[PATCH] D103492: [RS4GC] Treat inttoptr as base pointer

[Philip Reames via Phabricator via llvm-commits]

reames requested changes to this revision.
reames added a comment.
This revision now requires changes to proceed.

This is unsound.  If you have a test case where the optimizer is generating an inttoptr for a non-integral pointer from well defined IR, please share it so that we can fix the bug.

I'll note that I might be convinced to take something like this, but only with a clear comment describing it as explicit undefined behavior.  (i.e. not safe to rely on)  I do see an argument for robustness (or reliably crashing) in release builds though.  fatal_error might be an alternate approach.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D103492/new/

https://reviews.llvm.org/D103492