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
|
/*
* Copyright (c) 2020 Cisco and/or its affiliates.
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at:
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "vppinfra/string.h"
#include <vnet/vnet.h>
#include <vlibapi/api.h>
#include <vlibmemory/api.h>
#include <vnet/plugin/plugin.h>
#include <vpp/app/version.h>
#include <linux/limits.h>
#include <sys/ioctl.h>
#include <perfmon/perfmon.h>
static_always_inline void
perfmon_read_pmcs (u64 *counters, int *pmc_index, u8 n_counters)
{
switch (n_counters)
{
default:
case 7:
counters[6] = _rdpmc (pmc_index[6]);
case 6:
counters[5] = _rdpmc (pmc_index[5]);
case 5:
counters[4] = _rdpmc (pmc_index[4]);
case 4:
counters[3] = _rdpmc (pmc_index[3]);
case 3:
counters[2] = _rdpmc (pmc_index[2]);
case 2:
counters[1] = _rdpmc (pmc_index[1]);
case 1:
counters[0] = _rdpmc (pmc_index[0]);
break;
}
}
static_always_inline int
perfmon_calc_mmap_offset (perfmon_thread_runtime_t *tr, u8 i)
{
return (int) (tr->mmap_pages[i]->index + tr->mmap_pages[i]->offset);
}
static_always_inline int
perfmon_metric_index (perfmon_bundle_t *b, u8 i)
{
return (int) (b->metrics[i]);
}
uword
perfmon_dispatch_wrapper_mmap (vlib_main_t *vm, vlib_node_runtime_t *node,
vlib_frame_t *frame)
{
perfmon_main_t *pm = &perfmon_main;
perfmon_thread_runtime_t *rt =
vec_elt_at_index (pm->thread_runtimes, vm->thread_index);
perfmon_node_stats_t *s =
vec_elt_at_index (rt->node_stats, node->node_index);
u8 n_events = rt->n_events;
u64 before[PERF_MAX_EVENTS];
u64 after[PERF_MAX_EVENTS];
int pmc_index[PERF_MAX_EVENTS];
uword rv;
clib_prefetch_load (s);
switch (n_events)
{
default:
case 7:
pmc_index[6] = perfmon_calc_mmap_offset (rt, 6);
case 6:
pmc_index[5] = perfmon_calc_mmap_offset (rt, 5);
case 5:
pmc_index[4] = perfmon_calc_mmap_offset (rt, 4);
case 4:
pmc_index[3] = perfmon_calc_mmap_offset (rt, 3);
case 3:
pmc_index[2] = perfmon_calc_mmap_offset (rt, 2);
case 2:
pmc_index[1] = perfmon_calc_mmap_offset (rt, 1);
case 1:
pmc_index[0] = perfmon_calc_mmap_offset (rt, 0);
break;
}
perfmon_read_pmcs (&before[0], pmc_index, n_events);
rv = node->function (vm, node, frame);
perfmon_read_pmcs (&after[0], pmc_index, n_events);
if (rv == 0)
return rv;
s->n_calls += 1;
s->n_packets += rv;
for (int i = 0; i < n_events; i++)
s->value[i] += after[i] - before[i];
return rv;
}
uword
perfmon_dispatch_wrapper_metrics (vlib_main_t *vm, vlib_node_runtime_t *node,
vlib_frame_t *frame)
{
perfmon_main_t *pm = &perfmon_main;
perfmon_thread_runtime_t *rt =
vec_elt_at_index (pm->thread_runtimes, vm->thread_index);
perfmon_node_stats_t *s =
vec_elt_at_index (rt->node_stats, node->node_index);
u8 n_events = rt->n_events;
u64 before[PERF_MAX_EVENTS];
int pmc_index[PERF_MAX_EVENTS];
uword rv;
clib_prefetch_load (s);
switch (n_events)
{
default:
case 7:
pmc_index[6] = perfmon_metric_index (rt->bundle, 6);
case 6:
pmc_index[5] = perfmon_metric_index (rt->bundle, 5);
case 5:
pmc_index[4] = perfmon_metric_index (rt->bundle, 4);
case 4:
pmc_index[3] = perfmon_metric_index (rt->bundle, 3);
case 3:
pmc_index[2] = perfmon_metric_index (rt->bundle, 2);
case 2:
pmc_index[1] = perfmon_metric_index (rt->bundle, 1);
case 1:
pmc_index[0] = perfmon_metric_index (rt->bundle, 0);
break;
}
perfmon_read_pmcs (&before[0], pmc_index, n_events);
rv = node->function (vm, node, frame);
clib_memcpy_fast (&s->t[0].value[0], &before, sizeof (before));
perfmon_read_pmcs (&s->t[1].value[0], pmc_index, n_events);
if (rv == 0)
return rv;
s->n_calls += 1;
s->n_packets += rv;
return rv;
}
|
