exit.c 46.7 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>
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>
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() */
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>
49
#include <linux/init_task.h>
50
#include <linux/perf_event.h>
51
#include <trace/events/sched.h>
52
#include <linux/hw_breakpoint.h>
53
#include <linux/oom.h>
54
#include <linux/writeback.h>
L
Linus Torvalds 已提交
55 56 57 58 59 60

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

61 62
static void exit_mm(struct task_struct * tsk);

63
static void __unhash_process(struct task_struct *p, bool group_dead)
L
Linus Torvalds 已提交
64 65 66
{
	nr_threads--;
	detach_pid(p, PIDTYPE_PID);
67
	if (group_dead) {
L
Linus Torvalds 已提交
68 69
		detach_pid(p, PIDTYPE_PGID);
		detach_pid(p, PIDTYPE_SID);
70

71
		list_del_rcu(&p->tasks);
72
		list_del_init(&p->sibling);
73
		__this_cpu_dec(process_counts);
L
Linus Torvalds 已提交
74
	}
75
	list_del_rcu(&p->thread_group);
L
Linus Torvalds 已提交
76 77
}

78 79 80 81 82 83
/*
 * This function expects the tasklist_lock write-locked.
 */
static void __exit_signal(struct task_struct *tsk)
{
	struct signal_struct *sig = tsk->signal;
84
	bool group_dead = thread_group_leader(tsk);
85
	struct sighand_struct *sighand;
86
	struct tty_struct *uninitialized_var(tty);
87

88
	sighand = rcu_dereference_check(tsk->sighand,
89
					lockdep_tasklist_lock_is_held());
90 91 92
	spin_lock(&sighand->siglock);

	posix_cpu_timers_exit(tsk);
93
	if (group_dead) {
94
		posix_cpu_timers_exit_group(tsk);
95 96
		tty = sig->tty;
		sig->tty = NULL;
97
	} else {
98 99 100 101 102 103 104 105
		/*
		 * This can only happen if the caller is de_thread().
		 * FIXME: this is the temporary hack, we should teach
		 * posix-cpu-timers to handle this case correctly.
		 */
		if (unlikely(has_group_leader_pid(tsk)))
			posix_cpu_timers_exit_group(tsk);

106 107 108 109
		/*
		 * If there is any task waiting for the group exit
		 * then notify it:
		 */
110
		if (sig->notify_count > 0 && !--sig->notify_count)
111
			wake_up_process(sig->group_exit_task);
112

113 114 115 116 117 118 119 120 121 122 123 124
		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.
		 */
125 126 127
		sig->utime += tsk->utime;
		sig->stime += tsk->stime;
		sig->gtime += tsk->gtime;
128 129 130 131
		sig->min_flt += tsk->min_flt;
		sig->maj_flt += tsk->maj_flt;
		sig->nvcsw += tsk->nvcsw;
		sig->nivcsw += tsk->nivcsw;
132 133
		sig->inblock += task_io_get_inblock(tsk);
		sig->oublock += task_io_get_oublock(tsk);
134
		task_io_accounting_add(&sig->ioac, &tsk->ioac);
135
		sig->sum_sched_runtime += tsk->se.sum_exec_runtime;
136 137
	}

138
	sig->nr_threads--;
139
	__unhash_process(tsk, group_dead);
140

141 142 143 144 145
	/*
	 * Do this under ->siglock, we can race with another thread
	 * doing sigqueue_free() if we have SIGQUEUE_PREALLOC signals.
	 */
	flush_sigqueue(&tsk->pending);
146
	tsk->sighand = NULL;
147 148
	spin_unlock(&sighand->siglock);

149
	__cleanup_sighand(sighand);
150
	clear_tsk_thread_flag(tsk,TIF_SIGPENDING);
151
	if (group_dead) {
152
		flush_sigqueue(&sig->shared_pending);
153
		tty_kref_put(tty);
154 155 156
	}
}

157 158
static void delayed_put_task_struct(struct rcu_head *rhp)
{
159 160
	struct task_struct *tsk = container_of(rhp, struct task_struct, rcu);

161
	perf_event_delayed_put(tsk);
162 163
	trace_sched_process_free(tsk);
	put_task_struct(tsk);
164 165
}

R
Roland McGrath 已提交
166

L
Linus Torvalds 已提交
167 168
void release_task(struct task_struct * p)
{
169
	struct task_struct *leader;
L
Linus Torvalds 已提交
170
	int zap_leader;
171
repeat:
172
	/* don't need to get the RCU readlock here - the process is dead and
173 174
	 * can't be modifying its own credentials. But shut RCU-lockdep up */
	rcu_read_lock();
175
	atomic_dec(&__task_cred(p)->user->processes);
176
	rcu_read_unlock();
177

178
	proc_flush_task(p);
179

L
Linus Torvalds 已提交
180
	write_lock_irq(&tasklist_lock);
181
	ptrace_release_task(p);
L
Linus Torvalds 已提交
182
	__exit_signal(p);
183

L
Linus Torvalds 已提交
184 185 186 187 188 189 190 191 192 193 194 195
	/*
	 * If we are the last non-leader member of the thread
	 * group, and the leader is zombie, then notify the
	 * group leader's parent process. (if it wants notification.)
	 */
	zap_leader = 0;
	leader = p->group_leader;
	if (leader != p && thread_group_empty(leader) && leader->exit_state == EXIT_ZOMBIE) {
		/*
		 * If we were the last child thread and the leader has
		 * exited already, and the leader's parent ignores SIGCHLD,
		 * then we are the one who should release the leader.
R
Roland McGrath 已提交
196
		 */
197
		zap_leader = do_notify_parent(leader, leader->exit_signal);
R
Roland McGrath 已提交
198 199
		if (zap_leader)
			leader->exit_state = EXIT_DEAD;
L
Linus Torvalds 已提交
200 201 202 203
	}

	write_unlock_irq(&tasklist_lock);
	release_thread(p);
204
	call_rcu(&p->rcu, delayed_put_task_struct);
L
Linus Torvalds 已提交
205 206 207 208 209 210 211 212 213 214

	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...
215 216
 *
 * The caller must hold rcu lock or the tasklist lock.
L
Linus Torvalds 已提交
217
 */
218
struct pid *session_of_pgrp(struct pid *pgrp)
L
Linus Torvalds 已提交
219 220
{
	struct task_struct *p;
221
	struct pid *sid = NULL;
222

223
	p = pid_task(pgrp, PIDTYPE_PGID);
224
	if (p == NULL)
225
		p = pid_task(pgrp, PIDTYPE_PID);
226
	if (p != NULL)
227
		sid = task_session(p);
228

L
Linus Torvalds 已提交
229 230 231 232 233 234 235 236 237 238 239
	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?"
 */
240
static int will_become_orphaned_pgrp(struct pid *pgrp, struct task_struct *ignored_task)
L
Linus Torvalds 已提交
241 242 243
{
	struct task_struct *p;

244
	do_each_pid_task(pgrp, PIDTYPE_PGID, p) {
245 246 247
		if ((p == ignored_task) ||
		    (p->exit_state && thread_group_empty(p)) ||
		    is_global_init(p->real_parent))
L
Linus Torvalds 已提交
248
			continue;
249

250
		if (task_pgrp(p->real_parent) != pgrp &&
251 252
		    task_session(p->real_parent) == task_session(p))
			return 0;
253
	} while_each_pid_task(pgrp, PIDTYPE_PGID, p);
254 255

	return 1;
L
Linus Torvalds 已提交
256 257
}

258
int is_current_pgrp_orphaned(void)
L
Linus Torvalds 已提交
259 260 261 262
{
	int retval;

	read_lock(&tasklist_lock);
263
	retval = will_become_orphaned_pgrp(task_pgrp(current), NULL);
L
Linus Torvalds 已提交
264 265 266 267 268
	read_unlock(&tasklist_lock);

	return retval;
}

269
static bool has_stopped_jobs(struct pid *pgrp)
L
Linus Torvalds 已提交
270 271 272
{
	struct task_struct *p;

273
	do_each_pid_task(pgrp, PIDTYPE_PGID, p) {
274 275
		if (p->signal->flags & SIGNAL_STOP_STOPPED)
			return true;
276
	} while_each_pid_task(pgrp, PIDTYPE_PGID, p);
277 278

	return false;
L
Linus Torvalds 已提交
279 280
}

281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311
/*
 * 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
312
/**
313
 * reparent_to_kthreadd - Reparent the calling kernel thread to kthreadd
L
Linus Torvalds 已提交
314 315
 *
 * If a kernel thread is launched as a result of a system call, or if
316 317
 * 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
318 319 320 321
 *
 * 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.
 *
322
 * NOTE that reparent_to_kthreadd() gives the caller full capabilities.
L
Linus Torvalds 已提交
323
 */
324
static void reparent_to_kthreadd(void)
L
Linus Torvalds 已提交
325 326 327 328 329
{
	write_lock_irq(&tasklist_lock);

	ptrace_unlink(current);
	/* Reparent to init */
330
	current->real_parent = current->parent = kthreadd_task;
R
Roland McGrath 已提交
331
	list_move_tail(&current->sibling, &current->real_parent->children);
L
Linus Torvalds 已提交
332 333 334 335

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

336
	if (task_nice(current) < 0)
L
Linus Torvalds 已提交
337 338 339 340 341 342
		set_user_nice(current, 0);
	/* cpus_allowed? */
	/* rt_priority? */
	/* signals? */
	memcpy(current->signal->rlim, init_task.signal->rlim,
	       sizeof(current->signal->rlim));
343 344 345

	atomic_inc(&init_cred.usage);
	commit_creds(&init_cred);
L
Linus Torvalds 已提交
346 347 348
	write_unlock_irq(&tasklist_lock);
}

349
void __set_special_pids(struct pid *pid)
L
Linus Torvalds 已提交
350
{
351
	struct task_struct *curr = current->group_leader;
L
Linus Torvalds 已提交
352

353
	if (task_session(curr) != pid)
354
		change_pid(curr, PIDTYPE_SID, pid);
355 356

	if (task_pgrp(curr) != pid)
357
		change_pid(curr, PIDTYPE_PGID, pid);
L
Linus Torvalds 已提交
358 359
}

360
static void set_special_pids(struct pid *pid)
L
Linus Torvalds 已提交
361 362
{
	write_lock_irq(&tasklist_lock);
363
	__set_special_pids(pid);
L
Linus Torvalds 已提交
364 365 366 367
	write_unlock_irq(&tasklist_lock);
}

/*
368 369
 * 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
370 371 372
 */
int allow_signal(int sig)
{
373
	if (!valid_signal(sig) || sig < 1)
L
Linus Torvalds 已提交
374 375 376
		return -EINVAL;

	spin_lock_irq(&current->sighand->siglock);
377
	/* This is only needed for daemonize()'ed kthreads */
L
Linus Torvalds 已提交
378
	sigdelset(&current->blocked, sig);
379 380 381 382 383 384
	/*
	 * 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
385 386 387 388 389 390 391 392 393
	recalc_sigpending();
	spin_unlock_irq(&current->sighand->siglock);
	return 0;
}

EXPORT_SYMBOL(allow_signal);

int disallow_signal(int sig)
{
394
	if (!valid_signal(sig) || sig < 1)
L
Linus Torvalds 已提交
395 396 397
		return -EINVAL;

	spin_lock_irq(&current->sighand->siglock);
398
	current->sighand->action[(sig)-1].sa.sa_handler = SIG_IGN;
L
Linus Torvalds 已提交
399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425
	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);
426 427 428 429
	/*
	 * 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
430
	current->flags |= (PF_NOFREEZE | PF_KTHREAD);
L
Linus Torvalds 已提交
431

432 433 434 435
	if (current->nsproxy != &init_nsproxy) {
		get_nsproxy(&init_nsproxy);
		switch_task_namespaces(current, &init_nsproxy);
	}
436
	set_special_pids(&init_struct_pid);
437
	proc_clear_tty(current);
L
Linus Torvalds 已提交
438 439 440 441 442 443 444 445

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

	/* Become as one with the init task */

446
	daemonize_fs_struct();
447
	exit_files(current);
L
Linus Torvalds 已提交
448 449 450
	current->files = init_task.files;
	atomic_inc(&current->files->count);

451
	reparent_to_kthreadd();
L
Linus Torvalds 已提交
452 453 454 455
}

EXPORT_SYMBOL(daemonize);

456
static void close_files(struct files_struct * files)
L
Linus Torvalds 已提交
457 458
{
	int i, j;
459
	struct fdtable *fdt;
L
Linus Torvalds 已提交
460 461

	j = 0;
462 463 464 465

	/*
	 * It is safe to dereference the fd table without RCU or
	 * ->file_lock because this is the last reference to the
466
	 * files structure.  But use RCU to shut RCU-lockdep up.
467
	 */
468
	rcu_read_lock();
469
	fdt = files_fdtable(files);
470
	rcu_read_unlock();
L
Linus Torvalds 已提交
471 472 473
	for (;;) {
		unsigned long set;
		i = j * __NFDBITS;
474
		if (i >= fdt->max_fds)
L
Linus Torvalds 已提交
475
			break;
476
		set = fdt->open_fds->fds_bits[j++];
L
Linus Torvalds 已提交
477 478
		while (set) {
			if (set & 1) {
479
				struct file * file = xchg(&fdt->fd[i], NULL);
480
				if (file) {
L
Linus Torvalds 已提交
481
					filp_close(file, files);
482 483
					cond_resched();
				}
L
Linus Torvalds 已提交
484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503
			}
			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;
}

504
void put_files_struct(struct files_struct *files)
L
Linus Torvalds 已提交
505
{
506 507
	struct fdtable *fdt;

L
Linus Torvalds 已提交
508 509 510 511
	if (atomic_dec_and_test(&files->count)) {
		close_files(files);
		/*
		 * Free the fd and fdset arrays if we expanded them.
512 513 514
		 * 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
515
		 */
516
		rcu_read_lock();
517
		fdt = files_fdtable(files);
518
		if (fdt != &files->fdtab)
519
			kmem_cache_free(files_cachep, files);
520
		free_fdtable(fdt);
521
		rcu_read_unlock();
L
Linus Torvalds 已提交
522 523 524
	}
}

525
void reset_files_struct(struct files_struct *files)
526
{
527
	struct task_struct *tsk = current;
528 529 530 531 532 533 534 535 536
	struct files_struct *old;

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

537
void exit_files(struct task_struct *tsk)
L
Linus Torvalds 已提交
538 539 540 541 542 543 544 545 546 547 548
{
	struct files_struct * files = tsk->files;

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

549 550
#ifdef CONFIG_MM_OWNER
/*
551
 * A task is exiting.   If it owned this mm, find a new owner for the mm.
552 553 554 555 556 557
 */
void mm_update_next_owner(struct mm_struct *mm)
{
	struct task_struct *c, *g, *p = current;

retry:
558 559 560 561 562
	/*
	 * If the exiting or execing task is not the owner, it's
	 * someone else's problem.
	 */
	if (mm->owner != p)
563
		return;
564 565 566 567 568 569 570 571 572
	/*
	 * The current owner is exiting/execing and there are no other
	 * candidates.  Do not leave the mm pointing to a possibly
	 * freed task structure.
	 */
	if (atomic_read(&mm->mm_users) <= 1) {
		mm->owner = NULL;
		return;
	}
573 574 575 576 577 578 579 580 581 582 583 584 585

	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
	 */
586
	list_for_each_entry(c, &p->real_parent->children, sibling) {
587 588 589 590 591 592 593 594 595 596 597 598 599 600
		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);
601 602 603
	/*
	 * We found no owner yet mm_users > 1: this implies that we are
	 * most likely racing with swapoff (try_to_unuse()) or /proc or
604
	 * ptrace or page migration (get_task_mm()).  Mark owner as NULL.
605 606
	 */
	mm->owner = NULL;
607 608 609 610 611 612 613 614 615 616
	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);
617 618 619 620 621
	/*
	 * Delay read_unlock() till we have the task_lock()
	 * to ensure that c does not slip away underneath us
	 */
	read_unlock(&tasklist_lock);
622 623 624 625 626 627 628 629 630 631 632
	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
633 634 635 636
/*
 * Turn us into a lazy TLB process if we
 * aren't already..
 */
637
static void exit_mm(struct task_struct * tsk)
L
Linus Torvalds 已提交
638 639
{
	struct mm_struct *mm = tsk->mm;
640
	struct core_state *core_state;
L
Linus Torvalds 已提交
641 642 643 644 645 646

	mm_release(tsk, mm);
	if (!mm)
		return;
	/*
	 * Serialize with any possible pending coredump.
647
	 * We must hold mmap_sem around checking core_state
L
Linus Torvalds 已提交
648
	 * and clearing tsk->mm.  The core-inducing thread
649
	 * will increment ->nr_threads for each thread in the
L
Linus Torvalds 已提交
650 651 652
	 * group with ->mm != NULL.
	 */
	down_read(&mm->mmap_sem);
653 654 655
	core_state = mm->core_state;
	if (core_state) {
		struct core_thread self;
L
Linus Torvalds 已提交
656 657
		up_read(&mm->mmap_sem);

658 659 660 661 662 663 664 665
		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
666

667 668 669 670 671 672 673
		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
674 675 676
		down_read(&mm->mmap_sem);
	}
	atomic_inc(&mm->mm_count);
677
	BUG_ON(mm != tsk->active_mm);
L
Linus Torvalds 已提交
678 679 680 681 682 683
	/* more a memory barrier than a real lock */
	task_lock(tsk);
	tsk->mm = NULL;
	up_read(&mm->mmap_sem);
	enter_lazy_tlb(mm, current);
	task_unlock(tsk);
684
	mm_update_next_owner(mm);
L
Linus Torvalds 已提交
685 686 687 688 689 690 691
	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
692 693
 * the child reaper process (ie "init") in our pid
 * space.
L
Linus Torvalds 已提交
694
 */
695
static struct task_struct *find_new_reaper(struct task_struct *father)
696 697
	__releases(&tasklist_lock)
	__acquires(&tasklist_lock)
L
Linus Torvalds 已提交
698
{
699 700
	struct pid_namespace *pid_ns = task_active_pid_ns(father);
	struct task_struct *thread;
L
Linus Torvalds 已提交
701

702 703 704 705 706 707 708 709
	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
710

711 712 713 714
	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
715

716 717
		zap_pid_ns_processes(pid_ns);
		write_lock_irq(&tasklist_lock);
L
Linus Torvalds 已提交
718
		/*
719 720 721
		 * 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
722
		 */
723
		pid_ns->child_reaper = init_pid_ns.child_reaper;
L
Linus Torvalds 已提交
724
	}
725

726 727 728
	return pid_ns->child_reaper;
}

729 730 731
/*
* Any that need to be release_task'd are put on the @dead list.
 */
732
static void reparent_leader(struct task_struct *father, struct task_struct *p,
733 734 735 736
				struct list_head *dead)
{
	list_move_tail(&p->sibling, &p->real_parent->children);

737
	if (p->exit_state == EXIT_DEAD)
738 739 740 741 742 743 744 745 746 747 748 749
		return;
	/*
	 * If this is a threaded reparent there is no need to
	 * notify anyone anything has happened.
	 */
	if (same_thread_group(p->real_parent, father))
		return;

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

	/* If it has exited notify the new parent about this child's death. */
T
Tejun Heo 已提交
750
	if (!p->ptrace &&
751
	    p->exit_state == EXIT_ZOMBIE && thread_group_empty(p)) {
752
		if (do_notify_parent(p, p->exit_signal)) {
753 754 755 756 757 758 759 760
			p->exit_state = EXIT_DEAD;
			list_move_tail(&p->sibling, dead);
		}
	}

	kill_orphaned_pgrp(p, father);
}

761
static void forget_original_parent(struct task_struct *father)
L
Linus Torvalds 已提交
762
{
763
	struct task_struct *p, *n, *reaper;
764
	LIST_HEAD(dead_children);
765 766

	write_lock_irq(&tasklist_lock);
767 768 769 770 771
	/*
	 * Note that exit_ptrace() and find_new_reaper() might
	 * drop tasklist_lock and reacquire it.
	 */
	exit_ptrace(father);
772
	reaper = find_new_reaper(father);
R
Roland McGrath 已提交
773

774
	list_for_each_entry_safe(p, n, &father->children, sibling) {
775 776 777 778
		struct task_struct *t = p;
		do {
			t->real_parent = reaper;
			if (t->parent == father) {
T
Tejun Heo 已提交
779
				BUG_ON(t->ptrace);
780 781 782 783 784 785 786
				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
787
	}
788
	write_unlock_irq(&tasklist_lock);
789

790 791
	BUG_ON(!list_empty(&father->children));

792 793
	list_for_each_entry_safe(p, n, &dead_children, sibling) {
		list_del_init(&p->sibling);
794 795
		release_task(p);
	}
L
Linus Torvalds 已提交
796 797 798 799 800 801
}

/*
 * Send signals to all our closest relatives so that they know
 * to properly mourn us..
 */
802
static void exit_notify(struct task_struct *tsk, int group_dead)
L
Linus Torvalds 已提交
803
{
804
	bool autoreap;
L
Linus Torvalds 已提交
805 806 807 808 809 810 811 812 813

	/*
	 * 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)
	 */
814
	forget_original_parent(tsk);
815
	exit_task_namespaces(tsk);
L
Linus Torvalds 已提交
816

817
	write_lock_irq(&tasklist_lock);
818 819
	if (group_dead)
		kill_orphaned_pgrp(tsk->group_leader, NULL);
L
Linus Torvalds 已提交
820

821
	/* Let father know we died
L
Linus Torvalds 已提交
822 823
	 *
	 * Thread signals are configurable, but you aren't going to use
L
Lucas De Marchi 已提交
824
	 * that to send signals to arbitrary processes.
L
Linus Torvalds 已提交
825 826 827 828 829 830
	 * 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.
	 */
831
	if (thread_group_leader(tsk) && tsk->exit_signal != SIGCHLD &&
832
	    tsk->parent_exec_id != tsk->real_parent->self_exec_id)
L
Linus Torvalds 已提交
833 834
		tsk->exit_signal = SIGCHLD;

835 836 837 838 839 840 841 842 843 844 845 846
	if (unlikely(tsk->ptrace)) {
		int sig = thread_group_leader(tsk) &&
				thread_group_empty(tsk) &&
				!ptrace_reparented(tsk) ?
			tsk->exit_signal : SIGCHLD;
		autoreap = do_notify_parent(tsk, sig);
	} else if (thread_group_leader(tsk)) {
		autoreap = thread_group_empty(tsk) &&
			do_notify_parent(tsk, tsk->exit_signal);
	} else {
		autoreap = true;
	}
L
Linus Torvalds 已提交
847

848
	tsk->exit_state = autoreap ? EXIT_DEAD : EXIT_ZOMBIE;
L
Linus Torvalds 已提交
849

850 851
	/* mt-exec, de_thread() is waiting for group leader */
	if (unlikely(tsk->signal->notify_count < 0))
852
		wake_up_process(tsk->signal->group_exit_task);
L
Linus Torvalds 已提交
853 854 855
	write_unlock_irq(&tasklist_lock);

	/* If the process is dead, release it - nobody will wait for it */
856
	if (autoreap)
L
Linus Torvalds 已提交
857 858 859
		release_task(tsk);
}

860 861 862 863 864 865 866
#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;

867
	free = stack_not_used(current);
868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884

	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

885
void do_exit(long code)
L
Linus Torvalds 已提交
886 887 888 889 890 891
{
	struct task_struct *tsk = current;
	int group_dead;

	profile_task_exit(tsk);

892
	WARN_ON(blk_needs_flush_plug(tsk));
893

L
Linus Torvalds 已提交
894 895 896 897 898
	if (unlikely(in_interrupt()))
		panic("Aiee, killing interrupt handler!");
	if (unlikely(!tsk->pid))
		panic("Attempted to kill the idle task!");

899 900 901 902 903 904 905 906 907
	/*
	 * If do_exit is called because this processes oopsed, it's possible
	 * that get_fs() was left as KERNEL_DS, so reset it to USER_DS before
	 * continuing. Amongst other possible reasons, this is to prevent
	 * mm_release()->clear_child_tid() from writing to a user-controlled
	 * kernel address.
	 */
	set_fs(USER_DS);

908
	ptrace_event(PTRACE_EVENT_EXIT, code);
L
Linus Torvalds 已提交
909

910 911
	validate_creds_for_do_exit(tsk);

912 913 914 915 916 917 918
	/*
	 * 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");
919 920 921 922 923 924 925 926 927 928
		/*
		 * 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;
929 930 931 932
		set_current_state(TASK_UNINTERRUPTIBLE);
		schedule();
	}

933
	exit_signals(tsk);  /* sets PF_EXITING */
934 935 936 937
	/*
	 * tsk->flags are checked in the futex code to protect against
	 * an exiting task cleaning up the robust pi futexes.
	 */
938
	smp_mb();
939
	raw_spin_unlock_wait(&tsk->pi_lock);
L
Linus Torvalds 已提交
940

941 942
	exit_irq_thread();

L
Linus Torvalds 已提交
943 944
	if (unlikely(in_atomic()))
		printk(KERN_INFO "note: %s[%d] exited with preempt_count %d\n",
945
				current->comm, task_pid_nr(current),
L
Linus Torvalds 已提交
946 947 948
				preempt_count());

	acct_update_integrals(tsk);
949
	/* sync mm's RSS info before statistics gathering */
950 951
	if (tsk->mm)
		sync_mm_rss(tsk, tsk->mm);
L
Linus Torvalds 已提交
952
	group_dead = atomic_dec_and_test(&tsk->signal->live);
953
	if (group_dead) {
954
		hrtimer_cancel(&tsk->signal->real_timer);
955
		exit_itimers(tsk->signal);
956 957
		if (tsk->mm)
			setmax_mm_hiwater_rss(&tsk->signal->maxrss, tsk->mm);
958
	}
959
	acct_collect(code, group_dead);
960 961
	if (group_dead)
		tty_audit_exit();
962
	audit_free(tsk);
963

964
	tsk->exit_code = code;
965
	taskstats_exit(tsk, group_dead);
966

L
Linus Torvalds 已提交
967 968
	exit_mm(tsk);

969
	if (group_dead)
970
		acct_process();
971 972
	trace_sched_process_exit(tsk);

L
Linus Torvalds 已提交
973
	exit_sem(tsk);
974
	exit_shm(tsk);
975 976
	exit_files(tsk);
	exit_fs(tsk);
977
	check_stack_usage();
L
Linus Torvalds 已提交
978
	exit_thread();
979 980 981 982 983 984 985 986 987

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

988
	cgroup_exit(tsk, 1);
L
Linus Torvalds 已提交
989

990
	if (group_dead)
L
Linus Torvalds 已提交
991 992
		disassociate_ctty(1);

993
	module_put(task_thread_info(tsk)->exec_domain->module);
L
Linus Torvalds 已提交
994

995
	proc_exit_connector(tsk);
996

997 998 999
	/*
	 * FIXME: do that only when needed, using sched_exit tracepoint
	 */
1000
	ptrace_put_breakpoints(tsk);
1001

1002
	exit_notify(tsk, group_dead);
L
Linus Torvalds 已提交
1003
#ifdef CONFIG_NUMA
1004
	task_lock(tsk);
1005
	mpol_put(tsk->mempolicy);
L
Linus Torvalds 已提交
1006
	tsk->mempolicy = NULL;
1007
	task_unlock(tsk);
L
Linus Torvalds 已提交
1008
#endif
1009
#ifdef CONFIG_FUTEX
1010 1011
	if (unlikely(current->pi_state_cache))
		kfree(current->pi_state_cache);
1012
#endif
1013
	/*
1014
	 * Make sure we are holding no locks:
1015
	 */
1016
	debug_check_no_locks_held(tsk);
1017 1018 1019 1020 1021 1022
	/*
	 * 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
1023

1024
	if (tsk->io_context)
1025
		exit_io_context(tsk);
1026

1027 1028 1029
	if (tsk->splice_pipe)
		__free_pipe_info(tsk->splice_pipe);

1030 1031
	validate_creds_for_do_exit(tsk);

C
Coywolf Qi Hunt 已提交
1032
	preempt_disable();
1033 1034
	if (tsk->nr_dirtied)
		__this_cpu_add(dirty_throttle_leaks, tsk->nr_dirtied);
1035
	exit_rcu();
1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051

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

O
Oleg Nesterov 已提交
1052
	/* causes final put_task_struct in finish_task_switch(). */
1053
	tsk->state = TASK_DEAD;
1054
	tsk->flags |= PF_NOFREEZE;	/* tell freezer to ignore us */
L
Linus Torvalds 已提交
1055 1056 1057
	schedule();
	BUG();
	/* Avoid "noreturn function does return".  */
A
Alan Cox 已提交
1058 1059
	for (;;)
		cpu_relax();	/* For when BUG is null */
L
Linus Torvalds 已提交
1060 1061
}

1062 1063
EXPORT_SYMBOL_GPL(do_exit);

1064
void complete_and_exit(struct completion *comp, long code)
L
Linus Torvalds 已提交
1065 1066 1067
{
	if (comp)
		complete(comp);
O
Oleg Nesterov 已提交
1068

L
Linus Torvalds 已提交
1069 1070 1071 1072 1073
	do_exit(code);
}

EXPORT_SYMBOL(complete_and_exit);

1074
SYSCALL_DEFINE1(exit, int, error_code)
L
Linus Torvalds 已提交
1075 1076 1077 1078 1079 1080 1081 1082
{
	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).
 */
1083
void
L
Linus Torvalds 已提交
1084 1085
do_group_exit(int exit_code)
{
1086 1087
	struct signal_struct *sig = current->signal;

L
Linus Torvalds 已提交
1088 1089
	BUG_ON(exit_code & 0x80); /* core dumps don't get here */

1090 1091
	if (signal_group_exit(sig))
		exit_code = sig->group_exit_code;
L
Linus Torvalds 已提交
1092 1093 1094
	else if (!thread_group_empty(current)) {
		struct sighand_struct *const sighand = current->sighand;
		spin_lock_irq(&sighand->siglock);
1095
		if (signal_group_exit(sig))
L
Linus Torvalds 已提交
1096 1097 1098 1099
			/* Another thread got here before we took the lock.  */
			exit_code = sig->group_exit_code;
		else {
			sig->group_exit_code = exit_code;
1100
			sig->flags = SIGNAL_GROUP_EXIT;
L
Linus Torvalds 已提交
1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114
			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.
 */
1115
SYSCALL_DEFINE1(exit_group, int, error_code)
L
Linus Torvalds 已提交
1116 1117
{
	do_group_exit((error_code & 0xff) << 8);
1118 1119
	/* NOTREACHED */
	return 0;
L
Linus Torvalds 已提交
1120 1121
}

1122 1123 1124
struct wait_opts {
	enum pid_type		wo_type;
	int			wo_flags;
1125
	struct pid		*wo_pid;
1126 1127 1128 1129 1130

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

1131
	wait_queue_t		child_wait;
1132 1133 1134
	int			notask_error;
};

1135 1136
static inline
struct pid *task_pid_type(struct task_struct *task, enum pid_type type)
1137
{
1138 1139 1140
	if (type != PIDTYPE_PID)
		task = task->group_leader;
	return task->pids[type].pid;
1141 1142
}

1143
static int eligible_pid(struct wait_opts *wo, struct task_struct *p)
L
Linus Torvalds 已提交
1144
{
1145 1146 1147
	return	wo->wo_type == PIDTYPE_MAX ||
		task_pid_type(p, wo->wo_type) == wo->wo_pid;
}
L
Linus Torvalds 已提交
1148

1149 1150 1151 1152
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
1153 1154 1155 1156 1157
	/* 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.) */
1158 1159
	if (((p->exit_signal != SIGCHLD) ^ !!(wo->wo_flags & __WCLONE))
	    && !(wo->wo_flags & __WALL))
L
Linus Torvalds 已提交
1160 1161
		return 0;

1162
	return 1;
L
Linus Torvalds 已提交
1163 1164
}

1165 1166
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
1167
{
1168 1169 1170
	struct siginfo __user *infop;
	int retval = wo->wo_rusage
		? getrusage(p, RUSAGE_BOTH, wo->wo_rusage) : 0;
1171

L
Linus Torvalds 已提交
1172
	put_task_struct(p);
1173
	infop = wo->wo_info;
1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187
	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
1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198
	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.
 */
1199
static int wait_task_zombie(struct wait_opts *wo, struct task_struct *p)
L
Linus Torvalds 已提交
1200 1201
{
	unsigned long state;
1202
	int retval, status, traced;
1203
	pid_t pid = task_pid_vnr(p);
1204
	uid_t uid = __task_cred(p)->uid;
1205
	struct siginfo __user *infop;
L
Linus Torvalds 已提交
1206

1207
	if (!likely(wo->wo_flags & WEXITED))
R
Roland McGrath 已提交
1208 1209
		return 0;

1210
	if (unlikely(wo->wo_flags & WNOWAIT)) {
L
Linus Torvalds 已提交
1211
		int exit_code = p->exit_code;
1212
		int why;
L
Linus Torvalds 已提交
1213 1214 1215 1216 1217 1218 1219 1220 1221 1222

		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;
		}
1223
		return wait_noreap_copyout(wo, p, pid, uid, why, status);
L
Linus Torvalds 已提交
1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235
	}

	/*
	 * 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;
	}

1236
	traced = ptrace_reparented(p);
1237 1238
	/*
	 * It can be ptraced but not reparented, check
O
Oleg Nesterov 已提交
1239
	 * thread_group_leader() to filter out sub-threads.
1240
	 */
O
Oleg Nesterov 已提交
1241
	if (likely(!traced) && thread_group_leader(p)) {
1242 1243
		struct signal_struct *psig;
		struct signal_struct *sig;
1244
		unsigned long maxrss;
1245
		cputime_t tgutime, tgstime;
1246

L
Linus Torvalds 已提交
1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257
		/*
		 * 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
1258
		 * need to protect the access to parent->signal fields,
L
Linus Torvalds 已提交
1259 1260
		 * as other threads in the parent group can be right
		 * here reaping other children at the same time.
1261 1262 1263 1264
		 *
		 * 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
1265
		 */
1266
		thread_group_times(p, &tgutime, &tgstime);
1267 1268
		spin_lock_irq(&p->real_parent->sighand->siglock);
		psig = p->real_parent->signal;
1269
		sig = p->signal;
1270 1271 1272
		psig->cutime += tgutime + sig->cutime;
		psig->cstime += tgstime + sig->cstime;
		psig->cgtime += p->gtime + sig->gtime + sig->cgtime;
1273 1274 1275 1276 1277 1278 1279 1280
		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;
1281 1282 1283 1284 1285 1286
		psig->cinblock +=
			task_io_get_inblock(p) +
			sig->inblock + sig->cinblock;
		psig->coublock +=
			task_io_get_oublock(p) +
			sig->oublock + sig->coublock;
1287 1288 1289
		maxrss = max(sig->maxrss, sig->cmaxrss);
		if (psig->cmaxrss < maxrss)
			psig->cmaxrss = maxrss;
1290 1291
		task_io_accounting_add(&psig->ioac, &p->ioac);
		task_io_accounting_add(&psig->ioac, &sig->ioac);
1292
		spin_unlock_irq(&p->real_parent->sighand->siglock);
L
Linus Torvalds 已提交
1293 1294 1295 1296 1297 1298 1299 1300
	}

	/*
	 * 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);

1301 1302
	retval = wo->wo_rusage
		? getrusage(p, RUSAGE_BOTH, wo->wo_rusage) : 0;
L
Linus Torvalds 已提交
1303 1304
	status = (p->signal->flags & SIGNAL_GROUP_EXIT)
		? p->signal->group_exit_code : p->exit_code;
1305 1306 1307 1308
	if (!retval && wo->wo_stat)
		retval = put_user(status, wo->wo_stat);

	infop = wo->wo_info;
L
Linus Torvalds 已提交
1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327
	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)
1328
		retval = put_user(pid, &infop->si_pid);
L
Linus Torvalds 已提交
1329
	if (!retval && infop)
1330
		retval = put_user(uid, &infop->si_uid);
1331
	if (!retval)
1332
		retval = pid;
1333 1334

	if (traced) {
L
Linus Torvalds 已提交
1335
		write_lock_irq(&tasklist_lock);
1336 1337 1338
		/* We dropped tasklist, ptracer could die and untrace */
		ptrace_unlink(p);
		/*
1339 1340
		 * If this is not a sub-thread, notify the parent.
		 * If parent wants a zombie, don't release it now.
1341
		 */
1342 1343 1344 1345
		if (thread_group_leader(p) &&
		    !do_notify_parent(p, p->exit_signal)) {
			p->exit_state = EXIT_ZOMBIE;
			p = NULL;
L
Linus Torvalds 已提交
1346 1347 1348 1349 1350
		}
		write_unlock_irq(&tasklist_lock);
	}
	if (p != NULL)
		release_task(p);
1351

L
Linus Torvalds 已提交
1352 1353 1354
	return retval;
}

1355 1356 1357
static int *task_stopped_code(struct task_struct *p, bool ptrace)
{
	if (ptrace) {
1358 1359
		if (task_is_stopped_or_traced(p) &&
		    !(p->jobctl & JOBCTL_LISTENING))
1360 1361 1362 1363 1364 1365 1366 1367
			return &p->exit_code;
	} else {
		if (p->signal->flags & SIGNAL_STOP_STOPPED)
			return &p->signal->group_exit_code;
	}
	return NULL;
}

1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384
/**
 * wait_task_stopped - Wait for %TASK_STOPPED or %TASK_TRACED
 * @wo: wait options
 * @ptrace: is the wait for ptrace
 * @p: task to wait for
 *
 * Handle sys_wait4() work for %p in state %TASK_STOPPED or %TASK_TRACED.
 *
 * CONTEXT:
 * read_lock(&tasklist_lock), which is released if return value is
 * non-zero.  Also, grabs and releases @p->sighand->siglock.
 *
 * RETURNS:
 * 0 if wait condition didn't exist and search for other wait conditions
 * should continue.  Non-zero return, -errno on failure and @p's pid on
 * success, implies that tasklist_lock is released and wait condition
 * search should terminate.
L
Linus Torvalds 已提交
1385
 */
1386 1387
static int wait_task_stopped(struct wait_opts *wo,
				int ptrace, struct task_struct *p)
L
Linus Torvalds 已提交
1388
{
1389
	struct siginfo __user *infop;
1390
	int retval, exit_code, *p_code, why;
1391
	uid_t uid = 0; /* unneeded, required by compiler */
1392
	pid_t pid;
L
Linus Torvalds 已提交
1393

1394 1395 1396
	/*
	 * Traditionally we see ptrace'd stopped tasks regardless of options.
	 */
1397
	if (!ptrace && !(wo->wo_flags & WUNTRACED))
R
Roland McGrath 已提交
1398 1399
		return 0;

1400 1401 1402
	if (!task_stopped_code(p, ptrace))
		return 0;

1403 1404 1405
	exit_code = 0;
	spin_lock_irq(&p->sighand->siglock);

1406 1407
	p_code = task_stopped_code(p, ptrace);
	if (unlikely(!p_code))
1408 1409
		goto unlock_sig;

1410
	exit_code = *p_code;
1411 1412 1413
	if (!exit_code)
		goto unlock_sig;

1414
	if (!unlikely(wo->wo_flags & WNOWAIT))
1415
		*p_code = 0;
1416

1417
	uid = task_uid(p);
1418 1419 1420
unlock_sig:
	spin_unlock_irq(&p->sighand->siglock);
	if (!exit_code)
L
Linus Torvalds 已提交
1421 1422 1423 1424 1425 1426 1427 1428 1429 1430
		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);
1431
	pid = task_pid_vnr(p);
R
Roland McGrath 已提交
1432
	why = ptrace ? CLD_TRAPPED : CLD_STOPPED;
L
Linus Torvalds 已提交
1433 1434
	read_unlock(&tasklist_lock);

1435 1436 1437 1438 1439 1440 1441
	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
1442

1443
	infop = wo->wo_info;
L
Linus Torvalds 已提交
1444 1445 1446 1447 1448
	if (!retval && infop)
		retval = put_user(SIGCHLD, &infop->si_signo);
	if (!retval && infop)
		retval = put_user(0, &infop->si_errno);
	if (!retval && infop)
1449
		retval = put_user((short)why, &infop->si_code);
L
Linus Torvalds 已提交
1450 1451 1452
	if (!retval && infop)
		retval = put_user(exit_code, &infop->si_status);
	if (!retval && infop)
1453
		retval = put_user(pid, &infop->si_pid);
L
Linus Torvalds 已提交
1454
	if (!retval && infop)
1455
		retval = put_user(uid, &infop->si_uid);
L
Linus Torvalds 已提交
1456
	if (!retval)
1457
		retval = pid;
L
Linus Torvalds 已提交
1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469
	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.
 */
1470
static int wait_task_continued(struct wait_opts *wo, struct task_struct *p)
L
Linus Torvalds 已提交
1471 1472 1473 1474 1475
{
	int retval;
	pid_t pid;
	uid_t uid;

1476
	if (!unlikely(wo->wo_flags & WCONTINUED))
R
Roland McGrath 已提交
1477 1478
		return 0;

L
Linus Torvalds 已提交
1479 1480 1481 1482 1483 1484 1485 1486 1487
	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;
	}
1488
	if (!unlikely(wo->wo_flags & WNOWAIT))
L
Linus Torvalds 已提交
1489
		p->signal->flags &= ~SIGNAL_STOP_CONTINUED;
1490
	uid = task_uid(p);
L
Linus Torvalds 已提交
1491 1492
	spin_unlock_irq(&p->sighand->siglock);

1493
	pid = task_pid_vnr(p);
L
Linus Torvalds 已提交
1494 1495 1496
	get_task_struct(p);
	read_unlock(&tasklist_lock);

1497 1498 1499
	if (!wo->wo_info) {
		retval = wo->wo_rusage
			? getrusage(p, RUSAGE_BOTH, wo->wo_rusage) : 0;
L
Linus Torvalds 已提交
1500
		put_task_struct(p);
1501 1502
		if (!retval && wo->wo_stat)
			retval = put_user(0xffff, wo->wo_stat);
L
Linus Torvalds 已提交
1503
		if (!retval)
1504
			retval = pid;
L
Linus Torvalds 已提交
1505
	} else {
1506 1507
		retval = wait_noreap_copyout(wo, p, pid, uid,
					     CLD_CONTINUED, SIGCONT);
L
Linus Torvalds 已提交
1508 1509 1510 1511 1512 1513
		BUG_ON(retval == 0);
	}

	return retval;
}

R
Roland McGrath 已提交
1514 1515 1516
/*
 * Consider @p for a wait by @parent.
 *
1517
 * -ECHILD should be in ->notask_error before the first call.
R
Roland McGrath 已提交
1518 1519
 * Returns nonzero for a final return, when we have unlocked tasklist_lock.
 * Returns zero if the search for a child should continue;
1520
 * then ->notask_error is 0 if @p is an eligible child,
1521
 * or another error from security_task_wait(), or still -ECHILD.
R
Roland McGrath 已提交
1522
 */
1523 1524
static int wait_consider_task(struct wait_opts *wo, int ptrace,
				struct task_struct *p)
R
Roland McGrath 已提交
1525
{
1526
	int ret = eligible_child(wo, p);
1527
	if (!ret)
R
Roland McGrath 已提交
1528 1529
		return ret;

1530
	ret = security_task_wait(p);
1531 1532 1533 1534 1535 1536 1537 1538
	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.
		 */
1539 1540
		if (wo->notask_error)
			wo->notask_error = ret;
1541
		return 0;
1542 1543
	}

1544
	/* dead body doesn't have much to contribute */
1545 1546 1547 1548 1549 1550 1551
	if (unlikely(p->exit_state == EXIT_DEAD)) {
		/*
		 * But do not ignore this task until the tracer does
		 * wait_task_zombie()->do_notify_parent().
		 */
		if (likely(!ptrace) && unlikely(ptrace_reparented(p)))
			wo->notask_error = 0;
1552
		return 0;
1553
	}
1554

1555 1556
	/* slay zombie? */
	if (p->exit_state == EXIT_ZOMBIE) {
R
Roland McGrath 已提交
1557
		/*
1558 1559 1560
		 * A zombie ptracee is only visible to its ptracer.
		 * Notification and reaping will be cascaded to the real
		 * parent when the ptracer detaches.
R
Roland McGrath 已提交
1561
		 */
T
Tejun Heo 已提交
1562
		if (likely(!ptrace) && unlikely(p->ptrace)) {
1563 1564 1565 1566
			/* it will become visible, clear notask_error */
			wo->notask_error = 0;
			return 0;
		}
R
Roland McGrath 已提交
1567

1568 1569 1570
		/* we don't reap group leaders with subthreads */
		if (!delay_group_leader(p))
			return wait_task_zombie(wo, p);
R
Roland McGrath 已提交
1571

R
Roland McGrath 已提交
1572
		/*
1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594
		 * Allow access to stopped/continued state via zombie by
		 * falling through.  Clearing of notask_error is complex.
		 *
		 * When !@ptrace:
		 *
		 * If WEXITED is set, notask_error should naturally be
		 * cleared.  If not, subset of WSTOPPED|WCONTINUED is set,
		 * so, if there are live subthreads, there are events to
		 * wait for.  If all subthreads are dead, it's still safe
		 * to clear - this function will be called again in finite
		 * amount time once all the subthreads are released and
		 * will then return without clearing.
		 *
		 * When @ptrace:
		 *
		 * Stopped state is per-task and thus can't change once the
		 * target task dies.  Only continued and exited can happen.
		 * Clear notask_error if WCONTINUED | WEXITED.
		 */
		if (likely(!ptrace) || (wo->wo_flags & (WCONTINUED | WEXITED)))
			wo->notask_error = 0;
	} else {
1595 1596 1597 1598 1599 1600 1601 1602 1603 1604
		/*
		 * If @p is ptraced by a task in its real parent's group,
		 * hide group stop/continued state when looking at @p as
		 * the real parent; otherwise, a single stop can be
		 * reported twice as group and ptrace stops.
		 *
		 * If a ptracer wants to distinguish the two events for its
		 * own children, it should create a separate process which
		 * takes the role of real parent.
		 */
1605
		if (likely(!ptrace) && p->ptrace && !ptrace_reparented(p))
1606 1607
			return 0;

1608 1609 1610
		/*
		 * @p is alive and it's gonna stop, continue or exit, so
		 * there always is something to wait for.
R
Roland McGrath 已提交
1611
		 */
1612
		wo->notask_error = 0;
R
Roland McGrath 已提交
1613 1614
	}

R
Roland McGrath 已提交
1615
	/*
1616 1617
	 * Wait for stopped.  Depending on @ptrace, different stopped state
	 * is used and the two don't interact with each other.
R
Roland McGrath 已提交
1618
	 */
1619 1620 1621
	ret = wait_task_stopped(wo, ptrace, p);
	if (ret)
		return ret;
R
Roland McGrath 已提交
1622 1623

	/*
1624 1625 1626
	 * Wait for continued.  There's only one continued state and the
	 * ptracer can consume it which can confuse the real parent.  Don't
	 * use WCONTINUED from ptracer.  You don't need or want it.
R
Roland McGrath 已提交
1627
	 */
1628
	return wait_task_continued(wo, p);
R
Roland McGrath 已提交
1629 1630 1631 1632 1633
}

/*
 * Do the work of do_wait() for one thread in the group, @tsk.
 *
1634
 * -ECHILD should be in ->notask_error before the first call.
R
Roland McGrath 已提交
1635 1636
 * Returns nonzero for a final return, when we have unlocked tasklist_lock.
 * Returns zero if the search for a child should continue; then
1637
 * ->notask_error is 0 if there were any eligible children,
1638
 * or another error from security_task_wait(), or still -ECHILD.
R
Roland McGrath 已提交
1639
 */
1640
static int do_wait_thread(struct wait_opts *wo, struct task_struct *tsk)
R
Roland McGrath 已提交
1641 1642 1643 1644
{
	struct task_struct *p;

	list_for_each_entry(p, &tsk->children, sibling) {
1645 1646 1647
		int ret = wait_consider_task(wo, 0, p);
		if (ret)
			return ret;
R
Roland McGrath 已提交
1648 1649 1650 1651 1652
	}

	return 0;
}

1653
static int ptrace_do_wait(struct wait_opts *wo, struct task_struct *tsk)
R
Roland McGrath 已提交
1654 1655 1656
{
	struct task_struct *p;

R
Roland McGrath 已提交
1657
	list_for_each_entry(p, &tsk->ptraced, ptrace_entry) {
1658
		int ret = wait_consider_task(wo, 1, p);
R
Roland McGrath 已提交
1659
		if (ret)
R
Roland McGrath 已提交
1660 1661 1662 1663 1664 1665
			return ret;
	}

	return 0;
}

1666 1667 1668 1669 1670 1671 1672
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;

1673
	if (!eligible_pid(wo, p))
1674 1675
		return 0;

1676 1677 1678
	if ((wo->wo_flags & __WNOTHREAD) && wait->private != p->parent)
		return 0;

1679 1680 1681
	return default_wake_function(wait, mode, sync, key);
}

1682 1683
void __wake_up_parent(struct task_struct *p, struct task_struct *parent)
{
1684 1685
	__wake_up_sync_key(&parent->signal->wait_chldexit,
				TASK_INTERRUPTIBLE, 1, p);
1686 1687
}

1688
static long do_wait(struct wait_opts *wo)
L
Linus Torvalds 已提交
1689 1690
{
	struct task_struct *tsk;
R
Roland McGrath 已提交
1691
	int retval;
L
Linus Torvalds 已提交
1692

1693
	trace_sched_process_wait(wo->wo_pid);
1694

1695 1696 1697
	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
1698
repeat:
R
Roland McGrath 已提交
1699 1700
	/*
	 * If there is nothing that can match our critiera just get out.
1701 1702 1703
	 * 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
1704
	 */
1705
	wo->notask_error = -ECHILD;
1706 1707
	if ((wo->wo_type < PIDTYPE_MAX) &&
	   (!wo->wo_pid || hlist_empty(&wo->wo_pid->tasks[wo->wo_type])))
1708
		goto notask;
1709

1710
	set_current_state(TASK_INTERRUPTIBLE);
L
Linus Torvalds 已提交
1711 1712 1713
	read_lock(&tasklist_lock);
	tsk = current;
	do {
1714 1715 1716
		retval = do_wait_thread(wo, tsk);
		if (retval)
			goto end;
1717

1718 1719
		retval = ptrace_do_wait(wo, tsk);
		if (retval)
R
Roland McGrath 已提交
1720 1721
			goto end;

1722
		if (wo->wo_flags & __WNOTHREAD)
L
Linus Torvalds 已提交
1723
			break;
1724
	} while_each_thread(current, tsk);
L
Linus Torvalds 已提交
1725
	read_unlock(&tasklist_lock);
1726

1727
notask:
1728 1729
	retval = wo->notask_error;
	if (!retval && !(wo->wo_flags & WNOHANG)) {
L
Linus Torvalds 已提交
1730
		retval = -ERESTARTSYS;
R
Roland McGrath 已提交
1731 1732 1733 1734
		if (!signal_pending(current)) {
			schedule();
			goto repeat;
		}
L
Linus Torvalds 已提交
1735 1736
	}
end:
1737
	__set_current_state(TASK_RUNNING);
1738
	remove_wait_queue(&current->signal->wait_chldexit, &wo->child_wait);
L
Linus Torvalds 已提交
1739 1740 1741
	return retval;
}

1742 1743
SYSCALL_DEFINE5(waitid, int, which, pid_t, upid, struct siginfo __user *,
		infop, int, options, struct rusage __user *, ru)
L
Linus Torvalds 已提交
1744
{
1745
	struct wait_opts wo;
1746 1747
	struct pid *pid = NULL;
	enum pid_type type;
L
Linus Torvalds 已提交
1748 1749 1750 1751 1752 1753 1754 1755 1756
	long ret;

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

	switch (which) {
	case P_ALL:
1757
		type = PIDTYPE_MAX;
L
Linus Torvalds 已提交
1758 1759
		break;
	case P_PID:
1760 1761
		type = PIDTYPE_PID;
		if (upid <= 0)
L
Linus Torvalds 已提交
1762 1763 1764
			return -EINVAL;
		break;
	case P_PGID:
1765 1766
		type = PIDTYPE_PGID;
		if (upid <= 0)
L
Linus Torvalds 已提交
1767 1768 1769 1770 1771 1772
			return -EINVAL;
		break;
	default:
		return -EINVAL;
	}

1773 1774
	if (type < PIDTYPE_MAX)
		pid = find_get_pid(upid);
1775 1776 1777 1778 1779 1780 1781 1782

	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);
1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805

	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);
	}

1806
	put_pid(pid);
L
Linus Torvalds 已提交
1807 1808

	/* avoid REGPARM breakage on x86: */
1809
	asmlinkage_protect(5, ret, which, upid, infop, options, ru);
L
Linus Torvalds 已提交
1810 1811 1812
	return ret;
}

1813 1814
SYSCALL_DEFINE4(wait4, pid_t, upid, int __user *, stat_addr,
		int, options, struct rusage __user *, ru)
L
Linus Torvalds 已提交
1815
{
1816
	struct wait_opts wo;
1817 1818
	struct pid *pid = NULL;
	enum pid_type type;
L
Linus Torvalds 已提交
1819 1820 1821 1822 1823
	long ret;

	if (options & ~(WNOHANG|WUNTRACED|WCONTINUED|
			__WNOTHREAD|__WCLONE|__WALL))
		return -EINVAL;
1824 1825 1826 1827 1828 1829 1830 1831

	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;
1832
		pid = get_task_pid(current, PIDTYPE_PGID);
1833 1834 1835 1836 1837
	} else /* upid > 0 */ {
		type = PIDTYPE_PID;
		pid = find_get_pid(upid);
	}

1838 1839 1840 1841 1842 1843 1844
	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);
1845
	put_pid(pid);
L
Linus Torvalds 已提交
1846 1847

	/* avoid REGPARM breakage on x86: */
1848
	asmlinkage_protect(4, ret, upid, stat_addr, options, ru);
L
Linus Torvalds 已提交
1849 1850 1851 1852 1853 1854 1855 1856 1857
	return ret;
}

#ifdef __ARCH_WANT_SYS_WAITPID

/*
 * sys_waitpid() remains for compatibility. waitpid() should be
 * implemented by calling sys_wait4() from libc.a.
 */
1858
SYSCALL_DEFINE3(waitpid, pid_t, pid, int __user *, stat_addr, int, options)
L
Linus Torvalds 已提交
1859 1860 1861 1862 1863
{
	return sys_wait4(pid, stat_addr, options, NULL);
}

#endif