Index: llvm/trunk/lib/Target/AMDGPU/SIInstrInfo.cpp
===================================================================
--- llvm/trunk/lib/Target/AMDGPU/SIInstrInfo.cpp
+++ llvm/trunk/lib/Target/AMDGPU/SIInstrInfo.cpp
@@ -3210,6 +3210,8 @@
     return AMDGPU::V_SUB_I32_e32;
   case AMDGPU::S_SUBB_U32: return AMDGPU::V_SUBB_U32_e32;
   case AMDGPU::S_MUL_I32: return AMDGPU::V_MUL_LO_I32;
+  case AMDGPU::S_MUL_HI_U32: return AMDGPU::V_MUL_HI_U32;
+  case AMDGPU::S_MUL_HI_I32: return AMDGPU::V_MUL_HI_I32;
   case AMDGPU::S_AND_B32: return AMDGPU::V_AND_B32_e64;
   case AMDGPU::S_OR_B32: return AMDGPU::V_OR_B32_e64;
   case AMDGPU::S_XOR_B32: return AMDGPU::V_XOR_B32_e64;
@@ -3254,6 +3256,8 @@
   case AMDGPU::S_CBRANCH_SCC0: return AMDGPU::S_CBRANCH_VCCZ;
   case AMDGPU::S_CBRANCH_SCC1: return AMDGPU::S_CBRANCH_VCCNZ;
   }
+  llvm_unreachable(
+      "Unexpected scalar opcode without corresponding vector one!");
 }
 
 const TargetRegisterClass *SIInstrInfo::getOpRegClass(const MachineInstr &MI,
Index: llvm/trunk/lib/Target/AMDGPU/SOPInstructions.td
===================================================================
--- llvm/trunk/lib/Target/AMDGPU/SOPInstructions.td
+++ llvm/trunk/lib/Target/AMDGPU/SOPInstructions.td
@@ -548,8 +548,12 @@
     def S_LSHL4_ADD_U32 : SOP2_32<"s_lshl4_add_u32">;
   } // End Defs = [SCC]
 
-  def S_MUL_HI_U32 : SOP2_32<"s_mul_hi_u32">;
-  def S_MUL_HI_I32 : SOP2_32<"s_mul_hi_i32">;
+  let isCommutable = 1 in {
+    def S_MUL_HI_U32 : SOP2_32<"s_mul_hi_u32",
+      [(set i32:$sdst, (UniformBinFrag<mulhu> SSrc_b32:$src0, SSrc_b32:$src1))]>;
+    def S_MUL_HI_I32 : SOP2_32<"s_mul_hi_i32",
+      [(set i32:$sdst, (UniformBinFrag<mulhs> SSrc_b32:$src0, SSrc_b32:$src1))]>;
+  }
 }
 
 //===----------------------------------------------------------------------===//
Index: llvm/trunk/test/CodeGen/AMDGPU/mul.ll
===================================================================
--- llvm/trunk/test/CodeGen/AMDGPU/mul.ll
+++ llvm/trunk/test/CodeGen/AMDGPU/mul.ll
@@ -1,5 +1,6 @@
 ; RUN:  llc -amdgpu-scalarize-global-loads=false  -march=amdgcn -mcpu=verde -verify-machineinstrs < %s | FileCheck -allow-deprecated-dag-overlap -check-prefixes=GCN,SI,FUNC %s
 ; RUN:  llc -amdgpu-scalarize-global-loads=false  -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -allow-deprecated-dag-overlap -check-prefixes=GCN,VI,FUNC %s
+; RUN:  llc -amdgpu-scalarize-global-loads=false  -march=amdgcn -mcpu=gfx900 -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -allow-deprecated-dag-overlap -check-prefixes=FUNC,GFX9 %s
 ; RUN:  llc -amdgpu-scalarize-global-loads=false  -march=r600 -mcpu=redwood < %s | FileCheck -allow-deprecated-dag-overlap -check-prefixes=EG,FUNC %s
 
 ; mul24 and mad24 are affected
@@ -139,6 +140,11 @@
 ; crash with a 'failed to select' error.
 
 ; FUNC-LABEL: {{^}}s_mul_i64:
+; GFX9-DAG: s_mul_i32
+; GFX9-DAG: s_mul_hi_u32
+; GFX9-DAG: s_mul_i32
+; GFX9-DAG: s_mul_i32
+; GFX9: s_endpgm
 define amdgpu_kernel void @s_mul_i64(i64 addrspace(1)* %out, i64 %a, i64 %b) nounwind {
   %mul = mul i64 %a, %b
   store i64 %mul, i64 addrspace(1)* %out, align 8