posix-timers.c 30.5 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>
L
Linus Torvalds 已提交
38

39
#include <linux/uaccess.h>
L
Linus Torvalds 已提交
40 41 42
#include <linux/list.h>
#include <linux/init.h>
#include <linux/compiler.h>
43
#include <linux/hash.h>
44
#include <linux/posix-clock.h>
L
Linus Torvalds 已提交
45 46 47 48
#include <linux/posix-timers.h>
#include <linux/syscalls.h>
#include <linux/wait.h>
#include <linux/workqueue.h>
49
#include <linux/export.h>
50
#include <linux/hashtable.h>
L
Linus Torvalds 已提交
51

52 53
#include "timekeeping.h"

L
Linus Torvalds 已提交
54
/*
55 56 57 58 59 60
 * 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
61 62 63 64 65
 */

/*
 * Lets keep our timers in a slab cache :-)
 */
66
static struct kmem_cache *posix_timers_cache;
67 68 69

static DEFINE_HASHTABLE(posix_timers_hashtable, 9);
static DEFINE_SPINLOCK(hash_lock);
L
Linus Torvalds 已提交
70 71 72 73 74 75 76 77 78 79

/*
 * 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

80 81 82 83 84 85 86 87
/*
 * parisc wants ENOTSUP instead of EOPNOTSUPP
 */
#ifndef ENOTSUP
# define ENANOSLEEP_NOTSUP EOPNOTSUPP
#else
# define ENANOSLEEP_NOTSUP ENOTSUP
#endif
L
Linus Torvalds 已提交
88 89 90 91 92 93 94 95 96 97 98 99 100

/*
 * 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
101
 *	    clocks.
L
Linus Torvalds 已提交
102 103 104 105 106 107 108 109 110
 *
 * 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
111 112
 * FUNCTIONS: The CLOCKs structure defines possible functions to
 *	    handle various clock functions.
L
Linus Torvalds 已提交
113
 *
R
Richard Cochran 已提交
114 115 116 117
 *	    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
118 119 120 121 122 123 124 125 126 127
 *
 * 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];
128

L
Linus Torvalds 已提交
129
/*
130
 * These ones are defined below.
L
Linus Torvalds 已提交
131
 */
132 133
static int common_nsleep(const clockid_t, int flags, struct timespec *t,
			 struct timespec __user *rmtp);
134
static int common_timer_create(struct k_itimer *new_timer);
135 136 137 138
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
139

140
static enum hrtimer_restart posix_timer_fn(struct hrtimer *data);
L
Linus Torvalds 已提交
141

N
Namhyung Kim 已提交
142 143 144 145 146 147 148
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
149

150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199
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
200 201 202 203 204
static inline void unlock_timer(struct k_itimer *timr, unsigned long flags)
{
	spin_unlock_irqrestore(&timr->it_lock, flags);
}

205 206 207 208 209 210 211
/* Get clock_realtime */
static int posix_clock_realtime_get(clockid_t which_clock, struct timespec *tp)
{
	ktime_get_real_ts(tp);
	return 0;
}

212 213 214 215 216 217 218
/* Set clock_realtime */
static int posix_clock_realtime_set(const clockid_t which_clock,
				    const struct timespec *tp)
{
	return do_sys_settimeofday(tp, NULL);
}

219 220 221 222 223 224
static int posix_clock_realtime_adj(const clockid_t which_clock,
				    struct timex *t)
{
	return do_adjtimex(t);
}

225 226 227 228 229 230 231 232
/*
 * Get monotonic time for posix timers
 */
static int posix_ktime_get_ts(clockid_t which_clock, struct timespec *tp)
{
	ktime_get_ts(tp);
	return 0;
}
L
Linus Torvalds 已提交
233

234
/*
235
 * Get monotonic-raw time for posix timers
236 237 238 239 240 241 242
 */
static int posix_get_monotonic_raw(clockid_t which_clock, struct timespec *tp)
{
	getrawmonotonic(tp);
	return 0;
}

243 244 245 246 247 248 249 250 251 252 253 254 255 256

static int posix_get_realtime_coarse(clockid_t which_clock, struct timespec *tp)
{
	*tp = current_kernel_time();
	return 0;
}

static int posix_get_monotonic_coarse(clockid_t which_clock,
						struct timespec *tp)
{
	*tp = get_monotonic_coarse();
	return 0;
}

257
static int posix_get_coarse_res(const clockid_t which_clock, struct timespec *tp)
258 259 260 261
{
	*tp = ktime_to_timespec(KTIME_LOW_RES);
	return 0;
}
262 263 264 265 266 267 268

static int posix_get_boottime(const clockid_t which_clock, struct timespec *tp)
{
	get_monotonic_boottime(tp);
	return 0;
}

J
John Stultz 已提交
269 270 271 272 273
static int posix_get_tai(clockid_t which_clock, struct timespec *tp)
{
	timekeeping_clocktai(tp);
	return 0;
}
274

275 276 277 278 279 280 281
static int posix_get_hrtimer_res(clockid_t which_clock, struct timespec *tp)
{
	tp->tv_sec = 0;
	tp->tv_nsec = hrtimer_resolution;
	return 0;
}

L
Linus Torvalds 已提交
282 283 284 285 286
/*
 * Initialize everything, well, just everything in Posix clocks/timers ;)
 */
static __init int init_posix_timers(void)
{
287
	struct k_clock clock_realtime = {
288
		.clock_getres	= posix_get_hrtimer_res,
289
		.clock_get	= posix_clock_realtime_get,
290
		.clock_set	= posix_clock_realtime_set,
291
		.clock_adj	= posix_clock_realtime_adj,
292
		.nsleep		= common_nsleep,
293
		.nsleep_restart	= hrtimer_nanosleep_restart,
294
		.timer_create	= common_timer_create,
295
		.timer_set	= common_timer_set,
296
		.timer_get	= common_timer_get,
297
		.timer_del	= common_timer_del,
L
Linus Torvalds 已提交
298
	};
299
	struct k_clock clock_monotonic = {
300
		.clock_getres	= posix_get_hrtimer_res,
301
		.clock_get	= posix_ktime_get_ts,
302
		.nsleep		= common_nsleep,
303
		.nsleep_restart	= hrtimer_nanosleep_restart,
304
		.timer_create	= common_timer_create,
305
		.timer_set	= common_timer_set,
306
		.timer_get	= common_timer_get,
307
		.timer_del	= common_timer_del,
L
Linus Torvalds 已提交
308
	};
309
	struct k_clock clock_monotonic_raw = {
310
		.clock_getres	= posix_get_hrtimer_res,
311
		.clock_get	= posix_get_monotonic_raw,
312
	};
313
	struct k_clock clock_realtime_coarse = {
314 315
		.clock_getres	= posix_get_coarse_res,
		.clock_get	= posix_get_realtime_coarse,
316 317
	};
	struct k_clock clock_monotonic_coarse = {
318 319
		.clock_getres	= posix_get_coarse_res,
		.clock_get	= posix_get_monotonic_coarse,
320
	};
J
John Stultz 已提交
321
	struct k_clock clock_tai = {
322
		.clock_getres	= posix_get_hrtimer_res,
J
John Stultz 已提交
323
		.clock_get	= posix_get_tai,
324 325 326 327 328 329
		.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
330
	};
331
	struct k_clock clock_boottime = {
332
		.clock_getres	= posix_get_hrtimer_res,
333 334 335 336 337 338 339 340
		.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
341

342 343 344 345 346
	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);
347
	posix_timers_register_clock(CLOCK_BOOTTIME, &clock_boottime);
J
John Stultz 已提交
348
	posix_timers_register_clock(CLOCK_TAI, &clock_tai);
L
Linus Torvalds 已提交
349 350

	posix_timers_cache = kmem_cache_create("posix_timers_cache",
351 352
					sizeof (struct k_itimer), 0, SLAB_PANIC,
					NULL);
L
Linus Torvalds 已提交
353 354 355 356 357 358 359
	return 0;
}

__initcall(init_posix_timers);

static void schedule_next_timer(struct k_itimer *timr)
{
360 361
	struct hrtimer *timer = &timr->it.real.timer;

362
	if (timr->it.real.interval.tv64 == 0)
L
Linus Torvalds 已提交
363 364
		return;

D
Davide Libenzi 已提交
365 366 367
	timr->it_overrun += (unsigned int) hrtimer_forward(timer,
						timer->base->get_time(),
						timr->it.real.interval);
368

L
Linus Torvalds 已提交
369 370 371
	timr->it_overrun_last = timr->it_overrun;
	timr->it_overrun = -1;
	++timr->it_requeue_pending;
372
	hrtimer_restart(timer);
L
Linus Torvalds 已提交
373 374 375 376 377 378 379 380 381 382
}

/*
 * 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
383
 * To protect against the timer going away while the interrupt is queued,
L
Linus Torvalds 已提交
384 385 386 387 388 389 390 391 392
 * 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);

393 394 395 396 397
	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
398

399
		info->si_overrun += timr->it_overrun_last;
400 401
	}

402 403
	if (timr)
		unlock_timer(timr, flags);
L
Linus Torvalds 已提交
404 405
}

406
int posix_timer_event(struct k_itimer *timr, int si_private)
L
Linus Torvalds 已提交
407
{
408 409
	struct task_struct *task;
	int shared, ret = -1;
410 411 412 413 414 415 416 417 418 419 420
	/*
	 * 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
421 422
	timr->sigq->info.si_sys_private = si_private;

423 424 425 426 427 428 429
	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();
430 431
	/* If we failed to send the signal the timer stops. */
	return ret > 0;
L
Linus Torvalds 已提交
432 433 434 435 436 437 438 439 440 441
}
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.
 */
442
static enum hrtimer_restart posix_timer_fn(struct hrtimer *timer)
L
Linus Torvalds 已提交
443
{
444
	struct k_itimer *timr;
L
Linus Torvalds 已提交
445
	unsigned long flags;
446
	int si_private = 0;
447
	enum hrtimer_restart ret = HRTIMER_NORESTART;
L
Linus Torvalds 已提交
448

449
	timr = container_of(timer, struct k_itimer, it.real.timer);
L
Linus Torvalds 已提交
450 451
	spin_lock_irqsave(&timr->it_lock, flags);

452 453
	if (timr->it.real.interval.tv64 != 0)
		si_private = ++timr->it_requeue_pending;
L
Linus Torvalds 已提交
454

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.
		 */
		if (timr->it.real.interval.tv64 != 0) {
462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493
			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
			{
				ktime_t kj = ktime_set(0, NSEC_PER_SEC / HZ);

				if (timr->it.real.interval.tv64 < kj.tv64)
					now = ktime_add(now, kj);
			}
#endif
D
Davide Libenzi 已提交
494
			timr->it_overrun += (unsigned int)
495
				hrtimer_forward(timer, now,
496 497
						timr->it.real.interval);
			ret = HRTIMER_RESTART;
498
			++timr->it_requeue_pending;
L
Linus Torvalds 已提交
499 500 501
		}
	}

502 503 504
	unlock_timer(timr, flags);
	return ret;
}
L
Linus Torvalds 已提交
505

506
static struct pid *good_sigevent(sigevent_t * event)
L
Linus Torvalds 已提交
507 508 509 510
{
	struct task_struct *rtn = current->group_leader;

	if ((event->sigev_notify & SIGEV_THREAD_ID ) &&
511
		(!(rtn = find_task_by_vpid(event->sigev_notify_thread_id)) ||
512
		 !same_thread_group(rtn, current) ||
L
Linus Torvalds 已提交
513 514 515 516 517 518 519
		 (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;

520
	return task_pid(rtn);
L
Linus Torvalds 已提交
521 522
}

523 524
void posix_timers_register_clock(const clockid_t clock_id,
				 struct k_clock *new_clock)
L
Linus Torvalds 已提交
525 526
{
	if ((unsigned) clock_id >= MAX_CLOCKS) {
527 528 529 530 531 532 533 534 535 536 537 538
		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
539 540 541 542 543 544
		       clock_id);
		return;
	}

	posix_clocks[clock_id] = *new_clock;
}
545
EXPORT_SYMBOL_GPL(posix_timers_register_clock);
L
Linus Torvalds 已提交
546 547 548 549

static struct k_itimer * alloc_posix_timer(void)
{
	struct k_itimer *tmr;
550
	tmr = kmem_cache_zalloc(posix_timers_cache, GFP_KERNEL);
L
Linus Torvalds 已提交
551 552 553 554
	if (!tmr)
		return tmr;
	if (unlikely(!(tmr->sigq = sigqueue_alloc()))) {
		kmem_cache_free(posix_timers_cache, tmr);
555
		return NULL;
L
Linus Torvalds 已提交
556
	}
557
	memset(&tmr->sigq->info, 0, sizeof(siginfo_t));
L
Linus Torvalds 已提交
558 559 560
	return tmr;
}

E
Eric Dumazet 已提交
561 562 563 564 565 566 567
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
568 569 570 571 572 573
#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;
574 575 576
		spin_lock_irqsave(&hash_lock, flags);
		hlist_del_rcu(&tmr->t_hash);
		spin_unlock_irqrestore(&hash_lock, flags);
L
Linus Torvalds 已提交
577
	}
578
	put_pid(tmr->it_pid);
L
Linus Torvalds 已提交
579
	sigqueue_free(tmr->sigq);
E
Eric Dumazet 已提交
580
	call_rcu(&tmr->it.rcu, k_itimer_rcu_free);
L
Linus Torvalds 已提交
581 582
}

583 584 585
static struct k_clock *clockid_to_kclock(const clockid_t id)
{
	if (id < 0)
586 587
		return (id & CLOCKFD_MASK) == CLOCKFD ?
			&clock_posix_dynamic : &clock_posix_cpu;
588 589 590 591 592 593

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

594 595 596 597 598 599
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
600 601
/* Create a POSIX.1b interval timer. */

602 603 604
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
605
{
606
	struct k_clock *kc = clockid_to_kclock(which_clock);
607
	struct k_itimer *new_timer;
608
	int error, new_timer_id;
L
Linus Torvalds 已提交
609 610 611
	sigevent_t event;
	int it_id_set = IT_ID_NOT_SET;

612
	if (!kc)
L
Linus Torvalds 已提交
613
		return -EINVAL;
614 615
	if (!kc->timer_create)
		return -EOPNOTSUPP;
L
Linus Torvalds 已提交
616 617 618 619 620 621

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

	spin_lock_init(&new_timer->it_lock);
622 623 624
	new_timer_id = posix_timer_add(new_timer);
	if (new_timer_id < 0) {
		error = new_timer_id;
L
Linus Torvalds 已提交
625 626 627 628 629 630 631 632 633 634 635 636 637
		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;
		}
638
		rcu_read_lock();
639
		new_timer->it_pid = get_pid(good_sigevent(&event));
640
		rcu_read_unlock();
641
		if (!new_timer->it_pid) {
L
Linus Torvalds 已提交
642 643 644 645
			error = -EINVAL;
			goto out;
		}
	} else {
646
		memset(&event.sigev_value, 0, sizeof(event.sigev_value));
647 648 649
		event.sigev_notify = SIGEV_SIGNAL;
		event.sigev_signo = SIGALRM;
		event.sigev_value.sival_int = new_timer->it_id;
650
		new_timer->it_pid = get_pid(task_tgid(current));
L
Linus Torvalds 已提交
651 652
	}

653 654 655
	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;
656
	new_timer->sigq->info.si_tid   = new_timer->it_id;
657
	new_timer->sigq->info.si_code  = SI_TIMER;
658

659 660 661 662 663 664
	if (copy_to_user(created_timer_id,
			 &new_timer_id, sizeof (new_timer_id))) {
		error = -EFAULT;
		goto out;
	}

665
	error = kc->timer_create(new_timer);
666 667 668
	if (error)
		goto out;

669
	spin_lock_irq(&current->sighand->siglock);
670
	new_timer->it_signal = current->signal;
671 672
	list_add(&new_timer->list, &current->signal->posix_timers);
	spin_unlock_irq(&current->sighand->siglock);
673 674

	return 0;
675
	/*
L
Linus Torvalds 已提交
676 677 678 679 680 681
	 * 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:
682
	release_posix_timer(new_timer, it_id_set);
L
Linus Torvalds 已提交
683 684 685 686 687 688 689 690 691 692
	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
693
static struct k_itimer *__lock_timer(timer_t timer_id, unsigned long *flags)
L
Linus Torvalds 已提交
694 695
{
	struct k_itimer *timr;
E
Eric Dumazet 已提交
696

697 698 699 700 701 702 703
	/*
	 * 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
704
	rcu_read_lock();
705
	timr = posix_timer_by_id(timer_id);
L
Linus Torvalds 已提交
706
	if (timr) {
E
Eric Dumazet 已提交
707
		spin_lock_irqsave(&timr->it_lock, *flags);
708
		if (timr->it_signal == current->signal) {
E
Eric Dumazet 已提交
709
			rcu_read_unlock();
710 711
			return timr;
		}
E
Eric Dumazet 已提交
712
		spin_unlock_irqrestore(&timr->it_lock, *flags);
713
	}
E
Eric Dumazet 已提交
714
	rcu_read_unlock();
L
Linus Torvalds 已提交
715

716
	return NULL;
L
Linus Torvalds 已提交
717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737
}

/*
 * 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)
{
738
	ktime_t now, remaining, iv;
739
	struct hrtimer *timer = &timr->it.real.timer;
L
Linus Torvalds 已提交
740

741 742
	memset(cur_setting, 0, sizeof(struct itimerspec));

743 744
	iv = timr->it.real.interval;

745
	/* interval timer ? */
746 747 748 749
	if (iv.tv64)
		cur_setting->it_interval = ktime_to_timespec(iv);
	else if (!hrtimer_active(timer) &&
		 (timr->it_sigev_notify & ~SIGEV_THREAD_ID) != SIGEV_NONE)
750
		return;
751 752 753

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

754
	/*
755 756 757
	 * When a requeue is pending or this is a SIGEV_NONE
	 * timer move the expiry time forward by intervals, so
	 * expiry is > now.
758
	 */
759 760
	if (iv.tv64 && (timr->it_requeue_pending & REQUEUE_PENDING ||
	    (timr->it_sigev_notify & ~SIGEV_THREAD_ID) == SIGEV_NONE))
D
Davide Libenzi 已提交
761
		timr->it_overrun += (unsigned int) hrtimer_forward(timer, now, iv);
762

763
	remaining = __hrtimer_expires_remaining_adjusted(timer, now);
764
	/* Return 0 only, when the timer is expired and not pending */
765 766 767 768 769 770 771 772
	if (remaining.tv64 <= 0) {
		/*
		 * 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
773
		cur_setting->it_value = ktime_to_timespec(remaining);
L
Linus Torvalds 已提交
774 775 776
}

/* Get the time remaining on a POSIX.1b interval timer. */
777 778
SYSCALL_DEFINE2(timer_gettime, timer_t, timer_id,
		struct itimerspec __user *, setting)
L
Linus Torvalds 已提交
779 780
{
	struct itimerspec cur_setting;
781 782
	struct k_itimer *timr;
	struct k_clock *kc;
L
Linus Torvalds 已提交
783
	unsigned long flags;
784
	int ret = 0;
L
Linus Torvalds 已提交
785 786 787 788 789

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

790 791 792 793 794
	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
795 796 797

	unlock_timer(timr, flags);

798
	if (!ret && copy_to_user(setting, &cur_setting, sizeof (cur_setting)))
L
Linus Torvalds 已提交
799 800
		return -EFAULT;

801
	return ret;
L
Linus Torvalds 已提交
802
}
803

L
Linus Torvalds 已提交
804 805 806 807 808 809 810 811 812
/*
 * 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.
 */
813
SYSCALL_DEFINE1(timer_getoverrun, timer_t, timer_id)
L
Linus Torvalds 已提交
814 815 816
{
	struct k_itimer *timr;
	int overrun;
817
	unsigned long flags;
L
Linus Torvalds 已提交
818 819 820 821 822 823 824 825 826 827 828 829 830

	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. */
831
static int
L
Linus Torvalds 已提交
832 833 834
common_timer_set(struct k_itimer *timr, int flags,
		 struct itimerspec *new_setting, struct itimerspec *old_setting)
{
835
	struct hrtimer *timer = &timr->it.real.timer;
836
	enum hrtimer_mode mode;
L
Linus Torvalds 已提交
837 838 839 840 841

	if (old_setting)
		common_timer_get(timr, old_setting);

	/* disable the timer */
842
	timr->it.real.interval.tv64 = 0;
L
Linus Torvalds 已提交
843 844 845 846
	/*
	 * 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.
	 */
847
	if (hrtimer_try_to_cancel(timer) < 0)
L
Linus Torvalds 已提交
848 849 850 851 852 853
		return TIMER_RETRY;

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

854 855 856
	/* 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
857

858
	mode = flags & TIMER_ABSTIME ? HRTIMER_MODE_ABS : HRTIMER_MODE_REL;
859 860
	hrtimer_init(&timr->it.real.timer, timr->it_clock, mode);
	timr->it.real.timer.function = posix_timer_fn;
861

862
	hrtimer_set_expires(timer, timespec_to_ktime(new_setting->it_value));
863 864 865 866 867

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

	/* SIGEV_NONE timers are not queued ! See common_timer_get */
868 869
	if (((timr->it_sigev_notify & ~SIGEV_THREAD_ID) == SIGEV_NONE)) {
		/* Setup correct expiry time for relative timers */
870
		if (mode == HRTIMER_MODE_REL) {
871
			hrtimer_add_expires(timer, timer->base->get_time());
872
		}
873
		return 0;
874
	}
875

876
	hrtimer_start_expires(timer, mode);
L
Linus Torvalds 已提交
877 878 879 880
	return 0;
}

/* Set a POSIX.1b interval timer */
881 882 883
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
884 885 886 887
{
	struct k_itimer *timr;
	struct itimerspec new_spec, old_spec;
	int error = 0;
888
	unsigned long flag;
L
Linus Torvalds 已提交
889
	struct itimerspec *rtn = old_setting ? &old_spec : NULL;
890
	struct k_clock *kc;
L
Linus Torvalds 已提交
891 892 893 894 895 896 897

	if (!new_setting)
		return -EINVAL;

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

898 899
	if (!timespec_valid(&new_spec.it_interval) ||
	    !timespec_valid(&new_spec.it_value))
L
Linus Torvalds 已提交
900 901 902 903 904 905
		return -EINVAL;
retry:
	timr = lock_timer(timer_id, &flag);
	if (!timr)
		return -EINVAL;

906 907 908 909 910
	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
911 912 913 914 915 916 917

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

918 919
	if (old_setting && !error &&
	    copy_to_user(old_setting, &old_spec, sizeof (old_spec)))
L
Linus Torvalds 已提交
920 921 922 923 924
		error = -EFAULT;

	return error;
}

925
static int common_timer_del(struct k_itimer *timer)
L
Linus Torvalds 已提交
926
{
927
	timer->it.real.interval.tv64 = 0;
928

929
	if (hrtimer_try_to_cancel(&timer->it.real.timer) < 0)
L
Linus Torvalds 已提交
930 931 932 933 934 935
		return TIMER_RETRY;
	return 0;
}

static inline int timer_delete_hook(struct k_itimer *timer)
{
936 937 938 939 940
	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
941 942 943
}

/* Delete a POSIX.1b interval timer. */
944
SYSCALL_DEFINE1(timer_delete, timer_t, timer_id)
L
Linus Torvalds 已提交
945 946
{
	struct k_itimer *timer;
947
	unsigned long flags;
L
Linus Torvalds 已提交
948 949 950 951 952 953

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

954
	if (timer_delete_hook(timer) == TIMER_RETRY) {
L
Linus Torvalds 已提交
955 956 957
		unlock_timer(timer, flags);
		goto retry_delete;
	}
958

L
Linus Torvalds 已提交
959 960 961 962 963 964 965
	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).
	 */
966
	timer->it_signal = NULL;
967

L
Linus Torvalds 已提交
968 969 970 971
	unlock_timer(timer, flags);
	release_posix_timer(timer, IT_ID_SET);
	return 0;
}
972

L
Linus Torvalds 已提交
973 974 975
/*
 * return timer owned by the process, used by exit_itimers
 */
976
static void itimer_delete(struct k_itimer *timer)
L
Linus Torvalds 已提交
977 978 979 980 981 982
{
	unsigned long flags;

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

983
	if (timer_delete_hook(timer) == TIMER_RETRY) {
L
Linus Torvalds 已提交
984 985 986 987 988 989 990 991
		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).
	 */
992
	timer->it_signal = NULL;
993

L
Linus Torvalds 已提交
994 995 996 997 998
	unlock_timer(timer, flags);
	release_posix_timer(timer, IT_ID_SET);
}

/*
999
 * This is called by do_exit or de_thread, only when there are no more
L
Linus Torvalds 已提交
1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011
 * 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);
	}
}

1012 1013
SYSCALL_DEFINE2(clock_settime, const clockid_t, which_clock,
		const struct timespec __user *, tp)
L
Linus Torvalds 已提交
1014
{
1015
	struct k_clock *kc = clockid_to_kclock(which_clock);
L
Linus Torvalds 已提交
1016 1017
	struct timespec new_tp;

1018
	if (!kc || !kc->clock_set)
L
Linus Torvalds 已提交
1019
		return -EINVAL;
1020

L
Linus Torvalds 已提交
1021 1022 1023
	if (copy_from_user(&new_tp, tp, sizeof (*tp)))
		return -EFAULT;

1024
	return kc->clock_set(which_clock, &new_tp);
L
Linus Torvalds 已提交
1025 1026
}

1027 1028
SYSCALL_DEFINE2(clock_gettime, const clockid_t, which_clock,
		struct timespec __user *,tp)
L
Linus Torvalds 已提交
1029
{
1030
	struct k_clock *kc = clockid_to_kclock(which_clock);
L
Linus Torvalds 已提交
1031 1032 1033
	struct timespec kernel_tp;
	int error;

1034
	if (!kc)
L
Linus Torvalds 已提交
1035
		return -EINVAL;
1036 1037 1038

	error = kc->clock_get(which_clock, &kernel_tp);

L
Linus Torvalds 已提交
1039 1040 1041 1042 1043 1044
	if (!error && copy_to_user(tp, &kernel_tp, sizeof (kernel_tp)))
		error = -EFAULT;

	return error;
}

1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061
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);

1062
	if (err >= 0 && copy_to_user(utx, &ktx, sizeof(ktx)))
1063 1064 1065 1066 1067
		return -EFAULT;

	return err;
}

1068 1069
SYSCALL_DEFINE2(clock_getres, const clockid_t, which_clock,
		struct timespec __user *, tp)
L
Linus Torvalds 已提交
1070
{
1071
	struct k_clock *kc = clockid_to_kclock(which_clock);
L
Linus Torvalds 已提交
1072 1073 1074
	struct timespec rtn_tp;
	int error;

1075
	if (!kc)
L
Linus Torvalds 已提交
1076 1077
		return -EINVAL;

1078
	error = kc->clock_getres(which_clock, &rtn_tp);
L
Linus Torvalds 已提交
1079

1080
	if (!error && tp && copy_to_user(tp, &rtn_tp, sizeof (rtn_tp)))
L
Linus Torvalds 已提交
1081 1082 1083 1084 1085
		error = -EFAULT;

	return error;
}

1086 1087 1088 1089 1090 1091
/*
 * nanosleep for monotonic and realtime clocks
 */
static int common_nsleep(const clockid_t which_clock, int flags,
			 struct timespec *tsave, struct timespec __user *rmtp)
{
1092 1093 1094
	return hrtimer_nanosleep(tsave, rmtp, flags & TIMER_ABSTIME ?
				 HRTIMER_MODE_ABS : HRTIMER_MODE_REL,
				 which_clock);
1095
}
L
Linus Torvalds 已提交
1096

1097 1098 1099
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
1100
{
1101
	struct k_clock *kc = clockid_to_kclock(which_clock);
L
Linus Torvalds 已提交
1102 1103
	struct timespec t;

1104
	if (!kc)
L
Linus Torvalds 已提交
1105
		return -EINVAL;
1106 1107
	if (!kc->nsleep)
		return -ENANOSLEEP_NOTSUP;
L
Linus Torvalds 已提交
1108 1109 1110 1111

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

1112
	if (!timespec_valid(&t))
L
Linus Torvalds 已提交
1113 1114
		return -EINVAL;

1115
	return kc->nsleep(which_clock, flags, &t, rmtp);
L
Linus Torvalds 已提交
1116
}
1117 1118 1119 1120 1121

/*
 * This will restart clock_nanosleep. This is required only by
 * compat_clock_nanosleep_restart for now.
 */
1122
long clock_nanosleep_restart(struct restart_block *restart_block)
1123
{
1124
	clockid_t which_clock = restart_block->nanosleep.clockid;
1125 1126 1127 1128
	struct k_clock *kc = clockid_to_kclock(which_clock);

	if (WARN_ON_ONCE(!kc || !kc->nsleep_restart))
		return -EINVAL;
1129

1130
	return kc->nsleep_restart(restart_block);
1131
}