Index: lib/Target/AMDGPU/SILowerControlFlow.cpp
===================================================================
--- lib/Target/AMDGPU/SILowerControlFlow.cpp
+++ lib/Target/AMDGPU/SILowerControlFlow.cpp
@@ -729,14 +729,13 @@
 
           break;
 
-        case AMDGPU::S_ENDPGM: {
-          if (MF.getInfo<SIMachineFunctionInfo>()->returnsVoid())
-            break;
+        case AMDGPU::SI_RETURN: {
+          assert(!MF.getInfo<SIMachineFunctionInfo>()->returnsVoid());
 
           // Graphics shaders returning non-void shouldn't contain S_ENDPGM,
           // because external bytecode will be appended at the end.
           if (BI != --MF.end() || I != MBB.getFirstTerminator()) {
-            // S_ENDPGM is not the last instruction. Add an empty block at
+            // SI_RETURN is not the last instruction. Add an empty block at
             // the end and jump there.
             if (!EmptyMBBAtEnd) {
               EmptyMBBAtEnd = MF.CreateMachineBasicBlock();
@@ -746,9 +745,8 @@
             MBB.addSuccessor(EmptyMBBAtEnd);
             BuildMI(*BI, I, MI.getDebugLoc(), TII->get(AMDGPU::S_BRANCH))
                     .addMBB(EmptyMBBAtEnd);
+            I->eraseFromParent();
           }
-
-          I->eraseFromParent();
           break;
         }
       }
Index: test/CodeGen/AMDGPU/ret_jump.ll
===================================================================
--- test/CodeGen/AMDGPU/ret_jump.ll
+++ test/CodeGen/AMDGPU/ret_jump.ll
@@ -12,10 +12,11 @@
 ; GCN-NEXT: ; mask branch [[UNREACHABLE_BB:BB[0-9]+_[0-9]+]]
 
 ; GCN: [[RET_BB]]:
-; GCN-NEXT: ; return
+; GCN-NEXT: s_branch [[FINAL_BB:BB[0-9]+_[0-9]+]]
 
 ; GCN-NEXT: [[UNREACHABLE_BB]]:
 ; GCN-NEXT: s_or_b64 exec, exec, [[XOR_EXEC]]
+; GCN-NEXT: [[FINAL_BB]]:
 ; GCN-NEXT: .Lfunc_end0
 define amdgpu_ps <{ i32, i32, i32, i32, i32, i32, i32, i32, i32, float, float, float, float, float, float, float, float, float, float, float, float, float, float }> @main([9 x <16 x i8>] addrspace(2)* byval, [17 x <16 x i8>] addrspace(2)* byval, [17 x <8 x i32>] addrspace(2)* byval, i32 addrspace(2)* byval, float inreg, i32 inreg, <2 x i32>, <2 x i32>, <2 x i32>, <3 x i32>, <2 x i32>, <2 x i32>, <2 x i32>, float, float, float, float, float, i32, i32, float, i32) #0 {
 main_body: