This transform has caused a few issues with operations that can naturally be
extended. This patch just adds a debug option for disabling the transform,
useful for testing cases where it might not be profitable.

One big issue with DirectXShaderCompiler was test coverage: DXIL and
SPIR-V backends had their own tests. When a bug was found in one, the
other wasn't always checked. This lead to unequal support of HLSL for
both backends. We'd like to avoid those issues here, hence the
test-sharing.

By default, all the tests in this folder are marked as requiring
DirectX. But as SPIR-V support grows, each test drop this requirement,
and check the SPIR-V behavior.

I would have preferred to mark new tests as XFAIL for SPIR-V by default,
so we could differentiate real unsupported tests (as SPIR-V has no
equivalent), from newly added tests. But the way LIT is built, I don't
think this is possible.

---------

Signed-off-by: Nathan Gauër <brioche@google.com>

Now that VPInstruction can manage fast math flags via
VPRecipeWithIRFlags, use them directly to model the fast-math flags of
the select created for the final reduction value instead of adding them
late.

Code Object V2 has been deprecated for more than a year now. We can
safely remove it from LLVM.

- [clang] Remove support for the `-mcode-object-version=2` option.
- [lld] Remove/refactor tests that were still using COV2
- [llvm] Update AMDGPUUsage.rst
- Code Object V2 docs are left for informational purposes because those
code objects may still be supported by the runtime/loaders for a while.
- [AMDGPU] Remove COV2 emission capabilities.
- [AMDGPU] Remove `MetadataStreamerYamlV2` which was only used by COV2
- [AMDGPU] Update all tests that were still using COV2 - They are either
deleted or ported directly to code object v4 (as v3 is also planned to
be removed soon).

Dependent interfaces have been added a while ago and these TODOs can be
addressed now.

The RISCV-V-Spec[1] describes which RVV instructions are allowed to overlap between source operands and destination operands. Currently, most of these instructions are constrained with early-clobber.

This patch labels these instructions with TargetOverlapConstraintType in pseudo to indicate that these instructions can overlap.

The value of TargetOverlapConstraintType is:

1 -> default
2 -> narrow case
3 -> widen case

This information is useful for future optimization passes.

[1] https://github.com/riscv/riscv-v-spec/blob/master/v-spec.adoc#52-vector-operands

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D157588

This reverts commit 933ad5c8.

This caused 1 test failure and an unexpected pass on AArch64 Linux:
https://lab.llvm.org/buildbot/#/builders/96/builds/45765

Wasn't reported because the bot was already red at the time.

https://cplusplus.github.io/LWG/issue3655

Differential Revision: https://reviews.llvm.org/D144645


Co-authored-by: Louis Dionne <ldionne.2@gmail.com>

[libc][clang-tidy] Add llvm-header-guard to get consistant naming and prevent file copy/paste issues. (#66477)

After f108c6cd, (mul x, 1) is simplified to x, which can cause the select
for the final reduction value when tail-folding to use the reduction
value for both options. Relax the assertion to make sure this case is
allowed.

Note that the reduction is now redundant itself and could be further
simplified.

Fixes #66895.

This patch is a melting pot of changes picked up from
https://llvm.org/D61878. It adds a few tests checking corner cases of
unordered containers comparison and adds benchmarks for a few
unordered_set operations.

On Apple platforms, we always support the -nostdlib++ flag. Hence, it is
not necessary to manually link against system libraries. In fact, doing
so causes us to link against libSystem explicitly, which messes up with
the order of libraries we should use. Indeed:

   Before patch, using the system unwinder (LIBCXXABI_USE_LLVM_UNWINDER = OFF)
   ===========================================================================
   $ otool -L lib/{libc++.1.dylib,libc++abi.1.dylib,libunwind.1.dylib}
   lib/libc++.1.dylib:
         @rpath/libc++.1.dylib
         /usr/lib/libSystem.B.dylib
         @rpath/libc++abi.1.dylib
   lib/libc++abi.1.dylib:
         @rpath/libc++abi.1.dylib
         /usr/lib/libSystem.B.dylib
   lib/libunwind.1.dylib:
         @rpath/libunwind.1.dylib
         /usr/lib/libSystem.B.dylib

   After patch, using the system unwinder (LIBCXXABI_USE_LLVM_UNWINDER = OFF)
   ==================================...

`AffineForOp::getInitOperands` is renamed to `getInits` to be consistent
with MLIR operand getter naming conventions. ("get" + operand name)

This change is https://reviews.llvm.org/D49549 ported to the monorepo layout.

Patch by Arnaud Coomans.

Patch title and message:
```
Change 'clang-test' to 'check-clang' on the hacking webpage

The command to run tests was previously changed from 'clang-test' to 'check-clang'; with 'clang-test' remaining available as a legacy alias (see git-svn-id: https://reviews.llvm.org/rC159483 91177308-0d34-0410-b5e6-96231b3b80d8).

This commit changes the www documentation accordingly.
```

- make sure that the type of induction variable should be determined by the type of the lower bound type.

Reviewed By: ftynse

Differential Revision: https://reviews.llvm.org/D159534

When unswitching via invariant condition injection, we currently
mark the condition in the old loop, so that it does not get
unswitched again. However, if there are multiple branches for
which conditions can be injected, then we can do that for both
the old and new loop. This means that the number of unswitches
increases exponentially.

Change the handling to be more similar to partial unswitching,
where we instead mark the whole loop, rather than a single
condition. This means that we will only generate a linear number
of loops.

TBH I think even that is still highly undesirable, and we should
probably be unswitching all candidates at the same time, so that
we end up with only two loops. But at least this mitigates the
worst case.

The test case is a reduced variant that generates 1700 lines of IR
without this patch and 290 with it.

Fixes https://github.com/llvm/llvm-project/issues/66868.

Similar to D110763.

This is in preparation towards the move from recommonmark to myst

See #65664

Fixed #66047
Before fix,the following testcase expected true.
```cpp
TEST_F(StructuralEquivalenceStmtTest, DeclRefENoEq) {
  std::string Prefix = "enum Test { AAA, BBB };";
  auto t = makeStmts(
      Prefix + "void foo(int i) {if (i > 0) {i = AAA;} else {i = BBB;}}",
      Prefix + "void foo(int i) {if (i > 0) {i = BBB;} else {i = AAA;}}",
      Lang_CXX03, ifStmt());
  EXPECT_FALSE(testStructuralMatch(t)); // EXPECT_TRUE
}
```

`FixRISCVCallsPass` changes all different forms of calls to `PseudoCALL`
instructions. However, the original call's annotations were lost in the
process.

This patch fixes this by moving all annotations from the old to the new
call. `MCPlusBuilder::moveAnnotations` had to be made public for this.

Re-applies #65683 with a fix to always run
`Actions.ActOnEndOfTranslationUnit` regardless of incremental
processing.

Fixed: https://github.com/llvm/llvm-project/issues/62127
https://reviews.llvm.org/D116203 introduced several compiler builtin
equivalents of the unary type traits. In some cases (e.g. template)
those builtin will be dependent and need to be mangle.
This patch add the check for `u{builtin}I{type}E` to demangle it.

Reviewed By: rjmccall

Differential Revision: https://reviews.llvm.org/D148465

Fix a18ee8b7 to use a comparator
that actually works: assign an ordinal to registered section.

InvalidFreeReport prints invalid ptr/mem.

This reverts commit 7641c220.

improve AST comparasion on VarDecl & GotoStmt:
  1. VarDecl should not be ignored,
  2. GotoStmt has no children, it should be handle explicitly.

Reviewed By: donat.nagy

Differential Revision: https://reviews.llvm.org/D159519



Co-authored-by: huqizhi <836744285@qq.com>

The test is expecting watchpoint hits that are valid on aarch64
systems but not on Intel.  I need to update this test to match the
actual behavior on Intel macs, and don't want the CI bots to stay
broken while I get that figured out.

Enables the cost-benefit-analysis-based inliner by default if we have
sample profile.

No extra fix is required.

[Flang][OpenMP][MLIR] Fix memory leak caused by D149368 causing sanitizer error and fix iterator invalidation error

This patch fixes two issues introduced by the D149368 patch, one is
a memory leak from using the removeFromParent rather
than eraseFromParent (the erase also had to be moved to not create
use after deletes).

And the other is a possible iterator invalidation bug, better to be safe
than sorry.