[NVPTX] Implement __nvvm_atom_add_gen_d builtin.

Summary:
This just seems to have been an oversight.  We already supported the f64
atomic add with an explicit scope (e.g. "cta"), but not the scopeless
version.

Reviewers: tra

Subscribers: jholewinski, sanjoy, cfe-commits, llvm-commits, hiraditya

Differential Revision: https://reviews.llvm.org/D39638

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@317623 91177308-0d34-0410-b5e6-96231b3b80d8

1 files changed, 10 insertions, 0 deletions

diff --git a/lib/CodeGen/CGBuiltin.cpp b/lib/CodeGen/CGBuiltin.cpp
index f1b8e2e611..369240f316 100644
--- a/lib/CodeGen/CGBuiltin.cpp
+++ b/lib/CodeGen/CGBuiltin.cpp
@@ -9554,6 +9554,16 @@ Value *CodeGenFunction::EmitNVPTXBuiltinExpr(unsigned BuiltinID,
     return Builder.CreateCall(FnALAF32, {Ptr, Val});
   }
 
+  case NVPTX::BI__nvvm_atom_add_gen_d: {
+    Value *Ptr = EmitScalarExpr(E->getArg(0));
+    Value *Val = EmitScalarExpr(E->getArg(1));
+    // atomicrmw only deals with integer arguments, so we need to use
+    // LLVM's nvvm_atomic_load_add_f64 intrinsic.
+    Value *FnALAF64 =
+        CGM.getIntrinsic(Intrinsic::nvvm_atomic_load_add_f64, Ptr->getType());
+    return Builder.CreateCall(FnALAF64, {Ptr, Val});
+  }
+
   case NVPTX::BI__nvvm_atom_inc_gen_ui: {
     Value *Ptr = EmitScalarExpr(E->getArg(0));
     Value *Val = EmitScalarExpr(E->getArg(1));