[PATCH] D101980: [RFC] [Coroutines] Put the contents of byval argument to the frame

Tue May 18 18:51:02 PDT 2021

ChuanqiXu added a comment.

In D101980#2767416 <https://reviews.llvm.org/D101980#2767416>, @lxfind wrote:

> In D101980#2758843 <https://reviews.llvm.org/D101980#2758843>, @ChuanqiXu wrote:
>
>> In D101980#2758780 <https://reviews.llvm.org/D101980#2758780>, @lxfind wrote:
>>
>>> Something like this: D102465 <https://reviews.llvm.org/D102465>. Let me know what you think.
>>
>> Oh it is tricky. Since we add a new intrinsic use for the alloca and the intrinsic can't be analyzed by other passes. So the alloca we want wouldn't be optimized. The overall ideas looks good. For the details, I would like to emit the intrinsic for the arguments who are passed by value only. I think we could dive into it further.
>
> by the way, I could also proceed with D102465 <https://reviews.llvm.org/D102465> first while you figure out how to optimize it, if it's going to take a while. The sooner we fix the bug the better I think

I think D102465 <https://reviews.llvm.org/D102465> may be better than this patch to fix bug 48857. We could move on that one. Do you mean that it would take some time not to emit the intrinsic for who are not passed by value? Do you suggest that we should fix the bug first by mark all the arguments then we try to remove the unneeded marks in successive patch? If yes, I think we could try to look at it for about one week to find solutions. If we can't find, I am OK to mark all the arguments as a temporary fix. After all, correctness is most important.

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D101980/new/

https://reviews.llvm.org/D101980