[llvm-branch-commits] [llvm] [AMDGPU][Scheduler] Simplify scheduling revert logic (PR #177203)

[Matt Arsenault via llvm-branch-commits]

================
@@ -1773,66 +1775,56 @@ bool GCNSchedStage::mayCauseSpilling(unsigned WavesAfter) {
   return false;
 }
 
-void GCNSchedStage::revertScheduling() {
-  LLVM_DEBUG(dbgs() << "Attempting to revert scheduling.\n");
-  DAG.RegionEnd = DAG.RegionBegin;
-  int SkippedDebugInstr = 0;
-  for (MachineInstr *MI : Unsched) {
-    if (MI->isDebugInstr()) {
-      ++SkippedDebugInstr;
-      continue;
-    }
+void GCNSchedStage::modifyRegionSchedule(unsigned RegionIdx,
+                                         MachineBasicBlock *MBB,
+                                         ArrayRef<MachineInstr *> MIOrder) {
+  assert(std::distance(DAG.Regions[RegionIdx].first,
+                       DAG.Regions[RegionIdx].second) ==
+             static_cast<long>(MIOrder.size()) &&
----------------
arsenm wrote:

Not sure why you need this cast to long, don't both of these return size_t?

https://github.com/llvm/llvm-project/pull/177203