1 files changed, 29 insertions, 8 deletions

diff --git a/llvm/lib/Target/AMDGPU/GCNDPPCombine.cpp b/llvm/lib/Target/AMDGPU/GCNDPPCombine.cpp
index 10e2c3a263f17..719a968b83147 100644
--- a/llvm/lib/Target/AMDGPU/GCNDPPCombine.cpp
+++ b/llvm/lib/Target/AMDGPU/GCNDPPCombine.cpp
@@ -105,6 +105,11 @@ public:
     MachineFunctionPass::getAnalysisUsage(AU);
   }
 
+  MachineFunctionProperties getRequiredProperties() const override {
+    return MachineFunctionProperties()
+      .set(MachineFunctionProperties::Property::IsSSA);
+  }
+
 private:
   int getDPPOp(unsigned Op) const;
 };
@@ -168,7 +173,9 @@ MachineInstr *GCNDPPCombine::createDPPInst(MachineInstr &OrigMI,
   }
 
   auto DPPInst = BuildMI(*OrigMI.getParent(), OrigMI,
-                         OrigMI.getDebugLoc(), TII->get(DPPOp));
+                         OrigMI.getDebugLoc(), TII->get(DPPOp))
+    .setMIFlags(OrigMI.getFlags());
+
   bool Fail = false;
   do {
     auto *Dst = TII->getNamedOperand(OrigMI, AMDGPU::OpName::vdst);
@@ -506,15 +513,32 @@ bool GCNDPPCombine::combineDPPMov(MachineInstr &MovMI) const {
       break;
     }
 
+    auto *Src0 = TII->getNamedOperand(OrigMI, AMDGPU::OpName::src0);
+    auto *Src1 = TII->getNamedOperand(OrigMI, AMDGPU::OpName::src1);
+    if (Use != Src0 && !(Use == Src1 && OrigMI.isCommutable())) { // [1]
+      LLVM_DEBUG(dbgs() << "  failed: no suitable operands\n");
+      break;
+    }
+
+    assert(Src0 && "Src1 without Src0?");
+    if (Src1 && Src1->isIdenticalTo(*Src0)) {
+      assert(Src1->isReg());
+      LLVM_DEBUG(
+          dbgs()
+          << "  " << OrigMI
+          << "  failed: DPP register is used more than once per instruction\n");
+      break;
+    }
+
     LLVM_DEBUG(dbgs() << "  combining: " << OrigMI);
-    if (Use == TII->getNamedOperand(OrigMI, AMDGPU::OpName::src0)) {
+    if (Use == Src0) {
       if (auto *DPPInst = createDPPInst(OrigMI, MovMI, CombOldVGPR,
                                         OldOpndValue, CombBCZ)) {
         DPPMIs.push_back(DPPInst);
         Rollback = false;
       }
-    } else if (OrigMI.isCommutable() &&
-               Use == TII->getNamedOperand(OrigMI, AMDGPU::OpName::src1)) {
+    } else {
+      assert(Use == Src1 && OrigMI.isCommutable()); // by check [1]
       auto *BB = OrigMI.getParent();
       auto *NewMI = BB->getParent()->CloneMachineInstr(&OrigMI);
       BB->insert(OrigMI, NewMI);
@@ -528,8 +552,7 @@ bool GCNDPPCombine::combineDPPMov(MachineInstr &MovMI) const {
       } else
         LLVM_DEBUG(dbgs() << "  failed: cannot be commuted\n");
       NewMI->eraseFromParent();
-    } else
-      LLVM_DEBUG(dbgs() << "  failed: no suitable operands\n");
+    }
     if (Rollback)
       break;
     OrigMIs.push_back(&OrigMI);
@@ -562,8 +585,6 @@ bool GCNDPPCombine::runOnMachineFunction(MachineFunction &MF) {
   MRI = &MF.getRegInfo();
   TII = ST.getInstrInfo();
 
-  assert(MRI->isSSA() && "Must be run on SSA");
-
   bool Changed = false;
   for (auto &MBB : MF) {
     for (auto I = MBB.rbegin(), E = MBB.rend(); I != E;) {