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
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
|
// SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note
/*
*
* (C) COPYRIGHT 2020-2021 ARM Limited. All rights reserved.
*
* This program is free software and is provided to you under the terms of the
* GNU General Public License version 2 as published by the Free Software
* Foundation, and any use by you of this program is subject to the terms
* of such GNU license.
*
* This program is distributed in the hope that 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.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, you can access it online at
* http://www.gnu.org/licenses/gpl-2.0.html.
*
*/
#include <mali_kbase.h>
#include <mali_kbase_mem_linux.h>
#include <mali_kbase_defs.h>
#include <mali_kbase_trace_gpu_mem.h>
/**
* struct kbase_dma_buf - Object instantiated when a dma-buf imported allocation
* is mapped to GPU for the first time within a process.
* Another instantiation is done for the case when that
* allocation is mapped for the first time to GPU.
*
* @dma_buf: Reference to dma_buf been imported.
* @dma_buf_node: Link node to maintain a rb_tree of kbase_dma_buf.
* @import_count: The number of times the dma_buf was imported.
*/
struct kbase_dma_buf {
struct dma_buf *dma_buf;
struct rb_node dma_buf_node;
u32 import_count;
};
/**
* kbase_delete_dma_buf_mapping - Delete a dma buffer mapping.
*
* @kctx: Pointer to kbase context.
* @dma_buf: Pointer to a dma buffer mapping.
* @tree: Pointer to root of rb_tree containing the dma_buf's mapped.
*
* when we un-map any dma mapping we need to remove them from rb_tree,
* rb_tree is maintained at kbase_device level and kbase_process level
* by passing the root of kbase_device or kbase_process we can remove
* the node from the tree.
*/
static bool kbase_delete_dma_buf_mapping(struct kbase_context *kctx,
struct dma_buf *dma_buf,
struct rb_root *tree)
{
struct kbase_dma_buf *buf_node = NULL;
struct rb_node *node = tree->rb_node;
bool mapping_removed = false;
lockdep_assert_held(&kctx->kbdev->dma_buf_lock);
while (node) {
buf_node = rb_entry(node, struct kbase_dma_buf, dma_buf_node);
if (dma_buf == buf_node->dma_buf) {
WARN_ON(!buf_node->import_count);
buf_node->import_count--;
if (!buf_node->import_count) {
rb_erase(&buf_node->dma_buf_node, tree);
kfree(buf_node);
mapping_removed = true;
}
break;
}
if (dma_buf < buf_node->dma_buf)
node = node->rb_left;
else
node = node->rb_right;
}
WARN_ON(!buf_node);
return mapping_removed;
}
/**
* kbase_capture_dma_buf_mapping - capture a dma buffer mapping.
*
* @kctx: Pointer to kbase context.
* @dma_buf: Pointer to a dma buffer mapping.
* @root: Pointer to root of rb_tree containing the dma_buf's.
*
* We maintain a kbase_device level and kbase_process level rb_tree
* of all unique dma_buf's mapped to gpu memory. So when attach any
* dma_buf add it the rb_tree's. To add the unique mapping we need
* check if the mapping is not a duplicate and then add them.
*/
static bool kbase_capture_dma_buf_mapping(struct kbase_context *kctx,
struct dma_buf *dma_buf,
struct rb_root *root)
{
struct kbase_dma_buf *buf_node = NULL;
struct rb_node *node = root->rb_node;
bool unique_buf_imported = true;
lockdep_assert_held(&kctx->kbdev->dma_buf_lock);
while (node) {
buf_node = rb_entry(node, struct kbase_dma_buf, dma_buf_node);
if (dma_buf == buf_node->dma_buf) {
unique_buf_imported = false;
break;
}
if (dma_buf < buf_node->dma_buf)
node = node->rb_left;
else
node = node->rb_right;
}
if (unique_buf_imported) {
struct kbase_dma_buf *new_buf_node =
kzalloc(sizeof(*new_buf_node), GFP_KERNEL);
if (new_buf_node == NULL) {
dev_err(kctx->kbdev->dev, "Error allocating memory for kbase_dma_buf\n");
/* Dont account for it if we fail to allocate memory */
unique_buf_imported = false;
} else {
struct rb_node **new = &(root->rb_node), *parent = NULL;
new_buf_node->dma_buf = dma_buf;
new_buf_node->import_count = 1;
while (*new) {
struct kbase_dma_buf *new_node;
parent = *new;
new_node = rb_entry(parent, struct kbase_dma_buf,
dma_buf_node);
if (dma_buf < new_node->dma_buf)
new = &(*new)->rb_left;
else
new = &(*new)->rb_right;
}
rb_link_node(&new_buf_node->dma_buf_node, parent, new);
rb_insert_color(&new_buf_node->dma_buf_node, root);
}
} else if (!WARN_ON(!buf_node)) {
buf_node->import_count++;
}
return unique_buf_imported;
}
void kbase_remove_dma_buf_usage(struct kbase_context *kctx,
struct kbase_mem_phy_alloc *alloc)
{
struct kbase_device *kbdev = kctx->kbdev;
bool dev_mapping_removed, prcs_mapping_removed;
mutex_lock(&kbdev->dma_buf_lock);
dev_mapping_removed = kbase_delete_dma_buf_mapping(
kctx, alloc->imported.umm.dma_buf, &kbdev->dma_buf_root);
prcs_mapping_removed = kbase_delete_dma_buf_mapping(
kctx, alloc->imported.umm.dma_buf, &kctx->kprcs->dma_buf_root);
WARN_ON(dev_mapping_removed && !prcs_mapping_removed);
spin_lock(&kbdev->gpu_mem_usage_lock);
if (dev_mapping_removed)
kbdev->total_gpu_pages -= alloc->nents;
if (prcs_mapping_removed)
kctx->kprcs->total_gpu_pages -= alloc->nents;
if (dev_mapping_removed || prcs_mapping_removed)
kbase_trace_gpu_mem_usage(kbdev, kctx);
spin_unlock(&kbdev->gpu_mem_usage_lock);
mutex_unlock(&kbdev->dma_buf_lock);
}
void kbase_add_dma_buf_usage(struct kbase_context *kctx,
struct kbase_mem_phy_alloc *alloc)
{
struct kbase_device *kbdev = kctx->kbdev;
bool unique_dev_dmabuf, unique_prcs_dmabuf;
mutex_lock(&kbdev->dma_buf_lock);
/* add dma_buf to device and process. */
unique_dev_dmabuf = kbase_capture_dma_buf_mapping(
kctx, alloc->imported.umm.dma_buf, &kbdev->dma_buf_root);
unique_prcs_dmabuf = kbase_capture_dma_buf_mapping(
kctx, alloc->imported.umm.dma_buf, &kctx->kprcs->dma_buf_root);
WARN_ON(unique_dev_dmabuf && !unique_prcs_dmabuf);
spin_lock(&kbdev->gpu_mem_usage_lock);
if (unique_dev_dmabuf)
kbdev->total_gpu_pages += alloc->nents;
if (unique_prcs_dmabuf)
kctx->kprcs->total_gpu_pages += alloc->nents;
if (unique_prcs_dmabuf || unique_dev_dmabuf)
kbase_trace_gpu_mem_usage(kbdev, kctx);
spin_unlock(&kbdev->gpu_mem_usage_lock);
mutex_unlock(&kbdev->dma_buf_lock);
}
|
