LLVM/project 994af34llvm/lib/HTTP HTTPClient.cpp

[lldb][Windows] Try falling back to TLS 1.2 before erroring out (#206108)

TLS 1.3 is only supported on Windows Server 2022 and beyond. Windows
Server 2019 only supports up to TLS 1.2.

This causes test failures on CI runners which run on Windows Server
2019.

This patch allows falling back to TLS 1.2 if 1.3 is not available.
DeltaFile
+9-2llvm/lib/HTTP/HTTPClient.cpp
+9-21 files

LLVM/project c23492aflang/lib/Semantics check-omp-structure.cpp

Add another case that came from main
DeltaFile
+1-1flang/lib/Semantics/check-omp-structure.cpp
+1-11 files

LLVM/project dd16c87libcxx/docs/Status Cxx23Issues.csv

address review comments
DeltaFile
+1-1libcxx/docs/Status/Cxx23Issues.csv
+1-11 files

LLVM/project 55cc77allvm/lib/Target/X86 X86TargetTransformInfo.cpp, llvm/test/Analysis/CostModel/X86 reduce-add.ll reduce-fadd.ll

[CostModel][X86] Add more realistic v8i64/v16i32 + v8f64/v16f32 add reduction costs (#206124)

Fixes failure to fold to v16i32 reduction on ax512 targets

We still need to determine better CostKind values - but that can wait until #194621 is complete
DeltaFile
+12-33llvm/test/Transforms/PhaseOrdering/X86/horizontal-reduce-add.ll
+9-14llvm/test/Transforms/SLPVectorizer/X86/redux-feed-buildvector.ll
+8-13llvm/test/Transforms/VectorCombine/X86/reduction-of-truncations.ll
+20-0llvm/lib/Target/X86/X86TargetTransformInfo.cpp
+8-8llvm/test/Analysis/CostModel/X86/reduce-add.ll
+2-2llvm/test/Analysis/CostModel/X86/reduce-fadd.ll
+59-706 files

LLVM/project debfc46llvm/lib/Target/AMDGPU/Utils AMDGPUPALMetadata.cpp, llvm/test/CodeGen/ARM vector-lrint.ll fpclamptosat_vec.ll

Merge branch 'main' into users/kparzysz/locator-utils
DeltaFile
+1,833-1,841llvm/test/CodeGen/ARM/vector-lrint.ll
+1,197-1,198llvm/test/CodeGen/ARM/fpclamptosat_vec.ll
+921-940llvm/test/CodeGen/ARM/vector-llrint.ll
+528-528llvm/test/CodeGen/ARM/vrint.ll
+927-0mlir/lib/Dialect/XeGPU/Transforms/XeGPUContiguityAnalysis.cpp
+359-363llvm/lib/Target/AMDGPU/Utils/AMDGPUPALMetadata.cpp
+5,765-4,870266 files not shown
+14,215-8,501272 files

LLVM/project 5113f6dflang/lib/Parser openmp-parsers.cpp, flang/lib/Semantics resolve-names.cpp check-omp-structure.cpp

[flang][OpenMP] Properly resolve CRITICAL construct names (#205904)

Resolve the names of CRITICAL constructs even if they are reserved
names.
This also limits locator parsing to known reserved names.

Fixes https://github.com/llvm/llvm-project/issues/205855
DeltaFile
+31-21flang/lib/Semantics/resolve-names.cpp
+21-4flang/lib/Parser/openmp-parsers.cpp
+5-10flang/lib/Semantics/check-omp-structure.cpp
+0-6flang/test/Semantics/OpenMP/reserved-locator.f90
+0-1flang/lib/Semantics/check-omp-structure.h
+57-425 files

LLVM/project 9db072flibcxx/include/__ranges istream_view.h, libcxx/test/libcxx/ranges/range.factories/range.istream.view nodiscard.verify.cpp

[libc++][ranges] Apply `[[nodiscard]]` to `istream_view` (#205154)

Towards #172124

Co-authored-by: Hristo Hristov <zingam at outlook.com>
DeltaFile
+41-0libcxx/test/libcxx/ranges/range.factories/range.istream.view/nodiscard.verify.cpp
+4-4libcxx/include/__ranges/istream_view.h
+45-42 files

LLVM/project 1b7de6dlibcxx/include __config, libcxx/include/__configuration compiler.h

[libc++] Move compiler-specific configuration into <__configuration/compiler.h> (#205590)

These macros are essentially there to query compiler features, so they
should be moved into `<__configuration/compiler.h>`.
DeltaFile
+0-25libcxx/include/__config
+25-0libcxx/include/__configuration/compiler.h
+25-252 files

LLVM/project daa92c7clang/include/clang/DependencyScanning DependencyActionController.h, clang/include/clang/Frontend CompilerInvocation.h

[clang][deps] Avoid `CompilerInvocation` copies (#205632)

When constructing the dependency graph for compilation caching, the
dependency scanner needs to do some extra operations on the compiler
invocations. Historically, these have not utilized the copy-on-write
variant well. This patch takes care to minimize `CompilerInvocation`
copies, which improves incremental scans with populated up-to-date
scanning module cache by 16-18%. Together with
https://github.com/llvm/llvm-project/pull/203350 which operates in the
same space, wall-times are improved by 1.54x and instruction counts by
1.66x.
DeltaFile
+79-1clang/include/clang/Frontend/CompilerInvocation.h
+67-0clang/unittests/Frontend/CompilerInvocationTest.cpp
+23-0clang/lib/Frontend/CompilerInvocation.cpp
+10-2clang/lib/DependencyScanning/DependencyScannerImpl.cpp
+5-1clang/lib/Tooling/DependencyScanningTool.cpp
+1-1clang/include/clang/DependencyScanning/DependencyActionController.h
+185-56 files

LLVM/project 9c7f086clang/lib/StaticAnalyzer/Checkers MoveChecker.cpp, clang/test/Analysis use-after-move-iterator.cpp use-after-move.cpp

[clang][analyzer] Detect use-after-move for 3-arg std::move (#196602)

This implementation detects a use-after-move for the 3-arguments
std::move on containers. This PR fixes #137157.

Since my current implementation uses `IteratorModeling` which is in
alpha stage I mark this PR as draft.

When both the `IteratorModeling` and `MoveChecker` are enabled my
implementation works to detect the use-after-move for the 3 argument
std::move case.

```cpp
std::move(l1.begin(), l1.end(), std::back_inserter(l2));
std::cout << "l1: " << *l1.cbegin() << '\n'; // <--- should have a use-after-move
```

```text
move_iterator.cpp:14:28: warning: Method called on moved-from object 'l1' of

    [14 lines not shown]
DeltaFile
+122-5clang/lib/StaticAnalyzer/Checkers/MoveChecker.cpp
+54-0clang/test/Analysis/use-after-move-iterator.cpp
+24-0clang/test/Analysis/Inputs/system-header-simulator-cxx.h
+13-0clang/test/Analysis/use-after-move.cpp
+213-54 files

LLVM/project ca8808dlibcxx/include __config, libcxx/include/__cxx03 __config

[libc++] Remove _BitScanForward{,64} (#205551)

`std::countr_zero` can be used instead, which is a standard API.
DeltaFile
+0-18libcxx/src/include/ryu/ryu.h
+2-13libcxx/src/ryu/d2s.cpp
+0-6libcxx/include/__config
+2-2libcxx/src/ryu/f2s.cpp
+0-3libcxx/include/__cxx03/__config
+4-425 files

LLVM/project 7510be0lldb/packages/Python/lldbsuite/test/tools/lldb-server gdbremote_testcase.py

[lldb] send 0x0 size packet if LLDB_LAUNCH_FLAG_USE_PIPES is set (#206107)

`LLDB_LAUNCH_FLAG_USE_PIPES=1` is used in tests to run lldb without the
ConPTY on Windows. This reduces the flakyness of tests.

This patch ensures that we read the value of
`LLDB_LAUNCH_FLAG_USE_PIPES` when setting up gdbremote tests, to make
sure they don't use the ConPTY.

This fixes `tools/lldb-server/TestGdbRemote_qThreadStopInfo.py` on
https://ci-external.swift.org/job/lldb-windows/job/main/.
DeltaFile
+11-0lldb/packages/Python/lldbsuite/test/tools/lldb-server/gdbremote_testcase.py
+11-01 files

LLVM/project 01eab1bmlir/lib/Dialect/Linalg/Transforms Transforms.cpp, mlir/test/Dialect/Linalg transform-op-pack.mlir

[mlir][linalg] Guard pack tensor semantics (#206011)

Added a guard so the structured pack transform reports a normal tiling
failure when the target has already been bufferized, instead of reaching
a tensor-only path and asserting.
Fixes #205744
DeltaFile
+20-0mlir/test/Dialect/Linalg/transform-op-pack.mlir
+4-0mlir/lib/Dialect/Linalg/Transforms/Transforms.cpp
+24-02 files

LLVM/project e6be5c9libcxx/include __config fstream

[libc++] Move _LIBCPP_FOPEN_CLOEXEC_MODE to <fstream> (#205537)

The macro is only required inside `<fstream>`, so we can move it there
instead of having it as a general configuration macro.
DeltaFile
+0-11libcxx/include/__config
+11-0libcxx/include/fstream
+11-112 files

LLVM/project 6e0a76amlir/lib/Conversion/VectorToXeGPU VectorToXeGPU.cpp, mlir/test/Conversion/VectorToXeGPU transfer-read-to-xegpu.mlir

[MLIR][XeGPU][VectorToXeGPU] Minor fix for proper handling of 0D memrefs (#195877)

It fixes the following case:
```
   vector.transfer_read %arg0[], %0 : memref<f16>, vector<f16>
```
DeltaFile
+13-0mlir/test/Conversion/VectorToXeGPU/transfer-read-to-xegpu.mlir
+6-1mlir/lib/Conversion/VectorToXeGPU/VectorToXeGPU.cpp
+19-12 files

LLVM/project 62a9a8clibcxx/include/__type_traits aligned_union.h

[libc++][NFC] Simplify the implementation of aligned_union (#185449)

Instead of manually calculating the size and alignment of a union, we
can just generate an actual union and take the size and alignment of
that.

Co-authored-by: Louis Dionne <ldionne.2 at gmail.com>
DeltaFile
+10-13libcxx/include/__type_traits/aligned_union.h
+10-131 files

LLVM/project c4b81fcllvm/test/MC/AMDGPU gfx11_asm_vopc.s, llvm/test/MC/Disassembler/AMDGPU gfx11_dasm_vopc.txt gfx11_dasm_vopc-fake16.txt

[AMDGPU][NFC] Templatise and roundtrip gfx11_asm_vopc.s

Resolves another portion of differences vs the downstream True16
branch.
DeltaFile
+10,260-9,388llvm/test/MC/AMDGPU/gfx11_asm_vopc.s
+0-7,069llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vopc.txt
+5,907-0llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vopc-fake16.txt
+16,167-16,4573 files

LLVM/project 32ccf41flang/lib/Optimizer/OpenACC/Support FIROpenACCTypeInterfaces.cpp, flang/lib/Optimizer/OpenACC/Transforms ACCRecipeBufferization.cpp

[flang] Attach a placeholder `acc.var_name` to allocations in recipes. (#205939)

`ACCRecipeMaterialization` can replace the placeholder with the actual
variable name when materializing the recipe.

Assisted-by: Claude Code
DeltaFile
+39-39flang/test/Lower/OpenACC/acc-reduction.f90
+15-15flang/test/Lower/OpenACC/acc-private.f90
+12-2flang/lib/Optimizer/OpenACC/Support/FIROpenACCTypeInterfaces.cpp
+6-0flang/lib/Optimizer/OpenACC/Transforms/ACCRecipeBufferization.cpp
+72-564 files

LLVM/project 75f0099utils/bazel/llvm-project-overlay/mlir/test BUILD.bazel

[Bazel] Fixes dd5357d (#206122)

This fixes dd5357d38d6b73e3a687bcc5ea8cb3a858fb3fea.

Co-authored-by: Google Bazel Bot <google-bazel-bot at google.com>
DeltaFile
+1-0utils/bazel/llvm-project-overlay/mlir/test/BUILD.bazel
+1-01 files

LLVM/project c1be719libc/shared builtins.h, libc/shared/builtins addtf3.h

[libc] introduce shared compiler-rt builtins (#200094)

Introduce shared compiler-rt builtins to libc and addtf3 builtin

Split from #197950

Part of #197824
DeltaFile
+38-0libc/src/__support/builtins/addtf3.h
+35-0libc/shared/builtins/addtf3.h
+29-0libc/test/shared/shared_builtins_test.cpp
+22-0libc/shared/builtins.h
+10-0libc/test/shared/CMakeLists.txt
+9-0libc/src/__support/builtins/CMakeLists.txt
+143-03 files not shown
+153-09 files

LLVM/project dc32609libunwind/include/mach-o compact_unwind_encoding.h, libunwind/src UnwindCursor.hpp DwarfInstructions.hpp

pr202772
DeltaFile
+73-28libunwind/src/UnwindCursor.hpp
+25-13libunwind/include/mach-o/compact_unwind_encoding.h
+17-5libunwind/src/DwarfInstructions.hpp
+22-0libunwind/src/Registers.hpp
+17-2libunwind/src/CompactUnwinder.hpp
+154-485 files

LLVM/project 0a2a74dclang/lib/Headers ptrauth.h, clang/lib/Sema SemaChecking.cpp

pr202742
DeltaFile
+360-0llvm/test/CodeGen/AArch64/ptrauth-intrinsic-auth-with-pc-and-resign.ll
+101-31llvm/lib/Target/AArch64/AArch64AsmPrinter.cpp
+43-0llvm/lib/Target/AArch64/AArch64ISelDAGToDAG.cpp
+39-2clang/lib/Sema/SemaChecking.cpp
+38-0llvm/lib/Target/AArch64/GISel/AArch64InstructionSelector.cpp
+36-0clang/lib/Headers/ptrauth.h
+617-3310 files not shown
+746-3416 files

LLVM/project dd5357dmlir/lib/Dialect/XeGPU/IR XeGPUOps.cpp, mlir/lib/Dialect/XeGPU/Transforms XeGPUContiguityAnalysis.cpp

[mlir][XeGPU][Transform] Add XeGPU contiguity analysis. (#201684)

Add an AxisInfo-based (borrows the idea from Triton Axis Info analysis)
dataflow analysis that computes, for each
`xegpu.load` / `xegpu.store` gather/scatter, how many elements are
contiguous
along the innermost offsets dimension, and stamps that count as a
`contiguity` **operation attribute** (`OptionalAttr<I64Attr>`) on the
op.

`contiguity` is a target-independent property of the offsets, not a
request tied to any optimization — a consumer is free to use or ignore
it. The
analysis performs no rewrite. Turning the property into a concrete
`lane_layout` / `lane_data` split (which needs the subgroup size) and
the
actual memory-message rewrite are consumer concerns, handled by later
layout-propagation steps (subsequent PRs) or, for testing, by the apply
helper

    [4 lines not shown]
DeltaFile
+927-0mlir/lib/Dialect/XeGPU/Transforms/XeGPUContiguityAnalysis.cpp
+256-0mlir/test/Dialect/XeGPU/contiguity-analysis.mlir
+110-0mlir/test/Dialect/XeGPU/test-xegpu-coalesce-gather-scatter.mlir
+101-0mlir/test/lib/Dialect/XeGPU/TestXeGPUTransforms.cpp
+33-4mlir/lib/Dialect/XeGPU/IR/XeGPUOps.cpp
+17-0mlir/test/Dialect/XeGPU/ops.mlir
+1,444-48 files not shown
+1,493-1714 files

LLVM/project 17627f1mlir/lib/ExecutionEngine LevelZeroRuntimeWrappers.cpp

[mlir][gpu] Fix mgpuLaunchKernel sharedMemBytes type in LevelZero runtime (#206119)

The GPU launch lowering in SelectObjectAttr.cpp declares and calls
`mgpuLaunchKernel` with the dynamic shared memory size argument typed as
`i32`, but the Level Zero runtime wrapper declared the corresponding
parameter as `size_t` (8 bytes on 64-bit targets). Since these are
positional C-ABI arguments, the 4-byte vs 8-byte mismatch shifts the
layout of every following argument (stream, params, extra, paramsCount),
corrupting the call and crashing at launch.

Change the parameter to int32_t to match the codegen, consistent with
the CUDA and ROCm runtime wrappers which already use int32_t smem.

Co-authored-by: Claude Opus 4.8 <noreply at anthropic.com>
DeltaFile
+1-1mlir/lib/ExecutionEngine/LevelZeroRuntimeWrappers.cpp
+1-11 files

LLVM/project 9623f43mlir/lib/Dialect/Utils StaticValueUtils.cpp, mlir/test/Dialect/SCF transform-ops.mlir

[mlir][SCF]: promote one-iteration loops with equal ub and step values (#205826)

Adds a fast-path to `constantTripCount` to return 1 on and enables
promotion of single-iteration loops of the form:

```
scf.for %j = %c0 to %val step %val ... { ... }
```

Signed-off-by: Ege Beysel <beyselege at gmail.com>
DeltaFile
+14-0mlir/test/Dialect/SCF/transform-ops.mlir
+6-0mlir/lib/Dialect/Utils/StaticValueUtils.cpp
+20-02 files

LLVM/project 8f123ccllvm/lib/Target/AMDGPU AMDGPUSubtarget.cpp GCNSubtarget.cpp, llvm/test/CodeGen/AMDGPU occupancy-levels.ll

[AMDGPU] Align to LDS granularity in occupancy calculation (#205637)

Account for LDS allocation granularity by rounding per-workgroup LDS up
to the block size in getOccupancyWithWorkGroupSizes, fixing
overestimated occupancy.
DeltaFile
+4-1llvm/lib/Target/AMDGPU/AMDGPUSubtarget.cpp
+2-2llvm/test/CodeGen/AMDGPU/occupancy-levels.ll
+3-0llvm/lib/Target/AMDGPU/GCNSubtarget.cpp
+1-0llvm/lib/Target/AMDGPU/AMDGPUSubtarget.h
+10-34 files

LLVM/project 4290299clang/lib/CodeGen CGStmtOpenMP.cpp

[clang][OpenMP][NFC] Assert fused distribute loop invariant (#206113)

See
https://github.com/llvm/llvm-project/pull/201670#discussion_r3463060131
DeltaFile
+8-3clang/lib/CodeGen/CGStmtOpenMP.cpp
+8-31 files

LLVM/project ce206e9llvm/test/CodeGen/DirectX strip-debug-info.ll

Test stripping debug info for DirectX
DeltaFile
+92-0llvm/test/CodeGen/DirectX/strip-debug-info.ll
+92-01 files

LLVM/project 894cca4clang-tools-extra/clang-tidy/bugprone BranchCloneCheck.h

[NFC][clang-tidy] Extend doc-comment of BranchCloneCheck (#206116)

Commit 8ac2b77a11c9db9879557ce1c26e38628e1ef45f extended the check
bugprone-branch-clone with a new feature but forgot to mention this in
the doc-comment at the beginning of BranchCloneCheck.h.

Although I don't think that this comment is read too often, let's still
update it to provide accurate information.
DeltaFile
+4-0clang-tools-extra/clang-tidy/bugprone/BranchCloneCheck.h
+4-01 files

LLVM/project d3ea8e2llvm/test/CodeGen/AMDGPU div_v2i128.ll bf16.ll, llvm/test/CodeGen/AMDGPU/GlobalISel udiv.i64.ll urem.i64.ll

Rebase

Created using spr 1.3.7
DeltaFile
+2,592-2,587llvm/test/CodeGen/AMDGPU/div_v2i128.ll
+1,940-1,931llvm/test/CodeGen/AMDGPU/bf16.ll
+1,833-1,841llvm/test/CodeGen/ARM/vector-lrint.ll
+1,410-1,359llvm/test/CodeGen/AMDGPU/GlobalISel/udiv.i64.ll
+1,351-1,351llvm/test/CodeGen/AMDGPU/GlobalISel/urem.i64.ll
+1,197-1,198llvm/test/CodeGen/ARM/fpclamptosat_vec.ll
+10,323-10,2671,684 files not shown
+66,689-51,1961,690 files