Index: llvm/trunk/lib/Target/AMDGPU/SILowerControlFlow.cpp
===================================================================
--- llvm/trunk/lib/Target/AMDGPU/SILowerControlFlow.cpp
+++ llvm/trunk/lib/Target/AMDGPU/SILowerControlFlow.cpp
@@ -141,8 +141,9 @@
 
   // Add an implicit def of exec to discourage scheduling VALU after this which
   // will interfere with trying to form s_and_saveexec_b64 later.
+  unsigned CopyReg = MRI->createVirtualRegister(&AMDGPU::SReg_64RegClass);
   MachineInstr *CopyExec =
-    BuildMI(MBB, I, DL, TII->get(AMDGPU::COPY), SaveExecReg)
+    BuildMI(MBB, I, DL, TII->get(AMDGPU::COPY), CopyReg)
     .addReg(AMDGPU::EXEC)
     .addReg(AMDGPU::EXEC, RegState::ImplicitDefine);
 
@@ -150,7 +151,7 @@
 
   MachineInstr *And =
     BuildMI(MBB, I, DL, TII->get(AMDGPU::S_AND_B64), Tmp)
-    .addReg(SaveExecReg)
+    .addReg(CopyReg)
     //.addReg(AMDGPU::EXEC)
     .addReg(Cond.getReg());
   setImpSCCDefDead(*And, true);
@@ -158,7 +159,7 @@
   MachineInstr *Xor =
     BuildMI(MBB, I, DL, TII->get(AMDGPU::S_XOR_B64), SaveExecReg)
     .addReg(Tmp)
-    .addReg(SaveExecReg);
+    .addReg(CopyReg);
   setImpSCCDefDead(*Xor, ImpDefSCC.isDead());
 
   // Use a copy that is a terminator to get correct spill code placement it with
@@ -197,6 +198,7 @@
   LIS->removeInterval(SaveExecReg);
   LIS->createAndComputeVirtRegInterval(SaveExecReg);
   LIS->createAndComputeVirtRegInterval(Tmp);
+  LIS->createAndComputeVirtRegInterval(CopyReg);
 }
 
 void SILowerControlFlow::emitElse(MachineInstr &MI) {
@@ -212,14 +214,17 @@
   // We are running before TwoAddressInstructions, and si_else's operands are
   // tied. In order to correctly tie the registers, split this into a copy of
   // the src like it does.
-  BuildMI(MBB, Start, DL, TII->get(AMDGPU::COPY), DstReg)
+  unsigned CopyReg = MRI->createVirtualRegister(&AMDGPU::SReg_64RegClass);
+  BuildMI(MBB, Start, DL, TII->get(AMDGPU::COPY), CopyReg)
     .addOperand(MI.getOperand(1)); // Saved EXEC
 
   // This must be inserted before phis and any spill code inserted before the
   // else.
+  unsigned SaveReg = ExecModified ?
+    MRI->createVirtualRegister(&AMDGPU::SReg_64RegClass) : DstReg;
   MachineInstr *OrSaveExec =
-    BuildMI(MBB, Start, DL, TII->get(AMDGPU::S_OR_SAVEEXEC_B64), DstReg)
-    .addReg(DstReg);
+    BuildMI(MBB, Start, DL, TII->get(AMDGPU::S_OR_SAVEEXEC_B64), SaveReg)
+    .addReg(CopyReg);
 
   MachineBasicBlock *DestBB = MI.getOperand(2).getMBB();
 
@@ -229,7 +234,7 @@
     MachineInstr *And =
       BuildMI(MBB, ElsePt, DL, TII->get(AMDGPU::S_AND_B64), DstReg)
       .addReg(AMDGPU::EXEC)
-      .addReg(DstReg);
+      .addReg(SaveReg);
 
     if (LIS)
       LIS->InsertMachineInstrInMaps(*And);
@@ -260,6 +265,9 @@
   // src reg is tied to dst reg.
   LIS->removeInterval(DstReg);
   LIS->createAndComputeVirtRegInterval(DstReg);
+  LIS->createAndComputeVirtRegInterval(CopyReg);
+  if (ExecModified)
+    LIS->createAndComputeVirtRegInterval(SaveReg);
 
   // Let this be recomputed.
   LIS->removeRegUnit(*MCRegUnitIterator(AMDGPU::EXEC, TRI));
Index: llvm/trunk/test/CodeGen/AMDGPU/branch-condition-and.ll
===================================================================
--- llvm/trunk/test/CodeGen/AMDGPU/branch-condition-and.ll
+++ llvm/trunk/test/CodeGen/AMDGPU/branch-condition-and.ll
@@ -14,7 +14,7 @@
 ; GCN-DAG: v_cmp_lt_f32_e32 vcc,
 ; GCN: s_and_b64 [[AND:s\[[0-9]+:[0-9]+\]]], vcc, [[OTHERCC]]
 ; GCN: s_and_saveexec_b64 [[SAVED:s\[[0-9]+:[0-9]+\]]], [[AND]]
-; GCN: s_xor_b64 [[SAVED]], exec, [[SAVED]]
+; GCN: s_xor_b64 {{s\[[0-9]+:[0-9]+\]}}, exec, [[SAVED]]
 ;
 ; TODO: The following sequence is a bug (missing s_endpgm)!
 ;
Index: llvm/trunk/test/CodeGen/AMDGPU/valu-i1.ll
===================================================================
--- llvm/trunk/test/CodeGen/AMDGPU/valu-i1.ll
+++ llvm/trunk/test/CodeGen/AMDGPU/valu-i1.ll
@@ -9,8 +9,8 @@
 
 ; waitcnt should be inserted after exec modification
 ; SI: v_cmp_lt_i32_e32 vcc, 0,
-; SI-NEXT: s_and_saveexec_b64 [[SAVE:s\[[0-9]+:[0-9]+\]]], vcc
-; SI-NEXT: s_xor_b64 [[SAVE]], exec, [[SAVE]]
+; SI-NEXT: s_and_saveexec_b64 [[SAVE1:s\[[0-9]+:[0-9]+\]]], vcc
+; SI-NEXT: s_xor_b64 [[SAVE2:s\[[0-9]+:[0-9]+\]]], exec, [[SAVE1]]
 ; SI-NEXT: s_waitcnt lgkmcnt(0)
 ; SI-NEXT: ; mask branch [[FLOW_BB:BB[0-9]+_[0-9]+]]
 ; SI-NEXT: s_cbranch_execz [[FLOW_BB]]
@@ -24,9 +24,9 @@
 
 ; v_mov should be after exec modification
 ; SI: [[FLOW_BB]]:
-; SI-NEXT: s_or_saveexec_b64 [[SAVE]], [[SAVE]]
+; SI-NEXT: s_or_saveexec_b64 [[SAVE3:s\[[0-9]+:[0-9]+\]]], [[SAVE2]]
 ; SI-NEXT: v_mov_b32_e32 v{{[0-9]+}}
-; SI-NEXT: s_xor_b64 exec, exec, [[SAVE]]
+; SI-NEXT: s_xor_b64 exec, exec, [[SAVE3]]
 ; SI-NEXT: ; mask branch
 ;
 define void @test_if(i32 %b, i32 addrspace(1)* %src, i32 addrspace(1)* %dst) #1 {
@@ -152,7 +152,7 @@
 ; SI-DAG: v_cmp_ne_u32_e32 [[NEG1_CHECK_1:vcc]], -1, [[B]]
 ; SI: s_and_b64 [[ORNEG1:s\[[0-9]+:[0-9]+\]]], [[NEG1_CHECK_1]], [[NEG1_CHECK_0]]
 ; SI: s_and_saveexec_b64 [[ORNEG2:s\[[0-9]+:[0-9]+\]]], [[ORNEG1]]
-; SI: s_xor_b64 [[ORNEG2]], exec, [[ORNEG2]]
+; SI: s_xor_b64 [[ORNEG3:s\[[0-9]+:[0-9]+\]]], exec, [[ORNEG2]]
 ; SI: s_cbranch_execz [[LABEL_FLOW:BB[0-9]+_[0-9]+]]
 
 ; SI: BB{{[0-9]+_[0-9]+}}: ; %bb20
@@ -162,8 +162,8 @@
 
 ; SI: [[LABEL_FLOW]]:
 ; SI-NEXT: ; in Loop: Header=[[LABEL_LOOP]]
-; SI-NEXT: s_or_b64 exec, exec, [[ORNEG2]]
-; SI-NEXT: s_or_b64 [[COND_STATE]], [[ORNEG2]], [[TMP]]
+; SI-NEXT: s_or_b64 exec, exec, [[ORNEG3]]
+; SI-NEXT: s_or_b64 [[COND_STATE]], [[ORNEG3]], [[TMP]]
 ; SI-NEXT: s_andn2_b64 exec, exec, [[COND_STATE]]
 ; SI-NEXT: s_cbranch_execnz [[LABEL_LOOP]]