mirror of
https://github.com/NVIDIA/cuda-samples.git
synced 2024-11-28 14:19:15 +08:00
124 lines
4.7 KiB
C++
124 lines
4.7 KiB
C++
|
/* Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
|
||
|
*
|
||
|
* Redistribution and use in source and binary forms, with or without
|
||
|
* modification, are permitted provided that the following conditions
|
||
|
* are met:
|
||
|
* * Redistributions of source code must retain the above copyright
|
||
|
* notice, this list of conditions and the following disclaimer.
|
||
|
* * Redistributions in binary form must reproduce the above copyright
|
||
|
* notice, this list of conditions and the following disclaimer in the
|
||
|
* documentation and/or other materials provided with the distribution.
|
||
|
* * Neither the name of NVIDIA CORPORATION nor the names of its
|
||
|
* contributors may be used to endorse or promote products derived
|
||
|
* from this software without specific prior written permission.
|
||
|
*
|
||
|
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
|
||
|
* EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||
|
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
||
|
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
|
||
|
* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
|
||
|
* EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
|
||
|
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
|
||
|
* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
|
||
|
* OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
||
|
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
||
|
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
||
|
*/
|
||
|
|
||
|
// This sample demonstrates how to use the CUDA hook library to receive
|
||
|
// callbacks
|
||
|
|
||
|
#include <cuda.h>
|
||
|
#include <stdio.h>
|
||
|
#include <dlfcn.h>
|
||
|
|
||
|
#include "libcuhook.h"
|
||
|
|
||
|
#define ASSERT_COND(x, msg) \
|
||
|
do { \
|
||
|
if (!(x)) { \
|
||
|
fprintf(stderr, "Error: Condition (%s) failed at %s:%d\n", #x, __FILE__, \
|
||
|
__LINE__); \
|
||
|
fprintf(stderr, "cuHook sample failed (%s)\n", msg); \
|
||
|
exit(1); \
|
||
|
} \
|
||
|
} while (0)
|
||
|
|
||
|
/*
|
||
|
** Example of how to use the CUDA Interception Library, libcuhook.so
|
||
|
** The library has to be loaded via LD_PRELOAD, e.g.
|
||
|
*LD_PRELOAD=<full_path>/libcuhook.so.1 ./cuHook
|
||
|
*/
|
||
|
|
||
|
static int allocation_cb = 0;
|
||
|
static int free_cb = 0;
|
||
|
static int destroy_ctx_cb = 0;
|
||
|
|
||
|
CUresult device_allocation_callback(CUdeviceptr *dptr, size_t bytesize) {
|
||
|
fprintf(stdout, "Received memory allocation callback!\n");
|
||
|
allocation_cb++;
|
||
|
return CUDA_SUCCESS;
|
||
|
}
|
||
|
|
||
|
CUresult device_free_callback(CUdeviceptr dptr) {
|
||
|
fprintf(stdout, "Received memory de-allocation callback!\n");
|
||
|
free_cb++;
|
||
|
return CUDA_SUCCESS;
|
||
|
}
|
||
|
|
||
|
CUresult destroy_context_callback(CUcontext ctx) {
|
||
|
fprintf(stdout, "Received context destroy event!\n");
|
||
|
destroy_ctx_cb++;
|
||
|
return CUDA_SUCCESS;
|
||
|
}
|
||
|
|
||
|
int main() {
|
||
|
int count;
|
||
|
CUcontext ctx;
|
||
|
|
||
|
count = 0;
|
||
|
|
||
|
cuInit(0);
|
||
|
cuDeviceGetCount(&count);
|
||
|
ASSERT_COND(count > 0, "No suitable devices found");
|
||
|
|
||
|
// Load the cudaHookRegisterCallback symbol using the default library search
|
||
|
// order. If we found the symbol, then the hooking library has been loaded
|
||
|
fnCuHookRegisterCallback cuHook =
|
||
|
(fnCuHookRegisterCallback)dlsym(RTLD_DEFAULT, "cuHookRegisterCallback");
|
||
|
// ASSERT_COND(cuHook, dlerror());
|
||
|
if (cuHook) {
|
||
|
// CUDA Runtime symbols cannot be hooked but the underlying driver ones
|
||
|
// _can_. Example:
|
||
|
// - cudaFree() will trigger cuMemFree
|
||
|
// - cudaDeviceReset() will trigger a context change and you would need to
|
||
|
// intercept cuCtxGetCurrent/cuCtxSetCurrent
|
||
|
cuHook(CU_HOOK_MEM_ALLOC, POST_CALL_HOOK,
|
||
|
(void *)device_allocation_callback);
|
||
|
cuHook(CU_HOOK_MEM_FREE, PRE_CALL_HOOK, (void *)device_free_callback);
|
||
|
cuHook(CU_HOOK_CTX_DESTROY, POST_CALL_HOOK,
|
||
|
(void *)destroy_context_callback);
|
||
|
}
|
||
|
|
||
|
cuCtxCreate(&ctx, 0, 0);
|
||
|
{
|
||
|
CUresult status;
|
||
|
CUdeviceptr dptr;
|
||
|
|
||
|
status = cuMemAlloc(&dptr, 1024);
|
||
|
ASSERT_COND(status == CUDA_SUCCESS, "cuMemAlloc call failed");
|
||
|
|
||
|
status = cuMemFree(dptr);
|
||
|
ASSERT_COND(status == CUDA_SUCCESS, "cuMemFree call failed");
|
||
|
}
|
||
|
cuCtxDestroy(ctx);
|
||
|
|
||
|
ASSERT_COND(allocation_cb == 1, "Didn't receive the allocation callback");
|
||
|
ASSERT_COND(free_cb == 1, "Didn't receive the free callback");
|
||
|
ASSERT_COND(destroy_ctx_cb == 1,
|
||
|
"Didn't receive the destroy context callback");
|
||
|
|
||
|
fprintf(stdout, "Sample finished successfully.\n");
|
||
|
return (0);
|
||
|
}
|