posix-timers.c 30.7 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 54
#include "timekeeping.h"

L
Linus Torvalds 已提交
55
/*
56 57 58 59 60 61
 * 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
62 63 64 65 66
 */

/*
 * Lets keep our timers in a slab cache :-)
 */
67
static struct kmem_cache *posix_timers_cache;
68 69 70

static DEFINE_HASHTABLE(posix_timers_hashtable, 9);
static DEFINE_SPINLOCK(hash_lock);
L
Linus Torvalds 已提交
71 72 73 74 75 76 77 78 79 80

/*
 * 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

81 82 83 84 85 86 87 88
/*
 * parisc wants ENOTSUP instead of EOPNOTSUPP
 */
#ifndef ENOTSUP
# define ENANOSLEEP_NOTSUP EOPNOTSUPP
#else
# define ENANOSLEEP_NOTSUP ENOTSUP
#endif
L
Linus Torvalds 已提交
89 90 91 92 93 94 95 96 97 98 99 100 101

/*
 * 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
102
 *	    clocks.
L
Linus Torvalds 已提交
103 104 105 106 107 108 109 110 111
 *
 * 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
112 113
 * FUNCTIONS: The CLOCKs structure defines possible functions to
 *	    handle various clock functions.
L
Linus Torvalds 已提交
114
 *
R
Richard Cochran 已提交
115 116 117 118
 *	    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
119 120 121 122 123 124 125 126 127 128
 *
 * 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().
 */

static struct k_clock posix_clocks[MAX_CLOCKS];
129

L
Linus Torvalds 已提交
130
/*
131
 * These ones are defined below.
L
Linus Torvalds 已提交
132
 */
133 134
static int common_nsleep(const clockid_t, int flags, struct timespec *t,
			 struct timespec __user *rmtp);
135
static int common_timer_create(struct k_itimer *new_timer);
136 137 138 139
static void common_timer_get(struct k_itimer *, struct itimerspec *);
static int common_timer_set(struct k_itimer *, int,
			    struct itimerspec *, struct itimerspec *);
static int common_timer_del(struct k_itimer *timer);
L
Linus Torvalds 已提交
140

141
static enum hrtimer_restart posix_timer_fn(struct hrtimer *data);
L
Linus Torvalds 已提交
142

N
Namhyung Kim 已提交
143 144 145 146 147 148 149
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
150

151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200
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
201 202 203 204 205
static inline void unlock_timer(struct k_itimer *timr, unsigned long flags)
{
	spin_unlock_irqrestore(&timr->it_lock, flags);
}

206
/* Get clock_realtime */
207
static int posix_clock_realtime_get(clockid_t which_clock, struct timespec64 *tp)
208
{
209
	ktime_get_real_ts64(tp);
210 211 212
	return 0;
}

213 214
/* Set clock_realtime */
static int posix_clock_realtime_set(const clockid_t which_clock,
215
				    const struct timespec64 *tp)
216
{
217
	return do_sys_settimeofday64(tp, NULL);
218 219
}

220 221 222 223 224 225
static int posix_clock_realtime_adj(const clockid_t which_clock,
				    struct timex *t)
{
	return do_adjtimex(t);
}

226 227 228
/*
 * Get monotonic time for posix timers
 */
229
static int posix_ktime_get_ts(clockid_t which_clock, struct timespec64 *tp)
230
{
231
	ktime_get_ts64(tp);
232 233
	return 0;
}
L
Linus Torvalds 已提交
234

235
/*
236
 * Get monotonic-raw time for posix timers
237
 */
238
static int posix_get_monotonic_raw(clockid_t which_clock, struct timespec64 *tp)
239
{
240
	getrawmonotonic64(tp);
241 242 243
	return 0;
}

244

245
static int posix_get_realtime_coarse(clockid_t which_clock, struct timespec64 *tp)
246
{
247
	*tp = current_kernel_time64();
248 249 250 251
	return 0;
}

static int posix_get_monotonic_coarse(clockid_t which_clock,
252
						struct timespec64 *tp)
253
{
254
	*tp = get_monotonic_coarse64();
255 256 257
	return 0;
}

258
static int posix_get_coarse_res(const clockid_t which_clock, struct timespec64 *tp)
259
{
260
	*tp = ktime_to_timespec64(KTIME_LOW_RES);
261 262
	return 0;
}
263

264
static int posix_get_boottime(const clockid_t which_clock, struct timespec64 *tp)
265
{
266
	get_monotonic_boottime64(tp);
267 268 269
	return 0;
}

270
static int posix_get_tai(clockid_t which_clock, struct timespec64 *tp)
J
John Stultz 已提交
271
{
272
	timekeeping_clocktai64(tp);
J
John Stultz 已提交
273 274
	return 0;
}
275

276
static int posix_get_hrtimer_res(clockid_t which_clock, struct timespec64 *tp)
277 278 279 280 281 282
{
	tp->tv_sec = 0;
	tp->tv_nsec = hrtimer_resolution;
	return 0;
}

L
Linus Torvalds 已提交
283 284 285 286 287
/*
 * Initialize everything, well, just everything in Posix clocks/timers ;)
 */
static __init int init_posix_timers(void)
{
288
	struct k_clock clock_realtime = {
289
		.clock_getres	= posix_get_hrtimer_res,
290
		.clock_get	= posix_clock_realtime_get,
291
		.clock_set	= posix_clock_realtime_set,
292
		.clock_adj	= posix_clock_realtime_adj,
293
		.nsleep		= common_nsleep,
294
		.nsleep_restart	= hrtimer_nanosleep_restart,
295
		.timer_create	= common_timer_create,
296
		.timer_set	= common_timer_set,
297
		.timer_get	= common_timer_get,
298
		.timer_del	= common_timer_del,
L
Linus Torvalds 已提交
299
	};
300
	struct k_clock clock_monotonic = {
301
		.clock_getres	= posix_get_hrtimer_res,
302
		.clock_get	= posix_ktime_get_ts,
303
		.nsleep		= common_nsleep,
304
		.nsleep_restart	= hrtimer_nanosleep_restart,
305
		.timer_create	= common_timer_create,
306
		.timer_set	= common_timer_set,
307
		.timer_get	= common_timer_get,
308
		.timer_del	= common_timer_del,
L
Linus Torvalds 已提交
309
	};
310
	struct k_clock clock_monotonic_raw = {
311
		.clock_getres	= posix_get_hrtimer_res,
312
		.clock_get	= posix_get_monotonic_raw,
313
	};
314
	struct k_clock clock_realtime_coarse = {
315 316
		.clock_getres	= posix_get_coarse_res,
		.clock_get	= posix_get_realtime_coarse,
317 318
	};
	struct k_clock clock_monotonic_coarse = {
319 320
		.clock_getres	= posix_get_coarse_res,
		.clock_get	= posix_get_monotonic_coarse,
321
	};
J
John Stultz 已提交
322
	struct k_clock clock_tai = {
323
		.clock_getres	= posix_get_hrtimer_res,
J
John Stultz 已提交
324
		.clock_get	= posix_get_tai,
325 326 327 328 329 330
		.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,
J
John Stultz 已提交
331
	};
332
	struct k_clock clock_boottime = {
333
		.clock_getres	= posix_get_hrtimer_res,
334 335 336 337 338 339 340 341
		.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,
	};
L
Linus Torvalds 已提交
342

343 344 345 346 347
	posix_timers_register_clock(CLOCK_REALTIME, &clock_realtime);
	posix_timers_register_clock(CLOCK_MONOTONIC, &clock_monotonic);
	posix_timers_register_clock(CLOCK_MONOTONIC_RAW, &clock_monotonic_raw);
	posix_timers_register_clock(CLOCK_REALTIME_COARSE, &clock_realtime_coarse);
	posix_timers_register_clock(CLOCK_MONOTONIC_COARSE, &clock_monotonic_coarse);
348
	posix_timers_register_clock(CLOCK_BOOTTIME, &clock_boottime);
J
John Stultz 已提交
349
	posix_timers_register_clock(CLOCK_TAI, &clock_tai);
L
Linus Torvalds 已提交
350 351

	posix_timers_cache = kmem_cache_create("posix_timers_cache",
352 353
					sizeof (struct k_itimer), 0, SLAB_PANIC,
					NULL);
L
Linus Torvalds 已提交
354 355 356 357 358 359 360
	return 0;
}

__initcall(init_posix_timers);

static void schedule_next_timer(struct k_itimer *timr)
{
361 362
	struct hrtimer *timer = &timr->it.real.timer;

T
Thomas Gleixner 已提交
363
	if (timr->it.real.interval == 0)
L
Linus Torvalds 已提交
364 365
		return;

D
Davide Libenzi 已提交
366 367 368
	timr->it_overrun += (unsigned int) hrtimer_forward(timer,
						timer->base->get_time(),
						timr->it.real.interval);
369

L
Linus Torvalds 已提交
370 371 372
	timr->it_overrun_last = timr->it_overrun;
	timr->it_overrun = -1;
	++timr->it_requeue_pending;
373
	hrtimer_restart(timer);
L
Linus Torvalds 已提交
374 375 376 377 378 379 380 381 382 383
}

/*
 * 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
384
 * To protect against the timer going away while the interrupt is queued,
L
Linus Torvalds 已提交
385 386 387 388 389 390 391 392 393
 * 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);

394 395 396 397 398
	if (timr && timr->it_requeue_pending == info->si_sys_private) {
		if (timr->it_clock < 0)
			posix_cpu_timer_schedule(timr);
		else
			schedule_next_timer(timr);
L
Linus Torvalds 已提交
399

400
		info->si_overrun += timr->it_overrun_last;
401 402
	}

403 404
	if (timr)
		unlock_timer(timr, flags);
L
Linus Torvalds 已提交
405 406
}

407
int posix_timer_event(struct k_itimer *timr, int si_private)
L
Linus Torvalds 已提交
408
{
409 410
	struct task_struct *task;
	int shared, ret = -1;
411 412 413 414 415 416 417 418 419 420 421
	/*
	 * 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
422 423
	timr->sigq->info.si_sys_private = si_private;

424 425 426 427 428 429 430
	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();
431 432
	/* If we failed to send the signal the timer stops. */
	return ret > 0;
L
Linus Torvalds 已提交
433 434 435 436 437 438 439 440 441 442
}
EXPORT_SYMBOL_GPL(posix_timer_event);

/*
 * 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.
 */
443
static enum hrtimer_restart posix_timer_fn(struct hrtimer *timer)
L
Linus Torvalds 已提交
444
{
445
	struct k_itimer *timr;
L
Linus Torvalds 已提交
446
	unsigned long flags;
447
	int si_private = 0;
448
	enum hrtimer_restart ret = HRTIMER_NORESTART;
L
Linus Torvalds 已提交
449

450
	timr = container_of(timer, struct k_itimer, it.real.timer);
L
Linus Torvalds 已提交
451 452
	spin_lock_irqsave(&timr->it_lock, flags);

T
Thomas Gleixner 已提交
453
	if (timr->it.real.interval != 0)
454
		si_private = ++timr->it_requeue_pending;
L
Linus Torvalds 已提交
455

456 457 458 459 460 461
	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.
		 */
T
Thomas Gleixner 已提交
462
		if (timr->it.real.interval != 0) {
463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488
			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
489
				ktime_t kj = NSEC_PER_SEC / HZ;
490

T
Thomas Gleixner 已提交
491
				if (timr->it.real.interval < kj)
492 493 494
					now = ktime_add(now, kj);
			}
#endif
D
Davide Libenzi 已提交
495
			timr->it_overrun += (unsigned int)
496
				hrtimer_forward(timer, now,
497 498
						timr->it.real.interval);
			ret = HRTIMER_RESTART;
499
			++timr->it_requeue_pending;
L
Linus Torvalds 已提交
500 501 502
		}
	}

503 504 505
	unlock_timer(timr, flags);
	return ret;
}
L
Linus Torvalds 已提交
506

507
static struct pid *good_sigevent(sigevent_t * event)
L
Linus Torvalds 已提交
508 509 510 511
{
	struct task_struct *rtn = current->group_leader;

	if ((event->sigev_notify & SIGEV_THREAD_ID ) &&
512
		(!(rtn = find_task_by_vpid(event->sigev_notify_thread_id)) ||
513
		 !same_thread_group(rtn, current) ||
L
Linus Torvalds 已提交
514 515 516 517 518 519 520
		 (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;

521
	return task_pid(rtn);
L
Linus Torvalds 已提交
522 523
}

524 525
void posix_timers_register_clock(const clockid_t clock_id,
				 struct k_clock *new_clock)
L
Linus Torvalds 已提交
526 527
{
	if ((unsigned) clock_id >= MAX_CLOCKS) {
528 529 530 531 532 533 534 535 536 537 538 539
		printk(KERN_WARNING "POSIX clock register failed for clock_id %d\n",
		       clock_id);
		return;
	}

	if (!new_clock->clock_get) {
		printk(KERN_WARNING "POSIX clock id %d lacks clock_get()\n",
		       clock_id);
		return;
	}
	if (!new_clock->clock_getres) {
		printk(KERN_WARNING "POSIX clock id %d lacks clock_getres()\n",
L
Linus Torvalds 已提交
540 541 542 543 544 545
		       clock_id);
		return;
	}

	posix_clocks[clock_id] = *new_clock;
}
546
EXPORT_SYMBOL_GPL(posix_timers_register_clock);
L
Linus Torvalds 已提交
547 548 549 550

static struct k_itimer * alloc_posix_timer(void)
{
	struct k_itimer *tmr;
551
	tmr = kmem_cache_zalloc(posix_timers_cache, GFP_KERNEL);
L
Linus Torvalds 已提交
552 553 554 555
	if (!tmr)
		return tmr;
	if (unlikely(!(tmr->sigq = sigqueue_alloc()))) {
		kmem_cache_free(posix_timers_cache, tmr);
556
		return NULL;
L
Linus Torvalds 已提交
557
	}
558
	memset(&tmr->sigq->info, 0, sizeof(siginfo_t));
L
Linus Torvalds 已提交
559 560 561
	return tmr;
}

E
Eric Dumazet 已提交
562 563 564 565 566 567 568
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
569 570 571 572 573 574
#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;
575 576 577
		spin_lock_irqsave(&hash_lock, flags);
		hlist_del_rcu(&tmr->t_hash);
		spin_unlock_irqrestore(&hash_lock, flags);
L
Linus Torvalds 已提交
578
	}
579
	put_pid(tmr->it_pid);
L
Linus Torvalds 已提交
580
	sigqueue_free(tmr->sigq);
E
Eric Dumazet 已提交
581
	call_rcu(&tmr->it.rcu, k_itimer_rcu_free);
L
Linus Torvalds 已提交
582 583
}

584 585 586
static struct k_clock *clockid_to_kclock(const clockid_t id)
{
	if (id < 0)
587 588
		return (id & CLOCKFD_MASK) == CLOCKFD ?
			&clock_posix_dynamic : &clock_posix_cpu;
589 590 591 592 593 594

	if (id >= MAX_CLOCKS || !posix_clocks[id].clock_getres)
		return NULL;
	return &posix_clocks[id];
}

595 596 597 598 599 600
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
601 602
/* Create a POSIX.1b interval timer. */

603 604 605
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
606
{
607
	struct k_clock *kc = clockid_to_kclock(which_clock);
608
	struct k_itimer *new_timer;
609
	int error, new_timer_id;
L
Linus Torvalds 已提交
610 611 612
	sigevent_t event;
	int it_id_set = IT_ID_NOT_SET;

613
	if (!kc)
L
Linus Torvalds 已提交
614
		return -EINVAL;
615 616
	if (!kc->timer_create)
		return -EOPNOTSUPP;
L
Linus Torvalds 已提交
617 618 619 620 621 622

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

	spin_lock_init(&new_timer->it_lock);
623 624 625
	new_timer_id = posix_timer_add(new_timer);
	if (new_timer_id < 0) {
		error = new_timer_id;
L
Linus Torvalds 已提交
626 627 628 629 630 631 632 633 634 635 636 637 638
		goto out;
	}

	it_id_set = IT_ID_SET;
	new_timer->it_id = (timer_t) new_timer_id;
	new_timer->it_clock = which_clock;
	new_timer->it_overrun = -1;

	if (timer_event_spec) {
		if (copy_from_user(&event, timer_event_spec, sizeof (event))) {
			error = -EFAULT;
			goto out;
		}
639
		rcu_read_lock();
640
		new_timer->it_pid = get_pid(good_sigevent(&event));
641
		rcu_read_unlock();
642
		if (!new_timer->it_pid) {
L
Linus Torvalds 已提交
643 644 645 646
			error = -EINVAL;
			goto out;
		}
	} else {
647
		memset(&event.sigev_value, 0, sizeof(event.sigev_value));
648 649 650
		event.sigev_notify = SIGEV_SIGNAL;
		event.sigev_signo = SIGALRM;
		event.sigev_value.sival_int = new_timer->it_id;
651
		new_timer->it_pid = get_pid(task_tgid(current));
L
Linus Torvalds 已提交
652 653
	}

654 655 656
	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;
657
	new_timer->sigq->info.si_tid   = new_timer->it_id;
658
	new_timer->sigq->info.si_code  = SI_TIMER;
659

660 661 662 663 664 665
	if (copy_to_user(created_timer_id,
			 &new_timer_id, sizeof (new_timer_id))) {
		error = -EFAULT;
		goto out;
	}

666
	error = kc->timer_create(new_timer);
667 668 669
	if (error)
		goto out;

670
	spin_lock_irq(&current->sighand->siglock);
671
	new_timer->it_signal = current->signal;
672 673
	list_add(&new_timer->list, &current->signal->posix_timers);
	spin_unlock_irq(&current->sighand->siglock);
674 675

	return 0;
676
	/*
L
Linus Torvalds 已提交
677 678 679 680 681 682
	 * 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:
683
	release_posix_timer(new_timer, it_id_set);
L
Linus Torvalds 已提交
684 685 686 687 688 689 690 691 692 693
	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
694
static struct k_itimer *__lock_timer(timer_t timer_id, unsigned long *flags)
L
Linus Torvalds 已提交
695 696
{
	struct k_itimer *timr;
E
Eric Dumazet 已提交
697

698 699 700 701 702 703 704
	/*
	 * 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
705
	rcu_read_lock();
706
	timr = posix_timer_by_id(timer_id);
L
Linus Torvalds 已提交
707
	if (timr) {
E
Eric Dumazet 已提交
708
		spin_lock_irqsave(&timr->it_lock, *flags);
709
		if (timr->it_signal == current->signal) {
E
Eric Dumazet 已提交
710
			rcu_read_unlock();
711 712
			return timr;
		}
E
Eric Dumazet 已提交
713
		spin_unlock_irqrestore(&timr->it_lock, *flags);
714
	}
E
Eric Dumazet 已提交
715
	rcu_read_unlock();
L
Linus Torvalds 已提交
716

717
	return NULL;
L
Linus Torvalds 已提交
718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738
}

/*
 * 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
common_timer_get(struct k_itimer *timr, struct itimerspec *cur_setting)
{
739
	ktime_t now, remaining, iv;
740
	struct hrtimer *timer = &timr->it.real.timer;
L
Linus Torvalds 已提交
741

742 743
	memset(cur_setting, 0, sizeof(struct itimerspec));

744 745
	iv = timr->it.real.interval;

746
	/* interval timer ? */
T
Thomas Gleixner 已提交
747
	if (iv)
748 749 750
		cur_setting->it_interval = ktime_to_timespec(iv);
	else if (!hrtimer_active(timer) &&
		 (timr->it_sigev_notify & ~SIGEV_THREAD_ID) != SIGEV_NONE)
751
		return;
752 753 754

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

755
	/*
756 757 758
	 * When a requeue is pending or this is a SIGEV_NONE
	 * timer move the expiry time forward by intervals, so
	 * expiry is > now.
759
	 */
T
Thomas Gleixner 已提交
760 761
	if (iv && (timr->it_requeue_pending & REQUEUE_PENDING ||
		   (timr->it_sigev_notify & ~SIGEV_THREAD_ID) == SIGEV_NONE))
D
Davide Libenzi 已提交
762
		timr->it_overrun += (unsigned int) hrtimer_forward(timer, now, iv);
763

764
	remaining = __hrtimer_expires_remaining_adjusted(timer, now);
765
	/* Return 0 only, when the timer is expired and not pending */
T
Thomas Gleixner 已提交
766
	if (remaining <= 0) {
767 768 769 770 771 772 773
		/*
		 * 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
774
		cur_setting->it_value = ktime_to_timespec(remaining);
L
Linus Torvalds 已提交
775 776 777
}

/* Get the time remaining on a POSIX.1b interval timer. */
778 779
SYSCALL_DEFINE2(timer_gettime, timer_t, timer_id,
		struct itimerspec __user *, setting)
L
Linus Torvalds 已提交
780 781
{
	struct itimerspec cur_setting;
782 783
	struct k_itimer *timr;
	struct k_clock *kc;
L
Linus Torvalds 已提交
784
	unsigned long flags;
785
	int ret = 0;
L
Linus Torvalds 已提交
786 787 788 789 790

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

791 792 793 794 795
	kc = clockid_to_kclock(timr->it_clock);
	if (WARN_ON_ONCE(!kc || !kc->timer_get))
		ret = -EINVAL;
	else
		kc->timer_get(timr, &cur_setting);
L
Linus Torvalds 已提交
796 797 798

	unlock_timer(timr, flags);

799
	if (!ret && copy_to_user(setting, &cur_setting, sizeof (cur_setting)))
L
Linus Torvalds 已提交
800 801
		return -EFAULT;

802
	return ret;
L
Linus Torvalds 已提交
803
}
804

L
Linus Torvalds 已提交
805 806 807 808 809 810 811 812 813
/*
 * 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.
 */
814
SYSCALL_DEFINE1(timer_getoverrun, timer_t, timer_id)
L
Linus Torvalds 已提交
815 816 817
{
	struct k_itimer *timr;
	int overrun;
818
	unsigned long flags;
L
Linus Torvalds 已提交
819 820 821 822 823 824 825 826 827 828 829 830 831

	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. */
832
static int
L
Linus Torvalds 已提交
833 834 835
common_timer_set(struct k_itimer *timr, int flags,
		 struct itimerspec *new_setting, struct itimerspec *old_setting)
{
836
	struct hrtimer *timer = &timr->it.real.timer;
837
	enum hrtimer_mode mode;
L
Linus Torvalds 已提交
838 839 840 841 842

	if (old_setting)
		common_timer_get(timr, old_setting);

	/* disable the timer */
T
Thomas Gleixner 已提交
843
	timr->it.real.interval = 0;
L
Linus Torvalds 已提交
844 845 846 847
	/*
	 * 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.
	 */
848
	if (hrtimer_try_to_cancel(timer) < 0)
L
Linus Torvalds 已提交
849 850 851 852 853 854
		return TIMER_RETRY;

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

855 856 857
	/* 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
858

859
	mode = flags & TIMER_ABSTIME ? HRTIMER_MODE_ABS : HRTIMER_MODE_REL;
860 861
	hrtimer_init(&timr->it.real.timer, timr->it_clock, mode);
	timr->it.real.timer.function = posix_timer_fn;
862

863
	hrtimer_set_expires(timer, timespec_to_ktime(new_setting->it_value));
864 865 866 867 868

	/* Convert interval */
	timr->it.real.interval = timespec_to_ktime(new_setting->it_interval);

	/* SIGEV_NONE timers are not queued ! See common_timer_get */
869 870
	if (((timr->it_sigev_notify & ~SIGEV_THREAD_ID) == SIGEV_NONE)) {
		/* Setup correct expiry time for relative timers */
871
		if (mode == HRTIMER_MODE_REL) {
872
			hrtimer_add_expires(timer, timer->base->get_time());
873
		}
874
		return 0;
875
	}
876

877
	hrtimer_start_expires(timer, mode);
L
Linus Torvalds 已提交
878 879 880 881
	return 0;
}

/* Set a POSIX.1b interval timer */
882 883 884
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
885 886 887 888
{
	struct k_itimer *timr;
	struct itimerspec new_spec, old_spec;
	int error = 0;
889
	unsigned long flag;
L
Linus Torvalds 已提交
890
	struct itimerspec *rtn = old_setting ? &old_spec : NULL;
891
	struct k_clock *kc;
L
Linus Torvalds 已提交
892 893 894 895 896 897 898

	if (!new_setting)
		return -EINVAL;

	if (copy_from_user(&new_spec, new_setting, sizeof (new_spec)))
		return -EFAULT;

899 900
	if (!timespec_valid(&new_spec.it_interval) ||
	    !timespec_valid(&new_spec.it_value))
L
Linus Torvalds 已提交
901 902 903 904 905 906
		return -EINVAL;
retry:
	timr = lock_timer(timer_id, &flag);
	if (!timr)
		return -EINVAL;

907 908 909 910 911
	kc = clockid_to_kclock(timr->it_clock);
	if (WARN_ON_ONCE(!kc || !kc->timer_set))
		error = -EINVAL;
	else
		error = kc->timer_set(timr, flags, &new_spec, rtn);
L
Linus Torvalds 已提交
912 913 914 915 916 917 918

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

919 920
	if (old_setting && !error &&
	    copy_to_user(old_setting, &old_spec, sizeof (old_spec)))
L
Linus Torvalds 已提交
921 922 923 924 925
		error = -EFAULT;

	return error;
}

926
static int common_timer_del(struct k_itimer *timer)
L
Linus Torvalds 已提交
927
{
T
Thomas Gleixner 已提交
928
	timer->it.real.interval = 0;
929

930
	if (hrtimer_try_to_cancel(&timer->it.real.timer) < 0)
L
Linus Torvalds 已提交
931 932 933 934 935 936
		return TIMER_RETRY;
	return 0;
}

static inline int timer_delete_hook(struct k_itimer *timer)
{
937 938 939 940 941
	struct k_clock *kc = clockid_to_kclock(timer->it_clock);

	if (WARN_ON_ONCE(!kc || !kc->timer_del))
		return -EINVAL;
	return kc->timer_del(timer);
L
Linus Torvalds 已提交
942 943 944
}

/* Delete a POSIX.1b interval timer. */
945
SYSCALL_DEFINE1(timer_delete, timer_t, timer_id)
L
Linus Torvalds 已提交
946 947
{
	struct k_itimer *timer;
948
	unsigned long flags;
L
Linus Torvalds 已提交
949 950 951 952 953 954

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

955
	if (timer_delete_hook(timer) == TIMER_RETRY) {
L
Linus Torvalds 已提交
956 957 958
		unlock_timer(timer, flags);
		goto retry_delete;
	}
959

L
Linus Torvalds 已提交
960 961 962 963 964 965 966
	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).
	 */
967
	timer->it_signal = NULL;
968

L
Linus Torvalds 已提交
969 970 971 972
	unlock_timer(timer, flags);
	release_posix_timer(timer, IT_ID_SET);
	return 0;
}
973

L
Linus Torvalds 已提交
974 975 976
/*
 * return timer owned by the process, used by exit_itimers
 */
977
static void itimer_delete(struct k_itimer *timer)
L
Linus Torvalds 已提交
978 979 980 981 982 983
{
	unsigned long flags;

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

984
	if (timer_delete_hook(timer) == TIMER_RETRY) {
L
Linus Torvalds 已提交
985 986 987 988 989 990 991 992
		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).
	 */
993
	timer->it_signal = NULL;
994

L
Linus Torvalds 已提交
995 996 997 998 999
	unlock_timer(timer, flags);
	release_posix_timer(timer, IT_ID_SET);
}

/*
1000
 * This is called by do_exit or de_thread, only when there are no more
L
Linus Torvalds 已提交
1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012
 * 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);
	}
}

1013 1014
SYSCALL_DEFINE2(clock_settime, const clockid_t, which_clock,
		const struct timespec __user *, tp)
L
Linus Torvalds 已提交
1015
{
1016
	struct k_clock *kc = clockid_to_kclock(which_clock);
1017
	struct timespec64 new_tp64;
L
Linus Torvalds 已提交
1018 1019
	struct timespec new_tp;

1020
	if (!kc || !kc->clock_set)
L
Linus Torvalds 已提交
1021
		return -EINVAL;
1022

L
Linus Torvalds 已提交
1023 1024
	if (copy_from_user(&new_tp, tp, sizeof (*tp)))
		return -EFAULT;
1025
	new_tp64 = timespec_to_timespec64(new_tp);
L
Linus Torvalds 已提交
1026

1027
	return kc->clock_set(which_clock, &new_tp64);
L
Linus Torvalds 已提交
1028 1029
}

1030 1031
SYSCALL_DEFINE2(clock_gettime, const clockid_t, which_clock,
		struct timespec __user *,tp)
L
Linus Torvalds 已提交
1032
{
1033
	struct k_clock *kc = clockid_to_kclock(which_clock);
1034
	struct timespec64 kernel_tp64;
L
Linus Torvalds 已提交
1035 1036 1037
	struct timespec kernel_tp;
	int error;

1038
	if (!kc)
L
Linus Torvalds 已提交
1039
		return -EINVAL;
1040

1041 1042
	error = kc->clock_get(which_clock, &kernel_tp64);
	kernel_tp = timespec64_to_timespec(kernel_tp64);
1043

L
Linus Torvalds 已提交
1044 1045 1046 1047 1048 1049
	if (!error && copy_to_user(tp, &kernel_tp, sizeof (kernel_tp)))
		error = -EFAULT;

	return error;
}

1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066
SYSCALL_DEFINE2(clock_adjtime, const clockid_t, which_clock,
		struct timex __user *, utx)
{
	struct k_clock *kc = clockid_to_kclock(which_clock);
	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);

1067
	if (err >= 0 && copy_to_user(utx, &ktx, sizeof(ktx)))
1068 1069 1070 1071 1072
		return -EFAULT;

	return err;
}

1073 1074
SYSCALL_DEFINE2(clock_getres, const clockid_t, which_clock,
		struct timespec __user *, tp)
L
Linus Torvalds 已提交
1075
{
1076
	struct k_clock *kc = clockid_to_kclock(which_clock);
1077
	struct timespec64 rtn_tp64;
L
Linus Torvalds 已提交
1078 1079 1080
	struct timespec rtn_tp;
	int error;

1081
	if (!kc)
L
Linus Torvalds 已提交
1082 1083
		return -EINVAL;

1084 1085
	error = kc->clock_getres(which_clock, &rtn_tp64);
	rtn_tp = timespec64_to_timespec(rtn_tp64);
L
Linus Torvalds 已提交
1086

1087
	if (!error && tp && copy_to_user(tp, &rtn_tp, sizeof (rtn_tp)))
L
Linus Torvalds 已提交
1088 1089 1090 1091 1092
		error = -EFAULT;

	return error;
}

1093 1094 1095 1096 1097 1098
/*
 * nanosleep for monotonic and realtime clocks
 */
static int common_nsleep(const clockid_t which_clock, int flags,
			 struct timespec *tsave, struct timespec __user *rmtp)
{
1099 1100 1101
	return hrtimer_nanosleep(tsave, rmtp, flags & TIMER_ABSTIME ?
				 HRTIMER_MODE_ABS : HRTIMER_MODE_REL,
				 which_clock);
1102
}
L
Linus Torvalds 已提交
1103

1104 1105 1106
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
1107
{
1108
	struct k_clock *kc = clockid_to_kclock(which_clock);
L
Linus Torvalds 已提交
1109 1110
	struct timespec t;

1111
	if (!kc)
L
Linus Torvalds 已提交
1112
		return -EINVAL;
1113 1114
	if (!kc->nsleep)
		return -ENANOSLEEP_NOTSUP;
L
Linus Torvalds 已提交
1115 1116 1117 1118

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

1119
	if (!timespec_valid(&t))
L
Linus Torvalds 已提交
1120 1121
		return -EINVAL;

1122
	return kc->nsleep(which_clock, flags, &t, rmtp);
L
Linus Torvalds 已提交
1123
}
1124 1125 1126 1127 1128

/*
 * This will restart clock_nanosleep. This is required only by
 * compat_clock_nanosleep_restart for now.
 */
1129
long clock_nanosleep_restart(struct restart_block *restart_block)
1130
{
1131
	clockid_t which_clock = restart_block->nanosleep.clockid;
1132 1133 1134 1135
	struct k_clock *kc = clockid_to_kclock(which_clock);

	if (WARN_ON_ONCE(!kc || !kc->nsleep_restart))
		return -EINVAL;
1136

1137
	return kc->nsleep_restart(restart_block);
1138
}