Jason Molenda [Thu, 24 Mar 2022 21:35:46 +0000 (14:35 -0700)]
Don't search for sim SDK path until we know we need it
When iterating over all Platforms looking for the best one, on a Mac the
Simulator platforms (iOS, tvOS, watchOS) will first find their SDK
directory by calling xcrun, then decide if they should activate or not.
When that SDK is absent, the call to xcrun to find it can be very slow.
This patch delays that directory search until we know we're activating
this platform, so non-simulator environments don't pay a perf cost ever
time they go through the list of platforms.
Differential Revision: https://reviews.llvm.org/D122373
rdar://
87960090
Fangrui Song [Thu, 24 Mar 2022 22:26:40 +0000 (15:26 -0700)]
[Bazel] Add BLAKE3 files to llvm:Support
Hope someone can figure out how to leverage SSE42/AVE2/Arm Neon.
Peter Steinfeld [Wed, 23 Mar 2022 22:20:56 +0000 (15:20 -0700)]
[flang] Make not yet implemented messages more consistent
To make it easier to find things that are not yet implemented, I'm changing the
messages that appear in the compiler's output to all have the string "not yet
implemented:".
These changes apply to files in the front end. I have another set of changes
to files in the lowering code.
Differential Revision: https://reviews.llvm.org/D122355
Zixu Wang [Thu, 24 Mar 2022 22:09:25 +0000 (15:09 -0700)]
[NFC][clang][extract-api] Rename variable
Rename a local variable name to avoid potential ambiguity/conflict for
some compilers.
Tue Ly [Wed, 23 Mar 2022 19:37:19 +0000 (15:37 -0400)]
[libc] Improve the performance of exp2f.
Reduce the range-reduction table size from 128 entries down to 64 entries, and
reduce the polynomial's degree from 6 down to 4.
Currently we use a degree-6 minimax polynomial on an interval of length 2^-7
around 0 to compute exp2f. Based on the suggestion of @santoshn and the RLIBM
project (https://github.com/rutgers-apl/rlibm-prog/blob/main/libm/float/exp2.c)
it is possible to have a good polynomial of degree-4 on a subinterval of length
2^(-6) to approximate 2^x.
We did try to either reduce the degree of the polynomial down to 3 or increase
the interval size to 2^(-5), but in both cases the number of exceptional values
exploded. So we settle with using a degree-4 polynomial of the interval of
size 2^(-6) around 0.
Reviewed By: michaelrj, sivachandra, zimmermann6, santoshn
Differential Revision: https://reviews.llvm.org/D122346
Zixu Wang [Thu, 24 Mar 2022 21:30:14 +0000 (14:30 -0700)]
[NFC][clang][extract-api] Add missing virtual anchors
Add missing virtual method anchors for structs in ExtractAPI/API.h
Arthur Eubanks [Thu, 10 Feb 2022 23:39:29 +0000 (15:39 -0800)]
[opt] Remove -analyze option
This is legacy PM-specific, which is deprecated.
Uses of this should be replaced with a corresponding `-passes='print<foo>'`.
Reviewed By: asbirlea
Differential Revision: https://reviews.llvm.org/D122420
Lei Zhang [Thu, 24 Mar 2022 19:00:39 +0000 (15:00 -0400)]
[mlir][spirv] Add Apple into the vendor list
Reviewed By: ThomasRaoux, hanchung
Differential Revision: https://reviews.llvm.org/D122414
Jennifer Yu [Wed, 23 Mar 2022 14:34:28 +0000 (07:34 -0700)]
Eliminate extra set of simd variant function attribute.
Current clang generates extra set of simd variant function attribute
with extra 'v' encoding.
For example:
_ZGVbN2v__Z5add_1Pf vs _ZGVbN2vv__Z5add_1Pf
The problem is due to declaration of ParamAttrs following:
llvm::SmallVector<ParamAttrTy, 8> ParamAttrs(ParamPositions.size());
where ParamPositions.size() is grown after following assignment:
Pos = ParamPositions[PVD];
So the PVD is not find in ParamPositions.
The problem is ParamPositions need to set for each FD decl. To fix this
Move ParamPositions's init inside while loop for each FD.
Differential Revision: https://reviews.llvm.org/D122338
Stanislav Mekhanoshin [Wed, 23 Mar 2022 18:29:56 +0000 (11:29 -0700)]
[AMDGPU] Use GenericTable to classify DGEMM
Since there is a table introduced for MAI instructions extend it
to use for DGEMM classification.
Differential Revision: https://reviews.llvm.org/D122337
Stanislav Mekhanoshin [Wed, 23 Mar 2022 16:12:34 +0000 (09:12 -0700)]
[AMDGPU] gfx940 MAI hazard recognizer
Differential Revision: https://reviews.llvm.org/D122263
Valentin Clement [Thu, 24 Mar 2022 19:43:11 +0000 (20:43 +0100)]
[flang][NFC] Remove unused variable
Stanislav Mekhanoshin [Mon, 21 Mar 2022 19:59:33 +0000 (12:59 -0700)]
[AMDGPU] Support gfx940 smfmac instructions
Differential Revision: https://reviews.llvm.org/D122191
Kirill Stoimenov [Wed, 16 Mar 2022 14:50:04 +0000 (14:50 +0000)]
[ASan] Reland of D116182 to always link asan_static library.
After landing D121813 the binary size increase introduced by this change can be minimized by using --gc-sections link options. D121813 allows each individual callbacks to be optimized out if not used.
Reviewed By: vitalybuka, MaskRay
Differential Revision: https://reviews.llvm.org/D122407
Nico Weber [Thu, 24 Mar 2022 19:23:36 +0000 (15:23 -0400)]
Revert "[gn build] Manually port llvm/lib/Support/BLAKE3"
This reverts commit
8424d4f641b186f30126c33f76f6a0fb619f903e.
That approach doesn't work. I checked in something that kinda
works 30 min ago or so.
wangyihan [Thu, 24 Mar 2022 19:09:10 +0000 (12:09 -0700)]
[clang][CodeGen]Fix clang crash and add bitfield support in __builtin_dump_struct
Fix clang crash and add bitfield support in __builtin_dump_struct.
In clang13.0.x, a struct with three or more members and a bitfield at
the same time will cause a crash. In clang15.x, as long as the struct
has one bitfield, it will cause a crash in clang.
Open issue: https://github.com/llvm/llvm-project/issues/54462
Differential Revision: https://reviews.llvm.org/D122248
Yaxun (Sam) Liu [Wed, 16 Mar 2022 01:45:02 +0000 (21:45 -0400)]
[CUDA][HIP] Fix hostness check with -fopenmp
CUDA/HIP determines whether a function can be called based on
the device/host attributes of callee and caller. Clang assumes the
caller is CurContext. This is correct in most cases, however, it is
not correct in OpenMP parallel region when CUDA/HIP program
is compiled with -fopenmp. This causes incorrect overloading
resolution and missed diagnostics.
To get the correct caller, clang needs to chase the parent chain
of DeclContext starting from CurContext until a function decl
or a lambda decl is reached. Sema API is adapted to achieve that
and used to determine the caller in hostness check.
Reviewed by: Artem Belevich, Richard Smith
Differential Revision: https://reviews.llvm.org/D121765
Fraser Cormack [Wed, 16 Mar 2022 10:14:07 +0000 (10:14 +0000)]
[VectorCombine] Insert addrspacecast when crossing address space boundaries
We can not bitcast pointers across different address spaces. This was
previously fixed in D89577 but then in D93229 an enhancement was added
which peeks further through the ponter operand, opening up the
possibility that address-space violations could be introduced.
Instead of bailing as the previous fix did, simply insert an
addrspacecast cast instruction.
Reviewed By: lebedev.ri
Differential Revision: https://reviews.llvm.org/D121787
Stanislav Mekhanoshin [Fri, 18 Mar 2022 19:33:21 +0000 (12:33 -0700)]
[AMDGPU] New gfx940 mfma instructions
Differential Revision: https://reviews.llvm.org/D122044
LLVM GN Syncbot [Thu, 24 Mar 2022 19:11:19 +0000 (19:11 +0000)]
[gn build] Port
Fangrui Song [Thu, 24 Mar 2022 19:06:19 +0000 (12:06 -0700)]
[gn build] Manually port llvm/lib/Support/BLAKE3
Johannes Doerfert [Thu, 24 Mar 2022 19:04:41 +0000 (14:04 -0500)]
Revert "[Intrinsics] Add `nocallback` to the default intrinsic attributes"
This reverts commit
7aea3ea8c3b33c9bb338d5d6c0e4832be1d09ac3 as it
breaks the buildbots.
I didn't see these failures in the pre-merge checks, looking into it.
Nico Weber [Thu, 24 Mar 2022 18:56:42 +0000 (14:56 -0400)]
[gn build] ugly hack to work around sync script for now
David Blaikie [Thu, 24 Mar 2022 18:31:48 +0000 (18:31 +0000)]
DebugInfo: Classify noreturn function types as non-reconstructible
This information isn't preserved in the DWARF description of function
types (though probably should be - it's preserved on the function
declarations/definitions themselves through the DW_AT_noreturn attribute
- but we should move or also include that in the subroutine type itself
too - but for now, with it not being there, the DWARF is lossy and
can't be reconstructed)
Nico Weber [Thu, 24 Mar 2022 18:44:46 +0000 (14:44 -0400)]
[gn build] (manually) port
9aa701984d39 (BLAKE3)
Johannes Doerfert [Tue, 1 Feb 2022 07:03:06 +0000 (01:03 -0600)]
[Intrinsics] Add `nocallback` to the default intrinsic attributes
Most intrinsics, especially "default" ones, will not call back into the
IR module. `nocallback` encodes this nicely. As it was not used before,
this patch also makes use of `nocallback` in the Attributor which
results in many more `norecurse` deductions.
Tablegen part is mechanical, test updates by script.
Differential Revision: https://reviews.llvm.org/D118680
Sanjay Patel [Thu, 24 Mar 2022 18:39:04 +0000 (14:39 -0400)]
[InstCombine] add more tests for nsw propagation; NFC
Follow-up as suggested with
b6efd2510a1efaf2 -
show a couple of examples of general subtraction
(the earlier patch was all negations).
Argyrios Kyrtzidis [Thu, 24 Mar 2022 18:40:28 +0000 (11:40 -0700)]
[Support/BLAKE3] Temporarily disable building neon file to get builders back to green
Fangrui Song [Thu, 24 Mar 2022 18:31:39 +0000 (11:31 -0700)]
[ELF] Implement --build-id={md5,sha1} with truncated BLAKE3
--build-id was introduced as "approximation of true uniqueness across all
binaries that might be used by overlapping sets of people". It does not require
the some resistance mentioned below. In practice, people just use --build-id=md5
for 16-byte build ID and --build-id=sha1 for 20-byte build ID.
BLAKE3 has 256-bit key length, which provides 128-bit security against
(second-)preimage, collision, and differentiability attacks. Its portable
implementation is fast. It additionally provides Arm Neon/AVX2/AVX-512. Just
implement --build-id={md5,sha1} with truncated BLAKE3.
Linking clang 14 RelWithDebInfo with --threads=8 on a Skylake CPU:
* 1.13x as fast with --build-id=md5
* 1.15x as fast with --build-id=sha1
--threads=4 on Apple m1:
* 1.25x as fast with --build-id=md5
* 1.17x as fast with --build-id=sha1
Reviewed By: ikudrin
Differential Revision: https://reviews.llvm.org/D121531
Argyrios Kyrtzidis [Thu, 24 Mar 2022 18:28:24 +0000 (11:28 -0700)]
[Support/BLAKE3] Temporarily disable building the assembly files to get the builders back to green
Gulfem Savrun Yeniceri [Thu, 24 Mar 2022 18:10:08 +0000 (18:10 +0000)]
Revert "[clang][deps] NFC: De-duplicate clang-cl tests"
This reverts commit
30cb49b44e4e2af2e0627b2aa33df0ba57ab9e55.
It caused test failures on Fuchsia Toolchain Mac builds:
https://luci-milo.appspot.com/ui/p/fuchsia/builders/toolchain.ci/clang-mac-x64/
b8818888168677577537/overview
Simon Pilgrim [Thu, 24 Mar 2022 17:59:50 +0000 (17:59 +0000)]
[X86] combineSub - fold SUB(X,ADC(Y,0,W)) -> SBB(X,Y,W)
Arthur Eubanks [Thu, 24 Mar 2022 17:53:50 +0000 (10:53 -0700)]
[gn build] Manually port
0c86198b2
Daniel Grumberg [Wed, 23 Mar 2022 15:03:02 +0000 (15:03 +0000)]
[clang][extract-api] Stop allocating APIRecords via BumpPtrAllocator
Using a BumpPtrAllocator introduced memory leaks for APIRecords as they
contain a std::vector. This meant that we needed to always keep a
reference to the records in APISet and arrange for their destructor to
get called appropriately. This was further complicated by the need for
records to own sub-records as these subrecords would still need to be
allocated via the BumpPtrAllocator and the owning record would now need
to arrange for the destructor of its subrecords to be called
appropriately.
Since APIRecords contain a std::vector so whenever elements get added to
that there is an associated heap allocation regardless. Since
performance isn't currently our main priority it makes sense to use
regular unique_ptr to keep track of APIRecords, this way we don't need
to arrange for destructors to get called.
The BumpPtrAllocator is still used for strings such as USRs so that we
can easily de-duplicate them as necessary.
Differential Revision: https://reviews.llvm.org/D122331
Simon Pilgrim [Thu, 24 Mar 2022 17:41:24 +0000 (17:41 +0000)]
Fix unused variable warning by embedding inside assertion
Argyrios Kyrtzidis [Thu, 24 Mar 2022 01:16:56 +0000 (18:16 -0700)]
[Support/BLAKE3] LLVM-specific changes over the original BLAKE3 C implementation
Changes from original BLAKE3 sources:
* `blake.h`:
* Changes to avoid conflicts if a client also links with its own BLAKE3 version:
* Renamed the header macro guard with `LLVM_C_` prefix
* Renamed the C symbols to add the `llvm_` prefix
* Added a top header comment that references the CC0 license and points to the `LICENSE` file in the repo.
* `blake3_impl.h`: Added `#define`s to remove some of `llvm_` prefixes for the rest of the internal implementation.
* Implementation files:
* Added a top header comment for `blake.c`
* Used `llvm_` prefix for the C public API functions
* Used `LLVM_LIBRARY_VISIBILITY` for internal implementation functions
* Added `.private_extern`/`.hidden` in assembly files to reduce visibility of the internal implementation functions
* `README.md`:
* added a note about where the sources originated from
* Used the C++ BLAKE3 class and `llvm_` prefixed C API in place of examples and API documentation.
* Removed instructions about how to build the files.
Argyrios Kyrtzidis [Sat, 12 Mar 2022 01:32:27 +0000 (17:32 -0800)]
[Support] Introduce the BLAKE3 hashing function implementation
BLAKE3 is a cryptographic hash function that is secure and very performant.
The C implementation originates from https://github.com/BLAKE3-team/BLAKE3/tree/1.3.1/c
License is at https://github.com/BLAKE3-team/BLAKE3/blob/1.3.1/LICENSE
This patch adds:
* `llvm/include/llvm-c/blake3.h`: The BLAKE3 C API
* `llvm/include/llvm/Support/BLAKE3.h`: C++ wrapper of the C API
* `llvm/lib/Support/BLAKE3`: Directory containing the BLAKE3 C implementation files, including the `LICENSE` file
* `llvm/unittests/Support/BLAKE3Test.cpp`: unit tests for the BLAKE3 C++ wrapper
This initial patch contains the pristine BLAKE3 sources, a follow-up patch will introduce
LLVM-specific prefixes to avoid conflicts if a client also links with its own BLAKE3 version.
And here's some timings comparing BLAKE3 with LLVM's SHA1/SHA256/MD5.
Timings include `AVX512`, `AVX2`, `neon`, and the generic/portable implementations.
The table shows the speed-up multiplier of BLAKE3 for hashing 100 MBs:
| Processor | SHA1 | SHA256 | MD5 |
|-------------------------|-------|--------|------|
| Intel Xeon W (AVX512) | 10.4x | 27x | 9.4x |
| Intel Xeon W (AVX2) | 6.5x | 17x | 5.9x |
| Intel Xeon W (portable) | 1.3x | 3.3x | 1.1x |
| M1Pro (neon) | 2.1x | 4.7x | 2.8x |
| M1Pro (portable) | 1.1x | 2.4x | 1.5x |
Differential Revision: https://reviews.llvm.org/D121510
Shraiysh Vaishay [Thu, 24 Mar 2022 16:46:43 +0000 (22:16 +0530)]
[mlir][OpenMP][IRBuilder] Add support for nowait on single construct
This patch adds the nowait parameter to `createSingle` in
OpenMPIRBuilder and handling for IR generation from OpenMP Dialect.
Also added tests for the same.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D122371
Simon Pilgrim [Thu, 24 Mar 2022 17:16:46 +0000 (17:16 +0000)]
[X86] Add additional 'add/sub single bit' patterns
InstCombine can convert some cases to 'splat bit' sra(shl(x,y),bw-1) patterns which we need to handle as well
Michael Jones [Thu, 24 Mar 2022 00:00:45 +0000 (17:00 -0700)]
[libc][obvious] only test FILE on working platforms
The main code for the FILE struct is only enabled on platforms that it
works on, but before this patch the tests were included unconditionally.
Reviewed By: sivachandra, lntue
Differential Revision: https://reviews.llvm.org/D122363
Louis Dionne [Thu, 24 Mar 2022 13:03:56 +0000 (09:03 -0400)]
[libc++] Audit all uses of _LIBCPP_ASSERT and _LIBCPP_DEBUG_ASSERT
I audited all uses of _LIBCPP_ASSERT to make sure that we only used it
for "basic assertions", i.e. assertions with constant-time conditions.
I also audited all uses of _LIBCPP_DEBUG_ASSERT to make sure we used it
only for debug-mode assertions, and in one case had to change for
_LIBCPP_ASSERT instead.
As a fly-by, I also changed a couple of tests against nullptr or 0 to
be more explicit.
After this patch, all uses of _LIBCPP_ASSERT should be with constant-time
conditions, and all uses of _LIBCPP_DEBUG_ASSERT should be with conditions
that we only want to check when the debug mode is enabled.
Differential Revision: https://reviews.llvm.org/D122395
Shafik Yaghmour [Thu, 24 Mar 2022 16:58:58 +0000 (09:58 -0700)]
[LLDB] Cleanup for Fixing DWARFExpression handling of ValueType::FileAddress case for DW_OP_deref_size
Late review on https://reviews.llvm.org/D121408 spotted some nice quick clean-ups on this code.
Arthur Eubanks [Tue, 22 Mar 2022 00:41:06 +0000 (17:41 -0700)]
[polly] Remove last instances of -analyze
As mentioned in D120782, the loop block order can be different depending
on if LoopInfo is incrementally updated or freshly computed.
Reviewed By: Meinersbur
Differential Revision: https://reviews.llvm.org/D122195
Dmitry Vyukov [Thu, 24 Mar 2022 16:07:32 +0000 (17:07 +0100)]
[TSan] Add a runtime flag to print full thread creation stacks up to the main thread
Currently, we only print how threads involved in data race are created from their parent threads.
Add a runtime flag 'print_full_thread_history' to print thread creation stacks for the threads involved in the data race and their ancestors up to the main thread.
Reviewed By: dvyukov
Differential Revision: https://reviews.llvm.org/D122131
Mike Rice [Wed, 23 Mar 2022 22:37:06 +0000 (15:37 -0700)]
[OpenMP] Initial parsing/sema for the 'omp target parallel loop' construct
Adds basic parsing/sema/serialization support for the
#pragma omp target parallel loop directive.
Differential Revision: https://reviews.llvm.org/D122359
Momchil Velikov [Thu, 24 Mar 2022 15:27:04 +0000 (15:27 +0000)]
[AArch64] Async unwind - function prologues
Re-commit of
32e8b550e5439c7e4aafa73894faffd5f25d0d05
This patch rearranges emission of CFI instructions, so the resulting
DWARF and `.eh_frame` information is precise at every instruction.
The current state is that the unwind info is emitted only after the
function prologue. This is fine for synchronous (e.g. C++) exceptions,
but the information is generally incorrect when the program counter is
at an instruction in the prologue or the epilogue, for example:
```
stp x29, x30, [sp, #-16]! // 16-byte Folded Spill
mov x29, sp
.cfi_def_cfa w29, 16
...
```
after the `stp` is executed the (initial) rule for the CFA still says
the CFA is in the `sp`, even though it's already offset by 16 bytes
A correct unwind info could look like:
```
stp x29, x30, [sp, #-16]! // 16-byte Folded Spill
.cfi_def_cfa_offset 16
mov x29, sp
.cfi_def_cfa w29, 16
...
```
Having this information precise up to an instruction is useful for
sampling profilers that would like to get a stack backtrace. The end
goal (towards this patch is just a step) is to have fully working
`-fasynchronous-unwind-tables`.
Reviewed By: danielkiss, MaskRay
Differential Revision: https://reviews.llvm.org/D111411
Valentin Clement [Thu, 24 Mar 2022 16:15:00 +0000 (17:15 +0100)]
[flang[OpenACC] Lower wait directive
This patch adds lowering for the `!$acc wait` directive
from the PFT to OpenACC dialect.
This patch is part of the upstreaming effort from fir-dev branch.
Reviewed By: PeteSteinfeld
Differential Revision: https://reviews.llvm.org/D122399
Craig Topper [Thu, 24 Mar 2022 15:59:00 +0000 (08:59 -0700)]
[SelectionDAG] Add AssertAlign to AddNodeIDCustom so that it will CSE properly.
The alignment needs to be part of the folding set hash. This is
handled by getAssertAlign when nodes are created, but needs to repeated here.
No test case as I found it as part of a very early experimental patch.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D122279
Alexander Belyaev [Thu, 24 Mar 2022 15:29:21 +0000 (16:29 +0100)]
[llvm] Initialize and move UseUpRegs outside of `union` MemOp struct.
Asan complained about uninitialized bool
`invalid-bool-load`
llvm/lib/Target/X86/AsmParser/X86Operand.h:389:12: runtime error: load
of value 171, which is not a valid value for type 'bool'
Differential Revision: https://reviews.llvm.org/D122405
Aaron Ballman [Thu, 24 Mar 2022 15:51:31 +0000 (11:51 -0400)]
Diagnose use of a naked attribute on a member function in ms-mode
This matches the behavior of cl.exe.
Jakob Koschel [Thu, 24 Mar 2022 06:52:16 +0000 (07:52 +0100)]
Reland "[ELF] Enable new passmanager plugin support for LTO"
This is the orignal patch + a check that LLVM_BUILD_EXAMPLES is enabled before
adding a dependency on the 'Bye' example pass.
Original summary:
Add cli options for new passmanager plugin support to lld.
Currently it is not possible to load dynamic NewPM plugins with lld. This is an
incremental update to D76866. While that patch only added cli options for
llvm-lto2, this adds them for lld as well. This is especially useful for running
dynamic plugins on the linux kernel with LTO.
Reviewed By: MaskRay
Differential Revision: https://reviews.llvm.org/D120490
Florian Hahn [Thu, 24 Mar 2022 14:58:23 +0000 (14:58 +0000)]
[VPlan] Add VPWidenPointerInductionRecipe.
This patch moves pointer induction handling from VPWidenPHIRecipe to its
own recipe. In the process, it adds all information required to generate
code for pointer inductions without relying on Legal to access the list
of induction phis.
Alternatively VPWidenPHIRecipe could also take an optional pointer to InductionDescriptor.
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D121615
Aaron Ballman [Thu, 24 Mar 2022 14:38:37 +0000 (10:38 -0400)]
Fix a crash with variably-modified parameter types in a naked function
Naked functions have no prolog, so it's not valid to emit prolog code
to evaluate the variably-modified type. This fixes Issue 50541.
Valentin Clement [Thu, 24 Mar 2022 14:29:59 +0000 (15:29 +0100)]
[flang[OpenACC] Lower data directive
This patch adds lowering for the `!$acc data`
from the PFT to OpenACC dialect.
This patch is part of the upstreaming effort from fir-dev branch.
Depends on D122384
Reviewed By: PeteSteinfeld
Differential Revision: https://reviews.llvm.org/D122398
Sanjay Patel [Thu, 24 Mar 2022 14:01:41 +0000 (10:01 -0400)]
[InstCombine] merge shuffled vector negate and multiply
Add the "(0 - X) --> (X * -1)" reverse identity to the list of alternate form binops.
We need a little hack to make the existing logic work because it does not expect to
move constants from op0 to op1, but the code comment hopefully makes that clear.
I don't think there are any other identities like that.
Fixes #54364
Differential Revision: https://reviews.llvm.org/D122390
Louis Dionne [Thu, 24 Mar 2022 14:00:23 +0000 (10:00 -0400)]
[libc++][NFC] Refactor the ABI changelog
Add missing commit SHAs in a few cases, and use "All Platforms" when
the change affected all platforms instead of explicitly listing the
platforms, since there's too many to list exhaustively.
Kadir Cetinkaya [Wed, 23 Mar 2022 14:22:47 +0000 (15:22 +0100)]
[clangd] Retain main file fixes attached to diags from preamble
Clangd ignores fixits if the diagnsotics is outside the main file (e.g.
a note on a declaration from a header), but the fix might still be inside the
main file (e.g. change the function call).
This patch changes the logic to retain fixes that touch main file, if the
diagnostic owning them is still inside main file, even if they are attached to a
note outside the main file.
Differential Revision: https://reviews.llvm.org/D122315
Valentin Clement [Thu, 24 Mar 2022 14:18:29 +0000 (15:18 +0100)]
[flang][OpenACC] Lower update directive
This patch adds lowering for the `!$acc update`
from the PFT to OpenACC dialect.
This patch is part of the upstreaming effort from fir-dev branch.
Depends on D122387
Reviewed By: PeteSteinfeld
Differential Revision: https://reviews.llvm.org/D122396
Valentin Clement [Thu, 24 Mar 2022 14:16:46 +0000 (15:16 +0100)]
[flang][OpenACC] Lower init/shutdown directive
This patch adds lowering for the `!$acc init` and `!$acc shutdown`
from the PFT to OpenACC dialect.
This patch is part of the upstreaming effort from fir-dev branch.
Depends on D122384
Reviewed By: PeteSteinfeld
Differential Revision: https://reviews.llvm.org/D122387
Valentin Clement [Thu, 24 Mar 2022 14:15:40 +0000 (15:15 +0100)]
[flang][OpenACC] Lower exit data directive
This patch adds lowering for the `!$acc exit data` directive
from the PFT to OpenACC dialect.
This patch is part of the upstreaming effort from fir-dev branch.
Depends on D122384
Reviewed By: PeteSteinfeld
Differential Revision: https://reviews.llvm.org/D122386
Valentin Clement [Thu, 24 Mar 2022 14:00:52 +0000 (15:00 +0100)]
[flang][OpenACC] Lower enter data directive
This patch adds lowering for the `!$acc enter data` directive
from the PFT to OpenACC dialect.
This patch is part of the upstreaming effort from fir-dev branch.
Reviewed By: PeteSteinfeld
Differential Revision: https://reviews.llvm.org/D122384
Valentin Clement [Thu, 24 Mar 2022 13:58:27 +0000 (14:58 +0100)]
[flang][NFC] Add lowering test for array expression
This patches adds lowering tests for some array
expressions use cases.
This patch is part of the upstreaming effort from fir-dev branch.
Reviewed By: PeteSteinfeld
Differential Revision: https://reviews.llvm.org/D122380
Co-authored-by: Jean Perier <jperier@nvidia.com>
Co-authored-by: Eric Schweitz <eschweitz@nvidia.com>
Louis Dionne [Thu, 24 Mar 2022 13:57:28 +0000 (09:57 -0400)]
[libc++][NFC] Update ABI changelog to account for new assertion handler
Sven van Haastregt [Thu, 24 Mar 2022 13:55:41 +0000 (13:55 +0000)]
[OpenCL] opencl-c.h: remove x/y/z arg names
This simplifies completeness comparisons against OpenCLBuiltins.td and
also makes the header no longer "claim" the identifiers "x", "y" and
"z".
Continues the direction set out in D119560.
LLVM GN Syncbot [Thu, 24 Mar 2022 13:50:02 +0000 (13:50 +0000)]
[gn build] Port
62d5f254ccd0
Haojian Wu [Mon, 21 Mar 2022 14:32:15 +0000 (15:32 +0100)]
[pseudo] Introduce parse forest.
Parse forest is the output of the GLR parser, it is a tree-like DAG
which presents all possible parse trees without duplicating subparse structures.
This is a patch split from https://reviews.llvm.org/D121150.
Differential Revision: https://reviews.llvm.org/D122139
Djordje Todorovic [Wed, 23 Mar 2022 15:00:03 +0000 (16:00 +0100)]
NFC: [LICM] Update some stale comments
After removing the MaybePromotable, some comments
became stale. This improves them.
Differential Revision: https://reviews.llvm.org/D122319
Haojian Wu [Wed, 23 Mar 2022 10:38:49 +0000 (11:38 +0100)]
[pseudo] Sort nonterminals based on their reduction order.
Reductions need to be performed in a careful order in GLR parser, to
make sure we gather all alternatives before creating an ambigous forest
node.
This patch encodes the nonterminal order into the rule id, so that we
can efficiently to determinal ordering of reductions in GLR parser.
This patch also abstracts to a TestGrammar, which is shared among tests.
This is a part of the GLR parser, https://reviews.llvm.org/D121368,
https://reviews.llvm.org/D121150
Differential Revision: https://reviews.llvm.org/D122303
Qiu Chaofan [Thu, 24 Mar 2022 12:46:35 +0000 (20:46 +0800)]
[NFC] Format and uglify PowerPC intrinsics headers
This change formats PowerPC intrinsics wrapper headers into LLVM style,
and add extra prefix '__' to all variables to prevent conflict with user
code.
Nico Weber [Thu, 24 Mar 2022 13:07:46 +0000 (09:07 -0400)]
[gn build] remove a "from __future__" import not needed after
0ff3cc2087b9
Alexey Bataev [Thu, 24 Mar 2022 12:48:35 +0000 (05:48 -0700)]
[SLP][NFC]Fix param name in comments, NFC.
Nathan Sidwell [Wed, 23 Mar 2022 14:27:43 +0000 (07:27 -0700)]
[clang] Reformat
Reformat some misindentation that is coincidentally close to a piece
being worked on.
Reviewed By: dblaikie
Differential Revision: https://reviews.llvm.org/D122314
Aaron Ballman [Thu, 24 Mar 2022 12:55:43 +0000 (08:55 -0400)]
Typo fix in the documentation; NFC
Martin Storsjö [Thu, 24 Mar 2022 11:05:30 +0000 (13:05 +0200)]
[compiler-rt] [cmake] Fix a typo in a variable name
There's no other references to this variable name, but other files
refer to the same with the same name but with double Fs.
Jakob Johnson [Tue, 22 Mar 2022 13:15:56 +0000 (06:15 -0700)]
[trace][intelpt] Server side changes for TSC to wall time conversion
Update the response schema of the TraceGetState packet and add
Intel PT specific response structure that contains the TSC conversion,
if it exists. The IntelPTCollector loads the TSC conversion and caches
it to prevent unnecessary calls to perf_event_open. Move the TSC conversion
calculation from Perf.h to TraceIntelPTGDBRemotePackets.h to remove
dependency on Linux specific headers.
Differential Revision: https://reviews.llvm.org/D122246
LLVM GN Syncbot [Thu, 24 Mar 2022 12:12:11 +0000 (12:12 +0000)]
[gn build] Port
406bde9a1513
Nikita Popov [Thu, 24 Mar 2022 12:07:24 +0000 (13:07 +0100)]
[lldb] Avoid pointer element type accesses
Qiu Chaofan [Thu, 24 Mar 2022 12:03:08 +0000 (20:03 +0800)]
[PowerPC] [Clang] Add SSE4 and BMI intrinsics implementation
Reviewed By: jsji
Differential Revision: https://reviews.llvm.org/D119407
Nikita Popov [Thu, 24 Mar 2022 11:55:42 +0000 (12:55 +0100)]
[lldb] Use byval type
Query byval type instead of pointer element type.
Nikita Popov [Thu, 24 Mar 2022 11:53:02 +0000 (12:53 +0100)]
[lldb] Remove unused Module argument (NFC)
Nikita Popov [Thu, 24 Mar 2022 10:51:04 +0000 (11:51 +0100)]
[Docs] Update opaque pointers transition state (NFC)
Opaque pointer support in Clang is now complete.
Shengchen Kan [Thu, 24 Mar 2022 10:46:59 +0000 (18:46 +0800)]
[X86][NFC] Remove unused variable introduce by D121785
gysit [Thu, 24 Mar 2022 10:28:20 +0000 (10:28 +0000)]
[mlir][linalg] Support padding LinalgOps in use-def chain.
Previously, only LinalgOps whose operands are defined by an ExtractSliceOp could be padded. The revision supports walking a use-def chain of LinalgOps to find an ExtractSliceOp.
Reviewed By: hanchung
Differential Revision: https://reviews.llvm.org/D122116
Antonio Frighetto [Thu, 24 Mar 2022 10:42:47 +0000 (11:42 +0100)]
[NFC] Mark derived destructors as `override`
Derived destructors can be marked as override, in order to prevent
possible compilation failures of projects depending on those
headers (when compiled with flags -Wall, -Wsuggest-destructor-override,
-Winconsistent-missing-destructor-override).
Differential Revision: https://reviews.llvm.org/D121993
gysit [Thu, 24 Mar 2022 10:22:28 +0000 (10:22 +0000)]
[mlir][linalg] Do not fuse shape-only producers.
This revision introduces a heuristic to stop fusion for shape-only tensors. A shape-only tensor only defines the shape of the consumer computation while the data is not used. Pure producer consumer fusion thus shall not fuse the producer of a shape-only tensor. In particular, since the shape-only tensor will have other uses that actually consume the data.
The revision enables fusion for consumers that have two uses of the same tensor. One as input operand and one as shape-only output operand. In these cases, we want to fuse only the input operand and avoid output fusion via iteration argument.
Reviewed By: hanchung
Differential Revision: https://reviews.llvm.org/D120981
David Green [Thu, 24 Mar 2022 10:02:33 +0000 (10:02 +0000)]
[AArch64] Lower 3 and 4 sources buildvectors to TBL
The default expansion for buildvectors is to extract each element and
insert them into a new vector. That involves a lot of copying to/from
the GPR registers. TLB3 and TLB4 can be relatively slow instructions
with the mask needing to be loaded from a constant pool, but they should
always be better than all the moves to/from GPRs.
Differential Revision: https://reviews.llvm.org/D121137
Benjamin Kramer [Thu, 24 Mar 2022 10:01:08 +0000 (11:01 +0100)]
[bazel] glob LLVMCoroutines headers to make the build rule less fragile
Benjamin Kramer [Thu, 24 Mar 2022 09:55:02 +0000 (10:55 +0100)]
[bazel] Set the default of LLVM_UNREACHABLE_OPTIMIZE to 1
This matches the CMake default. LLVM_UNREACHABLE_OPTIMIZE=0 doesn't
build with gcc
Daniil Kovalev [Thu, 24 Mar 2022 09:42:59 +0000 (12:42 +0300)]
[CodeGen] Define ABI breaking class members correctly
Non-static class members declared under #ifndef NDEBUG should be declared
under #if LLVM_ENABLE_ABI_BREAKING_CHECKS to make headers library-friendly and
allow cross-linking, as discussed in D120714.
Differential Revision: https://reviews.llvm.org/D121549
Daniil Kovalev [Thu, 24 Mar 2022 09:34:25 +0000 (12:34 +0300)]
[NVPTX] Enhance vectorization of ld.param & st.param
Since function parameters and return values are passed via param space, we
can force special alignment for values hold in it which will add vectorization
options. This change may be done if the function has private or internal
linkage. Special alignment is forced during 2 phases.
1) Instruction selection lowering. Here we use special alignment for function
prototypes (changing both own return value and parameters alignment), call
lowering (changing both callee's return value and parameters alignment).
2) IR pass nvptx-lower-args. Here we change alignment of byval parameters that
belong to param space (or are casted to it). We only handle cases when all
uses of such parameters are loads from it. For such loads, we can change the
alignment according to special type alignment and the load offset. Then,
load-store-vectorizer IR pass will perform vectorization where alignment
allows it.
Special alignment calculated as maximum from default ABI type alignment and
alignment 16. Alignment 16 is chosen because it's the maximum size of
vectorized ld.param & st.param.
Before specifying such special alignment, we should check if it is a multiple
of the alignment that the type already has. For example, if a value has an
enforced alignment of 64, default ABI alignment of 4 and special alignment
of 16, we should preserve 64.
This patch will be followed by a refactoring patch that removes duplicating
code in handling byval and non-byval arguments.
Differential Revision: https://reviews.llvm.org/D120129
Dávid Bolvanský [Thu, 24 Mar 2022 09:32:57 +0000 (10:32 +0100)]
Relands "[Clang] -Wunused-but-set-variable warning - handle also pre/post unary operators"
Daniil Kovalev [Thu, 24 Mar 2022 09:30:28 +0000 (12:30 +0300)]
Revert "[NVPTX] Enhance vectorization of ld.param & st.param"
This reverts commit
f854434f0f2a01027bdaad8e6fdac5a782fce291.
Placed URL to wrong differential revision in commit message.
Dávid Bolvanský [Thu, 24 Mar 2022 09:31:11 +0000 (10:31 +0100)]
[NFCI] Fix set-but-unused warning in X86DisassemblerTables.cpp
Daniil Kovalev [Thu, 24 Mar 2022 08:57:37 +0000 (11:57 +0300)]
[NVPTX] Enhance vectorization of ld.param & st.param
Since function parameters and return values are passed via param space, we
can force special alignment for values hold in it which will add vectorization
options. This change may be done if the function has private or internal
linkage. Special alignment is forced during 2 phases.
1) Instruction selection lowering. Here we use special alignment for function
prototypes (changing both own return value and parameters alignment), call
lowering (changing both callee's return value and parameters alignment).
2) IR pass nvptx-lower-args. Here we change alignment of byval parameters that
belong to param space (or are casted to it). We only handle cases when all
uses of such parameters are loads from it. For such loads, we can change the
alignment according to special type alignment and the load offset. Then,
load-store-vectorizer IR pass will perform vectorization where alignment
allows it.
Special alignment calculated as maximum from default ABI type alignment and
alignment 16. Alignment 16 is chosen because it's the maximum size of
vectorized ld.param & st.param.
Before specifying such special alignment, we should check if it is a multiple
of the alignment that the type already has. For example, if a value has an
enforced alignment of 64, default ABI alignment of 4 and special alignment
of 16, we should preserve 64.
This patch will be followed by a refactoring patch that removes duplicating
code in handling byval and non-byval arguments.
Differential Revision: https://reviews.llvm.org/D121549
Krasimir Georgiev [Thu, 24 Mar 2022 09:14:13 +0000 (10:14 +0100)]
Revert "[clang-format] Correctly recognize arrays in template parameter list."
This reverts commit
126b37a713dc1c67cbc7dc8b5288b2f907c906a9.
Regressed some ObjC patterns, see comments on https://reviews.llvm.org/D121584.
Qiu Chaofan [Thu, 24 Mar 2022 08:58:58 +0000 (16:58 +0800)]
[Clang] Add option to set alternative toolchain path
In some cases, we need to set alternative toolchain path other than the
default with system (headers, libraries, dynamic linker prefix, ld path,
etc.), e.g., to pick up newer components, but keep sysroot at the same
time (to pick up extra packages).
This change introduces a new option --overlay-platform-toolchain to set
up such alternative toolchain path.
Reviewed By: hubert.reinterpretcast
Differential Revision: https://reviews.llvm.org/D121992
Raphael Isemann [Thu, 24 Mar 2022 08:56:51 +0000 (09:56 +0100)]
Revert "[ELF] Enable new passmanager plugin support for LTO"
This reverts commit
32012eb11b235e1560a253664095676ea8ebd027.
Broke CMake configuration.
Kristof Beyls [Thu, 24 Mar 2022 08:55:32 +0000 (09:55 +0100)]
Update my office hours
Adding LLVM security group as another topic I can talk about.
David Green [Thu, 24 Mar 2022 08:55:27 +0000 (08:55 +0000)]
[AArch64] Make some methods static. NFC
Peixin-Qiao [Thu, 24 Mar 2022 08:50:20 +0000 (16:50 +0800)]
[NFC][flang] Remove unused code in lowerExplicitLowerBounds
There is no need to lower the implicit lower bounds for assumed-shape
array in lowerExplicitLowerBounds. Remove the unused code.
Reviewed By: Jean Perier
Differential Revision: https://reviews.llvm.org/D122280