exit.c 42.7 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>
L
Linus Torvalds 已提交
23
#include <linux/binfmts.h>
S
Serge E. Hallyn 已提交
24
#include <linux/nsproxy.h>
25
#include <linux/pid_namespace.h>
L
Linus Torvalds 已提交
26 27 28 29
#include <linux/ptrace.h>
#include <linux/profile.h>
#include <linux/mount.h>
#include <linux/proc_fs.h>
30
#include <linux/kthread.h>
L
Linus Torvalds 已提交
31
#include <linux/mempolicy.h>
32
#include <linux/taskstats_kern.h>
33
#include <linux/delayacct.h>
34
#include <linux/freezer.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>
L
Linus Torvalds 已提交
52 53 54 55 56

#include <asm/uaccess.h>
#include <asm/unistd.h>
#include <asm/pgtable.h>
#include <asm/mmu_context.h>
D
David Howells 已提交
57
#include "cred-internals.h"
L
Linus Torvalds 已提交
58

A
Adrian Bunk 已提交
59 60
static void exit_mm(struct task_struct * tsk);

L
Linus Torvalds 已提交
61 62 63 64 65 66 67
static void __unhash_process(struct task_struct *p)
{
	nr_threads--;
	detach_pid(p, PIDTYPE_PID);
	if (thread_group_leader(p)) {
		detach_pid(p, PIDTYPE_PGID);
		detach_pid(p, PIDTYPE_SID);
68

69
		list_del_rcu(&p->tasks);
70
		__get_cpu_var(process_counts)--;
L
Linus Torvalds 已提交
71
	}
O
Oleg Nesterov 已提交
72
	list_del_rcu(&p->thread_group);
R
Roland McGrath 已提交
73
	list_del_init(&p->sibling);
L
Linus Torvalds 已提交
74 75
}

76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97
/*
 * This function expects the tasklist_lock write-locked.
 */
static void __exit_signal(struct task_struct *tsk)
{
	struct signal_struct *sig = tsk->signal;
	struct sighand_struct *sighand;

	BUG_ON(!sig);
	BUG_ON(!atomic_read(&sig->count));

	sighand = rcu_dereference(tsk->sighand);
	spin_lock(&sighand->siglock);

	posix_cpu_timers_exit(tsk);
	if (atomic_dec_and_test(&sig->count))
		posix_cpu_timers_exit_group(tsk);
	else {
		/*
		 * If there is any task waiting for the group exit
		 * then notify it:
		 */
98
		if (sig->group_exit_task && atomic_read(&sig->count) == sig->notify_count)
99
			wake_up_process(sig->group_exit_task);
100

101 102 103 104 105 106 107 108 109 110 111 112
		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.
		 */
113 114
		sig->utime = cputime_add(sig->utime, task_utime(tsk));
		sig->stime = cputime_add(sig->stime, task_stime(tsk));
115
		sig->gtime = cputime_add(sig->gtime, task_gtime(tsk));
116 117 118 119
		sig->min_flt += tsk->min_flt;
		sig->maj_flt += tsk->maj_flt;
		sig->nvcsw += tsk->nvcsw;
		sig->nivcsw += tsk->nivcsw;
120 121
		sig->inblock += task_io_get_inblock(tsk);
		sig->oublock += task_io_get_oublock(tsk);
122
		task_io_accounting_add(&sig->ioac, &tsk->ioac);
123
		sig->sum_sched_runtime += tsk->se.sum_exec_runtime;
124 125 126
		sig = NULL; /* Marker for below. */
	}

127 128
	__unhash_process(tsk);

129 130 131 132 133 134
	/*
	 * Do this under ->siglock, we can race with another thread
	 * doing sigqueue_free() if we have SIGQUEUE_PREALLOC signals.
	 */
	flush_sigqueue(&tsk->pending);

135
	tsk->signal = NULL;
136
	tsk->sighand = NULL;
137 138
	spin_unlock(&sighand->siglock);

139
	__cleanup_sighand(sighand);
140 141 142
	clear_tsk_thread_flag(tsk,TIF_SIGPENDING);
	if (sig) {
		flush_sigqueue(&sig->shared_pending);
143
		taskstats_tgid_free(sig);
144 145 146 147 148
		/*
		 * Make sure ->signal can't go away under rq->lock,
		 * see account_group_exec_runtime().
		 */
		task_rq_unlock_wait(tsk);
149 150 151 152
		__cleanup_signal(sig);
	}
}

153 154
static void delayed_put_task_struct(struct rcu_head *rhp)
{
155 156
	struct task_struct *tsk = container_of(rhp, struct task_struct, rcu);

157 158
#ifdef CONFIG_PERF_EVENTS
	WARN_ON_ONCE(tsk->perf_event_ctxp);
159
#endif
160 161
	trace_sched_process_free(tsk);
	put_task_struct(tsk);
162 163
}

R
Roland McGrath 已提交
164

L
Linus Torvalds 已提交
165 166
void release_task(struct task_struct * p)
{
167
	struct task_struct *leader;
L
Linus Torvalds 已提交
168
	int zap_leader;
169
repeat:
R
Roland McGrath 已提交
170
	tracehook_prepare_release_task(p);
171 172 173 174
	/* don't need to get the RCU readlock here - the process is dead and
	 * can't be modifying its own credentials */
	atomic_dec(&__task_cred(p)->user->processes);

175
	proc_flush_task(p);
176

L
Linus Torvalds 已提交
177
	write_lock_irq(&tasklist_lock);
R
Roland McGrath 已提交
178
	tracehook_finish_release_task(p);
L
Linus Torvalds 已提交
179
	__exit_signal(p);
180

L
Linus Torvalds 已提交
181 182 183 184 185 186 187 188
	/*
	 * 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) {
189
		BUG_ON(task_detached(leader));
L
Linus Torvalds 已提交
190 191 192 193 194 195 196 197 198
		do_notify_parent(leader, leader->exit_signal);
		/*
		 * 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.
		 *
		 * do_notify_parent() will have marked it self-reaping in
		 * that case.
		 */
199
		zap_leader = task_detached(leader);
R
Roland McGrath 已提交
200 201 202 203 204 205 206

		/*
		 * This maintains the invariant that release_task()
		 * only runs on a task in EXIT_DEAD, just for sanity.
		 */
		if (zap_leader)
			leader->exit_state = EXIT_DEAD;
L
Linus Torvalds 已提交
207 208 209 210
	}

	write_unlock_irq(&tasklist_lock);
	release_thread(p);
211
	call_rcu(&p->rcu, delayed_put_task_struct);
L
Linus Torvalds 已提交
212 213 214 215 216 217 218 219 220 221

	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...
222 223
 *
 * The caller must hold rcu lock or the tasklist lock.
L
Linus Torvalds 已提交
224
 */
225
struct pid *session_of_pgrp(struct pid *pgrp)
L
Linus Torvalds 已提交
226 227
{
	struct task_struct *p;
228
	struct pid *sid = NULL;
229

230
	p = pid_task(pgrp, PIDTYPE_PGID);
231
	if (p == NULL)
232
		p = pid_task(pgrp, PIDTYPE_PID);
233
	if (p != NULL)
234
		sid = task_session(p);
235

L
Linus Torvalds 已提交
236 237 238 239 240 241 242 243 244 245 246
	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?"
 */
247
static int will_become_orphaned_pgrp(struct pid *pgrp, struct task_struct *ignored_task)
L
Linus Torvalds 已提交
248 249 250
{
	struct task_struct *p;

251
	do_each_pid_task(pgrp, PIDTYPE_PGID, p) {
252 253 254
		if ((p == ignored_task) ||
		    (p->exit_state && thread_group_empty(p)) ||
		    is_global_init(p->real_parent))
L
Linus Torvalds 已提交
255
			continue;
256

257
		if (task_pgrp(p->real_parent) != pgrp &&
258 259
		    task_session(p->real_parent) == task_session(p))
			return 0;
260
	} while_each_pid_task(pgrp, PIDTYPE_PGID, p);
261 262

	return 1;
L
Linus Torvalds 已提交
263 264
}

265
int is_current_pgrp_orphaned(void)
L
Linus Torvalds 已提交
266 267 268 269
{
	int retval;

	read_lock(&tasklist_lock);
270
	retval = will_become_orphaned_pgrp(task_pgrp(current), NULL);
L
Linus Torvalds 已提交
271 272 273 274 275
	read_unlock(&tasklist_lock);

	return retval;
}

276
static int has_stopped_jobs(struct pid *pgrp)
L
Linus Torvalds 已提交
277 278 279 280
{
	int retval = 0;
	struct task_struct *p;

281
	do_each_pid_task(pgrp, PIDTYPE_PGID, p) {
M
Matthew Wilcox 已提交
282
		if (!task_is_stopped(p))
L
Linus Torvalds 已提交
283 284 285
			continue;
		retval = 1;
		break;
286
	} while_each_pid_task(pgrp, PIDTYPE_PGID, p);
L
Linus Torvalds 已提交
287 288 289
	return retval;
}

290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320
/*
 * 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);
	}
}

L
Linus Torvalds 已提交
321
/**
322
 * reparent_to_kthreadd - Reparent the calling kernel thread to kthreadd
L
Linus Torvalds 已提交
323 324
 *
 * If a kernel thread is launched as a result of a system call, or if
325 326
 * it ever exits, it should generally reparent itself to kthreadd so it
 * isn't in the way of other processes and is correctly cleaned up on exit.
L
Linus Torvalds 已提交
327 328 329 330
 *
 * The various task state such as scheduling policy and priority may have
 * been inherited from a user process, so we reset them to sane values here.
 *
331
 * NOTE that reparent_to_kthreadd() gives the caller full capabilities.
L
Linus Torvalds 已提交
332
 */
333
static void reparent_to_kthreadd(void)
L
Linus Torvalds 已提交
334 335 336 337 338
{
	write_lock_irq(&tasklist_lock);

	ptrace_unlink(current);
	/* Reparent to init */
339
	current->real_parent = current->parent = kthreadd_task;
R
Roland McGrath 已提交
340
	list_move_tail(&current->sibling, &current->real_parent->children);
L
Linus Torvalds 已提交
341 342 343 344

	/* Set the exit signal to SIGCHLD so we signal init on exit */
	current->exit_signal = SIGCHLD;

345
	if (task_nice(current) < 0)
L
Linus Torvalds 已提交
346 347 348 349 350 351
		set_user_nice(current, 0);
	/* cpus_allowed? */
	/* rt_priority? */
	/* signals? */
	memcpy(current->signal->rlim, init_task.signal->rlim,
	       sizeof(current->signal->rlim));
D
David Howells 已提交
352 353 354

	atomic_inc(&init_cred.usage);
	commit_creds(&init_cred);
L
Linus Torvalds 已提交
355 356 357
	write_unlock_irq(&tasklist_lock);
}

358
void __set_special_pids(struct pid *pid)
L
Linus Torvalds 已提交
359
{
360
	struct task_struct *curr = current->group_leader;
L
Linus Torvalds 已提交
361

362
	if (task_session(curr) != pid)
363
		change_pid(curr, PIDTYPE_SID, pid);
364 365

	if (task_pgrp(curr) != pid)
366
		change_pid(curr, PIDTYPE_PGID, pid);
L
Linus Torvalds 已提交
367 368
}

369
static void set_special_pids(struct pid *pid)
L
Linus Torvalds 已提交
370 371
{
	write_lock_irq(&tasklist_lock);
372
	__set_special_pids(pid);
L
Linus Torvalds 已提交
373 374 375 376
	write_unlock_irq(&tasklist_lock);
}

/*
377 378
 * 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
379 380 381
 */
int allow_signal(int sig)
{
382
	if (!valid_signal(sig) || sig < 1)
L
Linus Torvalds 已提交
383 384 385
		return -EINVAL;

	spin_lock_irq(&current->sighand->siglock);
386
	/* This is only needed for daemonize()'ed kthreads */
L
Linus Torvalds 已提交
387
	sigdelset(&current->blocked, sig);
388 389 390 391 392 393
	/*
	 * 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
394 395 396 397 398 399 400 401 402
	recalc_sigpending();
	spin_unlock_irq(&current->sighand->siglock);
	return 0;
}

EXPORT_SYMBOL(allow_signal);

int disallow_signal(int sig)
{
403
	if (!valid_signal(sig) || sig < 1)
L
Linus Torvalds 已提交
404 405 406
		return -EINVAL;

	spin_lock_irq(&current->sighand->siglock);
407
	current->sighand->action[(sig)-1].sa.sa_handler = SIG_IGN;
L
Linus Torvalds 已提交
408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434
	recalc_sigpending();
	spin_unlock_irq(&current->sighand->siglock);
	return 0;
}

EXPORT_SYMBOL(disallow_signal);

/*
 *	Put all the gunge required to become a kernel thread without
 *	attached user resources in one place where it belongs.
 */

void daemonize(const char *name, ...)
{
	va_list args;
	sigset_t blocked;

	va_start(args, name);
	vsnprintf(current->comm, sizeof(current->comm), name, args);
	va_end(args);

	/*
	 * If we were started as result of loading a module, close all of the
	 * user space pages.  We don't need them, and if we didn't close them
	 * they would be locked into memory.
	 */
	exit_mm(current);
435 436 437 438
	/*
	 * We don't want to have TIF_FREEZE set if the system-wide hibernation
	 * or suspend transition begins right now.
	 */
O
Oleg Nesterov 已提交
439
	current->flags |= (PF_NOFREEZE | PF_KTHREAD);
L
Linus Torvalds 已提交
440

441 442 443 444
	if (current->nsproxy != &init_nsproxy) {
		get_nsproxy(&init_nsproxy);
		switch_task_namespaces(current, &init_nsproxy);
	}
445
	set_special_pids(&init_struct_pid);
446
	proc_clear_tty(current);
L
Linus Torvalds 已提交
447 448 449 450 451 452 453 454

	/* Block and flush all signals */
	sigfillset(&blocked);
	sigprocmask(SIG_BLOCK, &blocked, NULL);
	flush_signals(current);

	/* Become as one with the init task */

455
	daemonize_fs_struct();
456
	exit_files(current);
L
Linus Torvalds 已提交
457 458 459
	current->files = init_task.files;
	atomic_inc(&current->files->count);

460
	reparent_to_kthreadd();
L
Linus Torvalds 已提交
461 462 463 464
}

EXPORT_SYMBOL(daemonize);

465
static void close_files(struct files_struct * files)
L
Linus Torvalds 已提交
466 467
{
	int i, j;
468
	struct fdtable *fdt;
L
Linus Torvalds 已提交
469 470

	j = 0;
471 472 473 474 475 476

	/*
	 * It is safe to dereference the fd table without RCU or
	 * ->file_lock because this is the last reference to the
	 * files structure.
	 */
477
	fdt = files_fdtable(files);
L
Linus Torvalds 已提交
478 479 480
	for (;;) {
		unsigned long set;
		i = j * __NFDBITS;
481
		if (i >= fdt->max_fds)
L
Linus Torvalds 已提交
482
			break;
483
		set = fdt->open_fds->fds_bits[j++];
L
Linus Torvalds 已提交
484 485
		while (set) {
			if (set & 1) {
486
				struct file * file = xchg(&fdt->fd[i], NULL);
487
				if (file) {
L
Linus Torvalds 已提交
488
					filp_close(file, files);
489 490
					cond_resched();
				}
L
Linus Torvalds 已提交
491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510
			}
			i++;
			set >>= 1;
		}
	}
}

struct files_struct *get_files_struct(struct task_struct *task)
{
	struct files_struct *files;

	task_lock(task);
	files = task->files;
	if (files)
		atomic_inc(&files->count);
	task_unlock(task);

	return files;
}

511
void put_files_struct(struct files_struct *files)
L
Linus Torvalds 已提交
512
{
513 514
	struct fdtable *fdt;

L
Linus Torvalds 已提交
515 516 517 518
	if (atomic_dec_and_test(&files->count)) {
		close_files(files);
		/*
		 * Free the fd and fdset arrays if we expanded them.
519 520 521
		 * If the fdtable was embedded, pass files for freeing
		 * at the end of the RCU grace period. Otherwise,
		 * you can free files immediately.
L
Linus Torvalds 已提交
522
		 */
523
		fdt = files_fdtable(files);
524
		if (fdt != &files->fdtab)
525
			kmem_cache_free(files_cachep, files);
526
		free_fdtable(fdt);
L
Linus Torvalds 已提交
527 528 529
	}
}

530
void reset_files_struct(struct files_struct *files)
531
{
532
	struct task_struct *tsk = current;
533 534 535 536 537 538 539 540 541
	struct files_struct *old;

	old = tsk->files;
	task_lock(tsk);
	tsk->files = files;
	task_unlock(tsk);
	put_files_struct(old);
}

542
void exit_files(struct task_struct *tsk)
L
Linus Torvalds 已提交
543 544 545 546 547 548 549 550 551 552 553
{
	struct files_struct * files = tsk->files;

	if (files) {
		task_lock(tsk);
		tsk->files = NULL;
		task_unlock(tsk);
		put_files_struct(files);
	}
}

554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591
#ifdef CONFIG_MM_OWNER
/*
 * Task p is exiting and it owned mm, lets find a new owner for it
 */
static inline int
mm_need_new_owner(struct mm_struct *mm, struct task_struct *p)
{
	/*
	 * If there are other users of the mm and the owner (us) is exiting
	 * we need to find a new owner to take on the responsibility.
	 */
	if (atomic_read(&mm->mm_users) <= 1)
		return 0;
	if (mm->owner != p)
		return 0;
	return 1;
}

void mm_update_next_owner(struct mm_struct *mm)
{
	struct task_struct *c, *g, *p = current;

retry:
	if (!mm_need_new_owner(mm, p))
		return;

	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
	 */
592
	list_for_each_entry(c, &p->real_parent->children, sibling) {
593 594 595 596 597 598 599 600 601 602 603 604 605 606
		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);
607 608 609
	/*
	 * We found no owner yet mm_users > 1: this implies that we are
	 * most likely racing with swapoff (try_to_unuse()) or /proc or
610
	 * ptrace or page migration (get_task_mm()).  Mark owner as NULL.
611 612
	 */
	mm->owner = NULL;
613 614 615 616 617 618 619 620 621 622
	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);
623 624 625 626 627
	/*
	 * Delay read_unlock() till we have the task_lock()
	 * to ensure that c does not slip away underneath us
	 */
	read_unlock(&tasklist_lock);
628 629 630 631 632 633 634 635 636 637 638
	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
639 640 641 642
/*
 * Turn us into a lazy TLB process if we
 * aren't already..
 */
A
Adrian Bunk 已提交
643
static void exit_mm(struct task_struct * tsk)
L
Linus Torvalds 已提交
644 645
{
	struct mm_struct *mm = tsk->mm;
646
	struct core_state *core_state;
L
Linus Torvalds 已提交
647 648 649 650 651 652

	mm_release(tsk, mm);
	if (!mm)
		return;
	/*
	 * Serialize with any possible pending coredump.
653
	 * We must hold mmap_sem around checking core_state
L
Linus Torvalds 已提交
654
	 * and clearing tsk->mm.  The core-inducing thread
655
	 * will increment ->nr_threads for each thread in the
L
Linus Torvalds 已提交
656 657 658
	 * group with ->mm != NULL.
	 */
	down_read(&mm->mmap_sem);
659 660 661
	core_state = mm->core_state;
	if (core_state) {
		struct core_thread self;
L
Linus Torvalds 已提交
662 663
		up_read(&mm->mmap_sem);

664 665 666 667 668 669 670 671
		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
672

O
Oleg Nesterov 已提交
673 674 675 676 677 678 679
		for (;;) {
			set_task_state(tsk, TASK_UNINTERRUPTIBLE);
			if (!self.task) /* see coredump_finish() */
				break;
			schedule();
		}
		__set_task_state(tsk, TASK_RUNNING);
L
Linus Torvalds 已提交
680 681 682
		down_read(&mm->mmap_sem);
	}
	atomic_inc(&mm->mm_count);
E
Eric Sesterhenn 已提交
683
	BUG_ON(mm != tsk->active_mm);
L
Linus Torvalds 已提交
684 685 686 687 688
	/* more a memory barrier than a real lock */
	task_lock(tsk);
	tsk->mm = NULL;
	up_read(&mm->mmap_sem);
	enter_lazy_tlb(mm, current);
689 690
	/* We don't want this task to be frozen prematurely */
	clear_freeze_flag(tsk);
L
Linus Torvalds 已提交
691
	task_unlock(tsk);
692
	mm_update_next_owner(mm);
L
Linus Torvalds 已提交
693 694 695 696 697 698 699
	mmput(mm);
}

/*
 * When we die, we re-parent all our children.
 * Try to give them to another thread in our thread
 * group, and if no such member exists, give it to
700 701
 * the child reaper process (ie "init") in our pid
 * space.
L
Linus Torvalds 已提交
702
 */
703
static struct task_struct *find_new_reaper(struct task_struct *father)
L
Linus Torvalds 已提交
704
{
705 706
	struct pid_namespace *pid_ns = task_active_pid_ns(father);
	struct task_struct *thread;
L
Linus Torvalds 已提交
707

708 709 710 711 712 713 714 715
	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
716

717 718 719 720
	if (unlikely(pid_ns->child_reaper == father)) {
		write_unlock_irq(&tasklist_lock);
		if (unlikely(pid_ns == &init_pid_ns))
			panic("Attempted to kill init!");
L
Linus Torvalds 已提交
721

722 723
		zap_pid_ns_processes(pid_ns);
		write_lock_irq(&tasklist_lock);
L
Linus Torvalds 已提交
724
		/*
725 726 727
		 * We can not clear ->child_reaper or leave it alone.
		 * There may by stealth EXIT_DEAD tasks on ->children,
		 * forget_original_parent() must move them somewhere.
L
Linus Torvalds 已提交
728
		 */
729
		pid_ns->child_reaper = init_pid_ns.child_reaper;
L
Linus Torvalds 已提交
730
	}
731

732 733 734
	return pid_ns->child_reaper;
}

735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758
/*
* Any that need to be release_task'd are put on the @dead list.
 */
static void reparent_thread(struct task_struct *father, struct task_struct *p,
				struct list_head *dead)
{
	if (p->pdeath_signal)
		group_send_sig_info(p->pdeath_signal, SEND_SIG_NOINFO, p);

	list_move_tail(&p->sibling, &p->real_parent->children);

	if (task_detached(p))
		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. */
759
	if (!task_ptrace(p) &&
760 761 762 763 764 765 766 767 768 769 770
	    p->exit_state == EXIT_ZOMBIE && thread_group_empty(p)) {
		do_notify_parent(p, p->exit_signal);
		if (task_detached(p)) {
			p->exit_state = EXIT_DEAD;
			list_move_tail(&p->sibling, dead);
		}
	}

	kill_orphaned_pgrp(p, father);
}

771
static void forget_original_parent(struct task_struct *father)
L
Linus Torvalds 已提交
772
{
773
	struct task_struct *p, *n, *reaper;
774
	LIST_HEAD(dead_children);
775

776 777
	exit_ptrace(father);

778
	write_lock_irq(&tasklist_lock);
779
	reaper = find_new_reaper(father);
R
Roland McGrath 已提交
780

781
	list_for_each_entry_safe(p, n, &father->children, sibling) {
782
		p->real_parent = reaper;
R
Roland McGrath 已提交
783
		if (p->parent == father) {
784
			BUG_ON(task_ptrace(p));
R
Roland McGrath 已提交
785 786
			p->parent = p->real_parent;
		}
787
		reparent_thread(father, p, &dead_children);
L
Linus Torvalds 已提交
788
	}
789
	write_unlock_irq(&tasklist_lock);
790

791 792
	BUG_ON(!list_empty(&father->children));

793 794
	list_for_each_entry_safe(p, n, &dead_children, sibling) {
		list_del_init(&p->sibling);
795 796
		release_task(p);
	}
L
Linus Torvalds 已提交
797 798 799 800 801 802
}

/*
 * Send signals to all our closest relatives so that they know
 * to properly mourn us..
 */
803
static void exit_notify(struct task_struct *tsk, int group_dead)
L
Linus Torvalds 已提交
804
{
R
Roland McGrath 已提交
805 806
	int signal;
	void *cookie;
L
Linus Torvalds 已提交
807 808 809 810 811 812 813 814 815

	/*
	 * 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)
	 */
816
	forget_original_parent(tsk);
817
	exit_task_namespaces(tsk);
L
Linus Torvalds 已提交
818

819
	write_lock_irq(&tasklist_lock);
820 821
	if (group_dead)
		kill_orphaned_pgrp(tsk->group_leader, NULL);
L
Linus Torvalds 已提交
822

823
	/* Let father know we died
L
Linus Torvalds 已提交
824 825
	 *
	 * Thread signals are configurable, but you aren't going to use
826
	 * that to send signals to arbitary processes.
L
Linus Torvalds 已提交
827 828 829 830 831 832 833 834 835 836
	 * That stops right now.
	 *
	 * If the parent exec id doesn't match the exec id we saved
	 * when we started then we know the parent has changed security
	 * domain.
	 *
	 * If our self_exec id doesn't match our parent_exec_id then
	 * we have changed execution domain as these two values started
	 * the same after a fork.
	 */
837
	if (tsk->exit_signal != SIGCHLD && !task_detached(tsk) &&
838
	    (tsk->parent_exec_id != tsk->real_parent->self_exec_id ||
839
	     tsk->self_exec_id != tsk->parent_exec_id))
L
Linus Torvalds 已提交
840 841
		tsk->exit_signal = SIGCHLD;

R
Roland McGrath 已提交
842
	signal = tracehook_notify_death(tsk, &cookie, group_dead);
843
	if (signal >= 0)
R
Roland McGrath 已提交
844
		signal = do_notify_parent(tsk, signal);
L
Linus Torvalds 已提交
845

846
	tsk->exit_state = signal == DEATH_REAP ? EXIT_DEAD : EXIT_ZOMBIE;
L
Linus Torvalds 已提交
847

848
	/* mt-exec, de_thread() is waiting for us */
849
	if (thread_group_leader(tsk) &&
850 851
	    tsk->signal->group_exit_task &&
	    tsk->signal->notify_count < 0)
852 853
		wake_up_process(tsk->signal->group_exit_task);

L
Linus Torvalds 已提交
854 855
	write_unlock_irq(&tasklist_lock);

R
Roland McGrath 已提交
856 857
	tracehook_report_death(tsk, signal, cookie, group_dead);

L
Linus Torvalds 已提交
858
	/* If the process is dead, release it - nobody will wait for it */
859
	if (signal == DEATH_REAP)
L
Linus Torvalds 已提交
860 861 862
		release_task(tsk);
}

863 864 865 866 867 868 869
#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;

870
	free = stack_not_used(current);
871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887

	if (free >= lowest_to_date)
		return;

	spin_lock(&low_water_lock);
	if (free < lowest_to_date) {
		printk(KERN_WARNING "%s used greatest stack depth: %lu bytes "
				"left\n",
				current->comm, free);
		lowest_to_date = free;
	}
	spin_unlock(&low_water_lock);
}
#else
static inline void check_stack_usage(void) {}
#endif

888
NORET_TYPE void do_exit(long code)
L
Linus Torvalds 已提交
889 890 891 892 893 894
{
	struct task_struct *tsk = current;
	int group_dead;

	profile_task_exit(tsk);

895 896
	WARN_ON(atomic_read(&tsk->fs_excl));

L
Linus Torvalds 已提交
897 898 899 900 901
	if (unlikely(in_interrupt()))
		panic("Aiee, killing interrupt handler!");
	if (unlikely(!tsk->pid))
		panic("Attempted to kill the idle task!");

R
Roland McGrath 已提交
902
	tracehook_report_exit(&code);
L
Linus Torvalds 已提交
903

904 905
	validate_creds_for_do_exit(tsk);

906 907 908 909 910 911 912
	/*
	 * 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");
913 914 915 916 917 918 919 920 921 922
		/*
		 * 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;
923 924 925 926
		set_current_state(TASK_UNINTERRUPTIBLE);
		schedule();
	}

927 928
	exit_irq_thread();

O
Oleg Nesterov 已提交
929
	exit_signals(tsk);  /* sets PF_EXITING */
930 931 932 933
	/*
	 * tsk->flags are checked in the futex code to protect against
	 * an exiting task cleaning up the robust pi futexes.
	 */
934 935
	smp_mb();
	spin_unlock_wait(&tsk->pi_lock);
L
Linus Torvalds 已提交
936 937 938

	if (unlikely(in_atomic()))
		printk(KERN_INFO "note: %s[%d] exited with preempt_count %d\n",
939
				current->comm, task_pid_nr(current),
L
Linus Torvalds 已提交
940 941 942
				preempt_count());

	acct_update_integrals(tsk);
943

L
Linus Torvalds 已提交
944
	group_dead = atomic_dec_and_test(&tsk->signal->live);
945
	if (group_dead) {
946
		hrtimer_cancel(&tsk->signal->real_timer);
947
		exit_itimers(tsk->signal);
948
	}
949
	acct_collect(code, group_dead);
M
Miloslav Trmac 已提交
950 951
	if (group_dead)
		tty_audit_exit();
952 953
	if (unlikely(tsk->audit_context))
		audit_free(tsk);
954

955
	tsk->exit_code = code;
956
	taskstats_exit(tsk, group_dead);
957

L
Linus Torvalds 已提交
958 959
	exit_mm(tsk);

960
	if (group_dead)
961
		acct_process();
962 963
	trace_sched_process_exit(tsk);

L
Linus Torvalds 已提交
964
	exit_sem(tsk);
965 966
	exit_files(tsk);
	exit_fs(tsk);
967
	check_stack_usage();
L
Linus Torvalds 已提交
968
	exit_thread();
969
	cgroup_exit(tsk, 1);
L
Linus Torvalds 已提交
970 971 972 973

	if (group_dead && tsk->signal->leader)
		disassociate_ctty(1);

A
Al Viro 已提交
974
	module_put(task_thread_info(tsk)->exec_domain->module);
L
Linus Torvalds 已提交
975 976 977
	if (tsk->binfmt)
		module_put(tsk->binfmt->module);

M
Matt Helsley 已提交
978
	proc_exit_connector(tsk);
979 980 981 982 983

	/*
	 * Flush inherited counters to the parent - before the parent
	 * gets woken up by child-exit notifications.
	 */
984
	perf_event_exit_task(tsk);
985

986
	exit_notify(tsk, group_dead);
L
Linus Torvalds 已提交
987
#ifdef CONFIG_NUMA
988
	mpol_put(tsk->mempolicy);
L
Linus Torvalds 已提交
989 990
	tsk->mempolicy = NULL;
#endif
991
#ifdef CONFIG_FUTEX
992 993 994 995
	if (unlikely(!list_empty(&tsk->pi_state_list)))
		exit_pi_state_list(tsk);
	if (unlikely(current->pi_state_cache))
		kfree(current->pi_state_cache);
996
#endif
997
	/*
998
	 * Make sure we are holding no locks:
999
	 */
1000
	debug_check_no_locks_held(tsk);
1001 1002 1003 1004 1005 1006
	/*
	 * 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
1007

1008 1009 1010
	if (tsk->io_context)
		exit_io_context();

1011 1012 1013
	if (tsk->splice_pipe)
		__free_pipe_info(tsk->splice_pipe);

1014 1015
	validate_creds_for_do_exit(tsk);

C
Coywolf Qi Hunt 已提交
1016
	preempt_disable();
1017
	exit_rcu();
O
Oleg Nesterov 已提交
1018
	/* causes final put_task_struct in finish_task_switch(). */
1019
	tsk->state = TASK_DEAD;
L
Linus Torvalds 已提交
1020 1021 1022
	schedule();
	BUG();
	/* Avoid "noreturn function does return".  */
A
Alan Cox 已提交
1023 1024
	for (;;)
		cpu_relax();	/* For when BUG is null */
L
Linus Torvalds 已提交
1025 1026
}

1027 1028
EXPORT_SYMBOL_GPL(do_exit);

L
Linus Torvalds 已提交
1029 1030 1031 1032
NORET_TYPE void complete_and_exit(struct completion *comp, long code)
{
	if (comp)
		complete(comp);
O
Oleg Nesterov 已提交
1033

L
Linus Torvalds 已提交
1034 1035 1036 1037 1038
	do_exit(code);
}

EXPORT_SYMBOL(complete_and_exit);

1039
SYSCALL_DEFINE1(exit, int, error_code)
L
Linus Torvalds 已提交
1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050
{
	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).
 */
NORET_TYPE void
do_group_exit(int exit_code)
{
1051 1052
	struct signal_struct *sig = current->signal;

L
Linus Torvalds 已提交
1053 1054
	BUG_ON(exit_code & 0x80); /* core dumps don't get here */

1055 1056
	if (signal_group_exit(sig))
		exit_code = sig->group_exit_code;
L
Linus Torvalds 已提交
1057 1058 1059
	else if (!thread_group_empty(current)) {
		struct sighand_struct *const sighand = current->sighand;
		spin_lock_irq(&sighand->siglock);
1060
		if (signal_group_exit(sig))
L
Linus Torvalds 已提交
1061 1062 1063 1064
			/* Another thread got here before we took the lock.  */
			exit_code = sig->group_exit_code;
		else {
			sig->group_exit_code = exit_code;
1065
			sig->flags = SIGNAL_GROUP_EXIT;
L
Linus Torvalds 已提交
1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079
			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.
 */
1080
SYSCALL_DEFINE1(exit_group, int, error_code)
L
Linus Torvalds 已提交
1081 1082
{
	do_group_exit((error_code & 0xff) << 8);
1083 1084
	/* NOTREACHED */
	return 0;
L
Linus Torvalds 已提交
1085 1086
}

1087 1088 1089
struct wait_opts {
	enum pid_type		wo_type;
	int			wo_flags;
1090
	struct pid		*wo_pid;
1091 1092 1093 1094 1095 1096 1097 1098

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

	int			notask_error;
};

E
Eric W. Biederman 已提交
1099 1100 1101 1102 1103 1104 1105 1106 1107 1108
static struct pid *task_pid_type(struct task_struct *task, enum pid_type type)
{
	struct pid *pid = NULL;
	if (type == PIDTYPE_PID)
		pid = task->pids[type].pid;
	else if (type < PIDTYPE_MAX)
		pid = task->group_leader->pids[type].pid;
	return pid;
}

1109
static int eligible_child(struct wait_opts *wo, struct task_struct *p)
L
Linus Torvalds 已提交
1110
{
1111 1112
	int err;

1113 1114
	if (wo->wo_type < PIDTYPE_MAX) {
		if (task_pid_type(p, wo->wo_type) != wo->wo_pid)
L
Linus Torvalds 已提交
1115 1116 1117 1118 1119 1120 1121 1122
			return 0;
	}

	/* 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.) */
1123 1124
	if (((p->exit_signal != SIGCHLD) ^ !!(wo->wo_flags & __WCLONE))
	    && !(wo->wo_flags & __WALL))
L
Linus Torvalds 已提交
1125 1126
		return 0;

1127
	err = security_task_wait(p);
1128 1129
	if (err)
		return err;
L
Linus Torvalds 已提交
1130

1131
	return 1;
L
Linus Torvalds 已提交
1132 1133
}

1134 1135
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
1136
{
1137 1138 1139
	struct siginfo __user *infop;
	int retval = wo->wo_rusage
		? getrusage(p, RUSAGE_BOTH, wo->wo_rusage) : 0;
1140

L
Linus Torvalds 已提交
1141
	put_task_struct(p);
1142
	infop = wo->wo_info;
L
Linus Torvalds 已提交
1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165
	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);
	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.
 */
1166
static int wait_task_zombie(struct wait_opts *wo, struct task_struct *p)
L
Linus Torvalds 已提交
1167 1168
{
	unsigned long state;
1169
	int retval, status, traced;
1170
	pid_t pid = task_pid_vnr(p);
1171
	uid_t uid = __task_cred(p)->uid;
1172
	struct siginfo __user *infop;
L
Linus Torvalds 已提交
1173

1174
	if (!likely(wo->wo_flags & WEXITED))
R
Roland McGrath 已提交
1175 1176
		return 0;

1177
	if (unlikely(wo->wo_flags & WNOWAIT)) {
L
Linus Torvalds 已提交
1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189
		int exit_code = p->exit_code;
		int why, status;

		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;
		}
1190
		return wait_noreap_copyout(wo, p, pid, uid, why, status);
L
Linus Torvalds 已提交
1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202
	}

	/*
	 * 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;
	}

1203
	traced = ptrace_reparented(p);
1204 1205 1206 1207 1208
	/*
	 * It can be ptraced but not reparented, check
	 * !task_detached() to filter out sub-threads.
	 */
	if (likely(!traced) && likely(!task_detached(p))) {
1209 1210 1211
		struct signal_struct *psig;
		struct signal_struct *sig;

L
Linus Torvalds 已提交
1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222
		/*
		 * 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
1223
		 * need to protect the access to parent->signal fields,
L
Linus Torvalds 已提交
1224 1225 1226
		 * as other threads in the parent group can be right
		 * here reaping other children at the same time.
		 */
1227 1228
		spin_lock_irq(&p->real_parent->sighand->siglock);
		psig = p->real_parent->signal;
1229 1230 1231
		sig = p->signal;
		psig->cutime =
			cputime_add(psig->cutime,
1232 1233 1234
			cputime_add(p->utime,
			cputime_add(sig->utime,
				    sig->cutime)));
1235 1236
		psig->cstime =
			cputime_add(psig->cstime,
1237 1238 1239
			cputime_add(p->stime,
			cputime_add(sig->stime,
				    sig->cstime)));
1240 1241 1242 1243 1244
		psig->cgtime =
			cputime_add(psig->cgtime,
			cputime_add(p->gtime,
			cputime_add(sig->gtime,
				    sig->cgtime)));
1245 1246 1247 1248 1249 1250 1251 1252
		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;
1253 1254 1255 1256 1257 1258
		psig->cinblock +=
			task_io_get_inblock(p) +
			sig->inblock + sig->cinblock;
		psig->coublock +=
			task_io_get_oublock(p) +
			sig->oublock + sig->coublock;
1259 1260
		task_io_accounting_add(&psig->ioac, &p->ioac);
		task_io_accounting_add(&psig->ioac, &sig->ioac);
1261
		spin_unlock_irq(&p->real_parent->sighand->siglock);
L
Linus Torvalds 已提交
1262 1263 1264 1265 1266 1267 1268 1269
	}

	/*
	 * 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);

1270 1271
	retval = wo->wo_rusage
		? getrusage(p, RUSAGE_BOTH, wo->wo_rusage) : 0;
L
Linus Torvalds 已提交
1272 1273
	status = (p->signal->flags & SIGNAL_GROUP_EXIT)
		? p->signal->group_exit_code : p->exit_code;
1274 1275 1276 1277
	if (!retval && wo->wo_stat)
		retval = put_user(status, wo->wo_stat);

	infop = wo->wo_info;
L
Linus Torvalds 已提交
1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296
	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)
1297
		retval = put_user(pid, &infop->si_pid);
L
Linus Torvalds 已提交
1298
	if (!retval && infop)
1299
		retval = put_user(uid, &infop->si_uid);
1300
	if (!retval)
1301
		retval = pid;
1302 1303

	if (traced) {
L
Linus Torvalds 已提交
1304
		write_lock_irq(&tasklist_lock);
1305 1306 1307 1308 1309 1310 1311
		/* We dropped tasklist, ptracer could die and untrace */
		ptrace_unlink(p);
		/*
		 * If this is not a detached task, notify the parent.
		 * If it's still not detached after that, don't release
		 * it now.
		 */
1312
		if (!task_detached(p)) {
1313
			do_notify_parent(p, p->exit_signal);
1314
			if (!task_detached(p)) {
1315 1316
				p->exit_state = EXIT_ZOMBIE;
				p = NULL;
L
Linus Torvalds 已提交
1317 1318 1319 1320 1321 1322
			}
		}
		write_unlock_irq(&tasklist_lock);
	}
	if (p != NULL)
		release_task(p);
1323

L
Linus Torvalds 已提交
1324 1325 1326
	return retval;
}

1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338
static int *task_stopped_code(struct task_struct *p, bool ptrace)
{
	if (ptrace) {
		if (task_is_stopped_or_traced(p))
			return &p->exit_code;
	} else {
		if (p->signal->flags & SIGNAL_STOP_STOPPED)
			return &p->signal->group_exit_code;
	}
	return NULL;
}

L
Linus Torvalds 已提交
1339 1340 1341 1342 1343 1344
/*
 * Handle sys_wait4 work for one task in state TASK_STOPPED.  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.
 */
1345 1346
static int wait_task_stopped(struct wait_opts *wo,
				int ptrace, struct task_struct *p)
L
Linus Torvalds 已提交
1347
{
1348
	struct siginfo __user *infop;
1349
	int retval, exit_code, *p_code, why;
1350
	uid_t uid = 0; /* unneeded, required by compiler */
1351
	pid_t pid;
L
Linus Torvalds 已提交
1352

1353 1354 1355
	/*
	 * Traditionally we see ptrace'd stopped tasks regardless of options.
	 */
1356
	if (!ptrace && !(wo->wo_flags & WUNTRACED))
R
Roland McGrath 已提交
1357 1358
		return 0;

1359 1360 1361
	exit_code = 0;
	spin_lock_irq(&p->sighand->siglock);

1362 1363
	p_code = task_stopped_code(p, ptrace);
	if (unlikely(!p_code))
1364 1365
		goto unlock_sig;

1366
	exit_code = *p_code;
1367 1368 1369
	if (!exit_code)
		goto unlock_sig;

1370
	if (!unlikely(wo->wo_flags & WNOWAIT))
1371
		*p_code = 0;
1372

1373 1374
	/* don't need the RCU readlock here as we're holding a spinlock */
	uid = __task_cred(p)->uid;
1375 1376 1377
unlock_sig:
	spin_unlock_irq(&p->sighand->siglock);
	if (!exit_code)
L
Linus Torvalds 已提交
1378 1379 1380 1381 1382 1383 1384 1385 1386 1387
		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);
1388
	pid = task_pid_vnr(p);
R
Roland McGrath 已提交
1389
	why = ptrace ? CLD_TRAPPED : CLD_STOPPED;
L
Linus Torvalds 已提交
1390 1391
	read_unlock(&tasklist_lock);

1392 1393 1394 1395 1396 1397 1398
	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
1399

1400
	infop = wo->wo_info;
L
Linus Torvalds 已提交
1401 1402 1403 1404 1405
	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
1406
		retval = put_user((short)why, &infop->si_code);
L
Linus Torvalds 已提交
1407 1408 1409
	if (!retval && infop)
		retval = put_user(exit_code, &infop->si_status);
	if (!retval && infop)
1410
		retval = put_user(pid, &infop->si_pid);
L
Linus Torvalds 已提交
1411
	if (!retval && infop)
1412
		retval = put_user(uid, &infop->si_uid);
L
Linus Torvalds 已提交
1413
	if (!retval)
1414
		retval = pid;
L
Linus Torvalds 已提交
1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426
	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.
 */
1427
static int wait_task_continued(struct wait_opts *wo, struct task_struct *p)
L
Linus Torvalds 已提交
1428 1429 1430 1431 1432
{
	int retval;
	pid_t pid;
	uid_t uid;

1433
	if (!unlikely(wo->wo_flags & WCONTINUED))
R
Roland McGrath 已提交
1434 1435
		return 0;

L
Linus Torvalds 已提交
1436 1437 1438 1439 1440 1441 1442 1443 1444
	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;
	}
1445
	if (!unlikely(wo->wo_flags & WNOWAIT))
L
Linus Torvalds 已提交
1446
		p->signal->flags &= ~SIGNAL_STOP_CONTINUED;
1447
	uid = __task_cred(p)->uid;
L
Linus Torvalds 已提交
1448 1449
	spin_unlock_irq(&p->sighand->siglock);

1450
	pid = task_pid_vnr(p);
L
Linus Torvalds 已提交
1451 1452 1453
	get_task_struct(p);
	read_unlock(&tasklist_lock);

1454 1455 1456
	if (!wo->wo_info) {
		retval = wo->wo_rusage
			? getrusage(p, RUSAGE_BOTH, wo->wo_rusage) : 0;
L
Linus Torvalds 已提交
1457
		put_task_struct(p);
1458 1459
		if (!retval && wo->wo_stat)
			retval = put_user(0xffff, wo->wo_stat);
L
Linus Torvalds 已提交
1460
		if (!retval)
1461
			retval = pid;
L
Linus Torvalds 已提交
1462
	} else {
1463 1464
		retval = wait_noreap_copyout(wo, p, pid, uid,
					     CLD_CONTINUED, SIGCONT);
L
Linus Torvalds 已提交
1465 1466 1467 1468 1469 1470
		BUG_ON(retval == 0);
	}

	return retval;
}

R
Roland McGrath 已提交
1471 1472 1473
/*
 * Consider @p for a wait by @parent.
 *
1474
 * -ECHILD should be in ->notask_error before the first call.
R
Roland McGrath 已提交
1475 1476
 * Returns nonzero for a final return, when we have unlocked tasklist_lock.
 * Returns zero if the search for a child should continue;
1477
 * then ->notask_error is 0 if @p is an eligible child,
1478
 * or another error from security_task_wait(), or still -ECHILD.
R
Roland McGrath 已提交
1479
 */
1480 1481
static int wait_consider_task(struct wait_opts *wo, struct task_struct *parent,
				int ptrace, struct task_struct *p)
R
Roland McGrath 已提交
1482
{
1483
	int ret = eligible_child(wo, p);
1484
	if (!ret)
R
Roland McGrath 已提交
1485 1486
		return ret;

1487 1488 1489 1490 1491 1492 1493 1494
	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.
		 */
1495 1496
		if (wo->notask_error)
			wo->notask_error = ret;
1497
		return 0;
1498 1499
	}

1500
	if (likely(!ptrace) && unlikely(task_ptrace(p))) {
R
Roland McGrath 已提交
1501 1502 1503 1504
		/*
		 * This child is hidden by ptrace.
		 * We aren't allowed to see it now, but eventually we will.
		 */
1505
		wo->notask_error = 0;
R
Roland McGrath 已提交
1506 1507 1508
		return 0;
	}

R
Roland McGrath 已提交
1509 1510 1511 1512 1513 1514 1515
	if (p->exit_state == EXIT_DEAD)
		return 0;

	/*
	 * We don't reap group leaders with subthreads.
	 */
	if (p->exit_state == EXIT_ZOMBIE && !delay_group_leader(p))
1516
		return wait_task_zombie(wo, p);
R
Roland McGrath 已提交
1517 1518 1519 1520 1521

	/*
	 * It's stopped or running now, so it might
	 * later continue, exit, or stop again.
	 */
1522
	wo->notask_error = 0;
R
Roland McGrath 已提交
1523

1524
	if (task_stopped_code(p, ptrace))
1525
		return wait_task_stopped(wo, ptrace, p);
R
Roland McGrath 已提交
1526

1527
	return wait_task_continued(wo, p);
R
Roland McGrath 已提交
1528 1529 1530 1531 1532
}

/*
 * Do the work of do_wait() for one thread in the group, @tsk.
 *
1533
 * -ECHILD should be in ->notask_error before the first call.
R
Roland McGrath 已提交
1534 1535
 * Returns nonzero for a final return, when we have unlocked tasklist_lock.
 * Returns zero if the search for a child should continue; then
1536
 * ->notask_error is 0 if there were any eligible children,
1537
 * or another error from security_task_wait(), or still -ECHILD.
R
Roland McGrath 已提交
1538
 */
1539
static int do_wait_thread(struct wait_opts *wo, struct task_struct *tsk)
R
Roland McGrath 已提交
1540 1541 1542 1543
{
	struct task_struct *p;

	list_for_each_entry(p, &tsk->children, sibling) {
R
Roland McGrath 已提交
1544 1545 1546 1547
		/*
		 * Do not consider detached threads.
		 */
		if (!task_detached(p)) {
1548
			int ret = wait_consider_task(wo, tsk, 0, p);
R
Roland McGrath 已提交
1549 1550 1551
			if (ret)
				return ret;
		}
R
Roland McGrath 已提交
1552 1553 1554 1555 1556
	}

	return 0;
}

1557
static int ptrace_do_wait(struct wait_opts *wo, struct task_struct *tsk)
R
Roland McGrath 已提交
1558 1559 1560
{
	struct task_struct *p;

R
Roland McGrath 已提交
1561
	list_for_each_entry(p, &tsk->ptraced, ptrace_entry) {
1562
		int ret = wait_consider_task(wo, tsk, 1, p);
R
Roland McGrath 已提交
1563
		if (ret)
R
Roland McGrath 已提交
1564 1565 1566 1567 1568 1569
			return ret;
	}

	return 0;
}

1570
static long do_wait(struct wait_opts *wo)
L
Linus Torvalds 已提交
1571 1572 1573
{
	DECLARE_WAITQUEUE(wait, current);
	struct task_struct *tsk;
R
Roland McGrath 已提交
1574
	int retval;
L
Linus Torvalds 已提交
1575

1576
	trace_sched_process_wait(wo->wo_pid);
1577

L
Linus Torvalds 已提交
1578 1579
	add_wait_queue(&current->signal->wait_chldexit,&wait);
repeat:
R
Roland McGrath 已提交
1580 1581
	/*
	 * If there is nothing that can match our critiera just get out.
1582 1583 1584
	 * 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
1585
	 */
1586
	wo->notask_error = -ECHILD;
1587 1588
	if ((wo->wo_type < PIDTYPE_MAX) &&
	   (!wo->wo_pid || hlist_empty(&wo->wo_pid->tasks[wo->wo_type])))
1589
		goto notask;
E
Eric W. Biederman 已提交
1590

1591
	set_current_state(TASK_INTERRUPTIBLE);
L
Linus Torvalds 已提交
1592 1593 1594
	read_lock(&tasklist_lock);
	tsk = current;
	do {
1595 1596 1597
		retval = do_wait_thread(wo, tsk);
		if (retval)
			goto end;
1598

1599 1600
		retval = ptrace_do_wait(wo, tsk);
		if (retval)
R
Roland McGrath 已提交
1601 1602
			goto end;

1603
		if (wo->wo_flags & __WNOTHREAD)
L
Linus Torvalds 已提交
1604
			break;
1605
	} while_each_thread(current, tsk);
L
Linus Torvalds 已提交
1606
	read_unlock(&tasklist_lock);
O
Oleg Nesterov 已提交
1607

1608
notask:
1609 1610
	retval = wo->notask_error;
	if (!retval && !(wo->wo_flags & WNOHANG)) {
L
Linus Torvalds 已提交
1611
		retval = -ERESTARTSYS;
R
Roland McGrath 已提交
1612 1613 1614 1615
		if (!signal_pending(current)) {
			schedule();
			goto repeat;
		}
L
Linus Torvalds 已提交
1616 1617
	}
end:
1618
	__set_current_state(TASK_RUNNING);
L
Linus Torvalds 已提交
1619
	remove_wait_queue(&current->signal->wait_chldexit,&wait);
1620 1621 1622
	if (wo->wo_info) {
		struct siginfo __user *infop = wo->wo_info;

L
Linus Torvalds 已提交
1623
		if (retval > 0)
1624
			retval = 0;
L
Linus Torvalds 已提交
1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647
		else {
			/*
			 * For a WNOHANG return, clear out all the fields
			 * we would set so the user can easily tell the
			 * difference.
			 */
			if (!retval)
				retval = put_user(0, &infop->si_signo);
			if (!retval)
				retval = put_user(0, &infop->si_errno);
			if (!retval)
				retval = put_user(0, &infop->si_code);
			if (!retval)
				retval = put_user(0, &infop->si_pid);
			if (!retval)
				retval = put_user(0, &infop->si_uid);
			if (!retval)
				retval = put_user(0, &infop->si_status);
		}
	}
	return retval;
}

1648 1649
SYSCALL_DEFINE5(waitid, int, which, pid_t, upid, struct siginfo __user *,
		infop, int, options, struct rusage __user *, ru)
L
Linus Torvalds 已提交
1650
{
1651
	struct wait_opts wo;
E
Eric W. Biederman 已提交
1652 1653
	struct pid *pid = NULL;
	enum pid_type type;
L
Linus Torvalds 已提交
1654 1655 1656 1657 1658 1659 1660 1661 1662
	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
1663
		type = PIDTYPE_MAX;
L
Linus Torvalds 已提交
1664 1665
		break;
	case P_PID:
E
Eric W. Biederman 已提交
1666 1667
		type = PIDTYPE_PID;
		if (upid <= 0)
L
Linus Torvalds 已提交
1668 1669 1670
			return -EINVAL;
		break;
	case P_PGID:
E
Eric W. Biederman 已提交
1671 1672
		type = PIDTYPE_PGID;
		if (upid <= 0)
L
Linus Torvalds 已提交
1673 1674 1675 1676 1677 1678
			return -EINVAL;
		break;
	default:
		return -EINVAL;
	}

E
Eric W. Biederman 已提交
1679 1680
	if (type < PIDTYPE_MAX)
		pid = find_get_pid(upid);
1681 1682 1683 1684 1685 1686 1687 1688

	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);
E
Eric W. Biederman 已提交
1689
	put_pid(pid);
L
Linus Torvalds 已提交
1690 1691

	/* avoid REGPARM breakage on x86: */
1692
	asmlinkage_protect(5, ret, which, upid, infop, options, ru);
L
Linus Torvalds 已提交
1693 1694 1695
	return ret;
}

1696 1697
SYSCALL_DEFINE4(wait4, pid_t, upid, int __user *, stat_addr,
		int, options, struct rusage __user *, ru)
L
Linus Torvalds 已提交
1698
{
1699
	struct wait_opts wo;
E
Eric W. Biederman 已提交
1700 1701
	struct pid *pid = NULL;
	enum pid_type type;
L
Linus Torvalds 已提交
1702 1703 1704 1705 1706
	long ret;

	if (options & ~(WNOHANG|WUNTRACED|WCONTINUED|
			__WNOTHREAD|__WCLONE|__WALL))
		return -EINVAL;
E
Eric W. Biederman 已提交
1707 1708 1709 1710 1711 1712 1713 1714

	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;
1715
		pid = get_task_pid(current, PIDTYPE_PGID);
E
Eric W. Biederman 已提交
1716 1717 1718 1719 1720
	} else /* upid > 0 */ {
		type = PIDTYPE_PID;
		pid = find_get_pid(upid);
	}

1721 1722 1723 1724 1725 1726 1727
	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
1728
	put_pid(pid);
L
Linus Torvalds 已提交
1729 1730

	/* avoid REGPARM breakage on x86: */
1731
	asmlinkage_protect(4, ret, upid, stat_addr, options, ru);
L
Linus Torvalds 已提交
1732 1733 1734 1735 1736 1737 1738 1739 1740
	return ret;
}

#ifdef __ARCH_WANT_SYS_WAITPID

/*
 * sys_waitpid() remains for compatibility. waitpid() should be
 * implemented by calling sys_wait4() from libc.a.
 */
1741
SYSCALL_DEFINE3(waitpid, pid_t, pid, int __user *, stat_addr, int, options)
L
Linus Torvalds 已提交
1742 1743 1744 1745 1746
{
	return sys_wait4(pid, stat_addr, options, NULL);
}

#endif