[Mlir-commits] [mlir] [Tosa] : Fix integer overflow for computing intmax+1 in tosa.cast to linalg. (PR #112455)

[Georgios Pinitas]

GeorgeARM wrote:

> So it's possible to add a verifier to error out for such tosa ops instead of producing incorrect code. However, I chose to fix the overflow issue to still be able to represent the deeplabv3 model with tosa ops in the above-mentioned pipeline. Open to suggestions if adding the verifier is more appropriate instead.

@sahas3 IMHO I think is better to fix; my expectation is that at some point TOSA will get `i64` support.
Moreover, as you mentioned is good to get models legalizing rather than explicitly restricting. 


https://github.com/llvm/llvm-project/pull/112455