Johannes Doerfert [Fri, 3 Feb 2023 09:42:30 +0000 (01:42 -0800)]
[Attributor][NFCI] Use queries without exclusion set whenever possible
If a query uses an exclusion set but we haven't used it to determine the
result, we can cache the query without exclusion set too. When we lookup
a cached result we can check for the non-exclusion set version first.
Johannes Doerfert [Wed, 8 Feb 2023 15:02:24 +0000 (07:02 -0800)]
[Attributor][FIX] Ensure we use the right set comparison
Johannes Doerfert [Tue, 7 Feb 2023 14:17:08 +0000 (06:17 -0800)]
[Attributor][NFC] Avoid unnecessary string operations
This caused multiple string operations which we don't need if we do not
create a profile.
Johannes Doerfert [Fri, 3 Feb 2023 09:42:10 +0000 (01:42 -0800)]
[Attributor][NFCI] Create a AAIsDead for the function eagerly
Johannes Doerfert [Fri, 3 Feb 2023 09:40:36 +0000 (01:40 -0800)]
[Attributor][NFCI] Avoid a temporary vector and exit early
This change simply avoids the temporary vector and processes the elments
right away.
Louis Dionne [Thu, 9 Feb 2023 18:52:20 +0000 (10:52 -0800)]
[libc++][NFC] Reorganize hash.h
- Add missing _LIBCPP_HIDE_FROM_ABI
- Implement inline functions in the class to simplify the code
- Add missing `const` to `operator()`
- Move _LIBCPP_DISABLE_UBSAN_UNSIGNED_INTEGER_CHECK to the usual location for function attributes
Differential Revision: https://reviews.llvm.org/
D143668
Michael Buch [Tue, 7 Feb 2023 13:53:30 +0000 (13:53 +0000)]
[lldb][DWARFASTParserClang] Attach linkage name to ctors/dtors if missing
**Summary**
This patch addresses the case where we have a `DW_AT_external`
subprogram for a constructor (and/or destructor) that doesn't carry
a `DW_AT_linkage_name` attribute. The corresponding DIE(s) that
represent the definition will have a linkage name, but if the name
contains constructs that LLDBs fallback mechanism for guessing mangled
names to resolve external symbols doesn't support (e.g., abi-tags)
then we end up failing to resolve the function call.
We address this by trying to find the linkage name before we create
the constructor/destructor decl, which will get attached using
an `AsmLabelAttr` to make symbol resolution easier.
**Testing**
* Added API test
Differential Revision: https://reviews.llvm.org/
D143652
Michael Buch [Fri, 10 Feb 2023 12:43:42 +0000 (12:43 +0000)]
Reland "[llvm][dsymutil] Add DW_TAG_imported_declaration to accelerator table"
This relands the commit previously reverted in
`
8570bee53a8ce0c5d04bc11f288e19a457474c4c` due to failures on linux.
The problem was that the test executable was built with absolute
OSO prefix paths. This re-commit adds a modified version of the
executable that strips the absolute OSO prefix paths and makes
sure the test appends the OSO prefix appropriately (via the appropriate
dsymutil flags).
Differential Revision: https://reviews.llvm.org/
D143458
Florian Hahn [Fri, 10 Feb 2023 17:12:21 +0000 (17:12 +0000)]
[ConstraintElim] Update getLastConstraint to return to last row. (NFC)
The current code incorrectly returned the first instead of the last row.
This fixes the debug output.
Slava Zakharin [Thu, 9 Feb 2023 22:38:24 +0000 (14:38 -0800)]
[flang] Fixed selective TargetRewrite.
Some conversions were still happening under no-complex/character-conversion
options. This change fixes that and adds a LIT test.
Differential Revision: https://reviews.llvm.org/
D143685
Markus Böck [Wed, 8 Feb 2023 14:37:17 +0000 (15:37 +0100)]
[mlir][OpenMP] Add support for using Opaque Pointers in the OpenMP Dialect
The current OpenMP implementation assumes the use of typed pointers (or rather typed pointer like types). Given the support for typed pointers in LLVM is now pending removal, the OpenMP Dialect should be able to support opaque pointers as well, given that any users of it must lower OpenMP through the LLVM Dialect.
This patch fixes the above and adds support for using LLVM opaque pointers with the OpenMP dialect. This is implemented by making all related code not make use of the element type of pointer arguments. The few (one) op requiring a pointer element type now use an explicit `TypeAttr` for representing the element type.
More concretely, the list of changes are:
* `omp.atomic.read` now has an extra `TypeAttr` (also in syntax) which is the element type of the values read and stored from the operands
* `omp.reduction` now has an type argument in the syntax for both the accmulator and operand since the operand type can no longer be inferred from the accumulator
* `OpenMPToLLVMIRTranslation.cpp` was rewritten to never query element types of pointers
* Adjusted the verifier to be able to handle pointers without element types
Differential Revision: https://reviews.llvm.org/
D143582
Markus Böck [Fri, 10 Feb 2023 11:35:49 +0000 (12:35 +0100)]
[mlir][Async] Add option to LLVM lowering to use opaque pointers
Part of https://discourse.llvm.org/t/rfc-switching-the-llvm-dialect-and-dialect-lowerings-to-opaque-pointers/68179
This patch adds the pass option 'use-opaque-pointers' to allow the dialect conversion from async to LLVM to create LLVM opaque pointers instead of typed pointers.
The gist of the changes boil down to having to propagate the choice of whether opaque or typed pointers should be used, to various helper functions that then either create typed pointers or opaque pointers.
This sadly creates a bit of a code duplication in comparison to other patches in this series, which I think is mostly unavoidable however, since a lot of the patterns in this lowering require the use of the AsyncTypeConverter, instead of the LLVMTypeConverter.
Besides that, the tests have been converter to opaque pointers with one file with typed pointer support having been created as regression tests.
Differential Revision: https://reviews.llvm.org/
D143661
Daniel Grumberg [Thu, 19 Jan 2023 11:25:17 +0000 (11:25 +0000)]
[clang] [extract-api] Don't crash for category in libclang APIs
Remove failure conditions for categories in libclang and return empty
content instead.
Differential Revision: https://reviews.llvm.org/
D142101
Florian Hahn [Fri, 10 Feb 2023 16:26:38 +0000 (16:26 +0000)]
[ConstraintElim] Improve debug test to show removed constraints (NFC).
The current checks show incorrect debug output.
Tom Eccles [Tue, 10 Jan 2023 16:08:58 +0000 (16:08 +0000)]
[mlir] Add function for checking if a block is inside a loop
This function returns whether a block is nested inside of a loop. There
can be three kinds of loop:
1) The block is nested inside of a LoopLikeOpInterface
2) The block is nested inside another block which is in a loop
3) There is a cycle in the control flow graph
This will be useful for Flang's stack arrays pass, which moves array
allocations from the heap to the stack. Special handling is needed when
allocations occur inside of loops to ensure additional stack space is
not allocated on each loop iteration.
Differential Revision: https://reviews.llvm.org/
D141401
Sanjay Patel [Fri, 10 Feb 2023 15:53:22 +0000 (10:53 -0500)]
[VectorCombine] fix insertion point of shuffles
As shown in issue #60649, the new shuffles were
being inserted before a phi, and that is invalid.
It seems like most test coverage for this fold
(foldSelectShuffle) lives in the AArch64 dir,
but this doesn't repro there for a base target.
John Brawn [Fri, 10 Feb 2023 14:51:54 +0000 (14:51 +0000)]
[extract_symbols.py] Adjust how the output of nm is interpreted
When looking for defined symbols, look for symbols that aren't of a
type that we don't want, instead of having specific list of symbol
types that we do want. This fixes a problem where (when using GNU
nm at least) there were some symbol types that we want to export but
which weren't in the list.
Yitzhak Mandelbaum [Fri, 10 Feb 2023 15:02:44 +0000 (15:02 +0000)]
[clang][dataflow] Convert nonnull pointer parameter to a reference.
The parameter in question belongs to a function that is only called once. This patch updates the API to use a reference and changes the caller accordingly.
Differential Revision: https://reviews.llvm.org/
D143735
Mircea Trofin [Fri, 10 Feb 2023 15:46:47 +0000 (07:46 -0800)]
Revert "[mlgo] Fix test after
D143624"
This reverts commit
dc4c3cfd78c01bef427fca0431fe66a6c6de7c35.
Reverting because
D143624 has been reverted.
Guillaume Chatelet [Fri, 10 Feb 2023 15:15:10 +0000 (15:15 +0000)]
[libc] Move likely/unlikely to the optimization header
Sanjay Patel [Fri, 10 Feb 2023 14:59:33 +0000 (09:59 -0500)]
[InstCombine] propagate FMF in exp2->ldexp fold
Sanjay Patel [Fri, 10 Feb 2023 14:54:22 +0000 (09:54 -0500)]
[InstCombine] copy tail markings in exp2->ldexp fold
Sanjay Patel [Fri, 10 Feb 2023 14:51:27 +0000 (09:51 -0500)]
[InstCombine] add 'tail' to exp2 call in test; NFC
Tail markings are not propagated if the transform succeeds.
David Green [Fri, 10 Feb 2023 15:01:49 +0000 (15:01 +0000)]
Revert "Inlining: Run the legacy AlwaysInliner before the regular inliner."
This seems to cause large regressions in existing code, as much as 75% slower
(4x the time taken). Small always inline functions seem to be used a lot in the
cmsis-dsp library.
I would add a phase ordering test to show the problems, but one already exists!
The llvm/test/Transforms/PhaseOrdering/ARM/arm_mult_q15.ll was just changed by
removing alwaysinline to hide the problems that existed.
This reverts commit
cae033dcf227aeecf58fca5af6fc7fde1fd2fb4f.
This reverts commit
8e33c41e72ad42e4c27f8cbc3ad2e02b169637a1.
Juan Manuel MARTINEZ CAAMAÑO [Fri, 10 Feb 2023 14:05:05 +0000 (15:05 +0100)]
[AMDGPU][MC] Generate relative relocations for allocatable (more particularly, eh_frame) sections
Reviewed By: scott.linder
Differential Revision: https://reviews.llvm.org/
D142453
Benjamin Maxwell [Tue, 7 Feb 2023 10:36:34 +0000 (10:36 +0000)]
[DebugInfo] Handle missed DW_FORM_addrx3 and DW_FORM_strx3 cases
This fixes a few places where the addrx3 and strx3 forms were missed.
Previously this meant if one of these forms appeared somewhere various
errors could occur. This now also adds an extra test case for the addrx3
form (which previously failed).
Differential Revision: https://reviews.llvm.org/
D143488
Denis Fatkulin [Fri, 10 Feb 2023 14:19:11 +0000 (17:19 +0300)]
[clangd] Move function body to out-of-line: unnamed class method incorrect moving
The refactoring !!Move function body to out-of-line!! produces incorrect code for methods of unnamed classes.
For this simple example
// foo.h
struct Foo {
struct {
void f^oo() {}
} Bar;
};
the refactoring generates code:
// foo.cpp
void Foo::(unnamed struct at D:\test\foo.h:2:3)foo() {}
Outplace definition for methods of unnamed classes is meaningless. The patch disables it.
Reviewed By: kadircet
Differential Revision: https://reviews.llvm.org/
D143638
Tobias Gysi [Fri, 10 Feb 2023 14:22:54 +0000 (15:22 +0100)]
[mlir][llvm] Store memory op metadata using op attributes.
The revision introduces operation attributes to store tbaa metadata on
load and store operations rather than relying using dialect attributes.
At the same time, the change also ensures the provided getters and
setters instead are used instead of a string based lookup. The latter
is done for the tbaa, access groups, and alias scope attributes.
The goal of this change is to ensure the metadata attributes are only
placed on operations that have the corresponding operation attributes.
This is imported since only these operations later on translate these
attributes to LLVM IR. Dialect attributes placed on other operations
are lost during the translation.
Reviewed By: vzakhari, Dinistro
Differential Revision: https://reviews.llvm.org/
D143654
Joseph Huber [Fri, 10 Feb 2023 14:22:05 +0000 (08:22 -0600)]
[amdgpu-arch] Fix build of amdgpu-arch with CLANG_LINK_CLANG_DYLIB
The function clang_target_link_libraries must only be used with real
Clang libraries; with CLANG_LINK_CLANG_DYLIB, it will instead link in
clang-cpp. We must use the standard CMake target_link_libraries for
the HSA library.
v1nh1shungry [Wed, 18 Jan 2023 14:15:28 +0000 (22:15 +0800)]
[clangd] fix wrong CalleeArgInfo in the hover
```
void foobar(int);
int main() {
foobar(1 + 2);
^
}
```
Currently the CalleeArgInfo will be "Passed by reference", which should
be "Passed by value".
Fixes https://github.com/clangd/clangd/issues/1467
Differential Revision: https://reviews.llvm.org/
D142014
LLVM GN Syncbot [Fri, 10 Feb 2023 14:07:17 +0000 (14:07 +0000)]
[gn build] Port
6470706bc0ed
Simon Pilgrim [Fri, 10 Feb 2023 13:14:38 +0000 (13:14 +0000)]
[X86] combineConcatVectorOps - concatenate AVX512 vselect nodes. NFC.
This also requires us to constant fold vXi1 concat_vector nodes
Guillaume Chatelet [Fri, 10 Feb 2023 13:49:03 +0000 (13:49 +0000)]
[libc] Fix exp2f and prevent misuse of likely/unlikely
Let's make sure that we only accept boolean expressions when using likely/unlikely.
Differential Revision: https://reviews.llvm.org/
D143732
Nico Weber [Fri, 10 Feb 2023 13:58:51 +0000 (08:58 -0500)]
OCHyams [Fri, 10 Feb 2023 12:33:16 +0000 (12:33 +0000)]
[Assignment Tracking] Fix fragment index error in getDerefOffsetInBytes
Without this patch `getDerefOffsetInBytes` incorrectly always returns
`std::nullopt` for expressions with fragments due to an off-by-one error with
fragment element indices.
Reviewed By: StephenTozer
Differential Revision: https://reviews.llvm.org/
D143567
Christian Ulmann [Fri, 10 Feb 2023 13:33:41 +0000 (14:33 +0100)]
[mlir][llvm] Adapt loop metadata to match llvm
This commit adds support for the "llvm.loop.isvectorized" metadata and
ensures that the unroll followups match llvm's naming.
Reviewed By: gysit
Differential Revision: https://reviews.llvm.org/
D143730
Tobias Hieta [Fri, 10 Feb 2023 13:42:48 +0000 (14:42 +0100)]
[NFC][compiler-rt] Shorten comment
Tobias Hieta [Fri, 27 Jan 2023 15:46:19 +0000 (16:46 +0100)]
[Workflow] Add Release Repo sync script
Adds a bash script that syncs llvm/llvm-project and llvm/llvm-project-release-prs.
This should run on pushes to any of the repositories release branches.
I will follow this up with a change to the github actions to run this
script.
This breaks out the sync script from: https://reviews.llvm.org/
D133476
so we can keep them separate.
Reviewed By: kwk
Differential Revision: https://reviews.llvm.org/
D142726
Tobias Hieta [Wed, 8 Feb 2023 08:26:49 +0000 (09:26 +0100)]
[CMake][compiler-rt] Support clang-cl in CompilerRTMockLLVMCMakeConfig
clang-cl doesn't support -dumpmachine directly, so we need to
preface it with /clang: in order to get this probing function
to work.
This is needed in order to run cmake directly on the runtimes
directory.
Reviewed By: hans
Differential Revision: https://reviews.llvm.org/
D143557
Job Noorman [Fri, 10 Feb 2023 11:50:26 +0000 (19:50 +0800)]
[RISCV] Add tests for .attribute stack_align
As suggested by @asb [here](https://reviews.llvm.org/
D143570#
4112877), we need tests for `.attribute stack_align`. This patch simply verifies a stack alignment of 16 is emitted for any combination of currently supported base ISA and extensions.
Reviewed By: asb, kito-cheng
Differential Revision: https://reviews.llvm.org/
D143639
Sanjay Patel [Fri, 10 Feb 2023 13:01:04 +0000 (08:01 -0500)]
[InstCombine] avoid crashing in pow->ldexp
Similar to
62a0a1b9eea7788c1f9dbae -
We have pow math intrinsics in IR, but no ldexp intrinsics
to handle vector types.
A patch for that was proposed in D14327, but it was not completed.
Issue #60605
Frederic Cambus [Fri, 10 Feb 2023 12:50:04 +0000 (13:50 +0100)]
[clang] Update Clang version from 16 to 17 in scan-build.1.
Similar to
D110763.
Sanjay Patel [Fri, 10 Feb 2023 12:03:56 +0000 (07:03 -0500)]
[InstCombine] avoid crashing in exp2->ldexp
We have exp2 math intrinsics in IR, but no ldexp intrinsics
to handle vector types.
A patch for that was proposed in D14327, but it was not completed.
Issue #60605
Sanjay Patel [Fri, 10 Feb 2023 11:55:59 +0000 (06:55 -0500)]
[InstCombine] add test for exp2->ldexp; NFC
Sanjay Patel [Fri, 10 Feb 2023 11:51:57 +0000 (06:51 -0500)]
[InstCombine] auto-generate check lines in test file; NFC
Tim Northover [Fri, 10 Feb 2023 11:40:20 +0000 (11:40 +0000)]
ARM: skip debug instructions when matching jump-table patterns.
When working out whether we can see a compressible jump-table pattern during
ConstantIslands, we were stopping when we saw a debug instruction. Instead it's
better to keep iterating backwards to the first real instruction.
https://reviews.llvm.org/
D142019
Ingo Müller [Fri, 10 Feb 2023 11:31:35 +0000 (11:31 +0000)]
[mlir][bufferization] Restrict function boundary buffer. to func.call.
The current bufferization on function boundaries works on `func.func`
and any call op implementing `CallOpInterface`. Then, an error is thrown
if there is a `CallOpInterface` op that is not `func.call`. This is
unnecessary and breaks the pass whenever such an op occurs (such as
`llvm.call`). This PR simply restricts the handling of call ops to
`func.call`.
Reviewed By: springerm
Differential Revision: https://reviews.llvm.org/
D143724
Guillaume Chatelet [Fri, 10 Feb 2023 11:40:14 +0000 (11:40 +0000)]
[bazel] Fix build and make all target respect layering_check
Tested with `bazelisk-linux-amd64 query //... + @llvm-project//... | xargs bazelisk-linux-amd64 test --config=ci --features=layering_check`
LiaoChunyu [Fri, 10 Feb 2023 11:34:37 +0000 (19:34 +0800)]
[RISCV] Add precommit tests for
D143646
Ivan Kosarev [Fri, 10 Feb 2023 11:07:17 +0000 (11:07 +0000)]
[AMDGPU][AsmParser] Fix matching immediate literals.
Prevents potential matching of literal offsets to non-literal operands.
Reviewed By: dp
Differential Revision: https://reviews.llvm.org/
D142194
Yingchi Long [Fri, 10 Feb 2023 11:26:14 +0000 (19:26 +0800)]
[InstCombine][NFC] regenerate tests for simple_phi_condition.ll
Max Kazantsev [Fri, 10 Feb 2023 10:55:01 +0000 (17:55 +0700)]
[Test] Add complementary tests for OR analogous for those for AND
Max Kazantsev [Fri, 10 Feb 2023 10:41:03 +0000 (17:41 +0700)]
[Test] Add test showing an opportunity to transform checks and hoist invariants
Marco Elver [Fri, 10 Feb 2023 10:22:17 +0000 (11:22 +0100)]
[SanitizerBinaryMetadata] Fix ignorelist test under Windows
Windows paths confuse the regular expression. Just use the test source
name directly.
Fixes: 421215b919d0 ("[SanitizerBinaryMetadata] Support ignore list")
Dominik Adamski [Tue, 17 Jan 2023 09:51:48 +0000 (03:51 -0600)]
Move SIMD alignment calculation to LLVM Frontend
Currently default simd alignment is defined by Clang specific TargetInfo class.
This class cannot be reused for LLVM Flang. That's why default simd alignment
calculation has been moved to OMPIRBuilder which is common for Flang and Clang.
Previous attempt: https://reviews.llvm.org/
D138496 was wrong because
the default alignment depended on the number of built LLVM targets.
If we wanted to calculate the default alignment for PPC and we hadn't specified
PPC LLVM target to build, then we would get 0 as the alignment because
OMPIRBuilder couldn't create PPCTargetMachine object and it returned 0 as
the default value.
If PPC LLVM target had been built earlier, then OMPIRBuilder could have created
PPCTargetMachine object and it would have returned 128.
Differential Revision: https://reviews.llvm.org/
D141910
Reviewed By: jdoerfert
Dmitry Makogon [Fri, 10 Feb 2023 09:00:16 +0000 (16:00 +0700)]
[LVI] Don't traverse uses when calculating range at use
This effectively reverts
5c38c6a and
4f772b0.
A recently introduced LazyValueInfo::getConstantRangeAtUse returns incorrect
ranges for values in certain cases. One such example is described in PR60629.
The issue has something to do with traversing PHI uses of a value transitively.
As nikic pointed out, we're effectively reasoning about values from different
loop iterations.
In the faulting test case, CVP made a miscompilation because the calculated
range for a shift argument was incorrect. It returned empty-set, however it is
clearly not a dead code. CVP then erased the shift instruction because
of empty range.
Dmitry Makogon [Fri, 10 Feb 2023 09:30:47 +0000 (16:30 +0700)]
[LVI][CVP] Add test for miscompilation by CVP due to incorrect range returned by LVI
This adds a test case from PR60629 which shows a miscompilation by CVP.
Archibald Elliott [Fri, 10 Feb 2023 09:59:46 +0000 (09:59 +0000)]
[NFC][TargetParser] Replace uses of llvm/Support/Host.h
The forwarding header is left in place because of its use in
`polly/lib/External/isl/interface/extract_interface.cc`, but I have
added a GCC warning about the fact it is deprecated, because it is used
in `isl` from where it is included by Polly.
OCHyams [Fri, 10 Feb 2023 09:27:37 +0000 (09:27 +0000)]
[Assignment Tracking][mem2reg] Remove overly defensive assert
The assert fires if a store to an alloca with no linked dbg.assigns has linked
dbg.assigns. This can happen in the wild due to optimisations dropping the
alloca's debug info so we shouldn't assert against it.
Reviewed By: jryans
Differential Revision: https://reviews.llvm.org/
D143153
OCHyams [Fri, 10 Feb 2023 09:26:26 +0000 (09:26 +0000)]
[Assignment Tracking][SROA] Delete dbg.assigns linked to rewritten stores
AggLoadStoreRewriter splits aggregate loads and stores into scalars (before the
alloca is split up). The new stores and debug intrinsics are already wired up
correctly - we just need to also delete the dbg.assign that is linked to the
split to-be-deleted store too.
Reviewed By: jmorse
Differential Revision: https://reviews.llvm.org/
D142882
Chuanqi Xu [Fri, 10 Feb 2023 09:46:18 +0000 (17:46 +0800)]
Revert "[C++20] [Modules] [ClangScanDeps] Enable to print make-style dependency file within P1689 format (4/4)"
This reverts commit
e1354763b6e641e45cc2371270883fcd26edf705.
There is a build failure in m68k-linux testing bot
(https://lab.llvm.org/buildbot/#/builders/192/builds/267), which is
weird. Revert this for now and look at the reasons.
Samuel Parker [Wed, 8 Feb 2023 14:09:06 +0000 (14:09 +0000)]
[WebAssembly] Additional patterns for pmin/pax
Each operation was missing their inverted condition using olt or ogt.
Also, as we don't need to discern +/-0, I think we should also be
able to use ole and oge.
Differential Revision: https://reviews.llvm.org/
D143581
Valentin Clement [Fri, 10 Feb 2023 09:25:01 +0000 (10:25 +0100)]
[flang][NFC] Clean up polymorphic-temp.f90
Marco Elver [Fri, 10 Feb 2023 08:34:08 +0000 (09:34 +0100)]
[SanitizerBinaryMetadata] Support ignore list
For large projects it will be required to opt out entire subdirectories.
In the absence of fine-grained control over the flags passed via the
build system, introduce -fexperimental-sanitize-metadata-ignorelist=.
The format is identical to other sanitizer ignore lists, and its effect
will be to simply not instrument either functions or entire modules
based on the rules in the ignore list file.
Reviewed By: dvyukov
Differential Revision: https://reviews.llvm.org/
D143664
OCHyams [Fri, 10 Feb 2023 09:20:53 +0000 (09:20 +0000)]
[NFC][Assignment Tracking] Remove lifetime intrinsics from some tests
The intrinsics don't add anything to the tests and the tests are easier to
debug without the additional noise.
Some SSA value names have changed as a result of no longer visiting the
intrinsics.
Reviewed By: jryans
Differential Revision: https://reviews.llvm.org/
D143141
Max Kazantsev [Fri, 10 Feb 2023 09:05:16 +0000 (16:05 +0700)]
[Test][SCEV] Add tests where multiple assumes can be used to improve trip multiple
Add tests combining assume on trip count divisibility (TV % VF) and minimum/maximum
values (TC > N, TC < M).
Patch by Alon Kom!
Differential Revision: https://reviews.llvm.org/
D143337
Benjamin Kramer [Fri, 10 Feb 2023 09:22:18 +0000 (10:22 +0100)]
[bazel] Add missing dependency after
687f78210d0
Tobias Hieta [Fri, 10 Feb 2023 07:33:32 +0000 (08:33 +0100)]
[NFC] convert compiler-rt/lib/dfsan/scripts/build-libc-list.py to python3 syntax
I found a few Python files not using Python 3 syntax in the tree
when trying to apply reformatting to them. This was converted with
2to3 and the changes seemed sane.
Reviewed By: MaskRay, browneee
Differential Revision: https://reviews.llvm.org/
D143701
Siva Chandra Reddy [Fri, 10 Feb 2023 09:03:05 +0000 (09:03 +0000)]
[libc][Obvious] Include MPFRWrapper and testutils only if tests are enabled.
Mikhail Goncharov [Fri, 10 Feb 2023 09:03:37 +0000 (10:03 +0100)]
[bazel] fix build for
6470706bc0ed
chenglin.bi [Fri, 10 Feb 2023 08:49:56 +0000 (16:49 +0800)]
[Reland][LSR] Hoist IVInc to loop header if its all uses are in the loop header
Original code will cause crash when the load/store memory type is structure because isIndexedLoadLegal/isIndexedStore doesn't support struct type.
So we limit the load/store memory type to integer.
Origin commit message:
When the latch block is different from header block, IVInc will be expanded in the latch loop. We can't generate the post index load/store this case.
But if the IVInc only used in the loop, actually we still can use the post index load/store because when exit loop we don't care the last IVInc value.
So, trying to hoist IVInc to help backend to generate more post index load/store.
Fix #53625
Reviewed By: eopXD
Differential Revision: https://reviews.llvm.org/
D138636
Siva Chandra Reddy [Thu, 9 Feb 2023 05:42:12 +0000 (05:42 +0000)]
[libc] Run libc tests unconditionally, even if there are no source changes.
Before this change, the test binaries were executed only if the deps
changed. This change makes the test binaries run unconditionally. This
is more in line with the rest of LLVM. Future patches will print test
statistics etc., much like the rest of LLVM.
Reviewed By: gchatelet
Differential Revision: https://reviews.llvm.org/
D143627
Matthias Springer [Fri, 10 Feb 2023 08:30:21 +0000 (09:30 +0100)]
[mlir][bufferization] Support multiple leaves in EmptyTensorElimination
Support cases where a source tensor can be traced back to multiple possible tensor.empty ops.
Differential Revision: https://reviews.llvm.org/
D142130
Sandeep Kosuri [Fri, 10 Feb 2023 08:32:50 +0000 (02:32 -0600)]
[OPENMP][NFC] added loop bind to the support page
Max Kazantsev [Fri, 10 Feb 2023 08:20:03 +0000 (15:20 +0700)]
[SimplifyIndVar][NFC] Refactor Binary Operator's flag strengthening
Extract complex logic of Binary Operator's flag strengthening to a separate
method in order to reuse it.
Patch by Aleksandr Popov!
Differential Revision: https://reviews.llvm.org/
D143562
Reviewed By: mkazantsev
Chuanqi Xu [Fri, 10 Feb 2023 08:13:46 +0000 (16:13 +0800)]
[NFC] Add test to show that preprocess an import decl won't require to load it actually
Close https://github.com/llvm/llvm-project/issues/59287
This is addressed in
D137526 but I forgot to add a test for it.
Uday Bondhugula [Fri, 10 Feb 2023 08:05:04 +0000 (13:35 +0530)]
[MLIR] Fold away divs and mods in affine ops with operand info
Fold away divs and mods in affine maps exploiting operand info during
canonicalization. This simplifies affine map applications such as the ones
below:
```
// Simple ones.
affine.for %i = 0 to 32 {
affine.load %A[%i floordiv 32]
affine.load %A[%i mod 32]
affine.load %A[2 * %i floordiv 64]
affine.load %A[(%i mod 16) floordiv 16]
...
}
// Others.
affine.for %i = -8 to 32 {
// Will be simplified %A[0].
affine.store %cst, %A[2 + (%i - 96) floordiv 64] : memref<64xf32>
}
```
Reviewed By: springerm
Differential Revision: https://reviews.llvm.org/
D143456
Chuanqi Xu [Fri, 10 Feb 2023 08:02:31 +0000 (16:02 +0800)]
[NFC] Remove the unused parameter introduced in
e135476
I introduced an unused parameter due to an oversight during rebasing.
Now the patch removes the parameter.
Jean Perier [Fri, 10 Feb 2023 07:56:53 +0000 (08:56 +0100)]
[flang][hlfir] Ramp-up support of implicit interface mismatches
There is a lot of Fortran code that takes advantage of F77 implicit
interface to pass arguments with a different type than those from
the subprogram definition (which is well defined if the storage
and passing convention are the same or compatible).
When the definition and calls are in different files, there is nothing
special to do: the actual arguments are already used to compute the
call interface.
The trouble for lowering comes when the definition is in the same
compilation unit (Semantics raises warning). Then, lowering will
be provided with the interface from the definition to prepare the
argument, and this leads to many ad-hoc handling (see
builder.convertWithSemantics) in the current lowering to cope
with the dummy/actual mismatches on a case by case basis. The
current lowering to FIR is not even complete for all mismatch cases that
can be found in the wild (see https://github.com/llvm/llvm-project/issues/60550),
it is crashing or hitting asserts for many of the added tests.
For HLFIR, instead of coping on a case by case basis, the call
interface will be recomputed according to the actual arguments when
calling an external procedure that can be called with an explicit
interface.
One extra case still has to be handled manually because it may happen
in calls with explicit interfaces: passing a character procedure
designator to a non character procedure dummy (and vice-versa) is widely
accepted even with explicit interfaces (and flang semantic accepts it).
Yet, this "mismatch" cannot be dealt with a simple fir.convert because
character dummy procedure are passed with a different passing
convention: an extra argument is hoisted for the result length (in FIR,
there is no extra argument yet, but the MLIR func argument is a
tuple<fir.boxproc, len>).
Differential Revision: https://reviews.llvm.org/
D143636
Christian Ulmann [Fri, 10 Feb 2023 07:21:46 +0000 (08:21 +0100)]
[mlir][llvm] Fix TBAA verfication crash
This commit fixes a crash of the TBAA verification that happened due to
accessing memory through invalid pointers. A DenseMap does not guarantee
that pointers to its elements remain valid after additional elements
are inserted.
A testcase that caused this crash had more than 100 TBAA metadata
operations and thus no test is added. Instead, there is now an assertion
that ensures that the graph class is used correctly.
Reviewed By: vzakhari
Differential Revision: https://reviews.llvm.org/
D143653
Pierre van Houtryve [Wed, 25 Jan 2023 13:06:09 +0000 (14:06 +0100)]
[RFC][GISel] Add a way to ignore COPY instructions in InstructionSelector
RFC to add a way to ignore COPY instructions when pattern-matching MIR in GISel.
- Add a new "GISelFlags" class to TableGen. Both `Pattern` and `PatFrags` defs can use it to alter matching behaviour.
- Flags start at zero and are scoped: the setter returns a `SaveAndRestore` object so that when the current scope ends, the flags are restored to their previous values. This allows child patterns to modify the flags without affecting the parent pattern.
- Child patterns always reuse the parent's pattern, but they can override its values. For more examples, see `GlobalISelEmitterFlags.td` tests.
- [AMDGPU] Use the IgnoreCopies flag in BFI patterns, which are known to be bothered by cross-regbank copies.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/
D136234
Pierre van Houtryve [Tue, 24 Jan 2023 09:33:29 +0000 (10:33 +0100)]
[AMDGPU] Run unmerge combines post regbankselect
RegBankSelect can insert G_UNMERGE_VALUES in a lot of places which
left us with a lot of unmerge/merge pairs that could be simplified.
These often got in the way of pattern matching and made codegen
worse.
This patch:
- Makes the necessary changes to the merge/unmerge combines so they can run post RegBankSelect
- Adds relevant unmerge combines to the list of RegBankSelect combines for AMDGPU
- Updates some tablegen patterns that were missing explicit cross-regbank copies (V_BFI patterns were causing constant bus violations with this change).
This seems to be mostly beneficial for code quality.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/
D142192
Max Kazantsev [Fri, 10 Feb 2023 06:56:33 +0000 (13:56 +0700)]
[Test] Add tests where we can replace IV check with invariant check basing on predicated backedge cond
wanglei [Fri, 10 Feb 2023 06:35:18 +0000 (14:35 +0800)]
[LoongArch] Convert test to opaque pointer. NFC
lizhijin [Fri, 10 Feb 2023 06:42:33 +0000 (14:42 +0800)]
[SVE] Add intrinsics for logical/bitwise operations that explicitly undefine the result for inactive lanes
This patch adds new intrinsics for logical/bitwise operations and
changes the lowering for the following builtins to emit calls to
the new aarch64.sve.###.u intrinsics.
svand_x
svand_n_x
svorr_x
svorr_n_x
sveor_x
sveor_n_x
svbic_x
svbic_n_x
Reviewed By: paulwalker-arm
Differential Revision: https://reviews.llvm.org/
D143499
gonglingqin [Fri, 10 Feb 2023 05:57:55 +0000 (13:57 +0800)]
[LoongArch] Implement the convertSelectOfConstantsToMath hook
Normally, the ISD::SELECT is lowered to maskeqz + masknez. This is
inefficient when operands are constants that can be computed easily.
To avoid that, override TargetLowering::convertSelectOfConstantsToMath.
Differential Revision: https://reviews.llvm.org/
D143621
Amara Emerson [Fri, 10 Feb 2023 05:56:02 +0000 (21:56 -0800)]
Inliner: Address missed review comments for
D143624
Max Kazantsev [Fri, 10 Feb 2023 04:56:08 +0000 (11:56 +0700)]
[SimpleLoopUnswitch] Inject loop-invariant conditions and unswitch them when it's profitable
Based on https://discourse.llvm.org/t/rfc-inject-invariant-conditions-to-loops-to-enable-unswitching-and-constraint-elimination
This transform attempts to handle the following loop:
```
for (...) {
x = <some variant>
if (x <u C1) {} else break;
if (x <u C2) {} else break;
}
```
Here `x` is some loop-variant value, and `C1` and `C2` are loop invariants.
As we see, this loop has no invariant checks we can unswitch on. However, there is an
invariant condition that can make the second check redundant. Specifically, it is `C1 <=u C2`.
We can modify this code in the following way:
```
for (...) {
x = <some variant>
if (x <u C1) {} else break;
if (C1 <=u C2) {
/* no check is required */
}
else {
// do the check normally
if (x <u C2) {} else break;
}
}
```
Now we have an invariant condition `C1 <=u C2` and can unswitch on it.
This patch introduces the basic version of this transform, with some limitations,
all of them seem liftable (but needs more work & testing):
- All checks are `ult` condition;
- All branches in question stay in loop if the said condition is true and leave it otherwise;
- All in-loop branches are hot enough;
There is also a room for improvement cost model. So far we evalutate the cost of
unswitching this newly injected invariant branch the same as if we would unswitch
on 2nd condition, which is not exactly precise (but also not grossly wrong).
Differential Revision: https://reviews.llvm.org/
D136233
Reviewed By: skatkov
Jonas Devlieghere [Fri, 10 Feb 2023 01:24:10 +0000 (17:24 -0800)]
[lldb] Hoist code to create StructuredData into DiagnosticEventData (NFC)
Hoist the code that creates a StructuredData dictionary from a
diagnostic event into the DiagnosticEventData. This addresses Ismail's
code review feedback from
D143687.
Differential revision: https://reviews.llvm.org/
D143694
Mircea Trofin [Fri, 10 Feb 2023 05:13:41 +0000 (21:13 -0800)]
[mlgo] Fix test after
D143624
Kai Luo [Fri, 10 Feb 2023 04:52:31 +0000 (04:52 +0000)]
[CMake] Fix -DBUILD_SHARED_LIBS=on builds after
D137534
Yingchi Long [Tue, 24 Jan 2023 13:25:50 +0000 (21:25 +0800)]
[InstCombine] combine intersection for inequality icmps
```
define i1 @src(i32 %A) {
%mask1 = and i32 %A, 15 ; 0x0f
%tst1 = icmp eq i32 %mask1, 3 ; 0x03
%mask2 = and i32 %A, 255 ; 0xff
%tst2 = icmp eq i32 %mask2, 243; 0xf3
%res = or i1 %tst1, %tst2
ret i1 %res
}
```
->
```
define i1 @tgt(i32 %A) {
%1 = and i32 %A, 15
%res = icmp eq i32 %1, 3
ret i1 %res
}
```
Proof: https://alive2.llvm.org/ce/z/4AyvcE
Assume that `(B & D) & (C ^ E) == 0`, and `(B & D) == D || (B & D) == B`,
transforms:
```
(icmp ne (A & B), C) & (icmp ne (A & D), E)
-> (icmp ne (A & (B&D)), (C&E))
```
Fixes: https://github.com/llvm/llvm-project/issues/59680
Reviewed By: spatel, bcl5980
Differential Revision: https://reviews.llvm.org/
D140666
Yingchi Long [Thu, 19 Jan 2023 09:14:15 +0000 (17:14 +0800)]
[InstCombine] precommit tests for
D140666
Differential Revision: https://reviews.llvm.org/
D142090
Dave Lee [Sat, 4 Feb 2023 05:44:07 +0000 (21:44 -0800)]
[lldb] Remove unused CommandObjectRegexCommand::m_max_matches (NFC)
Michael Buch [Fri, 10 Feb 2023 03:28:28 +0000 (03:28 +0000)]
Revert "[lldb][DWARFASTParserClang] Correctly resolve imported namespaces during expression evaluation"
This reverts commit
1ba6ec0ba906472661506ef7bd2f0bf4efe4657e.
Fails on linux buildbots.
Differential Revision: https://reviews.llvm.org/
D143398
Michael Buch [Fri, 10 Feb 2023 03:24:52 +0000 (03:24 +0000)]
Revert "[llvm][dsymutil] Add DW_TAG_imported_declaration to accelerator table"
This reverts commit
ccee9b7839a182fc692a65f563abf76f81dd436c.
This started failing on Linux buildbots. The executable wasn't generated
correctly.
Differential Revision: https://reviews.llvm.org/
D143458
Chuanqi Xu [Fri, 10 Feb 2023 03:22:33 +0000 (11:22 +0800)]
[C++20] [Modules] [ClangScanDeps] Enable to print make-style dependency file within P1689 format (4/4)
Required in https://reviews.llvm.org/
D137534.
The build systems needs the information to know that "header X changed,
scanning may have changed, so please rerun scanning". Although it is
possible to get the information by running clang-scan-deps for the
second time with make format, it is not user friendly clearly.
Reviewed By: jansvoboda11
Differential Revision: https://reviews.llvm.org/
D139168
Kai Luo [Fri, 10 Feb 2023 03:17:39 +0000 (03:17 +0000)]
[DwarfGenerator] Emit dwarf address section when dwarf version is not less than 4
Some targets (like AIX) might not feature dwarf address section due to low dwarf version. Check before emitting one.
Fixed DebugInfo unit test failures in https://lab.llvm.org/buildbot/#/builders/214/builds/5739.
Reviewed By: qiucf
Differential Revision: https://reviews.llvm.org/
D143561
Mehdi Amini [Fri, 10 Feb 2023 03:13:48 +0000 (19:13 -0800)]
Remove empty MLIRMemRefUtils library (NFC)
Shengchen Kan [Tue, 7 Feb 2023 08:04:05 +0000 (16:04 +0800)]
[X86][MC][NFC] Refine code in X86MCCodeEmitter.cpp about opcode prefix
1. Make code clearer by separating the logic of setting bits from the
logic of how a prefix is encoded
2. Extract common code into functions to avoid code duplication
3. Return a enum rather a boolean to ehance scalability and uniform
the behavior of functions
Reviewed By: pengfei, craig.topper
Differential Revision: https://reviews.llvm.org/
D143471