[Mlir-commits] [mlir] [mlir] Fix RemoveDeadRegionBranchOpSuccessorInputs producing invalid scf.for (LoopLikeOpInterface tie) (PR #216627)

Matthias Springer llvmlistbot at llvm.org
Wed Aug 19 00:14:00 PDT 2026


matthias-springer wrote:

Generally speaking, I think this PR is still useful. #188986 fixes the existing expensive-checks failures and improves compilation time, but the underlying problem is still there. If you have a pass that just uses `populateRegionBranchOpInterfaceCanonicalizationPatterns` but not the inlining pattern (`populateRegionBranchOpInterfaceInliningPattern`), you can still reproduce the failure. But you'd have to create a new test pass, probably not be worth it if you're just interested in removing XFAIL from the existing test suite. 

https://github.com/llvm/llvm-project/pull/216627


More information about the Mlir-commits mailing list