ojhunt wrote: > > > I'm concerned about this - I initially thought this was for the purpose > > > of the structure field protection in the frontend, but this is modifying > > > the actual pointer auth intrinsics in the backend which is very > > > concerning given the work we need to do to merging and protection etc. > > > > > > Not sure I understand your concerns, can you be more specific? > > I'm concerned about the interaction of these changes with ptrauth intrinsic > optimizations, and the ability for attackers to gain control of the > enablement flags. > > But that said, this is a backend change so @ahmedbougacha should be the main > reviewer.
I have checked in with @ahmedbougacha and his feeling is that this is fine as it requires a bunch of work to opt in, and for places where the security is important enough that we don't want people using this it's easy enough to block. https://github.com/llvm/llvm-project/pull/133537 _______________________________________________ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
