[qt/clang/llvm-project]: Summary of bulk changes made

Git repository change summary for qt/clang/llvm-project
Pushed by mirror-service into branch 'upstream/users/cchen/flang-metadirective-loop-core'.
Changed from cbfcc8420a9590500f8a3b4338489f497c9c17b1 to 6293a73ee198f54e7a6a64e25cfc51a8bc502633
Acknowledgement was received that this change introduces only existing code that has been pushed to another public open source repository.

This change contains the following new commits:

Git commit c5b02a931385d197ef634ae0c21102ea52626e18 by GitHub (on behalf of Dave Lee) on 03/08/2026 at 17:15..
[lldb][bytecode] Improve testing of bytecode section loading (#212568)

Add tests for loading embedded bytecode formatters from binaries.

Assisted-by: claude
https://invent.kde.org/qt/clang/llvm-project/-/commit/c5b02a931385d197ef634ae0c21102ea52626e18

Git commit dd9eb9a5c397d4e5f39a31b38013dc8731c5b64b by GitHub (on behalf of Justin Fargnoli) on 03/08/2026 at 17:20..
[LTO][CodeMetrics] Count indirect calls as inline candidates when `PrepareForLTO` is set (#192154)

Indirect calls may be resolved during the post-link LLVM pipeline. Thus, count them as potential inline candidates during the pre-link LTO phase.
https://invent.kde.org/qt/clang/llvm-project/-/commit/dd9eb9a5c397d4e5f39a31b38013dc8731c5b64b

Git commit 27cc956b00682b5e9162e7325edde51f264d69d6 by GitHub (on behalf of Alexey Bataev) on 03/08/2026 at 17:26..
[SLP]Keep tree-claimed peeled reassociated scalars alive

Peeled scalars are erased unconditionally, but a gather node may still
reference one in its generated buildvector. Keep such scalars (and the
peeled scalars in their operand chains) as plain scalar code.

Fixes #213674

Reviewers: 

Pull Request: https://github.com/llvm/llvm-project/pull/213724
https://invent.kde.org/qt/clang/llvm-project/-/commit/27cc956b00682b5e9162e7325edde51f264d69d6

Git commit 2ff5099be4ea2f88609c23037fdbe4e355739a04 by GitHub (on behalf of Florian Hahn) on 03/08/2026 at 17:34..
[ConstraintElim] Set NSW on sub when simplifying ssub.with.overflow. (#213532)

When simplifying ssub.with.overflow to a plain sub, we already proven
that the sub does not sign wrap. Replace it with `sub nsw`.

No changes on
https://github.com/dtcxzyw/llvm-opt-benchmark-nightly/pull/842, but
there is very little usage of ssub.with.overflow in default C/C++
builds.

It improves optimizations for code where ssub.with.overflow is used more
widely, like Clang + UBSan or Swift where checked arithmetic is the
default.

A simple C example is https://clang.godbolt.org/z/cv5MbYsrz

PR: https://github.com/llvm/llvm-project/pull/213532
https://invent.kde.org/qt/clang/llvm-project/-/commit/2ff5099be4ea2f88609c23037fdbe4e355739a04

Git commit fc1f3e7a26f202a3928daa4102027e4c31746513 by GitHub (on behalf of Arseniy Obolenskiy) on 03/08/2026 at 17:49..
Revert "Reland "[AMDGPU] Fix llvm.amdgcn.ballot with return width != wavefront size"" (#213713)

Reverts llvm/llvm-project#213635

buildbot failures are reported here:
https://github.com/llvm/llvm-project/pull/213635#issuecomment-5169081277
https://invent.kde.org/qt/clang/llvm-project/-/commit/fc1f3e7a26f202a3928daa4102027e4c31746513

Git commit 57965996196ca9caf62784392caf478a9d4133b3 by GitHub (on behalf of Alexey Bataev) on 03/08/2026 at 17:50..
[SLP]Allow unordered fadd reductions with reassoc only

The unordered reduction matcher required isAssociative() for fadd,
i.e. reassoc + nsz, a condition inherited from InstCombine-style
reassociation that also cancels and folds terms. Pure regrouping of
additions cannot change the sign of a zero result, so nsz is not
needed here: the reduction is seeded with the exact -0.0 identity,
the repeated-value multiplier preserves the sign of zero, and
constant folding is IEEE-exact. Brings the fadd requirement in line
with RecurrenceDescriptor, LV and with fmul, which accept reassoc
alone.

Reviewers: hiraditya, bababuck, RKSimon

Pull Request: https://github.com/llvm/llvm-project/pull/213261
https://invent.kde.org/qt/clang/llvm-project/-/commit/57965996196ca9caf62784392caf478a9d4133b3

Git commit d8952cde88ae46e2824dc12fefa2eb87c49686ca by GitHub (on behalf of Gábor Horváth) on 03/08/2026 at 18:02..
[clang][APINotes] Do not duplicate Escapable and Copyable attributes (#213716)

Applying them when they already present is a wasted allocation and can
also cause trouble in the Swift compiler that can complain about
duplicated attributes.

rdar://174868727
https://invent.kde.org/qt/clang/llvm-project/-/commit/d8952cde88ae46e2824dc12fefa2eb87c49686ca

Git commit 01935196807a3e816580d36616f5ec02ad8e50ef by GitHub (on behalf of Ramkumar Ramachandra) on 03/08/2026 at 18:04..
Partial-revert "[IR] Make semantics of strictfp consistent v2" (#213723)

This partially reverts 15bb4a97a7 ([IR] Make semantics of strictfp
consistent v2, #211769) due to a verifier failure in real-world code,
disabling just the verifier-check with a FIXME. The reason for a partial
manual-revert is because a clean revert doesn't apply cleanly: in
particular, the LangRef is untouched by this patch.

Ref:
https://github.com/llvm/llvm-project/pull/211769#issuecomment-5169150682
https://invent.kde.org/qt/clang/llvm-project/-/commit/01935196807a3e816580d36616f5ec02ad8e50ef

Git commit 79beb26b22e15932bfe85c91fcac3ce96ccc7c19 by GitHub (on behalf of Reid Kleckner) on 03/08/2026 at 18:17..
[clang-tidy][docs] Rename static analyzer check docs to Markdown (#211448)

Tracking issue: #201242
See the [migration guide] for more information. 

[migration guide]:
https://llvm.org/docs/SphinxQuickstartTemplate.html#markdown-migration-guidelines

This is the initial straight rename commit. It will probably break the
docs build, but it has to be a separate PR for blame preservation
purposes.
https://invent.kde.org/qt/clang/llvm-project/-/commit/79beb26b22e15932bfe85c91fcac3ce96ccc7c19

Git commit d2bfc1b9d50552533a099e4d5b6501068ec8b8d8 by GitHub (on behalf of Reid Kleckner) on 03/08/2026 at 18:34..
[clang-tidy][docs] Generate static analyzer check docs as Markdown (#211449)

Tracking issue: #201242
See the [migration guide] for more information. 

[migration guide]:
https://llvm.org/docs/SphinxQuickstartTemplate.html#markdown-migration-guidelines
This is a stacked PR based on #211448 , which will be a standalone
commit that
renames *.rst -> *.md before this PR lands for history preservation
purposes.

Unlike the other changes in this series, most of these changes are from
the generator script. Ignore all files under checks/ except `list.md`,
that is hand-modified. I confirmed that running the generator script
produces no changes.

Notably, the generator script *added new files*, since it seems there
are new checks since the script was run.
https://invent.kde.org/qt/clang/llvm-project/-/commit/d2bfc1b9d50552533a099e4d5b6501068ec8b8d8

Git commit c676c7b386d27f0d9a54799f050fc6a74f1b4166 by Chi-Chun, Chen on 03/08/2026 at 21:34..
[flang][OpenMP] Lower DO and SIMD variants in metadirectives

A standalone metadirective and its associated DO are sibling PFT evaluations,
so a selected loop replacement cannot directly reuse ordinary OpenMP loop
lowering. Runtime selection must also preserve exactly one copy of the loop in
each reachable branch. Temporarily associate the evaluations while lowering
to support DO, SIMD, and DO SIMD replacements without losing or duplicating the
ordinary fallback loop.

For example:

```fortran
!$omp metadirective &
!$omp& when(user={condition(flag)}: do) &
!$omp& otherwise(nothing)
do i = 1, n
  a(i) = i
end do
```

is represented as sibling evaluations:

```text
METADIRECTIVE
DO
`-- a(i) = i
```

and runtime selection lowers conceptually as:

```text
if (flag)
  omp.wsloop
    omp.loop_nest
else
  fir.do_loop
```

Build the statically applicable candidate set before lowering and use shared
selection utilities to preserve OpenMP ranking. Keep runtime guards separate
from static applicability, retain candidates reachable through a failed
condition, and apply loop-specific restrictions only after the reachable
replacements are known.

This change:

- Lowers DO, SIMD, and DO SIMD replacements through the existing loop-lowering
  path for standalone and begin/end metadirectives.
- Supports static selection and runtime user conditions while preserving a
  Fortran loop for NOTHING and other non-consuming replacements.
- Preserves compiler loop directives, statements following the associated
  loop, and evaluation ownership across alternate ENTRY lowering.
- Applies COLLAPSE and ORDERED to the required depth without leaking temporary
  loop-index attributes between replacement arms.
- Emits focused diagnostics when a reachable replacement requires unsupported
  data-environment, host-association, capture, or eager-privatization handling.

Initial loop support remains limited to DO, SIMD, and DO SIMD. Diagnose
combined data-environment loops such as PARALLEL DO, explicit data-sharing and
DEFAULT clauses, loops nested in an existing OpenMP data environment, TARGET
and mixed block/loop replacements, non-canonical or unstructured loops,
insufficient COLLAPSE or ORDERED depth, and unsupported loop-index entities.

Add lowering coverage for candidate ranking, runtime fallback paths, DO and
SIMD composition, affected-loop depth, compiler directives, ENTRY handling,
and the unsupported boundaries above.

Assisted with codex.
https://invent.kde.org/qt/clang/llvm-project/-/commit/c676c7b386d27f0d9a54799f050fc6a74f1b4166

Git commit 6293a73ee198f54e7a6a64e25cfc51a8bc502633 by Chi-Chun, Chen on 03/08/2026 at 22:03..
Consolidate metadirective lowering TODO tests

Group related block, loop data-environment, iteration-variable, and unsupported-replacement cases into split-file tests. This keeps each diagnostic isolated while reducing the number of TODO test files.
https://invent.kde.org/qt/clang/llvm-project/-/commit/6293a73ee198f54e7a6a64e25cfc51a8bc502633