Index: llvm/trunk/lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp
===================================================================
--- llvm/trunk/lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp
+++ llvm/trunk/lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp
@@ -193,13 +193,10 @@
   const SIMachineFunctionInfo &MFI = *MF->getInfo<SIMachineFunctionInfo>();
   if (!MFI.isEntryFunction())
     return;
-  if (IsaInfo::hasCodeObjectV3(getSTI()) &&
-      TM.getTargetTriple().getOS() == Triple::AMDHSA)
-    return;
 
   const GCNSubtarget &STM = MF->getSubtarget<GCNSubtarget>();
   const Function &F = MF->getFunction();
-  if (STM.isAmdCodeObjectV2(F) &&
+  if (!STM.hasCodeObjectV3() && STM.isAmdHsaOrMesa(F) &&
       (F.getCallingConv() == CallingConv::AMDGPU_KERNEL ||
        F.getCallingConv() == CallingConv::SPIR_KERNEL)) {
     amd_kernel_code_t KernelCode;
@@ -210,7 +207,8 @@
   if (TM.getTargetTriple().getOS() != Triple::AMDHSA)
     return;
 
-  HSAMetadataStream.emitKernel(*MF, CurrentProgramInfo);
+  if (!STM.hasCodeObjectV3() && STM.isAmdHsaOS())
+    HSAMetadataStream.emitKernel(*MF, CurrentProgramInfo);
 }
 
 void AMDGPUAsmPrinter::EmitFunctionBodyEnd() {
@@ -259,7 +257,7 @@
 
   const SIMachineFunctionInfo *MFI = MF->getInfo<SIMachineFunctionInfo>();
   const GCNSubtarget &STM = MF->getSubtarget<GCNSubtarget>();
-  if (MFI->isEntryFunction() && STM.isAmdCodeObjectV2(MF->getFunction())) {
+  if (MFI->isEntryFunction() && STM.isAmdHsaOrMesa(MF->getFunction())) {
     SmallString<128> SymbolName;
     getNameWithPrefix(SymbolName, &MF->getFunction()),
     getTargetStreamer()->EmitAMDGPUSymbolType(
Index: llvm/trunk/lib/Target/AMDGPU/AMDGPUSubtarget.h
===================================================================
--- llvm/trunk/lib/Target/AMDGPU/AMDGPUSubtarget.h
+++ llvm/trunk/lib/Target/AMDGPU/AMDGPUSubtarget.h
@@ -135,7 +135,7 @@
     return isMesa3DOS() && !AMDGPU::isShader(F.getCallingConv());
   }
 
-  bool isAmdCodeObjectV2(const Function &F) const {
+  bool isAmdHsaOrMesa(const Function &F) const {
     return isAmdHsaOS() || isMesaKernel(F);
   }
 
@@ -202,7 +202,7 @@
   /// Returns the offset in bytes from the start of the input buffer
   ///        of the first explicit kernel argument.
   unsigned getExplicitKernelArgOffset(const Function &F) const {
-    return isAmdCodeObjectV2(F) ? 0 : 36;
+    return isAmdHsaOrMesa(F) ? 0 : 36;
   }
 
   /// \returns Maximum number of work groups per compute unit supported by the
Index: llvm/trunk/lib/Target/AMDGPU/SIFrameLowering.cpp
===================================================================
--- llvm/trunk/lib/Target/AMDGPU/SIFrameLowering.cpp
+++ llvm/trunk/lib/Target/AMDGPU/SIFrameLowering.cpp
@@ -289,7 +289,7 @@
     AMDGPUFunctionArgInfo::PRIVATE_SEGMENT_WAVE_BYTE_OFFSET);
 
   unsigned PreloadedPrivateBufferReg = AMDGPU::NoRegister;
-  if (ST.isAmdCodeObjectV2(F)) {
+  if (ST.isAmdHsaOrMesa(F)) {
     PreloadedPrivateBufferReg = MFI->getPreloadedReg(
       AMDGPUFunctionArgInfo::PRIVATE_SEGMENT_BUFFER);
   }
@@ -308,7 +308,7 @@
   }
 
   if (ResourceRegUsed && PreloadedPrivateBufferReg != AMDGPU::NoRegister) {
-    assert(ST.isAmdCodeObjectV2(F) || ST.isMesaGfxShader(F));
+    assert(ST.isAmdHsaOrMesa(F) || ST.isMesaGfxShader(F));
     MRI.addLiveIn(PreloadedPrivateBufferReg);
     MBB.addLiveIn(PreloadedPrivateBufferReg);
   }
@@ -333,7 +333,7 @@
 
   bool CopyBuffer = ResourceRegUsed &&
     PreloadedPrivateBufferReg != AMDGPU::NoRegister &&
-    ST.isAmdCodeObjectV2(F) &&
+    ST.isAmdHsaOrMesa(F) &&
     ScratchRsrcReg != PreloadedPrivateBufferReg;
 
   // This needs to be careful of the copying order to avoid overwriting one of
@@ -433,7 +433,7 @@
   }
   if (ST.isMesaGfxShader(Fn)
       || (PreloadedPrivateBufferReg == AMDGPU::NoRegister)) {
-    assert(!ST.isAmdCodeObjectV2(Fn));
+    assert(!ST.isAmdHsaOrMesa(Fn));
     const MCInstrDesc &SMovB32 = TII->get(AMDGPU::S_MOV_B32);
 
     unsigned Rsrc2 = TRI->getSubReg(ScratchRsrcReg, AMDGPU::sub2);
Index: llvm/trunk/lib/Target/AMDGPU/SIISelLowering.cpp
===================================================================
--- llvm/trunk/lib/Target/AMDGPU/SIISelLowering.cpp
+++ llvm/trunk/lib/Target/AMDGPU/SIISelLowering.cpp
@@ -1653,7 +1653,7 @@
   bool RequiresStackAccess = HasStackObjects || MFI.hasCalls();
 
   const GCNSubtarget &ST = MF.getSubtarget<GCNSubtarget>();
-  if (ST.isAmdCodeObjectV2(MF.getFunction())) {
+  if (ST.isAmdHsaOrMesa(MF.getFunction())) {
     if (RequiresStackAccess) {
       // If we have stack objects, we unquestionably need the private buffer
       // resource. For the Code Object V2 ABI, this will be the first 4 user
@@ -4809,14 +4809,14 @@
 
   switch (IntrinsicID) {
   case Intrinsic::amdgcn_implicit_buffer_ptr: {
-    if (getSubtarget()->isAmdCodeObjectV2(MF.getFunction()))
+    if (getSubtarget()->isAmdHsaOrMesa(MF.getFunction()))
       return emitNonHSAIntrinsicError(DAG, DL, VT);
     return getPreloadedValue(DAG, *MFI, VT,
                              AMDGPUFunctionArgInfo::IMPLICIT_BUFFER_PTR);
   }
   case Intrinsic::amdgcn_dispatch_ptr:
   case Intrinsic::amdgcn_queue_ptr: {
-    if (!Subtarget->isAmdCodeObjectV2(MF.getFunction())) {
+    if (!Subtarget->isAmdHsaOrMesa(MF.getFunction())) {
       DiagnosticInfoUnsupported BadIntrin(
           MF.getFunction(), "unsupported hsa intrinsic without hsa target",
           DL.getDebugLoc());
Index: llvm/trunk/lib/Target/AMDGPU/SIMachineFunctionInfo.cpp
===================================================================
--- llvm/trunk/lib/Target/AMDGPU/SIMachineFunctionInfo.cpp
+++ llvm/trunk/lib/Target/AMDGPU/SIMachineFunctionInfo.cpp
@@ -137,8 +137,8 @@
     }
   }
 
-  bool IsCOV2 = ST.isAmdCodeObjectV2(F);
-  if (IsCOV2) {
+  bool isAmdHsaOrMesa = ST.isAmdHsaOrMesa(F);
+  if (isAmdHsaOrMesa) {
     if (HasStackObjects || MaySpill)
       PrivateSegmentBuffer = true;
 
@@ -158,7 +158,7 @@
   if (F.hasFnAttribute("amdgpu-kernarg-segment-ptr"))
     KernargSegmentPtr = true;
 
-  if (ST.hasFlatAddressSpace() && isEntryFunction() && IsCOV2) {
+  if (ST.hasFlatAddressSpace() && isEntryFunction() && isAmdHsaOrMesa) {
     // TODO: This could be refined a lot. The attribute is a poor way of
     // detecting calls that may require it before argument lowering.
     if (HasStackObjects || F.hasFnAttribute("amdgpu-flat-scratch"))
Index: llvm/trunk/test/CodeGen/AMDGPU/code-object-v3.ll
===================================================================
--- llvm/trunk/test/CodeGen/AMDGPU/code-object-v3.ll
+++ llvm/trunk/test/CodeGen/AMDGPU/code-object-v3.ll
@@ -2,6 +2,10 @@
 ; RUN: llc -filetype=obj -mtriple=amdgcn-amd-amdhsa -mcpu=gfx803 -mattr=+code-object-v3 < %s | llvm-readobj -elf-output-style=GNU -notes -relocations -sections -symbols | FileCheck --check-prefixes=ALL-ELF,OSABI-AMDHSA-ELF %s
 
 ; ALL-ASM-LABEL: {{^}}fadd:
+
+; OSABI-AMDHSA-ASM-NOT: .amdgpu_hsa_kernel
+; OSABI-AMDHSA-ASM-NOT: .amd_kernel_code_t
+
 ; OSABI-AMDHSA-ASM: s_endpgm
 ; OSABI-AMDHSA-ASM: .section .rodata,#alloc
 ; OSABI-AMDHSA-ASM: .p2align 6
@@ -16,6 +20,10 @@
 ; OSABI-AMDHSA-ASM: .text
 
 ; ALL-ASM-LABEL: {{^}}fsub:
+
+; OSABI-AMDHSA-ASM-NOT: .amdgpu_hsa_kernel
+; OSABI-AMDHSA-ASM-NOT: .amd_kernel_code_t
+
 ; OSABI-AMDHSA-ASM: s_endpgm
 ; OSABI-AMDHSA-ASM: .section .rodata,#alloc
 ; OSABI-AMDHSA-ASM: .p2align 6