posix-cpu-timers.c 37.5 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7
/*
 * Implement CPU time clocks for the POSIX clock interface.
 */

#include <linux/sched.h>
#include <linux/posix-timers.h>
#include <linux/errno.h>
R
Roman Zippel 已提交
8
#include <linux/math64.h>
9
#include <linux/uaccess.h>
10
#include <linux/kernel_stat.h>
11
#include <trace/events/timer.h>
12 13
#include <linux/tick.h>
#include <linux/workqueue.h>
L
Linus Torvalds 已提交
14

15
/*
16 17 18 19
 * Called after updating RLIMIT_CPU to run cpu timer and update
 * tsk->signal->cputime_expires expiration cache if necessary. Needs
 * siglock protection since other code may update expiration cache as
 * well.
20
 */
21
void update_rlimit_cpu(struct task_struct *task, unsigned long rlim_new)
22
{
23
	u64 nsecs = rlim_new * NSEC_PER_SEC;
24

25
	spin_lock_irq(&task->sighand->siglock);
26
	set_process_cpu_timer(task, CPUCLOCK_PROF, &nsecs, NULL);
27
	spin_unlock_irq(&task->sighand->siglock);
28 29
}

30
static int check_clock(const clockid_t which_clock)
L
Linus Torvalds 已提交
31 32 33 34 35 36 37 38 39 40 41
{
	int error = 0;
	struct task_struct *p;
	const pid_t pid = CPUCLOCK_PID(which_clock);

	if (CPUCLOCK_WHICH(which_clock) >= CPUCLOCK_MAX)
		return -EINVAL;

	if (pid == 0)
		return 0;

42
	rcu_read_lock();
43
	p = find_task_by_vpid(pid);
44
	if (!p || !(CPUCLOCK_PERTHREAD(which_clock) ?
45
		   same_thread_group(p, current) : has_group_leader_pid(p))) {
L
Linus Torvalds 已提交
46 47
		error = -EINVAL;
	}
48
	rcu_read_unlock();
L
Linus Torvalds 已提交
49 50 51 52 53 54 55 56

	return error;
}

/*
 * Update expiry time from increment, and increase overrun count,
 * given the current clock sample.
 */
57
static void bump_cpu_timer(struct k_itimer *timer, u64 now)
L
Linus Torvalds 已提交
58 59
{
	int i;
60
	u64 delta, incr;
L
Linus Torvalds 已提交
61

62
	if (timer->it.cpu.incr == 0)
L
Linus Torvalds 已提交
63 64
		return;

65 66
	if (now < timer->it.cpu.expires)
		return;
L
Linus Torvalds 已提交
67

68 69
	incr = timer->it.cpu.incr;
	delta = now + incr - timer->it.cpu.expires;
L
Linus Torvalds 已提交
70

71 72 73 74 75 76 77 78 79 80 81
	/* Don't use (incr*2 < delta), incr*2 might overflow. */
	for (i = 0; incr < delta - incr; i++)
		incr = incr << 1;

	for (; i >= 0; incr >>= 1, i--) {
		if (delta < incr)
			continue;

		timer->it.cpu.expires += incr;
		timer->it_overrun += 1 << i;
		delta -= incr;
L
Linus Torvalds 已提交
82 83 84
	}
}

85 86 87 88 89 90 91 92
/**
 * task_cputime_zero - Check a task_cputime struct for all zero fields.
 *
 * @cputime:	The struct to compare.
 *
 * Checks @cputime to see if all fields are zero.  Returns true if all fields
 * are zero, false if any field is nonzero.
 */
93
static inline int task_cputime_zero(const struct task_cputime *cputime)
94 95 96 97 98 99
{
	if (!cputime->utime && !cputime->stime && !cputime->sum_exec_runtime)
		return 1;
	return 0;
}

100
static inline u64 prof_ticks(struct task_struct *p)
L
Linus Torvalds 已提交
101
{
102
	u64 utime, stime;
103

104
	task_cputime(p, &utime, &stime);
105

106
	return utime + stime;
L
Linus Torvalds 已提交
107
}
108
static inline u64 virt_ticks(struct task_struct *p)
L
Linus Torvalds 已提交
109
{
110
	u64 utime, stime;
111

112
	task_cputime(p, &utime, &stime);
113

114
	return utime;
L
Linus Torvalds 已提交
115 116
}

117 118
static int
posix_cpu_clock_getres(const clockid_t which_clock, struct timespec *tp)
L
Linus Torvalds 已提交
119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135
{
	int error = check_clock(which_clock);
	if (!error) {
		tp->tv_sec = 0;
		tp->tv_nsec = ((NSEC_PER_SEC + HZ - 1) / HZ);
		if (CPUCLOCK_WHICH(which_clock) == CPUCLOCK_SCHED) {
			/*
			 * If sched_clock is using a cycle counter, we
			 * don't have any idea of its true resolution
			 * exported, but it is much more than 1s/HZ.
			 */
			tp->tv_nsec = 1;
		}
	}
	return error;
}

136 137
static int
posix_cpu_clock_set(const clockid_t which_clock, const struct timespec *tp)
L
Linus Torvalds 已提交
138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153
{
	/*
	 * You can never reset a CPU clock, but we check for other errors
	 * in the call before failing with EPERM.
	 */
	int error = check_clock(which_clock);
	if (error == 0) {
		error = -EPERM;
	}
	return error;
}


/*
 * Sample a per-thread clock for the given task.
 */
154 155
static int cpu_clock_sample(const clockid_t which_clock,
			    struct task_struct *p, u64 *sample)
L
Linus Torvalds 已提交
156 157 158 159 160
{
	switch (CPUCLOCK_WHICH(which_clock)) {
	default:
		return -EINVAL;
	case CPUCLOCK_PROF:
161
		*sample = prof_ticks(p);
L
Linus Torvalds 已提交
162 163
		break;
	case CPUCLOCK_VIRT:
164
		*sample = virt_ticks(p);
L
Linus Torvalds 已提交
165 166
		break;
	case CPUCLOCK_SCHED:
167
		*sample = task_sched_runtime(p);
L
Linus Torvalds 已提交
168 169 170 171 172
		break;
	}
	return 0;
}

173 174 175 176 177
/*
 * Set cputime to sum_cputime if sum_cputime > cputime. Use cmpxchg
 * to avoid race conditions with concurrent updates to cputime.
 */
static inline void __update_gt_cputime(atomic64_t *cputime, u64 sum_cputime)
178
{
179 180 181 182 183 184 185 186
	u64 curr_cputime;
retry:
	curr_cputime = atomic64_read(cputime);
	if (sum_cputime > curr_cputime) {
		if (atomic64_cmpxchg(cputime, curr_cputime, sum_cputime) != curr_cputime)
			goto retry;
	}
}
187

188
static void update_gt_cputime(struct task_cputime_atomic *cputime_atomic, struct task_cputime *sum)
189
{
190 191 192
	__update_gt_cputime(&cputime_atomic->utime, sum->utime);
	__update_gt_cputime(&cputime_atomic->stime, sum->stime);
	__update_gt_cputime(&cputime_atomic->sum_exec_runtime, sum->sum_exec_runtime);
193
}
194

195
/* Sample task_cputime_atomic values in "atomic_timers", store results in "times". */
196
static inline void sample_cputime_atomic(struct task_cputime *times,
197
					 struct task_cputime_atomic *atomic_times)
198
{
199 200 201
	times->utime = atomic64_read(&atomic_times->utime);
	times->stime = atomic64_read(&atomic_times->stime);
	times->sum_exec_runtime = atomic64_read(&atomic_times->sum_exec_runtime);
202 203
}

204
void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times)
205 206
{
	struct thread_group_cputimer *cputimer = &tsk->signal->cputimer;
207
	struct task_cputime sum;
208

209 210
	/* Check if cputimer isn't running. This is accessed without locking. */
	if (!READ_ONCE(cputimer->running)) {
211 212 213
		/*
		 * The POSIX timer interface allows for absolute time expiry
		 * values through the TIMER_ABSTIME flag, therefore we have
214
		 * to synchronize the timer to the clock every time we start it.
215
		 */
216
		thread_group_cputime(tsk, &sum);
217
		update_gt_cputime(&cputimer->cputime_atomic, &sum);
218 219 220 221 222 223 224 225

		/*
		 * We're setting cputimer->running without a lock. Ensure
		 * this only gets written to in one operation. We set
		 * running after update_gt_cputime() as a small optimization,
		 * but barriers are not required because update_gt_cputime()
		 * can handle concurrent updates.
		 */
226
		WRITE_ONCE(cputimer->running, true);
227
	}
228
	sample_cputime_atomic(times, &cputimer->cputime_atomic);
229 230
}

L
Linus Torvalds 已提交
231 232
/*
 * Sample a process (thread group) clock for the given group_leader task.
233 234
 * Must be called with task sighand lock held for safe while_each_thread()
 * traversal.
L
Linus Torvalds 已提交
235
 */
236 237
static int cpu_clock_sample_group(const clockid_t which_clock,
				  struct task_struct *p,
238
				  u64 *sample)
L
Linus Torvalds 已提交
239
{
240
	struct task_cputime cputime;
241

242
	switch (CPUCLOCK_WHICH(which_clock)) {
L
Linus Torvalds 已提交
243 244 245
	default:
		return -EINVAL;
	case CPUCLOCK_PROF:
246 247
		thread_group_cputime(p, &cputime);
		*sample = cputime.utime + cputime.stime;
L
Linus Torvalds 已提交
248 249
		break;
	case CPUCLOCK_VIRT:
250 251
		thread_group_cputime(p, &cputime);
		*sample = cputime.utime;
L
Linus Torvalds 已提交
252 253
		break;
	case CPUCLOCK_SCHED:
254
		thread_group_cputime(p, &cputime);
255
		*sample = cputime.sum_exec_runtime;
L
Linus Torvalds 已提交
256 257 258 259 260
		break;
	}
	return 0;
}

261 262 263 264 265
static int posix_cpu_clock_get_task(struct task_struct *tsk,
				    const clockid_t which_clock,
				    struct timespec *tp)
{
	int err = -EINVAL;
266
	u64 rtn;
267 268 269 270 271

	if (CPUCLOCK_PERTHREAD(which_clock)) {
		if (same_thread_group(tsk, current))
			err = cpu_clock_sample(which_clock, tsk, &rtn);
	} else {
272
		if (tsk == current || thread_group_leader(tsk))
273 274 275 276
			err = cpu_clock_sample_group(which_clock, tsk, &rtn);
	}

	if (!err)
277
		*tp = ns_to_timespec(rtn);
278 279 280 281

	return err;
}

L
Linus Torvalds 已提交
282

283
static int posix_cpu_clock_get(const clockid_t which_clock, struct timespec *tp)
L
Linus Torvalds 已提交
284 285
{
	const pid_t pid = CPUCLOCK_PID(which_clock);
286
	int err = -EINVAL;
L
Linus Torvalds 已提交
287 288 289 290 291 292

	if (pid == 0) {
		/*
		 * Special case constant value for our own clocks.
		 * We don't have to do any lookup to find ourselves.
		 */
293
		err = posix_cpu_clock_get_task(current, which_clock, tp);
L
Linus Torvalds 已提交
294 295 296 297 298 299
	} else {
		/*
		 * Find the given PID, and validate that the caller
		 * should be able to see it.
		 */
		struct task_struct *p;
300
		rcu_read_lock();
301
		p = find_task_by_vpid(pid);
302 303
		if (p)
			err = posix_cpu_clock_get_task(p, which_clock, tp);
304
		rcu_read_unlock();
L
Linus Torvalds 已提交
305 306
	}

307
	return err;
L
Linus Torvalds 已提交
308 309 310 311
}

/*
 * Validate the clockid_t for a new CPU-clock timer, and initialize the timer.
312 313
 * This is called from sys_timer_create() and do_cpu_nanosleep() with the
 * new timer already all-zeros initialized.
L
Linus Torvalds 已提交
314
 */
315
static int posix_cpu_timer_create(struct k_itimer *new_timer)
L
Linus Torvalds 已提交
316 317 318 319 320 321 322 323 324 325
{
	int ret = 0;
	const pid_t pid = CPUCLOCK_PID(new_timer->it_clock);
	struct task_struct *p;

	if (CPUCLOCK_WHICH(new_timer->it_clock) >= CPUCLOCK_MAX)
		return -EINVAL;

	INIT_LIST_HEAD(&new_timer->it.cpu.entry);

326
	rcu_read_lock();
L
Linus Torvalds 已提交
327 328 329 330
	if (CPUCLOCK_PERTHREAD(new_timer->it_clock)) {
		if (pid == 0) {
			p = current;
		} else {
331
			p = find_task_by_vpid(pid);
332
			if (p && !same_thread_group(p, current))
L
Linus Torvalds 已提交
333 334 335 336 337 338
				p = NULL;
		}
	} else {
		if (pid == 0) {
			p = current->group_leader;
		} else {
339
			p = find_task_by_vpid(pid);
340
			if (p && !has_group_leader_pid(p))
L
Linus Torvalds 已提交
341 342 343 344 345 346 347 348 349
				p = NULL;
		}
	}
	new_timer->it.cpu.task = p;
	if (p) {
		get_task_struct(p);
	} else {
		ret = -EINVAL;
	}
350
	rcu_read_unlock();
L
Linus Torvalds 已提交
351 352 353 354 355 356 357 358 359 360

	return ret;
}

/*
 * Clean up a CPU-clock timer that is about to be destroyed.
 * This is called from timer deletion with the timer already locked.
 * If we return TIMER_RETRY, it's necessary to release the timer's lock
 * and try again.  (This happens when the timer is in the middle of firing.)
 */
361
static int posix_cpu_timer_del(struct k_itimer *timer)
L
Linus Torvalds 已提交
362
{
363
	int ret = 0;
364 365 366
	unsigned long flags;
	struct sighand_struct *sighand;
	struct task_struct *p = timer->it.cpu.task;
L
Linus Torvalds 已提交
367

368
	WARN_ON_ONCE(p == NULL);
369

370 371 372 373 374 375
	/*
	 * Protect against sighand release/switch in exit/exec and process/
	 * thread timer list entry concurrent read/writes.
	 */
	sighand = lock_task_sighand(p, &flags);
	if (unlikely(sighand == NULL)) {
376 377 378 379
		/*
		 * We raced with the reaping of the task.
		 * The deletion should have cleared us off the list.
		 */
380
		WARN_ON_ONCE(!list_empty(&timer->it.cpu.entry));
381 382 383 384 385
	} else {
		if (timer->it.cpu.firing)
			ret = TIMER_RETRY;
		else
			list_del(&timer->it.cpu.entry);
386 387

		unlock_task_sighand(p, &flags);
L
Linus Torvalds 已提交
388
	}
389 390 391

	if (!ret)
		put_task_struct(p);
L
Linus Torvalds 已提交
392

393
	return ret;
L
Linus Torvalds 已提交
394 395
}

396
static void cleanup_timers_list(struct list_head *head)
397 398 399
{
	struct cpu_timer_list *timer, *next;

400
	list_for_each_entry_safe(timer, next, head, entry)
401 402 403
		list_del_init(&timer->entry);
}

L
Linus Torvalds 已提交
404 405 406 407 408 409
/*
 * Clean out CPU timers still ticking when a thread exited.  The task
 * pointer is cleared, and the expiry time is replaced with the residual
 * time for later timer_gettime calls to return.
 * This must be called with the siglock held.
 */
410
static void cleanup_timers(struct list_head *head)
L
Linus Torvalds 已提交
411
{
412 413 414
	cleanup_timers_list(head);
	cleanup_timers_list(++head);
	cleanup_timers_list(++head);
L
Linus Torvalds 已提交
415 416 417 418 419 420 421 422 423
}

/*
 * These are both called with the siglock held, when the current thread
 * is being reaped.  When the final (leader) thread in the group is reaped,
 * posix_cpu_timers_exit_group will be called after posix_cpu_timers_exit.
 */
void posix_cpu_timers_exit(struct task_struct *tsk)
{
424
	cleanup_timers(tsk->cpu_timers);
L
Linus Torvalds 已提交
425 426 427
}
void posix_cpu_timers_exit_group(struct task_struct *tsk)
{
428
	cleanup_timers(tsk->signal->cpu_timers);
L
Linus Torvalds 已提交
429 430
}

431
static inline int expires_gt(u64 expires, u64 new_exp)
432
{
433
	return expires == 0 || expires > new_exp;
434 435
}

L
Linus Torvalds 已提交
436 437
/*
 * Insert the timer on the appropriate list before any timers that
438
 * expire later.  This must be called with the sighand lock held.
L
Linus Torvalds 已提交
439
 */
440
static void arm_timer(struct k_itimer *timer)
L
Linus Torvalds 已提交
441 442 443
{
	struct task_struct *p = timer->it.cpu.task;
	struct list_head *head, *listpos;
444
	struct task_cputime *cputime_expires;
L
Linus Torvalds 已提交
445 446 447
	struct cpu_timer_list *const nt = &timer->it.cpu;
	struct cpu_timer_list *next;

448 449 450 451 452 453 454
	if (CPUCLOCK_PERTHREAD(timer->it_clock)) {
		head = p->cpu_timers;
		cputime_expires = &p->cputime_expires;
	} else {
		head = p->signal->cpu_timers;
		cputime_expires = &p->signal->cputime_expires;
	}
L
Linus Torvalds 已提交
455 456 457
	head += CPUCLOCK_WHICH(timer->it_clock);

	listpos = head;
458
	list_for_each_entry(next, head, entry) {
459
		if (nt->expires < next->expires)
460 461
			break;
		listpos = &next->entry;
L
Linus Torvalds 已提交
462 463 464 465
	}
	list_add(&nt->entry, listpos);

	if (listpos == head) {
466
		u64 exp = nt->expires;
467

L
Linus Torvalds 已提交
468
		/*
469 470 471 472
		 * We are the new earliest-expiring POSIX 1.b timer, hence
		 * need to update expiration cache. Take into account that
		 * for process timers we share expiration cache with itimers
		 * and RLIMIT_CPU and for thread timers with RLIMIT_RTTIME.
L
Linus Torvalds 已提交
473 474
		 */

475 476
		switch (CPUCLOCK_WHICH(timer->it_clock)) {
		case CPUCLOCK_PROF:
477 478
			if (expires_gt(cputime_expires->prof_exp, exp))
				cputime_expires->prof_exp = exp;
479 480
			break;
		case CPUCLOCK_VIRT:
481 482
			if (expires_gt(cputime_expires->virt_exp, exp))
				cputime_expires->virt_exp = exp;
483 484
			break;
		case CPUCLOCK_SCHED:
485
			if (expires_gt(cputime_expires->sched_exp, exp))
486
				cputime_expires->sched_exp = exp;
487
			break;
L
Linus Torvalds 已提交
488
		}
489 490 491 492
		if (CPUCLOCK_PERTHREAD(timer->it_clock))
			tick_dep_set_task(p, TICK_DEP_BIT_POSIX_TIMER);
		else
			tick_dep_set_signal(p->signal, TICK_DEP_BIT_POSIX_TIMER);
L
Linus Torvalds 已提交
493 494 495 496 497 498 499 500
	}
}

/*
 * The timer is locked, fire it and arrange for its reload.
 */
static void cpu_timer_fire(struct k_itimer *timer)
{
501 502 503 504
	if ((timer->it_sigev_notify & ~SIGEV_THREAD_ID) == SIGEV_NONE) {
		/*
		 * User don't want any signal.
		 */
505
		timer->it.cpu.expires = 0;
506
	} else if (unlikely(timer->sigq == NULL)) {
L
Linus Torvalds 已提交
507 508 509 510 511
		/*
		 * This a special case for clock_nanosleep,
		 * not a normal timer from sys_timer_create.
		 */
		wake_up_process(timer->it_process);
512 513
		timer->it.cpu.expires = 0;
	} else if (timer->it.cpu.incr == 0) {
L
Linus Torvalds 已提交
514 515 516 517
		/*
		 * One-shot timer.  Clear it as soon as it's fired.
		 */
		posix_timer_event(timer, 0);
518
		timer->it.cpu.expires = 0;
L
Linus Torvalds 已提交
519 520 521 522 523 524 525 526 527 528 529
	} else if (posix_timer_event(timer, ++timer->it_requeue_pending)) {
		/*
		 * The signal did not get queued because the signal
		 * was ignored, so we won't get any callback to
		 * reload the timer.  But we need to keep it
		 * ticking in case the signal is deliverable next time.
		 */
		posix_cpu_timer_schedule(timer);
	}
}

530 531
/*
 * Sample a process (thread group) timer for the given group_leader task.
532 533
 * Must be called with task sighand lock held for safe while_each_thread()
 * traversal.
534 535
 */
static int cpu_timer_sample_group(const clockid_t which_clock,
536
				  struct task_struct *p, u64 *sample)
537
{
538
	struct task_cputime cputime;
539 540 541 542 543 544

	thread_group_cputimer(p, &cputime);
	switch (CPUCLOCK_WHICH(which_clock)) {
	default:
		return -EINVAL;
	case CPUCLOCK_PROF:
545
		*sample = cputime.utime + cputime.stime;
546 547
		break;
	case CPUCLOCK_VIRT:
548
		*sample = cputime.utime;
549 550
		break;
	case CPUCLOCK_SCHED:
551
		*sample = cputime.sum_exec_runtime;
552 553 554 555 556
		break;
	}
	return 0;
}

L
Linus Torvalds 已提交
557 558 559 560 561 562
/*
 * Guts of sys_timer_settime for CPU timers.
 * This is called with the timer locked and interrupts disabled.
 * If we return TIMER_RETRY, it's necessary to release the timer's lock
 * and try again.  (This happens when the timer is in the middle of firing.)
 */
563
static int posix_cpu_timer_set(struct k_itimer *timer, int timer_flags,
564
			       struct itimerspec *new, struct itimerspec *old)
L
Linus Torvalds 已提交
565
{
566 567
	unsigned long flags;
	struct sighand_struct *sighand;
L
Linus Torvalds 已提交
568
	struct task_struct *p = timer->it.cpu.task;
569
	u64 old_expires, new_expires, old_incr, val;
L
Linus Torvalds 已提交
570 571
	int ret;

572
	WARN_ON_ONCE(p == NULL);
L
Linus Torvalds 已提交
573

574
	new_expires = timespec_to_ns(&new->it_value);
L
Linus Torvalds 已提交
575 576

	/*
577 578 579 580 581 582
	 * Protect against sighand release/switch in exit/exec and p->cpu_timers
	 * and p->signal->cpu_timers read/write in arm_timer()
	 */
	sighand = lock_task_sighand(p, &flags);
	/*
	 * If p has just been reaped, we can no
L
Linus Torvalds 已提交
583 584
	 * longer get any information about it at all.
	 */
585
	if (unlikely(sighand == NULL)) {
L
Linus Torvalds 已提交
586 587 588 589 590 591
		return -ESRCH;
	}

	/*
	 * Disarm any old timer after extracting its expiry time.
	 */
592
	WARN_ON_ONCE(!irqs_disabled());
593 594

	ret = 0;
595
	old_incr = timer->it.cpu.incr;
L
Linus Torvalds 已提交
596
	old_expires = timer->it.cpu.expires;
597 598 599 600 601
	if (unlikely(timer->it.cpu.firing)) {
		timer->it.cpu.firing = -1;
		ret = TIMER_RETRY;
	} else
		list_del_init(&timer->it.cpu.entry);
L
Linus Torvalds 已提交
602 603 604 605 606 607 608 609 610 611 612 613

	/*
	 * We need to sample the current value to convert the new
	 * value from to relative and absolute, and to convert the
	 * old value from absolute to relative.  To set a process
	 * timer, we need a sample to balance the thread expiry
	 * times (in arm_timer).  With an absolute time, we must
	 * check if it's already passed.  In short, we need a sample.
	 */
	if (CPUCLOCK_PERTHREAD(timer->it_clock)) {
		cpu_clock_sample(timer->it_clock, p, &val);
	} else {
614
		cpu_timer_sample_group(timer->it_clock, p, &val);
L
Linus Torvalds 已提交
615 616 617
	}

	if (old) {
618
		if (old_expires == 0) {
L
Linus Torvalds 已提交
619 620 621 622 623 624 625 626 627 628 629 630 631 632
			old->it_value.tv_sec = 0;
			old->it_value.tv_nsec = 0;
		} else {
			/*
			 * Update the timer in case it has
			 * overrun already.  If it has,
			 * we'll report it as having overrun
			 * and with the next reloaded timer
			 * already ticking, though we are
			 * swallowing that pending
			 * notification here to install the
			 * new setting.
			 */
			bump_cpu_timer(timer, val);
633 634
			if (val < timer->it.cpu.expires) {
				old_expires = timer->it.cpu.expires - val;
635
				old->it_value = ns_to_timespec(old_expires);
L
Linus Torvalds 已提交
636 637 638 639 640 641 642
			} else {
				old->it_value.tv_nsec = 1;
				old->it_value.tv_sec = 0;
			}
		}
	}

643
	if (unlikely(ret)) {
L
Linus Torvalds 已提交
644 645 646 647 648 649
		/*
		 * We are colliding with the timer actually firing.
		 * Punt after filling in the timer's old value, and
		 * disable this firing since we are already reporting
		 * it as an overrun (thanks to bump_cpu_timer above).
		 */
650
		unlock_task_sighand(p, &flags);
L
Linus Torvalds 已提交
651 652 653
		goto out;
	}

654
	if (new_expires != 0 && !(timer_flags & TIMER_ABSTIME)) {
655
		new_expires += val;
L
Linus Torvalds 已提交
656 657 658 659 660 661 662 663
	}

	/*
	 * Install the new expiry time (or zero).
	 * For a timer with no notification action, we don't actually
	 * arm the timer (we'll just fake it for timer_gettime).
	 */
	timer->it.cpu.expires = new_expires;
664
	if (new_expires != 0 && val < new_expires) {
665
		arm_timer(timer);
L
Linus Torvalds 已提交
666 667
	}

668
	unlock_task_sighand(p, &flags);
L
Linus Torvalds 已提交
669 670 671 672
	/*
	 * Install the new reload setting, and
	 * set up the signal and overrun bookkeeping.
	 */
673
	timer->it.cpu.incr = timespec_to_ns(&new->it_interval);
L
Linus Torvalds 已提交
674 675 676 677 678 679 680 681 682 683 684

	/*
	 * This acts as a modification timestamp for the timer,
	 * so any automatic reload attempt will punt on seeing
	 * that we have reset the timer manually.
	 */
	timer->it_requeue_pending = (timer->it_requeue_pending + 2) &
		~REQUEUE_PENDING;
	timer->it_overrun_last = 0;
	timer->it_overrun = -1;

685
	if (new_expires != 0 && !(val < new_expires)) {
L
Linus Torvalds 已提交
686 687 688 689 690 691 692 693 694 695
		/*
		 * The designated time already passed, so we notify
		 * immediately, even if the thread never runs to
		 * accumulate more time on this clock.
		 */
		cpu_timer_fire(timer);
	}

	ret = 0;
 out:
696 697
	if (old)
		old->it_interval = ns_to_timespec(old_incr);
698

L
Linus Torvalds 已提交
699 700 701
	return ret;
}

702
static void posix_cpu_timer_get(struct k_itimer *timer, struct itimerspec *itp)
L
Linus Torvalds 已提交
703
{
704
	u64 now;
L
Linus Torvalds 已提交
705 706
	struct task_struct *p = timer->it.cpu.task;

707 708
	WARN_ON_ONCE(p == NULL);

L
Linus Torvalds 已提交
709 710 711
	/*
	 * Easy part: convert the reload time.
	 */
712
	itp->it_interval = ns_to_timespec(timer->it.cpu.incr);
L
Linus Torvalds 已提交
713

714
	if (timer->it.cpu.expires == 0) {	/* Timer not armed at all.  */
L
Linus Torvalds 已提交
715 716 717 718 719 720 721 722 723 724
		itp->it_value.tv_sec = itp->it_value.tv_nsec = 0;
		return;
	}

	/*
	 * Sample the clock to take the difference with the expiry time.
	 */
	if (CPUCLOCK_PERTHREAD(timer->it_clock)) {
		cpu_clock_sample(timer->it_clock, p, &now);
	} else {
725 726 727 728 729 730
		struct sighand_struct *sighand;
		unsigned long flags;

		/*
		 * Protect against sighand release/switch in exit/exec and
		 * also make timer sampling safe if it ends up calling
731
		 * thread_group_cputime().
732 733 734
		 */
		sighand = lock_task_sighand(p, &flags);
		if (unlikely(sighand == NULL)) {
L
Linus Torvalds 已提交
735 736 737 738 739
			/*
			 * The process has been reaped.
			 * We can't even collect a sample any more.
			 * Call the timer disarmed, nothing else to do.
			 */
740
			timer->it.cpu.expires = 0;
741
			itp->it_value = ns_to_timespec(timer->it.cpu.expires);
742
			return;
L
Linus Torvalds 已提交
743
		} else {
744
			cpu_timer_sample_group(timer->it_clock, p, &now);
745
			unlock_task_sighand(p, &flags);
L
Linus Torvalds 已提交
746 747 748
		}
	}

749
	if (now < timer->it.cpu.expires) {
750
		itp->it_value = ns_to_timespec(timer->it.cpu.expires - now);
L
Linus Torvalds 已提交
751 752 753 754 755 756 757 758 759 760
	} else {
		/*
		 * The timer should have expired already, but the firing
		 * hasn't taken place yet.  Say it's just about to expire.
		 */
		itp->it_value.tv_nsec = 1;
		itp->it_value.tv_sec = 0;
	}
}

761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782
static unsigned long long
check_timers_list(struct list_head *timers,
		  struct list_head *firing,
		  unsigned long long curr)
{
	int maxfire = 20;

	while (!list_empty(timers)) {
		struct cpu_timer_list *t;

		t = list_first_entry(timers, struct cpu_timer_list, entry);

		if (!--maxfire || curr < t->expires)
			return t->expires;

		t->firing = 1;
		list_move_tail(&t->entry, firing);
	}

	return 0;
}

L
Linus Torvalds 已提交
783 784 785 786 787 788 789 790 791
/*
 * Check for any per-thread CPU timers that have fired and move them off
 * the tsk->cpu_timers[N] list onto the firing list.  Here we update the
 * tsk->it_*_expires values to reflect the remaining thread CPU timers.
 */
static void check_thread_timers(struct task_struct *tsk,
				struct list_head *firing)
{
	struct list_head *timers = tsk->cpu_timers;
792
	struct signal_struct *const sig = tsk->signal;
793 794
	struct task_cputime *tsk_expires = &tsk->cputime_expires;
	u64 expires;
795
	unsigned long soft;
L
Linus Torvalds 已提交
796

797 798 799 800 801 802 803
	/*
	 * If cputime_expires is zero, then there are no active
	 * per thread CPU timers.
	 */
	if (task_cputime_zero(&tsk->cputime_expires))
		return;

804
	expires = check_timers_list(timers, firing, prof_ticks(tsk));
805
	tsk_expires->prof_exp = expires;
L
Linus Torvalds 已提交
806

807
	expires = check_timers_list(++timers, firing, virt_ticks(tsk));
808
	tsk_expires->virt_exp = expires;
L
Linus Torvalds 已提交
809

810 811
	tsk_expires->sched_exp = check_timers_list(++timers, firing,
						   tsk->se.sum_exec_runtime);
812 813 814 815

	/*
	 * Check for the special case thread timers.
	 */
816
	soft = READ_ONCE(sig->rlim[RLIMIT_RTTIME].rlim_cur);
817
	if (soft != RLIM_INFINITY) {
818
		unsigned long hard =
819
			READ_ONCE(sig->rlim[RLIMIT_RTTIME].rlim_max);
820

821 822
		if (hard != RLIM_INFINITY &&
		    tsk->rt.timeout > DIV_ROUND_UP(hard, USEC_PER_SEC/HZ)) {
823 824 825 826 827 828 829
			/*
			 * At the hard limit, we just die.
			 * No need to calculate anything else now.
			 */
			__group_send_sig_info(SIGKILL, SEND_SIG_PRIV, tsk);
			return;
		}
830
		if (tsk->rt.timeout > DIV_ROUND_UP(soft, USEC_PER_SEC/HZ)) {
831 832 833
			/*
			 * At the soft limit, send a SIGXCPU every second.
			 */
834 835 836
			if (soft < hard) {
				soft += USEC_PER_SEC;
				sig->rlim[RLIMIT_RTTIME].rlim_cur = soft;
837
			}
838 839 840
			printk(KERN_INFO
				"RT Watchdog Timeout: %s[%d]\n",
				tsk->comm, task_pid_nr(tsk));
841 842 843
			__group_send_sig_info(SIGXCPU, SEND_SIG_PRIV, tsk);
		}
	}
844 845
	if (task_cputime_zero(tsk_expires))
		tick_dep_clear_task(tsk, TICK_DEP_BIT_POSIX_TIMER);
L
Linus Torvalds 已提交
846 847
}

848
static inline void stop_process_timers(struct signal_struct *sig)
849
{
850
	struct thread_group_cputimer *cputimer = &sig->cputimer;
851

852
	/* Turn off cputimer->running. This is done without locking. */
853
	WRITE_ONCE(cputimer->running, false);
854
	tick_dep_clear_signal(sig, TICK_DEP_BIT_POSIX_TIMER);
855 856
}

857
static void check_cpu_itimer(struct task_struct *tsk, struct cpu_itimer *it,
858
			     u64 *expires, u64 cur_time, int signo)
859
{
860
	if (!it->expires)
861 862
		return;

863 864
	if (cur_time >= it->expires) {
		if (it->incr)
865
			it->expires += it->incr;
866
		else
867
			it->expires = 0;
868

869 870 871
		trace_itimer_expire(signo == SIGPROF ?
				    ITIMER_PROF : ITIMER_VIRTUAL,
				    tsk->signal->leader_pid, cur_time);
872 873 874
		__group_send_sig_info(signo, SEND_SIG_PRIV, tsk);
	}

875 876
	if (it->expires && (!*expires || it->expires < *expires))
		*expires = it->expires;
877 878
}

L
Linus Torvalds 已提交
879 880 881 882 883 884 885 886 887
/*
 * Check for any per-thread CPU timers that have fired and move them
 * off the tsk->*_timers list onto the firing list.  Per-thread timers
 * have already been taken off.
 */
static void check_process_timers(struct task_struct *tsk,
				 struct list_head *firing)
{
	struct signal_struct *const sig = tsk->signal;
888 889
	u64 utime, ptime, virt_expires, prof_expires;
	u64 sum_sched_runtime, sched_expires;
L
Linus Torvalds 已提交
890
	struct list_head *timers = sig->cpu_timers;
891
	struct task_cputime cputime;
892
	unsigned long soft;
L
Linus Torvalds 已提交
893

894 895 896 897 898 899 900
	/*
	 * If cputimer is not running, then there are no active
	 * process wide timers (POSIX 1.b, itimers, RLIMIT_CPU).
	 */
	if (!READ_ONCE(tsk->signal->cputimer.running))
		return;

901 902 903 904 905 906
        /*
	 * Signify that a thread is checking for process timers.
	 * Write access to this field is protected by the sighand lock.
	 */
	sig->cputimer.checking_timer = true;

L
Linus Torvalds 已提交
907 908 909
	/*
	 * Collect the current process totals.
	 */
910
	thread_group_cputimer(tsk, &cputime);
911 912
	utime = cputime.utime;
	ptime = utime + cputime.stime;
913
	sum_sched_runtime = cputime.sum_exec_runtime;
L
Linus Torvalds 已提交
914

915 916 917
	prof_expires = check_timers_list(timers, firing, ptime);
	virt_expires = check_timers_list(++timers, firing, utime);
	sched_expires = check_timers_list(++timers, firing, sum_sched_runtime);
L
Linus Torvalds 已提交
918 919 920 921

	/*
	 * Check for the special case process timers.
	 */
922 923 924 925
	check_cpu_itimer(tsk, &sig->it[CPUCLOCK_PROF], &prof_expires, ptime,
			 SIGPROF);
	check_cpu_itimer(tsk, &sig->it[CPUCLOCK_VIRT], &virt_expires, utime,
			 SIGVTALRM);
926
	soft = READ_ONCE(sig->rlim[RLIMIT_CPU].rlim_cur);
927
	if (soft != RLIM_INFINITY) {
928
		unsigned long psecs = div_u64(ptime, NSEC_PER_SEC);
929
		unsigned long hard =
930
			READ_ONCE(sig->rlim[RLIMIT_CPU].rlim_max);
931
		u64 x;
932
		if (psecs >= hard) {
L
Linus Torvalds 已提交
933 934 935 936 937 938 939
			/*
			 * At the hard limit, we just die.
			 * No need to calculate anything else now.
			 */
			__group_send_sig_info(SIGKILL, SEND_SIG_PRIV, tsk);
			return;
		}
940
		if (psecs >= soft) {
L
Linus Torvalds 已提交
941 942 943 944
			/*
			 * At the soft limit, send a SIGXCPU every second.
			 */
			__group_send_sig_info(SIGXCPU, SEND_SIG_PRIV, tsk);
945 946 947
			if (soft < hard) {
				soft++;
				sig->rlim[RLIMIT_CPU].rlim_cur = soft;
L
Linus Torvalds 已提交
948 949
			}
		}
950 951
		x = soft * NSEC_PER_SEC;
		if (!prof_expires || x < prof_expires)
L
Linus Torvalds 已提交
952 953 954
			prof_expires = x;
	}

955 956
	sig->cputime_expires.prof_exp = prof_expires;
	sig->cputime_expires.virt_exp = virt_expires;
957 958 959
	sig->cputime_expires.sched_exp = sched_expires;
	if (task_cputime_zero(&sig->cputime_expires))
		stop_process_timers(sig);
960 961

	sig->cputimer.checking_timer = false;
L
Linus Torvalds 已提交
962 963 964 965 966 967 968 969
}

/*
 * This is called from the signal code (via do_schedule_next_timer)
 * when the last timer signal was delivered and we have to reload the timer.
 */
void posix_cpu_timer_schedule(struct k_itimer *timer)
{
970 971
	struct sighand_struct *sighand;
	unsigned long flags;
L
Linus Torvalds 已提交
972
	struct task_struct *p = timer->it.cpu.task;
973
	u64 now;
L
Linus Torvalds 已提交
974

975
	WARN_ON_ONCE(p == NULL);
L
Linus Torvalds 已提交
976 977 978 979 980 981 982

	/*
	 * Fetch the current sample and update the timer's expiry time.
	 */
	if (CPUCLOCK_PERTHREAD(timer->it_clock)) {
		cpu_clock_sample(timer->it_clock, p, &now);
		bump_cpu_timer(timer, now);
983
		if (unlikely(p->exit_state))
984
			goto out;
985

986 987 988 989
		/* Protect timer list r/w in arm_timer() */
		sighand = lock_task_sighand(p, &flags);
		if (!sighand)
			goto out;
L
Linus Torvalds 已提交
990
	} else {
991 992
		/*
		 * Protect arm_timer() and timer sampling in case of call to
993
		 * thread_group_cputime().
994 995 996
		 */
		sighand = lock_task_sighand(p, &flags);
		if (unlikely(sighand == NULL)) {
L
Linus Torvalds 已提交
997 998 999 1000
			/*
			 * The process has been reaped.
			 * We can't even collect a sample any more.
			 */
1001
			timer->it.cpu.expires = 0;
1002
			goto out;
L
Linus Torvalds 已提交
1003
		} else if (unlikely(p->exit_state) && thread_group_empty(p)) {
1004
			unlock_task_sighand(p, &flags);
1005
			/* Optimizations: if the process is dying, no need to rearm */
1006
			goto out;
L
Linus Torvalds 已提交
1007
		}
1008
		cpu_timer_sample_group(timer->it_clock, p, &now);
L
Linus Torvalds 已提交
1009
		bump_cpu_timer(timer, now);
1010
		/* Leave the sighand locked for the call below.  */
L
Linus Torvalds 已提交
1011 1012 1013 1014 1015
	}

	/*
	 * Now re-arm for the new expiry time.
	 */
1016
	WARN_ON_ONCE(!irqs_disabled());
1017
	arm_timer(timer);
1018
	unlock_task_sighand(p, &flags);
1019 1020 1021 1022 1023

out:
	timer->it_overrun_last = timer->it_overrun;
	timer->it_overrun = -1;
	++timer->it_requeue_pending;
L
Linus Torvalds 已提交
1024 1025
}

1026 1027 1028 1029 1030 1031 1032 1033 1034 1035
/**
 * task_cputime_expired - Compare two task_cputime entities.
 *
 * @sample:	The task_cputime structure to be checked for expiration.
 * @expires:	Expiration times, against which @sample will be checked.
 *
 * Checks @sample against @expires to see if any field of @sample has expired.
 * Returns true if any field of the former is greater than the corresponding
 * field of the latter if the latter field is set.  Otherwise returns false.
 */
1036 1037
static inline int task_cputime_expired(const struct task_cputime *sample,
					const struct task_cputime *expires)
1038
{
1039
	if (expires->utime && sample->utime >= expires->utime)
1040
		return 1;
1041
	if (expires->stime && sample->utime + sample->stime >= expires->stime)
1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053
		return 1;
	if (expires->sum_exec_runtime != 0 &&
	    sample->sum_exec_runtime >= expires->sum_exec_runtime)
		return 1;
	return 0;
}

/**
 * fastpath_timer_check - POSIX CPU timers fast path.
 *
 * @tsk:	The task (thread) being checked.
 *
1054 1055 1056 1057
 * Check the task and thread group timers.  If both are zero (there are no
 * timers set) return false.  Otherwise snapshot the task and thread group
 * timers and compare them with the corresponding expiration times.  Return
 * true if a timer has expired, else return false.
1058
 */
1059
static inline int fastpath_timer_check(struct task_struct *tsk)
1060
{
1061
	struct signal_struct *sig;
1062 1063

	if (!task_cputime_zero(&tsk->cputime_expires)) {
1064
		struct task_cputime task_sample;
1065

1066
		task_cputime(tsk, &task_sample.utime, &task_sample.stime);
1067
		task_sample.sum_exec_runtime = tsk->se.sum_exec_runtime;
1068 1069 1070
		if (task_cputime_expired(&task_sample, &tsk->cputime_expires))
			return 1;
	}
1071 1072

	sig = tsk->signal;
1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088
	/*
	 * Check if thread group timers expired when the cputimer is
	 * running and no other thread in the group is already checking
	 * for thread group cputimers. These fields are read without the
	 * sighand lock. However, this is fine because this is meant to
	 * be a fastpath heuristic to determine whether we should try to
	 * acquire the sighand lock to check/handle timers.
	 *
	 * In the worst case scenario, if 'running' or 'checking_timer' gets
	 * set but the current thread doesn't see the change yet, we'll wait
	 * until the next thread in the group gets a scheduler interrupt to
	 * handle the timer. This isn't an issue in practice because these
	 * types of delays with signals actually getting sent are expected.
	 */
	if (READ_ONCE(sig->cputimer.running) &&
	    !READ_ONCE(sig->cputimer.checking_timer)) {
1089
		struct task_cputime group_sample;
1090

1091
		sample_cputime_atomic(&group_sample, &sig->cputimer.cputime_atomic);
1092

1093 1094 1095
		if (task_cputime_expired(&group_sample, &sig->cputime_expires))
			return 1;
	}
1096

1097
	return 0;
1098 1099
}

L
Linus Torvalds 已提交
1100 1101 1102 1103 1104 1105 1106 1107 1108
/*
 * This is called from the timer interrupt handler.  The irq handler has
 * already updated our counts.  We need to check if any timers fire now.
 * Interrupts are disabled.
 */
void run_posix_cpu_timers(struct task_struct *tsk)
{
	LIST_HEAD(firing);
	struct k_itimer *timer, *next;
1109
	unsigned long flags;
L
Linus Torvalds 已提交
1110

1111
	WARN_ON_ONCE(!irqs_disabled());
L
Linus Torvalds 已提交
1112 1113

	/*
1114
	 * The fast path checks that there are no expired thread or thread
1115
	 * group timers.  If that's so, just return.
L
Linus Torvalds 已提交
1116
	 */
1117
	if (!fastpath_timer_check(tsk))
1118
		return;
1119

1120 1121
	if (!lock_task_sighand(tsk, &flags))
		return;
1122 1123 1124 1125 1126 1127
	/*
	 * Here we take off tsk->signal->cpu_timers[N] and
	 * tsk->cpu_timers[N] all the timers that are firing, and
	 * put them on the firing list.
	 */
	check_thread_timers(tsk, &firing);
1128 1129

	check_process_timers(tsk, &firing);
L
Linus Torvalds 已提交
1130

1131 1132 1133 1134 1135 1136 1137 1138
	/*
	 * We must release these locks before taking any timer's lock.
	 * There is a potential race with timer deletion here, as the
	 * siglock now protects our private firing list.  We have set
	 * the firing flag in each timer, so that a deletion attempt
	 * that gets the timer lock before we do will give it up and
	 * spin until we've taken care of that timer below.
	 */
1139
	unlock_task_sighand(tsk, &flags);
L
Linus Torvalds 已提交
1140 1141 1142

	/*
	 * Now that all the timers on our list have the firing flag,
L
Lucas De Marchi 已提交
1143
	 * no one will touch their list entries but us.  We'll take
L
Linus Torvalds 已提交
1144 1145 1146 1147
	 * each timer's lock before clearing its firing flag, so no
	 * timer call will interfere.
	 */
	list_for_each_entry_safe(timer, next, &firing, it.cpu.entry) {
1148 1149
		int cpu_firing;

L
Linus Torvalds 已提交
1150 1151
		spin_lock(&timer->it_lock);
		list_del_init(&timer->it.cpu.entry);
1152
		cpu_firing = timer->it.cpu.firing;
L
Linus Torvalds 已提交
1153 1154 1155 1156 1157 1158
		timer->it.cpu.firing = 0;
		/*
		 * The firing flag is -1 if we collided with a reset
		 * of the timer, which already reported this
		 * almost-firing as an overrun.  So don't generate an event.
		 */
1159
		if (likely(cpu_firing >= 0))
L
Linus Torvalds 已提交
1160 1161 1162 1163 1164 1165
			cpu_timer_fire(timer);
		spin_unlock(&timer->it_lock);
	}
}

/*
1166
 * Set one of the process-wide special case CPU timers or RLIMIT_CPU.
1167
 * The tsk->sighand->siglock must be held by the caller.
L
Linus Torvalds 已提交
1168 1169
 */
void set_process_cpu_timer(struct task_struct *tsk, unsigned int clock_idx,
1170
			   u64 *newval, u64 *oldval)
L
Linus Torvalds 已提交
1171
{
1172
	u64 now;
L
Linus Torvalds 已提交
1173

1174
	WARN_ON_ONCE(clock_idx == CPUCLOCK_SCHED);
1175
	cpu_timer_sample_group(clock_idx, tsk, &now);
L
Linus Torvalds 已提交
1176 1177

	if (oldval) {
1178 1179 1180 1181 1182
		/*
		 * We are setting itimer. The *oldval is absolute and we update
		 * it to be relative, *newval argument is relative and we update
		 * it to be absolute.
		 */
1183
		if (*oldval) {
1184
			if (*oldval <= now) {
L
Linus Torvalds 已提交
1185
				/* Just about to fire. */
1186
				*oldval = TICK_NSEC;
L
Linus Torvalds 已提交
1187
			} else {
1188
				*oldval -= now;
L
Linus Torvalds 已提交
1189 1190 1191
			}
		}

1192
		if (!*newval)
1193
			return;
1194
		*newval += now;
L
Linus Torvalds 已提交
1195 1196 1197
	}

	/*
1198 1199
	 * Update expiration cache if we are the earliest timer, or eventually
	 * RLIMIT_CPU limit is earlier than prof_exp cpu timer expire.
L
Linus Torvalds 已提交
1200
	 */
1201 1202
	switch (clock_idx) {
	case CPUCLOCK_PROF:
1203 1204
		if (expires_gt(tsk->signal->cputime_expires.prof_exp, *newval))
			tsk->signal->cputime_expires.prof_exp = *newval;
1205 1206
		break;
	case CPUCLOCK_VIRT:
1207 1208
		if (expires_gt(tsk->signal->cputime_expires.virt_exp, *newval))
			tsk->signal->cputime_expires.virt_exp = *newval;
1209
		break;
L
Linus Torvalds 已提交
1210
	}
1211 1212

	tick_dep_set_signal(tsk->signal, TICK_DEP_BIT_POSIX_TIMER);
L
Linus Torvalds 已提交
1213 1214
}

1215 1216
static int do_cpu_nanosleep(const clockid_t which_clock, int flags,
			    struct timespec *rqtp, struct itimerspec *it)
L
Linus Torvalds 已提交
1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231
{
	struct k_itimer timer;
	int error;

	/*
	 * Set up a temporary timer and then wait for it to go off.
	 */
	memset(&timer, 0, sizeof timer);
	spin_lock_init(&timer.it_lock);
	timer.it_clock = which_clock;
	timer.it_overrun = -1;
	error = posix_cpu_timer_create(&timer);
	timer.it_process = current;
	if (!error) {
		static struct itimerspec zero_it;
1232 1233 1234

		memset(it, 0, sizeof *it);
		it->it_value = *rqtp;
L
Linus Torvalds 已提交
1235 1236

		spin_lock_irq(&timer.it_lock);
1237
		error = posix_cpu_timer_set(&timer, flags, it, NULL);
L
Linus Torvalds 已提交
1238 1239 1240 1241 1242 1243
		if (error) {
			spin_unlock_irq(&timer.it_lock);
			return error;
		}

		while (!signal_pending(current)) {
1244
			if (timer.it.cpu.expires == 0) {
L
Linus Torvalds 已提交
1245
				/*
1246 1247
				 * Our timer fired and was reset, below
				 * deletion can not fail.
L
Linus Torvalds 已提交
1248
				 */
1249
				posix_cpu_timer_del(&timer);
L
Linus Torvalds 已提交
1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265
				spin_unlock_irq(&timer.it_lock);
				return 0;
			}

			/*
			 * Block until cpu_timer_fire (or a signal) wakes us.
			 */
			__set_current_state(TASK_INTERRUPTIBLE);
			spin_unlock_irq(&timer.it_lock);
			schedule();
			spin_lock_irq(&timer.it_lock);
		}

		/*
		 * We were interrupted by a signal.
		 */
1266
		*rqtp = ns_to_timespec(timer.it.cpu.expires);
1267 1268 1269 1270 1271 1272 1273
		error = posix_cpu_timer_set(&timer, 0, &zero_it, it);
		if (!error) {
			/*
			 * Timer is now unarmed, deletion can not fail.
			 */
			posix_cpu_timer_del(&timer);
		}
L
Linus Torvalds 已提交
1274 1275
		spin_unlock_irq(&timer.it_lock);

1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286
		while (error == TIMER_RETRY) {
			/*
			 * We need to handle case when timer was or is in the
			 * middle of firing. In other cases we already freed
			 * resources.
			 */
			spin_lock_irq(&timer.it_lock);
			error = posix_cpu_timer_del(&timer);
			spin_unlock_irq(&timer.it_lock);
		}

1287
		if ((it->it_value.tv_sec | it->it_value.tv_nsec) == 0) {
L
Linus Torvalds 已提交
1288 1289 1290 1291 1292 1293
			/*
			 * It actually did fire already.
			 */
			return 0;
		}

1294 1295 1296 1297 1298 1299
		error = -ERESTART_RESTARTBLOCK;
	}

	return error;
}

1300 1301 1302 1303
static long posix_cpu_nsleep_restart(struct restart_block *restart_block);

static int posix_cpu_nsleep(const clockid_t which_clock, int flags,
			    struct timespec *rqtp, struct timespec __user *rmtp)
1304
{
1305
	struct restart_block *restart_block = &current->restart_block;
1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320
	struct itimerspec it;
	int error;

	/*
	 * Diagnose required errors first.
	 */
	if (CPUCLOCK_PERTHREAD(which_clock) &&
	    (CPUCLOCK_PID(which_clock) == 0 ||
	     CPUCLOCK_PID(which_clock) == current->pid))
		return -EINVAL;

	error = do_cpu_nanosleep(which_clock, flags, rqtp, &it);

	if (error == -ERESTART_RESTARTBLOCK) {

1321
		if (flags & TIMER_ABSTIME)
1322
			return -ERESTARTNOHAND;
L
Linus Torvalds 已提交
1323
		/*
1324 1325 1326
		 * Report back to the user the time still remaining.
		 */
		if (rmtp && copy_to_user(rmtp, &it.it_value, sizeof *rmtp))
L
Linus Torvalds 已提交
1327 1328
			return -EFAULT;

1329
		restart_block->fn = posix_cpu_nsleep_restart;
1330
		restart_block->nanosleep.clockid = which_clock;
1331 1332
		restart_block->nanosleep.rmtp = rmtp;
		restart_block->nanosleep.expires = timespec_to_ns(rqtp);
L
Linus Torvalds 已提交
1333 1334 1335 1336
	}
	return error;
}

1337
static long posix_cpu_nsleep_restart(struct restart_block *restart_block)
L
Linus Torvalds 已提交
1338
{
1339
	clockid_t which_clock = restart_block->nanosleep.clockid;
1340
	struct timespec t;
1341 1342
	struct itimerspec it;
	int error;
1343

1344
	t = ns_to_timespec(restart_block->nanosleep.expires);
1345

1346 1347 1348
	error = do_cpu_nanosleep(which_clock, TIMER_ABSTIME, &t, &it);

	if (error == -ERESTART_RESTARTBLOCK) {
1349
		struct timespec __user *rmtp = restart_block->nanosleep.rmtp;
1350
		/*
1351 1352 1353
		 * Report back to the user the time still remaining.
		 */
		if (rmtp && copy_to_user(rmtp, &it.it_value, sizeof *rmtp))
1354 1355
			return -EFAULT;

1356
		restart_block->nanosleep.expires = timespec_to_ns(&t);
1357 1358 1359
	}
	return error;

L
Linus Torvalds 已提交
1360 1361 1362 1363 1364
}

#define PROCESS_CLOCK	MAKE_PROCESS_CPUCLOCK(0, CPUCLOCK_SCHED)
#define THREAD_CLOCK	MAKE_THREAD_CPUCLOCK(0, CPUCLOCK_SCHED)

1365 1366
static int process_cpu_clock_getres(const clockid_t which_clock,
				    struct timespec *tp)
L
Linus Torvalds 已提交
1367 1368 1369
{
	return posix_cpu_clock_getres(PROCESS_CLOCK, tp);
}
1370 1371
static int process_cpu_clock_get(const clockid_t which_clock,
				 struct timespec *tp)
L
Linus Torvalds 已提交
1372 1373 1374 1375 1376 1377 1378 1379
{
	return posix_cpu_clock_get(PROCESS_CLOCK, tp);
}
static int process_cpu_timer_create(struct k_itimer *timer)
{
	timer->it_clock = PROCESS_CLOCK;
	return posix_cpu_timer_create(timer);
}
1380
static int process_cpu_nsleep(const clockid_t which_clock, int flags,
1381 1382
			      struct timespec *rqtp,
			      struct timespec __user *rmtp)
L
Linus Torvalds 已提交
1383
{
1384
	return posix_cpu_nsleep(PROCESS_CLOCK, flags, rqtp, rmtp);
L
Linus Torvalds 已提交
1385
}
1386 1387 1388 1389
static long process_cpu_nsleep_restart(struct restart_block *restart_block)
{
	return -EINVAL;
}
1390 1391
static int thread_cpu_clock_getres(const clockid_t which_clock,
				   struct timespec *tp)
L
Linus Torvalds 已提交
1392 1393 1394
{
	return posix_cpu_clock_getres(THREAD_CLOCK, tp);
}
1395 1396
static int thread_cpu_clock_get(const clockid_t which_clock,
				struct timespec *tp)
L
Linus Torvalds 已提交
1397 1398 1399 1400 1401 1402 1403 1404 1405
{
	return posix_cpu_clock_get(THREAD_CLOCK, tp);
}
static int thread_cpu_timer_create(struct k_itimer *timer)
{
	timer->it_clock = THREAD_CLOCK;
	return posix_cpu_timer_create(timer);
}

1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417
struct k_clock clock_posix_cpu = {
	.clock_getres	= posix_cpu_clock_getres,
	.clock_set	= posix_cpu_clock_set,
	.clock_get	= posix_cpu_clock_get,
	.timer_create	= posix_cpu_timer_create,
	.nsleep		= posix_cpu_nsleep,
	.nsleep_restart	= posix_cpu_nsleep_restart,
	.timer_set	= posix_cpu_timer_set,
	.timer_del	= posix_cpu_timer_del,
	.timer_get	= posix_cpu_timer_get,
};

L
Linus Torvalds 已提交
1418 1419 1420
static __init int init_posix_cpu_timers(void)
{
	struct k_clock process = {
1421 1422 1423 1424 1425
		.clock_getres	= process_cpu_clock_getres,
		.clock_get	= process_cpu_clock_get,
		.timer_create	= process_cpu_timer_create,
		.nsleep		= process_cpu_nsleep,
		.nsleep_restart	= process_cpu_nsleep_restart,
L
Linus Torvalds 已提交
1426 1427
	};
	struct k_clock thread = {
1428 1429 1430
		.clock_getres	= thread_cpu_clock_getres,
		.clock_get	= thread_cpu_clock_get,
		.timer_create	= thread_cpu_timer_create,
L
Linus Torvalds 已提交
1431 1432
	};

1433 1434
	posix_timers_register_clock(CLOCK_PROCESS_CPUTIME_ID, &process);
	posix_timers_register_clock(CLOCK_THREAD_CPUTIME_ID, &thread);
L
Linus Torvalds 已提交
1435 1436 1437 1438

	return 0;
}
__initcall(init_posix_cpu_timers);