martinuy wrote: Thanks @kbeyls for having a look at this proposal. You raised a valid concern: a developer that is not explicitly adding the `__ptrauth` qualifier to their source code and is targeting an architecture that supports pointer authentication might be caught by surprise with this new warning.
My initial impression was that the [Pointer Authentication](https://clang.llvm.org/docs/PointerAuthentication.html) doc offers some clues. In particular, section _C function pointers_ says `On arm64e, C function pointers are currently signed with the IA key without address diversity and with a constant discriminator of 0. (...) The use of a uniform constant discriminator greatly simplifies the adoption of arm64e, but it is a significant weakness in the mitigation because it allows any C function pointer to be replaced with another.` @ojhunt what do you think of adding a reference to the new warning in that section so we make it easier for developers to connects the dots (e.g. when they search for the warning directly in the document or though a search engine)? https://github.com/llvm/llvm-project/pull/157779 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
