[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/main'.
Changed from e04cf35fc64cfadc921582c7e8d962dfcb2c4e3b to d814be1bb794baf0650158a6bd1dda23f4f86e99
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 21fb4bdd1e761021920790699efbfa56aaf081e2 by GitHub (on behalf of Nikhil Kotikalapudi) on 23/07/2026 at 05:31..
[M68k][CodeGen] pipeline failure fix (#211458)

Fixing pipeline mismatch on M68k (experimental target, failure shouldnt
have been reported).
Triggered upon merge of this PR:
https://github.com/llvm/llvm-project/pull/210826

Failure here:
https://lab.llvm.org/buildbot/#/builders/27/builds/2162/steps/5/logs/FAIL__LLVM__pipeline_ll
https://invent.kde.org/qt/clang/llvm-project/-/commit/21fb4bdd1e761021920790699efbfa56aaf081e2

Git commit 6aa4d5ef9b4eca1dc6b10a5ef4c17fec2676b9f4 by GitHub (on behalf of Mariusz Sikora) on 23/07/2026 at 05:37..
[AMDGPU] Add FeatureSubtarget for global async load/store lds (#211247)
https://invent.kde.org/qt/clang/llvm-project/-/commit/6aa4d5ef9b4eca1dc6b10a5ef4c17fec2676b9f4

Git commit 450cf0aafeab9fed6666b112b018d3913a36527c by GitHub (on behalf of HighW4y2H3ll) on 23/07/2026 at 05:46..
[llvm-profgen] Formatting StringRef iteration in isLBRSample (#211462)

Follow up on https://github.com/llvm/llvm-project/pull/211182 Cleanup
the StringRef iteration.
https://invent.kde.org/qt/clang/llvm-project/-/commit/450cf0aafeab9fed6666b112b018d3913a36527c

Git commit 22cb41c9ad9658319fb211b04f92806c504a8388 by GitHub (on behalf of Sean Clarke) on 23/07/2026 at 05:49..
[LoopIdiom] Use narrower bit widths where possible in `optimizeCRCLoopUsingClmul` (#210139)

The original implementation of `optimizeCRCLoopUsingClmul` (#203405)
uses a single conservative bit width for most operations, but this width
is not always necessary. Use more restrictive bit widths for each clmul
according to their inputs, and narrow the bit width for the initial
CRC/data setup.
https://invent.kde.org/qt/clang/llvm-project/-/commit/22cb41c9ad9658319fb211b04f92806c504a8388

Git commit 87304516a10cfbd148d0ad23fd03e7f838aed267 by GitHub (on behalf of Pankaj Dwivedi) on 23/07/2026 at 05:51..
[Reassociate] Use UniformityInfo to group uniform operands together (#198507)

On targets with branch divergence (TTI.hasBranchDivergence()), boost the
rank of divergent operands in ReassociateExpression() so they sort
towards the root of the expression tree. This clusters uniform operands
at the leaves, letting the backend lower the inner sub-expression to a
scalar ALU instruction (s_mul/s_or/s_and) instead of a vector.

Ref:
https://github.com/llvm/llvm-project/pull/175167#issuecomment-4476483692
https://invent.kde.org/qt/clang/llvm-project/-/commit/87304516a10cfbd148d0ad23fd03e7f838aed267

Git commit da27aac10e157e5659ba6dfa0e362b570a236978 by GitHub (on behalf of SiHuaN) on 23/07/2026 at 05:55..
[RISCV][P-ext] Add packed sign and zero extend intrinsics (#208685)

Add SelectionDAG and intrinsic support for the RISC-V P packed sign- and
zero-extend operations.

Sign-extend is matched through `sext_inreg` (folded from the `shl`/`ashr`
idiom by the generic combiner) and selects to `psext.*`. Zero-extend reuses
the existing `PPAIRE` node with a zero operand instead of a new target node,
since `pzext.* rd, rs` is an alias of `ppaire.* rd, rs, x0`: implementing
`isVectorClearMaskLegal` lets the generic `DAGCombiner::XformToShuffleWithZero`
rewrite the `(and vXi16, 0xff)` / `(and vXi32, 0xffff)` idiom into a shuffle
with a zero operand, which `lowerVECTOR_SHUFFLEAsPPair` folds to `PPAIRE`.

Also adds the LLVM IR intrinsics `llvm.riscv.psext.b`/`.h` and
`llvm.riscv.pzext.b`/`.h`. For RV64 32-bit packed results, the lowering widens
to the legal 64-bit type and extracts the low subvector.
https://invent.kde.org/qt/clang/llvm-project/-/commit/da27aac10e157e5659ba6dfa0e362b570a236978

Git commit f7ae0ec1b5821ff219e70218b6877bb977be06b7 by GitHub (on behalf of Damian Höster) on 23/07/2026 at 05:59..
[clang-format][clang-tools-extra] Fix leaking of host-level config into test environment (#206358)

These leaks were causing false test failures due to my global
.clang-format being in an ancestor directory of the build directory.
These changes prevent those issues when running the tests on my system.
I used MiMo v2.5 Pro agentically in Zed to find and fix the issue. But I
understand the changes and judge them to be safe and a clear improvement
over the status quo.
https://invent.kde.org/qt/clang/llvm-project/-/commit/f7ae0ec1b5821ff219e70218b6877bb977be06b7

Git commit 72b764559fd2d5f564f2c6f42edd782c8e20616d by GitHub (on behalf of Ritanya-B-Bharadwaj) on 23/07/2026 at 06:05..
[flang][OpenMP] Add lowering for assume and assumes directives (#205615)

Adds lowering for the assume and assumes directives. holds clauses are lowered to llvm.assume, the other assumption clauses are ignored, and assumes is a no-op. Hints are skipped under -fopenmp-simd.
https://invent.kde.org/qt/clang/llvm-project/-/commit/72b764559fd2d5f564f2c6f42edd782c8e20616d

Git commit 0a79ab0b7a7fe38f20d0f5df11b2530ff1c0c0cc by GitHub (on behalf of Kareem Ergawy) on 23/07/2026 at 06:06..
[flang][PFT-to-MLIR] Default wrap-unstructured-constructs-in-execute-region to off (#211290)
https://invent.kde.org/qt/clang/llvm-project/-/commit/0a79ab0b7a7fe38f20d0f5df11b2530ff1c0c0cc

Git commit 23de8d650e3d1a6f0e40121c87be3030c3005a32 by GitHub (on behalf of Santanu Das) on 23/07/2026 at 06:29..
Add missing opcode for offset validity check (#211186)

During offset validity check, the common API in HexagonInstrInfo does
not handle the opcode. Adding it fixes wherever offset calculation is
made for the instruction.

Co-authored-by: quic-santdas <[email protected]>
https://invent.kde.org/qt/clang/llvm-project/-/commit/23de8d650e3d1a6f0e40121c87be3030c3005a32

Git commit ec08887025038ab7e50f46bb2f9ea3b3c1fafce4 by GitHub (on behalf of Cullen Rhodes) on 23/07/2026 at 06:45..
[CodeGen] Precompute pressure-set register classes (#210751)

Profiling sqlite on aarch64-O3 shows ~1.26% of compile-time is spent in
RegisterClassInfo::computePSetLimit. This is called from
ScheduleDAGMILive::initRegPressure via
RegClassInfo->getRegPressureSetLimit.

computePSetLimit scans every register class and its pressure-set list to
find the class with the largest RegClassWeight::WeightLimit. This is
cached, but must be recomputed on every cache invalidation.

AArch64 has 530 register classes and 190 pressure sets. Populating all
pressure-set limits visits 530 x 190 = 100,700 register classes. This
can be precomputed in TableGen, while retaining the MF-dependent
reserved-register adjustment in computePSetLimit.

CTMark results [1]:
- stage1-aarch64-O3: -1.85% geomean; Bullet -4.01%.
- stage1-O3 (x86): -0.04%; Bullet -0.13%.

AMDGPU already suppresses pressure-set generation for most register
classes for compile-time reasons [2], so impact there should be minimal.

Assisted-by: codex

[1] https://llvm-compile-time-tracker.com/compare.php?from=4aa1590ad66a1bff9b3c74b4a2c7366473015b4c&to=d78521eab6e7aa17561d2decc0c4dff55127c955&stat=instructions%3Au
[2] https://reviews.llvm.org/D74744
https://invent.kde.org/qt/clang/llvm-project/-/commit/ec08887025038ab7e50f46bb2f9ea3b3c1fafce4

Git commit 5152c51477aad031eeac540a6f02ed1f6539c710 by GitHub (on behalf of Jeff Bailey) on 23/07/2026 at 06:48..
[libc] Add err.h to Linux public header target lists (#210995)

Add libc.include.err to TARGET_PUBLIC_HEADERS for Linux targets
(aarch64, arm, riscv, x86_64) so that err.h gets generated and
installed.

Assisted-by: Automated tooling, human reviewed.
https://invent.kde.org/qt/clang/llvm-project/-/commit/5152c51477aad031eeac540a6f02ed1f6539c710

Git commit e407c4b6778909a90582aa6b59735866554cab5c by GitHub (on behalf of Mariusz Sikora) on 23/07/2026 at 06:49..
[AMDGPU] Add gfx13 support for global_load_async_to_lds (#211248)
https://invent.kde.org/qt/clang/llvm-project/-/commit/e407c4b6778909a90582aa6b59735866554cab5c

Git commit bc3d3c9e3af37fab009d3fa10349f56b150e69d4 by GitHub (on behalf of Marco Elver) on 23/07/2026 at 06:52..
[AllocToken] Fix typeContainsPointer for references, PMFs, blocks, ObjC pointers (#211349)

When evaluating whether an allocated type contains a pointer to generate
the alloc_token metadata, typeContainsPointer() previously only checked
for PointerType.

Expand typeContainsPointer() to check isAnyPointerType(),
isReferenceType(), isMemberFunctionPointerType(), and
isBlockPointerType().
https://invent.kde.org/qt/clang/llvm-project/-/commit/bc3d3c9e3af37fab009d3fa10349f56b150e69d4

Git commit 44dd2956162b4727b4e5675a8160a0a5f80aa791 by GitHub (on behalf of Shoreshen) on 23/07/2026 at 06:54..
[AMDGPU] Using feature to define vmem cacheline size (#207884)

Co-authored-by: shore <[email protected]>
Co-authored-by: Matt Arsenault <[email protected]>
https://invent.kde.org/qt/clang/llvm-project/-/commit/44dd2956162b4727b4e5675a8160a0a5f80aa791

Git commit 520a930da26c01505fe012a71d24f6ad9443c9c9 by GitHub (on behalf of Ian Anderson) on 23/07/2026 at 07:13..
[llvm-objdump] Factor the load command prologue into a shared function (#210881)

Move the repeated code to print cmd and cmdsize into a shared function.
https://invent.kde.org/qt/clang/llvm-project/-/commit/520a930da26c01505fe012a71d24f6ad9443c9c9

Git commit c151d33acea677cee0c29ec15f9d5f6c7d31d0fd by GitHub (on behalf of Anutosh Bhat) on 23/07/2026 at 07:15..
[MLIR][Python] Do not install ExecutionEngine wrapper when disabled (#211250)

The `_mlirExecutionEngine` extension is declared only when
`MLIR_ENABLE_EXECUTION_ENGINE` is enabled. However,
`execution_engine.py` and its type stub are currently added to the
Python package unconditionally.

This leaves an installed `mlir.execution_engine` module that fails
immediately when imported because its native extension is absent. It
also exposes a type stub for an unavailable API.

Making sure they're added only if `MLIR_ENABLE_EXECUTION_ENGINE` is
enabled.
https://invent.kde.org/qt/clang/llvm-project/-/commit/c151d33acea677cee0c29ec15f9d5f6c7d31d0fd

Git commit b605aa6ac3fcd85a608e7ca4ab5b69109ae1272e by GitHub (on behalf of Madhur Amilkanthwar) on 23/07/2026 at 07:28..
[LoopDistribute][NFC] Autogenerate checks for some tests (#211471)
https://invent.kde.org/qt/clang/llvm-project/-/commit/b605aa6ac3fcd85a608e7ca4ab5b69109ae1272e

Git commit 3b7447e00baff23660b49fdcb67ce83a3bf94605 by GitHub (on behalf of Nikolas Klauser) on 23/07/2026 at 07:46..
[libc++] Optimize standard streams with sync_with_stdio(false) (#209161)

This updates the standard streams so that the buffers are replaced with
`basic_filebuf`s.

Fixes #21566

---------

Co-authored-by: Louis Dionne <[email protected]>
https://invent.kde.org/qt/clang/llvm-project/-/commit/3b7447e00baff23660b49fdcb67ce83a3bf94605

Git commit 355dac7c4997c891d0a917935dcccc80334f5392 by GitHub (on behalf of Lukas Sommer) on 23/07/2026 at 07:56..
[AMDGPU] Preserve liveness flags in ILP scheduler (#211190)

During DAG construction, liveness flags such as `undef` are removed from
the instructions. When the scheduler succeeds, the flags are recomputed
and restored.

However, so far, if the scheduler failed to meet the occupancy target,
it would not restore the liveness flags, leading to verification
failure/assertions later in the pipeline.

This change restores the liveness flags in cases where no schedule
meeting the occupancy target could be found.

---------

Signed-off-by: Lukas Sommer <[email protected]>
https://invent.kde.org/qt/clang/llvm-project/-/commit/355dac7c4997c891d0a917935dcccc80334f5392

Git commit 14371a71392de5c1426a5dd98a98deb2a5d26fbe by GitHub (on behalf of Stanislav Mekhanoshin) on 23/07/2026 at 08:08..
[AMDGPU] Simplify check for packed fp32/64 opsel bits. NFCI (#211374)

Call isLegalGFX12PlusPackedMathFP32or64BitOperand directly instead
of the whole isLegalOperand.
https://invent.kde.org/qt/clang/llvm-project/-/commit/14371a71392de5c1426a5dd98a98deb2a5d26fbe

Git commit beca2e90e6bc75f517fa4b78ac76de6141fc558f by GitHub (on behalf of Ricardo Jesus) on 23/07/2026 at 08:13..
[AArch64] Prefer (sub x, -c) over (add x, c) if -c is cheaper. (#211020)

The negation of an immediate can be cheaper to materialise than the
original immediate, in which case a subtraction is preferable.
https://invent.kde.org/qt/clang/llvm-project/-/commit/beca2e90e6bc75f517fa4b78ac76de6141fc558f

Git commit 64b593c2371b7f7225b0ec190a37cd8b672e4c5d by GitHub (on behalf of Balázs Benics) on 23/07/2026 at 08:13..
[clang][SSAF] Allow no Data for summary analysis (#211342)

Extractors might not always add summaries for all the TUs. If they don't
add any summaries to the TU result, then loading this summary for a
summary analysis would cause a failure.

This patch handles the situation gracefully by accepting such cases.

Split from: #209354

---------

Co-authored-by: Ziqing Luo <[email protected]>
https://invent.kde.org/qt/clang/llvm-project/-/commit/64b593c2371b7f7225b0ec190a37cd8b672e4c5d

Git commit d3fead5912f73a84bd33bf1034175f6fa8de4e76 by GitHub (on behalf of Nikolas Klauser) on 23/07/2026 at 08:17..
[libc++] Remove some adjacent_view::begin() benchmarks (#210265)

Benchmarking a bunch of very similar values has rather little benefit,
since they behave essentially the same.
https://invent.kde.org/qt/clang/llvm-project/-/commit/d3fead5912f73a84bd33bf1034175f6fa8de4e76

Git commit 0e780450034e4bf0c6aa6dd4b312e1e0b9a20689 by GitHub (on behalf of Florian Hahn) on 23/07/2026 at 08:21..
[Matrix] De-duplicate reshaped matrixes used as incoming values for phi. (#211210)

Phis can have multiple incoming entries for the same block. In that
case, all incoming values for the block must be the same.

Update visitPHI to avoid expanding the incoming matrix multiple times
for the some incoming block.

Fixes a verifier error for the newly added test case.

PR: https://github.com/llvm/llvm-project/pull/211210
https://invent.kde.org/qt/clang/llvm-project/-/commit/0e780450034e4bf0c6aa6dd4b312e1e0b9a20689

Git commit 27ec80f230b86f22bd7329d67965b48d41b18f92 by GitHub (on behalf of L-roro) on 23/07/2026 at 08:23..
[Hexagon] Avoid spurious high vmem for contained sub-HVX stores (#204661)

When a sub-HVX store is widened into a masked HVX store, the unaligned
masked store lowering can split it into two vector stores: one at Base
and one at Base+HwLen. For stores whose original memory size fits within
the guaranteed alignment, the high half predicate is known to be all
false, so the Base+HwLen store is unnecessary.

Even an all-false predicated vmem can still probe the TLB. Emitting the
high store can therefore fault when Base+HwLen is on an unmapped page,
even though no bytes should be written there.

Preserve the original memory VT when widening sub-HVX stores, and use
that size during masked-store lowering to elide the empty high vmem.

Relevant links:
https://docs.qualcomm.com/doc/80-N2040-60/topic/conditional-execution.html#consuming-scalar-predicates
and
https://docs.qualcomm.com/doc/80-N2040-61/topic/memory.html#permissions

Co-authored-by: L-roro <[email protected]>
https://invent.kde.org/qt/clang/llvm-project/-/commit/27ec80f230b86f22bd7329d67965b48d41b18f92

Git commit 1ee1c0edd6e11d13843b42846e9e5445232abc07 by GitHub (on behalf of Phoebe Wang) on 23/07/2026 at 08:27..
[X86][CCMP] Reorder Comparison Trees to Facilitate CSE (#211476)

This patch ports AArch64's CMP CSE optimization #168064

Assisted-by: Claude Opus 4.8
https://invent.kde.org/qt/clang/llvm-project/-/commit/1ee1c0edd6e11d13843b42846e9e5445232abc07

Git commit 5bf9d6127d3b46f4213882ac9454e932543f14be by GitHub (on behalf of Sander de Smalen) on 23/07/2026 at 08:32..
[AArch64] Add SVE lowering for FP_TO_[US]INT_SAT. (#207201)
https://invent.kde.org/qt/clang/llvm-project/-/commit/5bf9d6127d3b46f4213882ac9454e932543f14be

Git commit 9c5858eb699390c2cac5f7f032279be275fa6aef by GitHub (on behalf of Pengcheng Wang) on 23/07/2026 at 08:37..
[RISCV] Add TuneNoDefaultUnroll to generic CPUs (#135318)

Almost all CPUs have added this and experiments have shown
considerable gains.

And, we set a lower runtime unrolling count 4 for in-order models
to avoid potential regressions.

Fixes #134272.
https://invent.kde.org/qt/clang/llvm-project/-/commit/9c5858eb699390c2cac5f7f032279be275fa6aef

Git commit 4138dc431ad849bd8c0e76f907345c6d0507890b by GitHub (on behalf of Abid Qadeer) on 23/07/2026 at 08:57..
[MLIR][OpenMP] Preserve debug location in OpenMPIRBuilder calls. (#211254)

Many `OpenMPIRBuilder` entry points take an
`OpenMPIRBuilder::LocationDescription`. It has two relevant
constructors:

`LocationDescription(const IRBuilderBase &IRB)` // captures IP and debug
loc
`LocationDescription(const InsertPointTy &IP)` // captures IP only; DL
is empty

The OpenMP MLIR-to-LLVM-IR translation constructs the location from
`builder` in almost all places (~45 call sites), which selects the first
constructor and propagates both the insertion point and the current
debug location. A few call sites instead passed `builder.saveIP()`,
which selects the second constructor and silently drops the debug
location.

Change the 9 offending call sites to pass `builder` instead of
`builder.saveIP()`, so the debug location is preserved.
https://invent.kde.org/qt/clang/llvm-project/-/commit/4138dc431ad849bd8c0e76f907345c6d0507890b

Git commit a2b42b10a17f93392f13a67a35309e57bc87be0f by GitHub (on behalf of Wenju He) on 23/07/2026 at 08:59..
[IR] Reject inlining strictfp alwaysinline callee into non-strictfp caller (#210701)

Commit af0c933eda31 added a verifier check requiring every strictfp call
site to reside in a strictfp function.
It caused verify error `call site marked strictfp without caller
function marked strictfp` since AlwaysInliner force-inlines strictfp
callees into non-strictfp callers, bypassing checkStrictFP.

47b3b76825dc explictly not implemented inlining strictfp function into
non-strictfp function. But the restriction was lost in 8eb6757564cc.
8eb6757564cc was only NFC for the normal inliner pass. It regressed
AlwaysInliner path which bypasses checkStrictFP.

This PR restores the restriction.

---------

Co-authored-by: Jinsong Ji <[email protected]>
Co-authored-by: Claude Sonnet 5 <[email protected]>
Co-authored-by: Nikita Popov <[email protected]>
https://invent.kde.org/qt/clang/llvm-project/-/commit/a2b42b10a17f93392f13a67a35309e57bc87be0f

Git commit 09ab7a31cff482b5b0ce40a940e01ee068a89117 by GitHub (on behalf of Tom Eccles) on 23/07/2026 at 09:01..
[flang][Lower] Admit opaque terms to real sum reassociation (#211316)

First part of generalisations requested in #207377.

The split-sum eligibility check rejected any RHS containing explicit
parentheses or subtraction, even though the Add flattener already
preserves every non-Add subtree as one opaque term.

Remove those blanket guards and their unused predicates. Document the
opaque-term invariant and extend the test to cover parenthesized
additions and subtractions, whole-RHS parentheses, and subtraction
terms.

I did not observe any benchmark result changes as a result of this
patch.

Assisted-by: Codex
https://invent.kde.org/qt/clang/llvm-project/-/commit/09ab7a31cff482b5b0ce40a940e01ee068a89117

Git commit 58cebee609e42ba70c0b8642ba920d3fef7f9e78 by GitHub (on behalf of TelGome) on 23/07/2026 at 09:15..
[RISCV][P-ext] Support Packed "Q-format" Multiplication (#211236)
https://invent.kde.org/qt/clang/llvm-project/-/commit/58cebee609e42ba70c0b8642ba920d3fef7f9e78

Git commit 83530ce356dd55fdc981b5325ca1f5117b4a58f8 by GitHub (on behalf of Charles Zablit) on 23/07/2026 at 09:16..
[lldb][Windows] ignore loader breakpoints in system modules (#208233)

Currently, when debugging a program with `lldb-dap` on Windows and using
the `integratedTerminal` option, lldb-dap immediatly stops with an
`0x80000003` Exception. This is because `ntdll` executes an `int3`
breakpoint during process initialization when a debugger is attached.

This patch makes `lldb` and `lldb-server` skip the first `int3` after
launch when it originates from a system module (the loader's debugger
notification). Only that first loader breakpoint is skipped. Any later
int3, including `__debugbreak()`, `__builtin_debugtrap()` in the
debuggee's own code, still stops the debugger.

Fixes https://github.com/llvm/llvm-project/issues/198763
https://invent.kde.org/qt/clang/llvm-project/-/commit/83530ce356dd55fdc981b5325ca1f5117b4a58f8

Git commit 797ead5fc79645a09a22d72e13912bf7e7c8d884 by GitHub (on behalf of Mariusz Sikora) on 23/07/2026 at 09:36..
[AMDGPU] Add gfx13 support for v_cvt_pk_{fp8,bf8}_f16 (#211263)

GFX13 supports VOP1 and VOP3 forms of these instructions; GFX1250 keeps
its VOP3-only encoding.

Co-authored-by: Guo Chen <[email protected]>
https://invent.kde.org/qt/clang/llvm-project/-/commit/797ead5fc79645a09a22d72e13912bf7e7c8d884

Git commit 20d8cb474fc78892119e42099ccbad54a648b000 by GitHub (on behalf of Vladislav Dzhidzhoev) on 23/07/2026 at 10:16..
[docs][tools][DirectX] Document DXContainer debug info parts (#204021)

This patch adds ILDB, ILDN, VERS, SRCI parts format documentation. It
describes how to use LLVM tools to inspect them. Additionally,
llvm-pdbutil documentation has been expanded with information on how to
inspect PDB files generated by the DirectX compiler.
https://invent.kde.org/qt/clang/llvm-project/-/commit/20d8cb474fc78892119e42099ccbad54a648b000

Git commit 21f36e37286479a619d7010d938a97f50240e4c7 by GitHub (on behalf of Sean Clarke) on 23/07/2026 at 10:24..
[CostModel][X86] Add PCLMUL cost table (#211113)

The current cost modeling for X86 does not have `ISD::CLMUL` listed in
any cost table, so the cost for `llvm.clmul` falls back to the default
for `Custom`, which is twice the type legalization cost. Since `CLMUL`
uses the `xmm` registers with `pclmulqdq`, this operation is typically
much more expensive. Add a new table for PCLMUL to reflect this.

Note that AVX, AVX512, and VPCLMULQDQ are still not modeled correctly.
This is left as a FIXME for now.

Assisted-by: Claude Opus 4.8
https://invent.kde.org/qt/clang/llvm-project/-/commit/21f36e37286479a619d7010d938a97f50240e4c7

Git commit 280099e581ef165e3d2dafabcb856ec735321c81 by GitHub (on behalf of Florian Hahn) on 23/07/2026 at 10:33..
[VPlan] Widen reverse unit-stride accesses in makeMemOpWideningDesc (NFCI) (#208807)

Update widenConsecutiveMemOps to also widen reversed consecutive
accesses (stride == -1), by creating a VPVectorEndPointerRecipe via
VPBuilder, load and reverse.

PR: https://github.com/llvm/llvm-project/pull/208807
https://invent.kde.org/qt/clang/llvm-project/-/commit/280099e581ef165e3d2dafabcb856ec735321c81

Git commit c66c625d66c5dd6609e41b54bbe19dd85be02528 by GitHub (on behalf of Pengcheng Wang) on 23/07/2026 at 10:34..
[RISCV] Add TuneNoDefaultUnroll to spacemit-x60 (#211502)

This feature has been added to generic CPUs in #135318 and we don't
see any significant regression on `spacemit-x60`.
https://invent.kde.org/qt/clang/llvm-project/-/commit/c66c625d66c5dd6609e41b54bbe19dd85be02528

Git commit 492d726bcc169233d5e2c2077759977851c770ed by GitHub (on behalf of Phoebe Wang) on 23/07/2026 at 10:37..
[X86][AVX10.2] Add missing _mm_cvtts_ss/sd* intrinsics (#211185)

Fixes part of #210947

Assisted-by: Claude Opus 4.8
https://invent.kde.org/qt/clang/llvm-project/-/commit/492d726bcc169233d5e2c2077759977851c770ed

Git commit f953ed4ec657bef2b92b798faa2872b75f1f4e81 by GitHub (on behalf of guillem-bartrina-sonarsource) on 23/07/2026 at 10:40..
[analyzer][z3] Fix crash in Z3 SMTConv when casting atomic int (#211489)

CPP-7675
https://invent.kde.org/qt/clang/llvm-project/-/commit/f953ed4ec657bef2b92b798faa2872b75f1f4e81

Git commit 0037f09f094b0843368df63069b351ec3b952cf8 by GitHub (on behalf of Donát Nagy) on 23/07/2026 at 10:52..
[NFC][analyzer] Eliminate the NodeBuilder in VisitUnaryOperator (#211304)

This is part of the commit series to gradually remove the class
`NodeBuilder` from the analyzer codebase.

This was a relatively complex situation because there were many
different `generateNode` calls and two situations where the the node
from `CheckedSet` was left in `EvalSet` without changes (which is
implicit with the `NodeBuilder` but now requires explicit
`EvalSet.insert(N)` calls).
https://invent.kde.org/qt/clang/llvm-project/-/commit/0037f09f094b0843368df63069b351ec3b952cf8

Git commit cc1dd6bb25d00b3aa5420f43884bea3f4584cabf by GitHub (on behalf of David Green) on 23/07/2026 at 10:58..
[LoopUnroll] Make the list of supported recurrence kinds inclusive. (#210957)

This excludes Sub, AddChainWithSubs, FSub and FAddChainWithSubs
recurrence kinds by making the list of supported types opt-in as opposed
to opt-out. This will hopefully mean that as more recurrence kinds are
added in the future, the list supported by the unrolling remains valid.

Fixes #201065
https://invent.kde.org/qt/clang/llvm-project/-/commit/cc1dd6bb25d00b3aa5420f43884bea3f4584cabf

Git commit a6be8420d07734d80a695fb45b3531c7c7d832b7 by GitHub (on behalf of s1dd) on 23/07/2026 at 11:00..
[InstCombine] Fix typo in assert message for Fp state (#211244)
https://invent.kde.org/qt/clang/llvm-project/-/commit/a6be8420d07734d80a695fb45b3531c7c7d832b7

Git commit 1c825be24005af24237748ae98dae9cd4711bcdc by GitHub (on behalf of hanbeom) on 23/07/2026 at 11:00..
[VectorCombine] Check load offsets in APInt before narrowing shuffle indices (#210904)

vectorizeLoadInsert narrowed the APInt element offset to unsigned before
checking its range. Large offsets could wrap and select the wrong vector
element.

Check the offset range as an APInt before converting it to the shuffle
index. This prevents invalid load widening and is covered by
a regression test.

Fixes #210903
https://invent.kde.org/qt/clang/llvm-project/-/commit/1c825be24005af24237748ae98dae9cd4711bcdc

Git commit d814be1bb794baf0650158a6bd1dda23f4f86e99 by GitHub (on behalf of Yusuke MINATO) on 23/07/2026 at 11:03..
[NFC][clang][Driver] Add tests for --driver-mode=flang (#207658)

This patch intends to clarify the current behaviors, not to state the
expected/desirable behaviors.

---------

Co-authored-by: Tarun Prabhu <[email protected]>
https://invent.kde.org/qt/clang/llvm-project/-/commit/d814be1bb794baf0650158a6bd1dda23f4f86e99