[SystemZ/z/OS] Add va intrinsics for XPLINK

[Kai]

Summary

Add support for va intrinsics for the XPLINK ABI.
Only the extended vararg variant, which uses a pointer to next
argument, is supported. The standard variant will build on this.

[Kai]

Fix formatting.

[uweigand]

LGTM, thanks!

llvm/lib/Target/SystemZ/SystemZISelLowering.h

SDValue lowerGlobalTLSAddress(GlobalAddressSDNode *Node,
                                 SelectionDAG &DAG) const;
   SDValue lowerBlockAddress(BlockAddressSDNode *Node,
                             SelectionDAG &DAG) const;
   SDValue lowerJumpTable(JumpTableSDNode *JT, SelectionDAG &DAG) const;
   SDValue lowerConstantPool(ConstantPoolSDNode *CP, SelectionDAG &DAG) const;
   SDValue lowerFRAMEADDR(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerRETURNADDR(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerVASTART(SDValue Op, SelectionDAG &DAG) const;
+  SDValue lowerVASTART_ELF(SDValue Op, SelectionDAG &DAG) const;
+  SDValue lowerVASTART_XPLINK(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerVACOPY(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerDYNAMIC_STACKALLOC(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerDYNAMIC_STACKALLOC_ELF(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerDYNAMIC_STACKALLOC_XPLINK(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerGET_DYNAMIC_AREA_OFFSET(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerSMUL_LOHI(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerUMUL_LOHI(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerSDIVREM(SDValue Op, SelectionDAG &DAG) const;

llvm/lib/Target/SystemZ/SystemZISelLowering.cpp

SDValue Shift = DAG.getNode(ISD::SRL, DL, MVT::i64, Out64,
                                 DAG.getConstant(32, DL, MVT::i64));
     return DAG.getNode(ISD::TRUNCATE, DL, MVT::i32, Shift);
   }
   llvm_unreachable("Unexpected bitcast combination");
 }
 
 SDValue SystemZTargetLowering::lowerVASTART(SDValue Op,
                                             SelectionDAG &DAG) const {
+
+  if (Subtarget.isTargetXPLINK64())
+    return lowerVASTART_XPLINK(Op, DAG);
+  else
+    return lowerVASTART_ELF(Op, DAG);
+}
+
+SDValue SystemZTargetLowering::lowerVASTART_XPLINK(SDValue Op,
+                                                   SelectionDAG &DAG) const {
+  MachineFunction &MF = DAG.getMachineFunction();
+  SystemZMachineFunctionInfo *FuncInfo =
+      MF.getInfo<SystemZMachineFunctionInfo>();
+
+  SDLoc DL(Op);
+
+  // vastart just stores the address of the VarArgsFrameIndex slot into the
+  // memory location argument.
+  EVT PtrVT = getPointerTy(DAG.getDataLayout());
+  SDValue FR = DAG.getFrameIndex(FuncInfo->getVarArgsFrameIndex(), PtrVT);
+  const Value *SV = cast<SrcValueSDNode>(Op.getOperand(2))->getValue();
+  return DAG.getStore(Op.getOperand(0), DL, FR, Op.getOperand(1),
+                      MachinePointerInfo(SV));
+}
+
+SDValue SystemZTargetLowering::lowerVASTART_ELF(SDValue Op,
+                                                SelectionDAG &DAG) const {
   MachineFunction &MF = DAG.getMachineFunction();
   SystemZMachineFunctionInfo *FuncInfo =
     MF.getInfo<SystemZMachineFunctionInfo>();
   EVT PtrVT = getPointerTy(DAG.getDataLayout());
 
   SDValue Chain   = Op.getOperand(0);
   SDValue Addr    = Op.getOperand(1);
   const Value *SV = cast<SrcValueSDNode>(Op.getOperand(2))->getValue();
SDValue SystemZTargetLowering::lowerVACOPY(SDValue Op,
                                            SelectionDAG &DAG) const {
   SDValue Chain      = Op.getOperand(0);
   SDValue DstPtr     = Op.getOperand(1);
   SDValue SrcPtr     = Op.getOperand(2);
   const Value *DstSV = cast<SrcValueSDNode>(Op.getOperand(3))->getValue();
   const Value *SrcSV = cast<SrcValueSDNode>(Op.getOperand(4))->getValue();
   SDLoc DL(Op);
 
-  return DAG.getMemcpy(Chain, DL, DstPtr, SrcPtr, DAG.getIntPtrConstant(32, DL),
+  uint32_t Sz =
+      Subtarget.isTargetXPLINK64() ? getTargetMachine().getPointerSize(0) : 32;
+  return DAG.getMemcpy(Chain, DL, DstPtr, SrcPtr, DAG.getIntPtrConstant(Sz, DL),
                        Align(8), /*isVolatile*/ false, /*AlwaysInline*/ false,
                        /*isTailCall*/ false, MachinePointerInfo(DstSV),
                        MachinePointerInfo(SrcSV));
 }
 
 SDValue
 SystemZTargetLowering::lowerDYNAMIC_STACKALLOC(SDValue Op,
                                                SelectionDAG &DAG) const {

llvm/test/CodeGen/SystemZ/call-zos-vararg.ll

 ; CHECK-NOT: lg  3
 ; CHECK-NOT: lgr  3
 define i64 @call_vararg_integral0(i32 signext %arg0, i16 signext %arg1, i64 signext %arg2, i8 signext %arg3) {
 entry:
   %retval = call i64(...) @pass_vararg3(i32 signext %arg0, i16 signext %arg1, i64 signext %arg2, i8 signext %arg3)
   ret i64 %retval
 }
 
+; Derived from C source:
+; #define _VARARG_EXT_
+; #include <stdarg.h>
+;
+; long pass(long x, ...) {
+;   va_list va;
+;   va_start(va, x);
+;   long ret = va_arg(va, long);
+;   va_end(va);
+;   return ret;
+; }
+;
+; CHECK-LABEL: pass_vararg:
+; CHECK: aghi    4, -160
+; CHECK: la      0, 2208(4)
+; CHECK: stg     0, 2200(4)
+define hidden i64 @pass_vararg(i64 %x, ...) {
+entry:
+  %va = alloca i8*, align 8
+  %va1 = bitcast i8** %va to i8*
+  call void @llvm.va_start(i8* %va1)
+  %argp.cur = load i8*, i8** %va, align 8
+  %argp.next = getelementptr inbounds i8, i8* %argp.cur, i64 8
+  store i8* %argp.next, i8** %va, align 8
+  %0 = bitcast i8* %argp.cur to i64*
+  %ret = load i64, i64* %0, align 8
+  %va2 = bitcast i8** %va to i8*
+  call void @llvm.va_end(i8* %va2)
+  ret i64 %ret
+}
+
+declare void @llvm.va_start(i8*)
+declare void @llvm.va_end(i8*)
+
 declare i64 @pass_vararg0(i64 %arg0, i64 %arg1, ...)
 declare i64 @pass_vararg1(fp128 %arg0, ...)
 declare i64 @pass_vararg2(i64 %arg0, ...)
 declare i64 @pass_vararg3(...)