[NVPTX] Make tensor load/store intrinsics overloaded.

This way we can support address-space specific variants without explicitly
encoding the space in the name of the intrinsic. Less intrinsics to deal with ->
less boilerplate.

Added a bit of tablegen magic to match/replace an intrinsics with a pointer
argument in particular address space with the space-specific instruction
variant.

Updated tests to use non-default address spaces.

Differential Revision: https://reviews.llvm.org/D43268

llvm-svn: 328006

1 files changed, 3 insertions, 5 deletions

diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index 7186e882d52..7c8f111d10f 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -10527,8 +10527,7 @@ Value *CodeGenFunction::EmitNVPTXBuiltinExpr(unsigned BuiltinID,
       llvm_unreachable("Unexpected builtin ID.");
     }
     Value *Result =
-        Builder.CreateCall(CGM.getIntrinsic(IID),
-                           {Builder.CreatePointerCast(Src, VoidPtrTy), Ldm});
+        Builder.CreateCall(CGM.getIntrinsic(IID, Src->getType()), {Src, Ldm});
 
     // Save returned values.
     for (unsigned i = 0; i < NumResults; ++i) {
@@ -10567,10 +10566,9 @@ Value *CodeGenFunction::EmitNVPTXBuiltinExpr(unsigned BuiltinID,
     default:
       llvm_unreachable("Unexpected builtin ID.");
     }
-    Function *Intrinsic = CGM.getIntrinsic(IID);
+    Function *Intrinsic = CGM.getIntrinsic(IID, Dst->getType());
     llvm::Type *ParamType = Intrinsic->getFunctionType()->getParamType(1);
-    SmallVector<Value *, 10> Values;
-    Values.push_back(Builder.CreatePointerCast(Dst, VoidPtrTy));
+    SmallVector<Value *, 10> Values = {Dst};
     for (unsigned i = 0; i < NumResults; ++i) {
       Value *V = Builder.CreateAlignedLoad(
           Builder.CreateGEP(Src.getPointer(), llvm::ConstantInt::get(IntTy, i)),