[llvm] [AMDGPU][NFC] Minor source cleanups in SIInsertWaitcnts (PR #181095)

[Jay Foad via llvm-commits]

================
@@ -1002,6 +1002,15 @@ class WaitcntBrackets {
 
   // State of all counters at each async mark encountered so far.
   SmallVector<CounterValueArray> AsyncMarks;
+
+  // For each backedge in isolation, the algorithm reachs a fixed point after
+  // the first call to merge(). This is unchanged even with the AsyncMarks
----------------
jayfoad wrote:

I don't understand this. Are you saying that on a simple function with a simple loop with a single backedge, the fixed point iteration scheme in this pass would needs to iterate more than twice? I'm not sure that's true.

https://github.com/llvm/llvm-project/pull/181095