platform/upstream/llvm.git
2 years ago[ARM][AArch64] Introduce qrdmlah and qrdmlsh intrinsics
David Green [Thu, 27 Jan 2022 19:19:46 +0000 (19:19 +0000)]
[ARM][AArch64] Introduce qrdmlah and qrdmlsh intrinsics

Since it's introduction, the qrdmlah has been represented as a qrdmulh
and a sadd_sat. This doesn't produce the same result for all input
values though. This patch fixes that by introducing a qrdmlah (and
qrdmlsh) intrinsic specifically for the vqrdmlah and sqrdmlah
instructions. The old test cases will now produce a qrdmulh and sqadd,
as expected.

Fixes #53120 and #50905 and #51761.

Differential Revision: https://reviews.llvm.org/D117592

2 years agoTrivial update for debug location in LIT test.
Malhar Jajoo [Thu, 27 Jan 2022 19:07:47 +0000 (19:07 +0000)]
Trivial update for debug location in LIT test.

This just updates debug location of a loop in a LIT test to point
to the correct source line.

2 years ago[ARM][AArch64] Cleanup and autogenerate v8.1a vqdrmlah tests. NFC
David Green [Thu, 27 Jan 2022 18:43:06 +0000 (18:43 +0000)]
[ARM][AArch64] Cleanup and autogenerate v8.1a vqdrmlah tests. NFC

2 years ago[ConstraintElimination] Use constraints with 0 or 1 coefficients.
Florian Hahn [Thu, 27 Jan 2022 18:41:33 +0000 (18:41 +0000)]
[ConstraintElimination] Use constraints with 0 or 1 coefficients.

isConditionImplied is able to correctly handle 0 or 1 coefficients, so
let it handle those cases, rather than skipping them.

2 years ago[flang] Expand the semantics test for co_sum
Damian Rouson [Thu, 9 Dec 2021 22:23:20 +0000 (14:23 -0800)]
[flang] Expand the semantics test for co_sum

Increase the coverage of standard-conforming and non-conforming
co_sum calls.

Reviewed By: ktras

2 years ago[libc] Guard sqrt_80_bit_long_double.h header to not be included with aarch64.
Tue Ly [Thu, 27 Jan 2022 17:12:38 +0000 (12:12 -0500)]
[libc] Guard sqrt_80_bit_long_double.h header to not be included with aarch64.

Guard sqrt_80_bit_long_double.h header to not be included with aarch64.

Reviewed By: sivachandra

Differential Revision: https://reviews.llvm.org/D118389

2 years ago[X86] Fold MOVMSK(CONCAT(X,Y)) -> MOVMSK(AND/OR(X,Y)) for all_of/any_of patterns
Simon Pilgrim [Thu, 27 Jan 2022 17:58:29 +0000 (17:58 +0000)]
[X86] Fold MOVMSK(CONCAT(X,Y)) -> MOVMSK(AND/OR(X,Y)) for all_of/any_of patterns

Makes it easier for later folds and avoids unnecessary 256-bit ops (especially on AVX1-only targets where we miss a lot of integer instructions)

2 years ago[test] Use -passes syntax in SROA test cases
Bjorn Pettersson [Wed, 26 Jan 2022 17:15:53 +0000 (18:15 +0100)]
[test] Use -passes syntax in SROA test cases

Another step to move away from the legacy PM syntax when specifying
passes in opt.

Differential Revision: https://reviews.llvm.org/D118339

2 years ago[NewPM] Add debugify and check-debugify to the PassRegistry
Bjorn Pettersson [Thu, 27 Jan 2022 15:09:24 +0000 (16:09 +0100)]
[NewPM] Add debugify and check-debugify to the PassRegistry

Adding -debugify and -check-debugify in the PassRegistry will make
sure the passes are listed properly by -print-pipeline-passes as
well as -print-passes.

It also allows removal of the custom pipeline parsing callback that
has been used in the NewPMDriver.

Differential Revision: https://reviews.llvm.org/D118369

2 years ago[AMDGPU] SILoadStoreOptimizer: reject AGPR DS_WRITE sooner
Jay Foad [Thu, 27 Jan 2022 14:27:39 +0000 (14:27 +0000)]
[AMDGPU] SILoadStoreOptimizer: reject AGPR DS_WRITE sooner

Rejecting AGPR DS_WRITE instructions before adding them to any mergeable
list seems cleaner than adding them to the list and rejecting them
later.

Differential Revision: https://reviews.llvm.org/D118368

2 years ago[AMDGPU] SILoadStoreOptimizer: use separate lists for AGPR instructions
Jay Foad [Thu, 27 Jan 2022 14:48:08 +0000 (14:48 +0000)]
[AMDGPU] SILoadStoreOptimizer: use separate lists for AGPR instructions

Using separate lists for AGPR and non-AGPR instructions seems like a
cleaner solution than putting them all in the same list and then later
refusing to merge instructions of different AGPR-ness.

Differential Revision: https://reviews.llvm.org/D118367

2 years ago[AMDGPU] SILoadStoreOptimizer: tweak API of CombineInfo::setMI. NFC.
Jay Foad [Thu, 27 Jan 2022 14:39:42 +0000 (14:39 +0000)]
[AMDGPU] SILoadStoreOptimizer: tweak API of CombineInfo::setMI. NFC.

Change CombineInfo::setMI to take a reference to the
SILoadStoreOptimizer instance, for easy access to common fields like
TII and STM.

Differential Revision: https://reviews.llvm.org/D118366

2 years ago[InstrProf][Correlate] Verify debug info with llvm-profdata show
Ellis Hoag [Wed, 26 Jan 2022 17:59:57 +0000 (09:59 -0800)]
[InstrProf][Correlate] Verify debug info with llvm-profdata show

Use the `llvm-profdata show` command to verify debug info for profile correlation using the `--debug-info` option.

Reviewed By: kyulee

Differential Revision: https://reviews.llvm.org/D118181

2 years ago[lldb/test] Disable test_launch_scripted_process_stack_frames on x86_64
Med Ismail Bennani [Thu, 27 Jan 2022 17:39:55 +0000 (18:39 +0100)]
[lldb/test] Disable test_launch_scripted_process_stack_frames on x86_64

There seems to be an issue on x86_64 when launching a ScriptdProcess.
This disables temporarely the test that causes the bot to timeout until
I finish investigating the issue.

Signed-off-by: Med Ismail Bennani <medismail.bennani@gmail.com>
2 years ago[libc][NFC] Move the POSIX write function to the default build on linux.
Siva Chandra Reddy [Thu, 27 Jan 2022 17:11:06 +0000 (17:11 +0000)]
[libc][NFC] Move the POSIX write function to the default build on linux.

2 years ago[RISCV] Use const reference when looping over Exts in RISCVISAInfo.
Craig Topper [Thu, 27 Jan 2022 08:17:33 +0000 (00:17 -0800)]
[RISCV] Use const reference when looping over Exts in RISCVISAInfo.

Exts is a map of keyed by std::string with a extension info as
a value. Making copies of this wouldn't be cheap.

We had a mix of references and copies. This makes everything
consistently use a const reference to make it clear we aren't
modifying it.

Differential Revision: https://reviews.llvm.org/D118326

2 years ago[ConstraintElimination] Add additional GEP arithmetic tests.
Florian Hahn [Thu, 27 Jan 2022 17:15:43 +0000 (17:15 +0000)]
[ConstraintElimination] Add additional GEP arithmetic tests.

2 years ago[libc][NFC] Let var args be treated as a named type in the spec.
Siva Chandra Reddy [Thu, 27 Jan 2022 16:59:08 +0000 (16:59 +0000)]
[libc][NFC] Let var args be treated as a named type in the spec.

The type names in the spec are just sugar used by the header generator to
generate the function prototype. Giving the VarArgType a name of "..."
allows the header generator to treat and generate the "type name" for var
arg parameters similar to how it does for any other type.

2 years ago[clang-format] Fix AllowShortFunctionsOnASingleLine: InlineOnly with wrapping after...
Marek Kurdej [Thu, 27 Jan 2022 16:54:58 +0000 (17:54 +0100)]
[clang-format] Fix AllowShortFunctionsOnASingleLine: InlineOnly with wrapping after record.

Fixes https://github.com/llvm/llvm-project/issues/53430.

Initially, I had a quick and dirty approach, but it led to a myriad of special cases handling comments (that may add unwrapped lines).
So I added TT_RecordLBrace type annotations and it seems like a much nicer solution.
I think that in the future it will allow us to clean up some convoluted code that detects records.

Reviewed By: MyDeveloperDay, HazardyKnusperkeks

Differential Revision: https://reviews.llvm.org/D118337

2 years ago[libc] Refactor sqrt implementations and add tests for generic sqrt implementations.
Tue Ly [Tue, 25 Jan 2022 20:11:15 +0000 (15:11 -0500)]
[libc] Refactor sqrt implementations and add tests for generic sqrt implementations.

Refactor sqrt implementations:
- Move architecture specific instructions from `src/math/<arch>` to `src/__support/FPUtil/<arch>` folder.
- Move generic implementation of `sqrt` to `src/__support/FPUtil/generic` folder and add it as a header library.
- Use `src/__support/FPUtil/sqrt.h` for architecture/generic selections.
- Add unit tests for generic implementation of `sqrt`.

Reviewed By: sivachandra

Differential Revision: https://reviews.llvm.org/D118173

2 years ago[LTO] An option to disable automatic bitcode upgrade
Hongtao Yu [Wed, 26 Jan 2022 19:29:00 +0000 (11:29 -0800)]
[LTO] An option to disable automatic bitcode upgrade

Bitcode upgrade automatically takes in place when the current compiler version doesn't match the one building the input modules. The upgrade is very expansive for large applications and unnecessary when users are sure the two compilers do not have a version issue, even if their versions mismatch literally. An optional to disable the automatic bitcode upgrade will be handy in such case.

Reviewed By: wenlei

Differential Revision: https://reviews.llvm.org/D109869

2 years ago[clang-tidy] Add more documentation about check development (NFC)
Richard [Sat, 22 Jan 2022 00:57:40 +0000 (17:57 -0700)]
[clang-tidy] Add more documentation about check development (NFC)

- Mention pp-trace
- CMake configuration
- Overriding registerPPCallbacks
- Overriding isLanguageVersionSupported
- Check development tips
  - Guide to useful documentation
  - Using the Transformer library
  - Developing your check incrementally
  - Creating private matchers
  - Unit testing helper code
  - Making your check robust
  - Documenting your check
- Describe the Inputs test folder

Differential Revision: https://reviews.llvm.org/D117939

2 years ago[mlir][bufferize] Move arith BufferizableOpInterface impl to arith dialect
Matthias Springer [Thu, 27 Jan 2022 16:11:22 +0000 (01:11 +0900)]
[mlir][bufferize] Move arith BufferizableOpInterface impl to arith dialect

Also switch the implementation of `-arith-bufferize` to BufferizableOpInterface.

Differential Revision: https://reviews.llvm.org/D118325

2 years agoDon't trigger unused-parameter warnings on naked functions
MuAlphaOmegaEpsilon [Thu, 27 Jan 2022 16:39:05 +0000 (11:39 -0500)]
Don't trigger unused-parameter warnings on naked functions

This commit checks if a function is marked with the naked attribute
and, if it is, will silence the emission of any unused-parameter
warning.

Inside a naked function only the usage of basic ASM instructions is
expected. In this context the parameters can actually be used by
fetching them according to the underlying ABI. Since parameters might
be used through ASM instructions, the linter and the compiler will have
a hard time understanding if one of those is unused or not, therefore
no unused-parameter warning should ever be triggered whenever a
function is marked naked.

2 years ago[libc++][nfc] Add TEST_HAS_NO_INT128.
Mark de Wever [Sun, 23 Jan 2022 17:23:41 +0000 (18:23 +0100)]
[libc++][nfc] Add TEST_HAS_NO_INT128.

Avoid using the libc++ internal `_LIBCPP_HAS_NO_INT128` in our tests.

Reviewed By: #libc, ldionne

Differential Revision: https://reviews.llvm.org/D117992

2 years ago[libc++][nfc] Include test_macros.h in more tests.
Mark de Wever [Mon, 24 Jan 2022 18:04:06 +0000 (19:04 +0100)]
[libc++][nfc] Include test_macros.h in more tests.

This should fix the regressions detected in D117992.

This lands before D117992 to avoid breaking main.

Reviewed By: #libc, ldionne

Differential Revision: https://reviews.llvm.org/D118056

2 years agoAMDGPU: Handle addrspacecast of constant 32-bit to flat
Matt Arsenault [Thu, 27 Jan 2022 02:57:36 +0000 (21:57 -0500)]
AMDGPU: Handle addrspacecast of constant 32-bit to flat

I accidentally made this work on the GlobalISel path, and there's no
real reason not to handle this.

2 years agoAMDGPU: Fix broken check lines in test
Matt Arsenault [Thu, 27 Jan 2022 02:57:24 +0000 (21:57 -0500)]
AMDGPU: Fix broken check lines in test

2 years ago[AArch64][SVE] Folds VSELECT if the predicate is all active.
Sander de Smalen [Thu, 27 Jan 2022 13:01:07 +0000 (13:01 +0000)]
[AArch64][SVE] Folds VSELECT if the predicate is all active.

This adds the following changes:

* Fold: vselect(<all active predicate>, x, y) => x
* Extend isAllActivePredicate to take vscale_range into account, e.g.
  isAllActivePredicate(vl16) for nxv16i1 and vscale == 1 => true.
  isAllActivePredicate(vl32) for nxv16i1 and vscale == 2 => true.

Differential Revision: https://reviews.llvm.org/D118147

2 years ago[libc++][NFC] Move some functions from directory_iterator.cpp to filesystem_common.h
Louis Dionne [Wed, 26 Jan 2022 16:05:47 +0000 (11:05 -0500)]
[libc++][NFC] Move some functions from directory_iterator.cpp to filesystem_common.h

This move is going to be needed in order to reuse `posix_readdir` in
another translation unit. This doesn't change any of the code except
for removing an unused function parameter that otherwise triggers a
warning inside our tests.

Differential Revision: https://reviews.llvm.org/D118254

2 years ago[libc++] Fix bug in ranges::advance
Louis Dionne [Thu, 13 Jan 2022 17:16:30 +0000 (12:16 -0500)]
[libc++] Fix bug in ranges::advance

In `ranges::advance(iter, n, bound)`, we'd incorrectly handle the case
where bound < iter and n is 0:

    int a[10];
    int *p = a+5;
    int *bound = a+3;
    std::ranges::advance(p, 0, bound);
    assert(p - a == 5); // we'd return 3 before this patch

This was caused by an incorrect handling of 0 inside __magnitude_geq.

Differential Revision: https://reviews.llvm.org/D117240

2 years ago[libc++] Fix common_iterator for output_iterators
Louis Dionne [Tue, 18 Jan 2022 17:10:14 +0000 (12:10 -0500)]
[libc++] Fix common_iterator for output_iterators

We were missing a constraint in common_iterator's iterator_traits and
we were eagerly instantiating iter_value_t even when invalid.

Thanks to Casey Carter for finding this bug.

Differential Revision: https://reviews.llvm.org/D117449

2 years ago[PowerPC][AIX] Support toc-data attribute for read-only globals.
Yousuf Ali [Thu, 27 Jan 2022 14:57:04 +0000 (09:57 -0500)]
[PowerPC][AIX] Support toc-data attribute for read-only globals.

The patch handles the addition of constant global variables to the table
of contents.

Differential Revision: https://reviews.llvm.org/D116181

2 years agoAMDGPU/GlobalISel: Fix assert on invalid cond code for llvm.amdgcn.icmp
Matt Arsenault [Tue, 18 Jan 2022 23:37:27 +0000 (18:37 -0500)]
AMDGPU/GlobalISel: Fix assert on invalid cond code for llvm.amdgcn.icmp

2 years ago[flang] split character procedure arguments in target-rewrite pass
Jean Perier [Thu, 27 Jan 2022 13:59:38 +0000 (14:59 +0100)]
[flang] split character procedure arguments in target-rewrite pass

When passing a character procedure as a dummy procedure, the result
length must be passed along the function address. This is to cover
the cases where the dummy procedure is declared with assumed length
inside the scope that will call it (it will need the length to allocate
the result on the caller side).

To be compatible with other Fortran compiler, this length must be
appended after all other argument just like character objects
(fir.boxchar).

A fir.boxchar cannot be used to implement this feature because it
is meant to take an object address, not a function address.

Instead, argument like `tuple<function type, integer type> {fir.char_proc}`
will be recognized as being character dummy procedure in FIR. That way
lowering does not have to do the argument split.

This patch adds tools in Character.h to create this type and tuple
values as well as to recognize them and extract its tuple members.

It also updates the target rewrite pass to split these arguments like
fir.boxchar.

This part is part of fir-dev upstreaming. It was reviwed previously
in: https://github.com/flang-compiler/f18-llvm-project/pull/1393

Differential Revision: https://reviews.llvm.org/D118108

2 years agoAMDGPU/GlobalISel: Fix flat_scratch_init handling for shaders
Matt Arsenault [Tue, 18 Jan 2022 22:45:51 +0000 (17:45 -0500)]
AMDGPU/GlobalISel: Fix flat_scratch_init handling for shaders

I don't think this is actually defined for mesa, but this is what we
were doing on the DAG path.

2 years agoclang-format: [JS] sort import aliases.
Martin Probst [Thu, 27 Jan 2022 14:15:30 +0000 (15:15 +0100)]
clang-format: [JS] sort import aliases.

Users can define aliases for long symbols using import aliases:

    import X = A.B.C;

Previously, these were unhandled and would terminate import sorting.
With this change, aliases sort as their own group, coming last after all
other imports.

Aliases are not sorted within their group, as they may reference each
other, so order is significant.

Revision URI: https://reviews.llvm.org/D118361

2 years ago[X86][SSE] Add combineBitOpWithShift for BITOP(SHIFT(X,Z),SHIFT(Y,Z)) -> SHIFT(BITOP...
Simon Pilgrim [Thu, 27 Jan 2022 14:54:16 +0000 (14:54 +0000)]
[X86][SSE] Add combineBitOpWithShift for BITOP(SHIFT(X,Z),SHIFT(Y,Z)) -> SHIFT(BITOP(X,Y),Z) vector folds

InstCombine performs this more generally with SimplifyUsingDistributiveLaws, but we don't need anything that complex here - this is mainly to fix up cases where logic ops get created late on during lowering, often in conjunction with sext/zext ops for type legalization.

https://alive2.llvm.org/ce/z/gGpY5v

2 years ago[AMDGPU] SILoadStoreOptimizer: Allow merging across a swizzled access
Jay Foad [Wed, 26 Jan 2022 17:09:38 +0000 (17:09 +0000)]
[AMDGPU] SILoadStoreOptimizer: Allow merging across a swizzled access

Swizzled accesses are not merged, but there is no particular reason not
to merge two instructions if any of the intervening instructions happens
to be a swizzled access.

This moves the check for swizzled accesses out of checkAndPrepareMerge
into collectMergeableInsts where I think it makes more sense.

Differential Revision: https://reviews.llvm.org/D118267

2 years ago[AArch64] Add vector compare/select tests with UNE predicate.
Florian Hahn [Thu, 27 Jan 2022 14:20:30 +0000 (14:20 +0000)]
[AArch64] Add vector compare/select tests with UNE predicate.

Precommit some additional tests for D118256.

2 years ago[lldb] Delete TestBacktraceAll.py
Pavel Labath [Thu, 27 Jan 2022 13:55:49 +0000 (14:55 +0100)]
[lldb] Delete TestBacktraceAll.py

This test is completely nondeterministic, environment-dependent and does
not test what it was supposed to test (reverting the associated patch
does not make it fail).

I tried to figure out what the patch was meant to fix to see if I can
create a better test with the current tools, but I was not able to
understand the problem (it sounds like it has something to do with local
classes, but I don't understand the details).

2 years ago[ConstraintSystem] Mark function as const (NFC).
Florian Hahn [Thu, 27 Jan 2022 13:44:47 +0000 (13:44 +0000)]
[ConstraintSystem] Mark function as const (NFC).

2 years ago[OMPIRBuilder] Avoid pointer element type access
Nikita Popov [Thu, 27 Jan 2022 13:33:29 +0000 (14:33 +0100)]
[OMPIRBuilder] Avoid pointer element type access

Use isOpaqueOrPointeeTypeEquals() instead.

2 years ago[ConstraintElimination] Use simplified constraint for == 0.
Florian Hahn [Thu, 27 Jan 2022 13:31:23 +0000 (13:31 +0000)]
[ConstraintElimination] Use simplified constraint for == 0.

When checking x == 0, checking x u<= 0 is sufficient and simpler than
x u>= 0 && x u<= 0.

https://alive2.llvm.org/ce/z/btM7d3

    ----------------------------------------
    define i1 @src(i4 %a) {
    %0:
      %c = icmp eq i4 %a, 0
      ret i1 %c
    }
    =>
    define i1 @tgt(i4 %a) {
    %0:
       %c = icmp ule i4 %a, 0
       ret i1 %c
    }
    Transformation seems to be correct!

2 years ago[X86] Fold TESTZ(OR(LO(X),HI(X)),OR(LO(Y),HI(Y))) -> TESTZ(X,Y)
Simon Pilgrim [Thu, 27 Jan 2022 13:20:36 +0000 (13:20 +0000)]
[X86] Fold TESTZ(OR(LO(X),HI(X)),OR(LO(Y),HI(Y))) -> TESTZ(X,Y)

Helps fix a number of poor codegen cases for allof(cmp()) with 256-bit vectors on AVX1

2 years ago[X86] Add test showing failure to combine 128-bit PTEST of split vectors
Simon Pilgrim [Thu, 27 Jan 2022 12:56:50 +0000 (12:56 +0000)]
[X86] Add test showing failure to combine 128-bit PTEST of split vectors

2 years ago[AArch64][SVE] Avoid using ptrue for unpredicated predicate AND.
Sander de Smalen [Thu, 27 Jan 2022 12:15:12 +0000 (12:15 +0000)]
[AArch64][SVE] Avoid using ptrue for unpredicated predicate AND.

Reviewed By: david-arm

Differential Revision: https://reviews.llvm.org/D118146

2 years ago[mlir][BufferOptimization] Use datalayout instead of a flag to find index size
Benjamin Kramer [Thu, 27 Jan 2022 12:02:34 +0000 (13:02 +0100)]
[mlir][BufferOptimization] Use datalayout instead of a flag to find index size

This has the additional advantage of supporting more types.

Differential Revision: https://reviews.llvm.org/D118348

2 years ago[mlir][complex] Lower complex.constant to LLVM
Benjamin Kramer [Thu, 27 Jan 2022 12:08:08 +0000 (13:08 +0100)]
[mlir][complex] Lower complex.constant to LLVM

This fixes a regression from 480cd4cb8560532e544fc0c234749912dde759c6

Differential Revision: https://reviews.llvm.org/D118347

2 years ago[ConstraintElimination] Introduce struct to manage constraints. (NFC)
Florian Hahn [Thu, 27 Jan 2022 12:40:09 +0000 (12:40 +0000)]
[ConstraintElimination] Introduce struct to manage constraints. (NFC)

This patch adds a struct to manage a list of constraints. It simplifies
a follow-up change, that adds pre-conditions that must hold before a
list of constraints can be used.

2 years ago[lldb] [gdb-remote] Support getting siginfo via API
Michał Górny [Mon, 24 Jan 2022 17:52:49 +0000 (18:52 +0100)]
[lldb] [gdb-remote] Support getting siginfo via API

Add Thread::GetSiginfo() and SBThread::GetSiginfo() methods to retrieve
the siginfo value from server.

Differential Revision: https://reviews.llvm.org/D118055

2 years ago[lldb] [Platform] Support synthesizing siginfo_t
Michał Górny [Wed, 19 Jan 2022 19:16:07 +0000 (20:16 +0100)]
[lldb] [Platform] Support synthesizing siginfo_t

Support synthesizing the siginfo_t type from the Platform plugin.
This type is going to be used by LLDB client to process the raw siginfo
data received from lldb-server without the necessity of relying
on target's debug info being present.

Differential Revision: https://reviews.llvm.org/D117707

2 years ago[AMDGPU] Precommit test for swizzled store aliasing two loads
Jay Foad [Thu, 27 Jan 2022 12:10:48 +0000 (12:10 +0000)]
[AMDGPU] Precommit test for swizzled store aliasing two loads

2 years ago[Test] Add data layout to relevant tests + some wide-typed tests
Max Kazantsev [Thu, 27 Jan 2022 12:05:23 +0000 (19:05 +0700)]
[Test] Add data layout to relevant tests + some wide-typed tests

2 years ago[clangd][Hover] Suppress initializers with many tokens
Kadir Cetinkaya [Wed, 26 Jan 2022 16:55:52 +0000 (17:55 +0100)]
[clangd][Hover] Suppress initializers with many tokens

This results in excessive memory usage and eats a lot of screen estate.
Especially in the cases with lots of nested macro calls.

This patch tries to remedy it before the release cut by suppressing the
initializers. For better UX we should probably update the expression printer to
truncate those (behind some policy).

Fixes https://github.com/clangd/clangd/issues/917

Differential Revision: https://reviews.llvm.org/D118260

2 years ago[IR] Support ifuncs in opaque pointer mode
Nikita Popov [Thu, 27 Jan 2022 11:46:47 +0000 (12:46 +0100)]
[IR] Support ifuncs in opaque pointer mode

Relax the type assertion for opaque pointers, and enumerate the
value type in TypeFinder and ValueEnumerator.

2 years ago[Support] [Windows] Don't cancel delete if we failed to set delete
Shezan Baig [Thu, 27 Jan 2022 11:57:38 +0000 (13:57 +0200)]
[Support] [Windows] Don't cancel delete if we failed to set delete

Following up on commit 177176f75c6fa3f624d6d964b9d340ce39511565, if we
failed to setDeleteDisposition(true) during TempFile creation, then
don't try to setDeleteDisposition(false) during TempFile::keep, since it
will likely fail as well.

Instead of letting TempFile::keep just fail, we should let it go ahead
and try renaming the file.

This fixes an issue we are seeing when running clang-cl.exe through the
Incredibuild distributed build system.  We're seeing that renaming
temporary object files would fail here:
https://github.com/llvm/llvm-project/blob/5c1f7b296ac0dddeca02891976e6ab5cfc006719/clang/lib/Frontend/CompilerInstance.cpp#L789

Reviewed By: mstorsjo

Differential Revision: https://reviews.llvm.org/D118212

2 years agoRevert "[mlir][bufferize] Insert memref.cast ops during finalizing pass"
Matthias Springer [Thu, 27 Jan 2022 11:45:06 +0000 (20:45 +0900)]
Revert "[mlir][bufferize] Insert memref.cast ops during finalizing pass"

This reverts commit 1043107ce5e2dee38f6a9bf459549a75f78a83b2.

This commit caused a breakage in `finalizing-bufferize.mlir`.

2 years ago[AArch64][SVE] Avoid using ptrue for ptest in VECREDUCE_OR.
Sander de Smalen [Thu, 27 Jan 2022 11:01:56 +0000 (11:01 +0000)]
[AArch64][SVE] Avoid using ptrue for ptest in VECREDUCE_OR.

Reviewed By: david-arm

Differential Revision: https://reviews.llvm.org/D118145

2 years ago[IR] Handle opaque pointers in PtrToArgument mangling
Nikita Popov [Thu, 27 Jan 2022 11:36:25 +0000 (12:36 +0100)]
[IR] Handle opaque pointers in PtrToArgument mangling

It appears that this mangling type is currently unused. Make it
compatible with opaque pointers in case it becomes used again...

2 years ago[InstCombine] Avoid pointer element type access in PointerReplacer
Nikita Popov [Thu, 27 Jan 2022 11:26:32 +0000 (12:26 +0100)]
[InstCombine] Avoid pointer element type access in PointerReplacer

This code replaces the address space of the pointers while keeping
the element type. Use the appropriate helpers to make this work
with opaque pointers.

2 years ago[OpenCL] Add support of __opencl_c_device_enqueue feature macro.
Anton Zabaznov [Mon, 13 Dec 2021 15:43:37 +0000 (18:43 +0300)]
[OpenCL] Add support of __opencl_c_device_enqueue feature macro.

This feature requires support of __opencl_c_generic_address_space and
__opencl_c_program_scope_global_variables so diagnostics for that is provided as well.

Reviewed By: Anastasia

Differential Revision: https://reviews.llvm.org/D115640

2 years ago[libcxx] [test] Narrow down a MinGW bug workaround in rand.dist.uni.int/eval.pass.cpp
Martin Storsjö [Thu, 20 Jan 2022 22:33:56 +0000 (22:33 +0000)]
[libcxx] [test] Narrow down a MinGW bug workaround in rand.dist.uni.int/eval.pass.cpp

Differential Revision: https://reviews.llvm.org/D118235

2 years ago[AArch64][SVE] Implement missing lowering for extract_subvector for predicates.
Sander de Smalen [Thu, 27 Jan 2022 10:31:10 +0000 (10:31 +0000)]
[AArch64][SVE] Implement missing lowering for extract_subvector for predicates.

Reviewed By: efriedma

Differential Revision: https://reviews.llvm.org/D118057

2 years ago[DAG] SelectionDAG::getNode(N1,N2) - detect N2 constant vector splats as well as...
Simon Pilgrim [Thu, 27 Jan 2022 10:59:03 +0000 (10:59 +0000)]
[DAG] SelectionDAG::getNode(N1,N2) - detect N2 constant vector splats as well as scalars

We already perform some basic folds (add/sub with zero etc.) on scalar types, this patch adds some basic support for constant splats as well in a few cases (we can add more with future test coverage).

In the cases I've enabled, we can handle buildvector implicit truncation as we're not creating new constant nodes from the vector types - we're just returning existing nodes. This allows us to get a number of extra cases in the aarch64 tests.

I haven't enabled support for undefs in buildvector splats, as we're often checking for zero/allones patterns that return the original constant and we shouldn't be returning undef elements in some of these cases - we can enable this later if we're OK with creating new constants.

Differential Revision: https://reviews.llvm.org/D118264

2 years ago[AMDGPU] Update MachineMemOperands syntax in commented out tests
Jay Foad [Thu, 27 Jan 2022 10:56:35 +0000 (10:56 +0000)]
[AMDGPU] Update MachineMemOperands syntax in commented out tests

2 years ago[AMDGPU] Remove unused CI check lines
Jay Foad [Thu, 27 Jan 2022 10:52:59 +0000 (10:52 +0000)]
[AMDGPU] Remove unused CI check lines

2 years ago[ConstantFold] Disable gep of array bitcast fold with opaque pointers
Nikita Popov [Thu, 27 Jan 2022 10:50:27 +0000 (11:50 +0100)]
[ConstantFold] Disable gep of array bitcast fold with opaque pointers

Once again, this fold is meaningless with opaque pointers, as there
is no pointer element type to canonicalize. At some point, we may
want to do GEP type canonicalizations.

2 years ago[mlir][linalg][bufferize] Support arith.index_cast bufferization
Matthias Springer [Thu, 27 Jan 2022 10:37:58 +0000 (19:37 +0900)]
[mlir][linalg][bufferize] Support arith.index_cast bufferization

This is in preparation of switching `-tensor-constant-bufferize` and `-arith-bufferize` to BufferizableOpInterface-based implementations.

Differential Revision: https://reviews.llvm.org/D118324

2 years ago[AMDGPU] Remove unused GFX6 check lines
Jay Foad [Thu, 27 Jan 2022 10:48:15 +0000 (10:48 +0000)]
[AMDGPU] Remove unused GFX6 check lines

2 years ago[InstCombine] Mark element type access as non-opaque (NFC)
Nikita Popov [Thu, 27 Jan 2022 10:39:34 +0000 (11:39 +0100)]
[InstCombine] Mark element type access as non-opaque (NFC)

Also make the function static to make it more obvious that it is
only used in the one place.

2 years ago[OpenCL] opencl-c.h: add missing read_write image guards
Sven van Haastregt [Thu, 27 Jan 2022 10:33:12 +0000 (10:33 +0000)]
[OpenCL] opencl-c.h: add missing read_write image guards

The get_image_num_mip_levels overloads that take a read_write image
parameter were missing the __opencl_c_read_write_images guard.

2 years ago[mlir][tensor] Replace tensor-bufferize with BufferizableOpInterface impl
Matthias Springer [Thu, 27 Jan 2022 10:18:59 +0000 (19:18 +0900)]
[mlir][tensor] Replace tensor-bufferize with BufferizableOpInterface impl

This commit switches the `tensor-bufferize` pass over to BufferizableOpInterface-based bufferization.

Differential Revision: https://reviews.llvm.org/D118246

2 years ago[AArch64][SVE] Implement PFALSE with explicit AArch64ISD node.
Sander de Smalen [Thu, 27 Jan 2022 09:17:29 +0000 (09:17 +0000)]
[AArch64][SVE] Implement PFALSE with explicit AArch64ISD node.

The ISel patterns for PFALSE helps recognise the instructions as being
free of side-effects, which helps MachineCSE remove redundant
PFALSE instructions.

Reviewed By: dmgreen

Differential Revision: https://reviews.llvm.org/D118054

2 years ago[CMake][Fuchsia] Only build iossim runtimes for arm64
Petr Hosek [Thu, 27 Jan 2022 10:27:18 +0000 (02:27 -0800)]
[CMake][Fuchsia] Only build iossim runtimes for arm64

x86_64 isn't supported with the recent Xcode SDK.

2 years ago[AMDGPU] SILoadStoreOptimizer: Remove redundant check for volatile
Jay Foad [Wed, 26 Jan 2022 16:50:08 +0000 (16:50 +0000)]
[AMDGPU] SILoadStoreOptimizer: Remove redundant check for volatile

SILoadStoreOptimizer::collectMergeableInsts already ends the current
block if it sees a volatile (or ordered) memory access, so there is no
need to check for them again when scanning the instructions between two
pairing candidates in a block.

Differential Revision: https://reviews.llvm.org/D118266

2 years ago[mlir][bufferize] Insert memref.cast ops during finalizing pass
Matthias Springer [Thu, 27 Jan 2022 10:04:44 +0000 (19:04 +0900)]
[mlir][bufferize] Insert memref.cast ops during finalizing pass

The pass can currently not handle to_memref(to_tensor(x)) folding where a cast is necessary. This is required with the new unified bufferization. There is already a canonicalization pattern that handles such foldings and it should be used during this pass.

Differential Revision: https://reviews.llvm.org/D117988

2 years ago[NFC][CodeGen] Use llvm::DenseMap for DeferredDecls
Dawid Jurczak [Tue, 25 Jan 2022 15:53:05 +0000 (16:53 +0100)]
[NFC][CodeGen] Use llvm::DenseMap for DeferredDecls

CodeGenModule::DeferredDecls std::map::operator[] seem to be hot especially while code generating huge compilation units.
In such cases using DenseMap instead gives observable compile time improvement. Patch was tested on Linux build with default config acting as benchmark.
Build was performed on isolated CPU cores in silent x86-64 Linux environment following: https://llvm.org/docs/Benchmarking.html#linux rules.
Compile time statistics diff produced by perf and time before and after change are following:
instructions -0.15%, cycles -0.7%, max-rss +0.65%.
Using StringMap instead DenseMap doesn't bring any visible gains.

Differential Revision: https://reviews.llvm.org/D118169

2 years ago[LoopVersioning] Use IRBuilder for OR simplification.
Florian Hahn [Thu, 27 Jan 2022 09:55:38 +0000 (09:55 +0000)]
[LoopVersioning] Use IRBuilder for OR simplification.

2 years ago[LLDB] Add formatters for PointerIntPair, PointerUnion
Sergej Jaskiewicz [Thu, 20 Jan 2022 13:36:41 +0000 (16:36 +0300)]
[LLDB] Add formatters for PointerIntPair, PointerUnion

Also, add summaries for `SmallVector` and `ArrayRef`,
and fix the `StringRef` summary provider so it doesn't
ignore the `Length` field.

Differential Revision: https://reviews.llvm.org/D117779

2 years ago[InstCombine] Skip GEP of bitcast transform with opaque pointers
Nikita Popov [Thu, 27 Jan 2022 09:51:45 +0000 (10:51 +0100)]
[InstCombine] Skip GEP of bitcast transform with opaque pointers

This transform is fundamentally incompatible with opaque pointers.
Usually we would not hit it anyway because the bitcast is folded
away earlier, but due to worklist order it might survive until
here, so make sure we bail out explicitly.

2 years ago[InstCombine] Extract GEP of bitcast folds into separate function (NFC)
Nikita Popov [Thu, 27 Jan 2022 09:47:29 +0000 (10:47 +0100)]
[InstCombine] Extract GEP of bitcast folds into separate function (NFC)

2 years ago[RISCV] Fix test case expected output
Fraser Cormack [Thu, 27 Jan 2022 09:26:20 +0000 (09:26 +0000)]
[RISCV] Fix test case expected output

I didn't correctly update this before landing D118058.

2 years ago[BTFDebug] Avoid pointer element type access
Nikita Popov [Thu, 27 Jan 2022 09:30:21 +0000 (10:30 +0100)]
[BTFDebug] Avoid pointer element type access

Use the global value type instead.

2 years ago[InstCombine] Skip PromoteCastOfAllocation() transform under opaque pointers
Nikita Popov [Thu, 27 Jan 2022 09:24:15 +0000 (10:24 +0100)]
[InstCombine] Skip PromoteCastOfAllocation() transform under opaque pointers

I think this can't be hit anyway (because a ptr-to-ptr bitcast would
get folded earlier), but in the interest of being explicit skip
this transform for opaque pointers entirely.

2 years ago[lldb/test] Fix gnu-style-compression.yaml
Pavel Labath [Thu, 27 Jan 2022 09:16:56 +0000 (10:16 +0100)]
[lldb/test] Fix gnu-style-compression.yaml

In the rush to get the bot green, I did not realize I was building the
file with -gsplit-dwarf, and therefore the yaml ended up referring to a
file I did not check it.

This rebuilds the file without split dwarf.

2 years ago[InstCombine] Remove some pointer element type accesses
Nikita Popov [Thu, 27 Jan 2022 09:14:44 +0000 (10:14 +0100)]
[InstCombine] Remove some pointer element type accesses

One of these is guarded against opaque pointers, and the others
were accessing the call function type in a rather convoluted way.

2 years ago[SelectionDAG][VP] Provide expansion for VP_MERGE
Fraser Cormack [Mon, 24 Jan 2022 18:11:14 +0000 (18:11 +0000)]
[SelectionDAG][VP] Provide expansion for VP_MERGE

This patch adds support for expanding VP_MERGE through a sequence of
vector operations producing a full-length mask setting up the elements
past EVL/pivot to be false, combining this with the original mask, and
culminating in a full-length vector select.

This expansion should work for any data type, though the only use for
RVV is for boolean vectors, which themselves rely on an expansion for
the VSELECT.

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D118058

2 years ago[lldb] Convert ProcessWindowsLog to the new API
Pavel Labath [Thu, 27 Jan 2022 07:47:52 +0000 (08:47 +0100)]
[lldb] Convert ProcessWindowsLog to the new API

2 years ago[lldb/DWARF] Make manual dwarf index deterministic
Pavel Labath [Wed, 26 Jan 2022 15:29:51 +0000 (16:29 +0100)]
[lldb/DWARF] Make manual dwarf index deterministic

Currently, running the test suite with LLVM_ENABLE_EXPENSIVE_CHECKS=On
causes a couple of tests to fail. This happens because they expect a
certain order of variables (all of them happen to use the "target
variable" command, but other lookup functions should suffer from the
same issues), all of which have the same name. Sort algorithms often
preserve the order of equivalent elements (in this case the entries in
the NameToDIE map), but that not guaranteed, and
LLVM_ENABLE_EXPENSIVE_CHECKS stresses that by pre-shuffling all inputs
before sorting.

While this could easily be fixed by relaxing the test expectations,
having a deterministic output seems like a worthwhile goal,
particularly, as this could have bigger consequences than just a
different print order -- in some cases we just pick the first entry that
we find, whatever that is. Therefore this patch makes the sort
deterministic by introducing another sort key -- UniqueCString::Sort
gets a value comparator functor, which can be used to sort elements with
the same name -- in the DWARF case we use DIERef::operator<, which
roughly equals the order in which the entries appear in the debug info,
and matches the current "accidental" order.

Using a extra functor seemed preferable to using stable_sort, as the
latter allocates extra O(n) of temporary memory.

I observed no difference in debug info parsing speed with this patch
applied.

Differential Revision: https://reviews.llvm.org/D118251

2 years ago[lldb/test] Replace gnu-style-compression.cpp with a yaml file
Pavel Labath [Thu, 27 Jan 2022 09:01:59 +0000 (10:01 +0100)]
[lldb/test] Replace gnu-style-compression.cpp with a yaml file

In D117744, llvm removed writing support for this format, breaking the
test. We may eventually want to remove reading support as well, but for
now I have converted the test to a yaml file to maintain coverage.

2 years ago[clang-format] Format ParseOpenMP.cpp changes
Saiyedul Islam [Thu, 27 Jan 2022 08:55:08 +0000 (08:55 +0000)]
[clang-format] Format ParseOpenMP.cpp changes

Properly format D116549.

2 years ago[mlir][Bazel] Add TestMemRef target.
Adrian Kuegel [Thu, 27 Jan 2022 08:44:06 +0000 (09:44 +0100)]
[mlir][Bazel] Add TestMemRef target.

Needed after D118285

Differential Revision: https://reviews.llvm.org/D118330

2 years ago[MC] Add MCAsmParser::parseRParen to improve consistency and simplify code
Fangrui Song [Thu, 27 Jan 2022 08:37:49 +0000 (00:37 -0800)]
[MC] Add MCAsmParser::parseRParen to improve consistency and simplify code

Some diagnostics are more verbose but they don't seem to be more useful than
simple `expected ')'`

2 years ago[mlir][Bazel] Add missing dependency to ArithmeticTransforms.
Adrian Kuegel [Thu, 27 Jan 2022 08:31:55 +0000 (09:31 +0100)]
[mlir][Bazel] Add missing dependency to ArithmeticTransforms.

2 years agoAdd missing namespace to PPCLinux.cpp
Mike Hommey [Thu, 27 Jan 2022 08:26:16 +0000 (09:26 +0100)]
Add missing namespace to PPCLinux.cpp

This fixes a build failure with MSVC introduced in
https://reviews.llvm.org/D112906

Reviewed By: nemanjai

Differential Revision: https://reviews.llvm.org/D118211

2 years ago[MC] Remove unreachable .comm/.lcomm diagnostic
Fangrui Song [Thu, 27 Jan 2022 08:10:57 +0000 (00:10 -0800)]
[MC] Remove unreachable .comm/.lcomm diagnostic

and make another diagnostic math the prevailing format.

2 years ago[RISCV] fix dead code
Wu Xinlong [Thu, 27 Jan 2022 07:48:25 +0000 (15:48 +0800)]
[RISCV] fix dead code

fix dead code mentioned on https://reviews.llvm.org/D98136

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D118323

2 years ago[mlir] Don't emit unused labels
Marius Brehler [Tue, 25 Jan 2022 16:13:30 +0000 (16:13 +0000)]
[mlir] Don't emit unused labels

Stop the Cpp target from emitting unused labels. The previosly generated
code generated warning if `-Wunused-label` is passed to a compiler.

Co-authored-by: Simon Camphausen <simon.camphausen@iml.fraunhofer.de>
Reviewed By: jpienaar

Differential Revision: https://reviews.llvm.org/D118154

2 years ago[CSKY] Add floating operation support including float and double
Zi Xuan Wu [Thu, 27 Jan 2022 06:49:43 +0000 (14:49 +0800)]
[CSKY] Add floating operation support including float and double

CSKY arch has multiple FPU instruction versions such as FPU, FPUv2 and FPUv3 to implement floating operations.
For now, we just only support FPUv2 and FPUv3.

It includes the encoding, asm parsing of instructions and codegen of DAG nodes.