posix-timers.c 30.1 KB
Newer Older
L
Linus Torvalds 已提交
1
/*
2
 * linux/kernel/posix-timers.c
L
Linus Torvalds 已提交
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
 *
 *
 * 2002-10-15  Posix Clocks & timers
 *                           by George Anzinger george@mvista.com
 *
 *			     Copyright (C) 2002 2003 by MontaVista Software.
 *
 * 2004-06-01  Fix CLOCK_REALTIME clock/timer TIMER_ABSTIME bug.
 *			     Copyright (C) 2004 Boris Hu
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or (at
 * your option) any later version.
 *
 * This program is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
 * General Public License for more details.

 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 *
 * MontaVista Software | 1237 East Arques Avenue | Sunnyvale | CA 94085 | USA
 */

/* These are all the functions necessary to implement
 * POSIX clocks & timers
 */
#include <linux/mm.h>
#include <linux/interrupt.h>
#include <linux/slab.h>
#include <linux/time.h>
A
Arjan van de Ven 已提交
37
#include <linux/mutex.h>
38
#include <linux/sched/task.h>
L
Linus Torvalds 已提交
39

40
#include <linux/uaccess.h>
L
Linus Torvalds 已提交
41 42 43
#include <linux/list.h>
#include <linux/init.h>
#include <linux/compiler.h>
44
#include <linux/hash.h>
45
#include <linux/posix-clock.h>
L
Linus Torvalds 已提交
46 47 48 49
#include <linux/posix-timers.h>
#include <linux/syscalls.h>
#include <linux/wait.h>
#include <linux/workqueue.h>
50
#include <linux/export.h>
51
#include <linux/hashtable.h>
L
Linus Torvalds 已提交
52

53
#include "timekeeping.h"
54
#include "posix-timers.h"
55

L
Linus Torvalds 已提交
56
/*
57 58 59 60 61 62
 * Management arrays for POSIX timers. Timers are now kept in static hash table
 * with 512 entries.
 * Timer ids are allocated by local routine, which selects proper hash head by
 * key, constructed from current->signal address and per signal struct counter.
 * This keeps timer ids unique per process, but now they can intersect between
 * processes.
L
Linus Torvalds 已提交
63 64 65 66 67
 */

/*
 * Lets keep our timers in a slab cache :-)
 */
68
static struct kmem_cache *posix_timers_cache;
69 70 71

static DEFINE_HASHTABLE(posix_timers_hashtable, 9);
static DEFINE_SPINLOCK(hash_lock);
L
Linus Torvalds 已提交
72

73 74 75
static const struct k_clock * const posix_clocks[];
static const struct k_clock *clockid_to_kclock(const clockid_t id);

L
Linus Torvalds 已提交
76 77 78 79 80 81 82 83 84
/*
 * we assume that the new SIGEV_THREAD_ID shares no bits with the other
 * SIGEV values.  Here we put out an error if this assumption fails.
 */
#if SIGEV_THREAD_ID != (SIGEV_THREAD_ID & \
                       ~(SIGEV_SIGNAL | SIGEV_NONE | SIGEV_THREAD))
#error "SIGEV_THREAD_ID must not share bit with other SIGEV values!"
#endif

85 86 87 88 89 90 91 92
/*
 * parisc wants ENOTSUP instead of EOPNOTSUPP
 */
#ifndef ENOTSUP
# define ENANOSLEEP_NOTSUP EOPNOTSUPP
#else
# define ENANOSLEEP_NOTSUP ENOTSUP
#endif
L
Linus Torvalds 已提交
93 94 95 96 97 98 99 100 101 102 103 104 105

/*
 * The timer ID is turned into a timer address by idr_find().
 * Verifying a valid ID consists of:
 *
 * a) checking that idr_find() returns other than -1.
 * b) checking that the timer id matches the one in the timer itself.
 * c) that the timer owner is in the callers thread group.
 */

/*
 * CLOCKs: The POSIX standard calls for a couple of clocks and allows us
 *	    to implement others.  This structure defines the various
R
Richard Cochran 已提交
106
 *	    clocks.
L
Linus Torvalds 已提交
107 108 109 110 111 112 113 114 115
 *
 * RESOLUTION: Clock resolution is used to round up timer and interval
 *	    times, NOT to report clock times, which are reported with as
 *	    much resolution as the system can muster.  In some cases this
 *	    resolution may depend on the underlying clock hardware and
 *	    may not be quantifiable until run time, and only then is the
 *	    necessary code is written.	The standard says we should say
 *	    something about this issue in the documentation...
 *
R
Richard Cochran 已提交
116 117
 * FUNCTIONS: The CLOCKs structure defines possible functions to
 *	    handle various clock functions.
L
Linus Torvalds 已提交
118
 *
R
Richard Cochran 已提交
119 120 121 122
 *	    The standard POSIX timer management code assumes the
 *	    following: 1.) The k_itimer struct (sched.h) is used for
 *	    the timer.  2.) The list, it_lock, it_clock, it_id and
 *	    it_pid fields are not modified by timer code.
L
Linus Torvalds 已提交
123 124 125 126 127 128 129 130
 *
 * Permissions: It is assumed that the clock_settime() function defined
 *	    for each clock will take care of permission checks.	 Some
 *	    clocks may be set able by any user (i.e. local process
 *	    clocks) others not.	 Currently the only set able clock we
 *	    have is CLOCK_REALTIME and its high res counter part, both of
 *	    which we beg off on and pass to do_sys_settimeofday().
 */
N
Namhyung Kim 已提交
131 132 133 134 135 136 137
static struct k_itimer *__lock_timer(timer_t timer_id, unsigned long *flags);

#define lock_timer(tid, flags)						   \
({	struct k_itimer *__timr;					   \
	__cond_lock(&__timr->it_lock, __timr = __lock_timer(tid, flags));  \
	__timr;								   \
})
L
Linus Torvalds 已提交
138

139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188
static int hash(struct signal_struct *sig, unsigned int nr)
{
	return hash_32(hash32_ptr(sig) ^ nr, HASH_BITS(posix_timers_hashtable));
}

static struct k_itimer *__posix_timers_find(struct hlist_head *head,
					    struct signal_struct *sig,
					    timer_t id)
{
	struct k_itimer *timer;

	hlist_for_each_entry_rcu(timer, head, t_hash) {
		if ((timer->it_signal == sig) && (timer->it_id == id))
			return timer;
	}
	return NULL;
}

static struct k_itimer *posix_timer_by_id(timer_t id)
{
	struct signal_struct *sig = current->signal;
	struct hlist_head *head = &posix_timers_hashtable[hash(sig, id)];

	return __posix_timers_find(head, sig, id);
}

static int posix_timer_add(struct k_itimer *timer)
{
	struct signal_struct *sig = current->signal;
	int first_free_id = sig->posix_timer_id;
	struct hlist_head *head;
	int ret = -ENOENT;

	do {
		spin_lock(&hash_lock);
		head = &posix_timers_hashtable[hash(sig, sig->posix_timer_id)];
		if (!__posix_timers_find(head, sig, sig->posix_timer_id)) {
			hlist_add_head_rcu(&timer->t_hash, head);
			ret = sig->posix_timer_id;
		}
		if (++sig->posix_timer_id < 0)
			sig->posix_timer_id = 0;
		if ((sig->posix_timer_id == first_free_id) && (ret == -ENOENT))
			/* Loop over all possible ids completed */
			ret = -EAGAIN;
		spin_unlock(&hash_lock);
	} while (ret == -ENOENT);
	return ret;
}

L
Linus Torvalds 已提交
189 190 191 192 193
static inline void unlock_timer(struct k_itimer *timr, unsigned long flags)
{
	spin_unlock_irqrestore(&timr->it_lock, flags);
}

194
/* Get clock_realtime */
195
static int posix_clock_realtime_get(clockid_t which_clock, struct timespec64 *tp)
196
{
197
	ktime_get_real_ts64(tp);
198 199 200
	return 0;
}

201 202
/* Set clock_realtime */
static int posix_clock_realtime_set(const clockid_t which_clock,
203
				    const struct timespec64 *tp)
204
{
205
	return do_sys_settimeofday64(tp, NULL);
206 207
}

208 209 210 211 212 213
static int posix_clock_realtime_adj(const clockid_t which_clock,
				    struct timex *t)
{
	return do_adjtimex(t);
}

214 215 216
/*
 * Get monotonic time for posix timers
 */
217
static int posix_ktime_get_ts(clockid_t which_clock, struct timespec64 *tp)
218
{
219
	ktime_get_ts64(tp);
220 221
	return 0;
}
L
Linus Torvalds 已提交
222

223
/*
224
 * Get monotonic-raw time for posix timers
225
 */
226
static int posix_get_monotonic_raw(clockid_t which_clock, struct timespec64 *tp)
227
{
228
	getrawmonotonic64(tp);
229 230 231
	return 0;
}

232

233
static int posix_get_realtime_coarse(clockid_t which_clock, struct timespec64 *tp)
234
{
235
	*tp = current_kernel_time64();
236 237 238 239
	return 0;
}

static int posix_get_monotonic_coarse(clockid_t which_clock,
240
						struct timespec64 *tp)
241
{
242
	*tp = get_monotonic_coarse64();
243 244 245
	return 0;
}

246
static int posix_get_coarse_res(const clockid_t which_clock, struct timespec64 *tp)
247
{
248
	*tp = ktime_to_timespec64(KTIME_LOW_RES);
249 250
	return 0;
}
251

252
static int posix_get_boottime(const clockid_t which_clock, struct timespec64 *tp)
253
{
254
	get_monotonic_boottime64(tp);
255 256 257
	return 0;
}

258
static int posix_get_tai(clockid_t which_clock, struct timespec64 *tp)
J
John Stultz 已提交
259
{
260
	timekeeping_clocktai64(tp);
J
John Stultz 已提交
261 262
	return 0;
}
263

264
static int posix_get_hrtimer_res(clockid_t which_clock, struct timespec64 *tp)
265 266 267 268 269 270
{
	tp->tv_sec = 0;
	tp->tv_nsec = hrtimer_resolution;
	return 0;
}

L
Linus Torvalds 已提交
271 272 273 274 275 276
/*
 * Initialize everything, well, just everything in Posix clocks/timers ;)
 */
static __init int init_posix_timers(void)
{
	posix_timers_cache = kmem_cache_create("posix_timers_cache",
277 278
					sizeof (struct k_itimer), 0, SLAB_PANIC,
					NULL);
L
Linus Torvalds 已提交
279 280 281 282 283 284 285
	return 0;
}

__initcall(init_posix_timers);

static void schedule_next_timer(struct k_itimer *timr)
{
286 287
	struct hrtimer *timer = &timr->it.real.timer;

288
	if (!timr->it_interval)
L
Linus Torvalds 已提交
289 290
		return;

D
Davide Libenzi 已提交
291 292
	timr->it_overrun += (unsigned int) hrtimer_forward(timer,
						timer->base->get_time(),
293
						timr->it_interval);
294
	hrtimer_restart(timer);
L
Linus Torvalds 已提交
295 296 297 298 299 300 301 302 303 304
}

/*
 * This function is exported for use by the signal deliver code.  It is
 * called just prior to the info block being released and passes that
 * block to us.  It's function is to update the overrun entry AND to
 * restart the timer.  It should only be called if the timer is to be
 * restarted (i.e. we have flagged this in the sys_private entry of the
 * info block).
 *
L
Lucas De Marchi 已提交
305
 * To protect against the timer going away while the interrupt is queued,
L
Linus Torvalds 已提交
306 307 308 309 310 311 312 313
 * we require that the it_requeue_pending flag be set.
 */
void do_schedule_next_timer(struct siginfo *info)
{
	struct k_itimer *timr;
	unsigned long flags;

	timr = lock_timer(info->si_tid, &flags);
314 315
	if (!timr)
		return;
L
Linus Torvalds 已提交
316

317
	if (timr->it_requeue_pending == info->si_sys_private) {
318 319 320 321
		if (timr->it_clock < 0)
			posix_cpu_timer_schedule(timr);
		else
			schedule_next_timer(timr);
L
Linus Torvalds 已提交
322

323 324 325 326
		timr->it_overrun_last = timr->it_overrun;
		timr->it_overrun = -1;
		++timr->it_requeue_pending;

327
		info->si_overrun += timr->it_overrun_last;
328 329
	}

330
	unlock_timer(timr, flags);
L
Linus Torvalds 已提交
331 332
}

333
int posix_timer_event(struct k_itimer *timr, int si_private)
L
Linus Torvalds 已提交
334
{
335 336
	struct task_struct *task;
	int shared, ret = -1;
337 338 339 340 341 342 343 344 345 346 347
	/*
	 * FIXME: if ->sigq is queued we can race with
	 * dequeue_signal()->do_schedule_next_timer().
	 *
	 * If dequeue_signal() sees the "right" value of
	 * si_sys_private it calls do_schedule_next_timer().
	 * We re-queue ->sigq and drop ->it_lock().
	 * do_schedule_next_timer() locks the timer
	 * and re-schedules it while ->sigq is pending.
	 * Not really bad, but not that we want.
	 */
L
Linus Torvalds 已提交
348 349
	timr->sigq->info.si_sys_private = si_private;

350 351 352 353 354 355 356
	rcu_read_lock();
	task = pid_task(timr->it_pid, PIDTYPE_PID);
	if (task) {
		shared = !(timr->it_sigev_notify & SIGEV_THREAD_ID);
		ret = send_sigqueue(timr->sigq, task, shared);
	}
	rcu_read_unlock();
357 358
	/* If we failed to send the signal the timer stops. */
	return ret > 0;
L
Linus Torvalds 已提交
359 360 361 362 363 364 365 366 367
}

/*
 * This function gets called when a POSIX.1b interval timer expires.  It
 * is used as a callback from the kernel internal timer.  The
 * run_timer_list code ALWAYS calls with interrupts on.

 * This code is for CLOCK_REALTIME* and CLOCK_MONOTONIC* timers.
 */
368
static enum hrtimer_restart posix_timer_fn(struct hrtimer *timer)
L
Linus Torvalds 已提交
369
{
370
	struct k_itimer *timr;
L
Linus Torvalds 已提交
371
	unsigned long flags;
372
	int si_private = 0;
373
	enum hrtimer_restart ret = HRTIMER_NORESTART;
L
Linus Torvalds 已提交
374

375
	timr = container_of(timer, struct k_itimer, it.real.timer);
L
Linus Torvalds 已提交
376 377
	spin_lock_irqsave(&timr->it_lock, flags);

378
	if (timr->it_interval != 0)
379
		si_private = ++timr->it_requeue_pending;
L
Linus Torvalds 已提交
380

381 382 383 384 385 386
	if (posix_timer_event(timr, si_private)) {
		/*
		 * signal was not sent because of sig_ignor
		 * we will not get a call back to restart it AND
		 * it should be restarted.
		 */
387
		if (timr->it_interval != 0) {
388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413
			ktime_t now = hrtimer_cb_get_time(timer);

			/*
			 * FIXME: What we really want, is to stop this
			 * timer completely and restart it in case the
			 * SIG_IGN is removed. This is a non trivial
			 * change which involves sighand locking
			 * (sigh !), which we don't want to do late in
			 * the release cycle.
			 *
			 * For now we just let timers with an interval
			 * less than a jiffie expire every jiffie to
			 * avoid softirq starvation in case of SIG_IGN
			 * and a very small interval, which would put
			 * the timer right back on the softirq pending
			 * list. By moving now ahead of time we trick
			 * hrtimer_forward() to expire the timer
			 * later, while we still maintain the overrun
			 * accuracy, but have some inconsistency in
			 * the timer_gettime() case. This is at least
			 * better than a starved softirq. A more
			 * complex fix which solves also another related
			 * inconsistency is already in the pipeline.
			 */
#ifdef CONFIG_HIGH_RES_TIMERS
			{
T
Thomas Gleixner 已提交
414
				ktime_t kj = NSEC_PER_SEC / HZ;
415

416
				if (timr->it_interval < kj)
417 418 419
					now = ktime_add(now, kj);
			}
#endif
D
Davide Libenzi 已提交
420
			timr->it_overrun += (unsigned int)
421
				hrtimer_forward(timer, now,
422
						timr->it_interval);
423
			ret = HRTIMER_RESTART;
424
			++timr->it_requeue_pending;
L
Linus Torvalds 已提交
425 426 427
		}
	}

428 429 430
	unlock_timer(timr, flags);
	return ret;
}
L
Linus Torvalds 已提交
431

432
static struct pid *good_sigevent(sigevent_t * event)
L
Linus Torvalds 已提交
433 434 435 436
{
	struct task_struct *rtn = current->group_leader;

	if ((event->sigev_notify & SIGEV_THREAD_ID ) &&
437
		(!(rtn = find_task_by_vpid(event->sigev_notify_thread_id)) ||
438
		 !same_thread_group(rtn, current) ||
L
Linus Torvalds 已提交
439 440 441 442 443 444 445
		 (event->sigev_notify & ~SIGEV_THREAD_ID) != SIGEV_SIGNAL))
		return NULL;

	if (((event->sigev_notify & ~SIGEV_THREAD_ID) != SIGEV_NONE) &&
	    ((event->sigev_signo <= 0) || (event->sigev_signo > SIGRTMAX)))
		return NULL;

446
	return task_pid(rtn);
L
Linus Torvalds 已提交
447 448 449 450 451
}

static struct k_itimer * alloc_posix_timer(void)
{
	struct k_itimer *tmr;
452
	tmr = kmem_cache_zalloc(posix_timers_cache, GFP_KERNEL);
L
Linus Torvalds 已提交
453 454 455 456
	if (!tmr)
		return tmr;
	if (unlikely(!(tmr->sigq = sigqueue_alloc()))) {
		kmem_cache_free(posix_timers_cache, tmr);
457
		return NULL;
L
Linus Torvalds 已提交
458
	}
459
	memset(&tmr->sigq->info, 0, sizeof(siginfo_t));
L
Linus Torvalds 已提交
460 461 462
	return tmr;
}

E
Eric Dumazet 已提交
463 464 465 466 467 468 469
static void k_itimer_rcu_free(struct rcu_head *head)
{
	struct k_itimer *tmr = container_of(head, struct k_itimer, it.rcu);

	kmem_cache_free(posix_timers_cache, tmr);
}

L
Linus Torvalds 已提交
470 471 472 473 474 475
#define IT_ID_SET	1
#define IT_ID_NOT_SET	0
static void release_posix_timer(struct k_itimer *tmr, int it_id_set)
{
	if (it_id_set) {
		unsigned long flags;
476 477 478
		spin_lock_irqsave(&hash_lock, flags);
		hlist_del_rcu(&tmr->t_hash);
		spin_unlock_irqrestore(&hash_lock, flags);
L
Linus Torvalds 已提交
479
	}
480
	put_pid(tmr->it_pid);
L
Linus Torvalds 已提交
481
	sigqueue_free(tmr->sigq);
E
Eric Dumazet 已提交
482
	call_rcu(&tmr->it.rcu, k_itimer_rcu_free);
L
Linus Torvalds 已提交
483 484
}

485 486 487 488 489 490
static int common_timer_create(struct k_itimer *new_timer)
{
	hrtimer_init(&new_timer->it.real.timer, new_timer->it_clock, 0);
	return 0;
}

L
Linus Torvalds 已提交
491 492
/* Create a POSIX.1b interval timer. */

493 494 495
SYSCALL_DEFINE3(timer_create, const clockid_t, which_clock,
		struct sigevent __user *, timer_event_spec,
		timer_t __user *, created_timer_id)
L
Linus Torvalds 已提交
496
{
497
	const struct k_clock *kc = clockid_to_kclock(which_clock);
498
	struct k_itimer *new_timer;
499
	int error, new_timer_id;
L
Linus Torvalds 已提交
500 501 502
	sigevent_t event;
	int it_id_set = IT_ID_NOT_SET;

503
	if (!kc)
L
Linus Torvalds 已提交
504
		return -EINVAL;
505 506
	if (!kc->timer_create)
		return -EOPNOTSUPP;
L
Linus Torvalds 已提交
507 508 509 510 511 512

	new_timer = alloc_posix_timer();
	if (unlikely(!new_timer))
		return -EAGAIN;

	spin_lock_init(&new_timer->it_lock);
513 514 515
	new_timer_id = posix_timer_add(new_timer);
	if (new_timer_id < 0) {
		error = new_timer_id;
L
Linus Torvalds 已提交
516 517 518 519 520 521
		goto out;
	}

	it_id_set = IT_ID_SET;
	new_timer->it_id = (timer_t) new_timer_id;
	new_timer->it_clock = which_clock;
522
	new_timer->kclock = kc;
L
Linus Torvalds 已提交
523 524 525 526 527 528 529
	new_timer->it_overrun = -1;

	if (timer_event_spec) {
		if (copy_from_user(&event, timer_event_spec, sizeof (event))) {
			error = -EFAULT;
			goto out;
		}
530
		rcu_read_lock();
531
		new_timer->it_pid = get_pid(good_sigevent(&event));
532
		rcu_read_unlock();
533
		if (!new_timer->it_pid) {
L
Linus Torvalds 已提交
534 535 536 537
			error = -EINVAL;
			goto out;
		}
	} else {
538
		memset(&event.sigev_value, 0, sizeof(event.sigev_value));
539 540 541
		event.sigev_notify = SIGEV_SIGNAL;
		event.sigev_signo = SIGALRM;
		event.sigev_value.sival_int = new_timer->it_id;
542
		new_timer->it_pid = get_pid(task_tgid(current));
L
Linus Torvalds 已提交
543 544
	}

545 546 547
	new_timer->it_sigev_notify     = event.sigev_notify;
	new_timer->sigq->info.si_signo = event.sigev_signo;
	new_timer->sigq->info.si_value = event.sigev_value;
548
	new_timer->sigq->info.si_tid   = new_timer->it_id;
549
	new_timer->sigq->info.si_code  = SI_TIMER;
550

551 552 553 554 555 556
	if (copy_to_user(created_timer_id,
			 &new_timer_id, sizeof (new_timer_id))) {
		error = -EFAULT;
		goto out;
	}

557
	error = kc->timer_create(new_timer);
558 559 560
	if (error)
		goto out;

561
	spin_lock_irq(&current->sighand->siglock);
562
	new_timer->it_signal = current->signal;
563 564
	list_add(&new_timer->list, &current->signal->posix_timers);
	spin_unlock_irq(&current->sighand->siglock);
565 566

	return 0;
567
	/*
L
Linus Torvalds 已提交
568 569 570 571 572 573
	 * In the case of the timer belonging to another task, after
	 * the task is unlocked, the timer is owned by the other task
	 * and may cease to exist at any time.  Don't use or modify
	 * new_timer after the unlock call.
	 */
out:
574
	release_posix_timer(new_timer, it_id_set);
L
Linus Torvalds 已提交
575 576 577 578 579 580 581 582 583 584
	return error;
}

/*
 * Locking issues: We need to protect the result of the id look up until
 * we get the timer locked down so it is not deleted under us.  The
 * removal is done under the idr spinlock so we use that here to bridge
 * the find to the timer lock.  To avoid a dead lock, the timer id MUST
 * be release with out holding the timer lock.
 */
N
Namhyung Kim 已提交
585
static struct k_itimer *__lock_timer(timer_t timer_id, unsigned long *flags)
L
Linus Torvalds 已提交
586 587
{
	struct k_itimer *timr;
E
Eric Dumazet 已提交
588

589 590 591 592 593 594 595
	/*
	 * timer_t could be any type >= int and we want to make sure any
	 * @timer_id outside positive int range fails lookup.
	 */
	if ((unsigned long long)timer_id > INT_MAX)
		return NULL;

E
Eric Dumazet 已提交
596
	rcu_read_lock();
597
	timr = posix_timer_by_id(timer_id);
L
Linus Torvalds 已提交
598
	if (timr) {
E
Eric Dumazet 已提交
599
		spin_lock_irqsave(&timr->it_lock, *flags);
600
		if (timr->it_signal == current->signal) {
E
Eric Dumazet 已提交
601
			rcu_read_unlock();
602 603
			return timr;
		}
E
Eric Dumazet 已提交
604
		spin_unlock_irqrestore(&timr->it_lock, *flags);
605
	}
E
Eric Dumazet 已提交
606
	rcu_read_unlock();
L
Linus Torvalds 已提交
607

608
	return NULL;
L
Linus Torvalds 已提交
609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627
}

/*
 * Get the time remaining on a POSIX.1b interval timer.  This function
 * is ALWAYS called with spin_lock_irq on the timer, thus it must not
 * mess with irq.
 *
 * We have a couple of messes to clean up here.  First there is the case
 * of a timer that has a requeue pending.  These timers should appear to
 * be in the timer list with an expiry as if we were to requeue them
 * now.
 *
 * The second issue is the SIGEV_NONE timer which may be active but is
 * not really ever put in the timer list (to save system resources).
 * This timer may be expired, and if so, we will do it here.  Otherwise
 * it is the same as a requeue pending timer WRT to what we should
 * report.
 */
static void
628
common_timer_get(struct k_itimer *timr, struct itimerspec64 *cur_setting)
L
Linus Torvalds 已提交
629
{
630
	ktime_t now, remaining, iv;
631
	struct hrtimer *timer = &timr->it.real.timer;
L
Linus Torvalds 已提交
632

633
	memset(cur_setting, 0, sizeof(*cur_setting));
634

635
	iv = timr->it_interval;
636

637
	/* interval timer ? */
T
Thomas Gleixner 已提交
638
	if (iv)
639
		cur_setting->it_interval = ktime_to_timespec64(iv);
640 641
	else if (!hrtimer_active(timer) &&
		 (timr->it_sigev_notify & ~SIGEV_THREAD_ID) != SIGEV_NONE)
642
		return;
643 644 645

	now = timer->base->get_time();

646
	/*
647 648 649
	 * When a requeue is pending or this is a SIGEV_NONE
	 * timer move the expiry time forward by intervals, so
	 * expiry is > now.
650
	 */
T
Thomas Gleixner 已提交
651 652
	if (iv && (timr->it_requeue_pending & REQUEUE_PENDING ||
		   (timr->it_sigev_notify & ~SIGEV_THREAD_ID) == SIGEV_NONE))
D
Davide Libenzi 已提交
653
		timr->it_overrun += (unsigned int) hrtimer_forward(timer, now, iv);
654

655
	remaining = __hrtimer_expires_remaining_adjusted(timer, now);
656
	/* Return 0 only, when the timer is expired and not pending */
T
Thomas Gleixner 已提交
657
	if (remaining <= 0) {
658 659 660 661 662 663 664
		/*
		 * A single shot SIGEV_NONE timer must return 0, when
		 * it is expired !
		 */
		if ((timr->it_sigev_notify & ~SIGEV_THREAD_ID) != SIGEV_NONE)
			cur_setting->it_value.tv_nsec = 1;
	} else
665
		cur_setting->it_value = ktime_to_timespec64(remaining);
L
Linus Torvalds 已提交
666 667 668
}

/* Get the time remaining on a POSIX.1b interval timer. */
669 670
SYSCALL_DEFINE2(timer_gettime, timer_t, timer_id,
		struct itimerspec __user *, setting)
L
Linus Torvalds 已提交
671
{
672
	struct itimerspec64 cur_setting64;
L
Linus Torvalds 已提交
673
	struct itimerspec cur_setting;
674
	struct k_itimer *timr;
675
	const struct k_clock *kc;
L
Linus Torvalds 已提交
676
	unsigned long flags;
677
	int ret = 0;
L
Linus Torvalds 已提交
678 679 680 681 682

	timr = lock_timer(timer_id, &flags);
	if (!timr)
		return -EINVAL;

683
	kc = timr->kclock;
684 685 686
	if (WARN_ON_ONCE(!kc || !kc->timer_get))
		ret = -EINVAL;
	else
687
		kc->timer_get(timr, &cur_setting64);
L
Linus Torvalds 已提交
688 689 690

	unlock_timer(timr, flags);

691
	cur_setting = itimerspec64_to_itimerspec(&cur_setting64);
692
	if (!ret && copy_to_user(setting, &cur_setting, sizeof (cur_setting)))
L
Linus Torvalds 已提交
693 694
		return -EFAULT;

695
	return ret;
L
Linus Torvalds 已提交
696
}
697

L
Linus Torvalds 已提交
698 699 700 701 702 703 704 705 706
/*
 * Get the number of overruns of a POSIX.1b interval timer.  This is to
 * be the overrun of the timer last delivered.  At the same time we are
 * accumulating overruns on the next timer.  The overrun is frozen when
 * the signal is delivered, either at the notify time (if the info block
 * is not queued) or at the actual delivery time (as we are informed by
 * the call back to do_schedule_next_timer().  So all we need to do is
 * to pick up the frozen overrun.
 */
707
SYSCALL_DEFINE1(timer_getoverrun, timer_t, timer_id)
L
Linus Torvalds 已提交
708 709 710
{
	struct k_itimer *timr;
	int overrun;
711
	unsigned long flags;
L
Linus Torvalds 已提交
712 713 714 715 716 717 718 719 720 721 722 723 724

	timr = lock_timer(timer_id, &flags);
	if (!timr)
		return -EINVAL;

	overrun = timr->it_overrun_last;
	unlock_timer(timr, flags);

	return overrun;
}

/* Set a POSIX.1b interval timer. */
/* timr->it_lock is taken. */
725
static int
L
Linus Torvalds 已提交
726
common_timer_set(struct k_itimer *timr, int flags,
727
		 struct itimerspec64 *new_setting, struct itimerspec64 *old_setting)
L
Linus Torvalds 已提交
728
{
729
	struct hrtimer *timer = &timr->it.real.timer;
730
	enum hrtimer_mode mode;
L
Linus Torvalds 已提交
731 732 733 734 735

	if (old_setting)
		common_timer_get(timr, old_setting);

	/* disable the timer */
736
	timr->it_interval = 0;
L
Linus Torvalds 已提交
737 738 739 740
	/*
	 * careful here.  If smp we could be in the "fire" routine which will
	 * be spinning as we hold the lock.  But this is ONLY an SMP issue.
	 */
741
	if (hrtimer_try_to_cancel(timer) < 0)
L
Linus Torvalds 已提交
742 743 744 745 746 747
		return TIMER_RETRY;

	timr->it_requeue_pending = (timr->it_requeue_pending + 2) & 
		~REQUEUE_PENDING;
	timr->it_overrun_last = 0;

748 749 750
	/* switch off the timer when it_value is zero */
	if (!new_setting->it_value.tv_sec && !new_setting->it_value.tv_nsec)
		return 0;
L
Linus Torvalds 已提交
751

752
	mode = flags & TIMER_ABSTIME ? HRTIMER_MODE_ABS : HRTIMER_MODE_REL;
753 754
	hrtimer_init(&timr->it.real.timer, timr->it_clock, mode);
	timr->it.real.timer.function = posix_timer_fn;
755

756
	hrtimer_set_expires(timer, timespec64_to_ktime(new_setting->it_value));
757 758

	/* Convert interval */
759
	timr->it_interval = timespec64_to_ktime(new_setting->it_interval);
760 761

	/* SIGEV_NONE timers are not queued ! See common_timer_get */
762 763
	if (((timr->it_sigev_notify & ~SIGEV_THREAD_ID) == SIGEV_NONE)) {
		/* Setup correct expiry time for relative timers */
764
		if (mode == HRTIMER_MODE_REL) {
765
			hrtimer_add_expires(timer, timer->base->get_time());
766
		}
767
		return 0;
768
	}
769

770
	hrtimer_start_expires(timer, mode);
L
Linus Torvalds 已提交
771 772 773 774
	return 0;
}

/* Set a POSIX.1b interval timer */
775 776 777
SYSCALL_DEFINE4(timer_settime, timer_t, timer_id, int, flags,
		const struct itimerspec __user *, new_setting,
		struct itimerspec __user *, old_setting)
L
Linus Torvalds 已提交
778
{
779 780
	struct itimerspec64 new_spec64, old_spec64;
	struct itimerspec64 *rtn = old_setting ? &old_spec64 : NULL;
L
Linus Torvalds 已提交
781
	struct itimerspec new_spec, old_spec;
782
	struct k_itimer *timr;
783
	unsigned long flag;
784
	const struct k_clock *kc;
785
	int error = 0;
L
Linus Torvalds 已提交
786 787 788 789 790 791

	if (!new_setting)
		return -EINVAL;

	if (copy_from_user(&new_spec, new_setting, sizeof (new_spec)))
		return -EFAULT;
792
	new_spec64 = itimerspec_to_itimerspec64(&new_spec);
L
Linus Torvalds 已提交
793

794 795
	if (!timespec64_valid(&new_spec64.it_interval) ||
	    !timespec64_valid(&new_spec64.it_value))
L
Linus Torvalds 已提交
796 797 798 799 800 801
		return -EINVAL;
retry:
	timr = lock_timer(timer_id, &flag);
	if (!timr)
		return -EINVAL;

802
	kc = timr->kclock;
803 804 805
	if (WARN_ON_ONCE(!kc || !kc->timer_set))
		error = -EINVAL;
	else
806
		error = kc->timer_set(timr, flags, &new_spec64, rtn);
L
Linus Torvalds 已提交
807 808 809 810 811 812 813

	unlock_timer(timr, flag);
	if (error == TIMER_RETRY) {
		rtn = NULL;	// We already got the old time...
		goto retry;
	}

814
	old_spec = itimerspec64_to_itimerspec(&old_spec64);
815 816
	if (old_setting && !error &&
	    copy_to_user(old_setting, &old_spec, sizeof (old_spec)))
L
Linus Torvalds 已提交
817 818 819 820 821
		error = -EFAULT;

	return error;
}

822
static int common_timer_del(struct k_itimer *timer)
L
Linus Torvalds 已提交
823
{
824
	timer->it_interval = 0;
825

826
	if (hrtimer_try_to_cancel(&timer->it.real.timer) < 0)
L
Linus Torvalds 已提交
827 828 829 830 831 832
		return TIMER_RETRY;
	return 0;
}

static inline int timer_delete_hook(struct k_itimer *timer)
{
833
	const struct k_clock *kc = timer->kclock;
834 835 836 837

	if (WARN_ON_ONCE(!kc || !kc->timer_del))
		return -EINVAL;
	return kc->timer_del(timer);
L
Linus Torvalds 已提交
838 839 840
}

/* Delete a POSIX.1b interval timer. */
841
SYSCALL_DEFINE1(timer_delete, timer_t, timer_id)
L
Linus Torvalds 已提交
842 843
{
	struct k_itimer *timer;
844
	unsigned long flags;
L
Linus Torvalds 已提交
845 846 847 848 849 850

retry_delete:
	timer = lock_timer(timer_id, &flags);
	if (!timer)
		return -EINVAL;

851
	if (timer_delete_hook(timer) == TIMER_RETRY) {
L
Linus Torvalds 已提交
852 853 854
		unlock_timer(timer, flags);
		goto retry_delete;
	}
855

L
Linus Torvalds 已提交
856 857 858 859 860 861 862
	spin_lock(&current->sighand->siglock);
	list_del(&timer->list);
	spin_unlock(&current->sighand->siglock);
	/*
	 * This keeps any tasks waiting on the spin lock from thinking
	 * they got something (see the lock code above).
	 */
863
	timer->it_signal = NULL;
864

L
Linus Torvalds 已提交
865 866 867 868
	unlock_timer(timer, flags);
	release_posix_timer(timer, IT_ID_SET);
	return 0;
}
869

L
Linus Torvalds 已提交
870 871 872
/*
 * return timer owned by the process, used by exit_itimers
 */
873
static void itimer_delete(struct k_itimer *timer)
L
Linus Torvalds 已提交
874 875 876 877 878 879
{
	unsigned long flags;

retry_delete:
	spin_lock_irqsave(&timer->it_lock, flags);

880
	if (timer_delete_hook(timer) == TIMER_RETRY) {
L
Linus Torvalds 已提交
881 882 883 884 885 886 887 888
		unlock_timer(timer, flags);
		goto retry_delete;
	}
	list_del(&timer->list);
	/*
	 * This keeps any tasks waiting on the spin lock from thinking
	 * they got something (see the lock code above).
	 */
889
	timer->it_signal = NULL;
890

L
Linus Torvalds 已提交
891 892 893 894 895
	unlock_timer(timer, flags);
	release_posix_timer(timer, IT_ID_SET);
}

/*
896
 * This is called by do_exit or de_thread, only when there are no more
L
Linus Torvalds 已提交
897 898 899 900 901 902 903 904 905 906 907 908
 * references to the shared signal_struct.
 */
void exit_itimers(struct signal_struct *sig)
{
	struct k_itimer *tmr;

	while (!list_empty(&sig->posix_timers)) {
		tmr = list_entry(sig->posix_timers.next, struct k_itimer, list);
		itimer_delete(tmr);
	}
}

909 910
SYSCALL_DEFINE2(clock_settime, const clockid_t, which_clock,
		const struct timespec __user *, tp)
L
Linus Torvalds 已提交
911
{
912
	const struct k_clock *kc = clockid_to_kclock(which_clock);
913
	struct timespec64 new_tp64;
L
Linus Torvalds 已提交
914 915
	struct timespec new_tp;

916
	if (!kc || !kc->clock_set)
L
Linus Torvalds 已提交
917
		return -EINVAL;
918

L
Linus Torvalds 已提交
919 920
	if (copy_from_user(&new_tp, tp, sizeof (*tp)))
		return -EFAULT;
921
	new_tp64 = timespec_to_timespec64(new_tp);
L
Linus Torvalds 已提交
922

923
	return kc->clock_set(which_clock, &new_tp64);
L
Linus Torvalds 已提交
924 925
}

926 927
SYSCALL_DEFINE2(clock_gettime, const clockid_t, which_clock,
		struct timespec __user *,tp)
L
Linus Torvalds 已提交
928
{
929
	const struct k_clock *kc = clockid_to_kclock(which_clock);
930
	struct timespec64 kernel_tp64;
L
Linus Torvalds 已提交
931 932 933
	struct timespec kernel_tp;
	int error;

934
	if (!kc)
L
Linus Torvalds 已提交
935
		return -EINVAL;
936

937 938
	error = kc->clock_get(which_clock, &kernel_tp64);
	kernel_tp = timespec64_to_timespec(kernel_tp64);
939

L
Linus Torvalds 已提交
940 941 942 943 944 945
	if (!error && copy_to_user(tp, &kernel_tp, sizeof (kernel_tp)))
		error = -EFAULT;

	return error;
}

946 947 948
SYSCALL_DEFINE2(clock_adjtime, const clockid_t, which_clock,
		struct timex __user *, utx)
{
949
	const struct k_clock *kc = clockid_to_kclock(which_clock);
950 951 952 953 954 955 956 957 958 959 960 961 962
	struct timex ktx;
	int err;

	if (!kc)
		return -EINVAL;
	if (!kc->clock_adj)
		return -EOPNOTSUPP;

	if (copy_from_user(&ktx, utx, sizeof(ktx)))
		return -EFAULT;

	err = kc->clock_adj(which_clock, &ktx);

963
	if (err >= 0 && copy_to_user(utx, &ktx, sizeof(ktx)))
964 965 966 967 968
		return -EFAULT;

	return err;
}

969 970
SYSCALL_DEFINE2(clock_getres, const clockid_t, which_clock,
		struct timespec __user *, tp)
L
Linus Torvalds 已提交
971
{
972
	const struct k_clock *kc = clockid_to_kclock(which_clock);
973
	struct timespec64 rtn_tp64;
L
Linus Torvalds 已提交
974 975 976
	struct timespec rtn_tp;
	int error;

977
	if (!kc)
L
Linus Torvalds 已提交
978 979
		return -EINVAL;

980 981
	error = kc->clock_getres(which_clock, &rtn_tp64);
	rtn_tp = timespec64_to_timespec(rtn_tp64);
L
Linus Torvalds 已提交
982

983
	if (!error && tp && copy_to_user(tp, &rtn_tp, sizeof (rtn_tp)))
L
Linus Torvalds 已提交
984 985 986 987 988
		error = -EFAULT;

	return error;
}

989 990 991 992
/*
 * nanosleep for monotonic and realtime clocks
 */
static int common_nsleep(const clockid_t which_clock, int flags,
993
			 struct timespec64 *tsave, struct timespec __user *rmtp)
994
{
995 996 997
	return hrtimer_nanosleep(tsave, rmtp, flags & TIMER_ABSTIME ?
				 HRTIMER_MODE_ABS : HRTIMER_MODE_REL,
				 which_clock);
998
}
L
Linus Torvalds 已提交
999

1000 1001 1002
SYSCALL_DEFINE4(clock_nanosleep, const clockid_t, which_clock, int, flags,
		const struct timespec __user *, rqtp,
		struct timespec __user *, rmtp)
L
Linus Torvalds 已提交
1003
{
1004
	const struct k_clock *kc = clockid_to_kclock(which_clock);
1005
	struct timespec64 t64;
L
Linus Torvalds 已提交
1006 1007
	struct timespec t;

1008
	if (!kc)
L
Linus Torvalds 已提交
1009
		return -EINVAL;
1010 1011
	if (!kc->nsleep)
		return -ENANOSLEEP_NOTSUP;
L
Linus Torvalds 已提交
1012 1013 1014 1015

	if (copy_from_user(&t, rqtp, sizeof (struct timespec)))
		return -EFAULT;

1016 1017
	t64 = timespec_to_timespec64(t);
	if (!timespec64_valid(&t64))
L
Linus Torvalds 已提交
1018 1019
		return -EINVAL;

1020
	return kc->nsleep(which_clock, flags, &t64, rmtp);
L
Linus Torvalds 已提交
1021
}
1022 1023 1024 1025 1026

/*
 * This will restart clock_nanosleep. This is required only by
 * compat_clock_nanosleep_restart for now.
 */
1027
long clock_nanosleep_restart(struct restart_block *restart_block)
1028
{
1029
	clockid_t which_clock = restart_block->nanosleep.clockid;
1030
	const struct k_clock *kc = clockid_to_kclock(which_clock);
1031 1032 1033

	if (WARN_ON_ONCE(!kc || !kc->nsleep_restart))
		return -EINVAL;
1034

1035
	return kc->nsleep_restart(restart_block);
1036
}
1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122

static const struct k_clock clock_realtime = {
	.clock_getres	= posix_get_hrtimer_res,
	.clock_get	= posix_clock_realtime_get,
	.clock_set	= posix_clock_realtime_set,
	.clock_adj	= posix_clock_realtime_adj,
	.nsleep		= common_nsleep,
	.nsleep_restart	= hrtimer_nanosleep_restart,
	.timer_create	= common_timer_create,
	.timer_set	= common_timer_set,
	.timer_get	= common_timer_get,
	.timer_del	= common_timer_del,
};

static const struct k_clock clock_monotonic = {
	.clock_getres	= posix_get_hrtimer_res,
	.clock_get	= posix_ktime_get_ts,
	.nsleep		= common_nsleep,
	.nsleep_restart	= hrtimer_nanosleep_restart,
	.timer_create	= common_timer_create,
	.timer_set	= common_timer_set,
	.timer_get	= common_timer_get,
	.timer_del	= common_timer_del,
};

static const struct k_clock clock_monotonic_raw = {
	.clock_getres	= posix_get_hrtimer_res,
	.clock_get	= posix_get_monotonic_raw,
};

static const struct k_clock clock_realtime_coarse = {
	.clock_getres	= posix_get_coarse_res,
	.clock_get	= posix_get_realtime_coarse,
};

static const struct k_clock clock_monotonic_coarse = {
	.clock_getres	= posix_get_coarse_res,
	.clock_get	= posix_get_monotonic_coarse,
};

static const struct k_clock clock_tai = {
	.clock_getres	= posix_get_hrtimer_res,
	.clock_get	= posix_get_tai,
	.nsleep		= common_nsleep,
	.nsleep_restart	= hrtimer_nanosleep_restart,
	.timer_create	= common_timer_create,
	.timer_set	= common_timer_set,
	.timer_get	= common_timer_get,
	.timer_del	= common_timer_del,
};

static const struct k_clock clock_boottime = {
	.clock_getres	= posix_get_hrtimer_res,
	.clock_get	= posix_get_boottime,
	.nsleep		= common_nsleep,
	.nsleep_restart	= hrtimer_nanosleep_restart,
	.timer_create	= common_timer_create,
	.timer_set	= common_timer_set,
	.timer_get	= common_timer_get,
	.timer_del	= common_timer_del,
};

static const struct k_clock * const posix_clocks[] = {
	[CLOCK_REALTIME]		= &clock_realtime,
	[CLOCK_MONOTONIC]		= &clock_monotonic,
	[CLOCK_PROCESS_CPUTIME_ID]	= &clock_process,
	[CLOCK_THREAD_CPUTIME_ID]	= &clock_thread,
	[CLOCK_MONOTONIC_RAW]		= &clock_monotonic_raw,
	[CLOCK_REALTIME_COARSE]		= &clock_realtime_coarse,
	[CLOCK_MONOTONIC_COARSE]	= &clock_monotonic_coarse,
	[CLOCK_BOOTTIME]		= &clock_boottime,
	[CLOCK_REALTIME_ALARM]		= &alarm_clock,
	[CLOCK_BOOTTIME_ALARM]		= &alarm_clock,
	[CLOCK_TAI]			= &clock_tai,
};

static const struct k_clock *clockid_to_kclock(const clockid_t id)
{
	if (id < 0)
		return (id & CLOCKFD_MASK) == CLOCKFD ?
			&clock_posix_dynamic : &clock_posix_cpu;

	if (id >= ARRAY_SIZE(posix_clocks) || !posix_clocks[id])
		return NULL;
	return posix_clocks[id];
}