Merge branch 'users/skatrak/spr/loop-nest-02-wrapper-iface' into users/skatrak/spr/loop-nest-03-taskloop-mlir

Merge branch 'users/skatrak/spr/loop-nest-01-mlir' into users/skatrak/spr/loop-nest-02-wrapper-iface

This patch updates the definition of `omp.taskloop` to enforce the restrictions
of a wrapper operation.

This patch defines a common interface to be shared by all OpenMP loop wrapper
operations. The main restrictions these operations must meet in order to be
considered a wrapper are:

- They contain a single region.
- Their region contains a single block.
- Their block only contains another loop wrapper or `omp.loop_nest` and a
terminator.

The new interface is attached to the `omp.parallel`, `omp.wsloop`,
`omp.simdloop`, `omp.distribute` and `omp.taskloop` operations. It is not
currently enforced that these operations meet the wrapper restrictions, which
would break existing OpenMP loop-generating code. Rather, this will be
introduced progressively in subsequent patches.

This patch introduces an operation intended to hold loop information associated
to the `omp.distribute`, `omp.simdloop`, `omp.taskloop` and `omp.wsloop`
operations. This is a stopgap solution to unblock work on transitioning these
operations to becoming wrappers, as discussed in
[this RFC](https://discourse.llvm.org/t/rfc-representing-combined-composite-constructs-in-the-openmp-dialect/76986).

Long-term, this operation will likely be replaced by `omp.canonical_loop`,
which is being designed to address missing support for loop transformations,
etc.

This was preventing us from folding away the vmerge into its mask.

Remove the canonicalization of `trunc` to `i1` according to the
suggestion of
https://github.com/llvm/llvm-project/pull/83829#issuecomment-1986801166

https://github.com/llvm/llvm-project/blob/a84e66a92d7b97f68aa3ae7d2c5839f3fb0d291d/llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp#L737-L745

Alive2: https://alive2.llvm.org/ce/z/cacYVA

Reimplement bitfield access unit computation for SysV ABIs. Considers
expense of unaligned and non-power-of-2 accesses.

Promote ARM & AArch64's HasUnaligned to TargetInfo and set for all
targets.

Verify bitfield access units.

This upstreams https://github.com/apple/llvm-project/pull/7930.

This adds a test to verify that we can apply attributes to C++ class
templates using API Notes. Doing the same for function templates is not
currently possible and requires more work.

https://github.com/llvm/llvm-project/pull/85592

Summary:
This patch removes most of the special handling from the
`PluginAdaptorTy` in preparation for changing this to be the
`GenericPluginTy`. Doing this requires that the OpenMP specific handling
of stuff like device offsets be contained within the OpenMP plugin
manager. Generally this was uninvasive expect for the change to tracking
the offset and size of the used devices. The eaiest way I could think to
do this was to use some maps, which double as indicators for which
plugins have devices active. This should not affect the logic.

DAGCombiner (or InstCombine) will convert an add to an or if the bits
are disjoint, which can prevent what was originally an (add {s,z}ext,
{s,z}ext) from being selected as a vwadd.

This teaches combineBinOp_VLToVWBinOp_VL to recover it by treating it as
an add.

This reduces the include time of <format> from 691ms to 556ms.

Fix the calls to `PrintedDeclCXX98Matches` to take the lambda function
as the last argument.

This reverts commit c64a328c.
This broke Arm32 bit build on various LLVM buildbots.
For example:
https://lab.llvm.org/buildbot/#/builders/17/builds/51129

Introduce `Decl::isFromExplicitGlobalModule` to replace the
`D->getOwningModule() && D->getOwningModule()->isExplicitGlobalModule()`
pattern to save some typings.

Found with -Wundefined-func-template.

apx-spec: https://cdrdv2.intel.com/v1/dl/getContent/784266
apx-syntax-recommendation:
https://cdrdv2.intel.com/v1/dl/getContent/817241

Revert "Reapply "[clang][nullability] allow _Nonnull etc on nullable class types (#82705)"" (#87041)

This reverts commit bbbcc1d9.

This change is causing the following build bots to fail due to a missing
header file:
- https://lab.llvm.org/buildbot/#/builders/188/builds/43765
- https://lab.llvm.org/buildbot/#/builders/176/builds/9428
- https://lab.llvm.org/buildbot/#/builders/187/builds/14696
- https://lab.llvm.org/buildbot/#/builders/186/builds/15551
- https://lab.llvm.org/buildbot/#/builders/182/builds/9413
- https://lab.llvm.org/buildbot/#/builders/245/builds/22507
- https://lab.llvm.org/buildbot/#/builders/258/builds/16026
- https://lab.llvm.org/buildbot/#/builders/249/builds/17221
- https://lab.llvm.org/buildbot/#/builders/38/builds/18566
- https://lab.llvm.org/buildbot/#/builders/214/builds/11735
- https://lab.llvm.org/buildbot/#/builders/231/builds/21947
- https://lab.llvm.org/buildbot/#/builders/230/builds/26675
- https://lab.llvm.org/buildbot/#/builders/57/builds/33922
- https://lab.llvm.org/buildbot/#/builders/124/builds/10311
- https://lab.llvm.org/buildbot/#/builders/109/builds/86173
- https://lab.llvm.org/buildbot/#/builders/280/builds/1043
- https://lab.llvm.org/buildbot/#/builders/283/builds/440
- https://lab.llvm.org/buildbot/#/builders/247/builds/16034
- https://lab.llvm.org/buildbot/#/builders/139/builds/62423
- https://lab.llvm.org/buildbot/#/builders/216/builds/36718
- https://lab.llvm.org/buildbot/#/builders/259/builds/2039
- https://lab.llvm.org/buildbot/#/builders/36/builds/44091
- https://lab.llvm.org/buildbot/#/builders/272/builds/12629
- https://lab.llvm.org/buildbot/#/builders/271/builds/6020
- https://lab.llvm.org/buildbot/#/builders/236/builds/10319

We can avoid libcalls.

Fixes #86205

This is the second attempt of 32fcfcdc,
which had to be reverted.

This should be good to go now, after a number of cleanups to not expose
headers from multiple targets:

- 6e58efac
- f1dff836
- 89ef3130
- eb70b485
- 83e5a123
- fce046ca

by default

For reduced BMI, it is meaningless to record the local declarations in
functions if not required explicitly during the process of writing the
function bodies.

It wastes time for reduced BMI and may be problematic if we want to
avoid transiting unnecessary changes.

This patch adds the nuw (no unsigned wrap) and nsw (no signed wrap)
poison-generating flags to the trunc instruction.

Discourse thread:
https://discourse.llvm.org/t/rfc-add-nowrap-flags-to-trunc/77453

This adds the ability to create a Scalar from an APFloat, and to create
an APFloat from an APSInt or another APFloat.

We currently check if the source and promoted types are not equal before
generating truncate instructions. This does not work for RV64 where the
promoted type is i64 and this lead to a crash due to the generation of
truncate instructions from i32 to i64.

Fixes #86400

size_t in PatchItem eliminates the need for casts.

Move all the stale profile matching stuffs into new files so that it can
be shared for unit testing.

In python3.11 there is a new environment variable PYTHONSAFEPATH which
stops python from setting the current directory as the first entry in
sys.path. Bazel started setting this to ensure that python targets
don't accidentally access things that aren't in their dependency tree.
This resulted in lit tests breaking because sys.path didn't include the
directory to the lit source files. This is fixed by adding the lit
binary to the dependency tree and propagating the import path from it.

Fixes https://github.com/llvm/llvm-project/issues/75963

Also another move comment to correct place.

represented by bitwidth without analysis.

Need to check that initial ext/trunc nodes can be safely represented
using calculated bitwidth before applying it.

node.