[PATCH] D89768: [Coroutine] Properly determine whether an alloca should live on the frame

[JunMa via Phabricator via llvm-commits]

junparser added a comment.

Thanks for the patch! I think I generally agree with this patch. 
One thing is that the aliases seems to be over-analyzed. Can we just leave aliases on frame? Cause I have no idea about this effect on real workloads.



================
Comment at: llvm/lib/Transforms/Coroutines/CoroFrame.cpp:919-921
+    for (auto *BB1 : UserBBs)
+      for (auto *BB2 : UserBBs)
+        if (Checker.hasPathCrossingSuspendPoint(BB1, BB2))
----------------
this function implies dominate relation for arg1 and arg2, so we can not use hasPathCrossingSuspendPoint for two user basic blocks directly. 


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D89768/new/

https://reviews.llvm.org/D89768