platform/upstream/llvm.git
19 months ago[ThinLTO] Only import for non-prevailing interposable global variables
Shoaib Meenai [Sat, 25 Mar 2023 00:34:14 +0000 (17:34 -0700)]
[ThinLTO] Only import for non-prevailing interposable global variables

This logic was added in https://reviews.llvm.org/D95943 specifically to
handle an issue for non-prevailing global variables. It turns out that
it adds a new issue for prevailing glboal variables, since those could
be replaced by an available_externally definition and hence incorrectly
omitted from the output object file. Limit the import to non-prevailing
global variables to fix this, as suggested by @tejohnson.

The bulk of the diff is mechanical changes to thread isPrevailing
through to where it's needed and ensure it's available before the
relevant calls; the actual logic change itself is straightforward.

Fixes https://github.com/llvm/llvm-project/issues/61677

Reviewed By: tejohnson

Differential Revision: https://reviews.llvm.org/D146876

19 months ago[RISCV] Move PseudoRVVInitUndef pseudos to RISCVInstrInfoVPseudos.td. NFC
Craig Topper [Sun, 26 Mar 2023 02:18:15 +0000 (19:18 -0700)]
[RISCV] Move PseudoRVVInitUndef pseudos to RISCVInstrInfoVPseudos.td. NFC

19 months agoRevert "[llvm] Teach GlobalDCE about dso_local_equivalent"
Leonard Chan [Sun, 26 Mar 2023 01:47:10 +0000 (01:47 +0000)]
Revert "[llvm] Teach GlobalDCE about dso_local_equivalent"

This reverts commit 86dbcafd0cdc88ae85896c6f12ecaa6006aeba54.

Reverting since this depends on db288184765c0b4010060ebea1f6de3ac1f66445
which broke our lto builders reported by fxbug.dev/12380.

19 months agoRevert "[llvm] Teach whole program devirtualization about relative vtables"
Leonard Chan [Sun, 26 Mar 2023 01:38:08 +0000 (01:38 +0000)]
Revert "[llvm] Teach whole program devirtualization about relative vtables"

This reverts commit db288184765c0b4010060ebea1f6de3ac1f66445.

Reverting since it broke our lto builders reported by fxbug.dev/123807.

19 months ago[clang-format] Treat NTTP default values as expressions
Emilia Dreamer [Sun, 26 Mar 2023 01:39:03 +0000 (04:39 +0300)]
[clang-format] Treat NTTP default values as expressions

clang-format already has logic to threat the right-hand side of an
equals sign. This patch applies that logic to template defaults,
which are likely to be non-template type parameters in which case the
default value should be annotated as an expression.
This should mostly only ever apply to bool and &&.

Fixes https://github.com/llvm/llvm-project/issues/61664

Reviewed By: MyDeveloperDay, owenpan

Differential Revision: https://reviews.llvm.org/D146760

19 months ago[clang-format] Don't wrap struct return types as structs
Emilia Dreamer [Sun, 26 Mar 2023 01:38:42 +0000 (04:38 +0300)]
[clang-format] Don't wrap struct return types as structs

When using BraceWrapping.AfterClass or BraceWrapping.AfterStruct, the
token annotator relies on the first token of the line to determine if
we're dealing with a struct or class, however, this check is faulty if
it's actually a function with an elaborated struct/class return type, as
is common in C.

This patch skips the check if the brace is already annotated as
FunctionLBrace, in which case we already know it's a function and should
be treated as such.

Fixes https://github.com/llvm/llvm-project/issues/58527

Reviewed By: HazardyKnusperkeks, owenpan

Differential Revision: https://reviews.llvm.org/D146281

19 months ago[clang-format] Annotate lambdas with requires clauses.
Emilia Dreamer [Sun, 26 Mar 2023 01:37:59 +0000 (04:37 +0300)]
[clang-format] Annotate lambdas with requires clauses.

The C++ grammar allows lambdas to have a *requires-clause* in two
places, either directly after the *template-parameter-list*, such as:

`[] <typename T> requires foo<T> (T t) { ... };`

Or, at the end of the *lambda-declarator* (before the lambda's body):

`[] <typename T> (T t) requires foo<T> { ... };`

Previously, these cases weren't handled at all, resulting in weird
results.

Note that this commit only handles token annotation, so the actual
formatting still ends up suboptimal. This is mostly because I do not yet
know how to approach making the requires clause formatting of lambdas
match the formatting for functions.

Fixes https://github.com/llvm/llvm-project/issues/61269

Reviewed By: HazardyKnusperkeks, owenpan

Differential Revision: https://reviews.llvm.org/D145642

19 months ago[MLIR][Affine] add memory effect traits for dmaOp
lipracer [Sun, 26 Mar 2023 01:07:27 +0000 (06:37 +0530)]
[MLIR][Affine] add memory effect traits for dmaOp

DmaOp will read the source buffer and write the destination buffer so need to add some traits for it.

Reviewed By: bondhugula

Differential Revision: https://reviews.llvm.org/D144712

19 months ago[MachineCopyPropagation] Pass DestSourcePair to isBackwardPropagatableCopy. NFC
Craig Topper [Sun, 26 Mar 2023 00:20:08 +0000 (17:20 -0700)]
[MachineCopyPropagation] Pass DestSourcePair to isBackwardPropagatableCopy. NFC

Instead of calling isCopyInstr again, just pass the DestSourcePair
from the isCopyInstr call from the caller.

19 months ago[ARM] Add some tests for non-zero VCTP generation. NFC
David Green [Sat, 25 Mar 2023 23:33:25 +0000 (23:33 +0000)]
[ARM] Add some tests for non-zero VCTP generation. NFC

See D146517.

19 months ago[libc] Add missing cast in fputil sqrt code
Roland McGrath [Sat, 25 Mar 2023 21:30:02 +0000 (14:30 -0700)]
[libc] Add missing cast in fputil sqrt code

A cast is necessary to avoid implicit narrowing warnings
when those are enabled.

Reviewed By: abrachet

Differential Revision: https://reviews.llvm.org/D146886

19 months ago[MLIR][Affine] Fix/improve affine sibling fusion
Uday Bondhugula [Sat, 25 Mar 2023 22:05:03 +0000 (03:35 +0530)]
[MLIR][Affine] Fix/improve affine sibling fusion

The sibling fusion profitability checks shouldn't rely on the presence
of a store op in the sibling. The reuse is between the loads.

Fixes issues raised at https://discourse.llvm.org/t/understanding-the-affine-loop-fusion-pass/69452

Reviewed By: dcaballe

Differential Revision: https://reviews.llvm.org/D146763

19 months ago[Matrix] Extend test coverage for dot product lowering.
Florian Hahn [Sat, 25 Mar 2023 21:30:20 +0000 (21:30 +0000)]
[Matrix] Extend test coverage for dot product lowering.

Extra tests:
* result is used by instruction
* constant vector operands
* multiply fed by other math instructions
* extra test with larger stride

19 months ago[Matrix] Split up dot product tests into integer and float variants.
Florian Hahn [Sat, 25 Mar 2023 21:23:00 +0000 (21:23 +0000)]
[Matrix] Split up dot product tests into integer and float variants.

To avoid the individual files getting too big with further additions.

19 months ago[clang-format] Handle Verilog assign statements
sstwcw [Sat, 25 Mar 2023 21:12:49 +0000 (21:12 +0000)]
[clang-format] Handle Verilog assign statements

Reviewed By: MyDeveloperDay

Differential Revision: https://reviews.llvm.org/D146402

19 months ago[clang-format] More work on space around operators in Verilog
sstwcw [Sat, 25 Mar 2023 21:12:13 +0000 (21:12 +0000)]
[clang-format] More work on space around operators in Verilog

before:
```
(opcode *>o1) = 6.1;
a inside{b, c};
x = { >> {j}};
```

after:
```
(opcode *> o1) = 6.1;
a inside {b, c};
x = {>>{j}};
```

Reviewed By: MyDeveloperDay

Differential Revision: https://reviews.llvm.org/D146403

19 months ago[AMDGPUUnifyDivergentExitNodes] Add NewPM support
Anshil Gandhi [Sat, 25 Mar 2023 19:42:31 +0000 (13:42 -0600)]
[AMDGPUUnifyDivergentExitNodes] Add NewPM support

Meanwhile, use UniformityAnalysis instead of LegacyDivergenceAnalysis to collect divergence info.

Reviewed By: arsenm, sameerds

Differential Revision: https://reviews.llvm.org/D141355

19 months ago[clang][NFC] Fix location of 2>&1 in a few -print tests
Louis Dionne [Wed, 15 Mar 2023 17:46:51 +0000 (13:46 -0400)]
[clang][NFC] Fix location of 2>&1 in a few -print tests

While it's apparently valid to place Bash redirections anywhere in a
command-line, it is by far most frequently placed last. This changes
a few tests that did not conform to this convention and which I
originally thought were wrong.

Differential Revision: https://reviews.llvm.org/D146155

19 months ago[libc++] Add UNSUPPORTED annotations to more <format> tests on GCC 12
Louis Dionne [Sat, 25 Mar 2023 18:36:40 +0000 (14:36 -0400)]
[libc++] Add UNSUPPORTED annotations to more <format> tests on GCC 12

Those seem to have been failing for a while but we might not have noticed
because of the recent CI instability issues. I'm marking them as unsupported
to try to get the CI functional again, especially since the majority of
<format> tests are already not working on GCC 12.

19 months ago[libc] Define LLVM_LIBC_FUNCTION with a layer of macro expansion
Roland McGrath [Sat, 25 Mar 2023 03:29:08 +0000 (20:29 -0700)]
[libc] Define LLVM_LIBC_FUNCTION with a layer of macro expansion

Move the real LLVM_LIBC_FUNCTION macro definitions to
LLVM_LIBC_FUNCTION_IMPL and make LLVM_LIBC_FUNCTION a wrapper to
expand macros in its arguments.  This makes it possible to
compile libc implementation and test files with -Dfunc=othername.

Reviewed By: sivachandra

Differential Revision: https://reviews.llvm.org/D146863

19 months ago[libc++][NFC] Improve documentation for running BuildKite agent locally
Louis Dionne [Sat, 25 Mar 2023 17:55:04 +0000 (13:55 -0400)]
[libc++][NFC] Improve documentation for running BuildKite agent locally

19 months ago[Docs][llvm-mc] Add documentation on --filetype flag
Aiden Grossman [Sat, 25 Mar 2023 17:15:01 +0000 (17:15 +0000)]
[Docs][llvm-mc] Add documentation on --filetype flag

Currently the filetype flag is not documented, and knowing the behavior
of this flag is fairly important for doing anything other than
disassembling to text assembly.

Reviewed By: lattner

Differential Revision: https://reviews.llvm.org/D146878

19 months ago[lldb] Move #include out of namespace lldb. NFC.
Benjamin Kramer [Sat, 25 Mar 2023 12:33:17 +0000 (13:33 +0100)]
[lldb] Move #include out of namespace lldb. NFC.

Including system headers in a namespace is not safe.

19 months ago[Clang] Update DR status page to reflect Core Issues List 111.
Corentin Jabot [Sat, 25 Mar 2023 12:15:08 +0000 (13:15 +0100)]
[Clang] Update DR status page to reflect Core Issues List 111.

19 months ago[RISCV] Support vector crypto extension ISA string and assembly
4vtomat [Wed, 26 Oct 2022 02:09:21 +0000 (19:09 -0700)]
[RISCV] Support vector crypto extension ISA string and assembly

LLVM implements the 0.3 draft specification:
https://github.com/riscv/riscv-crypto/releases/download/v20230206/riscv-crypto-spec-vector.pdf
, and current vector crypto extension version can be found in:
https://github.com/riscv/riscv-crypto.

Differential Revision: https://reviews.llvm.org/D141672

19 months ago[mlir][Analysis] Fix assertion in FlatLinearConstraints
Matthias Springer [Sat, 25 Mar 2023 11:17:45 +0000 (12:17 +0100)]
[mlir][Analysis] Fix assertion in FlatLinearConstraints

Differential Revision: https://reviews.llvm.org/D146869

19 months ago[RISCV][RISCVISelLowering] Add tail agnostic policy operand to VECREDUCE instructions
Nitin John Raj [Thu, 23 Mar 2023 19:55:58 +0000 (12:55 -0700)]
[RISCV][RISCVISelLowering] Add tail agnostic policy operand to VECREDUCE instructions

Differential Revision: https://reviews.llvm.org/D146752

19 months ago[mlir][Vector] Use a RewriterBase for IR rewrites in VectorTransferOpTransforms
Nicolas Vasilache [Sat, 25 Mar 2023 07:25:28 +0000 (00:25 -0700)]
[mlir][Vector] Use a RewriterBase for IR rewrites in VectorTransferOpTransforms

19 months ago[X86] Fix the incorrect displacement for prolog/epilog
Luo, Yuanke [Sat, 25 Mar 2023 01:59:31 +0000 (09:59 +0800)]
[X86] Fix the incorrect displacement for prolog/epilog

The bug is introduced in rGe4ceb5a7bb9b which set the wrong offset from
the stack base. This patch is to fix the bug.

Differential Revision: https://reviews.llvm.org/D146862

19 months ago[libc] Remove unused aarch64 sqrt and sqrtf implementations
Roland McGrath [Sat, 25 Mar 2023 03:59:17 +0000 (20:59 -0700)]
[libc] Remove unused aarch64 sqrt and sqrtf implementations

These files are not used because the generic sqrt and sqrtf
functions already go through internal layers that reach the
machine-specific internal implemenations.

Reviewed By: sivachandra

Differential Revision: https://reviews.llvm.org/D146865

19 months ago[RISCV] Make RISCVMergeBaseOffsetOpt a class instead of a struct. NFC
Craig Topper [Sat, 25 Mar 2023 02:59:12 +0000 (19:59 -0700)]
[RISCV] Make RISCVMergeBaseOffsetOpt a class instead of a struct. NFC

This lets us remove 'private:' from the top of the class.
While there collect the other private member variable at the top
of the class.

19 months ago[Analysis] Fix use-after-scope in CFGElement dump
Sam McCall [Sat, 25 Mar 2023 03:32:52 +0000 (04:32 +0100)]
[Analysis] Fix use-after-scope in CFGElement dump

19 months ago[dataflow] handle missing case in value debug strings
Sam McCall [Wed, 22 Mar 2023 12:16:17 +0000 (13:16 +0100)]
[dataflow] handle missing case in value debug strings

Differential Revision: https://reviews.llvm.org/D146625

19 months agoRevert "[OpenMP] Ensure memory fences are created with barriers for AMDGPUs"
Ye Luo [Sat, 25 Mar 2023 02:10:03 +0000 (21:10 -0500)]
Revert "[OpenMP] Ensure memory fences are created with barriers for AMDGPUs"

This reverts commit 36d6217c4eb02c15168bf74c9f7ef44ea4fb7e41.

19 months ago[OpenMP] Ensure memory fences are created with barriers for AMDGPUs
Ye Luo [Sat, 25 Mar 2023 01:36:51 +0000 (20:36 -0500)]
[OpenMP] Ensure memory fences are created with barriers for AMDGPUs

It turns out that the `__builtin_amdgcn_s_barrier()` alone does not emit
a fence. We somehow got away with this and assumed it would work as it
(hopefully) is correct on the NVIDIA path where we just emit a
`__syncthreads`. After talking to @arsenm we now (mostly) align with the
OpenCL barrier implementation [1] and emit explicit fences for AMDGPUs.

It seems this was the underlying cause for #59759, but I am not 100%
certain. There is a chance this simply hides the problem.

Fixes: https://github.com/llvm/llvm-project/issues/59759

[1] https://github.com/RadeonOpenCompute/ROCm-Device-Libs/blob/07b347366eb2c6ebc3414af323c623cbbbafc854/opencl/src/workgroup/wgbarrier.cl#L21

Reviewed By: ye-luo

Differential Revision: https://reviews.llvm.org/D145290

19 months ago[libc] Implement the RPC client / server for NVPTX
Joseph Huber [Fri, 24 Mar 2023 20:53:05 +0000 (15:53 -0500)]
[libc] Implement the RPC client / server for NVPTX

This patch adds the necessary code to impelement the existing RPC client
/ server interface when targeting NVPTX GPUs. This follows closely to
the implementation in the AMDGPU version. This does not yet enable unit
testing as the `nvlink` linker does not support static libraries. So
that will need to be worked around.

I am ignoring the RPC duplication between the AMDGPU and NVPTX loaders. This
will be changed completely later so there's no point unifying the code at this
stage. The implementation was tested manually with the following file and
compilation flags.

```
namespace __llvm_libc {
void write_to_stderr(const char *msg);
void quick_exit(int);
} // namespace __llvm_libc

using namespace __llvm_libc;

int main(int argc, char **argv, char **envp) {
  for (int i = 0; i < argc; ++i) {
    write_to_stderr(argv[i]);
    write_to_stderr("\n");
  }
  quick_exit(255);
}
```

```
$ clang++ crt1.o rpc_client.o quick_exit.o io.o main.cpp --target=nvptx64-nvidia-cuda -march=sm_70 -o image
$ ./nvptx_loader image 1 2 3
image
1
2
3
$ echo $?
255
```

Depends on D146681

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D146846

19 months ago[libc] Add a loader utility for NVPTX architectures for testing
Joseph Huber [Thu, 23 Mar 2023 01:00:13 +0000 (20:00 -0500)]
[libc] Add a loader utility for NVPTX architectures for testing

This patch adds a loader utility targeting the CUDA driver API to launch
NVPTX images called `nvptx_loader`. This takes a GPU image on the
command line and launches the `_start` kernel with the appropriate
arguments. The `_start` kernel is provided by the already implemented
`nvptx/start.cpp`. So, an application with a `main` function can be
compiled and run as follows.

```
clang++ --target=nvptx64-nvidia-cuda main.cpp crt1.o -march=sm_70 -o image
./nvptx_loader image args to kernel
```

This implementation is not tested and does not yet support RPC. This
requires further development to work around NVIDIA specific limitations
in atomics and linking.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D146681

19 months ago[lldb][NFC] Remove outdated TODO in Log.h
Alex Langford [Sat, 25 Mar 2023 00:53:52 +0000 (17:53 -0700)]
[lldb][NFC] Remove outdated TODO in Log.h

The code this TODO was referring to was removed in
c34698a811b137b705738b7f8d193bc896027fb8.

19 months ago[lldb][NFC] Update and re-organize lldb-types.h
Alex Langford [Sat, 25 Mar 2023 00:28:13 +0000 (17:28 -0700)]
[lldb][NFC] Update and re-organize lldb-types.h

- Address the TODO by identifying and documenting all types needed by a
  host system in order for lldb to work correctly
- Reformatted the comments to be easier to read
- Put everything inside of one namespace declaration instead of having
  multiple of the same
- Move the macros up to be right under the accompanying definitions

19 months ago[lldb] Remove errant call to SBReproducer.SetWorkingDirectory
Jonas Devlieghere [Sat, 25 Mar 2023 00:31:39 +0000 (17:31 -0700)]
[lldb] Remove errant call to SBReproducer.SetWorkingDirectory

The old reproducer functionality has been removed. Remove this call as
it's now just a NO-OP.

19 months ago[RISCV] Made fsqrtv pseudoinstruction SEW-aware
Nitin John Raj [Fri, 24 Feb 2023 21:45:31 +0000 (13:45 -0800)]
[RISCV] Made fsqrtv pseudoinstruction SEW-aware

19 months ago[RISCV] Made division pseudoinstructions SEW-aware
Nitin John Raj [Fri, 24 Feb 2023 21:26:00 +0000 (13:26 -0800)]
[RISCV] Made division pseudoinstructions SEW-aware

19 months ago[RISCV] Made vrgather.vv and vrgatherei16 pseudoinstructions SEW-aware
Nitin John Raj [Fri, 24 Feb 2023 21:31:15 +0000 (13:31 -0800)]
[RISCV] Made vrgather.vv and vrgatherei16 pseudoinstructions SEW-aware

19 months ago[RISCV] Made vcompress pseudoinstruction SEW-aware
Nitin John Raj [Fri, 24 Feb 2023 19:40:38 +0000 (11:40 -0800)]
[RISCV] Made vcompress pseudoinstruction SEW-aware

19 months ago[RISCV][NFC] Remove SEW suffix from pseudoinstructions
Nitin John Raj [Fri, 24 Feb 2023 19:18:55 +0000 (11:18 -0800)]
[RISCV][NFC] Remove SEW suffix from pseudoinstructions

19 months ago[RISCV][NFC] Added possible SEWs associated with a given LMUL
Nitin John Raj [Fri, 24 Feb 2023 19:17:29 +0000 (11:17 -0800)]
[RISCV][NFC] Added possible SEWs associated with a given LMUL

19 months ago[RISCV][NFC] Broke ReadVRGatherVV into ReadVRGatherVV_data and ReadVRGatherVV_index...
Nitin John Raj [Mon, 6 Mar 2023 19:09:02 +0000 (11:09 -0800)]
[RISCV][NFC] Broke ReadVRGatherVV into ReadVRGatherVV_data and ReadVRGatherVV_index to separate the reads for VRGatherVV

Differential Revision: https://reviews.llvm.org/D145406

19 months ago[RISCV][NFC] Renamed [Read/Write]VGather* -> [Read/Write]VRGatherV*
Nitin John Raj [Mon, 6 Mar 2023 18:37:11 +0000 (10:37 -0800)]
[RISCV][NFC] Renamed [Read/Write]VGather* -> [Read/Write]VRGatherV*

Differential Revision: https://reviews.llvm.org/D145402

19 months ago[llvm-jitlink] Remove unnecessary header include left in 01bdd8cffca.
Lang Hames [Fri, 24 Mar 2023 23:14:58 +0000 (16:14 -0700)]
[llvm-jitlink] Remove unnecessary header include left in 01bdd8cffca.

19 months ago[llvm-jitlink] Rename -show-graph option to -show-graphs, make it a regex.
Lang Hames [Fri, 24 Mar 2023 21:43:32 +0000 (14:43 -0700)]
[llvm-jitlink] Rename -show-graph option to -show-graphs, make it a regex.

The original -show-graph option dumped the LinkGraph for all graphs loaded into
the session, but can make it difficult to see small graphs (e.g. reduced test
cases) among the surrounding larger files (especially the ORC runtime).

The new -show-graphs option takes a regex and dumps only those graphs matching
the regex. This allows testcases to specify exactly which graphs to dump.

19 months ago[Fuchsia] Add FUCHSIA_USE_MULTIPLE_DISTRIBUTIONS.
Daniel Thornburgh [Wed, 22 Mar 2023 18:02:41 +0000 (11:02 -0700)]
[Fuchsia] Add FUCHSIA_USE_MULTIPLE_DISTRIBUTIONS.

This flag causes the toolchain distribution to be built using LLVM
CMake's multiple distribution feature. The distribution* family of CMake
targets would be replaced with the toolchain-distribution* family.

This shouldn't otherwise affect the semantics of the build, but it sets
up the ability to split out the LLDB build from the main distribution
used by Fuchsia.

Reviewed By: phosek

Differential Revision: https://reviews.llvm.org/D146665

19 months ago[mlir][sparse] Factoring out LoopEmitter::isValidLevel
wren romano [Fri, 24 Mar 2023 22:24:02 +0000 (15:24 -0700)]
[mlir][sparse] Factoring out LoopEmitter::isValidLevel

Depends On D146674

Reviewed By: aartbik

Differential Revision: https://reviews.llvm.org/D146676

19 months ago[lldb][NFC] Remove outdated TODOs from API headers
Alex Langford [Fri, 24 Mar 2023 22:47:27 +0000 (15:47 -0700)]
[lldb][NFC] Remove outdated TODOs from API headers

There were added when I removed the swig interface files in
662548c82683bd8657a3179afee693c4965a3dfd. However, they mostly meant for
me to better track the differences between the existing API headers and
bindings interfaces. There's nothing actionable about these so I remove
them.

19 months ago[mlir][linalg] Convert input type to accumulator type in im2col patterns
Quinn Dawkins [Sun, 19 Mar 2023 00:14:57 +0000 (20:14 -0400)]
[mlir][linalg] Convert input type to accumulator type in im2col patterns

When the input types don't match the accumulator type in named
convolution ops there is supposed to be a conversion to the accumulator
type before the multiply and accumulate.

Differential Revision: https://reviews.llvm.org/D146824

19 months ago[mlir][vector] Hoist redundant singleton vector transfer reads
harsh-nod [Fri, 24 Mar 2023 17:34:17 +0000 (10:34 -0700)]
[mlir][vector] Hoist redundant singleton vector transfer reads

For singleton transfer reads, we allow hoisting them out
of the enclosing loop if its users are either transfer reads
or memory effect free.

Differential Revision: https://reviews.llvm.org/D146828

19 months ago[mlir][sparse] moving kInvalidId into "detail" namespace
wren romano [Fri, 24 Mar 2023 21:46:07 +0000 (14:46 -0700)]
[mlir][sparse] moving kInvalidId into "detail" namespace

In the next few commits I will be converting the various Merger identifier typedefs into newtypes; and once that's done, the `kInvalidId` constant will only be used internally and therefore does not need to be part of the public `mlir::sparse_tensor` namespace.

Depends On D146673

Reviewed By: aartbik

Differential Revision: https://reviews.llvm.org/D146674

19 months ago[mlir][sparse] Updating TensorExp ctor to catch unknown TensorExp::Kind
wren romano [Fri, 24 Mar 2023 21:34:20 +0000 (14:34 -0700)]
[mlir][sparse] Updating TensorExp ctor to catch unknown TensorExp::Kind

Depends On D146562

Reviewed By: aartbik

Differential Revision: https://reviews.llvm.org/D146673

19 months ago[mlir][sparse] Misc cleanup in Merger.h
wren romano [Fri, 24 Mar 2023 21:24:27 +0000 (14:24 -0700)]
[mlir][sparse] Misc cleanup in Merger.h

* Moving the `Children` class to be nested under `TensorExp`.
* Marking `TensorExp`, `TensorExp::Children`, and `LatPoint` as final.

Depends On D146083

Reviewed By: aartbik

Differential Revision: https://reviews.llvm.org/D146562

19 months ago[mlir][sparse] Updating the `Merger::{exp,lat,set}` methods to return const
wren romano [Fri, 24 Mar 2023 21:17:11 +0000 (14:17 -0700)]
[mlir][sparse] Updating the `Merger::{exp,lat,set}` methods to return const

This helps the `Merger` maintain invariants, as well as clarifying the immutability of the underlying objects (with the one exception of `TensorExp::val`).

Depends On: D146559

Reviewed By: aartbik

Differential Revision: https://reviews.llvm.org/D146083

19 months ago[lldb] Add ability to hide the root name of a value
Dave Lee [Fri, 24 Mar 2023 04:43:32 +0000 (21:43 -0700)]
[lldb] Add ability to hide the root name of a value

When printing a value, allow the root value's name to be elided, without omiting the
names of child values.

At the API level, this adds `SetHideRootName()`, which joins the existing
`SetHideName()` function.

This functionality is used by `dwim-print` and `expression`.

Fixes an issue identified by @jgorbe in https://reviews.llvm.org/D145609.

Differential Revision: https://reviews.llvm.org/D146783

19 months ago[libc][Obvious] Remove a compile opt to x86_64 longjmp in a previous commit.
Siva Chandra Reddy [Fri, 24 Mar 2023 20:24:57 +0000 (20:24 +0000)]
[libc][Obvious] Remove a compile opt to x86_64 longjmp in a previous commit.

The option -fno-omit-frame-pointer was accidentally added to the x86_64
longjmp target. This change not only removes it, but makes it
-fomit-frame-pointer.

19 months agoFix mlir/lib/Bindings/Python/IRTypes.cpp for Float8E4M3B11FNUZType
David Majnemer [Fri, 24 Mar 2023 20:42:47 +0000 (20:42 +0000)]
Fix mlir/lib/Bindings/Python/IRTypes.cpp for Float8E4M3B11FNUZType

19 months ago[APFloat] Add E4M3B11FNUZ
David Majnemer [Thu, 9 Mar 2023 23:10:57 +0000 (23:10 +0000)]
[APFloat] Add E4M3B11FNUZ

X. Sun et al. (https://dl.acm.org/doi/10.5555/3454287.3454728) published
a paper showing that an FP format with 4 bits of exponent, 3 bits of
significand and an exponent bias of 11 would work quite well for ML
applications.

Google hardware supports a variant of this format where 0x80 is used to
represent NaN, as in the Float8E4M3FNUZ format. Just like the
Float8E4M3FNUZ format, this format does not support -0 and values which
would map to it will become +0.

This format is proposed for inclusion in OpenXLA's StableHLO dialect: https://github.com/openxla/stablehlo/pull/1308

As part of inclusion in that dialect, APFloat needs to know how to
handle this format.

Differential Revision: https://reviews.llvm.org/D146441

19 months ago[Support] Fix warnings
Kazu Hirata [Fri, 24 Mar 2023 20:01:52 +0000 (13:01 -0700)]
[Support] Fix warnings

This patch fixes:

  llvm/unittests/Support/ScopedPrinterTest.cpp:519:20: error: unused
  variable 'InfDouble' [-Werror,-Wunused-variable]

  llvm/unittests/Support/ScopedPrinterTest.cpp:520:16: error: unused
  variable 'NaNDouble' [-Werror,-Wunused-variable]

  llvm/unittests/Support/ScopedPrinterTest.cpp:516:15: error: unused
  variable 'NaNFloat' [-Werror,-Wunused-variable]

  llvm/unittests/Support/ScopedPrinterTest.cpp:515:19: error: unused
  variable 'InfFloat' [-Werror,-Wunused-variable]

Since commit fa56e362af475e0758cfb41c42f78db50da7235c has temporarily
disabled tests involving these constants, this patch simply comments
them out instead of removing them.

19 months ago[JITLink] Fix MSVC build error: formatv can't handle support::ulittle16_t values
Stefan Gränitz [Fri, 24 Mar 2023 19:49:54 +0000 (20:49 +0100)]
[JITLink] Fix MSVC build error: formatv can't handle support::ulittle16_t values

The issue was reported with compiler output here: https://reviews.llvm.org/D144083#4219383

19 months ago[mlir][tosa] Improve performance of tosa.transpose constant folding
Spenser Bauman [Fri, 24 Mar 2023 19:42:12 +0000 (19:42 +0000)]
[mlir][tosa] Improve performance of tosa.transpose constant folding

Folding of the tosa.transpose operation is both time and memory
intensive as the underlying ElementsAttr is processed as a sequence of
Attributes. This change attempts operate on the underlying raw data of
the ElementsAttr.

In an example resnet50 network, this change reduces the time spent in
folding transpose ops from 35s to 1.5s.

Reviewed By: GeorgeARM, rsuderman, stellaraccident

Differential Revision: https://reviews.llvm.org/D146526

19 months ago[libc] Use `nvptx_kernel` attribute in NVPTX startup code
Joseph Huber [Fri, 24 Mar 2023 19:45:14 +0000 (14:45 -0500)]
[libc] Use `nvptx_kernel` attribute in NVPTX startup code

Summary:
A recent patch allowed us to emit a callable kernel from freestanding
NVPTX code. This allows us to move away from using the CUDA language.
This has several advantages in that it works around an entire assortment
of errors I was seeing while implementing RPC for Nvidia.

19 months ago[Pseudo Probe] Add the test for probe desc
wlei [Fri, 24 Mar 2023 17:21:38 +0000 (10:21 -0700)]
[Pseudo Probe] Add the test for probe desc

Added a test to https://reviews.llvm.org/D146657, make sure the guid and name are computed using the debug info name.

Reviewed By: hoy, wenlei

Differential Revision: https://reviews.llvm.org/D146826

19 months ago[lldb] Add header REPL.h to lldb_Expression
Alex Langford [Fri, 24 Mar 2023 19:38:50 +0000 (12:38 -0700)]
[lldb] Add header REPL.h to lldb_Expression

As of c5bfa3dafb3e7ccc871734a96b7a9188868d925a, REPL.h no longer has a
private implementation header in it. This TODO and the thing it marks
cdan be removed.

19 months ago[NVPTX] Introduce attribute to mark kernels without a language mode
Joseph Huber [Fri, 24 Mar 2023 18:10:22 +0000 (13:10 -0500)]
[NVPTX] Introduce attribute to mark kernels without a language mode

We may want to be able to mark certain regions as kernels even without
being in an accepted CUDA or OpenCL language mode. This patch introduces
a new attribute limited to `nvptx` targets called `nvptx_kernel` which
will perform the same metadata action as the existing CUDA ones. This
closely mimics the behaviour of the `amdgpu_kernel` attribute. This
allows for making executable NVPTX device images without using an
existing offloading language model.

I was unsure how to do this, I could potentially re-use all the CUDA
attributes and just replace the `CUDA` language requirement with an
`NVPTX` architecture requirement. Also I don't know if I should add more
than just this attribute.

Reviewed By: tra

Differential Revision: https://reviews.llvm.org/D140226

19 months ago[linalg] Fixed tosa-to-linalg-named for tosa.conv2d i8 with i8 bias
SJW [Fri, 24 Mar 2023 19:22:43 +0000 (19:22 +0000)]
[linalg] Fixed tosa-to-linalg-named for tosa.conv2d i8 with i8 bias

Missing sign extension.

Reviewed By: rsuderman

Differential Revision: https://reviews.llvm.org/D145744

19 months agoDon't expect what newlines look like - never works on Windows.
Jim Ingham [Fri, 24 Mar 2023 19:27:33 +0000 (12:27 -0700)]
Don't expect what newlines look like - never works on Windows.

19 months ago[Libomptarget] Update CMake messages if the tests aren't build
Joseph Huber [Fri, 24 Mar 2023 19:25:38 +0000 (14:25 -0500)]
[Libomptarget] Update CMake messages if the tests aren't build

Summary:
These messages have been wrong for quite some time. Update them to be
more descriptive of why the tests weren't built.

19 months ago[clang-tools-extra] Fix linking ClangdTests when using libclang-cpp
Michał Górny [Mon, 20 Mar 2023 15:00:10 +0000 (16:00 +0100)]
[clang-tools-extra] Fix linking ClangdTests when using libclang-cpp

Fix linking ClangdTests to specify the dependency on the private
clangTesting library via target_link_libraries() rather than
clang_target_link_libraries().  The latter uses libclang-cpp when
CLANG_LINK_CLANG_DYLIB is used, and clangTesting is not included
in this library.

This fixes d60d3455eb2b375d026a4aa74c4ba0c38f5d323c.

Differential Revision: https://reviews.llvm.org/D146427

19 months ago[libc] Add support for setjmp and longjmp in riscv
Mikhail R. Gadelha [Fri, 24 Mar 2023 19:13:19 +0000 (16:13 -0300)]
[libc] Add support for setjmp and longjmp in riscv

This patch implements setjmp and longjmp in riscv using inline asm. The
following changes were required:

* Omit frame pointer: otherwise gcc won't allow us to use s0
* Use __attribute__((naked)): otherwise both gcc and clang will generate
function prologue and epilogue in both functions. This doesn't happen
in x86_64, so we guard it to only riscv

Furthermore, using __attribute__((naked)) causes two problems: we
can't use `return 0` (both gcc and clang) and the function arguments in
the function body (clang only), so we had to use a0 and a1 directly.

Reviewed By: sivachandra

Differential Revision: https://reviews.llvm.org/D145584

19 months ago[ConstraintElim] Use GEPOperator instead of GetElementPtrInst.
Florian Hahn [Fri, 24 Mar 2023 19:14:34 +0000 (19:14 +0000)]
[ConstraintElim] Use GEPOperator instead of GetElementPtrInst.

The logic in ConstraintElimination should trivially apply to GEP
constant expressions as well, so update code to deal with GEPOperator
instead.

19 months ago[BOLT] Don't use section relocations when computing hash for data from other section
Denis Revunov [Fri, 24 Mar 2023 18:50:07 +0000 (21:50 +0300)]
[BOLT] Don't use section relocations when computing hash for data from other section

When computing symbol hashes in BinarySection::hash, we try to find relocations
in the section which reference the passed BinaryData. We do so by doing
lower_bound on data begin offset and upper_bound on data end offset. Since
offsets are relative to the current section, if it is a data from the previous
section, we get underflow when computing offset and lower_bound returns
Relocations.end(). If this data also ends where current section begins,
upper_bound on zero offset will return some valid iterator if we have any
relocations after the first byte. Then we'll try to iterate from lower_bound to
upper_bound, since they're not equal, which in that case means we'll dereference
Relocations.end(), increment it, and try to do so until we reach the second
valid iterator. Of course we reach segfault earlier. In this patch we stop BOLT
from searching relocations for symbols outside of the current section.

Reviewed By: rafauler

Differential Revision: https://reviews.llvm.org/D146620

19 months ago[SystemZ] Allow fp/int casting with inline assembly operands.
Jonas Paulsson [Tue, 14 Mar 2023 15:40:10 +0000 (16:40 +0100)]
[SystemZ] Allow fp/int casting with inline assembly operands.

Support bitcasting between int/fp/vector values and 'r'/'f'/'v' inline
assembly operands. This is intended to match GCCs beahvior.

Reviewed By: Ulrich Weigand

Differential Revision: https://reviews.llvm.org/D146059

19 months agoRevert "[Serialization] Place command line defines in the correct file"
Felipe de Azevedo Piovezan [Fri, 24 Mar 2023 18:49:47 +0000 (14:49 -0400)]
Revert "[Serialization] Place command line defines in the correct file"

This reverts commit 72073fc95cd4793a853925ddc8cc3fb2118808a5.

19 months ago[support] Temporarily disable Inf/NaN testing in PrintNumber
Paul Kirth [Fri, 24 Mar 2023 17:41:10 +0000 (17:41 +0000)]
[support] Temporarily disable Inf/NaN testing in PrintNumber

This is still breaking on some platforms. The underlying implementation
doesn't seem to be the cause, rather the test is not robust across
platforms. So, we'll just disable this for the time being, to unblock
builds until we have a proper fix.

Reviewed By: abhina.sreeskantharajan

Differential Revision: https://reviews.llvm.org/D146834

19 months agoAdd support for critical regions in device code.
Doru Bercea [Fri, 10 Mar 2023 23:03:51 +0000 (18:03 -0500)]
Add support for critical regions in device code.

Review: https://reviews.llvm.org/D145831

19 months ago[TSan] Refactor ExternalAccess() to avoid unnecessary pop/push tag [NFC]
Julian Lettner [Wed, 22 Mar 2023 22:07:37 +0000 (15:07 -0700)]
[TSan] Refactor ExternalAccess() to avoid unnecessary pop/push tag [NFC]

* Avoid unnecessary frame & tag push/pops if memory access is ignored
* Rename function and add comment to make it clearer what the code does
* Make helper functions static and move inside `#if !SANITIZER_GO`

Differential Revision: https://reviews.llvm.org/D146670

19 months ago[ARM] Add Thumb Attributes for thumb thunks created in SLSHarding
David Green [Fri, 24 Mar 2023 18:11:54 +0000 (18:11 +0000)]
[ARM] Add Thumb Attributes for thumb thunks created in SLSHarding

Without this the function will be use an Arm subtarget, meaning the
instructions in it will be invalid for the current subtarget.

Differential Revision: https://reviews.llvm.org/D144733

19 months ago[M68k] Fix CConvs for pointer type return values
Min-Yih Hsu [Fri, 24 Mar 2023 17:59:46 +0000 (10:59 -0700)]
[M68k] Fix CConvs for pointer type return values

Put the value into A0 instead of data registers. And remove the
redundant `RetCC_M68kCommon` as there aren't many rules shared between
existing CCs other than the pointer one.
This change is tested by existing tests.

19 months ago[lldb] Explicitly mark PlatformFreeBSD's dependency on PlatformPOSIX
Alex Langford [Fri, 24 Mar 2023 17:55:33 +0000 (10:55 -0700)]
[lldb] Explicitly mark PlatformFreeBSD's dependency on PlatformPOSIX

I accidentally broke the FreeBSD lldb-server build in 0c5cee7 because it
now depends on PlatformFreeBSD. PlatformFreeBSD depends on PlatformPOSIX
but this dependency was not explicitly tracked in CMake. As a result,
the FreeBSD lldb-server build broke.

Credit to John F. Carr <jfc@mit.edu> for pointing out the issue and
providing a fix.

19 months agoFix backtick handling in parsed commands.
Jim Ingham [Fri, 24 Mar 2023 17:38:56 +0000 (10:38 -0700)]
Fix backtick handling in parsed commands.

https://reviews.llvm.org/D146779

19 months ago[NVPTX] Fix integer overflow affecting array size calculation.
Artem Belevich [Thu, 23 Mar 2023 22:27:02 +0000 (15:27 -0700)]
[NVPTX] Fix integer overflow affecting array size calculation.

It turns out, 4GB+ large arrays are a thing and a 32-bit integer is just not
enough to handle them.

Differential Revision: https://reviews.llvm.org/D146767

19 months ago[RISCV] Model select and insertsubvector shuffle kinds
Luke Lau [Thu, 23 Mar 2023 17:45:30 +0000 (17:45 +0000)]
[RISCV] Model select and insertsubvector shuffle kinds

Selects get lowered to a vmerge with a mask, and insertsubvectors get
lowered to a vslideup.

Differential Revision: https://reviews.llvm.org/D146747

19 months ago[RISCV] Add test case for two equivalent reductions
Luke Lau [Fri, 24 Mar 2023 13:46:03 +0000 (13:46 +0000)]
[RISCV] Add test case for two equivalent reductions

They are functionally equivalent but currently one fails to vectorize
because the cost of an insert subvector shuffle is too expensive.
D146747 will update the cost of these types of shuffles, so add a test
case for it.

19 months ago[RISCV] Enable SLP in RISC-V SLP reduction tests
Luke Lau [Fri, 24 Mar 2023 13:43:32 +0000 (13:43 +0000)]
[RISCV] Enable SLP in RISC-V SLP reduction tests

Horizontal reduction can still kick in even when the max VF is set to 0,
but strange stuff can happen as it affects the cost model.
Enable it for these tests as eventually the goal will be to have SLP
enabled.

19 months ago[RISCV] Add test cases for modeling more shuffle kinds
Luke Lau [Fri, 24 Mar 2023 13:39:51 +0000 (13:39 +0000)]
[RISCV] Add test cases for modeling more shuffle kinds

These map to SK_InsertSubvector and SK_Select shuffle kinds

19 months ago[RISCV][NFC] Rename some test cases
Luke Lau [Fri, 24 Mar 2023 15:52:02 +0000 (15:52 +0000)]
[RISCV][NFC] Rename some test cases

Since they no longer involve a merge

19 months ago[RISCV] Lower insert subvector shuffles as vslideups
Luke Lau [Fri, 24 Mar 2023 12:03:41 +0000 (12:03 +0000)]
[RISCV] Lower insert subvector shuffles as vslideups

A shuffle with an insert subvector mask is functionally equivalent to:
(insert_subvector v0, (extract_subvector v1, len), index)
We can emulate by doing a vslideup on v1 into the right index, and
carefully selecting VL so that we don't overwrite any more destination
elements than what we have to.
This avoids the need for a select with a mask.

19 months ago[flang][driver][openmp] Write MLIR for -save-temps
Sergio Afonso [Tue, 14 Mar 2023 17:40:04 +0000 (17:40 +0000)]
[flang][driver][openmp] Write MLIR for -save-temps

This patch adds support for producing MLIR files when using -save-temps on
flang. One MLIR file will be produced before lowering and optimization passes,
containing the operations produced by the PFT-to-MLIR lowering bridge, and
another at the end of the process, just before LLVM IR generation.

This is accomplished by forwarding the -save-temps flag from the driver to the
frontend, and modifying it to output MLIR files accordingly.

Differential Revision: https://reviews.llvm.org/D146075

19 months ago[scudo] Use bytes-in-freelist as a hint of page release
Chia-hung Duan [Fri, 24 Mar 2023 17:06:29 +0000 (17:06 +0000)]
[scudo] Use bytes-in-freelist as a hint of page release

Tracking the pushed bytes between to releaseToOSMaybe calls may lead to
a overestimated case that if we do malloc 2KB -> free 2KB -> malloc 2KB
-> free 2KB, we may think we have released 4KB but it only releases 2KB
actually. Switch to use bytes-in-freelist excludes more cases that can't
release the pages

Reviewed By: cferris

Differential Revision: https://reviews.llvm.org/D146400

19 months agoRemove unused header file in VectorTransformOps.h
yijia1212 [Fri, 24 Mar 2023 17:05:07 +0000 (10:05 -0700)]
Remove unused header file in VectorTransformOps.h

Remove unused header file in VectorTransformOps.h

Differential Revision: https://reviews.llvm.org/D146825

19 months agoFix warnings
Kazu Hirata [Fri, 24 Mar 2023 16:52:17 +0000 (09:52 -0700)]
Fix warnings

This patch fixes:

  clang/lib/Driver/ToolChains/OHOS.cpp:410:18: warning: unused
  variable ‘A’ [-Wunused-variable]

  clang-tools-extra/clangd/refactor/tweaks/AddUsing.cpp:293:15:
  warning: unused variable ‘II’ [-Wunused-variable]

  llvm/lib/ProfileData/RawMemProfReader.cpp:364:68: warning: suggest
  parentheses around ‘&&’ within ‘||’ [-Wparentheses]

19 months agoMake test more explicit on failure.
Doru Bercea [Fri, 24 Mar 2023 14:18:27 +0000 (10:18 -0400)]
Make test more explicit on failure.

Patch: https://reviews.llvm.org/D146812

19 months ago[support] Attempt to fix PrintNumber test for Solaris
Paul Kirth [Fri, 24 Mar 2023 16:02:49 +0000 (16:02 +0000)]
[support] Attempt to fix PrintNumber test for Solaris

NaN and Inf are still causing some problems in a formatting test.
This patch makes the checked format string exactly match the internal
JSON format string. If there are still problems, we should disable
testing Inf and NaN values until we can come to a portable solution.

Reviewed By: abhina.sreeskantharajan

Differential Revision: https://reviews.llvm.org/D146818

19 months ago[extract_symbols.py] Export unique symbols
Tomas Matheson [Thu, 23 Mar 2023 15:18:34 +0000 (15:18 +0000)]
[extract_symbols.py] Export unique symbols

As described in a comment on D140637, PluginInlineOrderTest.NoInlineFoo
will fail with:

InlineOrderPlugin.so: undefined symbol: _ZN4llvm13AllAnalysesOnINS_6ModuleEE6SetKeyE

The symbol is unique in libLLVMCore and ends up local in AnalysisTests.
Fix this by exporting all unique symbols found in libraries.

AnalysisTests.symbols change in line count: 12464 -> 12499

Differential Revision: https://reviews.llvm.org/D146731