ojhunt wrote: > Not opposed but this is a pretty significant amount of change for being this > late in the rc cycles, and the changes haven't been upstream for very long. > How risky are these changes?
This is the upstreaming of code we've had deployed for a few years at this point, the only changes are style improvements and the brief period where we thought we could drop `LangOpts.PointerAuthBlockDescriptorPointers` from the upstreaming, before remembering that that's only ABI on a subset of our user-space environments. So I do recognize it is a much bigger change than would be ideal at this point, but it is self contained, and only actually applies to one family of platforms, which have significant amounts of deployment experience. e.g even if there is an error in it, it only impacts us, and not having it is even worse as instead of incorrect edge cases (because if there were non-edge case problems we have a number of tests that would presumably detect it) it would be a complete ABI mismatch. https://github.com/llvm/llvm-project/pull/153725 _______________________________________________ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
