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
281
282
283
284
285
286
287
288
289
290
291
292
|
/* BGP Keepalives.
* Implements a producer thread to generate BGP keepalives for peers.
* Copyright (C) 2017 Cumulus Networks, Inc.
* Quentin Young
*
* This file is part of FRRouting.
*
* FRRouting is free software; you can redistribute it and/or modify it under
* the terms of the GNU General Public License as published by the Free
* Software Foundation; either version 2, or (at your option) any later
* version.
*
* FRRouting 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; see the file COPYING; if not, write to the Free Software
* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
*/
/* clang-format off */
#include <zebra.h>
#include <pthread.h> // for pthread_mutex_lock, pthread_mutex_unlock
#include "frr_pthread.h" // for frr_pthread
#include "hash.h" // for hash, hash_clean, hash_create_size...
#include "log.h" // for zlog_debug
#include "memory.h" // for MTYPE_TMP, XFREE, XCALLOC, XMALLOC
#include "monotime.h" // for monotime, monotime_since
#include "bgpd/bgpd.h" // for peer, PEER_THREAD_KEEPALIVES_ON, peer...
#include "bgpd/bgp_debug.h" // for bgp_debug_neighbor_events
#include "bgpd/bgp_packet.h" // for bgp_keepalive_send
#include "bgpd/bgp_keepalives.h"
/* clang-format on */
/**
* Peer KeepAlive Timer.
* Associates a peer with the time of its last keepalive.
*/
struct pkat {
// the peer to send keepalives to
struct peer *peer;
// absolute time of last keepalive sent
struct timeval last;
};
/* List of peers we are sending keepalives for, and associated mutex. */
static pthread_mutex_t *peerhash_mtx;
static pthread_cond_t *peerhash_cond;
static struct hash *peerhash;
/* Thread control flag. */
bool bgp_keepalives_thread_run = false;
static struct pkat *pkat_new(struct peer *peer)
{
struct pkat *pkat = XMALLOC(MTYPE_TMP, sizeof(struct pkat));
pkat->peer = peer;
monotime(&pkat->last);
return pkat;
}
static void pkat_del(void *pkat)
{
XFREE(MTYPE_TMP, pkat);
}
/*
* Callback for hash_iterate. Determines if a peer needs a keepalive and if so,
* generates and sends it.
*
* For any given peer, if the elapsed time since its last keepalive exceeds its
* configured keepalive timer, a keepalive is sent to the peer and its
* last-sent time is reset. Additionally, If the elapsed time does not exceed
* the configured keepalive timer, but the time until the next keepalive is due
* is within a hardcoded tolerance, a keepalive is sent as if the configured
* timer was exceeded. Doing this helps alleviate nanosecond sleeps between
* ticks by grouping together peers who are due for keepalives at roughly the
* same time. This tolerance value is arbitrarily chosen to be 100ms.
*
* In addition, this function calculates the maximum amount of time that the
* keepalive thread can sleep before another tick needs to take place. This is
* equivalent to shortest time until a keepalive is due for any one peer.
*
* @return maximum time to wait until next update (0 if infinity)
*/
static void peer_process(struct hash_backet *hb, void *arg)
{
struct pkat *pkat = hb->data;
struct timeval *next_update = arg;
static struct timeval elapsed; // elapsed time since keepalive
static struct timeval ka = {0}; // peer->v_keepalive as a timeval
static struct timeval diff; // ka - elapsed
static struct timeval tolerance = {0, 100000};
// calculate elapsed time since last keepalive
monotime_since(&pkat->last, &elapsed);
// calculate difference between elapsed time and configured time
ka.tv_sec = pkat->peer->v_keepalive;
timersub(&ka, &elapsed, &diff);
int send_keepalive =
elapsed.tv_sec >= ka.tv_sec || timercmp(&diff, &tolerance, <);
if (send_keepalive) {
if (bgp_debug_neighbor_events(pkat->peer))
zlog_debug("%s [FSM] Timer (keepalive timer expire)",
pkat->peer->host);
bgp_keepalive_send(pkat->peer);
monotime(&pkat->last);
memset(&elapsed, 0x00, sizeof(struct timeval));
diff = ka; // time until next keepalive == peer keepalive time
}
// if calculated next update for this peer < current delay, use it
if (next_update->tv_sec <= 0 || timercmp(&diff, next_update, <))
*next_update = diff;
}
static int peer_hash_cmp(const void *f, const void *s)
{
const struct pkat *p1 = f;
const struct pkat *p2 = s;
return p1->peer == p2->peer;
}
static unsigned int peer_hash_key(void *arg)
{
struct pkat *pkat = arg;
return (uintptr_t)pkat->peer;
}
void bgp_keepalives_init()
{
peerhash_mtx = XCALLOC(MTYPE_TMP, sizeof(pthread_mutex_t));
peerhash_cond = XCALLOC(MTYPE_TMP, sizeof(pthread_cond_t));
// initialize mutex
pthread_mutex_init(peerhash_mtx, NULL);
// use monotonic clock with condition variable
pthread_condattr_t attrs;
pthread_condattr_init(&attrs);
pthread_condattr_setclock(&attrs, CLOCK_MONOTONIC);
pthread_cond_init(peerhash_cond, &attrs);
pthread_condattr_destroy(&attrs);
// initialize peer hashtable
peerhash = hash_create_size(2048, peer_hash_key, peer_hash_cmp, NULL);
}
static void bgp_keepalives_finish(void *arg)
{
bgp_keepalives_thread_run = false;
if (peerhash) {
hash_clean(peerhash, pkat_del);
hash_free(peerhash);
}
peerhash = NULL;
pthread_mutex_unlock(peerhash_mtx);
pthread_mutex_destroy(peerhash_mtx);
pthread_cond_destroy(peerhash_cond);
XFREE(MTYPE_TMP, peerhash_mtx);
XFREE(MTYPE_TMP, peerhash_cond);
}
/**
* Entry function for peer keepalive generation pthread.
*
* bgp_keepalives_init() must be called prior to this.
*/
void *bgp_keepalives_start(void *arg)
{
struct timeval currtime = {0, 0};
struct timeval aftertime = {0, 0};
struct timeval next_update = {0, 0};
struct timespec next_update_ts = {0, 0};
pthread_mutex_lock(peerhash_mtx);
// register cleanup handler
pthread_cleanup_push(&bgp_keepalives_finish, NULL);
bgp_keepalives_thread_run = true;
while (bgp_keepalives_thread_run) {
if (peerhash->count > 0)
pthread_cond_timedwait(peerhash_cond, peerhash_mtx,
&next_update_ts);
else
while (peerhash->count == 0
&& bgp_keepalives_thread_run)
pthread_cond_wait(peerhash_cond, peerhash_mtx);
monotime(&currtime);
next_update.tv_sec = -1;
hash_iterate(peerhash, peer_process, &next_update);
if (next_update.tv_sec == -1)
memset(&next_update, 0x00, sizeof(next_update));
monotime_since(&currtime, &aftertime);
timeradd(&currtime, &next_update, &next_update);
TIMEVAL_TO_TIMESPEC(&next_update, &next_update_ts);
}
// clean up
pthread_cleanup_pop(1);
return NULL;
}
/* --- thread external functions ------------------------------------------- */
void bgp_keepalives_on(struct peer *peer)
{
/* placeholder bucket data to use for fast key lookups */
static struct pkat holder = {0};
if (!peerhash_mtx) {
zlog_warn("%s: call bgp_keepalives_init() first", __func__);
return;
}
pthread_mutex_lock(peerhash_mtx);
{
holder.peer = peer;
if (!hash_lookup(peerhash, &holder)) {
struct pkat *pkat = pkat_new(peer);
hash_get(peerhash, pkat, hash_alloc_intern);
peer_lock(peer);
}
SET_FLAG(peer->thread_flags, PEER_THREAD_KEEPALIVES_ON);
}
pthread_mutex_unlock(peerhash_mtx);
bgp_keepalives_wake();
}
void bgp_keepalives_off(struct peer *peer)
{
/* placeholder bucket data to use for fast key lookups */
static struct pkat holder = {0};
if (!peerhash_mtx) {
zlog_warn("%s: call bgp_keepalives_init() first", __func__);
return;
}
pthread_mutex_lock(peerhash_mtx);
{
holder.peer = peer;
struct pkat *res = hash_release(peerhash, &holder);
if (res) {
pkat_del(res);
peer_unlock(peer);
}
UNSET_FLAG(peer->thread_flags, PEER_THREAD_KEEPALIVES_ON);
}
pthread_mutex_unlock(peerhash_mtx);
}
void bgp_keepalives_wake()
{
pthread_mutex_lock(peerhash_mtx);
{
pthread_cond_signal(peerhash_cond);
}
pthread_mutex_unlock(peerhash_mtx);
}
int bgp_keepalives_stop(void **result, struct frr_pthread *fpt)
{
bgp_keepalives_thread_run = false;
bgp_keepalives_wake();
pthread_join(fpt->thread, result);
return 0;
}
|