xref: /linux/kernel/workqueue.c (revision 776cfebb430c7b22c208b1b17add97f354d97cab)
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 	BUG_ON(strlen(name) > 10);
312 
313 	wq = kmalloc(sizeof(*wq), GFP_KERNEL);
314 	if (!wq)
315 		return NULL;
316 	memset(wq, 0, sizeof(*wq));
317 
318 	wq->name = name;
319 	/* We don't need the distraction of CPUs appearing and vanishing. */
320 	lock_cpu_hotplug();
321 	if (singlethread) {
322 		INIT_LIST_HEAD(&wq->list);
323 		p = create_workqueue_thread(wq, 0);
324 		if (!p)
325 			destroy = 1;
326 		else
327 			wake_up_process(p);
328 	} else {
329 		spin_lock(&workqueue_lock);
330 		list_add(&wq->list, &workqueues);
331 		spin_unlock(&workqueue_lock);
332 		for_each_online_cpu(cpu) {
333 			p = create_workqueue_thread(wq, cpu);
334 			if (p) {
335 				kthread_bind(p, cpu);
336 				wake_up_process(p);
337 			} else
338 				destroy = 1;
339 		}
340 	}
341 	unlock_cpu_hotplug();
342 
343 	/*
344 	 * Was there any error during startup? If yes then clean up:
345 	 */
346 	if (destroy) {
347 		destroy_workqueue(wq);
348 		wq = NULL;
349 	}
350 	return wq;
351 }
352 
353 static void cleanup_workqueue_thread(struct workqueue_struct *wq, int cpu)
354 {
355 	struct cpu_workqueue_struct *cwq;
356 	unsigned long flags;
357 	struct task_struct *p;
358 
359 	cwq = wq->cpu_wq + cpu;
360 	spin_lock_irqsave(&cwq->lock, flags);
361 	p = cwq->thread;
362 	cwq->thread = NULL;
363 	spin_unlock_irqrestore(&cwq->lock, flags);
364 	if (p)
365 		kthread_stop(p);
366 }
367 
368 void destroy_workqueue(struct workqueue_struct *wq)
369 {
370 	int cpu;
371 
372 	flush_workqueue(wq);
373 
374 	/* We don't need the distraction of CPUs appearing and vanishing. */
375 	lock_cpu_hotplug();
376 	if (is_single_threaded(wq))
377 		cleanup_workqueue_thread(wq, 0);
378 	else {
379 		for_each_online_cpu(cpu)
380 			cleanup_workqueue_thread(wq, cpu);
381 		spin_lock(&workqueue_lock);
382 		list_del(&wq->list);
383 		spin_unlock(&workqueue_lock);
384 	}
385 	unlock_cpu_hotplug();
386 	kfree(wq);
387 }
388 
389 static struct workqueue_struct *keventd_wq;
390 
391 int fastcall schedule_work(struct work_struct *work)
392 {
393 	return queue_work(keventd_wq, work);
394 }
395 
396 int fastcall schedule_delayed_work(struct work_struct *work, unsigned long delay)
397 {
398 	return queue_delayed_work(keventd_wq, work, delay);
399 }
400 
401 int schedule_delayed_work_on(int cpu,
402 			struct work_struct *work, unsigned long delay)
403 {
404 	int ret = 0;
405 	struct timer_list *timer = &work->timer;
406 
407 	if (!test_and_set_bit(0, &work->pending)) {
408 		BUG_ON(timer_pending(timer));
409 		BUG_ON(!list_empty(&work->entry));
410 		/* This stores keventd_wq for the moment, for the timer_fn */
411 		work->wq_data = keventd_wq;
412 		timer->expires = jiffies + delay;
413 		timer->data = (unsigned long)work;
414 		timer->function = delayed_work_timer_fn;
415 		add_timer_on(timer, cpu);
416 		ret = 1;
417 	}
418 	return ret;
419 }
420 
421 void flush_scheduled_work(void)
422 {
423 	flush_workqueue(keventd_wq);
424 }
425 
426 /**
427  * cancel_rearming_delayed_workqueue - reliably kill off a delayed
428  *			work whose handler rearms the delayed work.
429  * @wq:   the controlling workqueue structure
430  * @work: the delayed work struct
431  */
432 void cancel_rearming_delayed_workqueue(struct workqueue_struct *wq,
433 				       struct work_struct *work)
434 {
435 	while (!cancel_delayed_work(work))
436 		flush_workqueue(wq);
437 }
438 EXPORT_SYMBOL(cancel_rearming_delayed_workqueue);
439 
440 /**
441  * cancel_rearming_delayed_work - reliably kill off a delayed keventd
442  *			work whose handler rearms the delayed work.
443  * @work: the delayed work struct
444  */
445 void cancel_rearming_delayed_work(struct work_struct *work)
446 {
447 	cancel_rearming_delayed_workqueue(keventd_wq, work);
448 }
449 EXPORT_SYMBOL(cancel_rearming_delayed_work);
450 
451 int keventd_up(void)
452 {
453 	return keventd_wq != NULL;
454 }
455 
456 int current_is_keventd(void)
457 {
458 	struct cpu_workqueue_struct *cwq;
459 	int cpu = smp_processor_id();	/* preempt-safe: keventd is per-cpu */
460 	int ret = 0;
461 
462 	BUG_ON(!keventd_wq);
463 
464 	cwq = keventd_wq->cpu_wq + cpu;
465 	if (current == cwq->thread)
466 		ret = 1;
467 
468 	return ret;
469 
470 }
471 
472 #ifdef CONFIG_HOTPLUG_CPU
473 /* Take the work from this (downed) CPU. */
474 static void take_over_work(struct workqueue_struct *wq, unsigned int cpu)
475 {
476 	struct cpu_workqueue_struct *cwq = wq->cpu_wq + cpu;
477 	LIST_HEAD(list);
478 	struct work_struct *work;
479 
480 	spin_lock_irq(&cwq->lock);
481 	list_splice_init(&cwq->worklist, &list);
482 
483 	while (!list_empty(&list)) {
484 		printk("Taking work for %s\n", wq->name);
485 		work = list_entry(list.next,struct work_struct,entry);
486 		list_del(&work->entry);
487 		__queue_work(wq->cpu_wq + smp_processor_id(), work);
488 	}
489 	spin_unlock_irq(&cwq->lock);
490 }
491 
492 /* We're holding the cpucontrol mutex here */
493 static int __devinit workqueue_cpu_callback(struct notifier_block *nfb,
494 				  unsigned long action,
495 				  void *hcpu)
496 {
497 	unsigned int hotcpu = (unsigned long)hcpu;
498 	struct workqueue_struct *wq;
499 
500 	switch (action) {
501 	case CPU_UP_PREPARE:
502 		/* Create a new workqueue thread for it. */
503 		list_for_each_entry(wq, &workqueues, list) {
504 			if (create_workqueue_thread(wq, hotcpu) < 0) {
505 				printk("workqueue for %i failed\n", hotcpu);
506 				return NOTIFY_BAD;
507 			}
508 		}
509 		break;
510 
511 	case CPU_ONLINE:
512 		/* Kick off worker threads. */
513 		list_for_each_entry(wq, &workqueues, list) {
514 			kthread_bind(wq->cpu_wq[hotcpu].thread, hotcpu);
515 			wake_up_process(wq->cpu_wq[hotcpu].thread);
516 		}
517 		break;
518 
519 	case CPU_UP_CANCELED:
520 		list_for_each_entry(wq, &workqueues, list) {
521 			/* Unbind so it can run. */
522 			kthread_bind(wq->cpu_wq[hotcpu].thread,
523 				     smp_processor_id());
524 			cleanup_workqueue_thread(wq, hotcpu);
525 		}
526 		break;
527 
528 	case CPU_DEAD:
529 		list_for_each_entry(wq, &workqueues, list)
530 			cleanup_workqueue_thread(wq, hotcpu);
531 		list_for_each_entry(wq, &workqueues, list)
532 			take_over_work(wq, hotcpu);
533 		break;
534 	}
535 
536 	return NOTIFY_OK;
537 }
538 #endif
539 
540 void init_workqueues(void)
541 {
542 	hotcpu_notifier(workqueue_cpu_callback, 0);
543 	keventd_wq = create_workqueue("events");
544 	BUG_ON(!keventd_wq);
545 }
546 
547 EXPORT_SYMBOL_GPL(__create_workqueue);
548 EXPORT_SYMBOL_GPL(queue_work);
549 EXPORT_SYMBOL_GPL(queue_delayed_work);
550 EXPORT_SYMBOL_GPL(flush_workqueue);
551 EXPORT_SYMBOL_GPL(destroy_workqueue);
552 
553 EXPORT_SYMBOL(schedule_work);
554 EXPORT_SYMBOL(schedule_delayed_work);
555 EXPORT_SYMBOL(schedule_delayed_work_on);
556 EXPORT_SYMBOL(flush_scheduled_work);
557