efriedma-quic wrote: The performance isn't a big deal, I guess. Maybe we can mitigate the performance a bit by being a bit smarter about when we do the walk: the map becomes a lot smaller if we fill it before each statement, then clear it after each statement, instead of computing it for the whole function at once.
Can you put the recursive visit into clang/lib/CodeGenUtils/? https://github.com/llvm/llvm-project/pull/218775 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
