1 /*- 2 * Copyright (c) 2000 Doug Rabson 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 14 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 16 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 17 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 19 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 20 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 21 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 22 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 23 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 24 * SUCH DAMAGE. 25 */ 26 27 #include <sys/cdefs.h> 28 __FBSDID("$FreeBSD$"); 29 30 #include <sys/param.h> 31 #include <sys/systm.h> 32 #include <sys/bus.h> 33 #include <sys/cpuset.h> 34 #include <sys/interrupt.h> 35 #include <sys/kernel.h> 36 #include <sys/kthread.h> 37 #include <sys/libkern.h> 38 #include <sys/limits.h> 39 #include <sys/lock.h> 40 #include <sys/malloc.h> 41 #include <sys/mutex.h> 42 #include <sys/proc.h> 43 #include <sys/sched.h> 44 #include <sys/smp.h> 45 #include <sys/taskqueue.h> 46 #include <sys/unistd.h> 47 #include <machine/stdarg.h> 48 49 static MALLOC_DEFINE(M_TASKQUEUE, "taskqueue", "Task Queues"); 50 static void *taskqueue_giant_ih; 51 static void *taskqueue_ih; 52 static void taskqueue_fast_enqueue(void *); 53 static void taskqueue_swi_enqueue(void *); 54 static void taskqueue_swi_giant_enqueue(void *); 55 56 struct taskqueue_busy { 57 struct task *tb_running; 58 TAILQ_ENTRY(taskqueue_busy) tb_link; 59 }; 60 61 struct task * const TB_DRAIN_WAITER = (struct task *)0x1; 62 63 struct taskqueue { 64 STAILQ_HEAD(, task) tq_queue; 65 taskqueue_enqueue_fn tq_enqueue; 66 void *tq_context; 67 char *tq_name; 68 TAILQ_HEAD(, taskqueue_busy) tq_active; 69 struct mtx tq_mutex; 70 struct thread **tq_threads; 71 int tq_tcount; 72 int tq_spin; 73 int tq_flags; 74 int tq_callouts; 75 taskqueue_callback_fn tq_callbacks[TASKQUEUE_NUM_CALLBACKS]; 76 void *tq_cb_contexts[TASKQUEUE_NUM_CALLBACKS]; 77 }; 78 79 #define TQ_FLAGS_ACTIVE (1 << 0) 80 #define TQ_FLAGS_BLOCKED (1 << 1) 81 #define TQ_FLAGS_UNLOCKED_ENQUEUE (1 << 2) 82 83 #define DT_CALLOUT_ARMED (1 << 0) 84 #define DT_DRAIN_IN_PROGRESS (1 << 1) 85 86 #define TQ_LOCK(tq) \ 87 do { \ 88 if ((tq)->tq_spin) \ 89 mtx_lock_spin(&(tq)->tq_mutex); \ 90 else \ 91 mtx_lock(&(tq)->tq_mutex); \ 92 } while (0) 93 #define TQ_ASSERT_LOCKED(tq) mtx_assert(&(tq)->tq_mutex, MA_OWNED) 94 95 #define TQ_UNLOCK(tq) \ 96 do { \ 97 if ((tq)->tq_spin) \ 98 mtx_unlock_spin(&(tq)->tq_mutex); \ 99 else \ 100 mtx_unlock(&(tq)->tq_mutex); \ 101 } while (0) 102 #define TQ_ASSERT_UNLOCKED(tq) mtx_assert(&(tq)->tq_mutex, MA_NOTOWNED) 103 104 void 105 _timeout_task_init(struct taskqueue *queue, struct timeout_task *timeout_task, 106 int priority, task_fn_t func, void *context) 107 { 108 109 TASK_INIT(&timeout_task->t, priority, func, context); 110 callout_init_mtx(&timeout_task->c, &queue->tq_mutex, 111 CALLOUT_RETURNUNLOCKED); 112 timeout_task->q = queue; 113 timeout_task->f = 0; 114 } 115 116 static __inline int 117 TQ_SLEEP(struct taskqueue *tq, void *p, struct mtx *m, int pri, const char *wm, 118 int t) 119 { 120 if (tq->tq_spin) 121 return (msleep_spin(p, m, wm, t)); 122 return (msleep(p, m, pri, wm, t)); 123 } 124 125 static struct taskqueue * 126 _taskqueue_create(const char *name, int mflags, 127 taskqueue_enqueue_fn enqueue, void *context, 128 int mtxflags, const char *mtxname __unused) 129 { 130 struct taskqueue *queue; 131 char *tq_name; 132 133 tq_name = malloc(TASKQUEUE_NAMELEN, M_TASKQUEUE, mflags | M_ZERO); 134 if (tq_name == NULL) 135 return (NULL); 136 137 queue = malloc(sizeof(struct taskqueue), M_TASKQUEUE, mflags | M_ZERO); 138 if (queue == NULL) { 139 free(tq_name, M_TASKQUEUE); 140 return (NULL); 141 } 142 143 snprintf(tq_name, TASKQUEUE_NAMELEN, "%s", (name) ? name : "taskqueue"); 144 145 STAILQ_INIT(&queue->tq_queue); 146 TAILQ_INIT(&queue->tq_active); 147 queue->tq_enqueue = enqueue; 148 queue->tq_context = context; 149 queue->tq_name = tq_name; 150 queue->tq_spin = (mtxflags & MTX_SPIN) != 0; 151 queue->tq_flags |= TQ_FLAGS_ACTIVE; 152 if (enqueue == taskqueue_fast_enqueue || 153 enqueue == taskqueue_swi_enqueue || 154 enqueue == taskqueue_swi_giant_enqueue || 155 enqueue == taskqueue_thread_enqueue) 156 queue->tq_flags |= TQ_FLAGS_UNLOCKED_ENQUEUE; 157 mtx_init(&queue->tq_mutex, tq_name, NULL, mtxflags); 158 159 return (queue); 160 } 161 162 struct taskqueue * 163 taskqueue_create(const char *name, int mflags, 164 taskqueue_enqueue_fn enqueue, void *context) 165 { 166 167 return _taskqueue_create(name, mflags, enqueue, context, 168 MTX_DEF, name); 169 } 170 171 void 172 taskqueue_set_callback(struct taskqueue *queue, 173 enum taskqueue_callback_type cb_type, taskqueue_callback_fn callback, 174 void *context) 175 { 176 177 KASSERT(((cb_type >= TASKQUEUE_CALLBACK_TYPE_MIN) && 178 (cb_type <= TASKQUEUE_CALLBACK_TYPE_MAX)), 179 ("Callback type %d not valid, must be %d-%d", cb_type, 180 TASKQUEUE_CALLBACK_TYPE_MIN, TASKQUEUE_CALLBACK_TYPE_MAX)); 181 KASSERT((queue->tq_callbacks[cb_type] == NULL), 182 ("Re-initialization of taskqueue callback?")); 183 184 queue->tq_callbacks[cb_type] = callback; 185 queue->tq_cb_contexts[cb_type] = context; 186 } 187 188 /* 189 * Signal a taskqueue thread to terminate. 190 */ 191 static void 192 taskqueue_terminate(struct thread **pp, struct taskqueue *tq) 193 { 194 195 while (tq->tq_tcount > 0 || tq->tq_callouts > 0) { 196 wakeup(tq); 197 TQ_SLEEP(tq, pp, &tq->tq_mutex, PWAIT, "taskqueue_destroy", 0); 198 } 199 } 200 201 void 202 taskqueue_free(struct taskqueue *queue) 203 { 204 205 TQ_LOCK(queue); 206 queue->tq_flags &= ~TQ_FLAGS_ACTIVE; 207 taskqueue_terminate(queue->tq_threads, queue); 208 KASSERT(TAILQ_EMPTY(&queue->tq_active), ("Tasks still running?")); 209 KASSERT(queue->tq_callouts == 0, ("Armed timeout tasks")); 210 mtx_destroy(&queue->tq_mutex); 211 free(queue->tq_threads, M_TASKQUEUE); 212 free(queue->tq_name, M_TASKQUEUE); 213 free(queue, M_TASKQUEUE); 214 } 215 216 static int 217 taskqueue_enqueue_locked(struct taskqueue *queue, struct task *task) 218 { 219 struct task *ins; 220 struct task *prev; 221 222 KASSERT(task->ta_func != NULL, ("enqueueing task with NULL func")); 223 /* 224 * Count multiple enqueues. 225 */ 226 if (task->ta_pending) { 227 if (task->ta_pending < USHRT_MAX) 228 task->ta_pending++; 229 TQ_UNLOCK(queue); 230 return (0); 231 } 232 233 /* 234 * Optimise the case when all tasks have the same priority. 235 */ 236 prev = STAILQ_LAST(&queue->tq_queue, task, ta_link); 237 if (!prev || prev->ta_priority >= task->ta_priority) { 238 STAILQ_INSERT_TAIL(&queue->tq_queue, task, ta_link); 239 } else { 240 prev = NULL; 241 for (ins = STAILQ_FIRST(&queue->tq_queue); ins; 242 prev = ins, ins = STAILQ_NEXT(ins, ta_link)) 243 if (ins->ta_priority < task->ta_priority) 244 break; 245 246 if (prev) 247 STAILQ_INSERT_AFTER(&queue->tq_queue, prev, task, ta_link); 248 else 249 STAILQ_INSERT_HEAD(&queue->tq_queue, task, ta_link); 250 } 251 252 task->ta_pending = 1; 253 if ((queue->tq_flags & TQ_FLAGS_UNLOCKED_ENQUEUE) != 0) 254 TQ_UNLOCK(queue); 255 if ((queue->tq_flags & TQ_FLAGS_BLOCKED) == 0) 256 queue->tq_enqueue(queue->tq_context); 257 if ((queue->tq_flags & TQ_FLAGS_UNLOCKED_ENQUEUE) == 0) 258 TQ_UNLOCK(queue); 259 260 /* Return with lock released. */ 261 return (0); 262 } 263 264 int 265 taskqueue_enqueue(struct taskqueue *queue, struct task *task) 266 { 267 int res; 268 269 TQ_LOCK(queue); 270 res = taskqueue_enqueue_locked(queue, task); 271 /* The lock is released inside. */ 272 273 return (res); 274 } 275 276 static void 277 taskqueue_timeout_func(void *arg) 278 { 279 struct taskqueue *queue; 280 struct timeout_task *timeout_task; 281 282 timeout_task = arg; 283 queue = timeout_task->q; 284 KASSERT((timeout_task->f & DT_CALLOUT_ARMED) != 0, ("Stray timeout")); 285 timeout_task->f &= ~DT_CALLOUT_ARMED; 286 queue->tq_callouts--; 287 taskqueue_enqueue_locked(timeout_task->q, &timeout_task->t); 288 /* The lock is released inside. */ 289 } 290 291 int 292 taskqueue_enqueue_timeout_sbt(struct taskqueue *queue, 293 struct timeout_task *timeout_task, sbintime_t sbt, sbintime_t pr, int flags) 294 { 295 int res; 296 297 TQ_LOCK(queue); 298 KASSERT(timeout_task->q == NULL || timeout_task->q == queue, 299 ("Migrated queue")); 300 KASSERT(!queue->tq_spin, ("Timeout for spin-queue")); 301 timeout_task->q = queue; 302 res = timeout_task->t.ta_pending; 303 if (timeout_task->f & DT_DRAIN_IN_PROGRESS) { 304 /* Do nothing */ 305 TQ_UNLOCK(queue); 306 res = -1; 307 } else if (sbt == 0) { 308 taskqueue_enqueue_locked(queue, &timeout_task->t); 309 /* The lock is released inside. */ 310 } else { 311 if ((timeout_task->f & DT_CALLOUT_ARMED) != 0) { 312 res++; 313 } else { 314 queue->tq_callouts++; 315 timeout_task->f |= DT_CALLOUT_ARMED; 316 if (sbt < 0) 317 sbt = -sbt; /* Ignore overflow. */ 318 } 319 if (sbt > 0) { 320 callout_reset_sbt(&timeout_task->c, sbt, pr, 321 taskqueue_timeout_func, timeout_task, flags); 322 } 323 TQ_UNLOCK(queue); 324 } 325 return (res); 326 } 327 328 int 329 taskqueue_enqueue_timeout(struct taskqueue *queue, 330 struct timeout_task *ttask, int ticks) 331 { 332 333 return (taskqueue_enqueue_timeout_sbt(queue, ttask, ticks * tick_sbt, 334 0, 0)); 335 } 336 337 static void 338 taskqueue_task_nop_fn(void *context, int pending) 339 { 340 } 341 342 /* 343 * Block until all currently queued tasks in this taskqueue 344 * have begun execution. Tasks queued during execution of 345 * this function are ignored. 346 */ 347 static void 348 taskqueue_drain_tq_queue(struct taskqueue *queue) 349 { 350 struct task t_barrier; 351 352 if (STAILQ_EMPTY(&queue->tq_queue)) 353 return; 354 355 /* 356 * Enqueue our barrier after all current tasks, but with 357 * the highest priority so that newly queued tasks cannot 358 * pass it. Because of the high priority, we can not use 359 * taskqueue_enqueue_locked directly (which drops the lock 360 * anyway) so just insert it at tail while we have the 361 * queue lock. 362 */ 363 TASK_INIT(&t_barrier, USHRT_MAX, taskqueue_task_nop_fn, &t_barrier); 364 STAILQ_INSERT_TAIL(&queue->tq_queue, &t_barrier, ta_link); 365 t_barrier.ta_pending = 1; 366 367 /* 368 * Once the barrier has executed, all previously queued tasks 369 * have completed or are currently executing. 370 */ 371 while (t_barrier.ta_pending != 0) 372 TQ_SLEEP(queue, &t_barrier, &queue->tq_mutex, PWAIT, "-", 0); 373 } 374 375 /* 376 * Block until all currently executing tasks for this taskqueue 377 * complete. Tasks that begin execution during the execution 378 * of this function are ignored. 379 */ 380 static void 381 taskqueue_drain_tq_active(struct taskqueue *queue) 382 { 383 struct taskqueue_busy tb_marker, *tb_first; 384 385 if (TAILQ_EMPTY(&queue->tq_active)) 386 return; 387 388 /* Block taskq_terminate().*/ 389 queue->tq_callouts++; 390 391 /* 392 * Wait for all currently executing taskqueue threads 393 * to go idle. 394 */ 395 tb_marker.tb_running = TB_DRAIN_WAITER; 396 TAILQ_INSERT_TAIL(&queue->tq_active, &tb_marker, tb_link); 397 while (TAILQ_FIRST(&queue->tq_active) != &tb_marker) 398 TQ_SLEEP(queue, &tb_marker, &queue->tq_mutex, PWAIT, "-", 0); 399 TAILQ_REMOVE(&queue->tq_active, &tb_marker, tb_link); 400 401 /* 402 * Wakeup any other drain waiter that happened to queue up 403 * without any intervening active thread. 404 */ 405 tb_first = TAILQ_FIRST(&queue->tq_active); 406 if (tb_first != NULL && tb_first->tb_running == TB_DRAIN_WAITER) 407 wakeup(tb_first); 408 409 /* Release taskqueue_terminate(). */ 410 queue->tq_callouts--; 411 if ((queue->tq_flags & TQ_FLAGS_ACTIVE) == 0) 412 wakeup_one(queue->tq_threads); 413 } 414 415 void 416 taskqueue_block(struct taskqueue *queue) 417 { 418 419 TQ_LOCK(queue); 420 queue->tq_flags |= TQ_FLAGS_BLOCKED; 421 TQ_UNLOCK(queue); 422 } 423 424 void 425 taskqueue_unblock(struct taskqueue *queue) 426 { 427 428 TQ_LOCK(queue); 429 queue->tq_flags &= ~TQ_FLAGS_BLOCKED; 430 if (!STAILQ_EMPTY(&queue->tq_queue)) 431 queue->tq_enqueue(queue->tq_context); 432 TQ_UNLOCK(queue); 433 } 434 435 static void 436 taskqueue_run_locked(struct taskqueue *queue) 437 { 438 struct taskqueue_busy tb; 439 struct taskqueue_busy *tb_first; 440 struct task *task; 441 int pending; 442 443 KASSERT(queue != NULL, ("tq is NULL")); 444 TQ_ASSERT_LOCKED(queue); 445 tb.tb_running = NULL; 446 447 while (STAILQ_FIRST(&queue->tq_queue)) { 448 TAILQ_INSERT_TAIL(&queue->tq_active, &tb, tb_link); 449 450 /* 451 * Carefully remove the first task from the queue and 452 * zero its pending count. 453 */ 454 task = STAILQ_FIRST(&queue->tq_queue); 455 KASSERT(task != NULL, ("task is NULL")); 456 STAILQ_REMOVE_HEAD(&queue->tq_queue, ta_link); 457 pending = task->ta_pending; 458 task->ta_pending = 0; 459 tb.tb_running = task; 460 TQ_UNLOCK(queue); 461 462 KASSERT(task->ta_func != NULL, ("task->ta_func is NULL")); 463 task->ta_func(task->ta_context, pending); 464 465 TQ_LOCK(queue); 466 tb.tb_running = NULL; 467 wakeup(task); 468 469 TAILQ_REMOVE(&queue->tq_active, &tb, tb_link); 470 tb_first = TAILQ_FIRST(&queue->tq_active); 471 if (tb_first != NULL && 472 tb_first->tb_running == TB_DRAIN_WAITER) 473 wakeup(tb_first); 474 } 475 } 476 477 void 478 taskqueue_run(struct taskqueue *queue) 479 { 480 481 TQ_LOCK(queue); 482 taskqueue_run_locked(queue); 483 TQ_UNLOCK(queue); 484 } 485 486 static int 487 task_is_running(struct taskqueue *queue, struct task *task) 488 { 489 struct taskqueue_busy *tb; 490 491 TQ_ASSERT_LOCKED(queue); 492 TAILQ_FOREACH(tb, &queue->tq_active, tb_link) { 493 if (tb->tb_running == task) 494 return (1); 495 } 496 return (0); 497 } 498 499 /* 500 * Only use this function in single threaded contexts. It returns 501 * non-zero if the given task is either pending or running. Else the 502 * task is idle and can be queued again or freed. 503 */ 504 int 505 taskqueue_poll_is_busy(struct taskqueue *queue, struct task *task) 506 { 507 int retval; 508 509 TQ_LOCK(queue); 510 retval = task->ta_pending > 0 || task_is_running(queue, task); 511 TQ_UNLOCK(queue); 512 513 return (retval); 514 } 515 516 static int 517 taskqueue_cancel_locked(struct taskqueue *queue, struct task *task, 518 u_int *pendp) 519 { 520 521 if (task->ta_pending > 0) 522 STAILQ_REMOVE(&queue->tq_queue, task, task, ta_link); 523 if (pendp != NULL) 524 *pendp = task->ta_pending; 525 task->ta_pending = 0; 526 return (task_is_running(queue, task) ? EBUSY : 0); 527 } 528 529 int 530 taskqueue_cancel(struct taskqueue *queue, struct task *task, u_int *pendp) 531 { 532 int error; 533 534 TQ_LOCK(queue); 535 error = taskqueue_cancel_locked(queue, task, pendp); 536 TQ_UNLOCK(queue); 537 538 return (error); 539 } 540 541 int 542 taskqueue_cancel_timeout(struct taskqueue *queue, 543 struct timeout_task *timeout_task, u_int *pendp) 544 { 545 u_int pending, pending1; 546 int error; 547 548 TQ_LOCK(queue); 549 pending = !!(callout_stop(&timeout_task->c) > 0); 550 error = taskqueue_cancel_locked(queue, &timeout_task->t, &pending1); 551 if ((timeout_task->f & DT_CALLOUT_ARMED) != 0) { 552 timeout_task->f &= ~DT_CALLOUT_ARMED; 553 queue->tq_callouts--; 554 } 555 TQ_UNLOCK(queue); 556 557 if (pendp != NULL) 558 *pendp = pending + pending1; 559 return (error); 560 } 561 562 void 563 taskqueue_drain(struct taskqueue *queue, struct task *task) 564 { 565 566 if (!queue->tq_spin) 567 WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL, __func__); 568 569 TQ_LOCK(queue); 570 while (task->ta_pending != 0 || task_is_running(queue, task)) 571 TQ_SLEEP(queue, task, &queue->tq_mutex, PWAIT, "-", 0); 572 TQ_UNLOCK(queue); 573 } 574 575 void 576 taskqueue_drain_all(struct taskqueue *queue) 577 { 578 579 if (!queue->tq_spin) 580 WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL, __func__); 581 582 TQ_LOCK(queue); 583 taskqueue_drain_tq_queue(queue); 584 taskqueue_drain_tq_active(queue); 585 TQ_UNLOCK(queue); 586 } 587 588 void 589 taskqueue_drain_timeout(struct taskqueue *queue, 590 struct timeout_task *timeout_task) 591 { 592 593 /* 594 * Set flag to prevent timer from re-starting during drain: 595 */ 596 TQ_LOCK(queue); 597 KASSERT((timeout_task->f & DT_DRAIN_IN_PROGRESS) == 0, 598 ("Drain already in progress")); 599 timeout_task->f |= DT_DRAIN_IN_PROGRESS; 600 TQ_UNLOCK(queue); 601 602 callout_drain(&timeout_task->c); 603 taskqueue_drain(queue, &timeout_task->t); 604 605 /* 606 * Clear flag to allow timer to re-start: 607 */ 608 TQ_LOCK(queue); 609 timeout_task->f &= ~DT_DRAIN_IN_PROGRESS; 610 TQ_UNLOCK(queue); 611 } 612 613 static void 614 taskqueue_swi_enqueue(void *context) 615 { 616 swi_sched(taskqueue_ih, 0); 617 } 618 619 static void 620 taskqueue_swi_run(void *dummy) 621 { 622 taskqueue_run(taskqueue_swi); 623 } 624 625 static void 626 taskqueue_swi_giant_enqueue(void *context) 627 { 628 swi_sched(taskqueue_giant_ih, 0); 629 } 630 631 static void 632 taskqueue_swi_giant_run(void *dummy) 633 { 634 taskqueue_run(taskqueue_swi_giant); 635 } 636 637 static int 638 _taskqueue_start_threads(struct taskqueue **tqp, int count, int pri, 639 cpuset_t *mask, const char *name, va_list ap) 640 { 641 char ktname[MAXCOMLEN + 1]; 642 struct thread *td; 643 struct taskqueue *tq; 644 int i, error; 645 646 if (count <= 0) 647 return (EINVAL); 648 649 vsnprintf(ktname, sizeof(ktname), name, ap); 650 tq = *tqp; 651 652 tq->tq_threads = malloc(sizeof(struct thread *) * count, M_TASKQUEUE, 653 M_NOWAIT | M_ZERO); 654 if (tq->tq_threads == NULL) { 655 printf("%s: no memory for %s threads\n", __func__, ktname); 656 return (ENOMEM); 657 } 658 659 for (i = 0; i < count; i++) { 660 if (count == 1) 661 error = kthread_add(taskqueue_thread_loop, tqp, NULL, 662 &tq->tq_threads[i], RFSTOPPED, 0, "%s", ktname); 663 else 664 error = kthread_add(taskqueue_thread_loop, tqp, NULL, 665 &tq->tq_threads[i], RFSTOPPED, 0, 666 "%s_%d", ktname, i); 667 if (error) { 668 /* should be ok to continue, taskqueue_free will dtrt */ 669 printf("%s: kthread_add(%s): error %d", __func__, 670 ktname, error); 671 tq->tq_threads[i] = NULL; /* paranoid */ 672 } else 673 tq->tq_tcount++; 674 } 675 if (tq->tq_tcount == 0) { 676 free(tq->tq_threads, M_TASKQUEUE); 677 tq->tq_threads = NULL; 678 return (ENOMEM); 679 } 680 for (i = 0; i < count; i++) { 681 if (tq->tq_threads[i] == NULL) 682 continue; 683 td = tq->tq_threads[i]; 684 if (mask) { 685 error = cpuset_setthread(td->td_tid, mask); 686 /* 687 * Failing to pin is rarely an actual fatal error; 688 * it'll just affect performance. 689 */ 690 if (error) 691 printf("%s: curthread=%llu: can't pin; " 692 "error=%d\n", 693 __func__, 694 (unsigned long long) td->td_tid, 695 error); 696 } 697 thread_lock(td); 698 sched_prio(td, pri); 699 sched_add(td, SRQ_BORING); 700 thread_unlock(td); 701 } 702 703 return (0); 704 } 705 706 int 707 taskqueue_start_threads(struct taskqueue **tqp, int count, int pri, 708 const char *name, ...) 709 { 710 va_list ap; 711 int error; 712 713 va_start(ap, name); 714 error = _taskqueue_start_threads(tqp, count, pri, NULL, name, ap); 715 va_end(ap); 716 return (error); 717 } 718 719 int 720 taskqueue_start_threads_cpuset(struct taskqueue **tqp, int count, int pri, 721 cpuset_t *mask, const char *name, ...) 722 { 723 va_list ap; 724 int error; 725 726 va_start(ap, name); 727 error = _taskqueue_start_threads(tqp, count, pri, mask, name, ap); 728 va_end(ap); 729 return (error); 730 } 731 732 static inline void 733 taskqueue_run_callback(struct taskqueue *tq, 734 enum taskqueue_callback_type cb_type) 735 { 736 taskqueue_callback_fn tq_callback; 737 738 TQ_ASSERT_UNLOCKED(tq); 739 tq_callback = tq->tq_callbacks[cb_type]; 740 if (tq_callback != NULL) 741 tq_callback(tq->tq_cb_contexts[cb_type]); 742 } 743 744 void 745 taskqueue_thread_loop(void *arg) 746 { 747 struct taskqueue **tqp, *tq; 748 749 tqp = arg; 750 tq = *tqp; 751 taskqueue_run_callback(tq, TASKQUEUE_CALLBACK_TYPE_INIT); 752 TQ_LOCK(tq); 753 while ((tq->tq_flags & TQ_FLAGS_ACTIVE) != 0) { 754 /* XXX ? */ 755 taskqueue_run_locked(tq); 756 /* 757 * Because taskqueue_run() can drop tq_mutex, we need to 758 * check if the TQ_FLAGS_ACTIVE flag wasn't removed in the 759 * meantime, which means we missed a wakeup. 760 */ 761 if ((tq->tq_flags & TQ_FLAGS_ACTIVE) == 0) 762 break; 763 TQ_SLEEP(tq, tq, &tq->tq_mutex, 0, "-", 0); 764 } 765 taskqueue_run_locked(tq); 766 /* 767 * This thread is on its way out, so just drop the lock temporarily 768 * in order to call the shutdown callback. This allows the callback 769 * to look at the taskqueue, even just before it dies. 770 */ 771 TQ_UNLOCK(tq); 772 taskqueue_run_callback(tq, TASKQUEUE_CALLBACK_TYPE_SHUTDOWN); 773 TQ_LOCK(tq); 774 775 /* rendezvous with thread that asked us to terminate */ 776 tq->tq_tcount--; 777 wakeup_one(tq->tq_threads); 778 TQ_UNLOCK(tq); 779 kthread_exit(); 780 } 781 782 void 783 taskqueue_thread_enqueue(void *context) 784 { 785 struct taskqueue **tqp, *tq; 786 787 tqp = context; 788 tq = *tqp; 789 wakeup_one(tq); 790 } 791 792 TASKQUEUE_DEFINE(swi, taskqueue_swi_enqueue, NULL, 793 swi_add(NULL, "task queue", taskqueue_swi_run, NULL, SWI_TQ, 794 INTR_MPSAFE, &taskqueue_ih)); 795 796 TASKQUEUE_DEFINE(swi_giant, taskqueue_swi_giant_enqueue, NULL, 797 swi_add(NULL, "Giant taskq", taskqueue_swi_giant_run, 798 NULL, SWI_TQ_GIANT, 0, &taskqueue_giant_ih)); 799 800 TASKQUEUE_DEFINE_THREAD(thread); 801 802 struct taskqueue * 803 taskqueue_create_fast(const char *name, int mflags, 804 taskqueue_enqueue_fn enqueue, void *context) 805 { 806 return _taskqueue_create(name, mflags, enqueue, context, 807 MTX_SPIN, "fast_taskqueue"); 808 } 809 810 static void *taskqueue_fast_ih; 811 812 static void 813 taskqueue_fast_enqueue(void *context) 814 { 815 swi_sched(taskqueue_fast_ih, 0); 816 } 817 818 static void 819 taskqueue_fast_run(void *dummy) 820 { 821 taskqueue_run(taskqueue_fast); 822 } 823 824 TASKQUEUE_FAST_DEFINE(fast, taskqueue_fast_enqueue, NULL, 825 swi_add(NULL, "fast taskq", taskqueue_fast_run, NULL, 826 SWI_TQ_FAST, INTR_MPSAFE, &taskqueue_fast_ih)); 827 828 int 829 taskqueue_member(struct taskqueue *queue, struct thread *td) 830 { 831 int i, j, ret = 0; 832 833 for (i = 0, j = 0; ; i++) { 834 if (queue->tq_threads[i] == NULL) 835 continue; 836 if (queue->tq_threads[i] == td) { 837 ret = 1; 838 break; 839 } 840 if (++j >= queue->tq_tcount) 841 break; 842 } 843 return (ret); 844 } 845