[Mlir-commits] [mlir] [mlir][arith] Add more canonicalization and integration tests coverage (PR #92272)

[Jacob Yu]

pingshiyu wrote:

> Nice! Quick question, what kind of bugs did you find in the canonicalization patterns and the lowerings? Do you have issues/PR for them already?
> 
> Also, is your fuzzer/semantics open-source somewhere? It would be interesting for me to see them, to compare them with the one I currently have for arith (https://github.com/opencompl/xdsl-smt/blob/main/xdsl_smt/semantics/arith_semantics.py)

Yep! The bugs found by this method are here (some are as of now _unfixed_! :o)
https://github.com/llvm/llvm-project/issues/90296 (!!)
https://github.com/llvm/llvm-project/issues/90238 (!!)
https://github.com/llvm/llvm-project/issues/82788 (!!)
https://github.com/llvm/llvm-project/issues/84986 (!!)
https://github.com/llvm/llvm-project/issues/83079 
https://github.com/llvm/llvm-project/issues/88732
https://github.com/llvm/llvm-project/issues/89382

The fuzzer/semantics are unfortunately not open source at the moment, but I do plan on making it so in the near future! There's a bit more work I plan to work on before I publish it :)

https://github.com/llvm/llvm-project/pull/92272