andykaylor wrote: > @darkbuck I noticed that you have pushed a commit to `main` that resolves the > regression by changing the test code. But I believe the correct way to > resolve this problem would be to update the CIRGen code as in this patch. > Could you revert that commit?
I agree. Before the change in this PR, the order of the output could vary depending on the order in which arguments were evaluated. Changing the test fixed the failure for some targets, but it would break it for others. This PR moved the code to emit arguments out of the call, so the order is now determinant (but doesn't match the updated test). https://github.com/llvm/llvm-project/pull/154346 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
