[llvm] [AMDGPU][Scheduler] Scoring system for rematerialization candidates (PR #153092)

[Lucas Ramirez via llvm-commits]

================
@@ -2086,65 +2327,93 @@ void PreRARematStage::rematerialize() {
             assert((SR.LaneMask & UncoveredLanes).none());
         }
       }
+    }
 #endif
 
-      // The register is no longer a live-in in all regions but the one that
-      // contains the single use. In live-through regions, maximum register
-      // pressure decreases predictably so we can directly update it. In the
-      // using region, maximum RP may or may not decrease, so we will mark it
-      // for re-computation after all materializations have taken place.
-      LaneBitmask PrevMask = RegionLiveIns[Reg];
-      RegionLiveIns.erase(Reg);
-      RegMasks.insert({{I, Remat.RematMI->getOperand(0).getReg()}, PrevMask});
-      if (Remat.UseMI->getParent() != DAG.Regions[I].first->getParent())
-        DAG.Pressure[I].inc(Reg, PrevMask, LaneBitmask::getNone(), DAG.MRI);
-      else
-        RecomputeRP.insert(I);
+    // This save is guaranteed in regions in which the register is live-through
+    // and unused but optimistic in all other regions where the register is
+    // live.
+    RPTargets[I].saveReg(Reg, Remat.Mask, DAG.MRI);
+    DAG.LiveIns[I].erase(Reg);
+    DAG.RegionLiveOuts.getLiveRegsForRegionIdx(I).erase(Reg);
+    if (!Remat.isUnusedLiveThrough(I))
+      RecomputeRP.set(I);
+  }
+
+  DAG.deleteMI(Remat.DefRegion, &DefMI);
+  RescheduleRegions |= Remat.Live;
+}
+
+void PreRARematStage::rollback(const RollbackInfo &Rollback,
----------------
lucas-rami wrote:

I found a way to rollback without deleting the original MI by simply setting its opcode to DBG_VALUE. This makes the scheduler completely skip over the instruction and lets it handle its repositioning at its original position in case of a scheduling revert.

If the stage wants to roll back a rematerialization decision it just resets the original MI's opcode, re-uses its register, and deletes the rematerialization; if not then it just fully deletes the original MI at the end.

https://github.com/llvm/llvm-project/pull/153092