diff --git a/lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp b/lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp
index f892f7b95178..f1fcf2f036c0 100644
--- a/lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp
+++ b/lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp
@@ -770,6 +770,7 @@ class AMDGPUAsmParser : public MCTargetAsmParser {
   bool AddNextRegisterToList(unsigned& Reg, unsigned& RegWidth, RegisterKind RegKind, unsigned Reg1, unsigned RegNum);
   bool ParseAMDGPURegister(RegisterKind& RegKind, unsigned& Reg, unsigned& RegNum, unsigned& RegWidth, unsigned *DwordRegIndex);
   void cvtMubufImpl(MCInst &Inst, const OperandVector &Operands, bool IsAtomic, bool IsAtomicReturn);
+  void cvtDSImpl(MCInst &Inst, const OperandVector &Operands, bool IsGdsHardcoded);
 
 public:
   enum AMDGPUMatchResultTy {
@@ -888,7 +889,8 @@ class AMDGPUAsmParser : public MCTargetAsmParser {
   OperandMatchResultTy parseVReg32OrOff(OperandVector &Operands);
 
   void cvtDSOffset01(MCInst &Inst, const OperandVector &Operands);
-  void cvtDS(MCInst &Inst, const OperandVector &Operands);
+  void cvtDS(MCInst &Inst, const OperandVector &Operands) { cvtDSImpl(Inst, Operands, false); }
+  void cvtDSGds(MCInst &Inst, const OperandVector &Operands) { cvtDSImpl(Inst, Operands, true); }
   void cvtExp(MCInst &Inst, const OperandVector &Operands);
 
   bool parseCnt(int64_t &IntVal);
@@ -2350,9 +2352,8 @@ void AMDGPUAsmParser::cvtDSOffset01(MCInst &Inst,
   Inst.addOperand(MCOperand::createReg(AMDGPU::M0)); // m0
 }
 
-void AMDGPUAsmParser::cvtDS(MCInst &Inst, const OperandVector &Operands) {
+void AMDGPUAsmParser::cvtDSImpl(MCInst &Inst, const OperandVector &Operands, bool IsGdsHardcoded) {
   std::map<enum AMDGPUOperand::ImmTy, unsigned> OptionalIdx;
-  bool GDSOnly = false;
 
   for (unsigned i = 1, e = Operands.size(); i != e; ++i) {
     AMDGPUOperand &Op = ((AMDGPUOperand &)*Operands[i]);
@@ -2364,7 +2365,7 @@ void AMDGPUAsmParser::cvtDS(MCInst &Inst, const OperandVector &Operands) {
     }
 
     if (Op.isToken() && Op.getToken() == "gds") {
-      GDSOnly = true;
+      IsGdsHardcoded = true;
       continue;
     }
 
@@ -2373,7 +2374,7 @@ void AMDGPUAsmParser::cvtDS(MCInst &Inst, const OperandVector &Operands) {
   }
 
   addOptionalImmOperand(Inst, Operands, OptionalIdx, AMDGPUOperand::ImmTyOffset);
-  if (!GDSOnly) {
+  if (!IsGdsHardcoded) {
     addOptionalImmOperand(Inst, Operands, OptionalIdx, AMDGPUOperand::ImmTyGDS);
   }
   Inst.addOperand(MCOperand::createReg(AMDGPU::M0)); // m0
diff --git a/lib/Target/AMDGPU/DSInstructions.td b/lib/Target/AMDGPU/DSInstructions.td
index a077001df6bd..7f7aa18c6719 100644
--- a/lib/Target/AMDGPU/DSInstructions.td
+++ b/lib/Target/AMDGPU/DSInstructions.td
@@ -174,6 +174,7 @@ class DS_1A_RET_GDS <string opName> : DS_Pseudo<opName,
   let has_data1 = 0;
   let has_gds = 0;
   let gdsValue = 1;
+  let AsmMatchConverter = "cvtDSGds";
 }
 
 class DS_0A_RET <string opName> : DS_Pseudo<opName,
diff --git a/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp b/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp
index 70ed40ec3b0a..25e8f4b621c4 100644
--- a/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp
+++ b/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp
@@ -335,14 +335,14 @@ unsigned getMCReg(unsigned Reg, const MCSubtargetInfo &STI) {
 }
 
 bool isSISrcOperand(const MCInstrDesc &Desc, unsigned OpNo) {
-  assert(OpNo <= Desc.NumOperands);
+  assert(OpNo < Desc.NumOperands);
   unsigned OpType = Desc.OpInfo[OpNo].OperandType;
   return OpType >= AMDGPU::OPERAND_SRC_FIRST &&
          OpType <= AMDGPU::OPERAND_SRC_LAST;
 }
 
 bool isSISrcFPOperand(const MCInstrDesc &Desc, unsigned OpNo) {
-  assert(OpNo <= Desc.NumOperands);
+  assert(OpNo < Desc.NumOperands);
   unsigned OpType = Desc.OpInfo[OpNo].OperandType;
   switch (OpType) {
   case AMDGPU::OPERAND_REG_IMM_FP32:
@@ -358,7 +358,7 @@ bool isSISrcFPOperand(const MCInstrDesc &Desc, unsigned OpNo) {
 }
 
 bool isSISrcInlinableOperand(const MCInstrDesc &Desc, unsigned OpNo) {
-  assert(OpNo <= Desc.NumOperands);
+  assert(OpNo < Desc.NumOperands);
   unsigned OpType = Desc.OpInfo[OpNo].OperandType;
   return OpType >= AMDGPU::OPERAND_REG_INLINE_C_FIRST &&
          OpType <= AMDGPU::OPERAND_REG_INLINE_C_LAST;
@@ -402,7 +402,7 @@ unsigned getRegBitWidth(const MCRegisterClass &RC) {
 
 unsigned getRegOperandSize(const MCRegisterInfo *MRI, const MCInstrDesc &Desc,
                            unsigned OpNo) {
-  assert(OpNo <= Desc.NumOperands);
+  assert(OpNo < Desc.NumOperands);
   unsigned RCID = Desc.OpInfo[OpNo].RegClass;
   return getRegBitWidth(MRI->getRegClass(RCID)) / 8;
 }