af_unix.c 67.0 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3
/*
 * NET4:	Implementation of BSD Unix domain sockets.
 *
4
 * Authors:	Alan Cox, <alan@lxorguk.ukuu.org.uk>
L
Linus Torvalds 已提交
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82
 *
 *		This program is free software; you can redistribute it and/or
 *		modify it under the terms of the GNU General Public License
 *		as published by the Free Software Foundation; either version
 *		2 of the License, or (at your option) any later version.
 *
 * Fixes:
 *		Linus Torvalds	:	Assorted bug cures.
 *		Niibe Yutaka	:	async I/O support.
 *		Carsten Paeth	:	PF_UNIX check, address fixes.
 *		Alan Cox	:	Limit size of allocated blocks.
 *		Alan Cox	:	Fixed the stupid socketpair bug.
 *		Alan Cox	:	BSD compatibility fine tuning.
 *		Alan Cox	:	Fixed a bug in connect when interrupted.
 *		Alan Cox	:	Sorted out a proper draft version of
 *					file descriptor passing hacked up from
 *					Mike Shaver's work.
 *		Marty Leisner	:	Fixes to fd passing
 *		Nick Nevin	:	recvmsg bugfix.
 *		Alan Cox	:	Started proper garbage collector
 *		Heiko EiBfeldt	:	Missing verify_area check
 *		Alan Cox	:	Started POSIXisms
 *		Andreas Schwab	:	Replace inode by dentry for proper
 *					reference counting
 *		Kirk Petersen	:	Made this a module
 *	    Christoph Rohland	:	Elegant non-blocking accept/connect algorithm.
 *					Lots of bug fixes.
 *	     Alexey Kuznetosv	:	Repaired (I hope) bugs introduces
 *					by above two patches.
 *	     Andrea Arcangeli	:	If possible we block in connect(2)
 *					if the max backlog of the listen socket
 *					is been reached. This won't break
 *					old apps and it will avoid huge amount
 *					of socks hashed (this for unix_gc()
 *					performances reasons).
 *					Security fix that limits the max
 *					number of socks to 2*max_files and
 *					the number of skb queueable in the
 *					dgram receiver.
 *		Artur Skawina   :	Hash function optimizations
 *	     Alexey Kuznetsov   :	Full scale SMP. Lot of bugs are introduced 8)
 *	      Malcolm Beattie   :	Set peercred for socketpair
 *	     Michal Ostrowski   :       Module initialization cleanup.
 *	     Arnaldo C. Melo	:	Remove MOD_{INC,DEC}_USE_COUNT,
 *	     				the core infrastructure is doing that
 *	     				for all net proto families now (2.5.69+)
 *
 *
 * Known differences from reference BSD that was tested:
 *
 *	[TO FIX]
 *	ECONNREFUSED is not returned from one end of a connected() socket to the
 *		other the moment one end closes.
 *	fstat() doesn't return st_dev=0, and give the blksize as high water mark
 *		and a fake inode identifier (nor the BSD first socket fstat twice bug).
 *	[NOT TO FIX]
 *	accept() returns a path name even if the connecting socket has closed
 *		in the meantime (BSD loses the path and gives up).
 *	accept() returns 0 length path for an unbound connector. BSD returns 16
 *		and a null first byte in the path (but not for gethost/peername - BSD bug ??)
 *	socketpair(...SOCK_RAW..) doesn't panic the kernel.
 *	BSD af_unix apparently has connect forgetting to block properly.
 *		(need to check this with the POSIX spec in detail)
 *
 * Differences from 2.0.0-11-... (ANK)
 *	Bug fixes and improvements.
 *		- client shutdown killed server socket.
 *		- removed all useless cli/sti pairs.
 *
 *	Semantic changes/extensions.
 *		- generic control message passing.
 *		- SCM_CREDENTIALS control message.
 *		- "Abstract" (not FS based) socket bindings.
 *		  Abstract names are sequences of bytes (not zero terminated)
 *		  started by 0, so that this name space does not intersect
 *		  with BSD names.
 */

W
wangweidong 已提交
83 84
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

L
Linus Torvalds 已提交
85 86 87
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/signal.h>
88
#include <linux/sched/signal.h>
L
Linus Torvalds 已提交
89 90 91 92 93 94 95 96 97 98 99 100 101 102
#include <linux/errno.h>
#include <linux/string.h>
#include <linux/stat.h>
#include <linux/dcache.h>
#include <linux/namei.h>
#include <linux/socket.h>
#include <linux/un.h>
#include <linux/fcntl.h>
#include <linux/termios.h>
#include <linux/sockios.h>
#include <linux/net.h>
#include <linux/in.h>
#include <linux/fs.h>
#include <linux/slab.h>
103
#include <linux/uaccess.h>
L
Linus Torvalds 已提交
104 105
#include <linux/skbuff.h>
#include <linux/netdevice.h>
106
#include <net/net_namespace.h>
L
Linus Torvalds 已提交
107
#include <net/sock.h>
108
#include <net/tcp_states.h>
L
Linus Torvalds 已提交
109 110 111 112 113 114 115 116 117 118
#include <net/af_unix.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <net/scm.h>
#include <linux/init.h>
#include <linux/poll.h>
#include <linux/rtnetlink.h>
#include <linux/mount.h>
#include <net/checksum.h>
#include <linux/security.h>
119
#include <linux/freezer.h>
120
#include <linux/file.h>
L
Linus Torvalds 已提交
121

E
Eric Dumazet 已提交
122
struct hlist_head unix_socket_table[2 * UNIX_HASH_SIZE];
123 124 125
EXPORT_SYMBOL_GPL(unix_socket_table);
DEFINE_SPINLOCK(unix_table_lock);
EXPORT_SYMBOL_GPL(unix_table_lock);
E
Eric Dumazet 已提交
126
static atomic_long_t unix_nr_socks;
L
Linus Torvalds 已提交
127 128


E
Eric Dumazet 已提交
129 130 131 132 133 134 135 136 137 138 139
static struct hlist_head *unix_sockets_unbound(void *addr)
{
	unsigned long hash = (unsigned long)addr;

	hash ^= hash >> 16;
	hash ^= hash >> 8;
	hash %= UNIX_HASH_SIZE;
	return &unix_socket_table[UNIX_HASH_SIZE + hash];
}

#define UNIX_ABSTRACT(sk)	(unix_sk(sk)->addr->hash < UNIX_HASH_SIZE)
L
Linus Torvalds 已提交
140

C
Catherine Zhang 已提交
141
#ifdef CONFIG_SECURITY_NETWORK
142
static void unix_get_secdata(struct scm_cookie *scm, struct sk_buff *skb)
C
Catherine Zhang 已提交
143
{
144
	UNIXCB(skb).secid = scm->secid;
C
Catherine Zhang 已提交
145 146 147 148
}

static inline void unix_set_secdata(struct scm_cookie *scm, struct sk_buff *skb)
{
149 150 151 152 153 154
	scm->secid = UNIXCB(skb).secid;
}

static inline bool unix_secdata_eq(struct scm_cookie *scm, struct sk_buff *skb)
{
	return (scm->secid == UNIXCB(skb).secid);
C
Catherine Zhang 已提交
155 156
}
#else
157
static inline void unix_get_secdata(struct scm_cookie *scm, struct sk_buff *skb)
C
Catherine Zhang 已提交
158 159 160 161
{ }

static inline void unix_set_secdata(struct scm_cookie *scm, struct sk_buff *skb)
{ }
162 163 164 165 166

static inline bool unix_secdata_eq(struct scm_cookie *scm, struct sk_buff *skb)
{
	return true;
}
C
Catherine Zhang 已提交
167 168
#endif /* CONFIG_SECURITY_NETWORK */

L
Linus Torvalds 已提交
169 170
/*
 *  SMP locking strategy:
171
 *    hash table is protected with spinlock unix_table_lock
172
 *    each socket state is protected by separate spin lock.
L
Linus Torvalds 已提交
173 174
 */

175
static inline unsigned int unix_hash_fold(__wsum n)
L
Linus Torvalds 已提交
176
{
177
	unsigned int hash = (__force unsigned int)csum_fold(n);
178

L
Linus Torvalds 已提交
179 180 181 182 183 184 185 186 187 188 189 190 191
	hash ^= hash>>8;
	return hash&(UNIX_HASH_SIZE-1);
}

#define unix_peer(sk) (unix_sk(sk)->peer)

static inline int unix_our_peer(struct sock *sk, struct sock *osk)
{
	return unix_peer(osk) == sk;
}

static inline int unix_may_send(struct sock *sk, struct sock *osk)
{
E
Eric Dumazet 已提交
192
	return unix_peer(osk) == NULL || unix_our_peer(sk, osk);
L
Linus Torvalds 已提交
193 194
}

195 196 197 198 199
static inline int unix_recvq_full(struct sock const *sk)
{
	return skb_queue_len(&sk->sk_receive_queue) > sk->sk_max_ack_backlog;
}

200
struct sock *unix_peer_get(struct sock *s)
L
Linus Torvalds 已提交
201 202 203
{
	struct sock *peer;

204
	unix_state_lock(s);
L
Linus Torvalds 已提交
205 206 207
	peer = unix_peer(s);
	if (peer)
		sock_hold(peer);
208
	unix_state_unlock(s);
L
Linus Torvalds 已提交
209 210
	return peer;
}
211
EXPORT_SYMBOL_GPL(unix_peer_get);
L
Linus Torvalds 已提交
212 213 214

static inline void unix_release_addr(struct unix_address *addr)
{
215
	if (refcount_dec_and_test(&addr->refcnt))
L
Linus Torvalds 已提交
216 217 218 219 220 221 222 223 224
		kfree(addr);
}

/*
 *	Check unix socket name:
 *		- should be not zero length.
 *	        - if started by not zero, should be NULL terminated (FS object)
 *		- if started by zero, it is abstract name.
 */
225

226
static int unix_mkname(struct sockaddr_un *sunaddr, int len, unsigned int *hashp)
L
Linus Torvalds 已提交
227 228 229 230 231 232 233 234 235
{
	if (len <= sizeof(short) || len > sizeof(*sunaddr))
		return -EINVAL;
	if (!sunaddr || sunaddr->sun_family != AF_UNIX)
		return -EINVAL;
	if (sunaddr->sun_path[0]) {
		/*
		 * This may look like an off by one error but it is a bit more
		 * subtle. 108 is the longest valid AF_UNIX path for a binding.
L
Lucas De Marchi 已提交
236
		 * sun_path[108] doesn't as such exist.  However in kernel space
L
Linus Torvalds 已提交
237 238 239
		 * we are guaranteed that it is a valid memory location in our
		 * kernel address buffer.
		 */
240
		((char *)sunaddr)[len] = 0;
L
Linus Torvalds 已提交
241 242 243 244
		len = strlen(sunaddr->sun_path)+1+sizeof(short);
		return len;
	}

245
	*hashp = unix_hash_fold(csum_partial(sunaddr, len, 0));
L
Linus Torvalds 已提交
246 247 248 249 250 251 252 253 254 255
	return len;
}

static void __unix_remove_socket(struct sock *sk)
{
	sk_del_node_init(sk);
}

static void __unix_insert_socket(struct hlist_head *list, struct sock *sk)
{
256
	WARN_ON(!sk_unhashed(sk));
L
Linus Torvalds 已提交
257 258 259 260 261
	sk_add_node(sk, list);
}

static inline void unix_remove_socket(struct sock *sk)
{
262
	spin_lock(&unix_table_lock);
L
Linus Torvalds 已提交
263
	__unix_remove_socket(sk);
264
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
265 266 267 268
}

static inline void unix_insert_socket(struct hlist_head *list, struct sock *sk)
{
269
	spin_lock(&unix_table_lock);
L
Linus Torvalds 已提交
270
	__unix_insert_socket(list, sk);
271
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
272 273
}

274 275
static struct sock *__unix_find_socket_byname(struct net *net,
					      struct sockaddr_un *sunname,
276
					      int len, int type, unsigned int hash)
L
Linus Torvalds 已提交
277 278 279
{
	struct sock *s;

280
	sk_for_each(s, &unix_socket_table[hash ^ type]) {
L
Linus Torvalds 已提交
281 282
		struct unix_sock *u = unix_sk(s);

283
		if (!net_eq(sock_net(s), net))
284 285
			continue;

L
Linus Torvalds 已提交
286 287 288 289 290 291 292 293 294
		if (u->addr->len == len &&
		    !memcmp(u->addr->name, sunname, len))
			goto found;
	}
	s = NULL;
found:
	return s;
}

295 296
static inline struct sock *unix_find_socket_byname(struct net *net,
						   struct sockaddr_un *sunname,
L
Linus Torvalds 已提交
297
						   int len, int type,
298
						   unsigned int hash)
L
Linus Torvalds 已提交
299 300 301
{
	struct sock *s;

302
	spin_lock(&unix_table_lock);
303
	s = __unix_find_socket_byname(net, sunname, len, type, hash);
L
Linus Torvalds 已提交
304 305
	if (s)
		sock_hold(s);
306
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
307 308 309
	return s;
}

310
static struct sock *unix_find_socket_byinode(struct inode *i)
L
Linus Torvalds 已提交
311 312 313
{
	struct sock *s;

314
	spin_lock(&unix_table_lock);
315
	sk_for_each(s,
L
Linus Torvalds 已提交
316
		    &unix_socket_table[i->i_ino & (UNIX_HASH_SIZE - 1)]) {
A
Al Viro 已提交
317
		struct dentry *dentry = unix_sk(s)->path.dentry;
L
Linus Torvalds 已提交
318

319
		if (dentry && d_backing_inode(dentry) == i) {
L
Linus Torvalds 已提交
320 321 322 323 324 325
			sock_hold(s);
			goto found;
		}
	}
	s = NULL;
found:
326
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
327 328 329
	return s;
}

330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345
/* Support code for asymmetrically connected dgram sockets
 *
 * If a datagram socket is connected to a socket not itself connected
 * to the first socket (eg, /dev/log), clients may only enqueue more
 * messages if the present receive queue of the server socket is not
 * "too large". This means there's a second writeability condition
 * poll and sendmsg need to test. The dgram recv code will do a wake
 * up on the peer_wait wait queue of a socket upon reception of a
 * datagram which needs to be propagated to sleeping would-be writers
 * since these might not have sent anything so far. This can't be
 * accomplished via poll_wait because the lifetime of the server
 * socket might be less than that of its clients if these break their
 * association with it or if the server socket is closed while clients
 * are still connected to it and there's no way to inform "a polling
 * implementation" that it should let go of a certain wait queue
 *
346
 * In order to propagate a wake up, a wait_queue_entry_t of the client
347 348 349 350 351 352 353 354
 * socket is enqueued on the peer_wait queue of the server socket
 * whose wake function does a wake_up on the ordinary client socket
 * wait queue. This connection is established whenever a write (or
 * poll for write) hit the flow control condition and broken when the
 * association to the server socket is dissolved or after a wake up
 * was relayed.
 */

355
static int unix_dgram_peer_wake_relay(wait_queue_entry_t *q, unsigned mode, int flags,
356 357 358 359 360 361 362 363 364 365 366 367 368 369
				      void *key)
{
	struct unix_sock *u;
	wait_queue_head_t *u_sleep;

	u = container_of(q, struct unix_sock, peer_wake);

	__remove_wait_queue(&unix_sk(u->peer_wake.private)->peer_wait,
			    q);
	u->peer_wake.private = NULL;

	/* relaying can only happen while the wq still exists */
	u_sleep = sk_sleep(&u->sk);
	if (u_sleep)
A
Al Viro 已提交
370
		wake_up_interruptible_poll(u_sleep, key_to_poll(key));
371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417

	return 0;
}

static int unix_dgram_peer_wake_connect(struct sock *sk, struct sock *other)
{
	struct unix_sock *u, *u_other;
	int rc;

	u = unix_sk(sk);
	u_other = unix_sk(other);
	rc = 0;
	spin_lock(&u_other->peer_wait.lock);

	if (!u->peer_wake.private) {
		u->peer_wake.private = other;
		__add_wait_queue(&u_other->peer_wait, &u->peer_wake);

		rc = 1;
	}

	spin_unlock(&u_other->peer_wait.lock);
	return rc;
}

static void unix_dgram_peer_wake_disconnect(struct sock *sk,
					    struct sock *other)
{
	struct unix_sock *u, *u_other;

	u = unix_sk(sk);
	u_other = unix_sk(other);
	spin_lock(&u_other->peer_wait.lock);

	if (u->peer_wake.private == other) {
		__remove_wait_queue(&u_other->peer_wait, &u->peer_wake);
		u->peer_wake.private = NULL;
	}

	spin_unlock(&u_other->peer_wait.lock);
}

static void unix_dgram_peer_wake_disconnect_wakeup(struct sock *sk,
						   struct sock *other)
{
	unix_dgram_peer_wake_disconnect(sk, other);
	wake_up_interruptible_poll(sk_sleep(sk),
418 419 420
				   EPOLLOUT |
				   EPOLLWRNORM |
				   EPOLLWRBAND);
421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441
}

/* preconditions:
 *	- unix_peer(sk) == other
 *	- association is stable
 */
static int unix_dgram_peer_wake_me(struct sock *sk, struct sock *other)
{
	int connected;

	connected = unix_dgram_peer_wake_connect(sk, other);

	if (unix_recvq_full(other))
		return 1;

	if (connected)
		unix_dgram_peer_wake_disconnect(sk, other);

	return 0;
}

442
static int unix_writable(const struct sock *sk)
L
Linus Torvalds 已提交
443
{
444
	return sk->sk_state != TCP_LISTEN &&
445
	       (refcount_read(&sk->sk_wmem_alloc) << 2) <= sk->sk_sndbuf;
L
Linus Torvalds 已提交
446 447 448 449
}

static void unix_write_space(struct sock *sk)
{
450 451 452
	struct socket_wq *wq;

	rcu_read_lock();
L
Linus Torvalds 已提交
453
	if (unix_writable(sk)) {
454
		wq = rcu_dereference(sk->sk_wq);
H
Herbert Xu 已提交
455
		if (skwq_has_sleeper(wq))
E
Eric Dumazet 已提交
456
			wake_up_interruptible_sync_poll(&wq->wait,
457
				EPOLLOUT | EPOLLWRNORM | EPOLLWRBAND);
458
		sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
L
Linus Torvalds 已提交
459
	}
460
	rcu_read_unlock();
L
Linus Torvalds 已提交
461 462 463 464 465 466 467 468
}

/* When dgram socket disconnects (or changes its peer), we clear its receive
 * queue of packets arrived from previous peer. First, it allows to do
 * flow control based only on wmem_alloc; second, sk connected to peer
 * may receive messages only from that peer. */
static void unix_dgram_disconnected(struct sock *sk, struct sock *other)
{
469
	if (!skb_queue_empty(&sk->sk_receive_queue)) {
L
Linus Torvalds 已提交
470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489
		skb_queue_purge(&sk->sk_receive_queue);
		wake_up_interruptible_all(&unix_sk(sk)->peer_wait);

		/* If one link of bidirectional dgram pipe is disconnected,
		 * we signal error. Messages are lost. Do not make this,
		 * when peer was not connected to us.
		 */
		if (!sock_flag(other, SOCK_DEAD) && unix_peer(other) == sk) {
			other->sk_err = ECONNRESET;
			other->sk_error_report(other);
		}
	}
}

static void unix_sock_destructor(struct sock *sk)
{
	struct unix_sock *u = unix_sk(sk);

	skb_queue_purge(&sk->sk_receive_queue);

490
	WARN_ON(refcount_read(&sk->sk_wmem_alloc));
491 492
	WARN_ON(!sk_unhashed(sk));
	WARN_ON(sk->sk_socket);
L
Linus Torvalds 已提交
493
	if (!sock_flag(sk, SOCK_DEAD)) {
W
wangweidong 已提交
494
		pr_info("Attempt to release alive unix socket: %p\n", sk);
L
Linus Torvalds 已提交
495 496 497 498 499 500
		return;
	}

	if (u->addr)
		unix_release_addr(u->addr);

E
Eric Dumazet 已提交
501
	atomic_long_dec(&unix_nr_socks);
502
	local_bh_disable();
503
	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
504
	local_bh_enable();
L
Linus Torvalds 已提交
505
#ifdef UNIX_REFCNT_DEBUG
W
wangweidong 已提交
506
	pr_debug("UNIX %p is destroyed, %ld are still alive.\n", sk,
E
Eric Dumazet 已提交
507
		atomic_long_read(&unix_nr_socks));
L
Linus Torvalds 已提交
508 509 510
#endif
}

511
static void unix_release_sock(struct sock *sk, int embrion)
L
Linus Torvalds 已提交
512 513
{
	struct unix_sock *u = unix_sk(sk);
A
Al Viro 已提交
514
	struct path path;
L
Linus Torvalds 已提交
515 516 517 518 519 520 521
	struct sock *skpair;
	struct sk_buff *skb;
	int state;

	unix_remove_socket(sk);

	/* Clear state */
522
	unix_state_lock(sk);
L
Linus Torvalds 已提交
523 524
	sock_orphan(sk);
	sk->sk_shutdown = SHUTDOWN_MASK;
A
Al Viro 已提交
525 526 527
	path	     = u->path;
	u->path.dentry = NULL;
	u->path.mnt = NULL;
L
Linus Torvalds 已提交
528 529
	state = sk->sk_state;
	sk->sk_state = TCP_CLOSE;
530
	unix_state_unlock(sk);
L
Linus Torvalds 已提交
531 532 533

	wake_up_interruptible_all(&u->peer_wait);

534
	skpair = unix_peer(sk);
L
Linus Torvalds 已提交
535

536
	if (skpair != NULL) {
L
Linus Torvalds 已提交
537
		if (sk->sk_type == SOCK_STREAM || sk->sk_type == SOCK_SEQPACKET) {
538
			unix_state_lock(skpair);
L
Linus Torvalds 已提交
539 540 541 542
			/* No more writes */
			skpair->sk_shutdown = SHUTDOWN_MASK;
			if (!skb_queue_empty(&sk->sk_receive_queue) || embrion)
				skpair->sk_err = ECONNRESET;
543
			unix_state_unlock(skpair);
L
Linus Torvalds 已提交
544
			skpair->sk_state_change(skpair);
545
			sk_wake_async(skpair, SOCK_WAKE_WAITD, POLL_HUP);
L
Linus Torvalds 已提交
546
		}
547 548

		unix_dgram_peer_wake_disconnect(sk, skpair);
L
Linus Torvalds 已提交
549 550 551 552 553 554 555
		sock_put(skpair); /* It may now die */
		unix_peer(sk) = NULL;
	}

	/* Try to flush out this socket. Throw out buffers at least */

	while ((skb = skb_dequeue(&sk->sk_receive_queue)) != NULL) {
556
		if (state == TCP_LISTEN)
L
Linus Torvalds 已提交
557 558
			unix_release_sock(skb->sk, 1);
		/* passed fds are erased in the kfree_skb hook	      */
559
		UNIXCB(skb).consumed = skb->len;
L
Linus Torvalds 已提交
560 561 562
		kfree_skb(skb);
	}

A
Al Viro 已提交
563 564
	if (path.dentry)
		path_put(&path);
L
Linus Torvalds 已提交
565 566 567 568 569 570

	sock_put(sk);

	/* ---- Socket is dead now and most probably destroyed ---- */

	/*
A
Alan Cox 已提交
571
	 * Fixme: BSD difference: In BSD all sockets connected to us get
L
Linus Torvalds 已提交
572 573 574 575 576 577 578 579 580
	 *	  ECONNRESET and we die on the spot. In Linux we behave
	 *	  like files and pipes do and wait for the last
	 *	  dereference.
	 *
	 * Can't we simply set sock->err?
	 *
	 *	  What the above comment does talk about? --ANK(980817)
	 */

581
	if (unix_tot_inflight)
582
		unix_gc();		/* Garbage collect fds */
L
Linus Torvalds 已提交
583 584
}

585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602
static void init_peercred(struct sock *sk)
{
	put_pid(sk->sk_peer_pid);
	if (sk->sk_peer_cred)
		put_cred(sk->sk_peer_cred);
	sk->sk_peer_pid  = get_pid(task_tgid(current));
	sk->sk_peer_cred = get_current_cred();
}

static void copy_peercred(struct sock *sk, struct sock *peersk)
{
	put_pid(sk->sk_peer_pid);
	if (sk->sk_peer_cred)
		put_cred(sk->sk_peer_cred);
	sk->sk_peer_pid  = get_pid(peersk->sk_peer_pid);
	sk->sk_peer_cred = get_cred(peersk->sk_peer_cred);
}

L
Linus Torvalds 已提交
603 604 605 606 607
static int unix_listen(struct socket *sock, int backlog)
{
	int err;
	struct sock *sk = sock->sk;
	struct unix_sock *u = unix_sk(sk);
608
	struct pid *old_pid = NULL;
L
Linus Torvalds 已提交
609 610

	err = -EOPNOTSUPP;
E
Eric Dumazet 已提交
611 612
	if (sock->type != SOCK_STREAM && sock->type != SOCK_SEQPACKET)
		goto out;	/* Only stream/seqpacket sockets accept */
L
Linus Torvalds 已提交
613 614
	err = -EINVAL;
	if (!u->addr)
E
Eric Dumazet 已提交
615
		goto out;	/* No listens on an unbound socket */
616
	unix_state_lock(sk);
L
Linus Torvalds 已提交
617 618 619 620 621 622 623
	if (sk->sk_state != TCP_CLOSE && sk->sk_state != TCP_LISTEN)
		goto out_unlock;
	if (backlog > sk->sk_max_ack_backlog)
		wake_up_interruptible_all(&u->peer_wait);
	sk->sk_max_ack_backlog	= backlog;
	sk->sk_state		= TCP_LISTEN;
	/* set credentials so connect can copy them */
624
	init_peercred(sk);
L
Linus Torvalds 已提交
625 626 627
	err = 0;

out_unlock:
628
	unix_state_unlock(sk);
629
	put_pid(old_pid);
L
Linus Torvalds 已提交
630 631 632 633 634 635 636 637 638
out:
	return err;
}

static int unix_release(struct socket *);
static int unix_bind(struct socket *, struct sockaddr *, int);
static int unix_stream_connect(struct socket *, struct sockaddr *,
			       int addr_len, int flags);
static int unix_socketpair(struct socket *, struct socket *);
639
static int unix_accept(struct socket *, struct socket *, int, bool);
640
static int unix_getname(struct socket *, struct sockaddr *, int);
641 642
static __poll_t unix_poll_mask(struct socket *, __poll_t);
static __poll_t unix_dgram_poll_mask(struct socket *, __poll_t);
L
Linus Torvalds 已提交
643 644
static int unix_ioctl(struct socket *, unsigned int, unsigned long);
static int unix_shutdown(struct socket *, int);
645 646
static int unix_stream_sendmsg(struct socket *, struct msghdr *, size_t);
static int unix_stream_recvmsg(struct socket *, struct msghdr *, size_t, int);
647 648
static ssize_t unix_stream_sendpage(struct socket *, struct page *, int offset,
				    size_t size, int flags);
649 650 651
static ssize_t unix_stream_splice_read(struct socket *,  loff_t *ppos,
				       struct pipe_inode_info *, size_t size,
				       unsigned int flags);
652 653
static int unix_dgram_sendmsg(struct socket *, struct msghdr *, size_t);
static int unix_dgram_recvmsg(struct socket *, struct msghdr *, size_t, int);
L
Linus Torvalds 已提交
654 655
static int unix_dgram_connect(struct socket *, struct sockaddr *,
			      int, int);
656 657 658
static int unix_seqpacket_sendmsg(struct socket *, struct msghdr *, size_t);
static int unix_seqpacket_recvmsg(struct socket *, struct msghdr *, size_t,
				  int);
L
Linus Torvalds 已提交
659

660
static int unix_set_peek_off(struct sock *sk, int val)
661 662 663
{
	struct unix_sock *u = unix_sk(sk);

664
	if (mutex_lock_interruptible(&u->iolock))
665 666
		return -EINTR;

667
	sk->sk_peek_off = val;
668
	mutex_unlock(&u->iolock);
669 670

	return 0;
671 672 673
}


674
static const struct proto_ops unix_stream_ops = {
L
Linus Torvalds 已提交
675 676 677 678 679 680 681 682
	.family =	PF_UNIX,
	.owner =	THIS_MODULE,
	.release =	unix_release,
	.bind =		unix_bind,
	.connect =	unix_stream_connect,
	.socketpair =	unix_socketpair,
	.accept =	unix_accept,
	.getname =	unix_getname,
683
	.poll_mask =	unix_poll_mask,
L
Linus Torvalds 已提交
684 685 686 687 688 689 690 691
	.ioctl =	unix_ioctl,
	.listen =	unix_listen,
	.shutdown =	unix_shutdown,
	.setsockopt =	sock_no_setsockopt,
	.getsockopt =	sock_no_getsockopt,
	.sendmsg =	unix_stream_sendmsg,
	.recvmsg =	unix_stream_recvmsg,
	.mmap =		sock_no_mmap,
692
	.sendpage =	unix_stream_sendpage,
693
	.splice_read =	unix_stream_splice_read,
694
	.set_peek_off =	unix_set_peek_off,
L
Linus Torvalds 已提交
695 696
};

697
static const struct proto_ops unix_dgram_ops = {
L
Linus Torvalds 已提交
698 699 700 701 702 703 704 705
	.family =	PF_UNIX,
	.owner =	THIS_MODULE,
	.release =	unix_release,
	.bind =		unix_bind,
	.connect =	unix_dgram_connect,
	.socketpair =	unix_socketpair,
	.accept =	sock_no_accept,
	.getname =	unix_getname,
706
	.poll_mask =	unix_dgram_poll_mask,
L
Linus Torvalds 已提交
707 708 709 710 711 712 713 714 715
	.ioctl =	unix_ioctl,
	.listen =	sock_no_listen,
	.shutdown =	unix_shutdown,
	.setsockopt =	sock_no_setsockopt,
	.getsockopt =	sock_no_getsockopt,
	.sendmsg =	unix_dgram_sendmsg,
	.recvmsg =	unix_dgram_recvmsg,
	.mmap =		sock_no_mmap,
	.sendpage =	sock_no_sendpage,
716
	.set_peek_off =	unix_set_peek_off,
L
Linus Torvalds 已提交
717 718
};

719
static const struct proto_ops unix_seqpacket_ops = {
L
Linus Torvalds 已提交
720 721 722 723 724 725 726 727
	.family =	PF_UNIX,
	.owner =	THIS_MODULE,
	.release =	unix_release,
	.bind =		unix_bind,
	.connect =	unix_stream_connect,
	.socketpair =	unix_socketpair,
	.accept =	unix_accept,
	.getname =	unix_getname,
728
	.poll_mask =	unix_dgram_poll_mask,
L
Linus Torvalds 已提交
729 730 731 732 733 734
	.ioctl =	unix_ioctl,
	.listen =	unix_listen,
	.shutdown =	unix_shutdown,
	.setsockopt =	sock_no_setsockopt,
	.getsockopt =	sock_no_getsockopt,
	.sendmsg =	unix_seqpacket_sendmsg,
735
	.recvmsg =	unix_seqpacket_recvmsg,
L
Linus Torvalds 已提交
736 737
	.mmap =		sock_no_mmap,
	.sendpage =	sock_no_sendpage,
738
	.set_peek_off =	unix_set_peek_off,
L
Linus Torvalds 已提交
739 740 741
};

static struct proto unix_proto = {
742 743 744
	.name			= "UNIX",
	.owner			= THIS_MODULE,
	.obj_size		= sizeof(struct unix_sock),
L
Linus Torvalds 已提交
745 746
};

747
static struct sock *unix_create1(struct net *net, struct socket *sock, int kern)
L
Linus Torvalds 已提交
748 749 750 751
{
	struct sock *sk = NULL;
	struct unix_sock *u;

E
Eric Dumazet 已提交
752 753
	atomic_long_inc(&unix_nr_socks);
	if (atomic_long_read(&unix_nr_socks) > 2 * get_max_files())
L
Linus Torvalds 已提交
754 755
		goto out;

756
	sk = sk_alloc(net, PF_UNIX, GFP_KERNEL, &unix_proto, kern);
L
Linus Torvalds 已提交
757 758 759
	if (!sk)
		goto out;

E
Eric Dumazet 已提交
760
	sock_init_data(sock, sk);
L
Linus Torvalds 已提交
761

762
	sk->sk_allocation	= GFP_KERNEL_ACCOUNT;
L
Linus Torvalds 已提交
763
	sk->sk_write_space	= unix_write_space;
764
	sk->sk_max_ack_backlog	= net->unx.sysctl_max_dgram_qlen;
L
Linus Torvalds 已提交
765 766
	sk->sk_destruct		= unix_sock_destructor;
	u	  = unix_sk(sk);
A
Al Viro 已提交
767 768
	u->path.dentry = NULL;
	u->path.mnt = NULL;
769
	spin_lock_init(&u->lock);
A
Al Viro 已提交
770
	atomic_long_set(&u->inflight, 0);
771
	INIT_LIST_HEAD(&u->link);
772 773
	mutex_init(&u->iolock); /* single task reading lock */
	mutex_init(&u->bindlock); /* single task binding lock */
L
Linus Torvalds 已提交
774
	init_waitqueue_head(&u->peer_wait);
775
	init_waitqueue_func_entry(&u->peer_wake, unix_dgram_peer_wake_relay);
E
Eric Dumazet 已提交
776
	unix_insert_socket(unix_sockets_unbound(sk), sk);
L
Linus Torvalds 已提交
777
out:
778
	if (sk == NULL)
E
Eric Dumazet 已提交
779
		atomic_long_dec(&unix_nr_socks);
780 781
	else {
		local_bh_disable();
782
		sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
783 784
		local_bh_enable();
	}
L
Linus Torvalds 已提交
785 786 787
	return sk;
}

788 789
static int unix_create(struct net *net, struct socket *sock, int protocol,
		       int kern)
L
Linus Torvalds 已提交
790 791 792 793 794 795 796 797 798 799 800 801 802 803 804
{
	if (protocol && protocol != PF_UNIX)
		return -EPROTONOSUPPORT;

	sock->state = SS_UNCONNECTED;

	switch (sock->type) {
	case SOCK_STREAM:
		sock->ops = &unix_stream_ops;
		break;
		/*
		 *	Believe it or not BSD has AF_UNIX, SOCK_RAW though
		 *	nothing uses it.
		 */
	case SOCK_RAW:
805
		sock->type = SOCK_DGRAM;
806
		/* fall through */
L
Linus Torvalds 已提交
807 808 809 810 811 812 813 814 815 816
	case SOCK_DGRAM:
		sock->ops = &unix_dgram_ops;
		break;
	case SOCK_SEQPACKET:
		sock->ops = &unix_seqpacket_ops;
		break;
	default:
		return -ESOCKTNOSUPPORT;
	}

817
	return unix_create1(net, sock, kern) ? 0 : -ENOMEM;
L
Linus Torvalds 已提交
818 819 820 821 822 823 824 825 826
}

static int unix_release(struct socket *sock)
{
	struct sock *sk = sock->sk;

	if (!sk)
		return 0;

827
	unix_release_sock(sk, 0);
L
Linus Torvalds 已提交
828 829
	sock->sk = NULL;

830
	return 0;
L
Linus Torvalds 已提交
831 832 833 834 835
}

static int unix_autobind(struct socket *sock)
{
	struct sock *sk = sock->sk;
836
	struct net *net = sock_net(sk);
L
Linus Torvalds 已提交
837 838
	struct unix_sock *u = unix_sk(sk);
	static u32 ordernum = 1;
E
Eric Dumazet 已提交
839
	struct unix_address *addr;
L
Linus Torvalds 已提交
840
	int err;
841
	unsigned int retries = 0;
L
Linus Torvalds 已提交
842

843
	err = mutex_lock_interruptible(&u->bindlock);
844 845
	if (err)
		return err;
L
Linus Torvalds 已提交
846 847 848 849 850 851

	err = 0;
	if (u->addr)
		goto out;

	err = -ENOMEM;
852
	addr = kzalloc(sizeof(*addr) + sizeof(short) + 16, GFP_KERNEL);
L
Linus Torvalds 已提交
853 854 855 856
	if (!addr)
		goto out;

	addr->name->sun_family = AF_UNIX;
857
	refcount_set(&addr->refcnt, 1);
L
Linus Torvalds 已提交
858 859 860

retry:
	addr->len = sprintf(addr->name->sun_path+1, "%05x", ordernum) + 1 + sizeof(short);
861
	addr->hash = unix_hash_fold(csum_partial(addr->name, addr->len, 0));
L
Linus Torvalds 已提交
862

863
	spin_lock(&unix_table_lock);
L
Linus Torvalds 已提交
864 865
	ordernum = (ordernum+1)&0xFFFFF;

866
	if (__unix_find_socket_byname(net, addr->name, addr->len, sock->type,
L
Linus Torvalds 已提交
867
				      addr->hash)) {
868
		spin_unlock(&unix_table_lock);
869 870 871 872 873 874 875 876 877 878 879
		/*
		 * __unix_find_socket_byname() may take long time if many names
		 * are already in use.
		 */
		cond_resched();
		/* Give up if all names seems to be in use. */
		if (retries++ == 0xFFFFF) {
			err = -ENOSPC;
			kfree(addr);
			goto out;
		}
L
Linus Torvalds 已提交
880 881 882 883 884 885 886
		goto retry;
	}
	addr->hash ^= sk->sk_type;

	__unix_remove_socket(sk);
	u->addr = addr;
	__unix_insert_socket(&unix_socket_table[addr->hash], sk);
887
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
888 889
	err = 0;

890
out:	mutex_unlock(&u->bindlock);
L
Linus Torvalds 已提交
891 892 893
	return err;
}

894 895
static struct sock *unix_find_other(struct net *net,
				    struct sockaddr_un *sunname, int len,
896
				    int type, unsigned int hash, int *error)
L
Linus Torvalds 已提交
897 898
{
	struct sock *u;
899
	struct path path;
L
Linus Torvalds 已提交
900
	int err = 0;
901

L
Linus Torvalds 已提交
902
	if (sunname->sun_path[0]) {
903 904
		struct inode *inode;
		err = kern_path(sunname->sun_path, LOOKUP_FOLLOW, &path);
L
Linus Torvalds 已提交
905 906
		if (err)
			goto fail;
907
		inode = d_backing_inode(path.dentry);
908
		err = inode_permission(inode, MAY_WRITE);
L
Linus Torvalds 已提交
909 910 911 912
		if (err)
			goto put_fail;

		err = -ECONNREFUSED;
913
		if (!S_ISSOCK(inode->i_mode))
L
Linus Torvalds 已提交
914
			goto put_fail;
915
		u = unix_find_socket_byinode(inode);
L
Linus Torvalds 已提交
916 917 918 919
		if (!u)
			goto put_fail;

		if (u->sk_type == type)
A
Al Viro 已提交
920
			touch_atime(&path);
L
Linus Torvalds 已提交
921

922
		path_put(&path);
L
Linus Torvalds 已提交
923

924
		err = -EPROTOTYPE;
L
Linus Torvalds 已提交
925 926 927 928 929 930
		if (u->sk_type != type) {
			sock_put(u);
			goto fail;
		}
	} else {
		err = -ECONNREFUSED;
931
		u = unix_find_socket_byname(net, sunname, len, type, hash);
L
Linus Torvalds 已提交
932 933
		if (u) {
			struct dentry *dentry;
A
Al Viro 已提交
934
			dentry = unix_sk(u)->path.dentry;
L
Linus Torvalds 已提交
935
			if (dentry)
A
Al Viro 已提交
936
				touch_atime(&unix_sk(u)->path);
L
Linus Torvalds 已提交
937 938 939 940 941 942
		} else
			goto fail;
	}
	return u;

put_fail:
943
	path_put(&path);
L
Linus Torvalds 已提交
944
fail:
945
	*error = err;
L
Linus Torvalds 已提交
946 947 948
	return NULL;
}

949
static int unix_mknod(const char *sun_path, umode_t mode, struct path *res)
A
Al Viro 已提交
950
{
951 952 953 954 955 956 957 958 959 960 961
	struct dentry *dentry;
	struct path path;
	int err = 0;
	/*
	 * Get the parent directory, calculate the hash for last
	 * component.
	 */
	dentry = kern_path_create(AT_FDCWD, sun_path, &path, 0);
	err = PTR_ERR(dentry);
	if (IS_ERR(dentry))
		return err;
A
Al Viro 已提交
962

963 964 965 966
	/*
	 * All right, let's create it.
	 */
	err = security_path_mknod(&path, dentry, mode, 0);
A
Al Viro 已提交
967
	if (!err) {
968
		err = vfs_mknod(d_inode(path.dentry), dentry, mode, 0);
A
Al Viro 已提交
969
		if (!err) {
970
			res->mnt = mntget(path.mnt);
A
Al Viro 已提交
971 972 973
			res->dentry = dget(dentry);
		}
	}
974
	done_path_create(&path, dentry);
A
Al Viro 已提交
975 976
	return err;
}
L
Linus Torvalds 已提交
977 978 979 980

static int unix_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
{
	struct sock *sk = sock->sk;
981
	struct net *net = sock_net(sk);
L
Linus Torvalds 已提交
982
	struct unix_sock *u = unix_sk(sk);
983
	struct sockaddr_un *sunaddr = (struct sockaddr_un *)uaddr;
984
	char *sun_path = sunaddr->sun_path;
985
	int err;
986
	unsigned int hash;
L
Linus Torvalds 已提交
987 988
	struct unix_address *addr;
	struct hlist_head *list;
K
Kees Cook 已提交
989
	struct path path = { };
L
Linus Torvalds 已提交
990 991

	err = -EINVAL;
992 993
	if (addr_len < offsetofend(struct sockaddr_un, sun_family) ||
	    sunaddr->sun_family != AF_UNIX)
L
Linus Torvalds 已提交
994 995
		goto out;

996
	if (addr_len == sizeof(short)) {
L
Linus Torvalds 已提交
997 998 999 1000 1001 1002 1003 1004 1005
		err = unix_autobind(sock);
		goto out;
	}

	err = unix_mkname(sunaddr, addr_len, &hash);
	if (err < 0)
		goto out;
	addr_len = err;

1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016
	if (sun_path[0]) {
		umode_t mode = S_IFSOCK |
		       (SOCK_INODE(sock)->i_mode & ~current_umask());
		err = unix_mknod(sun_path, mode, &path);
		if (err) {
			if (err == -EEXIST)
				err = -EADDRINUSE;
			goto out;
		}
	}

1017
	err = mutex_lock_interruptible(&u->bindlock);
1018
	if (err)
1019
		goto out_put;
L
Linus Torvalds 已提交
1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032

	err = -EINVAL;
	if (u->addr)
		goto out_up;

	err = -ENOMEM;
	addr = kmalloc(sizeof(*addr)+addr_len, GFP_KERNEL);
	if (!addr)
		goto out_up;

	memcpy(addr->name, sunaddr, addr_len);
	addr->len = addr_len;
	addr->hash = hash ^ sk->sk_type;
1033
	refcount_set(&addr->refcnt, 1);
L
Linus Torvalds 已提交
1034

1035
	if (sun_path[0]) {
L
Linus Torvalds 已提交
1036
		addr->hash = UNIX_HASH_SIZE;
1037
		hash = d_backing_inode(path.dentry)->i_ino & (UNIX_HASH_SIZE - 1);
A
Al Viro 已提交
1038
		spin_lock(&unix_table_lock);
1039
		u->path = path;
A
Al Viro 已提交
1040 1041 1042
		list = &unix_socket_table[hash];
	} else {
		spin_lock(&unix_table_lock);
L
Linus Torvalds 已提交
1043
		err = -EADDRINUSE;
1044
		if (__unix_find_socket_byname(net, sunaddr, addr_len,
L
Linus Torvalds 已提交
1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058
					      sk->sk_type, hash)) {
			unix_release_addr(addr);
			goto out_unlock;
		}

		list = &unix_socket_table[addr->hash];
	}

	err = 0;
	__unix_remove_socket(sk);
	u->addr = addr;
	__unix_insert_socket(list, sk);

out_unlock:
1059
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
1060
out_up:
1061
	mutex_unlock(&u->bindlock);
1062 1063 1064
out_put:
	if (err)
		path_put(&path);
L
Linus Torvalds 已提交
1065 1066 1067 1068
out:
	return err;
}

1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093
static void unix_state_double_lock(struct sock *sk1, struct sock *sk2)
{
	if (unlikely(sk1 == sk2) || !sk2) {
		unix_state_lock(sk1);
		return;
	}
	if (sk1 < sk2) {
		unix_state_lock(sk1);
		unix_state_lock_nested(sk2);
	} else {
		unix_state_lock(sk2);
		unix_state_lock_nested(sk1);
	}
}

static void unix_state_double_unlock(struct sock *sk1, struct sock *sk2)
{
	if (unlikely(sk1 == sk2) || !sk2) {
		unix_state_unlock(sk1);
		return;
	}
	unix_state_unlock(sk1);
	unix_state_unlock(sk2);
}

L
Linus Torvalds 已提交
1094 1095 1096 1097
static int unix_dgram_connect(struct socket *sock, struct sockaddr *addr,
			      int alen, int flags)
{
	struct sock *sk = sock->sk;
1098
	struct net *net = sock_net(sk);
1099
	struct sockaddr_un *sunaddr = (struct sockaddr_un *)addr;
L
Linus Torvalds 已提交
1100
	struct sock *other;
1101
	unsigned int hash;
L
Linus Torvalds 已提交
1102 1103
	int err;

1104 1105 1106 1107
	err = -EINVAL;
	if (alen < offsetofend(struct sockaddr, sa_family))
		goto out;

L
Linus Torvalds 已提交
1108 1109 1110 1111 1112 1113 1114 1115 1116 1117
	if (addr->sa_family != AF_UNSPEC) {
		err = unix_mkname(sunaddr, alen, &hash);
		if (err < 0)
			goto out;
		alen = err;

		if (test_bit(SOCK_PASSCRED, &sock->flags) &&
		    !unix_sk(sk)->addr && (err = unix_autobind(sock)) != 0)
			goto out;

1118
restart:
1119
		other = unix_find_other(net, sunaddr, alen, sock->type, hash, &err);
L
Linus Torvalds 已提交
1120 1121 1122
		if (!other)
			goto out;

1123 1124 1125 1126 1127 1128 1129 1130
		unix_state_double_lock(sk, other);

		/* Apparently VFS overslept socket death. Retry. */
		if (sock_flag(other, SOCK_DEAD)) {
			unix_state_double_unlock(sk, other);
			sock_put(other);
			goto restart;
		}
L
Linus Torvalds 已提交
1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144

		err = -EPERM;
		if (!unix_may_send(sk, other))
			goto out_unlock;

		err = security_unix_may_send(sk->sk_socket, other->sk_socket);
		if (err)
			goto out_unlock;

	} else {
		/*
		 *	1003.1g breaking connected state with AF_UNSPEC
		 */
		other = NULL;
1145
		unix_state_double_lock(sk, other);
L
Linus Torvalds 已提交
1146 1147 1148 1149 1150 1151 1152
	}

	/*
	 * If it was connected, reconnect.
	 */
	if (unix_peer(sk)) {
		struct sock *old_peer = unix_peer(sk);
1153
		unix_peer(sk) = other;
1154 1155
		unix_dgram_peer_wake_disconnect_wakeup(sk, old_peer);

1156
		unix_state_double_unlock(sk, other);
L
Linus Torvalds 已提交
1157 1158 1159 1160 1161

		if (other != old_peer)
			unix_dgram_disconnected(sk, old_peer);
		sock_put(old_peer);
	} else {
1162
		unix_peer(sk) = other;
1163
		unix_state_double_unlock(sk, other);
L
Linus Torvalds 已提交
1164
	}
1165
	return 0;
L
Linus Torvalds 已提交
1166 1167

out_unlock:
1168
	unix_state_double_unlock(sk, other);
L
Linus Torvalds 已提交
1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183
	sock_put(other);
out:
	return err;
}

static long unix_wait_for_peer(struct sock *other, long timeo)
{
	struct unix_sock *u = unix_sk(other);
	int sched;
	DEFINE_WAIT(wait);

	prepare_to_wait_exclusive(&u->peer_wait, &wait, TASK_INTERRUPTIBLE);

	sched = !sock_flag(other, SOCK_DEAD) &&
		!(other->sk_shutdown & RCV_SHUTDOWN) &&
1184
		unix_recvq_full(other);
L
Linus Torvalds 已提交
1185

1186
	unix_state_unlock(other);
L
Linus Torvalds 已提交
1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197

	if (sched)
		timeo = schedule_timeout(timeo);

	finish_wait(&u->peer_wait, &wait);
	return timeo;
}

static int unix_stream_connect(struct socket *sock, struct sockaddr *uaddr,
			       int addr_len, int flags)
{
1198
	struct sockaddr_un *sunaddr = (struct sockaddr_un *)uaddr;
L
Linus Torvalds 已提交
1199
	struct sock *sk = sock->sk;
1200
	struct net *net = sock_net(sk);
L
Linus Torvalds 已提交
1201 1202 1203 1204
	struct unix_sock *u = unix_sk(sk), *newu, *otheru;
	struct sock *newsk = NULL;
	struct sock *other = NULL;
	struct sk_buff *skb = NULL;
1205
	unsigned int hash;
L
Linus Torvalds 已提交
1206 1207 1208 1209 1210 1211 1212 1213 1214
	int st;
	int err;
	long timeo;

	err = unix_mkname(sunaddr, addr_len, &hash);
	if (err < 0)
		goto out;
	addr_len = err;

1215 1216
	if (test_bit(SOCK_PASSCRED, &sock->flags) && !u->addr &&
	    (err = unix_autobind(sock)) != 0)
L
Linus Torvalds 已提交
1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228
		goto out;

	timeo = sock_sndtimeo(sk, flags & O_NONBLOCK);

	/* First of all allocate resources.
	   If we will make it after state is locked,
	   we will have to recheck all again in any case.
	 */

	err = -ENOMEM;

	/* create new sock for complete connection */
1229
	newsk = unix_create1(sock_net(sk), NULL, 0);
L
Linus Torvalds 已提交
1230 1231 1232 1233 1234 1235 1236 1237 1238 1239
	if (newsk == NULL)
		goto out;

	/* Allocate skb for sending to listening sock */
	skb = sock_wmalloc(newsk, 1, 0, GFP_KERNEL);
	if (skb == NULL)
		goto out;

restart:
	/*  Find listening sock. */
1240
	other = unix_find_other(net, sunaddr, addr_len, sk->sk_type, hash, &err);
L
Linus Torvalds 已提交
1241 1242 1243 1244
	if (!other)
		goto out;

	/* Latch state of peer */
1245
	unix_state_lock(other);
L
Linus Torvalds 已提交
1246 1247 1248

	/* Apparently VFS overslept socket death. Retry. */
	if (sock_flag(other, SOCK_DEAD)) {
1249
		unix_state_unlock(other);
L
Linus Torvalds 已提交
1250 1251 1252 1253 1254 1255 1256
		sock_put(other);
		goto restart;
	}

	err = -ECONNREFUSED;
	if (other->sk_state != TCP_LISTEN)
		goto out_unlock;
1257 1258
	if (other->sk_shutdown & RCV_SHUTDOWN)
		goto out_unlock;
L
Linus Torvalds 已提交
1259

1260
	if (unix_recvq_full(other)) {
L
Linus Torvalds 已提交
1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271
		err = -EAGAIN;
		if (!timeo)
			goto out_unlock;

		timeo = unix_wait_for_peer(other, timeo);

		err = sock_intr_errno(timeo);
		if (signal_pending(current))
			goto out;
		sock_put(other);
		goto restart;
1272
	}
L
Linus Torvalds 已提交
1273 1274 1275

	/* Latch our state.

D
Daniel Baluta 已提交
1276
	   It is tricky place. We need to grab our state lock and cannot
L
Linus Torvalds 已提交
1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299
	   drop lock on peer. It is dangerous because deadlock is
	   possible. Connect to self case and simultaneous
	   attempt to connect are eliminated by checking socket
	   state. other is TCP_LISTEN, if sk is TCP_LISTEN we
	   check this before attempt to grab lock.

	   Well, and we have to recheck the state after socket locked.
	 */
	st = sk->sk_state;

	switch (st) {
	case TCP_CLOSE:
		/* This is ok... continue with connect */
		break;
	case TCP_ESTABLISHED:
		/* Socket is already connected */
		err = -EISCONN;
		goto out_unlock;
	default:
		err = -EINVAL;
		goto out_unlock;
	}

1300
	unix_state_lock_nested(sk);
L
Linus Torvalds 已提交
1301 1302

	if (sk->sk_state != st) {
1303 1304
		unix_state_unlock(sk);
		unix_state_unlock(other);
L
Linus Torvalds 已提交
1305 1306 1307 1308
		sock_put(other);
		goto restart;
	}

1309
	err = security_unix_stream_connect(sk, other, newsk);
L
Linus Torvalds 已提交
1310
	if (err) {
1311
		unix_state_unlock(sk);
L
Linus Torvalds 已提交
1312 1313 1314 1315 1316 1317 1318 1319 1320
		goto out_unlock;
	}

	/* The way is open! Fastly set all the necessary fields... */

	sock_hold(sk);
	unix_peer(newsk)	= sk;
	newsk->sk_state		= TCP_ESTABLISHED;
	newsk->sk_type		= sk->sk_type;
1321
	init_peercred(newsk);
L
Linus Torvalds 已提交
1322
	newu = unix_sk(newsk);
1323
	RCU_INIT_POINTER(newsk->sk_wq, &newu->peer_wq);
L
Linus Torvalds 已提交
1324 1325 1326 1327
	otheru = unix_sk(other);

	/* copy address information from listening to new sock*/
	if (otheru->addr) {
1328
		refcount_inc(&otheru->addr->refcnt);
L
Linus Torvalds 已提交
1329 1330
		newu->addr = otheru->addr;
	}
A
Al Viro 已提交
1331 1332 1333
	if (otheru->path.dentry) {
		path_get(&otheru->path);
		newu->path = otheru->path;
L
Linus Torvalds 已提交
1334 1335 1336
	}

	/* Set credentials */
1337
	copy_peercred(sk, other);
L
Linus Torvalds 已提交
1338 1339 1340

	sock->state	= SS_CONNECTED;
	sk->sk_state	= TCP_ESTABLISHED;
1341 1342
	sock_hold(newsk);

1343
	smp_mb__after_atomic();	/* sock_hold() does an atomic_inc() */
1344
	unix_peer(sk)	= newsk;
L
Linus Torvalds 已提交
1345

1346
	unix_state_unlock(sk);
L
Linus Torvalds 已提交
1347 1348 1349 1350 1351

	/* take ten and and send info to listening sock */
	spin_lock(&other->sk_receive_queue.lock);
	__skb_queue_tail(&other->sk_receive_queue, skb);
	spin_unlock(&other->sk_receive_queue.lock);
1352
	unix_state_unlock(other);
1353
	other->sk_data_ready(other);
L
Linus Torvalds 已提交
1354 1355 1356 1357 1358
	sock_put(other);
	return 0;

out_unlock:
	if (other)
1359
		unix_state_unlock(other);
L
Linus Torvalds 已提交
1360 1361

out:
1362
	kfree_skb(skb);
L
Linus Torvalds 已提交
1363 1364 1365 1366 1367 1368 1369 1370 1371
	if (newsk)
		unix_release_sock(newsk, 0);
	if (other)
		sock_put(other);
	return err;
}

static int unix_socketpair(struct socket *socka, struct socket *sockb)
{
1372
	struct sock *ska = socka->sk, *skb = sockb->sk;
L
Linus Torvalds 已提交
1373 1374 1375 1376

	/* Join our sockets back to back */
	sock_hold(ska);
	sock_hold(skb);
1377 1378
	unix_peer(ska) = skb;
	unix_peer(skb) = ska;
1379 1380
	init_peercred(ska);
	init_peercred(skb);
L
Linus Torvalds 已提交
1381 1382 1383 1384 1385 1386 1387 1388 1389 1390

	if (ska->sk_type != SOCK_DGRAM) {
		ska->sk_state = TCP_ESTABLISHED;
		skb->sk_state = TCP_ESTABLISHED;
		socka->state  = SS_CONNECTED;
		sockb->state  = SS_CONNECTED;
	}
	return 0;
}

1391 1392 1393 1394 1395 1396 1397 1398 1399
static void unix_sock_inherit_flags(const struct socket *old,
				    struct socket *new)
{
	if (test_bit(SOCK_PASSCRED, &old->flags))
		set_bit(SOCK_PASSCRED, &new->flags);
	if (test_bit(SOCK_PASSSEC, &old->flags))
		set_bit(SOCK_PASSSEC, &new->flags);
}

1400 1401
static int unix_accept(struct socket *sock, struct socket *newsock, int flags,
		       bool kern)
L
Linus Torvalds 已提交
1402 1403 1404 1405 1406 1407 1408
{
	struct sock *sk = sock->sk;
	struct sock *tsk;
	struct sk_buff *skb;
	int err;

	err = -EOPNOTSUPP;
E
Eric Dumazet 已提交
1409
	if (sock->type != SOCK_STREAM && sock->type != SOCK_SEQPACKET)
L
Linus Torvalds 已提交
1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432
		goto out;

	err = -EINVAL;
	if (sk->sk_state != TCP_LISTEN)
		goto out;

	/* If socket state is TCP_LISTEN it cannot change (for now...),
	 * so that no locks are necessary.
	 */

	skb = skb_recv_datagram(sk, 0, flags&O_NONBLOCK, &err);
	if (!skb) {
		/* This means receive shutdown. */
		if (err == 0)
			err = -EINVAL;
		goto out;
	}

	tsk = skb->sk;
	skb_free_datagram(sk, skb);
	wake_up_interruptible(&unix_sk(sk)->peer_wait);

	/* attach accepted sock to socket */
1433
	unix_state_lock(tsk);
L
Linus Torvalds 已提交
1434
	newsock->state = SS_CONNECTED;
1435
	unix_sock_inherit_flags(sock, newsock);
L
Linus Torvalds 已提交
1436
	sock_graft(tsk, newsock);
1437
	unix_state_unlock(tsk);
L
Linus Torvalds 已提交
1438 1439 1440 1441 1442 1443 1444
	return 0;

out:
	return err;
}


1445
static int unix_getname(struct socket *sock, struct sockaddr *uaddr, int peer)
L
Linus Torvalds 已提交
1446 1447 1448
{
	struct sock *sk = sock->sk;
	struct unix_sock *u;
1449
	DECLARE_SOCKADDR(struct sockaddr_un *, sunaddr, uaddr);
L
Linus Torvalds 已提交
1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463
	int err = 0;

	if (peer) {
		sk = unix_peer_get(sk);

		err = -ENOTCONN;
		if (!sk)
			goto out;
		err = 0;
	} else {
		sock_hold(sk);
	}

	u = unix_sk(sk);
1464
	unix_state_lock(sk);
L
Linus Torvalds 已提交
1465 1466 1467
	if (!u->addr) {
		sunaddr->sun_family = AF_UNIX;
		sunaddr->sun_path[0] = 0;
1468
		err = sizeof(short);
L
Linus Torvalds 已提交
1469 1470 1471
	} else {
		struct unix_address *addr = u->addr;

1472 1473
		err = addr->len;
		memcpy(sunaddr, addr->name, addr->len);
L
Linus Torvalds 已提交
1474
	}
1475
	unix_state_unlock(sk);
L
Linus Torvalds 已提交
1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487
	sock_put(sk);
out:
	return err;
}

static void unix_detach_fds(struct scm_cookie *scm, struct sk_buff *skb)
{
	int i;

	scm->fp = UNIXCB(skb).fp;
	UNIXCB(skb).fp = NULL;

E
Eric Dumazet 已提交
1488
	for (i = scm->fp->count-1; i >= 0; i--)
1489
		unix_notinflight(scm->fp->user, scm->fp->fp[i]);
L
Linus Torvalds 已提交
1490 1491
}

1492
static void unix_destruct_scm(struct sk_buff *skb)
L
Linus Torvalds 已提交
1493 1494 1495
{
	struct scm_cookie scm;
	memset(&scm, 0, sizeof(scm));
1496 1497 1498
	scm.pid  = UNIXCB(skb).pid;
	if (UNIXCB(skb).fp)
		unix_detach_fds(&scm, skb);
L
Linus Torvalds 已提交
1499 1500 1501 1502 1503 1504 1505

	/* Alas, it calls VFS */
	/* So fscking what? fput() had been SMP-safe since the last Summer */
	scm_destroy(&scm);
	sock_wfree(skb);
}

1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520
/*
 * The "user->unix_inflight" variable is protected by the garbage
 * collection lock, and we just read it locklessly here. If you go
 * over the limit, there might be a tiny race in actually noticing
 * it across threads. Tough.
 */
static inline bool too_many_unix_fds(struct task_struct *p)
{
	struct user_struct *user = current_user();

	if (unlikely(user->unix_inflight > task_rlimit(p, RLIMIT_NOFILE)))
		return !capable(CAP_SYS_RESOURCE) && !capable(CAP_SYS_ADMIN);
	return false;
}

1521
static int unix_attach_fds(struct scm_cookie *scm, struct sk_buff *skb)
L
Linus Torvalds 已提交
1522 1523
{
	int i;
E
Eric Dumazet 已提交
1524

1525 1526 1527
	if (too_many_unix_fds(current))
		return -ETOOMANYREFS;

1528 1529 1530 1531 1532 1533 1534 1535 1536
	/*
	 * Need to duplicate file references for the sake of garbage
	 * collection.  Otherwise a socket in the fps might become a
	 * candidate for GC while the skb is not yet queued.
	 */
	UNIXCB(skb).fp = scm_fp_dup(scm->fp);
	if (!UNIXCB(skb).fp)
		return -ENOMEM;

1537
	for (i = scm->fp->count - 1; i >= 0; i--)
1538
		unix_inflight(scm->fp->user, scm->fp->fp[i]);
1539
	return 0;
L
Linus Torvalds 已提交
1540 1541
}

1542
static int unix_scm_to_skb(struct scm_cookie *scm, struct sk_buff *skb, bool send_fds)
1543 1544
{
	int err = 0;
1545

1546
	UNIXCB(skb).pid  = get_pid(scm->pid);
1547 1548
	UNIXCB(skb).uid = scm->creds.uid;
	UNIXCB(skb).gid = scm->creds.gid;
1549
	UNIXCB(skb).fp = NULL;
1550
	unix_get_secdata(scm, skb);
1551 1552 1553 1554 1555 1556 1557
	if (scm->fp && send_fds)
		err = unix_attach_fds(scm, skb);

	skb->destructor = unix_destruct_scm;
	return err;
}

1558 1559 1560 1561 1562 1563 1564 1565
static bool unix_passcred_enabled(const struct socket *sock,
				  const struct sock *other)
{
	return test_bit(SOCK_PASSCRED, &sock->flags) ||
	       !other->sk_socket ||
	       test_bit(SOCK_PASSCRED, &other->sk_socket->flags);
}

1566 1567 1568 1569 1570 1571 1572 1573
/*
 * Some apps rely on write() giving SCM_CREDENTIALS
 * We include credentials if source or destination socket
 * asserted SOCK_PASSCRED.
 */
static void maybe_add_creds(struct sk_buff *skb, const struct socket *sock,
			    const struct sock *other)
{
1574
	if (UNIXCB(skb).pid)
1575
		return;
1576
	if (unix_passcred_enabled(sock, other)) {
1577
		UNIXCB(skb).pid  = get_pid(task_tgid(current));
1578
		current_uid_gid(&UNIXCB(skb).uid, &UNIXCB(skb).gid);
1579 1580 1581
	}
}

1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610
static int maybe_init_creds(struct scm_cookie *scm,
			    struct socket *socket,
			    const struct sock *other)
{
	int err;
	struct msghdr msg = { .msg_controllen = 0 };

	err = scm_send(socket, &msg, scm, false);
	if (err)
		return err;

	if (unix_passcred_enabled(socket, other)) {
		scm->pid = get_pid(task_tgid(current));
		current_uid_gid(&scm->creds.uid, &scm->creds.gid);
	}
	return err;
}

static bool unix_skb_scm_eq(struct sk_buff *skb,
			    struct scm_cookie *scm)
{
	const struct unix_skb_parms *u = &UNIXCB(skb);

	return u->pid == scm->pid &&
	       uid_eq(u->uid, scm->creds.uid) &&
	       gid_eq(u->gid, scm->creds.gid) &&
	       unix_secdata_eq(scm, skb);
}

L
Linus Torvalds 已提交
1611 1612 1613 1614
/*
 *	Send AF_UNIX data.
 */

1615 1616
static int unix_dgram_sendmsg(struct socket *sock, struct msghdr *msg,
			      size_t len)
L
Linus Torvalds 已提交
1617 1618
{
	struct sock *sk = sock->sk;
1619
	struct net *net = sock_net(sk);
L
Linus Torvalds 已提交
1620
	struct unix_sock *u = unix_sk(sk);
1621
	DECLARE_SOCKADDR(struct sockaddr_un *, sunaddr, msg->msg_name);
L
Linus Torvalds 已提交
1622 1623 1624
	struct sock *other = NULL;
	int namelen = 0; /* fake GCC */
	int err;
1625
	unsigned int hash;
1626
	struct sk_buff *skb;
L
Linus Torvalds 已提交
1627
	long timeo;
C
Christoph Hellwig 已提交
1628
	struct scm_cookie scm;
1629
	int data_len = 0;
1630
	int sk_locked;
L
Linus Torvalds 已提交
1631

1632
	wait_for_unix_gc();
C
Christoph Hellwig 已提交
1633
	err = scm_send(sock, msg, &scm, false);
L
Linus Torvalds 已提交
1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653
	if (err < 0)
		return err;

	err = -EOPNOTSUPP;
	if (msg->msg_flags&MSG_OOB)
		goto out;

	if (msg->msg_namelen) {
		err = unix_mkname(sunaddr, msg->msg_namelen, &hash);
		if (err < 0)
			goto out;
		namelen = err;
	} else {
		sunaddr = NULL;
		err = -ENOTCONN;
		other = unix_peer_get(sk);
		if (!other)
			goto out;
	}

1654 1655
	if (test_bit(SOCK_PASSCRED, &sock->flags) && !u->addr
	    && (err = unix_autobind(sock)) != 0)
L
Linus Torvalds 已提交
1656 1657 1658 1659 1660 1661
		goto out;

	err = -EMSGSIZE;
	if (len > sk->sk_sndbuf - 32)
		goto out;

1662
	if (len > SKB_MAX_ALLOC) {
1663 1664 1665
		data_len = min_t(size_t,
				 len - SKB_MAX_ALLOC,
				 MAX_SKB_FRAGS * PAGE_SIZE);
1666 1667 1668 1669
		data_len = PAGE_ALIGN(data_len);

		BUILD_BUG_ON(SKB_MAX_ALLOC < PAGE_SIZE);
	}
1670 1671

	skb = sock_alloc_send_pskb(sk, len - data_len, data_len,
1672 1673
				   msg->msg_flags & MSG_DONTWAIT, &err,
				   PAGE_ALLOC_COSTLY_ORDER);
1674
	if (skb == NULL)
L
Linus Torvalds 已提交
1675 1676
		goto out;

C
Christoph Hellwig 已提交
1677
	err = unix_scm_to_skb(&scm, skb, true);
E
Eric Dumazet 已提交
1678
	if (err < 0)
1679
		goto out_free;
C
Catherine Zhang 已提交
1680

1681 1682 1683
	skb_put(skb, len - data_len);
	skb->data_len = data_len;
	skb->len = len;
A
Al Viro 已提交
1684
	err = skb_copy_datagram_from_iter(skb, 0, &msg->msg_iter, len);
L
Linus Torvalds 已提交
1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695
	if (err)
		goto out_free;

	timeo = sock_sndtimeo(sk, msg->msg_flags & MSG_DONTWAIT);

restart:
	if (!other) {
		err = -ECONNRESET;
		if (sunaddr == NULL)
			goto out_free;

1696
		other = unix_find_other(net, sunaddr, namelen, sk->sk_type,
L
Linus Torvalds 已提交
1697
					hash, &err);
1698
		if (other == NULL)
L
Linus Torvalds 已提交
1699 1700 1701
			goto out_free;
	}

A
Alban Crequy 已提交
1702 1703 1704 1705 1706 1707
	if (sk_filter(other, skb) < 0) {
		/* Toss the packet but do not return any error to the sender */
		err = len;
		goto out_free;
	}

1708
	sk_locked = 0;
1709
	unix_state_lock(other);
1710
restart_locked:
L
Linus Torvalds 已提交
1711 1712 1713 1714
	err = -EPERM;
	if (!unix_may_send(sk, other))
		goto out_unlock;

1715
	if (unlikely(sock_flag(other, SOCK_DEAD))) {
L
Linus Torvalds 已提交
1716 1717 1718 1719
		/*
		 *	Check with 1003.1g - what should
		 *	datagram error
		 */
1720
		unix_state_unlock(other);
L
Linus Torvalds 已提交
1721 1722
		sock_put(other);

1723 1724 1725
		if (!sk_locked)
			unix_state_lock(sk);

L
Linus Torvalds 已提交
1726 1727
		err = 0;
		if (unix_peer(sk) == other) {
1728
			unix_peer(sk) = NULL;
1729 1730
			unix_dgram_peer_wake_disconnect_wakeup(sk, other);

1731
			unix_state_unlock(sk);
L
Linus Torvalds 已提交
1732 1733 1734 1735 1736

			unix_dgram_disconnected(sk, other);
			sock_put(other);
			err = -ECONNREFUSED;
		} else {
1737
			unix_state_unlock(sk);
L
Linus Torvalds 已提交
1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755
		}

		other = NULL;
		if (err)
			goto out_free;
		goto restart;
	}

	err = -EPIPE;
	if (other->sk_shutdown & RCV_SHUTDOWN)
		goto out_unlock;

	if (sk->sk_type != SOCK_SEQPACKET) {
		err = security_unix_may_send(sk->sk_socket, other->sk_socket);
		if (err)
			goto out_unlock;
	}

1756 1757 1758 1759 1760 1761
	/* other == sk && unix_peer(other) != sk if
	 * - unix_peer(sk) == NULL, destination address bound to sk
	 * - unix_peer(sk) == sk by time of get but disconnected before lock
	 */
	if (other != sk &&
	    unlikely(unix_peer(other) != sk && unix_recvq_full(other))) {
1762 1763 1764 1765 1766 1767 1768 1769
		if (timeo) {
			timeo = unix_wait_for_peer(other, timeo);

			err = sock_intr_errno(timeo);
			if (signal_pending(current))
				goto out_free;

			goto restart;
L
Linus Torvalds 已提交
1770 1771
		}

1772 1773 1774 1775
		if (!sk_locked) {
			unix_state_unlock(other);
			unix_state_double_lock(sk, other);
		}
L
Linus Torvalds 已提交
1776

1777 1778 1779 1780 1781 1782
		if (unix_peer(sk) != other ||
		    unix_dgram_peer_wake_me(sk, other)) {
			err = -EAGAIN;
			sk_locked = 1;
			goto out_unlock;
		}
L
Linus Torvalds 已提交
1783

1784 1785 1786 1787
		if (!sk_locked) {
			sk_locked = 1;
			goto restart_locked;
		}
L
Linus Torvalds 已提交
1788 1789
	}

1790 1791 1792
	if (unlikely(sk_locked))
		unix_state_unlock(sk);

1793 1794
	if (sock_flag(other, SOCK_RCVTSTAMP))
		__net_timestamp(skb);
1795
	maybe_add_creds(skb, sock, other);
L
Linus Torvalds 已提交
1796
	skb_queue_tail(&other->sk_receive_queue, skb);
1797
	unix_state_unlock(other);
1798
	other->sk_data_ready(other);
L
Linus Torvalds 已提交
1799
	sock_put(other);
C
Christoph Hellwig 已提交
1800
	scm_destroy(&scm);
L
Linus Torvalds 已提交
1801 1802 1803
	return len;

out_unlock:
1804 1805
	if (sk_locked)
		unix_state_unlock(sk);
1806
	unix_state_unlock(other);
L
Linus Torvalds 已提交
1807 1808 1809 1810 1811
out_free:
	kfree_skb(skb);
out:
	if (other)
		sock_put(other);
C
Christoph Hellwig 已提交
1812
	scm_destroy(&scm);
L
Linus Torvalds 已提交
1813 1814 1815
	return err;
}

1816
/* We use paged skbs for stream sockets, and limit occupancy to 32768
1817
 * bytes, and a minimum of a full page.
1818 1819
 */
#define UNIX_SKB_FRAGS_SZ (PAGE_SIZE << get_order(32768))
1820

1821 1822
static int unix_stream_sendmsg(struct socket *sock, struct msghdr *msg,
			       size_t len)
L
Linus Torvalds 已提交
1823 1824 1825
{
	struct sock *sk = sock->sk;
	struct sock *other = NULL;
E
Eric Dumazet 已提交
1826
	int err, size;
1827
	struct sk_buff *skb;
1828
	int sent = 0;
C
Christoph Hellwig 已提交
1829
	struct scm_cookie scm;
1830
	bool fds_sent = false;
1831
	int data_len;
L
Linus Torvalds 已提交
1832

1833
	wait_for_unix_gc();
C
Christoph Hellwig 已提交
1834
	err = scm_send(sock, msg, &scm, false);
L
Linus Torvalds 已提交
1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846
	if (err < 0)
		return err;

	err = -EOPNOTSUPP;
	if (msg->msg_flags&MSG_OOB)
		goto out_err;

	if (msg->msg_namelen) {
		err = sk->sk_state == TCP_ESTABLISHED ? -EISCONN : -EOPNOTSUPP;
		goto out_err;
	} else {
		err = -ENOTCONN;
1847
		other = unix_peer(sk);
L
Linus Torvalds 已提交
1848 1849 1850 1851 1852 1853 1854
		if (!other)
			goto out_err;
	}

	if (sk->sk_shutdown & SEND_SHUTDOWN)
		goto pipe_err;

E
Eric Dumazet 已提交
1855
	while (sent < len) {
1856
		size = len - sent;
L
Linus Torvalds 已提交
1857 1858

		/* Keep two messages in the pipe so it schedules better */
1859
		size = min_t(int, size, (sk->sk_sndbuf >> 1) - 64);
L
Linus Torvalds 已提交
1860

1861 1862
		/* allow fallback to order-0 allocations */
		size = min_t(int, size, SKB_MAX_HEAD(0) + UNIX_SKB_FRAGS_SZ);
1863

1864
		data_len = max_t(int, 0, size - SKB_MAX_HEAD(0));
L
Linus Torvalds 已提交
1865

1866 1867
		data_len = min_t(size_t, size, PAGE_ALIGN(data_len));

1868
		skb = sock_alloc_send_pskb(sk, size - data_len, data_len,
1869 1870
					   msg->msg_flags & MSG_DONTWAIT, &err,
					   get_order(UNIX_SKB_FRAGS_SZ));
1871
		if (!skb)
L
Linus Torvalds 已提交
1872 1873
			goto out_err;

1874
		/* Only send the fds in the first buffer */
C
Christoph Hellwig 已提交
1875
		err = unix_scm_to_skb(&scm, skb, !fds_sent);
E
Eric Dumazet 已提交
1876
		if (err < 0) {
1877
			kfree_skb(skb);
1878
			goto out_err;
1879
		}
1880
		fds_sent = true;
L
Linus Torvalds 已提交
1881

1882 1883 1884
		skb_put(skb, size - data_len);
		skb->data_len = data_len;
		skb->len = size;
A
Al Viro 已提交
1885
		err = skb_copy_datagram_from_iter(skb, 0, &msg->msg_iter, size);
E
Eric Dumazet 已提交
1886
		if (err) {
L
Linus Torvalds 已提交
1887
			kfree_skb(skb);
1888
			goto out_err;
L
Linus Torvalds 已提交
1889 1890
		}

1891
		unix_state_lock(other);
L
Linus Torvalds 已提交
1892 1893 1894 1895 1896

		if (sock_flag(other, SOCK_DEAD) ||
		    (other->sk_shutdown & RCV_SHUTDOWN))
			goto pipe_err_free;

1897
		maybe_add_creds(skb, sock, other);
L
Linus Torvalds 已提交
1898
		skb_queue_tail(&other->sk_receive_queue, skb);
1899
		unix_state_unlock(other);
1900
		other->sk_data_ready(other);
1901
		sent += size;
L
Linus Torvalds 已提交
1902 1903
	}

C
Christoph Hellwig 已提交
1904
	scm_destroy(&scm);
L
Linus Torvalds 已提交
1905 1906 1907 1908

	return sent;

pipe_err_free:
1909
	unix_state_unlock(other);
L
Linus Torvalds 已提交
1910 1911
	kfree_skb(skb);
pipe_err:
E
Eric Dumazet 已提交
1912 1913
	if (sent == 0 && !(msg->msg_flags&MSG_NOSIGNAL))
		send_sig(SIGPIPE, current, 0);
L
Linus Torvalds 已提交
1914 1915
	err = -EPIPE;
out_err:
C
Christoph Hellwig 已提交
1916
	scm_destroy(&scm);
L
Linus Torvalds 已提交
1917 1918 1919
	return sent ? : err;
}

1920 1921 1922
static ssize_t unix_stream_sendpage(struct socket *socket, struct page *page,
				    int offset, size_t size, int flags)
{
1923 1924 1925 1926
	int err;
	bool send_sigpipe = false;
	bool init_scm = true;
	struct scm_cookie scm;
1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939
	struct sock *other, *sk = socket->sk;
	struct sk_buff *skb, *newskb = NULL, *tail = NULL;

	if (flags & MSG_OOB)
		return -EOPNOTSUPP;

	other = unix_peer(sk);
	if (!other || sk->sk_state != TCP_ESTABLISHED)
		return -ENOTCONN;

	if (false) {
alloc_skb:
		unix_state_unlock(other);
1940
		mutex_unlock(&unix_sk(other)->iolock);
1941 1942 1943
		newskb = sock_alloc_send_pskb(sk, 0, 0, flags & MSG_DONTWAIT,
					      &err, 0);
		if (!newskb)
1944
			goto err;
1945 1946
	}

1947
	/* we must acquire iolock as we modify already present
1948 1949
	 * skbs in the sk_receive_queue and mess with skb->len
	 */
1950
	err = mutex_lock_interruptible(&unix_sk(other)->iolock);
1951 1952 1953 1954 1955 1956 1957
	if (err) {
		err = flags & MSG_DONTWAIT ? -EAGAIN : -ERESTARTSYS;
		goto err;
	}

	if (sk->sk_shutdown & SEND_SHUTDOWN) {
		err = -EPIPE;
1958
		send_sigpipe = true;
1959 1960 1961 1962 1963 1964 1965 1966
		goto err_unlock;
	}

	unix_state_lock(other);

	if (sock_flag(other, SOCK_DEAD) ||
	    other->sk_shutdown & RCV_SHUTDOWN) {
		err = -EPIPE;
1967
		send_sigpipe = true;
1968 1969 1970
		goto err_state_unlock;
	}

1971 1972 1973 1974 1975 1976 1977
	if (init_scm) {
		err = maybe_init_creds(&scm, socket, other);
		if (err)
			goto err_state_unlock;
		init_scm = false;
	}

1978 1979 1980
	skb = skb_peek_tail(&other->sk_receive_queue);
	if (tail && tail == skb) {
		skb = newskb;
1981 1982
	} else if (!skb || !unix_skb_scm_eq(skb, &scm)) {
		if (newskb) {
1983
			skb = newskb;
1984 1985
		} else {
			tail = skb;
1986
			goto alloc_skb;
1987
		}
1988 1989 1990 1991 1992 1993
	} else if (newskb) {
		/* this is fast path, we don't necessarily need to
		 * call to kfree_skb even though with newskb == NULL
		 * this - does no harm
		 */
		consume_skb(newskb);
1994
		newskb = NULL;
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
	}

	if (skb_append_pagefrags(skb, page, offset, size)) {
		tail = skb;
		goto alloc_skb;
	}

	skb->len += size;
	skb->data_len += size;
	skb->truesize += size;
2005
	refcount_add(size, &sk->sk_wmem_alloc);
2006

2007
	if (newskb) {
2008 2009 2010
		err = unix_scm_to_skb(&scm, skb, false);
		if (err)
			goto err_state_unlock;
2011
		spin_lock(&other->sk_receive_queue.lock);
2012
		__skb_queue_tail(&other->sk_receive_queue, newskb);
2013 2014
		spin_unlock(&other->sk_receive_queue.lock);
	}
2015 2016

	unix_state_unlock(other);
2017
	mutex_unlock(&unix_sk(other)->iolock);
2018 2019

	other->sk_data_ready(other);
2020
	scm_destroy(&scm);
2021 2022 2023 2024 2025
	return size;

err_state_unlock:
	unix_state_unlock(other);
err_unlock:
2026
	mutex_unlock(&unix_sk(other)->iolock);
2027 2028 2029 2030
err:
	kfree_skb(newskb);
	if (send_sigpipe && !(flags & MSG_NOSIGNAL))
		send_sig(SIGPIPE, current, 0);
2031 2032
	if (!init_scm)
		scm_destroy(&scm);
2033 2034 2035
	return err;
}

2036 2037
static int unix_seqpacket_sendmsg(struct socket *sock, struct msghdr *msg,
				  size_t len)
L
Linus Torvalds 已提交
2038 2039 2040
{
	int err;
	struct sock *sk = sock->sk;
2041

L
Linus Torvalds 已提交
2042 2043 2044 2045 2046 2047 2048 2049 2050 2051
	err = sock_error(sk);
	if (err)
		return err;

	if (sk->sk_state != TCP_ESTABLISHED)
		return -ENOTCONN;

	if (msg->msg_namelen)
		msg->msg_namelen = 0;

2052
	return unix_dgram_sendmsg(sock, msg, len);
L
Linus Torvalds 已提交
2053
}
2054

2055 2056
static int unix_seqpacket_recvmsg(struct socket *sock, struct msghdr *msg,
				  size_t size, int flags)
2057 2058 2059 2060 2061 2062
{
	struct sock *sk = sock->sk;

	if (sk->sk_state != TCP_ESTABLISHED)
		return -ENOTCONN;

2063
	return unix_dgram_recvmsg(sock, msg, size, flags);
2064 2065
}

L
Linus Torvalds 已提交
2066 2067 2068 2069 2070 2071 2072 2073 2074 2075
static void unix_copy_addr(struct msghdr *msg, struct sock *sk)
{
	struct unix_sock *u = unix_sk(sk);

	if (u->addr) {
		msg->msg_namelen = u->addr->len;
		memcpy(msg->msg_name, u->addr->name, u->addr->len);
	}
}

2076 2077
static int unix_dgram_recvmsg(struct socket *sock, struct msghdr *msg,
			      size_t size, int flags)
L
Linus Torvalds 已提交
2078
{
C
Christoph Hellwig 已提交
2079
	struct scm_cookie scm;
L
Linus Torvalds 已提交
2080 2081
	struct sock *sk = sock->sk;
	struct unix_sock *u = unix_sk(sk);
2082 2083
	struct sk_buff *skb, *last;
	long timeo;
L
Linus Torvalds 已提交
2084
	int err;
2085
	int peeked, skip;
L
Linus Torvalds 已提交
2086 2087 2088 2089 2090

	err = -EOPNOTSUPP;
	if (flags&MSG_OOB)
		goto out;

2091
	timeo = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
L
Linus Torvalds 已提交
2092

2093
	do {
2094
		mutex_lock(&u->iolock);
2095

2096
		skip = sk_peek_offset(sk, flags);
2097 2098
		skb = __skb_try_recv_datagram(sk, flags, NULL, &peeked, &skip,
					      &err, &last);
2099 2100 2101
		if (skb)
			break;

2102
		mutex_unlock(&u->iolock);
2103 2104 2105 2106 2107 2108

		if (err != -EAGAIN)
			break;
	} while (timeo &&
		 !__skb_wait_for_more_packets(sk, &err, &timeo, last));

2109
	if (!skb) { /* implies iolock unlocked */
2110 2111 2112 2113 2114 2115
		unix_state_lock(sk);
		/* Signal EOF on disconnected non-blocking SEQPACKET socket. */
		if (sk->sk_type == SOCK_SEQPACKET && err == -EAGAIN &&
		    (sk->sk_shutdown & RCV_SHUTDOWN))
			err = 0;
		unix_state_unlock(sk);
2116
		goto out;
2117
	}
L
Linus Torvalds 已提交
2118

2119 2120
	if (wq_has_sleeper(&u->peer_wait))
		wake_up_interruptible_sync_poll(&u->peer_wait,
2121 2122
						EPOLLOUT | EPOLLWRNORM |
						EPOLLWRBAND);
L
Linus Torvalds 已提交
2123 2124 2125 2126

	if (msg->msg_name)
		unix_copy_addr(msg, skb->sk);

2127 2128 2129
	if (size > skb->len - skip)
		size = skb->len - skip;
	else if (size < skb->len - skip)
L
Linus Torvalds 已提交
2130 2131
		msg->msg_flags |= MSG_TRUNC;

2132
	err = skb_copy_datagram_msg(skb, skip, msg, size);
L
Linus Torvalds 已提交
2133 2134 2135
	if (err)
		goto out_free;

2136 2137 2138
	if (sock_flag(sk, SOCK_RCVTSTAMP))
		__sock_recv_timestamp(msg, sk, skb);

C
Christoph Hellwig 已提交
2139 2140 2141 2142
	memset(&scm, 0, sizeof(scm));

	scm_set_cred(&scm, UNIXCB(skb).pid, UNIXCB(skb).uid, UNIXCB(skb).gid);
	unix_set_secdata(&scm, skb);
L
Linus Torvalds 已提交
2143

E
Eric Dumazet 已提交
2144
	if (!(flags & MSG_PEEK)) {
L
Linus Torvalds 已提交
2145
		if (UNIXCB(skb).fp)
C
Christoph Hellwig 已提交
2146
			unix_detach_fds(&scm, skb);
2147 2148

		sk_peek_offset_bwd(sk, skb->len);
E
Eric Dumazet 已提交
2149
	} else {
L
Linus Torvalds 已提交
2150 2151 2152 2153 2154 2155
		/* It is questionable: on PEEK we could:
		   - do not return fds - good, but too simple 8)
		   - return fds, and do not return them on read (old strategy,
		     apparently wrong)
		   - clone fds (I chose it for now, it is the most universal
		     solution)
2156 2157 2158 2159 2160

		   POSIX 1003.1g does not actually define this clearly
		   at all. POSIX 1003.1g doesn't define a lot of things
		   clearly however!

L
Linus Torvalds 已提交
2161
		*/
2162 2163 2164

		sk_peek_offset_fwd(sk, size);

L
Linus Torvalds 已提交
2165
		if (UNIXCB(skb).fp)
C
Christoph Hellwig 已提交
2166
			scm.fp = scm_fp_dup(UNIXCB(skb).fp);
L
Linus Torvalds 已提交
2167
	}
2168
	err = (flags & MSG_TRUNC) ? skb->len - skip : size;
L
Linus Torvalds 已提交
2169

C
Christoph Hellwig 已提交
2170
	scm_recv(sock, msg, &scm, flags);
L
Linus Torvalds 已提交
2171 2172

out_free:
E
Eric Dumazet 已提交
2173
	skb_free_datagram(sk, skb);
2174
	mutex_unlock(&u->iolock);
L
Linus Torvalds 已提交
2175 2176 2177 2178 2179
out:
	return err;
}

/*
2180
 *	Sleep until more data has arrived. But check for races..
L
Linus Torvalds 已提交
2181
 */
2182
static long unix_stream_data_wait(struct sock *sk, long timeo,
2183 2184
				  struct sk_buff *last, unsigned int last_len,
				  bool freezable)
L
Linus Torvalds 已提交
2185
{
2186
	struct sk_buff *tail;
L
Linus Torvalds 已提交
2187 2188
	DEFINE_WAIT(wait);

2189
	unix_state_lock(sk);
L
Linus Torvalds 已提交
2190 2191

	for (;;) {
E
Eric Dumazet 已提交
2192
		prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
L
Linus Torvalds 已提交
2193

2194 2195 2196
		tail = skb_peek_tail(&sk->sk_receive_queue);
		if (tail != last ||
		    (tail && tail->len != last_len) ||
L
Linus Torvalds 已提交
2197 2198 2199 2200 2201 2202
		    sk->sk_err ||
		    (sk->sk_shutdown & RCV_SHUTDOWN) ||
		    signal_pending(current) ||
		    !timeo)
			break;

2203
		sk_set_bit(SOCKWQ_ASYNC_WAITDATA, sk);
2204
		unix_state_unlock(sk);
2205 2206 2207 2208
		if (freezable)
			timeo = freezable_schedule_timeout(timeo);
		else
			timeo = schedule_timeout(timeo);
2209
		unix_state_lock(sk);
2210 2211 2212 2213

		if (sock_flag(sk, SOCK_DEAD))
			break;

2214
		sk_clear_bit(SOCKWQ_ASYNC_WAITDATA, sk);
L
Linus Torvalds 已提交
2215 2216
	}

E
Eric Dumazet 已提交
2217
	finish_wait(sk_sleep(sk), &wait);
2218
	unix_state_unlock(sk);
L
Linus Torvalds 已提交
2219 2220 2221
	return timeo;
}

2222 2223 2224 2225 2226
static unsigned int unix_skb_len(const struct sk_buff *skb)
{
	return skb->len - UNIXCB(skb).consumed;
}

2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237
struct unix_stream_read_state {
	int (*recv_actor)(struct sk_buff *, int, int,
			  struct unix_stream_read_state *);
	struct socket *socket;
	struct msghdr *msg;
	struct pipe_inode_info *pipe;
	size_t size;
	int flags;
	unsigned int splice_flags;
};

2238 2239
static int unix_stream_read_generic(struct unix_stream_read_state *state,
				    bool freezable)
L
Linus Torvalds 已提交
2240
{
C
Christoph Hellwig 已提交
2241
	struct scm_cookie scm;
2242
	struct socket *sock = state->socket;
L
Linus Torvalds 已提交
2243 2244 2245
	struct sock *sk = sock->sk;
	struct unix_sock *u = unix_sk(sk);
	int copied = 0;
2246
	int flags = state->flags;
2247
	int noblock = flags & MSG_DONTWAIT;
2248
	bool check_creds = false;
L
Linus Torvalds 已提交
2249 2250 2251
	int target;
	int err = 0;
	long timeo;
2252
	int skip;
2253 2254
	size_t size = state->size;
	unsigned int last_len;
L
Linus Torvalds 已提交
2255

2256 2257
	if (unlikely(sk->sk_state != TCP_ESTABLISHED)) {
		err = -EINVAL;
L
Linus Torvalds 已提交
2258
		goto out;
2259
	}
L
Linus Torvalds 已提交
2260

2261 2262
	if (unlikely(flags & MSG_OOB)) {
		err = -EOPNOTSUPP;
L
Linus Torvalds 已提交
2263
		goto out;
2264
	}
L
Linus Torvalds 已提交
2265

2266
	target = sock_rcvlowat(sk, flags & MSG_WAITALL, size);
2267
	timeo = sock_rcvtimeo(sk, noblock);
L
Linus Torvalds 已提交
2268

2269 2270
	memset(&scm, 0, sizeof(scm));

L
Linus Torvalds 已提交
2271 2272 2273
	/* Lock the socket to prevent queue disordering
	 * while sleeps in memcpy_tomsg
	 */
2274
	mutex_lock(&u->iolock);
L
Linus Torvalds 已提交
2275

2276
	skip = max(sk_peek_offset(sk, flags), 0);
2277

E
Eric Dumazet 已提交
2278
	do {
L
Linus Torvalds 已提交
2279
		int chunk;
2280
		bool drop_skb;
2281
		struct sk_buff *skb, *last;
L
Linus Torvalds 已提交
2282

2283
redo:
2284
		unix_state_lock(sk);
2285 2286 2287 2288
		if (sock_flag(sk, SOCK_DEAD)) {
			err = -ECONNRESET;
			goto unlock;
		}
2289
		last = skb = skb_peek(&sk->sk_receive_queue);
2290
		last_len = last ? last->len : 0;
2291
again:
E
Eric Dumazet 已提交
2292
		if (skb == NULL) {
L
Linus Torvalds 已提交
2293
			if (copied >= target)
2294
				goto unlock;
L
Linus Torvalds 已提交
2295 2296 2297 2298

			/*
			 *	POSIX 1003.1g mandates this order.
			 */
2299

E
Eric Dumazet 已提交
2300 2301
			err = sock_error(sk);
			if (err)
2302
				goto unlock;
L
Linus Torvalds 已提交
2303
			if (sk->sk_shutdown & RCV_SHUTDOWN)
2304 2305 2306
				goto unlock;

			unix_state_unlock(sk);
2307 2308
			if (!timeo) {
				err = -EAGAIN;
L
Linus Torvalds 已提交
2309
				break;
2310 2311
			}

2312
			mutex_unlock(&u->iolock);
L
Linus Torvalds 已提交
2313

2314
			timeo = unix_stream_data_wait(sk, timeo, last,
2315
						      last_len, freezable);
L
Linus Torvalds 已提交
2316

2317
			if (signal_pending(current)) {
L
Linus Torvalds 已提交
2318
				err = sock_intr_errno(timeo);
2319
				scm_destroy(&scm);
L
Linus Torvalds 已提交
2320 2321
				goto out;
			}
2322

2323
			mutex_lock(&u->iolock);
2324
			goto redo;
2325
unlock:
2326 2327
			unix_state_unlock(sk);
			break;
L
Linus Torvalds 已提交
2328
		}
2329

2330 2331
		while (skip >= unix_skb_len(skb)) {
			skip -= unix_skb_len(skb);
2332
			last = skb;
2333
			last_len = skb->len;
2334
			skb = skb_peek_next(skb, &sk->sk_receive_queue);
2335 2336
			if (!skb)
				goto again;
2337 2338
		}

2339
		unix_state_unlock(sk);
L
Linus Torvalds 已提交
2340 2341 2342

		if (check_creds) {
			/* Never glue messages from different writers */
2343
			if (!unix_skb_scm_eq(skb, &scm))
L
Linus Torvalds 已提交
2344
				break;
2345
		} else if (test_bit(SOCK_PASSCRED, &sock->flags)) {
L
Linus Torvalds 已提交
2346
			/* Copy credentials */
C
Christoph Hellwig 已提交
2347
			scm_set_cred(&scm, UNIXCB(skb).pid, UNIXCB(skb).uid, UNIXCB(skb).gid);
2348
			unix_set_secdata(&scm, skb);
2349
			check_creds = true;
L
Linus Torvalds 已提交
2350 2351 2352
		}

		/* Copy address just once */
2353 2354 2355 2356
		if (state->msg && state->msg->msg_name) {
			DECLARE_SOCKADDR(struct sockaddr_un *, sunaddr,
					 state->msg->msg_name);
			unix_copy_addr(state->msg, skb->sk);
L
Linus Torvalds 已提交
2357 2358 2359
			sunaddr = NULL;
		}

2360
		chunk = min_t(unsigned int, unix_skb_len(skb) - skip, size);
2361
		skb_get(skb);
2362
		chunk = state->recv_actor(skb, skip, chunk, state);
2363 2364 2365
		drop_skb = !unix_skb_len(skb);
		/* skb is only safe to use if !drop_skb */
		consume_skb(skb);
2366
		if (chunk < 0) {
L
Linus Torvalds 已提交
2367 2368 2369 2370 2371 2372 2373
			if (copied == 0)
				copied = -EFAULT;
			break;
		}
		copied += chunk;
		size -= chunk;

2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385
		if (drop_skb) {
			/* the skb was touched by a concurrent reader;
			 * we should not expect anything from this skb
			 * anymore and assume it invalid - we can be
			 * sure it was dropped from the socket queue
			 *
			 * let's report a short read
			 */
			err = 0;
			break;
		}

L
Linus Torvalds 已提交
2386
		/* Mark read part of skb as used */
E
Eric Dumazet 已提交
2387
		if (!(flags & MSG_PEEK)) {
2388
			UNIXCB(skb).consumed += chunk;
L
Linus Torvalds 已提交
2389

2390 2391
			sk_peek_offset_bwd(sk, chunk);

L
Linus Torvalds 已提交
2392
			if (UNIXCB(skb).fp)
C
Christoph Hellwig 已提交
2393
				unix_detach_fds(&scm, skb);
L
Linus Torvalds 已提交
2394

2395
			if (unix_skb_len(skb))
L
Linus Torvalds 已提交
2396 2397
				break;

2398
			skb_unlink(skb, &sk->sk_receive_queue);
2399
			consume_skb(skb);
L
Linus Torvalds 已提交
2400

C
Christoph Hellwig 已提交
2401
			if (scm.fp)
L
Linus Torvalds 已提交
2402
				break;
E
Eric Dumazet 已提交
2403
		} else {
L
Linus Torvalds 已提交
2404 2405 2406
			/* It is questionable, see note in unix_dgram_recvmsg.
			 */
			if (UNIXCB(skb).fp)
C
Christoph Hellwig 已提交
2407
				scm.fp = scm_fp_dup(UNIXCB(skb).fp);
L
Linus Torvalds 已提交
2408

2409
			sk_peek_offset_fwd(sk, chunk);
2410

2411 2412 2413
			if (UNIXCB(skb).fp)
				break;

2414
			skip = 0;
2415 2416 2417 2418 2419 2420 2421
			last = skb;
			last_len = skb->len;
			unix_state_lock(sk);
			skb = skb_peek_next(skb, &sk->sk_receive_queue);
			if (skb)
				goto again;
			unix_state_unlock(sk);
L
Linus Torvalds 已提交
2422 2423 2424 2425
			break;
		}
	} while (size);

2426
	mutex_unlock(&u->iolock);
2427 2428 2429 2430
	if (state->msg)
		scm_recv(sock, state->msg, &scm, flags);
	else
		scm_destroy(&scm);
L
Linus Torvalds 已提交
2431 2432 2433 2434
out:
	return copied ? : err;
}

2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456
static int unix_stream_read_actor(struct sk_buff *skb,
				  int skip, int chunk,
				  struct unix_stream_read_state *state)
{
	int ret;

	ret = skb_copy_datagram_msg(skb, UNIXCB(skb).consumed + skip,
				    state->msg, chunk);
	return ret ?: chunk;
}

static int unix_stream_recvmsg(struct socket *sock, struct msghdr *msg,
			       size_t size, int flags)
{
	struct unix_stream_read_state state = {
		.recv_actor = unix_stream_read_actor,
		.socket = sock,
		.msg = msg,
		.size = size,
		.flags = flags
	};

2457
	return unix_stream_read_generic(&state, true);
2458 2459 2460 2461 2462 2463 2464 2465
}

static int unix_stream_splice_actor(struct sk_buff *skb,
				    int skip, int chunk,
				    struct unix_stream_read_state *state)
{
	return skb_splice_bits(skb, state->socket->sk,
			       UNIXCB(skb).consumed + skip,
A
Al Viro 已提交
2466
			       state->pipe, chunk, state->splice_flags);
2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487
}

static ssize_t unix_stream_splice_read(struct socket *sock,  loff_t *ppos,
				       struct pipe_inode_info *pipe,
				       size_t size, unsigned int flags)
{
	struct unix_stream_read_state state = {
		.recv_actor = unix_stream_splice_actor,
		.socket = sock,
		.pipe = pipe,
		.size = size,
		.splice_flags = flags,
	};

	if (unlikely(*ppos))
		return -ESPIPE;

	if (sock->file->f_flags & O_NONBLOCK ||
	    flags & SPLICE_F_NONBLOCK)
		state.flags = MSG_DONTWAIT;

2488
	return unix_stream_read_generic(&state, false);
2489 2490
}

L
Linus Torvalds 已提交
2491 2492 2493 2494 2495
static int unix_shutdown(struct socket *sock, int mode)
{
	struct sock *sk = sock->sk;
	struct sock *other;

2496 2497 2498 2499 2500 2501 2502 2503
	if (mode < SHUT_RD || mode > SHUT_RDWR)
		return -EINVAL;
	/* This maps:
	 * SHUT_RD   (0) -> RCV_SHUTDOWN  (1)
	 * SHUT_WR   (1) -> SEND_SHUTDOWN (2)
	 * SHUT_RDWR (2) -> SHUTDOWN_MASK (3)
	 */
	++mode;
2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529

	unix_state_lock(sk);
	sk->sk_shutdown |= mode;
	other = unix_peer(sk);
	if (other)
		sock_hold(other);
	unix_state_unlock(sk);
	sk->sk_state_change(sk);

	if (other &&
		(sk->sk_type == SOCK_STREAM || sk->sk_type == SOCK_SEQPACKET)) {

		int peer_mode = 0;

		if (mode&RCV_SHUTDOWN)
			peer_mode |= SEND_SHUTDOWN;
		if (mode&SEND_SHUTDOWN)
			peer_mode |= RCV_SHUTDOWN;
		unix_state_lock(other);
		other->sk_shutdown |= peer_mode;
		unix_state_unlock(other);
		other->sk_state_change(other);
		if (peer_mode == SHUTDOWN_MASK)
			sk_wake_async(other, SOCK_WAKE_WAITD, POLL_HUP);
		else if (peer_mode & RCV_SHUTDOWN)
			sk_wake_async(other, SOCK_WAKE_WAITD, POLL_IN);
L
Linus Torvalds 已提交
2530
	}
2531 2532 2533
	if (other)
		sock_put(other);

L
Linus Torvalds 已提交
2534 2535 2536
	return 0;
}

2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548
long unix_inq_len(struct sock *sk)
{
	struct sk_buff *skb;
	long amount = 0;

	if (sk->sk_state == TCP_LISTEN)
		return -EINVAL;

	spin_lock(&sk->sk_receive_queue.lock);
	if (sk->sk_type == SOCK_STREAM ||
	    sk->sk_type == SOCK_SEQPACKET) {
		skb_queue_walk(&sk->sk_receive_queue, skb)
2549
			amount += unix_skb_len(skb);
2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566
	} else {
		skb = skb_peek(&sk->sk_receive_queue);
		if (skb)
			amount = skb->len;
	}
	spin_unlock(&sk->sk_receive_queue.lock);

	return amount;
}
EXPORT_SYMBOL_GPL(unix_inq_len);

long unix_outq_len(struct sock *sk)
{
	return sk_wmem_alloc_get(sk);
}
EXPORT_SYMBOL_GPL(unix_outq_len);

2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603
static int unix_open_file(struct sock *sk)
{
	struct path path;
	struct file *f;
	int fd;

	if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
		return -EPERM;

	unix_state_lock(sk);
	path = unix_sk(sk)->path;
	if (!path.dentry) {
		unix_state_unlock(sk);
		return -ENOENT;
	}

	path_get(&path);
	unix_state_unlock(sk);

	fd = get_unused_fd_flags(O_CLOEXEC);
	if (fd < 0)
		goto out;

	f = dentry_open(&path, O_PATH, current_cred());
	if (IS_ERR(f)) {
		put_unused_fd(fd);
		fd = PTR_ERR(f);
		goto out;
	}

	fd_install(fd, f);
out:
	path_put(&path);

	return fd;
}

L
Linus Torvalds 已提交
2604 2605 2606
static int unix_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
{
	struct sock *sk = sock->sk;
2607
	long amount = 0;
L
Linus Torvalds 已提交
2608 2609
	int err;

E
Eric Dumazet 已提交
2610 2611
	switch (cmd) {
	case SIOCOUTQ:
2612
		amount = unix_outq_len(sk);
E
Eric Dumazet 已提交
2613 2614 2615
		err = put_user(amount, (int __user *)arg);
		break;
	case SIOCINQ:
2616 2617 2618 2619
		amount = unix_inq_len(sk);
		if (amount < 0)
			err = amount;
		else
L
Linus Torvalds 已提交
2620
			err = put_user(amount, (int __user *)arg);
2621
		break;
2622 2623 2624
	case SIOCUNIXFILE:
		err = unix_open_file(sk);
		break;
E
Eric Dumazet 已提交
2625 2626 2627
	default:
		err = -ENOIOCTLCMD;
		break;
L
Linus Torvalds 已提交
2628 2629 2630 2631
	}
	return err;
}

2632
static __poll_t unix_poll_mask(struct socket *sock, __poll_t events)
L
Linus Torvalds 已提交
2633 2634
{
	struct sock *sk = sock->sk;
2635
	__poll_t mask = 0;
L
Linus Torvalds 已提交
2636 2637 2638

	/* exceptional events? */
	if (sk->sk_err)
2639
		mask |= EPOLLERR;
L
Linus Torvalds 已提交
2640
	if (sk->sk_shutdown == SHUTDOWN_MASK)
2641
		mask |= EPOLLHUP;
2642
	if (sk->sk_shutdown & RCV_SHUTDOWN)
2643
		mask |= EPOLLRDHUP | EPOLLIN | EPOLLRDNORM;
L
Linus Torvalds 已提交
2644 2645

	/* readable? */
E
Eric Dumazet 已提交
2646
	if (!skb_queue_empty(&sk->sk_receive_queue))
2647
		mask |= EPOLLIN | EPOLLRDNORM;
L
Linus Torvalds 已提交
2648 2649

	/* Connection-based need to check for termination and startup */
E
Eric Dumazet 已提交
2650 2651
	if ((sk->sk_type == SOCK_STREAM || sk->sk_type == SOCK_SEQPACKET) &&
	    sk->sk_state == TCP_CLOSE)
2652
		mask |= EPOLLHUP;
L
Linus Torvalds 已提交
2653 2654 2655 2656 2657 2658

	/*
	 * we set writable also when the other side has shut down the
	 * connection. This prevents stuck sockets.
	 */
	if (unix_writable(sk))
2659
		mask |= EPOLLOUT | EPOLLWRNORM | EPOLLWRBAND;
L
Linus Torvalds 已提交
2660 2661 2662 2663

	return mask;
}

2664
static __poll_t unix_dgram_poll_mask(struct socket *sock, __poll_t events)
2665
{
2666
	struct sock *sk = sock->sk, *other;
2667 2668
	int writable;
	__poll_t mask = 0;
2669 2670 2671

	/* exceptional events? */
	if (sk->sk_err || !skb_queue_empty(&sk->sk_error_queue))
2672 2673
		mask |= EPOLLERR |
			(sock_flag(sk, SOCK_SELECT_ERR_QUEUE) ? EPOLLPRI : 0);
2674

2675
	if (sk->sk_shutdown & RCV_SHUTDOWN)
2676
		mask |= EPOLLRDHUP | EPOLLIN | EPOLLRDNORM;
2677
	if (sk->sk_shutdown == SHUTDOWN_MASK)
2678
		mask |= EPOLLHUP;
2679 2680

	/* readable? */
2681
	if (!skb_queue_empty(&sk->sk_receive_queue))
2682
		mask |= EPOLLIN | EPOLLRDNORM;
2683 2684 2685 2686

	/* Connection-based need to check for termination and startup */
	if (sk->sk_type == SOCK_SEQPACKET) {
		if (sk->sk_state == TCP_CLOSE)
2687
			mask |= EPOLLHUP;
2688 2689 2690 2691 2692
		/* connection hasn't started yet? */
		if (sk->sk_state == TCP_SYN_SENT)
			return mask;
	}

E
Eric Dumazet 已提交
2693
	/* No write status requested, avoid expensive OUT tests. */
2694
	if (!(events & (EPOLLWRBAND|EPOLLWRNORM|EPOLLOUT)))
E
Eric Dumazet 已提交
2695 2696
		return mask;

2697
	writable = unix_writable(sk);
2698 2699 2700 2701 2702 2703 2704 2705 2706 2707
	if (writable) {
		unix_state_lock(sk);

		other = unix_peer(sk);
		if (other && unix_peer(other) != sk &&
		    unix_recvq_full(other) &&
		    unix_dgram_peer_wake_me(sk, other))
			writable = 0;

		unix_state_unlock(sk);
2708 2709 2710
	}

	if (writable)
2711
		mask |= EPOLLOUT | EPOLLWRNORM | EPOLLWRBAND;
2712
	else
2713
		sk_set_bit(SOCKWQ_ASYNC_NOSPACE, sk);
2714 2715 2716

	return mask;
}
L
Linus Torvalds 已提交
2717 2718

#ifdef CONFIG_PROC_FS
2719

E
Eric Dumazet 已提交
2720 2721 2722 2723 2724
#define BUCKET_SPACE (BITS_PER_LONG - (UNIX_HASH_BITS + 1) - 1)

#define get_bucket(x) ((x) >> BUCKET_SPACE)
#define get_offset(x) ((x) & ((1L << BUCKET_SPACE) - 1))
#define set_bucket_offset(b, o) ((b) << BUCKET_SPACE | (o))
2725

E
Eric Dumazet 已提交
2726
static struct sock *unix_from_bucket(struct seq_file *seq, loff_t *pos)
L
Linus Torvalds 已提交
2727
{
E
Eric Dumazet 已提交
2728 2729 2730 2731
	unsigned long offset = get_offset(*pos);
	unsigned long bucket = get_bucket(*pos);
	struct sock *sk;
	unsigned long count = 0;
L
Linus Torvalds 已提交
2732

E
Eric Dumazet 已提交
2733 2734
	for (sk = sk_head(&unix_socket_table[bucket]); sk; sk = sk_next(sk)) {
		if (sock_net(sk) != seq_file_net(seq))
2735
			continue;
E
Eric Dumazet 已提交
2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754
		if (++count == offset)
			break;
	}

	return sk;
}

static struct sock *unix_next_socket(struct seq_file *seq,
				     struct sock *sk,
				     loff_t *pos)
{
	unsigned long bucket;

	while (sk > (struct sock *)SEQ_START_TOKEN) {
		sk = sk_next(sk);
		if (!sk)
			goto next_bucket;
		if (sock_net(sk) == seq_file_net(seq))
			return sk;
L
Linus Torvalds 已提交
2755
	}
E
Eric Dumazet 已提交
2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766

	do {
		sk = unix_from_bucket(seq, pos);
		if (sk)
			return sk;

next_bucket:
		bucket = get_bucket(*pos) + 1;
		*pos = set_bucket_offset(bucket, 1);
	} while (bucket < ARRAY_SIZE(unix_socket_table));

L
Linus Torvalds 已提交
2767 2768 2769 2770
	return NULL;
}

static void *unix_seq_start(struct seq_file *seq, loff_t *pos)
2771
	__acquires(unix_table_lock)
L
Linus Torvalds 已提交
2772
{
2773
	spin_lock(&unix_table_lock);
E
Eric Dumazet 已提交
2774 2775 2776 2777 2778 2779 2780 2781

	if (!*pos)
		return SEQ_START_TOKEN;

	if (get_bucket(*pos) >= ARRAY_SIZE(unix_socket_table))
		return NULL;

	return unix_next_socket(seq, NULL, pos);
L
Linus Torvalds 已提交
2782 2783 2784 2785 2786
}

static void *unix_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
	++*pos;
E
Eric Dumazet 已提交
2787
	return unix_next_socket(seq, v, pos);
L
Linus Torvalds 已提交
2788 2789 2790
}

static void unix_seq_stop(struct seq_file *seq, void *v)
2791
	__releases(unix_table_lock)
L
Linus Torvalds 已提交
2792
{
2793
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
2794 2795 2796 2797
}

static int unix_seq_show(struct seq_file *seq, void *v)
{
2798

J
Joe Perches 已提交
2799
	if (v == SEQ_START_TOKEN)
L
Linus Torvalds 已提交
2800 2801 2802 2803 2804
		seq_puts(seq, "Num       RefCount Protocol Flags    Type St "
			 "Inode Path\n");
	else {
		struct sock *s = v;
		struct unix_sock *u = unix_sk(s);
2805
		unix_state_lock(s);
L
Linus Torvalds 已提交
2806

D
Dan Rosenberg 已提交
2807
		seq_printf(seq, "%pK: %08X %08X %08X %04X %02X %5lu",
L
Linus Torvalds 已提交
2808
			s,
2809
			refcount_read(&s->sk_refcnt),
L
Linus Torvalds 已提交
2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830
			0,
			s->sk_state == TCP_LISTEN ? __SO_ACCEPTCON : 0,
			s->sk_type,
			s->sk_socket ?
			(s->sk_state == TCP_ESTABLISHED ? SS_CONNECTED : SS_UNCONNECTED) :
			(s->sk_state == TCP_ESTABLISHED ? SS_CONNECTING : SS_DISCONNECTING),
			sock_i_ino(s));

		if (u->addr) {
			int i, len;
			seq_putc(seq, ' ');

			i = 0;
			len = u->addr->len - sizeof(short);
			if (!UNIX_ABSTRACT(s))
				len--;
			else {
				seq_putc(seq, '@');
				i++;
			}
			for ( ; i < len; i++)
2831 2832
				seq_putc(seq, u->addr->name->sun_path[i] ?:
					 '@');
L
Linus Torvalds 已提交
2833
		}
2834
		unix_state_unlock(s);
L
Linus Torvalds 已提交
2835 2836 2837 2838 2839 2840
		seq_putc(seq, '\n');
	}

	return 0;
}

2841
static const struct seq_operations unix_seq_ops = {
L
Linus Torvalds 已提交
2842 2843 2844 2845 2846 2847 2848
	.start  = unix_seq_start,
	.next   = unix_seq_next,
	.stop   = unix_seq_stop,
	.show   = unix_seq_show,
};
#endif

2849
static const struct net_proto_family unix_family_ops = {
L
Linus Torvalds 已提交
2850 2851 2852 2853 2854
	.family = PF_UNIX,
	.create = unix_create,
	.owner	= THIS_MODULE,
};

2855

2856
static int __net_init unix_net_init(struct net *net)
2857 2858 2859
{
	int error = -ENOMEM;

2860
	net->unx.sysctl_max_dgram_qlen = 10;
2861 2862
	if (unix_sysctl_register(net))
		goto out;
2863

2864
#ifdef CONFIG_PROC_FS
2865 2866
	if (!proc_create_net("unix", 0, net->proc_net, &unix_seq_ops,
			sizeof(struct seq_net_private))) {
2867
		unix_sysctl_unregister(net);
2868
		goto out;
2869
	}
2870 2871 2872
#endif
	error = 0;
out:
2873
	return error;
2874 2875
}

2876
static void __net_exit unix_net_exit(struct net *net)
2877
{
2878
	unix_sysctl_unregister(net);
2879
	remove_proc_entry("unix", net->proc_net);
2880 2881 2882 2883 2884 2885 2886
}

static struct pernet_operations unix_net_ops = {
	.init = unix_net_init,
	.exit = unix_net_exit,
};

L
Linus Torvalds 已提交
2887 2888 2889 2890
static int __init af_unix_init(void)
{
	int rc = -1;

2891
	BUILD_BUG_ON(sizeof(struct unix_skb_parms) > FIELD_SIZEOF(struct sk_buff, cb));
L
Linus Torvalds 已提交
2892 2893

	rc = proto_register(&unix_proto, 1);
2894
	if (rc != 0) {
W
wangweidong 已提交
2895
		pr_crit("%s: Cannot create unix_sock SLAB cache!\n", __func__);
L
Linus Torvalds 已提交
2896 2897 2898 2899
		goto out;
	}

	sock_register(&unix_family_ops);
2900
	register_pernet_subsys(&unix_net_ops);
L
Linus Torvalds 已提交
2901 2902 2903 2904 2905 2906 2907 2908
out:
	return rc;
}

static void __exit af_unix_exit(void)
{
	sock_unregister(PF_UNIX);
	proto_unregister(&unix_proto);
2909
	unregister_pernet_subsys(&unix_net_ops);
L
Linus Torvalds 已提交
2910 2911
}

2912 2913 2914 2915 2916
/* Earlier than device_initcall() so that other drivers invoking
   request_module() don't end up in a loop when modprobe tries
   to use a UNIX socket. But later than subsys_initcall() because
   we depend on stuff initialised there */
fs_initcall(af_unix_init);
L
Linus Torvalds 已提交
2917 2918 2919 2920
module_exit(af_unix_exit);

MODULE_LICENSE("GPL");
MODULE_ALIAS_NETPROTO(PF_UNIX);