[Mlir-commits] [mlir] [MLIR] Add more ops support for flattening memref operands (PR #159841)

[Krzysztof Drewniak]

krzysz00 wrote:

Also, general notes
- We should be using `affine.linearize_index` where possible (this is part of the motivation for the contiguous layout stuff I've been fixing to get around to)
- Arithmetic ops should have `nsw`s on them

https://github.com/llvm/llvm-project/pull/159841