Most of CGExprConstant.cpp is using the CharUnits abstraction
and is using getCharWidth() (directly of indirectly) when converting
between size of a char and size in bits. This patch is making that
abstraction more consistent by adding CharTy to the CodeGenTypeCache
(honoring getCharWidth() when mapping from char to LLVM IR types,
instead of using Int8Ty directly).

Reviewed By: rjmccall

Differential Revision: https://reviews.llvm.org/D94979

When using getByteArrayType the requested size is calculated in
char units, but the type used for the array was hardcoded to the
Int8Ty. This patch is using getCharWIdth a bit more consistently
by using getIntNTy in combination with getCharWidth, instead
of explictly using getInt8Ty.

Reviewed By: rjmccall

Differential Revision: https://reviews.llvm.org/D94977

This revision adds support for using either operand or result types to anchor an optional group. It also removes the arbitrary restriction that type directives must refer to variables in the same group, which is overly limiting for a declarative format syntax.

Fixes PR#48784

Differential Revision: https://reviews.llvm.org/D95109

Remove a few of my own TODOs that I will not have time to fix from lit
code.

If the call result is unused, we should let it get DCEd rather
than replacing it. Also, don't try to replace an existing sincos
with another one (unless it's as part of combining sin and cos).

This avoids an infinite combine loop if the calls are not DCEd
as expected, which can happen with D94106 and lack of willreturn
annotation in hand-crafted IR.

This is meant to unblock Chrome, as discussed in https://llvm.org/D74489.

Differential Revision: https://reviews.llvm.org/D95177

Differential Revision: https://reviews.llvm.org/D95201

Differential Revision: https://reviews.llvm.org/D93086

During the review of https://reviews.llvm.org/D93912 we failed to notice
the implementation status wasn't updated. This rectifies the issue.

representing dependence from producer result to consumer.

With Linalg on tensors the dependence between operations can be from
the result of the producer to the consumer. This change just does a
NFC refactoring of the LinalgDependenceGraphElem to allow representing
both OpResult and OpOperand*.

Differential Revision: https://reviews.llvm.org/D95208

Adding a lld option to support emitting pseudo probe metadata in LTO mode.

Reviewed By: MaskRay, wmi, wenlei

Differential Revision: https://reviews.llvm.org/D95056

* Remove an unimplemented and unused member function declaration
* Remove a misleading comment about an unrelated constraint number
* Fix a comment
* Add f18 crash message to "flang" driver script

Differential Revision: https://reviews.llvm.org/D95180

This is a continuation of https://reviews.llvm.org/D94239. I missed some other spellings of the same error.

Reviewed By: muiez

Differential Revision: https://reviews.llvm.org/D95246

Correct the analysis of references to transformational intrinsic
functions that have different semantics based on the presence or
absence of a DIM= argument; add shape analysis for UNPACK().

Differential Revision: https://reviews.llvm.org/D94716

In the motivating cases from https://llvm.org/PR48816 ,
we have a trailing trunc. But that is not required to
reduce the abs width:
https://alive2.llvm.org/ce/z/ECaz-p
...as long as we clear the int-min-is-poison bit (nsw).

We have some existing tests that are affected, and I'm
not sure what the overall implications are, but in general
we favor narrowing operations over preserving nsw/nuw.

If that causes problems, we could restrict this transform
based on type (shouldChangeType() and/or vector vs. scalar).

Differential Revision: https://reviews.llvm.org/D95235

https://llvm.org/PR48816

spv.Ordered/spv.Unordered are meant for OpenCL Kernel capability.
For Vulkan Shader capability, we should use spv.IsNan to check
whether a number is NaN.

Add a new pattern for converting `std.cmpf ord|uno` to spv.IsNan
and bumped the pattern converting to spv.Ordered/spv.Unordered
to a higher benefit. The SPIR-V target environment will properly
select between these two patterns.

Reviewed By: mravishankar

Differential Revision: https://reviews.llvm.org/D95237

Expressions emitted to module files and error messages
sometimes contain conversions of integer results of inquiry
intrinsics; these are usually not needed, and can conflict
with "int" in the user's namespace.  Improve folding so that
these conversions don't appear, and do some other clean-up
in adjacent code.

Differential Revision: https://reviews.llvm.org/D95172

Previously we only autogen the availability for ops that are
direct instantiating `SPV_Op` and expected other subclasses of
`SPV_Op` to define aggregated availability for all ops. This is
quite error prone and we can miss capabilities for certain ops.
Also it's arguable to have multiple levels of subclasses and try
to deduplicate too much: having the availability directly in the
op can be quite explicit and clear. A few extra lines of
declarative code is fine.

Reviewed By: mravishankar

Differential Revision: https://reviews.llvm.org/D95236

ObjCBOOLSummaryProvider was incorrectly treating BOOL as unsigned and this is now fixed.
Also adding tests for one bit bit-fields of BOOL and unsigned char.

This PR only has coro intrinsics needed for the Async to LLVM lowering. Will add other intrinsics as needed in the followup PRs.

Reviewed By: mehdi_amini

Differential Revision: https://reviews.llvm.org/D95143

The buckets are initialized in __kmp_dephash_create but when they are extended
the memory is allocated but not NULL'd, potentially leaving some buckets
uninitialized after all entries have been copied into the new allocation.
This commit makes sure the buckets are properly initialized with NULL before
copying the entries.

Differential Revision: https://reviews.llvm.org/D95167

If foo is referenced in any object file, bitcode file or shared object,
`__wrap_foo` should be retained as the redirection target of sym
(f96ff3c0).

If the object file defining foo has foo references, we cannot easily distinguish
the case from cases where foo is not referenced (we haven't scanned
relocations). Retain `__wrap_foo` because we choose to wrap sym references
regardless of whether sym is defined to keep non-LTO/LTO/relocatable links' behaviors similar
https://sourceware.org/bugzilla/show_bug.cgi?id=26358 .

If foo is defined in a shared object, `__wrap_foo` can still be omitted
(`wrap-dynamic-undef.s`).

Reviewed By: andrewng

Differential Revision: https://reviews.llvm.org/D95152

- Fix arguments name for subview and subtensor.
- Fix a typo in a comment of subtensor's method.

Reviewed By: nicolasvasilache

Differential Revision: https://reviews.llvm.org/D95211

[MLIR] Add support for extracting an integer sample point (if one exists) from an unbounded FlatAffineConstraints.

With this, we have complete support for finding integer sample points in FlatAffineConstraints.

Reviewed By: ftynse

Differential Revision: https://reviews.llvm.org/D95047

This patch implements codegen for __managed__ variable attribute for HIP.

Diagnostics will be added later.

Differential Revision: https://reviews.llvm.org/D94814

On z/OS, the following error message is not matched correctly in lit tests. This patch updates the CHECK expression to match the end period successfully.
```
EDC5129I No such file or directory.
```

Differential Revision: https://reviews.llvm.org/D94239

Fold vperm2x128(movddup(x),movddup(y)) -> movddup(vperm2x128(x,y))

[X86][AVX] canonicalizeLaneShuffleWithRepeatedOps - handle unary vperm2x128(permute/shift(x,c),undef) cases

Fold vperm2x128(permute/shift(x,c),undef) -> permute/shift(vperm2x128(x,undef),c)

Simplify vperm2x128(concat(X,Y),concat(Z,W)) folding.

Use collectConcatOps / ISD::INSERT_SUBVECTOR to find the source subvectors instead of hardcoded immediate matching.

The title says it all.

Differential Revision: https://reviews.llvm.org/D95233

The existing code did not deal with atomic loads correctly. Such loads
are represented as MemoryDefs. Bail out on any MemoryAccess that is not
a MemoryUse.

Because we were not looking for the llvm.coro.id.async intrinsic in the
early coro pass which triggers follow-up passes we relied on the
llvm.coro.end intrinsic being present. This might not be the case in
functions that end in unreachable code.

Differential Revision: https://reviews.llvm.org/D95144

[libomptarget][devicertl] Drop templated atomic functions

The five __kmpc_atomic templates are instantiated a total of seven times.
This change replaces the template with explictly typed functions, which
have the same prototype for amdgcn and nvptx, and implements them with
the same code presently in use.

Rolls in the accepted but not yet landed D95085.

The unsigned long long type can be replaced with uint64_t when replacing
the cuda function. Until then, clang warns on casting a pointer to one to
a pointer to the other.

Reviewed By: tianshilei1992

Differential Revision: https://reviews.llvm.org/D95093

Does not build in XCode:
http://green.lab.llvm.org/green/job/clang-stage1-RA/17963/consoleFull#-1704658317a1ca8a51-895e-46c6-af87-ce24fa4cd561

This reverts commit aabed371.

Iff we know we can get rid of the inversions in the new pattern,
we can thus get rid of the inversion in the old pattern,
this decreasing instruction count.

Note that we could position this transformation as just hoisting
of the `not` (still, iff y is freely negatible), but the test changes
show a number of regressions, so let's not do that.

Iff we know we can get rid of the inversions in the new pattern,
we can thus get rid of the inversion in the old pattern,
this decreasing instruction count.