kernel/workqueue.c

   1 /*
   2  * linux/kernel/workqueue.c
   3  *
   4  * Generic mechanism for defining kernel helper threads for running
   5  * arbitrary tasks in process context.
   6  *
   7  * Started by Ingo Molnar, Copyright (C) 2002
   8  *
   9  * Derived from the taskqueue/keventd code by:
  10  *
  11  *   David Woodhouse <dwmw2@infradead.org>
  12  *   Andrew Morton <andrewm@uow.edu.au>
  13  *   Kai Petzke <wpp@marie.physik.tu-berlin.de>
  14  *   Theodore Ts'o <tytso@mit.edu>
  15  */
  16
  17 #include <linux/module.h>
  18 #include <linux/kernel.h>
  19 #include <linux/sched.h>
  20 #include <linux/init.h>
  21 #include <linux/signal.h>
  22 #include <linux/completion.h>
  23 #include <linux/workqueue.h>
  24 #include <linux/slab.h>
  25 #include <linux/cpu.h>
  26 #include <linux/notifier.h>
  27 #include <linux/kthread.h>
  28
  29 /*
  30  * The per-CPU workqueue (if single thread, we always use cpu 0's).
  31  *
  32  * The sequence counters are for flush_scheduled_work().  It wants to wait
  33  * until until all currently-scheduled works are completed, but it doesn't
  34  * want to be livelocked by new, incoming ones.  So it waits until
  35  * remove_sequence is >= the insert_sequence which pertained when
  36  * flush_scheduled_work() was called.
  37  */
  38 struct cpu_workqueue_struct {
  39
  40         spinlock_t lock;
  41
  42         long remove_sequence;   /* Least-recently added (next to run) */
  43         long insert_sequence;   /* Next to add */
  44
  45         struct list_head worklist;
  46         wait_queue_head_t more_work;
  47         wait_queue_head_t work_done;
  48
  49         struct workqueue_struct *wq;
  50         task_t *thread;
  51
  52         int run_depth;          /* Detect run_workqueue() recursion depth */
  53 } ____cacheline_aligned;
  54
  55 /*
  56  * The externally visible workqueue abstraction is an array of
  57  * per-CPU workqueues:
  58  */
  59 struct workqueue_struct {
  60         struct cpu_workqueue_struct cpu_wq[NR_CPUS];
  61         const char *name;
  62         struct list_head list;  /* Empty if single thread */
  63 };
  64
  65 /* All the per-cpu workqueues on the system, for hotplug cpu to add/remove
  66    threads to each one as cpus come/go. */
  67 static DEFINE_SPINLOCK(workqueue_lock);
  68 static LIST_HEAD(workqueues);
  69
  70 /* If it's single threaded, it isn't in the list of workqueues. */
  71 static inline int is_single_threaded(struct workqueue_struct *wq)
  72 {
  73         return list_empty(&wq->list);
  74 }
  75
  76 /* Preempt must be disabled. */
  77 static void __queue_work(struct cpu_workqueue_struct *cwq,
  78                          struct work_struct *work)
  79 {
  80         unsigned long flags;
  81
  82         spin_lock_irqsave(&cwq->lock, flags);
  83         work->wq_data = cwq;
  84         list_add_tail(&work->entry, &cwq->worklist);
  85         cwq->insert_sequence++;
  86         wake_up(&cwq->more_work);
  87         spin_unlock_irqrestore(&cwq->lock, flags);
  88 }
  89
  90 /*
  91  * Queue work on a workqueue. Return non-zero if it was successfully
  92  * added.
  93  *
  94  * We queue the work to the CPU it was submitted, but there is no
  95  * guarantee that it will be processed by that CPU.
  96  */
  97 int fastcall queue_work(struct workqueue_struct *wq, struct work_struct *work)
  98 {
  99         int ret = 0, cpu = get_cpu();
 100
 101         if (!test_and_set_bit(0, &work->pending)) {
 102                 if (unlikely(is_single_threaded(wq)))
 103                         cpu = 0;
 104                 BUG_ON(!list_empty(&work->entry));
 105                 __queue_work(wq->cpu_wq + cpu, work);
 106                 ret = 1;
 107         }
 108         put_cpu();
 109         return ret;
 110 }
 111
 112 static void delayed_work_timer_fn(unsigned long __data)
 113 {
 114         struct work_struct *work = (struct work_struct *)__data;
 115         struct workqueue_struct *wq = work->wq_data;
 116         int cpu = smp_processor_id();
 117
 118         if (unlikely(is_single_threaded(wq)))
 119                 cpu = 0;
 120
 121         __queue_work(wq->cpu_wq + cpu, work);
 122 }
 123
 124 int fastcall queue_delayed_work(struct workqueue_struct *wq,
 125                         struct work_struct *work, unsigned long delay)
 126 {
 127         int ret = 0;
 128         struct timer_list *timer = &work->timer;
 129
 130         if (!test_and_set_bit(0, &work->pending)) {
 131                 BUG_ON(timer_pending(timer));
 132                 BUG_ON(!list_empty(&work->entry));
 133
 134                 /* This stores wq for the moment, for the timer_fn */
 135                 work->wq_data = wq;
 136                 timer->expires = jiffies + delay;
 137                 timer->data = (unsigned long)work;
 138                 timer->function = delayed_work_timer_fn;
 139                 add_timer(timer);
 140                 ret = 1;
 141         }
 142         return ret;
 143 }
 144
 145 static inline void run_workqueue(struct cpu_workqueue_struct *cwq)
 146 {
 147         unsigned long flags;
 148
 149         /*
 150          * Keep taking off work from the queue until
 151          * done.
 152          */
 153         spin_lock_irqsave(&cwq->lock, flags);
 154         cwq->run_depth++;
 155         if (cwq->run_depth > 3) {
 156                 /* morton gets to eat his hat */
 157                 printk("%s: recursion depth exceeded: %d\n",
 158                         __FUNCTION__, cwq->run_depth);
 159                 dump_stack();
 160         }
 161         while (!list_empty(&cwq->worklist)) {
 162                 struct work_struct *work = list_entry(cwq->worklist.next,
 163                                                 struct work_struct, entry);
 164                 void (*f) (void *) = work->func;
 165                 void *data = work->data;
 166
 167                 list_del_init(cwq->worklist.next);
 168                 spin_unlock_irqrestore(&cwq->lock, flags);
 169
 170                 BUG_ON(work->wq_data != cwq);
 171                 clear_bit(0, &work->pending);
 172                 f(data);
 173
 174                 spin_lock_irqsave(&cwq->lock, flags);
 175                 cwq->remove_sequence++;
 176                 wake_up(&cwq->work_done);
 177         }
 178         cwq->run_depth--;
 179         spin_unlock_irqrestore(&cwq->lock, flags);
 180 }
 181
 182 static int worker_thread(void *__cwq)
 183 {
 184         struct cpu_workqueue_struct *cwq = __cwq;
 185         DECLARE_WAITQUEUE(wait, current);
 186         struct k_sigaction sa;
 187         sigset_t blocked;
 188
 189         current->flags |= PF_NOFREEZE;
 190
 191         set_user_nice(current, -5);
 192
 193         /* Block and flush all signals */
 194         sigfillset(&blocked);
 195         sigprocmask(SIG_BLOCK, &blocked, NULL);
 196         flush_signals(current);
 197
 198         /* SIG_IGN makes children autoreap: see do_notify_parent(). */
 199         sa.sa.sa_handler = SIG_IGN;
 200         sa.sa.sa_flags = 0;
 201         siginitset(&sa.sa.sa_mask, sigmask(SIGCHLD));
 202         do_sigaction(SIGCHLD, &sa, (struct k_sigaction *)0);
 203
 204         set_current_state(TASK_INTERRUPTIBLE);
 205         while (!kthread_should_stop()) {
 206                 add_wait_queue(&cwq->more_work, &wait);
 207                 if (list_empty(&cwq->worklist))
 208                         schedule();
 209                 else
 210                         __set_current_state(TASK_RUNNING);
 211                 remove_wait_queue(&cwq->more_work, &wait);
 212
 213                 if (!list_empty(&cwq->worklist))
 214                         run_workqueue(cwq);
 215                 set_current_state(TASK_INTERRUPTIBLE);
 216         }
 217         __set_current_state(TASK_RUNNING);
 218         return 0;
 219 }
 220
 221 static void flush_cpu_workqueue(struct cpu_workqueue_struct *cwq)
 222 {
 223         if (cwq->thread == current) {
 224                 /*
 225                  * Probably keventd trying to flush its own queue. So simply run
 226                  * it by hand rather than deadlocking.
 227                  */
 228                 run_workqueue(cwq);
 229         } else {
 230                 DEFINE_WAIT(wait);
 231                 long sequence_needed;
 232
 233                 spin_lock_irq(&cwq->lock);
 234                 sequence_needed = cwq->insert_sequence;
 235
 236                 while (sequence_needed - cwq->remove_sequence > 0) {
 237                         prepare_to_wait(&cwq->work_done, &wait,
 238                                         TASK_UNINTERRUPTIBLE);
 239                         spin_unlock_irq(&cwq->lock);
 240                         schedule();
 241                         spin_lock_irq(&cwq->lock);
 242                 }
 243                 finish_wait(&cwq->work_done, &wait);
 244                 spin_unlock_irq(&cwq->lock);
 245         }
 246 }
 247
 248 /*
 249  * flush_workqueue - ensure that any scheduled work has run to completion.
 250  *
 251  * Forces execution of the workqueue and blocks until its completion.
 252  * This is typically used in driver shutdown handlers.
 253  *
 254  * This function will sample each workqueue's current insert_sequence number and
 255  * will sleep until the head sequence is greater than or equal to that.  This
 256  * means that we sleep until all works which were queued on entry have been
 257  * handled, but we are not livelocked by new incoming ones.
 258  *
 259  * This function used to run the workqueues itself.  Now we just wait for the
 260  * helper threads to do it.
 261  */
 262 void fastcall flush_workqueue(struct workqueue_struct *wq)
 263 {
 264         might_sleep();
 265
 266         if (is_single_threaded(wq)) {
 267                 /* Always use cpu 0's area. */
 268                 flush_cpu_workqueue(wq->cpu_wq + 0);
 269         } else {
 270                 int cpu;
 271
 272                 lock_cpu_hotplug();
 273                 for_each_online_cpu(cpu)
 274                         flush_cpu_workqueue(wq->cpu_wq + cpu);
 275                 unlock_cpu_hotplug();
 276         }
 277 }
 278
 279 static struct task_struct *create_workqueue_thread(struct workqueue_struct *wq,
 280                                                    int cpu)
 281 {
 282         struct cpu_workqueue_struct *cwq = wq->cpu_wq + cpu;
 283         struct task_struct *p;
 284
 285         spin_lock_init(&cwq->lock);
 286         cwq->wq = wq;
 287         cwq->thread = NULL;
 288         cwq->insert_sequence = 0;
 289         cwq->remove_sequence = 0;
 290         INIT_LIST_HEAD(&cwq->worklist);
 291         init_waitqueue_head(&cwq->more_work);
 292         init_waitqueue_head(&cwq->work_done);
 293
 294         if (is_single_threaded(wq))
 295                 p = kthread_create(worker_thread, cwq, "%s", wq->name);
 296         else
 297                 p = kthread_create(worker_thread, cwq, "%s/%d", wq->name, cpu);
 298         if (IS_ERR(p))
 299                 return NULL;
 300         cwq->thread = p;
 301         return p;
 302 }
 303
 304 struct workqueue_struct *__create_workqueue(const char *name,
 305                                             int singlethread)
 306 {
 307         int cpu, destroy = 0;
 308         struct workqueue_struct *wq;
 309         struct task_struct *p;
 310
 311         wq = kzalloc(sizeof(*wq), GFP_KERNEL);
 312         if (!wq)
 313                 return NULL;
 314
 315         wq->name = name;
 316         /* We don't need the distraction of CPUs appearing and vanishing. */
 317         lock_cpu_hotplug();
 318         if (singlethread) {
 319                 INIT_LIST_HEAD(&wq->list);
 320                 p = create_workqueue_thread(wq, 0);
 321                 if (!p)
 322                         destroy = 1;
 323                 else
 324                         wake_up_process(p);
 325         } else {
 326                 spin_lock(&workqueue_lock);
 327                 list_add(&wq->list, &workqueues);
 328                 spin_unlock(&workqueue_lock);
 329                 for_each_online_cpu(cpu) {
 330                         p = create_workqueue_thread(wq, cpu);
 331                         if (p) {
 332                                 kthread_bind(p, cpu);
 333                                 wake_up_process(p);
 334                         } else
 335                                 destroy = 1;
 336                 }
 337         }
 338         unlock_cpu_hotplug();
 339
 340         /*
 341          * Was there any error during startup? If yes then clean up:
 342          */
 343         if (destroy) {
 344                 destroy_workqueue(wq);
 345                 wq = NULL;
 346         }
 347         return wq;
 348 }
 349
 350 static void cleanup_workqueue_thread(struct workqueue_struct *wq, int cpu)
 351 {
 352         struct cpu_workqueue_struct *cwq;
 353         unsigned long flags;
 354         struct task_struct *p;
 355
 356         cwq = wq->cpu_wq + cpu;
 357         spin_lock_irqsave(&cwq->lock, flags);
 358         p = cwq->thread;
 359         cwq->thread = NULL;
 360         spin_unlock_irqrestore(&cwq->lock, flags);
 361         if (p)
 362                 kthread_stop(p);
 363 }
 364
 365 void destroy_workqueue(struct workqueue_struct *wq)
 366 {
 367         int cpu;
 368
 369         flush_workqueue(wq);
 370
 371         /* We don't need the distraction of CPUs appearing and vanishing. */
 372         lock_cpu_hotplug();
 373         if (is_single_threaded(wq))
 374                 cleanup_workqueue_thread(wq, 0);
 375         else {
 376                 for_each_online_cpu(cpu)
 377                         cleanup_workqueue_thread(wq, cpu);
 378                 spin_lock(&workqueue_lock);
 379                 list_del(&wq->list);
 380                 spin_unlock(&workqueue_lock);
 381         }
 382         unlock_cpu_hotplug();
 383         kfree(wq);
 384 }
 385
 386 static struct workqueue_struct *keventd_wq;
 387
 388 int fastcall schedule_work(struct work_struct *work)
 389 {
 390         return queue_work(keventd_wq, work);
 391 }
 392
 393 int fastcall schedule_delayed_work(struct work_struct *work, unsigned long delay)
 394 {
 395         return queue_delayed_work(keventd_wq, work, delay);
 396 }
 397
 398 int schedule_delayed_work_on(int cpu,
 399                         struct work_struct *work, unsigned long delay)
 400 {
 401         int ret = 0;
 402         struct timer_list *timer = &work->timer;
 403
 404         if (!test_and_set_bit(0, &work->pending)) {
 405                 BUG_ON(timer_pending(timer));
 406                 BUG_ON(!list_empty(&work->entry));
 407                 /* This stores keventd_wq for the moment, for the timer_fn */
 408                 work->wq_data = keventd_wq;
 409                 timer->expires = jiffies + delay;
 410                 timer->data = (unsigned long)work;
 411                 timer->function = delayed_work_timer_fn;
 412                 add_timer_on(timer, cpu);
 413                 ret = 1;
 414         }
 415         return ret;
 416 }
 417
 418 void flush_scheduled_work(void)
 419 {
 420         flush_workqueue(keventd_wq);
 421 }
 422
 423 /**
 424  * cancel_rearming_delayed_workqueue - reliably kill off a delayed
 425  *                      work whose handler rearms the delayed work.
 426  * @wq:   the controlling workqueue structure
 427  * @work: the delayed work struct
 428  */
 429 void cancel_rearming_delayed_workqueue(struct workqueue_struct *wq,
 430                                        struct work_struct *work)
 431 {
 432         while (!cancel_delayed_work(work))
 433                 flush_workqueue(wq);
 434 }
 435 EXPORT_SYMBOL(cancel_rearming_delayed_workqueue);
 436
 437 /**
 438  * cancel_rearming_delayed_work - reliably kill off a delayed keventd
 439  *                      work whose handler rearms the delayed work.
 440  * @work: the delayed work struct
 441  */
 442 void cancel_rearming_delayed_work(struct work_struct *work)
 443 {
 444         cancel_rearming_delayed_workqueue(keventd_wq, work);
 445 }
 446 EXPORT_SYMBOL(cancel_rearming_delayed_work);
 447
 448 int keventd_up(void)
 449 {
 450         return keventd_wq != NULL;
 451 }
 452
 453 int current_is_keventd(void)
 454 {
 455         struct cpu_workqueue_struct *cwq;
 456         int cpu = smp_processor_id();   /* preempt-safe: keventd is per-cpu */
 457         int ret = 0;
 458
 459         BUG_ON(!keventd_wq);
 460
 461         cwq = keventd_wq->cpu_wq + cpu;
 462         if (current == cwq->thread)
 463                 ret = 1;
 464
 465         return ret;
 466
 467 }
 468
 469 #ifdef CONFIG_HOTPLUG_CPU
 470 /* Take the work from this (downed) CPU. */
 471 static void take_over_work(struct workqueue_struct *wq, unsigned int cpu)
 472 {
 473         struct cpu_workqueue_struct *cwq = wq->cpu_wq + cpu;
 474         LIST_HEAD(list);
 475         struct work_struct *work;
 476
 477         spin_lock_irq(&cwq->lock);
 478         list_splice_init(&cwq->worklist, &list);
 479
 480         while (!list_empty(&list)) {
 481                 printk("Taking work for %s\n", wq->name);
 482                 work = list_entry(list.next,struct work_struct,entry);
 483                 list_del(&work->entry);
 484                 __queue_work(wq->cpu_wq + smp_processor_id(), work);
 485         }
 486         spin_unlock_irq(&cwq->lock);
 487 }
 488
 489 /* We're holding the cpucontrol mutex here */
 490 static int __devinit workqueue_cpu_callback(struct notifier_block *nfb,
 491                                   unsigned long action,
 492                                   void *hcpu)
 493 {
 494         unsigned int hotcpu = (unsigned long)hcpu;
 495         struct workqueue_struct *wq;
 496
 497         switch (action) {
 498         case CPU_UP_PREPARE:
 499                 /* Create a new workqueue thread for it. */
 500                 list_for_each_entry(wq, &workqueues, list) {
 501                         if (!create_workqueue_thread(wq, hotcpu)) {
 502                                 printk("workqueue for %i failed\n", hotcpu);
 503                                 return NOTIFY_BAD;
 504                         }
 505                 }
 506                 break;
 507
 508         case CPU_ONLINE:
 509                 /* Kick off worker threads. */
 510                 list_for_each_entry(wq, &workqueues, list) {
 511                         kthread_bind(wq->cpu_wq[hotcpu].thread, hotcpu);
 512                         wake_up_process(wq->cpu_wq[hotcpu].thread);
 513                 }
 514                 break;
 515
 516         case CPU_UP_CANCELED:
 517                 list_for_each_entry(wq, &workqueues, list) {
 518                         /* Unbind so it can run. */
 519                         kthread_bind(wq->cpu_wq[hotcpu].thread,
 520                                      smp_processor_id());
 521                         cleanup_workqueue_thread(wq, hotcpu);
 522                 }
 523                 break;
 524
 525         case CPU_DEAD:
 526                 list_for_each_entry(wq, &workqueues, list)
 527                         cleanup_workqueue_thread(wq, hotcpu);
 528                 list_for_each_entry(wq, &workqueues, list)
 529                         take_over_work(wq, hotcpu);
 530                 break;
 531         }
 532
 533         return NOTIFY_OK;
 534 }
 535 #endif
 536
 537 void init_workqueues(void)
 538 {
 539         hotcpu_notifier(workqueue_cpu_callback, 0);
 540         keventd_wq = create_workqueue("events");
 541         BUG_ON(!keventd_wq);
 542 }
 543
 544 EXPORT_SYMBOL_GPL(__create_workqueue);
 545 EXPORT_SYMBOL_GPL(queue_work);
 546 EXPORT_SYMBOL_GPL(queue_delayed_work);
 547 EXPORT_SYMBOL_GPL(flush_workqueue);
 548 EXPORT_SYMBOL_GPL(destroy_workqueue);
 549
 550 EXPORT_SYMBOL(schedule_work);
 551 EXPORT_SYMBOL(schedule_delayed_work);
 552 EXPORT_SYMBOL(schedule_delayed_work_on);
 553 EXPORT_SYMBOL(flush_scheduled_work);