[AMDGPU][clang] Fix __builtin_nontemporal_store() failure on AMDGPU

[skc7]

Summary

__builtin_nontemporal_store() fails on AMDGPU. This is due to mismatch in address space while creating bitcast. This change matches addrspace of poiter type to src addrSpace in EmitNontemporalStore() in clang codegen.

[yaxunl]

clang/test/CodeGenOpenCL/amdgcn-non-temporal-store.cl

9

pls add newline

[skc7]

added new life at eof amdgcn-non-temporal-store.cl

[yaxunl]

LGTM. Thanks.

clang/lib/CodeGen/CGBuiltin.cpp

 }
 
 static Value *EmitNontemporalStore(CodeGenFunction &CGF, const CallExpr *E) {
   Value *Val = CGF.EmitScalarExpr(E->getArg(0));
   Value *Address = CGF.EmitScalarExpr(E->getArg(1));
 
   // Convert the type of the pointer to a pointer to the stored type.
   Val = CGF.EmitToMemory(Val, E->getArg(0)->getType());
+  unsigned SrcAddrSpace = Address->getType()->getPointerAddressSpace();
   Value *BC = CGF.Builder.CreateBitCast(
-      Address, llvm::PointerType::getUnqual(Val->getType()), "cast");
+      Address, llvm::PointerType::get(Val->getType(), SrcAddrSpace), "cast");
   LValue LV = CGF.MakeNaturalAlignAddrLValue(BC, E->getArg(0)->getType());
   LV.setNontemporal(true);
   CGF.EmitStoreOfScalar(Val, LV, false);
   return nullptr;
 }
 
 static Value *EmitNontemporalLoad(CodeGenFunction &CGF, const CallExpr *E) {
   Value *Address = CGF.EmitScalarExpr(E->getArg(0));

clang/test/CodeGenOpenCL/amdgcn-non-temporal-store.cl

+// REQUIRES: amdgpu-registered-target
+// RUN: %clang_cc1 -triple amdgcn-unknown-unknown -S -emit-llvm -o - %s | FileCheck %s
+// CHECK-LABEL: @test_non_temporal_store_kernel
+// CHECK: store i32 0, i32 addrspace(1)* %{{.*}}, align 4, !tbaa !{{.*}}, !nontemporal {{.*}}
+
+kernel void test_non_temporal_store_kernel(global unsigned int* io) {
+  __builtin_nontemporal_store(0, io);
+}