LLVM/project 7acde7allvm/lib/Analysis TargetTransformInfo.cpp, llvm/lib/Target/RISCV RISCVTargetTransformInfo.cpp RISCVSubtarget.h

[LV][RISCV][NFC] Add explicit LMUL controls via computeFeasibleMaxVF

Add components of maxVF and its support for scalable
vectorization. The default for unspecified RISCV is
LMUL=4 with this change, so some tests will have
the flag that controls max LMUL to extend to LMUL=8
when the request is made.
DeltaFile
+50-0llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+22-0llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+9-0llvm/lib/Target/RISCV/RISCVSubtarget.h
+4-4llvm/test/Transforms/LoopVectorize/RISCV/lmul.ll
+3-3llvm/test/Transforms/LoopVectorize/RISCV/reg-usage-maxbandwidth.ll
+5-0llvm/lib/Analysis/TargetTransformInfo.cpp
+93-75 files not shown
+110-811 files

LLVM/project f4bf729lldb/examples/darwin/heap_find heap.py, lldb/examples/python lldbtk.py diagnose_unwind.py

[lldb] Add exe_ctx to examples commands (#193347)

Updates example python commands to use `SBExecutionContext` where
beneficial. This was to remove uses of
`GetSelected{Target,Thread,Frame}`.

While making this changes, I also renamed `dict` parameters to
`internal_dict`. In examples, it's better not to shadow `dict`, which is
the builtin type and constructor for dictionaries.

Lastly, I removed a command (`section_ptr_refs`) whose function was
disabled, and non-functional (it calls a non-existent function
`load_dylib`).
DeltaFile
+35-120lldb/examples/darwin/heap_find/heap.py
+10-14lldb/examples/python/lldbtk.py
+8-8lldb/examples/python/diagnose_unwind.py
+4-9lldb/examples/python/diagnose_nsstring.py
+5-7lldb/examples/python/jump.py
+5-4lldb/examples/python/types.py
+67-1623 files not shown
+75-1739 files

LLVM/project a680361clang-tools-extra/clang-tidy/readability UseAnyOfAllOfCheck.cpp, clang-tools-extra/docs ReleaseNotes.rst

[clang-tidy] Suggest materializing temporary ranges in readability-use-anyofallof (#185791)

While the check still correctly identifies that the loop can be replaced
with `any_of`/`all_of`, reusing these temporaries directly in the
replacement is often unsafe or leads to compilation errors. So we
suggest a `note: ` for these cases.

Closes https://github.com/llvm/llvm-project/issues/185593
DeltaFile
+59-0clang-tools-extra/test/clang-tidy/checkers/readability/use-anyofallof-cpp20.cpp
+57-0clang-tools-extra/test/clang-tidy/checkers/readability/use-anyofallof.cpp
+45-3clang-tools-extra/docs/clang-tidy/checks/readability/use-anyofallof.rst
+33-13clang-tools-extra/clang-tidy/readability/UseAnyOfAllOfCheck.cpp
+5-0clang-tools-extra/docs/ReleaseNotes.rst
+199-165 files

LLVM/project b5f7bc5utils/bazel/llvm-project-overlay/llvm/unittests BUILD.bazel

[Bazel] Fixes 3081d52 (#193376)

This fixes 3081d52d8242a31620d5ce8111fdd2ad817574b7.

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

LLVM/project 2531a67clang/lib/CodeGen CGDebugInfo.cpp, clang/test/DebugInfo/CXX global-ctor-dtor.cpp atexit-stub.cpp

[clang][DebugInfo] Set linkage name for dynamic initializer/destructor debug info (#189794)

Compiler-generated dynamic initializers and atexit destructors for global
variables were missing DW_AT_linkage_name in their DWARF debug info.
For CodeView targets, different template instantiations of the same
variable all shared the same human-readable name, making them
indistinguishable in the debug info.

Set LinkageName to Fn->getName() if subprogram name does not match
the mangled name.
DeltaFile
+7-7clang/test/DebugInfo/CXX/global-ctor-dtor.cpp
+1-1clang/test/DebugInfo/CXX/atexit-stub.cpp
+2-0clang/lib/CodeGen/CGDebugInfo.cpp
+10-83 files

LLVM/project 653030bllvm/include/llvm/DebugInfo/PDB/Native SymbolCache.h NativeSession.h, llvm/lib/DebugInfo/PDB/Native SymbolCache.cpp NativeSession.cpp

[PDB] Refactor cache strategy for function symbol lookups (#188927)

The original algorithm only caches the symbols that are being queried.
The module needs to be decoded again and again even when looking up the
same symbol but different address. This is time consuming when looking
for a large amount of symbol info. This patch uses IntervalMap to cache
decoded symbols to avoid duplicate decoding. We tested the symbol lookup
time for all symbols in symtab for Blender. The time was shortened by
258x relative to the original algorithm. This will greatly improve the
experience of loading symbols for pseudo probe on Windows.
DeltaFile
+69-27llvm/lib/DebugInfo/PDB/Native/SymbolCache.cpp
+15-12llvm/lib/DebugInfo/PDB/Native/NativeSession.cpp
+11-6llvm/include/llvm/DebugInfo/PDB/Native/SymbolCache.h
+4-0llvm/include/llvm/DebugInfo/PDB/Native/NativeSession.h
+99-454 files

LLVM/project 523c26fclang-tools-extra/clang-tidy/tool check_update_docs.py, clang-tools-extra/test/clang-tidy/infrastructure update-checks-list.test

[clang-tidy][NFC] Add a unittest for checking list.rst (#193134)

Follow up of #192228
DeltaFile
+93-0clang-tools-extra/clang-tidy/tool/check_update_docs.py
+2-0clang-tools-extra/test/clang-tidy/infrastructure/update-checks-list.test
+95-02 files

LLVM/project 3081d52llvm/lib/MC MCDwarf.cpp, llvm/unittests/MC DwarfDebugFrameCIE.cpp CMakeLists.txt

[MC][debug_frame] Fix a bug in MCDwarfFrameEmitter::emit() so that per-function CIE can be generated when CIEs are different (#192727)

When CIEs of the .debug_frame section are different across multiple
functions, CIEs must not be deduplicated. This PR fixes a bug that has
prevented generating per-function CIE for `IsEH=false` when they are
different.

The test case (DwarfDebugFrameCIE.cpp) generation is assisted by LLM
agents.
DeltaFile
+251-0llvm/unittests/MC/DwarfDebugFrameCIE.cpp
+18-8llvm/lib/MC/MCDwarf.cpp
+2-0llvm/unittests/MC/CMakeLists.txt
+271-83 files

LLVM/project cb7f71fmlir/include/mlir/Dialect/AMDGPU/IR AMDGPUOps.td

nits

Signed-off-by: Eric Feng <Eric.Feng at amd.com>
DeltaFile
+1-2mlir/include/mlir/Dialect/AMDGPU/IR/AMDGPUOps.td
+1-21 files

LLVM/project 506c3f1clang-tools-extra/clang-tidy/readability ConvertMemberFunctionsToStaticCheck.cpp, clang-tools-extra/docs ReleaseNotes.rst

[clang-tidy] Fix false positive in readability-convert-member-functions-to-static for const overloads (#191712)

Don't suggest conversion of overloaded + same signature methods
differing in `const`ness to replace the `const` method with `static`.
    
E.g.;
```
void S::f();         // method1
void S::f() const;   // method2
```
    
method2 can't have it's `const` replaced with `static`.
    
Fixes #149152
DeltaFile
+56-0clang-tools-extra/test/clang-tidy/checkers/readability/convert-member-functions-to-static.cpp
+26-1clang-tools-extra/clang-tidy/readability/ConvertMemberFunctionsToStaticCheck.cpp
+5-0clang-tools-extra/docs/ReleaseNotes.rst
+87-13 files

LLVM/project 4f2707amlir/lib/Conversion/AMDGPUToROCDL AMDGPUToROCDL.cpp

nit

Signed-off-by: Eric Feng <Eric.Feng at amd.com>
DeltaFile
+1-2mlir/lib/Conversion/AMDGPUToROCDL/AMDGPUToROCDL.cpp
+1-21 files

LLVM/project b84d8famlir/test/Conversion/AMDGPUToROCDL dot-invalid.mlir

nits

Signed-off-by: Eric Feng <Eric.Feng at amd.com>
DeltaFile
+1-2mlir/test/Conversion/AMDGPUToROCDL/dot-invalid.mlir
+1-21 files

LLVM/project 0886010mlir/test/Conversion/AMDGPUToROCDL dot-gfx9.mlir dot-gfx11.mlir

nits

Signed-off-by: Eric Feng <Eric.Feng at amd.com>
DeltaFile
+0-4mlir/test/Conversion/AMDGPUToROCDL/dot-gfx9.mlir
+0-3mlir/test/Conversion/AMDGPUToROCDL/dot-gfx11.mlir
+0-72 files

LLVM/project 1c62f40mlir/include/mlir/Dialect/AMDGPU/IR AMDGPUOps.td, mlir/lib/Conversion/AMDGPUToROCDL AMDGPUToROCDL.cpp

format

Signed-off-by: Eric Feng <Eric.Feng at amd.com>
DeltaFile
+2-4mlir/lib/Conversion/AMDGPUToROCDL/AMDGPUToROCDL.cpp
+1-1mlir/include/mlir/Dialect/AMDGPU/IR/AMDGPUOps.td
+3-52 files

LLVM/project 2f07948mlir/lib/Conversion/AMDGPUToROCDL AMDGPUToROCDL.cpp, mlir/test/Conversion/AMDGPUToROCDL dot-gfx9.mlir dot-gfx11.mlir

amdgpu dot ops

Signed-off-by: Eric Feng <Eric.Feng at amd.com>
DeltaFile
+166-7mlir/lib/Conversion/AMDGPUToROCDL/AMDGPUToROCDL.cpp
+99-0mlir/test/Dialect/AMDGPU/invalid.mlir
+84-0mlir/unittests/Dialect/AMDGPU/AMDGPUUtilsTest.cpp
+74-0mlir/test/Dialect/AMDGPU/ops.mlir
+65-0mlir/test/Conversion/AMDGPUToROCDL/dot-gfx9.mlir
+62-0mlir/test/Conversion/AMDGPUToROCDL/dot-gfx11.mlir
+550-75 files not shown
+786-911 files

LLVM/project 66119a3lld/MachO ConcatOutputSection.cpp

remove dead code
DeltaFile
+0-6lld/MachO/ConcatOutputSection.cpp
+0-61 files

LLVM/project 3a2c242lld/MachO ConcatOutputSection.cpp Options.td, lld/test/MachO set-slop-scale.s

[lld][macho] Remove --slop_scale flag
DeltaFile
+34-10lld/MachO/ConcatOutputSection.cpp
+4-8lld/MachO/Options.td
+0-11lld/test/MachO/set-slop-scale.s
+0-8lld/MachO/Driver.cpp
+2-0lld/MachO/ConcatOutputSection.h
+0-1lld/MachO/Config.h
+40-386 files

LLVM/project 3e8cdeelld/MachO ConcatOutputSection.cpp

remove page logs
DeltaFile
+1-7lld/MachO/ConcatOutputSection.cpp
+1-71 files

LLVM/project 820654dclang/include/clang/Analysis/Analyses UnsafeBufferUsage.h, clang/lib/Analysis UnsafeBufferUsage.cpp

[UnsafeBufferUsage][SSAF] Change -Wunsafe-buffer-usage API for SSAF-based analysis (#191934)

Change -Wunsafe-buffer-usage API to match `Stmt`s instead of `Decl`s. It
is up to clients of the API to determine how to traversal a `Decl`. In
this change, the client is SSAF-based
UnsafeBufferUsageExtractor.

---------

Co-authored-by: Balázs Benics <benicsbalazs at gmail.com>
DeltaFile
+66-57clang/lib/Analysis/UnsafeBufferUsage.cpp
+2-25clang/lib/ScalableStaticAnalysisFramework/Analyses/UnsafeBufferUsage/UnsafeBufferUsageExtractor.cpp
+4-8clang/include/clang/Analysis/Analyses/UnsafeBufferUsage.h
+72-903 files

LLVM/project f2e4fcdllvm/test/Transforms/LowerTypeTests x86-jumptable-dbg.ll aarch64-jumptable-dbg.ll

[NFC][LowerTypeTests] Add AArch64 and X86 jump table tests with debug info (#192735) (#193358)

It just recommits test copied from non dbg version.

Implementation is https://github.com/llvm/llvm-project/pull/192736.

This is the same as #192735, accidentally merged into spr/users branch.
DeltaFile
+104-0llvm/test/Transforms/LowerTypeTests/x86-jumptable-dbg.ll
+72-0llvm/test/Transforms/LowerTypeTests/aarch64-jumptable-dbg.ll
+176-02 files

LLVM/project b7a4020clang/lib/ScalableStaticAnalysisFramework/Analyses SSAFAnalysesCommon.cpp

Add #include "clang/AST/ExprCXX.h"
DeltaFile
+1-0clang/lib/ScalableStaticAnalysisFramework/Analyses/SSAFAnalysesCommon.cpp
+1-01 files

LLVM/project 1cbd27fllvm/lib/Analysis ConstantFolding.cpp, llvm/test/Transforms/InstSimplify/ConstProp loads.ll

[ConstantFolding] Increase folding limit for vector loads to 128 bytes (#192775)

In FoldReinterpretLoadFromConst, ReadDataFromGlobal bails out when
BytesLoaded exceeds 32 bytes. This prevent folding in our downstream
OpenCL case where global constant is [16 x float] array and is loaded as
float16 vector, which is 64 bytes.

This PR increases BytesLoaded cap to 128 bytes, to accommodate large
vector support, e.g. double16 type in OpenCL. For scalar integer load,
the limit remains 32 bytes to avoid regression on load from string
literal.

---------

Co-authored-by: Nikita Popov <github at npopov.com>
DeltaFile
+62-0llvm/test/Transforms/InstSimplify/ConstProp/loads.ll
+16-6llvm/lib/Analysis/ConstantFolding.cpp
+78-62 files

LLVM/project 3c3f64fclang/include/clang/ScalableStaticAnalysisFramework/Analyses/PointerFlow PointerFlow.h, clang/lib/ScalableStaticAnalysisFramework/Analyses CMakeLists.txt

had Claude to give a review pass
DeltaFile
+87-88clang/lib/ScalableStaticAnalysisFramework/Analyses/PointerFlow/PointerFlowExtractor.cpp
+23-0clang/unittests/ScalableStaticAnalysisFramework/Analyses/PointerFlow/PointerFlowTest.cpp
+2-3clang/lib/ScalableStaticAnalysisFramework/Analyses/PointerFlow/PointerFlow.cpp
+2-2clang/lib/ScalableStaticAnalysisFramework/Analyses/CMakeLists.txt
+2-2clang/include/clang/ScalableStaticAnalysisFramework/Analyses/PointerFlow/PointerFlow.h
+0-3clang/test/Analysis/Scalable/PointerFlow/Inputs/tu-summary.json
+116-981 files not shown
+116-997 files

LLVM/project 4d83691lldb/include/lldb/Core EmulateInstruction.h, lldb/source/Core EmulateInstruction.cpp

[lldb] Fix a couple of return type / return value mismatches (#191464)

* `EmulateInstruction::ReadMemory()` returns a boolean value and is used
in boolean contexts, but the return type is specified as `size_t`.
Change it to `bool`. This also aligns it with `WriteMemory()`.
* `ClangExpressionDeclMap::GetSymbolAddress()` returns `false` if
`Target` is not available, but it is expected to return an address.
Change it to return `LLDB_INVALID_ADDRESS`.
* `ValueObject::GetPointeeData()` returns `true`, whereas a return value
of type `size_t` is expected. Change it to return 0 (this code is
unreachable).
DeltaFile
+2-2lldb/include/lldb/Core/EmulateInstruction.h
+2-2lldb/source/Core/EmulateInstruction.cpp
+2-1lldb/source/Plugins/ExpressionParser/Clang/ClangExpressionDeclMap.h
+1-1lldb/source/Plugins/ExpressionParser/Clang/ClangExpressionDeclMap.cpp
+1-1lldb/source/ValueObject/ValueObject.cpp
+8-75 files

LLVM/project 95aa046clang/lib/AST/ByteCode Compiler.cpp, clang/test/CodeGen/AArch64 neon-misc.c

rebase

Created using spr 1.3.7
DeltaFile
+868-0llvm/test/CodeGen/AArch64/hadd-combine-scalar.ll
+474-121clang/lib/AST/ByteCode/Compiler.cpp
+459-0llvm/test/CodeGen/AMDGPU/rewrite-vgpr-mfma-to-agpr-spill-multi-store-mir.mir
+0-428llvm/test/CodeGen/AArch64/xtn.ll
+355-0clang/test/CodeGen/AArch64/neon/vpaddl.c
+0-308clang/test/CodeGen/AArch64/neon-misc.c
+2,156-857262 files not shown
+6,328-2,081268 files

LLVM/project f99a8f1llvm/include/llvm/ADT Hashing.h

Update llvm/include/llvm/ADT/Hashing.h

Co-authored-by: Matt Arsenault <Matthew.Arsenault at amd.com>
DeltaFile
+1-1llvm/include/llvm/ADT/Hashing.h
+1-11 files

LLVM/project 8e63ff8llvm/include/llvm/ADT Hashing.h

Update llvm/include/llvm/ADT/Hashing.h

Co-authored-by: Matt Arsenault <Matthew.Arsenault at amd.com>
DeltaFile
+1-1llvm/include/llvm/ADT/Hashing.h
+1-11 files

LLVM/project c0584efllvm/include/llvm/ADT Hashing.h

Update llvm/include/llvm/ADT/Hashing.h

Co-authored-by: Matt Arsenault <Matthew.Arsenault at amd.com>
DeltaFile
+1-1llvm/include/llvm/ADT/Hashing.h
+1-11 files

LLVM/project 8082b53llvm/include/llvm/ADT Hashing.h

Update llvm/include/llvm/ADT/Hashing.h

Co-authored-by: Matt Arsenault <Matthew.Arsenault at amd.com>
DeltaFile
+1-1llvm/include/llvm/ADT/Hashing.h
+1-11 files

LLVM/project 7682cf8clang/include/clang/Parse Parser.h, clang/include/clang/Sema DeclSpec.h

Merge branch 'main' into users/ziqingluo/PR-172429193-2-split-4
DeltaFile
+162-141clang/include/clang/Sema/DeclSpec.h
+44-63clang/include/clang/Parse/Parser.h
+81-10llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ds.atomic.barrier.arrive.rtn.b64.ll
+33-20offload/plugins-nextgen/common/src/PluginInterface.cpp
+32-5llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ds.atomic.async.barrier.arrive.b64.ll
+29-8libc/utils/libctest/format.py
+381-24736 files not shown
+658-34242 files