exit.c 43.8 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>
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>
52
#include <linux/hw_breakpoint.h>
L
Linus Torvalds 已提交
53 54 55 56 57

#include <asm/uaccess.h>
#include <asm/unistd.h>
#include <asm/pgtable.h>
#include <asm/mmu_context.h>
D
David Howells 已提交
58
#include "cred-internals.h"
L
Linus Torvalds 已提交
59

A
Adrian Bunk 已提交
60 61
static void exit_mm(struct task_struct * tsk);

L
Linus Torvalds 已提交
62 63 64 65 66 67 68
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);
69

70
		list_del_rcu(&p->tasks);
71
		list_del_init(&p->sibling);
72
		__get_cpu_var(process_counts)--;
L
Linus Torvalds 已提交
73
	}
O
Oleg Nesterov 已提交
74
	list_del_rcu(&p->thread_group);
L
Linus Torvalds 已提交
75 76
}

77 78 79 80 81 82 83 84 85 86 87
/*
 * 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));

88 89 90
	sighand = rcu_dereference_check(tsk->sighand,
					rcu_read_lock_held() ||
					lockdep_is_held(&tasklist_lock));
91 92 93 94 95 96 97 98 99 100
	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:
		 */
101
		if (sig->group_exit_task && atomic_read(&sig->count) == sig->notify_count)
102
			wake_up_process(sig->group_exit_task);
103

104 105 106 107 108 109 110 111 112 113 114 115
		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.
		 */
116 117
		sig->utime = cputime_add(sig->utime, tsk->utime);
		sig->stime = cputime_add(sig->stime, tsk->stime);
H
Hidetoshi Seto 已提交
118
		sig->gtime = cputime_add(sig->gtime, tsk->gtime);
119 120 121 122
		sig->min_flt += tsk->min_flt;
		sig->maj_flt += tsk->maj_flt;
		sig->nvcsw += tsk->nvcsw;
		sig->nivcsw += tsk->nivcsw;
123 124
		sig->inblock += task_io_get_inblock(tsk);
		sig->oublock += task_io_get_oublock(tsk);
125
		task_io_accounting_add(&sig->ioac, &tsk->ioac);
126
		sig->sum_sched_runtime += tsk->se.sum_exec_runtime;
127 128 129
		sig = NULL; /* Marker for below. */
	}

130 131
	__unhash_process(tsk);

132 133 134 135 136 137
	/*
	 * Do this under ->siglock, we can race with another thread
	 * doing sigqueue_free() if we have SIGQUEUE_PREALLOC signals.
	 */
	flush_sigqueue(&tsk->pending);

138
	tsk->signal = NULL;
139
	tsk->sighand = NULL;
140 141
	spin_unlock(&sighand->siglock);

142
	__cleanup_sighand(sighand);
143 144 145
	clear_tsk_thread_flag(tsk,TIF_SIGPENDING);
	if (sig) {
		flush_sigqueue(&sig->shared_pending);
146
		taskstats_tgid_free(sig);
147 148 149 150 151
		/*
		 * Make sure ->signal can't go away under rq->lock,
		 * see account_group_exec_runtime().
		 */
		task_rq_unlock_wait(tsk);
152 153 154 155
		__cleanup_signal(sig);
	}
}

156 157
static void delayed_put_task_struct(struct rcu_head *rhp)
{
158 159
	struct task_struct *tsk = container_of(rhp, struct task_struct, rcu);

160 161
#ifdef CONFIG_PERF_EVENTS
	WARN_ON_ONCE(tsk->perf_event_ctxp);
162
#endif
163 164
	trace_sched_process_free(tsk);
	put_task_struct(tsk);
165 166
}

R
Roland McGrath 已提交
167

L
Linus Torvalds 已提交
168 169
void release_task(struct task_struct * p)
{
170
	struct task_struct *leader;
L
Linus Torvalds 已提交
171
	int zap_leader;
172
repeat:
R
Roland McGrath 已提交
173
	tracehook_prepare_release_task(p);
174
	/* don't need to get the RCU readlock here - the process is dead and
175 176
	 * can't be modifying its own credentials. But shut RCU-lockdep up */
	rcu_read_lock();
177
	atomic_dec(&__task_cred(p)->user->processes);
178
	rcu_read_unlock();
179

180
	proc_flush_task(p);
181

L
Linus Torvalds 已提交
182
	write_lock_irq(&tasklist_lock);
R
Roland McGrath 已提交
183
	tracehook_finish_release_task(p);
L
Linus Torvalds 已提交
184
	__exit_signal(p);
185

L
Linus Torvalds 已提交
186 187 188 189 190 191 192 193
	/*
	 * 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) {
194
		BUG_ON(task_detached(leader));
L
Linus Torvalds 已提交
195 196 197 198 199 200 201 202 203
		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.
		 */
204
		zap_leader = task_detached(leader);
R
Roland McGrath 已提交
205 206 207 208 209 210 211

		/*
		 * 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
212 213 214 215
	}

	write_unlock_irq(&tasklist_lock);
	release_thread(p);
216
	call_rcu(&p->rcu, delayed_put_task_struct);
L
Linus Torvalds 已提交
217 218 219 220 221 222 223 224 225 226

	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...
227 228
 *
 * The caller must hold rcu lock or the tasklist lock.
L
Linus Torvalds 已提交
229
 */
230
struct pid *session_of_pgrp(struct pid *pgrp)
L
Linus Torvalds 已提交
231 232
{
	struct task_struct *p;
233
	struct pid *sid = NULL;
234

235
	p = pid_task(pgrp, PIDTYPE_PGID);
236
	if (p == NULL)
237
		p = pid_task(pgrp, PIDTYPE_PID);
238
	if (p != NULL)
239
		sid = task_session(p);
240

L
Linus Torvalds 已提交
241 242 243 244 245 246 247 248 249 250 251
	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?"
 */
252
static int will_become_orphaned_pgrp(struct pid *pgrp, struct task_struct *ignored_task)
L
Linus Torvalds 已提交
253 254 255
{
	struct task_struct *p;

256
	do_each_pid_task(pgrp, PIDTYPE_PGID, p) {
257 258 259
		if ((p == ignored_task) ||
		    (p->exit_state && thread_group_empty(p)) ||
		    is_global_init(p->real_parent))
L
Linus Torvalds 已提交
260
			continue;
261

262
		if (task_pgrp(p->real_parent) != pgrp &&
263 264
		    task_session(p->real_parent) == task_session(p))
			return 0;
265
	} while_each_pid_task(pgrp, PIDTYPE_PGID, p);
266 267

	return 1;
L
Linus Torvalds 已提交
268 269
}

270
int is_current_pgrp_orphaned(void)
L
Linus Torvalds 已提交
271 272 273 274
{
	int retval;

	read_lock(&tasklist_lock);
275
	retval = will_become_orphaned_pgrp(task_pgrp(current), NULL);
L
Linus Torvalds 已提交
276 277 278 279 280
	read_unlock(&tasklist_lock);

	return retval;
}

281
static int has_stopped_jobs(struct pid *pgrp)
L
Linus Torvalds 已提交
282 283 284 285
{
	int retval = 0;
	struct task_struct *p;

286
	do_each_pid_task(pgrp, PIDTYPE_PGID, p) {
M
Matthew Wilcox 已提交
287
		if (!task_is_stopped(p))
L
Linus Torvalds 已提交
288 289 290
			continue;
		retval = 1;
		break;
291
	} while_each_pid_task(pgrp, PIDTYPE_PGID, p);
L
Linus Torvalds 已提交
292 293 294
	return retval;
}

295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325
/*
 * 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
326
/**
327
 * reparent_to_kthreadd - Reparent the calling kernel thread to kthreadd
L
Linus Torvalds 已提交
328 329
 *
 * If a kernel thread is launched as a result of a system call, or if
330 331
 * 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
332 333 334 335
 *
 * 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.
 *
336
 * NOTE that reparent_to_kthreadd() gives the caller full capabilities.
L
Linus Torvalds 已提交
337
 */
338
static void reparent_to_kthreadd(void)
L
Linus Torvalds 已提交
339 340 341 342 343
{
	write_lock_irq(&tasklist_lock);

	ptrace_unlink(current);
	/* Reparent to init */
344
	current->real_parent = current->parent = kthreadd_task;
R
Roland McGrath 已提交
345
	list_move_tail(&current->sibling, &current->real_parent->children);
L
Linus Torvalds 已提交
346 347 348 349

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

350
	if (task_nice(current) < 0)
L
Linus Torvalds 已提交
351 352 353 354 355 356
		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
357 358 359

	atomic_inc(&init_cred.usage);
	commit_creds(&init_cred);
L
Linus Torvalds 已提交
360 361 362
	write_unlock_irq(&tasklist_lock);
}

363
void __set_special_pids(struct pid *pid)
L
Linus Torvalds 已提交
364
{
365
	struct task_struct *curr = current->group_leader;
L
Linus Torvalds 已提交
366

367
	if (task_session(curr) != pid)
368
		change_pid(curr, PIDTYPE_SID, pid);
369 370

	if (task_pgrp(curr) != pid)
371
		change_pid(curr, PIDTYPE_PGID, pid);
L
Linus Torvalds 已提交
372 373
}

374
static void set_special_pids(struct pid *pid)
L
Linus Torvalds 已提交
375 376
{
	write_lock_irq(&tasklist_lock);
377
	__set_special_pids(pid);
L
Linus Torvalds 已提交
378 379 380 381
	write_unlock_irq(&tasklist_lock);
}

/*
382 383
 * 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
384 385 386
 */
int allow_signal(int sig)
{
387
	if (!valid_signal(sig) || sig < 1)
L
Linus Torvalds 已提交
388 389 390
		return -EINVAL;

	spin_lock_irq(&current->sighand->siglock);
391
	/* This is only needed for daemonize()'ed kthreads */
L
Linus Torvalds 已提交
392
	sigdelset(&current->blocked, sig);
393 394 395 396 397 398
	/*
	 * 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
399 400 401 402 403 404 405 406 407
	recalc_sigpending();
	spin_unlock_irq(&current->sighand->siglock);
	return 0;
}

EXPORT_SYMBOL(allow_signal);

int disallow_signal(int sig)
{
408
	if (!valid_signal(sig) || sig < 1)
L
Linus Torvalds 已提交
409 410 411
		return -EINVAL;

	spin_lock_irq(&current->sighand->siglock);
412
	current->sighand->action[(sig)-1].sa.sa_handler = SIG_IGN;
L
Linus Torvalds 已提交
413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439
	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);
440 441 442 443
	/*
	 * 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
444
	current->flags |= (PF_NOFREEZE | PF_KTHREAD);
L
Linus Torvalds 已提交
445

446 447 448 449
	if (current->nsproxy != &init_nsproxy) {
		get_nsproxy(&init_nsproxy);
		switch_task_namespaces(current, &init_nsproxy);
	}
450
	set_special_pids(&init_struct_pid);
451
	proc_clear_tty(current);
L
Linus Torvalds 已提交
452 453 454 455 456 457 458 459

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

	/* Become as one with the init task */

460
	daemonize_fs_struct();
461
	exit_files(current);
L
Linus Torvalds 已提交
462 463 464
	current->files = init_task.files;
	atomic_inc(&current->files->count);

465
	reparent_to_kthreadd();
L
Linus Torvalds 已提交
466 467 468 469
}

EXPORT_SYMBOL(daemonize);

470
static void close_files(struct files_struct * files)
L
Linus Torvalds 已提交
471 472
{
	int i, j;
473
	struct fdtable *fdt;
L
Linus Torvalds 已提交
474 475

	j = 0;
476 477 478 479

	/*
	 * It is safe to dereference the fd table without RCU or
	 * ->file_lock because this is the last reference to the
480
	 * files structure.  But use RCU to shut RCU-lockdep up.
481
	 */
482
	rcu_read_lock();
483
	fdt = files_fdtable(files);
484
	rcu_read_unlock();
L
Linus Torvalds 已提交
485 486 487
	for (;;) {
		unsigned long set;
		i = j * __NFDBITS;
488
		if (i >= fdt->max_fds)
L
Linus Torvalds 已提交
489
			break;
490
		set = fdt->open_fds->fds_bits[j++];
L
Linus Torvalds 已提交
491 492
		while (set) {
			if (set & 1) {
493
				struct file * file = xchg(&fdt->fd[i], NULL);
494
				if (file) {
L
Linus Torvalds 已提交
495
					filp_close(file, files);
496 497
					cond_resched();
				}
L
Linus Torvalds 已提交
498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517
			}
			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;
}

518
void put_files_struct(struct files_struct *files)
L
Linus Torvalds 已提交
519
{
520 521
	struct fdtable *fdt;

L
Linus Torvalds 已提交
522 523 524 525
	if (atomic_dec_and_test(&files->count)) {
		close_files(files);
		/*
		 * Free the fd and fdset arrays if we expanded them.
526 527 528
		 * 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
529
		 */
530
		rcu_read_lock();
531
		fdt = files_fdtable(files);
532
		if (fdt != &files->fdtab)
533
			kmem_cache_free(files_cachep, files);
534
		free_fdtable(fdt);
535
		rcu_read_unlock();
L
Linus Torvalds 已提交
536 537 538
	}
}

539
void reset_files_struct(struct files_struct *files)
540
{
541
	struct task_struct *tsk = current;
542 543 544 545 546 547 548 549 550
	struct files_struct *old;

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

551
void exit_files(struct task_struct *tsk)
L
Linus Torvalds 已提交
552 553 554 555 556 557 558 559 560 561 562
{
	struct files_struct * files = tsk->files;

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

563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600
#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
	 */
601
	list_for_each_entry(c, &p->real_parent->children, sibling) {
602 603 604 605 606 607 608 609 610 611 612 613 614 615
		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);
616 617 618
	/*
	 * We found no owner yet mm_users > 1: this implies that we are
	 * most likely racing with swapoff (try_to_unuse()) or /proc or
619
	 * ptrace or page migration (get_task_mm()).  Mark owner as NULL.
620 621
	 */
	mm->owner = NULL;
622 623 624 625 626 627 628 629 630 631
	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);
632 633 634 635 636
	/*
	 * Delay read_unlock() till we have the task_lock()
	 * to ensure that c does not slip away underneath us
	 */
	read_unlock(&tasklist_lock);
637 638 639 640 641 642 643 644 645 646 647
	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
648 649 650 651
/*
 * Turn us into a lazy TLB process if we
 * aren't already..
 */
A
Adrian Bunk 已提交
652
static void exit_mm(struct task_struct * tsk)
L
Linus Torvalds 已提交
653 654
{
	struct mm_struct *mm = tsk->mm;
655
	struct core_state *core_state;
L
Linus Torvalds 已提交
656 657 658 659 660 661

	mm_release(tsk, mm);
	if (!mm)
		return;
	/*
	 * Serialize with any possible pending coredump.
662
	 * We must hold mmap_sem around checking core_state
L
Linus Torvalds 已提交
663
	 * and clearing tsk->mm.  The core-inducing thread
664
	 * will increment ->nr_threads for each thread in the
L
Linus Torvalds 已提交
665 666 667
	 * group with ->mm != NULL.
	 */
	down_read(&mm->mmap_sem);
668 669 670
	core_state = mm->core_state;
	if (core_state) {
		struct core_thread self;
L
Linus Torvalds 已提交
671 672
		up_read(&mm->mmap_sem);

673 674 675 676 677 678 679 680
		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
681

O
Oleg Nesterov 已提交
682 683 684 685 686 687 688
		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
689 690 691
		down_read(&mm->mmap_sem);
	}
	atomic_inc(&mm->mm_count);
E
Eric Sesterhenn 已提交
692
	BUG_ON(mm != tsk->active_mm);
L
Linus Torvalds 已提交
693 694 695 696 697
	/* more a memory barrier than a real lock */
	task_lock(tsk);
	tsk->mm = NULL;
	up_read(&mm->mmap_sem);
	enter_lazy_tlb(mm, current);
698 699
	/* We don't want this task to be frozen prematurely */
	clear_freeze_flag(tsk);
L
Linus Torvalds 已提交
700
	task_unlock(tsk);
701
	mm_update_next_owner(mm);
L
Linus Torvalds 已提交
702 703 704 705 706 707 708
	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
709 710
 * the child reaper process (ie "init") in our pid
 * space.
L
Linus Torvalds 已提交
711
 */
712
static struct task_struct *find_new_reaper(struct task_struct *father)
L
Linus Torvalds 已提交
713
{
714 715
	struct pid_namespace *pid_ns = task_active_pid_ns(father);
	struct task_struct *thread;
L
Linus Torvalds 已提交
716

717 718 719 720 721 722 723 724
	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
725

726 727 728 729
	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
730

731 732
		zap_pid_ns_processes(pid_ns);
		write_lock_irq(&tasklist_lock);
L
Linus Torvalds 已提交
733
		/*
734 735 736
		 * 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
737
		 */
738
		pid_ns->child_reaper = init_pid_ns.child_reaper;
L
Linus Torvalds 已提交
739
	}
740

741 742 743
	return pid_ns->child_reaper;
}

744 745 746
/*
* Any that need to be release_task'd are put on the @dead list.
 */
747
static void reparent_leader(struct task_struct *father, struct task_struct *p,
748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764
				struct list_head *dead)
{
	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. */
765
	if (!task_ptrace(p) &&
766 767 768 769 770 771 772 773 774 775 776
	    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);
}

777
static void forget_original_parent(struct task_struct *father)
L
Linus Torvalds 已提交
778
{
779
	struct task_struct *p, *n, *reaper;
780
	LIST_HEAD(dead_children);
781

782 783
	exit_ptrace(father);

784
	write_lock_irq(&tasklist_lock);
785
	reaper = find_new_reaper(father);
R
Roland McGrath 已提交
786

787
	list_for_each_entry_safe(p, n, &father->children, sibling) {
788 789 790 791 792 793 794 795 796 797 798 799
		struct task_struct *t = p;
		do {
			t->real_parent = reaper;
			if (t->parent == father) {
				BUG_ON(task_ptrace(t));
				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
800
	}
801
	write_unlock_irq(&tasklist_lock);
802

803 804
	BUG_ON(!list_empty(&father->children));

805 806
	list_for_each_entry_safe(p, n, &dead_children, sibling) {
		list_del_init(&p->sibling);
807 808
		release_task(p);
	}
L
Linus Torvalds 已提交
809 810 811 812 813 814
}

/*
 * Send signals to all our closest relatives so that they know
 * to properly mourn us..
 */
815
static void exit_notify(struct task_struct *tsk, int group_dead)
L
Linus Torvalds 已提交
816
{
R
Roland McGrath 已提交
817 818
	int signal;
	void *cookie;
L
Linus Torvalds 已提交
819 820 821 822 823 824 825 826 827

	/*
	 * 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)
	 */
828
	forget_original_parent(tsk);
829
	exit_task_namespaces(tsk);
L
Linus Torvalds 已提交
830

831
	write_lock_irq(&tasklist_lock);
832 833
	if (group_dead)
		kill_orphaned_pgrp(tsk->group_leader, NULL);
L
Linus Torvalds 已提交
834

835
	/* Let father know we died
L
Linus Torvalds 已提交
836 837
	 *
	 * Thread signals are configurable, but you aren't going to use
838
	 * that to send signals to arbitary processes.
L
Linus Torvalds 已提交
839 840 841 842 843 844 845 846 847 848
	 * 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.
	 */
849
	if (tsk->exit_signal != SIGCHLD && !task_detached(tsk) &&
850
	    (tsk->parent_exec_id != tsk->real_parent->self_exec_id ||
851
	     tsk->self_exec_id != tsk->parent_exec_id))
L
Linus Torvalds 已提交
852 853
		tsk->exit_signal = SIGCHLD;

R
Roland McGrath 已提交
854
	signal = tracehook_notify_death(tsk, &cookie, group_dead);
855
	if (signal >= 0)
R
Roland McGrath 已提交
856
		signal = do_notify_parent(tsk, signal);
L
Linus Torvalds 已提交
857

858
	tsk->exit_state = signal == DEATH_REAP ? EXIT_DEAD : EXIT_ZOMBIE;
L
Linus Torvalds 已提交
859

860
	/* mt-exec, de_thread() is waiting for us */
861
	if (thread_group_leader(tsk) &&
862 863
	    tsk->signal->group_exit_task &&
	    tsk->signal->notify_count < 0)
864 865
		wake_up_process(tsk->signal->group_exit_task);

L
Linus Torvalds 已提交
866 867
	write_unlock_irq(&tasklist_lock);

R
Roland McGrath 已提交
868 869
	tracehook_report_death(tsk, signal, cookie, group_dead);

L
Linus Torvalds 已提交
870
	/* If the process is dead, release it - nobody will wait for it */
871
	if (signal == DEATH_REAP)
L
Linus Torvalds 已提交
872 873 874
		release_task(tsk);
}

875 876 877 878 879 880 881
#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;

882
	free = stack_not_used(current);
883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899

	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

900
NORET_TYPE void do_exit(long code)
L
Linus Torvalds 已提交
901 902 903 904 905 906
{
	struct task_struct *tsk = current;
	int group_dead;

	profile_task_exit(tsk);

907 908
	WARN_ON(atomic_read(&tsk->fs_excl));

L
Linus Torvalds 已提交
909 910 911 912 913
	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
914
	tracehook_report_exit(&code);
L
Linus Torvalds 已提交
915

916 917
	validate_creds_for_do_exit(tsk);

918 919 920 921 922 923 924
	/*
	 * 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");
925 926 927 928 929 930 931 932 933 934
		/*
		 * 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;
935 936 937 938
		set_current_state(TASK_UNINTERRUPTIBLE);
		schedule();
	}

939 940
	exit_irq_thread();

O
Oleg Nesterov 已提交
941
	exit_signals(tsk);  /* sets PF_EXITING */
942 943 944 945
	/*
	 * tsk->flags are checked in the futex code to protect against
	 * an exiting task cleaning up the robust pi futexes.
	 */
946
	smp_mb();
947
	raw_spin_unlock_wait(&tsk->pi_lock);
L
Linus Torvalds 已提交
948 949 950

	if (unlikely(in_atomic()))
		printk(KERN_INFO "note: %s[%d] exited with preempt_count %d\n",
951
				current->comm, task_pid_nr(current),
L
Linus Torvalds 已提交
952 953 954
				preempt_count());

	acct_update_integrals(tsk);
955

L
Linus Torvalds 已提交
956
	group_dead = atomic_dec_and_test(&tsk->signal->live);
957
	if (group_dead) {
958
		hrtimer_cancel(&tsk->signal->real_timer);
959
		exit_itimers(tsk->signal);
J
Jiri Pirko 已提交
960 961
		if (tsk->mm)
			setmax_mm_hiwater_rss(&tsk->signal->maxrss, tsk->mm);
962
	}
963
	acct_collect(code, group_dead);
M
Miloslav Trmac 已提交
964 965
	if (group_dead)
		tty_audit_exit();
966 967
	if (unlikely(tsk->audit_context))
		audit_free(tsk);
968

969
	tsk->exit_code = code;
970
	taskstats_exit(tsk, group_dead);
971

L
Linus Torvalds 已提交
972 973
	exit_mm(tsk);

974
	if (group_dead)
975
		acct_process();
976 977
	trace_sched_process_exit(tsk);

L
Linus Torvalds 已提交
978
	exit_sem(tsk);
979 980
	exit_files(tsk);
	exit_fs(tsk);
981
	check_stack_usage();
L
Linus Torvalds 已提交
982
	exit_thread();
983
	cgroup_exit(tsk, 1);
L
Linus Torvalds 已提交
984

985
	if (group_dead)
L
Linus Torvalds 已提交
986 987
		disassociate_ctty(1);

A
Al Viro 已提交
988
	module_put(task_thread_info(tsk)->exec_domain->module);
L
Linus Torvalds 已提交
989

M
Matt Helsley 已提交
990
	proc_exit_connector(tsk);
991

992 993 994 995
	/*
	 * FIXME: do that only when needed, using sched_exit tracepoint
	 */
	flush_ptrace_hw_breakpoint(tsk);
996 997 998 999
	/*
	 * Flush inherited counters to the parent - before the parent
	 * gets woken up by child-exit notifications.
	 */
1000
	perf_event_exit_task(tsk);
1001

1002
	exit_notify(tsk, group_dead);
L
Linus Torvalds 已提交
1003
#ifdef CONFIG_NUMA
1004
	mpol_put(tsk->mempolicy);
L
Linus Torvalds 已提交
1005 1006
	tsk->mempolicy = NULL;
#endif
1007
#ifdef CONFIG_FUTEX
1008 1009
	if (unlikely(current->pi_state_cache))
		kfree(current->pi_state_cache);
1010
#endif
1011
	/*
1012
	 * Make sure we are holding no locks:
1013
	 */
1014
	debug_check_no_locks_held(tsk);
1015 1016 1017 1018 1019 1020
	/*
	 * 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
1021

1022
	if (tsk->io_context)
1023
		exit_io_context(tsk);
1024

1025 1026 1027
	if (tsk->splice_pipe)
		__free_pipe_info(tsk->splice_pipe);

1028 1029
	validate_creds_for_do_exit(tsk);

C
Coywolf Qi Hunt 已提交
1030
	preempt_disable();
1031
	exit_rcu();
O
Oleg Nesterov 已提交
1032
	/* causes final put_task_struct in finish_task_switch(). */
1033
	tsk->state = TASK_DEAD;
L
Linus Torvalds 已提交
1034 1035 1036
	schedule();
	BUG();
	/* Avoid "noreturn function does return".  */
A
Alan Cox 已提交
1037 1038
	for (;;)
		cpu_relax();	/* For when BUG is null */
L
Linus Torvalds 已提交
1039 1040
}

1041 1042
EXPORT_SYMBOL_GPL(do_exit);

L
Linus Torvalds 已提交
1043 1044 1045 1046
NORET_TYPE void complete_and_exit(struct completion *comp, long code)
{
	if (comp)
		complete(comp);
O
Oleg Nesterov 已提交
1047

L
Linus Torvalds 已提交
1048 1049 1050 1051 1052
	do_exit(code);
}

EXPORT_SYMBOL(complete_and_exit);

1053
SYSCALL_DEFINE1(exit, int, error_code)
L
Linus Torvalds 已提交
1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064
{
	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)
{
1065 1066
	struct signal_struct *sig = current->signal;

L
Linus Torvalds 已提交
1067 1068
	BUG_ON(exit_code & 0x80); /* core dumps don't get here */

1069 1070
	if (signal_group_exit(sig))
		exit_code = sig->group_exit_code;
L
Linus Torvalds 已提交
1071 1072 1073
	else if (!thread_group_empty(current)) {
		struct sighand_struct *const sighand = current->sighand;
		spin_lock_irq(&sighand->siglock);
1074
		if (signal_group_exit(sig))
L
Linus Torvalds 已提交
1075 1076 1077 1078
			/* Another thread got here before we took the lock.  */
			exit_code = sig->group_exit_code;
		else {
			sig->group_exit_code = exit_code;
1079
			sig->flags = SIGNAL_GROUP_EXIT;
L
Linus Torvalds 已提交
1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093
			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.
 */
1094
SYSCALL_DEFINE1(exit_group, int, error_code)
L
Linus Torvalds 已提交
1095 1096
{
	do_group_exit((error_code & 0xff) << 8);
1097 1098
	/* NOTREACHED */
	return 0;
L
Linus Torvalds 已提交
1099 1100
}

1101 1102 1103
struct wait_opts {
	enum pid_type		wo_type;
	int			wo_flags;
1104
	struct pid		*wo_pid;
1105 1106 1107 1108 1109

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

1110
	wait_queue_t		child_wait;
1111 1112 1113
	int			notask_error;
};

1114 1115
static inline
struct pid *task_pid_type(struct task_struct *task, enum pid_type type)
E
Eric W. Biederman 已提交
1116
{
1117 1118 1119
	if (type != PIDTYPE_PID)
		task = task->group_leader;
	return task->pids[type].pid;
E
Eric W. Biederman 已提交
1120 1121
}

1122
static int eligible_pid(struct wait_opts *wo, struct task_struct *p)
L
Linus Torvalds 已提交
1123
{
1124 1125 1126
	return	wo->wo_type == PIDTYPE_MAX ||
		task_pid_type(p, wo->wo_type) == wo->wo_pid;
}
L
Linus Torvalds 已提交
1127

1128 1129 1130 1131
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
1132 1133 1134 1135 1136
	/* 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.) */
1137 1138
	if (((p->exit_signal != SIGCHLD) ^ !!(wo->wo_flags & __WCLONE))
	    && !(wo->wo_flags & __WALL))
L
Linus Torvalds 已提交
1139 1140
		return 0;

1141
	return 1;
L
Linus Torvalds 已提交
1142 1143
}

1144 1145
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
1146
{
1147 1148 1149
	struct siginfo __user *infop;
	int retval = wo->wo_rusage
		? getrusage(p, RUSAGE_BOTH, wo->wo_rusage) : 0;
1150

L
Linus Torvalds 已提交
1151
	put_task_struct(p);
1152
	infop = wo->wo_info;
1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166
	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
1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177
	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.
 */
1178
static int wait_task_zombie(struct wait_opts *wo, struct task_struct *p)
L
Linus Torvalds 已提交
1179 1180
{
	unsigned long state;
1181
	int retval, status, traced;
1182
	pid_t pid = task_pid_vnr(p);
1183
	uid_t uid = __task_cred(p)->uid;
1184
	struct siginfo __user *infop;
L
Linus Torvalds 已提交
1185

1186
	if (!likely(wo->wo_flags & WEXITED))
R
Roland McGrath 已提交
1187 1188
		return 0;

1189
	if (unlikely(wo->wo_flags & WNOWAIT)) {
L
Linus Torvalds 已提交
1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201
		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;
		}
1202
		return wait_noreap_copyout(wo, p, pid, uid, why, status);
L
Linus Torvalds 已提交
1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214
	}

	/*
	 * 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;
	}

1215
	traced = ptrace_reparented(p);
1216 1217 1218 1219 1220
	/*
	 * It can be ptraced but not reparented, check
	 * !task_detached() to filter out sub-threads.
	 */
	if (likely(!traced) && likely(!task_detached(p))) {
1221 1222
		struct signal_struct *psig;
		struct signal_struct *sig;
J
Jiri Pirko 已提交
1223
		unsigned long maxrss;
1224
		cputime_t tgutime, tgstime;
1225

L
Linus Torvalds 已提交
1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236
		/*
		 * 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
1237
		 * need to protect the access to parent->signal fields,
L
Linus Torvalds 已提交
1238 1239
		 * as other threads in the parent group can be right
		 * here reaping other children at the same time.
1240 1241 1242 1243
		 *
		 * We use thread_group_times() to get times for the thread
		 * group, which consolidates times for all threads in the
		 * group including the group leader.
L
Linus Torvalds 已提交
1244
		 */
1245
		thread_group_times(p, &tgutime, &tgstime);
1246 1247
		spin_lock_irq(&p->real_parent->sighand->siglock);
		psig = p->real_parent->signal;
1248 1249 1250
		sig = p->signal;
		psig->cutime =
			cputime_add(psig->cutime,
1251 1252
			cputime_add(tgutime,
				    sig->cutime));
1253 1254
		psig->cstime =
			cputime_add(psig->cstime,
1255 1256
			cputime_add(tgstime,
				    sig->cstime));
1257 1258 1259 1260 1261
		psig->cgtime =
			cputime_add(psig->cgtime,
			cputime_add(p->gtime,
			cputime_add(sig->gtime,
				    sig->cgtime)));
1262 1263 1264 1265 1266 1267 1268 1269
		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;
1270 1271 1272 1273 1274 1275
		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
1276 1277 1278
		maxrss = max(sig->maxrss, sig->cmaxrss);
		if (psig->cmaxrss < maxrss)
			psig->cmaxrss = maxrss;
1279 1280
		task_io_accounting_add(&psig->ioac, &p->ioac);
		task_io_accounting_add(&psig->ioac, &sig->ioac);
1281
		spin_unlock_irq(&p->real_parent->sighand->siglock);
L
Linus Torvalds 已提交
1282 1283 1284 1285 1286 1287 1288 1289
	}

	/*
	 * 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);

1290 1291
	retval = wo->wo_rusage
		? getrusage(p, RUSAGE_BOTH, wo->wo_rusage) : 0;
L
Linus Torvalds 已提交
1292 1293
	status = (p->signal->flags & SIGNAL_GROUP_EXIT)
		? p->signal->group_exit_code : p->exit_code;
1294 1295 1296 1297
	if (!retval && wo->wo_stat)
		retval = put_user(status, wo->wo_stat);

	infop = wo->wo_info;
L
Linus Torvalds 已提交
1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316
	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)
1317
		retval = put_user(pid, &infop->si_pid);
L
Linus Torvalds 已提交
1318
	if (!retval && infop)
1319
		retval = put_user(uid, &infop->si_uid);
1320
	if (!retval)
1321
		retval = pid;
1322 1323

	if (traced) {
L
Linus Torvalds 已提交
1324
		write_lock_irq(&tasklist_lock);
1325 1326 1327 1328 1329 1330 1331
		/* 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.
		 */
1332
		if (!task_detached(p)) {
1333
			do_notify_parent(p, p->exit_signal);
1334
			if (!task_detached(p)) {
1335 1336
				p->exit_state = EXIT_ZOMBIE;
				p = NULL;
L
Linus Torvalds 已提交
1337 1338 1339 1340 1341 1342
			}
		}
		write_unlock_irq(&tasklist_lock);
	}
	if (p != NULL)
		release_task(p);
1343

L
Linus Torvalds 已提交
1344 1345 1346
	return retval;
}

1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358
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
1359 1360 1361 1362 1363 1364
/*
 * 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.
 */
1365 1366
static int wait_task_stopped(struct wait_opts *wo,
				int ptrace, struct task_struct *p)
L
Linus Torvalds 已提交
1367
{
1368
	struct siginfo __user *infop;
1369
	int retval, exit_code, *p_code, why;
1370
	uid_t uid = 0; /* unneeded, required by compiler */
1371
	pid_t pid;
L
Linus Torvalds 已提交
1372

1373 1374 1375
	/*
	 * Traditionally we see ptrace'd stopped tasks regardless of options.
	 */
1376
	if (!ptrace && !(wo->wo_flags & WUNTRACED))
R
Roland McGrath 已提交
1377 1378
		return 0;

1379 1380 1381
	exit_code = 0;
	spin_lock_irq(&p->sighand->siglock);

1382 1383
	p_code = task_stopped_code(p, ptrace);
	if (unlikely(!p_code))
1384 1385
		goto unlock_sig;

1386
	exit_code = *p_code;
1387 1388 1389
	if (!exit_code)
		goto unlock_sig;

1390
	if (!unlikely(wo->wo_flags & WNOWAIT))
1391
		*p_code = 0;
1392

1393 1394
	/* don't need the RCU readlock here as we're holding a spinlock */
	uid = __task_cred(p)->uid;
1395 1396 1397
unlock_sig:
	spin_unlock_irq(&p->sighand->siglock);
	if (!exit_code)
L
Linus Torvalds 已提交
1398 1399 1400 1401 1402 1403 1404 1405 1406 1407
		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);
1408
	pid = task_pid_vnr(p);
R
Roland McGrath 已提交
1409
	why = ptrace ? CLD_TRAPPED : CLD_STOPPED;
L
Linus Torvalds 已提交
1410 1411
	read_unlock(&tasklist_lock);

1412 1413 1414 1415 1416 1417 1418
	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
1419

1420
	infop = wo->wo_info;
L
Linus Torvalds 已提交
1421 1422 1423 1424 1425
	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
1426
		retval = put_user((short)why, &infop->si_code);
L
Linus Torvalds 已提交
1427 1428 1429
	if (!retval && infop)
		retval = put_user(exit_code, &infop->si_status);
	if (!retval && infop)
1430
		retval = put_user(pid, &infop->si_pid);
L
Linus Torvalds 已提交
1431
	if (!retval && infop)
1432
		retval = put_user(uid, &infop->si_uid);
L
Linus Torvalds 已提交
1433
	if (!retval)
1434
		retval = pid;
L
Linus Torvalds 已提交
1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446
	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.
 */
1447
static int wait_task_continued(struct wait_opts *wo, struct task_struct *p)
L
Linus Torvalds 已提交
1448 1449 1450 1451 1452
{
	int retval;
	pid_t pid;
	uid_t uid;

1453
	if (!unlikely(wo->wo_flags & WCONTINUED))
R
Roland McGrath 已提交
1454 1455
		return 0;

L
Linus Torvalds 已提交
1456 1457 1458 1459 1460 1461 1462 1463 1464
	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;
	}
1465
	if (!unlikely(wo->wo_flags & WNOWAIT))
L
Linus Torvalds 已提交
1466
		p->signal->flags &= ~SIGNAL_STOP_CONTINUED;
1467
	uid = __task_cred(p)->uid;
L
Linus Torvalds 已提交
1468 1469
	spin_unlock_irq(&p->sighand->siglock);

1470
	pid = task_pid_vnr(p);
L
Linus Torvalds 已提交
1471 1472 1473
	get_task_struct(p);
	read_unlock(&tasklist_lock);

1474 1475 1476
	if (!wo->wo_info) {
		retval = wo->wo_rusage
			? getrusage(p, RUSAGE_BOTH, wo->wo_rusage) : 0;
L
Linus Torvalds 已提交
1477
		put_task_struct(p);
1478 1479
		if (!retval && wo->wo_stat)
			retval = put_user(0xffff, wo->wo_stat);
L
Linus Torvalds 已提交
1480
		if (!retval)
1481
			retval = pid;
L
Linus Torvalds 已提交
1482
	} else {
1483 1484
		retval = wait_noreap_copyout(wo, p, pid, uid,
					     CLD_CONTINUED, SIGCONT);
L
Linus Torvalds 已提交
1485 1486 1487 1488 1489 1490
		BUG_ON(retval == 0);
	}

	return retval;
}

R
Roland McGrath 已提交
1491 1492 1493
/*
 * Consider @p for a wait by @parent.
 *
1494
 * -ECHILD should be in ->notask_error before the first call.
R
Roland McGrath 已提交
1495 1496
 * Returns nonzero for a final return, when we have unlocked tasklist_lock.
 * Returns zero if the search for a child should continue;
1497
 * then ->notask_error is 0 if @p is an eligible child,
1498
 * or another error from security_task_wait(), or still -ECHILD.
R
Roland McGrath 已提交
1499
 */
1500 1501
static int wait_consider_task(struct wait_opts *wo, int ptrace,
				struct task_struct *p)
R
Roland McGrath 已提交
1502
{
1503
	int ret = eligible_child(wo, p);
1504
	if (!ret)
R
Roland McGrath 已提交
1505 1506
		return ret;

1507
	ret = security_task_wait(p);
1508 1509 1510 1511 1512 1513 1514 1515
	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.
		 */
1516 1517
		if (wo->notask_error)
			wo->notask_error = ret;
1518
		return 0;
1519 1520
	}

1521
	if (likely(!ptrace) && unlikely(task_ptrace(p))) {
R
Roland McGrath 已提交
1522 1523 1524 1525
		/*
		 * This child is hidden by ptrace.
		 * We aren't allowed to see it now, but eventually we will.
		 */
1526
		wo->notask_error = 0;
R
Roland McGrath 已提交
1527 1528 1529
		return 0;
	}

R
Roland McGrath 已提交
1530 1531 1532 1533 1534 1535 1536
	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))
1537
		return wait_task_zombie(wo, p);
R
Roland McGrath 已提交
1538 1539 1540 1541 1542

	/*
	 * It's stopped or running now, so it might
	 * later continue, exit, or stop again.
	 */
1543
	wo->notask_error = 0;
R
Roland McGrath 已提交
1544

1545
	if (task_stopped_code(p, ptrace))
1546
		return wait_task_stopped(wo, ptrace, p);
R
Roland McGrath 已提交
1547

1548
	return wait_task_continued(wo, p);
R
Roland McGrath 已提交
1549 1550 1551 1552 1553
}

/*
 * Do the work of do_wait() for one thread in the group, @tsk.
 *
1554
 * -ECHILD should be in ->notask_error before the first call.
R
Roland McGrath 已提交
1555 1556
 * Returns nonzero for a final return, when we have unlocked tasklist_lock.
 * Returns zero if the search for a child should continue; then
1557
 * ->notask_error is 0 if there were any eligible children,
1558
 * or another error from security_task_wait(), or still -ECHILD.
R
Roland McGrath 已提交
1559
 */
1560
static int do_wait_thread(struct wait_opts *wo, struct task_struct *tsk)
R
Roland McGrath 已提交
1561 1562 1563 1564
{
	struct task_struct *p;

	list_for_each_entry(p, &tsk->children, sibling) {
1565 1566 1567
		int ret = wait_consider_task(wo, 0, p);
		if (ret)
			return ret;
R
Roland McGrath 已提交
1568 1569 1570 1571 1572
	}

	return 0;
}

1573
static int ptrace_do_wait(struct wait_opts *wo, struct task_struct *tsk)
R
Roland McGrath 已提交
1574 1575 1576
{
	struct task_struct *p;

R
Roland McGrath 已提交
1577
	list_for_each_entry(p, &tsk->ptraced, ptrace_entry) {
1578
		int ret = wait_consider_task(wo, 1, p);
R
Roland McGrath 已提交
1579
		if (ret)
R
Roland McGrath 已提交
1580 1581 1582 1583 1584 1585
			return ret;
	}

	return 0;
}

1586 1587 1588 1589 1590 1591 1592
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;

1593
	if (!eligible_pid(wo, p))
1594 1595
		return 0;

1596 1597 1598
	if ((wo->wo_flags & __WNOTHREAD) && wait->private != p->parent)
		return 0;

1599 1600 1601
	return default_wake_function(wait, mode, sync, key);
}

1602 1603
void __wake_up_parent(struct task_struct *p, struct task_struct *parent)
{
1604 1605
	__wake_up_sync_key(&parent->signal->wait_chldexit,
				TASK_INTERRUPTIBLE, 1, p);
1606 1607
}

1608
static long do_wait(struct wait_opts *wo)
L
Linus Torvalds 已提交
1609 1610
{
	struct task_struct *tsk;
R
Roland McGrath 已提交
1611
	int retval;
L
Linus Torvalds 已提交
1612

1613
	trace_sched_process_wait(wo->wo_pid);
1614

1615 1616 1617
	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
1618
repeat:
R
Roland McGrath 已提交
1619 1620
	/*
	 * If there is nothing that can match our critiera just get out.
1621 1622 1623
	 * 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
1624
	 */
1625
	wo->notask_error = -ECHILD;
1626 1627
	if ((wo->wo_type < PIDTYPE_MAX) &&
	   (!wo->wo_pid || hlist_empty(&wo->wo_pid->tasks[wo->wo_type])))
1628
		goto notask;
E
Eric W. Biederman 已提交
1629

1630
	set_current_state(TASK_INTERRUPTIBLE);
L
Linus Torvalds 已提交
1631 1632 1633
	read_lock(&tasklist_lock);
	tsk = current;
	do {
1634 1635 1636
		retval = do_wait_thread(wo, tsk);
		if (retval)
			goto end;
1637

1638 1639
		retval = ptrace_do_wait(wo, tsk);
		if (retval)
R
Roland McGrath 已提交
1640 1641
			goto end;

1642
		if (wo->wo_flags & __WNOTHREAD)
L
Linus Torvalds 已提交
1643
			break;
1644
	} while_each_thread(current, tsk);
L
Linus Torvalds 已提交
1645
	read_unlock(&tasklist_lock);
O
Oleg Nesterov 已提交
1646

1647
notask:
1648 1649
	retval = wo->notask_error;
	if (!retval && !(wo->wo_flags & WNOHANG)) {
L
Linus Torvalds 已提交
1650
		retval = -ERESTARTSYS;
R
Roland McGrath 已提交
1651 1652 1653 1654
		if (!signal_pending(current)) {
			schedule();
			goto repeat;
		}
L
Linus Torvalds 已提交
1655 1656
	}
end:
1657
	__set_current_state(TASK_RUNNING);
1658
	remove_wait_queue(&current->signal->wait_chldexit, &wo->child_wait);
L
Linus Torvalds 已提交
1659 1660 1661
	return retval;
}

1662 1663
SYSCALL_DEFINE5(waitid, int, which, pid_t, upid, struct siginfo __user *,
		infop, int, options, struct rusage __user *, ru)
L
Linus Torvalds 已提交
1664
{
1665
	struct wait_opts wo;
E
Eric W. Biederman 已提交
1666 1667
	struct pid *pid = NULL;
	enum pid_type type;
L
Linus Torvalds 已提交
1668 1669 1670 1671 1672 1673 1674 1675 1676
	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
1677
		type = PIDTYPE_MAX;
L
Linus Torvalds 已提交
1678 1679
		break;
	case P_PID:
E
Eric W. Biederman 已提交
1680 1681
		type = PIDTYPE_PID;
		if (upid <= 0)
L
Linus Torvalds 已提交
1682 1683 1684
			return -EINVAL;
		break;
	case P_PGID:
E
Eric W. Biederman 已提交
1685 1686
		type = PIDTYPE_PGID;
		if (upid <= 0)
L
Linus Torvalds 已提交
1687 1688 1689 1690 1691 1692
			return -EINVAL;
		break;
	default:
		return -EINVAL;
	}

E
Eric W. Biederman 已提交
1693 1694
	if (type < PIDTYPE_MAX)
		pid = find_get_pid(upid);
1695 1696 1697 1698 1699 1700 1701 1702

	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);
1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725

	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
1726
	put_pid(pid);
L
Linus Torvalds 已提交
1727 1728

	/* avoid REGPARM breakage on x86: */
1729
	asmlinkage_protect(5, ret, which, upid, infop, options, ru);
L
Linus Torvalds 已提交
1730 1731 1732
	return ret;
}

1733 1734
SYSCALL_DEFINE4(wait4, pid_t, upid, int __user *, stat_addr,
		int, options, struct rusage __user *, ru)
L
Linus Torvalds 已提交
1735
{
1736
	struct wait_opts wo;
E
Eric W. Biederman 已提交
1737 1738
	struct pid *pid = NULL;
	enum pid_type type;
L
Linus Torvalds 已提交
1739 1740 1741 1742 1743
	long ret;

	if (options & ~(WNOHANG|WUNTRACED|WCONTINUED|
			__WNOTHREAD|__WCLONE|__WALL))
		return -EINVAL;
E
Eric W. Biederman 已提交
1744 1745 1746 1747 1748 1749 1750 1751

	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;
1752
		pid = get_task_pid(current, PIDTYPE_PGID);
E
Eric W. Biederman 已提交
1753 1754 1755 1756 1757
	} else /* upid > 0 */ {
		type = PIDTYPE_PID;
		pid = find_get_pid(upid);
	}

1758 1759 1760 1761 1762 1763 1764
	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
1765
	put_pid(pid);
L
Linus Torvalds 已提交
1766 1767

	/* avoid REGPARM breakage on x86: */
1768
	asmlinkage_protect(4, ret, upid, stat_addr, options, ru);
L
Linus Torvalds 已提交
1769 1770 1771 1772 1773 1774 1775 1776 1777
	return ret;
}

#ifdef __ARCH_WANT_SYS_WAITPID

/*
 * sys_waitpid() remains for compatibility. waitpid() should be
 * implemented by calling sys_wait4() from libc.a.
 */
1778
SYSCALL_DEFINE3(waitpid, pid_t, pid, int __user *, stat_addr, int, options)
L
Linus Torvalds 已提交
1779 1780 1781 1782 1783
{
	return sys_wait4(pid, stat_addr, options, NULL);
}

#endif