| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
RT #133204
Wean this function off int/I32 and onto UV/Size_t.
Also, replace all malloc-ish calls with a wrapper that does
overflow checks,
In particular, it was doing (nlen + vlen + 2) which could wrap when
the combined length of the environment variable name and value
exceeded around 0x7fffffff.
The wrapper check function is probably overkill, but belt and braces...
NB this function has several variant parts, #ifdef'ed by platform
type; I have blindly changed the parts that aren't compiled under linux.
(cherry picked from commit 34716e2a6ee2af96078d62b065b7785c001194be)
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
RT #131575
re_intuit_start() could calculate a maximum end position less than the
current start position. This used to get rejected by fbm_intr(), until
v5.23.3-110-g147f21b, which made fbm_intr() faster and removed unnecessary
checks.
This commits fixes re_intuit_start(), and adds an assert to fbm_intr().
(cherry picked from commit bb152a4b442f7718fd37d32cc558be675e8ae1ae)
|
|
|
|
|
|
| |
This reverts commit dd1b95f812312c85390f487887cdd55282fcd6ce.
Accidentally pushed work pending unfreeze.
|
|
|
|
|
|
| |
I think this has been outright broken for a long time, and it
doesnt make any sense to since mandatory randomization anyway,
so simply remove it.
|
|
|
|
|
|
|
| |
With v5.25.11-59-g7335cb8 I added an include guard. PERL_UTIL_H_,
but added it to util.c rather than util.h.
I am not a smart man....
|
|
|
|
|
|
| |
see RT #131110 for the bikeshedding on what name to use.
Based on an original patch by Jim Schneider
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
RT #130841
In general code, change this idiom:
PL_foo_max += size;
Renew(PL_foo, PL_foo_max, foo_t);
to
Renew(PL_foo, PL_foo_max + size, foo_t);
PL_foo_max += size;
so that if Renew dies, PL_foo_max won't be left hanging.
|
|
|
|
| |
These short functions are called in inner loops and regex backtracking.
|
|
|
|
|
|
| |
Some vars have been tagged as const because they do not change in their
new scopes. In pp_reverse in pp.c, I32 tmp is only used to hold a char,
so is changed to char.
|
|
|
|
| |
helps to test builds with the right options...
|
|
|
|
|
|
|
|
|
|
| |
In the unlikely case quadmath_snprintf() fails both sv_vcatpvfn_flags()
and my_snprintf() could leak the temp format string returned by
quadmath_format_single() if quadmath_format_single() had to rebuild
the format.
Getting quadmath_snprintf() to fail in this context seems impractical,
but future changes may make it happen, so clean up after ourselves.
|
|
|
|
|
|
|
| |
For the case where littlestr hasn't been FBM compiled (!SvVALID()), it
can't be SvTAIL(), so there's no need for optional \n handling.
Spotted by Coverity.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
RT #130621
In Perl_mess_sv(), don't try to add an "at foo line NN" to the error
message if PL_curcop is null.
In the ticket above, the reason that PL_curcop is null is the less
than optimal way that evals free their optree: ideally the optree should
be attached to the eval CV and freed when the CV is; instead a separate
SAVEFREEOP() is done. But that fix is for another time; regardless,
mess_sv() should have a PL_curcop != NULL guard anyway.
|
|
|
|
|
|
|
|
| |
This would not compile due to passing a thread context parameter to
Safefree, which isn't expecting it.
No claim is made by this commit that it works well under threads. No
claim is made in the documentation to that effect either.
|
|
|
|
|
|
| |
These variables either control or reveal information used in perl's
hash implementation that a careful user may not want controlled or
exposed.
|
|
|
|
|
|
|
| |
This function is equivalent to sv_setsv(sv, &PL_sv_undef), but more
efficient.
Also change the obvious places in the core to use the new idiom.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
C++11 requires space between the end of a string literal and a macro, so
that a feature can unambiguously be added to the language. Starting in
g++ 6.2, the compiler emits a warning when there isn't a space
(presumably so that future versions can support C++11). Unfortunately
there are many such instances in the perl core. This commit fixes
those, including those in ext/, but individual commits will be used for
the other modules, those in dist/ and cpan/.
This commit also inserts space at the end of a macro before a string
literal, even though that is not deprecated, and removes useless ""
literals following a macro (instead of inserting a blank). The result
is easier to read, making the macro stand out, and be clearer as to the
intention.
Code and modules included with the Perl core need to be compilable using
C++. This is so that perl can be embedded in C++ programs. (Actually,
only the hdr files need to be so compilable, but it would be hard to
test that just the hdrs are compilable.) So we need to accommodate
changes to the C++ language.
|
|
|
|
|
|
|
|
| |
Now that SvVALID() no longer just checks an SV flag, but instead checks
for the existence of a certain type of magic, avoid using this more
expensive macro when its not really needed.
Also, and an extra flag test to SvVALID() to make it fail quicker.
|
|
|
|
|
|
|
|
| |
This flag is set on an SV to indicate that it has PERL_MAGIC_bm
(fast Boyer-Moore) magic attached. Instead just directly check whether
it has such magic.
This frees up the 0x40000000 bit for anything except AVs and HVs
|
|
|
|
|
|
|
|
|
|
|
|
| |
(but keep SvTAIL())
This flag is only set on SVs that have Boyer-Moore magic attached.
Such SVs already re-purpose the unused IVX slot of that SV to store
BmUSEFUL. This commit repurposes the unused NVX slot to store this
boolean value instead.
Now that flag bit (0x80000000) is only used with AVs, HVs, RVs and
scalar SVs with IOK.
|
|
|
|
|
|
| |
Only use the SvTAIL() macro when we've already confirmed that
the SV is SvVALID() - this is in preparation for removing the
SVpbm_TAIL flag in the next commit
|
|
|
|
|
| |
Its a bit confusing having both S_delimcpy() and Perl_delimcpy()
functions.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
[perl #129064] heap-buffer-overflow S_scan_heredoc
[perl #129176] Conditional jump depends on uninitialized values in
S_scan_heredoc
Perl_delimcpy() is supposed to copy a delimited string to another buffer;
it handles \-<delimiter> escapes, but if the backslash is the last
character in the src buffer, it could overrun the end of the buffer
slightly.
Also document a bit better what this function is supposed to do.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
$ cat > foo
#!/usr/bin/perl
print "What?!\n"
^D
$ chmod +x foo
$ ./perl -Ilib -Te '$ENV{PATH}="."; exec "foo"'
Insecure directory in $ENV{PATH} while running with -T switch at -e line 1.
That is what I expect to see. But:
$ ./perl -Ilib -Te '$ENV{PATH}="/\\:."; exec "foo"'
What?!
Perl is allowing the \ to escape the :, but the \ is not treated as an
escape by the system, allowing a relative path in PATH to be consid-
ered safe.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
in the usequadmath branch, gcc is too clever for its own good:
PERL_UNUSED_ARG(ap);
gives:
util.c:5299:18: warning: ‘sizeof’ on array function parameter ‘ap’ will
return size of ‘__va_list_tag *’ [-Wsizeof-array-argument]
Stick in a void* cast to shut it up.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
fea1d2dd5d210564d4 turned instr into a macro. It also left the
actual function in util.c while commenting out the prototype
in proto.h (via the m flag in embed.fnc).
A function compiled without a prototype under C++ does not get
declared with extern "C" and thus gets mangled, which breaks the
build with a strict linker (VMS, possibly AIX) because the
expected symbol name is no longer produced. Without a strict
linker, it just breaks the binary compatibility that was presumably
the nominal reason for leaving the function around in the first
place.
So move the function into mathoms.c and put its prototype in the
extern "C"-guarded section at the top of the same file.
We also have to fake the PERL_ARGS_ASSERT_INSTR macro since its
original declaration in proto.h is commented out but the porting
test t/porting/args_assert.t will take revenge if it doesn't
find the macro being used somewhere.
|
|
|
|
|
|
|
|
| |
It had rotted a bit Well, more than one probably.
Move the declarations of the functions Perl_mem_log_alloc etc from handy.h
into embed.fnc where whey belong, and where Malloc_t will have already
been defined.
|
| |
|
| |
|
|
|
|
| |
Fix a thinko in 22ff3130.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This commit:
1. Renames the various dtrace probe macros into a consistent and
self-documenting pattern, e.g.
ENTRY_PROBE => PERL_DTRACE_PROBE_ENTRY
RETURN_PROBE => PERL_DTRACE_PROBE_RETURN
Since they're supposed to be defined only under PERL_CORE, this shouldn't
break anything that's not being naughty.
2. Implement the main body of these macros using a real function.
They were formerly defined along the lines of
if (PERL_SUB_ENTRY_ENABLED())
PERL_SUB_ENTRY(...);
The PERL_SUB_ENTRY() part is a macro generated by the dtrace system, which
for example on linux expands to a large bunch of assembly directives.
Replace the direct macro with a function wrapper, e.g.
if (PERL_SUB_ENTRY_ENABLED())
Perl_dtrace_probe_call(aTHX_ cv, TRUE);
This reduces to once the number of times the macro is expanded.
The new functions also take simpler args and then process the values they
need using intermediate temporary vars to avoid huge macro expansions.
For example
ENTRY_PROBE(CvNAMED(cv)
? HEK_KEY(CvNAME_HEK(cv))
: GvENAME(CvGV(cv)),
CopFILE((const COP *)CvSTART(cv)),
CopLINE((const COP *)CvSTART(cv)),
CopSTASHPV((const COP *)CvSTART(cv)));
is now
PERL_DTRACE_PROBE_ENTRY(cv);
This reduces the executable size by 1K on -O2 -Dusedtrace builds,
and by 45K on -DDEBUGGING -Dusedtrace builds.
|
|
|
|
|
|
|
| |
popen(): handle better the case where the popened external
might exit before the child process manages to start.
pclose(): protect with a semaphore.
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
| |
My change in e396210 was incomplete, that change was intended to
force use of setenv()/unsetenv() on Darwin, but ended up using putenv()
instead, which is a leaky mechanism.
Added darwin to the list of many others that work better with setenv()/
unsetenv().
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
util.c:2729:1: warning: mutex 'PL_perlio_mutex' is still held at the end of function [-Wthread-safety-analysis]
}
util.c:2729:1: warning: mutex 'PL_op_mutex' is still held at the end of function [-Wthread-safety-analysis]
}
util.c:2739:5: warning: releasing mutex 'PL_perlio_mutex' that was not held [-Wthread-safety-analysis]
MUTEX_UNLOCK(&PL_perlio_mutex);
util.c:2744:5: warning: releasing mutex 'PL_op_mutex' that was not held [-Wthread-safety-analysis]
OP_REFCNT_UNLOCK;
|
|
|
|
|
|
|
|
|
|
|
|
| |
http://clang.llvm.org/docs/ThreadSafetyAnalysis.html
Static (compile-time) annotations for declaring the multithreaded
behavior of functions, variables, and capabilities (like mutexes).
Available since about clang 3.5.
./Configure -des -Dusedevel -Dusethreads -Dcc=clang -Accflags='-Wthread-safety'
clang -Wthread-safety then checks the validity of the annotations.
|
|
|
|
|
|
|
|
| |
In 97466d2cbf895b I added a declaration at function scope, but in
some paths that was overridden at an inner scope, leaving the
function-level one declared but not used. So lets go back to the
original intent of 45a23732c73c8 and have a separate declaration
in each block that needs it.
|
|
|
|
|
|
|
| |
This reverts commit 97466d2cbf895b35ac41b8bf7c31db955b52d48e.
Adding a declaration at function scope causes unused variable
warnings when there are paths that redeclare it at an inner scope.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
TL;DR: on platforms with a libc memchr() implementation which makes good
use of underlying hardware support, patterns which include fixed
substrings will now often be much faster; for example with glibc on on a
recent x86_64 CPU, this:
$s = "a" x 1000 . "wxyz";
$s =~ /wxyz/ for 1..30000
is now about 7 times faster. On systems with slow memchr(), e.g. 32-bit
ARM Raspberry Pi, there will be a small or little speedup. Conversely,
some pathological cases, such as "ab" x 1000 =~ /aa/ will be slower now;
up to 3 times slower on the rPi, 1.5x slower on x86_64.
In detail:
The perl core includes a Boyer-Moore substring matcher, Perl_fbm_instr(),
which is used to quickly find a fixed substring within a longer string,
where a table of lookups is pre-computed from the substring. As well as
being used in index() when the substring is a constant, its main use
is in patterns. When the regex engine compiles a pattern, it typically
takes note of the two longest fixed substrings within the pattern; for
example in
/[01]abc\w+de\d+fghij+/
the two longest are "abc" and "fghij". The engine uses Perl_fbm_instr() to
scan for these two strings before running the full NFA. This often allows
the string to be quickly rejected, or to find a suitable minimum starting
point to run the NFA.
However, Perl_fbm_instr() was written about 16 years ago and has been
virtually untouched since, so it could do with some love.
It currently special-cases strings of length 1 and 2, using roll-your-own
loops along the lines of
while (s < end) { if (*s++ = c1) ... }
while strings of length 3+ use the Boyer-Moore algorithm. The big
advantage of BM is that in a best-case, where none of the characters from
the substring are found in this region of the string, it only has to test
every N'th char, where N is length of the substring. For example when
searching for wxyz in abcdefghikl..., it just reads and tests d,h,l,..
However these days some platforms have decent memchr() implementations.
For example, glibc has assembly-level implementations for i386, x86_64,
sparc32/64, powerpc32/64, s390-32/64, arm, m68k and ia64 by the looks of
it. These can often be substantially faster than a C-level implementation.
This commit makes Perl_fbm_instr() use memchr() where possible.
For the length == 1 special case, it just calls memchr() directly rather
than using a loop as previously.
For the length == 2 special case, it continues to distinguish the cases
where the two chars of the substring are the same or differ. For the
former it calls memchr() after an initial direct failure, i.e.
if (*s != c) { s++; s = memchr(....); ... }
For the latter case it does a similar direct test first (to avoid the
costly overhead of a call to memchr() when the next char is the one we
seek anyway), but in addition, on each failure to find the second char
following a found first char, it swaps which char it's searching for.
This means that in something like "aaaaaaa..." =~ /ab/, it wont keep
hopping 1 char position with memchar(s,'a'); after the first hop it
will do memchr(s,'b') and skip lots of chars in one go. This helps reduce
the number of pathological cases.
For the length >= 3 cases (normal BM), it keeps using BM, but after each
iteration where the pointer has been incremented by the skip determined by
the BM algorithm, it now does an additional
if (*s != c) { s++; s = memchr(....); ... }
step before running the next iteration of BM.
|
|
|
|
|
| |
Expand the commentary at the start of this function; add more blank lines
to separate chunks of code, and document what SVpbm_TAIL is for.
|
|
|
|
|
| |
The VMS-specific corner of the ifdef jungle ended up using
statbuf without declaring it, so add a declaration.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Perl_nextargv has to have access to the Stat_t that is written to inside
S_openn_cleanup or else run/switches.t, io/argv.t, io/inplace.t, and
io/iprefix.t will fail. Removing the uses of PL_statbuf that are using
PL_statbuf due to historical reason, and not using PL_statbuf to pass data
between different funcs/different callstacks. This patch makes it easier to
remove PL_statbuf in the future since the number uses of it has been
reduced.
-in Perl_apply move SETERRNO before tot--; so the branch can be combined
with other "tot--;" branches by CC optmizer
-combine 2 Perl_croak(aTHX_ "Illegal suidscript"); statements in
S_validate_suid to make code look simpler, drop my_perl arg for space
efficiency on threads of rarely executed code
|