2 files changed, 62 insertions, 1 deletions
diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index 174cc4602d9..a6e894b7cab 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -582,10 +582,12 @@ void X86TargetLowering::resetOperationActions() {
   // VASTART needs to be custom lowered to use the VarArgsFrameIndex
   setOperationAction(ISD::VASTART           , MVT::Other, Custom);
   setOperationAction(ISD::VAEND             , MVT::Other, Expand);
-  if (Subtarget->is64Bit()) {
+  if (Subtarget->is64Bit() && !Subtarget->isTargetWin64()) {
+    // TargetInfo::X86_64ABIBuiltinVaList
     setOperationAction(ISD::VAARG           , MVT::Other, Custom);
     setOperationAction(ISD::VACOPY          , MVT::Other, Custom);
   } else {
+    // TargetInfo::CharPtrBuiltinVaList
     setOperationAction(ISD::VAARG           , MVT::Other, Expand);
     setOperationAction(ISD::VACOPY          , MVT::Other, Expand);
   }
diff --git a/llvm/test/CodeGen/X86/win64_vararg.ll b/llvm/test/CodeGen/X86/win64_vararg.ll
index 52bc50922c2..07b7826c000 100644
--- a/llvm/test/CodeGen/X86/win64_vararg.ll
+++ b/llvm/test/CodeGen/X86/win64_vararg.ll
@@ -18,6 +18,7 @@ entry:
 }
 
 declare void @llvm.va_start(i8*) nounwind
+declare void @llvm.va_copy(i8*, i8*) nounwind
 
 ; CHECK: f5:
 ; CHECK: pushq
@@ -51,3 +52,61 @@ entry:
   call void @llvm.va_start(i8* %ap1)
   ret i8* %ap1
 }
+
+; WinX86_64 uses char* for va_list. Verify that the correct amount of bytes
+; are copied using va_copy.
+; CHECK: copy4:
+; CHECK: subq $16
+; CHECK: leaq 56(%rsp), [[REGISTER:%[a-z]+]]
+; CHECK: movq [[REGISTER]], 8(%rsp)
+; CHECK: movq [[REGISTER]], (%rsp)
+; CHECK: addq $16
+; CHECK: ret
+define void @copy4(i64 %a0, i64 %a1, i64 %a2, i64 %a3, ...) nounwind {
+entry:
+  %ap = alloca i8*, align 8
+  %cp = alloca i8*, align 8
+  %ap1 = bitcast i8** %ap to i8*
+  %cp1 = bitcast i8** %cp to i8*
+  call void @llvm.va_start(i8* %ap1)
+  call void @llvm.va_copy(i8* %cp1, i8* %ap1)
+  ret void
+}
+
+; CHECK: copy1:
+; CHECK: subq $16
+; CHECK: leaq 32(%rsp), [[REGISTER:%[a-z]+]]
+; CHECK: movq [[REGISTER]], 8(%rsp)
+; CHECK: movq [[REGISTER]], (%rsp)
+; CHECK: addq $16
+; CHECK: ret
+define void @copy1(i64 %a0, ...) nounwind {
+entry:
+  %ap = alloca i8*, align 8
+  %cp = alloca i8*, align 8
+  %ap1 = bitcast i8** %ap to i8*
+  %cp1 = bitcast i8** %cp to i8*
+  call void @llvm.va_start(i8* %ap1)
+  call void @llvm.va_copy(i8* %cp1, i8* %ap1)
+  ret void
+}
+
+; CHECK: arg4:
+; CHECK: pushq
+; va_start:
+; CHECK: leaq 48(%rsp), [[REG1:%[a-z]+]]
+; CHECK: movq [[REG1]], (%rsp)
+; va_arg:
+; CHECK: leaq 52(%rsp), [[REG2:%[a-z]+]]
+; CHECK: movq [[REG2]], (%rsp)
+; CHECK: movl 48(%rsp), %eax
+; CHECK: popq
+; CHECK: ret
+define i32 @arg4(i64 %a0, i64 %a1, i64 %a2, i64 %a3, ...) nounwind {
+entry:
+  %ap = alloca i8*, align 8
+  %ap1 = bitcast i8** %ap to i8*
+  call void @llvm.va_start(i8* %ap1)
+  %tmp = va_arg i8** %ap, i32
+  ret i32 %tmp
+}