[PATCH] D136918: [AMDGPU] Scheduler: fix RP calculation for a MBB with one successor

Fri Oct 28 10:20:07 PDT 2022

vpykhtin abandoned this revision.
vpykhtin added a comment.

In D136918#3892369 <https://reviews.llvm.org/D136918#3892369>, @foad wrote:

> In D136918#3892323 <https://reviews.llvm.org/D136918#3892323>, @vpykhtin wrote:
>
>> In D136918#3891234 <https://reviews.llvm.org/D136918#3891234>, @foad wrote:
>>
>>> I don't understand this. Why would the liveouts of A and B be different from the liveins of C?
>>
>> You're right - live-outs of A //and// B are live-ins to C, but in this case we're trying to use live-outs of only A or only B as live-ins to C.
>
> No, I mean that the live-outs of A should be identical to the live-outs of B. And both should be identical to the live-ins of C. That's how liveness works.

Agree, this patch hides other bug.

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D136918/new/

https://reviews.llvm.org/D136918