[llvm] [AMDGPU] Use range-based for loops (NFC) (PR #106328)

[Matt Arsenault via llvm-commits]

================
@@ -219,15 +216,10 @@ bool R600InstrInfo::readsLDSSrcReg(const MachineInstr &MI) const {
   if (!isALUInstr(MI.getOpcode())) {
     return false;
   }
-  for (MachineInstr::const_mop_iterator I = MI.operands_begin(),
-                                        E = MI.operands_end();
-       I != E; ++I) {
-    if (!I->isReg() || !I->isUse() || I->getReg().isVirtual())
-      continue;
-
-    if (R600::R600_LDS_SRC_REGRegClass.contains(I->getReg()))
+  for (const MachineOperand &MO : MI.all_uses())
+    if (!MO.getReg().isVirtual() &&
----------------
arsenm wrote:

Would be slightly better if this was an isPhysical check instead of !isVirtual

https://github.com/llvm/llvm-project/pull/106328