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
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
|
/*
* drivers/video/tegra/host/gk20a/semaphore_gk20a.c
*
* GK20A Semaphores
*
* Copyright (c) 2014-2015, NVIDIA CORPORATION. All rights reserved.
*
* This program is free software; you can redistribute it and/or modify it
* under the terms and conditions of the GNU General Public License,
* version 2, as published by the Free Software Foundation.
*
* This program is distributed in the hope it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
* more details.
*/
#include "semaphore_gk20a.h"
#include <linux/dma-mapping.h>
#include <linux/slab.h>
#include "gk20a.h"
#include "mm_gk20a.h"
static const int SEMAPHORE_SIZE = 16;
struct gk20a_semaphore_pool *gk20a_semaphore_pool_alloc(struct gk20a *g,
const char *unique_name, size_t capacity)
{
struct gk20a_semaphore_pool *p;
p = kzalloc(sizeof(*p), GFP_KERNEL);
if (!p)
return NULL;
kref_init(&p->ref);
INIT_LIST_HEAD(&p->maps);
mutex_init(&p->maps_mutex);
p->g = g;
/* Alloc one 4k page of semaphore per channel. */
if (gk20a_gmmu_alloc(g, roundup(capacity * SEMAPHORE_SIZE, PAGE_SIZE),
&p->mem))
goto clean_up;
/* Sacrifice one semaphore in the name of returning error codes. */
if (gk20a_allocator_init(&p->alloc, unique_name,
SEMAPHORE_SIZE, p->mem.size - SEMAPHORE_SIZE,
SEMAPHORE_SIZE))
goto clean_up;
gk20a_dbg_info("cpuva=%p iova=%llx phys=%llx", p->mem.cpu_va,
(u64)sg_dma_address(p->mem.sgt->sgl),
(u64)sg_phys(p->mem.sgt->sgl));
return p;
clean_up:
if (p->mem.size)
gk20a_gmmu_free(p->g, &p->mem);
kfree(p);
return NULL;
}
static void gk20a_semaphore_pool_free(struct kref *ref)
{
struct gk20a_semaphore_pool *p =
container_of(ref, struct gk20a_semaphore_pool, ref);
mutex_lock(&p->maps_mutex);
WARN_ON(!list_empty(&p->maps));
mutex_unlock(&p->maps_mutex);
gk20a_gmmu_free(p->g, &p->mem);
gk20a_allocator_destroy(&p->alloc);
kfree(p);
}
static void gk20a_semaphore_pool_get(struct gk20a_semaphore_pool *p)
{
kref_get(&p->ref);
}
void gk20a_semaphore_pool_put(struct gk20a_semaphore_pool *p)
{
kref_put(&p->ref, gk20a_semaphore_pool_free);
}
static struct gk20a_semaphore_pool_map *
gk20a_semaphore_pool_find_map_locked(struct gk20a_semaphore_pool *p,
struct vm_gk20a *vm)
{
struct gk20a_semaphore_pool_map *map, *found = NULL;
list_for_each_entry(map, &p->maps, list) {
if (map->vm == vm) {
found = map;
break;
}
}
return found;
}
int gk20a_semaphore_pool_map(struct gk20a_semaphore_pool *p,
struct vm_gk20a *vm,
enum gk20a_mem_rw_flag rw_flag)
{
struct gk20a_semaphore_pool_map *map;
map = kzalloc(sizeof(*map), GFP_KERNEL);
if (!map)
return -ENOMEM;
map->vm = vm;
map->rw_flag = rw_flag;
map->gpu_va = gk20a_gmmu_map(vm, &p->mem.sgt, p->mem.size,
0/*uncached*/, rw_flag,
false);
if (!map->gpu_va) {
kfree(map);
return -ENOMEM;
}
gk20a_vm_get(vm);
mutex_lock(&p->maps_mutex);
WARN_ON(gk20a_semaphore_pool_find_map_locked(p, vm));
list_add(&map->list, &p->maps);
mutex_unlock(&p->maps_mutex);
return 0;
}
void gk20a_semaphore_pool_unmap(struct gk20a_semaphore_pool *p,
struct vm_gk20a *vm)
{
struct gk20a_semaphore_pool_map *map;
WARN_ON(!vm);
mutex_lock(&p->maps_mutex);
map = gk20a_semaphore_pool_find_map_locked(p, vm);
if (map) {
gk20a_gmmu_unmap(vm, map->gpu_va, p->mem.size, map->rw_flag);
gk20a_vm_put(vm);
list_del(&map->list);
kfree(map);
}
mutex_unlock(&p->maps_mutex);
}
u64 gk20a_semaphore_pool_gpu_va(struct gk20a_semaphore_pool *p,
struct vm_gk20a *vm)
{
struct gk20a_semaphore_pool_map *map;
u64 gpu_va = 0;
mutex_lock(&p->maps_mutex);
map = gk20a_semaphore_pool_find_map_locked(p, vm);
if (map)
gpu_va = map->gpu_va;
mutex_unlock(&p->maps_mutex);
return gpu_va;
}
struct gk20a_semaphore *gk20a_semaphore_alloc(struct gk20a_semaphore_pool *pool)
{
struct gk20a_semaphore *s;
s = kzalloc(sizeof(*s), GFP_KERNEL);
if (!s)
return NULL;
s->offset = gk20a_balloc(&pool->alloc, SEMAPHORE_SIZE);
if (!s->offset) {
gk20a_err(dev_from_gk20a(pool->g),
"failed to allocate semaphore");
kfree(s);
return NULL;
}
gk20a_semaphore_pool_get(pool);
s->pool = pool;
kref_init(&s->ref);
/* Initially acquired. */
gk20a_mem_wr(s->pool->g, &s->pool->mem, s->offset, 0);
gk20a_dbg_info("created semaphore offset=%d, value=%d",
s->offset,
gk20a_mem_rd(s->pool->g, &s->pool->mem, s->offset));
return s;
}
static void gk20a_semaphore_free(struct kref *ref)
{
struct gk20a_semaphore *s =
container_of(ref, struct gk20a_semaphore, ref);
gk20a_bfree(&s->pool->alloc, s->offset);
gk20a_semaphore_pool_put(s->pool);
kfree(s);
}
void gk20a_semaphore_put(struct gk20a_semaphore *s)
{
kref_put(&s->ref, gk20a_semaphore_free);
}
void gk20a_semaphore_get(struct gk20a_semaphore *s)
{
kref_get(&s->ref);
}
|
