fcntl.c 23.1 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0
Linus Torvalds's avatar
Linus Torvalds committed
2 3 4 5 6 7 8 9 10
/*
 *  linux/fs/fcntl.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 */

#include <linux/syscalls.h>
#include <linux/init.h>
#include <linux/mm.h>
11
#include <linux/sched/task.h>
Linus Torvalds's avatar
Linus Torvalds committed
12 13
#include <linux/fs.h>
#include <linux/file.h>
Al Viro's avatar
Al Viro committed
14
#include <linux/fdtable.h>
15
#include <linux/capability.h>
Linus Torvalds's avatar
Linus Torvalds committed
16 17 18
#include <linux/dnotify.h>
#include <linux/slab.h>
#include <linux/module.h>
19
#include <linux/pipe_fs_i.h>
Linus Torvalds's avatar
Linus Torvalds committed
20 21
#include <linux/security.h>
#include <linux/ptrace.h>
22
#include <linux/signal.h>
23
#include <linux/rcupdate.h>
24
#include <linux/pid_namespace.h>
25
#include <linux/user_namespace.h>
26
#include <linux/memfd.h>
27
#include <linux/compat.h>
Linus Torvalds's avatar
Linus Torvalds committed
28

29
#include <linux/poll.h>
Linus Torvalds's avatar
Linus Torvalds committed
30
#include <asm/siginfo.h>
31
#include <linux/uaccess.h>
Linus Torvalds's avatar
Linus Torvalds committed
32

33
#define SETFL_MASK (O_APPEND | O_NONBLOCK | O_NDELAY | O_DIRECT | O_NOATIME)
Linus Torvalds's avatar
Linus Torvalds committed
34 35 36

static int setfl(int fd, struct file * filp, unsigned long arg)
{
Al Viro's avatar
Al Viro committed
37
	struct inode * inode = file_inode(filp);
Linus Torvalds's avatar
Linus Torvalds committed
38 39
	int error = 0;

40 41 42 43 44
	/*
	 * 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
45 46 47 48
		return -EPERM;

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

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

57
	/* Pipe packetized mode is controlled by O_DIRECT flag */
Al Viro's avatar
Al Viro committed
58
	if (!S_ISFIFO(inode->i_mode) && (arg & O_DIRECT)) {
Linus Torvalds's avatar
Linus Torvalds committed
59 60 61 62 63
		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
64
	if (filp->f_op->check_flags)
Linus Torvalds's avatar
Linus Torvalds committed
65 66 67 68
		error = filp->f_op->check_flags(arg);
	if (error)
		return error;

69
	/*
70
	 * ->fasync() is responsible for setting the FASYNC bit.
71
	 */
Al Viro's avatar
Al Viro committed
72
	if (((arg ^ filp->f_flags) & FASYNC) && filp->f_op->fasync) {
73 74 75
		error = filp->f_op->fasync(fd, filp, (arg & FASYNC) != 0);
		if (error < 0)
			goto out;
76 77
		if (error > 0)
			error = 0;
Linus Torvalds's avatar
Linus Torvalds committed
78
	}
79
	spin_lock(&filp->f_lock);
Linus Torvalds's avatar
Linus Torvalds committed
80
	filp->f_flags = (arg & SETFL_MASK) | (filp->f_flags & ~SETFL_MASK);
81
	spin_unlock(&filp->f_lock);
82

Linus Torvalds's avatar
Linus Torvalds committed
83 84 85 86
 out:
	return error;
}

87
static void f_modown(struct file *filp, struct pid *pid, enum pid_type type,
88
                     int force)
Linus Torvalds's avatar
Linus Torvalds committed
89
{
90
	write_lock_irq(&filp->f_owner.lock);
Linus Torvalds's avatar
Linus Torvalds committed
91
	if (force || !filp->f_owner.pid) {
92 93 94
		put_pid(filp->f_owner.pid);
		filp->f_owner.pid = get_pid(pid);
		filp->f_owner.pid_type = type;
95 96 97 98 99 100

		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
101
	}
102
	write_unlock_irq(&filp->f_owner.lock);
Linus Torvalds's avatar
Linus Torvalds committed
103 104
}

105
void __f_setown(struct file *filp, struct pid *pid, enum pid_type type,
106
		int force)
Linus Torvalds's avatar
Linus Torvalds committed
107
{
108
	security_file_set_fowner(filp);
109
	f_modown(filp, pid, type, force);
Linus Torvalds's avatar
Linus Torvalds committed
110
}
111
EXPORT_SYMBOL(__f_setown);
Linus Torvalds's avatar
Linus Torvalds committed
112

113
int f_setown(struct file *filp, unsigned long arg, int force)
114 115
{
	enum pid_type type;
116 117 118
	struct pid *pid = NULL;
	int who = arg, ret = 0;

119 120
	type = PIDTYPE_PID;
	if (who < 0) {
121 122 123 124
		/* avoid overflow below */
		if (who == INT_MIN)
			return -EINVAL;

125 126 127
		type = PIDTYPE_PGID;
		who = -who;
	}
128

129
	rcu_read_lock();
130 131 132 133 134 135 136 137
	if (who) {
		pid = find_vpid(who);
		if (!pid)
			ret = -ESRCH;
	}

	if (!ret)
		__f_setown(filp, pid, type, force);
138
	rcu_read_unlock();
139

140
	return ret;
141
}
Linus Torvalds's avatar
Linus Torvalds committed
142 143 144 145
EXPORT_SYMBOL(f_setown);

void f_delown(struct file *filp)
{
146
	f_modown(filp, NULL, PIDTYPE_PID, 1);
147 148 149 150 151
}

pid_t f_getown(struct file *filp)
{
	pid_t pid;
152
	read_lock(&filp->f_owner.lock);
153
	pid = pid_vnr(filp->f_owner.pid);
154 155
	if (filp->f_owner.pid_type == PIDTYPE_PGID)
		pid = -pid;
156
	read_unlock(&filp->f_owner.lock);
157
	return pid;
Linus Torvalds's avatar
Linus Torvalds committed
158 159
}

160 161
static int f_setown_ex(struct file *filp, unsigned long arg)
{
Al Viro's avatar
Al Viro committed
162
	struct f_owner_ex __user *owner_p = (void __user *)arg;
163 164 165 166 167 168 169
	struct f_owner_ex owner;
	struct pid *pid;
	int type;
	int ret;

	ret = copy_from_user(&owner, owner_p, sizeof(owner));
	if (ret)
170
		return -EFAULT;
171 172 173 174 175 176 177 178 179 180

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

	case F_OWNER_PID:
		type = PIDTYPE_PID;
		break;

181
	case F_OWNER_PGRP:
182 183 184 185 186 187 188 189 190 191 192 193
		type = PIDTYPE_PGID;
		break;

	default:
		return -EINVAL;
	}

	rcu_read_lock();
	pid = find_vpid(owner.pid);
	if (owner.pid && !pid)
		ret = -ESRCH;
	else
194
		 __f_setown(filp, pid, type, 1);
195 196 197 198 199 200 201
	rcu_read_unlock();

	return ret;
}

static int f_getown_ex(struct file *filp, unsigned long arg)
{
Al Viro's avatar
Al Viro committed
202
	struct f_owner_ex __user *owner_p = (void __user *)arg;
203 204 205 206 207 208 209 210 211 212 213 214 215 216 217
	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:
218
		owner.type = F_OWNER_PGRP;
219 220 221 222 223 224 225 226 227
		break;

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

228
	if (!ret) {
229
		ret = copy_to_user(owner_p, &owner, sizeof(owner));
230 231 232
		if (ret)
			ret = -EFAULT;
	}
233 234 235
	return ret;
}

236 237 238 239
#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
240
	uid_t __user *dst = (void __user *)arg;
241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260
	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

261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281
static bool rw_hint_valid(enum rw_hint hint)
{
	switch (hint) {
	case RWF_WRITE_LIFE_NOT_SET:
	case RWH_WRITE_LIFE_NONE:
	case RWH_WRITE_LIFE_SHORT:
	case RWH_WRITE_LIFE_MEDIUM:
	case RWH_WRITE_LIFE_LONG:
	case RWH_WRITE_LIFE_EXTREME:
		return true;
	default:
		return false;
	}
}

static long fcntl_rw_hint(struct file *file, unsigned int cmd,
			  unsigned long arg)
{
	struct inode *inode = file_inode(file);
	u64 *argp = (u64 __user *)arg;
	enum rw_hint hint;
282
	u64 h;
283 284 285

	switch (cmd) {
	case F_GET_FILE_RW_HINT:
286 287
		h = file_write_hint(file);
		if (copy_to_user(argp, &h, sizeof(*argp)))
288 289 290
			return -EFAULT;
		return 0;
	case F_SET_FILE_RW_HINT:
291
		if (copy_from_user(&h, argp, sizeof(h)))
292
			return -EFAULT;
293
		hint = (enum rw_hint) h;
294 295 296 297 298 299 300 301
		if (!rw_hint_valid(hint))
			return -EINVAL;

		spin_lock(&file->f_lock);
		file->f_write_hint = hint;
		spin_unlock(&file->f_lock);
		return 0;
	case F_GET_RW_HINT:
302 303
		h = inode->i_write_hint;
		if (copy_to_user(argp, &h, sizeof(*argp)))
304 305 306
			return -EFAULT;
		return 0;
	case F_SET_RW_HINT:
307
		if (copy_from_user(&h, argp, sizeof(h)))
308
			return -EFAULT;
309
		hint = (enum rw_hint) h;
310 311 312 313 314 315 316 317 318 319 320 321
		if (!rw_hint_valid(hint))
			return -EINVAL;

		inode_lock(inode);
		inode->i_write_hint = hint;
		inode_unlock(inode);
		return 0;
	default:
		return -EINVAL;
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
322 323 324
static long do_fcntl(int fd, unsigned int cmd, unsigned long arg,
		struct file *filp)
{
325 326
	void __user *argp = (void __user *)arg;
	struct flock flock;
Linus Torvalds's avatar
Linus Torvalds committed
327 328 329 330
	long err = -EINVAL;

	switch (cmd) {
	case F_DUPFD:
331 332
		err = f_dupfd(arg, filp, 0);
		break;
333
	case F_DUPFD_CLOEXEC:
Al Viro's avatar
Al Viro committed
334
		err = f_dupfd(arg, filp, O_CLOEXEC);
Linus Torvalds's avatar
Linus Torvalds committed
335 336 337 338 339 340 341 342 343 344 345 346 347 348
		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;
349 350
#if BITS_PER_LONG != 32
	/* 32-bit arches must use fcntl64() */
351
	case F_OFD_GETLK:
352
#endif
Linus Torvalds's avatar
Linus Torvalds committed
353
	case F_GETLK:
354 355 356 357 358
		if (copy_from_user(&flock, argp, sizeof(flock)))
			return -EFAULT;
		err = fcntl_getlk(filp, cmd, &flock);
		if (!err && copy_to_user(argp, &flock, sizeof(flock)))
			return -EFAULT;
Linus Torvalds's avatar
Linus Torvalds committed
359
		break;
360 361
#if BITS_PER_LONG != 32
	/* 32-bit arches must use fcntl64() */
362 363
	case F_OFD_SETLK:
	case F_OFD_SETLKW:
364 365
#endif
		/* Fallthrough */
Linus Torvalds's avatar
Linus Torvalds committed
366 367
	case F_SETLK:
	case F_SETLKW:
368 369 370
		if (copy_from_user(&flock, argp, sizeof(flock)))
			return -EFAULT;
		err = fcntl_setlk(fd, filp, cmd, &flock);
Linus Torvalds's avatar
Linus Torvalds committed
371 372 373 374 375 376 377 378 379
		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.
		 */
380
		err = f_getown(filp);
Linus Torvalds's avatar
Linus Torvalds committed
381 382 383
		force_successful_syscall_return();
		break;
	case F_SETOWN:
384
		err = f_setown(filp, arg, 1);
Linus Torvalds's avatar
Linus Torvalds committed
385
		break;
386 387 388 389 390 391
	case F_GETOWN_EX:
		err = f_getown_ex(filp, arg);
		break;
	case F_SETOWN_EX:
		err = f_setown_ex(filp, arg);
		break;
392 393 394
	case F_GETOWNER_UIDS:
		err = f_getowner_uids(filp, arg);
		break;
Linus Torvalds's avatar
Linus Torvalds committed
395 396 397 398 399
	case F_GETSIG:
		err = filp->f_owner.signum;
		break;
	case F_SETSIG:
		/* arg == 0 restores default behaviour. */
400
		if (!valid_signal(arg)) {
Linus Torvalds's avatar
Linus Torvalds committed
401 402 403 404 405 406 407 408 409 410 411 412 413 414
			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;
415 416 417 418
	case F_SETPIPE_SZ:
	case F_GETPIPE_SZ:
		err = pipe_fcntl(filp, cmd, arg);
		break;
David Herrmann's avatar
David Herrmann committed
419 420
	case F_ADD_SEALS:
	case F_GET_SEALS:
421
		err = memfd_fcntl(filp, cmd, arg);
David Herrmann's avatar
David Herrmann committed
422
		break;
423 424 425 426 427 428
	case F_GET_RW_HINT:
	case F_SET_RW_HINT:
	case F_GET_FILE_RW_HINT:
	case F_SET_FILE_RW_HINT:
		err = fcntl_rw_hint(filp, cmd, arg);
		break;
Linus Torvalds's avatar
Linus Torvalds committed
429 430 431 432 433 434
	default:
		break;
	}
	return err;
}

435 436 437 438 439 440 441 442 443 444 445 446 447
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;
}

448
SYSCALL_DEFINE3(fcntl, unsigned int, fd, unsigned int, cmd, unsigned long, arg)
Linus Torvalds's avatar
Linus Torvalds committed
449
{	
450
	struct fd f = fdget_raw(fd);
Linus Torvalds's avatar
Linus Torvalds committed
451 452
	long err = -EBADF;

453
	if (!f.file)
Linus Torvalds's avatar
Linus Torvalds committed
454 455
		goto out;

456
	if (unlikely(f.file->f_mode & FMODE_PATH)) {
457 458
		if (!check_fcntl_cmd(cmd))
			goto out1;
459 460
	}

461
	err = security_file_fcntl(f.file, cmd, arg);
462
	if (!err)
463
		err = do_fcntl(fd, cmd, arg, f.file);
Linus Torvalds's avatar
Linus Torvalds committed
464

465
out1:
466
 	fdput(f);
Linus Torvalds's avatar
Linus Torvalds committed
467 468 469 470 471
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
	void __user *argp = (void __user *)arg;
476
	struct fd f = fdget_raw(fd);
477
	struct flock64 flock;
478
	long err = -EBADF;
Linus Torvalds's avatar
Linus Torvalds committed
479

480
	if (!f.file)
Linus Torvalds's avatar
Linus Torvalds committed
481 482
		goto out;

483
	if (unlikely(f.file->f_mode & FMODE_PATH)) {
484 485
		if (!check_fcntl_cmd(cmd))
			goto out1;
486 487
	}

488
	err = security_file_fcntl(f.file, cmd, arg);
489 490
	if (err)
		goto out1;
Linus Torvalds's avatar
Linus Torvalds committed
491 492
	
	switch (cmd) {
493
	case F_GETLK64:
494
	case F_OFD_GETLK:
495 496 497 498 499 500
		err = -EFAULT;
		if (copy_from_user(&flock, argp, sizeof(flock)))
			break;
		err = fcntl_getlk64(f.file, cmd, &flock);
		if (!err && copy_to_user(argp, &flock, sizeof(flock)))
			err = -EFAULT;
501 502 503
		break;
	case F_SETLK64:
	case F_SETLKW64:
504 505
	case F_OFD_SETLK:
	case F_OFD_SETLKW:
506 507 508 509
		err = -EFAULT;
		if (copy_from_user(&flock, argp, sizeof(flock)))
			break;
		err = fcntl_setlk64(fd, f.file, cmd, &flock);
510 511 512 513
		break;
	default:
		err = do_fcntl(fd, cmd, arg, f.file);
		break;
Linus Torvalds's avatar
Linus Torvalds committed
514
	}
515
out1:
516
	fdput(f);
Linus Torvalds's avatar
Linus Torvalds committed
517 518 519 520 521
out:
	return err;
}
#endif

522
#ifdef CONFIG_COMPAT
523
/* careful - don't use anywhere else */
524 525 526 527 528 529 530 531
#define copy_flock_fields(dst, src)		\
	(dst)->l_type = (src)->l_type;		\
	(dst)->l_whence = (src)->l_whence;	\
	(dst)->l_start = (src)->l_start;	\
	(dst)->l_len = (src)->l_len;		\
	(dst)->l_pid = (src)->l_pid;

static int get_compat_flock(struct flock *kfl, const struct compat_flock __user *ufl)
532
{
533 534 535
	struct compat_flock fl;

	if (copy_from_user(&fl, ufl, sizeof(struct compat_flock)))
536
		return -EFAULT;
537
	copy_flock_fields(kfl, &fl);
538 539 540
	return 0;
}

541
static int get_compat_flock64(struct flock *kfl, const struct compat_flock64 __user *ufl)
542
{
543 544 545
	struct compat_flock64 fl;

	if (copy_from_user(&fl, ufl, sizeof(struct compat_flock64)))
546
		return -EFAULT;
547
	copy_flock_fields(kfl, &fl);
548 549 550
	return 0;
}

551
static int put_compat_flock(const struct flock *kfl, struct compat_flock __user *ufl)
552
{
553 554 555
	struct compat_flock fl;

	memset(&fl, 0, sizeof(struct compat_flock));
556
	copy_flock_fields(&fl, kfl);
557
	if (copy_to_user(ufl, &fl, sizeof(struct compat_flock)))
558 559 560 561
		return -EFAULT;
	return 0;
}

562
static int put_compat_flock64(const struct flock *kfl, struct compat_flock64 __user *ufl)
563
{
564 565
	struct compat_flock64 fl;

566 567 568
	BUILD_BUG_ON(sizeof(kfl->l_start) > sizeof(ufl->l_start));
	BUILD_BUG_ON(sizeof(kfl->l_len) > sizeof(ufl->l_len));

569
	memset(&fl, 0, sizeof(struct compat_flock64));
570
	copy_flock_fields(&fl, kfl);
571
	if (copy_to_user(ufl, &fl, sizeof(struct compat_flock64)))
572 573 574
		return -EFAULT;
	return 0;
}
575
#undef copy_flock_fields
576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591

static unsigned int
convert_fcntl_cmd(unsigned int cmd)
{
	switch (cmd) {
	case F_GETLK64:
		return F_GETLK;
	case F_SETLK64:
		return F_SETLK;
	case F_SETLKW64:
		return F_SETLKW;
	}

	return cmd;
}

592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609
/*
 * GETLK was successful and we need to return the data, but it needs to fit in
 * the compat structure.
 * l_start shouldn't be too big, unless the original start + end is greater than
 * COMPAT_OFF_T_MAX, in which case the app was asking for trouble, so we return
 * -EOVERFLOW in that case.  l_len could be too big, in which case we just
 * truncate it, and only allow the app to see that part of the conflicting lock
 * that might make sense to it anyway
 */
static int fixup_compat_flock(struct flock *flock)
{
	if (flock->l_start > COMPAT_OFF_T_MAX)
		return -EOVERFLOW;
	if (flock->l_len > COMPAT_OFF_T_MAX)
		flock->l_len = COMPAT_OFF_T_MAX;
	return 0;
}

610 611
static long do_compat_fcntl64(unsigned int fd, unsigned int cmd,
			     compat_ulong_t arg)
612
{
613 614 615 616 617 618 619 620 621 622 623 624 625 626 627
	struct fd f = fdget_raw(fd);
	struct flock flock;
	long err = -EBADF;

	if (!f.file)
		return err;

	if (unlikely(f.file->f_mode & FMODE_PATH)) {
		if (!check_fcntl_cmd(cmd))
			goto out_put;
	}

	err = security_file_fcntl(f.file, cmd, arg);
	if (err)
		goto out_put;
628 629 630

	switch (cmd) {
	case F_GETLK:
631 632 633 634 635 636 637
		err = get_compat_flock(&flock, compat_ptr(arg));
		if (err)
			break;
		err = fcntl_getlk(f.file, convert_fcntl_cmd(cmd), &flock);
		if (err)
			break;
		err = fixup_compat_flock(&flock);
638 639
		if (!err)
			err = put_compat_flock(&flock, compat_ptr(arg));
640 641 642 643 644 645 646
		break;
	case F_GETLK64:
	case F_OFD_GETLK:
		err = get_compat_flock64(&flock, compat_ptr(arg));
		if (err)
			break;
		err = fcntl_getlk(f.file, convert_fcntl_cmd(cmd), &flock);
647 648
		if (!err)
			err = put_compat_flock64(&flock, compat_ptr(arg));
649
		break;
650 651
	case F_SETLK:
	case F_SETLKW:
652 653
		err = get_compat_flock(&flock, compat_ptr(arg));
		if (err)
654
			break;
655
		err = fcntl_setlk(fd, f.file, convert_fcntl_cmd(cmd), &flock);
656 657 658 659 660
		break;
	case F_SETLK64:
	case F_SETLKW64:
	case F_OFD_SETLK:
	case F_OFD_SETLKW:
661 662
		err = get_compat_flock64(&flock, compat_ptr(arg));
		if (err)
663
			break;
664
		err = fcntl_setlk(fd, f.file, convert_fcntl_cmd(cmd), &flock);
665 666
		break;
	default:
667
		err = do_fcntl(fd, cmd, arg, f.file);
668 669
		break;
	}
670 671 672
out_put:
	fdput(f);
	return err;
673 674
}

675 676 677 678 679 680
COMPAT_SYSCALL_DEFINE3(fcntl64, unsigned int, fd, unsigned int, cmd,
		       compat_ulong_t, arg)
{
	return do_compat_fcntl64(fd, cmd, arg);
}

681 682 683 684 685 686 687 688 689 690 691 692
COMPAT_SYSCALL_DEFINE3(fcntl, unsigned int, fd, unsigned int, cmd,
		       compat_ulong_t, arg)
{
	switch (cmd) {
	case F_GETLK64:
	case F_SETLK64:
	case F_SETLKW64:
	case F_OFD_GETLK:
	case F_OFD_SETLK:
	case F_OFD_SETLKW:
		return -EINVAL;
	}
693
	return do_compat_fcntl64(fd, cmd, arg);
694 695 696
}
#endif

Linus Torvalds's avatar
Linus Torvalds committed
697 698
/* Table to convert sigio signal codes into poll band bitmaps */

699
static const __poll_t band_table[NSIGPOLL] = {
700 701 702 703 704 705
	EPOLLIN | EPOLLRDNORM,			/* POLL_IN */
	EPOLLOUT | EPOLLWRNORM | EPOLLWRBAND,	/* POLL_OUT */
	EPOLLIN | EPOLLRDNORM | EPOLLMSG,		/* POLL_MSG */
	EPOLLERR,				/* POLL_ERR */
	EPOLLPRI | EPOLLRDBAND,			/* POLL_PRI */
	EPOLLHUP | EPOLLERR			/* POLL_HUP */
Linus Torvalds's avatar
Linus Torvalds committed
706 707 708 709 710
};

static inline int sigio_perm(struct task_struct *p,
                             struct fown_struct *fown, int sig)
{
711 712 713 714 715
	const struct cred *cred;
	int ret;

	rcu_read_lock();
	cred = __task_cred(p);
716 717 718
	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)) &&
719 720 721
	       !security_file_send_sigiotask(p, fown, sig));
	rcu_read_unlock();
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
722 723 724
}

static void send_sigio_to_task(struct task_struct *p,
725
			       struct fown_struct *fown,
726
			       int fd, int reason, int group)
Linus Torvalds's avatar
Linus Torvalds committed
727
{
728 729 730 731
	/*
	 * F_SETSIG can change ->signum lockless in parallel, make
	 * sure we read it once and use the same value throughout.
	 */
732
	int signum = READ_ONCE(fown->signum);
733 734

	if (!sigio_perm(p, fown, signum))
Linus Torvalds's avatar
Linus Torvalds committed
735 736
		return;

737
	switch (signum) {
Linus Torvalds's avatar
Linus Torvalds committed
738 739 740 741 742 743 744 745
		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 */
746
			clear_siginfo(&si);
747
			si.si_signo = signum;
Linus Torvalds's avatar
Linus Torvalds committed
748 749
			si.si_errno = 0;
		        si.si_code  = reason;
750 751 752 753 754 755 756 757
			/*
			 * Posix definies POLL_IN and friends to be signal
			 * specific si_codes for SIG_POLL.  Linux extended
			 * these si_codes to other signals in a way that is
			 * ambiguous if other signals also have signal
			 * specific si_codes.  In that case use SI_SIGIO instead
			 * to remove the ambiguity.
			 */
758
			if ((signum != SIGPOLL) && sig_specific_sicodes(signum))
759 760
				si.si_code = SI_SIGIO;

Linus Torvalds's avatar
Linus Torvalds committed
761 762 763
			/* Make sure we are called with one of the POLL_*
			   reasons, otherwise we could leak kernel stack into
			   userspace.  */
764
			BUG_ON((reason < POLL_IN) || ((reason - POLL_IN) >= NSIGPOLL));
Linus Torvalds's avatar
Linus Torvalds committed
765 766 767
			if (reason - POLL_IN >= NSIGPOLL)
				si.si_band  = ~0L;
			else
768
				si.si_band = mangle_poll(band_table[reason - POLL_IN]);
Linus Torvalds's avatar
Linus Torvalds committed
769
			si.si_fd    = fd;
770
			if (!do_send_sig_info(signum, &si, p, group))
Linus Torvalds's avatar
Linus Torvalds committed
771 772 773
				break;
		/* fall-through: fall back on the old plain SIGIO signal */
		case 0:
774
			do_send_sig_info(SIGIO, SEND_SIG_PRIV, p, group);
Linus Torvalds's avatar
Linus Torvalds committed
775 776 777 778 779 780
	}
}

void send_sigio(struct fown_struct *fown, int fd, int band)
{
	struct task_struct *p;
781 782
	enum pid_type type;
	struct pid *pid;
783
	int group = 1;
Linus Torvalds's avatar
Linus Torvalds committed
784 785
	
	read_lock(&fown->lock);
786

787
	type = fown->pid_type;
788 789 790 791 792
	if (type == PIDTYPE_MAX) {
		group = 0;
		type = PIDTYPE_PID;
	}

Linus Torvalds's avatar
Linus Torvalds committed
793 794 795 796 797
	pid = fown->pid;
	if (!pid)
		goto out_unlock_fown;
	
	read_lock(&tasklist_lock);
798
	do_each_pid_task(pid, type, p) {
799
		send_sigio_to_task(p, fown, fd, band, group);
800
	} while_each_pid_task(pid, type, p);
Linus Torvalds's avatar
Linus Torvalds committed
801 802 803 804 805 806
	read_unlock(&tasklist_lock);
 out_unlock_fown:
	read_unlock(&fown->lock);
}

static void send_sigurg_to_task(struct task_struct *p,
807
				struct fown_struct *fown, int group)
Linus Torvalds's avatar
Linus Torvalds committed
808 809
{
	if (sigio_perm(p, fown, SIGURG))
810
		do_send_sig_info(SIGURG, SEND_SIG_PRIV, p, group);
Linus Torvalds's avatar
Linus Torvalds committed
811 812 813 814 815
}

int send_sigurg(struct fown_struct *fown)
{
	struct task_struct *p;
816 817
	enum pid_type type;
	struct pid *pid;
818
	int group = 1;
819
	int ret = 0;
Linus Torvalds's avatar
Linus Torvalds committed
820 821
	
	read_lock(&fown->lock);
822

823
	type = fown->pid_type;
824 825 826 827 828
	if (type == PIDTYPE_MAX) {
		group = 0;
		type = PIDTYPE_PID;
	}

Linus Torvalds's avatar
Linus Torvalds committed
829 830 831 832 833 834 835
	pid = fown->pid;
	if (!pid)
		goto out_unlock_fown;

	ret = 1;
	
	read_lock(&tasklist_lock);
836
	do_each_pid_task(pid, type, p) {
837
		send_sigurg_to_task(p, fown, group);
838
	} while_each_pid_task(pid, type, p);
Linus Torvalds's avatar
Linus Torvalds committed
839 840 841 842 843 844
	read_unlock(&tasklist_lock);
 out_unlock_fown:
	read_unlock(&fown->lock);
	return ret;
}

845
static DEFINE_SPINLOCK(fasync_lock);
846
static struct kmem_cache *fasync_cache __read_mostly;
Linus Torvalds's avatar
Linus Torvalds committed
847

848 849 850 851 852 853
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
854
/*
855 856 857 858 859 860 861
 * 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
862
 */
863
int fasync_remove_entry(struct file *filp, struct fasync_struct **fapp)
Linus Torvalds's avatar
Linus Torvalds committed
864 865 866 867
{
	struct fasync_struct *fa, **fp;
	int result = 0;

868
	spin_lock(&filp->f_lock);
869
	spin_lock(&fasync_lock);
870 871 872
	for (fp = fapp; (fa = *fp) != NULL; fp = &fa->fa_next) {
		if (fa->fa_file != filp)
			continue;
873

874
		write_lock_irq(&fa->fa_lock);
875
		fa->fa_file = NULL;
876
		write_unlock_irq(&fa->fa_lock);
877

878
		*fp = fa->fa_next;
879
		call_rcu(&fa->fa_rcu, fasync_free_rcu);
880 881 882
		filp->f_flags &= ~FASYNC;
		result = 1;
		break;
Linus Torvalds's avatar
Linus Torvalds committed
883
	}
884
	spin_unlock(&fasync_lock);
885 886 887 888
	spin_unlock(&filp->f_lock);
	return result;
}

889 890 891 892 893
struct fasync_struct *fasync_alloc(void)
{
	return kmem_cache_alloc(fasync_cache, GFP_KERNEL);
}

894
/*
895 896 897
 * 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.
898
 */
899
void fasync_free(struct fasync_struct *new)
900
{
901 902
	kmem_cache_free(fasync_cache, new);
}
903

904 905 906
/*
 * Insert a new entry into the fasync list.  Return the pointer to the
 * old one if we didn't use the new one.
907 908 909
 *
 * NOTE! It is very important that the FASYNC flag always
 * match the state "is the filp on a fasync list".
910 911 912 913
 */
struct fasync_struct *fasync_insert_entry(int fd, struct file *filp, struct fasync_struct **fapp, struct fasync_struct *new)
{
        struct fasync_struct *fa, **fp;
914 915

	spin_lock(&filp->f_lock);
916
	spin_lock(&fasync_lock);
Linus Torvalds's avatar
Linus Torvalds committed
917
	for (fp = fapp; (fa = *fp) != NULL; fp = &fa->fa_next) {
918 919
		if (fa->fa_file != filp)
			continue;
920

921
		write_lock_irq(&fa->fa_lock);
922
		fa->fa_fd = fd;
923
		write_unlock_irq(&fa->fa_lock);
924
		goto out;
Linus Torvalds's avatar
Linus Torvalds committed
925 926
	}

927
	rwlock_init(&new->fa_lock);
928 929 930 931
	new->magic = FASYNC_MAGIC;
	new->fa_file = filp;
	new->fa_fd = fd;
	new->fa_next = *fapp;
932
	rcu_assign_pointer(*fapp, new);
933 934
	filp->f_flags |= FASYNC;

Linus Torvalds's avatar
Linus Torvalds committed
935
out:
936
	spin_unlock(&fasync_lock);
937
	spin_unlock(&filp->f_lock);
938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965
	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
966 967
}

968 969 970 971 972 973 974 975 976 977 978 979 980
/*
 * 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
981 982
EXPORT_SYMBOL(fasync_helper);

983 984 985 986
/*
 * 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
987 988
{
	while (fa) {
989
		struct fown_struct *fown;
990

Linus Torvalds's avatar
Linus Torvalds committed
991 992 993 994 995
		if (fa->magic != FASYNC_MAGIC) {
			printk(KERN_ERR "kill_fasync: bad magic number in "
			       "fasync_struct!\n");
			return;
		}
996
		read_lock(&fa->fa_lock);
997 998 999 1000 1001 1002 1003 1004
		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);
		}
1005
		read_unlock(&fa->fa_lock);
1006
		fa = rcu_dereference(fa->fa_next);
Linus Torvalds's avatar
Linus Torvalds committed
1007 1008 1009 1010 1011 1012 1013 1014 1015
	}
}

void kill_fasync(struct fasync_struct **fp, int sig, int band)
{
	/* First a quick test without locking: usually
	 * the list is empty.
	 */
	if (*fp) {
1016 1017 1018
		rcu_read_lock();
		kill_fasync_rcu(rcu_dereference(*fp), sig, band);
		rcu_read_unlock();
Linus Torvalds's avatar
Linus Torvalds committed
1019 1020 1021 1022
	}
}
EXPORT_SYMBOL(kill_fasync);

1023
static int __init fcntl_init(void)
Linus Torvalds's avatar
Linus Torvalds committed
1024
{
1025 1026 1027 1028 1029
	/*
	 * 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.
	 */
1030 1031 1032 1033
	BUILD_BUG_ON(21 - 1 /* for O_RDONLY being 0 */ !=
		HWEIGHT32(
			(VALID_OPEN_FLAGS & ~(O_NONBLOCK | O_NDELAY)) |
			__FMODE_EXEC | __FMODE_NONOTIFY));
1034

Linus Torvalds's avatar
Linus Torvalds committed
1035
	fasync_cache = kmem_cache_create("fasync_cache",
1036
		sizeof(struct fasync_struct), 0, SLAB_PANIC, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
1037 1038 1039
	return 0;
}

1040
module_init(fcntl_init)