, uint64_t *)
+
+/******************************************************************************/
+
+CTL_RO_BOOL_CONFIG_GEN(config_debug)
+CTL_RO_BOOL_CONFIG_GEN(config_dss)
+CTL_RO_BOOL_CONFIG_GEN(config_fill)
+CTL_RO_BOOL_CONFIG_GEN(config_lazy_lock)
+CTL_RO_BOOL_CONFIG_GEN(config_mremap)
+CTL_RO_BOOL_CONFIG_GEN(config_munmap)
+CTL_RO_BOOL_CONFIG_GEN(config_prof)
+CTL_RO_BOOL_CONFIG_GEN(config_prof_libgcc)
+CTL_RO_BOOL_CONFIG_GEN(config_prof_libunwind)
+CTL_RO_BOOL_CONFIG_GEN(config_stats)
+CTL_RO_BOOL_CONFIG_GEN(config_tcache)
+CTL_RO_BOOL_CONFIG_GEN(config_tls)
+CTL_RO_BOOL_CONFIG_GEN(config_utrace)
+CTL_RO_BOOL_CONFIG_GEN(config_valgrind)
+CTL_RO_BOOL_CONFIG_GEN(config_xmalloc)
+
+/******************************************************************************/
+
+CTL_RO_NL_GEN(opt_abort, opt_abort, bool)
+CTL_RO_NL_GEN(opt_dss, opt_dss, const char *)
+CTL_RO_NL_GEN(opt_lg_chunk, opt_lg_chunk, size_t)
+CTL_RO_NL_GEN(opt_narenas, opt_narenas, size_t)
+CTL_RO_NL_GEN(opt_lg_dirty_mult, opt_lg_dirty_mult, ssize_t)
+CTL_RO_NL_GEN(opt_stats_print, opt_stats_print, bool)
+CTL_RO_NL_CGEN(config_fill, opt_junk, opt_junk, bool)
+CTL_RO_NL_CGEN(config_fill, opt_zero, opt_zero, bool)
+CTL_RO_NL_CGEN(config_fill, opt_quarantine, opt_quarantine, size_t)
+CTL_RO_NL_CGEN(config_fill, opt_redzone, opt_redzone, bool)
+CTL_RO_NL_CGEN(config_utrace, opt_utrace, opt_utrace, bool)
+CTL_RO_NL_CGEN(config_valgrind, opt_valgrind, opt_valgrind, bool)
+CTL_RO_NL_CGEN(config_xmalloc, opt_xmalloc, opt_xmalloc, bool)
+CTL_RO_NL_CGEN(config_tcache, opt_tcache, opt_tcache, bool)
+CTL_RO_NL_CGEN(config_tcache, opt_lg_tcache_max, opt_lg_tcache_max, ssize_t)
+CTL_RO_NL_CGEN(config_prof, opt_prof, opt_prof, bool)
+CTL_RO_NL_CGEN(config_prof, opt_prof_prefix, opt_prof_prefix, const char *)
+CTL_RO_CGEN(config_prof, opt_prof_active, opt_prof_active, bool) /* Mutable. */
+CTL_RO_NL_CGEN(config_prof, opt_lg_prof_sample, opt_lg_prof_sample, size_t)
+CTL_RO_NL_CGEN(config_prof, opt_lg_prof_interval, opt_lg_prof_interval, ssize_t)
+CTL_RO_NL_CGEN(config_prof, opt_prof_gdump, opt_prof_gdump, bool)
+CTL_RO_NL_CGEN(config_prof, opt_prof_final, opt_prof_final, bool)
+CTL_RO_NL_CGEN(config_prof, opt_prof_leak, opt_prof_leak, bool)
+CTL_RO_NL_CGEN(config_prof, opt_prof_accum, opt_prof_accum, bool)
+
+/******************************************************************************/
+
+/* ctl_mutex must be held during execution of this function. */
+static void
+arena_purge(unsigned arena_ind)
+{
+	VARIABLE_ARRAY(arena_t *, tarenas, ctl_stats.narenas);
+
+	malloc_mutex_lock(&arenas_lock);
+	memcpy(tarenas, arenas, sizeof(arena_t *) * ctl_stats.narenas);
+	malloc_mutex_unlock(&arenas_lock);
+
+	if (arena_ind == ctl_stats.narenas) {
+		unsigned i;
+		for (i = 0; i < ctl_stats.narenas; i++) {
+			if (tarenas[i] != NULL)
+				arena_purge_all(tarenas[i]);
+		}
+	} else {
+		assert(arena_ind < ctl_stats.narenas);
+		if (tarenas[arena_ind] != NULL)
+			arena_purge_all(tarenas[arena_ind]);
+	}
+}
+
+static int
+arena_i_purge_ctl(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen)
+{
+	int ret;
+
+	READONLY();
+	WRITEONLY();
+	malloc_mutex_lock(&ctl_mtx);
+	arena_purge(mib[1]);
+	malloc_mutex_unlock(&ctl_mtx);
+
+	ret = 0;
+label_return:
+	return (ret);
+}
+
+static int
+arena_i_dss_ctl(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen)
+{
+	int ret, i;
+	bool match, err;
+	const char *dss;
+	unsigned arena_ind = mib[1];
+	dss_prec_t dss_prec_old = dss_prec_limit;
+	dss_prec_t dss_prec = dss_prec_limit;
+
+	malloc_mutex_lock(&ctl_mtx);
+	WRITE(dss, const char *);
+	match = false;
+	for (i = 0; i < dss_prec_limit; i++) {
+		if (strcmp(dss_prec_names[i], dss) == 0) {
+			dss_prec = i;
+			match = true;
+			break;
+		}
+	}
+	if (match == false) {
+		ret = EINVAL;
+		goto label_return;
+	}
+
+	if (arena_ind < ctl_stats.narenas) {
+		arena_t *arena = arenas[arena_ind];
+		if (arena != NULL) {
+			dss_prec_old = arena_dss_prec_get(arena);
+			arena_dss_prec_set(arena, dss_prec);
+			err = false;
+		} else
+			err = true;
+	} else {
+		dss_prec_old = chunk_dss_prec_get();
+		err = chunk_dss_prec_set(dss_prec);
+	}
+	dss = dss_prec_names[dss_prec_old];
+	READ(dss, const char *);
+	if (err) {
+		ret = EFAULT;
+		goto label_return;
+	}
+
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(&ctl_mtx);
+	return (ret);
+}
+
+static const ctl_named_node_t *
+arena_i_index(const size_t *mib, size_t miblen, size_t i)
+{
+	const ctl_named_node_t * ret;
+
+	malloc_mutex_lock(&ctl_mtx);
+	if (i > ctl_stats.narenas) {
+		ret = NULL;
+		goto label_return;
+	}
+
+	ret = super_arena_i_node;
+label_return:
+	malloc_mutex_unlock(&ctl_mtx);
+	return (ret);
+}
+
+
+/******************************************************************************/
+
+CTL_RO_NL_GEN(arenas_bin_i_size, arena_bin_info[mib[2]].reg_size, size_t)
+CTL_RO_NL_GEN(arenas_bin_i_nregs, arena_bin_info[mib[2]].nregs, uint32_t)
+CTL_RO_NL_GEN(arenas_bin_i_run_size, arena_bin_info[mib[2]].run_size, size_t)
+static const ctl_named_node_t *
+arenas_bin_i_index(const size_t *mib, size_t miblen, size_t i)
+{
+
+	if (i > NBINS)
+		return (NULL);
+	return (super_arenas_bin_i_node);
+}
+
+CTL_RO_NL_GEN(arenas_lrun_i_size, ((mib[2]+1) << LG_PAGE), size_t)
+static const ctl_named_node_t *
+arenas_lrun_i_index(const size_t *mib, size_t miblen, size_t i)
+{
+
+	if (i > nlclasses)
+		return (NULL);
+	return (super_arenas_lrun_i_node);
+}
+
+static int
+arenas_narenas_ctl(const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen)
+{
+	int ret;
+	unsigned narenas;
+
+	malloc_mutex_lock(&ctl_mtx);
+	READONLY();
+	if (*oldlenp != sizeof(unsigned)) {
+		ret = EINVAL;
+		goto label_return;
+	}
+	narenas = ctl_stats.narenas;
+	READ(narenas, unsigned);
+
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(&ctl_mtx);
+	return (ret);
+}
+
+static int
+arenas_initialized_ctl(const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen)
+{
+	int ret;
+	unsigned nread, i;
+
+	malloc_mutex_lock(&ctl_mtx);
+	READONLY();
+	if (*oldlenp != ctl_stats.narenas * sizeof(bool)) {
+		ret = EINVAL;
+		nread = (*oldlenp < ctl_stats.narenas * sizeof(bool))
+		    ? (*oldlenp / sizeof(bool)) : ctl_stats.narenas;
+	} else {
+		ret = 0;
+		nread = ctl_stats.narenas;
+	}
+
+	for (i = 0; i < nread; i++)
+		((bool *)oldp)[i] = ctl_stats.arenas[i].initialized;
+
+label_return:
+	malloc_mutex_unlock(&ctl_mtx);
+	return (ret);
+}
+
+CTL_RO_NL_GEN(arenas_quantum, QUANTUM, size_t)
+CTL_RO_NL_GEN(arenas_page, PAGE, size_t)
+CTL_RO_NL_CGEN(config_tcache, arenas_tcache_max, tcache_maxclass, size_t)
+CTL_RO_NL_GEN(arenas_nbins, NBINS, unsigned)
+CTL_RO_NL_CGEN(config_tcache, arenas_nhbins, nhbins, unsigned)
+CTL_RO_NL_GEN(arenas_nlruns, nlclasses, size_t)
+
+static int
+arenas_purge_ctl(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen)
+{
+	int ret;
+	unsigned arena_ind;
+
+	malloc_mutex_lock(&ctl_mtx);
+	WRITEONLY();
+	arena_ind = UINT_MAX;
+	WRITE(arena_ind, unsigned);
+	if (newp != NULL && arena_ind >= ctl_stats.narenas)
+		ret = EFAULT;
+	else {
+		if (arena_ind == UINT_MAX)
+			arena_ind = ctl_stats.narenas;
+		arena_purge(arena_ind);
+		ret = 0;
+	}
+
+label_return:
+	malloc_mutex_unlock(&ctl_mtx);
+	return (ret);
+}
+
+static int
+arenas_extend_ctl(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen)
+{
+	int ret;
+	unsigned narenas;
+
+	malloc_mutex_lock(&ctl_mtx);
+	READONLY();
+	if (ctl_grow()) {
+		ret = EAGAIN;
+		goto label_return;
+	}
+	narenas = ctl_stats.narenas - 1;
+	READ(narenas, unsigned);
+
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(&ctl_mtx);
+	return (ret);
+}
+
+/******************************************************************************/
+
+static int
+prof_active_ctl(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen)
+{
+	int ret;
+	bool oldval;
+
+	if (config_prof == false)
+		return (ENOENT);
+
+	malloc_mutex_lock(&ctl_mtx); /* Protect opt_prof_active. */
+	oldval = opt_prof_active;
+	if (newp != NULL) {
+		/*
+		 * The memory barriers will tend to make opt_prof_active
+		 * propagate faster on systems with weak memory ordering.
+		 */
+		mb_write();
+		WRITE(opt_prof_active, bool);
+		mb_write();
+	}
+	READ(oldval, bool);
+
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(&ctl_mtx);
+	return (ret);
+}
+
+static int
+prof_dump_ctl(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen)
+{
+	int ret;
+	const char *filename = NULL;
+
+	if (config_prof == false)
+		return (ENOENT);
+
+	WRITEONLY();
+	WRITE(filename, const char *);
+
+	if (prof_mdump(filename)) {
+		ret = EFAULT;
+		goto label_return;
+	}
+
+	ret = 0;
+label_return:
+	return (ret);
+}
+
+CTL_RO_NL_CGEN(config_prof, prof_interval, prof_interval, uint64_t)
+
+/******************************************************************************/
+
+CTL_RO_CGEN(config_stats, stats_chunks_current, ctl_stats.chunks.current,
+    size_t)
+CTL_RO_CGEN(config_stats, stats_chunks_total, ctl_stats.chunks.total, uint64_t)
+CTL_RO_CGEN(config_stats, stats_chunks_high, ctl_stats.chunks.high, size_t)
+CTL_RO_CGEN(config_stats, stats_huge_allocated, huge_allocated, size_t)
+CTL_RO_CGEN(config_stats, stats_huge_nmalloc, huge_nmalloc, uint64_t)
+CTL_RO_CGEN(config_stats, stats_huge_ndalloc, huge_ndalloc, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_small_allocated,
+    ctl_stats.arenas[mib[2]].allocated_small, size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_small_nmalloc,
+    ctl_stats.arenas[mib[2]].nmalloc_small, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_small_ndalloc,
+    ctl_stats.arenas[mib[2]].ndalloc_small, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_small_nrequests,
+    ctl_stats.arenas[mib[2]].nrequests_small, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_large_allocated,
+    ctl_stats.arenas[mib[2]].astats.allocated_large, size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_large_nmalloc,
+    ctl_stats.arenas[mib[2]].astats.nmalloc_large, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_large_ndalloc,
+    ctl_stats.arenas[mib[2]].astats.ndalloc_large, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_large_nrequests,
+    ctl_stats.arenas[mib[2]].astats.nrequests_large, uint64_t)
+
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_allocated,
+    ctl_stats.arenas[mib[2]].bstats[mib[4]].allocated, size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nmalloc,
+    ctl_stats.arenas[mib[2]].bstats[mib[4]].nmalloc, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_ndalloc,
+    ctl_stats.arenas[mib[2]].bstats[mib[4]].ndalloc, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nrequests,
+    ctl_stats.arenas[mib[2]].bstats[mib[4]].nrequests, uint64_t)
+CTL_RO_CGEN(config_stats && config_tcache, stats_arenas_i_bins_j_nfills,
+    ctl_stats.arenas[mib[2]].bstats[mib[4]].nfills, uint64_t)
+CTL_RO_CGEN(config_stats && config_tcache, stats_arenas_i_bins_j_nflushes,
+    ctl_stats.arenas[mib[2]].bstats[mib[4]].nflushes, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nruns,
+    ctl_stats.arenas[mib[2]].bstats[mib[4]].nruns, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nreruns,
+    ctl_stats.arenas[mib[2]].bstats[mib[4]].reruns, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_curruns,
+    ctl_stats.arenas[mib[2]].bstats[mib[4]].curruns, size_t)
+
+static const ctl_named_node_t *
+stats_arenas_i_bins_j_index(const size_t *mib, size_t miblen, size_t j)
+{
+
+	if (j > NBINS)
+		return (NULL);
+	return (super_stats_arenas_i_bins_j_node);
+}
+
+CTL_RO_CGEN(config_stats, stats_arenas_i_lruns_j_nmalloc,
+    ctl_stats.arenas[mib[2]].lstats[mib[4]].nmalloc, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_lruns_j_ndalloc,
+    ctl_stats.arenas[mib[2]].lstats[mib[4]].ndalloc, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_lruns_j_nrequests,
+    ctl_stats.arenas[mib[2]].lstats[mib[4]].nrequests, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_lruns_j_curruns,
+    ctl_stats.arenas[mib[2]].lstats[mib[4]].curruns, size_t)
+
+static const ctl_named_node_t *
+stats_arenas_i_lruns_j_index(const size_t *mib, size_t miblen, size_t j)
+{
+
+	if (j > nlclasses)
+		return (NULL);
+	return (super_stats_arenas_i_lruns_j_node);
+}
+
+CTL_RO_GEN(stats_arenas_i_nthreads, ctl_stats.arenas[mib[2]].nthreads, unsigned)
+CTL_RO_GEN(stats_arenas_i_dss, ctl_stats.arenas[mib[2]].dss, const char *)
+CTL_RO_GEN(stats_arenas_i_pactive, ctl_stats.arenas[mib[2]].pactive, size_t)
+CTL_RO_GEN(stats_arenas_i_pdirty, ctl_stats.arenas[mib[2]].pdirty, size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_mapped,
+    ctl_stats.arenas[mib[2]].astats.mapped, size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_npurge,
+    ctl_stats.arenas[mib[2]].astats.npurge, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_nmadvise,
+    ctl_stats.arenas[mib[2]].astats.nmadvise, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_purged,
+    ctl_stats.arenas[mib[2]].astats.purged, uint64_t)
+
+static const ctl_named_node_t *
+stats_arenas_i_index(const size_t *mib, size_t miblen, size_t i)
+{
+	const ctl_named_node_t * ret;
+
+	malloc_mutex_lock(&ctl_mtx);
+	if (i > ctl_stats.narenas || ctl_stats.arenas[i].initialized == false) {
+		ret = NULL;
+		goto label_return;
+	}
+
+	ret = super_stats_arenas_i_node;
+label_return:
+	malloc_mutex_unlock(&ctl_mtx);
+	return (ret);
+}
+
+CTL_RO_CGEN(config_stats, stats_cactive, &stats_cactive, size_t *)
+CTL_RO_CGEN(config_stats, stats_allocated, ctl_stats.allocated, size_t)
+CTL_RO_CGEN(config_stats, stats_active, ctl_stats.active, size_t)
+CTL_RO_CGEN(config_stats, stats_mapped, ctl_stats.mapped, size_t)
diff --git a/extra/jemalloc/src/extent.c b/extra/jemalloc/src/extent.c
new file mode 100644
index 00000000000..8c09b486ed8
--- /dev/null
+++ b/extra/jemalloc/src/extent.c
@@ -0,0 +1,39 @@
+#define	JEMALLOC_EXTENT_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+
+static inline int
+extent_szad_comp(extent_node_t *a, extent_node_t *b)
+{
+	int ret;
+	size_t a_size = a->size;
+	size_t b_size = b->size;
+
+	ret = (a_size > b_size) - (a_size < b_size);
+	if (ret == 0) {
+		uintptr_t a_addr = (uintptr_t)a->addr;
+		uintptr_t b_addr = (uintptr_t)b->addr;
+
+		ret = (a_addr > b_addr) - (a_addr < b_addr);
+	}
+
+	return (ret);
+}
+
+/* Generate red-black tree functions. */
+rb_gen(, extent_tree_szad_, extent_tree_t, extent_node_t, link_szad,
+    extent_szad_comp)
+
+static inline int
+extent_ad_comp(extent_node_t *a, extent_node_t *b)
+{
+	uintptr_t a_addr = (uintptr_t)a->addr;
+	uintptr_t b_addr = (uintptr_t)b->addr;
+
+	return ((a_addr > b_addr) - (a_addr < b_addr));
+}
+
+/* Generate red-black tree functions. */
+rb_gen(, extent_tree_ad_, extent_tree_t, extent_node_t, link_ad,
+    extent_ad_comp)
diff --git a/extra/jemalloc/src/hash.c b/extra/jemalloc/src/hash.c
new file mode 100644
index 00000000000..cfa4da0275c
--- /dev/null
+++ b/extra/jemalloc/src/hash.c
@@ -0,0 +1,2 @@
+#define	JEMALLOC_HASH_C_
+#include "jemalloc/internal/jemalloc_internal.h"
diff --git a/extra/jemalloc/src/huge.c b/extra/jemalloc/src/huge.c
new file mode 100644
index 00000000000..aa08d43d362
--- /dev/null
+++ b/extra/jemalloc/src/huge.c
@@ -0,0 +1,313 @@
+#define	JEMALLOC_HUGE_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Data. */
+
+uint64_t	huge_nmalloc;
+uint64_t	huge_ndalloc;
+size_t		huge_allocated;
+
+malloc_mutex_t	huge_mtx;
+
+/******************************************************************************/
+
+/* Tree of chunks that are stand-alone huge allocations. */
+static extent_tree_t	huge;
+
+void *
+huge_malloc(size_t size, bool zero)
+{
+
+	return (huge_palloc(size, chunksize, zero));
+}
+
+void *
+huge_palloc(size_t size, size_t alignment, bool zero)
+{
+	void *ret;
+	size_t csize;
+	extent_node_t *node;
+	bool is_zeroed;
+
+	/* Allocate one or more contiguous chunks for this request. */
+
+	csize = CHUNK_CEILING(size);
+	if (csize == 0) {
+		/* size is large enough to cause size_t wrap-around. */
+		return (NULL);
+	}
+
+	/* Allocate an extent node with which to track the chunk. */
+	node = base_node_alloc();
+	if (node == NULL)
+		return (NULL);
+
+	/*
+	 * Copy zero into is_zeroed and pass the copy to chunk_alloc(), so that
+	 * it is possible to make correct junk/zero fill decisions below.
+	 */
+	is_zeroed = zero;
+	ret = chunk_alloc(csize, alignment, false, &is_zeroed,
+	    chunk_dss_prec_get());
+	if (ret == NULL) {
+		base_node_dealloc(node);
+		return (NULL);
+	}
+
+	/* Insert node into huge. */
+	node->addr = ret;
+	node->size = csize;
+
+	malloc_mutex_lock(&huge_mtx);
+	extent_tree_ad_insert(&huge, node);
+	if (config_stats) {
+		stats_cactive_add(csize);
+		huge_nmalloc++;
+		huge_allocated += csize;
+	}
+	malloc_mutex_unlock(&huge_mtx);
+
+	if (config_fill && zero == false) {
+		if (opt_junk)
+			memset(ret, 0xa5, csize);
+		else if (opt_zero && is_zeroed == false)
+			memset(ret, 0, csize);
+	}
+
+	return (ret);
+}
+
+void *
+huge_ralloc_no_move(void *ptr, size_t oldsize, size_t size, size_t extra)
+{
+
+	/*
+	 * Avoid moving the allocation if the size class can be left the same.
+	 */
+	if (oldsize > arena_maxclass
+	    && CHUNK_CEILING(oldsize) >= CHUNK_CEILING(size)
+	    && CHUNK_CEILING(oldsize) <= CHUNK_CEILING(size+extra)) {
+		assert(CHUNK_CEILING(oldsize) == oldsize);
+		if (config_fill && opt_junk && size < oldsize) {
+			memset((void *)((uintptr_t)ptr + size), 0x5a,
+			    oldsize - size);
+		}
+		return (ptr);
+	}
+
+	/* Reallocation would require a move. */
+	return (NULL);
+}
+
+void *
+huge_ralloc(void *ptr, size_t oldsize, size_t size, size_t extra,
+    size_t alignment, bool zero, bool try_tcache_dalloc)
+{
+	void *ret;
+	size_t copysize;
+
+	/* Try to avoid moving the allocation. */
+	ret = huge_ralloc_no_move(ptr, oldsize, size, extra);
+	if (ret != NULL)
+		return (ret);
+
+	/*
+	 * size and oldsize are different enough that we need to use a
+	 * different size class.  In that case, fall back to allocating new
+	 * space and copying.
+	 */
+	if (alignment > chunksize)
+		ret = huge_palloc(size + extra, alignment, zero);
+	else
+		ret = huge_malloc(size + extra, zero);
+
+	if (ret == NULL) {
+		if (extra == 0)
+			return (NULL);
+		/* Try again, this time without extra. */
+		if (alignment > chunksize)
+			ret = huge_palloc(size, alignment, zero);
+		else
+			ret = huge_malloc(size, zero);
+
+		if (ret == NULL)
+			return (NULL);
+	}
+
+	/*
+	 * Copy at most size bytes (not size+extra), since the caller has no
+	 * expectation that the extra bytes will be reliably preserved.
+	 */
+	copysize = (size < oldsize) ? size : oldsize;
+
+#ifdef JEMALLOC_MREMAP
+	/*
+	 * Use mremap(2) if this is a huge-->huge reallocation, and neither the
+	 * source nor the destination are in dss.
+	 */
+	if (oldsize >= chunksize && (config_dss == false || (chunk_in_dss(ptr)
+	    == false && chunk_in_dss(ret) == false))) {
+		size_t newsize = huge_salloc(ret);
+
+		/*
+		 * Remove ptr from the tree of huge allocations before
+		 * performing the remap operation, in order to avoid the
+		 * possibility of another thread acquiring that mapping before
+		 * this one removes it from the tree.
+		 */
+		huge_dalloc(ptr, false);
+		if (mremap(ptr, oldsize, newsize, MREMAP_MAYMOVE|MREMAP_FIXED,
+		    ret) == MAP_FAILED) {
+			/*
+			 * Assuming no chunk management bugs in the allocator,
+			 * the only documented way an error can occur here is
+			 * if the application changed the map type for a
+			 * portion of the old allocation.  This is firmly in
+			 * undefined behavior territory, so write a diagnostic
+			 * message, and optionally abort.
+			 */
+			char buf[BUFERROR_BUF];
+
+			buferror(buf, sizeof(buf));
+			malloc_printf("<jemalloc>: Error in mremap(): %s\n",
+			    buf);
+			if (opt_abort)
+				abort();
+			memcpy(ret, ptr, copysize);
+			chunk_dealloc_mmap(ptr, oldsize);
+		}
+	} else
+#endif
+	{
+		memcpy(ret, ptr, copysize);
+		iqallocx(ptr, try_tcache_dalloc);
+	}
+	return (ret);
+}
+
+void
+huge_dalloc(void *ptr, bool unmap)
+{
+	extent_node_t *node, key;
+
+	malloc_mutex_lock(&huge_mtx);
+
+	/* Extract from tree of huge allocations. */
+	key.addr = ptr;
+	node = extent_tree_ad_search(&huge, &key);
+	assert(node != NULL);
+	assert(node->addr == ptr);
+	extent_tree_ad_remove(&huge, node);
+
+	if (config_stats) {
+		stats_cactive_sub(node->size);
+		huge_ndalloc++;
+		huge_allocated -= node->size;
+	}
+
+	malloc_mutex_unlock(&huge_mtx);
+
+	if (unmap && config_fill && config_dss && opt_junk)
+		memset(node->addr, 0x5a, node->size);
+
+	chunk_dealloc(node->addr, node->size, unmap);
+
+	base_node_dealloc(node);
+}
+
+size_t
+huge_salloc(const void *ptr)
+{
+	size_t ret;
+	extent_node_t *node, key;
+
+	malloc_mutex_lock(&huge_mtx);
+
+	/* Extract from tree of huge allocations. */
+	key.addr = __DECONST(void *, ptr);
+	node = extent_tree_ad_search(&huge, &key);
+	assert(node != NULL);
+
+	ret = node->size;
+
+	malloc_mutex_unlock(&huge_mtx);
+
+	return (ret);
+}
+
+prof_ctx_t *
+huge_prof_ctx_get(const void *ptr)
+{
+	prof_ctx_t *ret;
+	extent_node_t *node, key;
+
+	malloc_mutex_lock(&huge_mtx);
+
+	/* Extract from tree of huge allocations. */
+	key.addr = __DECONST(void *, ptr);
+	node = extent_tree_ad_search(&huge, &key);
+	assert(node != NULL);
+
+	ret = node->prof_ctx;
+
+	malloc_mutex_unlock(&huge_mtx);
+
+	return (ret);
+}
+
+void
+huge_prof_ctx_set(const void *ptr, prof_ctx_t *ctx)
+{
+	extent_node_t *node, key;
+
+	malloc_mutex_lock(&huge_mtx);
+
+	/* Extract from tree of huge allocations. */
+	key.addr = __DECONST(void *, ptr);
+	node = extent_tree_ad_search(&huge, &key);
+	assert(node != NULL);
+
+	node->prof_ctx = ctx;
+
+	malloc_mutex_unlock(&huge_mtx);
+}
+
+bool
+huge_boot(void)
+{
+
+	/* Initialize chunks data. */
+	if (malloc_mutex_init(&huge_mtx))
+		return (true);
+	extent_tree_ad_new(&huge);
+
+	if (config_stats) {
+		huge_nmalloc = 0;
+		huge_ndalloc = 0;
+		huge_allocated = 0;
+	}
+
+	return (false);
+}
+
+void
+huge_prefork(void)
+{
+
+	malloc_mutex_prefork(&huge_mtx);
+}
+
+void
+huge_postfork_parent(void)
+{
+
+	malloc_mutex_postfork_parent(&huge_mtx);
+}
+
+void
+huge_postfork_child(void)
+{
+
+	malloc_mutex_postfork_child(&huge_mtx);
+}
diff --git a/extra/jemalloc/src/jemalloc.c b/extra/jemalloc/src/jemalloc.c
new file mode 100644
index 00000000000..bc350ed953b
--- /dev/null
+++ b/extra/jemalloc/src/jemalloc.c
@@ -0,0 +1,1868 @@
+#define	JEMALLOC_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Data. */
+
+malloc_tsd_data(, arenas, arena_t *, NULL)
+malloc_tsd_data(, thread_allocated, thread_allocated_t,
+    THREAD_ALLOCATED_INITIALIZER)
+
+/* Runtime configuration options. */
+const char	*je_malloc_conf;
+bool	opt_abort =
+#ifdef JEMALLOC_DEBUG
+    true
+#else
+    false
+#endif
+    ;
+bool	opt_junk =
+#if (defined(JEMALLOC_DEBUG) && defined(JEMALLOC_FILL))
+    true
+#else
+    false
+#endif
+    ;
+size_t	opt_quarantine = ZU(0);
+bool	opt_redzone = false;
+bool	opt_utrace = false;
+bool	opt_valgrind = false;
+bool	opt_xmalloc = false;
+bool	opt_zero = false;
+size_t	opt_narenas = 0;
+
+unsigned	ncpus;
+
+malloc_mutex_t		arenas_lock;
+arena_t			**arenas;
+unsigned		narenas_total;
+unsigned		narenas_auto;
+
+/* Set to true once the allocator has been initialized. */
+static bool		malloc_initialized = false;
+
+#ifdef JEMALLOC_THREADED_INIT
+/* Used to let the initializing thread recursively allocate. */
+#  define NO_INITIALIZER	((unsigned long)0)
+#  define INITIALIZER		pthread_self()
+#  define IS_INITIALIZER	(malloc_initializer == pthread_self())
+static pthread_t		malloc_initializer = NO_INITIALIZER;
+#else
+#  define NO_INITIALIZER	false
+#  define INITIALIZER		true
+#  define IS_INITIALIZER	malloc_initializer
+static bool			malloc_initializer = NO_INITIALIZER;
+#endif
+
+/* Used to avoid initialization races. */
+#ifdef _WIN32
+static malloc_mutex_t	init_lock;
+
+JEMALLOC_ATTR(constructor)
+static void WINAPI
+_init_init_lock(void)
+{
+
+	malloc_mutex_init(&init_lock);
+}
+
+#ifdef _MSC_VER
+#  pragma section(".CRT$XCU", read)
+JEMALLOC_SECTION(".CRT$XCU") JEMALLOC_ATTR(used)
+static const void (WINAPI *init_init_lock)(void) = _init_init_lock;
+#endif
+
+#else
+static malloc_mutex_t	init_lock = MALLOC_MUTEX_INITIALIZER;
+#endif
+
+typedef struct {
+	void	*p;	/* Input pointer (as in realloc(p, s)). */
+	size_t	s;	/* Request size. */
+	void	*r;	/* Result pointer. */
+} malloc_utrace_t;
+
+#ifdef JEMALLOC_UTRACE
+#  define UTRACE(a, b, c) do {						\
+	if (opt_utrace) {						\
+		int utrace_serrno = errno;				\
+		malloc_utrace_t ut;					\
+		ut.p = (a);						\
+		ut.s = (b);						\
+		ut.r = (c);						\
+		utrace(&ut, sizeof(ut));				\
+		errno = utrace_serrno;					\
+	}								\
+} while (0)
+#else
+#  define UTRACE(a, b, c)
+#endif
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static void	stats_print_atexit(void);
+static unsigned	malloc_ncpus(void);
+static bool	malloc_conf_next(char const **opts_p, char const **k_p,
+    size_t *klen_p, char const **v_p, size_t *vlen_p);
+static void	malloc_conf_error(const char *msg, const char *k, size_t klen,
+    const char *v, size_t vlen);
+static void	malloc_conf_init(void);
+static bool	malloc_init_hard(void);
+static int	imemalign(void **memptr, size_t alignment, size_t size,
+    size_t min_alignment);
+
+/******************************************************************************/
+/*
+ * Begin miscellaneous support functions.
+ */
+
+/* Create a new arena and insert it into the arenas array at index ind. */
+arena_t *
+arenas_extend(unsigned ind)
+{
+	arena_t *ret;
+
+	ret = (arena_t *)base_alloc(sizeof(arena_t));
+	if (ret != NULL && arena_new(ret, ind) == false) {
+		arenas[ind] = ret;
+		return (ret);
+	}
+	/* Only reached if there is an OOM error. */
+
+	/*
+	 * OOM here is quite inconvenient to propagate, since dealing with it
+	 * would require a check for failure in the fast path.  Instead, punt
+	 * by using arenas[0].  In practice, this is an extremely unlikely
+	 * failure.
+	 */
+	malloc_write("<jemalloc>: Error initializing arena\n");
+	if (opt_abort)
+		abort();
+
+	return (arenas[0]);
+}
+
+/* Slow path, called only by choose_arena(). */
+arena_t *
+choose_arena_hard(void)
+{
+	arena_t *ret;
+
+	if (narenas_auto > 1) {
+		unsigned i, choose, first_null;
+
+		choose = 0;
+		first_null = narenas_auto;
+		malloc_mutex_lock(&arenas_lock);
+		assert(arenas[0] != NULL);
+		for (i = 1; i < narenas_auto; i++) {
+			if (arenas[i] != NULL) {
+				/*
+				 * Choose the first arena that has the lowest
+				 * number of threads assigned to it.
+				 */
+				if (arenas[i]->nthreads <
+				    arenas[choose]->nthreads)
+					choose = i;
+			} else if (first_null == narenas_auto) {
+				/*
+				 * Record the index of the first uninitialized
+				 * arena, in case all extant arenas are in use.
+				 *
+				 * NB: It is possible for there to be
+				 * discontinuities in terms of initialized
+				 * versus uninitialized arenas, due to the
+				 * "thread.arena" mallctl.
+				 */
+				first_null = i;
+			}
+		}
+
+		if (arenas[choose]->nthreads == 0
+		    || first_null == narenas_auto) {
+			/*
+			 * Use an unloaded arena, or the least loaded arena if
+			 * all arenas are already initialized.
+			 */
+			ret = arenas[choose];
+		} else {
+			/* Initialize a new arena. */
+			ret = arenas_extend(first_null);
+		}
+		ret->nthreads++;
+		malloc_mutex_unlock(&arenas_lock);
+	} else {
+		ret = arenas[0];
+		malloc_mutex_lock(&arenas_lock);
+		ret->nthreads++;
+		malloc_mutex_unlock(&arenas_lock);
+	}
+
+	arenas_tsd_set(&ret);
+
+	return (ret);
+}
+
+static void
+stats_print_atexit(void)
+{
+
+	if (config_tcache && config_stats) {
+		unsigned narenas, i;
+
+		/*
+		 * Merge stats from extant threads.  This is racy, since
+		 * individual threads do not lock when recording tcache stats
+		 * events.  As a consequence, the final stats may be slightly
+		 * out of date by the time they are reported, if other threads
+		 * continue to allocate.
+		 */
+		for (i = 0, narenas = narenas_total_get(); i < narenas; i++) {
+			arena_t *arena = arenas[i];
+			if (arena != NULL) {
+				tcache_t *tcache;
+
+				/*
+				 * tcache_stats_merge() locks bins, so if any
+				 * code is introduced that acquires both arena
+				 * and bin locks in the opposite order,
+				 * deadlocks may result.
+				 */
+				malloc_mutex_lock(&arena->lock);
+				ql_foreach(tcache, &arena->tcache_ql, link) {
+					tcache_stats_merge(tcache, arena);
+				}
+				malloc_mutex_unlock(&arena->lock);
+			}
+		}
+	}
+	je_malloc_stats_print(NULL, NULL, NULL);
+}
+
+/*
+ * End miscellaneous support functions.
+ */
+/******************************************************************************/
+/*
+ * Begin initialization functions.
+ */
+
+static unsigned
+malloc_ncpus(void)
+{
+	unsigned ret;
+	long result;
+
+#ifdef _WIN32
+	SYSTEM_INFO si;
+	GetSystemInfo(&si);
+	result = si.dwNumberOfProcessors;
+#else
+	result = sysconf(_SC_NPROCESSORS_ONLN);
+#endif
+	if (result == -1) {
+		/* Error. */
+		ret = 1;
+	}  else {
+    ret = (unsigned)result;
+  }
+
+	return (ret);
+}
+
+void
+arenas_cleanup(void *arg)
+{
+	arena_t *arena = *(arena_t **)arg;
+
+	malloc_mutex_lock(&arenas_lock);
+	arena->nthreads--;
+	malloc_mutex_unlock(&arenas_lock);
+}
+
+static JEMALLOC_ATTR(always_inline) void
+malloc_thread_init(void)
+{
+
+	/*
+	 * TSD initialization can't be safely done as a side effect of
+	 * deallocation, because it is possible for a thread to do nothing but
+	 * deallocate its TLS data via free(), in which case writing to TLS
+	 * would cause write-after-free memory corruption.  The quarantine
+	 * facility *only* gets used as a side effect of deallocation, so make
+	 * a best effort attempt at initializing its TSD by hooking all
+	 * allocation events.
+	 */
+	if (config_fill && opt_quarantine)
+		quarantine_alloc_hook();
+}
+
+static JEMALLOC_ATTR(always_inline) bool
+malloc_init(void)
+{
+
+	if (malloc_initialized == false && malloc_init_hard())
+		return (true);
+	malloc_thread_init();
+
+	return (false);
+}
+
+static bool
+malloc_conf_next(char const **opts_p, char const **k_p, size_t *klen_p,
+    char const **v_p, size_t *vlen_p)
+{
+	bool accept;
+	const char *opts = *opts_p;
+
+	*k_p = opts;
+
+	for (accept = false; accept == false;) {
+		switch (*opts) {
+		case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
+		case 'G': case 'H': case 'I': case 'J': case 'K': case 'L':
+		case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R':
+		case 'S': case 'T': case 'U': case 'V': case 'W': case 'X':
+		case 'Y': case 'Z':
+		case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
+		case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
+		case 'm': case 'n': case 'o': case 'p': case 'q': case 'r':
+		case 's': case 't': case 'u': case 'v': case 'w': case 'x':
+		case 'y': case 'z':
+		case '0': case '1': case '2': case '3': case '4': case '5':
+		case '6': case '7': case '8': case '9':
+		case '_':
+			opts++;
+			break;
+		case ':':
+			opts++;
+			*klen_p = (uintptr_t)opts - 1 - (uintptr_t)*k_p;
+			*v_p = opts;
+			accept = true;
+			break;
+		case '\0':
+			if (opts != *opts_p) {
+				malloc_write("<jemalloc>: Conf string ends "
+				    "with key\n");
+			}
+			return (true);
+		default:
+			malloc_write("<jemalloc>: Malformed conf string\n");
+			return (true);
+		}
+	}
+
+	for (accept = false; accept == false;) {
+		switch (*opts) {
+		case ',':
+			opts++;
+			/*
+			 * Look ahead one character here, because the next time
+			 * this function is called, it will assume that end of
+			 * input has been cleanly reached if no input remains,
+			 * but we have optimistically already consumed the
+			 * comma if one exists.
+			 */
+			if (*opts == '\0') {
+				malloc_write("<jemalloc>: Conf string ends "
+				    "with comma\n");
+			}
+			*vlen_p = (uintptr_t)opts - 1 - (uintptr_t)*v_p;
+			accept = true;
+			break;
+		case '\0':
+			*vlen_p = (uintptr_t)opts - (uintptr_t)*v_p;
+			accept = true;
+			break;
+		default:
+			opts++;
+			break;
+		}
+	}
+
+	*opts_p = opts;
+	return (false);
+}
+
+static void
+malloc_conf_error(const char *msg, const char *k, size_t klen, const char *v,
+    size_t vlen)
+{
+
+	malloc_printf("<jemalloc>: %s: %.*s:%.*s\n", msg, (int)klen, k,
+	    (int)vlen, v);
+}
+
+static void
+malloc_conf_init(void)
+{
+	unsigned i;
+	char buf[PATH_MAX + 1];
+	const char *opts, *k, *v;
+	size_t klen, vlen;
+
+	/*
+	 * Automatically configure valgrind before processing options.  The
+	 * valgrind option remains in jemalloc 3.x for compatibility reasons.
+	 */
+	if (config_valgrind) {
+		opt_valgrind = (RUNNING_ON_VALGRIND != 0) ? true : false;
+		if (config_fill && opt_valgrind) {
+			opt_junk = false;
+			assert(opt_zero == false);
+			opt_quarantine = JEMALLOC_VALGRIND_QUARANTINE_DEFAULT;
+			opt_redzone = true;
+		}
+		if (config_tcache && opt_valgrind)
+			opt_tcache = false;
+	}
+
+	for (i = 0; i < 3; i++) {
+		/* Get runtime configuration. */
+		switch (i) {
+		case 0:
+			if (je_malloc_conf != NULL) {
+				/*
+				 * Use options that were compiled into the
+				 * program.
+				 */
+				opts = je_malloc_conf;
+			} else {
+				/* No configuration specified. */
+				buf[0] = '\0';
+				opts = buf;
+			}
+			break;
+		case 1: {
+#ifndef _WIN32
+			int linklen;
+			const char *linkname =
+#  ifdef JEMALLOC_PREFIX
+			    "/etc/"JEMALLOC_PREFIX"malloc.conf"
+#  else
+			    "/etc/malloc.conf"
+#  endif
+			    ;
+
+			if ((linklen = readlink(linkname, buf,
+			    sizeof(buf) - 1)) != -1) {
+				/*
+				 * Use the contents of the "/etc/malloc.conf"
+				 * symbolic link's name.
+				 */
+				buf[linklen] = '\0';
+				opts = buf;
+			} else
+#endif
+			{
+				/* No configuration specified. */
+				buf[0] = '\0';
+				opts = buf;
+			}
+			break;
+		} case 2: {
+			const char *envname =
+#ifdef JEMALLOC_PREFIX
+			    JEMALLOC_CPREFIX"MALLOC_CONF"
+#else
+			    "MALLOC_CONF"
+#endif
+			    ;
+
+			if ((opts = getenv(envname)) != NULL) {
+				/*
+				 * Do nothing; opts is already initialized to
+				 * the value of the MALLOC_CONF environment
+				 * variable.
+				 */
+			} else {
+				/* No configuration specified. */
+				buf[0] = '\0';
+				opts = buf;
+			}
+			break;
+		} default:
+			/* NOTREACHED */
+			assert(false);
+			buf[0] = '\0';
+			opts = buf;
+		}
+
+		while (*opts != '\0' && malloc_conf_next(&opts, &k, &klen, &v,
+		    &vlen) == false) {
+#define	CONF_HANDLE_BOOL(o, n)						\
+			if (sizeof(n)-1 == klen && strncmp(n, k,	\
+			    klen) == 0) {				\
+				if (strncmp("true", v, vlen) == 0 &&	\
+				    vlen == sizeof("true")-1)		\
+					o = true;			\
+				else if (strncmp("false", v, vlen) ==	\
+				    0 && vlen == sizeof("false")-1)	\
+					o = false;			\
+				else {					\
+					malloc_conf_error(		\
+					    "Invalid conf value",	\
+					    k, klen, v, vlen);		\
+				}					\
+				continue;				\
+			}
+#define	CONF_HANDLE_SIZE_T(o, n, min, max, clip)			\
+			if (sizeof(n)-1 == klen && strncmp(n, k,	\
+			    klen) == 0) {				\
+				uintmax_t um;				\
+				char *end;				\
+									\
+				set_errno(0);				\
+				um = malloc_strtoumax(v, &end, 0);	\
+				if (get_errno() != 0 || (uintptr_t)end -\
+				    (uintptr_t)v != vlen) {		\
+					malloc_conf_error(		\
+					    "Invalid conf value",	\
+					    k, klen, v, vlen);		\
+				} else if (clip) {			\
+					if (um < min)			\
+						o = min;		\
+					else if (um > max)		\
+						o = max;		\
+					else				\
+						o = um;			\
+				} else {				\
+					if (um < min || um > max) {	\
+						malloc_conf_error(	\
+						    "Out-of-range "	\
+						    "conf value",	\
+						    k, klen, v, vlen);	\
+					} else				\
+						o = um;			\
+				}					\
+				continue;				\
+			}
+#define	CONF_HANDLE_SSIZE_T(o, n, min, max)				\
+			if (sizeof(n)-1 == klen && strncmp(n, k,	\
+			    klen) == 0) {				\
+				long l;					\
+				char *end;				\
+									\
+				set_errno(0);				\
+				l = strtol(v, &end, 0);			\
+				if (get_errno() != 0 || (uintptr_t)end -\
+				    (uintptr_t)v != vlen) {		\
+					malloc_conf_error(		\
+					    "Invalid conf value",	\
+					    k, klen, v, vlen);		\
+				} else if (l < (ssize_t)min || l >	\
+				    (ssize_t)max) {			\
+					malloc_conf_error(		\
+					    "Out-of-range conf value",	\
+					    k, klen, v, vlen);		\
+				} else					\
+					o = l;				\
+				continue;				\
+			}
+#define	CONF_HANDLE_CHAR_P(o, n, d)					\
+			if (sizeof(n)-1 == klen && strncmp(n, k,	\
+			    klen) == 0) {				\
+				size_t cpylen = (vlen <=		\
+				    sizeof(o)-1) ? vlen :		\
+				    sizeof(o)-1;			\
+				strncpy(o, v, cpylen);			\
+				o[cpylen] = '\0';			\
+				continue;				\
+			}
+
+			CONF_HANDLE_BOOL(opt_abort, "abort")
+			/*
+			 * Chunks always require at least one header page, plus
+			 * one data page in the absence of redzones, or three
+			 * pages in the presence of redzones.  In order to
+			 * simplify options processing, fix the limit based on
+			 * config_fill.
+			 */
+			CONF_HANDLE_SIZE_T(opt_lg_chunk, "lg_chunk", LG_PAGE +
+			    (config_fill ? 2 : 1), (sizeof(size_t) << 3) - 1,
+			    true)
+			if (strncmp("dss", k, klen) == 0) {
+				int i;
+				bool match = false;
+				for (i = 0; i < dss_prec_limit; i++) {
+					if (strncmp(dss_prec_names[i], v, vlen)
+					    == 0) {
+						if (chunk_dss_prec_set(i)) {
+							malloc_conf_error(
+							    "Error setting dss",
+							    k, klen, v, vlen);
+						} else {
+							opt_dss =
+							    dss_prec_names[i];
+							match = true;
+							break;
+						}
+					}
+				}
+				if (match == false) {
+					malloc_conf_error("Invalid conf value",
+					    k, klen, v, vlen);
+				}
+				continue;
+			}
+			CONF_HANDLE_SIZE_T(opt_narenas, "narenas", 1,
+			    SIZE_T_MAX, false)
+			CONF_HANDLE_SSIZE_T(opt_lg_dirty_mult, "lg_dirty_mult",
+			    -1, (sizeof(size_t) << 3) - 1)
+			CONF_HANDLE_BOOL(opt_stats_print, "stats_print")
+			if (config_fill) {
+				CONF_HANDLE_BOOL(opt_junk, "junk")
+				CONF_HANDLE_SIZE_T(opt_quarantine, "quarantine",
+				    0, SIZE_T_MAX, false)
+				CONF_HANDLE_BOOL(opt_redzone, "redzone")
+				CONF_HANDLE_BOOL(opt_zero, "zero")
+			}
+			if (config_utrace) {
+				CONF_HANDLE_BOOL(opt_utrace, "utrace")
+			}
+			if (config_valgrind) {
+				CONF_HANDLE_BOOL(opt_valgrind, "valgrind")
+			}
+			if (config_xmalloc) {
+				CONF_HANDLE_BOOL(opt_xmalloc, "xmalloc")
+			}
+			if (config_tcache) {
+				CONF_HANDLE_BOOL(opt_tcache, "tcache")
+				CONF_HANDLE_SSIZE_T(opt_lg_tcache_max,
+				    "lg_tcache_max", -1,
+				    (sizeof(size_t) << 3) - 1)
+			}
+			if (config_prof) {
+				CONF_HANDLE_BOOL(opt_prof, "prof")
+				CONF_HANDLE_CHAR_P(opt_prof_prefix,
+				    "prof_prefix", "jeprof")
+				CONF_HANDLE_BOOL(opt_prof_active, "prof_active")
+				CONF_HANDLE_SSIZE_T(opt_lg_prof_sample,
+				    "lg_prof_sample", 0,
+				    (sizeof(uint64_t) << 3) - 1)
+				CONF_HANDLE_BOOL(opt_prof_accum, "prof_accum")
+				CONF_HANDLE_SSIZE_T(opt_lg_prof_interval,
+				    "lg_prof_interval", -1,
+				    (sizeof(uint64_t) << 3) - 1)
+				CONF_HANDLE_BOOL(opt_prof_gdump, "prof_gdump")
+				CONF_HANDLE_BOOL(opt_prof_final, "prof_final")
+				CONF_HANDLE_BOOL(opt_prof_leak, "prof_leak")
+			}
+			malloc_conf_error("Invalid conf pair", k, klen, v,
+			    vlen);
+#undef CONF_HANDLE_BOOL
+#undef CONF_HANDLE_SIZE_T
+#undef CONF_HANDLE_SSIZE_T
+#undef CONF_HANDLE_CHAR_P
+		}
+	}
+}
+
+static bool
+malloc_init_hard(void)
+{
+	arena_t *init_arenas[1];
+
+	malloc_mutex_lock(&init_lock);
+	if (malloc_initialized || IS_INITIALIZER) {
+		/*
+		 * Another thread initialized the allocator before this one
+		 * acquired init_lock, or this thread is the initializing
+		 * thread, and it is recursively allocating.
+		 */
+		malloc_mutex_unlock(&init_lock);
+		return (false);
+	}
+#ifdef JEMALLOC_THREADED_INIT
+	if (malloc_initializer != NO_INITIALIZER && IS_INITIALIZER == false) {
+		/* Busy-wait until the initializing thread completes. */
+		do {
+			malloc_mutex_unlock(&init_lock);
+			CPU_SPINWAIT;
+			malloc_mutex_lock(&init_lock);
+		} while (malloc_initialized == false);
+		malloc_mutex_unlock(&init_lock);
+		return (false);
+	}
+#endif
+	malloc_initializer = INITIALIZER;
+
+	malloc_tsd_boot();
+	if (config_prof)
+		prof_boot0();
+
+	malloc_conf_init();
+
+#if (!defined(JEMALLOC_MUTEX_INIT_CB) && !defined(JEMALLOC_ZONE) \
+    && !defined(_WIN32))
+	/* Register fork handlers. */
+	if (pthread_atfork(jemalloc_prefork, jemalloc_postfork_parent,
+	    jemalloc_postfork_child) != 0) {
+		malloc_write("<jemalloc>: Error in pthread_atfork()\n");
+		if (opt_abort)
+			abort();
+	}
+#endif
+
+	if (opt_stats_print) {
+		/* Print statistics at exit. */
+		if (atexit(stats_print_atexit) != 0) {
+			malloc_write("<jemalloc>: Error in atexit()\n");
+			if (opt_abort)
+				abort();
+		}
+	}
+
+	if (base_boot()) {
+		malloc_mutex_unlock(&init_lock);
+		return (true);
+	}
+
+	if (chunk_boot()) {
+		malloc_mutex_unlock(&init_lock);
+		return (true);
+	}
+
+	if (ctl_boot()) {
+		malloc_mutex_unlock(&init_lock);
+		return (true);
+	}
+
+	if (config_prof)
+		prof_boot1();
+
+	arena_boot();
+
+	if (config_tcache && tcache_boot0()) {
+		malloc_mutex_unlock(&init_lock);
+		return (true);
+	}
+
+	if (huge_boot()) {
+		malloc_mutex_unlock(&init_lock);
+		return (true);
+	}
+
+	if (malloc_mutex_init(&arenas_lock))
+		return (true);
+
+	/*
+	 * Create enough scaffolding to allow recursive allocation in
+	 * malloc_ncpus().
+	 */
+	narenas_total = narenas_auto = 1;
+	arenas = init_arenas;
+	memset(arenas, 0, sizeof(arena_t *) * narenas_auto);
+
+	/*
+	 * Initialize one arena here.  The rest are lazily created in
+	 * choose_arena_hard().
+	 */
+	arenas_extend(0);
+	if (arenas[0] == NULL) {
+		malloc_mutex_unlock(&init_lock);
+		return (true);
+	}
+
+	/* Initialize allocation counters before any allocations can occur. */
+	if (config_stats && thread_allocated_tsd_boot()) {
+		malloc_mutex_unlock(&init_lock);
+		return (true);
+	}
+
+	if (arenas_tsd_boot()) {
+		malloc_mutex_unlock(&init_lock);
+		return (true);
+	}
+
+	if (config_tcache && tcache_boot1()) {
+		malloc_mutex_unlock(&init_lock);
+		return (true);
+	}
+
+	if (config_fill && quarantine_boot()) {
+		malloc_mutex_unlock(&init_lock);
+		return (true);
+	}
+
+	if (config_prof && prof_boot2()) {
+		malloc_mutex_unlock(&init_lock);
+		return (true);
+	}
+
+	/* Get number of CPUs. */
+	malloc_mutex_unlock(&init_lock);
+	ncpus = malloc_ncpus();
+	malloc_mutex_lock(&init_lock);
+
+	if (mutex_boot()) {
+		malloc_mutex_unlock(&init_lock);
+		return (true);
+	}
+
+	if (opt_narenas == 0) {
+		/*
+		 * For SMP systems, create more than one arena per CPU by
+		 * default.
+		 */
+		if (ncpus > 1)
+			opt_narenas = ncpus << 2;
+		else
+			opt_narenas = 1;
+	}
+	narenas_auto = opt_narenas;
+	/*
+	 * Make sure that the arenas array can be allocated.  In practice, this
+	 * limit is enough to allow the allocator to function, but the ctl
+	 * machinery will fail to allocate memory at far lower limits.
+	 */
+	if (narenas_auto > chunksize / sizeof(arena_t *)) {
+		narenas_auto = chunksize / sizeof(arena_t *);
+		malloc_printf("<jemalloc>: Reducing narenas to limit (%d)\n",
+		    narenas_auto);
+	}
+	narenas_total = narenas_auto;
+
+	/* Allocate and initialize arenas. */
+	arenas = (arena_t **)base_alloc(sizeof(arena_t *) * narenas_total);
+	if (arenas == NULL) {
+		malloc_mutex_unlock(&init_lock);
+		return (true);
+	}
+	/*
+	 * Zero the array.  In practice, this should always be pre-zeroed,
+	 * since it was just mmap()ed, but let's be sure.
+	 */
+	memset(arenas, 0, sizeof(arena_t *) * narenas_total);
+	/* Copy the pointer to the one arena that was already initialized. */
+	arenas[0] = init_arenas[0];
+
+	malloc_initialized = true;
+	malloc_mutex_unlock(&init_lock);
+	return (false);
+}
+
+/*
+ * End initialization functions.
+ */
+/******************************************************************************/
+/*
+ * Begin malloc(3)-compatible functions.
+ */
+
+void *
+je_malloc(size_t size)
+{
+	void *ret;
+	size_t usize JEMALLOC_CC_SILENCE_INIT(0);
+	prof_thr_cnt_t *cnt JEMALLOC_CC_SILENCE_INIT(NULL);
+
+	if (malloc_init()) {
+		ret = NULL;
+		goto label_oom;
+	}
+
+	if (size == 0)
+		size = 1;
+
+	if (config_prof && opt_prof) {
+		usize = s2u(size);
+		PROF_ALLOC_PREP(1, usize, cnt);
+		if (cnt == NULL) {
+			ret = NULL;
+			goto label_oom;
+		}
+		if (prof_promote && (uintptr_t)cnt != (uintptr_t)1U && usize <=
+		    SMALL_MAXCLASS) {
+			ret = imalloc(SMALL_MAXCLASS+1);
+			if (ret != NULL)
+				arena_prof_promoted(ret, usize);
+		} else
+			ret = imalloc(size);
+	} else {
+		if (config_stats || (config_valgrind && opt_valgrind))
+			usize = s2u(size);
+		ret = imalloc(size);
+	}
+
+label_oom:
+	if (ret == NULL) {
+		if (config_xmalloc && opt_xmalloc) {
+			malloc_write("<jemalloc>: Error in malloc(): "
+			    "out of memory\n");
+			abort();
+		}
+		set_errno(ENOMEM);
+	}
+	if (config_prof && opt_prof && ret != NULL)
+		prof_malloc(ret, usize, cnt);
+	if (config_stats && ret != NULL) {
+		assert(usize == isalloc(ret, config_prof));
+		thread_allocated_tsd_get()->allocated += usize;
+	}
+	UTRACE(0, size, ret);
+	JEMALLOC_VALGRIND_MALLOC(ret != NULL, ret, usize, false);
+	return (ret);
+}
+
+JEMALLOC_ATTR(nonnull(1))
+#ifdef JEMALLOC_PROF
+/*
+ * Avoid any uncertainty as to how many backtrace frames to ignore in
+ * PROF_ALLOC_PREP().
+ */
+JEMALLOC_NOINLINE
+#endif
+static int
+imemalign(void **memptr, size_t alignment, size_t size,
+    size_t min_alignment)
+{
+	int ret;
+	size_t usize;
+	void *result;
+	prof_thr_cnt_t *cnt JEMALLOC_CC_SILENCE_INIT(NULL);
+
+	assert(min_alignment != 0);
+
+	if (malloc_init())
+		result = NULL;
+	else {
+		if (size == 0)
+			size = 1;
+
+		/* Make sure that alignment is a large enough power of 2. */
+		if (((alignment - 1) & alignment) != 0
+		    || (alignment < min_alignment)) {
+			if (config_xmalloc && opt_xmalloc) {
+				malloc_write("<jemalloc>: Error allocating "
+				    "aligned memory: invalid alignment\n");
+				abort();
+			}
+			result = NULL;
+			ret = EINVAL;
+			goto label_return;
+		}
+
+		usize = sa2u(size, alignment);
+		if (usize == 0) {
+			result = NULL;
+			ret = ENOMEM;
+			goto label_return;
+		}
+
+		if (config_prof && opt_prof) {
+			PROF_ALLOC_PREP(2, usize, cnt);
+			if (cnt == NULL) {
+				result = NULL;
+				ret = EINVAL;
+			} else {
+				if (prof_promote && (uintptr_t)cnt !=
+				    (uintptr_t)1U && usize <= SMALL_MAXCLASS) {
+					assert(sa2u(SMALL_MAXCLASS+1,
+					    alignment) != 0);
+					result = ipalloc(sa2u(SMALL_MAXCLASS+1,
+					    alignment), alignment, false);
+					if (result != NULL) {
+						arena_prof_promoted(result,
+						    usize);
+					}
+				} else {
+					result = ipalloc(usize, alignment,
+					    false);
+				}
+			}
+		} else
+			result = ipalloc(usize, alignment, false);
+	}
+
+	if (result == NULL) {
+		if (config_xmalloc && opt_xmalloc) {
+			malloc_write("<jemalloc>: Error allocating aligned "
+			    "memory: out of memory\n");
+			abort();
+		}
+		ret = ENOMEM;
+		goto label_return;
+	}
+
+	*memptr = result;
+	ret = 0;
+
+label_return:
+	if (config_stats && result != NULL) {
+		assert(usize == isalloc(result, config_prof));
+		thread_allocated_tsd_get()->allocated += usize;
+	}
+	if (config_prof && opt_prof && result != NULL)
+		prof_malloc(result, usize, cnt);
+	UTRACE(0, size, result);
+	return (ret);
+}
+
+int
+je_posix_memalign(void **memptr, size_t alignment, size_t size)
+{
+	int ret = imemalign(memptr, alignment, size, sizeof(void *));
+	JEMALLOC_VALGRIND_MALLOC(ret == 0, *memptr, isalloc(*memptr,
+	    config_prof), false);
+	return (ret);
+}
+
+void *
+je_aligned_alloc(size_t alignment, size_t size)
+{
+	void *ret;
+	int err;
+
+	if ((err = imemalign(&ret, alignment, size, 1)) != 0) {
+		ret = NULL;
+		set_errno(err);
+	}
+	JEMALLOC_VALGRIND_MALLOC(err == 0, ret, isalloc(ret, config_prof),
+	    false);
+	return (ret);
+}
+
+void *
+je_calloc(size_t num, size_t size)
+{
+	void *ret;
+	size_t num_size;
+	size_t usize JEMALLOC_CC_SILENCE_INIT(0);
+	prof_thr_cnt_t *cnt JEMALLOC_CC_SILENCE_INIT(NULL);
+
+	if (malloc_init()) {
+		num_size = 0;
+		ret = NULL;
+		goto label_return;
+	}
+
+	num_size = num * size;
+	if (num_size == 0) {
+		if (num == 0 || size == 0)
+			num_size = 1;
+		else {
+			ret = NULL;
+			goto label_return;
+		}
+	/*
+	 * Try to avoid division here.  We know that it isn't possible to
+	 * overflow during multiplication if neither operand uses any of the
+	 * most significant half of the bits in a size_t.
+	 */
+	} else if (((num | size) & (SIZE_T_MAX << (sizeof(size_t) << 2)))
+	    && (num_size / size != num)) {
+		/* size_t overflow. */
+		ret = NULL;
+		goto label_return;
+	}
+
+	if (config_prof && opt_prof) {
+		usize = s2u(num_size);
+		PROF_ALLOC_PREP(1, usize, cnt);
+		if (cnt == NULL) {
+			ret = NULL;
+			goto label_return;
+		}
+		if (prof_promote && (uintptr_t)cnt != (uintptr_t)1U && usize
+		    <= SMALL_MAXCLASS) {
+			ret = icalloc(SMALL_MAXCLASS+1);
+			if (ret != NULL)
+				arena_prof_promoted(ret, usize);
+		} else
+			ret = icalloc(num_size);
+	} else {
+		if (config_stats || (config_valgrind && opt_valgrind))
+			usize = s2u(num_size);
+		ret = icalloc(num_size);
+	}
+
+label_return:
+	if (ret == NULL) {
+		if (config_xmalloc && opt_xmalloc) {
+			malloc_write("<jemalloc>: Error in calloc(): out of "
+			    "memory\n");
+			abort();
+		}
+		set_errno(ENOMEM);
+	}
+
+	if (config_prof && opt_prof && ret != NULL)
+		prof_malloc(ret, usize, cnt);
+	if (config_stats && ret != NULL) {
+		assert(usize == isalloc(ret, config_prof));
+		thread_allocated_tsd_get()->allocated += usize;
+	}
+	UTRACE(0, num_size, ret);
+	JEMALLOC_VALGRIND_MALLOC(ret != NULL, ret, usize, true);
+	return (ret);
+}
+
+void *
+je_realloc(void *ptr, size_t size)
+{
+	void *ret;
+	size_t usize JEMALLOC_CC_SILENCE_INIT(0);
+	size_t old_size = 0;
+	size_t old_rzsize JEMALLOC_CC_SILENCE_INIT(0);
+	prof_thr_cnt_t *cnt JEMALLOC_CC_SILENCE_INIT(NULL);
+	prof_ctx_t *old_ctx JEMALLOC_CC_SILENCE_INIT(NULL);
+
+	if (size == 0) {
+		if (ptr != NULL) {
+			/* realloc(ptr, 0) is equivalent to free(p). */
+			assert(malloc_initialized || IS_INITIALIZER);
+			if (config_prof) {
+				old_size = isalloc(ptr, true);
+				if (config_valgrind && opt_valgrind)
+					old_rzsize = p2rz(ptr);
+			} else if (config_stats) {
+				old_size = isalloc(ptr, false);
+				if (config_valgrind && opt_valgrind)
+					old_rzsize = u2rz(old_size);
+			} else if (config_valgrind && opt_valgrind) {
+				old_size = isalloc(ptr, false);
+				old_rzsize = u2rz(old_size);
+			}
+			if (config_prof && opt_prof) {
+				old_ctx = prof_ctx_get(ptr);
+				cnt = NULL;
+			}
+			iqalloc(ptr);
+			ret = NULL;
+			goto label_return;
+		} else
+			size = 1;
+	}
+
+	if (ptr != NULL) {
+		assert(malloc_initialized || IS_INITIALIZER);
+		malloc_thread_init();
+
+		if (config_prof) {
+			old_size = isalloc(ptr, true);
+			if (config_valgrind && opt_valgrind)
+				old_rzsize = p2rz(ptr);
+		} else if (config_stats) {
+			old_size = isalloc(ptr, false);
+			if (config_valgrind && opt_valgrind)
+				old_rzsize = u2rz(old_size);
+		} else if (config_valgrind && opt_valgrind) {
+			old_size = isalloc(ptr, false);
+			old_rzsize = u2rz(old_size);
+		}
+		if (config_prof && opt_prof) {
+			usize = s2u(size);
+			old_ctx = prof_ctx_get(ptr);
+			PROF_ALLOC_PREP(1, usize, cnt);
+			if (cnt == NULL) {
+				old_ctx = NULL;
+				ret = NULL;
+				goto label_oom;
+			}
+			if (prof_promote && (uintptr_t)cnt != (uintptr_t)1U &&
+			    usize <= SMALL_MAXCLASS) {
+				ret = iralloc(ptr, SMALL_MAXCLASS+1, 0, 0,
+				    false, false);
+				if (ret != NULL)
+					arena_prof_promoted(ret, usize);
+				else
+					old_ctx = NULL;
+			} else {
+				ret = iralloc(ptr, size, 0, 0, false, false);
+				if (ret == NULL)
+					old_ctx = NULL;
+			}
+		} else {
+			if (config_stats || (config_valgrind && opt_valgrind))
+				usize = s2u(size);
+			ret = iralloc(ptr, size, 0, 0, false, false);
+		}
+
+label_oom:
+		if (ret == NULL) {
+			if (config_xmalloc && opt_xmalloc) {
+				malloc_write("<jemalloc>: Error in realloc(): "
+				    "out of memory\n");
+				abort();
+			}
+			set_errno(ENOMEM);
+		}
+	} else {
+		/* realloc(NULL, size) is equivalent to malloc(size). */
+		if (config_prof && opt_prof)
+			old_ctx = NULL;
+		if (malloc_init()) {
+			if (config_prof && opt_prof)
+				cnt = NULL;
+			ret = NULL;
+		} else {
+			if (config_prof && opt_prof) {
+				usize = s2u(size);
+				PROF_ALLOC_PREP(1, usize, cnt);
+				if (cnt == NULL)
+					ret = NULL;
+				else {
+					if (prof_promote && (uintptr_t)cnt !=
+					    (uintptr_t)1U && usize <=
+					    SMALL_MAXCLASS) {
+						ret = imalloc(SMALL_MAXCLASS+1);
+						if (ret != NULL) {
+							arena_prof_promoted(ret,
+							    usize);
+						}
+					} else
+						ret = imalloc(size);
+				}
+			} else {
+				if (config_stats || (config_valgrind &&
+				    opt_valgrind))
+					usize = s2u(size);
+				ret = imalloc(size);
+			}
+		}
+
+		if (ret == NULL) {
+			if (config_xmalloc && opt_xmalloc) {
+				malloc_write("<jemalloc>: Error in realloc(): "
+				    "out of memory\n");
+				abort();
+			}
+			set_errno(ENOMEM);
+		}
+	}
+
+label_return:
+	if (config_prof && opt_prof)
+		prof_realloc(ret, usize, cnt, old_size, old_ctx);
+	if (config_stats && ret != NULL) {
+		thread_allocated_t *ta;
+		assert(usize == isalloc(ret, config_prof));
+		ta = thread_allocated_tsd_get();
+		ta->allocated += usize;
+		ta->deallocated += old_size;
+	}
+	UTRACE(ptr, size, ret);
+	JEMALLOC_VALGRIND_REALLOC(ret, usize, ptr, old_size, old_rzsize, false);
+	return (ret);
+}
+
+void
+je_free(void *ptr)
+{
+
+	UTRACE(ptr, 0, 0);
+	if (ptr != NULL) {
+		size_t usize;
+		size_t rzsize JEMALLOC_CC_SILENCE_INIT(0);
+
+		assert(malloc_initialized || IS_INITIALIZER);
+
+		if (config_prof && opt_prof) {
+			usize = isalloc(ptr, config_prof);
+			prof_free(ptr, usize);
+		} else if (config_stats || config_valgrind)
+			usize = isalloc(ptr, config_prof);
+		if (config_stats)
+			thread_allocated_tsd_get()->deallocated += usize;
+		if (config_valgrind && opt_valgrind)
+			rzsize = p2rz(ptr);
+		iqalloc(ptr);
+		JEMALLOC_VALGRIND_FREE(ptr, rzsize);
+	}
+}
+
+/*
+ * End malloc(3)-compatible functions.
+ */
+/******************************************************************************/
+/*
+ * Begin non-standard override functions.
+ */
+
+#ifdef JEMALLOC_OVERRIDE_MEMALIGN
+void *
+je_memalign(size_t alignment, size_t size)
+{
+	void *ret JEMALLOC_CC_SILENCE_INIT(NULL);
+	imemalign(&ret, alignment, size, 1);
+	JEMALLOC_VALGRIND_MALLOC(ret != NULL, ret, size, false);
+	return (ret);
+}
+#endif
+
+#ifdef JEMALLOC_OVERRIDE_VALLOC
+void *
+je_valloc(size_t size)
+{
+	void *ret JEMALLOC_CC_SILENCE_INIT(NULL);
+	imemalign(&ret, PAGE, size, 1);
+	JEMALLOC_VALGRIND_MALLOC(ret != NULL, ret, size, false);
+	return (ret);
+}
+#endif
+
+/*
+ * is_malloc(je_malloc) is some macro magic to detect if jemalloc_defs.h has
+ * #define je_malloc malloc
+ */
+#define	malloc_is_malloc 1
+#define	is_malloc_(a) malloc_is_ ## a
+#define	is_malloc(a) is_malloc_(a)
+
+#if ((is_malloc(je_malloc) == 1) && defined(__GLIBC__) && !defined(__UCLIBC__))
+/*
+ * glibc provides the RTLD_DEEPBIND flag for dlopen which can make it possible
+ * to inconsistently reference libc's malloc(3)-compatible functions
+ * (https://bugzilla.mozilla.org/show_bug.cgi?id=493541).
+ *
+ * These definitions interpose hooks in glibc.  The functions are actually
+ * passed an extra argument for the caller return address, which will be
+ * ignored.
+ */
+JEMALLOC_EXPORT void (* __free_hook)(void *ptr) = je_free;
+JEMALLOC_EXPORT void *(* __malloc_hook)(size_t size) = je_malloc;
+JEMALLOC_EXPORT void *(* __realloc_hook)(void *ptr, size_t size) = je_realloc;
+JEMALLOC_EXPORT void *(* __memalign_hook)(size_t alignment, size_t size) =
+    je_memalign;
+#endif
+
+/*
+ * End non-standard override functions.
+ */
+/******************************************************************************/
+/*
+ * Begin non-standard functions.
+ */
+
+size_t
+je_malloc_usable_size(JEMALLOC_USABLE_SIZE_CONST void *ptr)
+{
+	size_t ret;
+
+	assert(malloc_initialized || IS_INITIALIZER);
+	malloc_thread_init();
+
+	if (config_ivsalloc)
+		ret = ivsalloc(ptr, config_prof);
+	else
+		ret = (ptr != NULL) ? isalloc(ptr, config_prof) : 0;
+
+	return (ret);
+}
+
+void
+je_malloc_stats_print(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *opts)
+{
+
+	stats_print(write_cb, cbopaque, opts);
+}
+
+int
+je_mallctl(const char *name, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen)
+{
+
+	if (malloc_init())
+		return (EAGAIN);
+
+	return (ctl_byname(name, oldp, oldlenp, newp, newlen));
+}
+
+int
+je_mallctlnametomib(const char *name, size_t *mibp, size_t *miblenp)
+{
+
+	if (malloc_init())
+		return (EAGAIN);
+
+	return (ctl_nametomib(name, mibp, miblenp));
+}
+
+int
+je_mallctlbymib(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,
+  void *newp, size_t newlen)
+{
+
+	if (malloc_init())
+		return (EAGAIN);
+
+	return (ctl_bymib(mib, miblen, oldp, oldlenp, newp, newlen));
+}
+
+/*
+ * End non-standard functions.
+ */
+/******************************************************************************/
+/*
+ * Begin experimental functions.
+ */
+#ifdef JEMALLOC_EXPERIMENTAL
+
+static JEMALLOC_ATTR(always_inline) void *
+iallocm(size_t usize, size_t alignment, bool zero, bool try_tcache,
+    arena_t *arena)
+{
+
+	assert(usize == ((alignment == 0) ? s2u(usize) : sa2u(usize,
+	    alignment)));
+
+	if (alignment != 0)
+		return (ipallocx(usize, alignment, zero, try_tcache, arena));
+	else if (zero)
+		return (icallocx(usize, try_tcache, arena));
+	else
+		return (imallocx(usize, try_tcache, arena));
+}
+
+int
+je_allocm(void **ptr, size_t *rsize, size_t size, int flags)
+{
+	void *p;
+	size_t usize;
+	size_t alignment = (ZU(1) << (flags & ALLOCM_LG_ALIGN_MASK)
+	    & (SIZE_T_MAX-1));
+	bool zero = flags & ALLOCM_ZERO;
+	unsigned arena_ind = ((unsigned)(flags >> 8)) - 1;
+	arena_t *arena;
+	bool try_tcache;
+
+	assert(ptr != NULL);
+	assert(size != 0);
+
+	if (malloc_init())
+		goto label_oom;
+
+	if (arena_ind != UINT_MAX) {
+		arena = arenas[arena_ind];
+		try_tcache = false;
+	} else {
+		arena = NULL;
+		try_tcache = true;
+	}
+
+	usize = (alignment == 0) ? s2u(size) : sa2u(size, alignment);
+	if (usize == 0)
+		goto label_oom;
+
+	if (config_prof && opt_prof) {
+		prof_thr_cnt_t *cnt;
+
+		PROF_ALLOC_PREP(1, usize, cnt);
+		if (cnt == NULL)
+			goto label_oom;
+		if (prof_promote && (uintptr_t)cnt != (uintptr_t)1U && usize <=
+		    SMALL_MAXCLASS) {
+			size_t usize_promoted = (alignment == 0) ?
+			    s2u(SMALL_MAXCLASS+1) : sa2u(SMALL_MAXCLASS+1,
+			    alignment);
+			assert(usize_promoted != 0);
+			p = iallocm(usize_promoted, alignment, zero,
+			    try_tcache, arena);
+			if (p == NULL)
+				goto label_oom;
+			arena_prof_promoted(p, usize);
+		} else {
+			p = iallocm(usize, alignment, zero, try_tcache, arena);
+			if (p == NULL)
+				goto label_oom;
+		}
+		prof_malloc(p, usize, cnt);
+	} else {
+		p = iallocm(usize, alignment, zero, try_tcache, arena);
+		if (p == NULL)
+			goto label_oom;
+	}
+	if (rsize != NULL)
+		*rsize = usize;
+
+	*ptr = p;
+	if (config_stats) {
+		assert(usize == isalloc(p, config_prof));
+		thread_allocated_tsd_get()->allocated += usize;
+	}
+	UTRACE(0, size, p);
+	JEMALLOC_VALGRIND_MALLOC(true, p, usize, zero);
+	return (ALLOCM_SUCCESS);
+label_oom:
+	if (config_xmalloc && opt_xmalloc) {
+		malloc_write("<jemalloc>: Error in allocm(): "
+		    "out of memory\n");
+		abort();
+	}
+	*ptr = NULL;
+	UTRACE(0, size, 0);
+	return (ALLOCM_ERR_OOM);
+}
+
+int
+je_rallocm(void **ptr, size_t *rsize, size_t size, size_t extra, int flags)
+{
+	void *p, *q;
+	size_t usize;
+	size_t old_size;
+	size_t old_rzsize JEMALLOC_CC_SILENCE_INIT(0);
+	size_t alignment = (ZU(1) << (flags & ALLOCM_LG_ALIGN_MASK)
+	    & (SIZE_T_MAX-1));
+	bool zero = flags & ALLOCM_ZERO;
+	bool no_move = flags & ALLOCM_NO_MOVE;
+	unsigned arena_ind = ((unsigned)(flags >> 8)) - 1;
+	bool try_tcache_alloc, try_tcache_dalloc;
+	arena_t *arena;
+
+	assert(ptr != NULL);
+	assert(*ptr != NULL);
+	assert(size != 0);
+	assert(SIZE_T_MAX - size >= extra);
+	assert(malloc_initialized || IS_INITIALIZER);
+	malloc_thread_init();
+
+	if (arena_ind != UINT_MAX) {
+		arena_chunk_t *chunk;
+		try_tcache_alloc = true;
+		chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(*ptr);
+		try_tcache_dalloc = (chunk == *ptr || chunk->arena !=
+		    arenas[arena_ind]);
+		arena = arenas[arena_ind];
+	} else {
+		try_tcache_alloc = true;
+		try_tcache_dalloc = true;
+		arena = NULL;
+	}
+
+	p = *ptr;
+	if (config_prof && opt_prof) {
+		prof_thr_cnt_t *cnt;
+
+		/*
+		 * usize isn't knowable before iralloc() returns when extra is
+		 * non-zero.  Therefore, compute its maximum possible value and
+		 * use that in PROF_ALLOC_PREP() to decide whether to capture a
+		 * backtrace.  prof_realloc() will use the actual usize to
+		 * decide whether to sample.
+		 */
+		size_t max_usize = (alignment == 0) ? s2u(size+extra) :
+		    sa2u(size+extra, alignment);
+		prof_ctx_t *old_ctx = prof_ctx_get(p);
+		old_size = isalloc(p, true);
+		if (config_valgrind && opt_valgrind)
+			old_rzsize = p2rz(p);
+		PROF_ALLOC_PREP(1, max_usize, cnt);
+		if (cnt == NULL)
+			goto label_oom;
+		/*
+		 * Use minimum usize to determine whether promotion may happen.
+		 */
+		if (prof_promote && (uintptr_t)cnt != (uintptr_t)1U
+		    && ((alignment == 0) ? s2u(size) : sa2u(size, alignment))
+		    <= SMALL_MAXCLASS) {
+			q = irallocx(p, SMALL_MAXCLASS+1, (SMALL_MAXCLASS+1 >=
+			    size+extra) ? 0 : size+extra - (SMALL_MAXCLASS+1),
+			    alignment, zero, no_move, try_tcache_alloc,
+			    try_tcache_dalloc, arena);
+			if (q == NULL)
+				goto label_err;
+			if (max_usize < PAGE) {
+				usize = max_usize;
+				arena_prof_promoted(q, usize);
+			} else
+				usize = isalloc(q, config_prof);
+		} else {
+			q = irallocx(p, size, extra, alignment, zero, no_move,
+			    try_tcache_alloc, try_tcache_dalloc, arena);
+			if (q == NULL)
+				goto label_err;
+			usize = isalloc(q, config_prof);
+		}
+		prof_realloc(q, usize, cnt, old_size, old_ctx);
+		if (rsize != NULL)
+			*rsize = usize;
+	} else {
+		if (config_stats) {
+			old_size = isalloc(p, false);
+			if (config_valgrind && opt_valgrind)
+				old_rzsize = u2rz(old_size);
+		} else if (config_valgrind && opt_valgrind) {
+			old_size = isalloc(p, false);
+			old_rzsize = u2rz(old_size);
+		}
+		q = irallocx(p, size, extra, alignment, zero, no_move,
+		    try_tcache_alloc, try_tcache_dalloc, arena);
+		if (q == NULL)
+			goto label_err;
+		if (config_stats)
+			usize = isalloc(q, config_prof);
+		if (rsize != NULL) {
+			if (config_stats == false)
+				usize = isalloc(q, config_prof);
+			*rsize = usize;
+		}
+	}
+
+	*ptr = q;
+	if (config_stats) {
+		thread_allocated_t *ta;
+		ta = thread_allocated_tsd_get();
+		ta->allocated += usize;
+		ta->deallocated += old_size;
+	}
+	UTRACE(p, size, q);
+	JEMALLOC_VALGRIND_REALLOC(q, usize, p, old_size, old_rzsize, zero);
+	return (ALLOCM_SUCCESS);
+label_err:
+	if (no_move) {
+		UTRACE(p, size, q);
+		return (ALLOCM_ERR_NOT_MOVED);
+	}
+label_oom:
+	if (config_xmalloc && opt_xmalloc) {
+		malloc_write("<jemalloc>: Error in rallocm(): "
+		    "out of memory\n");
+		abort();
+	}
+	UTRACE(p, size, 0);
+	return (ALLOCM_ERR_OOM);
+}
+
+int
+je_sallocm(const void *ptr, size_t *rsize, int flags)
+{
+	size_t sz;
+
+	assert(malloc_initialized || IS_INITIALIZER);
+	malloc_thread_init();
+
+	if (config_ivsalloc)
+		sz = ivsalloc(ptr, config_prof);
+	else {
+		assert(ptr != NULL);
+		sz = isalloc(ptr, config_prof);
+	}
+	assert(rsize != NULL);
+	*rsize = sz;
+
+	return (ALLOCM_SUCCESS);
+}
+
+int
+je_dallocm(void *ptr, int flags)
+{
+	size_t usize;
+	size_t rzsize JEMALLOC_CC_SILENCE_INIT(0);
+	unsigned arena_ind = ((unsigned)(flags >> 8)) - 1;
+	bool try_tcache;
+
+	assert(ptr != NULL);
+	assert(malloc_initialized || IS_INITIALIZER);
+
+	if (arena_ind != UINT_MAX) {
+		arena_chunk_t *chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+		try_tcache = (chunk == ptr || chunk->arena !=
+		    arenas[arena_ind]);
+	} else
+		try_tcache = true;
+
+	UTRACE(ptr, 0, 0);
+	if (config_stats || config_valgrind)
+		usize = isalloc(ptr, config_prof);
+	if (config_prof && opt_prof) {
+		if (config_stats == false && config_valgrind == false)
+			usize = isalloc(ptr, config_prof);
+		prof_free(ptr, usize);
+	}
+	if (config_stats)
+		thread_allocated_tsd_get()->deallocated += usize;
+	if (config_valgrind && opt_valgrind)
+		rzsize = p2rz(ptr);
+	iqallocx(ptr, try_tcache);
+	JEMALLOC_VALGRIND_FREE(ptr, rzsize);
+
+	return (ALLOCM_SUCCESS);
+}
+
+int
+je_nallocm(size_t *rsize, size_t size, int flags)
+{
+	size_t usize;
+	size_t alignment = (ZU(1) << (flags & ALLOCM_LG_ALIGN_MASK)
+	    & (SIZE_T_MAX-1));
+
+	assert(size != 0);
+
+	if (malloc_init())
+		return (ALLOCM_ERR_OOM);
+
+	usize = (alignment == 0) ? s2u(size) : sa2u(size, alignment);
+	if (usize == 0)
+		return (ALLOCM_ERR_OOM);
+
+	if (rsize != NULL)
+		*rsize = usize;
+	return (ALLOCM_SUCCESS);
+}
+
+#endif
+/*
+ * End experimental functions.
+ */
+/******************************************************************************/
+/*
+ * The following functions are used by threading libraries for protection of
+ * malloc during fork().
+ */
+
+/*
+ * If an application creates a thread before doing any allocation in the main
+ * thread, then calls fork(2) in the main thread followed by memory allocation
+ * in the child process, a race can occur that results in deadlock within the
+ * child: the main thread may have forked while the created thread had
+ * partially initialized the allocator.  Ordinarily jemalloc prevents
+ * fork/malloc races via the following functions it registers during
+ * initialization using pthread_atfork(), but of course that does no good if
+ * the allocator isn't fully initialized at fork time.  The following library
+ * constructor is a partial solution to this problem.  It may still possible to
+ * trigger the deadlock described above, but doing so would involve forking via
+ * a library constructor that runs before jemalloc's runs.
+ */
+JEMALLOC_ATTR(constructor)
+static void
+jemalloc_constructor(void)
+{
+
+	malloc_init();
+}
+
+#ifndef JEMALLOC_MUTEX_INIT_CB
+void
+jemalloc_prefork(void)
+#else
+JEMALLOC_EXPORT void
+_malloc_prefork(void)
+#endif
+{
+	unsigned i;
+
+#ifdef JEMALLOC_MUTEX_INIT_CB
+	if (malloc_initialized == false)
+		return;
+#endif
+	assert(malloc_initialized);
+
+	/* Acquire all mutexes in a safe order. */
+	ctl_prefork();
+	prof_prefork();
+	malloc_mutex_prefork(&arenas_lock);
+	for (i = 0; i < narenas_total; i++) {
+		if (arenas[i] != NULL)
+			arena_prefork(arenas[i]);
+	}
+	chunk_prefork();
+	base_prefork();
+	huge_prefork();
+}
+
+#ifndef JEMALLOC_MUTEX_INIT_CB
+void
+jemalloc_postfork_parent(void)
+#else
+JEMALLOC_EXPORT void
+_malloc_postfork(void)
+#endif
+{
+	unsigned i;
+
+#ifdef JEMALLOC_MUTEX_INIT_CB
+	if (malloc_initialized == false)
+		return;
+#endif
+	assert(malloc_initialized);
+
+	/* Release all mutexes, now that fork() has completed. */
+	huge_postfork_parent();
+	base_postfork_parent();
+	chunk_postfork_parent();
+	for (i = 0; i < narenas_total; i++) {
+		if (arenas[i] != NULL)
+			arena_postfork_parent(arenas[i]);
+	}
+	malloc_mutex_postfork_parent(&arenas_lock);
+	prof_postfork_parent();
+	ctl_postfork_parent();
+}
+
+void
+jemalloc_postfork_child(void)
+{
+	unsigned i;
+
+	assert(malloc_initialized);
+
+	/* Release all mutexes, now that fork() has completed. */
+	huge_postfork_child();
+	base_postfork_child();
+	chunk_postfork_child();
+	for (i = 0; i < narenas_total; i++) {
+		if (arenas[i] != NULL)
+			arena_postfork_child(arenas[i]);
+	}
+	malloc_mutex_postfork_child(&arenas_lock);
+	prof_postfork_child();
+	ctl_postfork_child();
+}
+
+/******************************************************************************/
+/*
+ * The following functions are used for TLS allocation/deallocation in static
+ * binaries on FreeBSD.  The primary difference between these and i[mcd]alloc()
+ * is that these avoid accessing TLS variables.
+ */
+
+static void *
+a0alloc(size_t size, bool zero)
+{
+
+	if (malloc_init())
+		return (NULL);
+
+	if (size == 0)
+		size = 1;
+
+	if (size <= arena_maxclass)
+		return (arena_malloc(arenas[0], size, zero, false));
+	else
+		return (huge_malloc(size, zero));
+}
+
+void *
+a0malloc(size_t size)
+{
+
+	return (a0alloc(size, false));
+}
+
+void *
+a0calloc(size_t num, size_t size)
+{
+
+	return (a0alloc(num * size, true));
+}
+
+void
+a0free(void *ptr)
+{
+	arena_chunk_t *chunk;
+
+	if (ptr == NULL)
+		return;
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+	if (chunk != ptr)
+		arena_dalloc(chunk->arena, chunk, ptr, false);
+	else
+		huge_dalloc(ptr, true);
+}
+
+/******************************************************************************/
diff --git a/extra/jemalloc/src/mb.c b/extra/jemalloc/src/mb.c
new file mode 100644
index 00000000000..dc2c0a256fd
--- /dev/null
+++ b/extra/jemalloc/src/mb.c
@@ -0,0 +1,2 @@
+#define	JEMALLOC_MB_C_
+#include "jemalloc/internal/jemalloc_internal.h"
diff --git a/extra/jemalloc/src/mutex.c b/extra/jemalloc/src/mutex.c
new file mode 100644
index 00000000000..55e18c23713
--- /dev/null
+++ b/extra/jemalloc/src/mutex.c
@@ -0,0 +1,149 @@
+#define	JEMALLOC_MUTEX_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+#if defined(JEMALLOC_LAZY_LOCK) && !defined(_WIN32)
+#include <dlfcn.h>
+#endif
+
+#ifndef _CRT_SPINCOUNT
+#define _CRT_SPINCOUNT 4000
+#endif
+
+/******************************************************************************/
+/* Data. */
+
+#ifdef JEMALLOC_LAZY_LOCK
+bool isthreaded = false;
+#endif
+#ifdef JEMALLOC_MUTEX_INIT_CB
+static bool		postpone_init = true;
+static malloc_mutex_t	*postponed_mutexes = NULL;
+#endif
+
+#if defined(JEMALLOC_LAZY_LOCK) && !defined(_WIN32)
+static void	pthread_create_once(void);
+#endif
+
+/******************************************************************************/
+/*
+ * We intercept pthread_create() calls in order to toggle isthreaded if the
+ * process goes multi-threaded.
+ */
+
+#if defined(JEMALLOC_LAZY_LOCK) && !defined(_WIN32)
+static int (*pthread_create_fptr)(pthread_t *__restrict, const pthread_attr_t *,
+    void *(*)(void *), void *__restrict);
+
+static void
+pthread_create_once(void)
+{
+
+	pthread_create_fptr = dlsym(RTLD_NEXT, "pthread_create");
+	if (pthread_create_fptr == NULL) {
+		malloc_write("<jemalloc>: Error in dlsym(RTLD_NEXT, "
+		    "\"pthread_create\")\n");
+		abort();
+	}
+
+	isthreaded = true;
+}
+
+JEMALLOC_EXPORT int
+pthread_create(pthread_t *__restrict thread,
+    const pthread_attr_t *__restrict attr, void *(*start_routine)(void *),
+    void *__restrict arg)
+{
+	static pthread_once_t once_control = PTHREAD_ONCE_INIT;
+
+	pthread_once(&once_control, pthread_create_once);
+
+	return (pthread_create_fptr(thread, attr, start_routine, arg));
+}
+#endif
+
+/******************************************************************************/
+
+#ifdef JEMALLOC_MUTEX_INIT_CB
+JEMALLOC_EXPORT int	_pthread_mutex_init_calloc_cb(pthread_mutex_t *mutex,
+    void *(calloc_cb)(size_t, size_t));
+#endif
+
+bool
+malloc_mutex_init(malloc_mutex_t *mutex)
+{
+
+#ifdef _WIN32
+	if (!InitializeCriticalSectionAndSpinCount(&mutex->lock,
+	    _CRT_SPINCOUNT))
+		return (true);
+#elif (defined(JEMALLOC_OSSPIN))
+	mutex->lock = 0;
+#elif (defined(JEMALLOC_MUTEX_INIT_CB))
+	if (postpone_init) {
+		mutex->postponed_next = postponed_mutexes;
+		postponed_mutexes = mutex;
+	} else {
+		if (_pthread_mutex_init_calloc_cb(&mutex->lock, base_calloc) !=
+		    0)
+			return (true);
+	}
+#else
+	pthread_mutexattr_t attr;
+
+	if (pthread_mutexattr_init(&attr) != 0)
+		return (true);
+	pthread_mutexattr_settype(&attr, MALLOC_MUTEX_TYPE);
+	if (pthread_mutex_init(&mutex->lock, &attr) != 0) {
+		pthread_mutexattr_destroy(&attr);
+		return (true);
+	}
+	pthread_mutexattr_destroy(&attr);
+#endif
+	return (false);
+}
+
+void
+malloc_mutex_prefork(malloc_mutex_t *mutex)
+{
+
+	malloc_mutex_lock(mutex);
+}
+
+void
+malloc_mutex_postfork_parent(malloc_mutex_t *mutex)
+{
+
+	malloc_mutex_unlock(mutex);
+}
+
+void
+malloc_mutex_postfork_child(malloc_mutex_t *mutex)
+{
+
+#ifdef JEMALLOC_MUTEX_INIT_CB
+	malloc_mutex_unlock(mutex);
+#else
+	if (malloc_mutex_init(mutex)) {
+		malloc_printf("<jemalloc>: Error re-initializing mutex in "
+		    "child\n");
+		if (opt_abort)
+			abort();
+	}
+#endif
+}
+
+bool
+mutex_boot(void)
+{
+
+#ifdef JEMALLOC_MUTEX_INIT_CB
+	postpone_init = false;
+	while (postponed_mutexes != NULL) {
+		if (_pthread_mutex_init_calloc_cb(&postponed_mutexes->lock,
+		    base_calloc) != 0)
+			return (true);
+		postponed_mutexes = postponed_mutexes->postponed_next;
+	}
+#endif
+	return (false);
+}
diff --git a/extra/jemalloc/src/prof.c b/extra/jemalloc/src/prof.c
new file mode 100644
index 00000000000..c133b95c2c6
--- /dev/null
+++ b/extra/jemalloc/src/prof.c
@@ -0,0 +1,1283 @@
+#define	JEMALLOC_PROF_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+/******************************************************************************/
+
+#ifdef JEMALLOC_PROF_LIBUNWIND
+#define	UNW_LOCAL_ONLY
+#include <libunwind.h>
+#endif
+
+#ifdef JEMALLOC_PROF_LIBGCC
+#include <unwind.h>
+#endif
+
+/******************************************************************************/
+/* Data. */
+
+malloc_tsd_data(, prof_tdata, prof_tdata_t *, NULL)
+
+bool		opt_prof = false;
+bool		opt_prof_active = true;
+size_t		opt_lg_prof_sample = LG_PROF_SAMPLE_DEFAULT;
+ssize_t		opt_lg_prof_interval = LG_PROF_INTERVAL_DEFAULT;
+bool		opt_prof_gdump = false;
+bool		opt_prof_final = true;
+bool		opt_prof_leak = false;
+bool		opt_prof_accum = false;
+char		opt_prof_prefix[PATH_MAX + 1];
+
+uint64_t	prof_interval = 0;
+bool		prof_promote;
+
+/*
+ * Table of mutexes that are shared among ctx's.  These are leaf locks, so
+ * there is no problem with using them for more than one ctx at the same time.
+ * The primary motivation for this sharing though is that ctx's are ephemeral,
+ * and destroying mutexes causes complications for systems that allocate when
+ * creating/destroying mutexes.
+ */
+static malloc_mutex_t	*ctx_locks;
+static unsigned		cum_ctxs; /* Atomic counter. */
+
+/*
+ * Global hash of (prof_bt_t *)-->(prof_ctx_t *).  This is the master data
+ * structure that knows about all backtraces currently captured.
+ */
+static ckh_t		bt2ctx;
+static malloc_mutex_t	bt2ctx_mtx;
+
+static malloc_mutex_t	prof_dump_seq_mtx;
+static uint64_t		prof_dump_seq;
+static uint64_t		prof_dump_iseq;
+static uint64_t		prof_dump_mseq;
+static uint64_t		prof_dump_useq;
+
+/*
+ * This buffer is rather large for stack allocation, so use a single buffer for
+ * all profile dumps.  The buffer is implicitly protected by bt2ctx_mtx, since
+ * it must be locked anyway during dumping.
+ */
+static char		prof_dump_buf[PROF_DUMP_BUFSIZE];
+static unsigned		prof_dump_buf_end;
+static int		prof_dump_fd;
+
+/* Do not dump any profiles until bootstrapping is complete. */
+static bool		prof_booted = false;
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static prof_bt_t	*bt_dup(prof_bt_t *bt);
+static void	bt_destroy(prof_bt_t *bt);
+#ifdef JEMALLOC_PROF_LIBGCC
+static _Unwind_Reason_Code	prof_unwind_init_callback(
+    struct _Unwind_Context *context, void *arg);
+static _Unwind_Reason_Code	prof_unwind_callback(
+    struct _Unwind_Context *context, void *arg);
+#endif
+static bool	prof_flush(bool propagate_err);
+static bool	prof_write(bool propagate_err, const char *s);
+static bool	prof_printf(bool propagate_err, const char *format, ...)
+    JEMALLOC_ATTR(format(printf, 2, 3));
+static void	prof_ctx_sum(prof_ctx_t *ctx, prof_cnt_t *cnt_all,
+    size_t *leak_nctx);
+static void	prof_ctx_destroy(prof_ctx_t *ctx);
+static void	prof_ctx_merge(prof_ctx_t *ctx, prof_thr_cnt_t *cnt);
+static bool	prof_dump_ctx(bool propagate_err, prof_ctx_t *ctx,
+    prof_bt_t *bt);
+static bool	prof_dump_maps(bool propagate_err);
+static bool	prof_dump(bool propagate_err, const char *filename,
+    bool leakcheck);
+static void	prof_dump_filename(char *filename, char v, int64_t vseq);
+static void	prof_fdump(void);
+static void	prof_bt_hash(const void *key, size_t r_hash[2]);
+static bool	prof_bt_keycomp(const void *k1, const void *k2);
+static malloc_mutex_t	*prof_ctx_mutex_choose(void);
+
+/******************************************************************************/
+
+void
+bt_init(prof_bt_t *bt, void **vec)
+{
+
+	cassert(config_prof);
+
+	bt->vec = vec;
+	bt->len = 0;
+}
+
+static void
+bt_destroy(prof_bt_t *bt)
+{
+
+	cassert(config_prof);
+
+	idalloc(bt);
+}
+
+static prof_bt_t *
+bt_dup(prof_bt_t *bt)
+{
+	prof_bt_t *ret;
+
+	cassert(config_prof);
+
+	/*
+	 * Create a single allocation that has space for vec immediately
+	 * following the prof_bt_t structure.  The backtraces that get
+	 * stored in the backtrace caches are copied from stack-allocated
+	 * temporary variables, so size is known at creation time.  Making this
+	 * a contiguous object improves cache locality.
+	 */
+	ret = (prof_bt_t *)imalloc(QUANTUM_CEILING(sizeof(prof_bt_t)) +
+	    (bt->len * sizeof(void *)));
+	if (ret == NULL)
+		return (NULL);
+	ret->vec = (void **)((uintptr_t)ret +
+	    QUANTUM_CEILING(sizeof(prof_bt_t)));
+	memcpy(ret->vec, bt->vec, bt->len * sizeof(void *));
+	ret->len = bt->len;
+
+	return (ret);
+}
+
+static inline void
+prof_enter(prof_tdata_t *prof_tdata)
+{
+
+	cassert(config_prof);
+
+	assert(prof_tdata->enq == false);
+	prof_tdata->enq = true;
+
+	malloc_mutex_lock(&bt2ctx_mtx);
+}
+
+static inline void
+prof_leave(prof_tdata_t *prof_tdata)
+{
+	bool idump, gdump;
+
+	cassert(config_prof);
+
+	malloc_mutex_unlock(&bt2ctx_mtx);
+
+	assert(prof_tdata->enq);
+	prof_tdata->enq = false;
+	idump = prof_tdata->enq_idump;
+	prof_tdata->enq_idump = false;
+	gdump = prof_tdata->enq_gdump;
+	prof_tdata->enq_gdump = false;
+
+	if (idump)
+		prof_idump();
+	if (gdump)
+		prof_gdump();
+}
+
+#ifdef JEMALLOC_PROF_LIBUNWIND
+void
+prof_backtrace(prof_bt_t *bt, unsigned nignore)
+{
+	unw_context_t uc;
+	unw_cursor_t cursor;
+	unsigned i;
+	int err;
+
+	cassert(config_prof);
+	assert(bt->len == 0);
+	assert(bt->vec != NULL);
+
+	unw_getcontext(&uc);
+	unw_init_local(&cursor, &uc);
+
+	/* Throw away (nignore+1) stack frames, if that many exist. */
+	for (i = 0; i < nignore + 1; i++) {
+		err = unw_step(&cursor);
+		if (err <= 0)
+			return;
+	}
+
+	/*
+	 * Iterate over stack frames until there are no more, or until no space
+	 * remains in bt.
+	 */
+	for (i = 0; i < PROF_BT_MAX; i++) {
+		unw_get_reg(&cursor, UNW_REG_IP, (unw_word_t *)&bt->vec[i]);
+		bt->len++;
+		err = unw_step(&cursor);
+		if (err <= 0)
+			break;
+	}
+}
+#elif (defined(JEMALLOC_PROF_LIBGCC))
+static _Unwind_Reason_Code
+prof_unwind_init_callback(struct _Unwind_Context *context, void *arg)
+{
+
+	cassert(config_prof);
+
+	return (_URC_NO_REASON);
+}
+
+static _Unwind_Reason_Code
+prof_unwind_callback(struct _Unwind_Context *context, void *arg)
+{
+	prof_unwind_data_t *data = (prof_unwind_data_t *)arg;
+
+	cassert(config_prof);
+
+	if (data->nignore > 0)
+		data->nignore--;
+	else {
+		data->bt->vec[data->bt->len] = (void *)_Unwind_GetIP(context);
+		data->bt->len++;
+		if (data->bt->len == data->max)
+			return (_URC_END_OF_STACK);
+	}
+
+	return (_URC_NO_REASON);
+}
+
+void
+prof_backtrace(prof_bt_t *bt, unsigned nignore)
+{
+	prof_unwind_data_t data = {bt, nignore, PROF_BT_MAX};
+
+	cassert(config_prof);
+
+	_Unwind_Backtrace(prof_unwind_callback, &data);
+}
+#elif (defined(JEMALLOC_PROF_GCC))
+void
+prof_backtrace(prof_bt_t *bt, unsigned nignore)
+{
+#define	BT_FRAME(i)							\
+	if ((i) < nignore + PROF_BT_MAX) {				\
+		void *p;						\
+		if (__builtin_frame_address(i) == 0)			\
+			return;						\
+		p = __builtin_return_address(i);			\
+		if (p == NULL)						\
+			return;						\
+		if (i >= nignore) {					\
+			bt->vec[(i) - nignore] = p;			\
+			bt->len = (i) - nignore + 1;			\
+		}							\
+	} else								\
+		return;
+
+	cassert(config_prof);
+	assert(nignore <= 3);
+
+	BT_FRAME(0)
+	BT_FRAME(1)
+	BT_FRAME(2)
+	BT_FRAME(3)
+	BT_FRAME(4)
+	BT_FRAME(5)
+	BT_FRAME(6)
+	BT_FRAME(7)
+	BT_FRAME(8)
+	BT_FRAME(9)
+
+	BT_FRAME(10)
+	BT_FRAME(11)
+	BT_FRAME(12)
+	BT_FRAME(13)
+	BT_FRAME(14)
+	BT_FRAME(15)
+	BT_FRAME(16)
+	BT_FRAME(17)
+	BT_FRAME(18)
+	BT_FRAME(19)
+
+	BT_FRAME(20)
+	BT_FRAME(21)
+	BT_FRAME(22)
+	BT_FRAME(23)
+	BT_FRAME(24)
+	BT_FRAME(25)
+	BT_FRAME(26)
+	BT_FRAME(27)
+	BT_FRAME(28)
+	BT_FRAME(29)
+
+	BT_FRAME(30)
+	BT_FRAME(31)
+	BT_FRAME(32)
+	BT_FRAME(33)
+	BT_FRAME(34)
+	BT_FRAME(35)
+	BT_FRAME(36)
+	BT_FRAME(37)
+	BT_FRAME(38)
+	BT_FRAME(39)
+
+	BT_FRAME(40)
+	BT_FRAME(41)
+	BT_FRAME(42)
+	BT_FRAME(43)
+	BT_FRAME(44)
+	BT_FRAME(45)
+	BT_FRAME(46)
+	BT_FRAME(47)
+	BT_FRAME(48)
+	BT_FRAME(49)
+
+	BT_FRAME(50)
+	BT_FRAME(51)
+	BT_FRAME(52)
+	BT_FRAME(53)
+	BT_FRAME(54)
+	BT_FRAME(55)
+	BT_FRAME(56)
+	BT_FRAME(57)
+	BT_FRAME(58)
+	BT_FRAME(59)
+
+	BT_FRAME(60)
+	BT_FRAME(61)
+	BT_FRAME(62)
+	BT_FRAME(63)
+	BT_FRAME(64)
+	BT_FRAME(65)
+	BT_FRAME(66)
+	BT_FRAME(67)
+	BT_FRAME(68)
+	BT_FRAME(69)
+
+	BT_FRAME(70)
+	BT_FRAME(71)
+	BT_FRAME(72)
+	BT_FRAME(73)
+	BT_FRAME(74)
+	BT_FRAME(75)
+	BT_FRAME(76)
+	BT_FRAME(77)
+	BT_FRAME(78)
+	BT_FRAME(79)
+
+	BT_FRAME(80)
+	BT_FRAME(81)
+	BT_FRAME(82)
+	BT_FRAME(83)
+	BT_FRAME(84)
+	BT_FRAME(85)
+	BT_FRAME(86)
+	BT_FRAME(87)
+	BT_FRAME(88)
+	BT_FRAME(89)
+
+	BT_FRAME(90)
+	BT_FRAME(91)
+	BT_FRAME(92)
+	BT_FRAME(93)
+	BT_FRAME(94)
+	BT_FRAME(95)
+	BT_FRAME(96)
+	BT_FRAME(97)
+	BT_FRAME(98)
+	BT_FRAME(99)
+
+	BT_FRAME(100)
+	BT_FRAME(101)
+	BT_FRAME(102)
+	BT_FRAME(103)
+	BT_FRAME(104)
+	BT_FRAME(105)
+	BT_FRAME(106)
+	BT_FRAME(107)
+	BT_FRAME(108)
+	BT_FRAME(109)
+
+	BT_FRAME(110)
+	BT_FRAME(111)
+	BT_FRAME(112)
+	BT_FRAME(113)
+	BT_FRAME(114)
+	BT_FRAME(115)
+	BT_FRAME(116)
+	BT_FRAME(117)
+	BT_FRAME(118)
+	BT_FRAME(119)
+
+	BT_FRAME(120)
+	BT_FRAME(121)
+	BT_FRAME(122)
+	BT_FRAME(123)
+	BT_FRAME(124)
+	BT_FRAME(125)
+	BT_FRAME(126)
+	BT_FRAME(127)
+
+	/* Extras to compensate for nignore. */
+	BT_FRAME(128)
+	BT_FRAME(129)
+	BT_FRAME(130)
+#undef BT_FRAME
+}
+#else
+void
+prof_backtrace(prof_bt_t *bt, unsigned nignore)
+{
+
+	cassert(config_prof);
+	assert(false);
+}
+#endif
+
+prof_thr_cnt_t *
+prof_lookup(prof_bt_t *bt)
+{
+	union {
+		prof_thr_cnt_t	*p;
+		void		*v;
+	} ret;
+	prof_tdata_t *prof_tdata;
+
+	cassert(config_prof);
+
+	prof_tdata = prof_tdata_get(false);
+	if ((uintptr_t)prof_tdata <= (uintptr_t)PROF_TDATA_STATE_MAX)
+		return (NULL);
+
+	if (ckh_search(&prof_tdata->bt2cnt, bt, NULL, &ret.v)) {
+		union {
+			prof_bt_t	*p;
+			void		*v;
+		} btkey;
+		union {
+			prof_ctx_t	*p;
+			void		*v;
+		} ctx;
+		bool new_ctx;
+
+		/*
+		 * This thread's cache lacks bt.  Look for it in the global
+		 * cache.
+		 */
+		prof_enter(prof_tdata);
+		if (ckh_search(&bt2ctx, bt, &btkey.v, &ctx.v)) {
+			/* bt has never been seen before.  Insert it. */
+			ctx.v = imalloc(sizeof(prof_ctx_t));
+			if (ctx.v == NULL) {
+				prof_leave(prof_tdata);
+				return (NULL);
+			}
+			btkey.p = bt_dup(bt);
+			if (btkey.v == NULL) {
+				prof_leave(prof_tdata);
+				idalloc(ctx.v);
+				return (NULL);
+			}
+			ctx.p->bt = btkey.p;
+			ctx.p->lock = prof_ctx_mutex_choose();
+			/*
+			 * Set nlimbo to 1, in order to avoid a race condition
+			 * with prof_ctx_merge()/prof_ctx_destroy().
+			 */
+			ctx.p->nlimbo = 1;
+			memset(&ctx.p->cnt_merged, 0, sizeof(prof_cnt_t));
+			ql_new(&ctx.p->cnts_ql);
+			if (ckh_insert(&bt2ctx, btkey.v, ctx.v)) {
+				/* OOM. */
+				prof_leave(prof_tdata);
+				idalloc(btkey.v);
+				idalloc(ctx.v);
+				return (NULL);
+			}
+			new_ctx = true;
+		} else {
+			/*
+			 * Increment nlimbo, in order to avoid a race condition
+			 * with prof_ctx_merge()/prof_ctx_destroy().
+			 */
+			malloc_mutex_lock(ctx.p->lock);
+			ctx.p->nlimbo++;
+			malloc_mutex_unlock(ctx.p->lock);
+			new_ctx = false;
+		}
+		prof_leave(prof_tdata);
+
+		/* Link a prof_thd_cnt_t into ctx for this thread. */
+		if (ckh_count(&prof_tdata->bt2cnt) == PROF_TCMAX) {
+			assert(ckh_count(&prof_tdata->bt2cnt) > 0);
+			/*
+			 * Flush the least recently used cnt in order to keep
+			 * bt2cnt from becoming too large.
+			 */
+			ret.p = ql_last(&prof_tdata->lru_ql, lru_link);
+			assert(ret.v != NULL);
+			if (ckh_remove(&prof_tdata->bt2cnt, ret.p->ctx->bt,
+			    NULL, NULL))
+				assert(false);
+			ql_remove(&prof_tdata->lru_ql, ret.p, lru_link);
+			prof_ctx_merge(ret.p->ctx, ret.p);
+			/* ret can now be re-used. */
+		} else {
+			assert(ckh_count(&prof_tdata->bt2cnt) < PROF_TCMAX);
+			/* Allocate and partially initialize a new cnt. */
+			ret.v = imalloc(sizeof(prof_thr_cnt_t));
+			if (ret.p == NULL) {
+				if (new_ctx)
+					prof_ctx_destroy(ctx.p);
+				return (NULL);
+			}
+			ql_elm_new(ret.p, cnts_link);
+			ql_elm_new(ret.p, lru_link);
+		}
+		/* Finish initializing ret. */
+		ret.p->ctx = ctx.p;
+		ret.p->epoch = 0;
+		memset(&ret.p->cnts, 0, sizeof(prof_cnt_t));
+		if (ckh_insert(&prof_tdata->bt2cnt, btkey.v, ret.v)) {
+			if (new_ctx)
+				prof_ctx_destroy(ctx.p);
+			idalloc(ret.v);
+			return (NULL);
+		}
+		ql_head_insert(&prof_tdata->lru_ql, ret.p, lru_link);
+		malloc_mutex_lock(ctx.p->lock);
+		ql_tail_insert(&ctx.p->cnts_ql, ret.p, cnts_link);
+		ctx.p->nlimbo--;
+		malloc_mutex_unlock(ctx.p->lock);
+	} else {
+		/* Move ret to the front of the LRU. */
+		ql_remove(&prof_tdata->lru_ql, ret.p, lru_link);
+		ql_head_insert(&prof_tdata->lru_ql, ret.p, lru_link);
+	}
+
+	return (ret.p);
+}
+
+static bool
+prof_flush(bool propagate_err)
+{
+	bool ret = false;
+	ssize_t err;
+
+	cassert(config_prof);
+
+	err = write(prof_dump_fd, prof_dump_buf, prof_dump_buf_end);
+	if (err == -1) {
+		if (propagate_err == false) {
+			malloc_write("<jemalloc>: write() failed during heap "
+			    "profile flush\n");
+			if (opt_abort)
+				abort();
+		}
+		ret = true;
+	}
+	prof_dump_buf_end = 0;
+
+	return (ret);
+}
+
+static bool
+prof_write(bool propagate_err, const char *s)
+{
+	unsigned i, slen, n;
+
+	cassert(config_prof);
+
+	i = 0;
+	slen = strlen(s);
+	while (i < slen) {
+		/* Flush the buffer if it is full. */
+		if (prof_dump_buf_end == PROF_DUMP_BUFSIZE)
+			if (prof_flush(propagate_err) && propagate_err)
+				return (true);
+
+		if (prof_dump_buf_end + slen <= PROF_DUMP_BUFSIZE) {
+			/* Finish writing. */
+			n = slen - i;
+		} else {
+			/* Write as much of s as will fit. */
+			n = PROF_DUMP_BUFSIZE - prof_dump_buf_end;
+		}
+		memcpy(&prof_dump_buf[prof_dump_buf_end], &s[i], n);
+		prof_dump_buf_end += n;
+		i += n;
+	}
+
+	return (false);
+}
+
+JEMALLOC_ATTR(format(printf, 2, 3))
+static bool
+prof_printf(bool propagate_err, const char *format, ...)
+{
+	bool ret;
+	va_list ap;
+	char buf[PROF_PRINTF_BUFSIZE];
+
+	va_start(ap, format);
+	malloc_vsnprintf(buf, sizeof(buf), format, ap);
+	va_end(ap);
+	ret = prof_write(propagate_err, buf);
+
+	return (ret);
+}
+
+static void
+prof_ctx_sum(prof_ctx_t *ctx, prof_cnt_t *cnt_all, size_t *leak_nctx)
+{
+	prof_thr_cnt_t *thr_cnt;
+	prof_cnt_t tcnt;
+
+	cassert(config_prof);
+
+	malloc_mutex_lock(ctx->lock);
+
+	memcpy(&ctx->cnt_summed, &ctx->cnt_merged, sizeof(prof_cnt_t));
+	ql_foreach(thr_cnt, &ctx->cnts_ql, cnts_link) {
+		volatile unsigned *epoch = &thr_cnt->epoch;
+
+		while (true) {
+			unsigned epoch0 = *epoch;
+
+			/* Make sure epoch is even. */
+			if (epoch0 & 1U)
+				continue;
+
+			memcpy(&tcnt, &thr_cnt->cnts, sizeof(prof_cnt_t));
+
+			/* Terminate if epoch didn't change while reading. */
+			if (*epoch == epoch0)
+				break;
+		}
+
+		ctx->cnt_summed.curobjs += tcnt.curobjs;
+		ctx->cnt_summed.curbytes += tcnt.curbytes;
+		if (opt_prof_accum) {
+			ctx->cnt_summed.accumobjs += tcnt.accumobjs;
+			ctx->cnt_summed.accumbytes += tcnt.accumbytes;
+		}
+	}
+
+	if (ctx->cnt_summed.curobjs != 0)
+		(*leak_nctx)++;
+
+	/* Add to cnt_all. */
+	cnt_all->curobjs += ctx->cnt_summed.curobjs;
+	cnt_all->curbytes += ctx->cnt_summed.curbytes;
+	if (opt_prof_accum) {
+		cnt_all->accumobjs += ctx->cnt_summed.accumobjs;
+		cnt_all->accumbytes += ctx->cnt_summed.accumbytes;
+	}
+
+	malloc_mutex_unlock(ctx->lock);
+}
+
+static void
+prof_ctx_destroy(prof_ctx_t *ctx)
+{
+	prof_tdata_t *prof_tdata;
+
+	cassert(config_prof);
+
+	/*
+	 * Check that ctx is still unused by any thread cache before destroying
+	 * it.  prof_lookup() increments ctx->nlimbo in order to avoid a race
+	 * condition with this function, as does prof_ctx_merge() in order to
+	 * avoid a race between the main body of prof_ctx_merge() and entry
+	 * into this function.
+	 */
+	prof_tdata = prof_tdata_get(false);
+	assert((uintptr_t)prof_tdata > (uintptr_t)PROF_TDATA_STATE_MAX);
+	prof_enter(prof_tdata);
+	malloc_mutex_lock(ctx->lock);
+	if (ql_first(&ctx->cnts_ql) == NULL && ctx->cnt_merged.curobjs == 0 &&
+	    ctx->nlimbo == 1) {
+		assert(ctx->cnt_merged.curbytes == 0);
+		assert(ctx->cnt_merged.accumobjs == 0);
+		assert(ctx->cnt_merged.accumbytes == 0);
+		/* Remove ctx from bt2ctx. */
+		if (ckh_remove(&bt2ctx, ctx->bt, NULL, NULL))
+			assert(false);
+		prof_leave(prof_tdata);
+		/* Destroy ctx. */
+		malloc_mutex_unlock(ctx->lock);
+		bt_destroy(ctx->bt);
+		idalloc(ctx);
+	} else {
+		/*
+		 * Compensate for increment in prof_ctx_merge() or
+		 * prof_lookup().
+		 */
+		ctx->nlimbo--;
+		malloc_mutex_unlock(ctx->lock);
+		prof_leave(prof_tdata);
+	}
+}
+
+static void
+prof_ctx_merge(prof_ctx_t *ctx, prof_thr_cnt_t *cnt)
+{
+	bool destroy;
+
+	cassert(config_prof);
+
+	/* Merge cnt stats and detach from ctx. */
+	malloc_mutex_lock(ctx->lock);
+	ctx->cnt_merged.curobjs += cnt->cnts.curobjs;
+	ctx->cnt_merged.curbytes += cnt->cnts.curbytes;
+	ctx->cnt_merged.accumobjs += cnt->cnts.accumobjs;
+	ctx->cnt_merged.accumbytes += cnt->cnts.accumbytes;
+	ql_remove(&ctx->cnts_ql, cnt, cnts_link);
+	if (opt_prof_accum == false && ql_first(&ctx->cnts_ql) == NULL &&
+	    ctx->cnt_merged.curobjs == 0 && ctx->nlimbo == 0) {
+		/*
+		 * Increment ctx->nlimbo in order to keep another thread from
+		 * winning the race to destroy ctx while this one has ctx->lock
+		 * dropped.  Without this, it would be possible for another
+		 * thread to:
+		 *
+		 * 1) Sample an allocation associated with ctx.
+		 * 2) Deallocate the sampled object.
+		 * 3) Successfully prof_ctx_destroy(ctx).
+		 *
+		 * The result would be that ctx no longer exists by the time
+		 * this thread accesses it in prof_ctx_destroy().
+		 */
+		ctx->nlimbo++;
+		destroy = true;
+	} else
+		destroy = false;
+	malloc_mutex_unlock(ctx->lock);
+	if (destroy)
+		prof_ctx_destroy(ctx);
+}
+
+static bool
+prof_dump_ctx(bool propagate_err, prof_ctx_t *ctx, prof_bt_t *bt)
+{
+	unsigned i;
+
+	cassert(config_prof);
+
+	/*
+	 * Current statistics can sum to 0 as a result of unmerged per thread
+	 * statistics.  Additionally, interval- and growth-triggered dumps can
+	 * occur between the time a ctx is created and when its statistics are
+	 * filled in.  Avoid dumping any ctx that is an artifact of either
+	 * implementation detail.
+	 */
+	if ((opt_prof_accum == false && ctx->cnt_summed.curobjs == 0) ||
+	    (opt_prof_accum && ctx->cnt_summed.accumobjs == 0)) {
+		assert(ctx->cnt_summed.curobjs == 0);
+		assert(ctx->cnt_summed.curbytes == 0);
+		assert(ctx->cnt_summed.accumobjs == 0);
+		assert(ctx->cnt_summed.accumbytes == 0);
+		return (false);
+	}
+
+	if (prof_printf(propagate_err, "%"PRId64": %"PRId64
+	    " [%"PRIu64": %"PRIu64"] @",
+	    ctx->cnt_summed.curobjs, ctx->cnt_summed.curbytes,
+	    ctx->cnt_summed.accumobjs, ctx->cnt_summed.accumbytes))
+		return (true);
+
+	for (i = 0; i < bt->len; i++) {
+		if (prof_printf(propagate_err, " %#"PRIxPTR,
+		    (uintptr_t)bt->vec[i]))
+			return (true);
+	}
+
+	if (prof_write(propagate_err, "\n"))
+		return (true);
+
+	return (false);
+}
+
+static bool
+prof_dump_maps(bool propagate_err)
+{
+	int mfd;
+	char filename[PATH_MAX + 1];
+
+	cassert(config_prof);
+
+	malloc_snprintf(filename, sizeof(filename), "/proc/%d/maps",
+	    (int)getpid());
+	mfd = open(filename, O_RDONLY);
+	if (mfd != -1) {
+		ssize_t nread;
+
+		if (prof_write(propagate_err, "\nMAPPED_LIBRARIES:\n") &&
+		    propagate_err)
+			return (true);
+		nread = 0;
+		do {
+			prof_dump_buf_end += nread;
+			if (prof_dump_buf_end == PROF_DUMP_BUFSIZE) {
+				/* Make space in prof_dump_buf before read(). */
+				if (prof_flush(propagate_err) && propagate_err)
+					return (true);
+			}
+			nread = read(mfd, &prof_dump_buf[prof_dump_buf_end],
+			    PROF_DUMP_BUFSIZE - prof_dump_buf_end);
+		} while (nread > 0);
+		close(mfd);
+	} else
+		return (true);
+
+	return (false);
+}
+
+static bool
+prof_dump(bool propagate_err, const char *filename, bool leakcheck)
+{
+	prof_tdata_t *prof_tdata;
+	prof_cnt_t cnt_all;
+	size_t tabind;
+	union {
+		prof_bt_t	*p;
+		void		*v;
+	} bt;
+	union {
+		prof_ctx_t	*p;
+		void		*v;
+	} ctx;
+	size_t leak_nctx;
+
+	cassert(config_prof);
+
+	prof_tdata = prof_tdata_get(false);
+	if ((uintptr_t)prof_tdata <= (uintptr_t)PROF_TDATA_STATE_MAX)
+		return (true);
+	prof_enter(prof_tdata);
+	prof_dump_fd = creat(filename, 0644);
+	if (prof_dump_fd == -1) {
+		if (propagate_err == false) {
+			malloc_printf(
+			    "<jemalloc>: creat(\"%s\"), 0644) failed\n",
+			    filename);
+			if (opt_abort)
+				abort();
+		}
+		goto label_error;
+	}
+
+	/* Merge per thread profile stats, and sum them in cnt_all. */
+	memset(&cnt_all, 0, sizeof(prof_cnt_t));
+	leak_nctx = 0;
+	for (tabind = 0; ckh_iter(&bt2ctx, &tabind, NULL, &ctx.v) == false;)
+		prof_ctx_sum(ctx.p, &cnt_all, &leak_nctx);
+
+	/* Dump profile header. */
+	if (opt_lg_prof_sample == 0) {
+		if (prof_printf(propagate_err,
+		    "heap profile: %"PRId64": %"PRId64
+		    " [%"PRIu64": %"PRIu64"] @ heapprofile\n",
+		    cnt_all.curobjs, cnt_all.curbytes,
+		    cnt_all.accumobjs, cnt_all.accumbytes))
+			goto label_error;
+	} else {
+		if (prof_printf(propagate_err,
+		    "heap profile: %"PRId64": %"PRId64
+		    " [%"PRIu64": %"PRIu64"] @ heap_v2/%"PRIu64"\n",
+		    cnt_all.curobjs, cnt_all.curbytes,
+		    cnt_all.accumobjs, cnt_all.accumbytes,
+		    ((uint64_t)1U << opt_lg_prof_sample)))
+			goto label_error;
+	}
+
+	/* Dump  per ctx profile stats. */
+	for (tabind = 0; ckh_iter(&bt2ctx, &tabind, &bt.v, &ctx.v)
+	    == false;) {
+		if (prof_dump_ctx(propagate_err, ctx.p, bt.p))
+			goto label_error;
+	}
+
+	/* Dump /proc/<pid>/maps if possible. */
+	if (prof_dump_maps(propagate_err))
+		goto label_error;
+
+	if (prof_flush(propagate_err))
+		goto label_error;
+	close(prof_dump_fd);
+	prof_leave(prof_tdata);
+
+	if (leakcheck && cnt_all.curbytes != 0) {
+		malloc_printf("<jemalloc>: Leak summary: %"PRId64" byte%s, %"
+		    PRId64" object%s, %zu context%s\n",
+		    cnt_all.curbytes, (cnt_all.curbytes != 1) ? "s" : "",
+		    cnt_all.curobjs, (cnt_all.curobjs != 1) ? "s" : "",
+		    leak_nctx, (leak_nctx != 1) ? "s" : "");
+		malloc_printf(
+		    "<jemalloc>: Run pprof on \"%s\" for leak detail\n",
+		    filename);
+	}
+
+	return (false);
+label_error:
+	prof_leave(prof_tdata);
+	return (true);
+}
+
+#define	DUMP_FILENAME_BUFSIZE	(PATH_MAX + 1)
+static void
+prof_dump_filename(char *filename, char v, int64_t vseq)
+{
+
+	cassert(config_prof);
+
+	if (vseq != UINT64_C(0xffffffffffffffff)) {
+	        /* "<prefix>.<pid>.<seq>.v<vseq>.heap" */
+		malloc_snprintf(filename, DUMP_FILENAME_BUFSIZE,
+		    "%s.%d.%"PRIu64".%c%"PRId64".heap",
+		    opt_prof_prefix, (int)getpid(), prof_dump_seq, v, vseq);
+	} else {
+	        /* "<prefix>.<pid>.<seq>.<v>.heap" */
+		malloc_snprintf(filename, DUMP_FILENAME_BUFSIZE,
+		    "%s.%d.%"PRIu64".%c.heap",
+		    opt_prof_prefix, (int)getpid(), prof_dump_seq, v);
+	}
+	prof_dump_seq++;
+}
+
+static void
+prof_fdump(void)
+{
+	char filename[DUMP_FILENAME_BUFSIZE];
+
+	cassert(config_prof);
+
+	if (prof_booted == false)
+		return;
+
+	if (opt_prof_final && opt_prof_prefix[0] != '\0') {
+		malloc_mutex_lock(&prof_dump_seq_mtx);
+		prof_dump_filename(filename, 'f', UINT64_C(0xffffffffffffffff));
+		malloc_mutex_unlock(&prof_dump_seq_mtx);
+		prof_dump(false, filename, opt_prof_leak);
+	}
+}
+
+void
+prof_idump(void)
+{
+	prof_tdata_t *prof_tdata;
+	char filename[PATH_MAX + 1];
+
+	cassert(config_prof);
+
+	if (prof_booted == false)
+		return;
+	prof_tdata = prof_tdata_get(false);
+	if ((uintptr_t)prof_tdata <= (uintptr_t)PROF_TDATA_STATE_MAX)
+		return;
+	if (prof_tdata->enq) {
+		prof_tdata->enq_idump = true;
+		return;
+	}
+
+	if (opt_prof_prefix[0] != '\0') {
+		malloc_mutex_lock(&prof_dump_seq_mtx);
+		prof_dump_filename(filename, 'i', prof_dump_iseq);
+		prof_dump_iseq++;
+		malloc_mutex_unlock(&prof_dump_seq_mtx);
+		prof_dump(false, filename, false);
+	}
+}
+
+bool
+prof_mdump(const char *filename)
+{
+	char filename_buf[DUMP_FILENAME_BUFSIZE];
+
+	cassert(config_prof);
+
+	if (opt_prof == false || prof_booted == false)
+		return (true);
+
+	if (filename == NULL) {
+		/* No filename specified, so automatically generate one. */
+		if (opt_prof_prefix[0] == '\0')
+			return (true);
+		malloc_mutex_lock(&prof_dump_seq_mtx);
+		prof_dump_filename(filename_buf, 'm', prof_dump_mseq);
+		prof_dump_mseq++;
+		malloc_mutex_unlock(&prof_dump_seq_mtx);
+		filename = filename_buf;
+	}
+	return (prof_dump(true, filename, false));
+}
+
+void
+prof_gdump(void)
+{
+	prof_tdata_t *prof_tdata;
+	char filename[DUMP_FILENAME_BUFSIZE];
+
+	cassert(config_prof);
+
+	if (prof_booted == false)
+		return;
+	prof_tdata = prof_tdata_get(false);
+	if ((uintptr_t)prof_tdata <= (uintptr_t)PROF_TDATA_STATE_MAX)
+		return;
+	if (prof_tdata->enq) {
+		prof_tdata->enq_gdump = true;
+		return;
+	}
+
+	if (opt_prof_prefix[0] != '\0') {
+		malloc_mutex_lock(&prof_dump_seq_mtx);
+		prof_dump_filename(filename, 'u', prof_dump_useq);
+		prof_dump_useq++;
+		malloc_mutex_unlock(&prof_dump_seq_mtx);
+		prof_dump(false, filename, false);
+	}
+}
+
+static void
+prof_bt_hash(const void *key, size_t r_hash[2])
+{
+	prof_bt_t *bt = (prof_bt_t *)key;
+
+	cassert(config_prof);
+
+	hash(bt->vec, bt->len * sizeof(void *), 0x94122f33U, r_hash);
+}
+
+static bool
+prof_bt_keycomp(const void *k1, const void *k2)
+{
+	const prof_bt_t *bt1 = (prof_bt_t *)k1;
+	const prof_bt_t *bt2 = (prof_bt_t *)k2;
+
+	cassert(config_prof);
+
+	if (bt1->len != bt2->len)
+		return (false);
+	return (memcmp(bt1->vec, bt2->vec, bt1->len * sizeof(void *)) == 0);
+}
+
+static malloc_mutex_t *
+prof_ctx_mutex_choose(void)
+{
+	unsigned nctxs = atomic_add_u(&cum_ctxs, 1);
+
+	return (&ctx_locks[(nctxs - 1) % PROF_NCTX_LOCKS]);
+}
+
+prof_tdata_t *
+prof_tdata_init(void)
+{
+	prof_tdata_t *prof_tdata;
+
+	cassert(config_prof);
+
+	/* Initialize an empty cache for this thread. */
+	prof_tdata = (prof_tdata_t *)imalloc(sizeof(prof_tdata_t));
+	if (prof_tdata == NULL)
+		return (NULL);
+
+	if (ckh_new(&prof_tdata->bt2cnt, PROF_CKH_MINITEMS,
+	    prof_bt_hash, prof_bt_keycomp)) {
+		idalloc(prof_tdata);
+		return (NULL);
+	}
+	ql_new(&prof_tdata->lru_ql);
+
+	prof_tdata->vec = imalloc(sizeof(void *) * PROF_BT_MAX);
+	if (prof_tdata->vec == NULL) {
+		ckh_delete(&prof_tdata->bt2cnt);
+		idalloc(prof_tdata);
+		return (NULL);
+	}
+
+	prof_tdata->prng_state = 0;
+	prof_tdata->threshold = 0;
+	prof_tdata->accum = 0;
+
+	prof_tdata->enq = false;
+	prof_tdata->enq_idump = false;
+	prof_tdata->enq_gdump = false;
+
+	prof_tdata_tsd_set(&prof_tdata);
+
+	return (prof_tdata);
+}
+
+void
+prof_tdata_cleanup(void *arg)
+{
+	prof_thr_cnt_t *cnt;
+	prof_tdata_t *prof_tdata = *(prof_tdata_t **)arg;
+
+	cassert(config_prof);
+
+	if (prof_tdata == PROF_TDATA_STATE_REINCARNATED) {
+		/*
+		 * Another destructor deallocated memory after this destructor
+		 * was called.  Reset prof_tdata to PROF_TDATA_STATE_PURGATORY
+		 * in order to receive another callback.
+		 */
+		prof_tdata = PROF_TDATA_STATE_PURGATORY;
+		prof_tdata_tsd_set(&prof_tdata);
+	} else if (prof_tdata == PROF_TDATA_STATE_PURGATORY) {
+		/*
+		 * The previous time this destructor was called, we set the key
+		 * to PROF_TDATA_STATE_PURGATORY so that other destructors
+		 * wouldn't cause re-creation of the prof_tdata.  This time, do
+		 * nothing, so that the destructor will not be called again.
+		 */
+	} else if (prof_tdata != NULL) {
+		/*
+		 * Delete the hash table.  All of its contents can still be
+		 * iterated over via the LRU.
+		 */
+		ckh_delete(&prof_tdata->bt2cnt);
+		/*
+		 * Iteratively merge cnt's into the global stats and delete
+		 * them.
+		 */
+		while ((cnt = ql_last(&prof_tdata->lru_ql, lru_link)) != NULL) {
+			ql_remove(&prof_tdata->lru_ql, cnt, lru_link);
+			prof_ctx_merge(cnt->ctx, cnt);
+			idalloc(cnt);
+		}
+		idalloc(prof_tdata->vec);
+		idalloc(prof_tdata);
+		prof_tdata = PROF_TDATA_STATE_PURGATORY;
+		prof_tdata_tsd_set(&prof_tdata);
+	}
+}
+
+void
+prof_boot0(void)
+{
+
+	cassert(config_prof);
+
+	memcpy(opt_prof_prefix, PROF_PREFIX_DEFAULT,
+	    sizeof(PROF_PREFIX_DEFAULT));
+}
+
+void
+prof_boot1(void)
+{
+
+	cassert(config_prof);
+
+	/*
+	 * opt_prof and prof_promote must be in their final state before any
+	 * arenas are initialized, so this function must be executed early.
+	 */
+
+	if (opt_prof_leak && opt_prof == false) {
+		/*
+		 * Enable opt_prof, but in such a way that profiles are never
+		 * automatically dumped.
+		 */
+		opt_prof = true;
+		opt_prof_gdump = false;
+	} else if (opt_prof) {
+		if (opt_lg_prof_interval >= 0) {
+			prof_interval = (((uint64_t)1U) <<
+			    opt_lg_prof_interval);
+		}
+	}
+
+	prof_promote = (opt_prof && opt_lg_prof_sample > LG_PAGE);
+}
+
+bool
+prof_boot2(void)
+{
+
+	cassert(config_prof);
+
+	if (opt_prof) {
+		unsigned i;
+
+		if (ckh_new(&bt2ctx, PROF_CKH_MINITEMS, prof_bt_hash,
+		    prof_bt_keycomp))
+			return (true);
+		if (malloc_mutex_init(&bt2ctx_mtx))
+			return (true);
+		if (prof_tdata_tsd_boot()) {
+			malloc_write(
+			    "<jemalloc>: Error in pthread_key_create()\n");
+			abort();
+		}
+
+		if (malloc_mutex_init(&prof_dump_seq_mtx))
+			return (true);
+
+		if (atexit(prof_fdump) != 0) {
+			malloc_write("<jemalloc>: Error in atexit()\n");
+			if (opt_abort)
+				abort();
+		}
+
+		ctx_locks = (malloc_mutex_t *)base_alloc(PROF_NCTX_LOCKS *
+		    sizeof(malloc_mutex_t));
+		if (ctx_locks == NULL)
+			return (true);
+		for (i = 0; i < PROF_NCTX_LOCKS; i++) {
+			if (malloc_mutex_init(&ctx_locks[i]))
+				return (true);
+		}
+	}
+
+#ifdef JEMALLOC_PROF_LIBGCC
+	/*
+	 * Cause the backtracing machinery to allocate its internal state
+	 * before enabling profiling.
+	 */
+	_Unwind_Backtrace(prof_unwind_init_callback, NULL);
+#endif
+
+	prof_booted = true;
+
+	return (false);
+}
+
+void
+prof_prefork(void)
+{
+
+	if (opt_prof) {
+		unsigned i;
+
+		malloc_mutex_lock(&bt2ctx_mtx);
+		malloc_mutex_lock(&prof_dump_seq_mtx);
+		for (i = 0; i < PROF_NCTX_LOCKS; i++)
+			malloc_mutex_lock(&ctx_locks[i]);
+	}
+}
+
+void
+prof_postfork_parent(void)
+{
+
+	if (opt_prof) {
+		unsigned i;
+
+		for (i = 0; i < PROF_NCTX_LOCKS; i++)
+			malloc_mutex_postfork_parent(&ctx_locks[i]);
+		malloc_mutex_postfork_parent(&prof_dump_seq_mtx);
+		malloc_mutex_postfork_parent(&bt2ctx_mtx);
+	}
+}
+
+void
+prof_postfork_child(void)
+{
+
+	if (opt_prof) {
+		unsigned i;
+
+		for (i = 0; i < PROF_NCTX_LOCKS; i++)
+			malloc_mutex_postfork_child(&ctx_locks[i]);
+		malloc_mutex_postfork_child(&prof_dump_seq_mtx);
+		malloc_mutex_postfork_child(&bt2ctx_mtx);
+	}
+}
+
+/******************************************************************************/
diff --git a/extra/jemalloc/src/quarantine.c b/extra/jemalloc/src/quarantine.c
new file mode 100644
index 00000000000..f96a948d5c7
--- /dev/null
+++ b/extra/jemalloc/src/quarantine.c
@@ -0,0 +1,190 @@
+#define	JEMALLOC_QUARANTINE_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/*
+ * quarantine pointers close to NULL are used to encode state information that
+ * is used for cleaning up during thread shutdown.
+ */
+#define	QUARANTINE_STATE_REINCARNATED	((quarantine_t *)(uintptr_t)1)
+#define	QUARANTINE_STATE_PURGATORY	((quarantine_t *)(uintptr_t)2)
+#define	QUARANTINE_STATE_MAX		QUARANTINE_STATE_PURGATORY
+
+/******************************************************************************/
+/* Data. */
+
+malloc_tsd_data(, quarantine, quarantine_t *, NULL)
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static quarantine_t	*quarantine_grow(quarantine_t *quarantine);
+static void	quarantine_drain_one(quarantine_t *quarantine);
+static void	quarantine_drain(quarantine_t *quarantine, size_t upper_bound);
+
+/******************************************************************************/
+
+quarantine_t *
+quarantine_init(size_t lg_maxobjs)
+{
+	quarantine_t *quarantine;
+
+	quarantine = (quarantine_t *)imalloc(offsetof(quarantine_t, objs) +
+	    ((ZU(1) << lg_maxobjs) * sizeof(quarantine_obj_t)));
+	if (quarantine == NULL)
+		return (NULL);
+	quarantine->curbytes = 0;
+	quarantine->curobjs = 0;
+	quarantine->first = 0;
+	quarantine->lg_maxobjs = lg_maxobjs;
+
+	quarantine_tsd_set(&quarantine);
+
+	return (quarantine);
+}
+
+static quarantine_t *
+quarantine_grow(quarantine_t *quarantine)
+{
+	quarantine_t *ret;
+
+	ret = quarantine_init(quarantine->lg_maxobjs + 1);
+	if (ret == NULL) {
+		quarantine_drain_one(quarantine);
+		return (quarantine);
+	}
+
+	ret->curbytes = quarantine->curbytes;
+	ret->curobjs = quarantine->curobjs;
+	if (quarantine->first + quarantine->curobjs <= (ZU(1) <<
+	    quarantine->lg_maxobjs)) {
+		/* objs ring buffer data are contiguous. */
+		memcpy(ret->objs, &quarantine->objs[quarantine->first],
+		    quarantine->curobjs * sizeof(quarantine_obj_t));
+	} else {
+		/* objs ring buffer data wrap around. */
+		size_t ncopy_a = (ZU(1) << quarantine->lg_maxobjs) -
+		    quarantine->first;
+		size_t ncopy_b = quarantine->curobjs - ncopy_a;
+
+		memcpy(ret->objs, &quarantine->objs[quarantine->first], ncopy_a
+		    * sizeof(quarantine_obj_t));
+		memcpy(&ret->objs[ncopy_a], quarantine->objs, ncopy_b *
+		    sizeof(quarantine_obj_t));
+	}
+	idalloc(quarantine);
+
+	return (ret);
+}
+
+static void
+quarantine_drain_one(quarantine_t *quarantine)
+{
+	quarantine_obj_t *obj = &quarantine->objs[quarantine->first];
+	assert(obj->usize == isalloc(obj->ptr, config_prof));
+	idalloc(obj->ptr);
+	quarantine->curbytes -= obj->usize;
+	quarantine->curobjs--;
+	quarantine->first = (quarantine->first + 1) & ((ZU(1) <<
+	    quarantine->lg_maxobjs) - 1);
+}
+
+static void
+quarantine_drain(quarantine_t *quarantine, size_t upper_bound)
+{
+
+	while (quarantine->curbytes > upper_bound && quarantine->curobjs > 0)
+		quarantine_drain_one(quarantine);
+}
+
+void
+quarantine(void *ptr)
+{
+	quarantine_t *quarantine;
+	size_t usize = isalloc(ptr, config_prof);
+
+	cassert(config_fill);
+	assert(opt_quarantine);
+
+	quarantine = *quarantine_tsd_get();
+	if ((uintptr_t)quarantine <= (uintptr_t)QUARANTINE_STATE_MAX) {
+		if (quarantine == QUARANTINE_STATE_PURGATORY) {
+			/*
+			 * Make a note that quarantine() was called after
+			 * quarantine_cleanup() was called.
+			 */
+			quarantine = QUARANTINE_STATE_REINCARNATED;
+			quarantine_tsd_set(&quarantine);
+		}
+		idalloc(ptr);
+		return;
+	}
+	/*
+	 * Drain one or more objects if the quarantine size limit would be
+	 * exceeded by appending ptr.
+	 */
+	if (quarantine->curbytes + usize > opt_quarantine) {
+		size_t upper_bound = (opt_quarantine >= usize) ? opt_quarantine
+		    - usize : 0;
+		quarantine_drain(quarantine, upper_bound);
+	}
+	/* Grow the quarantine ring buffer if it's full. */
+	if (quarantine->curobjs == (ZU(1) << quarantine->lg_maxobjs))
+		quarantine = quarantine_grow(quarantine);
+	/* quarantine_grow() must free a slot if it fails to grow. */
+	assert(quarantine->curobjs < (ZU(1) << quarantine->lg_maxobjs));
+	/* Append ptr if its size doesn't exceed the quarantine size. */
+	if (quarantine->curbytes + usize <= opt_quarantine) {
+		size_t offset = (quarantine->first + quarantine->curobjs) &
+		    ((ZU(1) << quarantine->lg_maxobjs) - 1);
+		quarantine_obj_t *obj = &quarantine->objs[offset];
+		obj->ptr = ptr;
+		obj->usize = usize;
+		quarantine->curbytes += usize;
+		quarantine->curobjs++;
+		if (opt_junk)
+			memset(ptr, 0x5a, usize);
+	} else {
+		assert(quarantine->curbytes == 0);
+		idalloc(ptr);
+	}
+}
+
+void
+quarantine_cleanup(void *arg)
+{
+	quarantine_t *quarantine = *(quarantine_t **)arg;
+
+	if (quarantine == QUARANTINE_STATE_REINCARNATED) {
+		/*
+		 * Another destructor deallocated memory after this destructor
+		 * was called.  Reset quarantine to QUARANTINE_STATE_PURGATORY
+		 * in order to receive another callback.
+		 */
+		quarantine = QUARANTINE_STATE_PURGATORY;
+		quarantine_tsd_set(&quarantine);
+	} else if (quarantine == QUARANTINE_STATE_PURGATORY) {
+		/*
+		 * The previous time this destructor was called, we set the key
+		 * to QUARANTINE_STATE_PURGATORY so that other destructors
+		 * wouldn't cause re-creation of the quarantine.  This time, do
+		 * nothing, so that the destructor will not be called again.
+		 */
+	} else if (quarantine != NULL) {
+		quarantine_drain(quarantine, 0);
+		idalloc(quarantine);
+		quarantine = QUARANTINE_STATE_PURGATORY;
+		quarantine_tsd_set(&quarantine);
+	}
+}
+
+bool
+quarantine_boot(void)
+{
+
+	cassert(config_fill);
+
+	if (quarantine_tsd_boot())
+		return (true);
+
+	return (false);
+}
diff --git a/extra/jemalloc/src/rtree.c b/extra/jemalloc/src/rtree.c
new file mode 100644
index 00000000000..90c6935a0ed
--- /dev/null
+++ b/extra/jemalloc/src/rtree.c
@@ -0,0 +1,67 @@
+#define	JEMALLOC_RTREE_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+rtree_t *
+rtree_new(unsigned bits)
+{
+	rtree_t *ret;
+	unsigned bits_per_level, height, i;
+
+	bits_per_level = ffs(pow2_ceil((RTREE_NODESIZE / sizeof(void *)))) - 1;
+	height = bits / bits_per_level;
+	if (height * bits_per_level != bits)
+		height++;
+	assert(height * bits_per_level >= bits);
+
+	ret = (rtree_t*)base_alloc(offsetof(rtree_t, level2bits) +
+	    (sizeof(unsigned) * height));
+	if (ret == NULL)
+		return (NULL);
+	memset(ret, 0, offsetof(rtree_t, level2bits) + (sizeof(unsigned) *
+	    height));
+
+	if (malloc_mutex_init(&ret->mutex)) {
+		/* Leak the rtree. */
+		return (NULL);
+	}
+	ret->height = height;
+	if (bits_per_level * height > bits)
+		ret->level2bits[0] = bits % bits_per_level;
+	else
+		ret->level2bits[0] = bits_per_level;
+	for (i = 1; i < height; i++)
+		ret->level2bits[i] = bits_per_level;
+
+	ret->root = (void**)base_alloc(sizeof(void *) << ret->level2bits[0]);
+	if (ret->root == NULL) {
+		/*
+		 * We leak the rtree here, since there's no generic base
+		 * deallocation.
+		 */
+		return (NULL);
+	}
+	memset(ret->root, 0, sizeof(void *) << ret->level2bits[0]);
+
+	return (ret);
+}
+
+void
+rtree_prefork(rtree_t *rtree)
+{
+
+	malloc_mutex_prefork(&rtree->mutex);
+}
+
+void
+rtree_postfork_parent(rtree_t *rtree)
+{
+
+	malloc_mutex_postfork_parent(&rtree->mutex);
+}
+
+void
+rtree_postfork_child(rtree_t *rtree)
+{
+
+	malloc_mutex_postfork_child(&rtree->mutex);
+}
diff --git a/extra/jemalloc/src/stats.c b/extra/jemalloc/src/stats.c
new file mode 100644
index 00000000000..43f87af6700
--- /dev/null
+++ b/extra/jemalloc/src/stats.c
@@ -0,0 +1,549 @@
+#define	JEMALLOC_STATS_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+#define	CTL_GET(n, v, t) do {						\
+	size_t sz = sizeof(t);						\
+	xmallctl(n, v, &sz, NULL, 0);					\
+} while (0)
+
+#define	CTL_I_GET(n, v, t) do {						\
+	size_t mib[6];							\
+	size_t miblen = sizeof(mib) / sizeof(size_t);			\
+	size_t sz = sizeof(t);						\
+	xmallctlnametomib(n, mib, &miblen);				\
+	mib[2] = i;							\
+	xmallctlbymib(mib, miblen, v, &sz, NULL, 0);			\
+} while (0)
+
+#define	CTL_J_GET(n, v, t) do {						\
+	size_t mib[6];							\
+	size_t miblen = sizeof(mib) / sizeof(size_t);			\
+	size_t sz = sizeof(t);						\
+	xmallctlnametomib(n, mib, &miblen);				\
+	mib[2] = j;							\
+	xmallctlbymib(mib, miblen, v, &sz, NULL, 0);			\
+} while (0)
+
+#define	CTL_IJ_GET(n, v, t) do {					\
+	size_t mib[6];							\
+	size_t miblen = sizeof(mib) / sizeof(size_t);			\
+	size_t sz = sizeof(t);						\
+	xmallctlnametomib(n, mib, &miblen);				\
+	mib[2] = i;							\
+	mib[4] = j;							\
+	xmallctlbymib(mib, miblen, v, &sz, NULL, 0);			\
+} while (0)
+
+/******************************************************************************/
+/* Data. */
+
+bool	opt_stats_print = false;
+
+size_t	stats_cactive = 0;
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static void	stats_arena_bins_print(void (*write_cb)(void *, const char *),
+    void *cbopaque, unsigned i);
+static void	stats_arena_lruns_print(void (*write_cb)(void *, const char *),
+    void *cbopaque, unsigned i);
+static void	stats_arena_print(void (*write_cb)(void *, const char *),
+    void *cbopaque, unsigned i, bool bins, bool large);
+
+/******************************************************************************/
+
+static void
+stats_arena_bins_print(void (*write_cb)(void *, const char *), void *cbopaque,
+    unsigned i)
+{
+	size_t page;
+	bool config_tcache;
+	unsigned nbins, j, gap_start;
+
+	CTL_GET("arenas.page", &page, size_t);
+
+	CTL_GET("config.tcache", &config_tcache, bool);
+	if (config_tcache) {
+		malloc_cprintf(write_cb, cbopaque,
+		    "bins:     bin  size regs pgs    allocated      nmalloc"
+		    "      ndalloc    nrequests       nfills     nflushes"
+		    "      newruns       reruns      curruns\n");
+	} else {
+		malloc_cprintf(write_cb, cbopaque,
+		    "bins:     bin  size regs pgs    allocated      nmalloc"
+		    "      ndalloc      newruns       reruns      curruns\n");
+	}
+	CTL_GET("arenas.nbins", &nbins, unsigned);
+	for (j = 0, gap_start = UINT_MAX; j < nbins; j++) {
+		uint64_t nruns;
+
+		CTL_IJ_GET("stats.arenas.0.bins.0.nruns", &nruns, uint64_t);
+		if (nruns == 0) {
+			if (gap_start == UINT_MAX)
+				gap_start = j;
+		} else {
+			size_t reg_size, run_size, allocated;
+			uint32_t nregs;
+			uint64_t nmalloc, ndalloc, nrequests, nfills, nflushes;
+			uint64_t reruns;
+			size_t curruns;
+
+			if (gap_start != UINT_MAX) {
+				if (j > gap_start + 1) {
+					/* Gap of more than one size class. */
+					malloc_cprintf(write_cb, cbopaque,
+					    "[%u..%u]\n", gap_start,
+					    j - 1);
+				} else {
+					/* Gap of one size class. */
+					malloc_cprintf(write_cb, cbopaque,
+					    "[%u]\n", gap_start);
+				}
+				gap_start = UINT_MAX;
+			}
+			CTL_J_GET("arenas.bin.0.size", &reg_size, size_t);
+			CTL_J_GET("arenas.bin.0.nregs", &nregs, uint32_t);
+			CTL_J_GET("arenas.bin.0.run_size", &run_size, size_t);
+			CTL_IJ_GET("stats.arenas.0.bins.0.allocated",
+			    &allocated, size_t);
+			CTL_IJ_GET("stats.arenas.0.bins.0.nmalloc",
+			    &nmalloc, uint64_t);
+			CTL_IJ_GET("stats.arenas.0.bins.0.ndalloc",
+			    &ndalloc, uint64_t);
+			if (config_tcache) {
+				CTL_IJ_GET("stats.arenas.0.bins.0.nrequests",
+				    &nrequests, uint64_t);
+				CTL_IJ_GET("stats.arenas.0.bins.0.nfills",
+				    &nfills, uint64_t);
+				CTL_IJ_GET("stats.arenas.0.bins.0.nflushes",
+				    &nflushes, uint64_t);
+			}
+			CTL_IJ_GET("stats.arenas.0.bins.0.nreruns", &reruns,
+			    uint64_t);
+			CTL_IJ_GET("stats.arenas.0.bins.0.curruns", &curruns,
+			    size_t);
+			if (config_tcache) {
+				malloc_cprintf(write_cb, cbopaque,
+				    "%13u %5zu %4u %3zu %12zu %12"PRIu64
+				    " %12"PRIu64" %12"PRIu64" %12"PRIu64
+				    " %12"PRIu64" %12"PRIu64" %12"PRIu64
+				    " %12zu\n",
+				    j, reg_size, nregs, run_size / page,
+				    allocated, nmalloc, ndalloc, nrequests,
+				    nfills, nflushes, nruns, reruns, curruns);
+			} else {
+				malloc_cprintf(write_cb, cbopaque,
+				    "%13u %5zu %4u %3zu %12zu %12"PRIu64
+				    " %12"PRIu64" %12"PRIu64" %12"PRIu64
+				    " %12zu\n",
+				    j, reg_size, nregs, run_size / page,
+				    allocated, nmalloc, ndalloc, nruns, reruns,
+				    curruns);
+			}
+		}
+	}
+	if (gap_start != UINT_MAX) {
+		if (j > gap_start + 1) {
+			/* Gap of more than one size class. */
+			malloc_cprintf(write_cb, cbopaque, "[%u..%u]\n",
+			    gap_start, j - 1);
+		} else {
+			/* Gap of one size class. */
+			malloc_cprintf(write_cb, cbopaque, "[%u]\n", gap_start);
+		}
+	}
+}
+
+static void
+stats_arena_lruns_print(void (*write_cb)(void *, const char *), void *cbopaque,
+    unsigned i)
+{
+	size_t page, nlruns, j;
+	ssize_t gap_start;
+
+	CTL_GET("arenas.page", &page, size_t);
+
+	malloc_cprintf(write_cb, cbopaque,
+	    "large:   size pages      nmalloc      ndalloc    nrequests"
+	    "      curruns\n");
+	CTL_GET("arenas.nlruns", &nlruns, size_t);
+	for (j = 0, gap_start = -1; j < nlruns; j++) {
+		uint64_t nmalloc, ndalloc, nrequests;
+		size_t run_size, curruns;
+
+		CTL_IJ_GET("stats.arenas.0.lruns.0.nmalloc", &nmalloc,
+		    uint64_t);
+		CTL_IJ_GET("stats.arenas.0.lruns.0.ndalloc", &ndalloc,
+		    uint64_t);
+		CTL_IJ_GET("stats.arenas.0.lruns.0.nrequests", &nrequests,
+		    uint64_t);
+		if (nrequests == 0) {
+			if (gap_start == -1)
+				gap_start = j;
+		} else {
+			CTL_J_GET("arenas.lrun.0.size", &run_size, size_t);
+			CTL_IJ_GET("stats.arenas.0.lruns.0.curruns", &curruns,
+			    size_t);
+			if (gap_start != -1) {
+				malloc_cprintf(write_cb, cbopaque, "[%zu]\n",
+				    j - gap_start);
+				gap_start = -1;
+			}
+			malloc_cprintf(write_cb, cbopaque,
+			    "%13zu %5zu %12"PRIu64" %12"PRIu64" %12"PRIu64
+			    " %12zu\n",
+			    run_size, run_size / page, nmalloc, ndalloc,
+			    nrequests, curruns);
+		}
+	}
+	if (gap_start != -1)
+		malloc_cprintf(write_cb, cbopaque, "[%zu]\n", j - gap_start);
+}
+
+static void
+stats_arena_print(void (*write_cb)(void *, const char *), void *cbopaque,
+    unsigned i, bool bins, bool large)
+{
+	unsigned nthreads;
+	const char *dss;
+	size_t page, pactive, pdirty, mapped;
+	uint64_t npurge, nmadvise, purged;
+	size_t small_allocated;
+	uint64_t small_nmalloc, small_ndalloc, small_nrequests;
+	size_t large_allocated;
+	uint64_t large_nmalloc, large_ndalloc, large_nrequests;
+
+	CTL_GET("arenas.page", &page, size_t);
+
+	CTL_I_GET("stats.arenas.0.nthreads", &nthreads, unsigned);
+	malloc_cprintf(write_cb, cbopaque,
+	    "assigned threads: %u\n", nthreads);
+	CTL_I_GET("stats.arenas.0.dss", &dss, const char *);
+	malloc_cprintf(write_cb, cbopaque, "dss allocation precedence: %s\n",
+	    dss);
+	CTL_I_GET("stats.arenas.0.pactive", &pactive, size_t);
+	CTL_I_GET("stats.arenas.0.pdirty", &pdirty, size_t);
+	CTL_I_GET("stats.arenas.0.npurge", &npurge, uint64_t);
+	CTL_I_GET("stats.arenas.0.nmadvise", &nmadvise, uint64_t);
+	CTL_I_GET("stats.arenas.0.purged", &purged, uint64_t);
+	malloc_cprintf(write_cb, cbopaque,
+	    "dirty pages: %zu:%zu active:dirty, %"PRIu64" sweep%s,"
+	    " %"PRIu64" madvise%s, %"PRIu64" purged\n",
+	    pactive, pdirty, npurge, npurge == 1 ? "" : "s",
+	    nmadvise, nmadvise == 1 ? "" : "s", purged);
+
+	malloc_cprintf(write_cb, cbopaque,
+	    "            allocated      nmalloc      ndalloc    nrequests\n");
+	CTL_I_GET("stats.arenas.0.small.allocated", &small_allocated, size_t);
+	CTL_I_GET("stats.arenas.0.small.nmalloc", &small_nmalloc, uint64_t);
+	CTL_I_GET("stats.arenas.0.small.ndalloc", &small_ndalloc, uint64_t);
+	CTL_I_GET("stats.arenas.0.small.nrequests", &small_nrequests, uint64_t);
+	malloc_cprintf(write_cb, cbopaque,
+	    "small:   %12zu %12"PRIu64" %12"PRIu64" %12"PRIu64"\n",
+	    small_allocated, small_nmalloc, small_ndalloc, small_nrequests);
+	CTL_I_GET("stats.arenas.0.large.allocated", &large_allocated, size_t);
+	CTL_I_GET("stats.arenas.0.large.nmalloc", &large_nmalloc, uint64_t);
+	CTL_I_GET("stats.arenas.0.large.ndalloc", &large_ndalloc, uint64_t);
+	CTL_I_GET("stats.arenas.0.large.nrequests", &large_nrequests, uint64_t);
+	malloc_cprintf(write_cb, cbopaque,
+	    "large:   %12zu %12"PRIu64" %12"PRIu64" %12"PRIu64"\n",
+	    large_allocated, large_nmalloc, large_ndalloc, large_nrequests);
+	malloc_cprintf(write_cb, cbopaque,
+	    "total:   %12zu %12"PRIu64" %12"PRIu64" %12"PRIu64"\n",
+	    small_allocated + large_allocated,
+	    small_nmalloc + large_nmalloc,
+	    small_ndalloc + large_ndalloc,
+	    small_nrequests + large_nrequests);
+	malloc_cprintf(write_cb, cbopaque, "active:  %12zu\n", pactive * page);
+	CTL_I_GET("stats.arenas.0.mapped", &mapped, size_t);
+	malloc_cprintf(write_cb, cbopaque, "mapped:  %12zu\n", mapped);
+
+	if (bins)
+		stats_arena_bins_print(write_cb, cbopaque, i);
+	if (large)
+		stats_arena_lruns_print(write_cb, cbopaque, i);
+}
+
+void
+stats_print(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *opts)
+{
+	int err;
+	uint64_t epoch;
+	size_t u64sz;
+	bool general = true;
+	bool merged = true;
+	bool unmerged = true;
+	bool bins = true;
+	bool large = true;
+
+	/*
+	 * Refresh stats, in case mallctl() was called by the application.
+	 *
+	 * Check for OOM here, since refreshing the ctl cache can trigger
+	 * allocation.  In practice, none of the subsequent mallctl()-related
+	 * calls in this function will cause OOM if this one succeeds.
+	 * */
+	epoch = 1;
+	u64sz = sizeof(uint64_t);
+	err = je_mallctl("epoch", &epoch, &u64sz, &epoch, sizeof(uint64_t));
+	if (err != 0) {
+		if (err == EAGAIN) {
+			malloc_write("<jemalloc>: Memory allocation failure in "
+			    "mallctl(\"epoch\", ...)\n");
+			return;
+		}
+		malloc_write("<jemalloc>: Failure in mallctl(\"epoch\", "
+		    "...)\n");
+		abort();
+	}
+
+	if (opts != NULL) {
+		unsigned i;
+
+		for (i = 0; opts[i] != '\0'; i++) {
+			switch (opts[i]) {
+			case 'g':
+				general = false;
+				break;
+			case 'm':
+				merged = false;
+				break;
+			case 'a':
+				unmerged = false;
+				break;
+			case 'b':
+				bins = false;
+				break;
+			case 'l':
+				large = false;
+				break;
+			default:;
+			}
+		}
+	}
+
+	malloc_cprintf(write_cb, cbopaque,
+	    "___ Begin jemalloc statistics ___\n");
+	if (general) {
+		int err;
+		const char *cpv;
+		bool bv;
+		unsigned uv;
+		ssize_t ssv;
+		size_t sv, bsz, ssz, sssz, cpsz;
+
+		bsz = sizeof(bool);
+		ssz = sizeof(size_t);
+		sssz = sizeof(ssize_t);
+		cpsz = sizeof(const char *);
+
+		CTL_GET("version", &cpv, const char *);
+		malloc_cprintf(write_cb, cbopaque, "Version: %s\n", cpv);
+		CTL_GET("config.debug", &bv, bool);
+		malloc_cprintf(write_cb, cbopaque, "Assertions %s\n",
+		    bv ? "enabled" : "disabled");
+
+#define OPT_WRITE_BOOL(n)						\
+		if ((err = je_mallctl("opt."#n, &bv, &bsz, NULL, 0))	\
+		    == 0) {						\
+			malloc_cprintf(write_cb, cbopaque,		\
+			    "  opt."#n": %s\n", bv ? "true" : "false");	\
+		}
+#define OPT_WRITE_SIZE_T(n)						\
+		if ((err = je_mallctl("opt."#n, &sv, &ssz, NULL, 0))	\
+		    == 0) {						\
+			malloc_cprintf(write_cb, cbopaque,		\
+			"  opt."#n": %zu\n", sv);			\
+		}
+#define OPT_WRITE_SSIZE_T(n)						\
+		if ((err = je_mallctl("opt."#n, &ssv, &sssz, NULL, 0))	\
+		    == 0) {						\
+			malloc_cprintf(write_cb, cbopaque,		\
+			    "  opt."#n": %zd\n", ssv);			\
+		}
+#define OPT_WRITE_CHAR_P(n)						\
+		if ((err = je_mallctl("opt."#n, &cpv, &cpsz, NULL, 0))	\
+		    == 0) {						\
+			malloc_cprintf(write_cb, cbopaque,		\
+			    "  opt."#n": \"%s\"\n", cpv);		\
+		}
+
+		malloc_cprintf(write_cb, cbopaque,
+		    "Run-time option settings:\n");
+		OPT_WRITE_BOOL(abort)
+		OPT_WRITE_SIZE_T(lg_chunk)
+		OPT_WRITE_CHAR_P(dss)
+		OPT_WRITE_SIZE_T(narenas)
+		OPT_WRITE_SSIZE_T(lg_dirty_mult)
+		OPT_WRITE_BOOL(stats_print)
+		OPT_WRITE_BOOL(junk)
+		OPT_WRITE_SIZE_T(quarantine)
+		OPT_WRITE_BOOL(redzone)
+		OPT_WRITE_BOOL(zero)
+		OPT_WRITE_BOOL(utrace)
+		OPT_WRITE_BOOL(valgrind)
+		OPT_WRITE_BOOL(xmalloc)
+		OPT_WRITE_BOOL(tcache)
+		OPT_WRITE_SSIZE_T(lg_tcache_max)
+		OPT_WRITE_BOOL(prof)
+		OPT_WRITE_CHAR_P(prof_prefix)
+		OPT_WRITE_BOOL(prof_active)
+		OPT_WRITE_SSIZE_T(lg_prof_sample)
+		OPT_WRITE_BOOL(prof_accum)
+		OPT_WRITE_SSIZE_T(lg_prof_interval)
+		OPT_WRITE_BOOL(prof_gdump)
+		OPT_WRITE_BOOL(prof_final)
+		OPT_WRITE_BOOL(prof_leak)
+
+#undef OPT_WRITE_BOOL
+#undef OPT_WRITE_SIZE_T
+#undef OPT_WRITE_SSIZE_T
+#undef OPT_WRITE_CHAR_P
+
+		malloc_cprintf(write_cb, cbopaque, "CPUs: %u\n", ncpus);
+
+		CTL_GET("arenas.narenas", &uv, unsigned);
+		malloc_cprintf(write_cb, cbopaque, "Arenas: %u\n", uv);
+
+		malloc_cprintf(write_cb, cbopaque, "Pointer size: %zu\n",
+		    sizeof(void *));
+
+		CTL_GET("arenas.quantum", &sv, size_t);
+		malloc_cprintf(write_cb, cbopaque, "Quantum size: %zu\n", sv);
+
+		CTL_GET("arenas.page", &sv, size_t);
+		malloc_cprintf(write_cb, cbopaque, "Page size: %zu\n", sv);
+
+		CTL_GET("opt.lg_dirty_mult", &ssv, ssize_t);
+		if (ssv >= 0) {
+			malloc_cprintf(write_cb, cbopaque,
+			    "Min active:dirty page ratio per arena: %u:1\n",
+			    (1U << ssv));
+		} else {
+			malloc_cprintf(write_cb, cbopaque,
+			    "Min active:dirty page ratio per arena: N/A\n");
+		}
+		if ((err = je_mallctl("arenas.tcache_max", &sv, &ssz, NULL, 0))
+		    == 0) {
+			malloc_cprintf(write_cb, cbopaque,
+			    "Maximum thread-cached size class: %zu\n", sv);
+		}
+		if ((err = je_mallctl("opt.prof", &bv, &bsz, NULL, 0)) == 0 &&
+		    bv) {
+			CTL_GET("opt.lg_prof_sample", &sv, size_t);
+			malloc_cprintf(write_cb, cbopaque,
+			    "Average profile sample interval: %"PRIu64
+			    " (2^%zu)\n", (((uint64_t)1U) << sv), sv);
+
+			CTL_GET("opt.lg_prof_interval", &ssv, ssize_t);
+			if (ssv >= 0) {
+				malloc_cprintf(write_cb, cbopaque,
+				    "Average profile dump interval: %"PRIu64
+				    " (2^%zd)\n",
+				    (((uint64_t)1U) << ssv), ssv);
+			} else {
+				malloc_cprintf(write_cb, cbopaque,
+				    "Average profile dump interval: N/A\n");
+			}
+		}
+		CTL_GET("opt.lg_chunk", &sv, size_t);
+		malloc_cprintf(write_cb, cbopaque, "Chunk size: %zu (2^%zu)\n",
+		    (ZU(1) << sv), sv);
+	}
+
+	if (config_stats) {
+		size_t *cactive;
+		size_t allocated, active, mapped;
+		size_t chunks_current, chunks_high;
+		uint64_t chunks_total;
+		size_t huge_allocated;
+		uint64_t huge_nmalloc, huge_ndalloc;
+
+		CTL_GET("stats.cactive", &cactive, size_t *);
+		CTL_GET("stats.allocated", &allocated, size_t);
+		CTL_GET("stats.active", &active, size_t);
+		CTL_GET("stats.mapped", &mapped, size_t);
+		malloc_cprintf(write_cb, cbopaque,
+		    "Allocated: %zu, active: %zu, mapped: %zu\n",
+		    allocated, active, mapped);
+		malloc_cprintf(write_cb, cbopaque,
+		    "Current active ceiling: %zu\n", atomic_read_z(cactive));
+
+		/* Print chunk stats. */
+		CTL_GET("stats.chunks.total", &chunks_total, uint64_t);
+		CTL_GET("stats.chunks.high", &chunks_high, size_t);
+		CTL_GET("stats.chunks.current", &chunks_current, size_t);
+		malloc_cprintf(write_cb, cbopaque, "chunks: nchunks   "
+		    "highchunks    curchunks\n");
+		malloc_cprintf(write_cb, cbopaque,
+		    "  %13"PRIu64" %12zu %12zu\n",
+		    chunks_total, chunks_high, chunks_current);
+
+		/* Print huge stats. */
+		CTL_GET("stats.huge.nmalloc", &huge_nmalloc, uint64_t);
+		CTL_GET("stats.huge.ndalloc", &huge_ndalloc, uint64_t);
+		CTL_GET("stats.huge.allocated", &huge_allocated, size_t);
+		malloc_cprintf(write_cb, cbopaque,
+		    "huge: nmalloc      ndalloc    allocated\n");
+		malloc_cprintf(write_cb, cbopaque,
+		    " %12"PRIu64" %12"PRIu64" %12zu\n",
+		    huge_nmalloc, huge_ndalloc, huge_allocated);
+
+		if (merged) {
+			unsigned narenas;
+
+			CTL_GET("arenas.narenas", &narenas, unsigned);
+			{
+				VARIABLE_ARRAY(bool, initialized, narenas);
+				size_t isz;
+				unsigned i, ninitialized;
+
+				isz = sizeof(bool) * narenas;
+				xmallctl("arenas.initialized", initialized,
+				    &isz, NULL, 0);
+				for (i = ninitialized = 0; i < narenas; i++) {
+					if (initialized[i])
+						ninitialized++;
+				}
+
+				if (ninitialized > 1 || unmerged == false) {
+					/* Print merged arena stats. */
+					malloc_cprintf(write_cb, cbopaque,
+					    "\nMerged arenas stats:\n");
+					stats_arena_print(write_cb, cbopaque,
+					    narenas, bins, large);
+				}
+			}
+		}
+
+		if (unmerged) {
+			unsigned narenas;
+
+			/* Print stats for each arena. */
+
+			CTL_GET("arenas.narenas", &narenas, unsigned);
+			{
+				VARIABLE_ARRAY(bool, initialized, narenas);
+				size_t isz;
+				unsigned i;
+
+				isz = sizeof(bool) * narenas;
+				xmallctl("arenas.initialized", initialized,
+				    &isz, NULL, 0);
+
+				for (i = 0; i < narenas; i++) {
+					if (initialized[i]) {
+						malloc_cprintf(write_cb,
+						    cbopaque,
+						    "\narenas[%u]:\n", i);
+						stats_arena_print(write_cb,
+						    cbopaque, i, bins, large);
+					}
+				}
+			}
+		}
+	}
+	malloc_cprintf(write_cb, cbopaque, "--- End jemalloc statistics ---\n");
+}
diff --git a/extra/jemalloc/src/tcache.c b/extra/jemalloc/src/tcache.c
new file mode 100644
index 00000000000..98ed19edd52
--- /dev/null
+++ b/extra/jemalloc/src/tcache.c
@@ -0,0 +1,476 @@
+#define	JEMALLOC_TCACHE_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Data. */
+
+malloc_tsd_data(, tcache, tcache_t *, NULL)
+malloc_tsd_data(, tcache_enabled, tcache_enabled_t, tcache_enabled_default)
+
+bool	opt_tcache = true;
+ssize_t	opt_lg_tcache_max = LG_TCACHE_MAXCLASS_DEFAULT;
+
+tcache_bin_info_t	*tcache_bin_info;
+static unsigned		stack_nelms; /* Total stack elms per tcache. */
+
+size_t			nhbins;
+size_t			tcache_maxclass;
+
+/******************************************************************************/
+
+size_t	tcache_salloc(const void *ptr)
+{
+
+	return (arena_salloc(ptr, false));
+}
+
+void
+tcache_event_hard(tcache_t *tcache)
+{
+	size_t binind = tcache->next_gc_bin;
+	tcache_bin_t *tbin = &tcache->tbins[binind];
+	tcache_bin_info_t *tbin_info = &tcache_bin_info[binind];
+
+	if (tbin->low_water > 0) {
+		/*
+		 * Flush (ceiling) 3/4 of the objects below the low water mark.
+		 */
+		if (binind < NBINS) {
+			tcache_bin_flush_small(tbin, binind, tbin->ncached -
+			    tbin->low_water + (tbin->low_water >> 2), tcache);
+		} else {
+			tcache_bin_flush_large(tbin, binind, tbin->ncached -
+			    tbin->low_water + (tbin->low_water >> 2), tcache);
+		}
+		/*
+		 * Reduce fill count by 2X.  Limit lg_fill_div such that the
+		 * fill count is always at least 1.
+		 */
+		if ((tbin_info->ncached_max >> (tbin->lg_fill_div+1)) >= 1)
+			tbin->lg_fill_div++;
+	} else if (tbin->low_water < 0) {
+		/*
+		 * Increase fill count by 2X.  Make sure lg_fill_div stays
+		 * greater than 0.
+		 */
+		if (tbin->lg_fill_div > 1)
+			tbin->lg_fill_div--;
+	}
+	tbin->low_water = tbin->ncached;
+
+	tcache->next_gc_bin++;
+	if (tcache->next_gc_bin == nhbins)
+		tcache->next_gc_bin = 0;
+	tcache->ev_cnt = 0;
+}
+
+void *
+tcache_alloc_small_hard(tcache_t *tcache, tcache_bin_t *tbin, size_t binind)
+{
+	void *ret;
+
+	arena_tcache_fill_small(tcache->arena, tbin, binind,
+	    config_prof ? tcache->prof_accumbytes : 0);
+	if (config_prof)
+		tcache->prof_accumbytes = 0;
+	ret = tcache_alloc_easy(tbin);
+
+	return (ret);
+}
+
+void
+tcache_bin_flush_small(tcache_bin_t *tbin, size_t binind, unsigned rem,
+    tcache_t *tcache)
+{
+	void *ptr;
+	unsigned i, nflush, ndeferred;
+	bool merged_stats = false;
+
+	assert(binind < NBINS);
+	assert(rem <= tbin->ncached);
+
+	for (nflush = tbin->ncached - rem; nflush > 0; nflush = ndeferred) {
+		/* Lock the arena bin associated with the first object. */
+		arena_chunk_t *chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(
+		    tbin->avail[0]);
+		arena_t *arena = chunk->arena;
+		arena_bin_t *bin = &arena->bins[binind];
+
+		if (config_prof && arena == tcache->arena) {
+			if (arena_prof_accum(arena, tcache->prof_accumbytes))
+				prof_idump();
+			tcache->prof_accumbytes = 0;
+		}
+
+		malloc_mutex_lock(&bin->lock);
+		if (config_stats && arena == tcache->arena) {
+			assert(merged_stats == false);
+			merged_stats = true;
+			bin->stats.nflushes++;
+			bin->stats.nrequests += tbin->tstats.nrequests;
+			tbin->tstats.nrequests = 0;
+		}
+		ndeferred = 0;
+		for (i = 0; i < nflush; i++) {
+			ptr = tbin->avail[i];
+			assert(ptr != NULL);
+			chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+			if (chunk->arena == arena) {
+				size_t pageind = ((uintptr_t)ptr -
+				    (uintptr_t)chunk) >> LG_PAGE;
+				arena_chunk_map_t *mapelm =
+				    arena_mapp_get(chunk, pageind);
+				if (config_fill && opt_junk) {
+					arena_alloc_junk_small(ptr,
+					    &arena_bin_info[binind], true);
+				}
+				arena_dalloc_bin_locked(arena, chunk, ptr,
+				    mapelm);
+			} else {
+				/*
+				 * This object was allocated via a different
+				 * arena bin than the one that is currently
+				 * locked.  Stash the object, so that it can be
+				 * handled in a future pass.
+				 */
+				tbin->avail[ndeferred] = ptr;
+				ndeferred++;
+			}
+		}
+		malloc_mutex_unlock(&bin->lock);
+	}
+	if (config_stats && merged_stats == false) {
+		/*
+		 * The flush loop didn't happen to flush to this thread's
+		 * arena, so the stats didn't get merged.  Manually do so now.
+		 */
+		arena_bin_t *bin = &tcache->arena->bins[binind];
+		malloc_mutex_lock(&bin->lock);
+		bin->stats.nflushes++;
+		bin->stats.nrequests += tbin->tstats.nrequests;
+		tbin->tstats.nrequests = 0;
+		malloc_mutex_unlock(&bin->lock);
+	}
+
+	memmove(tbin->avail, &tbin->avail[tbin->ncached - rem],
+	    rem * sizeof(void *));
+	tbin->ncached = rem;
+	if ((int)tbin->ncached < tbin->low_water)
+		tbin->low_water = tbin->ncached;
+}
+
+void
+tcache_bin_flush_large(tcache_bin_t *tbin, size_t binind, unsigned rem,
+    tcache_t *tcache)
+{
+	void *ptr;
+	unsigned i, nflush, ndeferred;
+	bool merged_stats = false;
+
+	assert(binind < nhbins);
+	assert(rem <= tbin->ncached);
+
+	for (nflush = tbin->ncached - rem; nflush > 0; nflush = ndeferred) {
+		/* Lock the arena associated with the first object. */
+		arena_chunk_t *chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(
+		    tbin->avail[0]);
+		arena_t *arena = chunk->arena;
+		UNUSED bool idump;
+
+		if (config_prof)
+			idump = false;
+		malloc_mutex_lock(&arena->lock);
+		if ((config_prof || config_stats) && arena == tcache->arena) {
+			if (config_prof) {
+				idump = arena_prof_accum_locked(arena,
+				    tcache->prof_accumbytes);
+				tcache->prof_accumbytes = 0;
+			}
+			if (config_stats) {
+				merged_stats = true;
+				arena->stats.nrequests_large +=
+				    tbin->tstats.nrequests;
+				arena->stats.lstats[binind - NBINS].nrequests +=
+				    tbin->tstats.nrequests;
+				tbin->tstats.nrequests = 0;
+			}
+		}
+		ndeferred = 0;
+		for (i = 0; i < nflush; i++) {
+			ptr = tbin->avail[i];
+			assert(ptr != NULL);
+			chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+			if (chunk->arena == arena)
+				arena_dalloc_large_locked(arena, chunk, ptr);
+			else {
+				/*
+				 * This object was allocated via a different
+				 * arena than the one that is currently locked.
+				 * Stash the object, so that it can be handled
+				 * in a future pass.
+				 */
+				tbin->avail[ndeferred] = ptr;
+				ndeferred++;
+			}
+		}
+		malloc_mutex_unlock(&arena->lock);
+		if (config_prof && idump)
+			prof_idump();
+	}
+	if (config_stats && merged_stats == false) {
+		/*
+		 * The flush loop didn't happen to flush to this thread's
+		 * arena, so the stats didn't get merged.  Manually do so now.
+		 */
+		arena_t *arena = tcache->arena;
+		malloc_mutex_lock(&arena->lock);
+		arena->stats.nrequests_large += tbin->tstats.nrequests;
+		arena->stats.lstats[binind - NBINS].nrequests +=
+		    tbin->tstats.nrequests;
+		tbin->tstats.nrequests = 0;
+		malloc_mutex_unlock(&arena->lock);
+	}
+
+	memmove(tbin->avail, &tbin->avail[tbin->ncached - rem],
+	    rem * sizeof(void *));
+	tbin->ncached = rem;
+	if ((int)tbin->ncached < tbin->low_water)
+		tbin->low_water = tbin->ncached;
+}
+
+void
+tcache_arena_associate(tcache_t *tcache, arena_t *arena)
+{
+
+	if (config_stats) {
+		/* Link into list of extant tcaches. */
+		malloc_mutex_lock(&arena->lock);
+		ql_elm_new(tcache, link);
+		ql_tail_insert(&arena->tcache_ql, tcache, link);
+		malloc_mutex_unlock(&arena->lock);
+	}
+	tcache->arena = arena;
+}
+
+void
+tcache_arena_dissociate(tcache_t *tcache)
+{
+
+	if (config_stats) {
+		/* Unlink from list of extant tcaches. */
+		malloc_mutex_lock(&tcache->arena->lock);
+		ql_remove(&tcache->arena->tcache_ql, tcache, link);
+		malloc_mutex_unlock(&tcache->arena->lock);
+		tcache_stats_merge(tcache, tcache->arena);
+	}
+}
+
+tcache_t *
+tcache_create(arena_t *arena)
+{
+	tcache_t *tcache;
+	size_t size, stack_offset;
+	unsigned i;
+
+	size = offsetof(tcache_t, tbins) + (sizeof(tcache_bin_t) * nhbins);
+	/* Naturally align the pointer stacks. */
+	size = PTR_CEILING(size);
+	stack_offset = size;
+	size += stack_nelms * sizeof(void *);
+	/*
+	 * Round up to the nearest multiple of the cacheline size, in order to
+	 * avoid the possibility of false cacheline sharing.
+	 *
+	 * That this works relies on the same logic as in ipalloc(), but we
+	 * cannot directly call ipalloc() here due to tcache bootstrapping
+	 * issues.
+	 */
+	size = (size + CACHELINE_MASK) & (-CACHELINE);
+
+	if (size <= SMALL_MAXCLASS)
+		tcache = (tcache_t *)arena_malloc_small(arena, size, true);
+	else if (size <= tcache_maxclass)
+		tcache = (tcache_t *)arena_malloc_large(arena, size, true);
+	else
+		tcache = (tcache_t *)icallocx(size, false, arena);
+
+	if (tcache == NULL)
+		return (NULL);
+
+	tcache_arena_associate(tcache, arena);
+
+	assert((TCACHE_NSLOTS_SMALL_MAX & 1U) == 0);
+	for (i = 0; i < nhbins; i++) {
+		tcache->tbins[i].lg_fill_div = 1;
+		tcache->tbins[i].avail = (void **)((uintptr_t)tcache +
+		    (uintptr_t)stack_offset);
+		stack_offset += tcache_bin_info[i].ncached_max * sizeof(void *);
+	}
+
+	tcache_tsd_set(&tcache);
+
+	return (tcache);
+}
+
+void
+tcache_destroy(tcache_t *tcache)
+{
+	unsigned i;
+	size_t tcache_size;
+
+	tcache_arena_dissociate(tcache);
+
+	for (i = 0; i < NBINS; i++) {
+		tcache_bin_t *tbin = &tcache->tbins[i];
+		tcache_bin_flush_small(tbin, i, 0, tcache);
+
+		if (config_stats && tbin->tstats.nrequests != 0) {
+			arena_t *arena = tcache->arena;
+			arena_bin_t *bin = &arena->bins[i];
+			malloc_mutex_lock(&bin->lock);
+			bin->stats.nrequests += tbin->tstats.nrequests;
+			malloc_mutex_unlock(&bin->lock);
+		}
+	}
+
+	for (; i < nhbins; i++) {
+		tcache_bin_t *tbin = &tcache->tbins[i];
+		tcache_bin_flush_large(tbin, i, 0, tcache);
+
+		if (config_stats && tbin->tstats.nrequests != 0) {
+			arena_t *arena = tcache->arena;
+			malloc_mutex_lock(&arena->lock);
+			arena->stats.nrequests_large += tbin->tstats.nrequests;
+			arena->stats.lstats[i - NBINS].nrequests +=
+			    tbin->tstats.nrequests;
+			malloc_mutex_unlock(&arena->lock);
+		}
+	}
+
+	if (config_prof && tcache->prof_accumbytes > 0 &&
+	    arena_prof_accum(tcache->arena, tcache->prof_accumbytes))
+		prof_idump();
+
+	tcache_size = arena_salloc(tcache, false);
+	if (tcache_size <= SMALL_MAXCLASS) {
+		arena_chunk_t *chunk = CHUNK_ADDR2BASE(tcache);
+		arena_t *arena = chunk->arena;
+		size_t pageind = ((uintptr_t)tcache - (uintptr_t)chunk) >>
+		    LG_PAGE;
+		arena_chunk_map_t *mapelm = arena_mapp_get(chunk, pageind);
+
+		arena_dalloc_bin(arena, chunk, tcache, pageind, mapelm);
+	} else if (tcache_size <= tcache_maxclass) {
+		arena_chunk_t *chunk = CHUNK_ADDR2BASE(tcache);
+		arena_t *arena = chunk->arena;
+
+		arena_dalloc_large(arena, chunk, tcache);
+	} else
+		idallocx(tcache, false);
+}
+
+void
+tcache_thread_cleanup(void *arg)
+{
+	tcache_t *tcache = *(tcache_t **)arg;
+
+	if (tcache == TCACHE_STATE_DISABLED) {
+		/* Do nothing. */
+	} else if (tcache == TCACHE_STATE_REINCARNATED) {
+		/*
+		 * Another destructor called an allocator function after this
+		 * destructor was called.  Reset tcache to
+		 * TCACHE_STATE_PURGATORY in order to receive another callback.
+		 */
+		tcache = TCACHE_STATE_PURGATORY;
+		tcache_tsd_set(&tcache);
+	} else if (tcache == TCACHE_STATE_PURGATORY) {
+		/*
+		 * The previous time this destructor was called, we set the key
+		 * to TCACHE_STATE_PURGATORY so that other destructors wouldn't
+		 * cause re-creation of the tcache.  This time, do nothing, so
+		 * that the destructor will not be called again.
+		 */
+	} else if (tcache != NULL) {
+		assert(tcache != TCACHE_STATE_PURGATORY);
+		tcache_destroy(tcache);
+		tcache = TCACHE_STATE_PURGATORY;
+		tcache_tsd_set(&tcache);
+	}
+}
+
+void
+tcache_stats_merge(tcache_t *tcache, arena_t *arena)
+{
+	unsigned i;
+
+	/* Merge and reset tcache stats. */
+	for (i = 0; i < NBINS; i++) {
+		arena_bin_t *bin = &arena->bins[i];
+		tcache_bin_t *tbin = &tcache->tbins[i];
+		malloc_mutex_lock(&bin->lock);
+		bin->stats.nrequests += tbin->tstats.nrequests;
+		malloc_mutex_unlock(&bin->lock);
+		tbin->tstats.nrequests = 0;
+	}
+
+	for (; i < nhbins; i++) {
+		malloc_large_stats_t *lstats = &arena->stats.lstats[i - NBINS];
+		tcache_bin_t *tbin = &tcache->tbins[i];
+		arena->stats.nrequests_large += tbin->tstats.nrequests;
+		lstats->nrequests += tbin->tstats.nrequests;
+		tbin->tstats.nrequests = 0;
+	}
+}
+
+bool
+tcache_boot0(void)
+{
+	unsigned i;
+
+	/*
+	 * If necessary, clamp opt_lg_tcache_max, now that arena_maxclass is
+	 * known.
+	 */
+	if (opt_lg_tcache_max < 0 || (1U << opt_lg_tcache_max) < SMALL_MAXCLASS)
+		tcache_maxclass = SMALL_MAXCLASS;
+	else if ((1U << opt_lg_tcache_max) > arena_maxclass)
+		tcache_maxclass = arena_maxclass;
+	else
+		tcache_maxclass = (1U << opt_lg_tcache_max);
+
+	nhbins = NBINS + (tcache_maxclass >> LG_PAGE);
+
+	/* Initialize tcache_bin_info. */
+	tcache_bin_info = (tcache_bin_info_t *)base_alloc(nhbins *
+	    sizeof(tcache_bin_info_t));
+	if (tcache_bin_info == NULL)
+		return (true);
+	stack_nelms = 0;
+	for (i = 0; i < NBINS; i++) {
+		if ((arena_bin_info[i].nregs << 1) <= TCACHE_NSLOTS_SMALL_MAX) {
+			tcache_bin_info[i].ncached_max =
+			    (arena_bin_info[i].nregs << 1);
+		} else {
+			tcache_bin_info[i].ncached_max =
+			    TCACHE_NSLOTS_SMALL_MAX;
+		}
+		stack_nelms += tcache_bin_info[i].ncached_max;
+	}
+	for (; i < nhbins; i++) {
+		tcache_bin_info[i].ncached_max = TCACHE_NSLOTS_LARGE;
+		stack_nelms += tcache_bin_info[i].ncached_max;
+	}
+
+	return (false);
+}
+
+bool
+tcache_boot1(void)
+{
+
+	if (tcache_tsd_boot() || tcache_enabled_tsd_boot())
+		return (true);
+
+	return (false);
+}
diff --git a/extra/jemalloc/src/tsd.c b/extra/jemalloc/src/tsd.c
new file mode 100644
index 00000000000..961a546329c
--- /dev/null
+++ b/extra/jemalloc/src/tsd.c
@@ -0,0 +1,107 @@
+#define	JEMALLOC_TSD_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Data. */
+
+static unsigned ncleanups;
+static malloc_tsd_cleanup_t cleanups[MALLOC_TSD_CLEANUPS_MAX];
+
+/******************************************************************************/
+
+void *
+malloc_tsd_malloc(size_t size)
+{
+
+	/* Avoid choose_arena() in order to dodge bootstrapping issues. */
+	return (arena_malloc(arenas[0], size, false, false));
+}
+
+void
+malloc_tsd_dalloc(void *wrapper)
+{
+
+	idalloc(wrapper);
+}
+
+void
+malloc_tsd_no_cleanup(void *arg)
+{
+
+	not_reached();
+}
+
+#if defined(JEMALLOC_MALLOC_THREAD_CLEANUP) || defined(_WIN32)
+#ifndef _WIN32
+JEMALLOC_EXPORT
+#endif
+void
+_malloc_thread_cleanup(void)
+{
+	bool pending[MALLOC_TSD_CLEANUPS_MAX], again;
+	unsigned i;
+
+	for (i = 0; i < ncleanups; i++)
+		pending[i] = true;
+
+	do {
+		again = false;
+		for (i = 0; i < ncleanups; i++) {
+			if (pending[i]) {
+				pending[i] = cleanups[i]();
+				if (pending[i])
+					again = true;
+			}
+		}
+	} while (again);
+}
+#endif
+
+void
+malloc_tsd_cleanup_register(bool (*f)(void))
+{
+
+	assert(ncleanups < MALLOC_TSD_CLEANUPS_MAX);
+	cleanups[ncleanups] = f;
+	ncleanups++;
+}
+
+void
+malloc_tsd_boot(void)
+{
+
+	ncleanups = 0;
+}
+
+#ifdef _WIN32
+static BOOL WINAPI
+_tls_callback(HINSTANCE hinstDLL, DWORD fdwReason, LPVOID lpvReserved)
+{
+
+	switch (fdwReason) {
+#ifdef JEMALLOC_LAZY_LOCK
+	case DLL_THREAD_ATTACH:
+		isthreaded = true;
+		break;
+#endif
+	case DLL_THREAD_DETACH:
+		_malloc_thread_cleanup();
+		break;
+	default:
+		break;
+	}
+	return (true);
+}
+
+#ifdef _MSC_VER
+#  ifdef _M_IX86
+#    pragma comment(linker, "/INCLUDE:__tls_used")
+#  else
+#    pragma comment(linker, "/INCLUDE:_tls_used")
+#  endif
+#  pragma section(".CRT$XLY",long,read)
+#endif
+JEMALLOC_SECTION(".CRT$XLY") JEMALLOC_ATTR(used)
+static const BOOL	(WINAPI *tls_callback)(HINSTANCE hinstDLL,
+    DWORD fdwReason, LPVOID lpvReserved) = _tls_callback;
+#endif
diff --git a/extra/jemalloc/src/util.c b/extra/jemalloc/src/util.c
new file mode 100644
index 00000000000..b3a01143698
--- /dev/null
+++ b/extra/jemalloc/src/util.c
@@ -0,0 +1,641 @@
+#define	assert(e) do {							\
+	if (config_debug && !(e)) {					\
+		malloc_write("<jemalloc>: Failed assertion\n");		\
+		abort();						\
+	}								\
+} while (0)
+
+#define	not_reached() do {						\
+	if (config_debug) {						\
+		malloc_write("<jemalloc>: Unreachable code reached\n");	\
+		abort();						\
+	}								\
+} while (0)
+
+#define	not_implemented() do {						\
+	if (config_debug) {						\
+		malloc_write("<jemalloc>: Not implemented\n");		\
+		abort();						\
+	}								\
+} while (0)
+
+#define	JEMALLOC_UTIL_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static void	wrtmessage(void *cbopaque, const char *s);
+#define	U2S_BUFSIZE	((1U << (LG_SIZEOF_INTMAX_T + 3)) + 1)
+static char	*u2s(uintmax_t x, unsigned base, bool uppercase, char *s,
+    size_t *slen_p);
+#define	D2S_BUFSIZE	(1 + U2S_BUFSIZE)
+static char	*d2s(intmax_t x, char sign, char *s, size_t *slen_p);
+#define	O2S_BUFSIZE	(1 + U2S_BUFSIZE)
+static char	*o2s(uintmax_t x, bool alt_form, char *s, size_t *slen_p);
+#define	X2S_BUFSIZE	(2 + U2S_BUFSIZE)
+static char	*x2s(uintmax_t x, bool alt_form, bool uppercase, char *s,
+    size_t *slen_p);
+
+/******************************************************************************/
+
+/* malloc_message() setup. */
+static void
+wrtmessage(void *cbopaque, const char *s)
+{
+
+#ifdef SYS_write
+	/*
+	 * Use syscall(2) rather than write(2) when possible in order to avoid
+	 * the possibility of memory allocation within libc.  This is necessary
+	 * on FreeBSD; most operating systems do not have this problem though.
+	 */
+	UNUSED int result = syscall(SYS_write, STDERR_FILENO, s, strlen(s));
+#else
+	UNUSED int result = write(STDERR_FILENO, s, strlen(s));
+#endif
+}
+
+JEMALLOC_EXPORT void	(*je_malloc_message)(void *, const char *s);
+
+/*
+ * Wrapper around malloc_message() that avoids the need for
+ * je_malloc_message(...) throughout the code.
+ */
+void
+malloc_write(const char *s)
+{
+
+	if (je_malloc_message != NULL)
+		je_malloc_message(NULL, s);
+	else
+		wrtmessage(NULL, s);
+}
+
+/*
+ * glibc provides a non-standard strerror_r() when _GNU_SOURCE is defined, so
+ * provide a wrapper.
+ */
+int
+buferror(char *buf, size_t buflen)
+{
+
+#ifdef _WIN32
+	FormatMessageA(FORMAT_MESSAGE_FROM_SYSTEM, NULL, GetLastError(), 0,
+	    (LPSTR)buf, buflen, NULL);
+	return (0);
+#elif defined(_GNU_SOURCE)
+	char *b = strerror_r(errno, buf, buflen);
+	if (b != buf) {
+		strncpy(buf, b, buflen);
+		buf[buflen-1] = '\0';
+	}
+	return (0);
+#else
+	return (strerror_r(errno, buf, buflen));
+#endif
+}
+
+uintmax_t
+malloc_strtoumax(const char *nptr, char **endptr, int base)
+{
+	uintmax_t ret, digit;
+	int b;
+	bool neg;
+	const char *p, *ns;
+
+	if (base < 0 || base == 1 || base > 36) {
+		set_errno(EINVAL);
+		return (UINTMAX_MAX);
+	}
+	b = base;
+
+	/* Swallow leading whitespace and get sign, if any. */
+	neg = false;
+	p = nptr;
+	while (true) {
+		switch (*p) {
+		case '\t': case '\n': case '\v': case '\f': case '\r': case ' ':
+			p++;
+			break;
+		case '-':
+			neg = true;
+			/* Fall through. */
+		case '+':
+			p++;
+			/* Fall through. */
+		default:
+			goto label_prefix;
+		}
+	}
+
+	/* Get prefix, if any. */
+	label_prefix:
+	/*
+	 * Note where the first non-whitespace/sign character is so that it is
+	 * possible to tell whether any digits are consumed (e.g., "  0" vs.
+	 * "  -x").
+	 */
+	ns = p;
+	if (*p == '0') {
+		switch (p[1]) {
+		case '0': case '1': case '2': case '3': case '4': case '5':
+		case '6': case '7':
+			if (b == 0)
+				b = 8;
+			if (b == 8)
+				p++;
+			break;
+		case 'x':
+			switch (p[2]) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+			case 'A': case 'B': case 'C': case 'D': case 'E':
+			case 'F':
+			case 'a': case 'b': case 'c': case 'd': case 'e':
+			case 'f':
+				if (b == 0)
+					b = 16;
+				if (b == 16)
+					p += 2;
+				break;
+			default:
+				break;
+			}
+			break;
+		default:
+			break;
+		}
+	}
+	if (b == 0)
+		b = 10;
+
+	/* Convert. */
+	ret = 0;
+	while ((*p >= '0' && *p <= '9' && (digit = *p - '0') < b)
+	    || (*p >= 'A' && *p <= 'Z' && (digit = 10 + *p - 'A') < b)
+	    || (*p >= 'a' && *p <= 'z' && (digit = 10 + *p - 'a') < b)) {
+		uintmax_t pret = ret;
+		ret *= b;
+		ret += digit;
+		if (ret < pret) {
+			/* Overflow. */
+			set_errno(ERANGE);
+			return (UINTMAX_MAX);
+		}
+		p++;
+	}
+	if (neg)
+		ret = -ret;
+
+	if (endptr != NULL) {
+		if (p == ns) {
+			/* No characters were converted. */
+			*endptr = (char *)nptr;
+		} else
+			*endptr = (char *)p;
+	}
+
+	return (ret);
+}
+
+static char *
+u2s(uintmax_t x, unsigned base, bool uppercase, char *s, size_t *slen_p)
+{
+	unsigned i;
+
+	i = U2S_BUFSIZE - 1;
+	s[i] = '\0';
+	switch (base) {
+	case 10:
+		do {
+			i--;
+			s[i] = "0123456789"[x % (uint64_t)10];
+			x /= (uint64_t)10;
+		} while (x > 0);
+		break;
+	case 16: {
+		const char *digits = (uppercase)
+		    ? "0123456789ABCDEF"
+		    : "0123456789abcdef";
+
+		do {
+			i--;
+			s[i] = digits[x & 0xf];
+			x >>= 4;
+		} while (x > 0);
+		break;
+	} default: {
+		const char *digits = (uppercase)
+		    ? "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"
+		    : "0123456789abcdefghijklmnopqrstuvwxyz";
+
+		assert(base >= 2 && base <= 36);
+		do {
+			i--;
+			s[i] = digits[x % (uint64_t)base];
+			x /= (uint64_t)base;
+		} while (x > 0);
+	}}
+
+	*slen_p = U2S_BUFSIZE - 1 - i;
+	return (&s[i]);
+}
+
+static char *
+d2s(intmax_t x, char sign, char *s, size_t *slen_p)
+{
+	bool neg;
+
+	if ((neg = (x < 0)))
+		x = -x;
+	s = u2s(x, 10, false, s, slen_p);
+	if (neg)
+		sign = '-';
+	switch (sign) {
+	case '-':
+		if (neg == false)
+			break;
+		/* Fall through. */
+	case ' ':
+	case '+':
+		s--;
+		(*slen_p)++;
+		*s = sign;
+		break;
+	default: not_reached();
+	}
+	return (s);
+}
+
+static char *
+o2s(uintmax_t x, bool alt_form, char *s, size_t *slen_p)
+{
+
+	s = u2s(x, 8, false, s, slen_p);
+	if (alt_form && *s != '0') {
+		s--;
+		(*slen_p)++;
+		*s = '0';
+	}
+	return (s);
+}
+
+static char *
+x2s(uintmax_t x, bool alt_form, bool uppercase, char *s, size_t *slen_p)
+{
+
+	s = u2s(x, 16, uppercase, s, slen_p);
+	if (alt_form) {
+		s -= 2;
+		(*slen_p) += 2;
+		memcpy(s, uppercase ? "0X" : "0x", 2);
+	}
+	return (s);
+}
+
+int
+malloc_vsnprintf(char *str, size_t size, const char *format, va_list ap)
+{
+	int ret;
+	size_t i;
+	const char *f;
+
+#define	APPEND_C(c) do {						\
+	if (i < size)							\
+		str[i] = (c);						\
+	i++;								\
+} while (0)
+#define	APPEND_S(s, slen) do {						\
+	if (i < size) {							\
+		size_t cpylen = (slen <= size - i) ? slen : size - i;	\
+		memcpy(&str[i], s, cpylen);				\
+	}								\
+	i += slen;							\
+} while (0)
+#define	APPEND_PADDED_S(s, slen, width, left_justify) do {		\
+	/* Left padding. */						\
+	size_t pad_len = (width == -1) ? 0 : ((slen < (size_t)width) ?	\
+	    (size_t)width - slen : 0);					\
+	if (left_justify == false && pad_len != 0) {			\
+		size_t j;						\
+		for (j = 0; j < pad_len; j++)				\
+			APPEND_C(' ');					\
+	}								\
+	/* Value. */							\
+	APPEND_S(s, slen);						\
+	/* Right padding. */						\
+	if (left_justify && pad_len != 0) {				\
+		size_t j;						\
+		for (j = 0; j < pad_len; j++)				\
+			APPEND_C(' ');					\
+	}								\
+} while (0)
+#define GET_ARG_NUMERIC(val, len) do {					\
+	switch (len) {							\
+	case '?':							\
+		val = va_arg(ap, int);					\
+		break;							\
+	case '?' | 0x80:						\
+		val = va_arg(ap, unsigned int);				\
+		break;							\
+	case 'l':							\
+		val = va_arg(ap, long);					\
+		break;							\
+	case 'l' | 0x80:						\
+		val = va_arg(ap, unsigned long);			\
+		break;							\
+	case 'q':							\
+		val = va_arg(ap, long long);				\
+		break;							\
+	case 'q' | 0x80:						\
+		val = va_arg(ap, unsigned long long);			\
+		break;							\
+	case 'j':							\
+		val = va_arg(ap, intmax_t);				\
+		break;							\
+	case 't':							\
+		val = va_arg(ap, ptrdiff_t);				\
+		break;							\
+	case 'z':							\
+		val = va_arg(ap, ssize_t);				\
+		break;							\
+	case 'z' | 0x80:						\
+		val = va_arg(ap, size_t);				\
+		break;							\
+	case 'p': /* Synthetic; used for %p. */				\
+		val = va_arg(ap, uintptr_t);				\
+		break;							\
+	default: not_reached();						\
+	}								\
+} while (0)
+
+	i = 0;
+	f = format;
+	while (true) {
+		switch (*f) {
+		case '\0': goto label_out;
+		case '%': {
+			bool alt_form = false;
+			bool left_justify = false;
+			bool plus_space = false;
+			bool plus_plus = false;
+			int prec = -1;
+			int width = -1;
+			unsigned char len = '?';
+
+			f++;
+			if (*f == '%') {
+				/* %% */
+				APPEND_C(*f);
+				break;
+			}
+			/* Flags. */
+			while (true) {
+				switch (*f) {
+				case '#':
+					assert(alt_form == false);
+					alt_form = true;
+					break;
+				case '-':
+					assert(left_justify == false);
+					left_justify = true;
+					break;
+				case ' ':
+					assert(plus_space == false);
+					plus_space = true;
+					break;
+				case '+':
+					assert(plus_plus == false);
+					plus_plus = true;
+					break;
+				default: goto label_width;
+				}
+				f++;
+			}
+			/* Width. */
+			label_width:
+			switch (*f) {
+			case '*':
+				width = va_arg(ap, int);
+				f++;
+				break;
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9': {
+				uintmax_t uwidth;
+				set_errno(0);
+				uwidth = malloc_strtoumax(f, (char **)&f, 10);
+				assert(uwidth != UINTMAX_MAX || get_errno() !=
+				    ERANGE);
+				width = (int)uwidth;
+				if (*f == '.') {
+					f++;
+					goto label_precision;
+				} else
+					goto label_length;
+				break;
+			} case '.':
+				f++;
+				goto label_precision;
+			default: goto label_length;
+			}
+			/* Precision. */
+			label_precision:
+			switch (*f) {
+			case '*':
+				prec = va_arg(ap, int);
+				f++;
+				break;
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9': {
+				uintmax_t uprec;
+				set_errno(0);
+				uprec = malloc_strtoumax(f, (char **)&f, 10);
+				assert(uprec != UINTMAX_MAX || get_errno() !=
+				    ERANGE);
+				prec = (int)uprec;
+				break;
+			}
+			default: break;
+			}
+			/* Length. */
+			label_length:
+			switch (*f) {
+			case 'l':
+				f++;
+				if (*f == 'l') {
+					len = 'q';
+					f++;
+				} else
+					len = 'l';
+				break;
+			case 'j':
+				len = 'j';
+				f++;
+				break;
+			case 't':
+				len = 't';
+				f++;
+				break;
+			case 'z':
+				len = 'z';
+				f++;
+				break;
+			default: break;
+			}
+			/* Conversion specifier. */
+			switch (*f) {
+				char *s;
+				size_t slen;
+			case 'd': case 'i': {
+				intmax_t val JEMALLOC_CC_SILENCE_INIT(0);
+				char buf[D2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, len);
+				s = d2s(val, (plus_plus ? '+' : (plus_space ?
+				    ' ' : '-')), buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} case 'o': {
+				uintmax_t val JEMALLOC_CC_SILENCE_INIT(0);
+				char buf[O2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, len | 0x80);
+				s = o2s(val, alt_form, buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} case 'u': {
+				uintmax_t val JEMALLOC_CC_SILENCE_INIT(0);
+				char buf[U2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, len | 0x80);
+				s = u2s(val, 10, false, buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} case 'x': case 'X': {
+				uintmax_t val JEMALLOC_CC_SILENCE_INIT(0);
+				char buf[X2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, len | 0x80);
+				s = x2s(val, alt_form, *f == 'X', buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} case 'c': {
+				unsigned char val;
+				char buf[2];
+
+				assert(len == '?' || len == 'l');
+				assert_not_implemented(len != 'l');
+				val = va_arg(ap, int);
+				buf[0] = val;
+				buf[1] = '\0';
+				APPEND_PADDED_S(buf, 1, width, left_justify);
+				f++;
+				break;
+			} case 's':
+				assert(len == '?' || len == 'l');
+				assert_not_implemented(len != 'l');
+				s = va_arg(ap, char *);
+				slen = (prec == -1) ? strlen(s) : prec;
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			case 'p': {
+				uintmax_t val;
+				char buf[X2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, 'p');
+				s = x2s(val, true, false, buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			}
+			default: not_implemented();
+			}
+			break;
+		} default: {
+			APPEND_C(*f);
+			f++;
+			break;
+		}}
+	}
+	label_out:
+	if (i < size)
+		str[i] = '\0';
+	else
+		str[size - 1] = '\0';
+	ret = i;
+
+#undef APPEND_C
+#undef APPEND_S
+#undef APPEND_PADDED_S
+#undef GET_ARG_NUMERIC
+	return (ret);
+}
+
+JEMALLOC_ATTR(format(printf, 3, 4))
+int
+malloc_snprintf(char *str, size_t size, const char *format, ...)
+{
+	int ret;
+	va_list ap;
+
+	va_start(ap, format);
+	ret = malloc_vsnprintf(str, size, format, ap);
+	va_end(ap);
+
+	return (ret);
+}
+
+void
+malloc_vcprintf(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *format, va_list ap)
+{
+	char buf[MALLOC_PRINTF_BUFSIZE];
+
+	if (write_cb == NULL) {
+		/*
+		 * The caller did not provide an alternate write_cb callback
+		 * function, so use the default one.  malloc_write() is an
+		 * inline function, so use malloc_message() directly here.
+		 */
+		write_cb = (je_malloc_message != NULL) ? je_malloc_message :
+		    wrtmessage;
+		cbopaque = NULL;
+	}
+
+	malloc_vsnprintf(buf, sizeof(buf), format, ap);
+	write_cb(cbopaque, buf);
+}
+
+/*
+ * Print to a callback function in such a way as to (hopefully) avoid memory
+ * allocation.
+ */
+JEMALLOC_ATTR(format(printf, 3, 4))
+void
+malloc_cprintf(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *format, ...)
+{
+	va_list ap;
+
+	va_start(ap, format);
+	malloc_vcprintf(write_cb, cbopaque, format, ap);
+	va_end(ap);
+}
+
+/* Print to stderr in such a way as to avoid memory allocation. */
+JEMALLOC_ATTR(format(printf, 1, 2))
+void
+malloc_printf(const char *format, ...)
+{
+	va_list ap;
+
+	va_start(ap, format);
+	malloc_vcprintf(NULL, NULL, format, ap);
+	va_end(ap);
+}
diff --git a/extra/jemalloc/src/zone.c b/extra/jemalloc/src/zone.c
new file mode 100644
index 00000000000..c62c183f65e
--- /dev/null
+++ b/extra/jemalloc/src/zone.c
@@ -0,0 +1,258 @@
+#include "jemalloc/internal/jemalloc_internal.h"
+#ifndef JEMALLOC_ZONE
+#  error "This source file is for zones on Darwin (OS X)."
+#endif
+
+/*
+ * The malloc_default_purgeable_zone function is only available on >= 10.6.
+ * We need to check whether it is present at runtime, thus the weak_import.
+ */
+extern malloc_zone_t *malloc_default_purgeable_zone(void)
+JEMALLOC_ATTR(weak_import);
+
+/******************************************************************************/
+/* Data. */
+
+static malloc_zone_t zone;
+static struct malloc_introspection_t zone_introspect;
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static size_t	zone_size(malloc_zone_t *zone, void *ptr);
+static void	*zone_malloc(malloc_zone_t *zone, size_t size);
+static void	*zone_calloc(malloc_zone_t *zone, size_t num, size_t size);
+static void	*zone_valloc(malloc_zone_t *zone, size_t size);
+static void	zone_free(malloc_zone_t *zone, void *ptr);
+static void	*zone_realloc(malloc_zone_t *zone, void *ptr, size_t size);
+#if (JEMALLOC_ZONE_VERSION >= 5)
+static void	*zone_memalign(malloc_zone_t *zone, size_t alignment,
+#endif
+#if (JEMALLOC_ZONE_VERSION >= 6)
+    size_t size);
+static void	zone_free_definite_size(malloc_zone_t *zone, void *ptr,
+    size_t size);
+#endif
+static void	*zone_destroy(malloc_zone_t *zone);
+static size_t	zone_good_size(malloc_zone_t *zone, size_t size);
+static void	zone_force_lock(malloc_zone_t *zone);
+static void	zone_force_unlock(malloc_zone_t *zone);
+
+/******************************************************************************/
+/*
+ * Functions.
+ */
+
+static size_t
+zone_size(malloc_zone_t *zone, void *ptr)
+{
+
+	/*
+	 * There appear to be places within Darwin (such as setenv(3)) that
+	 * cause calls to this function with pointers that *no* zone owns.  If
+	 * we knew that all pointers were owned by *some* zone, we could split
+	 * our zone into two parts, and use one as the default allocator and
+	 * the other as the default deallocator/reallocator.  Since that will
+	 * not work in practice, we must check all pointers to assure that they
+	 * reside within a mapped chunk before determining size.
+	 */
+	return (ivsalloc(ptr, config_prof));
+}
+
+static void *
+zone_malloc(malloc_zone_t *zone, size_t size)
+{
+
+	return (je_malloc(size));
+}
+
+static void *
+zone_calloc(malloc_zone_t *zone, size_t num, size_t size)
+{
+
+	return (je_calloc(num, size));
+}
+
+static void *
+zone_valloc(malloc_zone_t *zone, size_t size)
+{
+	void *ret = NULL; /* Assignment avoids useless compiler warning. */
+
+	je_posix_memalign(&ret, PAGE, size);
+
+	return (ret);
+}
+
+static void
+zone_free(malloc_zone_t *zone, void *ptr)
+{
+
+	if (ivsalloc(ptr, config_prof) != 0) {
+		je_free(ptr);
+		return;
+	}
+
+	free(ptr);
+}
+
+static void *
+zone_realloc(malloc_zone_t *zone, void *ptr, size_t size)
+{
+
+	if (ivsalloc(ptr, config_prof) != 0)
+		return (je_realloc(ptr, size));
+
+	return (realloc(ptr, size));
+}
+
+#if (JEMALLOC_ZONE_VERSION >= 5)
+static void *
+zone_memalign(malloc_zone_t *zone, size_t alignment, size_t size)
+{
+	void *ret = NULL; /* Assignment avoids useless compiler warning. */
+
+	je_posix_memalign(&ret, alignment, size);
+
+	return (ret);
+}
+#endif
+
+#if (JEMALLOC_ZONE_VERSION >= 6)
+static void
+zone_free_definite_size(malloc_zone_t *zone, void *ptr, size_t size)
+{
+
+	if (ivsalloc(ptr, config_prof) != 0) {
+		assert(ivsalloc(ptr, config_prof) == size);
+		je_free(ptr);
+		return;
+	}
+
+	free(ptr);
+}
+#endif
+
+static void *
+zone_destroy(malloc_zone_t *zone)
+{
+
+	/* This function should never be called. */
+	assert(false);
+	return (NULL);
+}
+
+static size_t
+zone_good_size(malloc_zone_t *zone, size_t size)
+{
+
+	if (size == 0)
+		size = 1;
+	return (s2u(size));
+}
+
+static void
+zone_force_lock(malloc_zone_t *zone)
+{
+
+	if (isthreaded)
+		jemalloc_prefork();
+}
+
+static void
+zone_force_unlock(malloc_zone_t *zone)
+{
+
+	if (isthreaded)
+		jemalloc_postfork_parent();
+}
+
+JEMALLOC_ATTR(constructor)
+void
+register_zone(void)
+{
+
+	/*
+	 * If something else replaced the system default zone allocator, don't
+	 * register jemalloc's.
+	 */
+	malloc_zone_t *default_zone = malloc_default_zone();
+	if (!default_zone->zone_name ||
+	    strcmp(default_zone->zone_name, "DefaultMallocZone") != 0) {
+		return;
+	}
+
+	zone.size = (void *)zone_size;
+	zone.malloc = (void *)zone_malloc;
+	zone.calloc = (void *)zone_calloc;
+	zone.valloc = (void *)zone_valloc;
+	zone.free = (void *)zone_free;
+	zone.realloc = (void *)zone_realloc;
+	zone.destroy = (void *)zone_destroy;
+	zone.zone_name = "jemalloc_zone";
+	zone.batch_malloc = NULL;
+	zone.batch_free = NULL;
+	zone.introspect = &zone_introspect;
+	zone.version = JEMALLOC_ZONE_VERSION;
+#if (JEMALLOC_ZONE_VERSION >= 5)
+	zone.memalign = zone_memalign;
+#endif
+#if (JEMALLOC_ZONE_VERSION >= 6)
+	zone.free_definite_size = zone_free_definite_size;
+#endif
+#if (JEMALLOC_ZONE_VERSION >= 8)
+	zone.pressure_relief = NULL;
+#endif
+
+	zone_introspect.enumerator = NULL;
+	zone_introspect.good_size = (void *)zone_good_size;
+	zone_introspect.check = NULL;
+	zone_introspect.print = NULL;
+	zone_introspect.log = NULL;
+	zone_introspect.force_lock = (void *)zone_force_lock;
+	zone_introspect.force_unlock = (void *)zone_force_unlock;
+	zone_introspect.statistics = NULL;
+#if (JEMALLOC_ZONE_VERSION >= 6)
+	zone_introspect.zone_locked = NULL;
+#endif
+#if (JEMALLOC_ZONE_VERSION >= 7)
+	zone_introspect.enable_discharge_checking = NULL;
+	zone_introspect.disable_discharge_checking = NULL;
+	zone_introspect.discharge = NULL;
+#ifdef __BLOCKS__
+	zone_introspect.enumerate_discharged_pointers = NULL;
+#else
+	zone_introspect.enumerate_unavailable_without_blocks = NULL;
+#endif
+#endif
+
+	/*
+	 * The default purgeable zone is created lazily by OSX's libc.  It uses
+	 * the default zone when it is created for "small" allocations
+	 * (< 15 KiB), but assumes the default zone is a scalable_zone.  This
+	 * obviously fails when the default zone is the jemalloc zone, so
+	 * malloc_default_purgeable_zone is called beforehand so that the
+	 * default purgeable zone is created when the default zone is still
+	 * a scalable_zone.  As purgeable zones only exist on >= 10.6, we need
+	 * to check for the existence of malloc_default_purgeable_zone() at
+	 * run time.
+	 */
+	if (malloc_default_purgeable_zone != NULL)
+		malloc_default_purgeable_zone();
+
+	/* Register the custom zone.  At this point it won't be the default. */
+	malloc_zone_register(&zone);
+
+	/*
+	 * Unregister and reregister the default zone.  On OSX >= 10.6,
+	 * unregistering takes the last registered zone and places it at the
+	 * location of the specified zone.  Unregistering the default zone thus
+	 * makes the last registered one the default.  On OSX < 10.6,
+	 * unregistering shifts all registered zones.  The first registered zone
+	 * then becomes the default.
+	 */
+	do {
+		default_zone = malloc_default_zone();
+		malloc_zone_unregister(default_zone);
+		malloc_zone_register(default_zone);
+	} while (malloc_default_zone() != &zone);
+}
diff --git a/extra/jemalloc/test/ALLOCM_ARENA.c b/extra/jemalloc/test/ALLOCM_ARENA.c
new file mode 100644
index 00000000000..2c52485e890
--- /dev/null
+++ b/extra/jemalloc/test/ALLOCM_ARENA.c
@@ -0,0 +1,67 @@
+#define	JEMALLOC_MANGLE
+#include "jemalloc_test.h"
+
+#define	NTHREADS 10
+
+void *
+je_thread_start(void *arg)
+{
+	unsigned thread_ind = (unsigned)(uintptr_t)arg;
+	unsigned arena_ind;
+	int r;
+	void *p;
+	size_t rsz, sz;
+
+	sz = sizeof(arena_ind);
+	if (mallctl("arenas.extend", &arena_ind, &sz, NULL, 0)
+	    != 0) {
+		malloc_printf("Error in arenas.extend\n");
+		abort();
+	}
+
+	if (thread_ind % 4 != 3) {
+		size_t mib[3];
+		size_t miblen = sizeof(mib) / sizeof(size_t);
+		const char *dss_precs[] = {"disabled", "primary", "secondary"};
+		const char *dss = dss_precs[thread_ind % 4];
+		if (mallctlnametomib("arena.0.dss", mib, &miblen) != 0) {
+			malloc_printf("Error in mallctlnametomib()\n");
+			abort();
+		}
+		mib[1] = arena_ind;
+		if (mallctlbymib(mib, miblen, NULL, NULL, (void *)&dss,
+		    sizeof(const char *))) {
+			malloc_printf("Error in mallctlbymib()\n");
+			abort();
+		}
+	}
+
+	r = allocm(&p, &rsz, 1, ALLOCM_ARENA(arena_ind));
+	if (r != ALLOCM_SUCCESS) {
+		malloc_printf("Unexpected allocm() error\n");
+		abort();
+	}
+	dallocm(p, 0);
+
+	return (NULL);
+}
+
+int
+main(void)
+{
+	je_thread_t threads[NTHREADS];
+	unsigned i;
+
+	malloc_printf("Test begin\n");
+
+	for (i = 0; i < NTHREADS; i++) {
+		je_thread_create(&threads[i], je_thread_start,
+		    (void *)(uintptr_t)i);
+	}
+
+	for (i = 0; i < NTHREADS; i++)
+		je_thread_join(threads[i], NULL);
+
+	malloc_printf("Test end\n");
+	return (0);
+}
diff --git a/extra/jemalloc/test/ALLOCM_ARENA.exp b/extra/jemalloc/test/ALLOCM_ARENA.exp
new file mode 100644
index 00000000000..369a88dd240
--- /dev/null
+++ b/extra/jemalloc/test/ALLOCM_ARENA.exp
@@ -0,0 +1,2 @@
+Test begin
+Test end
diff --git a/extra/jemalloc/test/aligned_alloc.c b/extra/jemalloc/test/aligned_alloc.c
new file mode 100644
index 00000000000..5a9b0caea78
--- /dev/null
+++ b/extra/jemalloc/test/aligned_alloc.c
@@ -0,0 +1,119 @@
+#define	JEMALLOC_MANGLE
+#include "jemalloc_test.h"
+
+#define CHUNK 0x400000
+/* #define MAXALIGN ((size_t)UINT64_C(0x80000000000)) */
+#define MAXALIGN ((size_t)0x2000000LU)
+#define NITER 4
+
+int
+main(void)
+{
+	size_t alignment, size, total;
+	unsigned i;
+	void *p, *ps[NITER];
+
+	malloc_printf("Test begin\n");
+
+	/* Test error conditions. */
+	alignment = 0;
+	set_errno(0);
+	p = aligned_alloc(alignment, 1);
+	if (p != NULL || get_errno() != EINVAL) {
+		malloc_printf(
+		    "Expected error for invalid alignment %zu\n", alignment);
+	}
+
+	for (alignment = sizeof(size_t); alignment < MAXALIGN;
+	    alignment <<= 1) {
+		set_errno(0);
+		p = aligned_alloc(alignment + 1, 1);
+		if (p != NULL || get_errno() != EINVAL) {
+			malloc_printf(
+			    "Expected error for invalid alignment %zu\n",
+			    alignment + 1);
+		}
+	}
+
+#if LG_SIZEOF_PTR == 3
+	alignment = UINT64_C(0x8000000000000000);
+	size      = UINT64_C(0x8000000000000000);
+#else
+	alignment = 0x80000000LU;
+	size      = 0x80000000LU;
+#endif
+	set_errno(0);
+	p = aligned_alloc(alignment, size);
+	if (p != NULL || get_errno() != ENOMEM) {
+		malloc_printf(
+		    "Expected error for aligned_alloc(%zu, %zu)\n",
+		    alignment, size);
+	}
+
+#if LG_SIZEOF_PTR == 3
+	alignment = UINT64_C(0x4000000000000000);
+	size      = UINT64_C(0x8400000000000001);
+#else
+	alignment = 0x40000000LU;
+	size      = 0x84000001LU;
+#endif
+	set_errno(0);
+	p = aligned_alloc(alignment, size);
+	if (p != NULL || get_errno() != ENOMEM) {
+		malloc_printf(
+		    "Expected error for aligned_alloc(%zu, %zu)\n",
+		    alignment, size);
+	}
+
+	alignment = 0x10LU;
+#if LG_SIZEOF_PTR == 3
+	size = UINT64_C(0xfffffffffffffff0);
+#else
+	size = 0xfffffff0LU;
+#endif
+	set_errno(0);
+	p = aligned_alloc(alignment, size);
+	if (p != NULL || get_errno() != ENOMEM) {
+		malloc_printf(
+		    "Expected error for aligned_alloc(&p, %zu, %zu)\n",
+		    alignment, size);
+	}
+
+	for (i = 0; i < NITER; i++)
+		ps[i] = NULL;
+
+	for (alignment = 8;
+	    alignment <= MAXALIGN;
+	    alignment <<= 1) {
+		total = 0;
+		malloc_printf("Alignment: %zu\n", alignment);
+		for (size = 1;
+		    size < 3 * alignment && size < (1U << 31);
+		    size += (alignment >> (LG_SIZEOF_PTR-1)) - 1) {
+			for (i = 0; i < NITER; i++) {
+				ps[i] = aligned_alloc(alignment, size);
+				if (ps[i] == NULL) {
+					char buf[BUFERROR_BUF];
+
+					buferror(buf, sizeof(buf));
+					malloc_printf(
+					    "Error for size %zu (%#zx): %s\n",
+					    size, size, buf);
+					exit(1);
+				}
+				total += malloc_usable_size(ps[i]);
+				if (total >= (MAXALIGN << 1))
+					break;
+			}
+			for (i = 0; i < NITER; i++) {
+				if (ps[i] != NULL) {
+					free(ps[i]);
+					ps[i] = NULL;
+				}
+			}
+		}
+	}
+
+	malloc_printf("Test end\n");
+	return (0);
+}
diff --git a/extra/jemalloc/test/aligned_alloc.exp b/extra/jemalloc/test/aligned_alloc.exp
new file mode 100644
index 00000000000..b5061c7277e
--- /dev/null
+++ b/extra/jemalloc/test/aligned_alloc.exp
@@ -0,0 +1,25 @@
+Test begin
+Alignment: 8
+Alignment: 16
+Alignment: 32
+Alignment: 64
+Alignment: 128
+Alignment: 256
+Alignment: 512
+Alignment: 1024
+Alignment: 2048
+Alignment: 4096
+Alignment: 8192
+Alignment: 16384
+Alignment: 32768
+Alignment: 65536
+Alignment: 131072
+Alignment: 262144
+Alignment: 524288
+Alignment: 1048576
+Alignment: 2097152
+Alignment: 4194304
+Alignment: 8388608
+Alignment: 16777216
+Alignment: 33554432
+Test end
diff --git a/extra/jemalloc/test/allocated.c b/extra/jemalloc/test/allocated.c
new file mode 100644
index 00000000000..9884905d810
--- /dev/null
+++ b/extra/jemalloc/test/allocated.c
@@ -0,0 +1,118 @@
+#define	JEMALLOC_MANGLE
+#include "jemalloc_test.h"
+
+void *
+je_thread_start(void *arg)
+{
+	int err;
+	void *p;
+	uint64_t a0, a1, d0, d1;
+	uint64_t *ap0, *ap1, *dp0, *dp1;
+	size_t sz, usize;
+
+	sz = sizeof(a0);
+	if ((err = mallctl("thread.allocated", &a0, &sz, NULL, 0))) {
+		if (err == ENOENT) {
+#ifdef JEMALLOC_STATS
+			assert(false);
+#endif
+			goto label_return;
+		}
+		malloc_printf("%s(): Error in mallctl(): %s\n", __func__,
+		    strerror(err));
+		exit(1);
+	}
+	sz = sizeof(ap0);
+	if ((err = mallctl("thread.allocatedp", &ap0, &sz, NULL, 0))) {
+		if (err == ENOENT) {
+#ifdef JEMALLOC_STATS
+			assert(false);
+#endif
+			goto label_return;
+		}
+		malloc_printf("%s(): Error in mallctl(): %s\n", __func__,
+		    strerror(err));
+		exit(1);
+	}
+	assert(*ap0 == a0);
+
+	sz = sizeof(d0);
+	if ((err = mallctl("thread.deallocated", &d0, &sz, NULL, 0))) {
+		if (err == ENOENT) {
+#ifdef JEMALLOC_STATS
+			assert(false);
+#endif
+			goto label_return;
+		}
+		malloc_printf("%s(): Error in mallctl(): %s\n", __func__,
+		    strerror(err));
+		exit(1);
+	}
+	sz = sizeof(dp0);
+	if ((err = mallctl("thread.deallocatedp", &dp0, &sz, NULL, 0))) {
+		if (err == ENOENT) {
+#ifdef JEMALLOC_STATS
+			assert(false);
+#endif
+			goto label_return;
+		}
+		malloc_printf("%s(): Error in mallctl(): %s\n", __func__,
+		    strerror(err));
+		exit(1);
+	}
+	assert(*dp0 == d0);
+
+	p = malloc(1);
+	if (p == NULL) {
+		malloc_printf("%s(): Error in malloc()\n", __func__);
+		exit(1);
+	}
+
+	sz = sizeof(a1);
+	mallctl("thread.allocated", &a1, &sz, NULL, 0);
+	sz = sizeof(ap1);
+	mallctl("thread.allocatedp", &ap1, &sz, NULL, 0);
+	assert(*ap1 == a1);
+	assert(ap0 == ap1);
+
+	usize = malloc_usable_size(p);
+	assert(a0 + usize <= a1);
+
+	free(p);
+
+	sz = sizeof(d1);
+	mallctl("thread.deallocated", &d1, &sz, NULL, 0);
+	sz = sizeof(dp1);
+	mallctl("thread.deallocatedp", &dp1, &sz, NULL, 0);
+	assert(*dp1 == d1);
+	assert(dp0 == dp1);
+
+	assert(d0 + usize <= d1);
+
+label_return:
+	return (NULL);
+}
+
+int
+main(void)
+{
+	int ret = 0;
+	je_thread_t thread;
+
+	malloc_printf("Test begin\n");
+
+	je_thread_start(NULL);
+
+	je_thread_create(&thread, je_thread_start, NULL);
+	je_thread_join(thread, (void *)&ret);
+
+	je_thread_start(NULL);
+
+	je_thread_create(&thread, je_thread_start, NULL);
+	je_thread_join(thread, (void *)&ret);
+
+	je_thread_start(NULL);
+
+	malloc_printf("Test end\n");
+	return (ret);
+}
diff --git a/extra/jemalloc/test/allocated.exp b/extra/jemalloc/test/allocated.exp
new file mode 100644
index 00000000000..369a88dd240
--- /dev/null
+++ b/extra/jemalloc/test/allocated.exp
@@ -0,0 +1,2 @@
+Test begin
+Test end
diff --git a/extra/jemalloc/test/allocm.c b/extra/jemalloc/test/allocm.c
new file mode 100644
index 00000000000..80be673b8fd
--- /dev/null
+++ b/extra/jemalloc/test/allocm.c
@@ -0,0 +1,194 @@
+#define	JEMALLOC_MANGLE
+#include "jemalloc_test.h"
+
+#define CHUNK 0x400000
+/* #define MAXALIGN ((size_t)UINT64_C(0x80000000000)) */
+#define MAXALIGN ((size_t)0x2000000LU)
+#define NITER 4
+
+int
+main(void)
+{
+	int r;
+	void *p;
+	size_t nsz, rsz, sz, alignment, total;
+	unsigned i;
+	void *ps[NITER];
+
+	malloc_printf("Test begin\n");
+
+	sz = 42;
+	nsz = 0;
+	r = nallocm(&nsz, sz, 0);
+	if (r != ALLOCM_SUCCESS) {
+		malloc_printf("Unexpected nallocm() error\n");
+		abort();
+	}
+	rsz = 0;
+	r = allocm(&p, &rsz, sz, 0);
+	if (r != ALLOCM_SUCCESS) {
+		malloc_printf("Unexpected allocm() error\n");
+		abort();
+	}
+	if (rsz < sz)
+		malloc_printf("Real size smaller than expected\n");
+	if (nsz != rsz)
+		malloc_printf("nallocm()/allocm() rsize mismatch\n");
+	if (dallocm(p, 0) != ALLOCM_SUCCESS)
+		malloc_printf("Unexpected dallocm() error\n");
+
+	r = allocm(&p, NULL, sz, 0);
+	if (r != ALLOCM_SUCCESS) {
+		malloc_printf("Unexpected allocm() error\n");
+		abort();
+	}
+	if (dallocm(p, 0) != ALLOCM_SUCCESS)
+		malloc_printf("Unexpected dallocm() error\n");
+
+	nsz = 0;
+	r = nallocm(&nsz, sz, ALLOCM_ZERO);
+	if (r != ALLOCM_SUCCESS) {
+		malloc_printf("Unexpected nallocm() error\n");
+		abort();
+	}
+	rsz = 0;
+	r = allocm(&p, &rsz, sz, ALLOCM_ZERO);
+	if (r != ALLOCM_SUCCESS) {
+		malloc_printf("Unexpected allocm() error\n");
+		abort();
+	}
+	if (nsz != rsz)
+		malloc_printf("nallocm()/allocm() rsize mismatch\n");
+	if (dallocm(p, 0) != ALLOCM_SUCCESS)
+		malloc_printf("Unexpected dallocm() error\n");
+
+#if LG_SIZEOF_PTR == 3
+	alignment = UINT64_C(0x8000000000000000);
+	sz        = UINT64_C(0x8000000000000000);
+#else
+	alignment = 0x80000000LU;
+	sz        = 0x80000000LU;
+#endif
+	nsz = 0;
+	r = nallocm(&nsz, sz, ALLOCM_ALIGN(alignment));
+	if (r == ALLOCM_SUCCESS) {
+		malloc_printf(
+		    "Expected error for nallocm(&nsz, %zu, %#x)\n",
+		    sz, ALLOCM_ALIGN(alignment));
+	}
+	rsz = 0;
+	r = allocm(&p, &rsz, sz, ALLOCM_ALIGN(alignment));
+	if (r == ALLOCM_SUCCESS) {
+		malloc_printf(
+		    "Expected error for allocm(&p, %zu, %#x)\n",
+		    sz, ALLOCM_ALIGN(alignment));
+	}
+	if (nsz != rsz)
+		malloc_printf("nallocm()/allocm() rsize mismatch\n");
+
+#if LG_SIZEOF_PTR == 3
+	alignment = UINT64_C(0x4000000000000000);
+	sz        = UINT64_C(0x8400000000000001);
+#else
+	alignment = 0x40000000LU;
+	sz        = 0x84000001LU;
+#endif
+	nsz = 0;
+	r = nallocm(&nsz, sz, ALLOCM_ALIGN(alignment));
+	if (r != ALLOCM_SUCCESS)
+		malloc_printf("Unexpected nallocm() error\n");
+	rsz = 0;
+	r = allocm(&p, &rsz, sz, ALLOCM_ALIGN(alignment));
+	if (r == ALLOCM_SUCCESS) {
+		malloc_printf(
+		    "Expected error for allocm(&p, %zu, %#x)\n",
+		    sz, ALLOCM_ALIGN(alignment));
+	}
+
+	alignment = 0x10LU;
+#if LG_SIZEOF_PTR == 3
+	sz = UINT64_C(0xfffffffffffffff0);
+#else
+	sz = 0xfffffff0LU;
+#endif
+	nsz = 0;
+	r = nallocm(&nsz, sz, ALLOCM_ALIGN(alignment));
+	if (r == ALLOCM_SUCCESS) {
+		malloc_printf(
+		    "Expected error for nallocm(&nsz, %zu, %#x)\n",
+		    sz, ALLOCM_ALIGN(alignment));
+	}
+	rsz = 0;
+	r = allocm(&p, &rsz, sz, ALLOCM_ALIGN(alignment));
+	if (r == ALLOCM_SUCCESS) {
+		malloc_printf(
+		    "Expected error for allocm(&p, %zu, %#x)\n",
+		    sz, ALLOCM_ALIGN(alignment));
+	}
+	if (nsz != rsz)
+		malloc_printf("nallocm()/allocm() rsize mismatch\n");
+
+	for (i = 0; i < NITER; i++)
+		ps[i] = NULL;
+
+	for (alignment = 8;
+	    alignment <= MAXALIGN;
+	    alignment <<= 1) {
+		total = 0;
+		malloc_printf("Alignment: %zu\n", alignment);
+		for (sz = 1;
+		    sz < 3 * alignment && sz < (1U << 31);
+		    sz += (alignment >> (LG_SIZEOF_PTR-1)) - 1) {
+			for (i = 0; i < NITER; i++) {
+				nsz = 0;
+				r = nallocm(&nsz, sz,
+				    ALLOCM_ALIGN(alignment) | ALLOCM_ZERO);
+				if (r != ALLOCM_SUCCESS) {
+					malloc_printf(
+					    "nallocm() error for size %zu"
+					    " (%#zx): %d\n",
+					    sz, sz, r);
+					exit(1);
+				}
+				rsz = 0;
+				r = allocm(&ps[i], &rsz, sz,
+				    ALLOCM_ALIGN(alignment) | ALLOCM_ZERO);
+				if (r != ALLOCM_SUCCESS) {
+					malloc_printf(
+					    "allocm() error for size %zu"
+					    " (%#zx): %d\n",
+					    sz, sz, r);
+					exit(1);
+				}
+				if (rsz < sz) {
+					malloc_printf(
+					    "Real size smaller than"
+					    " expected\n");
+				}
+				if (nsz != rsz) {
+					malloc_printf(
+					    "nallocm()/allocm() rsize"
+					    " mismatch\n");
+				}
+				if ((uintptr_t)p & (alignment-1)) {
+					malloc_printf(
+					    "%p inadequately aligned for"
+					    " alignment: %zu\n", p, alignment);
+				}
+				sallocm(ps[i], &rsz, 0);
+				total += rsz;
+				if (total >= (MAXALIGN << 1))
+					break;
+			}
+			for (i = 0; i < NITER; i++) {
+				if (ps[i] != NULL) {
+					dallocm(ps[i], 0);
+					ps[i] = NULL;
+				}
+			}
+		}
+	}
+
+	malloc_printf("Test end\n");
+	return (0);
+}
diff --git a/extra/jemalloc/test/allocm.exp b/extra/jemalloc/test/allocm.exp
new file mode 100644
index 00000000000..b5061c7277e
--- /dev/null
+++ b/extra/jemalloc/test/allocm.exp
@@ -0,0 +1,25 @@
+Test begin
+Alignment: 8
+Alignment: 16
+Alignment: 32
+Alignment: 64
+Alignment: 128
+Alignment: 256
+Alignment: 512
+Alignment: 1024
+Alignment: 2048
+Alignment: 4096
+Alignment: 8192
+Alignment: 16384
+Alignment: 32768
+Alignment: 65536
+Alignment: 131072
+Alignment: 262144
+Alignment: 524288
+Alignment: 1048576
+Alignment: 2097152
+Alignment: 4194304
+Alignment: 8388608
+Alignment: 16777216
+Alignment: 33554432
+Test end
diff --git a/extra/jemalloc/test/bitmap.c b/extra/jemalloc/test/bitmap.c
new file mode 100644
index 00000000000..b2cb63004bc
--- /dev/null
+++ b/extra/jemalloc/test/bitmap.c
@@ -0,0 +1,153 @@
+#define	JEMALLOC_MANGLE
+#include "jemalloc_test.h"
+
+#if (LG_BITMAP_MAXBITS > 12)
+#  define MAXBITS	4500
+#else
+#  define MAXBITS	(1U << LG_BITMAP_MAXBITS)
+#endif
+
+static void
+test_bitmap_size(void)
+{
+	size_t i, prev_size;
+
+	prev_size = 0;
+	for (i = 1; i <= MAXBITS; i++) {
+		size_t size = bitmap_size(i);
+		assert(size >= prev_size);
+		prev_size = size;
+	}
+}
+
+static void
+test_bitmap_init(void)
+{
+	size_t i;
+
+	for (i = 1; i <= MAXBITS; i++) {
+		bitmap_info_t binfo;
+		bitmap_info_init(&binfo, i);
+		{
+			size_t j;
+			bitmap_t *bitmap = malloc(sizeof(bitmap_t) *
+				bitmap_info_ngroups(&binfo));
+			bitmap_init(bitmap, &binfo);
+
+			for (j = 0; j < i; j++)
+				assert(bitmap_get(bitmap, &binfo, j) == false);
+			free(bitmap);
+
+		}
+	}
+}
+
+static void
+test_bitmap_set(void)
+{
+	size_t i;
+
+	for (i = 1; i <= MAXBITS; i++) {
+		bitmap_info_t binfo;
+		bitmap_info_init(&binfo, i);
+		{
+			size_t j;
+			bitmap_t *bitmap = malloc(sizeof(bitmap_t) *
+				bitmap_info_ngroups(&binfo));
+			bitmap_init(bitmap, &binfo);
+
+			for (j = 0; j < i; j++)
+				bitmap_set(bitmap, &binfo, j);
+			assert(bitmap_full(bitmap, &binfo));
+			free(bitmap);
+		}
+	}
+}
+
+static void
+test_bitmap_unset(void)
+{
+	size_t i;
+
+	for (i = 1; i <= MAXBITS; i++) {
+		bitmap_info_t binfo;
+		bitmap_info_init(&binfo, i);
+		{
+			size_t j;
+			bitmap_t *bitmap = malloc(sizeof(bitmap_t) *
+				bitmap_info_ngroups(&binfo));
+			bitmap_init(bitmap, &binfo);
+
+			for (j = 0; j < i; j++)
+				bitmap_set(bitmap, &binfo, j);
+			assert(bitmap_full(bitmap, &binfo));
+			for (j = 0; j < i; j++)
+				bitmap_unset(bitmap, &binfo, j);
+			for (j = 0; j < i; j++)
+				bitmap_set(bitmap, &binfo, j);
+			assert(bitmap_full(bitmap, &binfo));
+			free(bitmap);
+		}
+	}
+}
+
+static void
+test_bitmap_sfu(void)
+{
+	size_t i;
+
+	for (i = 1; i <= MAXBITS; i++) {
+		bitmap_info_t binfo;
+		bitmap_info_init(&binfo, i);
+		{
+			ssize_t j;
+			bitmap_t *bitmap = malloc(sizeof(bitmap_t) *
+				bitmap_info_ngroups(&binfo));
+			bitmap_init(bitmap, &binfo);
+
+			/* Iteratively set bits starting at the beginning. */
+			for (j = 0; j < i; j++)
+				assert(bitmap_sfu(bitmap, &binfo) == j);
+			assert(bitmap_full(bitmap, &binfo));
+
+			/*
+			 * Iteratively unset bits starting at the end, and
+			 * verify that bitmap_sfu() reaches the unset bits.
+			 */
+			for (j = i - 1; j >= 0; j--) {
+				bitmap_unset(bitmap, &binfo, j);
+				assert(bitmap_sfu(bitmap, &binfo) == j);
+				bitmap_unset(bitmap, &binfo, j);
+			}
+			assert(bitmap_get(bitmap, &binfo, 0) == false);
+
+			/*
+			 * Iteratively set bits starting at the beginning, and
+			 * verify that bitmap_sfu() looks past them.
+			 */
+			for (j = 1; j < i; j++) {
+				bitmap_set(bitmap, &binfo, j - 1);
+				assert(bitmap_sfu(bitmap, &binfo) == j);
+				bitmap_unset(bitmap, &binfo, j);
+			}
+			assert(bitmap_sfu(bitmap, &binfo) == i - 1);
+			assert(bitmap_full(bitmap, &binfo));
+			free(bitmap);
+		}
+	}
+}
+
+int
+main(void)
+{
+	malloc_printf("Test begin\n");
+
+	test_bitmap_size();
+	test_bitmap_init();
+	test_bitmap_set();
+	test_bitmap_unset();
+	test_bitmap_sfu();
+
+	malloc_printf("Test end\n");
+	return (0);
+}
diff --git a/extra/jemalloc/test/bitmap.exp b/extra/jemalloc/test/bitmap.exp
new file mode 100644
index 00000000000..369a88dd240
--- /dev/null
+++ b/extra/jemalloc/test/bitmap.exp
@@ -0,0 +1,2 @@
+Test begin
+Test end
diff --git a/extra/jemalloc/test/jemalloc_test.h.in b/extra/jemalloc/test/jemalloc_test.h.in
new file mode 100644
index 00000000000..e38b48efa41
--- /dev/null
+++ b/extra/jemalloc/test/jemalloc_test.h.in
@@ -0,0 +1,53 @@
+/*
+ * This header should be included by tests, rather than directly including
+ * jemalloc/jemalloc.h, because --with-install-suffix may cause the header to
+ * have a different name.
+ */
+#include "jemalloc/jemalloc@install_suffix@.h"
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/* Abstraction layer for threading in tests */
+#ifdef _WIN32
+#include <windows.h>
+
+typedef HANDLE je_thread_t;
+
+void
+je_thread_create(je_thread_t *thread, void *(*proc)(void *), void *arg)
+{
+	LPTHREAD_START_ROUTINE routine = (LPTHREAD_START_ROUTINE)proc;
+	*thread = CreateThread(NULL, 0, routine, arg, 0, NULL);
+	if (*thread == NULL) {
+		malloc_printf("Error in CreateThread()\n");
+		exit(1);
+	}
+}
+
+void
+je_thread_join(je_thread_t thread, void **ret)
+{
+	WaitForSingleObject(thread, INFINITE);
+}
+
+#else
+#include <pthread.h>
+
+typedef pthread_t je_thread_t;
+
+void
+je_thread_create(je_thread_t *thread, void *(*proc)(void *), void *arg)
+{
+
+	if (pthread_create(thread, NULL, proc, arg) != 0) {
+		malloc_printf("Error in pthread_create()\n");
+		exit(1);
+	}
+}
+
+void
+je_thread_join(je_thread_t thread, void **ret)
+{
+
+	pthread_join(thread, ret);
+}
+#endif
diff --git a/extra/jemalloc/test/mremap.c b/extra/jemalloc/test/mremap.c
new file mode 100644
index 00000000000..47efa7c415b
--- /dev/null
+++ b/extra/jemalloc/test/mremap.c
@@ -0,0 +1,60 @@
+#define	JEMALLOC_MANGLE
+#include "jemalloc_test.h"
+
+int
+main(void)
+{
+	int ret, err;
+	size_t sz, lg_chunk, chunksize, i;
+	char *p, *q;
+
+	malloc_printf("Test begin\n");
+
+	sz = sizeof(lg_chunk);
+	if ((err = mallctl("opt.lg_chunk", &lg_chunk, &sz, NULL, 0))) {
+		assert(err != ENOENT);
+		malloc_printf("%s(): Error in mallctl(): %s\n", __func__,
+		    strerror(err));
+		ret = 1;
+		goto label_return;
+	}
+	chunksize = ((size_t)1U) << lg_chunk;
+
+	p = (char *)malloc(chunksize);
+	if (p == NULL) {
+		malloc_printf("malloc(%zu) --> %p\n", chunksize, p);
+		ret = 1;
+		goto label_return;
+	}
+	memset(p, 'a', chunksize);
+
+	q = (char *)realloc(p, chunksize * 2);
+	if (q == NULL) {
+		malloc_printf("realloc(%p, %zu) --> %p\n", p, chunksize * 2,
+		    q);
+		ret = 1;
+		goto label_return;
+	}
+	for (i = 0; i < chunksize; i++) {
+		assert(q[i] == 'a');
+	}
+
+	p = q;
+
+	q = (char *)realloc(p, chunksize);
+	if (q == NULL) {
+		malloc_printf("realloc(%p, %zu) --> %p\n", p, chunksize, q);
+		ret = 1;
+		goto label_return;
+	}
+	for (i = 0; i < chunksize; i++) {
+		assert(q[i] == 'a');
+	}
+
+	free(q);
+
+	ret = 0;
+label_return:
+	malloc_printf("Test end\n");
+	return (ret);
+}
diff --git a/extra/jemalloc/test/mremap.exp b/extra/jemalloc/test/mremap.exp
new file mode 100644
index 00000000000..369a88dd240
--- /dev/null
+++ b/extra/jemalloc/test/mremap.exp
@@ -0,0 +1,2 @@
+Test begin
+Test end
diff --git a/extra/jemalloc/test/posix_memalign.c b/extra/jemalloc/test/posix_memalign.c
new file mode 100644
index 00000000000..2185bcf762a
--- /dev/null
+++ b/extra/jemalloc/test/posix_memalign.c
@@ -0,0 +1,115 @@
+#define	JEMALLOC_MANGLE
+#include "jemalloc_test.h"
+
+#define CHUNK 0x400000
+/* #define MAXALIGN ((size_t)UINT64_C(0x80000000000)) */
+#define MAXALIGN ((size_t)0x2000000LU)
+#define NITER 4
+
+int
+main(void)
+{
+	size_t alignment, size, total;
+	unsigned i;
+	int err;
+	void *p, *ps[NITER];
+
+	malloc_printf("Test begin\n");
+
+	/* Test error conditions. */
+	for (alignment = 0; alignment < sizeof(void *); alignment++) {
+		err = posix_memalign(&p, alignment, 1);
+		if (err != EINVAL) {
+			malloc_printf(
+			    "Expected error for invalid alignment %zu\n",
+			    alignment);
+		}
+	}
+
+	for (alignment = sizeof(size_t); alignment < MAXALIGN;
+	    alignment <<= 1) {
+		err = posix_memalign(&p, alignment + 1, 1);
+		if (err == 0) {
+			malloc_printf(
+			    "Expected error for invalid alignment %zu\n",
+			    alignment + 1);
+		}
+	}
+
+#if LG_SIZEOF_PTR == 3
+	alignment = UINT64_C(0x8000000000000000);
+	size      = UINT64_C(0x8000000000000000);
+#else
+	alignment = 0x80000000LU;
+	size      = 0x80000000LU;
+#endif
+	err = posix_memalign(&p, alignment, size);
+	if (err == 0) {
+		malloc_printf(
+		    "Expected error for posix_memalign(&p, %zu, %zu)\n",
+		    alignment, size);
+	}
+
+#if LG_SIZEOF_PTR == 3
+	alignment = UINT64_C(0x4000000000000000);
+	size      = UINT64_C(0x8400000000000001);
+#else
+	alignment = 0x40000000LU;
+	size      = 0x84000001LU;
+#endif
+	err = posix_memalign(&p, alignment, size);
+	if (err == 0) {
+		malloc_printf(
+		    "Expected error for posix_memalign(&p, %zu, %zu)\n",
+		    alignment, size);
+	}
+
+	alignment = 0x10LU;
+#if LG_SIZEOF_PTR == 3
+	size = UINT64_C(0xfffffffffffffff0);
+#else
+	size = 0xfffffff0LU;
+#endif
+	err = posix_memalign(&p, alignment, size);
+	if (err == 0) {
+		malloc_printf(
+		    "Expected error for posix_memalign(&p, %zu, %zu)\n",
+		    alignment, size);
+	}
+
+	for (i = 0; i < NITER; i++)
+		ps[i] = NULL;
+
+	for (alignment = 8;
+	    alignment <= MAXALIGN;
+	    alignment <<= 1) {
+		total = 0;
+		malloc_printf("Alignment: %zu\n", alignment);
+		for (size = 1;
+		    size < 3 * alignment && size < (1U << 31);
+		    size += (alignment >> (LG_SIZEOF_PTR-1)) - 1) {
+			for (i = 0; i < NITER; i++) {
+				err = posix_memalign(&ps[i],
+				    alignment, size);
+				if (err) {
+					malloc_printf(
+					    "Error for size %zu (%#zx): %s\n",
+					    size, size, strerror(err));
+					exit(1);
+				}
+				total += malloc_usable_size(ps[i]);
+				if (total >= (MAXALIGN << 1))
+					break;
+			}
+			for (i = 0; i < NITER; i++) {
+				if (ps[i] != NULL) {
+					free(ps[i]);
+					ps[i] = NULL;
+				}
+			}
+		}
+	}
+
+	malloc_printf("Test end\n");
+	return (0);
+}
diff --git a/extra/jemalloc/test/posix_memalign.exp b/extra/jemalloc/test/posix_memalign.exp
new file mode 100644
index 00000000000..b5061c7277e
--- /dev/null
+++ b/extra/jemalloc/test/posix_memalign.exp
@@ -0,0 +1,25 @@
+Test begin
+Alignment: 8
+Alignment: 16
+Alignment: 32
+Alignment: 64
+Alignment: 128
+Alignment: 256
+Alignment: 512
+Alignment: 1024
+Alignment: 2048
+Alignment: 4096
+Alignment: 8192
+Alignment: 16384
+Alignment: 32768
+Alignment: 65536
+Alignment: 131072
+Alignment: 262144
+Alignment: 524288
+Alignment: 1048576
+Alignment: 2097152
+Alignment: 4194304
+Alignment: 8388608
+Alignment: 16777216
+Alignment: 33554432
+Test end
diff --git a/extra/jemalloc/test/rallocm.c b/extra/jemalloc/test/rallocm.c
new file mode 100644
index 00000000000..c5dedf48d7b
--- /dev/null
+++ b/extra/jemalloc/test/rallocm.c
@@ -0,0 +1,127 @@
+#define	JEMALLOC_MANGLE
+#include "jemalloc_test.h"
+
+int
+main(void)
+{
+	size_t pagesize;
+	void *p, *q;
+	size_t sz, tsz;
+	int r;
+
+	malloc_printf("Test begin\n");
+
+	/* Get page size. */
+	{
+#ifdef _WIN32
+		SYSTEM_INFO si;
+		GetSystemInfo(&si);
+		pagesize = (size_t)si.dwPageSize;
+#else
+		long result = sysconf(_SC_PAGESIZE);
+		assert(result != -1);
+		pagesize = (size_t)result;
+#endif
+	}
+
+	r = allocm(&p, &sz, 42, 0);
+	if (r != ALLOCM_SUCCESS) {
+		malloc_printf("Unexpected allocm() error\n");
+		abort();
+	}
+
+	q = p;
+	r = rallocm(&q, &tsz, sz, 0, ALLOCM_NO_MOVE);
+	if (r != ALLOCM_SUCCESS)
+		malloc_printf("Unexpected rallocm() error\n");
+	if (q != p)
+		malloc_printf("Unexpected object move\n");
+	if (tsz != sz) {
+		malloc_printf("Unexpected size change: %zu --> %zu\n",
+		    sz, tsz);
+	}
+
+	q = p;
+	r = rallocm(&q, &tsz, sz, 5, ALLOCM_NO_MOVE);
+	if (r != ALLOCM_SUCCESS)
+		malloc_printf("Unexpected rallocm() error\n");
+	if (q != p)
+		malloc_printf("Unexpected object move\n");
+	if (tsz != sz) {
+		malloc_printf("Unexpected size change: %zu --> %zu\n",
+		    sz, tsz);
+	}
+
+	q = p;
+	r = rallocm(&q, &tsz, sz + 5, 0, ALLOCM_NO_MOVE);
+	if (r != ALLOCM_ERR_NOT_MOVED)
+		malloc_printf("Unexpected rallocm() result\n");
+	if (q != p)
+		malloc_printf("Unexpected object move\n");
+	if (tsz != sz) {
+		malloc_printf("Unexpected size change: %zu --> %zu\n",
+		    sz, tsz);
+	}
+
+	q = p;
+	r = rallocm(&q, &tsz, sz + 5, 0, 0);
+	if (r != ALLOCM_SUCCESS)
+		malloc_printf("Unexpected rallocm() error\n");
+	if (q == p)
+		malloc_printf("Expected object move\n");
+	if (tsz == sz) {
+		malloc_printf("Expected size change: %zu --> %zu\n",
+		    sz, tsz);
+	}
+	p = q;
+	sz = tsz;
+
+	r = rallocm(&q, &tsz, pagesize*2, 0, 0);
+	if (r != ALLOCM_SUCCESS)
+		malloc_printf("Unexpected rallocm() error\n");
+	if (q == p)
+		malloc_printf("Expected object move\n");
+	if (tsz == sz) {
+		malloc_printf("Expected size change: %zu --> %zu\n",
+		    sz, tsz);
+	}
+	p = q;
+	sz = tsz;
+
+	r = rallocm(&q, &tsz, pagesize*4, 0, 0);
+	if (r != ALLOCM_SUCCESS)
+		malloc_printf("Unexpected rallocm() error\n");
+	if (tsz == sz) {
+		malloc_printf("Expected size change: %zu --> %zu\n",
+		    sz, tsz);
+	}
+	p = q;
+	sz = tsz;
+
+	r = rallocm(&q, &tsz, pagesize*2, 0, ALLOCM_NO_MOVE);
+	if (r != ALLOCM_SUCCESS)
+		malloc_printf("Unexpected rallocm() error\n");
+	if (q != p)
+		malloc_printf("Unexpected object move\n");
+	if (tsz == sz) {
+		malloc_printf("Expected size change: %zu --> %zu\n",
+		    sz, tsz);
+	}
+	sz = tsz;
+
+	r = rallocm(&q, &tsz, pagesize*4, 0, ALLOCM_NO_MOVE);
+	if (r != ALLOCM_SUCCESS)
+		malloc_printf("Unexpected rallocm() error\n");
+	if (q != p)
+		malloc_printf("Unexpected object move\n");
+	if (tsz == sz) {
+		malloc_printf("Expected size change: %zu --> %zu\n",
+		    sz, tsz);
+	}
+	sz = tsz;
+
+	dallocm(p, 0);
+
+	malloc_printf("Test end\n");
+	return (0);
+}
diff --git a/extra/jemalloc/test/rallocm.exp b/extra/jemalloc/test/rallocm.exp
new file mode 100644
index 00000000000..369a88dd240
--- /dev/null
+++ b/extra/jemalloc/test/rallocm.exp
@@ -0,0 +1,2 @@
+Test begin
+Test end
diff --git a/extra/jemalloc/test/thread_arena.c b/extra/jemalloc/test/thread_arena.c
new file mode 100644
index 00000000000..c5a21fa0c70
--- /dev/null
+++ b/extra/jemalloc/test/thread_arena.c
@@ -0,0 +1,81 @@
+#define	JEMALLOC_MANGLE
+#include "jemalloc_test.h"
+
+#define	NTHREADS 10
+
+void *
+je_thread_start(void *arg)
+{
+	unsigned main_arena_ind = *(unsigned *)arg;
+	void *p;
+	unsigned arena_ind;
+	size_t size;
+	int err;
+
+	p = malloc(1);
+	if (p == NULL) {
+		malloc_printf("%s(): Error in malloc()\n", __func__);
+		return (void *)1;
+	}
+	free(p);
+
+	size = sizeof(arena_ind);
+	if ((err = mallctl("thread.arena", &arena_ind, &size, &main_arena_ind,
+	    sizeof(main_arena_ind)))) {
+		malloc_printf("%s(): Error in mallctl(): %s\n", __func__,
+		    strerror(err));
+		return (void *)1;
+	}
+
+	size = sizeof(arena_ind);
+	if ((err = mallctl("thread.arena", &arena_ind, &size, NULL,
+	    0))) {
+		malloc_printf("%s(): Error in mallctl(): %s\n", __func__,
+		    strerror(err));
+		return (void *)1;
+	}
+	assert(arena_ind == main_arena_ind);
+
+	return (NULL);
+}
+
+int
+main(void)
+{
+	int ret = 0;
+	void *p;
+	unsigned arena_ind;
+	size_t size;
+	int err;
+	je_thread_t threads[NTHREADS];
+	unsigned i;
+
+	malloc_printf("Test begin\n");
+
+	p = malloc(1);
+	if (p == NULL) {
+		malloc_printf("%s(): Error in malloc()\n", __func__);
+		ret = 1;
+		goto label_return;
+	}
+
+	size = sizeof(arena_ind);
+	if ((err = mallctl("thread.arena", &arena_ind, &size, NULL, 0))) {
+		malloc_printf("%s(): Error in mallctl(): %s\n", __func__,
+		    strerror(err));
+		ret = 1;
+		goto label_return;
+	}
+
+	for (i = 0; i < NTHREADS; i++) {
+		je_thread_create(&threads[i], je_thread_start,
+		    (void *)&arena_ind);
+	}
+
+	for (i = 0; i < NTHREADS; i++)
+		je_thread_join(threads[i], (void *)&ret);
+
+label_return:
+	malloc_printf("Test end\n");
+	return (ret);
+}
diff --git a/extra/jemalloc/test/thread_arena.exp b/extra/jemalloc/test/thread_arena.exp
new file mode 100644
index 00000000000..369a88dd240
--- /dev/null
+++ b/extra/jemalloc/test/thread_arena.exp
@@ -0,0 +1,2 @@
+Test begin
+Test end
diff --git a/extra/jemalloc/test/thread_tcache_enabled.c b/extra/jemalloc/test/thread_tcache_enabled.c
new file mode 100644
index 00000000000..2061b7bbaff
--- /dev/null
+++ b/extra/jemalloc/test/thread_tcache_enabled.c
@@ -0,0 +1,91 @@
+#define	JEMALLOC_MANGLE
+#include "jemalloc_test.h"
+
+void *
+je_thread_start(void *arg)
+{
+	int err;
+	size_t sz;
+	bool e0, e1;
+
+	sz = sizeof(bool);
+	if ((err = mallctl("thread.tcache.enabled", &e0, &sz, NULL, 0))) {
+		if (err == ENOENT) {
+#ifdef JEMALLOC_TCACHE
+			assert(false);
+#endif
+		}
+		goto label_return;
+	}
+
+	if (e0) {
+		e1 = false;
+		assert(mallctl("thread.tcache.enabled", &e0, &sz, &e1, sz)
+		    == 0);
+		assert(e0);
+	}
+
+	e1 = true;
+	assert(mallctl("thread.tcache.enabled", &e0, &sz, &e1, sz) == 0);
+	assert(e0 == false);
+
+	e1 = true;
+	assert(mallctl("thread.tcache.enabled", &e0, &sz, &e1, sz) == 0);
+	assert(e0);
+
+	e1 = false;
+	assert(mallctl("thread.tcache.enabled", &e0, &sz, &e1, sz) == 0);
+	assert(e0);
+
+	e1 = false;
+	assert(mallctl("thread.tcache.enabled", &e0, &sz, &e1, sz) == 0);
+	assert(e0 == false);
+
+	free(malloc(1));
+	e1 = true;
+	assert(mallctl("thread.tcache.enabled", &e0, &sz, &e1, sz) == 0);
+	assert(e0 == false);
+
+	free(malloc(1));
+	e1 = true;
+	assert(mallctl("thread.tcache.enabled", &e0, &sz, &e1, sz) == 0);
+	assert(e0);
+
+	free(malloc(1));
+	e1 = false;
+	assert(mallctl("thread.tcache.enabled", &e0, &sz, &e1, sz) == 0);
+	assert(e0);
+
+	free(malloc(1));
+	e1 = false;
+	assert(mallctl("thread.tcache.enabled", &e0, &sz, &e1, sz) == 0);
+	assert(e0 == false);
+
+	free(malloc(1));
+label_return:
+	return (NULL);
+}
+
+int
+main(void)
+{
+	int ret = 0;
+	je_thread_t thread;
+
+	malloc_printf("Test begin\n");
+
+	je_thread_start(NULL);
+
+	je_thread_create(&thread, je_thread_start, NULL);
+	je_thread_join(thread, (void *)&ret);
+
+	je_thread_start(NULL);
+
+	je_thread_create(&thread, je_thread_start, NULL);
+	je_thread_join(thread, (void *)&ret);
+
+	je_thread_start(NULL);
+
+	malloc_printf("Test end\n");
+	return (ret);
+}
diff --git a/extra/jemalloc/test/thread_tcache_enabled.exp b/extra/jemalloc/test/thread_tcache_enabled.exp
new file mode 100644
index 00000000000..369a88dd240
--- /dev/null
+++ b/extra/jemalloc/test/thread_tcache_enabled.exp
@@ -0,0 +1,2 @@
+Test begin
+Test end