[Mlir-commits] [mlir] [mlir][arith] Add constraints to the MulIOp for preventing type mismatch while folding (PR #136093)

[Mehdi Amini]

joker-eph wrote:

> I would like to understand why we proceeded with this isolated workaround 

Because the underlying issue seems hard to fix fundamentally?

> without explaining what the underlying issue is

What kind of explanation are you looking for other than "some constant returns an attribute that does not match the SSA value type"?

> referencing previous investigations

There is a bug where we tried to look at `llvm.mlir.constant` behavior, as mentioned above, I can't find it (I must be bad at GitHub search, or the tool isn't very good...).



https://github.com/llvm/llvm-project/pull/136093