[llvm] [AArch64][SME] Implement the SME ABI (ZA state management) in Machine IR (PR #149062)

Tue Aug 19 02:12:16 PDT 2025

llvm-ci wrote:

LLVM Buildbot has detected a new failure on builder `ml-opt-devrel-x86-64` running on `ml-opt-devrel-x86-64-b1` while building `llvm` at step 5 "build-unified-tree".

Full details are available at: https://lab.llvm.org/buildbot/#/builders/175/builds/23684

<details>
<summary>Here is the relevant piece of the build log for the reference</summary>

```
Step 5 (build-unified-tree) failure: build (failure)
...
251.228 [911/64/2959] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/R600RegisterInfo.cpp.o
251.435 [910/64/2960] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/R600Subtarget.cpp.o
254.161 [909/64/2961] Building CXX object lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/GISel/AArch64PreLegalizerCombiner.cpp.o
256.733 [908/64/2962] Building CXX object lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/AArch64TargetObjectFile.cpp.o
257.168 [907/64/2963] Building CXX object lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/AArch64StackTaggingPreRA.cpp.o
257.664 [906/64/2964] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/SIFixSGPRCopies.cpp.o
258.095 [905/64/2965] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/SIFixVGPRCopies.cpp.o
258.174 [904/64/2966] Building CXX object lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/AArch64FrameLowering.cpp.o
259.381 [903/64/2967] Building CXX object lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/SMEPeepholeOpt.cpp.o
259.661 [902/64/2968] Building CXX object lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/MachineSMEABIPass.cpp.o
FAILED: lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/MachineSMEABIPass.cpp.o 
ccache /usr/bin/c++ -DCPUINFO_SUPPORTED_PLATFORM=1 -DGTEST_HAS_RTTI=0 -D_DEBUG -D_GLIBCXX_ASSERTIONS -D_GNU_SOURCE -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -D__STDC_LIMIT_MACROS -I/b/ml-opt-devrel-x86-64-b1/build/lib/Target/AArch64 -I/b/ml-opt-devrel-x86-64-b1/llvm-project/llvm/lib/Target/AArch64 -I/var/lib/buildbot/.local/lib/python3.7/site-packages/tensorflow/include -I/b/ml-opt-devrel-x86-64-b1/build/include -I/b/ml-opt-devrel-x86-64-b1/llvm-project/llvm/include -isystem /tmp/tflitebuild/tensorflow/include -isystem /tmp/tflitebuild/eigen/include/eigen3 -isystem /tmp/tflitebuild/abseil-cpp/include -isystem /tmp/tflitebuild/flatbuffers/include -isystem /tmp/tflitebuild/gemmlowp/include/gemmlowp -isystem /tmp/tflitebuild/ml_dtypes/src/ml_dtypes -isystem /tmp/tflitebuild/ml_dtypes/src/ml_dtypes/ml_dtypes -isystem /tmp/tflitebuild/ruy/include -isystem /tmp/tflitebuild/cpuinfo/include -isystem /tmp/tflitebuild/ARM_NEON_2_x86_SSE/include -fPIC -fno-semantic-interposition -fvisibility-inlines-hidden -Werror=date-time -Wall -Wextra -Wno-unused-parameter -Wwrite-strings -Wcast-qual -Wno-missing-field-initializers -pedantic -Wno-long-long -Wimplicit-fallthrough -Wno-uninitialized -Wno-nonnull -Wno-class-memaccess -Wno-redundant-move -Wno-pessimizing-move -Wno-noexcept-type -Wdelete-non-virtual-dtor -Wsuggest-override -Wno-comment -Wno-misleading-indentation -Wctad-maybe-unsupported -fdiagnostics-color -ffunction-sections -fdata-sections -O3 -DNDEBUG -fvisibility=hidden  -fno-exceptions -funwind-tables -fno-rtti -UNDEBUG -DEIGEN_NEON_GEBP_NR=4 -DTFL_STATIC_LIBRARY_BUILD -std=c++17 -MD -MT lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/MachineSMEABIPass.cpp.o -MF lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/MachineSMEABIPass.cpp.o.d -o lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/MachineSMEABIPass.cpp.o -c /b/ml-opt-devrel-x86-64-b1/llvm-project/llvm/lib/Target/AArch64/MachineSMEABIPass.cpp
/b/ml-opt-devrel-x86-64-b1/llvm-project/llvm/lib/Target/AArch64/MachineSMEABIPass.cpp:255: error: declaration of ‘llvm::EdgeBundles* {anonymous}::MachineSMEABI::EdgeBundles’ changes meaning of ‘EdgeBundles’ [-fpermissive]
  255 |   EdgeBundles *EdgeBundles = nullptr;
      | 
In file included from /b/ml-opt-devrel-x86-64-b1/llvm-project/llvm/lib/Target/AArch64/MachineSMEABIPass.cpp:62:
/b/ml-opt-devrel-x86-64-b1/llvm-project/llvm/include/llvm/CodeGen/EdgeBundles.h:27: note: ‘EdgeBundles’ declared here as ‘class llvm::EdgeBundles’
   27 | class EdgeBundles {
      | 
260.267 [902/63/2969] Building CXX object lib/CodeGen/AsmPrinter/CMakeFiles/LLVMAsmPrinter.dir/AsmPrinter.cpp.o
262.569 [902/62/2970] Building CXX object lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/AArch64RegisterInfo.cpp.o
265.945 [902/61/2971] Building CXX object lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/AArch64FastISel.cpp.o
266.620 [902/60/2972] Building CXX object lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/AArch64InstrInfo.cpp.o
269.825 [902/59/2973] Building CXX object lib/LTO/CMakeFiles/LLVMLTO.dir/LTO.cpp.o
271.206 [902/58/2974] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUUnifyDivergentExitNodes.cpp.o
274.356 [902/57/2975] Building CXX object lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/AArch64Subtarget.cpp.o
274.783 [902/56/2976] Building CXX object lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/AArch64StackTagging.cpp.o
280.846 [902/55/2977] Building CXX object lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/AArch64ISelDAGToDAG.cpp.o
282.878 [902/54/2978] Building CXX object lib/CodeGen/CMakeFiles/LLVMCodeGen.dir/CodeGenPrepare.cpp.o
288.100 [902/53/2979] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUAlwaysInlinePass.cpp.o
296.140 [902/52/2980] Building CXX object lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/AArch64TargetMachine.cpp.o
297.900 [902/51/2981] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUInstCombineIntrinsic.cpp.o
298.138 [902/50/2982] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUImageIntrinsicOptimizer.cpp.o
298.688 [902/49/2983] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPULateCodeGenPrepare.cpp.o
299.409 [902/48/2984] Building CXX object lib/Target/AArch64/CMakeFiles/LLVMAArch64CodeGen.dir/GISel/AArch64InstructionSelector.cpp.o
299.686 [902/47/2985] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUPrepareAGPRAlloc.cpp.o
300.618 [902/46/2986] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUCallLowering.cpp.o
301.558 [902/45/2987] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUHSAMetadataStreamer.cpp.o
302.350 [902/44/2988] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUMIRFormatter.cpp.o
302.737 [902/43/2989] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUCodeGenPrepare.cpp.o
/b/ml-opt-devrel-x86-64-b1/llvm-project/llvm/lib/Target/AMDGPU/AMDGPUCodeGenPrepare.cpp:312: warning: ‘bool {anonymous}::AMDGPUCodeGenPrepareImpl::isSigned(const llvm::SelectInst&) const’ defined but not used [-Wunused-function]
  312 | bool AMDGPUCodeGenPrepareImpl::isSigned(const SelectInst &I) const {
      | 
/b/ml-opt-devrel-x86-64-b1/llvm-project/llvm/lib/Target/AMDGPU/AMDGPUCodeGenPrepare.cpp:307: warning: ‘bool {anonymous}::AMDGPUCodeGenPrepareImpl::isSigned(const llvm::BinaryOperator&) const’ defined but not used [-Wunused-function]
  307 | bool AMDGPUCodeGenPrepareImpl::isSigned(const BinaryOperator &I) const {
      | 
302.885 [902/42/2990] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUAsmPrinter.cpp.o
303.490 [902/41/2991] Building CXX object lib/CodeGen/SelectionDAG/CMakeFiles/LLVMSelectionDAG.dir/SelectionDAGBuilder.cpp.o
303.871 [902/40/2992] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUPerfHintAnalysis.cpp.o

```

</details>

https://github.com/llvm/llvm-project/pull/149062