Craig Topper [Fri, 23 Jul 2021 06:20:54 +0000 (23:20 -0700)]
[X86] Add test case simplified from PR51175. NFC
Fraser Cormack [Thu, 22 Jul 2021 16:06:51 +0000 (17:06 +0100)]
[SelectionDAG][RISCV] Add tests showing missed scalable-splat optimizations
These tests show missed opportunities in the SelectionDAG layer when
dealing with scalable-vector splats. All of these are handled for the
equivalent `ISD::BUILD_VECTOR` code, and the tests have largely been
translated from the equivalent X86 tests.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D106574
Siva Chandra Reddy [Thu, 22 Jul 2021 22:05:34 +0000 (22:05 +0000)]
[libc] Raise denormal exception if the libc defines __FE_DENORM.
Reviewed By: michaelrj
Differential Revision: https://reviews.llvm.org/D106604
Johannes Doerfert [Thu, 22 Jul 2021 17:18:46 +0000 (12:18 -0500)]
[OpenMP] Simplify the ThreadStackTy for globalization fallback
With D106496 we can make the globalization fallback stack much simpler
and this version doesn't seem to experience the spurious failures and
deadlocks we have seen before.
Differential Revision: https://reviews.llvm.org/D106576
Johannes Doerfert [Thu, 22 Jul 2021 05:36:31 +0000 (00:36 -0500)]
[Attributor] If provided, only look at simplification callbacks not IR
A simplification callback can mean that the IR value is modified beyond
the apparent IR semantics. That is, a `i1 true` could be replaced by an
`i1 false` based on high-level domain-specific information. If a user
provides a simplification callback we will not look at the IR but
instead give up if the callback returns a nullptr.
Hsiangkai Wang [Tue, 29 Jun 2021 07:22:18 +0000 (15:22 +0800)]
[RISCV] Add FrameSetup/FrameDestroy flag to prologue/epilog instructions.
Differential Revision: https://reviews.llvm.org/D105086
Jacques Pienaar [Fri, 23 Jul 2021 03:04:40 +0000 (20:04 -0700)]
[mlir] Fix comment typos
Tom Stellard [Fri, 23 Jul 2021 02:58:05 +0000 (19:58 -0700)]
cmake: Remove unused property on some targets: LLVM_LINK_LIBS
This doesn't appear to be used anywhere.
Reviewed By: serge-sans-paille
Differential Revision: https://reviews.llvm.org/D100021
Leonard Grey [Fri, 23 Jul 2021 02:47:22 +0000 (22:47 -0400)]
[lld-macho] Disambiguate bitcode files with the same name by archive name/offset in archive
Ported from COFF/ELF; test is adapted from
test/COFF/thinlto-archivecollision.ll
LTO expects every bitcode file to have a unique name. If given multiple bitcode
files with the same name, it errors with "Expected at most one ThinLTO module
per bitcode file".
This change incorporates the archive name, to disambiguate members with the
same name in different archives and the offset in archive to disambiguate
members with the same name in the same archive.
Differential Revision: https://reviews.llvm.org/D106179
Joseph Huber [Fri, 23 Jul 2021 01:50:09 +0000 (21:50 -0400)]
[OpenMP][NFC] Fix formatting in CUDA plugin
Joseph Huber [Fri, 23 Jul 2021 01:19:14 +0000 (21:19 -0400)]
[OpenMP] Add environment variables to change stack / heap size in the CUDA plugin
This patch adds support for two environment variables to configure the device.
``LIBOMPTARGET_STACK_SIZE`` sets the amount of memory in bytes that each thread
has for its stack. ``LIBOMPTARGET_HEAP_SIZE`` sets the amount of heap memory
that can be allocated using malloc / free on the device.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D106627
Nico Weber [Fri, 23 Jul 2021 01:38:00 +0000 (21:38 -0400)]
[gn build] Allow use_asan=true on macOS
Seems to work.
(I only tried macOS, not iOS, but need to allow both because
the iOS toolchain used to build compiler-rt asserts otherwise.)
Nico Weber [Fri, 23 Jul 2021 01:35:35 +0000 (21:35 -0400)]
[gn build] Reformat all gn files
Ran `git ls-files '*.gn' '*.gni' | xargs llvm/utils/gn/gn.py format`.
Nico Weber [Fri, 23 Jul 2021 01:31:32 +0000 (21:31 -0400)]
[lld/mac] Remove "else" after return
No behavior change
Giorgis Georgakoudis [Fri, 23 Jul 2021 01:20:14 +0000 (18:20 -0700)]
[Attributor][Fix] Add overrides for AA2HS analysis
Kai Luo [Fri, 23 Jul 2021 00:57:53 +0000 (00:57 +0000)]
[PowerPC] Implement XL compatible behavior of __compare_and_swap
According to https://www.ibm.com/docs/en/xl-c-and-cpp-aix/16.1?topic=functions-compare-swap-compare-swaplp
XL's `__compare_and_swap` has a weird behavior that
> In either case, the contents of the memory location specified by addr are copied into the memory location specified by old_val_addr.
(unlike c11 `atomic_compare_exchange` specified in http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1548.pdf)
This patch let clang's implementation follow this behavior.
Reviewed By: jsji
Differential Revision: https://reviews.llvm.org/D106344
Giorgis Georgakoudis [Thu, 22 Jul 2021 01:36:17 +0000 (18:36 -0700)]
[OpenMP] Use AAHeapToStack/AAHeapToShared analysis in SPMDization
SPMDization D102307 detects incompatible OpenMP runtime calls to abort converting a target region to SPMD mode. Calls to memory allocation/de-allocation routines kmpc_alloc_shared, kmpc_free_shared are incompatible unless they are removed by AAHeapToStack/AAHeapToShared analysis. This patch extends SPMDization detection to include AAHeapToStack/AAHeapToShared analysis results for enlarging the scope of possible SPMDized regions detected.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D105634
Vitaly Buka [Fri, 23 Jul 2021 00:36:32 +0000 (17:36 -0700)]
[compiler-rt] Fix lld dependency for sanitizers
TARGET lld is always false there.
Vitaly Buka [Fri, 23 Jul 2021 00:07:44 +0000 (17:07 -0700)]
[NFC][asan] Always pass Dominator Trees into forAllReachableExits
Rob Suderman [Fri, 23 Jul 2021 00:22:11 +0000 (17:22 -0700)]
[mlir][tosa] Make tosa MakeBroadcastable pass handle unreanked tensors.
If this pass executes without shape inference its possible for unranked tensors
to appear in the IR. This pass should gracefully handle unranked tensors.
Differential Revision: https://reviews.llvm.org/D106617
Thomas Johnson [Thu, 22 Jul 2021 23:26:46 +0000 (16:26 -0700)]
[ARC] Add tablegen definition for the Find Leading Set (FLS) instruction
Differential Revision: https://reviews.llvm.org/D106602
Gulfem Savrun Yeniceri [Thu, 6 May 2021 16:09:12 +0000 (16:09 +0000)]
[profile] Add binary id into profiles
This patch adds binary id into profiles to easily associate binaries
with the corresponding profiles. There is an RFC that discusses
the motivation, design and implementation in more detail:
https://lists.llvm.org/pipermail/llvm-dev/2021-June/151154.html
Differential Revision: https://reviews.llvm.org/D102039
Fangrui Song [Fri, 23 Jul 2021 00:09:23 +0000 (17:09 -0700)]
[ELF] --gc-sections: allow GC on reserved sections in a group
This generalizes D70146 (SHT_NOTE) to more reserved sections and makes our rules
more consistent. Now SHF_GROUP is more similar to SHF_LINK_ORDER.
For SHT_INIT_ARRAY/SHT_FINI_ARRAY, the rule will be closer to PE/COFF link.exe.
Previously sanitizers use llvm.global_ctors to make module_ctor a GC
root, which is considered an abuse.
https://groups.google.com/g/generic-abi/c/TpleUEkNoQI
We can squeak through on compatibility issues because compilers otherwise don't
use SHF_GROUP special sections.
Fangrui Song [Fri, 23 Jul 2021 00:04:54 +0000 (17:04 -0700)]
[ELF][test] Add a test about GCable SHF_LINK_ORDER SHT_INIT_ARRAY
Hongtao Yu [Thu, 22 Jul 2021 21:41:55 +0000 (14:41 -0700)]
[CSSPGO] Fix a typo in SampleContextTracker
Fixing a typo in SampleContextTracker to use debug name when debug linkage name is no present. This should only affect C programs.
Saw 0.6% perf win on Cinder which is mostly C code.
Reviewed By: wenlei, wmi
Differential Revision: https://reviews.llvm.org/D106599
Nico Weber [Thu, 22 Jul 2021 15:20:36 +0000 (11:20 -0400)]
[lld/mac] Handle symbols from -U in treatUndefinedSymbol()
In ld64, `-U section$start$FOO$bar` handles `section$start$FOO$bar`
as a regular `section$start` symbol, that is section$start processing
happens before -U processing.
Likely, nobody uses that in practice so it doesn't seem very important
to be compatible with this, but it also moves the -U handling code next
to the `-undefined dynamic_lookup` handling code, which is nice because
they do the same thing. And, in fact, this did identify a bug in a corner
case in the intersection of `-undefined dynamic_lookup` and dead-stripping
(fix for that in D106565).
Vaguely related to PR50760.
No interesting behavior change.
Differential Revision: https://reviews.llvm.org/D106566
Nico Weber [Thu, 22 Jul 2021 23:38:19 +0000 (19:38 -0400)]
[gn build] (manually) port
f8c6515554cc (libLLVMDWP)
Mara Sophie Grosch [Thu, 22 Jul 2021 23:33:20 +0000 (16:33 -0700)]
Add llvm-readobj and binutils symlinks to LLVM_TOOLCHAIN_TOOLS
This patch adds llvm-readobj and the binutils symlink for readelf to
LLVM_TOOLCHAIN_TOOLS.
Tvoid *thread, void *attr,hey are required by some (most?)
autoconf-built libraries, adding these allows me to build newlib with
the toolchain generated this way.
Also opened an issue for that some days ago, see
https://bugs.llvm.org/show_bug.cgi?id=50698
Reviewed By: sbc100
Differential Revision: https://reviews.llvm.org/D104957
Thomas Lively [Thu, 22 Jul 2021 23:31:12 +0000 (16:31 -0700)]
[WebAssembly][NFC] Update test expectations labels after
db7efcab7dd9
Commit
db7efcab7dd9 changed the implementations of the wasm_*_extract_lane and
wasm_*_replace_lane intrinsics from using builtin functions to using the
standard vector extensions. This did not change the resulting IR, but it changes
how update_cc_test_checks.py labels values in the IR. This commit simply updates
those labels.
Differential Revision: https://reviews.llvm.org/D106611
Nico Weber [Thu, 22 Jul 2021 15:37:29 +0000 (11:37 -0400)]
[lld/mac] Fix bug in interaction of -dead_strip and -undefined dynamic_lookup
We lost the `used` bit on the Undefined when we replaced it with a DylibSymbol
in treatUndefined().
Differential Revision: https://reviews.llvm.org/D106565
Christopher Di Bella [Tue, 29 Jun 2021 03:53:44 +0000 (03:53 +0000)]
[libcxx][modularisation] properly modularises advance, next, and prev
`__function_like` wasn't being exported, so certain properties of the
`ranges` functions weren't being propagated in modules land.
Differential Revision: https://reviews.llvm.org/D105078
Mircea Trofin [Thu, 22 Jul 2021 23:26:46 +0000 (16:26 -0700)]
[MLGO] Strip TF_PIP cmake variable
This should fix build breaks for 'development' mode. The other modes
were unaffected - 'release' because it doesn't use TFUtils.cpp, and the
mixed mode because the AOT compiled code brings in the necessary include
dirs anyway.
Shilei Tian [Thu, 22 Jul 2021 23:20:57 +0000 (19:20 -0400)]
[OpenMP] Refined the logic to give a regular task from a hidden helper task
In current implementation, if a regular task depends on a hidden helper task,
and when the hidden helper task is releasing its dependences, it directly calls
`__kmp_omp_task`. This could cause a problem that if `__kmp_push_task` returns
`TASK_NOT_PUSHED`, the task will be executed immediately. However, the hidden
helper threads are assumed to only execute hidden helper tasks. This could cause
problems because when calling `__kmp_omp_task`, the encountering gtid, which is
not the real one of the thread, is passed.
This patch uses `__kmp_give_task`, but because it is a static function, a new
wrapper `__kmpc_give_task` is added.
Reviewed By: AndreyChurbanov
Differential Revision: https://reviews.llvm.org/D106572
Florian Mayer [Thu, 22 Jul 2021 21:19:21 +0000 (14:19 -0700)]
[hwasan] Use stack safety analysis.
This avoids unnecessary instrumentation.
Reviewed By: eugenis, vitalybuka
Differential Revision: https://reviews.llvm.org/D105703
Geoffrey Martin-Noble [Thu, 22 Jul 2021 23:05:40 +0000 (16:05 -0700)]
[Bazel] Rename LLVMDWP to DWP
Since Bazel has packages, we don't prefix target names with LLVM. I
forgot when translating the CMake.
Mircea Trofin [Wed, 28 Apr 2021 16:09:01 +0000 (09:09 -0700)]
[docs] Add the compiler-rt requirement to the test suite doc
Differential Revision: https://reviews.llvm.org/D101467
Michael McLoughlin [Thu, 22 Jul 2021 23:01:49 +0000 (16:01 -0700)]
[Bazel] Change external_zlib attribute to string
When using `llvm_zlib_external` rule with `external_zlib` attribute set to a
label referring to the main repository, like `@//third_party/zlib`, it will be
replaced with `//third_party/zlib` after template substitution. This will then
attempt to find `//third_party/zlib` within the local repository
`@llvm_zlib//third_party/zlib`, which does not exist, rather than the intended
reference back to the main repository. The issue appears to be that the
conversion of `Label` type to string with
`str(repository_ctx.attr.external_zlib)`, which is causing the main repository
qualifier to be lost.
This diff fixes the issue by changing the `external_zlib` attribute to
`attr.string` type rather than `attr.label`.
In future a more elegant solution may be possible that preserves use of the
`Label` type, depending on resolution of the issue
https://github.com/bazelbuild/bazel/issues/13731.
Ported from Github PR https://github.com/google/llvm-bazel/pull/236.
Reviewed By: GMNGeoffrey
Differential Revision: https://reviews.llvm.org/D106606
Rob Suderman [Tue, 20 Jul 2021 21:41:37 +0000 (14:41 -0700)]
[mlir][tosa] Quantized Conv2DOp lowering to linalg added.
Includes a version of a quantized conv2D operations with a lowering from TOSA
to linalg with corresponding test. We keep the quantized and quantized variants
as separate named ops to avoid the additional operations for non-quantized
convolutions.
Differential Revision: https://reviews.llvm.org/D106407
George Balatsouras [Fri, 16 Jul 2021 21:53:32 +0000 (14:53 -0700)]
[dfsan] Add wrappers for v*printf functions
Functions `vsnprintf`, `vsprintf` and `vfprintf` commonly occur in DFSan warnings.
Reviewed By: stephan.yichao.zhao
Differential Revision: https://reviews.llvm.org/D106195
Jose M Monsalve Diaz [Thu, 22 Jul 2021 22:17:09 +0000 (18:17 -0400)]
[OpenMP] Renaming RT functions `GetNumberOfBlocksInKernel` and `GetNumberOfThreadsInBlock`
These functions should follow the camel case convention. These are really easy to change
and are needed for D106033.
Reviewed By: JonChesterfield
Differential Revision: https://reviews.llvm.org/D106390
Jim Ingham [Thu, 22 Jul 2021 19:03:12 +0000 (12:03 -0700)]
Fix the logic so stop-hooks get run after a breakpoint that ran an expression
Code was added to Target::RunStopHook to make sure that we don't run stop hooks when
you stop after an expression evaluation. But the way it was done was to check that we
hadn't run an expression since the last natural stop. That failed in the case where you
stopped for a breakpoint which had run an expression, because the stop-hooks get run
after the breakpoint actions, and so by the time we got to running the stop-hooks,
we had already run a user expression.
I fixed this by adding a target ivar tracking the last natural stop ID at which we had
run a stop-hook. Then we keep track of this and make sure we run the stop-hooks only
once per natural stop.
Differential Revision: https://reviews.llvm.org/D106514
Eli Friedman [Thu, 22 Jul 2021 22:03:05 +0000 (15:03 -0700)]
[AArch64] Regenerate test arm64-ccmp.ll
Geoffrey Martin-Noble [Thu, 22 Jul 2021 21:51:04 +0000 (14:51 -0700)]
[Bazel] Update for
f8c6515554
Update Bazel for
https://github.com/llvm/llvm-project/commit/
f8c6515554 by splitting out
an LLVMDWP library target.
Amy Huang [Thu, 22 Jul 2021 18:26:03 +0000 (11:26 -0700)]
[DebugInfo] Add -fno-ctor-homing for as counterpart to -fuse-ctor-homing
Add an opt out flag for constructor homing.
Differential Revision: https://reviews.llvm.org/D106582
David Blaikie [Thu, 22 Jul 2021 21:23:32 +0000 (14:23 -0700)]
PR51158: Don't emit -Wswitch or -Wcovered-switch-default for empty enums
An empty enum is used to implement C++'s new-ish "byte" type (to make
sure it's a separate type for overloading, etc - compared to a typedef)
- without any enumerators. Some clang warnings don't make sense in this
sort of situation, so let's skip them for empty enums.
It's arguable that possibly some situations of enumerations without
enumerators might want the previous-to-this-patch behavior (if the enum
is autogenerated and in some cases comes up empty, then maybe a default
in an empty switch would still be considered problematic - so that when
you add the first enumeration you do get a -Wswitch warning). But I
think that's niche enough & this std::byte case is mainstream enough
that we should prioritize the latter over the former.
If someone's got a middle ground proposal to account for both of those
situations, I'm open to patches/suggestions/etc.
Jianzhou Zhao [Thu, 22 Jul 2021 21:00:45 +0000 (21:00 +0000)]
[dfsan] Make warn_unimplemented off by default
Because almost all internal use cases need to turn warn_unimplemented off.
Roman Lebedev [Thu, 22 Jul 2021 21:13:46 +0000 (00:13 +0300)]
[SimplifyCFG] SimplifyCondBranchToTwoReturns(): really only deal with different ret blocks
This function is called when some predecessor of an empty return block
ends with a conditional branch, with both successors being empty ret blocks.
Now, because of the way SimplifyCFG works, it might happen to simplify
one of the blocks in a way that makes a conditional branch
into an unconditional one, since it's destinations are now identical,
but it might not have actually simplified said conditional branch
into an unconditional one yet.
So, we have to check that ourselves first,
especially now that SimplifyCFG aggressively tail-merges
all ret and resume blocks.
Even if it was an unconditional branch already,
`SimplifyCFGOpt::simplifyReturn()` doesn't call `FoldReturnIntoUncondBranch()`
by default.
Roman Lebedev [Thu, 22 Jul 2021 21:28:43 +0000 (00:28 +0300)]
[NFC][LoopDeletion] Autogenerate checlines in simplify-then-delete.ll test
Roman Lebedev [Thu, 22 Jul 2021 21:09:44 +0000 (00:09 +0300)]
[NFC][SimplifyCFG] Add test for SimplifyCondBranchToTwoReturns() mishandling
Alexander Yermolovich [Thu, 22 Jul 2021 21:11:49 +0000 (14:11 -0700)]
[DWP] Refactoring llvm-dwp in to a library part 2
This is follow up to https://reviews.llvm.org/D106198 where llvm-dwp was refactored in to multiple files.
In this patch moving them in to lib/include directories.
Reviewed By: dblaikie
Differential Revision: https://reviews.llvm.org/D106493
Kirill Stoimenov [Wed, 21 Jul 2021 16:20:35 +0000 (16:20 +0000)]
[asan] Modified ASAN_MEMORY_ACCESS_CALLBACK to use a function call to ReportGenericErrorWrapper.
This change eliminate the stack frame for the fast path and improves runtime performance.
Reviewed By: vitalybuka
Differential Revision: https://reviews.llvm.org/D106505
Shu-Chun Weng [Thu, 22 Jul 2021 21:17:58 +0000 (14:17 -0700)]
[NFC] Fix test build breakage on Darwin
Nick Fitzgerald [Thu, 22 Jul 2021 21:12:50 +0000 (14:12 -0700)]
Reland: "[WebAssembly] Deduplicate imports of the same module name, field name, and type"
When two symbols import the same thing, only one import should be
emitted in the Wasm file.
Fixes https://bugs.llvm.org/show_bug.cgi?id=50938
Reverted in:
16aac493e59519377071e900d119ba2e7e5b525d.
Reviewed By: sbc100
Differential Revision: https://reviews.llvm.org/D105519
Jacques Pienaar [Thu, 22 Jul 2021 20:57:36 +0000 (13:57 -0700)]
[mlir] Also update inferReturnTensorTypes (NFC)
Missed this one in the first go.
Med Ismail Bennani [Thu, 22 Jul 2021 20:54:27 +0000 (20:54 +0000)]
[lldb] Fix build failure introduced by
3d4cadfb26437bd686ca8177f5454a366fed59eb
This patch updates the `ScriptedProcess::GetGenericInteger` return type
to `llvm::Optional<unsigned long long>` to match implementation.
Differential Revision: https://reviews.llvm.org/D105788
Signed-off-by: Med Ismail Bennani <medismail.bennani@gmail.com>
Med Ismail Bennani [Thu, 22 Jul 2021 20:47:25 +0000 (20:47 +0000)]
[lldb/Interpreter] Conform ScriptedProcessPythonInterface to SWIG python types
This patch should address the compiler warnings due to mismatch type
comparaison.
Differential Revision: https://reviews.llvm.org/D105788
Signed-off-by: Med Ismail Bennani <medismail.bennani@gmail.com>
Martin Storsjö [Sun, 18 Jul 2021 19:52:32 +0000 (22:52 +0300)]
[LLD] [COFF] Make -export-all-symbols work as intended for EXEs
If some symbols are marked with dllexport, we still want to export
all symbols if -export-all-symbols is specified. Previously, this
only worked as it should for DLL output, not for EXE.
This should fix downstream bug
https://github.com/msys2/MINGW-packages/issues/9163.
Differential Revision: https://reviews.llvm.org/D106245
Mircea Trofin [Thu, 22 Jul 2021 20:24:28 +0000 (13:24 -0700)]
[MLGO] Correct protobuf path
Felix Berger [Fri, 16 Jul 2021 18:38:47 +0000 (14:38 -0400)]
[clang-tidy] performance-unnecessary-copy-initialization: Create option to exclude container types from triggering the check.
Add string list option of type names analagous to `AllowedTypes` which lets
users specify a list of ExcludedContainerTypes.
Types matching this list will not trigger the check when an expensive variable
is copy initialized from a const accessor method they provide, i.e.:
```
ExcludedContainerTypes = 'ExcludedType'
void foo() {
ExcludedType<ExpensiveToCopy> Container;
const ExpensiveToCopy NecessaryCopy = Container.get();
}
```
Even though an expensive to copy variable is copy initialized the check does not
trigger because the container type is excluded.
This is useful for container types that don't own their data, such as view types
where modification of the returned references in other places cannot be reliably
tracked, or const incorrect types.
Differential Revision: https://reviews.llvm.org/D106173
Reviewed-by: ymandel
Paulo Matos [Thu, 22 Jul 2021 19:11:39 +0000 (21:11 +0200)]
[WebAssembly] Implementation of global.get/set for reftypes in LLVM IR
Reland of
31859f896.
This change implements new DAG notes GLOBAL_GET/GLOBAL_SET, and
lowering methods for load and stores of reference types from IR
globals. Once the lowering creates the new nodes, tablegen pattern
matches those and converts them to Wasm global.get/set.
Reviewed By: tlively
Differential Revision: https://reviews.llvm.org/D104797
Mircea Trofin [Thu, 22 Jul 2021 19:47:49 +0000 (12:47 -0700)]
[NFC][MLGO] Fix vector sizing
The bots only build release mode, and the use of `reserve` instead of
`resize`, while not causing invalid memory accesses, is incorrect.
Jake Egan [Thu, 22 Jul 2021 19:44:26 +0000 (15:44 -0400)]
[AIX] Define __LONGDOUBLE64 macro
This patch defines the macro __LONGDOUBLE64 for AIX when long double is 8 bytes.
Reviewed By: cebowleratibm
Differential Revision: https://reviews.llvm.org/D105477
Roman Lebedev [Thu, 22 Jul 2021 19:59:22 +0000 (22:59 +0300)]
[NFCI][TLI] prepare[US]REMEqFold(): don't add nonsensical 'exact' flag to rotates created
As pointed out by Craig Topper.
Eric Astor [Thu, 22 Jul 2021 19:50:37 +0000 (15:50 -0400)]
[ms] [llvm-ml] Fix macro case-insensitivity
We previously had issues identifying macros not registered with a lowercase name.
Reviewed By: mstorsjo, thakis
Differential Revision: https://reviews.llvm.org/D106453
Anjan Kumar Guttahalli Krishna [Thu, 22 Jul 2021 04:20:11 +0000 (00:20 -0400)]
[AIX] Generate large code model relocations when mcmodel=medium on AIX
This patch makes the changes in the driver that converts the medium code
model to large.
Reviewed By: hubert.reinterpretcast
Differential Revision: https://reviews.llvm.org/D106371
Shu-Chun Weng [Wed, 21 Jul 2021 17:59:09 +0000 (10:59 -0700)]
Fix TSAN signal interceptor out-of-bound access
signal(2) and sigaction(2) have defined behaviors for invalid signal number
(EINVAL) and some programs rely on it.
The added test case also reveals that MSAN is too strict in this regard.
Test case passed on x86_64 Linux and AArch64 Linux.
Reviewed By: vitalybuka
Differential Revision: https://reviews.llvm.org/D106468
Nikita Popov [Thu, 22 Jul 2021 19:06:39 +0000 (21:06 +0200)]
[LICM][SCCP] Regenerate test checks (NFC)
Anjan Kumar Guttahalli Krishna [Thu, 22 Jul 2021 19:32:06 +0000 (15:32 -0400)]
[AIX] Clang's library integration support for 128-bit long double is incomplete on AIX.
Emit the unsupported option error until the Clang's library integration support for 128-bit long double is available for AIX.
Reviewed By: Whitney, cebowleratibm
Differential Revision: https://reviews.llvm.org/D106074
Jacques Pienaar [Thu, 22 Jul 2021 19:24:48 +0000 (12:24 -0700)]
[mlir] Update to use ValueShapeRange (NFC)
Update to use alias in preparation for changing it to not just be a pure alias.
Jon Chesterfield [Thu, 22 Jul 2021 19:22:49 +0000 (20:22 +0100)]
[libomptarget][amdgpu][nfc] Normalise license headers
Reviewed By: gregrodgers, jdoerfert
Differential Revision: https://reviews.llvm.org/D106581
Roman Lebedev [Thu, 22 Jul 2021 19:11:46 +0000 (22:11 +0300)]
[SimplifyCFG] FoldTwoEntryPHINode(): bailout on inverted logical and/or (PR51149)
The logical (select) form of and/or will now be a source of problems.
We don't really account for it's inverted form, yet it exists,
and presumably we should treat it just like non-inverted form:
https://alive2.llvm.org/ce/z/BU9AXk
https://bugs.llvm.org/show_bug.cgi?id=51149 reports a reportedly-serious
perf regression that will hopefully be mitigated by this.
Roman Lebedev [Thu, 22 Jul 2021 19:07:23 +0000 (22:07 +0300)]
[NFC][SimplifyCFG] Add some more tests w/ two-entry PHI nodes and
Felix Berger [Wed, 14 Jul 2021 20:11:55 +0000 (16:11 -0400)]
[clang-tidy] performance-unnecessary-copy-initialization: Disable check when variable and initializer have different replaced template param types.
This can happen when a template with two parameter types is instantiated with a
single type. The fix would only be valid for this instantiation but fail for
others that rely on an implicit type conversion.
The test cases illustrate when the check should trigger and when not.
Differential Revision: https://reviews.llvm.org/D106011
Jon Chesterfield [Thu, 22 Jul 2021 19:16:53 +0000 (20:16 +0100)]
[nfc] Fix typo in comment, s/node/note
Simon Pilgrim [Thu, 22 Jul 2021 19:07:18 +0000 (20:07 +0100)]
[CostModel][X86] Adjust shift SSE4 legalized costs based on llvm-mca reports.
Update shl/lshr/ashr costs based on the worst case costs from the script in D103695 - many of the 128-bit shifts (usually where integer multiplies aren't used) have similar behaviour to AVX1 so we can merge them.
Simon Pilgrim [Thu, 22 Jul 2021 19:00:05 +0000 (20:00 +0100)]
[CostModel][X86] Fix funnel shift check prefixes
We'd lost AVX1 test coverage due to bulldozer (XOP) trying to use the same check prefixes - we really need to fix the update script to avoid this!
Jon Chesterfield [Thu, 22 Jul 2021 19:04:13 +0000 (20:04 +0100)]
[libomptarget][amdgpu][nfc] Replace use of gelf.h with libelf.h
AMDGPU can assume Elf64 so doesn't need to abstract over Elf32
Drop a few other unused headers at the same time. Now only llvm elf
and libelf are used by the plugin.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D106579
Aaron Ballman [Thu, 22 Jul 2021 18:49:21 +0000 (14:49 -0400)]
Correctly diagnose taking the address of a register variable in C
We caught the cases where the user would explicitly use the & operator,
but we were missing implicit conversions such as array decay.
Fixes PR26336. Thanks to Samuel Neves for inspiration for the patch.
LLVM GN Syncbot [Thu, 22 Jul 2021 18:41:45 +0000 (18:41 +0000)]
[gn build] Port
3959c95deb11
Simon Pilgrim [Thu, 22 Jul 2021 18:26:07 +0000 (19:26 +0100)]
[X86] Fix SLM FP<->INT throughputs.
Noticed while trying to clean up the shift costs model for SSE4 targets using the script in D10369 - SLM double-pumps all the 128-bit vector conversion ops and only use FP0 pipe - numbers taken from Intel AOM + Agner.
Thomas Johnson [Thu, 22 Jul 2021 18:15:24 +0000 (11:15 -0700)]
[ARC] Add disassembly for the conditioned RSUB immediate instruction
Differential Revision: https://reviews.llvm.org/D106497
Fangrui Song [Thu, 22 Jul 2021 18:33:01 +0000 (11:33 -0700)]
[Matrix] Fix -Wunused-variable in -DLLVM_ENABLE_ASSERTIONS=off build after D106457. NFC
Louis Dionne [Mon, 19 Jul 2021 16:12:49 +0000 (12:12 -0400)]
[libc++] Add helper type non-propagating-cache
Differential Revision: https://reviews.llvm.org/D102121
zoecarver [Thu, 22 Jul 2021 18:15:24 +0000 (11:15 -0700)]
[libc++][docs] Take lock for range.single.view.
Mark this item as in progress and assigned to me.
Alex Lorenz [Thu, 22 Jul 2021 18:12:38 +0000 (11:12 -0700)]
[clang][test] Add -fuse-ld= to test case added in
2542c1a5a130 to resolve test failure with CLANG_DEFAULT_LINKER=lld
Adam Nemet [Wed, 21 Jul 2021 16:46:47 +0000 (09:46 -0700)]
[Matrix] Fix miscompile for NT matmul if the transpose has other use
We should only add the fake lowering entry for the matrix remark if the
transpose is not lowered on its own. `MapVector::insert` is used to insert
the entry during proper lowering which does not overwrite the fake entry in
the map.
We actually had test coverage for this but the reference output code was
wrong; it was storing undef rather than the transposed column.
Also add an assert that would have caught this.
Differential Revision: https://reviews.llvm.org/D106457
Marius Brehler [Thu, 22 Jul 2021 12:06:52 +0000 (12:06 +0000)]
[mlir] Improve description of interface options
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D106539
Krishna Kariya [Thu, 22 Jul 2021 17:18:57 +0000 (13:18 -0400)]
[InstCombine][test] add coverage for possible fabs folds; NFC
This goes with D101727 (adds FMF to the select).
Differential Revision: https://reviews.llvm.org/D106563
Alex Lorenz [Tue, 13 Jul 2021 19:33:04 +0000 (12:33 -0700)]
[clang][driver][darwin] Add driver support for Mac Catalyst
This commit adds driver support for the Mac Catalyst target,
as supported by the Apple clang compile
Differential Revision: https://reviews.llvm.org/D105960
David Green [Thu, 22 Jul 2021 17:19:54 +0000 (18:19 +0100)]
[AArch64] Adjust the cost of integer sum reductions
This changes the cost to (LT.first-1) * cost(add) + 2, where the cost of
an add is assumed to be 1. This brings it inline with the other
reductions.
Differential Revision: https://reviews.llvm.org/D106240
Simon Pilgrim [Thu, 22 Jul 2021 16:40:44 +0000 (17:40 +0100)]
[CostModel][X86] Adjust shift SSE legalized costs based on llvm-mca reports.
Update shl/lshr/ashr costs based on the worst case costs from the script in D103695.
Shilei Tian [Thu, 22 Jul 2021 16:37:43 +0000 (12:37 -0400)]
[OpenMPOpt] Add support for BooleanStateWithSetVector
D101977 added `BooleanStateWithPtrSetVector` to store pointers to a set meanwhile
tracking boolean state. One of the limitation is that it can only store pointer.
We might want it to store other types of values, such as integer for parallel
level. This patch generalizes the idea and create `BooleanStateWithSetVector`.
`BooleanStateWithPtrSetVector` therefore becomes a type alias of `BooleanStateWithSetVector`.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D106149
Fangrui Song [Thu, 22 Jul 2021 16:37:01 +0000 (09:37 -0700)]
[test] Add llvm-stress to LLVM_TEST_DEPENDS and lit substitutions
D106430 added a test which needs LLVM_TEST_DEPENDS and lit substitution.
Rahul Joshi [Thu, 22 Jul 2021 00:13:40 +0000 (17:13 -0700)]
[MLIR][memref] Fix findDealloc() to handle > 1 dealloc for the given alloc.
- Change findDealloc() to return Optional<Operation *> and return None if > 1
dealloc is associated with the given alloc.
- Add findDeallocs() to return all deallocs associated with the given alloc.
- Fix current uses of findDealloc() to bail out if > 1 dealloc is found.
Differential Revision: https://reviews.llvm.org/D106456
Jon Chesterfield [Thu, 22 Jul 2021 15:54:10 +0000 (16:54 +0100)]
[libomptarget][amdgpu] Implement dlopen of libhsa
AMDGPU plugin equivalent of D95155, build without HSA installed locally
Compiles a new file, plugins/amdgpu/dynamic_hsa/hsa.cpp, to an object file that
exposes the same symbols that the plugin presently uses from hsa. The object
file contains dlopen of hsa and cached dlsym calls. Also provides header files
corresponding to the subset that is used.
This is behind a feature flag, LIBOMPTARGET_FORCE_DLOPEN_LIBHSA, default off.
That allows developers to build against the dlopen/dlsym implementation, e.g.
while testing this mode.
Enabling by default will cause this plugin to build on a wider variety of
machines than it does at present so may break some CI builds. That risk can
be minimised by reviewing the header dependencies of the library and ensuring
it doesn't use any libraries that are not already used by libomptarget.
Separating the implementation from enabling by default in case the latter needs
to be rolled back after wider CI results.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D106559
Victor Huang [Thu, 22 Jul 2021 15:05:30 +0000 (10:05 -0500)]
[PowerPC] Add PowerPC "__stbcx" builtin and intrinsic for XL compatibility
This patch is in a series of patches to provide builtins for compatibility
with the XL compiler. This patch adds the builtin and intrinsic for "__stbcx".
Reviewed By: nemanjai, #powerpc
Differential revision: https://reviews.llvm.org/D106484
Anastasia Stulova [Thu, 22 Jul 2021 15:43:18 +0000 (16:43 +0100)]
[OpenCL][NFC] Refactors lang version check in test.
Fixed test to use predefined version marco instead
of passing extra macro in the command line.
Patch by Topotuna (Justas Janickas)!
Differential Revision: https://reviews.llvm.org/D106254
Alexey Bataev [Thu, 22 Jul 2021 15:11:49 +0000 (08:11 -0700)]
[OPENMP]Fix PR49787: Codegen for calling __tgt_target_teams_nowait_mapper has too few arguments.
Added missed arguments in
__tgt_target_teams_nowait_mapper/__tgt_target_nowait_mapper runtime
functions calls.
Differential Revision: https://reviews.llvm.org/D106542
Nico Weber [Thu, 22 Jul 2021 14:31:39 +0000 (10:31 -0400)]
[lld/mac] Move handling of special undefineds later
treatUndefinedSymbol() was previously called before gatherInputSections()
and markLive() for these special symbols, but after them for normal
undefineds.
For PR50760, treatUndefinedSymbol() will have to potentially create
sections, so it's good to move treatUndefinedSymbol() for special
undefineds later, so that it can assume that gatherInputSections()
and markLive() has already been called always.
No intended behavior change, but part of PR50760 (and covered in
tests in the patch for the full feature).
Differential Revision: https://reviews.llvm.org/D106552
Alexey Bataev [Thu, 22 Jul 2021 15:06:03 +0000 (08:06 -0700)]
Revert "[OPENMP]Fix PR49787: Codegen for calling __tgt_target_teams_nowait_mapper has too few arguments."
This reverts commit
b455f7f22564a096c043b02fa159ab16669c121c to fix
buildbots.