nikic wrote: > I'm not actively aware of any cases where the non-integrality is being used > to block invalid optimizations ... but I do know that, for example, > SeparateConstOffsetFromGEP has a mode (which I think Nvidia uses) that turns > pointer arithmetic into ptrtoint/inttoptr pairs.
This one already guards against pointer vs index size mismatch. But after looking at this, I've put up https://github.com/llvm/llvm-project/pull/151477 to drop this code entirely. This is not a reasonable thing to do, any it doesn't look like it's actually used anymore. https://github.com/llvm/llvm-project/pull/105735 _______________________________________________ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
