af_unix.c 58.7 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 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/signal.h>
#include <linux/sched.h>
#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>
#include <asm/uaccess.h>
#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>
L
Linus Torvalds 已提交
120

E
Eric Dumazet 已提交
121
struct hlist_head unix_socket_table[2 * UNIX_HASH_SIZE];
122 123 124
EXPORT_SYMBOL_GPL(unix_socket_table);
DEFINE_SPINLOCK(unix_table_lock);
EXPORT_SYMBOL_GPL(unix_table_lock);
E
Eric Dumazet 已提交
125
static atomic_long_t unix_nr_socks;
L
Linus Torvalds 已提交
126 127


E
Eric Dumazet 已提交
128 129 130 131 132 133 134 135 136 137 138
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
139

C
Catherine Zhang 已提交
140
#ifdef CONFIG_SECURITY_NETWORK
141
static void unix_get_secdata(struct scm_cookie *scm, struct sk_buff *skb)
C
Catherine Zhang 已提交
142
{
143
	memcpy(UNIXSID(skb), &scm->secid, sizeof(u32));
C
Catherine Zhang 已提交
144 145 146 147
}

static inline void unix_set_secdata(struct scm_cookie *scm, struct sk_buff *skb)
{
148
	scm->secid = *UNIXSID(skb);
C
Catherine Zhang 已提交
149 150
}
#else
151
static inline void unix_get_secdata(struct scm_cookie *scm, struct sk_buff *skb)
C
Catherine Zhang 已提交
152 153 154 155 156 157
{ }

static inline void unix_set_secdata(struct scm_cookie *scm, struct sk_buff *skb)
{ }
#endif /* CONFIG_SECURITY_NETWORK */

L
Linus Torvalds 已提交
158 159
/*
 *  SMP locking strategy:
160
 *    hash table is protected with spinlock unix_table_lock
161
 *    each socket state is protected by separate spin lock.
L
Linus Torvalds 已提交
162 163
 */

164
static inline unsigned int unix_hash_fold(__wsum n)
L
Linus Torvalds 已提交
165
{
166
	unsigned int hash = (__force unsigned int)csum_fold(n);
167

L
Linus Torvalds 已提交
168 169 170 171 172 173 174 175 176 177 178 179 180
	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
181
	return unix_peer(osk) == NULL || unix_our_peer(sk, osk);
L
Linus Torvalds 已提交
182 183
}

184 185 186 187 188
static inline int unix_recvq_full(struct sock const *sk)
{
	return skb_queue_len(&sk->sk_receive_queue) > sk->sk_max_ack_backlog;
}

189
struct sock *unix_peer_get(struct sock *s)
L
Linus Torvalds 已提交
190 191 192
{
	struct sock *peer;

193
	unix_state_lock(s);
L
Linus Torvalds 已提交
194 195 196
	peer = unix_peer(s);
	if (peer)
		sock_hold(peer);
197
	unix_state_unlock(s);
L
Linus Torvalds 已提交
198 199
	return peer;
}
200
EXPORT_SYMBOL_GPL(unix_peer_get);
L
Linus Torvalds 已提交
201 202 203 204 205 206 207 208 209 210 211 212 213

static inline void unix_release_addr(struct unix_address *addr)
{
	if (atomic_dec_and_test(&addr->refcnt))
		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.
 */
214

215
static int unix_mkname(struct sockaddr_un *sunaddr, int len, unsigned int *hashp)
L
Linus Torvalds 已提交
216 217 218 219 220 221 222 223 224
{
	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
225
		 * sun_path[108] doesn't as such exist.  However in kernel space
L
Linus Torvalds 已提交
226 227 228
		 * we are guaranteed that it is a valid memory location in our
		 * kernel address buffer.
		 */
229
		((char *)sunaddr)[len] = 0;
L
Linus Torvalds 已提交
230 231 232 233
		len = strlen(sunaddr->sun_path)+1+sizeof(short);
		return len;
	}

234
	*hashp = unix_hash_fold(csum_partial(sunaddr, len, 0));
L
Linus Torvalds 已提交
235 236 237 238 239 240 241 242 243 244
	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)
{
245
	WARN_ON(!sk_unhashed(sk));
L
Linus Torvalds 已提交
246 247 248 249 250
	sk_add_node(sk, list);
}

static inline void unix_remove_socket(struct sock *sk)
{
251
	spin_lock(&unix_table_lock);
L
Linus Torvalds 已提交
252
	__unix_remove_socket(sk);
253
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
254 255 256 257
}

static inline void unix_insert_socket(struct hlist_head *list, struct sock *sk)
{
258
	spin_lock(&unix_table_lock);
L
Linus Torvalds 已提交
259
	__unix_insert_socket(list, sk);
260
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
261 262
}

263 264
static struct sock *__unix_find_socket_byname(struct net *net,
					      struct sockaddr_un *sunname,
265
					      int len, int type, unsigned int hash)
L
Linus Torvalds 已提交
266 267 268
{
	struct sock *s;

269
	sk_for_each(s, &unix_socket_table[hash ^ type]) {
L
Linus Torvalds 已提交
270 271
		struct unix_sock *u = unix_sk(s);

272
		if (!net_eq(sock_net(s), net))
273 274
			continue;

L
Linus Torvalds 已提交
275 276 277 278 279 280 281 282 283
		if (u->addr->len == len &&
		    !memcmp(u->addr->name, sunname, len))
			goto found;
	}
	s = NULL;
found:
	return s;
}

284 285
static inline struct sock *unix_find_socket_byname(struct net *net,
						   struct sockaddr_un *sunname,
L
Linus Torvalds 已提交
286
						   int len, int type,
287
						   unsigned int hash)
L
Linus Torvalds 已提交
288 289 290
{
	struct sock *s;

291
	spin_lock(&unix_table_lock);
292
	s = __unix_find_socket_byname(net, sunname, len, type, hash);
L
Linus Torvalds 已提交
293 294
	if (s)
		sock_hold(s);
295
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
296 297 298
	return s;
}

299
static struct sock *unix_find_socket_byinode(struct inode *i)
L
Linus Torvalds 已提交
300 301 302
{
	struct sock *s;

303
	spin_lock(&unix_table_lock);
304
	sk_for_each(s,
L
Linus Torvalds 已提交
305
		    &unix_socket_table[i->i_ino & (UNIX_HASH_SIZE - 1)]) {
A
Al Viro 已提交
306
		struct dentry *dentry = unix_sk(s)->path.dentry;
L
Linus Torvalds 已提交
307

308
		if (dentry && d_backing_inode(dentry) == i) {
L
Linus Torvalds 已提交
309 310 311 312 313 314
			sock_hold(s);
			goto found;
		}
	}
	s = NULL;
found:
315
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
316 317 318 319 320 321 322 323 324 325
	return s;
}

static inline int unix_writable(struct sock *sk)
{
	return (atomic_read(&sk->sk_wmem_alloc) << 2) <= sk->sk_sndbuf;
}

static void unix_write_space(struct sock *sk)
{
326 327 328
	struct socket_wq *wq;

	rcu_read_lock();
L
Linus Torvalds 已提交
329
	if (unix_writable(sk)) {
330 331
		wq = rcu_dereference(sk->sk_wq);
		if (wq_has_sleeper(wq))
E
Eric Dumazet 已提交
332 333
			wake_up_interruptible_sync_poll(&wq->wait,
				POLLOUT | POLLWRNORM | POLLWRBAND);
334
		sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
L
Linus Torvalds 已提交
335
	}
336
	rcu_read_unlock();
L
Linus Torvalds 已提交
337 338 339 340 341 342 343 344
}

/* 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)
{
345
	if (!skb_queue_empty(&sk->sk_receive_queue)) {
L
Linus Torvalds 已提交
346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365
		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);

366 367 368
	WARN_ON(atomic_read(&sk->sk_wmem_alloc));
	WARN_ON(!sk_unhashed(sk));
	WARN_ON(sk->sk_socket);
L
Linus Torvalds 已提交
369
	if (!sock_flag(sk, SOCK_DEAD)) {
W
wangweidong 已提交
370
		pr_info("Attempt to release alive unix socket: %p\n", sk);
L
Linus Torvalds 已提交
371 372 373 374 375 376
		return;
	}

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

E
Eric Dumazet 已提交
377
	atomic_long_dec(&unix_nr_socks);
378
	local_bh_disable();
379
	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
380
	local_bh_enable();
L
Linus Torvalds 已提交
381
#ifdef UNIX_REFCNT_DEBUG
W
wangweidong 已提交
382
	pr_debug("UNIX %p is destroyed, %ld are still alive.\n", sk,
E
Eric Dumazet 已提交
383
		atomic_long_read(&unix_nr_socks));
L
Linus Torvalds 已提交
384 385 386
#endif
}

387
static void unix_release_sock(struct sock *sk, int embrion)
L
Linus Torvalds 已提交
388 389
{
	struct unix_sock *u = unix_sk(sk);
A
Al Viro 已提交
390
	struct path path;
L
Linus Torvalds 已提交
391 392 393 394 395 396 397
	struct sock *skpair;
	struct sk_buff *skb;
	int state;

	unix_remove_socket(sk);

	/* Clear state */
398
	unix_state_lock(sk);
L
Linus Torvalds 已提交
399 400
	sock_orphan(sk);
	sk->sk_shutdown = SHUTDOWN_MASK;
A
Al Viro 已提交
401 402 403
	path	     = u->path;
	u->path.dentry = NULL;
	u->path.mnt = NULL;
L
Linus Torvalds 已提交
404 405
	state = sk->sk_state;
	sk->sk_state = TCP_CLOSE;
406
	unix_state_unlock(sk);
L
Linus Torvalds 已提交
407 408 409

	wake_up_interruptible_all(&u->peer_wait);

410
	skpair = unix_peer(sk);
L
Linus Torvalds 已提交
411

412
	if (skpair != NULL) {
L
Linus Torvalds 已提交
413
		if (sk->sk_type == SOCK_STREAM || sk->sk_type == SOCK_SEQPACKET) {
414
			unix_state_lock(skpair);
L
Linus Torvalds 已提交
415 416 417 418
			/* No more writes */
			skpair->sk_shutdown = SHUTDOWN_MASK;
			if (!skb_queue_empty(&sk->sk_receive_queue) || embrion)
				skpair->sk_err = ECONNRESET;
419
			unix_state_unlock(skpair);
L
Linus Torvalds 已提交
420
			skpair->sk_state_change(skpair);
421
			sk_wake_async(skpair, SOCK_WAKE_WAITD, POLL_HUP);
L
Linus Torvalds 已提交
422 423 424 425 426 427 428 429
		}
		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) {
430
		if (state == TCP_LISTEN)
L
Linus Torvalds 已提交
431 432 433 434 435
			unix_release_sock(skb->sk, 1);
		/* passed fds are erased in the kfree_skb hook	      */
		kfree_skb(skb);
	}

A
Al Viro 已提交
436 437
	if (path.dentry)
		path_put(&path);
L
Linus Torvalds 已提交
438 439 440 441 442 443

	sock_put(sk);

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

	/*
A
Alan Cox 已提交
444
	 * Fixme: BSD difference: In BSD all sockets connected to us get
L
Linus Torvalds 已提交
445 446 447 448 449 450 451 452 453
	 *	  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)
	 */

454
	if (unix_tot_inflight)
455
		unix_gc();		/* Garbage collect fds */
L
Linus Torvalds 已提交
456 457
}

458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475
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
476 477 478 479 480
static int unix_listen(struct socket *sock, int backlog)
{
	int err;
	struct sock *sk = sock->sk;
	struct unix_sock *u = unix_sk(sk);
481
	struct pid *old_pid = NULL;
L
Linus Torvalds 已提交
482 483

	err = -EOPNOTSUPP;
E
Eric Dumazet 已提交
484 485
	if (sock->type != SOCK_STREAM && sock->type != SOCK_SEQPACKET)
		goto out;	/* Only stream/seqpacket sockets accept */
L
Linus Torvalds 已提交
486 487
	err = -EINVAL;
	if (!u->addr)
E
Eric Dumazet 已提交
488
		goto out;	/* No listens on an unbound socket */
489
	unix_state_lock(sk);
L
Linus Torvalds 已提交
490 491 492 493 494 495 496
	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 */
497
	init_peercred(sk);
L
Linus Torvalds 已提交
498 499 500
	err = 0;

out_unlock:
501
	unix_state_unlock(sk);
502
	put_pid(old_pid);
L
Linus Torvalds 已提交
503 504 505 506 507 508 509 510 511 512 513 514
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 *);
static int unix_accept(struct socket *, struct socket *, int);
static int unix_getname(struct socket *, struct sockaddr *, int *, int);
static unsigned int unix_poll(struct file *, struct socket *, poll_table *);
515 516
static unsigned int unix_dgram_poll(struct file *, struct socket *,
				    poll_table *);
L
Linus Torvalds 已提交
517 518
static int unix_ioctl(struct socket *, unsigned int, unsigned long);
static int unix_shutdown(struct socket *, int);
519 520
static int unix_stream_sendmsg(struct socket *, struct msghdr *, size_t);
static int unix_stream_recvmsg(struct socket *, struct msghdr *, size_t, int);
521 522
static ssize_t unix_stream_sendpage(struct socket *, struct page *, int offset,
				    size_t size, int flags);
523 524
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
525 526
static int unix_dgram_connect(struct socket *, struct sockaddr *,
			      int, int);
527 528 529
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
530

531
static int unix_set_peek_off(struct sock *sk, int val)
532 533 534
{
	struct unix_sock *u = unix_sk(sk);

535 536 537
	if (mutex_lock_interruptible(&u->readlock))
		return -EINTR;

538 539
	sk->sk_peek_off = val;
	mutex_unlock(&u->readlock);
540 541

	return 0;
542 543 544
}


545
static const struct proto_ops unix_stream_ops = {
L
Linus Torvalds 已提交
546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562
	.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,
	.poll =		unix_poll,
	.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,
563
	.sendpage =	unix_stream_sendpage,
564
	.set_peek_off =	unix_set_peek_off,
L
Linus Torvalds 已提交
565 566
};

567
static const struct proto_ops unix_dgram_ops = {
L
Linus Torvalds 已提交
568 569 570 571 572 573 574 575
	.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,
576
	.poll =		unix_dgram_poll,
L
Linus Torvalds 已提交
577 578 579 580 581 582 583 584 585
	.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,
586
	.set_peek_off =	unix_set_peek_off,
L
Linus Torvalds 已提交
587 588
};

589
static const struct proto_ops unix_seqpacket_ops = {
L
Linus Torvalds 已提交
590 591 592 593 594 595 596 597
	.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,
598
	.poll =		unix_dgram_poll,
L
Linus Torvalds 已提交
599 600 601 602 603 604
	.ioctl =	unix_ioctl,
	.listen =	unix_listen,
	.shutdown =	unix_shutdown,
	.setsockopt =	sock_no_setsockopt,
	.getsockopt =	sock_no_getsockopt,
	.sendmsg =	unix_seqpacket_sendmsg,
605
	.recvmsg =	unix_seqpacket_recvmsg,
L
Linus Torvalds 已提交
606 607
	.mmap =		sock_no_mmap,
	.sendpage =	sock_no_sendpage,
608
	.set_peek_off =	unix_set_peek_off,
L
Linus Torvalds 已提交
609 610 611
};

static struct proto unix_proto = {
612 613 614
	.name			= "UNIX",
	.owner			= THIS_MODULE,
	.obj_size		= sizeof(struct unix_sock),
L
Linus Torvalds 已提交
615 616
};

617 618 619 620 621 622 623 624
/*
 * AF_UNIX sockets do not interact with hardware, hence they
 * dont trigger interrupts - so it's safe for them to have
 * bh-unsafe locking for their sk_receive_queue.lock. Split off
 * this special lock-class by reinitializing the spinlock key:
 */
static struct lock_class_key af_unix_sk_receive_queue_lock_key;

625
static struct sock *unix_create1(struct net *net, struct socket *sock, int kern)
L
Linus Torvalds 已提交
626 627 628 629
{
	struct sock *sk = NULL;
	struct unix_sock *u;

E
Eric Dumazet 已提交
630 631
	atomic_long_inc(&unix_nr_socks);
	if (atomic_long_read(&unix_nr_socks) > 2 * get_max_files())
L
Linus Torvalds 已提交
632 633
		goto out;

634
	sk = sk_alloc(net, PF_UNIX, GFP_KERNEL, &unix_proto, kern);
L
Linus Torvalds 已提交
635 636 637
	if (!sk)
		goto out;

E
Eric Dumazet 已提交
638
	sock_init_data(sock, sk);
639 640
	lockdep_set_class(&sk->sk_receive_queue.lock,
				&af_unix_sk_receive_queue_lock_key);
L
Linus Torvalds 已提交
641 642

	sk->sk_write_space	= unix_write_space;
643
	sk->sk_max_ack_backlog	= net->unx.sysctl_max_dgram_qlen;
L
Linus Torvalds 已提交
644 645
	sk->sk_destruct		= unix_sock_destructor;
	u	  = unix_sk(sk);
A
Al Viro 已提交
646 647
	u->path.dentry = NULL;
	u->path.mnt = NULL;
648
	spin_lock_init(&u->lock);
A
Al Viro 已提交
649
	atomic_long_set(&u->inflight, 0);
650
	INIT_LIST_HEAD(&u->link);
I
Ingo Molnar 已提交
651
	mutex_init(&u->readlock); /* single task reading lock */
L
Linus Torvalds 已提交
652
	init_waitqueue_head(&u->peer_wait);
E
Eric Dumazet 已提交
653
	unix_insert_socket(unix_sockets_unbound(sk), sk);
L
Linus Torvalds 已提交
654
out:
655
	if (sk == NULL)
E
Eric Dumazet 已提交
656
		atomic_long_dec(&unix_nr_socks);
657 658
	else {
		local_bh_disable();
659
		sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
660 661
		local_bh_enable();
	}
L
Linus Torvalds 已提交
662 663 664
	return sk;
}

665 666
static int unix_create(struct net *net, struct socket *sock, int protocol,
		       int kern)
L
Linus Torvalds 已提交
667 668 669 670 671 672 673 674 675 676 677 678 679 680 681
{
	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:
682
		sock->type = SOCK_DGRAM;
L
Linus Torvalds 已提交
683 684 685 686 687 688 689 690 691 692
	case SOCK_DGRAM:
		sock->ops = &unix_dgram_ops;
		break;
	case SOCK_SEQPACKET:
		sock->ops = &unix_seqpacket_ops;
		break;
	default:
		return -ESOCKTNOSUPPORT;
	}

693
	return unix_create1(net, sock, kern) ? 0 : -ENOMEM;
L
Linus Torvalds 已提交
694 695 696 697 698 699 700 701 702
}

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

	if (!sk)
		return 0;

703
	unix_release_sock(sk, 0);
L
Linus Torvalds 已提交
704 705
	sock->sk = NULL;

706
	return 0;
L
Linus Torvalds 已提交
707 708 709 710 711
}

static int unix_autobind(struct socket *sock)
{
	struct sock *sk = sock->sk;
712
	struct net *net = sock_net(sk);
L
Linus Torvalds 已提交
713 714
	struct unix_sock *u = unix_sk(sk);
	static u32 ordernum = 1;
E
Eric Dumazet 已提交
715
	struct unix_address *addr;
L
Linus Torvalds 已提交
716
	int err;
717
	unsigned int retries = 0;
L
Linus Torvalds 已提交
718

719 720 721
	err = mutex_lock_interruptible(&u->readlock);
	if (err)
		return err;
L
Linus Torvalds 已提交
722 723 724 725 726 727

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

	err = -ENOMEM;
728
	addr = kzalloc(sizeof(*addr) + sizeof(short) + 16, GFP_KERNEL);
L
Linus Torvalds 已提交
729 730 731 732 733 734 735 736
	if (!addr)
		goto out;

	addr->name->sun_family = AF_UNIX;
	atomic_set(&addr->refcnt, 1);

retry:
	addr->len = sprintf(addr->name->sun_path+1, "%05x", ordernum) + 1 + sizeof(short);
737
	addr->hash = unix_hash_fold(csum_partial(addr->name, addr->len, 0));
L
Linus Torvalds 已提交
738

739
	spin_lock(&unix_table_lock);
L
Linus Torvalds 已提交
740 741
	ordernum = (ordernum+1)&0xFFFFF;

742
	if (__unix_find_socket_byname(net, addr->name, addr->len, sock->type,
L
Linus Torvalds 已提交
743
				      addr->hash)) {
744
		spin_unlock(&unix_table_lock);
745 746 747 748 749 750 751 752 753 754 755
		/*
		 * __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
756 757 758 759 760 761 762
		goto retry;
	}
	addr->hash ^= sk->sk_type;

	__unix_remove_socket(sk);
	u->addr = addr;
	__unix_insert_socket(&unix_socket_table[addr->hash], sk);
763
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
764 765
	err = 0;

I
Ingo Molnar 已提交
766
out:	mutex_unlock(&u->readlock);
L
Linus Torvalds 已提交
767 768 769
	return err;
}

770 771
static struct sock *unix_find_other(struct net *net,
				    struct sockaddr_un *sunname, int len,
772
				    int type, unsigned int hash, int *error)
L
Linus Torvalds 已提交
773 774
{
	struct sock *u;
775
	struct path path;
L
Linus Torvalds 已提交
776
	int err = 0;
777

L
Linus Torvalds 已提交
778
	if (sunname->sun_path[0]) {
779 780
		struct inode *inode;
		err = kern_path(sunname->sun_path, LOOKUP_FOLLOW, &path);
L
Linus Torvalds 已提交
781 782
		if (err)
			goto fail;
783
		inode = d_backing_inode(path.dentry);
784
		err = inode_permission(inode, MAY_WRITE);
L
Linus Torvalds 已提交
785 786 787 788
		if (err)
			goto put_fail;

		err = -ECONNREFUSED;
789
		if (!S_ISSOCK(inode->i_mode))
L
Linus Torvalds 已提交
790
			goto put_fail;
791
		u = unix_find_socket_byinode(inode);
L
Linus Torvalds 已提交
792 793 794 795
		if (!u)
			goto put_fail;

		if (u->sk_type == type)
A
Al Viro 已提交
796
			touch_atime(&path);
L
Linus Torvalds 已提交
797

798
		path_put(&path);
L
Linus Torvalds 已提交
799

800
		err = -EPROTOTYPE;
L
Linus Torvalds 已提交
801 802 803 804 805 806
		if (u->sk_type != type) {
			sock_put(u);
			goto fail;
		}
	} else {
		err = -ECONNREFUSED;
807
		u = unix_find_socket_byname(net, sunname, len, type, hash);
L
Linus Torvalds 已提交
808 809
		if (u) {
			struct dentry *dentry;
A
Al Viro 已提交
810
			dentry = unix_sk(u)->path.dentry;
L
Linus Torvalds 已提交
811
			if (dentry)
A
Al Viro 已提交
812
				touch_atime(&unix_sk(u)->path);
L
Linus Torvalds 已提交
813 814 815 816 817 818
		} else
			goto fail;
	}
	return u;

put_fail:
819
	path_put(&path);
L
Linus Torvalds 已提交
820
fail:
821
	*error = err;
L
Linus Torvalds 已提交
822 823 824
	return NULL;
}

A
Al Viro 已提交
825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843
static int unix_mknod(const char *sun_path, umode_t mode, struct path *res)
{
	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;

	/*
	 * All right, let's create it.
	 */
	err = security_path_mknod(&path, dentry, mode, 0);
	if (!err) {
844
		err = vfs_mknod(d_inode(path.dentry), dentry, mode, 0);
A
Al Viro 已提交
845 846 847 848 849 850 851 852
		if (!err) {
			res->mnt = mntget(path.mnt);
			res->dentry = dget(dentry);
		}
	}
	done_path_create(&path, dentry);
	return err;
}
L
Linus Torvalds 已提交
853 854 855 856

static int unix_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
{
	struct sock *sk = sock->sk;
857
	struct net *net = sock_net(sk);
L
Linus Torvalds 已提交
858
	struct unix_sock *u = unix_sk(sk);
859
	struct sockaddr_un *sunaddr = (struct sockaddr_un *)uaddr;
860
	char *sun_path = sunaddr->sun_path;
L
Linus Torvalds 已提交
861
	int err;
862
	unsigned int hash;
L
Linus Torvalds 已提交
863 864 865 866 867 868 869
	struct unix_address *addr;
	struct hlist_head *list;

	err = -EINVAL;
	if (sunaddr->sun_family != AF_UNIX)
		goto out;

870
	if (addr_len == sizeof(short)) {
L
Linus Torvalds 已提交
871 872 873 874 875 876 877 878 879
		err = unix_autobind(sock);
		goto out;
	}

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

880 881 882
	err = mutex_lock_interruptible(&u->readlock);
	if (err)
		goto out;
L
Linus Torvalds 已提交
883 884 885 886 887 888 889 890 891 892 893 894 895 896 897

	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;
	atomic_set(&addr->refcnt, 1);

898
	if (sun_path[0]) {
A
Al Viro 已提交
899 900
		struct path path;
		umode_t mode = S_IFSOCK |
A
Al Viro 已提交
901
		       (SOCK_INODE(sock)->i_mode & ~current_umask());
A
Al Viro 已提交
902 903 904 905 906 907 908
		err = unix_mknod(sun_path, mode, &path);
		if (err) {
			if (err == -EEXIST)
				err = -EADDRINUSE;
			unix_release_addr(addr);
			goto out_up;
		}
L
Linus Torvalds 已提交
909
		addr->hash = UNIX_HASH_SIZE;
910
		hash = d_backing_inode(path.dentry)->i_ino & (UNIX_HASH_SIZE-1);
A
Al Viro 已提交
911 912 913 914 915
		spin_lock(&unix_table_lock);
		u->path = path;
		list = &unix_socket_table[hash];
	} else {
		spin_lock(&unix_table_lock);
L
Linus Torvalds 已提交
916
		err = -EADDRINUSE;
917
		if (__unix_find_socket_byname(net, sunaddr, addr_len,
L
Linus Torvalds 已提交
918 919 920 921 922 923 924 925 926 927 928 929 930 931
					      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:
932
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
933
out_up:
I
Ingo Molnar 已提交
934
	mutex_unlock(&u->readlock);
L
Linus Torvalds 已提交
935 936 937 938
out:
	return err;
}

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
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
964 965 966 967
static int unix_dgram_connect(struct socket *sock, struct sockaddr *addr,
			      int alen, int flags)
{
	struct sock *sk = sock->sk;
968
	struct net *net = sock_net(sk);
969
	struct sockaddr_un *sunaddr = (struct sockaddr_un *)addr;
L
Linus Torvalds 已提交
970
	struct sock *other;
971
	unsigned int hash;
L
Linus Torvalds 已提交
972 973 974 975 976 977 978 979 980 981 982 983
	int err;

	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;

984
restart:
985
		other = unix_find_other(net, sunaddr, alen, sock->type, hash, &err);
L
Linus Torvalds 已提交
986 987 988
		if (!other)
			goto out;

989 990 991 992 993 994 995 996
		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
997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010

		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;
1011
		unix_state_double_lock(sk, other);
L
Linus Torvalds 已提交
1012 1013 1014 1015 1016 1017 1018
	}

	/*
	 * If it was connected, reconnect.
	 */
	if (unix_peer(sk)) {
		struct sock *old_peer = unix_peer(sk);
1019
		unix_peer(sk) = other;
1020
		unix_state_double_unlock(sk, other);
L
Linus Torvalds 已提交
1021 1022 1023 1024 1025

		if (other != old_peer)
			unix_dgram_disconnected(sk, old_peer);
		sock_put(old_peer);
	} else {
1026
		unix_peer(sk) = other;
1027
		unix_state_double_unlock(sk, other);
L
Linus Torvalds 已提交
1028
	}
1029
	return 0;
L
Linus Torvalds 已提交
1030 1031

out_unlock:
1032
	unix_state_double_unlock(sk, other);
L
Linus Torvalds 已提交
1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047
	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) &&
1048
		unix_recvq_full(other);
L
Linus Torvalds 已提交
1049

1050
	unix_state_unlock(other);
L
Linus Torvalds 已提交
1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061

	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)
{
1062
	struct sockaddr_un *sunaddr = (struct sockaddr_un *)uaddr;
L
Linus Torvalds 已提交
1063
	struct sock *sk = sock->sk;
1064
	struct net *net = sock_net(sk);
L
Linus Torvalds 已提交
1065 1066 1067 1068
	struct unix_sock *u = unix_sk(sk), *newu, *otheru;
	struct sock *newsk = NULL;
	struct sock *other = NULL;
	struct sk_buff *skb = NULL;
1069
	unsigned int hash;
L
Linus Torvalds 已提交
1070 1071 1072 1073 1074 1075 1076 1077 1078
	int st;
	int err;
	long timeo;

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

1079 1080
	if (test_bit(SOCK_PASSCRED, &sock->flags) && !u->addr &&
	    (err = unix_autobind(sock)) != 0)
L
Linus Torvalds 已提交
1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092
		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 */
1093
	newsk = unix_create1(sock_net(sk), NULL, 0);
L
Linus Torvalds 已提交
1094 1095 1096 1097 1098 1099 1100 1101 1102 1103
	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. */
1104
	other = unix_find_other(net, sunaddr, addr_len, sk->sk_type, hash, &err);
L
Linus Torvalds 已提交
1105 1106 1107 1108
	if (!other)
		goto out;

	/* Latch state of peer */
1109
	unix_state_lock(other);
L
Linus Torvalds 已提交
1110 1111 1112

	/* Apparently VFS overslept socket death. Retry. */
	if (sock_flag(other, SOCK_DEAD)) {
1113
		unix_state_unlock(other);
L
Linus Torvalds 已提交
1114 1115 1116 1117 1118 1119 1120
		sock_put(other);
		goto restart;
	}

	err = -ECONNREFUSED;
	if (other->sk_state != TCP_LISTEN)
		goto out_unlock;
1121 1122
	if (other->sk_shutdown & RCV_SHUTDOWN)
		goto out_unlock;
L
Linus Torvalds 已提交
1123

1124
	if (unix_recvq_full(other)) {
L
Linus Torvalds 已提交
1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135
		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;
1136
	}
L
Linus Torvalds 已提交
1137 1138 1139

	/* Latch our state.

D
Daniel Baluta 已提交
1140
	   It is tricky place. We need to grab our state lock and cannot
L
Linus Torvalds 已提交
1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163
	   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;
	}

1164
	unix_state_lock_nested(sk);
L
Linus Torvalds 已提交
1165 1166

	if (sk->sk_state != st) {
1167 1168
		unix_state_unlock(sk);
		unix_state_unlock(other);
L
Linus Torvalds 已提交
1169 1170 1171 1172
		sock_put(other);
		goto restart;
	}

1173
	err = security_unix_stream_connect(sk, other, newsk);
L
Linus Torvalds 已提交
1174
	if (err) {
1175
		unix_state_unlock(sk);
L
Linus Torvalds 已提交
1176 1177 1178 1179 1180 1181 1182 1183 1184
		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;
1185
	init_peercred(newsk);
L
Linus Torvalds 已提交
1186
	newu = unix_sk(newsk);
1187
	RCU_INIT_POINTER(newsk->sk_wq, &newu->peer_wq);
L
Linus Torvalds 已提交
1188 1189 1190 1191 1192 1193 1194
	otheru = unix_sk(other);

	/* copy address information from listening to new sock*/
	if (otheru->addr) {
		atomic_inc(&otheru->addr->refcnt);
		newu->addr = otheru->addr;
	}
A
Al Viro 已提交
1195 1196 1197
	if (otheru->path.dentry) {
		path_get(&otheru->path);
		newu->path = otheru->path;
L
Linus Torvalds 已提交
1198 1199 1200
	}

	/* Set credentials */
1201
	copy_peercred(sk, other);
L
Linus Torvalds 已提交
1202 1203 1204

	sock->state	= SS_CONNECTED;
	sk->sk_state	= TCP_ESTABLISHED;
1205 1206
	sock_hold(newsk);

1207
	smp_mb__after_atomic();	/* sock_hold() does an atomic_inc() */
1208
	unix_peer(sk)	= newsk;
L
Linus Torvalds 已提交
1209

1210
	unix_state_unlock(sk);
L
Linus Torvalds 已提交
1211 1212 1213 1214 1215

	/* 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);
1216
	unix_state_unlock(other);
1217
	other->sk_data_ready(other);
L
Linus Torvalds 已提交
1218 1219 1220 1221 1222
	sock_put(other);
	return 0;

out_unlock:
	if (other)
1223
		unix_state_unlock(other);
L
Linus Torvalds 已提交
1224 1225

out:
1226
	kfree_skb(skb);
L
Linus Torvalds 已提交
1227 1228 1229 1230 1231 1232 1233 1234 1235
	if (newsk)
		unix_release_sock(newsk, 0);
	if (other)
		sock_put(other);
	return err;
}

static int unix_socketpair(struct socket *socka, struct socket *sockb)
{
1236
	struct sock *ska = socka->sk, *skb = sockb->sk;
L
Linus Torvalds 已提交
1237 1238 1239 1240

	/* Join our sockets back to back */
	sock_hold(ska);
	sock_hold(skb);
1241 1242
	unix_peer(ska) = skb;
	unix_peer(skb) = ska;
1243 1244
	init_peercred(ska);
	init_peercred(skb);
L
Linus Torvalds 已提交
1245 1246 1247 1248 1249 1250 1251 1252 1253 1254

	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;
}

1255 1256 1257 1258 1259 1260 1261 1262 1263
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);
}

L
Linus Torvalds 已提交
1264 1265 1266 1267 1268 1269 1270 1271
static int unix_accept(struct socket *sock, struct socket *newsock, int flags)
{
	struct sock *sk = sock->sk;
	struct sock *tsk;
	struct sk_buff *skb;
	int err;

	err = -EOPNOTSUPP;
E
Eric Dumazet 已提交
1272
	if (sock->type != SOCK_STREAM && sock->type != SOCK_SEQPACKET)
L
Linus Torvalds 已提交
1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295
		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 */
1296
	unix_state_lock(tsk);
L
Linus Torvalds 已提交
1297
	newsock->state = SS_CONNECTED;
1298
	unix_sock_inherit_flags(sock, newsock);
L
Linus Torvalds 已提交
1299
	sock_graft(tsk, newsock);
1300
	unix_state_unlock(tsk);
L
Linus Torvalds 已提交
1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311
	return 0;

out:
	return err;
}


static int unix_getname(struct socket *sock, struct sockaddr *uaddr, int *uaddr_len, int peer)
{
	struct sock *sk = sock->sk;
	struct unix_sock *u;
1312
	DECLARE_SOCKADDR(struct sockaddr_un *, sunaddr, uaddr);
L
Linus Torvalds 已提交
1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326
	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);
1327
	unix_state_lock(sk);
L
Linus Torvalds 已提交
1328 1329 1330 1331 1332 1333 1334 1335 1336 1337
	if (!u->addr) {
		sunaddr->sun_family = AF_UNIX;
		sunaddr->sun_path[0] = 0;
		*uaddr_len = sizeof(short);
	} else {
		struct unix_address *addr = u->addr;

		*uaddr_len = addr->len;
		memcpy(sunaddr, addr->name, *uaddr_len);
	}
1338
	unix_state_unlock(sk);
L
Linus Torvalds 已提交
1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350
	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
1351
	for (i = scm->fp->count-1; i >= 0; i--)
L
Linus Torvalds 已提交
1352 1353 1354
		unix_notinflight(scm->fp->fp[i]);
}

1355
static void unix_destruct_scm(struct sk_buff *skb)
L
Linus Torvalds 已提交
1356 1357 1358
{
	struct scm_cookie scm;
	memset(&scm, 0, sizeof(scm));
1359 1360 1361
	scm.pid  = UNIXCB(skb).pid;
	if (UNIXCB(skb).fp)
		unix_detach_fds(&scm, skb);
L
Linus Torvalds 已提交
1362 1363 1364 1365 1366 1367 1368

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

E
Eric Dumazet 已提交
1369 1370
#define MAX_RECURSION_LEVEL 4

1371
static int unix_attach_fds(struct scm_cookie *scm, struct sk_buff *skb)
L
Linus Torvalds 已提交
1372 1373
{
	int i;
E
Eric Dumazet 已提交
1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387
	unsigned char max_level = 0;
	int unix_sock_count = 0;

	for (i = scm->fp->count - 1; i >= 0; i--) {
		struct sock *sk = unix_get_socket(scm->fp->fp[i]);

		if (sk) {
			unix_sock_count++;
			max_level = max(max_level,
					unix_sk(sk)->recursion_level);
		}
	}
	if (unlikely(max_level > MAX_RECURSION_LEVEL))
		return -ETOOMANYREFS;
1388 1389 1390 1391 1392 1393 1394 1395 1396 1397

	/*
	 * 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;

E
Eric Dumazet 已提交
1398 1399 1400 1401 1402
	if (unix_sock_count) {
		for (i = scm->fp->count - 1; i >= 0; i--)
			unix_inflight(scm->fp->fp[i]);
	}
	return max_level;
L
Linus Torvalds 已提交
1403 1404
}

1405
static int unix_scm_to_skb(struct scm_cookie *scm, struct sk_buff *skb, bool send_fds)
1406 1407
{
	int err = 0;
1408

1409
	UNIXCB(skb).pid  = get_pid(scm->pid);
1410 1411
	UNIXCB(skb).uid = scm->creds.uid;
	UNIXCB(skb).gid = scm->creds.gid;
1412 1413 1414 1415 1416 1417 1418 1419
	UNIXCB(skb).fp = NULL;
	if (scm->fp && send_fds)
		err = unix_attach_fds(scm, skb);

	skb->destructor = unix_destruct_scm;
	return err;
}

1420 1421 1422 1423 1424 1425 1426 1427
/*
 * 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)
{
1428
	if (UNIXCB(skb).pid)
1429 1430
		return;
	if (test_bit(SOCK_PASSCRED, &sock->flags) ||
1431 1432
	    !other->sk_socket ||
	    test_bit(SOCK_PASSCRED, &other->sk_socket->flags)) {
1433
		UNIXCB(skb).pid  = get_pid(task_tgid(current));
1434
		current_uid_gid(&UNIXCB(skb).uid, &UNIXCB(skb).gid);
1435 1436 1437
	}
}

L
Linus Torvalds 已提交
1438 1439 1440 1441
/*
 *	Send AF_UNIX data.
 */

1442 1443
static int unix_dgram_sendmsg(struct socket *sock, struct msghdr *msg,
			      size_t len)
L
Linus Torvalds 已提交
1444 1445
{
	struct sock *sk = sock->sk;
1446
	struct net *net = sock_net(sk);
L
Linus Torvalds 已提交
1447
	struct unix_sock *u = unix_sk(sk);
1448
	DECLARE_SOCKADDR(struct sockaddr_un *, sunaddr, msg->msg_name);
L
Linus Torvalds 已提交
1449 1450 1451
	struct sock *other = NULL;
	int namelen = 0; /* fake GCC */
	int err;
1452
	unsigned int hash;
1453
	struct sk_buff *skb;
L
Linus Torvalds 已提交
1454
	long timeo;
C
Christoph Hellwig 已提交
1455
	struct scm_cookie scm;
E
Eric Dumazet 已提交
1456
	int max_level;
1457
	int data_len = 0;
L
Linus Torvalds 已提交
1458

1459
	wait_for_unix_gc();
C
Christoph Hellwig 已提交
1460
	err = scm_send(sock, msg, &scm, false);
L
Linus Torvalds 已提交
1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480
	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;
	}

1481 1482
	if (test_bit(SOCK_PASSCRED, &sock->flags) && !u->addr
	    && (err = unix_autobind(sock)) != 0)
L
Linus Torvalds 已提交
1483 1484 1485 1486 1487 1488
		goto out;

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

1489
	if (len > SKB_MAX_ALLOC) {
1490 1491 1492
		data_len = min_t(size_t,
				 len - SKB_MAX_ALLOC,
				 MAX_SKB_FRAGS * PAGE_SIZE);
1493 1494 1495 1496
		data_len = PAGE_ALIGN(data_len);

		BUILD_BUG_ON(SKB_MAX_ALLOC < PAGE_SIZE);
	}
1497 1498

	skb = sock_alloc_send_pskb(sk, len - data_len, data_len,
1499 1500
				   msg->msg_flags & MSG_DONTWAIT, &err,
				   PAGE_ALLOC_COSTLY_ORDER);
1501
	if (skb == NULL)
L
Linus Torvalds 已提交
1502 1503
		goto out;

C
Christoph Hellwig 已提交
1504
	err = unix_scm_to_skb(&scm, skb, true);
E
Eric Dumazet 已提交
1505
	if (err < 0)
1506
		goto out_free;
E
Eric Dumazet 已提交
1507
	max_level = err + 1;
C
Christoph Hellwig 已提交
1508
	unix_get_secdata(&scm, skb);
C
Catherine Zhang 已提交
1509

1510 1511 1512
	skb_put(skb, len - data_len);
	skb->data_len = data_len;
	skb->len = len;
A
Al Viro 已提交
1513
	err = skb_copy_datagram_from_iter(skb, 0, &msg->msg_iter, len);
L
Linus Torvalds 已提交
1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524
	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;

1525
		other = unix_find_other(net, sunaddr, namelen, sk->sk_type,
L
Linus Torvalds 已提交
1526
					hash, &err);
1527
		if (other == NULL)
L
Linus Torvalds 已提交
1528 1529 1530
			goto out_free;
	}

A
Alban Crequy 已提交
1531 1532 1533 1534 1535 1536
	if (sk_filter(other, skb) < 0) {
		/* Toss the packet but do not return any error to the sender */
		err = len;
		goto out_free;
	}

1537
	unix_state_lock(other);
L
Linus Torvalds 已提交
1538 1539 1540 1541 1542 1543 1544 1545 1546
	err = -EPERM;
	if (!unix_may_send(sk, other))
		goto out_unlock;

	if (sock_flag(other, SOCK_DEAD)) {
		/*
		 *	Check with 1003.1g - what should
		 *	datagram error
		 */
1547
		unix_state_unlock(other);
L
Linus Torvalds 已提交
1548 1549 1550
		sock_put(other);

		err = 0;
1551
		unix_state_lock(sk);
L
Linus Torvalds 已提交
1552
		if (unix_peer(sk) == other) {
1553
			unix_peer(sk) = NULL;
1554
			unix_state_unlock(sk);
L
Linus Torvalds 已提交
1555 1556 1557 1558 1559

			unix_dgram_disconnected(sk, other);
			sock_put(other);
			err = -ECONNREFUSED;
		} else {
1560
			unix_state_unlock(sk);
L
Linus Torvalds 已提交
1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578
		}

		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;
	}

1579
	if (unix_peer(other) != sk && unix_recvq_full(other)) {
L
Linus Torvalds 已提交
1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593
		if (!timeo) {
			err = -EAGAIN;
			goto out_unlock;
		}

		timeo = unix_wait_for_peer(other, timeo);

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

		goto restart;
	}

1594 1595
	if (sock_flag(other, SOCK_RCVTSTAMP))
		__net_timestamp(skb);
1596
	maybe_add_creds(skb, sock, other);
L
Linus Torvalds 已提交
1597
	skb_queue_tail(&other->sk_receive_queue, skb);
E
Eric Dumazet 已提交
1598 1599
	if (max_level > unix_sk(other)->recursion_level)
		unix_sk(other)->recursion_level = max_level;
1600
	unix_state_unlock(other);
1601
	other->sk_data_ready(other);
L
Linus Torvalds 已提交
1602
	sock_put(other);
C
Christoph Hellwig 已提交
1603
	scm_destroy(&scm);
L
Linus Torvalds 已提交
1604 1605 1606
	return len;

out_unlock:
1607
	unix_state_unlock(other);
L
Linus Torvalds 已提交
1608 1609 1610 1611 1612
out_free:
	kfree_skb(skb);
out:
	if (other)
		sock_put(other);
C
Christoph Hellwig 已提交
1613
	scm_destroy(&scm);
L
Linus Torvalds 已提交
1614 1615 1616
	return err;
}

1617 1618 1619 1620
/* We use paged skbs for stream sockets, and limit occupancy to 32768
 * bytes, and a minimun of a full page.
 */
#define UNIX_SKB_FRAGS_SZ (PAGE_SIZE << get_order(32768))
1621

1622 1623
static int unix_stream_sendmsg(struct socket *sock, struct msghdr *msg,
			       size_t len)
L
Linus Torvalds 已提交
1624 1625 1626
{
	struct sock *sk = sock->sk;
	struct sock *other = NULL;
E
Eric Dumazet 已提交
1627
	int err, size;
1628
	struct sk_buff *skb;
1629
	int sent = 0;
C
Christoph Hellwig 已提交
1630
	struct scm_cookie scm;
1631
	bool fds_sent = false;
E
Eric Dumazet 已提交
1632
	int max_level;
1633
	int data_len;
L
Linus Torvalds 已提交
1634

1635
	wait_for_unix_gc();
C
Christoph Hellwig 已提交
1636
	err = scm_send(sock, msg, &scm, false);
L
Linus Torvalds 已提交
1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648
	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;
1649
		other = unix_peer(sk);
L
Linus Torvalds 已提交
1650 1651 1652 1653 1654 1655 1656
		if (!other)
			goto out_err;
	}

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

E
Eric Dumazet 已提交
1657
	while (sent < len) {
1658
		size = len - sent;
L
Linus Torvalds 已提交
1659 1660

		/* Keep two messages in the pipe so it schedules better */
1661
		size = min_t(int, size, (sk->sk_sndbuf >> 1) - 64);
L
Linus Torvalds 已提交
1662

1663 1664
		/* allow fallback to order-0 allocations */
		size = min_t(int, size, SKB_MAX_HEAD(0) + UNIX_SKB_FRAGS_SZ);
1665

1666
		data_len = max_t(int, 0, size - SKB_MAX_HEAD(0));
L
Linus Torvalds 已提交
1667

1668 1669
		data_len = min_t(size_t, size, PAGE_ALIGN(data_len));

1670
		skb = sock_alloc_send_pskb(sk, size - data_len, data_len,
1671 1672
					   msg->msg_flags & MSG_DONTWAIT, &err,
					   get_order(UNIX_SKB_FRAGS_SZ));
1673
		if (!skb)
L
Linus Torvalds 已提交
1674 1675
			goto out_err;

1676
		/* Only send the fds in the first buffer */
C
Christoph Hellwig 已提交
1677
		err = unix_scm_to_skb(&scm, skb, !fds_sent);
E
Eric Dumazet 已提交
1678
		if (err < 0) {
1679
			kfree_skb(skb);
1680
			goto out_err;
1681
		}
E
Eric Dumazet 已提交
1682
		max_level = err + 1;
1683
		fds_sent = true;
L
Linus Torvalds 已提交
1684

1685 1686 1687
		skb_put(skb, size - data_len);
		skb->data_len = data_len;
		skb->len = size;
A
Al Viro 已提交
1688
		err = skb_copy_datagram_from_iter(skb, 0, &msg->msg_iter, size);
E
Eric Dumazet 已提交
1689
		if (err) {
L
Linus Torvalds 已提交
1690
			kfree_skb(skb);
1691
			goto out_err;
L
Linus Torvalds 已提交
1692 1693
		}

1694
		unix_state_lock(other);
L
Linus Torvalds 已提交
1695 1696 1697 1698 1699

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

1700
		maybe_add_creds(skb, sock, other);
L
Linus Torvalds 已提交
1701
		skb_queue_tail(&other->sk_receive_queue, skb);
E
Eric Dumazet 已提交
1702 1703
		if (max_level > unix_sk(other)->recursion_level)
			unix_sk(other)->recursion_level = max_level;
1704
		unix_state_unlock(other);
1705
		other->sk_data_ready(other);
1706
		sent += size;
L
Linus Torvalds 已提交
1707 1708
	}

C
Christoph Hellwig 已提交
1709
	scm_destroy(&scm);
L
Linus Torvalds 已提交
1710 1711 1712 1713

	return sent;

pipe_err_free:
1714
	unix_state_unlock(other);
L
Linus Torvalds 已提交
1715 1716
	kfree_skb(skb);
pipe_err:
E
Eric Dumazet 已提交
1717 1718
	if (sent == 0 && !(msg->msg_flags&MSG_NOSIGNAL))
		send_sig(SIGPIPE, current, 0);
L
Linus Torvalds 已提交
1719 1720
	err = -EPIPE;
out_err:
C
Christoph Hellwig 已提交
1721
	scm_destroy(&scm);
L
Linus Torvalds 已提交
1722 1723 1724
	return sent ? : err;
}

1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819
static ssize_t unix_stream_sendpage(struct socket *socket, struct page *page,
				    int offset, size_t size, int flags)
{
	int err = 0;
	bool send_sigpipe = true;
	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);
		mutex_unlock(&unix_sk(other)->readlock);
		newskb = sock_alloc_send_pskb(sk, 0, 0, flags & MSG_DONTWAIT,
					      &err, 0);
		if (!newskb)
			return err;
	}

	/* we must acquire readlock as we modify already present
	 * skbs in the sk_receive_queue and mess with skb->len
	 */
	err = mutex_lock_interruptible(&unix_sk(other)->readlock);
	if (err) {
		err = flags & MSG_DONTWAIT ? -EAGAIN : -ERESTARTSYS;
		send_sigpipe = false;
		goto err;
	}

	if (sk->sk_shutdown & SEND_SHUTDOWN) {
		err = -EPIPE;
		goto err_unlock;
	}

	unix_state_lock(other);

	if (sock_flag(other, SOCK_DEAD) ||
	    other->sk_shutdown & RCV_SHUTDOWN) {
		err = -EPIPE;
		goto err_state_unlock;
	}

	skb = skb_peek_tail(&other->sk_receive_queue);
	if (tail && tail == skb) {
		skb = newskb;
	} else if (!skb) {
		if (newskb)
			skb = newskb;
		else
			goto alloc_skb;
	} 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);
	}

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

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

	if (newskb)
		__skb_queue_tail(&other->sk_receive_queue, newskb);

	unix_state_unlock(other);
	mutex_unlock(&unix_sk(other)->readlock);

	other->sk_data_ready(other);

	return size;

err_state_unlock:
	unix_state_unlock(other);
err_unlock:
	mutex_unlock(&unix_sk(other)->readlock);
err:
	kfree_skb(newskb);
	if (send_sigpipe && !(flags & MSG_NOSIGNAL))
		send_sig(SIGPIPE, current, 0);
	return err;
}

1820 1821
static int unix_seqpacket_sendmsg(struct socket *sock, struct msghdr *msg,
				  size_t len)
L
Linus Torvalds 已提交
1822 1823 1824
{
	int err;
	struct sock *sk = sock->sk;
1825

L
Linus Torvalds 已提交
1826 1827 1828 1829 1830 1831 1832 1833 1834 1835
	err = sock_error(sk);
	if (err)
		return err;

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

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

1836
	return unix_dgram_sendmsg(sock, msg, len);
L
Linus Torvalds 已提交
1837
}
1838

1839 1840
static int unix_seqpacket_recvmsg(struct socket *sock, struct msghdr *msg,
				  size_t size, int flags)
1841 1842 1843 1844 1845 1846
{
	struct sock *sk = sock->sk;

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

1847
	return unix_dgram_recvmsg(sock, msg, size, flags);
1848 1849
}

L
Linus Torvalds 已提交
1850 1851 1852 1853 1854 1855 1856 1857 1858 1859
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);
	}
}

1860 1861
static int unix_dgram_recvmsg(struct socket *sock, struct msghdr *msg,
			      size_t size, int flags)
L
Linus Torvalds 已提交
1862
{
C
Christoph Hellwig 已提交
1863
	struct scm_cookie scm;
L
Linus Torvalds 已提交
1864 1865 1866 1867 1868
	struct sock *sk = sock->sk;
	struct unix_sock *u = unix_sk(sk);
	int noblock = flags & MSG_DONTWAIT;
	struct sk_buff *skb;
	int err;
1869
	int peeked, skip;
L
Linus Torvalds 已提交
1870 1871 1872 1873 1874

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

1875
	err = mutex_lock_interruptible(&u->readlock);
1876 1877 1878 1879 1880
	if (unlikely(err)) {
		/* recvmsg() in non blocking mode is supposed to return -EAGAIN
		 * sk_rcvtimeo is not honored by mutex_lock_interruptible()
		 */
		err = noblock ? -EAGAIN : -ERESTARTSYS;
1881 1882
		goto out;
	}
L
Linus Torvalds 已提交
1883

1884 1885 1886
	skip = sk_peek_offset(sk, flags);

	skb = __skb_recv_datagram(sk, flags, &peeked, &skip, &err);
1887 1888 1889 1890 1891 1892 1893
	if (!skb) {
		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);
L
Linus Torvalds 已提交
1894
		goto out_unlock;
1895
	}
L
Linus Torvalds 已提交
1896

E
Eric Dumazet 已提交
1897 1898
	wake_up_interruptible_sync_poll(&u->peer_wait,
					POLLOUT | POLLWRNORM | POLLWRBAND);
L
Linus Torvalds 已提交
1899 1900 1901 1902

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

1903 1904 1905
	if (size > skb->len - skip)
		size = skb->len - skip;
	else if (size < skb->len - skip)
L
Linus Torvalds 已提交
1906 1907
		msg->msg_flags |= MSG_TRUNC;

1908
	err = skb_copy_datagram_msg(skb, skip, msg, size);
L
Linus Torvalds 已提交
1909 1910 1911
	if (err)
		goto out_free;

1912 1913 1914
	if (sock_flag(sk, SOCK_RCVTSTAMP))
		__sock_recv_timestamp(msg, sk, skb);

C
Christoph Hellwig 已提交
1915 1916 1917 1918
	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
1919

E
Eric Dumazet 已提交
1920
	if (!(flags & MSG_PEEK)) {
L
Linus Torvalds 已提交
1921
		if (UNIXCB(skb).fp)
C
Christoph Hellwig 已提交
1922
			unix_detach_fds(&scm, skb);
1923 1924

		sk_peek_offset_bwd(sk, skb->len);
E
Eric Dumazet 已提交
1925
	} else {
L
Linus Torvalds 已提交
1926 1927 1928 1929 1930 1931
		/* 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)
1932 1933 1934 1935 1936

		   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
1937
		*/
1938 1939 1940

		sk_peek_offset_fwd(sk, size);

L
Linus Torvalds 已提交
1941
		if (UNIXCB(skb).fp)
C
Christoph Hellwig 已提交
1942
			scm.fp = scm_fp_dup(UNIXCB(skb).fp);
L
Linus Torvalds 已提交
1943
	}
1944
	err = (flags & MSG_TRUNC) ? skb->len - skip : size;
L
Linus Torvalds 已提交
1945

C
Christoph Hellwig 已提交
1946
	scm_recv(sock, msg, &scm, flags);
L
Linus Torvalds 已提交
1947 1948

out_free:
E
Eric Dumazet 已提交
1949
	skb_free_datagram(sk, skb);
L
Linus Torvalds 已提交
1950
out_unlock:
I
Ingo Molnar 已提交
1951
	mutex_unlock(&u->readlock);
L
Linus Torvalds 已提交
1952 1953 1954 1955 1956
out:
	return err;
}

/*
1957
 *	Sleep until more data has arrived. But check for races..
L
Linus Torvalds 已提交
1958
 */
1959 1960
static long unix_stream_data_wait(struct sock *sk, long timeo,
				  struct sk_buff *last)
L
Linus Torvalds 已提交
1961 1962 1963
{
	DEFINE_WAIT(wait);

1964
	unix_state_lock(sk);
L
Linus Torvalds 已提交
1965 1966

	for (;;) {
E
Eric Dumazet 已提交
1967
		prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
L
Linus Torvalds 已提交
1968

1969
		if (skb_peek_tail(&sk->sk_receive_queue) != last ||
L
Linus Torvalds 已提交
1970 1971 1972 1973 1974 1975 1976
		    sk->sk_err ||
		    (sk->sk_shutdown & RCV_SHUTDOWN) ||
		    signal_pending(current) ||
		    !timeo)
			break;

		set_bit(SOCK_ASYNC_WAITDATA, &sk->sk_socket->flags);
1977
		unix_state_unlock(sk);
1978
		timeo = freezable_schedule_timeout(timeo);
1979
		unix_state_lock(sk);
L
Linus Torvalds 已提交
1980 1981 1982
		clear_bit(SOCK_ASYNC_WAITDATA, &sk->sk_socket->flags);
	}

E
Eric Dumazet 已提交
1983
	finish_wait(sk_sleep(sk), &wait);
1984
	unix_state_unlock(sk);
L
Linus Torvalds 已提交
1985 1986 1987
	return timeo;
}

1988 1989 1990 1991 1992
static unsigned int unix_skb_len(const struct sk_buff *skb)
{
	return skb->len - UNIXCB(skb).consumed;
}

1993 1994
static int unix_stream_recvmsg(struct socket *sock, struct msghdr *msg,
			       size_t size, int flags)
L
Linus Torvalds 已提交
1995
{
C
Christoph Hellwig 已提交
1996
	struct scm_cookie scm;
L
Linus Torvalds 已提交
1997 1998
	struct sock *sk = sock->sk;
	struct unix_sock *u = unix_sk(sk);
1999
	DECLARE_SOCKADDR(struct sockaddr_un *, sunaddr, msg->msg_name);
L
Linus Torvalds 已提交
2000
	int copied = 0;
2001
	int noblock = flags & MSG_DONTWAIT;
L
Linus Torvalds 已提交
2002 2003 2004 2005
	int check_creds = 0;
	int target;
	int err = 0;
	long timeo;
2006
	int skip;
L
Linus Torvalds 已提交
2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

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

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

	target = sock_rcvlowat(sk, flags&MSG_WAITALL, size);
2017
	timeo = sock_rcvtimeo(sk, noblock);
L
Linus Torvalds 已提交
2018 2019 2020 2021 2022

	/* Lock the socket to prevent queue disordering
	 * while sleeps in memcpy_tomsg
	 */

C
Christoph Hellwig 已提交
2023
	memset(&scm, 0, sizeof(scm));
L
Linus Torvalds 已提交
2024

2025
	err = mutex_lock_interruptible(&u->readlock);
2026 2027 2028 2029 2030
	if (unlikely(err)) {
		/* recvmsg() in non blocking mode is supposed to return -EAGAIN
		 * sk_rcvtimeo is not honored by mutex_lock_interruptible()
		 */
		err = noblock ? -EAGAIN : -ERESTARTSYS;
2031 2032
		goto out;
	}
L
Linus Torvalds 已提交
2033

E
Eric Dumazet 已提交
2034
	do {
L
Linus Torvalds 已提交
2035
		int chunk;
2036
		struct sk_buff *skb, *last;
L
Linus Torvalds 已提交
2037

2038
		unix_state_lock(sk);
2039
		last = skb = skb_peek(&sk->sk_receive_queue);
2040
again:
E
Eric Dumazet 已提交
2041
		if (skb == NULL) {
E
Eric Dumazet 已提交
2042
			unix_sk(sk)->recursion_level = 0;
L
Linus Torvalds 已提交
2043
			if (copied >= target)
2044
				goto unlock;
L
Linus Torvalds 已提交
2045 2046 2047 2048

			/*
			 *	POSIX 1003.1g mandates this order.
			 */
2049

E
Eric Dumazet 已提交
2050 2051
			err = sock_error(sk);
			if (err)
2052
				goto unlock;
L
Linus Torvalds 已提交
2053
			if (sk->sk_shutdown & RCV_SHUTDOWN)
2054 2055 2056
				goto unlock;

			unix_state_unlock(sk);
L
Linus Torvalds 已提交
2057 2058 2059
			err = -EAGAIN;
			if (!timeo)
				break;
I
Ingo Molnar 已提交
2060
			mutex_unlock(&u->readlock);
L
Linus Torvalds 已提交
2061

2062
			timeo = unix_stream_data_wait(sk, timeo, last);
L
Linus Torvalds 已提交
2063

2064 2065
			if (signal_pending(current)
			    ||  mutex_lock_interruptible(&u->readlock)) {
L
Linus Torvalds 已提交
2066 2067 2068
				err = sock_intr_errno(timeo);
				goto out;
			}
2069

L
Linus Torvalds 已提交
2070
			continue;
2071 2072 2073
 unlock:
			unix_state_unlock(sk);
			break;
L
Linus Torvalds 已提交
2074
		}
2075

2076
		skip = sk_peek_offset(sk, flags);
2077 2078
		while (skip >= unix_skb_len(skb)) {
			skip -= unix_skb_len(skb);
2079
			last = skb;
2080
			skb = skb_peek_next(skb, &sk->sk_receive_queue);
2081 2082
			if (!skb)
				goto again;
2083 2084
		}

2085
		unix_state_unlock(sk);
L
Linus Torvalds 已提交
2086 2087 2088

		if (check_creds) {
			/* Never glue messages from different writers */
C
Christoph Hellwig 已提交
2089 2090 2091
			if ((UNIXCB(skb).pid  != scm.pid) ||
			    !uid_eq(UNIXCB(skb).uid, scm.creds.uid) ||
			    !gid_eq(UNIXCB(skb).gid, scm.creds.gid))
L
Linus Torvalds 已提交
2092
				break;
2093
		} else if (test_bit(SOCK_PASSCRED, &sock->flags)) {
L
Linus Torvalds 已提交
2094
			/* Copy credentials */
C
Christoph Hellwig 已提交
2095
			scm_set_cred(&scm, UNIXCB(skb).pid, UNIXCB(skb).uid, UNIXCB(skb).gid);
L
Linus Torvalds 已提交
2096 2097 2098 2099
			check_creds = 1;
		}

		/* Copy address just once */
E
Eric Dumazet 已提交
2100
		if (sunaddr) {
L
Linus Torvalds 已提交
2101 2102 2103 2104
			unix_copy_addr(msg, skb->sk);
			sunaddr = NULL;
		}

2105
		chunk = min_t(unsigned int, unix_skb_len(skb) - skip, size);
2106 2107
		if (skb_copy_datagram_msg(skb, UNIXCB(skb).consumed + skip,
					  msg, chunk)) {
L
Linus Torvalds 已提交
2108 2109 2110 2111 2112 2113 2114 2115
			if (copied == 0)
				copied = -EFAULT;
			break;
		}
		copied += chunk;
		size -= chunk;

		/* Mark read part of skb as used */
E
Eric Dumazet 已提交
2116
		if (!(flags & MSG_PEEK)) {
2117
			UNIXCB(skb).consumed += chunk;
L
Linus Torvalds 已提交
2118

2119 2120
			sk_peek_offset_bwd(sk, chunk);

L
Linus Torvalds 已提交
2121
			if (UNIXCB(skb).fp)
C
Christoph Hellwig 已提交
2122
				unix_detach_fds(&scm, skb);
L
Linus Torvalds 已提交
2123

2124
			if (unix_skb_len(skb))
L
Linus Torvalds 已提交
2125 2126
				break;

2127
			skb_unlink(skb, &sk->sk_receive_queue);
2128
			consume_skb(skb);
L
Linus Torvalds 已提交
2129

C
Christoph Hellwig 已提交
2130
			if (scm.fp)
L
Linus Torvalds 已提交
2131
				break;
E
Eric Dumazet 已提交
2132
		} else {
L
Linus Torvalds 已提交
2133 2134 2135
			/* It is questionable, see note in unix_dgram_recvmsg.
			 */
			if (UNIXCB(skb).fp)
C
Christoph Hellwig 已提交
2136
				scm.fp = scm_fp_dup(UNIXCB(skb).fp);
L
Linus Torvalds 已提交
2137

2138 2139
			sk_peek_offset_fwd(sk, chunk);

L
Linus Torvalds 已提交
2140 2141 2142 2143
			break;
		}
	} while (size);

I
Ingo Molnar 已提交
2144
	mutex_unlock(&u->readlock);
C
Christoph Hellwig 已提交
2145
	scm_recv(sock, msg, &scm, flags);
L
Linus Torvalds 已提交
2146 2147 2148 2149 2150 2151 2152 2153 2154
out:
	return copied ? : err;
}

static int unix_shutdown(struct socket *sock, int mode)
{
	struct sock *sk = sock->sk;
	struct sock *other;

2155 2156 2157 2158 2159 2160 2161 2162
	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;
2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188

	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
2189
	}
2190 2191 2192
	if (other)
		sock_put(other);

L
Linus Torvalds 已提交
2193 2194 2195
	return 0;
}

2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207
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)
2208
			amount += unix_skb_len(skb);
2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225
	} 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);

L
Linus Torvalds 已提交
2226 2227 2228
static int unix_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
{
	struct sock *sk = sock->sk;
2229
	long amount = 0;
L
Linus Torvalds 已提交
2230 2231
	int err;

E
Eric Dumazet 已提交
2232 2233
	switch (cmd) {
	case SIOCOUTQ:
2234
		amount = unix_outq_len(sk);
E
Eric Dumazet 已提交
2235 2236 2237
		err = put_user(amount, (int __user *)arg);
		break;
	case SIOCINQ:
2238 2239 2240 2241
		amount = unix_inq_len(sk);
		if (amount < 0)
			err = amount;
		else
L
Linus Torvalds 已提交
2242
			err = put_user(amount, (int __user *)arg);
2243
		break;
E
Eric Dumazet 已提交
2244 2245 2246
	default:
		err = -ENOIOCTLCMD;
		break;
L
Linus Torvalds 已提交
2247 2248 2249 2250
	}
	return err;
}

E
Eric Dumazet 已提交
2251
static unsigned int unix_poll(struct file *file, struct socket *sock, poll_table *wait)
L
Linus Torvalds 已提交
2252 2253 2254 2255
{
	struct sock *sk = sock->sk;
	unsigned int mask;

E
Eric Dumazet 已提交
2256
	sock_poll_wait(file, sk_sleep(sk), wait);
L
Linus Torvalds 已提交
2257 2258 2259 2260 2261 2262 2263
	mask = 0;

	/* exceptional events? */
	if (sk->sk_err)
		mask |= POLLERR;
	if (sk->sk_shutdown == SHUTDOWN_MASK)
		mask |= POLLHUP;
2264
	if (sk->sk_shutdown & RCV_SHUTDOWN)
E
Eric Dumazet 已提交
2265
		mask |= POLLRDHUP | POLLIN | POLLRDNORM;
L
Linus Torvalds 已提交
2266 2267

	/* readable? */
E
Eric Dumazet 已提交
2268
	if (!skb_queue_empty(&sk->sk_receive_queue))
L
Linus Torvalds 已提交
2269 2270 2271
		mask |= POLLIN | POLLRDNORM;

	/* Connection-based need to check for termination and startup */
E
Eric Dumazet 已提交
2272 2273
	if ((sk->sk_type == SOCK_STREAM || sk->sk_type == SOCK_SEQPACKET) &&
	    sk->sk_state == TCP_CLOSE)
L
Linus Torvalds 已提交
2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285
		mask |= POLLHUP;

	/*
	 * we set writable also when the other side has shut down the
	 * connection. This prevents stuck sockets.
	 */
	if (unix_writable(sk))
		mask |= POLLOUT | POLLWRNORM | POLLWRBAND;

	return mask;
}

2286 2287
static unsigned int unix_dgram_poll(struct file *file, struct socket *sock,
				    poll_table *wait)
2288
{
2289 2290
	struct sock *sk = sock->sk, *other;
	unsigned int mask, writable;
2291

E
Eric Dumazet 已提交
2292
	sock_poll_wait(file, sk_sleep(sk), wait);
2293 2294 2295 2296
	mask = 0;

	/* exceptional events? */
	if (sk->sk_err || !skb_queue_empty(&sk->sk_error_queue))
2297
		mask |= POLLERR |
2298
			(sock_flag(sk, SOCK_SELECT_ERR_QUEUE) ? POLLPRI : 0);
2299

2300
	if (sk->sk_shutdown & RCV_SHUTDOWN)
2301
		mask |= POLLRDHUP | POLLIN | POLLRDNORM;
2302 2303 2304 2305
	if (sk->sk_shutdown == SHUTDOWN_MASK)
		mask |= POLLHUP;

	/* readable? */
2306
	if (!skb_queue_empty(&sk->sk_receive_queue))
2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317
		mask |= POLLIN | POLLRDNORM;

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

E
Eric Dumazet 已提交
2318
	/* No write status requested, avoid expensive OUT tests. */
2319
	if (!(poll_requested_events(wait) & (POLLWRBAND|POLLWRNORM|POLLOUT)))
E
Eric Dumazet 已提交
2320 2321
		return mask;

2322
	writable = unix_writable(sk);
2323 2324 2325 2326 2327 2328
	other = unix_peer_get(sk);
	if (other) {
		if (unix_peer(other) != sk) {
			sock_poll_wait(file, &unix_sk(other)->peer_wait, wait);
			if (unix_recvq_full(other))
				writable = 0;
2329
		}
2330
		sock_put(other);
2331 2332 2333
	}

	if (writable)
2334 2335 2336 2337 2338 2339
		mask |= POLLOUT | POLLWRNORM | POLLWRBAND;
	else
		set_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags);

	return mask;
}
L
Linus Torvalds 已提交
2340 2341

#ifdef CONFIG_PROC_FS
2342

E
Eric Dumazet 已提交
2343 2344 2345 2346 2347
#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))
2348

E
Eric Dumazet 已提交
2349
static struct sock *unix_from_bucket(struct seq_file *seq, loff_t *pos)
L
Linus Torvalds 已提交
2350
{
E
Eric Dumazet 已提交
2351 2352 2353 2354
	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
2355

E
Eric Dumazet 已提交
2356 2357
	for (sk = sk_head(&unix_socket_table[bucket]); sk; sk = sk_next(sk)) {
		if (sock_net(sk) != seq_file_net(seq))
2358
			continue;
E
Eric Dumazet 已提交
2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377
		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
2378
	}
E
Eric Dumazet 已提交
2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389

	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
2390 2391 2392 2393
	return NULL;
}

static void *unix_seq_start(struct seq_file *seq, loff_t *pos)
2394
	__acquires(unix_table_lock)
L
Linus Torvalds 已提交
2395
{
2396
	spin_lock(&unix_table_lock);
E
Eric Dumazet 已提交
2397 2398 2399 2400 2401 2402 2403 2404

	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
2405 2406 2407 2408 2409
}

static void *unix_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
	++*pos;
E
Eric Dumazet 已提交
2410
	return unix_next_socket(seq, v, pos);
L
Linus Torvalds 已提交
2411 2412 2413
}

static void unix_seq_stop(struct seq_file *seq, void *v)
2414
	__releases(unix_table_lock)
L
Linus Torvalds 已提交
2415
{
2416
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
2417 2418 2419 2420
}

static int unix_seq_show(struct seq_file *seq, void *v)
{
2421

J
Joe Perches 已提交
2422
	if (v == SEQ_START_TOKEN)
L
Linus Torvalds 已提交
2423 2424 2425 2426 2427
		seq_puts(seq, "Num       RefCount Protocol Flags    Type St "
			 "Inode Path\n");
	else {
		struct sock *s = v;
		struct unix_sock *u = unix_sk(s);
2428
		unix_state_lock(s);
L
Linus Torvalds 已提交
2429

D
Dan Rosenberg 已提交
2430
		seq_printf(seq, "%pK: %08X %08X %08X %04X %02X %5lu",
L
Linus Torvalds 已提交
2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455
			s,
			atomic_read(&s->sk_refcnt),
			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++)
				seq_putc(seq, u->addr->name->sun_path[i]);
		}
2456
		unix_state_unlock(s);
L
Linus Torvalds 已提交
2457 2458 2459 2460 2461 2462
		seq_putc(seq, '\n');
	}

	return 0;
}

2463
static const struct seq_operations unix_seq_ops = {
L
Linus Torvalds 已提交
2464 2465 2466 2467 2468 2469 2470 2471
	.start  = unix_seq_start,
	.next   = unix_seq_next,
	.stop   = unix_seq_stop,
	.show   = unix_seq_show,
};

static int unix_seq_open(struct inode *inode, struct file *file)
{
2472
	return seq_open_net(inode, file, &unix_seq_ops,
E
Eric Dumazet 已提交
2473
			    sizeof(struct seq_net_private));
L
Linus Torvalds 已提交
2474 2475
}

2476
static const struct file_operations unix_seq_fops = {
L
Linus Torvalds 已提交
2477 2478 2479 2480
	.owner		= THIS_MODULE,
	.open		= unix_seq_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
2481
	.release	= seq_release_net,
L
Linus Torvalds 已提交
2482 2483 2484 2485
};

#endif

2486
static const struct net_proto_family unix_family_ops = {
L
Linus Torvalds 已提交
2487 2488 2489 2490 2491
	.family = PF_UNIX,
	.create = unix_create,
	.owner	= THIS_MODULE,
};

2492

2493
static int __net_init unix_net_init(struct net *net)
2494 2495 2496
{
	int error = -ENOMEM;

2497
	net->unx.sysctl_max_dgram_qlen = 10;
2498 2499
	if (unix_sysctl_register(net))
		goto out;
2500

2501
#ifdef CONFIG_PROC_FS
2502
	if (!proc_create("unix", 0, net->proc_net, &unix_seq_fops)) {
2503
		unix_sysctl_unregister(net);
2504
		goto out;
2505
	}
2506 2507 2508
#endif
	error = 0;
out:
2509
	return error;
2510 2511
}

2512
static void __net_exit unix_net_exit(struct net *net)
2513
{
2514
	unix_sysctl_unregister(net);
2515
	remove_proc_entry("unix", net->proc_net);
2516 2517 2518 2519 2520 2521 2522
}

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

L
Linus Torvalds 已提交
2523 2524 2525 2526
static int __init af_unix_init(void)
{
	int rc = -1;

2527
	BUILD_BUG_ON(sizeof(struct unix_skb_parms) > FIELD_SIZEOF(struct sk_buff, cb));
L
Linus Torvalds 已提交
2528 2529

	rc = proto_register(&unix_proto, 1);
2530
	if (rc != 0) {
W
wangweidong 已提交
2531
		pr_crit("%s: Cannot create unix_sock SLAB cache!\n", __func__);
L
Linus Torvalds 已提交
2532 2533 2534 2535
		goto out;
	}

	sock_register(&unix_family_ops);
2536
	register_pernet_subsys(&unix_net_ops);
L
Linus Torvalds 已提交
2537 2538 2539 2540 2541 2542 2543 2544
out:
	return rc;
}

static void __exit af_unix_exit(void)
{
	sock_unregister(PF_UNIX);
	proto_unregister(&unix_proto);
2545
	unregister_pernet_subsys(&unix_net_ops);
L
Linus Torvalds 已提交
2546 2547
}

2548 2549 2550 2551 2552
/* 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
2553 2554 2555 2556
module_exit(af_unix_exit);

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