https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82951
--- Comment #3 from Marc Glisse <glisse at gcc dot gnu.org> --- (In reply to Richard Biener from comment #2) > Hmm, on RTL signs do not matter. I'd say strip the bogus cases from the > existing tests and "duplicate" them into new variants using -fwrapv? Thus > only apply -fwrapv to those cases that need it. As I mentioned with the patch, the bogus cases are quite inconvenient to isolate: there are 12 tests, the whole set is run for 12 values, and only a few tests are invalid for only a few values. So if I move them, I'll move a superset of the invalid cases (say all tests that are invalid for at least one value), and then it sounds easier to move all of them to -fwrapv (the testcase was introduced to test RTL anyway), so that's what I posted. I'll post something intermediate though since that was asked.
