Add ret instruction to PTX backend

llvm-svn: 114788

llvm/include/llvm/CallingConv.h

@@ -79,7 +79,15 @@ namespace CallingConv {
     /// X86_ThisCall - Similar to X86_StdCall. Passes first argument in ECX,
     /// others via stack. Callee is responsible for stack cleaning. MSVC uses
     /// this by default for methods in its ABI.
-    X86_ThisCall = 70
+    X86_ThisCall = 70,
+
+    /// PTX_Kernel - Call to a PTX kernel.
+    /// Passes all arguments in parameter space.
+    PTX_Kernel = 71,
+
+    /// PTX_Device - Call to a PTX device function.
+    /// Passes all arguments in register or parameter space.
+    PTX_Device = 72
   };
 } // End CallingConv namespace
 

llvm/lib/AsmParser/LLLexer.cpp

@@ -544,6 +544,8 @@ lltok::Kind LLLexer::LexIdentifier() {
   KEYWORD(arm_aapcscc);
   KEYWORD(arm_aapcs_vfpcc);
   KEYWORD(msp430_intrcc);
+  KEYWORD(ptx_kernel);
+  KEYWORD(ptx_device);
 
   KEYWORD(cc);
   KEYWORD(c);

llvm/lib/AsmParser/LLParser.cpp

@@ -1084,6 +1084,8 @@ bool LLParser::ParseOptionalVisibility(unsigned &Res) {
 ///   ::= 'arm_aapcscc'
 ///   ::= 'arm_aapcs_vfpcc'
 ///   ::= 'msp430_intrcc'
+///   ::= 'ptx_kernel'
+///   ::= 'ptx_device'
 ///   ::= 'cc' UINT
 ///
 bool LLParser::ParseOptionalCallingConv(CallingConv::ID &CC) {
@@ -1099,6 +1101,8 @@ bool LLParser::ParseOptionalCallingConv(CallingConv::ID &CC) {
   case lltok::kw_arm_aapcscc:    CC = CallingConv::ARM_AAPCS; break;
   case lltok::kw_arm_aapcs_vfpcc:CC = CallingConv::ARM_AAPCS_VFP; break;
   case lltok::kw_msp430_intrcc:  CC = CallingConv::MSP430_INTR; break;
+  case lltok::kw_ptx_kernel:     CC = CallingConv::PTX_Kernel; break;
+  case lltok::kw_ptx_device:     CC = CallingConv::PTX_Device; break;
   case lltok::kw_cc: {
       unsigned ArbitraryCC;
       Lex.Lex();

llvm/lib/AsmParser/LLToken.h

@@ -72,6 +72,7 @@ namespace lltok {
     kw_x86_stdcallcc, kw_x86_fastcallcc, kw_x86_thiscallcc,
     kw_arm_apcscc, kw_arm_aapcscc, kw_arm_aapcs_vfpcc,
     kw_msp430_intrcc,
+    kw_ptx_kernel, kw_ptx_device,
 
     kw_signext,
     kw_zeroext,

llvm/lib/Target/PTX/PTXISelLowering.cpp

@@ -32,6 +32,7 @@ const char *PTXTargetLowering::getTargetNodeName(unsigned Opcode) const {
   switch (Opcode) {
     default:           llvm_unreachable("Unknown opcode");
     case PTXISD::EXIT: return "PTXISD::EXIT";
+    case PTXISD::RET:  return "PTXISD::RET";
   }
 }
 
@@ -58,5 +59,25 @@ SDValue PTXTargetLowering::
               const SmallVectorImpl<SDValue> &OutVals,
               DebugLoc dl,
               SelectionDAG &DAG) const {
-  return DAG.getNode(PTXISD::EXIT, dl, MVT::Other, Chain);
+  assert(!isVarArg && "PTX does not support var args.");
+
+  switch (CallConv) {
+    default:
+      llvm_unreachable("Unsupported calling convention.");
+    case CallingConv::PTX_Kernel:
+      assert(Outs.size() == 0 && "Kernel must return void.");
+      return DAG.getNode(PTXISD::EXIT, dl, MVT::Other, Chain);
+    case CallingConv::PTX_Device:
+      assert(Outs.size() <= 1 && "Can at most return one value.");
+      break;
+  }
+
+  // PTX_Device
+
+  if (Outs.size() == 0)
+    return DAG.getNode(PTXISD::RET, dl, MVT::Other, Chain);
+
+  // TODO: allocate return register
+  SDValue Flag;
+  return DAG.getNode(PTXISD::RET, dl, MVT::Other, Chain, Flag);
 }

llvm/lib/Target/PTX/PTXISelLowering.h

@@ -24,7 +24,8 @@ class PTXTargetMachine;
 namespace PTXISD {
   enum NodeType {
     FIRST_NUMBER = ISD::BUILTIN_OP_END,
-    EXIT
+    EXIT,
+    RET
   };
 } // namespace PTXISD
 

llvm/lib/Target/PTX/PTXInstrInfo.td

@@ -23,6 +23,8 @@ include "PTXInstrFormats.td"
 
 def PTXexit
   : SDNode<"PTXISD::EXIT", SDTNone, [SDNPHasChain]>;
+def PTXret
+  : SDNode<"PTXISD::RET",  SDTNone, [SDNPHasChain]>;
 
 //===----------------------------------------------------------------------===//
 // Instructions
@@ -30,4 +32,5 @@ def PTXexit
 
 let isReturn = 1, isTerminator = 1, isBarrier = 1 in {
   def EXIT : InstPTX<(outs), (ins), "exit", [(PTXexit)]>;
+  def RET  : InstPTX<(outs), (ins), "ret",  [(PTXret)]>;
 }

llvm/lib/VMCore/AsmWriter.cpp

@@ -1579,6 +1579,8 @@ void AssemblyWriter::printFunction(const Function *F) {
   case CallingConv::ARM_AAPCS:    Out << "arm_aapcscc "; break;
   case CallingConv::ARM_AAPCS_VFP:Out << "arm_aapcs_vfpcc "; break;
   case CallingConv::MSP430_INTR:  Out << "msp430_intrcc "; break;
+  case CallingConv::PTX_Kernel:   Out << "ptx_kernel"; break;
+  case CallingConv::PTX_Device:   Out << "ptx_device"; break;
   default: Out << "cc" << F->getCallingConv() << " "; break;
   }
 
@@ -1847,6 +1849,8 @@ void AssemblyWriter::printInstruction(const Instruction &I) {
     case CallingConv::ARM_AAPCS:    Out << " arm_aapcscc "; break;
     case CallingConv::ARM_AAPCS_VFP:Out << " arm_aapcs_vfpcc "; break;
     case CallingConv::MSP430_INTR:  Out << " msp430_intrcc "; break;
+    case CallingConv::PTX_Kernel:   Out << " ptx_kernel"; break;
+    case CallingConv::PTX_Device:   Out << " ptx_device"; break;
     default: Out << " cc" << CI->getCallingConv(); break;
     }
 
@@ -1901,6 +1905,8 @@ void AssemblyWriter::printInstruction(const Instruction &I) {
     case CallingConv::ARM_AAPCS:    Out << " arm_aapcscc "; break;
     case CallingConv::ARM_AAPCS_VFP:Out << " arm_aapcs_vfpcc "; break;
     case CallingConv::MSP430_INTR:  Out << " msp430_intrcc "; break;
+    case CallingConv::PTX_Kernel:   Out << " ptx_kernel"; break;
+    case CallingConv::PTX_Device:   Out << " ptx_device"; break;
     default: Out << " cc" << II->getCallingConv(); break;
     }
 

llvm/lib/VMCore/Verifier.cpp

@@ -685,6 +685,8 @@ void Verifier::visitFunction(Function &F) {
   case CallingConv::Cold:
   case CallingConv::X86_FastCall:
   case CallingConv::X86_ThisCall:
+  case CallingConv::PTX_Kernel:
+  case CallingConv::PTX_Device:
     Assert1(!F.isVarArg(),
             "Varargs functions must have C calling conventions!", &F);
     break;

llvm/test/CodeGen/PTX/exit.ll

@@ -1,6 +1,6 @@
 ; RUN: llc < %s -march=ptx | FileCheck %s
 
-define void @t1() {
+define ptx_kernel void @t1() {
 ;CHECK: exit;
 	ret void
 }

llvm/utils/vim/llvm.vim

@@ -49,6 +49,7 @@ syn keyword llvmKeyword hidden protected default
 syn keyword llvmKeyword except deplibs
 syn keyword llvmKeyword volatile fastcc coldcc cc ccc
 syn keyword llvmKeyword x86_stdcallcc x86_fastcallcc
+syn keyword llvmKeyword ptx_kernel ptx_device
 syn keyword llvmKeyword signext zeroext inreg sret nounwind noreturn
 syn keyword llvmKeyword nocapture byval nest readnone readonly noalias
 syn keyword llvmKeyword inlinehint noinline alwaysinline optsize ssp sspreq