summaryrefslogtreecommitdiff
path: root/rts
Commit message (Collapse)AuthorAgeFilesLines
* mmapForLinkerMarkExecutable: do nothing when len = 0Travis Whitaker2021-11-121-0/+6
|
* Make: Do not generate ghc.* headers in stage0John Ericson2021-11-121-17/+9
| | | | | GHC should get everything it needs from the RTS, which for stage0 is the "old" RTS that comes from the bootstrap compiler.
* Don't expose bignum backend in ghc --info (#20495)Sylvain Henry2021-11-091-1/+0
| | | | | | | GHC is bignum backend agnostic and shouldn't report this information as in the future ghc-bignum will be reinstallable potentially with a different backend that GHC is unaware of. Moreover as #20495 shows the returned information may be wrong currently.
* RTS: open timerfd synchronously (#20618)Sylvain Henry2021-11-091-19/+32
|
* Avoid GHC_STAGE and other include bitsJohn Ericson2021-11-051-2/+0
| | | | | | | | | We should strive to make our includes in terms of the RTS as much as possible. One place there that is not possible, the llvm version, we make a new tiny header Stage numbers are somewhat arbitrary, if we simple need a newer RTS, we should say so.
* make: Futher systematize handling of generated headersJohn Ericson2021-11-052-14/+8
| | | | | This will make it easier to add and remove generated headers, as we will do when we add a configure script for the RTS.
* Correct load_load_barrier for risc-vTakenobu Tani2021-11-041-1/+1
| | | | | | | | | This patch corrects the instruction for load_load_barrier(). Current load_load_barrier() incorrectly uses `fence w,r`. It means a store-load barrier. See also linux-kernel's smp_rmb() implementation: https://github.com/torvalds/linux/blob/v5.14/arch/riscv/include/asm/barrier.h#L27
* Treat generated RTS headers in a more consistent mannerJohn Ericson2021-11-022-1/+24
| | | | We can depend on all of them at once the same way.
* Make build system: Put make generated include's in RTS distdirsJohn Ericson2021-10-313-13/+12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | These are best thought of as being part of the RTS. - After !6791, `ghcautoconf.h` won't be used by the compiler inappropriately. - `ghcversion.h` is only used once outside the RTS, which is `compiler/cbits/genSym.c`. Except we *do* mean the RTS GHC is built against there, so it's better if we always get get the installed version. - `ghcplatform.h` alone is used extensively outside the RTS, but since we no longer have a target platform it is perfectly safe/correct to get the info from the previous RTS. All 3 are exported from the RTS currently and in the bootstrap window. This commit just swaps directories around, such that the new headers may continue to be used in stage 0 despite the reasoning above, but the idea is that we can subsequently make more interesting changes doubling down on the reasoning above. In particular, in !6803 we'll start "morally" moving `ghcautonconf.h` over, introducing an RTS configure script and temporary header of its `AC_DEFINE`s until the top-level configure script doesn't define any more. Progress towards #17191
* make build system: RTS should use dist-install not distJohn Ericson2021-10-292-78/+78
| | | | | | | | | | | | | | | | | | | | | | | | | | This is the following find and replace: - `rts/dist` -> `rts/dist-install` # for paths - `rts_dist` -> `rts_dist-install` # for make rules and vars - `,dist` -> `,dist-install` # for make, just in rts/ghc.mk` Why do this? Does it matter when the RTS is just built once? The answer is, yes, I think it does, because I want the distdir--stage correspondence to be consistent. In particular, for #17191 and continuing from d5de970dafd5876ef30601697576167f56b9c132 I am going to make the headers (`rts/includes`) increasingly the responsibility of the RTS (hence their new location). However, those headers are current made for multiple stages. This will probably become unnecessary as work on #17191 progresses and the compiler proper becomes more of a freestanding cabal package (e.g. a library that can be downloaded from Hackage and built without any autoconf). However, until that is finished, we have will transitional period where the RTS and headers need to agree on dirs for multiple stages. I know the make build system is going away, but it's not going yet, so I need to change it to unblock things :).
* Make build system: Generalize and/or document distdirsJohn Ericson2021-10-271-1/+1
| | | | | | | | | `manual-package-config` should not hard-code the distdir, and no longer does Elsewhere, we must continue to hard-code due to inconsitent distdir names across stages, so we document this referring to the existing note "inconsistent distdirs".
* Fix dangling reference to RtsConfig.hJohn Ericson2021-10-241-1/+1
| | | | | It hasn't existed since a2a67cd520b9841114d69a87a423dabcb3b4368e -- in 2009!
* Null eventlog writerOleg Grenrus2021-10-155-0/+37
|
* Check for libatomic dependency for atomic operationsHaochen Tong2021-10-122-0/+8
| | | | | | | Some platforms (e.g. RISC-V) require linking against libatomic for some (e.g. sub-word-sized) atomic operations. Fixes #19119.
* rts/RtsSymbols: fini arrayBen Gamari2021-10-121-0/+13
|
* rts/RtsSymbols: Declare atexit as a strong symbolBen Gamari2021-10-121-1/+10
|
* rts/linker: Introduce a notion of strong symbolsBen Gamari2021-10-124-26/+51
|
* rts/RtsSymbols: Add environBen Gamari2021-10-121-0/+1
|
* rts/linker: Resolve __fini_array_* symbols to NULLGHC GitLab CI2021-10-121-0/+4
| | | | | If the __fini_array_{start,end} symbols are not defined (e.g. as is often the case when linking against musl) then resolve them to NULL.
* rts/linker: Define _DYNAMIC when necessaryBen Gamari2021-10-121-1/+13
| | | | | | Usually the dynamic linker would define _DYNAMIC. However, when dynamic linking is not supported (e.g. on musl) it is safe to define it to be NULL.
* rts/Linker: Fix __dso_handle handlingGHC GitLab CI2021-10-121-10/+27
| | | | | | | | | | | | Previously the linker's handling of __dso_handle was quite wrong. Not only did we claim that __dso_handle could be NULL when statically linking (which it can not), we didn't even implement this mislead theory faithfully and instead resolved the symbol to a random pointer. This lead to the failing relocations on AArch64 noted in #20493. Here we try to implement __dso_handle as a dynamic linker would do, choosing an address within the loaded object (specifically its start address) to serve as the object's handle.
* rts/nonmoving: Enable selector optimisation by defaultBen Gamari2021-10-123-12/+1
|
* rts/primops: Fix write barrier in stg_atomicModifyMutVarzuzhBen Gamari2021-10-121-4/+4
| | | | | | | Previously the call to dirty_MUT_VAR in stg_atomicModifyMutVarzuzh was missing its final argument. Fixes #20414.
* rts/nonmoving: Rename mark_* to trace_*Ben Gamari2021-10-121-42/+42
| | | | These functions really do no marking; they merely trace pointers.
* nonmoving: Fix and factor out mark_trec_chunkBen Gamari2021-10-121-22/+17
| | | | | We need to ensure that the TRecChunk itself is marked, in addition to the TRecs it contains.
* Fix nonmoving gen label in gc stats reportTeo Camarasu2021-10-081-3/+6
| | | | | | | | The current code assumes the non-moving generation is always generation 1, but this isn't the case if the amount of generations is greater than 2 Fixes #20461
* fix non-moving gc heap space requirements estimateTeo Camarasu2021-10-071-1/+1
| | | | | | | | | The space requirements of the non-moving gc are comparable to the compacting gc, not the copying gc. The copying gc requires a much larger overhead. Fixes #20475
* Corrected types of thread ids obtained from the RTSMann mit Hut2021-10-0611-47/+54
| | | | | | | | | | | | | | While the thread ids had been changed to 64 bit words in e57b7cc6d8b1222e0939d19c265b51d2c3c2b4c0 the return type of the foreign import function used to retrieve these ids - namely 'GHC.Conc.Sync.getThreadId' - was never updated accordingly. In order to fix that this function returns now a 'CUULong'. In addition to that the types used in the thread labeling subsystem were adjusted as well and several format strings were modified throughout the whole RTS to display thread ids in a consistent and correct way. Fixes #16761
* rts: Unify stack dirtiness checkBen Gamari2021-10-022-6/+6
| | | | | | This fixes an inconsistency where one dirtiness check would not mask out the STACK_DIRTY flag, meaning it may also be affected by the STACK_SANE flag.
* rts: Add missing write barriers in MVar wake-up pathsBen Gamari2021-10-022-12/+37
| | | | | | | | | | | Previously PerformPut failed to respect the non-moving collector's snapshot invariant, hiding references to an MVar and its new value by overwriting a stack frame without dirtying the stack. Fix this. PerformTake exhibited a similar bug, failing to dirty (and therefore mark) the blocked stack before mutating it. Closes #20399.
* Fix comment typosKirill Zaborsky2021-09-291-2/+2
|
* Remove special case for large objects in allocateForCompactFabian Thorand2021-09-291-11/+0
| | | | | | | | | | | | | | | | | | | | | | | | allocateForCompact() is called when the current allocation for the compact region does not fit in the nursery. It previously had a special case for objects exceeding the large object threshold. In that case, it would allocate a new compact region block just for that object. That led to a lot of small blocks being allocated in compact regions with a larger default block size (`autoBlockW`). This commit removes this special case because having a lot of small compact region blocks contributes significantly to memory fragmentation. The removal should be valid because - a more generic case for allocating a new compact region block follows at the end of allocateForCompact(), and that one takes `autoBlockW` into account - the reason for allocating separate blocks for large objects in the main heap seems to be to avoid copying during GCs, but once inside the compact region, the object will never be copied anyway. Fixes #18757. A regression test T18757 was added.
* rts: Ensure that headers don't refer to undefined __STDC_VERSION__Ben Gamari2021-09-231-0/+6
| | | | | | | | | Previously the C/C++ language version check in STG could throw an undefined macro warning due to __STDC_VERSION__ when compiled with a C++ compiler. Fix this by defining __STDC_VERSION__==0 when compiling with a C++ compiler. Fixes #20394.
* configure: Clarify meaning of CabalHaveLibffiBen Gamari2021-09-232-4/+2
| | | | | Previously the meaning of this flag was unclear and as a result I suspect that CabalHaveLibffi could be incorrectly False.
* Use Info Table Provenances to decode cloned stack (#18163)Sven Tennie2021-09-237-24/+137
| | | | | | | | | | | | | | | | Emit an Info Table Provenance Entry (IPE) for every stack represeted info table if -finfo-table-map is turned on. To decode a cloned stack, lookupIPE() is used. It provides a mapping between info tables and their source location. Please see these notes for details: - [Stacktraces from Info Table Provenance Entries (IPE based stack unwinding)] - [Mapping Info Tables to Source Positions] Metric Increase: T12545
* Introduce stack snapshotting / cloning (#18741)Sven Tennie2021-09-2310-1/+165
| | | | | | | | | | | | | | Add `StackSnapshot#` primitive type that represents a cloned stack (StgStack). The cloning interface consists of two functions, that clone either the treads own stack (cloneMyStack) or another threads stack (cloneThreadStack). The stack snapshot is offline/cold, i.e. it isn't evaluated any further. This is useful for analyses as it prevents concurrent modifications. For technical details, please see Note [Stack Cloning]. Co-authored-by: Ben Gamari <bgamari.foss@gmail.com> Co-authored-by: Matthew Pickering <matthewtpickering@gmail.com>
* Link with libm dynamically (#19877)Sylvain Henry2021-09-221-0/+1
| | | | The compiler should be independent of the target.
* Update error message to suggest the user consider OOM over RTS bug.Thomas M. DuBuisson2021-09-171-1/+3
| | | | Fix #17039
* rts: Factor out TRACE_ cache update logicBen Gamari2021-09-081-8/+18
| | | | Just a small refactoring to perhaps enable code reuse later.
* hadrian: Ensure that settings is regenerated during bindist installationBen Gamari2021-09-071-0/+1
| | | | | | | | | | | | | | | | | | | | | | | | Previously Hadrian would simply install the settings file generated in the build environment during the binary distribution installation. This is wrong since these environments may differ (e.g. different `cc` versions). We noticed on Darwin when installation of a binary distribution produced on a newer Darwin release resulted in a broken compiler due to the installed `settings` file incorrectly claiming that `cc` supported `-no-pie`. Fixing this sadly requires a bit of code duplication since `settings` is produced by Hadrian and not `configure`. For now I have simply duplicated the `settings` generation logic used by the Make build system into Hadrian's bindist Makefile. Ultimately the solution will probably involve shipping a freestanding utility to replace `configure`'s toolchain probing logic and generate a toolchain description file (similar to `settings`) as described in #19877. Fixes #20253.
* Add -Wl,-U,___darwin_check_fd_set_overflow to rts/package.conf.inMatthew Pickering2021-08-191-0/+4
| | | | | | | | | The make build system apparently uses this special package.conf rather than generating it from the cabal file. Ticket: #19950 (cherry picked from commit e316a0f3e7a733fac0c30633767487db086c4cd0)
* Optimize Info Table Provenance Entries (IPEs) Map creation and lookupSven Tennie2021-08-116-57/+199
| | | | | | | | | | | | | | | Using a hash map reduces the complexity of lookupIPE(), making it non linear. On registration each IPE list is added to a temporary IPE lists buffer, reducing registration time. The hash map is built lazily on first lookup. IPE event output to stderr is added with tests. For details, please see Note [The Info Table Provenance Entry (IPE) Map]. A performance test for IPE registration and lookup can be found here: https://gitlab.haskell.org/ghc/ghc/-/merge_requests/5724#note_370806
* Move `/includes` to `/rts/include`, sort per package betterJohn Ericson2021-08-0986-20/+12173
| | | | | | | | | | | | | | | | | | | | | | | | | | | In order to make the packages in this repo "reinstallable", we need to associate source code with a specific packages. Having a top level `/includes` dir that mixes concerns (which packages' includes?) gets in the way of this. To start, I have moved everything to `rts/`, which is mostly correct. There are a few things however that really don't belong in the rts (like the generated constants haskell type, `CodeGen.Platform.h`). Those needed to be manually adjusted. Things of note: - No symlinking for sake of windows, so we hard-link at configure time. - `CodeGen.Platform.h` no longer as `.hs` extension (in addition to being moved to `compiler/`) so as not to confuse anyone, since it is next to Haskell files. - Blanket `-Iincludes` is gone in both build systems, include paths now more strictly respect per-package dependencies. - `deriveConstants` has been taught to not require a `--target-os` flag when generating the platform-agnostic Haskell type. Make takes advantage of this, but Hadrian has yet to.
* Make `PosixSource.h` installed and under `rts/`John Ericson2021-08-0990-135/+97
| | | | | | is used outside of the rts so we do this rather than just fish it out of the repo in ad-hoc way, in order to make packages in this repo more self-contained.
* rts: Fix use of sized array in Heap.hBen Gamari2021-08-091-2/+2
| | | | | | Sized arrays cannot be used in headers that might be imported from C++. Fixes #20199.
* Fix ASSERTS_ENABLED CPPSylvain Henry2021-08-031-1/+1
|
* rts/OSThreads: Fix reference clock of timedWaitConditionBen Gamari2021-08-021-12/+33
| | | | | | | | | | | | | | | Previously `timedWaitCondition` assumed that timeouts were referenced against `CLOCK_MONOTONIC`. This is wrong; by default `pthread_cond_timedwait` references against `CLOCK_REALTIME`, although this can be overridden using `pthread_condattr_setclock`. Fix this and add support for using `CLOCK_MONOTONIC` whenever possible as it is more robust against system time changes and is likely cheaper to query. Unfortunately, this is complicated by the fact that older versions of Darwin did not provide `clock_gettime`, which means we also need to introduce a fallback path using `gettimeofday`. Fixes #20144.
* PrimOps: Add CAS op for all int sizesPeter Trommler2021-08-024-6/+58
| | | | | | | | | | | PPC NCG: Implement CAS inline for 32 and 64 bit testsuite: Add tests for smaller atomic CAS X86 NCG: Catch calls to CAS C fallback Primops: Add atomicCasWord[8|16|32|64]Addr# Add tests for atomicCasWord[8|16|32|64]Addr# Add changelog entry for new primops X86 NCG: Fix MO-Cmpxchg W64 on 32-bit arch ghc-prim: 64-bit CAS C fallback on all archs
* Add configure flag to enable ASSERTs in all waysDaniel Gröber2021-07-294-5/+4
| | | | | | | | Running the test suite with asserts enabled is somewhat tricky at the moment as running it with a GHC compiled the DEBUG way has some hundred failures from the start. These seem to be unrelated to assertions though. So this provides a toggle to make it easier to debug failing assertions using the test suite.
* rts: Allow building with ASSERTs on in non-DEBUG wayDaniel Gröber2021-07-293-8/+2
| | | | | | | | | | We have a couple of places where the conditions in asserts depend on code ifdefed out when DEBUG is off. I'd like to allow compiling assertions into non-DEBUG RTSen so that won't do. Currently if we remove the conditional around the definition of ASSERT() the build will not actually work due to a deadlock caused by initMutex not initializing mutexes with PTHREAD_MUTEX_ERRORCHECK because DEBUG is off.
View Lorry Controller Status