[all-commits] [llvm/llvm-project] 3a1ae2: [mlir][vector] Fix invalid `LoadOp` indices being ...

[Rik Huijzer via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 3a1ae2f46db473cfde4baa6e1b090f5dae67e8db
      https://github.com/llvm/llvm-project/commit/3a1ae2f46db473cfde4baa6e1b090f5dae67e8db
  Author: Rik Huijzer <github at huijzer.xyz>
  Date:   2023-12-17 (Sun, 17 Dec 2023)

  Changed paths:
    M mlir/lib/Conversion/VectorToSCF/VectorToSCF.cpp
    M mlir/lib/Dialect/MemRef/IR/MemRefOps.cpp
    M mlir/test/Conversion/VectorToSCF/vector-to-scf.mlir
    M mlir/test/Dialect/MemRef/invalid.mlir

  Log Message:
  -----------
  [mlir][vector] Fix invalid `LoadOp` indices being created (#75519)

Fixes https://github.com/llvm/llvm-project/issues/71326.

The cause of the issue was that a new `LoadOp` was created which looked
something like:
```mlir
%arg4 = 
func.func main(%arg1 : index, %arg2 : index) {
  %alloca_0 = memref.alloca() : memref<vector<1x32xi1>>
  %1 = vector.type_cast %alloca_0 : memref<vector<1x32xi1>> to memref<1xvector<32xi1>>
  %2 = memref.load %1[%arg1, %arg2] : memref<1xvector<32xi1>>
  return
}
```
which crashed inside the `LoadOp::verify`. Note here that `%alloca_0` is
0 dimensional, `%1` has one dimension, but `memref.load` tries to index
`%1` with two indices.

This is now fixed by using the fact that `unpackOneDim` always unpacks
one dim


https://github.com/llvm/llvm-project/blob/1bce61e6b01b38e04260be4f422bbae59c34c766/mlir/lib/Conversion/VectorToSCF/VectorToSCF.cpp#L897-L903

and so the `loadOp` should just index only one dimension.

---------

Co-authored-by: Benjamin Maxwell <macdue at dueutil.tech>