Index: llvm/trunk/lib/Target/AMDGPU/AMDGPUISelDAGToDAG.cpp
===================================================================
--- llvm/trunk/lib/Target/AMDGPU/AMDGPUISelDAGToDAG.cpp
+++ llvm/trunk/lib/Target/AMDGPU/AMDGPUISelDAGToDAG.cpp
@@ -876,7 +876,7 @@
                                      Zero, Addr.getOperand(1));
 
           Base = SDValue(MachineSub, 0);
-          Offset = Addr.getOperand(0);
+          Offset = CurDAG->getTargetConstant(ByteOffset, DL, MVT::i16);
           return true;
         }
       }
@@ -894,7 +894,7 @@
       MachineSDNode *MovZero = CurDAG->getMachineNode(AMDGPU::V_MOV_B32_e32,
                                  DL, MVT::i32, Zero);
       Base = SDValue(MovZero, 0);
-      Offset = Addr;
+      Offset = CurDAG->getTargetConstant(CAddr->getZExtValue(), DL, MVT::i16);
       return true;
     }
   }
Index: llvm/trunk/test/CodeGen/AMDGPU/llvm.amdgcn.ds.bpermute.ll
===================================================================
--- llvm/trunk/test/CodeGen/AMDGPU/llvm.amdgcn.ds.bpermute.ll
+++ llvm/trunk/test/CodeGen/AMDGPU/llvm.amdgcn.ds.bpermute.ll
@@ -21,4 +21,13 @@
   ret void
 }
 
+; CHECK-LABEL: {{^}}ds_bpermute_imm_index:
+; CHECK: ds_bpermute_b32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} offset:64
+; CHECK: s_waitcnt lgkmcnt
+define void @ds_bpermute_imm_index(i32 addrspace(1)* %out, i32 %base_index, i32 %src) nounwind {
+  %bpermute = call i32 @llvm.amdgcn.ds.bpermute(i32 64, i32 %src) #0
+  store i32 %bpermute, i32 addrspace(1)* %out, align 4
+  ret void
+}
+
 attributes #0 = { nounwind readnone convergent }