]>
Commit | Line | Data |
---|---|---|
718e3744 | 1 | /* Thread management routine |
2 | * Copyright (C) 1998, 2000 Kunihiro Ishiguro <kunihiro@zebra.org> | |
3 | * | |
4 | * This file is part of GNU Zebra. | |
5 | * | |
6 | * GNU Zebra is free software; you can redistribute it and/or modify it | |
7 | * under the terms of the GNU General Public License as published by the | |
8 | * Free Software Foundation; either version 2, or (at your option) any | |
9 | * later version. | |
10 | * | |
11 | * GNU Zebra is distributed in the hope that it will be useful, but | |
12 | * WITHOUT ANY WARRANTY; without even the implied warranty of | |
13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
14 | * General Public License for more details. | |
15 | * | |
896014f4 DL |
16 | * You should have received a copy of the GNU General Public License along |
17 | * with this program; see the file COPYING; if not, write to the Free Software | |
18 | * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA | |
718e3744 | 19 | */ |
20 | ||
21 | /* #define DEBUG */ | |
22 | ||
23 | #include <zebra.h> | |
308d14ae | 24 | #include <sys/resource.h> |
718e3744 | 25 | |
26 | #include "thread.h" | |
27 | #include "memory.h" | |
28 | #include "log.h" | |
e04ab74d | 29 | #include "hash.h" |
4becea72 | 30 | #include "pqueue.h" |
e04ab74d | 31 | #include "command.h" |
05c447dd | 32 | #include "sigevent.h" |
3bf2673b | 33 | #include "network.h" |
bd74dc61 | 34 | #include "jhash.h" |
fbcac826 | 35 | #include "frratomic.h" |
d6be5fb9 | 36 | |
d62a17ae | 37 | DEFINE_MTYPE_STATIC(LIB, THREAD, "Thread") |
4a1ab8e4 | 38 | DEFINE_MTYPE_STATIC(LIB, THREAD_MASTER, "Thread master") |
d62a17ae | 39 | DEFINE_MTYPE_STATIC(LIB, THREAD_STATS, "Thread stats") |
4a1ab8e4 | 40 | |
3b96b781 HT |
41 | #if defined(__APPLE__) |
42 | #include <mach/mach.h> | |
43 | #include <mach/mach_time.h> | |
44 | #endif | |
45 | ||
d62a17ae | 46 | #define AWAKEN(m) \ |
47 | do { \ | |
48 | static unsigned char wakebyte = 0x01; \ | |
49 | write(m->io_pipe[1], &wakebyte, 1); \ | |
50 | } while (0); | |
3bf2673b | 51 | |
62f44022 | 52 | /* control variable for initializer */ |
e0bebc7c | 53 | pthread_once_t init_once = PTHREAD_ONCE_INIT; |
e0bebc7c | 54 | pthread_key_t thread_current; |
6b0655a2 | 55 | |
62f44022 QY |
56 | pthread_mutex_t masters_mtx = PTHREAD_MUTEX_INITIALIZER; |
57 | static struct list *masters; | |
58 | ||
6b0655a2 | 59 | |
62f44022 | 60 | /* CLI start ---------------------------------------------------------------- */ |
d62a17ae | 61 | static unsigned int cpu_record_hash_key(struct cpu_thread_history *a) |
e04ab74d | 62 | { |
996c9314 | 63 | int size = sizeof(&a->func); |
bd74dc61 DS |
64 | |
65 | return jhash(&a->func, size, 0); | |
e04ab74d | 66 | } |
67 | ||
d62a17ae | 68 | static int cpu_record_hash_cmp(const struct cpu_thread_history *a, |
69 | const struct cpu_thread_history *b) | |
e04ab74d | 70 | { |
d62a17ae | 71 | return a->func == b->func; |
e04ab74d | 72 | } |
73 | ||
d62a17ae | 74 | static void *cpu_record_hash_alloc(struct cpu_thread_history *a) |
e04ab74d | 75 | { |
d62a17ae | 76 | struct cpu_thread_history *new; |
77 | new = XCALLOC(MTYPE_THREAD_STATS, sizeof(struct cpu_thread_history)); | |
78 | new->func = a->func; | |
79 | new->funcname = a->funcname; | |
80 | return new; | |
e04ab74d | 81 | } |
82 | ||
d62a17ae | 83 | static void cpu_record_hash_free(void *a) |
228da428 | 84 | { |
d62a17ae | 85 | struct cpu_thread_history *hist = a; |
86 | ||
87 | XFREE(MTYPE_THREAD_STATS, hist); | |
228da428 CC |
88 | } |
89 | ||
d62a17ae | 90 | static void vty_out_cpu_thread_history(struct vty *vty, |
91 | struct cpu_thread_history *a) | |
e04ab74d | 92 | { |
d1667f53 | 93 | vty_out(vty, "%5d %10lu.%03lu %9u %8lu %9lu %8lu %9lu", a->total_active, |
d62a17ae | 94 | a->cpu.total / 1000, a->cpu.total % 1000, a->total_calls, |
95 | a->cpu.total / a->total_calls, a->cpu.max, | |
96 | a->real.total / a->total_calls, a->real.max); | |
97 | vty_out(vty, " %c%c%c%c%c %s\n", | |
98 | a->types & (1 << THREAD_READ) ? 'R' : ' ', | |
99 | a->types & (1 << THREAD_WRITE) ? 'W' : ' ', | |
100 | a->types & (1 << THREAD_TIMER) ? 'T' : ' ', | |
101 | a->types & (1 << THREAD_EVENT) ? 'E' : ' ', | |
102 | a->types & (1 << THREAD_EXECUTE) ? 'X' : ' ', a->funcname); | |
e04ab74d | 103 | } |
104 | ||
d62a17ae | 105 | static void cpu_record_hash_print(struct hash_backet *bucket, void *args[]) |
e04ab74d | 106 | { |
d62a17ae | 107 | struct cpu_thread_history *totals = args[0]; |
fbcac826 | 108 | struct cpu_thread_history copy; |
d62a17ae | 109 | struct vty *vty = args[1]; |
fbcac826 | 110 | uint8_t *filter = args[2]; |
d62a17ae | 111 | |
112 | struct cpu_thread_history *a = bucket->data; | |
113 | ||
fbcac826 QY |
114 | copy.total_active = |
115 | atomic_load_explicit(&a->total_active, memory_order_seq_cst); | |
116 | copy.total_calls = | |
117 | atomic_load_explicit(&a->total_calls, memory_order_seq_cst); | |
118 | copy.cpu.total = | |
119 | atomic_load_explicit(&a->cpu.total, memory_order_seq_cst); | |
120 | copy.cpu.max = atomic_load_explicit(&a->cpu.max, memory_order_seq_cst); | |
121 | copy.real.total = | |
122 | atomic_load_explicit(&a->real.total, memory_order_seq_cst); | |
123 | copy.real.max = | |
124 | atomic_load_explicit(&a->real.max, memory_order_seq_cst); | |
125 | copy.types = atomic_load_explicit(&a->types, memory_order_seq_cst); | |
126 | copy.funcname = a->funcname; | |
127 | ||
128 | if (!(copy.types & *filter)) | |
d62a17ae | 129 | return; |
fbcac826 QY |
130 | |
131 | vty_out_cpu_thread_history(vty, ©); | |
132 | totals->total_active += copy.total_active; | |
133 | totals->total_calls += copy.total_calls; | |
134 | totals->real.total += copy.real.total; | |
135 | if (totals->real.max < copy.real.max) | |
136 | totals->real.max = copy.real.max; | |
137 | totals->cpu.total += copy.cpu.total; | |
138 | if (totals->cpu.max < copy.cpu.max) | |
139 | totals->cpu.max = copy.cpu.max; | |
e04ab74d | 140 | } |
141 | ||
fbcac826 | 142 | static void cpu_record_print(struct vty *vty, uint8_t filter) |
e04ab74d | 143 | { |
d62a17ae | 144 | struct cpu_thread_history tmp; |
145 | void *args[3] = {&tmp, vty, &filter}; | |
146 | struct thread_master *m; | |
147 | struct listnode *ln; | |
148 | ||
149 | memset(&tmp, 0, sizeof tmp); | |
150 | tmp.funcname = "TOTAL"; | |
151 | tmp.types = filter; | |
152 | ||
153 | pthread_mutex_lock(&masters_mtx); | |
154 | { | |
155 | for (ALL_LIST_ELEMENTS_RO(masters, ln, m)) { | |
156 | const char *name = m->name ? m->name : "main"; | |
157 | ||
158 | char underline[strlen(name) + 1]; | |
159 | memset(underline, '-', sizeof(underline)); | |
4f113d60 | 160 | underline[sizeof(underline) - 1] = '\0'; |
d62a17ae | 161 | |
162 | vty_out(vty, "\n"); | |
163 | vty_out(vty, "Showing statistics for pthread %s\n", | |
164 | name); | |
165 | vty_out(vty, "-------------------------------%s\n", | |
166 | underline); | |
167 | vty_out(vty, "%21s %18s %18s\n", "", | |
168 | "CPU (user+system):", "Real (wall-clock):"); | |
169 | vty_out(vty, | |
170 | "Active Runtime(ms) Invoked Avg uSec Max uSecs"); | |
171 | vty_out(vty, " Avg uSec Max uSecs"); | |
172 | vty_out(vty, " Type Thread\n"); | |
173 | ||
174 | if (m->cpu_record->count) | |
175 | hash_iterate( | |
176 | m->cpu_record, | |
177 | (void (*)(struct hash_backet *, | |
178 | void *))cpu_record_hash_print, | |
179 | args); | |
180 | else | |
181 | vty_out(vty, "No data to display yet.\n"); | |
182 | ||
183 | vty_out(vty, "\n"); | |
184 | } | |
185 | } | |
186 | pthread_mutex_unlock(&masters_mtx); | |
187 | ||
188 | vty_out(vty, "\n"); | |
189 | vty_out(vty, "Total thread statistics\n"); | |
190 | vty_out(vty, "-------------------------\n"); | |
191 | vty_out(vty, "%21s %18s %18s\n", "", | |
192 | "CPU (user+system):", "Real (wall-clock):"); | |
193 | vty_out(vty, "Active Runtime(ms) Invoked Avg uSec Max uSecs"); | |
194 | vty_out(vty, " Avg uSec Max uSecs"); | |
195 | vty_out(vty, " Type Thread\n"); | |
196 | ||
197 | if (tmp.total_calls > 0) | |
198 | vty_out_cpu_thread_history(vty, &tmp); | |
e04ab74d | 199 | } |
200 | ||
d62a17ae | 201 | static void cpu_record_hash_clear(struct hash_backet *bucket, void *args[]) |
e276eb82 | 202 | { |
fbcac826 | 203 | uint8_t *filter = args[0]; |
d62a17ae | 204 | struct hash *cpu_record = args[1]; |
205 | ||
206 | struct cpu_thread_history *a = bucket->data; | |
62f44022 | 207 | |
d62a17ae | 208 | if (!(a->types & *filter)) |
209 | return; | |
f48f65d2 | 210 | |
d62a17ae | 211 | hash_release(cpu_record, bucket->data); |
e276eb82 PJ |
212 | } |
213 | ||
fbcac826 | 214 | static void cpu_record_clear(uint8_t filter) |
e276eb82 | 215 | { |
fbcac826 | 216 | uint8_t *tmp = &filter; |
d62a17ae | 217 | struct thread_master *m; |
218 | struct listnode *ln; | |
219 | ||
220 | pthread_mutex_lock(&masters_mtx); | |
221 | { | |
222 | for (ALL_LIST_ELEMENTS_RO(masters, ln, m)) { | |
223 | pthread_mutex_lock(&m->mtx); | |
224 | { | |
225 | void *args[2] = {tmp, m->cpu_record}; | |
226 | hash_iterate( | |
227 | m->cpu_record, | |
228 | (void (*)(struct hash_backet *, | |
229 | void *))cpu_record_hash_clear, | |
230 | args); | |
231 | } | |
232 | pthread_mutex_unlock(&m->mtx); | |
233 | } | |
234 | } | |
235 | pthread_mutex_unlock(&masters_mtx); | |
62f44022 QY |
236 | } |
237 | ||
fbcac826 | 238 | static uint8_t parse_filter(const char *filterstr) |
62f44022 | 239 | { |
d62a17ae | 240 | int i = 0; |
241 | int filter = 0; | |
242 | ||
243 | while (filterstr[i] != '\0') { | |
244 | switch (filterstr[i]) { | |
245 | case 'r': | |
246 | case 'R': | |
247 | filter |= (1 << THREAD_READ); | |
248 | break; | |
249 | case 'w': | |
250 | case 'W': | |
251 | filter |= (1 << THREAD_WRITE); | |
252 | break; | |
253 | case 't': | |
254 | case 'T': | |
255 | filter |= (1 << THREAD_TIMER); | |
256 | break; | |
257 | case 'e': | |
258 | case 'E': | |
259 | filter |= (1 << THREAD_EVENT); | |
260 | break; | |
261 | case 'x': | |
262 | case 'X': | |
263 | filter |= (1 << THREAD_EXECUTE); | |
264 | break; | |
265 | default: | |
266 | break; | |
267 | } | |
268 | ++i; | |
269 | } | |
270 | return filter; | |
62f44022 QY |
271 | } |
272 | ||
273 | DEFUN (show_thread_cpu, | |
274 | show_thread_cpu_cmd, | |
275 | "show thread cpu [FILTER]", | |
276 | SHOW_STR | |
277 | "Thread information\n" | |
278 | "Thread CPU usage\n" | |
279 | "Display filter (rwtexb)\n") | |
280 | { | |
fbcac826 | 281 | uint8_t filter = (uint8_t)-1U; |
d62a17ae | 282 | int idx = 0; |
283 | ||
284 | if (argv_find(argv, argc, "FILTER", &idx)) { | |
285 | filter = parse_filter(argv[idx]->arg); | |
286 | if (!filter) { | |
287 | vty_out(vty, | |
288 | "Invalid filter \"%s\" specified; must contain at least" | |
289 | "one of 'RWTEXB'\n", | |
290 | argv[idx]->arg); | |
291 | return CMD_WARNING; | |
292 | } | |
293 | } | |
294 | ||
295 | cpu_record_print(vty, filter); | |
296 | return CMD_SUCCESS; | |
e276eb82 PJ |
297 | } |
298 | ||
49d41a26 DS |
299 | DEFUN (clear_thread_cpu, |
300 | clear_thread_cpu_cmd, | |
301 | "clear thread cpu [FILTER]", | |
62f44022 | 302 | "Clear stored data in all pthreads\n" |
49d41a26 DS |
303 | "Thread information\n" |
304 | "Thread CPU usage\n" | |
305 | "Display filter (rwtexb)\n") | |
e276eb82 | 306 | { |
fbcac826 | 307 | uint8_t filter = (uint8_t)-1U; |
d62a17ae | 308 | int idx = 0; |
309 | ||
310 | if (argv_find(argv, argc, "FILTER", &idx)) { | |
311 | filter = parse_filter(argv[idx]->arg); | |
312 | if (!filter) { | |
313 | vty_out(vty, | |
314 | "Invalid filter \"%s\" specified; must contain at least" | |
315 | "one of 'RWTEXB'\n", | |
316 | argv[idx]->arg); | |
317 | return CMD_WARNING; | |
318 | } | |
319 | } | |
320 | ||
321 | cpu_record_clear(filter); | |
322 | return CMD_SUCCESS; | |
e276eb82 | 323 | } |
6b0655a2 | 324 | |
d62a17ae | 325 | void thread_cmd_init(void) |
0b84f294 | 326 | { |
d62a17ae | 327 | install_element(VIEW_NODE, &show_thread_cpu_cmd); |
328 | install_element(ENABLE_NODE, &clear_thread_cpu_cmd); | |
0b84f294 | 329 | } |
62f44022 QY |
330 | /* CLI end ------------------------------------------------------------------ */ |
331 | ||
0b84f294 | 332 | |
d62a17ae | 333 | static int thread_timer_cmp(void *a, void *b) |
4becea72 | 334 | { |
d62a17ae | 335 | struct thread *thread_a = a; |
336 | struct thread *thread_b = b; | |
337 | ||
338 | if (timercmp(&thread_a->u.sands, &thread_b->u.sands, <)) | |
339 | return -1; | |
340 | if (timercmp(&thread_a->u.sands, &thread_b->u.sands, >)) | |
341 | return 1; | |
342 | return 0; | |
4becea72 CF |
343 | } |
344 | ||
d62a17ae | 345 | static void thread_timer_update(void *node, int actual_position) |
4becea72 | 346 | { |
d62a17ae | 347 | struct thread *thread = node; |
4becea72 | 348 | |
d62a17ae | 349 | thread->index = actual_position; |
4becea72 CF |
350 | } |
351 | ||
d62a17ae | 352 | static void cancelreq_del(void *cr) |
63ccb9cb | 353 | { |
d62a17ae | 354 | XFREE(MTYPE_TMP, cr); |
63ccb9cb QY |
355 | } |
356 | ||
e0bebc7c | 357 | /* initializer, only ever called once */ |
d62a17ae | 358 | static void initializer() |
e0bebc7c | 359 | { |
d62a17ae | 360 | pthread_key_create(&thread_current, NULL); |
e0bebc7c QY |
361 | } |
362 | ||
d62a17ae | 363 | struct thread_master *thread_master_create(const char *name) |
718e3744 | 364 | { |
d62a17ae | 365 | struct thread_master *rv; |
366 | struct rlimit limit; | |
367 | ||
368 | pthread_once(&init_once, &initializer); | |
369 | ||
370 | rv = XCALLOC(MTYPE_THREAD_MASTER, sizeof(struct thread_master)); | |
371 | if (rv == NULL) | |
372 | return NULL; | |
373 | ||
374 | /* Initialize master mutex */ | |
375 | pthread_mutex_init(&rv->mtx, NULL); | |
376 | pthread_cond_init(&rv->cancel_cond, NULL); | |
377 | ||
378 | /* Set name */ | |
379 | rv->name = name ? XSTRDUP(MTYPE_THREAD_MASTER, name) : NULL; | |
380 | ||
381 | /* Initialize I/O task data structures */ | |
382 | getrlimit(RLIMIT_NOFILE, &limit); | |
383 | rv->fd_limit = (int)limit.rlim_cur; | |
384 | rv->read = | |
385 | XCALLOC(MTYPE_THREAD, sizeof(struct thread *) * rv->fd_limit); | |
386 | if (rv->read == NULL) { | |
387 | XFREE(MTYPE_THREAD_MASTER, rv); | |
388 | return NULL; | |
389 | } | |
390 | rv->write = | |
391 | XCALLOC(MTYPE_THREAD, sizeof(struct thread *) * rv->fd_limit); | |
392 | if (rv->write == NULL) { | |
393 | XFREE(MTYPE_THREAD, rv->read); | |
394 | XFREE(MTYPE_THREAD_MASTER, rv); | |
395 | return NULL; | |
396 | } | |
397 | ||
bd74dc61 | 398 | rv->cpu_record = hash_create_size( |
996c9314 | 399 | 8, (unsigned int (*)(void *))cpu_record_hash_key, |
bd74dc61 DS |
400 | (int (*)(const void *, const void *))cpu_record_hash_cmp, |
401 | "Thread Hash"); | |
d62a17ae | 402 | |
403 | ||
404 | /* Initialize the timer queues */ | |
405 | rv->timer = pqueue_create(); | |
406 | rv->timer->cmp = thread_timer_cmp; | |
407 | rv->timer->update = thread_timer_update; | |
408 | ||
409 | /* Initialize thread_fetch() settings */ | |
410 | rv->spin = true; | |
411 | rv->handle_signals = true; | |
412 | ||
413 | /* Set pthread owner, should be updated by actual owner */ | |
414 | rv->owner = pthread_self(); | |
415 | rv->cancel_req = list_new(); | |
416 | rv->cancel_req->del = cancelreq_del; | |
417 | rv->canceled = true; | |
418 | ||
419 | /* Initialize pipe poker */ | |
420 | pipe(rv->io_pipe); | |
421 | set_nonblocking(rv->io_pipe[0]); | |
422 | set_nonblocking(rv->io_pipe[1]); | |
423 | ||
424 | /* Initialize data structures for poll() */ | |
425 | rv->handler.pfdsize = rv->fd_limit; | |
426 | rv->handler.pfdcount = 0; | |
427 | rv->handler.pfds = XCALLOC(MTYPE_THREAD_MASTER, | |
428 | sizeof(struct pollfd) * rv->handler.pfdsize); | |
429 | rv->handler.copy = XCALLOC(MTYPE_THREAD_MASTER, | |
430 | sizeof(struct pollfd) * rv->handler.pfdsize); | |
431 | ||
eff09c66 | 432 | /* add to list of threadmasters */ |
d62a17ae | 433 | pthread_mutex_lock(&masters_mtx); |
434 | { | |
eff09c66 QY |
435 | if (!masters) |
436 | masters = list_new(); | |
437 | ||
d62a17ae | 438 | listnode_add(masters, rv); |
439 | } | |
440 | pthread_mutex_unlock(&masters_mtx); | |
441 | ||
442 | return rv; | |
718e3744 | 443 | } |
444 | ||
d8a8a8de QY |
445 | void thread_master_set_name(struct thread_master *master, const char *name) |
446 | { | |
447 | pthread_mutex_lock(&master->mtx); | |
448 | { | |
449 | if (master->name) | |
450 | XFREE(MTYPE_THREAD_MASTER, master->name); | |
451 | master->name = XSTRDUP(MTYPE_THREAD_MASTER, name); | |
452 | } | |
453 | pthread_mutex_unlock(&master->mtx); | |
454 | } | |
455 | ||
718e3744 | 456 | /* Add a new thread to the list. */ |
d62a17ae | 457 | static void thread_list_add(struct thread_list *list, struct thread *thread) |
718e3744 | 458 | { |
d62a17ae | 459 | thread->next = NULL; |
460 | thread->prev = list->tail; | |
461 | if (list->tail) | |
462 | list->tail->next = thread; | |
463 | else | |
464 | list->head = thread; | |
465 | list->tail = thread; | |
466 | list->count++; | |
718e3744 | 467 | } |
468 | ||
718e3744 | 469 | /* Delete a thread from the list. */ |
d62a17ae | 470 | static struct thread *thread_list_delete(struct thread_list *list, |
471 | struct thread *thread) | |
718e3744 | 472 | { |
d62a17ae | 473 | if (thread->next) |
474 | thread->next->prev = thread->prev; | |
475 | else | |
476 | list->tail = thread->prev; | |
477 | if (thread->prev) | |
478 | thread->prev->next = thread->next; | |
479 | else | |
480 | list->head = thread->next; | |
481 | thread->next = thread->prev = NULL; | |
482 | list->count--; | |
483 | return thread; | |
718e3744 | 484 | } |
485 | ||
495f0b13 | 486 | /* Thread list is empty or not. */ |
d62a17ae | 487 | static int thread_empty(struct thread_list *list) |
495f0b13 | 488 | { |
d62a17ae | 489 | return list->head ? 0 : 1; |
495f0b13 DS |
490 | } |
491 | ||
492 | /* Delete top of the list and return it. */ | |
d62a17ae | 493 | static struct thread *thread_trim_head(struct thread_list *list) |
495f0b13 | 494 | { |
d62a17ae | 495 | if (!thread_empty(list)) |
496 | return thread_list_delete(list, list->head); | |
497 | return NULL; | |
495f0b13 DS |
498 | } |
499 | ||
718e3744 | 500 | /* Move thread to unuse list. */ |
d62a17ae | 501 | static void thread_add_unuse(struct thread_master *m, struct thread *thread) |
718e3744 | 502 | { |
d62a17ae | 503 | assert(m != NULL && thread != NULL); |
504 | assert(thread->next == NULL); | |
505 | assert(thread->prev == NULL); | |
506 | thread->ref = NULL; | |
507 | ||
508 | thread->type = THREAD_UNUSED; | |
509 | thread->hist->total_active--; | |
510 | thread_list_add(&m->unuse, thread); | |
718e3744 | 511 | } |
512 | ||
513 | /* Free all unused thread. */ | |
d62a17ae | 514 | static void thread_list_free(struct thread_master *m, struct thread_list *list) |
718e3744 | 515 | { |
d62a17ae | 516 | struct thread *t; |
517 | struct thread *next; | |
518 | ||
519 | for (t = list->head; t; t = next) { | |
520 | next = t->next; | |
521 | XFREE(MTYPE_THREAD, t); | |
522 | list->count--; | |
523 | m->alloc--; | |
524 | } | |
718e3744 | 525 | } |
526 | ||
d62a17ae | 527 | static void thread_array_free(struct thread_master *m, |
528 | struct thread **thread_array) | |
308d14ae | 529 | { |
d62a17ae | 530 | struct thread *t; |
531 | int index; | |
532 | ||
533 | for (index = 0; index < m->fd_limit; ++index) { | |
534 | t = thread_array[index]; | |
535 | if (t) { | |
536 | thread_array[index] = NULL; | |
537 | XFREE(MTYPE_THREAD, t); | |
538 | m->alloc--; | |
539 | } | |
540 | } | |
541 | XFREE(MTYPE_THREAD, thread_array); | |
308d14ae DV |
542 | } |
543 | ||
d62a17ae | 544 | static void thread_queue_free(struct thread_master *m, struct pqueue *queue) |
4becea72 | 545 | { |
d62a17ae | 546 | int i; |
4becea72 | 547 | |
d62a17ae | 548 | for (i = 0; i < queue->size; i++) |
549 | XFREE(MTYPE_THREAD, queue->array[i]); | |
4becea72 | 550 | |
d62a17ae | 551 | m->alloc -= queue->size; |
552 | pqueue_delete(queue); | |
4becea72 CF |
553 | } |
554 | ||
495f0b13 DS |
555 | /* |
556 | * thread_master_free_unused | |
557 | * | |
558 | * As threads are finished with they are put on the | |
559 | * unuse list for later reuse. | |
560 | * If we are shutting down, Free up unused threads | |
561 | * So we can see if we forget to shut anything off | |
562 | */ | |
d62a17ae | 563 | void thread_master_free_unused(struct thread_master *m) |
495f0b13 | 564 | { |
d62a17ae | 565 | pthread_mutex_lock(&m->mtx); |
566 | { | |
567 | struct thread *t; | |
568 | while ((t = thread_trim_head(&m->unuse)) != NULL) { | |
569 | pthread_mutex_destroy(&t->mtx); | |
570 | XFREE(MTYPE_THREAD, t); | |
571 | } | |
572 | } | |
573 | pthread_mutex_unlock(&m->mtx); | |
495f0b13 DS |
574 | } |
575 | ||
718e3744 | 576 | /* Stop thread scheduler. */ |
d62a17ae | 577 | void thread_master_free(struct thread_master *m) |
718e3744 | 578 | { |
d62a17ae | 579 | pthread_mutex_lock(&masters_mtx); |
580 | { | |
581 | listnode_delete(masters, m); | |
eff09c66 | 582 | if (masters->count == 0) { |
acdf5e25 | 583 | list_delete_and_null(&masters); |
eff09c66 | 584 | } |
d62a17ae | 585 | } |
586 | pthread_mutex_unlock(&masters_mtx); | |
587 | ||
588 | thread_array_free(m, m->read); | |
589 | thread_array_free(m, m->write); | |
590 | thread_queue_free(m, m->timer); | |
591 | thread_list_free(m, &m->event); | |
592 | thread_list_free(m, &m->ready); | |
593 | thread_list_free(m, &m->unuse); | |
594 | pthread_mutex_destroy(&m->mtx); | |
33844bbe | 595 | pthread_cond_destroy(&m->cancel_cond); |
d62a17ae | 596 | close(m->io_pipe[0]); |
597 | close(m->io_pipe[1]); | |
affe9e99 | 598 | list_delete_and_null(&m->cancel_req); |
1a0a92ea | 599 | m->cancel_req = NULL; |
d62a17ae | 600 | |
601 | hash_clean(m->cpu_record, cpu_record_hash_free); | |
602 | hash_free(m->cpu_record); | |
603 | m->cpu_record = NULL; | |
604 | ||
4e1000a1 QY |
605 | if (m->name) |
606 | XFREE(MTYPE_THREAD_MASTER, m->name); | |
d62a17ae | 607 | XFREE(MTYPE_THREAD_MASTER, m->handler.pfds); |
608 | XFREE(MTYPE_THREAD_MASTER, m->handler.copy); | |
609 | XFREE(MTYPE_THREAD_MASTER, m); | |
718e3744 | 610 | } |
611 | ||
718e3744 | 612 | /* Return remain time in second. */ |
d62a17ae | 613 | unsigned long thread_timer_remain_second(struct thread *thread) |
718e3744 | 614 | { |
d62a17ae | 615 | int64_t remain; |
1189d95f | 616 | |
d62a17ae | 617 | pthread_mutex_lock(&thread->mtx); |
618 | { | |
619 | remain = monotime_until(&thread->u.sands, NULL) / 1000000LL; | |
620 | } | |
621 | pthread_mutex_unlock(&thread->mtx); | |
1189d95f | 622 | |
d62a17ae | 623 | return remain < 0 ? 0 : remain; |
718e3744 | 624 | } |
625 | ||
9c7753e4 DL |
626 | #define debugargdef const char *funcname, const char *schedfrom, int fromln |
627 | #define debugargpass funcname, schedfrom, fromln | |
e04ab74d | 628 | |
d62a17ae | 629 | struct timeval thread_timer_remain(struct thread *thread) |
6ac44687 | 630 | { |
d62a17ae | 631 | struct timeval remain; |
632 | pthread_mutex_lock(&thread->mtx); | |
633 | { | |
634 | monotime_until(&thread->u.sands, &remain); | |
635 | } | |
636 | pthread_mutex_unlock(&thread->mtx); | |
637 | return remain; | |
6ac44687 CF |
638 | } |
639 | ||
718e3744 | 640 | /* Get new thread. */ |
d7c0a89a | 641 | static struct thread *thread_get(struct thread_master *m, uint8_t type, |
d62a17ae | 642 | int (*func)(struct thread *), void *arg, |
643 | debugargdef) | |
718e3744 | 644 | { |
d62a17ae | 645 | struct thread *thread = thread_trim_head(&m->unuse); |
646 | struct cpu_thread_history tmp; | |
647 | ||
648 | if (!thread) { | |
649 | thread = XCALLOC(MTYPE_THREAD, sizeof(struct thread)); | |
650 | /* mutex only needs to be initialized at struct creation. */ | |
651 | pthread_mutex_init(&thread->mtx, NULL); | |
652 | m->alloc++; | |
653 | } | |
654 | ||
655 | thread->type = type; | |
656 | thread->add_type = type; | |
657 | thread->master = m; | |
658 | thread->arg = arg; | |
659 | thread->index = -1; | |
660 | thread->yield = THREAD_YIELD_TIME_SLOT; /* default */ | |
661 | thread->ref = NULL; | |
662 | ||
663 | /* | |
664 | * So if the passed in funcname is not what we have | |
665 | * stored that means the thread->hist needs to be | |
666 | * updated. We keep the last one around in unused | |
667 | * under the assumption that we are probably | |
668 | * going to immediately allocate the same | |
669 | * type of thread. | |
670 | * This hopefully saves us some serious | |
671 | * hash_get lookups. | |
672 | */ | |
673 | if (thread->funcname != funcname || thread->func != func) { | |
674 | tmp.func = func; | |
675 | tmp.funcname = funcname; | |
676 | thread->hist = | |
677 | hash_get(m->cpu_record, &tmp, | |
678 | (void *(*)(void *))cpu_record_hash_alloc); | |
679 | } | |
680 | thread->hist->total_active++; | |
681 | thread->func = func; | |
682 | thread->funcname = funcname; | |
683 | thread->schedfrom = schedfrom; | |
684 | thread->schedfrom_line = fromln; | |
685 | ||
686 | return thread; | |
718e3744 | 687 | } |
688 | ||
d62a17ae | 689 | static int fd_poll(struct thread_master *m, struct pollfd *pfds, nfds_t pfdsize, |
690 | nfds_t count, const struct timeval *timer_wait) | |
209a72a6 | 691 | { |
d62a17ae | 692 | /* If timer_wait is null here, that means poll() should block |
693 | * indefinitely, | |
694 | * unless the thread_master has overriden it by setting | |
695 | * ->selectpoll_timeout. | |
696 | * If the value is positive, it specifies the maximum number of | |
697 | * milliseconds | |
698 | * to wait. If the timeout is -1, it specifies that we should never wait | |
699 | * and | |
700 | * always return immediately even if no event is detected. If the value | |
701 | * is | |
702 | * zero, the behavior is default. */ | |
703 | int timeout = -1; | |
704 | ||
705 | /* number of file descriptors with events */ | |
706 | int num; | |
707 | ||
708 | if (timer_wait != NULL | |
709 | && m->selectpoll_timeout == 0) // use the default value | |
710 | timeout = (timer_wait->tv_sec * 1000) | |
711 | + (timer_wait->tv_usec / 1000); | |
712 | else if (m->selectpoll_timeout > 0) // use the user's timeout | |
713 | timeout = m->selectpoll_timeout; | |
714 | else if (m->selectpoll_timeout | |
715 | < 0) // effect a poll (return immediately) | |
716 | timeout = 0; | |
717 | ||
718 | /* add poll pipe poker */ | |
719 | assert(count + 1 < pfdsize); | |
720 | pfds[count].fd = m->io_pipe[0]; | |
721 | pfds[count].events = POLLIN; | |
722 | pfds[count].revents = 0x00; | |
723 | ||
724 | num = poll(pfds, count + 1, timeout); | |
725 | ||
726 | unsigned char trash[64]; | |
727 | if (num > 0 && pfds[count].revents != 0 && num--) | |
728 | while (read(m->io_pipe[0], &trash, sizeof(trash)) > 0) | |
729 | ; | |
730 | ||
731 | return num; | |
209a72a6 DS |
732 | } |
733 | ||
718e3744 | 734 | /* Add new read thread. */ |
d62a17ae | 735 | struct thread *funcname_thread_add_read_write(int dir, struct thread_master *m, |
736 | int (*func)(struct thread *), | |
737 | void *arg, int fd, | |
738 | struct thread **t_ptr, | |
739 | debugargdef) | |
718e3744 | 740 | { |
d62a17ae | 741 | struct thread *thread = NULL; |
742 | ||
743 | pthread_mutex_lock(&m->mtx); | |
744 | { | |
745 | if (t_ptr | |
746 | && *t_ptr) // thread is already scheduled; don't reschedule | |
747 | { | |
748 | pthread_mutex_unlock(&m->mtx); | |
749 | return NULL; | |
750 | } | |
751 | ||
752 | /* default to a new pollfd */ | |
753 | nfds_t queuepos = m->handler.pfdcount; | |
754 | ||
755 | /* if we already have a pollfd for our file descriptor, find and | |
756 | * use it */ | |
757 | for (nfds_t i = 0; i < m->handler.pfdcount; i++) | |
758 | if (m->handler.pfds[i].fd == fd) { | |
759 | queuepos = i; | |
760 | break; | |
761 | } | |
762 | ||
763 | /* make sure we have room for this fd + pipe poker fd */ | |
764 | assert(queuepos + 1 < m->handler.pfdsize); | |
765 | ||
766 | thread = thread_get(m, dir, func, arg, debugargpass); | |
767 | ||
768 | m->handler.pfds[queuepos].fd = fd; | |
769 | m->handler.pfds[queuepos].events |= | |
770 | (dir == THREAD_READ ? POLLIN : POLLOUT); | |
771 | ||
772 | if (queuepos == m->handler.pfdcount) | |
773 | m->handler.pfdcount++; | |
774 | ||
775 | if (thread) { | |
776 | pthread_mutex_lock(&thread->mtx); | |
777 | { | |
778 | thread->u.fd = fd; | |
779 | if (dir == THREAD_READ) | |
780 | m->read[thread->u.fd] = thread; | |
781 | else | |
782 | m->write[thread->u.fd] = thread; | |
783 | } | |
784 | pthread_mutex_unlock(&thread->mtx); | |
785 | ||
786 | if (t_ptr) { | |
787 | *t_ptr = thread; | |
788 | thread->ref = t_ptr; | |
789 | } | |
790 | } | |
791 | ||
792 | AWAKEN(m); | |
793 | } | |
794 | pthread_mutex_unlock(&m->mtx); | |
795 | ||
796 | return thread; | |
718e3744 | 797 | } |
798 | ||
56a94b36 | 799 | static struct thread * |
d62a17ae | 800 | funcname_thread_add_timer_timeval(struct thread_master *m, |
801 | int (*func)(struct thread *), int type, | |
802 | void *arg, struct timeval *time_relative, | |
803 | struct thread **t_ptr, debugargdef) | |
718e3744 | 804 | { |
d62a17ae | 805 | struct thread *thread; |
806 | struct pqueue *queue; | |
807 | ||
808 | assert(m != NULL); | |
809 | ||
810 | assert(type == THREAD_TIMER); | |
811 | assert(time_relative); | |
812 | ||
813 | pthread_mutex_lock(&m->mtx); | |
814 | { | |
815 | if (t_ptr | |
816 | && *t_ptr) // thread is already scheduled; don't reschedule | |
817 | { | |
818 | pthread_mutex_unlock(&m->mtx); | |
819 | return NULL; | |
820 | } | |
821 | ||
822 | queue = m->timer; | |
823 | thread = thread_get(m, type, func, arg, debugargpass); | |
824 | ||
825 | pthread_mutex_lock(&thread->mtx); | |
826 | { | |
827 | monotime(&thread->u.sands); | |
828 | timeradd(&thread->u.sands, time_relative, | |
829 | &thread->u.sands); | |
830 | pqueue_enqueue(thread, queue); | |
831 | if (t_ptr) { | |
832 | *t_ptr = thread; | |
833 | thread->ref = t_ptr; | |
834 | } | |
835 | } | |
836 | pthread_mutex_unlock(&thread->mtx); | |
837 | ||
838 | AWAKEN(m); | |
839 | } | |
840 | pthread_mutex_unlock(&m->mtx); | |
841 | ||
842 | return thread; | |
9e867fe6 | 843 | } |
844 | ||
98c91ac6 | 845 | |
846 | /* Add timer event thread. */ | |
d62a17ae | 847 | struct thread *funcname_thread_add_timer(struct thread_master *m, |
848 | int (*func)(struct thread *), | |
849 | void *arg, long timer, | |
850 | struct thread **t_ptr, debugargdef) | |
9e867fe6 | 851 | { |
d62a17ae | 852 | struct timeval trel; |
9e867fe6 | 853 | |
d62a17ae | 854 | assert(m != NULL); |
9e867fe6 | 855 | |
d62a17ae | 856 | trel.tv_sec = timer; |
857 | trel.tv_usec = 0; | |
9e867fe6 | 858 | |
d62a17ae | 859 | return funcname_thread_add_timer_timeval(m, func, THREAD_TIMER, arg, |
860 | &trel, t_ptr, debugargpass); | |
98c91ac6 | 861 | } |
9e867fe6 | 862 | |
98c91ac6 | 863 | /* Add timer event thread with "millisecond" resolution */ |
d62a17ae | 864 | struct thread *funcname_thread_add_timer_msec(struct thread_master *m, |
865 | int (*func)(struct thread *), | |
866 | void *arg, long timer, | |
867 | struct thread **t_ptr, | |
868 | debugargdef) | |
98c91ac6 | 869 | { |
d62a17ae | 870 | struct timeval trel; |
9e867fe6 | 871 | |
d62a17ae | 872 | assert(m != NULL); |
718e3744 | 873 | |
d62a17ae | 874 | trel.tv_sec = timer / 1000; |
875 | trel.tv_usec = 1000 * (timer % 1000); | |
98c91ac6 | 876 | |
d62a17ae | 877 | return funcname_thread_add_timer_timeval(m, func, THREAD_TIMER, arg, |
878 | &trel, t_ptr, debugargpass); | |
a48b4e6d | 879 | } |
880 | ||
d03c4cbd | 881 | /* Add timer event thread with "millisecond" resolution */ |
d62a17ae | 882 | struct thread *funcname_thread_add_timer_tv(struct thread_master *m, |
883 | int (*func)(struct thread *), | |
884 | void *arg, struct timeval *tv, | |
885 | struct thread **t_ptr, debugargdef) | |
d03c4cbd | 886 | { |
d62a17ae | 887 | return funcname_thread_add_timer_timeval(m, func, THREAD_TIMER, arg, tv, |
888 | t_ptr, debugargpass); | |
d03c4cbd DL |
889 | } |
890 | ||
718e3744 | 891 | /* Add simple event thread. */ |
d62a17ae | 892 | struct thread *funcname_thread_add_event(struct thread_master *m, |
893 | int (*func)(struct thread *), | |
894 | void *arg, int val, | |
895 | struct thread **t_ptr, debugargdef) | |
718e3744 | 896 | { |
d62a17ae | 897 | struct thread *thread; |
898 | ||
899 | assert(m != NULL); | |
900 | ||
901 | pthread_mutex_lock(&m->mtx); | |
902 | { | |
903 | if (t_ptr | |
904 | && *t_ptr) // thread is already scheduled; don't reschedule | |
905 | { | |
906 | pthread_mutex_unlock(&m->mtx); | |
907 | return NULL; | |
908 | } | |
909 | ||
910 | thread = thread_get(m, THREAD_EVENT, func, arg, debugargpass); | |
911 | pthread_mutex_lock(&thread->mtx); | |
912 | { | |
913 | thread->u.val = val; | |
914 | thread_list_add(&m->event, thread); | |
915 | } | |
916 | pthread_mutex_unlock(&thread->mtx); | |
917 | ||
918 | if (t_ptr) { | |
919 | *t_ptr = thread; | |
920 | thread->ref = t_ptr; | |
921 | } | |
922 | ||
923 | AWAKEN(m); | |
924 | } | |
925 | pthread_mutex_unlock(&m->mtx); | |
926 | ||
927 | return thread; | |
718e3744 | 928 | } |
929 | ||
63ccb9cb QY |
930 | /* Thread cancellation ------------------------------------------------------ */ |
931 | ||
8797240e QY |
932 | /** |
933 | * NOT's out the .events field of pollfd corresponding to the given file | |
934 | * descriptor. The event to be NOT'd is passed in the 'state' parameter. | |
935 | * | |
936 | * This needs to happen for both copies of pollfd's. See 'thread_fetch' | |
937 | * implementation for details. | |
938 | * | |
939 | * @param master | |
940 | * @param fd | |
941 | * @param state the event to cancel. One or more (OR'd together) of the | |
942 | * following: | |
943 | * - POLLIN | |
944 | * - POLLOUT | |
945 | */ | |
d62a17ae | 946 | static void thread_cancel_rw(struct thread_master *master, int fd, short state) |
0a95a0d0 | 947 | { |
42d74538 QY |
948 | bool found = false; |
949 | ||
d62a17ae | 950 | /* Cancel POLLHUP too just in case some bozo set it */ |
951 | state |= POLLHUP; | |
952 | ||
953 | /* find the index of corresponding pollfd */ | |
954 | nfds_t i; | |
955 | ||
956 | for (i = 0; i < master->handler.pfdcount; i++) | |
42d74538 QY |
957 | if (master->handler.pfds[i].fd == fd) { |
958 | found = true; | |
d62a17ae | 959 | break; |
42d74538 QY |
960 | } |
961 | ||
962 | if (!found) { | |
963 | zlog_debug( | |
964 | "[!] Received cancellation request for nonexistent rw job"); | |
965 | zlog_debug("[!] threadmaster: %s | fd: %d", | |
996c9314 | 966 | master->name ? master->name : "", fd); |
42d74538 QY |
967 | return; |
968 | } | |
d62a17ae | 969 | |
970 | /* NOT out event. */ | |
971 | master->handler.pfds[i].events &= ~(state); | |
972 | ||
973 | /* If all events are canceled, delete / resize the pollfd array. */ | |
974 | if (master->handler.pfds[i].events == 0) { | |
975 | memmove(master->handler.pfds + i, master->handler.pfds + i + 1, | |
976 | (master->handler.pfdcount - i - 1) | |
977 | * sizeof(struct pollfd)); | |
978 | master->handler.pfdcount--; | |
979 | } | |
980 | ||
981 | /* If we have the same pollfd in the copy, perform the same operations, | |
982 | * otherwise return. */ | |
983 | if (i >= master->handler.copycount) | |
984 | return; | |
985 | ||
986 | master->handler.copy[i].events &= ~(state); | |
987 | ||
988 | if (master->handler.copy[i].events == 0) { | |
989 | memmove(master->handler.copy + i, master->handler.copy + i + 1, | |
990 | (master->handler.copycount - i - 1) | |
991 | * sizeof(struct pollfd)); | |
992 | master->handler.copycount--; | |
993 | } | |
0a95a0d0 DS |
994 | } |
995 | ||
1189d95f | 996 | /** |
63ccb9cb | 997 | * Process cancellation requests. |
1189d95f | 998 | * |
63ccb9cb QY |
999 | * This may only be run from the pthread which owns the thread_master. |
1000 | * | |
1001 | * @param master the thread master to process | |
1002 | * @REQUIRE master->mtx | |
1189d95f | 1003 | */ |
d62a17ae | 1004 | static void do_thread_cancel(struct thread_master *master) |
718e3744 | 1005 | { |
d62a17ae | 1006 | struct thread_list *list = NULL; |
1007 | struct pqueue *queue = NULL; | |
1008 | struct thread **thread_array = NULL; | |
1009 | struct thread *thread; | |
1010 | ||
1011 | struct cancel_req *cr; | |
1012 | struct listnode *ln; | |
1013 | for (ALL_LIST_ELEMENTS_RO(master->cancel_req, ln, cr)) { | |
1014 | /* If this is an event object cancellation, linear search | |
1015 | * through event | |
1016 | * list deleting any events which have the specified argument. | |
1017 | * We also | |
1018 | * need to check every thread in the ready queue. */ | |
1019 | if (cr->eventobj) { | |
1020 | struct thread *t; | |
1021 | thread = master->event.head; | |
1022 | ||
1023 | while (thread) { | |
1024 | t = thread; | |
1025 | thread = t->next; | |
1026 | ||
1027 | if (t->arg == cr->eventobj) { | |
1028 | thread_list_delete(&master->event, t); | |
1029 | if (t->ref) | |
1030 | *t->ref = NULL; | |
1031 | thread_add_unuse(master, t); | |
1032 | } | |
1033 | } | |
1034 | ||
1035 | thread = master->ready.head; | |
1036 | while (thread) { | |
1037 | t = thread; | |
1038 | thread = t->next; | |
1039 | ||
1040 | if (t->arg == cr->eventobj) { | |
1041 | thread_list_delete(&master->ready, t); | |
1042 | if (t->ref) | |
1043 | *t->ref = NULL; | |
1044 | thread_add_unuse(master, t); | |
1045 | } | |
1046 | } | |
1047 | continue; | |
1048 | } | |
1049 | ||
1050 | /* The pointer varies depending on whether the cancellation | |
1051 | * request was | |
1052 | * made asynchronously or not. If it was, we need to check | |
1053 | * whether the | |
1054 | * thread even exists anymore before cancelling it. */ | |
1055 | thread = (cr->thread) ? cr->thread : *cr->threadref; | |
1056 | ||
1057 | if (!thread) | |
1058 | continue; | |
1059 | ||
1060 | /* Determine the appropriate queue to cancel the thread from */ | |
1061 | switch (thread->type) { | |
1062 | case THREAD_READ: | |
1063 | thread_cancel_rw(master, thread->u.fd, POLLIN); | |
1064 | thread_array = master->read; | |
1065 | break; | |
1066 | case THREAD_WRITE: | |
1067 | thread_cancel_rw(master, thread->u.fd, POLLOUT); | |
1068 | thread_array = master->write; | |
1069 | break; | |
1070 | case THREAD_TIMER: | |
1071 | queue = master->timer; | |
1072 | break; | |
1073 | case THREAD_EVENT: | |
1074 | list = &master->event; | |
1075 | break; | |
1076 | case THREAD_READY: | |
1077 | list = &master->ready; | |
1078 | break; | |
1079 | default: | |
1080 | continue; | |
1081 | break; | |
1082 | } | |
1083 | ||
1084 | if (queue) { | |
1085 | assert(thread->index >= 0); | |
522f7f99 DS |
1086 | assert(thread == queue->array[thread->index]); |
1087 | pqueue_remove_at(thread->index, queue); | |
d62a17ae | 1088 | } else if (list) { |
1089 | thread_list_delete(list, thread); | |
1090 | } else if (thread_array) { | |
1091 | thread_array[thread->u.fd] = NULL; | |
1092 | } else { | |
1093 | assert(!"Thread should be either in queue or list or array!"); | |
1094 | } | |
1095 | ||
1096 | if (thread->ref) | |
1097 | *thread->ref = NULL; | |
1098 | ||
1099 | thread_add_unuse(thread->master, thread); | |
1100 | } | |
1101 | ||
1102 | /* Delete and free all cancellation requests */ | |
1103 | list_delete_all_node(master->cancel_req); | |
1104 | ||
1105 | /* Wake up any threads which may be blocked in thread_cancel_async() */ | |
1106 | master->canceled = true; | |
1107 | pthread_cond_broadcast(&master->cancel_cond); | |
718e3744 | 1108 | } |
1109 | ||
63ccb9cb QY |
1110 | /** |
1111 | * Cancel any events which have the specified argument. | |
1112 | * | |
1113 | * MT-Unsafe | |
1114 | * | |
1115 | * @param m the thread_master to cancel from | |
1116 | * @param arg the argument passed when creating the event | |
1117 | */ | |
d62a17ae | 1118 | void thread_cancel_event(struct thread_master *master, void *arg) |
718e3744 | 1119 | { |
d62a17ae | 1120 | assert(master->owner == pthread_self()); |
1121 | ||
1122 | pthread_mutex_lock(&master->mtx); | |
1123 | { | |
1124 | struct cancel_req *cr = | |
1125 | XCALLOC(MTYPE_TMP, sizeof(struct cancel_req)); | |
1126 | cr->eventobj = arg; | |
1127 | listnode_add(master->cancel_req, cr); | |
1128 | do_thread_cancel(master); | |
1129 | } | |
1130 | pthread_mutex_unlock(&master->mtx); | |
63ccb9cb | 1131 | } |
1189d95f | 1132 | |
63ccb9cb QY |
1133 | /** |
1134 | * Cancel a specific task. | |
1135 | * | |
1136 | * MT-Unsafe | |
1137 | * | |
1138 | * @param thread task to cancel | |
1139 | */ | |
d62a17ae | 1140 | void thread_cancel(struct thread *thread) |
63ccb9cb | 1141 | { |
d62a17ae | 1142 | assert(thread->master->owner == pthread_self()); |
1143 | ||
1144 | pthread_mutex_lock(&thread->master->mtx); | |
1145 | { | |
1146 | struct cancel_req *cr = | |
1147 | XCALLOC(MTYPE_TMP, sizeof(struct cancel_req)); | |
1148 | cr->thread = thread; | |
1149 | listnode_add(thread->master->cancel_req, cr); | |
1150 | do_thread_cancel(thread->master); | |
1151 | } | |
1152 | pthread_mutex_unlock(&thread->master->mtx); | |
63ccb9cb | 1153 | } |
1189d95f | 1154 | |
63ccb9cb QY |
1155 | /** |
1156 | * Asynchronous cancellation. | |
1157 | * | |
8797240e QY |
1158 | * Called with either a struct thread ** or void * to an event argument, |
1159 | * this function posts the correct cancellation request and blocks until it is | |
1160 | * serviced. | |
63ccb9cb QY |
1161 | * |
1162 | * If the thread is currently running, execution blocks until it completes. | |
1163 | * | |
8797240e QY |
1164 | * The last two parameters are mutually exclusive, i.e. if you pass one the |
1165 | * other must be NULL. | |
1166 | * | |
1167 | * When the cancellation procedure executes on the target thread_master, the | |
1168 | * thread * provided is checked for nullity. If it is null, the thread is | |
1169 | * assumed to no longer exist and the cancellation request is a no-op. Thus | |
1170 | * users of this API must pass a back-reference when scheduling the original | |
1171 | * task. | |
1172 | * | |
63ccb9cb QY |
1173 | * MT-Safe |
1174 | * | |
8797240e QY |
1175 | * @param master the thread master with the relevant event / task |
1176 | * @param thread pointer to thread to cancel | |
1177 | * @param eventobj the event | |
63ccb9cb | 1178 | */ |
d62a17ae | 1179 | void thread_cancel_async(struct thread_master *master, struct thread **thread, |
1180 | void *eventobj) | |
63ccb9cb | 1181 | { |
d62a17ae | 1182 | assert(!(thread && eventobj) && (thread || eventobj)); |
1183 | assert(master->owner != pthread_self()); | |
1184 | ||
1185 | pthread_mutex_lock(&master->mtx); | |
1186 | { | |
1187 | master->canceled = false; | |
1188 | ||
1189 | if (thread) { | |
1190 | struct cancel_req *cr = | |
1191 | XCALLOC(MTYPE_TMP, sizeof(struct cancel_req)); | |
1192 | cr->threadref = thread; | |
1193 | listnode_add(master->cancel_req, cr); | |
1194 | } else if (eventobj) { | |
1195 | struct cancel_req *cr = | |
1196 | XCALLOC(MTYPE_TMP, sizeof(struct cancel_req)); | |
1197 | cr->eventobj = eventobj; | |
1198 | listnode_add(master->cancel_req, cr); | |
1199 | } | |
1200 | AWAKEN(master); | |
1201 | ||
1202 | while (!master->canceled) | |
1203 | pthread_cond_wait(&master->cancel_cond, &master->mtx); | |
1204 | } | |
1205 | pthread_mutex_unlock(&master->mtx); | |
718e3744 | 1206 | } |
63ccb9cb | 1207 | /* ------------------------------------------------------------------------- */ |
718e3744 | 1208 | |
d62a17ae | 1209 | static struct timeval *thread_timer_wait(struct pqueue *queue, |
1210 | struct timeval *timer_val) | |
718e3744 | 1211 | { |
d62a17ae | 1212 | if (queue->size) { |
1213 | struct thread *next_timer = queue->array[0]; | |
1214 | monotime_until(&next_timer->u.sands, timer_val); | |
1215 | return timer_val; | |
1216 | } | |
1217 | return NULL; | |
718e3744 | 1218 | } |
718e3744 | 1219 | |
d62a17ae | 1220 | static struct thread *thread_run(struct thread_master *m, struct thread *thread, |
1221 | struct thread *fetch) | |
718e3744 | 1222 | { |
d62a17ae | 1223 | *fetch = *thread; |
1224 | thread_add_unuse(m, thread); | |
1225 | return fetch; | |
718e3744 | 1226 | } |
1227 | ||
d62a17ae | 1228 | static int thread_process_io_helper(struct thread_master *m, |
1229 | struct thread *thread, short state, int pos) | |
5d4ccd4e | 1230 | { |
d62a17ae | 1231 | struct thread **thread_array; |
1232 | ||
1233 | if (!thread) | |
1234 | return 0; | |
1235 | ||
1236 | if (thread->type == THREAD_READ) | |
1237 | thread_array = m->read; | |
1238 | else | |
1239 | thread_array = m->write; | |
1240 | ||
1241 | thread_array[thread->u.fd] = NULL; | |
1242 | thread_list_add(&m->ready, thread); | |
1243 | thread->type = THREAD_READY; | |
1244 | /* if another pthread scheduled this file descriptor for the event we're | |
1245 | * responding to, no problem; we're getting to it now */ | |
1246 | thread->master->handler.pfds[pos].events &= ~(state); | |
1247 | return 1; | |
5d4ccd4e DS |
1248 | } |
1249 | ||
8797240e QY |
1250 | /** |
1251 | * Process I/O events. | |
1252 | * | |
1253 | * Walks through file descriptor array looking for those pollfds whose .revents | |
1254 | * field has something interesting. Deletes any invalid file descriptors. | |
1255 | * | |
1256 | * @param m the thread master | |
1257 | * @param num the number of active file descriptors (return value of poll()) | |
1258 | */ | |
d62a17ae | 1259 | static void thread_process_io(struct thread_master *m, unsigned int num) |
0a95a0d0 | 1260 | { |
d62a17ae | 1261 | unsigned int ready = 0; |
1262 | struct pollfd *pfds = m->handler.copy; | |
1263 | ||
1264 | for (nfds_t i = 0; i < m->handler.copycount && ready < num; ++i) { | |
1265 | /* no event for current fd? immediately continue */ | |
1266 | if (pfds[i].revents == 0) | |
1267 | continue; | |
1268 | ||
1269 | ready++; | |
1270 | ||
1271 | /* Unless someone has called thread_cancel from another pthread, | |
1272 | * the only | |
1273 | * thing that could have changed in m->handler.pfds while we | |
1274 | * were | |
1275 | * asleep is the .events field in a given pollfd. Barring | |
1276 | * thread_cancel() | |
1277 | * that value should be a superset of the values we have in our | |
1278 | * copy, so | |
1279 | * there's no need to update it. Similarily, barring deletion, | |
1280 | * the fd | |
1281 | * should still be a valid index into the master's pfds. */ | |
1282 | if (pfds[i].revents & (POLLIN | POLLHUP)) | |
1283 | thread_process_io_helper(m, m->read[pfds[i].fd], POLLIN, | |
1284 | i); | |
1285 | if (pfds[i].revents & POLLOUT) | |
1286 | thread_process_io_helper(m, m->write[pfds[i].fd], | |
1287 | POLLOUT, i); | |
1288 | ||
1289 | /* if one of our file descriptors is garbage, remove the same | |
1290 | * from | |
1291 | * both pfds + update sizes and index */ | |
1292 | if (pfds[i].revents & POLLNVAL) { | |
1293 | memmove(m->handler.pfds + i, m->handler.pfds + i + 1, | |
1294 | (m->handler.pfdcount - i - 1) | |
1295 | * sizeof(struct pollfd)); | |
1296 | m->handler.pfdcount--; | |
1297 | ||
1298 | memmove(pfds + i, pfds + i + 1, | |
1299 | (m->handler.copycount - i - 1) | |
1300 | * sizeof(struct pollfd)); | |
1301 | m->handler.copycount--; | |
1302 | ||
1303 | i--; | |
1304 | } | |
1305 | } | |
718e3744 | 1306 | } |
1307 | ||
8b70d0b0 | 1308 | /* Add all timers that have popped to the ready list. */ |
d62a17ae | 1309 | static unsigned int thread_process_timers(struct pqueue *queue, |
1310 | struct timeval *timenow) | |
a48b4e6d | 1311 | { |
d62a17ae | 1312 | struct thread *thread; |
1313 | unsigned int ready = 0; | |
1314 | ||
1315 | while (queue->size) { | |
1316 | thread = queue->array[0]; | |
1317 | if (timercmp(timenow, &thread->u.sands, <)) | |
1318 | return ready; | |
1319 | pqueue_dequeue(queue); | |
1320 | thread->type = THREAD_READY; | |
1321 | thread_list_add(&thread->master->ready, thread); | |
1322 | ready++; | |
1323 | } | |
1324 | return ready; | |
a48b4e6d | 1325 | } |
1326 | ||
2613abe6 | 1327 | /* process a list en masse, e.g. for event thread lists */ |
d62a17ae | 1328 | static unsigned int thread_process(struct thread_list *list) |
2613abe6 | 1329 | { |
d62a17ae | 1330 | struct thread *thread; |
1331 | struct thread *next; | |
1332 | unsigned int ready = 0; | |
1333 | ||
1334 | for (thread = list->head; thread; thread = next) { | |
1335 | next = thread->next; | |
1336 | thread_list_delete(list, thread); | |
1337 | thread->type = THREAD_READY; | |
1338 | thread_list_add(&thread->master->ready, thread); | |
1339 | ready++; | |
1340 | } | |
1341 | return ready; | |
2613abe6 PJ |
1342 | } |
1343 | ||
1344 | ||
718e3744 | 1345 | /* Fetch next ready thread. */ |
d62a17ae | 1346 | struct thread *thread_fetch(struct thread_master *m, struct thread *fetch) |
718e3744 | 1347 | { |
d62a17ae | 1348 | struct thread *thread = NULL; |
1349 | struct timeval now; | |
1350 | struct timeval zerotime = {0, 0}; | |
1351 | struct timeval tv; | |
1352 | struct timeval *tw = NULL; | |
1353 | ||
1354 | int num = 0; | |
1355 | ||
1356 | do { | |
1357 | /* Handle signals if any */ | |
1358 | if (m->handle_signals) | |
1359 | quagga_sigevent_process(); | |
1360 | ||
1361 | pthread_mutex_lock(&m->mtx); | |
1362 | ||
1363 | /* Process any pending cancellation requests */ | |
1364 | do_thread_cancel(m); | |
1365 | ||
e3c9529e QY |
1366 | /* |
1367 | * Attempt to flush ready queue before going into poll(). | |
1368 | * This is performance-critical. Think twice before modifying. | |
1369 | */ | |
1370 | if ((thread = thread_trim_head(&m->ready))) { | |
1371 | fetch = thread_run(m, thread, fetch); | |
1372 | if (fetch->ref) | |
1373 | *fetch->ref = NULL; | |
1374 | pthread_mutex_unlock(&m->mtx); | |
1375 | break; | |
1376 | } | |
1377 | ||
1378 | /* otherwise, tick through scheduling sequence */ | |
1379 | ||
bca37d17 QY |
1380 | /* |
1381 | * Post events to ready queue. This must come before the | |
1382 | * following block since events should occur immediately | |
1383 | */ | |
d62a17ae | 1384 | thread_process(&m->event); |
1385 | ||
bca37d17 QY |
1386 | /* |
1387 | * If there are no tasks on the ready queue, we will poll() | |
1388 | * until a timer expires or we receive I/O, whichever comes | |
1389 | * first. The strategy for doing this is: | |
d62a17ae | 1390 | * |
1391 | * - If there are events pending, set the poll() timeout to zero | |
1392 | * - If there are no events pending, but there are timers | |
1393 | * pending, set the | |
1394 | * timeout to the smallest remaining time on any timer | |
1395 | * - If there are neither timers nor events pending, but there | |
1396 | * are file | |
1397 | * descriptors pending, block indefinitely in poll() | |
1398 | * - If nothing is pending, it's time for the application to die | |
1399 | * | |
1400 | * In every case except the last, we need to hit poll() at least | |
bca37d17 QY |
1401 | * once per loop to avoid starvation by events |
1402 | */ | |
d62a17ae | 1403 | if (m->ready.count == 0) |
1404 | tw = thread_timer_wait(m->timer, &tv); | |
1405 | ||
1406 | if (m->ready.count != 0 || (tw && !timercmp(tw, &zerotime, >))) | |
1407 | tw = &zerotime; | |
1408 | ||
1409 | if (!tw && m->handler.pfdcount == 0) { /* die */ | |
1410 | pthread_mutex_unlock(&m->mtx); | |
1411 | fetch = NULL; | |
1412 | break; | |
1413 | } | |
1414 | ||
bca37d17 QY |
1415 | /* |
1416 | * Copy pollfd array + # active pollfds in it. Not necessary to | |
1417 | * copy the array size as this is fixed. | |
1418 | */ | |
d62a17ae | 1419 | m->handler.copycount = m->handler.pfdcount; |
1420 | memcpy(m->handler.copy, m->handler.pfds, | |
1421 | m->handler.copycount * sizeof(struct pollfd)); | |
1422 | ||
e3c9529e QY |
1423 | pthread_mutex_unlock(&m->mtx); |
1424 | { | |
1425 | num = fd_poll(m, m->handler.copy, m->handler.pfdsize, | |
1426 | m->handler.copycount, tw); | |
1427 | } | |
1428 | pthread_mutex_lock(&m->mtx); | |
d764d2cc | 1429 | |
e3c9529e QY |
1430 | /* Handle any errors received in poll() */ |
1431 | if (num < 0) { | |
1432 | if (errno == EINTR) { | |
d62a17ae | 1433 | pthread_mutex_unlock(&m->mtx); |
e3c9529e QY |
1434 | /* loop around to signal handler */ |
1435 | continue; | |
d62a17ae | 1436 | } |
1437 | ||
e3c9529e QY |
1438 | /* else die */ |
1439 | zlog_warn("poll() error: %s", safe_strerror(errno)); | |
1440 | pthread_mutex_unlock(&m->mtx); | |
1441 | fetch = NULL; | |
1442 | break; | |
bca37d17 | 1443 | } |
d62a17ae | 1444 | |
1445 | /* Post timers to ready queue. */ | |
1446 | monotime(&now); | |
1447 | thread_process_timers(m->timer, &now); | |
1448 | ||
1449 | /* Post I/O to ready queue. */ | |
1450 | if (num > 0) | |
1451 | thread_process_io(m, num); | |
1452 | ||
d62a17ae | 1453 | pthread_mutex_unlock(&m->mtx); |
1454 | ||
1455 | } while (!thread && m->spin); | |
1456 | ||
1457 | return fetch; | |
718e3744 | 1458 | } |
1459 | ||
d62a17ae | 1460 | static unsigned long timeval_elapsed(struct timeval a, struct timeval b) |
62f44022 | 1461 | { |
d62a17ae | 1462 | return (((a.tv_sec - b.tv_sec) * TIMER_SECOND_MICRO) |
1463 | + (a.tv_usec - b.tv_usec)); | |
62f44022 QY |
1464 | } |
1465 | ||
d62a17ae | 1466 | unsigned long thread_consumed_time(RUSAGE_T *now, RUSAGE_T *start, |
1467 | unsigned long *cputime) | |
718e3744 | 1468 | { |
d62a17ae | 1469 | /* This is 'user + sys' time. */ |
1470 | *cputime = timeval_elapsed(now->cpu.ru_utime, start->cpu.ru_utime) | |
1471 | + timeval_elapsed(now->cpu.ru_stime, start->cpu.ru_stime); | |
1472 | return timeval_elapsed(now->real, start->real); | |
8b70d0b0 | 1473 | } |
1474 | ||
50596be0 DS |
1475 | /* We should aim to yield after yield milliseconds, which defaults |
1476 | to THREAD_YIELD_TIME_SLOT . | |
8b70d0b0 | 1477 | Note: we are using real (wall clock) time for this calculation. |
1478 | It could be argued that CPU time may make more sense in certain | |
1479 | contexts. The things to consider are whether the thread may have | |
1480 | blocked (in which case wall time increases, but CPU time does not), | |
1481 | or whether the system is heavily loaded with other processes competing | |
d62a17ae | 1482 | for CPU time. On balance, wall clock time seems to make sense. |
8b70d0b0 | 1483 | Plus it has the added benefit that gettimeofday should be faster |
1484 | than calling getrusage. */ | |
d62a17ae | 1485 | int thread_should_yield(struct thread *thread) |
718e3744 | 1486 | { |
d62a17ae | 1487 | int result; |
1488 | pthread_mutex_lock(&thread->mtx); | |
1489 | { | |
1490 | result = monotime_since(&thread->real, NULL) | |
1491 | > (int64_t)thread->yield; | |
1492 | } | |
1493 | pthread_mutex_unlock(&thread->mtx); | |
1494 | return result; | |
50596be0 DS |
1495 | } |
1496 | ||
d62a17ae | 1497 | void thread_set_yield_time(struct thread *thread, unsigned long yield_time) |
50596be0 | 1498 | { |
d62a17ae | 1499 | pthread_mutex_lock(&thread->mtx); |
1500 | { | |
1501 | thread->yield = yield_time; | |
1502 | } | |
1503 | pthread_mutex_unlock(&thread->mtx); | |
718e3744 | 1504 | } |
1505 | ||
d62a17ae | 1506 | void thread_getrusage(RUSAGE_T *r) |
db9c0df9 | 1507 | { |
d62a17ae | 1508 | monotime(&r->real); |
1509 | getrusage(RUSAGE_SELF, &(r->cpu)); | |
db9c0df9 PJ |
1510 | } |
1511 | ||
fbcac826 QY |
1512 | /* |
1513 | * Call a thread. | |
1514 | * | |
1515 | * This function will atomically update the thread's usage history. At present | |
1516 | * this is the only spot where usage history is written. Nevertheless the code | |
1517 | * has been written such that the introduction of writers in the future should | |
1518 | * not need to update it provided the writers atomically perform only the | |
1519 | * operations done here, i.e. updating the total and maximum times. In | |
1520 | * particular, the maximum real and cpu times must be monotonically increasing | |
1521 | * or this code is not correct. | |
1522 | */ | |
d62a17ae | 1523 | void thread_call(struct thread *thread) |
718e3744 | 1524 | { |
fbcac826 QY |
1525 | _Atomic unsigned long realtime, cputime; |
1526 | unsigned long exp; | |
1527 | unsigned long helper; | |
d62a17ae | 1528 | RUSAGE_T before, after; |
cc8b13a0 | 1529 | |
d62a17ae | 1530 | GETRUSAGE(&before); |
1531 | thread->real = before.real; | |
718e3744 | 1532 | |
d62a17ae | 1533 | pthread_setspecific(thread_current, thread); |
1534 | (*thread->func)(thread); | |
1535 | pthread_setspecific(thread_current, NULL); | |
718e3744 | 1536 | |
d62a17ae | 1537 | GETRUSAGE(&after); |
718e3744 | 1538 | |
fbcac826 QY |
1539 | realtime = thread_consumed_time(&after, &before, &helper); |
1540 | cputime = helper; | |
1541 | ||
1542 | /* update realtime */ | |
1543 | atomic_fetch_add_explicit(&thread->hist->real.total, realtime, | |
1544 | memory_order_seq_cst); | |
1545 | exp = atomic_load_explicit(&thread->hist->real.max, | |
1546 | memory_order_seq_cst); | |
1547 | while (exp < realtime | |
1548 | && !atomic_compare_exchange_weak_explicit( | |
1549 | &thread->hist->real.max, &exp, realtime, | |
1550 | memory_order_seq_cst, memory_order_seq_cst)) | |
1551 | ; | |
1552 | ||
1553 | /* update cputime */ | |
1554 | atomic_fetch_add_explicit(&thread->hist->cpu.total, cputime, | |
1555 | memory_order_seq_cst); | |
1556 | exp = atomic_load_explicit(&thread->hist->cpu.max, | |
1557 | memory_order_seq_cst); | |
1558 | while (exp < cputime | |
1559 | && !atomic_compare_exchange_weak_explicit( | |
1560 | &thread->hist->cpu.max, &exp, cputime, | |
1561 | memory_order_seq_cst, memory_order_seq_cst)) | |
1562 | ; | |
1563 | ||
1564 | atomic_fetch_add_explicit(&thread->hist->total_calls, 1, | |
1565 | memory_order_seq_cst); | |
1566 | atomic_fetch_or_explicit(&thread->hist->types, 1 << thread->add_type, | |
1567 | memory_order_seq_cst); | |
718e3744 | 1568 | |
924b9229 | 1569 | #ifdef CONSUMED_TIME_CHECK |
d62a17ae | 1570 | if (realtime > CONSUMED_TIME_CHECK) { |
1571 | /* | |
1572 | * We have a CPU Hog on our hands. | |
1573 | * Whinge about it now, so we're aware this is yet another task | |
1574 | * to fix. | |
1575 | */ | |
1576 | zlog_warn( | |
1577 | "SLOW THREAD: task %s (%lx) ran for %lums (cpu time %lums)", | |
1578 | thread->funcname, (unsigned long)thread->func, | |
1579 | realtime / 1000, cputime / 1000); | |
1580 | } | |
924b9229 | 1581 | #endif /* CONSUMED_TIME_CHECK */ |
718e3744 | 1582 | } |
1583 | ||
1584 | /* Execute thread */ | |
d62a17ae | 1585 | void funcname_thread_execute(struct thread_master *m, |
1586 | int (*func)(struct thread *), void *arg, int val, | |
1587 | debugargdef) | |
718e3744 | 1588 | { |
d62a17ae | 1589 | struct cpu_thread_history tmp; |
1590 | struct thread dummy; | |
718e3744 | 1591 | |
d62a17ae | 1592 | memset(&dummy, 0, sizeof(struct thread)); |
718e3744 | 1593 | |
d62a17ae | 1594 | pthread_mutex_init(&dummy.mtx, NULL); |
1595 | dummy.type = THREAD_EVENT; | |
1596 | dummy.add_type = THREAD_EXECUTE; | |
1597 | dummy.master = NULL; | |
1598 | dummy.arg = arg; | |
1599 | dummy.u.val = val; | |
9c7753e4 | 1600 | |
d62a17ae | 1601 | tmp.func = dummy.func = func; |
1602 | tmp.funcname = dummy.funcname = funcname; | |
1603 | dummy.hist = hash_get(m->cpu_record, &tmp, | |
1604 | (void *(*)(void *))cpu_record_hash_alloc); | |
f7c62e11 | 1605 | |
d62a17ae | 1606 | dummy.schedfrom = schedfrom; |
1607 | dummy.schedfrom_line = fromln; | |
9c7753e4 | 1608 | |
d62a17ae | 1609 | thread_call(&dummy); |
718e3744 | 1610 | } |