fix live interval empty issue #1

huaatian · 2025-02-28T07:57:55Z

[RISCV] Rename function name to start with prefix vpreduce for consistency. (NFC)
[RISCV] The test for vp.reduce.fminimum/fmaximum with fixed-length should stay in fixed-vectors-reduction-fp-vp.ll. (NFC)
[RISCV] Use Priv tablegen class for sf.cease instruction.
PeepholeOpt: Immediately check if a reg_sequence compose supports a subregister (PeepholeOpt: Immediately check if a reg_sequence compose supports a subregister llvm/llvm-project#128279)
[clang-format] Allow breaking before kw___attribute ([clang-format] Allow breaking before kw___attribute llvm/llvm-project#128623)
[LTO][Pipelines][Coro] De-duplicate Coro passes ([LTO][Pipelines][Coro] De-duplicate Coro passes llvm/llvm-project#128654)
[AMDGPU][NewPM] Port AMDGPUInsertDelayAlu to NPM ([AMDGPU][NewPM] Port AMDGPUInsertDelayAlu to NPM llvm/llvm-project#128003)
[RISCV] Merge some of the Sifive decoder tables. ([RISCV] Merge some of the Sifive decoder tables. llvm/llvm-project#128794)
Reland "[AArch64][NPM] Chalk out the CodeGenPassBuilder for NPM (opt -indvars with -verify-scev-strict causes Trip Count Changed! error llvm/llvm-project#128… (Reland "[AArch64][NPM] Chalk out the CodeGenPassBuilder for NPM (#128… llvm/llvm-project#128662)
Revert "Reland "[AArch64][NPM] Chalk out the CodeGenPassBuilder for NPM (opt -indvars with -verify-scev-strict causes Trip Count Changed! error llvm/llvm-project#128…" (Revert "Reland "[AArch64][NPM] Chalk out the CodeGenPassBuilder for NPM (#128…" llvm/llvm-project#128819)
[mlir][Vector] Move vector.extract canonicalizers for DenseElementsAttr to folders ([mlir][Vector] Move vector.extract canonicalizers for DenseElementsAttr to folders llvm/llvm-project#127995)
VirtRegRewriter: Fix verifier errors after regalloc failures (VirtRegRewriter: Fix verifier errors after regalloc failures llvm/llvm-project#128280)
RegAllocFast: Fix verifier errors after assigning to reserved registers (RegAllocFast: Fix verifier errors after assigning to reserved registers llvm/llvm-project#128281)
[CodeGen][NewPM] Port RegAllocGreedy to NPM ([CodeGen][NewPM] Port RegAllocGreedy to NPM llvm/llvm-project#119540)
Support: Do not check if a file exists before executing (Support: Do not check if a file exists before executing llvm/llvm-project#128821)
[clang][bytecode] Fix initing incomplete arrays from ImplicitValueIni… ([clang][bytecode] Fix initing incomplete arrays from ImplicitValueIni… llvm/llvm-project#128729)
AMDGPU: Add baseline tests for bitcast + readlane intrinsics (AMDGPU: Add baseline tests for bitcast + readlane intrinsics llvm/llvm-project#128493)
[Clang] Implement CWG2918 'Consideration of constraints for address of overloaded function' ([Clang] Implement CWG2918 'Consideration of constraints for address of overloaded function' llvm/llvm-project#127773)
[Passes] Fix a warning
[mlir][vector] Move tests for rewriteAlignedSubByteInt{Ext|Trunc} (nfc) ([mlir][vector] Move tests for rewriteAlignedSubByteInt{Ext|Trunc} (nfc) llvm/llvm-project#126416)
[clangd] [C++20] [Modules] Add scanning cache ([clangd] [C++20] [Modules] Add scanning cache llvm/llvm-project#125988)
[LoongArch] Pre-commit tests for vector sext & zext ([LoongArch] Pre-commit tests for vector sext & zext llvm/llvm-project#128835)
Reapply "RegAlloc: Fix verifier error after failed allocation (RegAlloc: Fix verifier error after failed allocation llvm/llvm-project#119690)" (Reapply "RegAlloc: Fix verifier error after failed allocation (#119690)" llvm/llvm-project#128400)
RegAlloc: Use new approach to handling failed allocations (RegAlloc: Use new approach to handling failed allocations llvm/llvm-project#128469)
DAG: Preserve range metadata when load is narrowed (DAG: Preserve range metadata when load is narrowed llvm/llvm-project#128144)
[Clang] Fix an integer overflow issue in computing CTAD's parameter depth ([Clang] Fix an integer overflow issue in computing CTAD's parameter depth llvm/llvm-project#128704)
[DWARFLinker] Avoid repeated hash lookups (NFC) ([DWARFLinker] Avoid repeated hash lookups (NFC) llvm/llvm-project#128825)
[DebugInfo] Avoid repeated map lookups (NFC) ([DebugInfo] Avoid repeated map lookups (NFC) llvm/llvm-project#128826)
[ExecutionEngine] Avoid repeated hash lookups (NFC) ([ExecutionEngine] Avoid repeated hash lookups (NFC) llvm/llvm-project#128827)
[Passes] Avoid repeated hash lookups (NFC) ([Passes] Avoid repeated hash lookups (NFC) llvm/llvm-project#128828)
[ProfileData] Avoid repeated hash lookups (NFC) ([ProfileData] Avoid repeated hash lookups (NFC) llvm/llvm-project#128829)
[MLIR][Bufferization] Remove GEN_PASS_DEF_BUFFERIZATIONBUFFERIZE ([MLIR][Bufferization] Remove GEN_PASS_DEF_BUFFERIZATIONBUFFERIZE llvm/llvm-project#128842)
[bazel] add missing header for RelayoutOptInterface
[lldb] Modernize ABI-based unwind plan creation ([lldb] Modernize ABI-based unwind plan creation llvm/llvm-project#128505)
[InstCombine] Test for trunc to i1 in foldLogOpOfMaskedICmps.
[X86] Handle multiple use freeze(undef) in LowerAVXCONCAT_VECTORS as zero vectors ([X86] Handle multiple use freeze(undef) in LowerAVXCONCAT_VECTORS as zero vectors llvm/llvm-project#128830)
[mlir][tosa] Enhance the conv2d verifier ([mlir][tosa] Enhance the conv2d verifier llvm/llvm-project#128693)
[LLVM] Port a few InstCombine tests to use splat instead of shufflevector.
[LLVM][AArch64] Reduce uses of "undef" in SVE InstCombine tests.
[LLVM][AArch64] Change SVE CodeGen tests to use splat().
[LLVM][AArch64] Reduce uses of "undef" in SVE CodeGen tests.
[libclc] Move __clc_ldexp to CLC library ([libclc] Move __clc_ldexp to CLC library llvm/llvm-project#126078)
[MergeFunc] Add linkonce test with discardable functions.
[CostModel] Handle vector struct results and cost llvm.sincos ([CostModel] Handle vector struct results and cost llvm.sincos llvm/llvm-project#123210)
[libclc] Make CLC library warning-free ([libclc] Make CLC library warning-free llvm/llvm-project#128864)
[AMDGPU] Do not allow M0 as v_readfirstlane_b32 dst ([AMDGPU] Do not allow M0 as v_readfirstlane_b32 dst llvm/llvm-project#128851)
[clang-tidy]improve performance-unnecessary-value-param performance ([clang-tidy]improve performance-unnecessary-value-param performance llvm/llvm-project#128383)
[analyzer] Update the undefined assignment checker diagnostics to not use the term 'garbage' ([analyzer] Update the undefined assignment checker diagnostics to not use the term 'garbage' llvm/llvm-project#126596)
[RISCV] Xqcia 0.4 The spec was recently updated, this changes the name in the TD files associated and increments the Extension number in the clang driver. This is mostly a MC change as there is no other generated code for these instructions yet.
[AMDGPU] Do not allow M0 as v_readlane_b32 dst ([AMDGPU] Do not allow M0 as v_readlane_b32 dst llvm/llvm-project#128867)
PeepholeOpt: Remove pointless check for subregister def (PeepholeOpt: Remove pointless check for subregister def llvm/llvm-project#128850)
[MergeFunc] Add tests showing incorrect handling of metadata call args.
[AArch64] Improve urem by constant costs ([AArch64] Improve urem by constant costs llvm/llvm-project#122236)
[AArch64][SVE] Lower unpredicated loads/stores as LDR/STR. ([AArch64][SVE] Lower unpredicated loads/stores as LDR/STR. llvm/llvm-project#127837)
[VPlan] Introduce explicit broadcasts for live-ins. ([VPlan] Introduce explicit broadcasts for live-ins. llvm/llvm-project#124644)
RegAllocFast: Stop reading uninitalized memory
[HLSL] Allow EmptyDecl in cbuffer/tbuffer ([HLSL] Allow EmptyDecl in cbuffer/tbuffer llvm/llvm-project#128250)
Simplify flip() for std::bitset (Simplify flip() for std::bitset llvm/llvm-project#120807)
RegAllocFast: Fix 8634635 to not trip assertions
Add unsigned integer overloads for abs (Add unsigned integer overloads for abs llvm/llvm-project#128257)
[Clang] Add BuiltinTemplates.td to generate code for builtin templates ([Clang] Add BuiltinTemplates.td to generate code for builtin templates llvm/llvm-project#123736)
[MemCpyOpt] Add stack move test with ret-only capture (NFC)
[X86] Allow select(cond,pshufb,pshufb) -> or(pshufb,pshufb) fold to peek through bitcasts ([X86] Allow select(cond,pshufb,pshufb) -> or(pshufb,pshufb) fold to peek through bitcasts llvm/llvm-project#128876)
[RISCV] Delete dead COPYs to vmv0 during vmv0 elimination
[MLIR][Affine] Make isValidLoopInterchangePermutation efficient ([MLIR][Affine] Make isValidLoopInterchangePermutation efficient llvm/llvm-project#128863)
[bazel] Port 8dd8e5f
[bazel] Export BuiltinTemplates.inc from clang:basic
Thread Safety Analysis: Handle address-of followed by dereference
Thread Safety Analysis: Support warning on passing/returning pointers to guarded variables
[X86] Fix a warning
[lldb] Deindent UnwindAssemblyInstEmulation ([lldb] Deindent UnwindAssemblyInstEmulation llvm/llvm-project#128874)
[AMDGPU][True16][CodeGen] true16 codegen for valu op ([AMDGPU][True16][CodeGen] true16 codegen for valu op llvm/llvm-project#124797)
[clang][bytecode] Handle UsingDirectiveDecls ([clang][bytecode] Handle UsingDirectiveDecls llvm/llvm-project#128888)
[lldb] Build the API unittests with -Wdocumentation ([lldb] Build the API unittests with -Wdocumentation llvm/llvm-project#128893)
[MachineOutliner] Add skipModule call for opt-bisect-limit. ([MachineOutliner] Add skipModule call for opt-bisect-limit. llvm/llvm-project#128836)
[libc++][test] Augment ranges::{fill, fill_n, find} with missing tests ([libc++][test] Augment ranges::{fill, fill_n, find} with missing tests llvm/llvm-project#121209)
Match .exe on Windows (Match .exe on Windows llvm/llvm-project#128894)
[NVPTX] Convert vector function nvvm.annotations to attributes ([NVPTX] Convert vector function nvvm.annotations to attributes llvm/llvm-project#127736)
[libc++][test] Refactor tests for ranges::swap_range algorithms ([libc++][test] Refactor tests for ranges::swap_range algorithms llvm/llvm-project#121138)
[libc++] Updates ostream's println LWG status. ([libc++] Updates ostream's println LWG status. llvm/llvm-project#128214)
[lib++][print] Don't pad the ostream output. ([lib++][print] Don't pad the ostream output. llvm/llvm-project#128354)
[libc++][format] Disables narrow string to wide string formatters. ([libc++][format] Disables narrow string to wide string formatters. llvm/llvm-project#128355)
[AMDGPU][True16][CodeGen] fix test for true16 codegen valu op ([AMDGPU][True16][CodeGen] fix test for true16 codegen valu op llvm/llvm-project#128905)
[RISCV][MC] Add assembler support for XRivosVisni ([RISCV][MC] Add assembler support for XRivosVisni llvm/llvm-project#128773)
[libc++][test] Refactor tests for rotate and rotate_copy ([libc++][test] Refactor tests for rotate and rotate_copy llvm/llvm-project#126458)
[RISCV] Update MicroOpBufferSize in P400 & P600 scheduling models ([RISCV] Update MicroOpBufferSize in P400 & P600 scheduling models llvm/llvm-project#128786)
[MLIR][OPENMP] Relax requirement about branches as terminator of private alloc ([MLIR][OPENMP] Relax requirement about branches as terminator of priv… llvm/llvm-project#128481)
[LLD][ELF] Generically report "address assignment did not converge" ([LLD][ELF] Generically report "address assignment did not converge" llvm/llvm-project#128774)
[libc++] Optimize ranges::equal for vector::iterator ([libc++] Optimize ranges::equal for vector<bool>::iterator llvm/llvm-project#121084)
[flang][Semantics] Ensure deterministic mod file output ([flang][Semantics] Ensure deterministic mod file output llvm/llvm-project#128655)
[RISCV][Docs] RISCV -> RISC-V in RISCVUsage.rst. NFC ([RISCV][Docs] RISCV -> RISC-V in RISCVUsage.rst. NFC llvm/llvm-project#128906)
[DirectX] Support the CBufferLoadLegacy operation ([DirectX] Support the CBufferLoadLegacy operation llvm/llvm-project#128699)
[lldb-dap] Avoid a std::string allocation for the help output (NFC)
[llvm][telemetry]Change Telemetry-disabling mechanism. ([llvm][telemetry]Change Telemetry-disabling mechanism. llvm/llvm-project#128534)
[TableGen] Update comment for size of NumToSkip field in DecoderEmitter. NFC
[Telemetry] Fix a warning
Fix the schedule of vectorizer improvement monthly sync
[Bazel] Port 128541 ([Bazel] Port 128541 llvm/llvm-project#128809)
[mlir][TosaToLinalg] Fix TosaToLinalg to restrict tosa.cast types to integer or float ([mlir][TosaToLinalg] Fix TosaToLinalg to restrict tosa.cast types to integer or float llvm/llvm-project#128859)
[ctxprof] don't inline weak symbols after instrumentation ([ctxprof] don't inline weak symbols after instrumentation llvm/llvm-project#128811)
[gn build] Port 8dd8e5f (BuiltinTemplates.td)
[lldb-dap] Use existing lldb::IOObjectSP for DAP IO (NFC). ([lldb-dap] Use existing lldb::IOObjectSP for DAP IO (NFC). llvm/llvm-project#128750)
[libcxx] Add LWG4135: The helper lambda of std::erase for list should specify return type as bool ([libcxx] Add LWG4135: The helper lambda of std::erase for list should specify return type as bool llvm/llvm-project#128358)
[LV] Generate check lines for if-conversion.ll
[mlir][tosa] Change zero points of convolution ops to required inputs ([mlir][tosa] Change zero points of convolution ops to required inputs llvm/llvm-project#127679)
[mlir][tosa] Rename ReduceProd to ReduceProduct ([mlir][tosa] Rename ReduceProd to ReduceProduct llvm/llvm-project#128751)
[libc][bazel] Add targets for stdbit functions ([libc][bazel] Add targets for stdbit functions llvm/llvm-project#128934)
[LV] Remove stray check lines after be28365.
[AMDGPU][True16][CodeGen] 16bit spill support in true16 mode ([AMDGPU][True16][CodeGen] 16bit spill support in true16 mode llvm/llvm-project#128060)
[clang modules] Setting DebugCompilationDir when it is safe to ignore current working directory ([clang modules] Setting DebugCompilationDir when it is safe to ignore current working directory llvm/llvm-project#128446)
[SLP]Do not use node, if it is a subvector or buildvector node
[flang][cuda] Add more math intrinsic interfaces in cudadevice ([flang][cuda] Add more math intrinsic interfaces in cudadevice llvm/llvm-project#128931)
[MLIR][Vector]: Generalize conversion of vector.insert to LLVM in line with vector.extract ([MLIR][Vector]: Generalize conversion of vector.insert to LLVM in line with vector.extract llvm/llvm-project#128915)
[mlir][AMDGPU] Plumb address space 7 through MLIR, add address_space attr. ([mlir][AMDGPU] Plumb address space 7 through MLIR, add address_space attr. llvm/llvm-project#125594)
[AMDGPU] Handle memcpy()-like ops in LowerBufferFatPointers ([AMDGPU] Handle memcpy()-like ops in LowerBufferFatPointers llvm/llvm-project#126621)
[WebAssemblyLowerEmscriptenEHSjLj] Avoid setting import_name where possible ([WebAssemblyLowerEmscriptenEHSjLj] Avoid setting import_name where possible llvm/llvm-project#128564)
Revert "DAG: Preserve range metadata when load is narrowed" (Revert "DAG: Preserve range metadata when load is narrowed" llvm/llvm-project#128948)
[SLP]Check if the operand for removal is the reduction operand, awaiting for the reduction
[clang][modules] Separate parsing of modulemaps ([clang][modules] Separate parsing of modulemaps llvm/llvm-project#119740)
[MLIR][LLVMIR] Import unregistered intrinsics via llvm.intrinsic_call ([MLIR][LLVMIR] Import unregistered intrinsics via llvm.intrinsic_call llvm/llvm-project#128626)
Revert "[AMDGPU] Handle memcpy()-like ops in LowerBufferFatPointers ([AMDGPU] Handle memcpy()-like ops in LowerBufferFatPointers llvm/llvm-project#126621)"
[RISCV] Adding missing P600 sched model test for RVV segmented loads/stores
[gn build] Port 8fb88f5
[compiler-rt][rtsan] truncate/ftruncate interception. ([compiler-rt][rtsan] truncate/ftruncate interception. llvm/llvm-project#128904)
[memprof] std::move matchings (NFC) ([memprof] std::move matchings (NFC) llvm/llvm-project#128933)
[libc][bazel] Add targets for strfrom ([libc][bazel] Add targets for strfrom<float> llvm/llvm-project#128956)
[libc][hdrgen] Allow to treat hdrgen Python code as a Python module. ([libc][hdrgen] Allow to treat hdrgen Python code as a Python module. llvm/llvm-project#128955)
AMDGPU: Fix si-fix-sgpr-copies asserting on VReg_1 phi (AMDGPU: Fix si-fix-sgpr-copies asserting on VReg_1 phi llvm/llvm-project#128903)
Revert "[MLIR][LLVMIR] Import unregistered intrinsics via llvm.intrinsic_call" (Revert "[MLIR][LLVMIR] Import unregistered intrinsics via llvm.intrinsic_call" llvm/llvm-project#128973)
[lldb-dap] Return an llvm::Error instead of calling exit directly (NFC) ([lldb-dap] Return an llvm::Error instead of calling exit directly (NFC) llvm/llvm-project#128951)
AMDGPU: Do not try to commute instruction with same input register (AMDGPU: Do not try to commute instruction with same input register llvm/llvm-project#127562)
AMDGPU: Fix overly conservative immediate operand check (AMDGPU: Fix overly conservative immediate operand check llvm/llvm-project#127563)
[clang][CodeGen] Additional fixes for [clang][CodeGen] sret args should always point to the alloca AS, so use that llvm/llvm-project#114062 ([clang][CodeGen] Additional fixes for #114062 llvm/llvm-project#128166)
[clang-format] Fix a bug that changes keyword or to an identifier ([clang-format] Fix a bug that changes keyword or to an identifier llvm/llvm-project#128410)
[clang-format] Don't break before *const ([clang-format] Don't break before *const llvm/llvm-project#128817)
[NFC] Fix Sanitizer breakage introduced in [clang][CodeGen] Additional fixes for #114062 llvm/llvm-project#128166 ([NFC] Fix Sanitizer breakage introduced in #128166 llvm/llvm-project#128990)
[ORC] De-duplicate some logic for handling MachO::dylib-based load commands.
[ORC] Support adding LC_LOAD_WEAK_DYLIB commands to MachO JITDylib headers.
[ORC] Sink include into implementation file.
[X86][GlobalISel] Enable Trigonometric functions with libcall mapping ([X86][GlobalISel] Enable Trigonometric functions with libcall mapping llvm/llvm-project#126931)
[lldb] Also show session history in fzf_history ([lldb] Also show session history in fzf_history llvm/llvm-project#128986)
LangRef: Clarify llvm.minnum and llvm.maxnum about sNaN and signed zero (LangRef: Clarify llvm.minnum and llvm.maxnum about sNaN and signed zero llvm/llvm-project#112852)
[mlir][Tosa] Add unreachable case for bad Extension type in TosaProfileCompliance ([mlir][Tosa] Add unreachable case for bad Extension type in TosaProfileCompliance llvm/llvm-project#128889)
[ctxprof] Override type of instrumentation if -profile-context-root is specified ([ctxprof] Override type of instrumentation if -profile-context-root is specified llvm/llvm-project#128940)
[RISCV] Add Xqccmp 0.1 Assembly Support ([RISCV] Add Xqccmp Assembly Support llvm/llvm-project#128731)
[flang][cuda] Handle floats in atomiccas ([flang][cuda] Handle floats in atomiccas llvm/llvm-project#128970)
[CIR] Function type return type improvements ([CIR] Function type return type improvements llvm/llvm-project#128787)
[compiler-rt][sanitizer_common] copy_file_range syscall interception. ([compiler-rt][sanitizer_common] copy_file_range syscall interception. llvm/llvm-project#125816)
[msan] Generalize handlePairwiseShadowOrIntrinsic, and handle x86 pairwise add/sub ([msan] Generalize handlePairwiseShadowOrIntrinsic, and handle x86 pairwise add/sub llvm/llvm-project#127567)
APFloat: Fix maxnum and minnum with sNaN (APFloat: Fix maxnum and minnum with sNaN llvm/llvm-project#112854)
[RISCV] Simplify createRISCVELFStreamer registration
[AMDGPU] Verify SdwaSel value range ([AMDGPU] Verify SdwaSel value range llvm/llvm-project#128898)
[compiler-rt][sanitizer_common] fix copy_file_range test. ([compiler-rt][sanitizer_common] fix copy_file_range test. llvm/llvm-project#129010)
[MLIR][NFC] Retire let constructor for Shape and MLProgram ([MLIR][NFC] Retire let constructor for Shape and MLProgram llvm/llvm-project#128869)
[AArch64] Simplify ELFStreamer and WinCOFFStreamer
[RISCV] Correct RISCVTTIImpl::getIntImmCostInst for Zba ([RISCV] Correct RISCVTTIImpl::getIntImmCostInst for Zba llvm/llvm-project#128174)
[bazel] port 42526d2
[InstCombine] matchOrConcat - return Value* not Instruction* ([InstCombine] matchOrConcat - return Value* not Instruction* llvm/llvm-project#128921)
Reapply [CaptureTracking][FunctionAttrs] Add support for CaptureInfo ([CaptureTracking][FunctionAttrs] Add support for CaptureInfo llvm/llvm-project#125880) (Reapply [CaptureTracking][FunctionAttrs] Add support for CaptureInfo (#125880) llvm/llvm-project#128020)
[DAG] replaceShuffleOfInsert - add support for shuffle_vector(scalar_to_vector(x),y) -> insert_vector_elt(y,x,c) ([DAG] replaceShuffleOfInsert - add support for shuffle_vector(scalar_to_vector(x),y) -> insert_vector_elt(y,x,c) llvm/llvm-project#127210)
[X86] combineINSERT_SUBVECTOR - use getBROADCAST_LOAD helper in insert_subvector(undef, broadcast(p), hi) -> broadcast(p) fold ([X86] combineINSERT_SUBVECTOR - use getBROADCAST_LOAD helper in insert_subvector(undef, broadcast(p), hi) -> broadcast(p) fold llvm/llvm-project#128900)
[lldb] Reimplement LineTable::FindLineEntryByAddress on top of lower_bound ([lldb] Reimplement LineTable::FindLineEntryByAddress on top of lower_bound llvm/llvm-project#127799)
[AArch64] Add codesize test coverage. NFC
[mlir][OpenMP] initialize (first)private variables before task exec ([mlir][OpenMP] initialize (first)private variables before task exec llvm/llvm-project#125304)
[Docs] Fix typo in GetElementPtr.rst (Update GetElementPtr.rst llvm/llvm-project#127393)
[mlir][OpenMP] Pack task private variables into a heap-allocated context struct ([mlir][OpenMP] Pack task private variables into a heap-allocated context struct llvm/llvm-project#125307)
[LLVM][NVPTX] Add codegen support for tcgen05.{ld, st} instructions ([LLVM][NVPTX] Add codegen support for tcgen05.{ld, st} instructions llvm/llvm-project#126740)
[X86] Add custom operation actions for f16: FABS, FNEG, and FCOPYSIGN ([X86] Add custom operation actions for f16: FABS, FNEG, and FCOPYSIGN llvm/llvm-project#128877)
[LV] Teach the loop vectorizer llvm.sincos is trivially vectorizable ([LV] Teach the loop vectorizer llvm.sincos is trivially vectorizable llvm/llvm-project#128035)
[Bitcode] Avoid repeated hash lookups (NFC) ([Bitcode] Avoid repeated hash lookups (NFC) llvm/llvm-project#128824)
[ARM] Avoid repeated hash lookups (NFC) ([ARM] Avoid repeated hash lookups (NFC) llvm/llvm-project#128994)
[AsmPrinter] Avoid repeated hash lookups (NFC) ([AsmPrinter] Avoid repeated hash lookups (NFC) llvm/llvm-project#128995)
[ExecutionEngine] Avoid repeated hash lookups (NFC) ([ExecutionEngine] Avoid repeated hash lookups (NFC) llvm/llvm-project#128997)
[IR] Avoid repeated hash lookups (NFC) ([IR] Avoid repeated hash lookups (NFC) llvm/llvm-project#128998)
[SelectionDAG] Avoid repeated hash lookups (NFC) ([SelectionDAG] Avoid repeated hash lookups (NFC) llvm/llvm-project#128999)
[Support] Avoid repeated hash lookups (NFC) ([Support] Avoid repeated hash lookups (NFC) llvm/llvm-project#129000)
[X86] Merge insertsubvector(load(p0),load_subv(p0),hi) -> subvbroadcast(p0) if either load is oneuse ([X86] Merge insertsubvector(load(p0),load_subv(p0),hi) -> subvbroadcast(p0) if either load is oneuse llvm/llvm-project#128857)
[lldb] Assorted improvements to the Pipe class ([lldb] Assorted improvements to the Pipe class llvm/llvm-project#128719)
[MachineScheduler][AMDGPU] Allow scheduling of single-MI regions ([MachineScheduler][AMDGPU] Allow scheduling of single-MI regions llvm/llvm-project#128739)
[NVPTX] Add Intrinsics for applypriority.* ([NVPTX] Add Intrinsics for applypriority.* llvm/llvm-project#127989)
[NFC] [C++20] [Modules] Add a test for no transitive changes
[LLVM][SVE] Add isel for bfloat based (de)interleave operations. ([LLVM][SVE] Add isel for bfloat based (de)interleave operations. llvm/llvm-project#128875)
[LoopVectorize] Use CodeSize as the cost kind for minsize ([LoopVectorize] Use CodeSize as the cost kind for minsize llvm/llvm-project#124119)
[flang] Extend omp loop semantic checks for reduction ([flang] Extend omp loop semantic checks for reduction llvm/llvm-project#128823)
[Clang][Sema] Add special handling of mfloat8 in initializer lists ([Clang][Sema] Add special handling of mfloat8 in initializer lists llvm/llvm-project#125097)
[clang-tidy] Fix performance-move-const-arg false negative in ternary… ([clang-tidy] Fix performance-move-const-arg false negative in ternary… llvm/llvm-project#128402)
[CLANG]Update svget, svset, svcreate, svundef to have FP8 variants ([CLANG]Update svget, svset, svcreate, svundef to have FP8 variants llvm/llvm-project#126754)
[clang-tidy] Add new check bugprone-unintended-char-ostream-output ([clang-tidy] Add new check bugprone-unintended-char-ostream-output llvm/llvm-project#127720)
[AArch64] Do not split bfloat HFA args between regs and stack ([AArch64] Do not split bfloat HFA args between regs and stack llvm/llvm-project#128909)
[LV] Fix tests after 8150ab9.
[lldb] Re-skip PipeTest on windows for now
[libclc] Move sqrt to CLC library ([libclc] Move sqrt to CLC library llvm/llvm-project#128748)
[gn build] Port 56762b7
[LoopVectorize][NFC] Fix formatting issue with a comment ([LoopVectorize][NFC] Fix formatting issue with a comment llvm/llvm-project#129033)
AMDGPU: Factor agpr reg_sequence folding into a function (AMDGPU: Factor agpr reg_sequence folding into a function llvm/llvm-project#129002)
AMDGPU: Add a mir variant of a regalloc failure test
AMDGPU: Fix a test typo reading a partially undefined vector
AMDGPU: Fold bitcasts into readfirstlane, readlane, and permlane64 (AMDGPU: Fold bitcasts into readfirstlane, readlane, and permlane64 llvm/llvm-project#128494)
[NFC][analyzer] Fix header comment in CreateCheckerManager.cpp ([NFC][analyzer] Fix header comment in CreateCheckerManager.cpp llvm/llvm-project#129055)
[AArch64] Runtime-unroll small multi-exit loops on Apple Silicon. ([AArch64] Runtime-unroll small multi-exit loops on Apple Silicon. llvm/llvm-project#124751)
[SPIR-V] Support 2 more instructions from SPV_INTEL_long_composites ([SPIR-V] Support 2 more instructions from SPV_INTEL_long_composites llvm/llvm-project#128190)
[X86] getFauxShuffleMask - insert_subvector - skip undemanded subvectors ([X86] getFauxShuffleMask - insert_subvector - skip undemanded subvectors llvm/llvm-project#129042)
[MergeFunc] Remove discardables function before writing alias or thunk. ([MergeFunc] Remove discardables function before writing alias or thunk. llvm/llvm-project#128865)
[DirectX] initialize registers properties by calling addRegisterClass and computeRegisterProperties ([DirectX] initialize registers properties by calling addRegisterClass and computeRegisterProperties llvm/llvm-project#128818)
[clang-tidy] Add a release note about unchecked-optional-access smart pointer caching ([clang-tidy] Add a release note about unchecked-optional-access smart pointer caching llvm/llvm-project#122290)
[gn build] Port dc764f5
Add clang atomic control options and attribute (Add clang atomic control options and attribute llvm/llvm-project#114841)
[libclc] Move rsqrt to the CLC library ([libclc] Move rsqrt to the CLC library llvm/llvm-project#129045)
[AMDGPU][True16][MC] true16 for v_alignbit_b32 ([AMDGPU][True16][MC] true16 for v_alignbit_b32 llvm/llvm-project#119409)
[StackProtector] Add test for atomicrmw xchg (NFC)
[StackProtector] Handle atomicrmw xchg in HasAddressTaken heuristic
[clang] Ignore GCC 11 [[malloc(x)]] attribute
[CodeGen][NVPTX] Add a TRI function get the Dwarf register number for a virtual register. ([CodeGen][NVPTX] Add a TRI function get the Dwarf register number for a virtual register. llvm/llvm-project#129017)
[MLIR][OpenMP]Add prescriptiveness-modifier support to granularity clauses of taskloop construct ([MLIR][OpenMP]Add prescriptiveness-modifier support to grainsize and … llvm/llvm-project#128477)
[SLP]Add a test with incorrect bitwidth after minbitwidth analysis, NFC
[MemProf] Fix handling of recursive edges during func assignment ([MemProf] Fix handling of recursive edges during func assignment llvm/llvm-project#129066)
[SLP]Check for potential safety of the truncation for vectorized scalars with multi uses
[NFC][libc++] Guard against operator& hijacking. ([NFC][libc++] Guard against operator& hijacking. llvm/llvm-project#128351)
[Flang] Generate math.erfc op for non-precise erfc interinsic calls ([Flang] Generate math.erfc op for non-precise erfc interinsic calls llvm/llvm-project#128897)
[verify] Improve the error messages with multiple active prefixes ([verify] Improve the error messages with multiple active prefixes llvm/llvm-project#126068)
[AMDGPU][MC] Disassembler warning for v_cmpx instructions ([AMDGPU][MC] Disassembler warning for v_cmpx instructions llvm/llvm-project#127925)
Reapply "[AMDGPU] Handle memcpy()-like ops in LowerBufferFatPointers ([AMDGPU] Handle memcpy()-like ops in LowerBufferFatPointers llvm/llvm-project#126621)" (Reapply "[AMDGPU] Handle memcpy()-like ops in LowerBufferFatPointers (#126621)" llvm/llvm-project#129078)
[NFC][analyzer] Simplify ownership of checker objects ([NFC][analyzer] Simplify ownership of checker objects llvm/llvm-project#128887)
[VPlan] Simplify BLEND %a, %b, NOT(%m) -> BLEND %b, %a, %m. ([VPlan] Simplify BLEND %a, %b, NOT(%m) -> BLEND %b, %a, %m. llvm/llvm-project#128375)
[OpenMP][OMPT][OMPD] Fix frame flags for OpenMP tool APIs ([OpenMP][OMPT][OMPD] Fix frame flags for OpenMP tool APIs llvm/llvm-project#114118)
AMDGPU: Use helper function for use/def chain walk (AMDGPU: Use helper function for use/def chain walk llvm/llvm-project#129052)
[libc][stdfix] Implement fixed point bitsfx functions in llvm libc ([libc][stdfix] Implement fixed point bitsfx functions in llvm libc llvm/llvm-project#128413)
[clang][deps] Propagate the entire service ([clang][deps] Propagate the entire service llvm/llvm-project#128959)
[MLIR][LLVMIR] Add support for atan2 intrinsics op ([MLIR][LLVMIR] Add support for atan2 intrinsics op llvm/llvm-project#127970)
[SandboxVec][Scheduler] Enforce scheduling SchedBundle instrs back-to-back ([SandboxVec][Scheduler] Enforce scheduling SchedBundle instrs back-to-back llvm/llvm-project#128092)
[SandboxVec] Fix unused variables warnings
[lldb-dap] Gardening in lldb-dap.cpp (NFC) ([lldb-dap] Gardening in lldb-dap.cpp (NFC) llvm/llvm-project#128949)
[docs] Fix typo in GettingStarted.rst Unlinke -> Unlike (NFC) ([docs] Fix typo in GettingStarted.rst Unlinke -> Unlike (NFC) llvm/llvm-project#128616)
[VPlan] Update VPBranchOnMaskRecipe to always set the mask (NFC).
[libc] Fix sqrtf128 smoke test for riscv32. ([libc] Fix sqrtf128 smoke test for riscv32. llvm/llvm-project#129094)
[clangd] Reduce superfluous rename conflicts ([clangd] Reduce superfluous rename conflicts llvm/llvm-project#121515)
[LLDB][SBProgress] Fix bad optional in sbprogress ([LLDB][SBProgress] Fix bad optional in sbprogress llvm/llvm-project#128971)
[mlir][spirv] Fix incorrect error message in processCapability ([mlir][spirv] Fix incorrect error message in processCapability llvm/llvm-project#129079)
[MLIR][ROCDL] Add conversion of math.erfc to AMD GPU library calls ([MLIR][ROCDL] Add conversion of math.erfc to AMD GPU library calls llvm/llvm-project#128899)
[clang] Alias cc modifier to c ([clang] Alias cc modifier to c llvm/llvm-project#127719)
[flang] Modifications to ieee_support_standard ([flang] Modifications to ieee_support_standard llvm/llvm-project#128895)
[stdfix] Check fxbits as complete ([stdfix] Check fxbits as complete llvm/llvm-project#129107)
Revert "[compiler-rt][sanitizer_common] copy_file_range syscall interception. ([compiler-rt][sanitizer_common] copy_file_range syscall interception. llvm/llvm-project#125816)" and fix
[VPlan] Preserve DebugLoc for VPBranchOnMaskRecipe.
[libc] implement l64a ([libc] implement l64a llvm/llvm-project#129099)
[mlir][AMDGPU] Add int4 intrinsics, mixed-type fp8 to handle gfx12 ([mlir][AMDGPU] Add int4 intrinsics, mixed-type fp8 to handle gfx12 llvm/llvm-project#128963)
[LLD][COFF] Use primary symbol table machine in Writer::writeHeader (NFC) ([LLD][COFF] Use primary symbol table machine in Writer::writeHeader (NFC) llvm/llvm-project#128442)
[LLD][COFF] Support CF guards on ARM64X ([LLD][COFF] Support CF guards on ARM64X llvm/llvm-project#128440)
Reland copy file range san (Reland copy file range san llvm/llvm-project#129114)
[libc][bazel] Rephrase list comp for downstream ([libc][bazel] Rephrase list comp for downstream llvm/llvm-project#129119)
[RISCV] Order the implicit defs/uses of vl/vtype on MC instructions the same as the pseudo version. ([RISCV] Order the implicit defs/uses of vl/vtype on MC instructions the same as the pseudo version. llvm/llvm-project#129104)
[SandboxVec][BottomUpVec] Add -sbvec-stop-at flag for debugging ([SandboxVec][BottomUpVec] Add -sbvec-stop-at flag for debugging llvm/llvm-project#129097)
[TypeID] Update private typeid definition in DeferredLocInfo ([TypeID] Update private typeid definition in DeferredLocInfo llvm/llvm-project#128968)
[llvm-objcopy] Let --change-section-lma change segments wth filesz=0,… ([llvm-objcopy] Let --change-section-lma change segments wth filesz=0,… llvm/llvm-project#127724)
[stdfix] Update function names ([stdfix] Update function names llvm/llvm-project#129129)
[libc++] Silence CMake's install messages in the CI ([libc++] Silence CMake's install messages in the CI llvm/llvm-project#128872)
[libc++] Diagnose when nullptrs are passed to string APIs ([libc++] Diagnose when nullptrs are passed to string APIs llvm/llvm-project#122790)
[NFC] [clang] [sanitize] add autogen test for array-bounds debuginfo ([NFC] [clang] [sanitize] add autogen test for array-bounds debuginfo llvm/llvm-project#128976)
[SandboxVec] Add option -sbvec-allow-file for bisection debugging ([SandboxVec] Add option -sbvec-allow-file for bisection debugging llvm/llvm-project#129127)
[SystemZ] Handle scalar to vector bitcasts. ([SystemZ] Handle scalar to vector bitcasts. llvm/llvm-project#128628)
[CIR] Upstream basic alloca and load support ([CIR] Upstream basic alloca and load support llvm/llvm-project#128792)
[SandboxIR][Region][NFC] Fix windows build issue ([SandboxIR][Region][NFC] Fix windows build issue llvm/llvm-project#129082)
[flang] Refine handling of NULL() actual to non-optional allocatable … ([flang] Refine handling of NULL() actual to non-optional allocatable … llvm/llvm-project#116126)
[flang] Support COSHAPE() intrinsic function ([flang] Support COSHAPE() intrinsic function llvm/llvm-project#125286)
[flang] Catch more semantic errors with coarrays ([flang] Catch more semantic errors with coarrays llvm/llvm-project#125536)
[flang] Don't flag CLASS(*) ASSOCIATED() pointer or target as error ([flang] Don't flag CLASS(*) ASSOCIATED() pointer or target as error llvm/llvm-project#125890)
[flang] Fix bogus error on defined I/O procedure. ([flang] Fix bogus error on defined I/O procedure. llvm/llvm-project#125898)
[flang] Silence warnings from hermetic module files ([flang] Silence warnings from hermetic module files llvm/llvm-project#128763)
[flang] Account for accessibility in extensibility check ([flang] Account for accessibility in extensibility check llvm/llvm-project#128765)
[flang] Accept proc ptr function result as actual argument without IN… ([flang] Accept proc ptr function result as actual argument without IN… llvm/llvm-project#128771)
[flang] Silence spurious error ([flang] Silence spurious error llvm/llvm-project#128777)
[flang] Refine handling of SELECT TYPE associations in analyses ([flang] Refine handling of SELECT TYPE associations in analyses llvm/llvm-project#128935)
[flang] Enforce C1503 ([flang] Enforce C1503 llvm/llvm-project#128962)
[flang] Catch usage of : and * lengths in array c'tors ([flang] Catch usage of : and * lengths in array c'tors llvm/llvm-project#128974)
[flang] Catch type-bound generic with inherited indistinguishable spe… ([flang] Catch type-bound generic with inherited indistinguishable spe… llvm/llvm-project#128980)
[flang] Fix a warning
[RISCV] Consolidate some DecoderNamespaces for standard extensions. ([RISCV] Consolidate some DecoderNamespaces for standard extensions. llvm/llvm-project#128954)
[RISCV] Reduce dynamic relocations for RISCVOpcodesList table. NFC
[JumpThreading] Remove deleted BB from Unreachable ([JumpThreading] Remove deleted BB from Unreachable llvm/llvm-project#126984)
IR, CodeGen: Add command line flags for dumping instruction addresses and debug locations.
[NVPTX] Combine addressing-mode variants of ld, st, wmma ([NVPTX] Combine addressing-mode variants of ld, st, wmma llvm/llvm-project#129102)
[MCA][RISCV] Mark one of the internal CustomBehavior functions static. NFC
[BOLT][instr] Avoid WX segment ([BOLT][instr] Avoid WX segment llvm/llvm-project#128982)
[flang][runtime] Detect byte order reversal problems ([flang][runtime] Detect byte order reversal problems llvm/llvm-project#129093)
[flang] Catch more defined I/O conflicts ([flang] Catch more defined I/O conflicts llvm/llvm-project#129115)
[WebAssembly] Generate __clang_call_terminate for Emscripten EH ([WebAssembly] Generate __clang_call_terminate for Emscripten EH llvm/llvm-project#129020)
[X86][AVX10.2] Add comments for the avx10_2convertintrin.h file ([X86][AVX10.2] Add comments for the avx10_2convertintrin.h file llvm/llvm-project#120766)
[flang][docs][NFC] Fix Markdown /*comments*/ ([flang][docs][NFC] Fix Markdown /*comments*/ llvm/llvm-project#129018)
[RISCV] Move RISCVVInversePseudosTable from RISCVMCTargetDesc.cpp to RISCVBaseInfo.cpp. NFC
[asan][win] Fix CreateThread leak ([asan][win] Fix CreateThread leak llvm/llvm-project#126738)
[lldb-dap] Adaptor -> Adapter (NFC) ([lldb-dap] Adaptor -> Adapter (NFC) llvm/llvm-project#129110)
[mlir] Add two clone methods about encoding to RankedTensorType. ([mlir] Add two clone methods about encoding to RankedTensorType. llvm/llvm-project#127709)
[llvm][CodeGen] Fix the empty interval issue in Window Scheduler(Assertion `VNI && "No live value at use" llvm/llvm-project#128714)

These are very common when using intrinsics (e.g. ARM NEON). For more context: ClangIR has currently been blocked on such intrinsics emission because of this lacking capability.

…lvm#124624) It is known that for vector whose element fits in i16 will be split and scalarized in SelectionDag's type legalizer (see SIISelLowering::getPreferredVectorAction). LRO attempts to undo the scalarizing of vectors across basic block boundary and shoehorn Values in VGPRs. LRO is beneficial for operations that natively work on illegal vector types to prevent flip-flopping between unpacked and packed. If we know that operations on vector will be split and scalarized, then we don't want to shoehorn them back to packed VGPR. Operations that we know to work natively on illegal vector types usually come in the form of intrinsics (MFMA, DOT8), buffer store, shuffle, phi nodes to name a few.

… prefetches. ld64 doesn't currently support the PAGEOFF relocations on anything but load/stores so we need to bail-out here to fix the build failures on greendragon. rdar://145495288

…unctions. (llvm#129831)

…lvm#129948) Consistently use LLDB_INVALID_LINE_NUMBER & LLDB_INVALID_COLUMN_NUMBER when parsing line and column numbers respectively.

Static analysis flags the final return statement in `ReadExtensionBlock` as unreachable and indeed it is since there is no way to exit the `while(true)` loop besides a *return statement*. So I am converting it into a `llvm_unreachable` to explicitly document this.

…lvm#129737) Currently, we error on non-variable or non-local variable declarations in `for` loops such as `for (struct S {}; 0; ) {}`. However, this is valid in C23, so this patch changes the error to a compatibilty warning and also allows this as an extension in earlier language modes. This also matches GCC’s behaviour.

…L op (llvm#127137) Fixes llvm#99205. - Implements the HLSL intrinsic `AddUint64` used to perform unsigned 64-bit integer addition by using pairs of unsigned 32-bit integers instead of native 64-bit types - The LLVM intrinsic `uadd_with_overflow` is used in the implementation of `AddUint64` in `CGBuiltin.cpp` - The DXIL op `UAddc` was defined in `DXIL.td`, and a lowering of the LLVM intrinsic `uadd_with_overflow` to the `UAddc` DXIL op was implemented in `DXILOpLowering.cpp` Notes: - `__builtin_addc` was not able to be used to implement `AddUint64` in `hlsl_intrinsics.h` because its `CarryOut` argument is a pointer, and pointers are not supported in HLSL - A lowering of the LLVM intrinsic `uadd_with_overflow` to SPIR-V [already exists](https://github.com/llvm/llvm-project/blob/main/llvm/test/CodeGen/SPIRV/llvm-intrinsics/uadd.with.overflow.ll) - When lowering the LLVM intrinsic `uadd_with_overflow` to the `UAddc` DXIL op, the anonymous struct type `{ i32, i1 }` is replaced with a named struct type `%dx.types.i32c`. This aspect of the implementation may be changed when issue llvm#113192 gets addressed - Fixes issues mentioned in the comments on the original PR llvm#125319 --------- Co-authored-by: Finn Plummer <[email protected]> Co-authored-by: Farzon Lotfi <[email protected]> Co-authored-by: Chris B <[email protected]> Co-authored-by: Justin Bogner <[email protected]>

Summary: Somehow these got the `!` dropped and it wasn't tested because the existing test only used the 32-bit variant.

…on` (llvm#129839)

Instead of hardcoding the decision on what mangling scheme to use based on targets, use TargetInfo to make the decision.

…m#129950)

We should not try to overwrite the pointer of struct, also need to add 1 for end of line.

…lvm#128034) This provides a range to decide how to subdivide the vector register budget on gfx90a+. A single value declares the minimum AGPRs that should be allocatable. Eventually this should replace amdgpu-no-agpr. I want this primarily for testing agpr allocation behavior. We should have a heuristic try to detect a reasonable number of AGPRs to keep allocatable.

This performs the minimal replacment of amdgpu-no-agpr to amdgpu-agpr-alloc=0. Most of the test diffs are due to the new attribute sorting later alphabetically. We could do better by trying to perform range merging in the attributor, and trying to pick non-0 values.

According to the commit history, the constructors removed by LWG4140 have never been added to libc++. Existence of non-public or deleted default constructor is observable, this patch tests that there's no such default constructor at all.

Forked from llvm/test/CodeGen/AArch64/arm64-ld1.ll Incorrectly handled by handleUnknownInstruction: - llvm.aarch64.neon.ld1x2, llvm.aarch64.neon.ld1x3, llvm.aarch64.neon.ld1x4 - llvm.aarch64.neon.ld2, llvm.aarch64.neon.ld3, llvm.aarch64.neon.ld4 - llvm.aarch64.neon.ld2lane, llvm.aarch64.neon.ld3lane, llvm.aarch64.neon.ld4lane - llvm.aarch64.neon.ld2r, llvm.aarch64.neon.ld3r, llvm.aarch64.neon.ld4r

Follow up of llvm#129922

…lvm#129969) https://reviews.llvm.org/D156069 has supported it.

…FC) (llvm#129972) https://reviews.llvm.org/D122655 has supported it.

…rCreator. ExecutionSession can provide the Triple, so this argument has been redundant for a while, and no in-tree clients use it.

In order for the union APFloat::Storage to permit access to the semantics field when another union member is stored there, all members of Storage must be standard layout. This is not necessarily the case for DoubleAPFloat which may be non-standard layout because there is no requirement that its std::unique_ptr member is standard layout. Fix this by converting Floats to a raw pointer. Reviewers: arsenm Reviewed By: arsenm Pull Request: llvm#129981

…king..." This reverts commit f905bf3 while I fix some compile errors reported on the buildbots (see e.g. https://lab.llvm.org/buildbot/#/builders/53/builds/13369).

The StringRef overload is often error-prone as users might forget to register the MCSymbol. Add comments to MCTargetExpr and MCSymbolRefExpr::VariantKind. In the distant future the VariantKind parameter might be removed.

) llvm#125983

…th fixes. This re-applies f905bf3, which was reverted in c861c1a due to compiler errors, with a fix for MLIR.

) This extension adds thirty eight bit manipulation instructions. The current spec can be found at: https://github.com/quic/riscv-unified-db/releases/tag/Xqci-0.6 This patch adds assembler only support. Co-authored-by: Sudharsan Veeravalli <[email protected]>

llvm#129980) We used to filter out relocations corresponding to NOP+ADR instruction pairs that were a result of linker "relaxation" optimization. However, these relocations will be useful for reversing the linker optimization. Keep the relocations and ignore them while symbolizing ADR instruction operands.

This patch fixes: llvm/lib/Target/X86/X86ISelLowering.cpp:31886:11: error: unused variable 'M' [-Werror,-Wunused-variable]

Deprecate the `match` and `rewrite` functions. They mainly exist for historic reasons. This PR also updates all remaining uses of in the MLIR codebase. This is addressing a [comment](llvm#129861 (review)) on an earlier PR. Note for LLVM integration: `SplitMatchAndRewrite` will be deleted soon, update your patterns to use `matchAndRewrite` instead of separate `match` / `rewrite`. --------- Co-authored-by: Jakub Kuderski <[email protected]>

Strengthen out-of-bounds guarantees for buffer accesses by disallowing buffer accesses with alignment lower than natural alignment. This is needed to specifically address the edge case where an access starts out-of-bounds and then enters in-bounds, as the hardware would treat the entire access as being out-of-bounds. This is normally not needed for most users, but at least one graphics device extension (VK_EXT_robustness2) has very strict requirements - in-bounds accesses must return correct value, and out-of-bounds accesses must return zero. The direct consequence of the patch is that a buffer access at negative address is not merged by load-store-vectorizer with one at a positive address, which fixes a CTS test. Targets that do not care about the new behavior are advised to use the new target feature relaxed-buffer-oob-mode that maintains the state from before the patch.

…xpr (llvm#129198) Track whether a LambdaExpr is an immediate operand of a CXXOperatorCallExpr using a new flag, isInCXXOperatorCall. This enables special handling of capture initializations to detect uninitialized variable uses, such as in `S s = [&]() { return s; }();`. Fix llvm#128058

Previously, commit 042f07e claimed that P0767R1 was implemented in LLVM 7.0, but no deprecation warning was implemented. This patch adds the missing warnings.

Introduce RISCVLoadStoreOptimizer MIR Pass that will do the optimization. The load/store pairing pass identifies adjacent load/store instructions operating on consecutive memory locations and merges them into a single paired instruction. This is part of MIPS extensions for the p8700 CPU. Production of ldp/sdp instructions is OFF by default, since it is beneficial for -Os only in the case of p8700 CPU.

…29531) This change means that llvm-strip no longer exits immediately upon encountering an error when modifying a file and will instead continue modifying the other inputs. Fixes llvm#129412

…130152) Adjust for llvm#129868.

This reverts commit f3dc358. This causes a large compile-time regression: https://llvm-compile-time-tracker.com/compare.php?from=267403442264959f6b06e227ff450c385f4b3ef2&to=f3dc358953a13caf7521fc615a08f6317930351c&stat=instructions:u

…#128714) The interval of newly generated reg in ModuloScheduleExpander is empty. This will cause crush at some corner case. This patch recalculate the live intervals of these regs.

These are macOS tests only and are currently failing on the x86_64 CI and on arm64 on recent versions of macOS/Xcode. The tests are failing because we're stopping in: ``` Process 17458 stopped * thread #1: tid = 0xbda69a, 0x00000002735bd000 libsystem_malloc.dylib`purgeable_print_self.cold.1, stop reason = EXC_BREAKPOINT (code=1, subcode=0x2735bd000) ``` instead of the libsanitizers library. This seems to be related to `-fsanitize-trivial-abi` support Skip these for now until we figure out the root cause.

bcardosolopes and others added 30 commits March 5, 2025 15:21

[MLIR][LLVMIR] Add elementtype attribute (llvm#129918)

aea7403

These are very common when using intrinsics (e.g. ARM NEON). For more context: ClangIR has currently been blocked on such intrinsics emission because of this lacking capability.

[AArch64][GlobalISel] On Darwin don't fold globals into the offset of…

5422e2c

… prefetches. ld64 doesn't currently support the PAGEOFF relocations on anything but load/stores so we need to bail-out here to fix the build failures on greendragon. rdar://145495288

[libc][math] Add skip accurate pass option for exp*, log*, and powf f…

f2bebdc

…unctions. (llvm#129831)

[lldb-dap] Use LLDB_INVALID_LINE_NUMBER & LLDB_INVALID_COLUMN_NUMBER (l…

b5e70d0

…lvm#129948) Consistently use LLDB_INVALID_LINE_NUMBER & LLDB_INVALID_COLUMN_NUMBER when parsing line and column numbers respectively.

[Clang] Fix incorrect condition on ballot

12c5a46

Summary: Somehow these got the `!` dropped and it wasn't tested because the existing test only used the 32-bit variant.

[libc++] Implement part of P2562R1: constexpr `ranges::stable_partiti…

c28c508

…on` (llvm#129839)

[clang] Use TargetInfo to decide Mangling for C (llvm#129920)

45ca613

Instead of hardcoding the decision on what mangling scheme to use based on targets, use TargetInfo to make the decision.

[flang][cuda] Make sure allocator id is set for pointer allocate (llv…

2130285

…m#129950)

c-index-test: fix buffer overflow (llvm#129922)

560cfd5

We should not try to overwrite the pointer of struct, also need to add 1 for end of line.

[NFC][c-index-test] factor data len out (llvm#129971)

e4c3d25

Follow up of llvm#129922

[RISCV] Remove the TODO for fmaximum/fminimum from the tests. (NFC) (l…

463a096

…lvm#129969) https://reviews.llvm.org/D156069 has supported it.

[RISCV] Remove the TODO for folding bswap and shift from the test. (N…

316f68f

…FC) (llvm#129972) https://reviews.llvm.org/D122655 has supported it.

[ORC-RT] Fix type name in comment. NFC.

a22881c

[ORC] Remove the Triple argument from LLJITBuilder::ObjectLinkingLaye…

f905bf3

…rCreator. ExecutionSession can provide the Triple, so this argument has been redundant for a while, and no in-tree clients use it.

Revert "[ORC] Remove the Triple argument from LLJITBuilder::ObjectLin…

c861c1a

…king..." This reverts commit f905bf3 while I fix some compile errors reported on the buildbots (see e.g. https://lab.llvm.org/buildbot/#/builders/53/builds/13369).

[clang-format][NFC] Use better names for a couple of data members

a6ccda2

[NFC][BOLT] Make file-local cl::opt global variables static (llvm#126472

038fff3

) llvm#125983

Re-apply "[ORC] Remove the Triple argument from LLJITBuilder::..." wi…

b18e5b6

…th fixes. This re-applies f905bf3, which was reverted in c861c1a due to compiler errors, with a fix for MLIR.

kazutakahirata and others added 24 commits March 6, 2025 23:28

[X86] Fix a warning

f83eeac

This patch fixes: llvm/lib/Target/X86/X86ISelLowering.cpp:31886:11: error: unused variable 'M' [-Werror,-Wunused-variable]

[ImplicitNullChecks] Use Register. NFC

59245b4

[MachineTraceMetrics] Use Register::id(). NFC

fc4bce3

[CriticalAntiDepBreaker] Use Register and MCRegister. NFC

6b09402

[CriticalAntiDepBreaker] Attempt to fix MSVC build error. NFC

fc1450c

[AMDGPU] Avoid repeated hash lookups (NFC) (llvm#130235)

6cb2f6d

[libc++] Deprecate is_pod(_v) since C++20 (llvm#129471)

94714fb

Previously, commit 042f07e claimed that P0767R1 was implemented in LLVM 7.0, but no deprecation warning was implemented. This patch adds the missing warnings.

[CriticalAntiDepBreaker] Fix another MSVC build error. NFC

749d68b

[gn build] Port 5048a08

951353d

[llvm-strip] Let llvm-strip continue on encountering an error (llvm#1…

bd5f29c

…29531) This change means that llvm-strip no longer exits immediately upon encountering an error when modifying a file and will instead continue modifying the other inputs. Fixes llvm#129412

[JITListener] Fix build after Module::getTargetTriple() change (llvm#…

d7f409d

…130152) Adjust for llvm#129868.

[Scalar] Avoid repeated hash lookups (NFC) (llvm#129989)

b32cf76

[Analysis] Avoid repeated hash lookups (NFC) (llvm#130236)

17aac7c

[CodeGen] Avoid repeated hash lookups (NFC) (llvm#130237)

616f277

[Transforms] Avoid repeated hash lookups (NFC) (llvm#130238)

bcec6c5

[ExecutionEngine] Avoid repeated hash lookups (NFC) (llvm#130239)

8bf13af

[SPIRV] Avoid repeated hash lookups (NFC) (llvm#130241)

8a855d6

[llvm][CodeGen] Fix the empty interval issue in Window Scheduler(llvm…

40c675f

…#128714) The interval of newly generated reg in ModuloScheduleExpander is empty. This will cause crush at some corner case. This patch recalculate the live intervals of these regs.

[llvm][CodeGen] Modifications made based on review comments 1

a5a28f9

huaatian force-pushed the fix_live_interval_empty_issue branch from 0a088de to a5a28f9 Compare March 7, 2025 11:09

huaatian added 4 commits March 10, 2025 17:27

[llvm][CodeGen] Modifications made based on review comments 2

10934db

[llvm][CodeGen] Modifications made based on review comments 3

f602aa6

[llvm][CodeGen] Modifications made based on review comments 4

c71564e

[llvm][CodeGen] Modifications made based on review comments 5

5e252bc

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fix live interval empty issue #1

fix live interval empty issue #1

huaatian commented Feb 28, 2025

fix live interval empty issue #1

Are you sure you want to change the base?

fix live interval empty issue #1

Conversation

huaatian commented Feb 28, 2025