[HIP] Fix device malloc/free

[yaxunl]

Summary

ROCm 4.5 device library introduced __ockl_dm_alloc and __ockl_dm_dealloc
for supporting device side malloc/free.

This patch redefines device malloc/free to use these functions.

It also fixes a bug in the wrapper header which incorrectly defines free
with return type void* instead of void.

[tra]

clang/lib/Headers/__clang_hip_runtime_wrapper.h

80

Nit: perhaps something like this would express the intent a bit more directly:

# if HIP_VERSION_MAJOR*100+HIP_VERSION_MINOR*10 > 450

[yaxunl]

clang/lib/Headers/__clang_hip_runtime_wrapper.h

80

We had ROCm 3.10, so the minor version may be 10 or greater.

Probably use HIP_VERSION_MAJOR*100+HIP_VERSION_MINOR > 405 instead?

[tra]

clang/lib/Headers/__clang_hip_runtime_wrapper.h

80

LGTM.

clang/lib/Headers/__clang_hip_runtime_wrapper.h

extern "C" {
   }
 }
 #endif //__cplusplus
 
 #if !defined(__HIPCC_RTC__)
 #include <cmath>
 #include <cstdlib>
 #include <stdlib.h>
+#if __has_include("hip/hip_version.h")
+#include "hip/hip_version.h"
+#endif // __has_include("hip/hip_version.h")
 #else
 typedef __SIZE_TYPE__ size_t;
 // Define macros which are needed to declare HIP device API's without standard
 // C/C++ headers. This is for readability so that these API's can be written
 // the same way as non-hipRTC use case. These macros need to be popped so that
 // they do not pollute users' name space.
 #pragma push_macro("NULL")
 #pragma push_macro("uint32_t")
 #pragma push_macro("uint64_t")
 #pragma push_macro("CHAR_BIT")
 #pragma push_macro("INT_MAX")
 #define NULL (void *)0
 #define uint32_t __UINT32_TYPE__
 #define uint64_t __UINT64_TYPE__
 #define CHAR_BIT __CHAR_BIT__
 #define INT_MAX __INTMAX_MAX__
 #endif // __HIPCC_RTC__
 
 typedef __SIZE_TYPE__ __hip_size_t;
 
 #ifdef __cplusplus
 extern "C" {
 #endif //__cplusplus
 
+#if HIP_VERSION_MAJOR * 100 + HIP_VERSION_MINOR >= 405
+extern "C" __device__ unsigned long long __ockl_dm_alloc(unsigned long long __size);
+extern "C" __device__ void __ockl_dm_dealloc(unsigned long long __addr);
+__attribute__((weak)) inline __device__ void *malloc(__hip_size_t __size) {
+  return (void *) __ockl_dm_alloc(__size);
+}
+__attribute__((weak)) inline __device__ void free(void *__ptr) {
+  __ockl_dm_dealloc((unsigned long long)__ptr);
+}
+#else  // HIP version check
 #if __HIP_ENABLE_DEVICE_MALLOC__
 __device__ void *__hip_malloc(__hip_size_t __size);
 __device__ void *__hip_free(void *__ptr);
 __attribute__((weak)) inline __device__ void *malloc(__hip_size_t __size) {
   return __hip_malloc(__size);
 }
-__attribute__((weak)) inline __device__ void *free(void *__ptr) {
-  return __hip_free(__ptr);
+__attribute__((weak)) inline __device__ void free(void *__ptr) {
+  __hip_free(__ptr);
 }
 #else
 __attribute__((weak)) inline __device__ void *malloc(__hip_size_t __size) {
   __builtin_trap();
   return (void *)0;
 }
-__attribute__((weak)) inline __device__ void *free(void *__ptr) {
+__attribute__((weak)) inline __device__ void free(void *__ptr) {
   __builtin_trap();
-  return (void *)0;
 }
 #endif
+#endif // HIP version check
 
 #ifdef __cplusplus
 } // extern "C"
 #endif //__cplusplus
 
 #include <__clang_hip_libdevice_declares.h>
 #include <__clang_hip_math.h>
 

clang/test/Headers/hip-header.hip

 // REQUIRES: amdgpu-registered-target
 // RUN: %clang_cc1 -include __clang_hip_runtime_wrapper.h \
 // RUN:   -internal-isystem %S/../../lib/Headers/cuda_wrappers \
 // RUN:   -internal-isystem %S/Inputs/include \
 // RUN:   -triple amdgcn-amd-amdhsa -aux-triple x86_64-unknown-unknown \
 // RUN:   -target-cpu gfx906 -emit-llvm %s -fcuda-is-device -o - \
-// RUN:   -D__HIPCC_RTC__ | FileCheck %s
+// RUN:   -D__HIPCC_RTC__ | FileCheck -check-prefixes=CHECK,NOMALLOC %s
 // RUN: %clang_cc1 -include __clang_hip_runtime_wrapper.h \
 // RUN:   -internal-isystem %S/../../lib/Headers/cuda_wrappers \
 // RUN:   -internal-isystem %S/Inputs/include \
 // RUN:   -include cmath \
 // RUN:   -triple amdgcn-amd-amdhsa -aux-triple x86_64-unknown-unknown \
 // RUN:   -target-cpu gfx906 -emit-llvm %s -fcuda-is-device -o - \
 // RUN:   -D__HIPCC_RTC__ | FileCheck %s  -check-prefixes=AMD_BOOL_RETURN
 // RUN: %clang_cc1 -include __clang_hip_runtime_wrapper.h \
 // RUN:   -internal-isystem %S/../../lib/Headers/cuda_wrappers \
 // RUN:   -internal-isystem %S/Inputs/include \
 // RUN:   -include cmath \
 // RUN:   -triple amdgcn-amd-amdhsa -aux-triple x86_64-unknown-unknown \
 // RUN:   -target-cpu gfx906 -emit-llvm %s -fcuda-is-device -o - \
 // RUN:   -D__HIPCC_RTC__ -DUSE_ISNAN_WITH_INT_RETURN | FileCheck %s -check-prefixes=AMD_INT_RETURN
 // RUN: %clang_cc1 -include __clang_hip_runtime_wrapper.h \
 // RUN:   -internal-isystem %S/../../lib/Headers/cuda_wrappers \
 // RUN:   -internal-isystem %S/Inputs/include \
 // RUN:   -triple amdgcn-amd-amdhsa -aux-triple x86_64-unknown-unknown \
 // RUN:   -target-cpu gfx906 -emit-llvm %s -fcuda-is-device -o - \
 // RUN:   -D__HIPCC_RTC__ -std=c++14 | FileCheck -check-prefixes=CHECK,CXX14 %s
+// RUN: %clang_cc1 -include __clang_hip_runtime_wrapper.h \
+// RUN:   -internal-isystem %S/../../lib/Headers/cuda_wrappers \
+// RUN:   -internal-isystem %S/Inputs/include \
+// RUN:   -triple amdgcn-amd-amdhsa -aux-triple x86_64-unknown-unknown \
+// RUN:   -target-cpu gfx906 -emit-llvm %s -fcuda-is-device -o - \
+// RUN:   -DHIP_VERSION_MAJOR=4 -DHIP_VERSION_MINOR=5 \
+// RUN:   -D__HIPCC_RTC__ | FileCheck -check-prefixes=CHECK,MALLOC %s
 
 // expected-no-diagnostics
 
 // Check support for pure and deleted virtual functions
 struct base {
   __host__
   __device__
   virtual void pv() = 0;
__device__ double test_isnan() {
 
   return r ;
 }
 
 // Check that device malloc and free do not conflict with std headers.
 #include <cstdlib>
 // CHECK-LABEL: define{{.*}}@_Z11test_malloc
 // CHECK: call {{.*}}i8* @malloc(i64
-// CHECK: define weak {{.*}}i8* @malloc(i64
+// CHECK-LABEL: define weak {{.*}}i8* @malloc(i64
+// MALLOC:  call i64 @__ockl_dm_alloc
+// NOMALLOC:  call void @llvm.trap
 __device__ void test_malloc(void *a) {
   a = malloc(42);
 }
 
 // CHECK-LABEL: define{{.*}}@_Z9test_free
-// CHECK: call {{.*}}i8* @free(i8*
-// CHECK: define weak {{.*}}i8* @free(i8*
+// CHECK: call {{.*}}void @free(i8*
+// CHECK-LABEL: define weak {{.*}}void @free(i8*
+// MALLOC:  call void @__ockl_dm_dealloc
+// NOMALLOC: call void @llvm.trap
 __device__ void test_free(void *a) {
   free(a);
 }