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

Tue Oct 27 07:53:22 PDT 2020

junparser added a comment.

In D89768#2355576 <https://reviews.llvm.org/D89768#2355576>, @junparser wrote:

> In D89768#2354119 <https://reviews.llvm.org/D89768#2354119>, @lxfind wrote:
>
>> In D89768#2353488 <https://reviews.llvm.org/D89768#2353488>, @junparser wrote:
>>
>>> 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.
>>
>> What do you mean by "leave aliases on frame"?
>
> Just setEscaped for pointer cast used before coro.begin,  or can we do something like  llvm::PointerMayBeCapturedBefore to check the pointer directly? I'm not sure about this.

@lxfind,  sorry for the confusing question, please ignore it. After I read about the D86859 <https://reviews.llvm.org/D86859>,  one of the suggestion is that after we find all of the aliases, we can do same thing like rewriteMaterializableInstructions

Repository:
  rG LLVM Github Monorepo

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

https://reviews.llvm.org/D89768