exit.c 42.8 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10
/*
 *  linux/kernel/exit.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 */

#include <linux/mm.h>
#include <linux/slab.h>
#include <linux/interrupt.h>
#include <linux/module.h>
11
#include <linux/capability.h>
L
Linus Torvalds 已提交
12 13 14
#include <linux/completion.h>
#include <linux/personality.h>
#include <linux/tty.h>
15
#include <linux/iocontext.h>
L
Linus Torvalds 已提交
16 17 18 19
#include <linux/key.h>
#include <linux/security.h>
#include <linux/cpu.h>
#include <linux/acct.h>
20
#include <linux/tsacct_kern.h>
L
Linus Torvalds 已提交
21
#include <linux/file.h>
A
Al Viro 已提交
22
#include <linux/fdtable.h>
23
#include <linux/freezer.h>
L
Linus Torvalds 已提交
24
#include <linux/binfmts.h>
S
Serge E. Hallyn 已提交
25
#include <linux/nsproxy.h>
26
#include <linux/pid_namespace.h>
L
Linus Torvalds 已提交
27 28 29 30
#include <linux/ptrace.h>
#include <linux/profile.h>
#include <linux/mount.h>
#include <linux/proc_fs.h>
31
#include <linux/kthread.h>
L
Linus Torvalds 已提交
32
#include <linux/mempolicy.h>
33
#include <linux/taskstats_kern.h>
34
#include <linux/delayacct.h>
35
#include <linux/cgroup.h>
L
Linus Torvalds 已提交
36
#include <linux/syscalls.h>
37
#include <linux/signal.h>
38
#include <linux/posix-timers.h>
M
Matt Helsley 已提交
39
#include <linux/cn_proc.h>
40
#include <linux/mutex.h>
41
#include <linux/futex.h>
42
#include <linux/pipe_fs_i.h>
43
#include <linux/audit.h> /* for audit_free() */
A
Adrian Bunk 已提交
44
#include <linux/resource.h>
45
#include <linux/blkdev.h>
46
#include <linux/task_io_accounting_ops.h>
R
Roland McGrath 已提交
47
#include <linux/tracehook.h>
48
#include <linux/fs_struct.h>
D
David Howells 已提交
49
#include <linux/init_task.h>
50
#include <linux/perf_event.h>
51
#include <trace/events/sched.h>
52
#include <linux/hw_breakpoint.h>
Y
Ying Han 已提交
53
#include <linux/oom.h>
54
#include <linux/writeback.h>
A
Al Viro 已提交
55
#include <linux/shm.h>
L
Linus Torvalds 已提交
56 57 58 59 60 61

#include <asm/uaccess.h>
#include <asm/unistd.h>
#include <asm/pgtable.h>
#include <asm/mmu_context.h>

A
Adrian Bunk 已提交
62 63
static void exit_mm(struct task_struct * tsk);

64
static void __unhash_process(struct task_struct *p, bool group_dead)
L
Linus Torvalds 已提交
65 66
{
	nr_threads--;
67
	detach_pid(p, PIDTYPE_PID);
68
	if (group_dead) {
L
Linus Torvalds 已提交
69 70
		detach_pid(p, PIDTYPE_PGID);
		detach_pid(p, PIDTYPE_SID);
71

72
		list_del_rcu(&p->tasks);
73
		list_del_init(&p->sibling);
74
		__this_cpu_dec(process_counts);
L
Linus Torvalds 已提交
75
	}
O
Oleg Nesterov 已提交
76
	list_del_rcu(&p->thread_group);
L
Linus Torvalds 已提交
77 78
}

79 80 81 82 83 84
/*
 * This function expects the tasklist_lock write-locked.
 */
static void __exit_signal(struct task_struct *tsk)
{
	struct signal_struct *sig = tsk->signal;
85
	bool group_dead = thread_group_leader(tsk);
86
	struct sighand_struct *sighand;
87
	struct tty_struct *uninitialized_var(tty);
88
	cputime_t utime, stime;
89

90
	sighand = rcu_dereference_check(tsk->sighand,
91
					lockdep_tasklist_lock_is_held());
92 93 94
	spin_lock(&sighand->siglock);

	posix_cpu_timers_exit(tsk);
95
	if (group_dead) {
96
		posix_cpu_timers_exit_group(tsk);
97 98
		tty = sig->tty;
		sig->tty = NULL;
99
	} else {
100 101 102 103 104 105 106 107
		/*
		 * This can only happen if the caller is de_thread().
		 * FIXME: this is the temporary hack, we should teach
		 * posix-cpu-timers to handle this case correctly.
		 */
		if (unlikely(has_group_leader_pid(tsk)))
			posix_cpu_timers_exit_group(tsk);

108 109 110 111
		/*
		 * If there is any task waiting for the group exit
		 * then notify it:
		 */
112
		if (sig->notify_count > 0 && !--sig->notify_count)
113
			wake_up_process(sig->group_exit_task);
114

115 116 117 118 119 120 121 122 123 124 125 126
		if (tsk == sig->curr_target)
			sig->curr_target = next_thread(tsk);
		/*
		 * Accumulate here the counters for all threads but the
		 * group leader as they die, so they can be added into
		 * the process-wide totals when those are taken.
		 * The group leader stays around as a zombie as long
		 * as there are other threads.  When it gets reaped,
		 * the exit.c code will add its counts into these totals.
		 * We won't ever get here for the group leader, since it
		 * will have been the last reference on the signal_struct.
		 */
127 128 129 130
		task_cputime(tsk, &utime, &stime);
		sig->utime += utime;
		sig->stime += stime;
		sig->gtime += task_gtime(tsk);
131 132 133 134
		sig->min_flt += tsk->min_flt;
		sig->maj_flt += tsk->maj_flt;
		sig->nvcsw += tsk->nvcsw;
		sig->nivcsw += tsk->nivcsw;
135 136
		sig->inblock += task_io_get_inblock(tsk);
		sig->oublock += task_io_get_oublock(tsk);
137
		task_io_accounting_add(&sig->ioac, &tsk->ioac);
138
		sig->sum_sched_runtime += tsk->se.sum_exec_runtime;
139 140
	}

141
	sig->nr_threads--;
142
	__unhash_process(tsk, group_dead);
143

144 145 146 147 148
	/*
	 * Do this under ->siglock, we can race with another thread
	 * doing sigqueue_free() if we have SIGQUEUE_PREALLOC signals.
	 */
	flush_sigqueue(&tsk->pending);
149
	tsk->sighand = NULL;
150 151
	spin_unlock(&sighand->siglock);

152
	__cleanup_sighand(sighand);
153
	clear_tsk_thread_flag(tsk,TIF_SIGPENDING);
154
	if (group_dead) {
155
		flush_sigqueue(&sig->shared_pending);
156
		tty_kref_put(tty);
157 158 159
	}
}

160 161
static void delayed_put_task_struct(struct rcu_head *rhp)
{
162 163
	struct task_struct *tsk = container_of(rhp, struct task_struct, rcu);

164
	perf_event_delayed_put(tsk);
165 166
	trace_sched_process_free(tsk);
	put_task_struct(tsk);
167 168
}

R
Roland McGrath 已提交
169

L
Linus Torvalds 已提交
170 171
void release_task(struct task_struct * p)
{
172
	struct task_struct *leader;
L
Linus Torvalds 已提交
173
	int zap_leader;
174
repeat:
175
	/* don't need to get the RCU readlock here - the process is dead and
176 177
	 * can't be modifying its own credentials. But shut RCU-lockdep up */
	rcu_read_lock();
178
	atomic_dec(&__task_cred(p)->user->processes);
179
	rcu_read_unlock();
180

181
	proc_flush_task(p);
182

L
Linus Torvalds 已提交
183
	write_lock_irq(&tasklist_lock);
T
Tejun Heo 已提交
184
	ptrace_release_task(p);
L
Linus Torvalds 已提交
185
	__exit_signal(p);
186

L
Linus Torvalds 已提交
187 188 189 190 191 192 193 194 195 196 197 198
	/*
	 * If we are the last non-leader member of the thread
	 * group, and the leader is zombie, then notify the
	 * group leader's parent process. (if it wants notification.)
	 */
	zap_leader = 0;
	leader = p->group_leader;
	if (leader != p && thread_group_empty(leader) && leader->exit_state == EXIT_ZOMBIE) {
		/*
		 * If we were the last child thread and the leader has
		 * exited already, and the leader's parent ignores SIGCHLD,
		 * then we are the one who should release the leader.
R
Roland McGrath 已提交
199
		 */
200
		zap_leader = do_notify_parent(leader, leader->exit_signal);
R
Roland McGrath 已提交
201 202
		if (zap_leader)
			leader->exit_state = EXIT_DEAD;
L
Linus Torvalds 已提交
203 204 205 206
	}

	write_unlock_irq(&tasklist_lock);
	release_thread(p);
207
	call_rcu(&p->rcu, delayed_put_task_struct);
L
Linus Torvalds 已提交
208 209 210 211 212 213 214 215 216 217

	p = leader;
	if (unlikely(zap_leader))
		goto repeat;
}

/*
 * This checks not only the pgrp, but falls back on the pid if no
 * satisfactory pgrp is found. I dunno - gdb doesn't work correctly
 * without this...
218 219
 *
 * The caller must hold rcu lock or the tasklist lock.
L
Linus Torvalds 已提交
220
 */
221
struct pid *session_of_pgrp(struct pid *pgrp)
L
Linus Torvalds 已提交
222 223
{
	struct task_struct *p;
224
	struct pid *sid = NULL;
225

226
	p = pid_task(pgrp, PIDTYPE_PGID);
227
	if (p == NULL)
228
		p = pid_task(pgrp, PIDTYPE_PID);
229
	if (p != NULL)
230
		sid = task_session(p);
231

L
Linus Torvalds 已提交
232 233 234 235 236 237 238 239 240 241 242
	return sid;
}

/*
 * Determine if a process group is "orphaned", according to the POSIX
 * definition in 2.2.2.52.  Orphaned process groups are not to be affected
 * by terminal-generated stop signals.  Newly orphaned process groups are
 * to receive a SIGHUP and a SIGCONT.
 *
 * "I ask you, have you ever known what it is to be an orphan?"
 */
243
static int will_become_orphaned_pgrp(struct pid *pgrp, struct task_struct *ignored_task)
L
Linus Torvalds 已提交
244 245 246
{
	struct task_struct *p;

247
	do_each_pid_task(pgrp, PIDTYPE_PGID, p) {
248 249 250
		if ((p == ignored_task) ||
		    (p->exit_state && thread_group_empty(p)) ||
		    is_global_init(p->real_parent))
L
Linus Torvalds 已提交
251
			continue;
252

253
		if (task_pgrp(p->real_parent) != pgrp &&
254 255
		    task_session(p->real_parent) == task_session(p))
			return 0;
256
	} while_each_pid_task(pgrp, PIDTYPE_PGID, p);
257 258

	return 1;
L
Linus Torvalds 已提交
259 260
}

261
int is_current_pgrp_orphaned(void)
L
Linus Torvalds 已提交
262 263 264 265
{
	int retval;

	read_lock(&tasklist_lock);
266
	retval = will_become_orphaned_pgrp(task_pgrp(current), NULL);
L
Linus Torvalds 已提交
267 268 269 270 271
	read_unlock(&tasklist_lock);

	return retval;
}

272
static bool has_stopped_jobs(struct pid *pgrp)
L
Linus Torvalds 已提交
273 274 275
{
	struct task_struct *p;

276
	do_each_pid_task(pgrp, PIDTYPE_PGID, p) {
277 278
		if (p->signal->flags & SIGNAL_STOP_STOPPED)
			return true;
279
	} while_each_pid_task(pgrp, PIDTYPE_PGID, p);
280 281

	return false;
L
Linus Torvalds 已提交
282 283
}

284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314
/*
 * Check to see if any process groups have become orphaned as
 * a result of our exiting, and if they have any stopped jobs,
 * send them a SIGHUP and then a SIGCONT. (POSIX 3.2.2.2)
 */
static void
kill_orphaned_pgrp(struct task_struct *tsk, struct task_struct *parent)
{
	struct pid *pgrp = task_pgrp(tsk);
	struct task_struct *ignored_task = tsk;

	if (!parent)
		 /* exit: our father is in a different pgrp than
		  * we are and we were the only connection outside.
		  */
		parent = tsk->real_parent;
	else
		/* reparent: our child is in a different pgrp than
		 * we are, and it was the only connection outside.
		 */
		ignored_task = NULL;

	if (task_pgrp(parent) != pgrp &&
	    task_session(parent) == task_session(tsk) &&
	    will_become_orphaned_pgrp(pgrp, ignored_task) &&
	    has_stopped_jobs(pgrp)) {
		__kill_pgrp_info(SIGHUP, SEND_SIG_PRIV, pgrp);
		__kill_pgrp_info(SIGCONT, SEND_SIG_PRIV, pgrp);
	}
}

315
void __set_special_pids(struct pid *pid)
L
Linus Torvalds 已提交
316
{
317
	struct task_struct *curr = current->group_leader;
L
Linus Torvalds 已提交
318

319
	if (task_session(curr) != pid)
320
		change_pid(curr, PIDTYPE_SID, pid);
321 322

	if (task_pgrp(curr) != pid)
323
		change_pid(curr, PIDTYPE_PGID, pid);
L
Linus Torvalds 已提交
324 325 326
}

/*
327 328
 * Let kernel threads use this to say that they allow a certain signal.
 * Must not be used if kthread was cloned with CLONE_SIGHAND.
L
Linus Torvalds 已提交
329 330 331
 */
int allow_signal(int sig)
{
332
	if (!valid_signal(sig) || sig < 1)
L
Linus Torvalds 已提交
333 334 335
		return -EINVAL;

	spin_lock_irq(&current->sighand->siglock);
336
	/* This is only needed for daemonize()'ed kthreads */
L
Linus Torvalds 已提交
337
	sigdelset(&current->blocked, sig);
338 339 340 341 342 343
	/*
	 * Kernel threads handle their own signals. Let the signal code
	 * know it'll be handled, so that they don't get converted to
	 * SIGKILL or just silently dropped.
	 */
	current->sighand->action[(sig)-1].sa.sa_handler = (void __user *)2;
L
Linus Torvalds 已提交
344 345 346 347 348 349 350 351 352
	recalc_sigpending();
	spin_unlock_irq(&current->sighand->siglock);
	return 0;
}

EXPORT_SYMBOL(allow_signal);

int disallow_signal(int sig)
{
353
	if (!valid_signal(sig) || sig < 1)
L
Linus Torvalds 已提交
354 355 356
		return -EINVAL;

	spin_lock_irq(&current->sighand->siglock);
357
	current->sighand->action[(sig)-1].sa.sa_handler = SIG_IGN;
L
Linus Torvalds 已提交
358 359 360 361 362 363 364
	recalc_sigpending();
	spin_unlock_irq(&current->sighand->siglock);
	return 0;
}

EXPORT_SYMBOL(disallow_signal);

365 366
#ifdef CONFIG_MM_OWNER
/*
367
 * A task is exiting.   If it owned this mm, find a new owner for the mm.
368 369 370 371 372 373
 */
void mm_update_next_owner(struct mm_struct *mm)
{
	struct task_struct *c, *g, *p = current;

retry:
374 375 376 377 378
	/*
	 * If the exiting or execing task is not the owner, it's
	 * someone else's problem.
	 */
	if (mm->owner != p)
379
		return;
380 381 382 383 384 385 386 387 388
	/*
	 * The current owner is exiting/execing and there are no other
	 * candidates.  Do not leave the mm pointing to a possibly
	 * freed task structure.
	 */
	if (atomic_read(&mm->mm_users) <= 1) {
		mm->owner = NULL;
		return;
	}
389 390 391 392 393 394 395 396 397 398 399 400 401

	read_lock(&tasklist_lock);
	/*
	 * Search in the children
	 */
	list_for_each_entry(c, &p->children, sibling) {
		if (c->mm == mm)
			goto assign_new_owner;
	}

	/*
	 * Search in the siblings
	 */
402
	list_for_each_entry(c, &p->real_parent->children, sibling) {
403 404 405 406 407 408 409 410 411 412 413 414 415 416
		if (c->mm == mm)
			goto assign_new_owner;
	}

	/*
	 * Search through everything else. We should not get
	 * here often
	 */
	do_each_thread(g, c) {
		if (c->mm == mm)
			goto assign_new_owner;
	} while_each_thread(g, c);

	read_unlock(&tasklist_lock);
417 418 419
	/*
	 * We found no owner yet mm_users > 1: this implies that we are
	 * most likely racing with swapoff (try_to_unuse()) or /proc or
420
	 * ptrace or page migration (get_task_mm()).  Mark owner as NULL.
421 422
	 */
	mm->owner = NULL;
423 424 425 426 427 428 429 430 431 432
	return;

assign_new_owner:
	BUG_ON(c == p);
	get_task_struct(c);
	/*
	 * The task_lock protects c->mm from changing.
	 * We always want mm->owner->mm == mm
	 */
	task_lock(c);
433 434 435 436 437
	/*
	 * Delay read_unlock() till we have the task_lock()
	 * to ensure that c does not slip away underneath us
	 */
	read_unlock(&tasklist_lock);
438 439 440 441 442 443 444 445 446 447 448
	if (c->mm != mm) {
		task_unlock(c);
		put_task_struct(c);
		goto retry;
	}
	mm->owner = c;
	task_unlock(c);
	put_task_struct(c);
}
#endif /* CONFIG_MM_OWNER */

L
Linus Torvalds 已提交
449 450 451 452
/*
 * Turn us into a lazy TLB process if we
 * aren't already..
 */
A
Adrian Bunk 已提交
453
static void exit_mm(struct task_struct * tsk)
L
Linus Torvalds 已提交
454 455
{
	struct mm_struct *mm = tsk->mm;
456
	struct core_state *core_state;
L
Linus Torvalds 已提交
457

458
	mm_release(tsk, mm);
L
Linus Torvalds 已提交
459 460
	if (!mm)
		return;
461
	sync_mm_rss(mm);
L
Linus Torvalds 已提交
462 463
	/*
	 * Serialize with any possible pending coredump.
464
	 * We must hold mmap_sem around checking core_state
L
Linus Torvalds 已提交
465
	 * and clearing tsk->mm.  The core-inducing thread
466
	 * will increment ->nr_threads for each thread in the
L
Linus Torvalds 已提交
467 468 469
	 * group with ->mm != NULL.
	 */
	down_read(&mm->mmap_sem);
470 471 472
	core_state = mm->core_state;
	if (core_state) {
		struct core_thread self;
L
Linus Torvalds 已提交
473 474
		up_read(&mm->mmap_sem);

475 476 477 478 479 480 481 482
		self.task = tsk;
		self.next = xchg(&core_state->dumper.next, &self);
		/*
		 * Implies mb(), the result of xchg() must be visible
		 * to core_state->dumper.
		 */
		if (atomic_dec_and_test(&core_state->nr_threads))
			complete(&core_state->startup);
L
Linus Torvalds 已提交
483

O
Oleg Nesterov 已提交
484 485 486 487
		for (;;) {
			set_task_state(tsk, TASK_UNINTERRUPTIBLE);
			if (!self.task) /* see coredump_finish() */
				break;
488
			freezable_schedule();
O
Oleg Nesterov 已提交
489 490
		}
		__set_task_state(tsk, TASK_RUNNING);
L
Linus Torvalds 已提交
491 492 493
		down_read(&mm->mmap_sem);
	}
	atomic_inc(&mm->mm_count);
E
Eric Sesterhenn 已提交
494
	BUG_ON(mm != tsk->active_mm);
L
Linus Torvalds 已提交
495 496 497 498 499 500
	/* more a memory barrier than a real lock */
	task_lock(tsk);
	tsk->mm = NULL;
	up_read(&mm->mmap_sem);
	enter_lazy_tlb(mm, current);
	task_unlock(tsk);
501
	mm_update_next_owner(mm);
L
Linus Torvalds 已提交
502 503 504 505
	mmput(mm);
}

/*
506 507 508 509 510
 * When we die, we re-parent all our children, and try to:
 * 1. give them to another thread in our thread group, if such a member exists
 * 2. give it to the first ancestor process which prctl'd itself as a
 *    child_subreaper for its children (like a service manager)
 * 3. give it to the init process (PID 1) in our pid namespace
L
Linus Torvalds 已提交
511
 */
512
static struct task_struct *find_new_reaper(struct task_struct *father)
513 514
	__releases(&tasklist_lock)
	__acquires(&tasklist_lock)
L
Linus Torvalds 已提交
515
{
516 517
	struct pid_namespace *pid_ns = task_active_pid_ns(father);
	struct task_struct *thread;
L
Linus Torvalds 已提交
518

519 520 521 522 523 524 525 526
	thread = father;
	while_each_thread(father, thread) {
		if (thread->flags & PF_EXITING)
			continue;
		if (unlikely(pid_ns->child_reaper == father))
			pid_ns->child_reaper = thread;
		return thread;
	}
L
Linus Torvalds 已提交
527

528 529
	if (unlikely(pid_ns->child_reaper == father)) {
		write_unlock_irq(&tasklist_lock);
530 531 532 533 534
		if (unlikely(pid_ns == &init_pid_ns)) {
			panic("Attempted to kill init! exitcode=0x%08x\n",
				father->signal->group_exit_code ?:
					father->exit_code);
		}
L
Linus Torvalds 已提交
535

536 537
		zap_pid_ns_processes(pid_ns);
		write_lock_irq(&tasklist_lock);
538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560
	} else if (father->signal->has_child_subreaper) {
		struct task_struct *reaper;

		/*
		 * Find the first ancestor marked as child_subreaper.
		 * Note that the code below checks same_thread_group(reaper,
		 * pid_ns->child_reaper).  This is what we need to DTRT in a
		 * PID namespace. However we still need the check above, see
		 * http://marc.info/?l=linux-kernel&m=131385460420380
		 */
		for (reaper = father->real_parent;
		     reaper != &init_task;
		     reaper = reaper->real_parent) {
			if (same_thread_group(reaper, pid_ns->child_reaper))
				break;
			if (!reaper->signal->is_child_subreaper)
				continue;
			thread = reaper;
			do {
				if (!(thread->flags & PF_EXITING))
					return reaper;
			} while_each_thread(reaper, thread);
		}
L
Linus Torvalds 已提交
561
	}
562

563 564 565
	return pid_ns->child_reaper;
}

566 567 568
/*
* Any that need to be release_task'd are put on the @dead list.
 */
569
static void reparent_leader(struct task_struct *father, struct task_struct *p,
570 571 572 573
				struct list_head *dead)
{
	list_move_tail(&p->sibling, &p->real_parent->children);

574
	if (p->exit_state == EXIT_DEAD)
575 576 577 578 579 580 581 582 583 584 585 586
		return;
	/*
	 * If this is a threaded reparent there is no need to
	 * notify anyone anything has happened.
	 */
	if (same_thread_group(p->real_parent, father))
		return;

	/* We don't want people slaying init.  */
	p->exit_signal = SIGCHLD;

	/* If it has exited notify the new parent about this child's death. */
T
Tejun Heo 已提交
587
	if (!p->ptrace &&
588
	    p->exit_state == EXIT_ZOMBIE && thread_group_empty(p)) {
589
		if (do_notify_parent(p, p->exit_signal)) {
590 591 592 593 594 595 596 597
			p->exit_state = EXIT_DEAD;
			list_move_tail(&p->sibling, dead);
		}
	}

	kill_orphaned_pgrp(p, father);
}

598
static void forget_original_parent(struct task_struct *father)
L
Linus Torvalds 已提交
599
{
600
	struct task_struct *p, *n, *reaper;
601
	LIST_HEAD(dead_children);
602 603

	write_lock_irq(&tasklist_lock);
604 605 606 607 608
	/*
	 * Note that exit_ptrace() and find_new_reaper() might
	 * drop tasklist_lock and reacquire it.
	 */
	exit_ptrace(father);
609
	reaper = find_new_reaper(father);
R
Roland McGrath 已提交
610

611
	list_for_each_entry_safe(p, n, &father->children, sibling) {
612 613 614 615
		struct task_struct *t = p;
		do {
			t->real_parent = reaper;
			if (t->parent == father) {
T
Tejun Heo 已提交
616
				BUG_ON(t->ptrace);
617 618 619 620 621 622 623
				t->parent = t->real_parent;
			}
			if (t->pdeath_signal)
				group_send_sig_info(t->pdeath_signal,
						    SEND_SIG_NOINFO, t);
		} while_each_thread(p, t);
		reparent_leader(father, p, &dead_children);
L
Linus Torvalds 已提交
624
	}
625
	write_unlock_irq(&tasklist_lock);
626

627 628
	BUG_ON(!list_empty(&father->children));

629 630
	list_for_each_entry_safe(p, n, &dead_children, sibling) {
		list_del_init(&p->sibling);
631 632
		release_task(p);
	}
L
Linus Torvalds 已提交
633 634 635 636 637 638
}

/*
 * Send signals to all our closest relatives so that they know
 * to properly mourn us..
 */
639
static void exit_notify(struct task_struct *tsk, int group_dead)
L
Linus Torvalds 已提交
640
{
641
	bool autoreap;
L
Linus Torvalds 已提交
642 643 644 645 646 647 648 649 650

	/*
	 * This does two things:
	 *
  	 * A.  Make init inherit all the child processes
	 * B.  Check to see if any process groups have become orphaned
	 *	as a result of our exiting, and if they have any stopped
	 *	jobs, send them a SIGHUP and then a SIGCONT.  (POSIX 3.2.2.2)
	 */
651
	forget_original_parent(tsk);
652
	exit_task_namespaces(tsk);
L
Linus Torvalds 已提交
653

654
	write_lock_irq(&tasklist_lock);
655 656
	if (group_dead)
		kill_orphaned_pgrp(tsk->group_leader, NULL);
L
Linus Torvalds 已提交
657

O
Oleg Nesterov 已提交
658 659 660 661 662 663 664 665 666 667 668 669
	if (unlikely(tsk->ptrace)) {
		int sig = thread_group_leader(tsk) &&
				thread_group_empty(tsk) &&
				!ptrace_reparented(tsk) ?
			tsk->exit_signal : SIGCHLD;
		autoreap = do_notify_parent(tsk, sig);
	} else if (thread_group_leader(tsk)) {
		autoreap = thread_group_empty(tsk) &&
			do_notify_parent(tsk, tsk->exit_signal);
	} else {
		autoreap = true;
	}
L
Linus Torvalds 已提交
670

671
	tsk->exit_state = autoreap ? EXIT_DEAD : EXIT_ZOMBIE;
L
Linus Torvalds 已提交
672

673 674
	/* mt-exec, de_thread() is waiting for group leader */
	if (unlikely(tsk->signal->notify_count < 0))
675
		wake_up_process(tsk->signal->group_exit_task);
L
Linus Torvalds 已提交
676 677 678
	write_unlock_irq(&tasklist_lock);

	/* If the process is dead, release it - nobody will wait for it */
679
	if (autoreap)
L
Linus Torvalds 已提交
680 681 682
		release_task(tsk);
}

683 684 685 686 687 688 689
#ifdef CONFIG_DEBUG_STACK_USAGE
static void check_stack_usage(void)
{
	static DEFINE_SPINLOCK(low_water_lock);
	static int lowest_to_date = THREAD_SIZE;
	unsigned long free;

690
	free = stack_not_used(current);
691 692 693 694 695 696

	if (free >= lowest_to_date)
		return;

	spin_lock(&low_water_lock);
	if (free < lowest_to_date) {
697 698 699
		printk(KERN_WARNING "%s (%d) used greatest stack depth: "
				"%lu bytes left\n",
				current->comm, task_pid_nr(current), free);
700 701 702 703 704 705 706 707
		lowest_to_date = free;
	}
	spin_unlock(&low_water_lock);
}
#else
static inline void check_stack_usage(void) {}
#endif

708
void do_exit(long code)
L
Linus Torvalds 已提交
709 710 711 712 713 714
{
	struct task_struct *tsk = current;
	int group_dead;

	profile_task_exit(tsk);

715
	WARN_ON(blk_needs_flush_plug(tsk));
716

L
Linus Torvalds 已提交
717 718 719 720 721
	if (unlikely(in_interrupt()))
		panic("Aiee, killing interrupt handler!");
	if (unlikely(!tsk->pid))
		panic("Attempted to kill the idle task!");

722 723 724 725 726 727 728 729 730
	/*
	 * If do_exit is called because this processes oopsed, it's possible
	 * that get_fs() was left as KERNEL_DS, so reset it to USER_DS before
	 * continuing. Amongst other possible reasons, this is to prevent
	 * mm_release()->clear_child_tid() from writing to a user-controlled
	 * kernel address.
	 */
	set_fs(USER_DS);

T
Tejun Heo 已提交
731
	ptrace_event(PTRACE_EVENT_EXIT, code);
L
Linus Torvalds 已提交
732

733 734
	validate_creds_for_do_exit(tsk);

735 736 737 738 739 740 741
	/*
	 * We're taking recursive faults here in do_exit. Safest is to just
	 * leave this task alone and wait for reboot.
	 */
	if (unlikely(tsk->flags & PF_EXITING)) {
		printk(KERN_ALERT
			"Fixing recursive fault but reboot is needed!\n");
742 743 744 745 746 747 748 749 750 751
		/*
		 * We can do this unlocked here. The futex code uses
		 * this flag just to verify whether the pi state
		 * cleanup has been done or not. In the worst case it
		 * loops once more. We pretend that the cleanup was
		 * done as there is no way to return. Either the
		 * OWNER_DIED bit is set by now or we push the blocked
		 * task into the wait for ever nirwana as well.
		 */
		tsk->flags |= PF_EXITPIDONE;
752 753 754 755
		set_current_state(TASK_UNINTERRUPTIBLE);
		schedule();
	}

O
Oleg Nesterov 已提交
756
	exit_signals(tsk);  /* sets PF_EXITING */
757 758
	/*
	 * tsk->flags are checked in the futex code to protect against
759
	 * an exiting task cleaning up the robust pi futexes.
760
	 */
761
	smp_mb();
762
	raw_spin_unlock_wait(&tsk->pi_lock);
L
Linus Torvalds 已提交
763 764 765

	if (unlikely(in_atomic()))
		printk(KERN_INFO "note: %s[%d] exited with preempt_count %d\n",
766
				current->comm, task_pid_nr(current),
L
Linus Torvalds 已提交
767 768 769
				preempt_count());

	acct_update_integrals(tsk);
770 771 772
	/* sync mm's RSS info before statistics gathering */
	if (tsk->mm)
		sync_mm_rss(tsk->mm);
L
Linus Torvalds 已提交
773
	group_dead = atomic_dec_and_test(&tsk->signal->live);
774
	if (group_dead) {
775
		hrtimer_cancel(&tsk->signal->real_timer);
776
		exit_itimers(tsk->signal);
J
Jiri Pirko 已提交
777 778
		if (tsk->mm)
			setmax_mm_hiwater_rss(&tsk->signal->maxrss, tsk->mm);
779
	}
780
	acct_collect(code, group_dead);
M
Miloslav Trmac 已提交
781 782
	if (group_dead)
		tty_audit_exit();
783
	audit_free(tsk);
784

785
	tsk->exit_code = code;
786
	taskstats_exit(tsk, group_dead);
787

L
Linus Torvalds 已提交
788 789
	exit_mm(tsk);

790
	if (group_dead)
791
		acct_process();
792 793
	trace_sched_process_exit(tsk);

L
Linus Torvalds 已提交
794
	exit_sem(tsk);
795
	exit_shm(tsk);
796 797
	exit_files(tsk);
	exit_fs(tsk);
798
	exit_task_work(tsk);
799
	check_stack_usage();
L
Linus Torvalds 已提交
800
	exit_thread();
801 802 803 804 805 806 807 808 809

	/*
	 * Flush inherited counters to the parent - before the parent
	 * gets woken up by child-exit notifications.
	 *
	 * because of cgroup mode, must be called before cgroup_exit()
	 */
	perf_event_exit_task(tsk);

810
	cgroup_exit(tsk, 1);
L
Linus Torvalds 已提交
811

812
	if (group_dead)
L
Linus Torvalds 已提交
813 814
		disassociate_ctty(1);

A
Al Viro 已提交
815
	module_put(task_thread_info(tsk)->exec_domain->module);
L
Linus Torvalds 已提交
816

M
Matt Helsley 已提交
817
	proc_exit_connector(tsk);
818

819 820 821
	/*
	 * FIXME: do that only when needed, using sched_exit tracepoint
	 */
822
	ptrace_put_breakpoints(tsk);
823

824
	exit_notify(tsk, group_dead);
L
Linus Torvalds 已提交
825
#ifdef CONFIG_NUMA
826
	task_lock(tsk);
827
	mpol_put(tsk->mempolicy);
L
Linus Torvalds 已提交
828
	tsk->mempolicy = NULL;
829
	task_unlock(tsk);
L
Linus Torvalds 已提交
830
#endif
831
#ifdef CONFIG_FUTEX
832 833
	if (unlikely(current->pi_state_cache))
		kfree(current->pi_state_cache);
834
#endif
835
	/*
836
	 * Make sure we are holding no locks:
837
	 */
838
	debug_check_no_locks_held(tsk);
839 840 841 842 843 844
	/*
	 * We can do this unlocked here. The futex code uses this flag
	 * just to verify whether the pi state cleanup has been done
	 * or not. In the worst case it loops once more.
	 */
	tsk->flags |= PF_EXITPIDONE;
L
Linus Torvalds 已提交
845

846
	if (tsk->io_context)
847
		exit_io_context(tsk);
848

849
	if (tsk->splice_pipe)
850
		free_pipe_info(tsk->splice_pipe);
851

852 853 854
	if (tsk->task_frag.page)
		put_page(tsk->task_frag.page);

855 856
	validate_creds_for_do_exit(tsk);

C
Coywolf Qi Hunt 已提交
857
	preempt_disable();
858 859
	if (tsk->nr_dirtied)
		__this_cpu_add(dirty_throttle_leaks, tsk->nr_dirtied);
860
	exit_rcu();
861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876

	/*
	 * The setting of TASK_RUNNING by try_to_wake_up() may be delayed
	 * when the following two conditions become true.
	 *   - There is race condition of mmap_sem (It is acquired by
	 *     exit_mm()), and
	 *   - SMI occurs before setting TASK_RUNINNG.
	 *     (or hypervisor of virtual machine switches to other guest)
	 *  As a result, we may become TASK_RUNNING after becoming TASK_DEAD
	 *
	 * To avoid it, we have to wait for releasing tsk->pi_lock which
	 * is held by try_to_wake_up()
	 */
	smp_mb();
	raw_spin_unlock_wait(&tsk->pi_lock);

O
Oleg Nesterov 已提交
877
	/* causes final put_task_struct in finish_task_switch(). */
878
	tsk->state = TASK_DEAD;
879
	tsk->flags |= PF_NOFREEZE;	/* tell freezer to ignore us */
L
Linus Torvalds 已提交
880 881 882
	schedule();
	BUG();
	/* Avoid "noreturn function does return".  */
A
Alan Cox 已提交
883 884
	for (;;)
		cpu_relax();	/* For when BUG is null */
L
Linus Torvalds 已提交
885 886
}

887 888
EXPORT_SYMBOL_GPL(do_exit);

889
void complete_and_exit(struct completion *comp, long code)
L
Linus Torvalds 已提交
890 891 892
{
	if (comp)
		complete(comp);
O
Oleg Nesterov 已提交
893

L
Linus Torvalds 已提交
894 895 896 897 898
	do_exit(code);
}

EXPORT_SYMBOL(complete_and_exit);

899
SYSCALL_DEFINE1(exit, int, error_code)
L
Linus Torvalds 已提交
900 901 902 903 904 905 906 907
{
	do_exit((error_code&0xff)<<8);
}

/*
 * Take down every thread in the group.  This is called by fatal signals
 * as well as by sys_exit_group (below).
 */
908
void
L
Linus Torvalds 已提交
909 910
do_group_exit(int exit_code)
{
911 912
	struct signal_struct *sig = current->signal;

L
Linus Torvalds 已提交
913 914
	BUG_ON(exit_code & 0x80); /* core dumps don't get here */

915 916
	if (signal_group_exit(sig))
		exit_code = sig->group_exit_code;
L
Linus Torvalds 已提交
917 918 919
	else if (!thread_group_empty(current)) {
		struct sighand_struct *const sighand = current->sighand;
		spin_lock_irq(&sighand->siglock);
920
		if (signal_group_exit(sig))
L
Linus Torvalds 已提交
921 922 923 924
			/* Another thread got here before we took the lock.  */
			exit_code = sig->group_exit_code;
		else {
			sig->group_exit_code = exit_code;
925
			sig->flags = SIGNAL_GROUP_EXIT;
L
Linus Torvalds 已提交
926 927 928 929 930 931 932 933 934 935 936 937 938 939
			zap_other_threads(current);
		}
		spin_unlock_irq(&sighand->siglock);
	}

	do_exit(exit_code);
	/* NOTREACHED */
}

/*
 * this kills every thread in the thread group. Note that any externally
 * wait4()-ing process will get the correct exit code - even if this
 * thread is not the thread group leader.
 */
940
SYSCALL_DEFINE1(exit_group, int, error_code)
L
Linus Torvalds 已提交
941 942
{
	do_group_exit((error_code & 0xff) << 8);
943 944
	/* NOTREACHED */
	return 0;
L
Linus Torvalds 已提交
945 946
}

947 948 949
struct wait_opts {
	enum pid_type		wo_type;
	int			wo_flags;
950
	struct pid		*wo_pid;
951 952 953 954 955

	struct siginfo __user	*wo_info;
	int __user		*wo_stat;
	struct rusage __user	*wo_rusage;

956
	wait_queue_t		child_wait;
957 958 959
	int			notask_error;
};

960 961
static inline
struct pid *task_pid_type(struct task_struct *task, enum pid_type type)
E
Eric W. Biederman 已提交
962
{
963 964 965
	if (type != PIDTYPE_PID)
		task = task->group_leader;
	return task->pids[type].pid;
E
Eric W. Biederman 已提交
966 967
}

968
static int eligible_pid(struct wait_opts *wo, struct task_struct *p)
L
Linus Torvalds 已提交
969
{
970 971 972
	return	wo->wo_type == PIDTYPE_MAX ||
		task_pid_type(p, wo->wo_type) == wo->wo_pid;
}
L
Linus Torvalds 已提交
973

974 975 976 977
static int eligible_child(struct wait_opts *wo, struct task_struct *p)
{
	if (!eligible_pid(wo, p))
		return 0;
L
Linus Torvalds 已提交
978 979 980 981 982
	/* Wait for all children (clone and not) if __WALL is set;
	 * otherwise, wait for clone children *only* if __WCLONE is
	 * set; otherwise, wait for non-clone children *only*.  (Note:
	 * A "clone" child here is one that reports to its parent
	 * using a signal other than SIGCHLD.) */
983 984
	if (((p->exit_signal != SIGCHLD) ^ !!(wo->wo_flags & __WCLONE))
	    && !(wo->wo_flags & __WALL))
L
Linus Torvalds 已提交
985 986
		return 0;

987
	return 1;
L
Linus Torvalds 已提交
988 989
}

990 991
static int wait_noreap_copyout(struct wait_opts *wo, struct task_struct *p,
				pid_t pid, uid_t uid, int why, int status)
L
Linus Torvalds 已提交
992
{
993 994 995
	struct siginfo __user *infop;
	int retval = wo->wo_rusage
		? getrusage(p, RUSAGE_BOTH, wo->wo_rusage) : 0;
996

L
Linus Torvalds 已提交
997
	put_task_struct(p);
998
	infop = wo->wo_info;
999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012
	if (infop) {
		if (!retval)
			retval = put_user(SIGCHLD, &infop->si_signo);
		if (!retval)
			retval = put_user(0, &infop->si_errno);
		if (!retval)
			retval = put_user((short)why, &infop->si_code);
		if (!retval)
			retval = put_user(pid, &infop->si_pid);
		if (!retval)
			retval = put_user(uid, &infop->si_uid);
		if (!retval)
			retval = put_user(status, &infop->si_status);
	}
L
Linus Torvalds 已提交
1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023
	if (!retval)
		retval = pid;
	return retval;
}

/*
 * Handle sys_wait4 work for one task in state EXIT_ZOMBIE.  We hold
 * read_lock(&tasklist_lock) on entry.  If we return zero, we still hold
 * the lock and this task is uninteresting.  If we return nonzero, we have
 * released the lock and the system call should return.
 */
1024
static int wait_task_zombie(struct wait_opts *wo, struct task_struct *p)
L
Linus Torvalds 已提交
1025 1026
{
	unsigned long state;
1027
	int retval, status, traced;
1028
	pid_t pid = task_pid_vnr(p);
1029
	uid_t uid = from_kuid_munged(current_user_ns(), task_uid(p));
1030
	struct siginfo __user *infop;
L
Linus Torvalds 已提交
1031

1032
	if (!likely(wo->wo_flags & WEXITED))
R
Roland McGrath 已提交
1033 1034
		return 0;

1035
	if (unlikely(wo->wo_flags & WNOWAIT)) {
L
Linus Torvalds 已提交
1036
		int exit_code = p->exit_code;
1037
		int why;
L
Linus Torvalds 已提交
1038 1039 1040 1041 1042 1043 1044 1045 1046 1047

		get_task_struct(p);
		read_unlock(&tasklist_lock);
		if ((exit_code & 0x7f) == 0) {
			why = CLD_EXITED;
			status = exit_code >> 8;
		} else {
			why = (exit_code & 0x80) ? CLD_DUMPED : CLD_KILLED;
			status = exit_code & 0x7f;
		}
1048
		return wait_noreap_copyout(wo, p, pid, uid, why, status);
L
Linus Torvalds 已提交
1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060
	}

	/*
	 * Try to move the task's state to DEAD
	 * only one thread is allowed to do this:
	 */
	state = xchg(&p->exit_state, EXIT_DEAD);
	if (state != EXIT_ZOMBIE) {
		BUG_ON(state != EXIT_DEAD);
		return 0;
	}

1061
	traced = ptrace_reparented(p);
1062 1063
	/*
	 * It can be ptraced but not reparented, check
O
Oleg Nesterov 已提交
1064
	 * thread_group_leader() to filter out sub-threads.
1065
	 */
O
Oleg Nesterov 已提交
1066
	if (likely(!traced) && thread_group_leader(p)) {
1067 1068
		struct signal_struct *psig;
		struct signal_struct *sig;
J
Jiri Pirko 已提交
1069
		unsigned long maxrss;
1070
		cputime_t tgutime, tgstime;
1071

L
Linus Torvalds 已提交
1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082
		/*
		 * The resource counters for the group leader are in its
		 * own task_struct.  Those for dead threads in the group
		 * are in its signal_struct, as are those for the child
		 * processes it has previously reaped.  All these
		 * accumulate in the parent's signal_struct c* fields.
		 *
		 * We don't bother to take a lock here to protect these
		 * p->signal fields, because they are only touched by
		 * __exit_signal, which runs with tasklist_lock
		 * write-locked anyway, and so is excluded here.  We do
1083
		 * need to protect the access to parent->signal fields,
L
Linus Torvalds 已提交
1084 1085
		 * as other threads in the parent group can be right
		 * here reaping other children at the same time.
1086
		 *
1087
		 * We use thread_group_cputime_adjusted() to get times for the thread
1088 1089
		 * group, which consolidates times for all threads in the
		 * group including the group leader.
L
Linus Torvalds 已提交
1090
		 */
1091
		thread_group_cputime_adjusted(p, &tgutime, &tgstime);
1092 1093
		spin_lock_irq(&p->real_parent->sighand->siglock);
		psig = p->real_parent->signal;
1094
		sig = p->signal;
1095 1096
		psig->cutime += tgutime + sig->cutime;
		psig->cstime += tgstime + sig->cstime;
1097
		psig->cgtime += task_gtime(p) + sig->gtime + sig->cgtime;
1098 1099 1100 1101 1102 1103 1104 1105
		psig->cmin_flt +=
			p->min_flt + sig->min_flt + sig->cmin_flt;
		psig->cmaj_flt +=
			p->maj_flt + sig->maj_flt + sig->cmaj_flt;
		psig->cnvcsw +=
			p->nvcsw + sig->nvcsw + sig->cnvcsw;
		psig->cnivcsw +=
			p->nivcsw + sig->nivcsw + sig->cnivcsw;
1106 1107 1108 1109 1110 1111
		psig->cinblock +=
			task_io_get_inblock(p) +
			sig->inblock + sig->cinblock;
		psig->coublock +=
			task_io_get_oublock(p) +
			sig->oublock + sig->coublock;
J
Jiri Pirko 已提交
1112 1113 1114
		maxrss = max(sig->maxrss, sig->cmaxrss);
		if (psig->cmaxrss < maxrss)
			psig->cmaxrss = maxrss;
1115 1116
		task_io_accounting_add(&psig->ioac, &p->ioac);
		task_io_accounting_add(&psig->ioac, &sig->ioac);
1117
		spin_unlock_irq(&p->real_parent->sighand->siglock);
L
Linus Torvalds 已提交
1118 1119 1120 1121 1122 1123 1124 1125
	}

	/*
	 * Now we are sure this task is interesting, and no other
	 * thread can reap it because we set its state to EXIT_DEAD.
	 */
	read_unlock(&tasklist_lock);

1126 1127
	retval = wo->wo_rusage
		? getrusage(p, RUSAGE_BOTH, wo->wo_rusage) : 0;
L
Linus Torvalds 已提交
1128 1129
	status = (p->signal->flags & SIGNAL_GROUP_EXIT)
		? p->signal->group_exit_code : p->exit_code;
1130 1131 1132 1133
	if (!retval && wo->wo_stat)
		retval = put_user(status, wo->wo_stat);

	infop = wo->wo_info;
L
Linus Torvalds 已提交
1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152
	if (!retval && infop)
		retval = put_user(SIGCHLD, &infop->si_signo);
	if (!retval && infop)
		retval = put_user(0, &infop->si_errno);
	if (!retval && infop) {
		int why;

		if ((status & 0x7f) == 0) {
			why = CLD_EXITED;
			status >>= 8;
		} else {
			why = (status & 0x80) ? CLD_DUMPED : CLD_KILLED;
			status &= 0x7f;
		}
		retval = put_user((short)why, &infop->si_code);
		if (!retval)
			retval = put_user(status, &infop->si_status);
	}
	if (!retval && infop)
1153
		retval = put_user(pid, &infop->si_pid);
L
Linus Torvalds 已提交
1154
	if (!retval && infop)
1155
		retval = put_user(uid, &infop->si_uid);
1156
	if (!retval)
1157
		retval = pid;
1158 1159

	if (traced) {
L
Linus Torvalds 已提交
1160
		write_lock_irq(&tasklist_lock);
1161 1162 1163
		/* We dropped tasklist, ptracer could die and untrace */
		ptrace_unlink(p);
		/*
1164 1165
		 * If this is not a sub-thread, notify the parent.
		 * If parent wants a zombie, don't release it now.
1166
		 */
1167 1168 1169 1170
		if (thread_group_leader(p) &&
		    !do_notify_parent(p, p->exit_signal)) {
			p->exit_state = EXIT_ZOMBIE;
			p = NULL;
L
Linus Torvalds 已提交
1171 1172 1173 1174 1175
		}
		write_unlock_irq(&tasklist_lock);
	}
	if (p != NULL)
		release_task(p);
1176

L
Linus Torvalds 已提交
1177 1178 1179
	return retval;
}

1180 1181 1182
static int *task_stopped_code(struct task_struct *p, bool ptrace)
{
	if (ptrace) {
T
Tejun Heo 已提交
1183 1184
		if (task_is_stopped_or_traced(p) &&
		    !(p->jobctl & JOBCTL_LISTENING))
1185 1186 1187 1188 1189 1190 1191 1192
			return &p->exit_code;
	} else {
		if (p->signal->flags & SIGNAL_STOP_STOPPED)
			return &p->signal->group_exit_code;
	}
	return NULL;
}

1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209
/**
 * wait_task_stopped - Wait for %TASK_STOPPED or %TASK_TRACED
 * @wo: wait options
 * @ptrace: is the wait for ptrace
 * @p: task to wait for
 *
 * Handle sys_wait4() work for %p in state %TASK_STOPPED or %TASK_TRACED.
 *
 * CONTEXT:
 * read_lock(&tasklist_lock), which is released if return value is
 * non-zero.  Also, grabs and releases @p->sighand->siglock.
 *
 * RETURNS:
 * 0 if wait condition didn't exist and search for other wait conditions
 * should continue.  Non-zero return, -errno on failure and @p's pid on
 * success, implies that tasklist_lock is released and wait condition
 * search should terminate.
L
Linus Torvalds 已提交
1210
 */
1211 1212
static int wait_task_stopped(struct wait_opts *wo,
				int ptrace, struct task_struct *p)
L
Linus Torvalds 已提交
1213
{
1214
	struct siginfo __user *infop;
1215
	int retval, exit_code, *p_code, why;
1216
	uid_t uid = 0; /* unneeded, required by compiler */
1217
	pid_t pid;
L
Linus Torvalds 已提交
1218

1219 1220 1221
	/*
	 * Traditionally we see ptrace'd stopped tasks regardless of options.
	 */
1222
	if (!ptrace && !(wo->wo_flags & WUNTRACED))
R
Roland McGrath 已提交
1223 1224
		return 0;

1225 1226 1227
	if (!task_stopped_code(p, ptrace))
		return 0;

1228 1229 1230
	exit_code = 0;
	spin_lock_irq(&p->sighand->siglock);

1231 1232
	p_code = task_stopped_code(p, ptrace);
	if (unlikely(!p_code))
1233 1234
		goto unlock_sig;

1235
	exit_code = *p_code;
1236 1237 1238
	if (!exit_code)
		goto unlock_sig;

1239
	if (!unlikely(wo->wo_flags & WNOWAIT))
1240
		*p_code = 0;
1241

1242
	uid = from_kuid_munged(current_user_ns(), task_uid(p));
1243 1244 1245
unlock_sig:
	spin_unlock_irq(&p->sighand->siglock);
	if (!exit_code)
L
Linus Torvalds 已提交
1246 1247 1248 1249 1250 1251 1252 1253 1254 1255
		return 0;

	/*
	 * Now we are pretty sure this task is interesting.
	 * Make sure it doesn't get reaped out from under us while we
	 * give up the lock and then examine it below.  We don't want to
	 * keep holding onto the tasklist_lock while we call getrusage and
	 * possibly take page faults for user memory.
	 */
	get_task_struct(p);
1256
	pid = task_pid_vnr(p);
R
Roland McGrath 已提交
1257
	why = ptrace ? CLD_TRAPPED : CLD_STOPPED;
L
Linus Torvalds 已提交
1258 1259
	read_unlock(&tasklist_lock);

1260 1261 1262 1263 1264 1265 1266
	if (unlikely(wo->wo_flags & WNOWAIT))
		return wait_noreap_copyout(wo, p, pid, uid, why, exit_code);

	retval = wo->wo_rusage
		? getrusage(p, RUSAGE_BOTH, wo->wo_rusage) : 0;
	if (!retval && wo->wo_stat)
		retval = put_user((exit_code << 8) | 0x7f, wo->wo_stat);
L
Linus Torvalds 已提交
1267

1268
	infop = wo->wo_info;
L
Linus Torvalds 已提交
1269 1270 1271 1272 1273
	if (!retval && infop)
		retval = put_user(SIGCHLD, &infop->si_signo);
	if (!retval && infop)
		retval = put_user(0, &infop->si_errno);
	if (!retval && infop)
R
Roland McGrath 已提交
1274
		retval = put_user((short)why, &infop->si_code);
L
Linus Torvalds 已提交
1275 1276 1277
	if (!retval && infop)
		retval = put_user(exit_code, &infop->si_status);
	if (!retval && infop)
1278
		retval = put_user(pid, &infop->si_pid);
L
Linus Torvalds 已提交
1279
	if (!retval && infop)
1280
		retval = put_user(uid, &infop->si_uid);
L
Linus Torvalds 已提交
1281
	if (!retval)
1282
		retval = pid;
L
Linus Torvalds 已提交
1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294
	put_task_struct(p);

	BUG_ON(!retval);
	return retval;
}

/*
 * Handle do_wait work for one task in a live, non-stopped state.
 * read_lock(&tasklist_lock) on entry.  If we return zero, we still hold
 * the lock and this task is uninteresting.  If we return nonzero, we have
 * released the lock and the system call should return.
 */
1295
static int wait_task_continued(struct wait_opts *wo, struct task_struct *p)
L
Linus Torvalds 已提交
1296 1297 1298 1299 1300
{
	int retval;
	pid_t pid;
	uid_t uid;

1301
	if (!unlikely(wo->wo_flags & WCONTINUED))
R
Roland McGrath 已提交
1302 1303
		return 0;

L
Linus Torvalds 已提交
1304 1305 1306 1307 1308 1309 1310 1311 1312
	if (!(p->signal->flags & SIGNAL_STOP_CONTINUED))
		return 0;

	spin_lock_irq(&p->sighand->siglock);
	/* Re-check with the lock held.  */
	if (!(p->signal->flags & SIGNAL_STOP_CONTINUED)) {
		spin_unlock_irq(&p->sighand->siglock);
		return 0;
	}
1313
	if (!unlikely(wo->wo_flags & WNOWAIT))
L
Linus Torvalds 已提交
1314
		p->signal->flags &= ~SIGNAL_STOP_CONTINUED;
1315
	uid = from_kuid_munged(current_user_ns(), task_uid(p));
L
Linus Torvalds 已提交
1316 1317
	spin_unlock_irq(&p->sighand->siglock);

1318
	pid = task_pid_vnr(p);
L
Linus Torvalds 已提交
1319 1320 1321
	get_task_struct(p);
	read_unlock(&tasklist_lock);

1322 1323 1324
	if (!wo->wo_info) {
		retval = wo->wo_rusage
			? getrusage(p, RUSAGE_BOTH, wo->wo_rusage) : 0;
L
Linus Torvalds 已提交
1325
		put_task_struct(p);
1326 1327
		if (!retval && wo->wo_stat)
			retval = put_user(0xffff, wo->wo_stat);
L
Linus Torvalds 已提交
1328
		if (!retval)
1329
			retval = pid;
L
Linus Torvalds 已提交
1330
	} else {
1331 1332
		retval = wait_noreap_copyout(wo, p, pid, uid,
					     CLD_CONTINUED, SIGCONT);
L
Linus Torvalds 已提交
1333 1334 1335 1336 1337 1338
		BUG_ON(retval == 0);
	}

	return retval;
}

R
Roland McGrath 已提交
1339 1340 1341
/*
 * Consider @p for a wait by @parent.
 *
1342
 * -ECHILD should be in ->notask_error before the first call.
R
Roland McGrath 已提交
1343 1344
 * Returns nonzero for a final return, when we have unlocked tasklist_lock.
 * Returns zero if the search for a child should continue;
1345
 * then ->notask_error is 0 if @p is an eligible child,
1346
 * or another error from security_task_wait(), or still -ECHILD.
R
Roland McGrath 已提交
1347
 */
1348 1349
static int wait_consider_task(struct wait_opts *wo, int ptrace,
				struct task_struct *p)
R
Roland McGrath 已提交
1350
{
1351
	int ret = eligible_child(wo, p);
1352
	if (!ret)
R
Roland McGrath 已提交
1353 1354
		return ret;

1355
	ret = security_task_wait(p);
1356 1357 1358 1359 1360 1361 1362 1363
	if (unlikely(ret < 0)) {
		/*
		 * If we have not yet seen any eligible child,
		 * then let this error code replace -ECHILD.
		 * A permission error will give the user a clue
		 * to look for security policy problems, rather
		 * than for mysterious wait bugs.
		 */
1364 1365
		if (wo->notask_error)
			wo->notask_error = ret;
1366
		return 0;
1367 1368
	}

1369
	/* dead body doesn't have much to contribute */
1370 1371 1372 1373 1374 1375 1376
	if (unlikely(p->exit_state == EXIT_DEAD)) {
		/*
		 * But do not ignore this task until the tracer does
		 * wait_task_zombie()->do_notify_parent().
		 */
		if (likely(!ptrace) && unlikely(ptrace_reparented(p)))
			wo->notask_error = 0;
1377
		return 0;
1378
	}
1379

1380 1381
	/* slay zombie? */
	if (p->exit_state == EXIT_ZOMBIE) {
R
Roland McGrath 已提交
1382
		/*
1383 1384 1385
		 * A zombie ptracee is only visible to its ptracer.
		 * Notification and reaping will be cascaded to the real
		 * parent when the ptracer detaches.
R
Roland McGrath 已提交
1386
		 */
T
Tejun Heo 已提交
1387
		if (likely(!ptrace) && unlikely(p->ptrace)) {
1388 1389 1390 1391
			/* it will become visible, clear notask_error */
			wo->notask_error = 0;
			return 0;
		}
R
Roland McGrath 已提交
1392

1393 1394 1395
		/* we don't reap group leaders with subthreads */
		if (!delay_group_leader(p))
			return wait_task_zombie(wo, p);
R
Roland McGrath 已提交
1396

R
Roland McGrath 已提交
1397
		/*
1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419
		 * Allow access to stopped/continued state via zombie by
		 * falling through.  Clearing of notask_error is complex.
		 *
		 * When !@ptrace:
		 *
		 * If WEXITED is set, notask_error should naturally be
		 * cleared.  If not, subset of WSTOPPED|WCONTINUED is set,
		 * so, if there are live subthreads, there are events to
		 * wait for.  If all subthreads are dead, it's still safe
		 * to clear - this function will be called again in finite
		 * amount time once all the subthreads are released and
		 * will then return without clearing.
		 *
		 * When @ptrace:
		 *
		 * Stopped state is per-task and thus can't change once the
		 * target task dies.  Only continued and exited can happen.
		 * Clear notask_error if WCONTINUED | WEXITED.
		 */
		if (likely(!ptrace) || (wo->wo_flags & (WCONTINUED | WEXITED)))
			wo->notask_error = 0;
	} else {
1420 1421 1422 1423 1424 1425 1426 1427 1428 1429
		/*
		 * If @p is ptraced by a task in its real parent's group,
		 * hide group stop/continued state when looking at @p as
		 * the real parent; otherwise, a single stop can be
		 * reported twice as group and ptrace stops.
		 *
		 * If a ptracer wants to distinguish the two events for its
		 * own children, it should create a separate process which
		 * takes the role of real parent.
		 */
1430
		if (likely(!ptrace) && p->ptrace && !ptrace_reparented(p))
1431 1432
			return 0;

1433 1434 1435
		/*
		 * @p is alive and it's gonna stop, continue or exit, so
		 * there always is something to wait for.
R
Roland McGrath 已提交
1436
		 */
1437
		wo->notask_error = 0;
R
Roland McGrath 已提交
1438 1439
	}

R
Roland McGrath 已提交
1440
	/*
1441 1442
	 * Wait for stopped.  Depending on @ptrace, different stopped state
	 * is used and the two don't interact with each other.
R
Roland McGrath 已提交
1443
	 */
1444 1445 1446
	ret = wait_task_stopped(wo, ptrace, p);
	if (ret)
		return ret;
R
Roland McGrath 已提交
1447 1448

	/*
1449 1450 1451
	 * Wait for continued.  There's only one continued state and the
	 * ptracer can consume it which can confuse the real parent.  Don't
	 * use WCONTINUED from ptracer.  You don't need or want it.
R
Roland McGrath 已提交
1452
	 */
1453
	return wait_task_continued(wo, p);
R
Roland McGrath 已提交
1454 1455 1456 1457 1458
}

/*
 * Do the work of do_wait() for one thread in the group, @tsk.
 *
1459
 * -ECHILD should be in ->notask_error before the first call.
R
Roland McGrath 已提交
1460 1461
 * Returns nonzero for a final return, when we have unlocked tasklist_lock.
 * Returns zero if the search for a child should continue; then
1462
 * ->notask_error is 0 if there were any eligible children,
1463
 * or another error from security_task_wait(), or still -ECHILD.
R
Roland McGrath 已提交
1464
 */
1465
static int do_wait_thread(struct wait_opts *wo, struct task_struct *tsk)
R
Roland McGrath 已提交
1466 1467 1468 1469
{
	struct task_struct *p;

	list_for_each_entry(p, &tsk->children, sibling) {
1470 1471 1472
		int ret = wait_consider_task(wo, 0, p);
		if (ret)
			return ret;
R
Roland McGrath 已提交
1473 1474 1475 1476 1477
	}

	return 0;
}

1478
static int ptrace_do_wait(struct wait_opts *wo, struct task_struct *tsk)
R
Roland McGrath 已提交
1479 1480 1481
{
	struct task_struct *p;

R
Roland McGrath 已提交
1482
	list_for_each_entry(p, &tsk->ptraced, ptrace_entry) {
1483
		int ret = wait_consider_task(wo, 1, p);
R
Roland McGrath 已提交
1484
		if (ret)
R
Roland McGrath 已提交
1485 1486 1487 1488 1489 1490
			return ret;
	}

	return 0;
}

1491 1492 1493 1494 1495 1496 1497
static int child_wait_callback(wait_queue_t *wait, unsigned mode,
				int sync, void *key)
{
	struct wait_opts *wo = container_of(wait, struct wait_opts,
						child_wait);
	struct task_struct *p = key;

1498
	if (!eligible_pid(wo, p))
1499 1500
		return 0;

1501 1502 1503
	if ((wo->wo_flags & __WNOTHREAD) && wait->private != p->parent)
		return 0;

1504 1505 1506
	return default_wake_function(wait, mode, sync, key);
}

1507 1508
void __wake_up_parent(struct task_struct *p, struct task_struct *parent)
{
1509 1510
	__wake_up_sync_key(&parent->signal->wait_chldexit,
				TASK_INTERRUPTIBLE, 1, p);
1511 1512
}

1513
static long do_wait(struct wait_opts *wo)
L
Linus Torvalds 已提交
1514 1515
{
	struct task_struct *tsk;
R
Roland McGrath 已提交
1516
	int retval;
L
Linus Torvalds 已提交
1517

1518
	trace_sched_process_wait(wo->wo_pid);
1519

1520 1521 1522
	init_waitqueue_func_entry(&wo->child_wait, child_wait_callback);
	wo->child_wait.private = current;
	add_wait_queue(&current->signal->wait_chldexit, &wo->child_wait);
L
Linus Torvalds 已提交
1523
repeat:
R
Roland McGrath 已提交
1524 1525
	/*
	 * If there is nothing that can match our critiera just get out.
1526 1527 1528
	 * We will clear ->notask_error to zero if we see any child that
	 * might later match our criteria, even if we are not able to reap
	 * it yet.
R
Roland McGrath 已提交
1529
	 */
1530
	wo->notask_error = -ECHILD;
1531 1532
	if ((wo->wo_type < PIDTYPE_MAX) &&
	   (!wo->wo_pid || hlist_empty(&wo->wo_pid->tasks[wo->wo_type])))
1533
		goto notask;
E
Eric W. Biederman 已提交
1534

1535
	set_current_state(TASK_INTERRUPTIBLE);
L
Linus Torvalds 已提交
1536 1537 1538
	read_lock(&tasklist_lock);
	tsk = current;
	do {
1539 1540 1541
		retval = do_wait_thread(wo, tsk);
		if (retval)
			goto end;
1542

1543 1544
		retval = ptrace_do_wait(wo, tsk);
		if (retval)
R
Roland McGrath 已提交
1545 1546
			goto end;

1547
		if (wo->wo_flags & __WNOTHREAD)
L
Linus Torvalds 已提交
1548
			break;
1549
	} while_each_thread(current, tsk);
L
Linus Torvalds 已提交
1550
	read_unlock(&tasklist_lock);
O
Oleg Nesterov 已提交
1551

1552
notask:
1553 1554
	retval = wo->notask_error;
	if (!retval && !(wo->wo_flags & WNOHANG)) {
L
Linus Torvalds 已提交
1555
		retval = -ERESTARTSYS;
R
Roland McGrath 已提交
1556 1557 1558 1559
		if (!signal_pending(current)) {
			schedule();
			goto repeat;
		}
L
Linus Torvalds 已提交
1560 1561
	}
end:
1562
	__set_current_state(TASK_RUNNING);
1563
	remove_wait_queue(&current->signal->wait_chldexit, &wo->child_wait);
L
Linus Torvalds 已提交
1564 1565 1566
	return retval;
}

1567 1568
SYSCALL_DEFINE5(waitid, int, which, pid_t, upid, struct siginfo __user *,
		infop, int, options, struct rusage __user *, ru)
L
Linus Torvalds 已提交
1569
{
1570
	struct wait_opts wo;
E
Eric W. Biederman 已提交
1571 1572
	struct pid *pid = NULL;
	enum pid_type type;
L
Linus Torvalds 已提交
1573 1574 1575 1576 1577 1578 1579 1580 1581
	long ret;

	if (options & ~(WNOHANG|WNOWAIT|WEXITED|WSTOPPED|WCONTINUED))
		return -EINVAL;
	if (!(options & (WEXITED|WSTOPPED|WCONTINUED)))
		return -EINVAL;

	switch (which) {
	case P_ALL:
E
Eric W. Biederman 已提交
1582
		type = PIDTYPE_MAX;
L
Linus Torvalds 已提交
1583 1584
		break;
	case P_PID:
E
Eric W. Biederman 已提交
1585 1586
		type = PIDTYPE_PID;
		if (upid <= 0)
L
Linus Torvalds 已提交
1587 1588 1589
			return -EINVAL;
		break;
	case P_PGID:
E
Eric W. Biederman 已提交
1590 1591
		type = PIDTYPE_PGID;
		if (upid <= 0)
L
Linus Torvalds 已提交
1592 1593 1594 1595 1596 1597
			return -EINVAL;
		break;
	default:
		return -EINVAL;
	}

E
Eric W. Biederman 已提交
1598 1599
	if (type < PIDTYPE_MAX)
		pid = find_get_pid(upid);
1600 1601 1602 1603 1604 1605 1606 1607

	wo.wo_type	= type;
	wo.wo_pid	= pid;
	wo.wo_flags	= options;
	wo.wo_info	= infop;
	wo.wo_stat	= NULL;
	wo.wo_rusage	= ru;
	ret = do_wait(&wo);
1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630

	if (ret > 0) {
		ret = 0;
	} else if (infop) {
		/*
		 * For a WNOHANG return, clear out all the fields
		 * we would set so the user can easily tell the
		 * difference.
		 */
		if (!ret)
			ret = put_user(0, &infop->si_signo);
		if (!ret)
			ret = put_user(0, &infop->si_errno);
		if (!ret)
			ret = put_user(0, &infop->si_code);
		if (!ret)
			ret = put_user(0, &infop->si_pid);
		if (!ret)
			ret = put_user(0, &infop->si_uid);
		if (!ret)
			ret = put_user(0, &infop->si_status);
	}

E
Eric W. Biederman 已提交
1631
	put_pid(pid);
L
Linus Torvalds 已提交
1632 1633 1634
	return ret;
}

1635 1636
SYSCALL_DEFINE4(wait4, pid_t, upid, int __user *, stat_addr,
		int, options, struct rusage __user *, ru)
L
Linus Torvalds 已提交
1637
{
1638
	struct wait_opts wo;
E
Eric W. Biederman 已提交
1639 1640
	struct pid *pid = NULL;
	enum pid_type type;
L
Linus Torvalds 已提交
1641 1642 1643 1644 1645
	long ret;

	if (options & ~(WNOHANG|WUNTRACED|WCONTINUED|
			__WNOTHREAD|__WCLONE|__WALL))
		return -EINVAL;
E
Eric W. Biederman 已提交
1646 1647 1648 1649 1650 1651 1652 1653

	if (upid == -1)
		type = PIDTYPE_MAX;
	else if (upid < 0) {
		type = PIDTYPE_PGID;
		pid = find_get_pid(-upid);
	} else if (upid == 0) {
		type = PIDTYPE_PGID;
1654
		pid = get_task_pid(current, PIDTYPE_PGID);
E
Eric W. Biederman 已提交
1655 1656 1657 1658 1659
	} else /* upid > 0 */ {
		type = PIDTYPE_PID;
		pid = find_get_pid(upid);
	}

1660 1661 1662 1663 1664 1665 1666
	wo.wo_type	= type;
	wo.wo_pid	= pid;
	wo.wo_flags	= options | WEXITED;
	wo.wo_info	= NULL;
	wo.wo_stat	= stat_addr;
	wo.wo_rusage	= ru;
	ret = do_wait(&wo);
E
Eric W. Biederman 已提交
1667
	put_pid(pid);
L
Linus Torvalds 已提交
1668 1669 1670 1671 1672 1673 1674 1675 1676 1677

	return ret;
}

#ifdef __ARCH_WANT_SYS_WAITPID

/*
 * sys_waitpid() remains for compatibility. waitpid() should be
 * implemented by calling sys_wait4() from libc.a.
 */
1678
SYSCALL_DEFINE3(waitpid, pid_t, pid, int __user *, stat_addr, int, options)
L
Linus Torvalds 已提交
1679 1680 1681 1682 1683
{
	return sys_wait4(pid, stat_addr, options, NULL);
}

#endif