Apparently afn doesn't allow you to drop the denormal handling
according to OpenCL conformance. This was hidden by losing the flags
during the library linking process. Fast log is still broken and needs
more work.

https://reviews.llvm.org/D157936

This reverts commit 5e28d30f.

This reverts commit b37c7ed0.

This is a follow-up to https://reviews.llvm.org/D156889

Downstream projects may have more complicated ops than the control flow ops upstream and therefore need a more powerful interface to support the lifting process. Use cases include the propagation of (inherent) metadata that was previously on the control flow ops and now needs to be lifted to structured control flow ops.
Since the lifting process is inherently non-local in respect to the function-body, we require stronger guarantees from the interface.

This patch therefore makes two changes to the interface:
* Passes the terminator that is being replaced to `createStructuredBranchRegionTerminatorOp`
* Adds as precondition to `createCFGSwitchOp` that its predecessors are already correctly established

Asserts have been added to verify these were it makes sense and to correctly state intent. I have not added tests purely because testing preconditions like these is not really feasible (and incredibly specific).

Differential Revision: https://reviews.llvm.org/D157981

Abort the transform instead of asserting.

Fixes https://github.com/llvm/llvm-project/issues/64661.

Align operand bundles can contain non-power-of-two alignments,
but LLVM otherwise does not support them. Bail out in that case
instead of crashing.

Fixes https://github.com/llvm/llvm-project/issues/64687.

Currently, the `-fprofile-udpate` is ignored when `-fprofile-generate` is in effect. This patch enables `-fprofile-update` for `-fprofile-generate`. This patch continues the work from https://reviews.llvm.org/D87737, which added `-fprofile-update` in the first place.

Reviewed By: MaskRay

Differential Revision: https://reviews.llvm.org/D157280

This patch adds code emission in emitTargetCall to call the OpenMP runtime to
launch an kernel, and to call the fallback host implementation if the launch
fails.

Reviewed By: TIFitis, kiranchandramohan, jdoerfert

Differential Revision: https://reviews.llvm.org/D155633

This is useful for any downstream users who may just want to use slightly different ops than the pass or need to take into account other ops in the input while still dealing with ControlFlow ops for the most part.
This also helps writing test implementations for `transformCFGToSCF`.

Since the implementation is now public, comments noting the post conditions of the methods have been added to ensure downstream users can stay compatible with the existing implementation.

Differential Revision: https://reviews.llvm.org/D157971

Reapply: [MemCpyOpt] implement single BB stack-move optimization which unify the static unescaped allocas

This reverts commit 20771802.

This revision is needed to support bufferization of `cf.br`/`cf.cond_br`. It will also be useful for better analysis of loop ops.

This revision generalizes `getAliasingOpResults` to `getAliasingValues`. An OpOperand can now not only alias with OpResults but also with BlockArguments. In the case of `cf.br` (will be added in a later revision): a `cf.br` operand will alias with the corresponding argument of the destination block.

If an op does not implement the `BufferizableOpInterface`, the analysis in conservative. It previously assumed that an OpOperand may alias with each OpResult. It now assumes that an OpOperand may alias with each OpResult and each BlockArgument of the entry block.

Differential Revision: https://reviews.llvm.org/D157957

Add a pattern that splits one dealloc operation into multiple dealloc operation
depending on static aliasing information of the values in the `memref` operand
list. This reduces the total number of aliasing checks required at runtime and
can enable futher canonicalizations of the new and simplified dealloc
operations.

Depends on D157407

Reviewed By: springerm

Differential Revision: https://reviews.llvm.org/D157508

This patch adds support for translating `teams` OpenMP directives to MLIR, when
appearing as either loop or block constructs and as part of combined constructs
or on its own.

The current Fortran parser does not allow the specification of the optional
lower bound for the "num_teams" clause, so only the `num_teams_upper` MLIR
argument is set by this patch.

Depends on D156809

Differential Revision: https://reviews.llvm.org/D156884

This patch extracts MLIR codegen logic from various types of OpenMP constructs
and places it into operation-specific functions. This refactoring mainly
targets block constructs and unifying logic for directives that can appear on
their own as well as combined with others.

The processing of clauses that do not apply to the directive being processed is
avoided and code duplication for combined constructs is reduced.

Depends on D156455.

Differential Revision: https://reviews.llvm.org/D156809

The attributes changes were left out of Clang 17.
Attributes that used to take a string literal now accept an unevaluated
string literal instead, which means they reject numeric escape sequences
and strings literal with an encoding prefix - but the later was already
ill-formed in most cases.

We need to know that we are going to parse an unevaluated string literal
before we do - so we can reject numeric escape sequence,
so we derive from Attrs.td which attributes parameters are expected
to be string literals.

Reviewed By: aaron.ballman

Differential Revision: https://reviews.llvm.org/D156237

This patch adds support for lowering target-related combined constructs from
PFT to MLIR. The lowering of OpenMP loop constructs is generalized in
preparation for later supporting different combinations of target, teams,
distribute, parallel and simd directives.

Currently enabled by this patch are the following combined constructs:
- do simd
- parallel do simd
- target parallel do simd
- target parallel do
- target simd

Depends on D155981 and D157090.

Differential Revision: https://reviews.llvm.org/D156455

This specifies the developer policy on adding links in source & test
files and commit messages, related to discussion at:
https://discourse.llvm.org/t/code-review-reminder-about-links-in-code-commit-messages/71847

The intent is to discourage adding links to resources that are not
available to the community as a whole (dead links, links to internal
documentation, links to internal bug trackers, etc) from source and
test files, while still allowing such links to appear in other contexts
as needed. It suggests to instead add sufficient context in the
surrounding comments to make such links unnecessary.

It also clarifies that these links can appear in commit messages as
metadata (similar to how we already have Differential Revision and
Fixes metadata with links).

Differential Revision: https://reviews.llvm.org/D155081

This patch adds semantics checks for REQUIRES directives appearing after other
directives that are affected by them. In particular, it adds checks for device
constructs appearing after device-related REQUIRES directives and for the
`atomic_default_mem_order` clause appearing after atomic operations where the
memory order is not explicitly specified.

This is patch 2/5 of a series splitting D149337 to simplify review.

Depends on D157710.

Differential Revision: https://reviews.llvm.org/D157722

Revert "Reland "[Clang][SemaCXX] Add unused warning for variables declared in condition expressions""

This causes a lot of warning in sanitizer build:
https://lab.llvm.org/buildbot/#/builders/258/builds/5424
https://lab.llvm.org/buildbot/#/builders/36/builds/36560

This reverts commit 8e329caa.

This patch fixes a problem in the representation of the clause list for the
REQUIRES directive. Location information was not present, making error messages
not very descriptive.

This is patch 1/5 of a series splitting D149337 to simplify review.

Differential Revision: https://reviews.llvm.org/D157710

The bufferization implementation of `func.func` and `func.call` can be simplified. It still contained code that was necessary when One-Shot Bufferize removed return values. This functionality has been extracted into a separate pass a while ago.

Differential Revision: https://reviews.llvm.org/D157893

If fmul's constant operand is the reciprocal of a power of 2 (i.e 1/2^n) or
fdiv's constant operand is power of 2, we can try to match patterns with
[su]int_to_fp for [su]cvtf.

Differential Revision: https://reviews.llvm.org/D156538

Report an error when trying to bufferize an op that contains unstructured control flow but for ops for which the bufferization implementation does not support unstructured control flow. At the moment, there are no ops for which unstructured control flow is supported.

Differential Revision: https://reviews.llvm.org/D157893

Differential Revision: https://reviews.llvm.org/D157633

This patch marks the declarations with initializations in condition expressions such as
if (int var = init) as unreferenced so that -Wunused can warn on them.

Fixes https://github.com/llvm/llvm-project/issues/61681

Reviewed By: cor3ntin
Differential Revision: https://reviews.llvm.org/D152495

This reverts commit 4e3b8948, with
fixes for places I'd missed updating in lld and lldb. I've also
renamed OptionVisibility::Default to "DefaultVis" to avoid ambiguity
since the undecorated name has to be available anywhere Options.inc is
included.

Original message follows:

This splits OptTable's "Flags" field into "Flags" and "Visibility",
updates the places where we instantiate Option tables, and adds
variants of the OptTable APIs that use Visibility mask instead of
Include/Exclude flags.

We need to do this to clean up a bunch of complexity in the clang
driver's option handling - there's a whole slew of flags like
CoreOption, NoDriverOption, and FlangOnlyOption there today to try to
handle all of the permutations of flags that the various drivers need,
but it really doesn't scale well, as can be seen by things like the
somewhat recently introduced CLDXCOption.

Instead, we'll provide an add...

An empty struct is handled as a struct with a dummy i8, on all targets.

Most targets treat an empty struct return value as essentially
void - but some don't. (Currently, at least x86_64-windows-* and
powerpc64le-* don't treat it as void.)

When intializing a struct with such a no_unique_address member,
make sure we don't write the dummy i8 into the struct where there's
no space allocated for it.

Previously it would clobber the actual valid data of the struct.

Fixes https://github.com/llvm/llvm-project/issues/64253, and
possibly https://github.com/llvm/llvm-project/issues/64077
and https://github.com/llvm/llvm-project/issues/64427 as well.

We should omit the store for any empty record (not only ones
declared with no_unique_address); we can have a situation where a
class doesn't have the no_unique_address attribute, but is embedded
in an outer struct with the no_unique_address attribute - like this:

    struct S {};
    S f();
    struct S2 : public S { S2();};
    S2::S2() : S(f()) {}
    struct S3 { int x; [[no_unique_address]] S2 y; S3(); };
    S3::S3() : x(1), y() {}

Here, the problematic store (which this patch omits) is in
the constructor of S2. In the case of S3, S2 has no valid storage
and aliases x - thus the constructor of S2 should omit the dummy
store.

Differential Revision: https://reviews.llvm.org/D157332

Modify check to ignore any parent typeLoc and
other unevaluated context.

Fixes: #35857

Reviewed By: carlosgalvezp

Differential Revision: https://reviews.llvm.org/D157374

Use isStatic instead of isStaticStorageClass to properly
handle a out-of-line definitions.

Fixes: #51861

Reviewed By: carlosgalvezp

Differential Revision: https://reviews.llvm.org/D157326

Follow patches will add implementation.

For https://github.com/llvm/llvm-project/issues/30098

Reviewed By: Enna1, phosek

Differential Revision: https://reviews.llvm.org/D157921

The vector arguments are split but assignments won't be pending.

Fixes #64645

Reviewed By: asb

Differential Revision: https://reviews.llvm.org/D157847

D144048 has added preferred function and loop alignment to
RISCVSubtarget, but now we need to set them manually for
different processors.

Tune features that set preferred function/loop align to
[2, 64] bytes (align 1 is not here since the min align is 2)
are added. These features can be used in processor
definitions.

Reviewed By: asb

Differential Revision: https://reviews.llvm.org/D157832

Add canonicalization pattern for index::CmpOp

Differential Revision: https://reviews.llvm.org/D157903

`llvm-cov convert-for-testing` is used to build the .covmapping files used in its regression tests. However the current implementation only works when there's only one source file in the mapping information data. If there are more than 1 source files, `llvm-cov convert-for-testing` can still produce a .covmapping file, but when read it back, `llvm-cov` will report:

```
error: Failed to load coverage: 'main.covmapping': Malformed coverage data
```

This is because the output .covmapping file doesn't have any mark to indicate the boundary between file records and function records, and current implementation jsut assume there's only one file record in the .covmapping file.

Changes to the code:

- Make `llvm-cov convert-for-testing` output a LEB128 number before file records to indicate its size.
- Change the testing format parsing code correspondingly.
- Update existing .covmapping files.

Reviewed By: gulfem

Differential Revision: https://reviews.llvm.org/D156611

Differential Revision: https://reviews.llvm.org/D157771