[llvm][PowerPC] Correct handling of spill slots for SPE when EXPENSIVE_CHECKS is enabled #73940
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…E_CHECKS is enabled This was modifying a container as it iterated it, which tripped a check in libstdc++'s debug checks. Instead, just assign to the item via the reference we already have. This fixes the following expensive checks failures on my machine: LLVM :: CodeGen/PowerPC/fp-strict.ll LLVM :: CodeGen/PowerPC/pr55463.ll LLVM :: CodeGen/PowerPC/register-pressure.ll LLVM :: CodeGen/PowerPC/spe.ll Which are some of the tests noted by llvm#68594.
|
@arsenm On https://reviews.llvm.org/D152437 you said: I can address that too if you want. Looks like MCPhysReg is passed around by copy is that right? |
Yes, MCPhysReg is just an int16_t |
fadlyas07
pushed a commit
to greenforce-project/llvm-project
that referenced
this pull request
Dec 2, 2023
* llvm-project/main:
[llvm] Adjust Autoupdater's llvm prefix detection (#74142)
[clang] Avoid recalculating TBAA base type info (#73264)
mailmap: add another entry for myself
[libc++][test] add more benchmarks for `stop_token` (#69590)
[VPlan] Replace IR based truncateToMinimalBitwidths with VPlan version.
[bazel] Remove :CAPIGPU as a dependency of :MLIRPythonBindingsCore. (#74133)
[VPlan] Properly update reduction live-out after placing select.
AMDGPU: Support llvm.exp10 (#65860)
AMDGPU: Add more tests for rootn libcall handling
[X86][CodeGen] Prefer KMOVkk_EVEX than KMOVkk when EGPR is supported (#74048)
AMDGPU: Regenerate test checks
[libc++] Update libcpp-uglify-attributes to diagnose all attributes in C++03
Revert "[Clang] Eagerly instantiate used constexpr function upon definition. (#73463)"
[mlir][tosa] Improve lowering of tosa.conv2d (#74143)
[lld][COFF][NFC] Store pdata range as ChunkRange. (#74024)
[AArch64] Fix -Wunused-variable in AArch64FrameLowering.cpp (NFC)
[clang-fuzzer] Remove GCC 4.x pre GCC 4.9 workaround to silence warning (#73974)
[AArch64] Stack probing for dynamic allocas in SelectionDAG (#66525)
Reland "[LoongArch] Support CTLZ with lsx/lasx"
Revert "[LoongArch] Support CTLZ with lsx/lasx"
[LoongArch] Support CTLZ with lsx/lasx
[llvm] Stop including array (NFC)
[llvm] Stop including optional (NFC)
[llvm-objdump] Stop including llvm/ADT/IndexedMap.h (NFC)
[LoongArch] Override TargetLowering::isShuffleMaskLegal
[MLIR][SCF] Handle more cases in pipelining transform (#74007)
Revert HWASAN failure (#74163)
[NFC][asan] Clang format a few lines
[NFC][ASAN] Declare few functions as static
[gn build] Port c45a66ecd4cb
[clang-format] ObjCPropertyAttributeOrder to sort ObjC property attributes
[mlir][transform] Remove `cachedNames` expensive check (#73961)
[mlir][Linalg] Fix foldFillPackIntoFillOp to work for general cases (#74148)
[mlir][tensor] Fix ReifyResultShapes implementation for tensor.concat (#74157)
[Driver] Remove tests for NetBSD 7. No longer supported.
[OpenMP][NFC] Remove PluginAdaptorManagerTy
[OpenMP][FIX] Fixup test
[OpenMP][NFCI] Organize offload entry logic
[OpenMP][NFC] Extract device image handling into a class/header (#74129)
[cmake] Fix relative paths in prefix map (#74132)
[OpenMP] Separate Requirements into a standalone header (#74126)
[scudo] Fix realloc hooks behavior (#74149)
Revert "[asan][win][msvc] override new and delete and seperate TUs (#68754)"
Reland [X86] With large code model, put functions into .ltext with large section flag (#73037)
[mlir][sparse] fix crash when generate rotated convolution kernels. (#74146)
[RISCV] Remove null terminator from CSRegs in determineCalleeSaves. NFC (#74131)
[CUDA][HIP] allow trivial ctor/dtor in device var init (#73140)
[HIP] fix stack marking for -fgpu-rdc (#72782)
[mlir][linalg] Fix weight dimension ordering in 2D grouped conv (#73855)
[OpenMP] Re-enable KMP_HAVE_QUAD on NetBSD 10.0 with GCC 10.5 (#73478)
[lldb] Make use of LD_LIBRARY_PATH on OpenBSD (#74017)
[lldb] Simplify pipe2(2) handling on FreeBSD (#74019)
[lldb/test] Add OpenBSD to _get_platform_os (#74036)
[PowerPC] Fix the register class for the instruction paddi. (#73995)
[libc++] Correct libcxx default linker script behavior (#74130)
[BOLT][NFC] Remove unused code for CDSplit (#74136)
[mlir][tensor] Add a tensor.concat operation (#72779)
Support soft failure on DWP section overflow, producing a truncated but valid DWP(#71902)
[asan][win][msvc] override new and delete and seperate TUs (#68754)
[nvptx] Fix autoupdater's intrinsic matcher (#73330)
[llvm][NFC] Autoupdater AMD intrinsic detection (#73331)
[libc] Allocate fine-grained memory for the RPC host symbol
Allow libc++ image tag to be specified via enviroment variables.
[DebugInfo][RemoveDIs] Handle a debug-info splicing corner case (#73810)
Avoid nullptr+0 in Regex (#73071)
[SLP]Improve gathering of the scalars used in the graph.
[RISCV][GISel] Support G_ABS with Zbb. (#72939)
[mlir][tensor] Fold padding_value away for pack ops when possible. (#74005)
Revert "[libc] Explicitly pin memory for the client symbol lookup (#73988)"
[AArch64] Regenerate fmin/fmax/memcpy legalization tests. NFC
[LLD][RISCV] Fix incorrect call relaxation when mixing +c and -c objects (#73977)
[RISCV] Collapse fast unaligned access into a single feature [nfc-ish] (#73971)
[OpenMP] Add an INFO message for data transfer of kernel launch env. (#74030)
[OpenMP][NFC] Extract OffloadPolicy into a helper class (#74029)
[AIX][tests] Disable mixed-source.ll test using debug_addr section
[OpenMP][NFC] Modernize the plugin handling (#74034)
Report pass name when -llvm-verify-each reports breakage (#71447)
Fix MSVC signed/unsigned mismatch warning. NFC.
Fix MSVC signed/unsigned mismatch warning. NFC.
[libc] Move the pointer to pin off the stack to the heap (#74118)
[libc] Exclude Fuchsia from float128 detection (#73985)
[RISCV][GISel] Lower G_FCONSTANT to constant pool load without F or D. (#73034)
[BOLT][utils] Bump default time threshold to 2s in nfc-stat-parser
[BOLT][utils] Remove heatmap mode detection from wrapper script
[mlir][sve][nfc] Merge the integration tests for linalg.matmul (#74059)
[Flang][Clang] Add support for frame pointers in Flang
[mlgo] Fix test post PR #73899
[OpenMP][flang] Adding more tests for commonblock with target map (#71146)
[llvm][IR] Remove method IRBuilderBase::getInt8PtrTy (#74096)
[Clang] Implement P2308R1 - Template Parameter Initialization. (#73103)
[amdgpu] Default to 1.0, instead of unspecified, for dynamic hsa (#74098)
Revert "Reland [X86] With large code model, put functions into .ltext with large section flag (#73037)"
TargetInstrInfo: squelch a signedness warning on MSVC (#74078)
[clang][ExtractAPI] Ensure LocationFileChecker doesn't try to traverse VFS when determining file path (#74071)
[AArch64][SME] Remove implicit-def's on smstart (#69012)
[mlir][tosa] Fix lowering of tosa.conv2d (#73240)
[mlir][tosa] Improve lowering to tosa.fully_connected (#73049)
Reapply [InstCombine] Support inverting lshr with non-negative operand
[InstCombine] Add additional test for invert of lshr (NFC)
[mlir][tosa] Improve tosa-infer-shapes for ops consumed by non-TOSA operators (#72715)
Revert "[InstCombine] Support inverting lshr with non-negative operand"
[mlir][vector] Fix unit dim dropping pattern for masked writes (#74038)
[InstCombine] Support inverting lshr with non-negative operand
[InstCombine] Add tests for invert of lshr (NFC)
[mlir][gpu] Generate multiple rank-specializations for tensor map cre… (#74082)
[InstCombiner] Make isFreeToInvert() and friends instance functions (NFC)
[AArch64][SME2] Add SME2 builtins for zero { zt0 } (#72274)
[mlir][Vector] Add fold transpose(shape_cast) -> shape_cast (#73951)
[libc++] Revert "Compile MSAN/TSAN failing test with -O1 (#73555)"
[clang][NFC] Refactor expected directives in C++ DRs 100-199 (#74061)
[llvm][MC][ARM][Assembly] Emit relocations for ADRs and big-endian targets (#73834)
[ValueTracking] Reduce duplication in haveNoCommonBitsSet() (NFC)
[libc][NFC] Rename SPECIAL_X86_LONG_DOUBLE in LIBC_LONG_DOUBLE_IS_X86_FLOAT80 (#73950)
[libc][NFC] Rename LIBC_LONG_DOUBLE_IS_IEEE754_BIN128 to LIBC_LONG_DOUBLE_IS_FLOAT128 (#74052)
[libc][NFC] rename LONG_DOUBLE_IS_DOUBLE into LIBC_LONG_DOUBLE_IS_FLOAT64 (#73948)
[OpenMP] Fix libomptarget build issue (#74067)
[NFC][LLVMContext] Clean up DenseMapInfo classes used for APInt & APFloat.
[Driver] Add exclusive-group feature to multilib.yaml. (#69447)
[NFC] Remove a space in CMake.rst
[mlir] update linalg transform ops docs
TargetInstrInfo: make getOperandLatency return optional (NFC) (#73769)
[InstSimplify] Check common operand with constant earlier
[InstSimplify] Add test for implied cond with equal ops and constant (NFC)
[InstSimplify] Fix or disjoint miscompile with op replacement
[InstSimplify] Add test for disjoint or miscompile (NFC)
[clang][NFC] Fill in historical data on when C++ DRs 100-199 were fixed
[llvm][PowerPC] Correct handling of spill slots for SPE when EXPENSIVE_CHECKS is enabled (#73940)
[mlir][ArmSME] Add option to only enable streaming mode/ZA if required (#73931)
[libc++][NFC] Update the remaining old license headers
[mlir][sve][nfc] Update a test to use transform-interpreter (#73771)
[X86][MC] Support Enc/Dec for EGPR for promoted BMI instructions (#73899)
[AArch64][SME2] Add ldr_zt, str_zt builtins and intrinsics (#72849)
[libc][NFC] Remove named_pair (#73952)
AMDGPU/load-global-i32: regenerate test using UTC (NFC) (#73962)
[mlir][nvvm] Introduce `cp.async.bulk.commit.group`
[mlir] notify insertion of parent op first when cloning (#73806)
Reapply "ValueTracking: Identify implied fp classes by general fcmp (#66505)"
[libc] Fix _Float16 detection for x86 (#73947)
[MLIR][Vector] Refactor tests for contract -> OP transforms (4/N) (#73807)
[libc] Fix lint message (#73956)
PowerPC/aix-cc-abi: regenerate test using UTC (NFC) (#73963)
[ValueTracking] Support srem/urem for isKnownNonNullFromDominatingCondition (#74021)
[X86][MC] Not emit {evex} for VEX-promoted instructions with GPR operands (#74039)
[AMDGPU] NFC. Add test for debug info on CFG annotation instructions. (#73959)
[llvm-reduce] Stop including llvm/ADT/SetVector.h (NFC)
[llvm] Stop including llvm/ADT/StringSet.h (NFC)
[CodeGen][NFC] Sort and format MachinePassRegistry.def (#74044)
clang/OpenCL: set sqrt fp accuracy on call to Z4sqrt (#66651)
[X86][NFC] Extract ReplaceableInstrs to a separate file and clang-format X86InstrInfo.cpp
[clang] Use the materialized temporary's type while creating the APValue (#73355)
[RISCV] Make InitUndef handle undef operand (#65755)
[ADT] Make use of the endian.h header on NetBSD and DragonFly (#74037)
[MemProf] Expand optimization scope to internal linkage function (#73236)
[NFC] Typo fix (#74033)
[RISCV] Minimally modify incoming state in transferBefore (#72352)
[NVPTX] Lower 16xi8 and 8xi8 stores efficiently (#73646)
[LoongArch] Add some binary IR instructions testcases for LASX (#74031)
[RISCV] Move AVL coalescing logic upwards into computeInfoForInstr. NFC (#73909)
Reland "[CodeGen] Port SafeStack to new pass manager (#74027)
[libc] remove fragile test from mincore (#74026)
[clang-format][NFC] Reformat source code with clang-format style
[BOLT] CDSplit main logic part 2/2 (#74032)
[lldb] Remove pre GCC 4.8 workaround (#73981)
[GISel][Mips] Infer alignment when creating memory operand for G_VASTART. (#74004)
[mlir] Fix the link of libcuda.so in MLIRGPUTransforms to not use fully qualified path (#74018)
[RISCV] default enable splitting regalloc between RVV and other (#72950)
[LoongArch] Add more and/or/xor patterns for vector types
[BOLT] CDSplit main logic part 1/2 (#73895)
[NFC][indvars] Remove unused code in WidenIV::widenLoopCompare (#73506)
[libc][mincore] use correct page_size for test (#73984)
[ASAN][AMDGPU] NFC. Improve instrumentation tests. (#73919)
[lldb] pipe2(2) is also supported by OpenBSD (#74012)
[LoongArch] Custom lowering `ISD::BUILD_VECTOR`
[LoongArch] Add codegen support for insertelement
[clang][llvm][fatlto] Avoid cloning modules in FatLTO (#72180)
[OpenMP][NFC] Move mapping related logic into Mapping.h (#74009)
[ThinLTO] Fix assembly dumping of vtable type ids (#73997)
[OpenMP][NFC] Encapsulate Devices.size() (#74010)
[mlir][sparse] refactoring: using util functions to query the index to load from position array for slice-driven loop. (#73986)
Fix ARM Sanitizer build
[OpenMP][NFC] Encapsulate profiling logic (#74003)
[mlir] Expose type and attribute names in the MLIRContext and abstract type/attr classes (#72189)
Reland "[llvm-exegesis] Set stack pointer register after starting perf counter (#72489)"
[OpenMP][NFC] Separate Envar (environment variable) handling (#73994)
Reland [X86] With large code model, put functions into .ltext with large section flag (#73037)
Allow for mixing source/no-source DIFiles in one CU
Revert "[DWARFv5] Verify all-or-nothing constraint on DIFile source"
[test] Fix expected line number in `TestRunner.py` (#73996)
[gn build] Port c73a3f16f81a
Correctly disable hardware watchpoints after a fork event
[lldb] try to fix build on linux after fc6b72523f3d7
Include <vector> in WatchpointResource.h
[lldb] [mostly NFC] Large WP foundation: WatchpointResources (#68845)
[Driver] Remove support for FreeBSD 11.x (#73392)
[Flang] Testcase changes to switch directories before rm (NFC) (#73602)
[BOLT][NFC] Fix comment (#73983)
[flang][openacc] Support early return in acc.loop (#73841)
[mlir][sparse] bug fix on all-dense lex insertion (#73987)
[OpenMP] Replace copy and paste code with instantiation (#73991)
[libc++][test] `ADDITIONAL_COMPILE_FLAGS` should be a space-separated list (#73541)
[Libomptarget] Output the DeviceRTL alongside the other libraries (#73705)
[OpenMP] Start organizing PluginManager, PluginAdaptors (#73875)
[AMDGPU] Rework dot4 signedness checks (#68757)
[MemProf][NFC] Simplify test case (#73979)
[libc] Explicitly pin memory for the client symbol lookup (#73988)
[flang] Detect more misparsed statement functions (same name as funct… (#73852)
[flang] Handle preprocessor macro expansion edge case (#73835)
[flang][openacc] Allow multiple clauses when in preceded by device_type (#73976)
Revert "[OpenMP] Use simple VLA implementation to replace uses of actual VLA"
[flang] Suppress error meant to prevent discontiguous association whe… (#73175)
[flang] Address case of under-processed array symbol (#73169)
[flang] Move internal Fortran::ISO namespace out of user-facing ISO_F… (#72909)
[flang][runtime] Allow already-documented missing 'w' on edit descriptor (#72901)
[clang][DependencyScanner] Include the working directory in the context hash (#73719)
[gn build] Port 248446980317
[libc] fix getchar_unlocked (#73874)
[flang][runtime] Fix BACKSPACE-WRITE on variable-length unformatted file (#72732)
Revert "[BPF] Attribute preserve_static_offset for structs"
[flang] Improve procedure interface compatibility checking for dummy … (#72704)
[SPIR-V] Mark currently failing tests as XFAIL (#73858)
[lldb-dap] Fixing a type encoding issue with dap Stopped events. (#72292)
[clang] Remove no-op ptr-to-ptr bitcasts (NFC)
Revert "[X86] Canonicalize fp zero vectors from bitcasted integer zero vectors"
[flang] Fix IsVariable() to be false for procedure pointers (#72577)
[coro][pgo] Don't promote pgo counters in the suspend basic block (#71263)
[libc] Explicitly pin memory for the HSA memory transfer (#73973)
[flang] Move and extend REDUCE() compile-time checking (#72570)
[Driver] Mark OpenBSD-specific -nopie as TargetSpecific after #72578
[clang-offload-bundler][NFC] Remove blank line in doc (#73968)
[mlir][vector] Add support for vector.maskedstore sub-type emulation. (#73871)
[clang-format] Don't skip stringizing when determining brace kind (#73886)
[flang] Update to: Adjust checks of ICHAR/IACHAR argument length (#73972)
[libc][SysMMan] implement mincore (#73704)
[clang][OpenMP] Fix missing DI for __kmpc_global_thread_num (#73856)
[llvm] Replace calls to Type::getPointerTo (NFC)
[flang] Adjust checks of ICHAR/IACHAR argument length (#72312)
[RISCV][GISel] Use customFor instead of customIf by pulling out a subtarget check. NFC
[flang][openacc] Disable CUDA argument checks in OpenACC regions (#72310)
[X86] Canonicalize fp zero vectors from bitcasted integer zero vectors
[X86] X86FixupVectorConstantsPass - consistently use non-DQI 128/256-bit subvector broadcasts
[gn build] Port cb13e9286b6d
[ThinLTO] Don't mark calloc function dead (#72673)
[RISCV][GISel] Remove unnecessary Observer notifications from legalizeVAStart.
Revert "[CodeGen] Port SafeStack to new pass manager (#73747)" (#73965)
Send an explicit interrupt to cancel an attach waitfor. (#72565)
[BPF] Attribute preserve_static_offset for structs
[AArch64] Stack probing for function prologues (#66524)
[BOLT][DWARF] Add support to create path (#73884)
[mlir][sparse] move all COO related methods into SparseTensorType (#73881)
[clang][NFC] Refactor expected directives in C++ DRs 1-99 (#73879)
[RISCV][GISEL] lowerFormalArguments for variadic arguments (#73064)
[RISCV][GISEL] Legalize G_VASTART using custom legalization (#73063)
[AArch64] Teach areMemAccessesTriviallyDisjoint about scalable widths. (#73655)
[RISCV][GISEL] legalize, regbankselect, and instruction-select G_PTRMASK (#73062)
[RISCV][GISEL] legalize, regbankselect, and instruction-select for G_… (#73061)
[RISCV][GISEL] regbankselect and instruction-select for G_IMPLICIT_DEF (#73060)
[mlir][Vector] Add a rewrite pattern for gather over a strided memref (#72991)
[llvm-reduce] Remove unreachable branch (NFC)
[DebugInfo][RemoveDIs] Handle DPValues at remaining dbg.value using sites (#73788)
[IR] Support or disjoint in hasPoisonGeneratingFlags()
[InstSimplify] Add test for incorrect freeze of or disjoint (NFC)
[libc] Mark operator== const to avoid ambiguity in C++20. (#73954)
[clang][NFC] Fix typo in comment about Mac Catalyst
Add test coverage for -Wunknown-directives; NFC
[InstSimplify] Fix select bit test miscompile with disjoint
[InstSimplify] Add test for or disjoint miscompile (NFC)
[AArch64][MC] Fix run line in Armv9.5-A's FEAT_CPA test
[RISCV] Add combines to form binop from tail insert idioms (#72675)
[AArch64][SME2] Add multi-vector SEL (x2, x4) ACLE builtins & intrinsics (#73188)
[DebugInfo][RemoveDIs] Support maintaining DPValues in CodeGenPrepare (#73660)
[AMDGPU] Add test for GCNRegPressure tracker bug (#73786)
[InstCombine] Fix phi or icmp fold with disjoint flag
[InstCombine] Add test for "or disjoint" miscompile (NFC)
[X86] X86InstrFoldTables.cpp - add Op4 Broadcast Fold/Unfold table entries
[SLP][NFC] Unify code for cost estimation/codegen for buildvector, NFC. (#73182)
[AArch64] Warn when calling a NEON builtin in a streaming function (#73672)
[SLP]Fix/improve minbitwidth mapping to use TreeEntry as a key.
[libc++] picolibc: avoid warning in __locale (#73937)
[X86][tablgen] Auto-gen broadcast tables (#73654)
[llvm][AArch64] Preserve regmask when expanding the BLR_BTI pseudo instruction (#73927)
Fix bazel build (#73942)
[libc++] Workaround linker errors in floating-point atomic tests (#73398)
[AArch64] Fix predicates for FEAT_CPA's SVE-specific instructions (#73923)
[MLIR][CUDA] Update export macros in CudaRuntimeWrappers (#73932)
[InstCombine] Add KnownBits consistency assertion behind option (NFC)
[DebugInfo] Set all dbg.value intrinsics to be tail-calls (#73661)
[LoongArch] Add some binary IR instructions testcases for LSX (#73929)
MachineVerifier: Reject extra non-register operands on instructions (#73758)
[InstCombine] Require non-demanded known bits to be accurate (NFC)
[InstCombine] Use analyzeKnownBitsFromAndXorOr() in multi-use demanded bits
[InstCombine] Switch to use FileCheck as UTC was favoured (NFC)
[ValueTracking] Switch analyzeKnownBitsFromAndXorOr() to use SimplifyQuery (NFC)
[DebugInfo][RemoveDIs] Have LICM insert at iterator positions (#73671)
[InstCombine] Use pointer alignment in SimplifyDemandedBits
[libc] Add more functions in CPP/bit.h (#73814)
[DebugInfo][RemoveDIs] Emulate inserting insts in dbg.value sequences (#73350)
[libc] Add the digits property to numeric_limits (#73926)
[clang] Fix sanitizer bot failure after 14ca8d4 (#73928)
[mlir][vector] Fix a `target-rank=0` unrolling (#73365)
[DebugInfo][RemoveDIs] Have getInsertionPtAfterDef return an iterator (#73149)
[AArch64] Remove invalid check lines from sme-aarch64-svcount.ll. NFC
[LLVM][SVE] Honour calling convention when using SVE for fixed length vectors. (#70847)
[InstCombine] Regenerate test checks (NFC)
[X86] Enable v8f16/v16f16/v32f16 FCOPYSIGN custom lowering on SSE2/AVX/AVX512 targets
[X86] Add fcopysign vector test coverage
[flang] Remove useless ConvertExpr.h includes in Optimizer
[clang][AArch64] Pass down stack clash protection options to LLVM/Backend (#68993)
[clang][DebugInfo][test] Remove unchecked member variable in debug-info-static-inline-member
[InstCombine] Handle equality comparison when flooring by constant 2
[X86][NFC] Refine code in X86FoldTablesEmitter.cpp
[clang][DebugInfo] Improve heuristic to determine whether to evaluate a static variable's initializer (#72974)
[clang] Use current rounding mode for float inc/dec (#73770)
[mlir][docs] Add missing .md into Transform and Passes docs(NFC) (#73916)
[mlir][ArmSME] Switch to an attribute-based tile allocation scheme (#73253)
Minor refactororing of ASTContext::getDeclAlign() (NFC) (#72977)
[AArch64] Assembly support for the Checked Pointer Arithmetic Extension (#73777)
Fix MSVC "not all control paths return a value" warning. NFC.
[X86] Enable v32f16 FNEG custom lowering on AVX512 targets
[X86] Enable v16f16 FNEG custom lowering on AVX targets
[X86] Enable v8f16 FABS custom lowering on SSE2 targets
[X86] vec_fabs.ll - add SSE test coverage
[X86] Enable v8f16 FNEG custom lowering
[X86][Codegen] Correct the domain of VP2INTERSECT
[clang] Fix a bug with qualified name lookup into current instantiation (#73018)
[mlir][docgen] Add ops source link (#73657)
[LoongArch] Add codegen support for extractelement (#73759)
[LPM] Set gen_crash_diag=false for non-MSSA pass in MSSA pipeline
[mlir] Fix two `CHECK:` typos (#73803)
[X86][NFC] Clang-format X86FoldTablesUtils.h
[mlir][vector][doc] 0D vectors (#73792)
[X86][MC][CodeGen] Support EGPR for KMOV (#73781)
[clang] Remove extra ';' in MultiplexExternalSemaSource.cpp (NFC)
[Clang] Eagerly instantiate used constexpr function upon definition. (#73463)
[AMDGPU] Don't create mulhi_24 in CGP (#72983)
[RISCV] Add helper to copy the AVL of another VSETVLIInfo. NFC
[Flang] Add multiline error message support to pass-plugin-not-found (NFC) (#73601)
[Driver] Support -mcmodel= for LoongArch (#72514)
[RISCV] Remove old FIXMEs from test. NFC
[flang][openacc] Keep CYCLE check for compute and data construct (#73897)
[mlir] Add missing BUILD dependency for GPUToGPURuntimeTransforms
[mlir] Fix BUILD.bazel for CAPITarget
[flang] Fix Funderscoring TC failure on AIX [NFC] (#73543)
[lldb][DWARFASTParserClang][NFCI] Make ParsedDWARFTypeAttributes parameter const (#73833)
[lldb][DWARFASTParserClang][NFC] Remove redundant parameter to AddMethodToObjCObjectType (#73832)
[PowerPC] Enhance test for PR #73609. NFC.
[Clang][Lex] Fix parsing of nested requirement to prevent flowing off the end of token stream (#73691)
[llvm][NFC] Refactor autoupdater's 'c' intrinsics (#73333)
[CodeGen] Port SafeStack to new pass manager (#73747)
[BOLT] Create .text.warm for 3-way splitting (#73863)
[clang-offload-bundler] Add support for -check-input-archive (#73709)
[AMDGPU] Preserve existing alias info on LDS DMA intrinsics in MIR (#73870)
[libunwind][WebAssembly] Omit unused parts of libunwind.cpp for Wasm (#73196)
[llvm][OpenMPOpt] Remove no-op ptr-to-ptr bitcast (NFC) (#73869)
[OpenMP] Add an 'stddef.h' include to 'omp.h' (#73876)
[CodeGen] Add conditional to module cloning in bitcode linking (#72478)
[clang][NFC] Bring back missing lines from DR52 test
Fix bazel build (#73880)
[BOLT] Refactor fixBranches() (#73752)
[RISCV] Add RISCVISD::SLLW to computeKnownBitsForTargetNode.
[OpenMP][NFC] Move out plugin API and APITypes into standalone headers (#73868)
[lld][WebAssembly] Convert bitcode test to assembly. NFC (#73716)
[flang][openacc] Do not check for unlabelled CYCLE branching (#73839)
[clang][NFC] Fill in historical data on when C++ DR 1-99 were fixed
[clang][CodeGen] Emit annotations for function declarations. (#66716)
[libc++] Reenable codecvt in the dylib. (#73679)
Compile MSAN/TSAN failing test with -O1 (#73555)
[clang-tidy][NFC] Fix bugprone-suspicious-enum-usage tests
[OpenMP] Avoid initializing the KernelLaunchEnvironment if possible (#73864)
Add a scheduled build for libc++ (#73848)
[libc++] Remove linux Buildkite builders entirely (#73825)
[libc++] Run picolibc tests with qemu
[RISCV] Disable clang-format around the RISCVISD opcode enum. NFC
[Sample Profile Loader] Fix potential invalidated reference (#73181)
[workflows] Add top-level permissions for libcxx-build-and-test.yaml (#73689)
[mlir][sve] Add an e2e for linalg.matmul with mixed types (#73773)
[flang][driver][nfc] Move the definition of SemanticsContext (#73669)
[libc][NFC] Fix missing dep in bazel (#73854)
[clang][CGStmtOpenMP] Remove unneeded calls to CreatePointerBitCastOrAddrSpaceCast (NFC)
Fix stale comment (#73846)
[RISCV] Minor improvements/cleanup to target attribute handling. NFC (#73851)
[clang-format] Finalize children after formatting them (#73753)
[clang-format][NFC] Extend isProto() to also cover LK_TextProto (#73582)
[LEB128] Mark error condition with LLVM_UNLIKELY
[LEB128] Don't handle edge cases in every loop iteration
[LEB128] Factor out redundant code
[LEB128] Don't initialize error on success
[BOLT] Add structure of CDSplit to SplitFunctions (#73430)
Revert "[LEB128] Don't initialize error on success"
Revert "[LEB128] Factor out redundant code"
Revert "[LEB128] Don't handle edge cases in every loop iteration"
Revert "[LEB128] Mark error condition with LLVM_UNLIKELY"
[RISCV] Fix a couple copy/paste mistakes in riscv_crypto.h. NFC
[LV] Add test variant without sdiv by undef and uses.
[Clang] Implement P2864R2 Remove Deprecated Arithmetic Conversion on Enumerations (#73105)
[JITLink][AArch32] Test out-of-range and interworking errors for Thumb (#73397)
[mlir] Fix build with shared libs (missing cmake link dependency) (NFC)
[LEB128] Mark error condition with LLVM_UNLIKELY
[LEB128] Don't handle edge cases in every loop iteration
[LEB128] Factor out redundant code
[LEB128] Don't initialize error on success
[llvm][docs] Add example use of `llvm-reduce`. (#73237)
[msan] Intercept mallinfo2 only on GLIBC 2.33+
[LSR][TTI][RISCV] Enable terminator folding for RISC-V
[flang][openacc] Representing Fortran descriptor management with OpenACC. (#72725)
[mlir] Adopt `ConvertToLLVMPatternInterface` GpuToLLVMConversionPass to align with `convert-to-llvm` (#73761)
[mlir][spirv] Add canon patterns for IAddCarry/[S|U]MulExtended (#73340)
Allow lldb to load .dwp files with large .debug_info or .debug_types. (#73736)
[RISCV] Remove reference to non-existant __builtin_riscv_sha512sum0l/__builtin_riscv_sha512sum1l.
[msan] Intercept mallinfo2 (#73729)
Reapply "[mlir] Add mlirTranslateModuleToLLVMIR to MLIR-C (#73627)" (#73749) (#73751)
clang/APINotes: fix parenthesization of &&, ||
[mlir][spirv] Add folding for [S|U]Mod, [S|U]Div, SRem (#73341)
[MemProf][NFC] Correct comment about stripping of suffixes in profile (#73840)
[OpenMP][NFC] Separate OpenMP/OpenACC specific mapping code (#73817)
[NFC] Fix failing test introduced in #67700
Enable custom lowering of fabs_v16f16 with AVX and fabs_v32f16 with A… (#73565)
[HLSL] Support vector swizzles on scalars (#67700)
[OpenMP][NFC] Move more declarations out of private.h (#73823)
[flang] Cleanup of NYI messages (#73740)
Remove an outdated comment; NFC
[OpenMP][NFC] Put ExponentialBackoff in a Utils header (#73816)
Revert "[flang] Enable alias tags pass by default (#73111)" (#73821)
Correct the documentation category for the 'cleanup' attribute
[flang] fix codegen warning from #73641 (#73808)
[libc++][format] Fixes formatting code units as integers. (#73396)
[OpenMP][NFC] Rename OmptCallback.cpp into OpenMP/OMPT/Callback.cpp (#73813)
[OpenMP][NFC] Replace unnecessary typedefs (#73815)
[OpenMP][NFC] Move OMPT headers into OpenMP/OMPT (#73718)
[llvm] Replace uses of Type::getPointerTo (NFC)
[DebugInfo][RemoveDIs] Extract DPValues in CodeExtractor like dbg.values (#73252)
[X86InstrInfo] support memfold on spillable inline asm (#70832)
[flang] Make adapt.valuebyref attribute work again (#73658)
Fix accsessing "PresentModifierLocs" array beyond its end. (#73579)
[OpenMP][NFC] Flatten plugin-nextgen/common folder sturcture (#73725)
[LICM] Add test for #64897 (NFC)
[OpenMP][NFC] Extract timescope profile support into its own header (#73727)
[workflows] Remove sync-release-repo.yml (#73682)
[C API] Add support for setting/getting new nneg flag on zext instructions (#73592)
[libc++] Add initial support for picolibc
[libc] Fix cast semantic in integer_to_string (#73804)
[ELF,LTO] Test calloc defined in a lazy bitcode file for (malloc+memset => calloc) libcall optimization
[clang-format] Indent Verilog case statements with comments (#71353)
[clang-format] Add spaces around the Verilog implication operator (#71352)
[libc] Fix the GPU build for the hashing support (#73799)
[X86] Add fneg vector test coverage
[X86] Rename vec_fneg.ll to combine-fneg.ll
[RISCV] Pre-commit test for FrameIndex handling in getMemOperandsWithOffsetWidth
[gn build] Port 1a5af34e6fcb
[AArch64][SME2] Enable CLAMP multi-vector builtins for SME2 (#72272)
[libc++] Speed up classic locale (take 2) (#73533)
[libc++] Build the dylib with sanitizers when requested (#73656)
[mlir][nvvm] Introduce `setmaxregister.sync.aligned` Op (#73780)
[libc++] Remove unused Python imports (#73724)
[MLIR][Vector] Refactor tests for contract -> OP transforms (3/N) (#73447)
[X86] Add getTargetConstantPoolFromBasePtr helper. NFC.
[OpenMP] New Openmp device RTL functions (#73225)
[OpenMP][libomptarget] Use two SDMA engines (#73633)
[DebugInfo][RemoveDIs] Make debugify pass convert to/from RemoveDIs mode (#73251)
Move documentation about -verify from a header to public docs (#73694)
[flang] Handle absent optional in LOC (#73530)
[flang] Fix test ctofortran (#73738)
[SelectionDAG] Fix assertion for widening of mask operand in MSTORE. (#73295)
[clang][analyzer] Support `fgets` in the SteamChecker (#73638)
[NFC][TLI] Improve tests for ArmPL and SLEEF Intrinsics. (#73352)
[DAG] getNode() - relax (zext (trunc x)) -> x fold iff the upper bits are known zero.
[X86] Add test case showing failure to remove unnecessary zext from address math
[InstCombine] Regenerate test checks (NFC)
[ValueTracking] Convert MaskedValueIsZero() to use SimplifyQuery (NFC)
[ValueTracking] Convert isKnownNegative() to use SimplifyQuery (NFC)
[ValueTracking] Convert isKnownPositive() to use SimplifyQuery (NFC)
Revert "[llvm-exegesis] Set stack pointer register after starting perf counter (#72489)"
[RISCV] Support load clustering in the MachineScheduler (off by default) (#73754)
[ValueTracking] Convert isKnownNonNegative() to use SimplifyQuery (NFC)
AArch64: add support for currently released Apple CPUs. (#73499)
[NFC][MachineScheduler] Rename NumLoads parameter of shouldClusterMemOps to ClusterSize (#73757)
[OpenMP] Support for `nothing` in `metadirective` (#73690)
[llvm-exegesis] Set stack pointer register after starting perf counter (#72489)
[libc][NFC] Fix missing LIBC_INLINE + style (#73659)
[flang] add fir.box_offset operation (#73641)
[Pass][NFC] Sort and format passes in PassRegistery.def (#73762)
AArch64: switch Apple CPUs (that support it) to v8.6a. (#73497)
[Flang] Add code-object-version option (#72638)
Revert "Fix bazel build (#73741)"
[NFC] Update X86 frem CodeGen case
[GitHub] Use proper Markdown syntax and avoid child list items (#73700)
[AArch64] Load/store optimizer fixes and cleanup.
[RISCV][GISel] Support s64 G_SELECT on RV32 with D extension.
[LoongArch] Fix pattern for FNMSUB_{S/D} instructions (#73742)
[TableGen][NFC] Format getOpcodeString and remove unreachable breaks
[BOLT][DWARF] Fix output ranges for deleted code (#73464)
[RISCV][GISel] Support G_FCOPYSIGN with F and D extension.
[AMDGPU] Fix -Wc++98-compat-extra-semi after c1511a65d5c09f7cff15feba91ce9bda23d74b6e
[lldb][test] Remove `reason` from `unittest2.expectedFailure` usage (#73028)
[runtimes] Pass-through CLANG_RESOURCE_DIR to cmake invocations (#73185)
[RISCV] Refine pattern (select_cc seteq (and x, C), 0, 0, A) with Zbs. (#73746)
[msan][aarch64] Fix mallinfo interceptor (#73728)
Revert "[mlir] Re-Add mlirTranslateModuleToLLVMIR to MLIR-C (#73627)" (#73749)
[RISCV] Implement RISCVInstrInfo::getMemOperandsWithOffsetWidth (#73681)
[libc] add PREFER_GENERIC flag (#73744)
[AMDGPU] Folding imm offset in more cases for scratch access (#70634)
[clang][docs] Fix emphasis syntax in attribute documentation (#73737)
[RISCV] Generate bexti for (select(setcc eq (and x, c))) where c is power of 2. (#73649)
Fix bazel build (#73741)
[C++20] [Modules] Handling capturing strucuted bindings
[libc] fix build on aarch64 (#73739)
[MemProf] Add interface for reseting the profile file descriptor (#73714)
[Clang] CGCoroutines skip emitting try block for value returning `noexcept` init `await_resume` calls (#73160)
[GISel][Docs] Add a little bit of documentation for G_FENCE. (#73722)
[libc] [search] implement hcreate(_r)/hsearch(_r)/hdestroy(_r) (#73469)
[InstCombine] Canonicalise SextADD + GEP (#69581)
[mlir][sparse] remove COO test from trait and encoding (#73733)
[lld][WebAssembly] Fix bitcode LTO order in archive parsing (#73095)
[OpenMP][NFC] Move debug declares into CMAKE out of "private.h" (#73732)
[mlir] Re-Add mlirTranslateModuleToLLVMIR to MLIR-C (#73627)
[clang][ASTImporter] IdentifierInfo of Attribute should be set using 'ToASTContext' (#73290)
[CodeGen] Port CallBrPrepare to new pass manager (#73630)
[BOLT][DWARF][NFC] Added const to variable (#73731)
[lldb][test] TestConstStaticIntegralMember: relax assertion on number of global variables (#73707)
[mlir][sparse] refactor sparse tensor traits (#73726)
Revert "[libc][bazel] Add copts to libc_support_library" (#73723)
[clang][CodeGen] Handle template parameter objects with explicit address spaces (#69266)
[OpenMP] Use simple VLA implementation to replace uses of actual VLA
[mlir][sparse] move toCOOType into SparseTensorType class (#73708)
Revert "[OpenMP] Use simple VLA implementation to replace uses of actual VLA"
[OpenMP] Use simple VLA implementation to replace uses of actual VLA
[BOLT][DWARF] Fix handling of DWARF5 DWP (#72729)
[libc++] Properly guard std::filesystem with >= C++17 (#72701)
[OpenMP][NFC] Create an "OpenMP" folder in the include folder (#73713)
Revert "[OpenMP] Use simple VLA implementation to replace uses of actual VLA (#71412)"
[CMake] Always define runtimes-test-depends (#73629)
[OpenMP] Use simple VLA implementation to replace uses of actual VLA (#71412)
[libc++][test] Change forbidden `extents<char>` to `extents<signed char>` (#73535)
[libc][NFC] unify nextafter and nexttoward code (#73698)
[OpenMP][NFC] Move Environment.h and SourceInfo.h into "Shared" folder (#73703)
[compiler-rt][sanitizer_common] Increase min user-map/freearray round… (#73600)
Revert "[X86] With large code model, put functions into .ltext with large section flag (#73037)"
Add support for parsing type unit entries in .debug_names. (#72952)
[OpenMP][NFC] Move Utils.h and Debug.h into a "Shared" include folder (#73701)
Fix clang Attribute doc build
[libcxxabi][ItaniumDemangle] Demangle explicitly named object parameters (#72881)
[gn build] Port ed27a4edb038
[HLSL] Parameter modifier parsing and AST (#72139)
[libc++][PSTL] Implement std::equal (#72448)
Mark some std::string functions noinline. (#72869)
[X86] With large code model, put functions into .ltext with large section flag (#73037)
[llvm][HWASan] Replace calls to Type::getPointerTo (NFC)
[flang][Driver] Let the linker fail on multiple definitions of main() (#73124)
[mlir] Add subbyte emulation support for `memref.store`. (#73174)
[ASan] Allow for passing AddressSanitizer command line options through the AddressSanitizerOptions struct. (#72439)
[SLP][NFC]Use DenseSet instead of SetVector, NFC.
Revert "[Flang] Add code-object-version option (#72638)"
[NFC][docs] Add AMDGPU documentation for `LIBOMPTARGET_STACK_SIZE`
[Flang] Add code-object-version option (#72638)
[RISCV] Work on subreg for insert_vector_elt when vlen is known (#72666) (#73680)
[lldb][progress] Always report progress upon Progress object destruction (#73605)
[clang] Non-object types are non-trivially relocatable (#69734)
[SLP][NFC]Fix loops variables names, NFC.
[NFC][libc++] Refactors the time.cal tests. (#73356)
[OpenMP][NFC] Merge MemoryManager into PluginInterface (#73678)
[libc][docs] Update implementation docs (#73590)
[MsgPack] Handle Expected<T> errors in document reader (#73183)
[asan] Fix Windows i386 regression (#73650)
[OpenMP][NFC] Merge elf_common into PluginInterface (#73677)
[clang] Remove unused argument. NFC. (#73594)
[SLP][NFC]Fix naming of variables/functions, NFC.
[AMDGPU] Fix folding of v2i16/v2f16 splat imms (#72709)
[AMDGPU] Clarify description of _HI relocation types (#73663)
[BOLT][utils] Bump default time threshold to 1s in nfc-stat-parser
[flang][openacc] Enforce no branching out of compute region for combined construct (#73581)
[RISCV] Precommit test coverage for insert_vector_elt with exact VLEN
[AArch64] Assembly support for Armv9.5-A Debug/PMU Extensions (#73537)
[JITLink][AArch32] Split out error test for invalid edge in applyFixup()
[clang][DebugInfo] Revert to attaching DW_AT_const_value on static member declarations (#73626)
[RISCV] Split build_vector into vreg sized pieces when exact VLEN is known (#73606)
[SLP][NFC]Improve/fix auto declarations, NFC.
[analyzer] Use AllocaRegion in MallocChecker (#72402)
[RISCV] Minor style cleanup to cf17a24 [nfc]
[SLP][NFC]Improve aliasing support in SLP, NFC.
[readtapi][tests] add descriptions to tests (#73576)
[MLIR][Vector] Refactor tests for contract -> OP transforms (2/N) (#73445)
[LV] Add support for uniform parameters on vectorized function variants (#72891)
[BOLT][test] Update perf2bolt/perf_test.test (#73482)
[SimpleLoopUnswitch] Regenerate test checks (NFC)
[LoopVectorize] Regenerate test checks (NFC)
[InstCombine] Regenerate test checks (NFC)
[LoopPeel] Regenerate test checks (NFC)
clang: Add pragma clang fp reciprocal (#68267)
AMDGPU: Remove let Predicates from PredicateControl'd InstAliases (#73474)
[AMDGPU] Document that PAL uses Elf64_Rel relocation records (#73648)
[AMDGPU] Allow folding to FMAAK with SGPR and immediate operand on GFX10+ (#72266)
[clang] fix typo (#73644)
[NFC][NVPTX] Add a simpler test case for 0b80288e9e0b (#73379)
[ValueTracking] Rename computeKnownBitsFrom{Assume -> Context} (NFC)
[X86][fold-table] Add an assertion in addEntryWithFlags, NFCI
Improve clang-format-diff help output (#73491)
[AssumptionCache] Remove unnecessary bitcast/not handling
[OpenMP][Runtime][test] Fix ompt task testcase fail randomly (#72337)
[clang][ExtractAPI] Add support for blocks in declaration fragments (#73369)
[Clang] CWG2789 Overload resolution with implicit and explicit object… (#73493)
[clang][analyzer] Support `fputs` in the StreamChecker (#73335)
[AArch64][SVE] Remove pseudo from LD1_IMM (#73631)
[InstCombine] Use SimplifyQuery for computeKnownBits() (NFC)
[ValueTracking] Remove unused computeKnownBits() overload (NFC)
[X86] Rename X86MemoryFoldTableEntry to X86FoldTableEntry, NFCI
[RISCV] Order includes. NFC
Revert "[llvm] Disable HandleLLVMOptions in runtimes mode (#73031)"
[RISCV] Avoid looking up base opcode when converting vmerge -> vmv. NFC
[LLDB] Respect the DW_AT_alignment attribute. (#73307)
[X86] Rename some variables for memory fold and format code, NFCI
[Clang] Fix compilation with GCC 7.5
[AMDGPU] PromoteAlloca - bail always if load/store is volatile (#73228)
[gn build] Port dd0973be58b8
[TableGen] Split GlobalISelCombinerEmitter into multiple files (#73325)
[X86] X86DAGToDAGISel - fix typo in #73126
[UTC] Support arm64-apple-macosx in update_llc_test_checks.py. (#73568)
[mlir][linalg] Add TransposeConv2D Transform Op (#68567)
[gn build] Port b0af8a1ede89
Revert "[lldb] [mostly NFC] Large WP foundation: WatchpointResources (#68845)"
[lldb][PDB] Fix message order in test case
[Clang] Regenerate test checks (NFC)
[mlir][Python] Add filegroup for files in mlir/extras.
[CodeGen] Port DwarfEHPrepare to new pass manager (#72500)
[AMDGPU] Fixed folding of inline imm into dot w/o opsel (#73589)
[InstCombine] Make indexed compare fold GEP source type independent (#71663)
[DomTree] Store ReverseChildren as indices (NFC) (#73505)
Reland "[gn] port 92b821f2dcdd"
Revert "[Bazel] Fixes for c43c88501e3bc273a7c1074a19e86dc305ad7234"
Reland "[Bazel] Fix llvm-exegesis build post 12b0ab2"
[mlir][affine] Fix dim index out of bounds crash (#73266)
[X86][CodeGen] Remove duplicated code for the table checks, NFCI
[libc][math] Add unit tests for raising excepts in nextafter (#73556)
[Frontend] Remove unneeded -frtti check. NFC
[X86][CodeGen] Cleanup code for EVEX2VEX pass, NFCI
[mlir,polly] Replace uses of IRBuilder::getInt8PtrTy with getPtrTy. NFC
Reland "Reland "[llvm-exegesis] Switch from MCJIT to LLJIT (#72838)"
[RISCV][GISel] Select G_FENCE. (#73184)
[X86][CodeGen] Not compress EVEX into VEX when R16-R31 is used (#73604)
[PowerPC] Precommit test of building vector via load and zeros. NFC.
[Clang] Improve support for expression messages in `static_assert` (#73234)
[X86][CodeGen] Remove CodeSize settings for instructions, NFCI
[BPF] Add asm support for JSET insn (#73161)
[gn] port 93a2be26e7cb
[X86][MC] Keep backward compatibility in inline asm for constraints (#73529)
[mlir][sparse] code simplification: always use synthetical tensor for… (#73597)
[libc++][test] Avoid preprocessor directives in macro argument lists (#73440)
[sanitizer_symbolizer] Add initial symbolizer markup support for linux.
Add `REQUIRES: systemz-registered-target` to test added in #68926 (9a38a72).
[mlir][sparse] remove unused COO method (#73595)
[mlir][python] enable registering dialects with the default `Context` (#72488)
[libc++][test] Avoid using `allocator<const T>` (#73545)
[libc][bazel] Add copts to libc_support_library (#73591)
[RISCV] Precommit tests for buildvector lowering with exact VLEN
[mlir][arith] doc updates for ub semantics, and int representations (#72932)
[DAG] Fix a miscompile in insert_subvector undef (insert_subvector undef, ..), idx combine (#73587)
[NFC sanitizer_symbolizer] Move Fuchsia specific code.
[mlir] Add support for `memref.alloca` sub-byte emulation (#73138)
[OpenMP][NFC] Simplify code
[OpenMP][NFC] Fix diagnostic warnings
[OpenMP][NFC] Remove else after return
[OpenMP][NFC] Remove unused variable
[OpenMP][Docs] Cleanup libomptarget README
[gn build] Port fc6b72523f3d
[gn build] Port 4e0c48b907f1
[lldb] try to fix build on linux after fc6b72523f3d7
[RISCV][GISel] Select trap and debugtrap. (#73171)
[gn] port 12bb2910c3e3 more
[gn] add a bunch of missing libc++ headers
[gn] port 12bb2910c3e3
[docs] Fix issues in SourceLevelDebugging (#73528)
Revert "[mlir] Add mlirTranslateModuleToLLVMIR to MLIR-C (#73117)"
[NewPM] Remove LazyValueInfoPrinter Pass (#73408)
[mlir][sparse] code cleanup, remove FIXMEs (#73575)
[vim] set commentstring (#71838)
[BOLT] Use deterministic xxh3 for computing BF/BB hashes (#72542)
[NFC] Map changed file through envar
[RISCV] Use subreg extract for extract_vector_elt when vlen is known (#72666)
[mlir][sparse] rename DimLevelType to LevelType (#73561)
[readtapi] Add Extract & Remove architecture functionality (#72657)
[sanitizer][nfc] Reformat sanitizer_linux sources (#73573)
[mlir][python] add type wrappers (#71218)
[GitHub] Add @hanhanW to CODEOWNERS for relevant files. (#73571)
[OpenMP] Temporarily disable OMPT tests for x64
[sanitizer] Add re-execution on FreeBSD when ASLR is detected (#73439)
Include <vector> in WatchpointResource.h
[libc] Move in_use into OptionalStorage (#73569)
[SystemZ][z/OS] This change adds support for the PPA2 section in zOS (#68926)
[lldb] [mostly NFC] Large WP foundation: WatchpointResources (#68845)
[mlir][vector] Distribute all non-permutation or broadcasted masked transfer reads (#73539)
[NFC] Cleanup and sort hlsl_intrinsics.h (#72414)
[cross-project-tests] Fix struct-dse example so that it fails again (#73566)
[RISCV][GISel] Legalize and select G_BRINDIRECT. (#73059)
[clang-format] Add BreakAdjacentStringLiterals option (#73432)
[RISCV][GISel] Fix 2 indirect call bugs. (#73170)
[InstCombine] Set disjoint flag when turning Add into Or. (#72702)
[libc++] Move compiler-detection Lit features first (#73544)
[OpenMP] Enable position independent code for libomptarget
[OpenMP] Add back implicit flags manually
[Bazel] Fix for 7eccd5284290fea26df44fbc5a604b54e7b115a2
Revert "[libc++] Speed up classic locale (#72112)"
[AArch64] Update comment on GHC CC (#72761)
[llvm] Disable HandleLLVMOptions in runtimes mode (#73031)
[libc++] Add missing REQUIRES for exception handling test
[OpenMP] Tear down GenericDeviceTy's with GenericPluginTy (#73557)
[unittest][Support] Fix bad negation of signed integer in LEB128Test.SLEB128Size (#72700)
Fix ISel crash when lowering BUILD_VECTOR (#73186)
[OpenMP][NFC] Minor name and code simplification
[OpenMP][NFC] Remove no-op __tgt_rtl_deinit_plugin
[OpenMP][NFC] Remove unused (un)register_lib plugin API
[BOLT][test] Replace /dev/null with temp file (#73485)
[mlir] Add narrow type emulation for `memref.reinterpret_cast` (#73144)
Revert "[MCP] Enhance MCP copy Instruction removal for special case (#70778)"
[RISCV] Add register bank and instruction selection support for FP G_SELECT. (#72726)
[RISCV] Fix a warning
[OpenMP][Tools] Have sort(1) not use long name parameters (#73477)
[runtime] Have the runtime use the compiler builtin for alloca on NetBSD (#73480)
[RISCV][GISel] Use setDesc in some cases instead of creating new instructions. (#72769)
[ORC] Add a public unsafe-operations helper for SymbolStringPtr.
[ORC-RT] Add SPS serialization for std::tuple.
[mlir][vector] Fix patterns for dropping leading unit dims from masks (#73525)
[mlir][spirv] Simplify gpu reduction to spirv logic (#73546)
[libc++] Fix UTF-8 decoding in codecvts (#68442)
Changes to support running tests for Windows arm64 asan (#66973)
[llvm-objdump] Fix lma display issue for non-bss sections (#72141)
Revert "[llvm] Disable HandleLLVMOptions in runtimes mode (#73031)"
[RISCV][GISel] Merge selectGlobalValue and selectJumpTable. NFC (#72759)
[RISCV] Add Zbs Write classes to SiFive7AnyToGPRBypass. (#72560)
[clang] Fix a new test to not write temp files to the git repository
[llvm] Replace calls to Type::getPointerTo (NFC)
Reland "[mlir][gpu] Align reduction operations with vector combining kinds (#73423)"
[ValueTracking] Pass unpacked arguments to computeKnownBitsFromCmp() (NFC)
[RISCV] Partially move doPeepholeMaskedRVV into RISCVFoldMasks (#72441)
Revert "[mlir][gpu] Align reduction operations with vector combining kinds (#73423)"
[unittest,examples] Replace uses of IRBuilder::getInt8PtrTy with getPtrTy. NFC
[AArch64] Add a test to show scheduling aliasing between SVE loads and stores. NFC
[mlir][gpu] Align reduction operations with vector combining kinds (#73423)
[mlir][spirv] Handle non-innerprod float vector add reductions (#73476)
[libc][NFC] Decouple FP properties from C++ types (#73517)
[Flang][OpenMP] Fix issue with empty critical or critical without surrounding context (#71944)
[libc++] Remove experimental pmr headers now shipped in mainline (#73172)
[InstCombine] Remove over-generalization from computeKnownBitsFromCmp() (#72637)
[libc++][NFC] Fix typo in comment
ms inline asm: Fix {call,jmp} fptr (#73207)
[libc++][test] Avoid C++23 Core features that MSVC lacks (#73438)
[OpenMP] Fix missing CMake function in runtimes build
[libc] Fix forward octal prefix (#73526)
[clangtidy] Allow safe suspensions in coroutine-hostile-raii check (#72954)
[libc++] fix some inconsistencies on libcxx status pages (#73471)
[flang] Enable alias tags pass by default (#73111)
[libc] Add detection support for float16 (#73372)
[X86] vector-interleaved tests - add AVX512F/AVX512DQ/AVX512BW/AVX512DQBW-ONLY common prefixes to merge more SLOW/FAST checks
Revert "[Bazel] Fix llvm-exegesis build post 12b0ab2"
[OpenACC] Implement 'routine' construct parsing (#73143)
[libc] Use file lock to join newline on RPC puts call (#73373)
[MachineLICM] Fix incorrect CSE on hoisted const load (#73007)
[libcxxabi][test][NFC] Turn off clang-format for demangler test-case array (#73503)
[llvm] Disable HandleLLVMOptions in runtimes mode (#73031)
[DomTree] Avoid duplicate hash lookup (NFC)
[llvm][TypeSize] Consider TypeSize of '0' to be fixed/scalable-agnostic. (#72994)
[clang][dataflow] Strengthen widening of boolean values. (#73484)
[X86] Regenerate vector-interleaved-store-i64-stride-4.ll
[X86][CodeGen] Teach frame lowering to spill/reload registers w/ PUSHP/POPP, PUSH2[P]/POP2[P] (#73292)
[clang][AArch64][NFC] Remove trailing space in SME intriniscs header
Revert "[runtimes] Add missing test dependencies to check-all (#72955)"
Re-commit "[MachineSink][AArch64] Enable sink-and-fold by default (#72132)"
[mlir][cuda] Guard mgpuLaunchClusterKernel for Cuda 12.0+ (NFC) (#73495)
Revert "[gn] port 92b821f2dcdd"
Revert "Reland "[llvm-exegesis] Switch from MCJIT to LLJIT (#72838)""
[X86] X86DAGToDAGISel - attempt to merge XMM/YMM loads with YMM/ZMM loads of the same ptr (#73126)
[clang][Interp][NFC] const qualify a local variable
[mlir][spirv][doc] Remove duplicate syntax formats (#73386)
[DomTree] Remove unnecessary domtree level check in SemiNCA (NFC) (#73107)
[LAA] Check HasSameSize before couldPreventStoreLoadForward.
[clang] Print static_assert values of arithmetic binary operators (#71671)
[LAA] Auto-generate checks for forward-loop-carried.ll
[mlir][gpu] Support Cluster of Thread Blocks in `gpu.launch_func` (#72871)
[Clang][SME2] Add outer product and accumulate/subtract builtins (#71176)
AArch64: remove duplicate SHA3 feature from Apple CPUs. NFC.
[lldb][AArch64][Linux] Correct name of FPCR field
[lldb] Improve error message for script commands when there's no interpreter (#73321)
[ConstraintElim] Refactor GEP offset collection.
[InstCombine] Add more inbounds tests for indexed compare fold (NFC)
[DomTree] Reduce number of hash table lookups (NFC) (#73097)
[AArch64][GlobalISel] Better vecreduce.fadd lowering. (PR #73294)
[JumpThreading] Remove LVI printer flag (#73426)
[CVP] Don't try to fold load/store operands to constant (#73338)
[mlir][affine][gpu] Replace DivSIOp to CeilDivSIOp when lowering to GPU launch (#73328)
[X86][MC] Allow to specify any of the 8/16/32/64 register names interchangeably for R16-R31 (#73421)
[clang-format][NFC] Improve an `if` conditional in the annotator
[gn build] Port f8afc53d641c
[libc++] Speed up classic locale (#72112)
[RISCV] Don't combine store of vmv.x.s/vfmv.f.s to vp_store with VL of 1 when it's indexed store (#73219)
[mlir] Fix a warning
[mlir][spirv] Add missing group non-uniform bitwise and logical ops (#73475)
[SelectionDAG] Add instantiated OPC_EmitInteger and OPC_EmitStringInteger (#73241)
[Driver] Allow -e entry but reject -eentry (#72804)
[X86][MC] Update condition about ExplicitVEXPrefix (#73312)
[llvm][SROA] Replace calls to Type::getPointerTo (NFC)
AMDGPU: Move intrinsic definition out of subtarget specific section
[libc++][test] Use `LIBCPP_STATIC_ASSERT` for `std::__mdspan_detail` (#73436)
[libc++][test] Cleanup typos and unnecessary semicolons (#73435)
Revert "[LLDB] Add more helper functions to CompilerType class."
[mlir][spirv] Update integer dot product op syntax (#73468)
[LLDB] Add more helper functions to CompilerType class. (#73467)
[mlir][spirv] Add floating point dot product (#73466)
[clang-format] Fix a bug in formating `#define A x:` (#73220)
Fix for building llvm-flang with gcc 7.5.0 (minimum LLVM required gcc version (#73265)
[libc++] Add missing headers to the modulemap (#71127)
[DebugInfo][RemoveDIs] Instrument loop-rotate for DPValues (#72997)
Suppressed unused-var warning from c672ba7dde
[gn] manually re-sync llvm-exegesis tool deps
[gn] port 92b821f2dcdd
[DebugInfo][RemoveDIs] Instrument inliner for non-instr debug-info (#72884)
[NewPM] Remove StripGCRelocatesLegacyPass (#73403)
[LVI] Add NewPM printer pass (#73425)
[libc++][NFC] Use __construct_at and __destroy_at instead of using preprocessor conditionals (#70866)
[libc++] Fix a few tests that are missing proper suffixes (#73444)
[OpenMP] atomic compare fail : Parser & AST support
[readtapi] Use consistent error handling diagnostics (#73419)
[libc++][test] Fix unused and nodiscard warnings (#73437)
[GitHub] Update bot message to suggest GitHub pull requests (#65914)
test-release.sh: Default to letting ninja select the number of build jobs (#73187)
[llvm][Coroutines] Remove no-op ptr-to-ptr bitcasts (NFC) (#73427)
[NewPM] Remove CFGPrinterLegacyPass (#73414)
[Bazel] Fix llvm-exegesis build post 12b0ab2
[BOLT] Reset output addresses for deleted blocks (#73429)
Add "REQUIRES: asserts" to test as it requires the compiler to hit an assertion failure to pass and was failing in release builds.
[BOLT][NFC] Refactor function state check (#73420)
[XCOFF] make related SD symbols as isFunction (#69553)
[X86][MC] Remove duplicated code in getX86SubSuperRegister by using macros, NFCI
[X86[MC][test] Error for EGPR in 32-bit mode
[NewPM] Remove CFGOnlyPrinterLegacyPass (#73412)
[NewPM] Remove Legacy RedudantDbgInstEliminationPass (#73405)
[NewPM] Remove CFGOnlyViewerLegacyPass (#73411)
[NewPM] Remove CFGViewerLegacyPass (#73410)
[NewPM] Remove CallGraphPrinterLegacyPass (#73409)
[NewPM] Remove PredicateInfoPrinterLegacyPass (#73407)
[NewPM] Remove Reg2Mem Legacy Pass (#73404)
[DWARFLinkerParallel] fix build on 32-bit platform. (#73388)
Reland "[llvm-exegesis] Switch from MCJIT to LLJIT (#72838)"
Reland "[clang][Sema] Use original template pattern when declaring implicit deduction guides for nested template classes" (#73087)
[RISCV][GISel] Add simplest case of folding add with immediate into load/store address.
[RISCV][GISel] Test G_FRAME_INDEX folding into store address. NFC
[CGOpenMPRuntimeGPU] Remove no-op ptr-to-ptr bitcasts (NFC)
Change-Id: Icbafb6ee92c94a6e5c2fed5ed73f3e8facabfb2b
Signed-off-by: greenforce-auto-merge <greenforce-auto-merge@users.noreply.github.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This was modifying a container as it iterated it, which tripped a check in libstdc++'s debug checks.
Instead, just assign to the item via the reference we already have.
This fixes the following expensive checks failures on my machine:
LLVM :: CodeGen/PowerPC/fp-strict.ll
LLVM :: CodeGen/PowerPC/pr55463.ll
LLVM :: CodeGen/PowerPC/register-pressure.ll
LLVM :: CodeGen/PowerPC/spe.ll
Which are some of the tests noted by #68594.