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

[Renato Golin]

rengolin wrote:

> 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.

Still worried we're forcing the maps to be in a canonical shape without verifying that that was the case in the original generic. Transpose, broadcast, weird access.

Also thinking that, if a generic has "non-canonical" loop order, then maybe it's done on purpose? Some other compiler pass is doing this to optimize memory access, and then we go on and revert that change? So, perhaps, if the `generic` has non-canonical order, we shouldn't convert into a `matmul`?

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