Index: lib/Target/AMDGPU/AMDGPUAnnotateKernelFeatures.cpp
===================================================================
--- lib/Target/AMDGPU/AMDGPUAnnotateKernelFeatures.cpp
+++ lib/Target/AMDGPU/AMDGPUAnnotateKernelFeatures.cpp
@@ -191,7 +191,8 @@
     { "llvm.amdgcn.dispatch.ptr", "amdgpu-dispatch-ptr" },
     { "llvm.amdgcn.queue.ptr", "amdgpu-queue-ptr" },
     { "llvm.amdgcn.dispatch.id", "amdgpu-dispatch-id" },
-	{ "llvm.trap", "amdgpu-queue-ptr" }
+    { "llvm.trap", "amdgpu-queue-ptr" },
+    { "llvm.debugtrap", "amdgpu-queue-ptr" }
   };
 
   // TODO: We should not add the attributes if the known compile time workgroup
Index: lib/Target/AMDGPU/SIISelLowering.cpp
===================================================================
--- lib/Target/AMDGPU/SIISelLowering.cpp
+++ lib/Target/AMDGPU/SIISelLowering.cpp
@@ -272,7 +272,11 @@
 
   // On SI this is s_memtime and s_memrealtime on VI.
   setOperationAction(ISD::READCYCLECOUNTER, MVT::i64, Legal);
-  setOperationAction(ISD::TRAP, MVT::Other, Legal);
+
+  if (Subtarget->isAmdHsaOS())
+    setOperationAction(ISD::TRAP, MVT::Other, Legal);
+  else
+    setOperationAction(ISD::TRAP, MVT::Other, Custom);
 
   setOperationAction(ISD::FMINNUM, MVT::f64, Legal);
   setOperationAction(ISD::FMAXNUM, MVT::f64, Legal);
@@ -1786,24 +1790,25 @@
   }
 
   switch (MI.getOpcode()) {
-   case AMDGPU::S_TRAP_PSEUDO: {
-	DebugLoc DL = MI.getDebugLoc();
-	BuildMI(*BB, MI, DL, TII->get(AMDGPU::V_MOV_B32_e32), AMDGPU::VGPR0)
-     .addImm(1);
+  case AMDGPU::S_TRAP_PSEUDO: {
+    DebugLoc DL = MI.getDebugLoc();
+    BuildMI(*BB, MI, DL, TII->get(AMDGPU::V_MOV_B32_e32), AMDGPU::VGPR0)
+      .addImm(1);
 
     MachineFunction *MF = BB->getParent();
     SIMachineFunctionInfo *Info = MF->getInfo<SIMachineFunctionInfo>();
     unsigned UserSGPR = Info->getQueuePtrUserSGPR();
+    Info->enableHSATrapHandler();
     assert(UserSGPR != AMDGPU::NoRegister);
 
     if (!BB->isLiveIn(UserSGPR))
       BB->addLiveIn(UserSGPR);
 
     BuildMI(*BB, MI, DL, TII->get(AMDGPU::COPY), AMDGPU::SGPR0_SGPR1)
-     .addReg(UserSGPR);
+      .addReg(UserSGPR);
     BuildMI(*BB, MI, DL, TII->get(AMDGPU::S_TRAP)).addImm(0x1)
-     .addReg(AMDGPU::VGPR0, RegState::Implicit)
-     .addReg(AMDGPU::SGPR0_SGPR1, RegState::Implicit);
+      .addReg(AMDGPU::VGPR0, RegState::Implicit)
+      .addReg(AMDGPU::SGPR0_SGPR1, RegState::Implicit);
 
     MI.eraseFromParent();
     return BB;
@@ -1978,6 +1983,7 @@
     return lowerINSERT_VECTOR_ELT(Op, DAG);
   case ISD::EXTRACT_VECTOR_ELT:
     return lowerEXTRACT_VECTOR_ELT(Op, DAG);
+  case ISD::TRAP: return lowerTRAP(Op, DAG);
   case ISD::FP_ROUND:
     return lowerFP_ROUND(Op, DAG);
   }
@@ -2451,6 +2457,23 @@
                          MachineMemOperand::MOInvariant);
 }
 
+SDValue SITargetLowering::lowerTRAP(SDValue Op,
+                                    SelectionDAG &DAG) const {
+  const MachineFunction &MF = DAG.getMachineFunction();
+  DiagnosticInfoUnsupported NoTrap(*MF.getFunction(),
+                                   "trap handler not supported",
+                                   Op.getDebugLoc(),
+                                   DS_Warning);
+  DAG.getContext()->diagnose(NoTrap);
+
+  // Emit s_endpgm.
+
+  // FIXME: This should really be selected to s_trap, but that requires
+  // setting up the trap handler for it o do anything.
+  return DAG.getNode(AMDGPUISD::ENDPGM, SDLoc(Op), MVT::Other,
+                     Op.getOperand(0));
+}
+
 SDValue SITargetLowering::copyToM0(SelectionDAG &DAG, SDValue Chain,
                                    const SDLoc &DL, SDValue V) const {
   // We can't use S_MOV_B32 directly, because there is no way to specify m0 as
Index: lib/Target/AMDGPU/SIMachineFunctionInfo.h
===================================================================
--- lib/Target/AMDGPU/SIMachineFunctionInfo.h
+++ lib/Target/AMDGPU/SIMachineFunctionInfo.h
@@ -151,6 +151,7 @@
   bool PrivateSegmentBuffer : 1;
   bool DispatchPtr : 1;
   bool QueuePtr : 1;
+  bool TrapHandlerPtr : 1;
   bool KernargSegmentPtr : 1;
   bool DispatchID : 1;
   bool FlatScratchInit : 1;
@@ -249,6 +250,10 @@
     PrivateSegmentWaveByteOffsetSystemSGPR = Reg;
   }
 
+  bool enableHSATrapHandler() {
+    return TrapHandlerPtr = true;
+  }
+
   bool hasPrivateSegmentBuffer() const {
     return PrivateSegmentBuffer;
   }
Index: lib/Target/AMDGPU/SIMachineFunctionInfo.cpp
===================================================================
--- lib/Target/AMDGPU/SIMachineFunctionInfo.cpp
+++ lib/Target/AMDGPU/SIMachineFunctionInfo.cpp
@@ -64,6 +64,7 @@
     PrivateSegmentBuffer(false),
     DispatchPtr(false),
     QueuePtr(false),
+    TrapHandlerPtr(false),
     KernargSegmentPtr(false),
     DispatchID(false),
     FlatScratchInit(false),
Index: test/CodeGen/AMDGPU/trap.ll
===================================================================
--- test/CodeGen/AMDGPU/trap.ll
+++ test/CodeGen/AMDGPU/trap.ll
@@ -1,11 +1,32 @@
-; RUN: llc -mtriple=amdgcn--amdhsa -verify-machineinstrs < %s 2>&1 | FileCheck -check-prefix=GCN %s
+; RUN: llc -mtriple=amdgcn--amdhsa -verify-machineinstrs < %s 2>&1 | FileCheck -check-prefix=HSA %s
+; RUN: llc -march=amdgcn -verify-machineinstrs < %s 2>&1 | FileCheck -check-prefix=GCN %s
 
 declare void @llvm.trap() #0
+declare void @llvm.debugtrap() #0
 
+; HSA-LABEL: {{^}}hsa_trap:
+; HSA: v_mov_b32_e32 v0, 1
+; HSA: s_mov_b64 s[0:1], s[4:5]
+; HSA: s_trap 1
+define void @hsa_trap() {
+  call void @llvm.trap()
+  ret void
+}
+
+; HSA-LABEL: {{^}}hsa_debugtrap:
+; HSA: v_mov_b32_e32 v0, 1
+; HSA: s_mov_b64 s[0:1], s[4:5]
+; HSA: s_trap 1
+define void @hsa_debugtrap() {
+  call void @llvm.debugtrap()
+  ret void
+}
+
+; For non-HSA path
+; GCN: warning: <unknown>:0:0: in function trap void (): trap handler not supported
 ; GCN-LABEL: {{^}}trap:
-; GCN: v_mov_b32_e32 v0, 1
-; GCN: s_mov_b64 s[0:1], s[4:5]
-; GCN: s_trap 1
+; GCN: s_endpgm
+; GCN-NEXT: s_endpgm
 define void @trap() {
   call void @llvm.trap()
   ret void