linux_schedule.c revision 330851 
1/*-
2 * Copyright (c) 2017 Mark Johnston <markj@FreeBSD.org>
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 conds
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice unmodified, this list of conds, and the following
10 *    disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 *    notice, this list of conds and the following disclaimer in the
13 *    documentation and/or other materials provided with the distribution.
14 *
15 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
16 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
17 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
18 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
19 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
20 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
21 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
22 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
23 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
24 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
25 */
26
27#include <sys/cdefs.h>
28__FBSDID("$FreeBSD: stable/11/sys/compat/linuxkpi/common/src/linux_schedule.c 330851 2018-03-13 16:17:36Z hselasky $");
29
30#include <sys/param.h>
31#include <sys/systm.h>
32#include <sys/proc.h>
33#include <sys/signalvar.h>
34#include <sys/sleepqueue.h>
35
36#include <linux/delay.h>
37#include <linux/errno.h>
38#include <linux/kernel.h>
39#include <linux/list.h>
40#include <linux/sched.h>
41#include <linux/spinlock.h>
42#include <linux/wait.h>
43
44static int
45linux_add_to_sleepqueue(void *wchan, struct task_struct *task,
46    const char *wmesg, int timeout, int state)
47{
48	int flags, ret;
49
50	MPASS((state & ~TASK_NORMAL) == 0);
51
52	flags = SLEEPQ_SLEEP | ((state & TASK_INTERRUPTIBLE) != 0 ?
53	    SLEEPQ_INTERRUPTIBLE : 0);
54
55	sleepq_add(wchan, NULL, wmesg, flags, 0);
56	if (timeout != 0)
57		sleepq_set_timeout(wchan, timeout);
58	if ((state & TASK_INTERRUPTIBLE) != 0) {
59		if (timeout == 0)
60			ret = -sleepq_wait_sig(wchan, 0);
61		else
62			ret = -sleepq_timedwait_sig(wchan, 0);
63	} else {
64		if (timeout == 0) {
65			sleepq_wait(wchan, 0);
66			ret = 0;
67		} else
68			ret = -sleepq_timedwait(wchan, 0);
69	}
70	/* filter return value */
71	if (ret != 0 && ret != -EWOULDBLOCK) {
72		linux_schedule_save_interrupt_value(task, ret);
73		ret = -ERESTARTSYS;
74	}
75	return (ret);
76}
77
78unsigned int
79linux_msleep_interruptible(unsigned int ms)
80{
81	int ret;
82
83	/* guard against invalid values */
84	if (ms == 0)
85		ms = 1;
86	ret = -pause_sbt("lnxsleep", mstosbt(ms), 0, C_HARDCLOCK | C_CATCH);
87
88	switch (ret) {
89	case -EWOULDBLOCK:
90		return (0);
91	default:
92		linux_schedule_save_interrupt_value(current, ret);
93		return (ms);
94	}
95}
96
97static int
98wake_up_task(struct task_struct *task, unsigned int state)
99{
100	int ret, wakeup_swapper;
101
102	ret = wakeup_swapper = 0;
103	sleepq_lock(task);
104	if ((atomic_read(&task->state) & state) != 0) {
105		set_task_state(task, TASK_WAKING);
106		wakeup_swapper = sleepq_signal(task, SLEEPQ_SLEEP, 0, 0);
107		ret = 1;
108	}
109	sleepq_release(task);
110	if (wakeup_swapper)
111		kick_proc0();
112	return (ret);
113}
114
115bool
116linux_signal_pending(struct task_struct *task)
117{
118	struct thread *td;
119	sigset_t pending;
120
121	td = task->task_thread;
122	PROC_LOCK(td->td_proc);
123	pending = td->td_siglist;
124	SIGSETOR(pending, td->td_proc->p_siglist);
125	SIGSETNAND(pending, td->td_sigmask);
126	PROC_UNLOCK(td->td_proc);
127	return (!SIGISEMPTY(pending));
128}
129
130bool
131linux_fatal_signal_pending(struct task_struct *task)
132{
133	struct thread *td;
134	bool ret;
135
136	td = task->task_thread;
137	PROC_LOCK(td->td_proc);
138	ret = SIGISMEMBER(td->td_siglist, SIGKILL) ||
139	    SIGISMEMBER(td->td_proc->p_siglist, SIGKILL);
140	PROC_UNLOCK(td->td_proc);
141	return (ret);
142}
143
144bool
145linux_signal_pending_state(long state, struct task_struct *task)
146{
147
148	MPASS((state & ~TASK_NORMAL) == 0);
149
150	if ((state & TASK_INTERRUPTIBLE) == 0)
151		return (false);
152	return (linux_signal_pending(task));
153}
154
155void
156linux_send_sig(int signo, struct task_struct *task)
157{
158	struct thread *td;
159
160	td = task->task_thread;
161	PROC_LOCK(td->td_proc);
162	tdsignal(td, signo);
163	PROC_UNLOCK(td->td_proc);
164}
165
166int
167autoremove_wake_function(wait_queue_t *wq, unsigned int state, int flags,
168    void *key __unused)
169{
170	struct task_struct *task;
171	int ret;
172
173	task = wq->private;
174	if ((ret = wake_up_task(task, state)) != 0)
175		list_del_init(&wq->task_list);
176	return (ret);
177}
178
179void
180linux_wake_up(wait_queue_head_t *wqh, unsigned int state, int nr, bool locked)
181{
182	wait_queue_t *pos, *next;
183
184	if (!locked)
185		spin_lock(&wqh->lock);
186	list_for_each_entry_safe(pos, next, &wqh->task_list, task_list) {
187		if (pos->func == NULL) {
188			if (wake_up_task(pos->private, state) != 0 && --nr == 0)
189				break;
190		} else {
191			if (pos->func(pos, state, 0, NULL) != 0 && --nr == 0)
192				break;
193		}
194	}
195	if (!locked)
196		spin_unlock(&wqh->lock);
197}
198
199void
200linux_prepare_to_wait(wait_queue_head_t *wqh, wait_queue_t *wq, int state)
201{
202
203	spin_lock(&wqh->lock);
204	if (list_empty(&wq->task_list))
205		__add_wait_queue(wqh, wq);
206	set_task_state(current, state);
207	spin_unlock(&wqh->lock);
208}
209
210void
211linux_finish_wait(wait_queue_head_t *wqh, wait_queue_t *wq)
212{
213
214	spin_lock(&wqh->lock);
215	set_task_state(current, TASK_RUNNING);
216	if (!list_empty(&wq->task_list)) {
217		__remove_wait_queue(wqh, wq);
218		INIT_LIST_HEAD(&wq->task_list);
219	}
220	spin_unlock(&wqh->lock);
221}
222
223bool
224linux_waitqueue_active(wait_queue_head_t *wqh)
225{
226	bool ret;
227
228	spin_lock(&wqh->lock);
229	ret = !list_empty(&wqh->task_list);
230	spin_unlock(&wqh->lock);
231	return (ret);
232}
233
234int
235linux_wait_event_common(wait_queue_head_t *wqh, wait_queue_t *wq, int timeout,
236    unsigned int state, spinlock_t *lock)
237{
238	struct task_struct *task;
239	int ret;
240
241	if (lock != NULL)
242		spin_unlock_irq(lock);
243
244	DROP_GIANT();
245
246	/* range check timeout */
247	if (timeout < 1)
248		timeout = 1;
249	else if (timeout == MAX_SCHEDULE_TIMEOUT)
250		timeout = 0;
251
252	task = current;
253
254	/*
255	 * Our wait queue entry is on the stack - make sure it doesn't
256	 * get swapped out while we sleep.
257	 */
258	PHOLD(task->task_thread->td_proc);
259	sleepq_lock(task);
260	if (atomic_read(&task->state) != TASK_WAKING) {
261		ret = linux_add_to_sleepqueue(task, task, "wevent", timeout, state);
262	} else {
263		sleepq_release(task);
264		ret = 0;
265	}
266	PRELE(task->task_thread->td_proc);
267
268	PICKUP_GIANT();
269
270	if (lock != NULL)
271		spin_lock_irq(lock);
272	return (ret);
273}
274
275int
276linux_schedule_timeout(int timeout)
277{
278	struct task_struct *task;
279	int ret;
280	int state;
281	int remainder;
282
283	task = current;
284
285	/* range check timeout */
286	if (timeout < 1)
287		timeout = 1;
288	else if (timeout == MAX_SCHEDULE_TIMEOUT)
289		timeout = 0;
290
291	remainder = ticks + timeout;
292
293	DROP_GIANT();
294
295	sleepq_lock(task);
296	state = atomic_read(&task->state);
297	if (state != TASK_WAKING) {
298		ret = linux_add_to_sleepqueue(task, task, "sched", timeout, state);
299	} else {
300		sleepq_release(task);
301		ret = 0;
302	}
303	set_task_state(task, TASK_RUNNING);
304
305	PICKUP_GIANT();
306
307	if (timeout == 0)
308		return (MAX_SCHEDULE_TIMEOUT);
309
310	/* range check return value */
311	remainder -= ticks;
312
313	/* range check return value */
314	if (ret == -ERESTARTSYS && remainder < 1)
315		remainder = 1;
316	else if (remainder < 0)
317		remainder = 0;
318	else if (remainder > timeout)
319		remainder = timeout;
320	return (remainder);
321}
322
323static void
324wake_up_sleepers(void *wchan)
325{
326	int wakeup_swapper;
327
328	sleepq_lock(wchan);
329	wakeup_swapper = sleepq_signal(wchan, SLEEPQ_SLEEP, 0, 0);
330	sleepq_release(wchan);
331	if (wakeup_swapper)
332		kick_proc0();
333}
334
335#define	bit_to_wchan(word, bit)	((void *)(((uintptr_t)(word) << 6) | (bit)))
336
337void
338linux_wake_up_bit(void *word, int bit)
339{
340
341	wake_up_sleepers(bit_to_wchan(word, bit));
342}
343
344int
345linux_wait_on_bit_timeout(unsigned long *word, int bit, unsigned int state,
346    int timeout)
347{
348	struct task_struct *task;
349	void *wchan;
350	int ret;
351
352	DROP_GIANT();
353
354	/* range check timeout */
355	if (timeout < 1)
356		timeout = 1;
357	else if (timeout == MAX_SCHEDULE_TIMEOUT)
358		timeout = 0;
359
360	task = current;
361	wchan = bit_to_wchan(word, bit);
362	for (;;) {
363		sleepq_lock(wchan);
364		if ((*word & (1 << bit)) == 0) {
365			sleepq_release(wchan);
366			ret = 0;
367			break;
368		}
369		set_task_state(task, state);
370		ret = linux_add_to_sleepqueue(wchan, task, "wbit", timeout, state);
371		if (ret != 0)
372			break;
373	}
374	set_task_state(task, TASK_RUNNING);
375
376	PICKUP_GIANT();
377
378	return (ret);
379}
380
381void
382linux_wake_up_atomic_t(atomic_t *a)
383{
384
385	wake_up_sleepers(a);
386}
387
388int
389linux_wait_on_atomic_t(atomic_t *a, unsigned int state)
390{
391	struct task_struct *task;
392	void *wchan;
393	int ret;
394
395	DROP_GIANT();
396
397	task = current;
398	wchan = a;
399	for (;;) {
400		sleepq_lock(wchan);
401		if (atomic_read(a) == 0) {
402			sleepq_release(wchan);
403			ret = 0;
404			break;
405		}
406		set_task_state(task, state);
407		ret = linux_add_to_sleepqueue(wchan, task, "watomic", 0, state);
408		if (ret != 0)
409			break;
410	}
411	set_task_state(task, TASK_RUNNING);
412
413	PICKUP_GIANT();
414
415	return (ret);
416}
417
418bool
419linux_wake_up_state(struct task_struct *task, unsigned int state)
420{
421
422	return (wake_up_task(task, state) != 0);
423}
424