[llvm] [AMDGPU] Skip terminators when forcing emit zero flag (PR #112116)

[Jay Foad via llvm-commits]

jayfoad wrote:

> > In the test case, why does SIInsertWaitcnts want to put a wait between the two branch instructions?
> 
> Because we force emit zero via `amdgpu-waitcnt-forcezero`. That is a debug switch, but didn't work properly in some cases.

Oh, I see. That option is weird. I thought it meant "if any waitcnt is required, wait for the counter to be 0". But actually it waits for the counter to be zero after every instruction, even instructions that have nothing to do with wait counts.

https://github.com/llvm/llvm-project/pull/112116