1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
|
#include "../include/cuda.h"
#include "../include/library_loader.h"
#include <string.h>
bool gsr_cuda_load(gsr_cuda *self, Display *display, bool do_overclock) {
memset(self, 0, sizeof(gsr_cuda));
self->do_overclock = do_overclock;
dlerror(); /* clear */
void *lib = dlopen("libcuda.so.1", RTLD_LAZY);
if(!lib) {
lib = dlopen("libcuda.so", RTLD_LAZY);
if(!lib) {
fprintf(stderr, "gsr error: gsr_cuda_load failed: failed to load libcuda.so/libcuda.so.1, error: %s\n", dlerror());
return false;
}
}
dlsym_assign required_dlsym[] = {
{ (void**)&self->cuInit, "cuInit" },
{ (void**)&self->cuDeviceGetCount, "cuDeviceGetCount" },
{ (void**)&self->cuDeviceGet, "cuDeviceGet" },
{ (void**)&self->cuCtxCreate_v2, "cuCtxCreate_v2" },
{ (void**)&self->cuCtxDestroy_v2, "cuCtxDestroy_v2" },
{ (void**)&self->cuCtxPushCurrent_v2, "cuCtxPushCurrent_v2" },
{ (void**)&self->cuCtxPopCurrent_v2, "cuCtxPopCurrent_v2" },
{ (void**)&self->cuGetErrorString, "cuGetErrorString" },
{ (void**)&self->cuMemsetD8_v2, "cuMemsetD8_v2" },
{ (void**)&self->cuMemcpy2D_v2, "cuMemcpy2D_v2" },
{ (void**)&self->cuGraphicsGLRegisterImage, "cuGraphicsGLRegisterImage" },
{ (void**)&self->cuGraphicsResourceSetMapFlags, "cuGraphicsResourceSetMapFlags" },
{ (void**)&self->cuGraphicsMapResources, "cuGraphicsMapResources" },
{ (void**)&self->cuGraphicsUnmapResources, "cuGraphicsUnmapResources" },
{ (void**)&self->cuGraphicsUnregisterResource, "cuGraphicsUnregisterResource" },
{ (void**)&self->cuGraphicsSubResourceGetMappedArray, "cuGraphicsSubResourceGetMappedArray" },
{ NULL, NULL }
};
CUresult res;
if(!dlsym_load_list(lib, required_dlsym)) {
fprintf(stderr, "gsr error: gsr_cuda_load failed: missing required symbols in libcuda.so/libcuda.so.1\n");
goto fail;
}
res = self->cuInit(0);
if(res != CUDA_SUCCESS) {
const char *err_str = "unknown";
self->cuGetErrorString(res, &err_str);
fprintf(stderr, "gsr error: gsr_cuda_load failed: cuInit failed, error: %s (result: %d)\n", err_str, res);
goto fail;
}
int nGpu = 0;
self->cuDeviceGetCount(&nGpu);
if(nGpu <= 0) {
fprintf(stderr, "gsr error: gsr_cuda_load failed: no cuda supported devices found\n");
goto fail;
}
CUdevice cu_dev;
res = self->cuDeviceGet(&cu_dev, 0);
if(res != CUDA_SUCCESS) {
const char *err_str = "unknown";
self->cuGetErrorString(res, &err_str);
fprintf(stderr, "gsr error: gsr_cuda_load failed: unable to get CUDA device, error: %s (result: %d)\n", err_str, res);
goto fail;
}
res = self->cuCtxCreate_v2(&self->cu_ctx, CU_CTX_SCHED_AUTO, cu_dev);
if(res != CUDA_SUCCESS) {
const char *err_str = "unknown";
self->cuGetErrorString(res, &err_str);
fprintf(stderr, "gsr error: gsr_cuda_load failed: unable to create CUDA context, error: %s (result: %d)\n", err_str, res);
goto fail;
}
if(self->do_overclock) {
if(gsr_overclock_load(&self->overclock, display))
gsr_overclock_start(&self->overclock);
else
fprintf(stderr, "gsr warning: gsr_cuda_load: failed to load xnvctrl, failed to overclock memory transfer rate\n");
}
self->library = lib;
return true;
fail:
dlclose(lib);
memset(self, 0, sizeof(gsr_cuda));
return false;
}
void gsr_cuda_unload(gsr_cuda *self) {
if(self->library) {
if(self->cu_ctx) {
self->cuCtxDestroy_v2(self->cu_ctx);
self->cu_ctx = 0;
}
dlclose(self->library);
}
if(self->do_overclock && self->overclock.xnvctrl.library) {
gsr_overclock_stop(&self->overclock);
gsr_overclock_unload(&self->overclock);
}
memset(self, 0, sizeof(gsr_cuda));
}
|
