fcntl.c 18.7 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>
Linus Torvalds's avatar
Linus Torvalds committed
23 24 25 26 27

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

28
void set_close_on_exec(unsigned int fd, int flag)
Linus Torvalds's avatar
Linus Torvalds committed
29 30
{
	struct files_struct *files = current->files;
31
	struct fdtable *fdt;
Linus Torvalds's avatar
Linus Torvalds committed
32
	spin_lock(&files->file_lock);
33
	fdt = files_fdtable(files);
Linus Torvalds's avatar
Linus Torvalds committed
34
	if (flag)
35
		__set_close_on_exec(fd, fdt);
Linus Torvalds's avatar
Linus Torvalds committed
36
	else
37
		__clear_close_on_exec(fd, fdt);
Linus Torvalds's avatar
Linus Torvalds committed
38 39 40
	spin_unlock(&files->file_lock);
}

41
static bool get_close_on_exec(unsigned int fd)
Linus Torvalds's avatar
Linus Torvalds committed
42 43
{
	struct files_struct *files = current->files;
44
	struct fdtable *fdt;
45
	bool res;
46
	rcu_read_lock();
47
	fdt = files_fdtable(files);
48
	res = close_on_exec(fd, fdt);
49
	rcu_read_unlock();
Linus Torvalds's avatar
Linus Torvalds committed
50 51 52
	return res;
}

53
SYSCALL_DEFINE3(dup3, unsigned int, oldfd, unsigned int, newfd, int, flags)
Linus Torvalds's avatar
Linus Torvalds committed
54 55 56 57
{
	int err = -EBADF;
	struct file * file, *tofree;
	struct files_struct * files = current->files;
58
	struct fdtable *fdt;
Linus Torvalds's avatar
Linus Torvalds committed
59

Ulrich Drepper's avatar
Ulrich Drepper committed
60 61 62
	if ((flags & ~O_CLOEXEC) != 0)
		return -EINVAL;

63 64 65
	if (unlikely(oldfd == newfd))
		return -EINVAL;

Linus Torvalds's avatar
Linus Torvalds committed
66 67
	spin_lock(&files->file_lock);
	err = expand_files(files, newfd);
Al Viro's avatar
Al Viro committed
68 69 70
	file = fcheck(oldfd);
	if (unlikely(!file))
		goto Ebadf;
Al Viro's avatar
Al Viro committed
71 72
	if (unlikely(err < 0)) {
		if (err == -EMFILE)
Al Viro's avatar
Al Viro committed
73 74
			goto Ebadf;
		goto out_unlock;
Al Viro's avatar
Al Viro committed
75
	}
Al Viro's avatar
Al Viro committed
76 77 78 79 80 81 82 83 84 85 86 87 88 89
	/*
	 * We need to detect attempts to do dup2() over allocated but still
	 * not finished descriptor.  NB: OpenBSD avoids that at the price of
	 * extra work in their equivalent of fget() - they insert struct
	 * file immediately after grabbing descriptor, mark it larval if
	 * more work (e.g. actual opening) is needed and make sure that
	 * fget() treats larval files as absent.  Potentially interesting,
	 * but while extra work in fget() is trivial, locking implications
	 * and amount of surgery on open()-related paths in VFS are not.
	 * FreeBSD fails with -EBADF in the same situation, NetBSD "solution"
	 * deadlocks in rather amusing ways, AFAICS.  All of that is out of
	 * scope of POSIX or SUS, since neither considers shared descriptor
	 * tables and this condition does not arise without those.
	 */
Linus Torvalds's avatar
Linus Torvalds committed
90
	err = -EBUSY;
91 92
	fdt = files_fdtable(files);
	tofree = fdt->fd[newfd];
93
	if (!tofree && fd_is_open(newfd, fdt))
Al Viro's avatar
Al Viro committed
94 95
		goto out_unlock;
	get_file(file);
96
	rcu_assign_pointer(fdt->fd[newfd], file);
97
	__set_open_fd(newfd, fdt);
Ulrich Drepper's avatar
Ulrich Drepper committed
98
	if (flags & O_CLOEXEC)
99
		__set_close_on_exec(newfd, fdt);
Ulrich Drepper's avatar
Ulrich Drepper committed
100
	else
101
		__clear_close_on_exec(newfd, fdt);
Linus Torvalds's avatar
Linus Torvalds committed
102 103 104 105 106
	spin_unlock(&files->file_lock);

	if (tofree)
		filp_close(tofree, files);

Al Viro's avatar
Al Viro committed
107 108 109 110 111
	return newfd;

Ebadf:
	err = -EBADF;
out_unlock:
Linus Torvalds's avatar
Linus Torvalds committed
112
	spin_unlock(&files->file_lock);
Al Viro's avatar
Al Viro committed
113
	return err;
Linus Torvalds's avatar
Linus Torvalds committed
114
}
Ulrich Drepper's avatar
Ulrich Drepper committed
115

116
SYSCALL_DEFINE2(dup2, unsigned int, oldfd, unsigned int, newfd)
Ulrich Drepper's avatar
Ulrich Drepper committed
117
{
118 119
	if (unlikely(newfd == oldfd)) { /* corner case */
		struct files_struct *files = current->files;
120 121
		int retval = oldfd;

122 123
		rcu_read_lock();
		if (!fcheck_files(files, oldfd))
124
			retval = -EBADF;
125
		rcu_read_unlock();
126
		return retval;
127
	}
Ulrich Drepper's avatar
Ulrich Drepper committed
128 129
	return sys_dup3(oldfd, newfd, 0);
}
Linus Torvalds's avatar
Linus Torvalds committed
130

131
SYSCALL_DEFINE1(dup, unsigned int, fildes)
Linus Torvalds's avatar
Linus Torvalds committed
132 133
{
	int ret = -EBADF;
134
	struct file *file = fget_raw(fildes);
135 136 137 138 139 140 141 142

	if (file) {
		ret = get_unused_fd();
		if (ret >= 0)
			fd_install(ret, file);
		else
			fput(file);
	}
Linus Torvalds's avatar
Linus Torvalds committed
143 144 145
	return ret;
}

146
#define SETFL_MASK (O_APPEND | O_NONBLOCK | O_NDELAY | O_DIRECT | O_NOATIME)
Linus Torvalds's avatar
Linus Torvalds committed
147 148 149

static int setfl(int fd, struct file * filp, unsigned long arg)
{
150
	struct inode * inode = filp->f_path.dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
151 152
	int error = 0;

153 154 155 156 157
	/*
	 * 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
158 159 160 161
		return -EPERM;

	/* O_NOATIME can only be set by the owner or superuser */
	if ((arg & O_NOATIME) && !(filp->f_flags & O_NOATIME))
162
		if (!inode_owner_or_capable(inode))
Linus Torvalds's avatar
Linus Torvalds committed
163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180
			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;
	}

	if (filp->f_op && filp->f_op->check_flags)
		error = filp->f_op->check_flags(arg);
	if (error)
		return error;

181
	/*
182
	 * ->fasync() is responsible for setting the FASYNC bit.
183
	 */
184 185 186 187 188
	if (((arg ^ filp->f_flags) & FASYNC) && filp->f_op &&
			filp->f_op->fasync) {
		error = filp->f_op->fasync(fd, filp, (arg & FASYNC) != 0);
		if (error < 0)
			goto out;
189 190
		if (error > 0)
			error = 0;
Linus Torvalds's avatar
Linus Torvalds committed
191
	}
192
	spin_lock(&filp->f_lock);
Linus Torvalds's avatar
Linus Torvalds committed
193
	filp->f_flags = (arg & SETFL_MASK) | (filp->f_flags & ~SETFL_MASK);
194
	spin_unlock(&filp->f_lock);
195

Linus Torvalds's avatar
Linus Torvalds committed
196 197 198 199
 out:
	return error;
}

200
static void f_modown(struct file *filp, struct pid *pid, enum pid_type type,
201
                     int force)
Linus Torvalds's avatar
Linus Torvalds committed
202
{
203
	write_lock_irq(&filp->f_owner.lock);
Linus Torvalds's avatar
Linus Torvalds committed
204
	if (force || !filp->f_owner.pid) {
205 206 207
		put_pid(filp->f_owner.pid);
		filp->f_owner.pid = get_pid(pid);
		filp->f_owner.pid_type = type;
208 209 210 211 212 213

		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
214
	}
215
	write_unlock_irq(&filp->f_owner.lock);
Linus Torvalds's avatar
Linus Torvalds committed
216 217
}

218 219
int __f_setown(struct file *filp, struct pid *pid, enum pid_type type,
		int force)
Linus Torvalds's avatar
Linus Torvalds committed
220 221
{
	int err;
222

Linus Torvalds's avatar
Linus Torvalds committed
223 224 225 226
	err = security_file_set_fowner(filp);
	if (err)
		return err;

227
	f_modown(filp, pid, type, force);
Linus Torvalds's avatar
Linus Torvalds committed
228 229
	return 0;
}
230
EXPORT_SYMBOL(__f_setown);
Linus Torvalds's avatar
Linus Torvalds committed
231

232 233 234 235 236 237 238 239 240 241 242 243
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();
244
	pid = find_vpid(who);
245 246 247 248
	result = __f_setown(filp, pid, type, force);
	rcu_read_unlock();
	return result;
}
Linus Torvalds's avatar
Linus Torvalds committed
249 250 251 252
EXPORT_SYMBOL(f_setown);

void f_delown(struct file *filp)
{
253
	f_modown(filp, NULL, PIDTYPE_PID, 1);
254 255 256 257 258
}

pid_t f_getown(struct file *filp)
{
	pid_t pid;
259
	read_lock(&filp->f_owner.lock);
260
	pid = pid_vnr(filp->f_owner.pid);
261 262
	if (filp->f_owner.pid_type == PIDTYPE_PGID)
		pid = -pid;
263
	read_unlock(&filp->f_owner.lock);
264
	return pid;
Linus Torvalds's avatar
Linus Torvalds committed
265 266
}

Peter Zijlstra's avatar
Peter Zijlstra committed
267 268 269 270 271 272 273 274 275 276
static int f_setown_ex(struct file *filp, unsigned long arg)
{
	struct f_owner_ex * __user owner_p = (void * __user)arg;
	struct f_owner_ex owner;
	struct pid *pid;
	int type;
	int ret;

	ret = copy_from_user(&owner, owner_p, sizeof(owner));
	if (ret)
277
		return -EFAULT;
Peter Zijlstra's avatar
Peter Zijlstra committed
278 279 280 281 282 283 284 285 286 287

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

	case F_OWNER_PID:
		type = PIDTYPE_PID;
		break;

288
	case F_OWNER_PGRP:
Peter Zijlstra's avatar
Peter Zijlstra committed
289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324
		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)
{
	struct f_owner_ex * __user owner_p = (void * __user)arg;
	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:
325
		owner.type = F_OWNER_PGRP;
Peter Zijlstra's avatar
Peter Zijlstra committed
326 327 328 329 330 331 332 333 334
		break;

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

335
	if (!ret) {
Peter Zijlstra's avatar
Peter Zijlstra committed
336
		ret = copy_to_user(owner_p, &owner, sizeof(owner));
337 338 339
		if (ret)
			ret = -EFAULT;
	}
Peter Zijlstra's avatar
Peter Zijlstra committed
340 341 342
	return ret;
}

Linus Torvalds's avatar
Linus Torvalds committed
343 344 345 346 347 348 349
static long do_fcntl(int fd, unsigned int cmd, unsigned long arg,
		struct file *filp)
{
	long err = -EINVAL;

	switch (cmd) {
	case F_DUPFD:
350
	case F_DUPFD_CLOEXEC:
Jiri Slaby's avatar
Jiri Slaby committed
351
		if (arg >= rlimit(RLIMIT_NOFILE))
Al Viro's avatar
Al Viro committed
352
			break;
353 354 355 356 357
		err = alloc_fd(arg, cmd == F_DUPFD_CLOEXEC ? O_CLOEXEC : 0);
		if (err >= 0) {
			get_file(filp);
			fd_install(err, filp);
		}
Linus Torvalds's avatar
Linus Torvalds committed
358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376
		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;
	case F_GETLK:
		err = fcntl_getlk(filp, (struct flock __user *) arg);
		break;
	case F_SETLK:
	case F_SETLKW:
377
		err = fcntl_setlk(fd, filp, cmd, (struct flock __user *) arg);
Linus Torvalds's avatar
Linus Torvalds committed
378 379 380 381 382 383 384 385 386
		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.
		 */
387
		err = f_getown(filp);
Linus Torvalds's avatar
Linus Torvalds committed
388 389 390 391 392
		force_successful_syscall_return();
		break;
	case F_SETOWN:
		err = f_setown(filp, arg, 1);
		break;
Peter Zijlstra's avatar
Peter Zijlstra committed
393 394 395 396 397 398
	case F_GETOWN_EX:
		err = f_getown_ex(filp, arg);
		break;
	case F_SETOWN_EX:
		err = f_setown_ex(filp, arg);
		break;
Linus Torvalds's avatar
Linus Torvalds committed
399 400 401 402 403
	case F_GETSIG:
		err = filp->f_owner.signum;
		break;
	case F_SETSIG:
		/* arg == 0 restores default behaviour. */
404
		if (!valid_signal(arg)) {
Linus Torvalds's avatar
Linus Torvalds committed
405 406 407 408 409 410 411 412 413 414 415 416 417 418
			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;
419 420 421 422
	case F_SETPIPE_SZ:
	case F_GETPIPE_SZ:
		err = pipe_fcntl(filp, cmd, arg);
		break;
Linus Torvalds's avatar
Linus Torvalds committed
423 424 425 426 427 428
	default:
		break;
	}
	return err;
}

429 430 431 432 433 434 435 436 437 438 439 440 441
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;
}

442
SYSCALL_DEFINE3(fcntl, unsigned int, fd, unsigned int, cmd, unsigned long, arg)
Linus Torvalds's avatar
Linus Torvalds committed
443 444 445 446
{	
	struct file *filp;
	long err = -EBADF;

447
	filp = fget_raw(fd);
Linus Torvalds's avatar
Linus Torvalds committed
448 449 450
	if (!filp)
		goto out;

451 452 453 454 455 456 457
	if (unlikely(filp->f_mode & FMODE_PATH)) {
		if (!check_fcntl_cmd(cmd)) {
			fput(filp);
			goto out;
		}
	}

Linus Torvalds's avatar
Linus Torvalds committed
458 459 460 461 462 463 464 465 466 467 468 469 470 471
	err = security_file_fcntl(filp, cmd, arg);
	if (err) {
		fput(filp);
		return err;
	}

	err = do_fcntl(fd, cmd, arg, filp);

 	fput(filp);
out:
	return err;
}

#if BITS_PER_LONG == 32
472 473
SYSCALL_DEFINE3(fcntl64, unsigned int, fd, unsigned int, cmd,
		unsigned long, arg)
Linus Torvalds's avatar
Linus Torvalds committed
474 475 476 477 478
{	
	struct file * filp;
	long err;

	err = -EBADF;
479
	filp = fget_raw(fd);
Linus Torvalds's avatar
Linus Torvalds committed
480 481 482
	if (!filp)
		goto out;

483 484 485 486 487 488 489
	if (unlikely(filp->f_mode & FMODE_PATH)) {
		if (!check_fcntl_cmd(cmd)) {
			fput(filp);
			goto out;
		}
	}

Linus Torvalds's avatar
Linus Torvalds committed
490 491 492 493 494 495 496 497 498 499 500 501 502
	err = security_file_fcntl(filp, cmd, arg);
	if (err) {
		fput(filp);
		return err;
	}
	err = -EBADF;
	
	switch (cmd) {
		case F_GETLK64:
			err = fcntl_getlk64(filp, (struct flock64 __user *) arg);
			break;
		case F_SETLK64:
		case F_SETLKW64:
503 504
			err = fcntl_setlk64(fd, filp, cmd,
					(struct flock64 __user *) arg);
Linus Torvalds's avatar
Linus Torvalds committed
505 506 507 508 509 510 511 512 513 514 515 516 517
			break;
		default:
			err = do_fcntl(fd, cmd, arg, filp);
			break;
	}
	fput(filp);
out:
	return err;
}
#endif

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

518
static const long band_table[NSIGPOLL] = {
Linus Torvalds's avatar
Linus Torvalds committed
519 520 521 522 523 524 525 526 527 528 529
	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)
{
530 531 532 533 534
	const struct cred *cred;
	int ret;

	rcu_read_lock();
	cred = __task_cred(p);
535 536 537
	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)) &&
538 539 540
	       !security_file_send_sigiotask(p, fown, sig));
	rcu_read_unlock();
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
541 542 543
}

static void send_sigio_to_task(struct task_struct *p,
544
			       struct fown_struct *fown,
Peter Zijlstra's avatar
Peter Zijlstra committed
545
			       int fd, int reason, int group)
Linus Torvalds's avatar
Linus Torvalds committed
546
{
547 548 549 550 551 552 553
	/*
	 * 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
554 555
		return;

556
	switch (signum) {
Linus Torvalds's avatar
Linus Torvalds committed
557 558 559 560 561 562 563 564
		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 */
565
			si.si_signo = signum;
Linus Torvalds's avatar
Linus Torvalds committed
566 567 568 569 570
			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.  */
571
			BUG_ON((reason & __SI_MASK) != __SI_POLL);
Linus Torvalds's avatar
Linus Torvalds committed
572 573 574 575 576
			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
577
			if (!do_send_sig_info(signum, &si, p, group))
Linus Torvalds's avatar
Linus Torvalds committed
578 579 580
				break;
		/* fall-through: fall back on the old plain SIGIO signal */
		case 0:
Peter Zijlstra's avatar
Peter Zijlstra committed
581
			do_send_sig_info(SIGIO, SEND_SIG_PRIV, p, group);
Linus Torvalds's avatar
Linus Torvalds committed
582 583 584 585 586 587
	}
}

void send_sigio(struct fown_struct *fown, int fd, int band)
{
	struct task_struct *p;
588 589
	enum pid_type type;
	struct pid *pid;
Peter Zijlstra's avatar
Peter Zijlstra committed
590
	int group = 1;
Linus Torvalds's avatar
Linus Torvalds committed
591 592
	
	read_lock(&fown->lock);
Peter Zijlstra's avatar
Peter Zijlstra committed
593

594
	type = fown->pid_type;
Peter Zijlstra's avatar
Peter Zijlstra committed
595 596 597 598 599
	if (type == PIDTYPE_MAX) {
		group = 0;
		type = PIDTYPE_PID;
	}

Linus Torvalds's avatar
Linus Torvalds committed
600 601 602 603 604
	pid = fown->pid;
	if (!pid)
		goto out_unlock_fown;
	
	read_lock(&tasklist_lock);
605
	do_each_pid_task(pid, type, p) {
Peter Zijlstra's avatar
Peter Zijlstra committed
606
		send_sigio_to_task(p, fown, fd, band, group);
607
	} while_each_pid_task(pid, type, p);
Linus Torvalds's avatar
Linus Torvalds committed
608 609 610 611 612 613
	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
614
				struct fown_struct *fown, int group)
Linus Torvalds's avatar
Linus Torvalds committed
615 616
{
	if (sigio_perm(p, fown, SIGURG))
Peter Zijlstra's avatar
Peter Zijlstra committed
617
		do_send_sig_info(SIGURG, SEND_SIG_PRIV, p, group);
Linus Torvalds's avatar
Linus Torvalds committed
618 619 620 621 622
}

int send_sigurg(struct fown_struct *fown)
{
	struct task_struct *p;
623 624
	enum pid_type type;
	struct pid *pid;
Peter Zijlstra's avatar
Peter Zijlstra committed
625
	int group = 1;
626
	int ret = 0;
Linus Torvalds's avatar
Linus Torvalds committed
627 628
	
	read_lock(&fown->lock);
Peter Zijlstra's avatar
Peter Zijlstra committed
629

630
	type = fown->pid_type;
Peter Zijlstra's avatar
Peter Zijlstra committed
631 632 633 634 635
	if (type == PIDTYPE_MAX) {
		group = 0;
		type = PIDTYPE_PID;
	}

Linus Torvalds's avatar
Linus Torvalds committed
636 637 638 639 640 641 642
	pid = fown->pid;
	if (!pid)
		goto out_unlock_fown;

	ret = 1;
	
	read_lock(&tasklist_lock);
643
	do_each_pid_task(pid, type, p) {
Peter Zijlstra's avatar
Peter Zijlstra committed
644
		send_sigurg_to_task(p, fown, group);
645
	} while_each_pid_task(pid, type, p);
Linus Torvalds's avatar
Linus Torvalds committed
646 647 648 649 650 651
	read_unlock(&tasklist_lock);
 out_unlock_fown:
	read_unlock(&fown->lock);
	return ret;
}

652
static DEFINE_SPINLOCK(fasync_lock);
653
static struct kmem_cache *fasync_cache __read_mostly;
Linus Torvalds's avatar
Linus Torvalds committed
654

655 656 657 658 659 660
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
661
/*
662 663 664 665 666 667 668
 * 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
669
 */
670
int fasync_remove_entry(struct file *filp, struct fasync_struct **fapp)
Linus Torvalds's avatar
Linus Torvalds committed
671 672 673 674
{
	struct fasync_struct *fa, **fp;
	int result = 0;

675
	spin_lock(&filp->f_lock);
676
	spin_lock(&fasync_lock);
677 678 679
	for (fp = fapp; (fa = *fp) != NULL; fp = &fa->fa_next) {
		if (fa->fa_file != filp)
			continue;
680 681 682 683 684

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

685
		*fp = fa->fa_next;
686
		call_rcu(&fa->fa_rcu, fasync_free_rcu);
687 688 689
		filp->f_flags &= ~FASYNC;
		result = 1;
		break;
Linus Torvalds's avatar
Linus Torvalds committed
690
	}
691
	spin_unlock(&fasync_lock);
692 693 694 695
	spin_unlock(&filp->f_lock);
	return result;
}

696 697 698 699 700
struct fasync_struct *fasync_alloc(void)
{
	return kmem_cache_alloc(fasync_cache, GFP_KERNEL);
}

701
/*
702 703 704
 * 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.
705
 */
706
void fasync_free(struct fasync_struct *new)
707
{
708 709
	kmem_cache_free(fasync_cache, new);
}
710

711 712 713
/*
 * Insert a new entry into the fasync list.  Return the pointer to the
 * old one if we didn't use the new one.
714 715 716
 *
 * NOTE! It is very important that the FASYNC flag always
 * match the state "is the filp on a fasync list".
717 718 719 720
 */
struct fasync_struct *fasync_insert_entry(int fd, struct file *filp, struct fasync_struct **fapp, struct fasync_struct *new)
{
        struct fasync_struct *fa, **fp;
721 722

	spin_lock(&filp->f_lock);
723
	spin_lock(&fasync_lock);
Linus Torvalds's avatar
Linus Torvalds committed
724
	for (fp = fapp; (fa = *fp) != NULL; fp = &fa->fa_next) {
725 726
		if (fa->fa_file != filp)
			continue;
727 728

		spin_lock_irq(&fa->fa_lock);
729
		fa->fa_fd = fd;
730
		spin_unlock_irq(&fa->fa_lock);
731
		goto out;
Linus Torvalds's avatar
Linus Torvalds committed
732 733
	}

734
	spin_lock_init(&new->fa_lock);
735 736 737 738
	new->magic = FASYNC_MAGIC;
	new->fa_file = filp;
	new->fa_fd = fd;
	new->fa_next = *fapp;
739
	rcu_assign_pointer(*fapp, new);
740 741
	filp->f_flags |= FASYNC;

Linus Torvalds's avatar
Linus Torvalds committed
742
out:
743
	spin_unlock(&fasync_lock);
744
	spin_unlock(&filp->f_lock);
745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772
	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
773 774
}

775 776 777 778 779 780 781 782 783 784 785 786 787
/*
 * 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
788 789
EXPORT_SYMBOL(fasync_helper);

790 791 792 793
/*
 * 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
794 795
{
	while (fa) {
796
		struct fown_struct *fown;
797 798
		unsigned long flags;

Linus Torvalds's avatar
Linus Torvalds committed
799 800 801 802 803
		if (fa->magic != FASYNC_MAGIC) {
			printk(KERN_ERR "kill_fasync: bad magic number in "
			       "fasync_struct!\n");
			return;
		}
804
		spin_lock_irqsave(&fa->fa_lock, flags);
805 806 807 808 809 810 811 812
		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);
		}
813
		spin_unlock_irqrestore(&fa->fa_lock, flags);
814
		fa = rcu_dereference(fa->fa_next);
Linus Torvalds's avatar
Linus Torvalds committed
815 816 817 818 819 820 821 822 823
	}
}

void kill_fasync(struct fasync_struct **fp, int sig, int band)
{
	/* First a quick test without locking: usually
	 * the list is empty.
	 */
	if (*fp) {
824 825 826
		rcu_read_lock();
		kill_fasync_rcu(rcu_dereference(*fp), sig, band);
		rcu_read_unlock();
Linus Torvalds's avatar
Linus Torvalds committed
827 828 829 830
	}
}
EXPORT_SYMBOL(kill_fasync);

831
static int __init fcntl_init(void)
Linus Torvalds's avatar
Linus Torvalds committed
832
{
833 834 835 836 837
	/*
	 * 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.
	 */
838
	BUILD_BUG_ON(19 - 1 /* for O_RDONLY being 0 */ != HWEIGHT32(
839 840
		O_RDONLY	| O_WRONLY	| O_RDWR	|
		O_CREAT		| O_EXCL	| O_NOCTTY	|
841
		O_TRUNC		| O_APPEND	| /* O_NONBLOCK	| */
842 843 844
		__O_SYNC	| O_DSYNC	| FASYNC	|
		O_DIRECT	| O_LARGEFILE	| O_DIRECTORY	|
		O_NOFOLLOW	| O_NOATIME	| O_CLOEXEC	|
845
		__FMODE_EXEC	| O_PATH
846 847
		));

Linus Torvalds's avatar
Linus Torvalds committed
848
	fasync_cache = kmem_cache_create("fasync_cache",
849
		sizeof(struct fasync_struct), 0, SLAB_PANIC, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
850 851 852
	return 0;
}

853
module_init(fcntl_init)