| Issue |
180309
|
| Summary |
mlir-opt crash: bufferizesToMemoryRead not implemented in sparse_tensor.new during --eliminate-empty-tensors
|
| Labels |
new issue
|
| Assignees |
|
| Reporter |
xvv4402
|
**Bug Op**: sparse_tensor.new
**Bug Pass**: --eliminate-empty-tensors
**Detailed Description**: The mlir-opt tool hits an UNREACHABLE assertion during the --eliminate-empty-tensors pass. The crash occurs because the sparse_tensor.new operation lacks an implementation for the bufferizesToMemoryRead method of the BufferizableOpInterface.
```mlir
module {
func.func @main() {
%0 = gpu.cluster_id x
%1 = shape.const_witness true
affine.if affine_set<(d0) : (d0 >= 0)>(%0) {
}
%2 = sparse_tensor.new %1 : !shape.witness to tensor<?xindex, #sparse_tensor.encoding<{ map = (d0) -> (d0 : compressed) }>>
%3 = shape.assuming %1 -> tensor<?xindex, #sparse_tensor.encoding<{ map = (d0) -> (d0 : compressed) }>> {
shape.assuming_yield %2 : tensor<?xindex, #sparse_tensor.encoding<{ map = (d0) -> (d0 : compressed) }>>
}
%4 = sparse_tensor.new %3 : tensor<?xindex, #sparse_tensor.encoding<{ map = (d0) -> (d0 : compressed) }>> to tensor<?xindex, #sparse_tensor.encoding<{ map = (d0) -> (d0 : compressed) }>>
return
}
}
```
```
mlir-opt --eliminate-empty-tensors temp.mlir
```
```
bufferizesToMemoryRead not implemented
UNREACHABLE executed at llvm-project-llvmorg-22.1.0-rc1/build/tools/mlir/include/mlir/Dialect/Bufferization/IR/BufferizableOpInterface.h.inc:1166!
PLEASE submit a bug report to https://github.com/llvm/llvm-project/issues/ and include the crash backtrace and instructions to reproduce the bug.
Stack dump:
0. Program arguments: llvmorg-22.1.0-install/bin/mlir-opt --eliminate-empty-tensors temp.mlir
#0 0x000064f5ccf65e32 llvm::sys::PrintStackTrace(llvm::raw_ostream&, int) (llvmorg-22.1.0-install/bin/mlir-opt+0x252fe32)
#1 0x000064f5ccf625af llvm::sys::RunSignalHandlers() (llvmorg-22.1.0-install/bin/mlir-opt+0x252c5af)
#2 0x000064f5ccf626fc SignalHandler(int, siginfo_t*, void*) Signals.cpp:0:0
#3 0x00007c6a8ae45330 (/lib/x86_64-linux-gnu/libc.so.6+0x45330)
#4 0x00007c6a8ae9eb2c __pthread_kill_implementation ./nptl/pthread_kill.c:44:76
#5 0x00007c6a8ae9eb2c __pthread_kill_internal ./nptl/pthread_kill.c:78:10
#6 0x00007c6a8ae9eb2c pthread_kill ./nptl/pthread_kill.c:89:10
#7 0x00007c6a8ae4527e raise ./signal/../sysdeps/posix/raise.c:27:6
#8 0x00007c6a8ae288ff abort ./stdlib/abort.c:81:7
#9 0x000064f5ccf4512e (llvmorg-22.1.0-install/bin/mlir-opt+0x250f12e)
#10 0x000064f5cdb9d262 (llvmorg-22.1.0-install/bin/mlir-opt+0x3167262)
#11 0x000064f5d32b6b94 mlir::bufferization::OneShotAnalysisState::applyOnAliases(mlir::Value, llvm::function_ref<void (mlir::Value)>) const (.constprop.3) OneShotAnalysis.cpp:0:0
#12 0x000064f5d32b6f85 wouldCreateReadAfterWriteInterference(mlir::OpOperand&, mlir::DominanceInfo const&, mlir::bufferization::OneShotAnalysisState&, bool) OneShotAnalysis.cpp:0:0
#13 0x000064f5d32b72d7 checkPreBufferizationAssumptions(mlir::Operation*, mlir::DominanceInfo const&, mlir::bufferization::OneShotAnalysisState&)::'lambda0'(mlir::bufferization::BufferizableOpInterface)::operator()(mlir::bufferization::BufferizableOpInterface) const (.isra.0) OneShotAnalysis.cpp:0:0
#14 0x000064f5d32b7511 mlir::WalkResult mlir::detail::walk<mlir::ForwardIterator>(mlir::Operation*, llvm::function_ref<mlir::WalkResult (mlir::Operation*)>, mlir::WalkOrder) (.constprop.1) OneShotAnalysis.cpp:0:0
#15 0x000064f5d32b74d0 mlir::WalkResult mlir::detail::walk<mlir::ForwardIterator>(mlir::Operation*, llvm::function_ref<mlir::WalkResult (mlir::Operation*)>, mlir::WalkOrder) (.constprop.1) OneShotAnalysis.cpp:0:0
#16 0x000064f5d32b74d0 mlir::WalkResult mlir::detail::walk<mlir::ForwardIterator>(mlir::Operation*, llvm::function_ref<mlir::WalkResult (mlir::Operation*)>, mlir::WalkOrder) (.constprop.1) OneShotAnalysis.cpp:0:0
```
![]()
_______________________________________________
llvm-bugs mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
