efriedma-quic wrote: I think if we're trying to automatically identify whether an alloca has one of two lifetimes, and one of those lifetimes isn't a subset of the other, we're guaranteed to run into trouble again, eventually. We have to pick the right lifetime 100% of the time... but without assistance from the frontend, it's impossible to write an algorithm that doesn't give up in some cases, given a pointer that escapes. So it's not really a question of "improving" the existing algorithm; we need markers in the IR, like coro_outside_frame, and the algorithm should be based on that.
https://github.com/llvm/llvm-project/pull/127653 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
