[Mlir-commits] [mlir] cpdl (PR #95829)

Mon Jun 17 11:57:29 PDT 2024

llvmbot wrote:




@llvm/pr-subscribers-mlir

Author: Jacques Pienaar (jpienaar)

<details>
<summary>Changes</summary>

- **[mlir][bzl] Add missing dep**
- **[llvm-profdata] Clean up traverseAllValueSites (NFC) (#95467)**
- **[MC] flushPendingLabels: revert setAtom change**
- **[mlir][spirv] Implement SPIR-V lowering for `vector.deinterleave` (#95313)**
- **[MC] Move MCFragment::Atom to MCSectionMachO::Atoms**
- **[LLDB][Windows] Fix watchpoints for Windows (#95446)**
- **[HWASan] disable hwasan_symbolize_stack_uas on x86**
- **[libc][stdlib] Fix UB in freelist (#95330)**
- **[BOLT] Fix duplicate diagnostic message (#95167)**
- **[mlir][mesh] Fix wrong argument passed to targetShardingInUnsplitLastAxis (#95059)**
- **[flang] Address missed cases for REDUCE change**
- **[lldb] Use packaging module instead of pkg_resources (#93712)**
- **[Transforms] Migrate to a new version of getValueProfDataFromInst (#95477)**
- **Reland '[hwasan] Add fixed_shadow_base flag (#73980)' (#95445)**
- **[HWASan] comment why hwasan_symbolize_stack_uas is arm64 only**
- **[llvm][SelectionDAG] Fix up chains in lowerInvokeable. rdar://113994760 (#94004)**
- **[compiler-rt][AArch64][FMV] Use the hw.optional.arm.caps fast path (#95275)**
- **[X86][AsmParser] Avoid duplicated code in MatchAndEmit(ATT/Intel)Instruction**
- **[RISCV] Don't use SEW=16 .vf instructions to move scalar bf16 into a vector.**
- **[Transforms] Migrate to a new version of getValueProfDataFromInst (#95485)**
- **[Codegen] (NFC) Faster algorithm for MachineBlockPlacement (#91843)**
- **[X86][MC] Add missing support for pseudo rex/rex2 prefix in assembler**
- **[RISCV] Remove unused check prefixes. NFC**
- **[RISCV] Remove duplicate bf16 testing. NFC**
- **StreamChecker.cpp: Use isa<> (for #93408) [-Wunused-but-set-variable]**
- **Reland "[libc] fix aarch64 linux full build (#95358)" (#95423)**
- **[mlir][scf]: Expose emitNormalizedLoopBounds/denormalizeInductionVariable util functions (#94429)**
- **[MLIR][Flang][OpenMP] Implement lowering simd aligned to MLIR (#95198)**
- **[RISCV] Remove unnecessary bf16 -mattr from vfmv.v.f.ll. NFC**
- **[MLIR][GPU] Add gpu.cluster_dim_blocks and gpu.cluster_block_id Ops (#95245)**
- **[RISCV] Remove partially duplicate riscv_vfmv_v_f_vl patterns.**
- **[RISCV] Remove vfmerge.vf patterns with bf16 types.**
- **[RISCV] Add vselect pattern with SelectFPImm.**
- **[Safe Buffers] Serialize unsafe_buffer_usage pragmas  (#92031)**
- **[libc] Provide vprintf for baremetal (#95363)**
- **[libc] printf**
- **[lldb][TypeSystemClang][NFCI] Factor completion logic for individual types out of GetCompleteQualType (#95402)**
- **[NFCI][AMDGPU] Try to use PressureDiff to Calculate RegPressure. (#94221)**
- **[clang][Fuchsia] Use unsigned int for wint_t on *-fuchsia targets (#95499)**
- **[lld][ELF] Add basic TLSDESC support for LoongArch**
- **[ARM][AArch64] Bail out if CandidatesWithoutStackFixups is empty (#95410)**
- **[Driver] Support -Wa**
- **[llvm-exegesis] Only link/initialize supported targets (NFC) (#95421)**
- **[Driver] Support -Wa**
- **[CodeGen] Remove target SubRegLiveness flags (#95437)**
- **[SDAG] Lower range attribute to AssertZext (#95450)**
- **MathExtras/test: increase coverage (#95425)**
- **[clang-cl] Map /Ot to -O3 instead of -O2 (#95406)**
- **[SimplifyCFG] More accurate use legality check for sinking (#94462)**
- **[SPIRV] Silence unused variable warnings (#95492)**
- **[APFloat] Add APFloat support for FP4 data type (#95392)**
- **Revert "[MC/DC][Coverage] Loosen the limit of NumConds from 6 (#82448)"**
- **Reapply#4 "[RemoveDIs] Load into new debug info format by default in LLVM (#89799)"**
- **[SimplifyCFG] Add tests for sinking with multiple uses (NFC)**
- **[Flang][OpenMP] NFC: Check omp.loop_nest in compound construct lowering (#95404)**
- **[flang][OpenMP] Support reduction of POINTER variables (#95148)**
- **[clang-cl] Support the /Ob3 flag**
- **[libclang/python] Fix bugs in custom enum implementation and add tests (#95381)**
- **[MC] Cache MCRegAliasIterator (#93510)**
- **[AArch64] LowerAVG - fallback to default expansion (#95416)**
- **[Error] Add non-consuming toString (#95375)**
- **[SDPatternMatch] Only match ISD::SIGN_EXTEND in m_SExt (#95415)**
- **[SROA] Preserve all GEP flags during speculation**
- **[AArch64] Add i128 and fp128 tests to fptoi. NFC**
- **[AMDGPULowerBufferFatPointers] Don't try to preserve flags for constant expressions**
- **Reapply: [MC/DC][Coverage] Loosen the limit of NumConds from 6 (#82448)**
- **[lldb][test] Add test for completing ObjCObjectType (#95405)**
- **Recommit "[VPlan] First step towards VPlan cost modeling. (#92555)"**
- **[AArch64] Extend SVE diagnostics. (#94976)**
- ** [AArch64][SME] Enable subreg liveness tracking for AArch64 (#92142)**
- **[AMDGPU] Extend llvm.amdgcn.set.inactive intrinsic to support Reg32/Reg64 types (#94457)**
- **[llvm-c] Move LLVMAttributeIndex to a more apropriate place. NFC.**
- **[Flang][OpenMP] Fix type in getBaseObject causing crashes in certain scenarios (#95472)**
- **[llvm-exegesis] Fix typos in cmake file**
- **[flang] Implement !DIR$ VECTOR ALWAYS (#93830)**
- **[X86] Add scalar test coverage for ISD::AVG nodes on 32 and 64-bit targets**
- **[AMDGPU] Change getLdStRegisterOperand to !cond for better diagnostic (#95475)**
- **[ProfileData] Migrate to getValueArrayForSite (#95493)**
- **[include-cleaner] Pass WorkingDir to suggestPathToFileForDiagnostics (#95114)**
- **[DAG] fold `avgs(sext(x)**
- **Fix silent truncation of inline ASM `srcloc` cookie when going through a `DiagnosticInfoSrcMgr` (#84559)**
- **[AMDGPU] Make use of composeSubRegIndices. NFCI. (#95548)**
- ** [RemoveDIs] Print IR with debug records by default (#91724)**
- **[clang][HeaderSearch] Fix handling of relative file-paths in suggestPathToFileForDiagnostics (#95121)**
- **[mlir][spirv] Add integration test for `vector.deinterleave` (#95469)**
- **[VPlan] Pre-compute cost for all instrs only feeding exit conditions.**
- **[OpenMP] Add Environment Variable to disable Reuse of Blocks for High Loop Trip Counts (#89239)**
- **[ARM] Add tan intrinsic lowering (#95439)**
- **[AMDGPULowerBufferFatPointers] Fix offset-only ptrtoint (#95543)**
- **[mlir][bzl] Fix broken BUILD due to typo in `CAPITransformsObjects` in BUILD.bazel.**
- **[MC][X86] addConstantComments - add mul vXi16 comments**
- **[CLANG][DWARF] Handle DIE offset collision in DW_IDX_parent (#95339)**
- **[AMDGPU] NFC: Remove duplicate VOP_DPP_Pseudo TableGen definitions (#95370)**
- **[mlir][Transform] Extend transform.foreach to take multiple arguments (#93705)**
- **[libcxx][test][AIX] address more platform differences in locale tests (#94826)**
- **[mlir][test] Fix filecheck annotation typos [2/n] (#93476)**
- **[GlobalISel][AArch64] AArch64O0PreLegalizerCombiner: Disable fixed-point iteration (#94291)**
- **[ProfileData] Fix the order of tests (#95549)**
- **[ProfileData] Sink the length checks (#95559)**
- **[libc][math][c23] Add f16fmaf C23 math function (#95483)**
- **[RISCV] Add isel patterns for bf16 riscv_vfmv_v_f_vl of FP constant.**
- **[lldb] Stop testing LLDB on Clang changes in pre-commit CI (#95537)**
- **[Transforms] Migrate to a new version of getValueProfDataFromInst (#95442)**
- **[Flang] Hoisting constant-sized allocas at flang codegen. (#95310)**
- **[libc++] Add default copy ctor to "__chrono/exception.h" (#95338)**
- **[MC] Add MCFragment allocation helpers**
- **[libc++] Rename workflow that restarts preempted jobs**
- **Make diagnostic pragma override -Werror=foo and DefaultError warnings**
- **[Analysis] Migrate to a new version of getValueProfDataFromInst (#95561)**
- **[MC] Aligned bundling: remove special handling for RelaxAll**
- **[AMDGPU] Adding multiple use analysis to SIPeepholeSDWA (#94800)**
- **[libc] Fixed NamedType usage in Fenv HeaderSpec Types (#95487)**
- **[flang] Lower REDUCE intrinsic for reduction op with args by value (#95353)**
- **[flang] Add stack reclaim pass to reclaim allocas in loop (#95309)**
- **[mlir][sparse] add canonicalization patterns for IterateOp. (#95569)**
- **Revert "[libc] printf**
- **Revert "Recommit "[VPlan] First step towards VPlan cost modeling. (#92555)""**
- **[flang][cuda] Propagate data attribute to global with initialization (#95504)**
- **[NFC][PowerPC] Add test to check lanemasks for subregisters. (#94363)**
- **[compiler-rt] Remove 'memprof_meminfoblock.h' from MEMPROF_HEADERS (NFC) (#95334)**
- **[HLSL] Use hlsl vector template in type printer (#95489)**
- **Check whether EvaluatedStmt::Value is valid in VarDecl::hasInit (#94515)**
- **[libc++] Revert temporary attempt to implement LWG 4110 (#95263)**
- **[clang-doc][cmake] Copy assets to build directory (#95187)**
- **[libc++] Revert LWG3223 Broken requirements for shared_ptr converting constructors (#93071)**
- **[flang] Lower function/subroutine attribute to func op (#95468)**
- **Revert " [AArch64][SME] Enable subreg liveness tracking when SME is available" (#95574)**
- **[lldb] Tweak Python interpreter workaround on macOS (#95582)**
- **[mlir][scf]: Removed LoopParams struct and used Range instead (NFC) (#95501)**
- **[GlobalIsel] Import GEP flags (#93850)**
- **[NFC][PowerPC] Update the option to -enable-subreg-liveness.**
- **[libc][stdlib] Make the FreeListHeap constant-initializable (#95453)**
- **Set Support system_libs if WIN32**
- **[clang-interp] Use -fno-sized-deallocation in two tests (#95546)**
- **[lldb][test] Force dwarf4 usage in test requiring it (#95449)**
- **[InstallAPI] Pick up input headers by directory traversal (#94508)**
- **[gn build] Port feed66f3eae5**
- **[Clang][NFC] Avoid opening namespace std (#95470)**
- **[libc] Remove unnecessary include from putchar.h (#95576)**
- **[libc] add LIBC_INLINE annotations to BlockStore (#95573)**
- **[libc] add rwlock (#94156)**
- **[mlir] Do not merge blocks during canonicalization by default (#95057)**
- **[flang] Fix shared libs build for alloca change. (#95597)**
- **[libc] fix preferred_type attribute detection (#95599)**
- **[AMDGPU] Fix lowering of abs for i16 vectors with more than 2 elements (#95413)**
- **Revert "[mlir][spirv] Add integration test for `vector.deinterleave`" (#95607)**
- **[VPlan] Support extracting any lane of uniform value.**
- **[CodeGen] Support vectors across all backends (#95518)**
- **[lldb] Fix the semantics of SupportFile equivalence (#95606)**
- **[libc] fix build errors (#95600)**
- **[clang-tidy] avoid false positive when overload for bugprone-return-const-ref-from-parameter (#95434)**
- **[ProfileData] Migrate to a new version of getValueProfDataFromInst (#95568)**
- **[ProfileData] Sink the length checks (#95604)**
- **[libc] fix build errors (#95613)**
- **[mlir][tosa] Only match rfft2d of floats in linalg conversion (#93432)**
- **Reapply "[libc] printf**
- **[libc] Add the implementation of the fdopen function (#94186)**
- **[flang] Fixing PPC lit failure due to alloca reordering. (#95621)**
- **[MLIR][Arith][Resubmit] add fastMathAttr on arith::extf and arith::truncf (#95346)**
- **[AMDGPU] Fix lit failure (#95620)**
- **[Support] Import SipHash c reference implementation. (#94393)**
- **[Support] Integrate SipHash.cpp into libSupport.  (#94394)**
- **[gn build] Port cfbed2c0e611**
- **[llvm][AArch64] Support -mcpu=apple-m4 (#95478)**
- **[llvm][AArch64] Rearrange Apple CPUs by generation**
- **[Support] Add SipHash-based 16-bit ptrauth ABI-stable hash. (#93902)**
- **[lldb] Adds additional fields to ProcessInfo (#91544)**
- **[libc][__support][bit] Switch popcount to Brian Kernighan’s Algorithm (#95625)**
- **[mlir][scf]: Copy old attributes of old ForOp in replaceWithAdditionalYields (#95502)**
- **[clang][Interp] Use different inline descriptors for global variables**
- **[clang-format] Don't over-indent comment below unbraced body (#95354)**
- **(New) Add option to generate additional debug info for expression dereferencing pointer to pointers (#95298)**
- **[clang][Interp] Fix calling lambdas with explicit instance pointers...**
- **[clang][Interp][test] Move explicit object parameter test to cxx23.cpp**
- **[llvm] Use llvm::unique (NFC) (#95628)**
- **[ELF] -r: force -Bstatic**
- **AMDGPU: Legalize atomicrmw fadd for v2f16/v2bf16 for local memory (#95393)**
- **AMDGPU: Handle legal v2f16/v2bf16 atomicrmw fadd for global/flat (#95394)**
- **[ARM] Remove duplicate custom SDag node (NFCI) (#93419)**
- **MathExtras: rewrite some methods to never overflow (#95556)**
- **[clang][Interp] Fix checking null pointers for initialization**
- **[Debug Info] Fix debug info ptr to ptr test (#95637)**
- **[GISel] Unify multiple instances of getTypeForLLT (NFC) (#95577)**
- **[AMDGPU][GISel] Use datalayout alignment for buffer-load legalization (#95578)**
- **[clang-tidy]fix false positives of the result of std::move() is used as rvalue for performance-move-const-arg (#95633)**
- **[X86] Lower vXi8 multiplies by constant using PMADDUBSW on SSSE3+ targets (#95403)**
- **[clang][Interp][NFC] Handle AddressSpaceConversion casts**
- **[clang][Interp] Support ExtVectorElementExprs**
- **[RISCV] Use isCompatible when we need runtime VSETVLIInfo equality. NFC (#94340)**
- **[x86] Fix -Wunused-value in X86ISelLowering.cpp (NFC)**
- **[X86] Add test showing failure to constant fold (V)PMADDWD nodes**
- **[X86] computeKnownBitsForTargetNode - add handling for (V)PMADDWD nodes**
- **[clang][modules] HeaderSearch::MarkFileModuleHeader sets textual headers' HeaderFileInfo non-external when it shouldn't (#89005)**
- **[RISCV] Add scalar type test coverage for ISD::AVG nodes**
- **[Clang][Driver] Expose `-fno-eliminate-unused-debug-types` to clang-cl (#95259)**
- **Revert "AMDGPU: Handle legal v2f16/v2bf16 atomicrmw fadd for global/flat (#95394)"**
- **[X86] Add test showing failure to constant fold (V)PMADDUBSW nodes**
- **[X86] computeKnownBitsForTargetNode - add handling for (V)PMADDUBSW nodes**
- **[mlir][gpu] Add builder to `gpu.launch_func` (#95541)**
- **[libc++][TZDB] Disables a failing test. (#95659)**
- **[lldb] Remove dead code block (NFC) (#94775)**
- **[llvm][CodeGen] Fix failure in window scheduler caused by weak dependencies (#95636)**
- **[DAG] Fold Y = sra (X**
- **[MachineSink] Use SmallDenseMap (NFC) (#95676)**
- **[LV] Add tests with multiple conditions feedin exit branches.**
- **[JumpThreading] Use SmallPtrSet (NFC) (#95674)**
- **[TargetLibraryInfo] Use std::move (NFC) (#95671)**
- **Cleanup MC/DC intrinsics for #82448 (#95496)**
- **[clang-tidy] fix false negatives for performance-inefficient-vector-operation (#95667)**
- **[CostModel][X86] Update vXi8 mul costs for AVX512BW/AVX2/AVX1/SSE**
- **[TargetLibraryInfo] Use the default move constructor/assignment operator (#95685)**
- **[StackColoring] Declare BitVector outside the loop (#95688)**
- **[X86] Fix throughput of AVX2/AVX512VL vector extension/truncations**
- **[clang] Clean up macOS version flags (#95374)**
- **Reformat**
- **mmapForContinuousMode: Calculate FileOffsetToBitmap from FileOffsetToCounters. NFC.**
- **[Clang] Instantiate local constexpr functions eagerly (#95660)**
- **[Clang][AArch64] Generalise streaming mode checks for builtins. (#93802)**
- **AMDGPU: Don't use amdgpu_ps for some atomic tests**
- **AMDGPU: Expand raw ptr buffer atomic fadd tests**
- **AMDGPU: Fix legalization for llvm.amdgcn.raw.buffer.atomic.fadd.v2bf16**
- **[VPlan] Use VPTransformState::UF in vectorizeInterleaveGroup (NFCI).**
- **[libcxx] [ci] Simplify the mingw x86_64 configs (#95609)**
- **[clang-format] Handle AttributeMacro before access modifiers (#95634)**
- **[clang-format] Add DiagHandler parameter to format::getStyle() (#91317)**
- **[clang][NFC] Update C++ DR issues list**
- **[Transforms] Replace incorrect uses of m_Deferred with m_Specific (#95719)**
- **[clang-format][NFC] Suppress diagnostic noise in GetStyleOfFile test**
- **[DebugInfo][Reassociate] Fix missing debug location drop (#95355)**
- **[clang-format] Handle Verilog delay control (#95703)**
- **[C++20] [Modules] [Itanium ABI] Generate the vtable in the module unit of dynamic classes (#75912)**
- **[LLDB] Remove dead code (NFC) (#95713)**
- **[InstSimplify] Implement simple folds for `ucmp`/`scmp` intrinsics (#95601)**
- **[InstCombine] simplify average of lsb (#95684)**
- **[SimplifyCFG] Add support for sinking instructions with multiple uses (#95521)**
- **[clang][AArch64] Add validation for Global Register Variable. (#94271)**
- **LangRef: getelementptr: inbounds is about the object the pointer is 'based on' (#95650)**
- **[AMDGPULowerBufferFatPointers] Expand const exprs using fat pointers (#95558)**
- **[AMDGPU][LIT] Added a MIR LIT showing the SGPR spills (#94584)**
- **[RISCV] Remove getOffsetOfLocalArea() (#93765)**
- **[InstCombine] Prefer source over result element type (NFC)**
- **[SelectionDAG] Add support for the 3-way comparison intrinsics [US]CMP (#91871)**
- **[mlir][ArmSVE] Lower predicate-sized vector.create_masks to whilelt (#95531)**
- **AMDGPU: Cleanup struct buffer atomic fadd intrinsic tests**
- **mlir: fix incorrect usages of divideCeilSigned (#95680)**
- **[lldb] Add register field enum class (#90063)**
- **[flang][debug] Support fixed size character type. (#95462)**
- ** [DebugInfo] Handle DW_OP_LLVM_extract_bits in SROA (#94638)**
- **[libc++] Mark more types as trivially relocatable (#89724)**
- **[LLVM][DAGCombiner] Extend coverage for insert_subv(undef**
- **[Xtensa] Fix register asm parsing. (#95551)**
- **[clang][CodeGen] Return RValue from `EmitVAArg` (#94635)**
- **[clang][CodeGen] Fix EmitInvariantStart for non-zero addrspace (#94346)**
- **[MachineLICM] Correctly Apply Register Masks (#95746)**
- **[mlir][emitc] Add EmitC index types (#93155)**
- **[DebugInfo] Update sroa-extract-bits.ll test (#95774)**
- **[CodeGen] Do not include $noreg in any regmask operands. NFCI. (#95775)**
- **[flang][debug] Support allocatables. (#95557)**
- **[AArch64] Refactor creation of a shuffle mask for TBL (NFC) (#92529)**
- **AMDGPU: Fix legalization for llvm.amdgcn.struct.buffer.atomic.fadd.v2bf16**
- **[libc++] Guard transitive include of `<locale>` with availability macro (#95686)**
- **[flang] Add -mlink-builtin-bitcode option to fc1 (#94763)**
- **[InstCombine] Add test for #95547 (NFC)**
- **[InstCombine] Don't preserve context across div**
- **[mlir][Target] Improve ROCDL gpu serialization API (#95456)**
- **[clang][Interp] Implement Complex-complex multiplication (#94891)**
- **Revert [mlir][Target] Improve ROCDL gpu serialization API (#95790)**
- **[VectorCombine] Change shuffleToIdentity to use Use. NFC**
- **Reapply "AMDGPU: Handle legal v2f16/v2bf16 atomicrmw fadd for global/flat (#95394)"**
- **[TableGen][Docs] Fix `!range` markup (#95540)**
- **[mlirc] Add missing extern C**


---
Full diff: https://github.com/llvm/llvm-project/pull/95829.diff


1 Files Affected:

- (modified) mlir/include/mlir-c/Rewrite.h (+8) 


``````````diff

diff --git a/mlir/include/mlir-c/Rewrite.h b/mlir/include/mlir-c/Rewrite.h
index 45218a1cd4ebd..bed93045f4b50 100644
--- a/mlir/include/mlir-c/Rewrite.h
+++ b/mlir/include/mlir-c/Rewrite.h
@@ -19,6 +19,10 @@
 #include "mlir-c/Support.h"
 #include "mlir/Config/mlir-config.h"
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+
 //===----------------------------------------------------------------------===//
 /// Opaque type declarations (see mlir-c/IR.h for more details).
 //===----------------------------------------------------------------------===//
@@ -57,4 +61,8 @@ mlirRewritePatternSetFromPDLPatternModule(MlirPDLPatternModule op);
 
 #undef DEFINE_C_API_STRUCT
 
+#ifdef __cplusplus
+}
+#endif
+
 #endif // MLIR_C_REWRITE_H

``````````

</details>


https://github.com/llvm/llvm-project/pull/95829
