MaskRay added a comment. In D148665#4316310 <https://reviews.llvm.org/D148665#4316310>, @peter.smith wrote:
> My apologies for not responding. If I've got this right there are 4 related > patches: > D148573 <https://reviews.llvm.org/D148573> (approved) > D148785 <https://reviews.llvm.org/D148785> Use type hashes rather than RTTI > D148827 <https://reviews.llvm.org/D148827> support C > D148665 <https://reviews.llvm.org/D148665> (this one) > My initial impressions is that this makes -fsanitize=function look more like > -fsanitize=kcfi which makes it accessible from C and available to more > targets. Is there anything that we lose in the process of making these > changes? For example I would expect RTTI to have more information available > than a type hash, although this might not make any functional difference. > > I'll try and look over the next few days and leave some comments, apologies a > bit busy at work at the moment so I can't promise anything speedy. Thanks! `-fsanitize=function` will indeed become more like `-fsanitize=kcfi`. There is a big difference that `-fsanitize=function` instrumented code has a signature check for compatibility with object files not compiled with `-fsanitize=function` (and old implementations of `-fsanitize=function` with a difference location to place the signature). -fsanitize=kcfi doesn't have the compatibility guarantee. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D148665/new/ https://reviews.llvm.org/D148665 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
