[llvm] [AMDGPU] Introduce custom MIR formatting for s_wait_alu (PR #176316)

[via llvm-commits]

================
@@ -177,7 +177,14 @@ inline unsigned getVaSsrcBitWidth() { return 1; }
 inline unsigned getVaSsrcBitShift() { return 8; }
 
 /// \returns HoldCnt bit shift
-inline unsigned getHoldCntWidth() { return 1; }
+inline unsigned getHoldCntWidth(unsigned VersionMajor, unsigned VersionMinor) {
+  static constexpr const unsigned MinMajor = 10;
+  static constexpr const unsigned MinMinor = 3;
+  return (VersionMajor == MinMajor && VersionMinor >= MinMinor) ||
----------------
vporpo wrote:

Yes that makes it much more readable too.
Done.

https://github.com/llvm/llvm-project/pull/176316