[clang] [libc] [Clang] Implement resource directory headers for common GPU intrinsics (PR #110179)

[Johannes Doerfert via cfe-commits]

================
@@ -0,0 +1,154 @@
+//===-- nvptxintrin.h - NVPTX intrinsic functions -------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef __NVPTXINTRIN_H
+#define __NVPTXINTRIN_H
+
+#ifndef __NVPTX__
+#error "This file is intended for NVPTX targets or offloading to NVPTX"
+#endif
+
+#include <stdbool.h>
+#include <stdint.h>
+
+#if defined(__HIP__) || defined(__CUDA__)
+#define _DEFAULT_ATTRS __attribute__((device))
+#elif !defined(_DEFAULT_ATTRS)
+#define _DEFAULT_ATTRS
+#endif
+
+#pragma omp begin declare target device_type(nohost)
+#pragma omp begin declare variant match(device = {arch(nvptx64)})
+
+// Type aliases to the address spaces used by the NVPTX backend.
+#define __gpu_private __attribute__((opencl_private))
+#define __gpu_constant __attribute__((opencl_constant))
+#define __gpu_local __attribute__((opencl_local))
+#define __gpu_global __attribute__((opencl_global))
+#define __gpu_generic __attribute__((opencl_generic))
+
+// Attribute to declare a function as a kernel.
+#define __gpu_kernel __attribute__((amdgpu_kernel, visibility("protected")))
----------------
jdoerfert wrote:

shouldn't we expose this to the user as well? I mean, explicitly in the gpuintr.h as defines or as comment.

Also, amdgpu_kernel? 

https://github.com/llvm/llvm-project/pull/110179