1 files changed, 24 insertions, 2 deletions
diff --git a/llvm/lib/CodeGen/InlineSpiller.cpp b/llvm/lib/CodeGen/InlineSpiller.cpp
index b7ab404070b..68f2a2ed9a1 100644
--- a/llvm/lib/CodeGen/InlineSpiller.cpp
+++ b/llvm/lib/CodeGen/InlineSpiller.cpp
@@ -857,14 +857,36 @@ void InlineSpiller::insertReload(unsigned NewVReg,
   ++NumReloads;
 }
 
+/// Check if \p Def fully defines a VReg with an undefined value.
+/// If that's the case, that means the value of VReg is actually
+/// not relevant.
+static bool isFullUndefDef(const MachineInstr &Def) {
+  if (!Def.isImplicitDef())
+    return false;
+  assert(Def.getNumOperands() == 1 &&
+         "Implicit def with more than one definition");
+  // We can say that the VReg defined by Def is undef, only if it is
+  // fully defined by Def. Otherwise, some of the lanes may not be
+  // undef and the value of the VReg matters.
+  return !Def.getOperand(0).getSubReg();
+}
+
 /// insertSpill - Insert a spill of NewVReg after MI.
 void InlineSpiller::insertSpill(unsigned NewVReg, bool isKill,
                                  MachineBasicBlock::iterator MI) {
   MachineBasicBlock &MBB = *MI->getParent();
 
   MachineInstrSpan MIS(MI);
-  TII.storeRegToStackSlot(MBB, std::next(MI), NewVReg, isKill, StackSlot,
-                          MRI.getRegClass(NewVReg), &TRI);
+  if (isFullUndefDef(*MI))
+    // Don't spill undef value.
+    // Anything works for undef, in particular keeping the memory
+    // uninitialized is a viable option and it saves code size and
+    // run time.
+    BuildMI(MBB, std::next(MI), MI->getDebugLoc(), TII.get(TargetOpcode::KILL))
+        .addReg(NewVReg, getKillRegState(isKill));
+  else
+    TII.storeRegToStackSlot(MBB, std::next(MI), NewVReg, isKill, StackSlot,
+                            MRI.getRegClass(NewVReg), &TRI);
 
   LIS.InsertMachineInstrRangeInMaps(std::next(MI), MIS.end());