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
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
|
/*
* Copyright 2019 Advanced Micro Devices, Inc.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
* OTHER DEALINGS IN THE SOFTWARE.
*
* Author: Jonathan Kim <jonathan.kim@amd.com>
*
*/
#include <linux/perf_event.h>
#include <linux/init.h>
#include "amdgpu.h"
#include "amdgpu_pmu.h"
#include "df_v3_6.h"
#define PMU_NAME_SIZE 32
/* record to keep track of pmu entry per pmu type per device */
struct amdgpu_pmu_entry {
struct list_head entry;
struct amdgpu_device *adev;
struct pmu pmu;
unsigned int pmu_perf_type;
};
static LIST_HEAD(amdgpu_pmu_list);
/* initialize perf counter */
static int amdgpu_perf_event_init(struct perf_event *event)
{
struct hw_perf_event *hwc = &event->hw;
/* test the event attr type check for PMU enumeration */
if (event->attr.type != event->pmu->type)
return -ENOENT;
/* update the hw_perf_event struct with config data */
hwc->conf = event->attr.config;
return 0;
}
/* start perf counter */
static void amdgpu_perf_start(struct perf_event *event, int flags)
{
struct hw_perf_event *hwc = &event->hw;
struct amdgpu_pmu_entry *pe = container_of(event->pmu,
struct amdgpu_pmu_entry,
pmu);
if (WARN_ON_ONCE(!(hwc->state & PERF_HES_STOPPED)))
return;
WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE));
hwc->state = 0;
switch (pe->pmu_perf_type) {
case PERF_TYPE_AMDGPU_DF:
if (!(flags & PERF_EF_RELOAD))
pe->adev->df_funcs->pmc_start(pe->adev, hwc->conf, 1);
pe->adev->df_funcs->pmc_start(pe->adev, hwc->conf, 0);
break;
default:
break;
}
perf_event_update_userpage(event);
}
/* read perf counter */
static void amdgpu_perf_read(struct perf_event *event)
{
struct hw_perf_event *hwc = &event->hw;
struct amdgpu_pmu_entry *pe = container_of(event->pmu,
struct amdgpu_pmu_entry,
pmu);
u64 count, prev;
do {
prev = local64_read(&hwc->prev_count);
switch (pe->pmu_perf_type) {
case PERF_TYPE_AMDGPU_DF:
pe->adev->df_funcs->pmc_get_count(pe->adev, hwc->conf,
&count);
break;
default:
count = 0;
break;
};
} while (local64_cmpxchg(&hwc->prev_count, prev, count) != prev);
local64_add(count - prev, &event->count);
}
/* stop perf counter */
static void amdgpu_perf_stop(struct perf_event *event, int flags)
{
struct hw_perf_event *hwc = &event->hw;
struct amdgpu_pmu_entry *pe = container_of(event->pmu,
struct amdgpu_pmu_entry,
pmu);
if (hwc->state & PERF_HES_UPTODATE)
return;
switch (pe->pmu_perf_type) {
case PERF_TYPE_AMDGPU_DF:
pe->adev->df_funcs->pmc_stop(pe->adev, hwc->conf, 0);
break;
default:
break;
};
WARN_ON_ONCE(hwc->state & PERF_HES_STOPPED);
hwc->state |= PERF_HES_STOPPED;
if (hwc->state & PERF_HES_UPTODATE)
return;
amdgpu_perf_read(event);
hwc->state |= PERF_HES_UPTODATE;
}
/* add perf counter */
static int amdgpu_perf_add(struct perf_event *event, int flags)
{
struct hw_perf_event *hwc = &event->hw;
int retval;
struct amdgpu_pmu_entry *pe = container_of(event->pmu,
struct amdgpu_pmu_entry,
pmu);
event->hw.state = PERF_HES_UPTODATE | PERF_HES_STOPPED;
switch (pe->pmu_perf_type) {
case PERF_TYPE_AMDGPU_DF:
retval = pe->adev->df_funcs->pmc_start(pe->adev, hwc->conf, 1);
break;
default:
return 0;
};
if (retval)
return retval;
if (flags & PERF_EF_START)
amdgpu_perf_start(event, PERF_EF_RELOAD);
return retval;
}
/* delete perf counter */
static void amdgpu_perf_del(struct perf_event *event, int flags)
{
struct hw_perf_event *hwc = &event->hw;
struct amdgpu_pmu_entry *pe = container_of(event->pmu,
struct amdgpu_pmu_entry,
pmu);
amdgpu_perf_stop(event, PERF_EF_UPDATE);
switch (pe->pmu_perf_type) {
case PERF_TYPE_AMDGPU_DF:
pe->adev->df_funcs->pmc_stop(pe->adev, hwc->conf, 1);
break;
default:
break;
};
perf_event_update_userpage(event);
}
/* vega20 pmus */
/* init pmu tracking per pmu type */
static int init_pmu_by_type(struct amdgpu_device *adev,
const struct attribute_group *attr_groups[],
char *pmu_type_name, char *pmu_file_prefix,
unsigned int pmu_perf_type,
unsigned int num_counters)
{
char pmu_name[PMU_NAME_SIZE];
struct amdgpu_pmu_entry *pmu_entry;
int ret = 0;
pmu_entry = kzalloc(sizeof(struct amdgpu_pmu_entry), GFP_KERNEL);
if (!pmu_entry)
return -ENOMEM;
pmu_entry->adev = adev;
pmu_entry->pmu = (struct pmu){
.event_init = amdgpu_perf_event_init,
.add = amdgpu_perf_add,
.del = amdgpu_perf_del,
.start = amdgpu_perf_start,
.stop = amdgpu_perf_stop,
.read = amdgpu_perf_read,
.task_ctx_nr = perf_invalid_context,
};
pmu_entry->pmu.attr_groups = attr_groups;
pmu_entry->pmu_perf_type = pmu_perf_type;
snprintf(pmu_name, PMU_NAME_SIZE, "%s_%d",
pmu_file_prefix, adev->ddev->primary->index);
ret = perf_pmu_register(&pmu_entry->pmu, pmu_name, -1);
if (ret) {
kfree(pmu_entry);
pr_warn("Error initializing AMDGPU %s PMUs.\n", pmu_type_name);
return ret;
}
pr_info("Detected AMDGPU %s Counters. # of Counters = %d.\n",
pmu_type_name, num_counters);
list_add_tail(&pmu_entry->entry, &amdgpu_pmu_list);
return 0;
}
/* init amdgpu_pmu */
int amdgpu_pmu_init(struct amdgpu_device *adev)
{
int ret = 0;
switch (adev->asic_type) {
case CHIP_VEGA20:
/* init df */
ret = init_pmu_by_type(adev, df_v3_6_attr_groups,
"DF", "amdgpu_df", PERF_TYPE_AMDGPU_DF,
DF_V3_6_MAX_COUNTERS);
/* other pmu types go here*/
break;
default:
return 0;
}
return 0;
}
/* destroy all pmu data associated with target device */
void amdgpu_pmu_fini(struct amdgpu_device *adev)
{
struct amdgpu_pmu_entry *pe, *temp;
list_for_each_entry_safe(pe, temp, &amdgpu_pmu_list, entry) {
if (pe->adev == adev) {
list_del(&pe->entry);
perf_pmu_unregister(&pe->pmu);
kfree(pe);
}
}
}
|