AaronBallman wrote: > I've switched to `checkBuiltinArgument`, but it causes an issue that I'd like > input on before I continue. > `clang/test/CodeGen/catch-alignment-assumption-ignorelist.c` tests that UBSan > doesn't validate the alignment of pointers to volatile. [Several UBSan checks > have this > exception](https://clang.llvm.org/docs/UndefinedBehaviorSanitizer.html#volatile), > but I don't understand the motivation behind it. With this change, the > `volatile`-ness of the pointer is lost when passed into > `__builtin_assume_aligned`, so the UBSan checks are generated. To _me_ this > seems like an improvement; after all, checking alignment does not dereference > the pointer, so why does it matter if it points to volatile?
CC @zygoloid @vitalybuka for opinions https://github.com/llvm/llvm-project/pull/153552 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
