| Issue |
177825
|
| Summary |
mlir-opt --convert-arith-to-emitc crashes on vector.mask with constant mask
|
| Labels |
new issue
|
| Assignees |
|
| Reporter |
Subway2023
|
# Description
mlir-opt version: 21.1.8 and 22.1.0-rc1
Running mlir-opt with the --convert-arith-to-emitc pass on a module containing a vector.mask operation with a constant mask triggers a crash in the fold hook
This bug causes mlir-opt to crash on version 21.1.8 and to hang on 22.1.0-rc1, preventing compilation from completing.
# Reproduce
location: If the **vector.mask** is commented out, the crash no longer occurs.
```
module {
func.func @main() {
%0 = arith.constant 10 : i32
%1 = vector.vscale
%2 = vector.constant_mask [10, 8] : vector<16x8xi1>
vector.mask %2 { vector.yield } : vector<16x8xi1>
vector.print %0 : i32
return
}
}
```
```
mlir-opt --convert-arith-to-emitc test.mlir
```
```
PLEASE submit a bug report to https://github.com/llvm/llvm-project/issues/ and include the crash backtrace.
Stack dump:
0. Program arguments: /mnt/sdd1/sbw/mlir/tool/llvm-project-llvmorg-21.1.8-install/bin/mlir-opt --convert-arith-to-emitc /mnt/sdd1/sbw/mlir/data/genProgram_1224/250219_e9c4/mlir_steps/250219_e9c4_4.mlir
#0 0x0000602f99e5b612 llvm::sys::PrintStackTrace(llvm::raw_ostream&, int) (/mnt/sdd1/sbw/mlir/tool/llvm-project-llvmorg-21.1.8-install/bin/mlir-opt+0x20a1612)
#1 0x0000602f99e57d7f llvm::sys::RunSignalHandlers() (/mnt/sdd1/sbw/mlir/tool/llvm-project-llvmorg-21.1.8-install/bin/mlir-opt+0x209dd7f)
#2 0x0000602f99e57ecc SignalHandler(int, siginfo_t*, void*) Signals.cpp:0:0
#3 0x000078a76a645330 (/lib/x86_64-linux-gnu/libc.so.6+0x45330)
#4 0x0000602f9cdd37a5 bool llvm::detail::UniqueFunctionBase<bool, mlir::TypeID>::CallImpl<mlir::Op<mlir::vector::ConstantMaskOp, mlir::OpTrait::ZeroRegions, mlir::OpTrait::OneResult, mlir::OpTrait::OneTypedResult<mlir::VectorType>::Impl, mlir::OpTrait::ZeroSuccessors, mlir::OpTrait::ZeroOperands, mlir::OpTrait::OpInvariants, mlir::BytecodeOpInterface::Trait, mlir::ConditionallySpeculatable::Trait, mlir::OpTrait::AlwaysSpeculatableImplTrait, mlir::MemoryEffectOpInterface::Trait>::getHasTraitFn()::'lambda'(mlir::TypeID) const>(void*, mlir::TypeID) (/mnt/sdd1/sbw/mlir/tool/llvm-project-llvmorg-21.1.8-install/bin/mlir-opt+0x50197a5)
#5 0x0000602f9cdb6639 mlir::RegisteredOperationName::Model<mlir::vector::ConstantMaskOp>::hasTrait(mlir::TypeID) (/mnt/sdd1/sbw/mlir/tool/llvm-project-llvmorg-21.1.8-install/bin/mlir-opt+0x4ffc639)
#6 0x0000602f9d835a64 mlir::Operation::fold(llvm::SmallVectorImpl<mlir::OpFoldResult>&) (/mnt/sdd1/sbw/mlir/tool/llvm-project-llvmorg-21.1.8-install/bin/mlir-opt+0x5a7ba64)
#7 0x0000602f9d76b364 mlir::OpBuilder::tryFold(mlir::Operation*, llvm::SmallVectorImpl<mlir::Value>&, llvm::SmallVectorImpl<mlir::Operation*>*) (/mnt/sdd1/sbw/mlir/tool/llvm-project-llvmorg-21.1.8-install/bin/mlir-opt+0x59b1364)
#8 0x0000602f9d6bb411 (anonymous namespace)::OperationLegalizer::legalize(mlir::Operation*, mlir::ConversionPatternRewriter&) DialectConversion.cpp:0:0
#9 0x0000602f9d6bb906 (anonymous namespace)::OperationLegalizer::legalize(mlir::Operation*, mlir::ConversionPatternRewriter&) DialectConversion.cpp:0:0
```
![]()
_______________________________________________
llvm-bugs mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
