[Mlir-commits] [mlir] [mlir][scf] support front peeling for unsigned loop (PR #233238)

Matthias Springer llvmlistbot at llvm.org
Tue Oct 20 17:01:29 PDT 2026


matthias-springer wrote:

> while testing the peeling i noticed that the scf.for verifier doesn’t reject a constant negative step.

That's expected. The step is an SSA value and the verifier is not supposed to look up the defining op of an SSA value ([guidelines](https://mlir.llvm.org/getting_started/DeveloperGuide/#ir-verifier)).

https://github.com/llvm/llvm-project/pull/233238


More information about the Mlir-commits mailing list