Meinersbur wrote: @tblah I don't think it is blocking cancellation, we just don't know what is going on. The `omp.par.outlined.exit` apparently was once need for cancellation points. Looking at the `cancel_codegen.cpp` test case, it seems to evantually jump to `omp.par.exit.exitStub`. That one is created by CodeExtractor, which is supposed to create a return value in this but doesn't for unknown reasons. Finding what change "fixed" it might help, but I also don't think it is worth spending time one finding it. Investigation will need to find out details once somebody is going to make cancellation work (again?).
https://github.com/llvm/llvm-project/pull/129872 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
