[llvm] [AMDGPU] Refine operand iterators in the SIInsertWaitcnts. NFCI. (PR #108884)

[Jay Foad via llvm-commits]

================
@@ -891,12 +868,9 @@ void WaitcntBrackets::updateByEvent(const SIInstrInfo *TII,
           }
         }
       }
-      for (unsigned I = 0, E = Inst.getNumOperands(); I != E; ++I) {
-        MachineOperand &MO = Inst.getOperand(I);
-        if (MO.isReg() && !MO.isDef() &&
-            TRI->isVectorRegister(*MRI, MO.getReg())) {
-          setExpScore(&Inst, TII, TRI, MRI, I, CurrScore);
-        }
+      for (const MachineOperand &Op : Inst.uses()) {
----------------
jayfoad wrote:

Looks like the old code also handled implicit uses, so this should be `Inst.all_uses()`?

https://github.com/llvm/llvm-project/pull/108884