fcntl.c 16.6 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1 2 3 4 5 6 7 8 9 10 11
/*
 *  linux/fs/fcntl.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 */

#include <linux/syscalls.h>
#include <linux/init.h>
#include <linux/mm.h>
#include <linux/fs.h>
#include <linux/file.h>
Al Viro's avatar
Al Viro committed
12
#include <linux/fdtable.h>
13
#include <linux/capability.h>
Linus Torvalds's avatar
Linus Torvalds committed
14 15 16
#include <linux/dnotify.h>
#include <linux/slab.h>
#include <linux/module.h>
17
#include <linux/pipe_fs_i.h>
Linus Torvalds's avatar
Linus Torvalds committed
18 19
#include <linux/security.h>
#include <linux/ptrace.h>
20
#include <linux/signal.h>
21
#include <linux/rcupdate.h>
22
#include <linux/pid_namespace.h>
23
#include <linux/user_namespace.h>
Linus Torvalds's avatar
Linus Torvalds committed
24 25 26 27 28

#include <asm/poll.h>
#include <asm/siginfo.h>
#include <asm/uaccess.h>

29
#define SETFL_MASK (O_APPEND | O_NONBLOCK | O_NDELAY | O_DIRECT | O_NOATIME)
Linus Torvalds's avatar
Linus Torvalds committed
30 31 32

static int setfl(int fd, struct file * filp, unsigned long arg)
{
Al Viro's avatar
Al Viro committed
33
	struct inode * inode = file_inode(filp);
Linus Torvalds's avatar
Linus Torvalds committed
34 35
	int error = 0;

36 37 38 39 40
	/*
	 * O_APPEND cannot be cleared if the file is marked as append-only
	 * and the file is open for write.
	 */
	if (((arg ^ filp->f_flags) & O_APPEND) && IS_APPEND(inode))
Linus Torvalds's avatar
Linus Torvalds committed
41 42 43 44
		return -EPERM;

	/* O_NOATIME can only be set by the owner or superuser */
	if ((arg & O_NOATIME) && !(filp->f_flags & O_NOATIME))
45
		if (!inode_owner_or_capable(inode))
Linus Torvalds's avatar
Linus Torvalds committed
46 47 48 49 50 51 52 53 54 55 56 57 58
			return -EPERM;

	/* required for strict SunOS emulation */
	if (O_NONBLOCK != O_NDELAY)
	       if (arg & O_NDELAY)
		   arg |= O_NONBLOCK;

	if (arg & O_DIRECT) {
		if (!filp->f_mapping || !filp->f_mapping->a_ops ||
			!filp->f_mapping->a_ops->direct_IO)
				return -EINVAL;
	}

Al Viro's avatar
Al Viro committed
59
	if (filp->f_op->check_flags)
Linus Torvalds's avatar
Linus Torvalds committed
60 61 62 63
		error = filp->f_op->check_flags(arg);
	if (error)
		return error;

64
	/*
65
	 * ->fasync() is responsible for setting the FASYNC bit.
66
	 */
Al Viro's avatar
Al Viro committed
67
	if (((arg ^ filp->f_flags) & FASYNC) && filp->f_op->fasync) {
68 69 70
		error = filp->f_op->fasync(fd, filp, (arg & FASYNC) != 0);
		if (error < 0)
			goto out;
71 72
		if (error > 0)
			error = 0;
Linus Torvalds's avatar
Linus Torvalds committed
73
	}
74
	spin_lock(&filp->f_lock);
Linus Torvalds's avatar
Linus Torvalds committed
75
	filp->f_flags = (arg & SETFL_MASK) | (filp->f_flags & ~SETFL_MASK);
76
	spin_unlock(&filp->f_lock);
77

Linus Torvalds's avatar
Linus Torvalds committed
78 79 80 81
 out:
	return error;
}

82
static void f_modown(struct file *filp, struct pid *pid, enum pid_type type,
83
                     int force)
Linus Torvalds's avatar
Linus Torvalds committed
84
{
85
	write_lock_irq(&filp->f_owner.lock);
Linus Torvalds's avatar
Linus Torvalds committed
86
	if (force || !filp->f_owner.pid) {
87 88 89
		put_pid(filp->f_owner.pid);
		filp->f_owner.pid = get_pid(pid);
		filp->f_owner.pid_type = type;
90 91 92 93 94 95

		if (pid) {
			const struct cred *cred = current_cred();
			filp->f_owner.uid = cred->uid;
			filp->f_owner.euid = cred->euid;
		}
Linus Torvalds's avatar
Linus Torvalds committed
96
	}
97
	write_unlock_irq(&filp->f_owner.lock);
Linus Torvalds's avatar
Linus Torvalds committed
98 99
}

100 101
int __f_setown(struct file *filp, struct pid *pid, enum pid_type type,
		int force)
Linus Torvalds's avatar
Linus Torvalds committed
102 103
{
	int err;
104

Linus Torvalds's avatar
Linus Torvalds committed
105 106 107 108
	err = security_file_set_fowner(filp);
	if (err)
		return err;

109
	f_modown(filp, pid, type, force);
Linus Torvalds's avatar
Linus Torvalds committed
110 111
	return 0;
}
112
EXPORT_SYMBOL(__f_setown);
Linus Torvalds's avatar
Linus Torvalds committed
113

114 115 116 117 118 119 120 121 122 123 124 125
int f_setown(struct file *filp, unsigned long arg, int force)
{
	enum pid_type type;
	struct pid *pid;
	int who = arg;
	int result;
	type = PIDTYPE_PID;
	if (who < 0) {
		type = PIDTYPE_PGID;
		who = -who;
	}
	rcu_read_lock();
126
	pid = find_vpid(who);
127 128 129 130
	result = __f_setown(filp, pid, type, force);
	rcu_read_unlock();
	return result;
}
Linus Torvalds's avatar
Linus Torvalds committed
131 132 133 134
EXPORT_SYMBOL(f_setown);

void f_delown(struct file *filp)
{
135
	f_modown(filp, NULL, PIDTYPE_PID, 1);
136 137 138 139 140
}

pid_t f_getown(struct file *filp)
{
	pid_t pid;
141
	read_lock(&filp->f_owner.lock);
142
	pid = pid_vnr(filp->f_owner.pid);
143 144
	if (filp->f_owner.pid_type == PIDTYPE_PGID)
		pid = -pid;
145
	read_unlock(&filp->f_owner.lock);
146
	return pid;
Linus Torvalds's avatar
Linus Torvalds committed
147 148
}

Peter Zijlstra's avatar
Peter Zijlstra committed
149 150
static int f_setown_ex(struct file *filp, unsigned long arg)
{
Al Viro's avatar
Al Viro committed
151
	struct f_owner_ex __user *owner_p = (void __user *)arg;
Peter Zijlstra's avatar
Peter Zijlstra committed
152 153 154 155 156 157 158
	struct f_owner_ex owner;
	struct pid *pid;
	int type;
	int ret;

	ret = copy_from_user(&owner, owner_p, sizeof(owner));
	if (ret)
159
		return -EFAULT;
Peter Zijlstra's avatar
Peter Zijlstra committed
160 161 162 163 164 165 166 167 168 169

	switch (owner.type) {
	case F_OWNER_TID:
		type = PIDTYPE_MAX;
		break;

	case F_OWNER_PID:
		type = PIDTYPE_PID;
		break;

170
	case F_OWNER_PGRP:
Peter Zijlstra's avatar
Peter Zijlstra committed
171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190
		type = PIDTYPE_PGID;
		break;

	default:
		return -EINVAL;
	}

	rcu_read_lock();
	pid = find_vpid(owner.pid);
	if (owner.pid && !pid)
		ret = -ESRCH;
	else
		ret = __f_setown(filp, pid, type, 1);
	rcu_read_unlock();

	return ret;
}

static int f_getown_ex(struct file *filp, unsigned long arg)
{
Al Viro's avatar
Al Viro committed
191
	struct f_owner_ex __user *owner_p = (void __user *)arg;
Peter Zijlstra's avatar
Peter Zijlstra committed
192 193 194 195 196 197 198 199 200 201 202 203 204 205 206
	struct f_owner_ex owner;
	int ret = 0;

	read_lock(&filp->f_owner.lock);
	owner.pid = pid_vnr(filp->f_owner.pid);
	switch (filp->f_owner.pid_type) {
	case PIDTYPE_MAX:
		owner.type = F_OWNER_TID;
		break;

	case PIDTYPE_PID:
		owner.type = F_OWNER_PID;
		break;

	case PIDTYPE_PGID:
207
		owner.type = F_OWNER_PGRP;
Peter Zijlstra's avatar
Peter Zijlstra committed
208 209 210 211 212 213 214 215 216
		break;

	default:
		WARN_ON(1);
		ret = -EINVAL;
		break;
	}
	read_unlock(&filp->f_owner.lock);

217
	if (!ret) {
Peter Zijlstra's avatar
Peter Zijlstra committed
218
		ret = copy_to_user(owner_p, &owner, sizeof(owner));
219 220 221
		if (ret)
			ret = -EFAULT;
	}
Peter Zijlstra's avatar
Peter Zijlstra committed
222 223 224
	return ret;
}

225 226 227 228
#ifdef CONFIG_CHECKPOINT_RESTORE
static int f_getowner_uids(struct file *filp, unsigned long arg)
{
	struct user_namespace *user_ns = current_user_ns();
Al Viro's avatar
Al Viro committed
229
	uid_t __user *dst = (void __user *)arg;
230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249
	uid_t src[2];
	int err;

	read_lock(&filp->f_owner.lock);
	src[0] = from_kuid(user_ns, filp->f_owner.uid);
	src[1] = from_kuid(user_ns, filp->f_owner.euid);
	read_unlock(&filp->f_owner.lock);

	err  = put_user(src[0], &dst[0]);
	err |= put_user(src[1], &dst[1]);

	return err;
}
#else
static int f_getowner_uids(struct file *filp, unsigned long arg)
{
	return -EINVAL;
}
#endif

Linus Torvalds's avatar
Linus Torvalds committed
250 251 252 253 254 255 256
static long do_fcntl(int fd, unsigned int cmd, unsigned long arg,
		struct file *filp)
{
	long err = -EINVAL;

	switch (cmd) {
	case F_DUPFD:
257 258
		err = f_dupfd(arg, filp, 0);
		break;
259
	case F_DUPFD_CLOEXEC:
Al Viro's avatar
Al Viro committed
260
		err = f_dupfd(arg, filp, O_CLOEXEC);
Linus Torvalds's avatar
Linus Torvalds committed
261 262 263 264 265 266 267 268 269 270 271 272 273 274
		break;
	case F_GETFD:
		err = get_close_on_exec(fd) ? FD_CLOEXEC : 0;
		break;
	case F_SETFD:
		err = 0;
		set_close_on_exec(fd, arg & FD_CLOEXEC);
		break;
	case F_GETFL:
		err = filp->f_flags;
		break;
	case F_SETFL:
		err = setfl(fd, filp, arg);
		break;
275 276 277 278
#if BITS_PER_LONG != 32
	/* 32-bit arches must use fcntl64() */
	case F_GETLKP:
#endif
Linus Torvalds's avatar
Linus Torvalds committed
279
	case F_GETLK:
280
		err = fcntl_getlk(filp, cmd, (struct flock __user *) arg);
Linus Torvalds's avatar
Linus Torvalds committed
281
		break;
282 283 284 285 286 287
#if BITS_PER_LONG != 32
	/* 32-bit arches must use fcntl64() */
	case F_SETLKP:
	case F_SETLKPW:
#endif
		/* Fallthrough */
Linus Torvalds's avatar
Linus Torvalds committed
288 289
	case F_SETLK:
	case F_SETLKW:
290
		err = fcntl_setlk(fd, filp, cmd, (struct flock __user *) arg);
Linus Torvalds's avatar
Linus Torvalds committed
291 292 293 294 295 296 297 298 299
		break;
	case F_GETOWN:
		/*
		 * XXX If f_owner is a process group, the
		 * negative return value will get converted
		 * into an error.  Oops.  If we keep the
		 * current syscall conventions, the only way
		 * to fix this will be in libc.
		 */
300
		err = f_getown(filp);
Linus Torvalds's avatar
Linus Torvalds committed
301 302 303 304 305
		force_successful_syscall_return();
		break;
	case F_SETOWN:
		err = f_setown(filp, arg, 1);
		break;
Peter Zijlstra's avatar
Peter Zijlstra committed
306 307 308 309 310 311
	case F_GETOWN_EX:
		err = f_getown_ex(filp, arg);
		break;
	case F_SETOWN_EX:
		err = f_setown_ex(filp, arg);
		break;
312 313 314
	case F_GETOWNER_UIDS:
		err = f_getowner_uids(filp, arg);
		break;
Linus Torvalds's avatar
Linus Torvalds committed
315 316 317 318 319
	case F_GETSIG:
		err = filp->f_owner.signum;
		break;
	case F_SETSIG:
		/* arg == 0 restores default behaviour. */
320
		if (!valid_signal(arg)) {
Linus Torvalds's avatar
Linus Torvalds committed
321 322 323 324 325 326 327 328 329 330 331 332 333 334
			break;
		}
		err = 0;
		filp->f_owner.signum = arg;
		break;
	case F_GETLEASE:
		err = fcntl_getlease(filp);
		break;
	case F_SETLEASE:
		err = fcntl_setlease(fd, filp, arg);
		break;
	case F_NOTIFY:
		err = fcntl_dirnotify(fd, filp, arg);
		break;
335 336 337 338
	case F_SETPIPE_SZ:
	case F_GETPIPE_SZ:
		err = pipe_fcntl(filp, cmd, arg);
		break;
Linus Torvalds's avatar
Linus Torvalds committed
339 340 341 342 343 344
	default:
		break;
	}
	return err;
}

345 346 347 348 349 350 351 352 353 354 355 356 357
static int check_fcntl_cmd(unsigned cmd)
{
	switch (cmd) {
	case F_DUPFD:
	case F_DUPFD_CLOEXEC:
	case F_GETFD:
	case F_SETFD:
	case F_GETFL:
		return 1;
	}
	return 0;
}

358
SYSCALL_DEFINE3(fcntl, unsigned int, fd, unsigned int, cmd, unsigned long, arg)
Linus Torvalds's avatar
Linus Torvalds committed
359
{	
360
	struct fd f = fdget_raw(fd);
Linus Torvalds's avatar
Linus Torvalds committed
361 362
	long err = -EBADF;

363
	if (!f.file)
Linus Torvalds's avatar
Linus Torvalds committed
364 365
		goto out;

366
	if (unlikely(f.file->f_mode & FMODE_PATH)) {
367 368
		if (!check_fcntl_cmd(cmd))
			goto out1;
369 370
	}

371
	err = security_file_fcntl(f.file, cmd, arg);
372
	if (!err)
373
		err = do_fcntl(fd, cmd, arg, f.file);
Linus Torvalds's avatar
Linus Torvalds committed
374

375
out1:
376
 	fdput(f);
Linus Torvalds's avatar
Linus Torvalds committed
377 378 379 380 381
out:
	return err;
}

#if BITS_PER_LONG == 32
382 383
SYSCALL_DEFINE3(fcntl64, unsigned int, fd, unsigned int, cmd,
		unsigned long, arg)
Linus Torvalds's avatar
Linus Torvalds committed
384
{	
385
	struct fd f = fdget_raw(fd);
386
	long err = -EBADF;
Linus Torvalds's avatar
Linus Torvalds committed
387

388
	if (!f.file)
Linus Torvalds's avatar
Linus Torvalds committed
389 390
		goto out;

391
	if (unlikely(f.file->f_mode & FMODE_PATH)) {
392 393
		if (!check_fcntl_cmd(cmd))
			goto out1;
394 395
	}

396
	err = security_file_fcntl(f.file, cmd, arg);
397 398
	if (err)
		goto out1;
Linus Torvalds's avatar
Linus Torvalds committed
399 400
	
	switch (cmd) {
401 402 403 404 405 406 407 408 409 410 411 412 413 414
	case F_GETLK64:
	case F_GETLKP:
		err = fcntl_getlk64(f.file, cmd, (struct flock64 __user *) arg);
		break;
	case F_SETLK64:
	case F_SETLKW64:
	case F_SETLKP:
	case F_SETLKPW:
		err = fcntl_setlk64(fd, f.file, cmd,
				(struct flock64 __user *) arg);
		break;
	default:
		err = do_fcntl(fd, cmd, arg, f.file);
		break;
Linus Torvalds's avatar
Linus Torvalds committed
415
	}
416
out1:
417
	fdput(f);
Linus Torvalds's avatar
Linus Torvalds committed
418 419 420 421 422 423 424
out:
	return err;
}
#endif

/* Table to convert sigio signal codes into poll band bitmaps */

425
static const long band_table[NSIGPOLL] = {
Linus Torvalds's avatar
Linus Torvalds committed
426 427 428 429 430 431 432 433 434 435 436
	POLLIN | POLLRDNORM,			/* POLL_IN */
	POLLOUT | POLLWRNORM | POLLWRBAND,	/* POLL_OUT */
	POLLIN | POLLRDNORM | POLLMSG,		/* POLL_MSG */
	POLLERR,				/* POLL_ERR */
	POLLPRI | POLLRDBAND,			/* POLL_PRI */
	POLLHUP | POLLERR			/* POLL_HUP */
};

static inline int sigio_perm(struct task_struct *p,
                             struct fown_struct *fown, int sig)
{
437 438 439 440 441
	const struct cred *cred;
	int ret;

	rcu_read_lock();
	cred = __task_cred(p);
442 443 444
	ret = ((uid_eq(fown->euid, GLOBAL_ROOT_UID) ||
		uid_eq(fown->euid, cred->suid) || uid_eq(fown->euid, cred->uid) ||
		uid_eq(fown->uid,  cred->suid) || uid_eq(fown->uid,  cred->uid)) &&
445 446 447
	       !security_file_send_sigiotask(p, fown, sig));
	rcu_read_unlock();
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
448 449 450
}

static void send_sigio_to_task(struct task_struct *p,
451
			       struct fown_struct *fown,
Peter Zijlstra's avatar
Peter Zijlstra committed
452
			       int fd, int reason, int group)
Linus Torvalds's avatar
Linus Torvalds committed
453
{
454 455 456 457 458 459 460
	/*
	 * F_SETSIG can change ->signum lockless in parallel, make
	 * sure we read it once and use the same value throughout.
	 */
	int signum = ACCESS_ONCE(fown->signum);

	if (!sigio_perm(p, fown, signum))
Linus Torvalds's avatar
Linus Torvalds committed
461 462
		return;

463
	switch (signum) {
Linus Torvalds's avatar
Linus Torvalds committed
464 465 466 467 468 469 470 471
		siginfo_t si;
		default:
			/* Queue a rt signal with the appropriate fd as its
			   value.  We use SI_SIGIO as the source, not 
			   SI_KERNEL, since kernel signals always get 
			   delivered even if we can't queue.  Failure to
			   queue in this case _should_ be reported; we fall
			   back to SIGIO in that case. --sct */
472
			si.si_signo = signum;
Linus Torvalds's avatar
Linus Torvalds committed
473 474 475 476 477
			si.si_errno = 0;
		        si.si_code  = reason;
			/* Make sure we are called with one of the POLL_*
			   reasons, otherwise we could leak kernel stack into
			   userspace.  */
478
			BUG_ON((reason & __SI_MASK) != __SI_POLL);
Linus Torvalds's avatar
Linus Torvalds committed
479 480 481 482 483
			if (reason - POLL_IN >= NSIGPOLL)
				si.si_band  = ~0L;
			else
				si.si_band = band_table[reason - POLL_IN];
			si.si_fd    = fd;
Peter Zijlstra's avatar
Peter Zijlstra committed
484
			if (!do_send_sig_info(signum, &si, p, group))
Linus Torvalds's avatar
Linus Torvalds committed
485 486 487
				break;
		/* fall-through: fall back on the old plain SIGIO signal */
		case 0:
Peter Zijlstra's avatar
Peter Zijlstra committed
488
			do_send_sig_info(SIGIO, SEND_SIG_PRIV, p, group);
Linus Torvalds's avatar
Linus Torvalds committed
489 490 491 492 493 494
	}
}

void send_sigio(struct fown_struct *fown, int fd, int band)
{
	struct task_struct *p;
495 496
	enum pid_type type;
	struct pid *pid;
Peter Zijlstra's avatar
Peter Zijlstra committed
497
	int group = 1;
Linus Torvalds's avatar
Linus Torvalds committed
498 499
	
	read_lock(&fown->lock);
Peter Zijlstra's avatar
Peter Zijlstra committed
500

501
	type = fown->pid_type;
Peter Zijlstra's avatar
Peter Zijlstra committed
502 503 504 505 506
	if (type == PIDTYPE_MAX) {
		group = 0;
		type = PIDTYPE_PID;
	}

Linus Torvalds's avatar
Linus Torvalds committed
507 508 509 510 511
	pid = fown->pid;
	if (!pid)
		goto out_unlock_fown;
	
	read_lock(&tasklist_lock);
512
	do_each_pid_task(pid, type, p) {
Peter Zijlstra's avatar
Peter Zijlstra committed
513
		send_sigio_to_task(p, fown, fd, band, group);
514
	} while_each_pid_task(pid, type, p);
Linus Torvalds's avatar
Linus Torvalds committed
515 516 517 518 519 520
	read_unlock(&tasklist_lock);
 out_unlock_fown:
	read_unlock(&fown->lock);
}

static void send_sigurg_to_task(struct task_struct *p,
Peter Zijlstra's avatar
Peter Zijlstra committed
521
				struct fown_struct *fown, int group)
Linus Torvalds's avatar
Linus Torvalds committed
522 523
{
	if (sigio_perm(p, fown, SIGURG))
Peter Zijlstra's avatar
Peter Zijlstra committed
524
		do_send_sig_info(SIGURG, SEND_SIG_PRIV, p, group);
Linus Torvalds's avatar
Linus Torvalds committed
525 526 527 528 529
}

int send_sigurg(struct fown_struct *fown)
{
	struct task_struct *p;
530 531
	enum pid_type type;
	struct pid *pid;
Peter Zijlstra's avatar
Peter Zijlstra committed
532
	int group = 1;
533
	int ret = 0;
Linus Torvalds's avatar
Linus Torvalds committed
534 535
	
	read_lock(&fown->lock);
Peter Zijlstra's avatar
Peter Zijlstra committed
536

537
	type = fown->pid_type;
Peter Zijlstra's avatar
Peter Zijlstra committed
538 539 540 541 542
	if (type == PIDTYPE_MAX) {
		group = 0;
		type = PIDTYPE_PID;
	}

Linus Torvalds's avatar
Linus Torvalds committed
543 544 545 546 547 548 549
	pid = fown->pid;
	if (!pid)
		goto out_unlock_fown;

	ret = 1;
	
	read_lock(&tasklist_lock);
550
	do_each_pid_task(pid, type, p) {
Peter Zijlstra's avatar
Peter Zijlstra committed
551
		send_sigurg_to_task(p, fown, group);
552
	} while_each_pid_task(pid, type, p);
Linus Torvalds's avatar
Linus Torvalds committed
553 554 555 556 557 558
	read_unlock(&tasklist_lock);
 out_unlock_fown:
	read_unlock(&fown->lock);
	return ret;
}

559
static DEFINE_SPINLOCK(fasync_lock);
560
static struct kmem_cache *fasync_cache __read_mostly;
Linus Torvalds's avatar
Linus Torvalds committed
561

562 563 564 565 566 567
static void fasync_free_rcu(struct rcu_head *head)
{
	kmem_cache_free(fasync_cache,
			container_of(head, struct fasync_struct, fa_rcu));
}

Linus Torvalds's avatar
Linus Torvalds committed
568
/*
569 570 571 572 573 574 575
 * Remove a fasync entry. If successfully removed, return
 * positive and clear the FASYNC flag. If no entry exists,
 * do nothing and return 0.
 *
 * NOTE! It is very important that the FASYNC flag always
 * match the state "is the filp on a fasync list".
 *
Linus Torvalds's avatar
Linus Torvalds committed
576
 */
577
int fasync_remove_entry(struct file *filp, struct fasync_struct **fapp)
Linus Torvalds's avatar
Linus Torvalds committed
578 579 580 581
{
	struct fasync_struct *fa, **fp;
	int result = 0;

582
	spin_lock(&filp->f_lock);
583
	spin_lock(&fasync_lock);
584 585 586
	for (fp = fapp; (fa = *fp) != NULL; fp = &fa->fa_next) {
		if (fa->fa_file != filp)
			continue;
587 588 589 590 591

		spin_lock_irq(&fa->fa_lock);
		fa->fa_file = NULL;
		spin_unlock_irq(&fa->fa_lock);

592
		*fp = fa->fa_next;
593
		call_rcu(&fa->fa_rcu, fasync_free_rcu);
594 595 596
		filp->f_flags &= ~FASYNC;
		result = 1;
		break;
Linus Torvalds's avatar
Linus Torvalds committed
597
	}
598
	spin_unlock(&fasync_lock);
599 600 601 602
	spin_unlock(&filp->f_lock);
	return result;
}

603 604 605 606 607
struct fasync_struct *fasync_alloc(void)
{
	return kmem_cache_alloc(fasync_cache, GFP_KERNEL);
}

608
/*
609 610 611
 * NOTE! This can be used only for unused fasync entries:
 * entries that actually got inserted on the fasync list
 * need to be released by rcu - see fasync_remove_entry.
612
 */
613
void fasync_free(struct fasync_struct *new)
614
{
615 616
	kmem_cache_free(fasync_cache, new);
}
617

618 619 620
/*
 * Insert a new entry into the fasync list.  Return the pointer to the
 * old one if we didn't use the new one.
621 622 623
 *
 * NOTE! It is very important that the FASYNC flag always
 * match the state "is the filp on a fasync list".
624 625 626 627
 */
struct fasync_struct *fasync_insert_entry(int fd, struct file *filp, struct fasync_struct **fapp, struct fasync_struct *new)
{
        struct fasync_struct *fa, **fp;
628 629

	spin_lock(&filp->f_lock);
630
	spin_lock(&fasync_lock);
Linus Torvalds's avatar
Linus Torvalds committed
631
	for (fp = fapp; (fa = *fp) != NULL; fp = &fa->fa_next) {
632 633
		if (fa->fa_file != filp)
			continue;
634 635

		spin_lock_irq(&fa->fa_lock);
636
		fa->fa_fd = fd;
637
		spin_unlock_irq(&fa->fa_lock);
638
		goto out;
Linus Torvalds's avatar
Linus Torvalds committed
639 640
	}

641
	spin_lock_init(&new->fa_lock);
642 643 644 645
	new->magic = FASYNC_MAGIC;
	new->fa_file = filp;
	new->fa_fd = fd;
	new->fa_next = *fapp;
646
	rcu_assign_pointer(*fapp, new);
647 648
	filp->f_flags |= FASYNC;

Linus Torvalds's avatar
Linus Torvalds committed
649
out:
650
	spin_unlock(&fasync_lock);
651
	spin_unlock(&filp->f_lock);
652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679
	return fa;
}

/*
 * Add a fasync entry. Return negative on error, positive if
 * added, and zero if did nothing but change an existing one.
 */
static int fasync_add_entry(int fd, struct file *filp, struct fasync_struct **fapp)
{
	struct fasync_struct *new;

	new = fasync_alloc();
	if (!new)
		return -ENOMEM;

	/*
	 * fasync_insert_entry() returns the old (update) entry if
	 * it existed.
	 *
	 * So free the (unused) new entry and return 0 to let the
	 * caller know that we didn't add any new fasync entries.
	 */
	if (fasync_insert_entry(fd, filp, fapp, new)) {
		fasync_free(new);
		return 0;
	}

	return 1;
Linus Torvalds's avatar
Linus Torvalds committed
680 681
}

682 683 684 685 686 687 688 689 690 691 692 693 694
/*
 * fasync_helper() is used by almost all character device drivers
 * to set up the fasync queue, and for regular files by the file
 * lease code. It returns negative on error, 0 if it did no changes
 * and positive if it added/deleted the entry.
 */
int fasync_helper(int fd, struct file * filp, int on, struct fasync_struct **fapp)
{
	if (!on)
		return fasync_remove_entry(filp, fapp);
	return fasync_add_entry(fd, filp, fapp);
}

Linus Torvalds's avatar
Linus Torvalds committed
695 696
EXPORT_SYMBOL(fasync_helper);

697 698 699 700
/*
 * rcu_read_lock() is held
 */
static void kill_fasync_rcu(struct fasync_struct *fa, int sig, int band)
Linus Torvalds's avatar
Linus Torvalds committed
701 702
{
	while (fa) {
703
		struct fown_struct *fown;
704 705
		unsigned long flags;

Linus Torvalds's avatar
Linus Torvalds committed
706 707 708 709 710
		if (fa->magic != FASYNC_MAGIC) {
			printk(KERN_ERR "kill_fasync: bad magic number in "
			       "fasync_struct!\n");
			return;
		}
711
		spin_lock_irqsave(&fa->fa_lock, flags);
712 713 714 715 716 717 718 719
		if (fa->fa_file) {
			fown = &fa->fa_file->f_owner;
			/* Don't send SIGURG to processes which have not set a
			   queued signum: SIGURG has its own default signalling
			   mechanism. */
			if (!(sig == SIGURG && fown->signum == 0))
				send_sigio(fown, fa->fa_fd, band);
		}
720
		spin_unlock_irqrestore(&fa->fa_lock, flags);
721
		fa = rcu_dereference(fa->fa_next);
Linus Torvalds's avatar
Linus Torvalds committed
722 723 724 725 726 727 728 729 730
	}
}

void kill_fasync(struct fasync_struct **fp, int sig, int band)
{
	/* First a quick test without locking: usually
	 * the list is empty.
	 */
	if (*fp) {
731 732 733
		rcu_read_lock();
		kill_fasync_rcu(rcu_dereference(*fp), sig, band);
		rcu_read_unlock();
Linus Torvalds's avatar
Linus Torvalds committed
734 735 736 737
	}
}
EXPORT_SYMBOL(kill_fasync);

738
static int __init fcntl_init(void)
Linus Torvalds's avatar
Linus Torvalds committed
739
{
740 741 742 743 744
	/*
	 * Please add new bits here to ensure allocation uniqueness.
	 * Exceptions: O_NONBLOCK is a two bit define on parisc; O_NDELAY
	 * is defined as O_NONBLOCK on some platforms and not on others.
	 */
745
	BUILD_BUG_ON(20 - 1 /* for O_RDONLY being 0 */ != HWEIGHT32(
746 747
		O_RDONLY	| O_WRONLY	| O_RDWR	|
		O_CREAT		| O_EXCL	| O_NOCTTY	|
748
		O_TRUNC		| O_APPEND	| /* O_NONBLOCK	| */
749 750 751
		__O_SYNC	| O_DSYNC	| FASYNC	|
		O_DIRECT	| O_LARGEFILE	| O_DIRECTORY	|
		O_NOFOLLOW	| O_NOATIME	| O_CLOEXEC	|
752
		__FMODE_EXEC	| O_PATH	| __O_TMPFILE
753 754
		));

Linus Torvalds's avatar
Linus Torvalds committed
755
	fasync_cache = kmem_cache_create("fasync_cache",
756
		sizeof(struct fasync_struct), 0, SLAB_PANIC, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
757 758 759
	return 0;
}

760
module_init(fcntl_init)