LLVM/project 50241dcllvm/lib/Target/AMDGPU AMDGPURewriteAGPRCopyMFMA.cpp, llvm/test/CodeGen/AMDGPU rewrite-vgpr-mfma-to-agpr-spill-multi-store-mir.mir rewrite-vgpr-mfma-to-agpr-spill-multi-store.ll

[AMDGPU] Reland "Fixed verifier crash because of multiple live range components." (#190719) (#193286)

PR #190719 was reverted by PR #193135 because a .s file was left behind
by a test. This patch fixes it by adding -filetype=null to the RUN line
of the test rewrite-vgpr-mfma-to-agpr-spill-multi-store.ll.

In Rewrite AGPR-Copy-MFMA pass, after replacing spill instructions, the
replacement register may have multiple live range components when the
spill slot was stored to more than once. The verifier crashes with a bad
machine code error. This patch fixes the problem by splitting a live
range but assigning the same physical register in this scenario. A new
test has been added that verifies the absence of this verifier error.

Assisted-by: Claude Opus
DeltaFile
+459-0llvm/test/CodeGen/AMDGPU/rewrite-vgpr-mfma-to-agpr-spill-multi-store-mir.mir
+148-0llvm/test/CodeGen/AMDGPU/rewrite-vgpr-mfma-to-agpr-spill-multi-store.ll
+19-0llvm/lib/Target/AMDGPU/AMDGPURewriteAGPRCopyMFMA.cpp
+626-03 files

LLVM/project dbcd082llvm/lib/Passes PassBuilderPipelines.cpp, llvm/test/Other new-pm-defaults.ll new-pm-thinlto-prelink-samplepgo-defaults.ll

Revert "Revert "[JTS][Passes] Enable JTS By Default" (#192894)"

This reverts commit 832d6dba668d8ea559dd21d52b264d38f97f0ae6.
DeltaFile
+1-5llvm/test/Other/new-pm-defaults.ll
+2-2llvm/lib/Passes/PassBuilderPipelines.cpp
+1-0llvm/test/Other/new-pm-thinlto-prelink-samplepgo-defaults.ll
+1-0llvm/test/Other/new-pm-thinlto-postlink-pgo-defaults.ll
+1-0llvm/test/Other/new-pm-thinlto-postlink-defaults.ll
+1-0llvm/test/Other/new-pm-thinlto-postlink-samplepgo-defaults.ll
+7-72 files not shown
+9-78 files

LLVM/project 6a9ed45llvm/lib/Transforms/Scalar JumpTableToSwitch.cpp, llvm/test/Transforms/JumpTableToSwitch profile-no-guid-metadata.ll

[JTS] Add a temporary workaround for multiple zero GUIDs (#193292)

We can sometimes end up with multiple zero values in the VP metadata if
profile loading cannot map multiple addresses to their functions. We
should probably eventually fix this case in the profile loader, but as a
temporary workaround, handle this in JTS explicitly by dropping the
assertion, similar to how this is handled in ICP.

This is similar to #192905, but with more researched justification.

Co-authored-by: Alexander Shaposhnikov <ashaposhnikov at google.com>
DeltaFile
+34-0llvm/test/Transforms/JumpTableToSwitch/profile-no-guid-metadata.ll
+4-1llvm/lib/Transforms/Scalar/JumpTableToSwitch.cpp
+38-12 files

LLVM/project b0fe500lldb/include/lldb/Target StackFrame.h, lldb/include/lldb/ValueObject DILEval.h

[lldb] Make global lookup in DIL configurable by caller (#192592)

DIL has supported unlimited global lookup, which is in contrast with
long standing `frame variable` behavior, which supports only globals
from the current file.

This is semantically incorrect for frame variable, and has shown to
produce bugs in `dwim-print`. For these reasons, this change makes
global lookup opt-out, controlled by a new
`eExpressionPathOptionsDisallowGlobals` flag.

For `dwim-print`, the bug manifests when a global shadows a computed
property of the instance variable (`self`). As an example, if a global
named `text` exists, and a property named `text` exists (ie
`self.text`), then running `dwim-print text` will unexpectedly print the
global, not `self.text`.

rdar://174834020

Assisted-by: claude
DeltaFile
+3-2lldb/source/Commands/CommandObjectDWIMPrint.cpp
+4-1lldb/source/ValueObject/DILEval.cpp
+2-1lldb/include/lldb/Target/StackFrame.h
+2-1lldb/include/lldb/ValueObject/DILEval.h
+1-1lldb/test/API/commands/dwim-print/TestDWIMPrint.py
+12-65 files

LLVM/project bf8cf4bcompiler-rt/lib/asan asan_errors.cpp asan_report.cpp, compiler-rt/test/asan/TestCases assume_dereferenceable.cpp assume_dereferenceable_pass.cpp

[compiler-rt][asan] Add asan checks for __builtin_assume_dereferencable (#190871)

This checks that the range covered by this intrinsic is dereferencable.
Specifically it checks for `llvm.assume` intrinsics using the
`dereferencable` operator bundle and just asserts the shadow for this
range is zero.

The bulk of this PR was made by gemini but it was thoroughly edited and
reviewed to the best of my ability.
DeltaFile
+87-0compiler-rt/test/asan/TestCases/assume_dereferenceable.cpp
+42-0llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp
+30-8compiler-rt/lib/asan/asan_errors.cpp
+35-0compiler-rt/test/asan/TestCases/assume_dereferenceable_pass.cpp
+22-0compiler-rt/test/asan/TestCases/assume_dereferenceable_halt_on_error.cpp
+19-2compiler-rt/lib/asan/asan_report.cpp
+235-109 files not shown
+313-1815 files

LLVM/project 1e2175eclang/docs ReleaseNotes.rst, clang/include/clang/Basic DiagnosticSemaKinds.td

[Clang] Diagnose UB and emit error when identifier has both internal and external linkage (#192116)

C11 6.2.2p7 says if the same identifier appears with both internal and
external linkage in a translation unit, the behavior is undefined.

This patch adds a check in `Sema::MergeVarDecl` that uses
`LookupResult::isShadowed()` to detect when a block-scope `extern` found
a file-scope `static` through a shadow, and emits an error.

Fixes #54215
DeltaFile
+36-0clang/test/Sema/linkage-internal-extern.c
+19-0clang/lib/Sema/SemaDecl.cpp
+19-0clang/test/Sema/linkage-internal-extern.cpp
+4-6clang/test/C/C2y/n3410.c
+6-0clang/docs/ReleaseNotes.rst
+3-0clang/include/clang/Basic/DiagnosticSemaKinds.td
+87-61 files not shown
+88-77 files

LLVM/project facb9abllvm/test/Transforms/LoopVectorize optimal-epilog-vectorization-liveout.ll, llvm/test/Transforms/LoopVectorize/AArch64 epilog-iv-live-outs.ll

[LV] Remove IV use restrictions for epilogue vectorization. (#190552)

There have been a number of improvements to epilogue handling over the
last year, including improved resume value handling and consistent final
induction value handling via ExitingIVValue.

With those, the restriction can be removed, as direct uses of induction
phis are already handled correctly.

PR: https://github.com/llvm/llvm-project/pull/190552
DeltaFile
+197-85llvm/test/Transforms/LoopVectorize/optimal-epilog-vectorization-liveout.ll
+148-45llvm/test/Transforms/LoopVectorize/AArch64/epilog-iv-live-outs.ll
+56-24llvm/test/Transforms/LoopVectorize/X86/float-induction-x86.ll
+49-23llvm/test/Transforms/LoopVectorize/X86/iv-live-outs.ll
+47-22llvm/test/Transforms/LoopVectorize/X86/transform-narrow-interleave-to-widen-memory.ll
+45-21llvm/test/Transforms/LoopVectorize/X86/transform-narrow-interleave-to-widen-memory-live-outs.ll
+542-2203 files not shown
+569-2829 files

LLVM/project 0a8ed87clang/include/clang/DependencyScanning DependencyScanningFilesystem.h, clang/lib/DependencyScanning DependencyScanningFilesystem.cpp DependencyScannerImpl.cpp

[clang][deps] Simplify scanner VFS (#190843)

This PR removes a hack from the scanning VFS that introduced a bypass
mechanism for module cache queries. Now that we go through the
`ModuleCache` interface for implicitly-built modules, this is no longer
necessary.
DeltaFile
+0-13clang/include/clang/DependencyScanning/DependencyScanningFilesystem.h
+0-13clang/lib/DependencyScanning/DependencyScanningFilesystem.cpp
+0-11clang/unittests/DependencyScanning/DependencyScanningFilesystemTest.cpp
+1-10clang/lib/DependencyScanning/DependencyScannerImpl.cpp
+1-474 files

LLVM/project 6997cc8llvm/docs/CommandGuide llvm-profgen.rst, llvm/tools/llvm-profgen llvm-profgen.cpp

[docs] Add missing command line options to llvm-profgen.rst (#192890)

Resolves #192867

This PR fix some of the undocumented options that are visible in
llvm-profgen --help and --help-hidden
DeltaFile
+39-0llvm/docs/CommandGuide/llvm-profgen.rst
+11-4llvm/tools/llvm-profgen/llvm-profgen.cpp
+50-42 files

LLVM/project 8cc12bcclang/lib/AST/ByteCode Compiler.cpp Compiler.h, clang/test/SemaHLSL/Types/Arithmetic half_size.hlsl literal_suffixes_202x.hlsl

[clang][bytecode][HLSL] Complete the HLSL aggregate splat and elementwise cast implementations, and enable the new constant interpreter on all HLSL tests with static asserts (#189126)

This PR fixes https://github.com/llvm/llvm-project/issues/183426,
completing the implementations of `CK_HLSLAggregateSplatCast` and
`CK_HLSLElementwiseCast` in Clang's new bytecode-based constant
expression evaluation engine / interpreter.

This PR also adds new RUN lines with
`-fexperimental-new-constant-interpreter` to all HLSL tests that have
static assertions.

The tests exercising the `CK_HLSLAggregateSplatCast` and
`CK_HLSLElementwiseCast` implementations in the new constant interpreter
are

- `clang/test/SemaHLSL/Types/AggregateSplatConstantExpr.hlsl` and
- `clang/test/SemaHLSL/Types/ElementWiseConstantExpr.hlsl`

respectively.

    [5 lines not shown]
DeltaFile
+474-121clang/lib/AST/ByteCode/Compiler.cpp
+23-0clang/lib/AST/ByteCode/Compiler.h
+4-0clang/test/SemaHLSL/Types/Arithmetic/half_size.hlsl
+4-0clang/test/SemaHLSL/Types/Arithmetic/literal_suffixes_202x.hlsl
+2-0clang/test/SemaHLSL/Types/Traits/ScalarizedLayoutCompatible.hlsl
+2-0clang/test/SemaHLSL/Types/Traits/IsIntangibleType.hlsl
+509-12115 files not shown
+525-12121 files

LLVM/project dd5632fllvm/utils/gn/secondary/lldb/test BUILD.gn

[gn] port 40fcd2517a110 (#193293)
DeltaFile
+1-0llvm/utils/gn/secondary/lldb/test/BUILD.gn
+1-01 files

LLVM/project d4e2850.github/workflows libcxx-build-and-test.yaml premerge.yaml

Update [Github] Update GHA Dependencies (#176676)

> ℹ️ **Note**
> 
> This PR body was truncated due to platform limits.

This PR contains the following updates:

| Package | Type | Update | Change | Pending |
|---|---|---|---|---|
|
[EnricoMi/publish-unit-test-result-action](https://redirect.github.com/EnricoMi/publish-unit-test-result-action)
| action | minor | `v2.22.0` → `v2.23.0` | |
| [actions/checkout](https://redirect.github.com/actions/checkout) |
action | patch | `v6.0.1` → `v6.0.2` | |
| [actions/setup-node](https://redirect.github.com/actions/setup-node) |
action | minor | `v6.1.0` → `v6.3.0` | `v6.4.0` |
|
[actions/setup-python](https://redirect.github.com/actions/setup-python)

    [1406 lines not shown]
DeltaFile
+10-10.github/workflows/libcxx-build-and-test.yaml
+8-8.github/workflows/premerge.yaml
+7-7.github/workflows/llvm-abi-tests.yml
+6-6.github/workflows/release-binaries.yml
+6-6.github/workflows/release-tasks.yml
+6-6.github/workflows/libclang-abi-tests.yml
+43-4347 files not shown
+144-14453 files

LLVM/project 99457c3clang/lib/CIR/CodeGen CIRGenExprScalar.cpp, clang/test/CIR/CodeGenOpenCL as_type.cl

[CIR] Upstream VectorType __builtin_astype (#192859)

Upstream support for VectorType __builtin_astype

Issue #192311
DeltaFile
+90-2clang/lib/CIR/CodeGen/CIRGenExprScalar.cpp
+54-0clang/test/CIR/CodeGenOpenCL/as_type.cl
+144-22 files

LLVM/project 9c2e677llvm/lib/Target/X86/GISel X86RegisterBankInfo.cpp, llvm/test/CodeGen/X86/GlobalISel regbankselect-dbg-value-physreg-crash.mir

[X86][GlobalISel] Ignore non-vregs in regbank mapping (#182880)

`X86RegisterBankInfo`'s regbank-mapping helpers work under the
assumption that every register operand was a typed virtual register.

This caused `RegBankSelect` crashes when such operands reached these
helpers:
* `getInstrPartialMappingIdxs` called `MRI.getType()` on a non-vreg
operand.
* `getInstrValueMapping` then called `getValueMapping(PMI_None, ...)`
for it.

Skip non-virtual register operands in both helpers. This keeps non-vregs
out of LLT/mapping logic while still mapping real vreg operands.

Fixes https://github.com/llvm/llvm-project/issues/182735
DeltaFile
+34-0llvm/test/CodeGen/X86/GlobalISel/regbankselect-dbg-value-physreg-crash.mir
+2-2llvm/lib/Target/X86/GISel/X86RegisterBankInfo.cpp
+36-22 files

LLVM/project 5299e00llvm/utils/TableGen/Basic RISCVTargetDefEmitter.cpp

[RISCV][TableGen] Use ArrayRef instead of vector&. NFC (#193267)

The vector isn't modified so it should be passed as const vector& or
ArrayRef.
DeltaFile
+2-2llvm/utils/TableGen/Basic/RISCVTargetDefEmitter.cpp
+2-21 files

LLVM/project 0a59b51llvm/lib/Target/RISCV RISCVAsmPrinter.cpp

[RISCV] Add a getTargetStreamer helper to RISCVAsmPrinter. NFC (#193250)

This avoids a repeated static_cast to RISCVTargetStreamer.
DeltaFile
+13-16llvm/lib/Target/RISCV/RISCVAsmPrinter.cpp
+13-161 files

LLVM/project bde6226.github/workflows libclang-python-tests.yml

[Github] Set persist-credentials in libclang-python-tests.yml (#193282)

Not exactly sure how this slipped through the crack in the mass cleanup,
but fix it now.
DeltaFile
+2-0.github/workflows/libclang-python-tests.yml
+2-01 files

LLVM/project 4cdd988clang/include/clang/Basic BuiltinsX86.td, clang/lib/AST ExprConstant.cpp

[Clang] Allow VDBPSADBW intrinsics in constexpr (#188887)

Add constexpr evaluation support for the VDBPSADBW intrinsics (`__builtin_ia32_dbpsadbw128/256/512`) in both Clang constant evaluators.

Fixes #188747
DeltaFile
+73-0clang/lib/AST/ByteCode/InterpBuiltin.cpp
+58-6clang/test/CodeGen/X86/avx512vlbw-builtins.c
+61-0clang/lib/AST/ExprConstant.cpp
+48-3clang/test/CodeGen/X86/avx512bw-builtins.c
+6-3clang/include/clang/Basic/BuiltinsX86.td
+246-125 files

LLVM/project bddd3d3lld/MachO InputFiles.cpp, lld/test/MachO catalyst-arm64e-reexport.s

[lld/mac] For catalyst outputs, tolerate implicitly linking against arm64e mac tbd files (#193065)

Some mac libraries published in the Xcode SDK are now arm64e only, and
catalyst applications should be able to link against these even if
building for arm64.

This matches ld-prime behavior.

https://reviews.llvm.org/D124336 allows linking against MacOS libraries,
but only if the architecture matches exactly.
https://reviews.llvm.org/D130683 allows linking against tbds with ABI
compatibility architectures, but the logic isn't used for this
particular case.

[Assisted-by](https://t.ly/Dkjjk): [Claude Opus
4.6](https://www.anthropic.com/news/claude-opus-4-6)

---------

Co-authored-by: Nuri Amari <nuriamari at fb.com>
DeltaFile
+50-0lld/test/MachO/catalyst-arm64e-reexport.s
+12-9lld/MachO/InputFiles.cpp
+62-92 files

LLVM/project ad31f8ellvm/test/CodeGen/AMDGPU amdgcn.bitcast.1024bit.ll amdgcn.bitcast.512bit.ll, llvm/test/CodeGen/RISCV/rvv vfadd-vp.ll vitofp-sdnode.ll

Merge remote-tracking branch 'origin/main' into users/ziqingluo/PR-172429193-2-split-2

 Conflicts:
        clang/lib/ScalableStaticAnalysisFramework/Analyses/UnsafeBufferUsage/UnsafeBufferUsageExtractor.cpp
DeltaFile
+4,805-4,811llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.1024bit.ll
+1,871-1,882llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.512bit.ll
+764-1,425llvm/test/CodeGen/RISCV/rvv/vfadd-vp.ll
+837-855llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.256bit.ll
+1,139-456llvm/test/CodeGen/AMDGPU/llvm.amdgcn.intersect_ray.ll
+878-428llvm/test/CodeGen/RISCV/rvv/vitofp-sdnode.ll
+10,294-9,857579 files not shown
+37,323-19,191585 files

LLVM/project d794e04clang/include/clang/Basic BuiltinsPPC.def, clang/lib/CodeGen/TargetBuiltins PPC.cpp

[Clang][PowerPC] Add DMF crypto builtins for extended mnemonics (#185961)

This commit adds new Clang builtins for the crypto extended mnemonics: 
void __builtin_dmsha256hash (__dmr1024 *, __dmr1024 *)
void __builtin_dmsha512hash (__dmr1024 *, __dmr1024 *)
void __builtin_dmsha3dw (__dmr2048 *)
void __builtin_dmcryshash (__dmr2048 *)
void __builtin_dmxxsha3512pad (__dmr1024 *, vec_t, uint1)
void __builtin_dmxxsha3384pad (__dmr1024 *, vec_t, uint1)
void __builtin_dmxxsha3256pad (__dmr1024 *, vec_t, uint1)
void __builtin_dmxxsha3224pad (__dmr1024 *, vec_t, uint1)
void __builtin_dmxxshake256pad (__dmr1024 *, vec_t, uint1)
void __builtin_dmxxshake128pad (__dmr1024 *, vec_t, uint1)
void __builtin_dmxxsha384512pad (__dmr1024 *, vec_t)
void __builtin_dmxxsha224256pad (__dmr1024 *, vec_t)
DeltaFile
+536-0llvm/test/CodeGen/PowerPC/mmaplus-crypto.ll
+336-66clang/test/CodeGen/PowerPC/builtins-ppc-dmf.c
+69-5clang/lib/CodeGen/TargetBuiltins/PPC.cpp
+30-0clang/test/Sema/builtins-ppc-crypto.c
+24-0clang/include/clang/Basic/BuiltinsPPC.def
+24-0clang/test/CodeGen/PowerPC/ppc-dmf-mma-builtin-err.c
+1,019-716 files

LLVM/project 06b85c8clang/include/clang/ScalableStaticAnalysisFramework SSAFBuiltinForceLinker.h, clang/include/clang/ScalableStaticAnalysisFramework/Analyses/EntityPointerLevel EntityPointerLevel.h

[SSAF][UnsafeBufferUsage] Make UnsafeBufferUsageExtractor a registered ASTConsumer (#191931)

- Removed UnsafeBufferUsageExtractor.h
- Registered UnsafeBufferUsageExtractor
- Changed unit tests to enable the extractor with
`HandleTranslationUnit`
- Fixed bugs in `UnsafeBufferUsageExtractor::HandleTranslationUnit`

---------

Co-authored-by: Balázs Benics <benicsbalazs at gmail.com>
DeltaFile
+156-132clang/unittests/ScalableStaticAnalysisFramework/Analyses/UnsafeBufferUsage/UnsafeBufferUsageTest.cpp
+47-45clang/lib/ScalableStaticAnalysisFramework/Analyses/UnsafeBufferUsage/UnsafeBufferUsageExtractor.cpp
+0-35clang/include/clang/ScalableStaticAnalysisFramework/Analyses/UnsafeBufferUsage/UnsafeBufferUsageExtractor.h
+9-0clang/include/clang/ScalableStaticAnalysisFramework/SSAFBuiltinForceLinker.h
+3-3clang/include/clang/ScalableStaticAnalysisFramework/Analyses/EntityPointerLevel/EntityPointerLevel.h
+1-1clang/lib/ScalableStaticAnalysisFramework/Analyses/EntityPointerLevel/EntityPointerLevel.cpp
+216-2166 files

LLVM/project 849de61clang/docs ReleaseNotes.rst, clang/include/clang/APINotes Types.h

[APINotes][unsafe-buffer-usage] Add [[clang::unsafe_buffer_usage]] support in APINotes (#189775)

Support the ``[[clang::unsafe_buffer_usage]]`` attribute in APINotes,
e.g.,
```
    Functions:
      - Name: myUnsafeFunction
        UnsafeBufferUsage: true
```

rdar://171859135
DeltaFile
+54-0clang/test/APINotes/unsafe-buffer-usage.cpp
+10-0clang/test/APINotes/Inputs/Headers/UnsafeBufferUsage.apinotes
+9-0clang/docs/ReleaseNotes.rst
+8-0clang/lib/Sema/SemaAPINotes.cpp
+6-1clang/include/clang/APINotes/Types.h
+5-0clang/test/APINotes/Inputs/Headers/module.modulemap
+92-16 files not shown
+106-212 files

LLVM/project 0d45876mlir/include/mlir/Dialect/LLVMIR ROCDLOps.td, mlir/test/Dialect/LLVMIR rocdl.mlir

[ROCDL] Add dot intrinsics to rocdl (#193129)

This patch adds dot intrinsic support to the rocdl dialect. Having these
(inc. follow up `amdgpu` wrapper) as first class citizens in MLIR will
allow us to lower thread local reductions involving `<=16bit` data more
effectively. This is in line with the spirit of `dot` intrinsic support
wrt existing edge dialects (`x86`, `nvvm`, `spirv`).

Assisted by: Claude

---------

Signed-off-by: Eric Feng <Eric.Feng at amd.com>
DeltaFile
+119-0mlir/include/mlir/Dialect/LLVMIR/ROCDLOps.td
+101-0mlir/test/Dialect/LLVMIR/rocdl.mlir
+94-0mlir/test/Target/LLVMIR/rocdl.mlir
+314-03 files

LLVM/project 40fcd25lldb/packages/Python/lldbsuite/test lldbplatformutil.py dotest_args.py, lldb/packages/Python/lldbsuite/test/builders builder.py

[lldb][test] Add support for building Wasm test inferiors (#192872)

This PR adds support for building the test inferiors to WebAssembly.
Specifically, it allows you to configure a sysroot and resource dir
(pointing at the WASI SDK). The Wasm runtime can be configured through
the `LLDB_TEST_USER_ARGS`.

```
LLDB_TEST_TRIPLE:STRING=wasm32-wasip1
LLDB_TEST_SYSROOT:PATH=/path/to/wasi-sdk-32.0-arm64-macos/share/wasi-sysroot
LLDB_TEST_RESOURCE_DIR:PATH=/path/to/wasi-sdk-32.0-arm64-macos/lib/clang/22/
LLDB_TEST_USER_ARGS:STRING=--setting;platform.plugin.wasm.runtime-path=/path/to/iwasm;--setting;platform.plugin.wasm.runtime-args=--heap-size=1048576;--setting;platform.plugin.wasm.port-arg=-g=
```

With the configuration listed above I was able to confirm that I could
build and run a handful of C and C++ tests. To set expectations: lots of
tests are unsupported because they rely on things not available in Wasm
(e.g. shared libraries) or they use features currently unsupported in
LLDB (most notably: expression evaluation).
DeltaFile
+18-1lldb/packages/Python/lldbsuite/test/lldbplatformutil.py
+14-0lldb/packages/Python/lldbsuite/test/make/WASI.rules
+9-1lldb/packages/Python/lldbsuite/test/make/Makefile.rules
+9-1lldb/packages/Python/lldbsuite/test/builders/builder.py
+9-0lldb/packages/Python/lldbsuite/test/dotest_args.py
+3-0lldb/packages/Python/lldbsuite/test/configuration.py
+62-35 files not shown
+70-411 files

LLVM/project 8d21e4ellvm/lib/Target/AMDGPU AMDGPURegBankLegalizeHelper.cpp AMDGPURegBankLegalizeRules.cpp, llvm/test/CodeGen/AMDGPU llvm.amdgcn.intersect_ray.ll llvm.amdgcn.dual_intersect_ray.ll

AMDGPU/GlobalISel: RegBankLegalize rules for BVH intersect ray (#192583)
DeltaFile
+1,139-456llvm/test/CodeGen/AMDGPU/llvm.amdgcn.intersect_ray.ll
+144-2llvm/test/CodeGen/AMDGPU/llvm.amdgcn.dual_intersect_ray.ll
+141-1llvm/test/CodeGen/AMDGPU/llvm.amdgcn.bvh8_intersect_ray.ll
+29-19llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeHelper.cpp
+4-4llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.intersect_ray.ll
+6-0llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.cpp
+1,463-4822 files not shown
+1,465-4838 files

LLVM/project febd3declang/include/clang/Basic Module.h, clang/include/clang/Serialization ModuleCache.h ModuleManager.h

[clang] Get the directory identity from `ModuleCache` instead of `FileManager` (#193070)

Using `FileManager`'s caching and deduplication functionality for
assigning identity to the module cache is handy, but it relies on two
assumptions:
* the rest of the compiler consistently calls
`FileManager::getOptionalDirectoryRef()` with `/*CacheFailure=*/false`
for the module cache path,
* the VFS is not caching failed stats for the module cache path.

This PR implements this functionality in the `ModuleCache` interface,
which is conceptually the right place for it. This PR enables us to land
the VFS simplifications in
https://github.com/llvm/llvm-project/pull/190843.
DeltaFile
+9-18clang/include/clang/Basic/Module.h
+19-2clang/lib/Serialization/ModuleManager.cpp
+0-20clang/lib/Basic/Module.cpp
+20-0clang/include/clang/Serialization/ModuleCache.h
+17-0clang/lib/Serialization/ModuleCache.cpp
+6-0clang/include/clang/Serialization/ModuleManager.h
+71-401 files not shown
+74-417 files

LLVM/project c7eea85llvm/lib/Debuginfod BuildIDFetcher.cpp, llvm/lib/Object BuildID.cpp

Revert "[llvm-cov] Fix error propagation in CoverageMapping::load() (… (#193266)

…#193197)"

This reverts commit b7cfcfe03deb679befe821d0c7a9c302f8645763.

Revert "[llvm] Errorize DebuginfodFetcher for inspection at call-sites
(#191191)"

This reverts commit 337ad44a3e585f62bcf2e30b5766146cb8aacaca.

Reason for revert: Caused debuginfod tests failed in profile runtime:

https://luci-milo.appspot.com/ui/p/fuchsia/builders/toolchain.ci/clang-linux-x64/b8683917826498677969/overview
DeltaFile
+11-13llvm/lib/ProfileData/Coverage/CoverageMapping.cpp
+9-11llvm/lib/Debuginfod/BuildIDFetcher.cpp
+3-9llvm/tools/llvm-objdump/llvm-objdump.cpp
+3-8llvm/lib/ProfileData/InstrProfCorrelator.cpp
+4-6llvm/tools/llvm-debuginfod-find/llvm-debuginfod-find.cpp
+2-6llvm/lib/Object/BuildID.cpp
+32-533 files not shown
+37-619 files

LLVM/project e268fb2clang/include/clang/AST ASTContext.h, clang/lib/AST ASTContext.cpp ItaniumMangle.cpp

[clang] implement CWG2064: ignore value dependence for decltype

The 'decltype' for a value-dependent (but non-type-dependent) should be known,
so this patch makes them non-opaque instead.

This patch also implements what's neceessary to allow overloading
on pure differences in instantiation dependence, making `std::void_t`
usable for SFINAE purposes.

This also readds a few test cases from da98651, which was a previous attempt
at resolving CWG2064.

Fixes #8740
Fixes #61818
Fixes #190388
DeltaFile
+907-175clang/lib/AST/ASTContext.cpp
+312-12clang/test/SemaTemplate/instantiation-dependence.cpp
+143-88clang/lib/AST/ItaniumMangle.cpp
+76-68clang/lib/AST/Type.cpp
+76-48clang/lib/Sema/SemaTemplate.cpp
+95-16clang/include/clang/AST/ASTContext.h
+1,609-40783 files not shown
+2,374-77289 files

LLVM/project 981a9e5llvm/lib/Target/AMDGPU AMDGPURegBankLegalizeRules.cpp, llvm/test/CodeGen/AMDGPU/GlobalISel llvm.amdgcn.ballot.i64.ll llvm.amdgcn.ballot.i32.ll

AMDGPU/GlobalISel: RegBankLegalize rules for amdgcn_ballot (#193105)

Disabled tests involving ctpop and added FIXME to re-enable after ctpop
support.
DeltaFile
+111-119llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i64.ll
+112-117llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i32.ll
+12-12llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-amdgcn.ballot.i64.mir
+4-0llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.cpp
+239-2484 files