[Mlir-commits] [mlir] [mlir] [linalg] fix failure on specializing matmul with permuted loops (PR #184294)

[Adam Siemieniuk]

adam-smnk wrote:

Wording is a bit tricky but it's about `iterator_types` i.e., order of the nested loops.
So while existing validation checks RHS of the indexing maps, this is about their LHS.

Overall LGTM, clearly needed addition.
I still wonder if these new maps could be constructed through a more generic way using affine map composition. But not sure if that's easily done to the bound LHS dims set. Manual construction as proposed might be the simplest.

https://github.com/llvm/llvm-project/pull/184294