af_unix.c 68.0 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 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
	UNIXCB(skb).secid = scm->secid;
C
Catherine Zhang 已提交
144 145 146 147
}

static inline void unix_set_secdata(struct scm_cookie *scm, struct sk_buff *skb)
{
148 149 150 151 152 153
	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
154 155
}
#else
156
static inline void unix_get_secdata(struct scm_cookie *scm, struct sk_buff *skb)
C
Catherine Zhang 已提交
157 158 159 160
{ }

static inline void unix_set_secdata(struct scm_cookie *scm, struct sk_buff *skb)
{ }
161 162 163 164 165

static inline bool unix_secdata_eq(struct scm_cookie *scm, struct sk_buff *skb)
{
	return true;
}
C
Catherine Zhang 已提交
166 167
#endif /* CONFIG_SECURITY_NETWORK */

L
Linus Torvalds 已提交
168 169
/*
 *  SMP locking strategy:
170
 *    hash table is protected with spinlock unix_table_lock
171
 *    each socket state is protected by separate spin lock.
L
Linus Torvalds 已提交
172 173
 */

174
static inline unsigned int unix_hash_fold(__wsum n)
L
Linus Torvalds 已提交
175
{
176
	unsigned int hash = (__force unsigned int)csum_fold(n);
177

L
Linus Torvalds 已提交
178 179 180 181 182 183 184 185 186 187 188 189 190
	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
191
	return unix_peer(osk) == NULL || unix_our_peer(sk, osk);
L
Linus Torvalds 已提交
192 193
}

194 195 196 197 198
static inline int unix_recvq_full(struct sock const *sk)
{
	return skb_queue_len(&sk->sk_receive_queue) > sk->sk_max_ack_backlog;
}

199
struct sock *unix_peer_get(struct sock *s)
L
Linus Torvalds 已提交
200 201 202
{
	struct sock *peer;

203
	unix_state_lock(s);
L
Linus Torvalds 已提交
204 205 206
	peer = unix_peer(s);
	if (peer)
		sock_hold(peer);
207
	unix_state_unlock(s);
L
Linus Torvalds 已提交
208 209
	return peer;
}
210
EXPORT_SYMBOL_GPL(unix_peer_get);
L
Linus Torvalds 已提交
211 212 213 214 215 216 217 218 219 220 221 222 223

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.
 */
224

225
static int unix_mkname(struct sockaddr_un *sunaddr, int len, unsigned int *hashp)
L
Linus Torvalds 已提交
226 227 228 229 230 231 232 233 234
{
	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
235
		 * sun_path[108] doesn't as such exist.  However in kernel space
L
Linus Torvalds 已提交
236 237 238
		 * we are guaranteed that it is a valid memory location in our
		 * kernel address buffer.
		 */
239
		((char *)sunaddr)[len] = 0;
L
Linus Torvalds 已提交
240 241 242 243
		len = strlen(sunaddr->sun_path)+1+sizeof(short);
		return len;
	}

244
	*hashp = unix_hash_fold(csum_partial(sunaddr, len, 0));
L
Linus Torvalds 已提交
245 246 247 248 249 250 251 252 253 254
	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)
{
255
	WARN_ON(!sk_unhashed(sk));
L
Linus Torvalds 已提交
256 257 258 259 260
	sk_add_node(sk, list);
}

static inline void unix_remove_socket(struct sock *sk)
{
261
	spin_lock(&unix_table_lock);
L
Linus Torvalds 已提交
262
	__unix_remove_socket(sk);
263
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
264 265 266 267
}

static inline void unix_insert_socket(struct hlist_head *list, struct sock *sk)
{
268
	spin_lock(&unix_table_lock);
L
Linus Torvalds 已提交
269
	__unix_insert_socket(list, sk);
270
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
271 272
}

273 274
static struct sock *__unix_find_socket_byname(struct net *net,
					      struct sockaddr_un *sunname,
275
					      int len, int type, unsigned int hash)
L
Linus Torvalds 已提交
276 277 278
{
	struct sock *s;

279
	sk_for_each(s, &unix_socket_table[hash ^ type]) {
L
Linus Torvalds 已提交
280 281
		struct unix_sock *u = unix_sk(s);

282
		if (!net_eq(sock_net(s), net))
283 284
			continue;

L
Linus Torvalds 已提交
285 286 287 288 289 290 291 292 293
		if (u->addr->len == len &&
		    !memcmp(u->addr->name, sunname, len))
			goto found;
	}
	s = NULL;
found:
	return s;
}

294 295
static inline struct sock *unix_find_socket_byname(struct net *net,
						   struct sockaddr_un *sunname,
L
Linus Torvalds 已提交
296
						   int len, int type,
297
						   unsigned int hash)
L
Linus Torvalds 已提交
298 299 300
{
	struct sock *s;

301
	spin_lock(&unix_table_lock);
302
	s = __unix_find_socket_byname(net, sunname, len, type, hash);
L
Linus Torvalds 已提交
303 304
	if (s)
		sock_hold(s);
305
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
306 307 308
	return s;
}

309
static struct sock *unix_find_socket_byinode(struct inode *i)
L
Linus Torvalds 已提交
310 311 312
{
	struct sock *s;

313
	spin_lock(&unix_table_lock);
314
	sk_for_each(s,
L
Linus Torvalds 已提交
315
		    &unix_socket_table[i->i_ino & (UNIX_HASH_SIZE - 1)]) {
A
Al Viro 已提交
316
		struct dentry *dentry = unix_sk(s)->path.dentry;
L
Linus Torvalds 已提交
317

318
		if (dentry && d_backing_inode(dentry) == i) {
L
Linus Torvalds 已提交
319 320 321 322 323 324
			sock_hold(s);
			goto found;
		}
	}
	s = NULL;
found:
325
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
326 327 328
	return s;
}

329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440
/* 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
 *
 * In order to propagate a wake up, a wait_queue_t of the client
 * 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.
 */

static int unix_dgram_peer_wake_relay(wait_queue_t *q, unsigned mode, int flags,
				      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)
		wake_up_interruptible_poll(u_sleep, key);

	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),
				   POLLOUT |
				   POLLWRNORM |
				   POLLWRBAND);
}

/* 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;
}

441
static int unix_writable(const struct sock *sk)
L
Linus Torvalds 已提交
442
{
443 444
	return sk->sk_state != TCP_LISTEN &&
	       (atomic_read(&sk->sk_wmem_alloc) << 2) <= sk->sk_sndbuf;
L
Linus Torvalds 已提交
445 446 447 448
}

static void unix_write_space(struct sock *sk)
{
449 450 451
	struct socket_wq *wq;

	rcu_read_lock();
L
Linus Torvalds 已提交
452
	if (unix_writable(sk)) {
453
		wq = rcu_dereference(sk->sk_wq);
H
Herbert Xu 已提交
454
		if (skwq_has_sleeper(wq))
E
Eric Dumazet 已提交
455 456
			wake_up_interruptible_sync_poll(&wq->wait,
				POLLOUT | POLLWRNORM | POLLWRBAND);
457
		sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
L
Linus Torvalds 已提交
458
	}
459
	rcu_read_unlock();
L
Linus Torvalds 已提交
460 461 462 463 464 465 466 467
}

/* 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)
{
468
	if (!skb_queue_empty(&sk->sk_receive_queue)) {
L
Linus Torvalds 已提交
469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488
		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);

489 490 491
	WARN_ON(atomic_read(&sk->sk_wmem_alloc));
	WARN_ON(!sk_unhashed(sk));
	WARN_ON(sk->sk_socket);
L
Linus Torvalds 已提交
492
	if (!sock_flag(sk, SOCK_DEAD)) {
W
wangweidong 已提交
493
		pr_info("Attempt to release alive unix socket: %p\n", sk);
L
Linus Torvalds 已提交
494 495 496 497 498 499
		return;
	}

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

E
Eric Dumazet 已提交
500
	atomic_long_dec(&unix_nr_socks);
501
	local_bh_disable();
502
	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
503
	local_bh_enable();
L
Linus Torvalds 已提交
504
#ifdef UNIX_REFCNT_DEBUG
W
wangweidong 已提交
505
	pr_debug("UNIX %p is destroyed, %ld are still alive.\n", sk,
E
Eric Dumazet 已提交
506
		atomic_long_read(&unix_nr_socks));
L
Linus Torvalds 已提交
507 508 509
#endif
}

510
static void unix_release_sock(struct sock *sk, int embrion)
L
Linus Torvalds 已提交
511 512
{
	struct unix_sock *u = unix_sk(sk);
A
Al Viro 已提交
513
	struct path path;
L
Linus Torvalds 已提交
514 515 516 517 518 519 520
	struct sock *skpair;
	struct sk_buff *skb;
	int state;

	unix_remove_socket(sk);

	/* Clear state */
521
	unix_state_lock(sk);
L
Linus Torvalds 已提交
522 523
	sock_orphan(sk);
	sk->sk_shutdown = SHUTDOWN_MASK;
A
Al Viro 已提交
524 525 526
	path	     = u->path;
	u->path.dentry = NULL;
	u->path.mnt = NULL;
L
Linus Torvalds 已提交
527 528
	state = sk->sk_state;
	sk->sk_state = TCP_CLOSE;
529
	unix_state_unlock(sk);
L
Linus Torvalds 已提交
530 531 532

	wake_up_interruptible_all(&u->peer_wait);

533
	skpair = unix_peer(sk);
L
Linus Torvalds 已提交
534

535
	if (skpair != NULL) {
L
Linus Torvalds 已提交
536
		if (sk->sk_type == SOCK_STREAM || sk->sk_type == SOCK_SEQPACKET) {
537
			unix_state_lock(skpair);
L
Linus Torvalds 已提交
538 539 540 541
			/* No more writes */
			skpair->sk_shutdown = SHUTDOWN_MASK;
			if (!skb_queue_empty(&sk->sk_receive_queue) || embrion)
				skpair->sk_err = ECONNRESET;
542
			unix_state_unlock(skpair);
L
Linus Torvalds 已提交
543
			skpair->sk_state_change(skpair);
544
			sk_wake_async(skpair, SOCK_WAKE_WAITD, POLL_HUP);
L
Linus Torvalds 已提交
545
		}
546 547

		unix_dgram_peer_wake_disconnect(sk, skpair);
L
Linus Torvalds 已提交
548 549 550 551 552 553 554
		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) {
555
		if (state == TCP_LISTEN)
L
Linus Torvalds 已提交
556 557
			unix_release_sock(skb->sk, 1);
		/* passed fds are erased in the kfree_skb hook	      */
558
		UNIXCB(skb).consumed = skb->len;
L
Linus Torvalds 已提交
559 560 561
		kfree_skb(skb);
	}

A
Al Viro 已提交
562 563
	if (path.dentry)
		path_put(&path);
L
Linus Torvalds 已提交
564 565 566 567 568 569

	sock_put(sk);

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

	/*
A
Alan Cox 已提交
570
	 * Fixme: BSD difference: In BSD all sockets connected to us get
L
Linus Torvalds 已提交
571 572 573 574 575 576 577 578 579
	 *	  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)
	 */

580
	if (unix_tot_inflight)
581
		unix_gc();		/* Garbage collect fds */
L
Linus Torvalds 已提交
582 583
}

584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601
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
602 603 604 605 606
static int unix_listen(struct socket *sock, int backlog)
{
	int err;
	struct sock *sk = sock->sk;
	struct unix_sock *u = unix_sk(sk);
607
	struct pid *old_pid = NULL;
L
Linus Torvalds 已提交
608 609

	err = -EOPNOTSUPP;
E
Eric Dumazet 已提交
610 611
	if (sock->type != SOCK_STREAM && sock->type != SOCK_SEQPACKET)
		goto out;	/* Only stream/seqpacket sockets accept */
L
Linus Torvalds 已提交
612 613
	err = -EINVAL;
	if (!u->addr)
E
Eric Dumazet 已提交
614
		goto out;	/* No listens on an unbound socket */
615
	unix_state_lock(sk);
L
Linus Torvalds 已提交
616 617 618 619 620 621 622
	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 */
623
	init_peercred(sk);
L
Linus Torvalds 已提交
624 625 626
	err = 0;

out_unlock:
627
	unix_state_unlock(sk);
628
	put_pid(old_pid);
L
Linus Torvalds 已提交
629 630 631 632 633 634 635 636 637 638 639 640
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 *);
641 642
static unsigned int unix_dgram_poll(struct file *, struct socket *,
				    poll_table *);
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);

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

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

	return 0;
671 672 673
}


674
static const struct proto_ops unix_stream_ops = {
L
Linus Torvalds 已提交
675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691
	.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,
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 =		unix_dgram_poll,
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 =		unix_dgram_poll,
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
};

747 748 749 750 751 752 753 754
/*
 * 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;

755
static struct sock *unix_create1(struct net *net, struct socket *sock, int kern)
L
Linus Torvalds 已提交
756 757 758 759
{
	struct sock *sk = NULL;
	struct unix_sock *u;

E
Eric Dumazet 已提交
760 761
	atomic_long_inc(&unix_nr_socks);
	if (atomic_long_read(&unix_nr_socks) > 2 * get_max_files())
L
Linus Torvalds 已提交
762 763
		goto out;

764
	sk = sk_alloc(net, PF_UNIX, GFP_KERNEL, &unix_proto, kern);
L
Linus Torvalds 已提交
765 766 767
	if (!sk)
		goto out;

E
Eric Dumazet 已提交
768
	sock_init_data(sock, sk);
769 770
	lockdep_set_class(&sk->sk_receive_queue.lock,
				&af_unix_sk_receive_queue_lock_key);
L
Linus Torvalds 已提交
771 772

	sk->sk_write_space	= unix_write_space;
773
	sk->sk_max_ack_backlog	= net->unx.sysctl_max_dgram_qlen;
L
Linus Torvalds 已提交
774 775
	sk->sk_destruct		= unix_sock_destructor;
	u	  = unix_sk(sk);
A
Al Viro 已提交
776 777
	u->path.dentry = NULL;
	u->path.mnt = NULL;
778
	spin_lock_init(&u->lock);
A
Al Viro 已提交
779
	atomic_long_set(&u->inflight, 0);
780
	INIT_LIST_HEAD(&u->link);
I
Ingo Molnar 已提交
781
	mutex_init(&u->readlock); /* single task reading lock */
L
Linus Torvalds 已提交
782
	init_waitqueue_head(&u->peer_wait);
783
	init_waitqueue_func_entry(&u->peer_wake, unix_dgram_peer_wake_relay);
E
Eric Dumazet 已提交
784
	unix_insert_socket(unix_sockets_unbound(sk), sk);
L
Linus Torvalds 已提交
785
out:
786
	if (sk == NULL)
E
Eric Dumazet 已提交
787
		atomic_long_dec(&unix_nr_socks);
788 789
	else {
		local_bh_disable();
790
		sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
791 792
		local_bh_enable();
	}
L
Linus Torvalds 已提交
793 794 795
	return sk;
}

796 797
static int unix_create(struct net *net, struct socket *sock, int protocol,
		       int kern)
L
Linus Torvalds 已提交
798 799 800 801 802 803 804 805 806 807 808 809 810 811 812
{
	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:
813
		sock->type = SOCK_DGRAM;
L
Linus Torvalds 已提交
814 815 816 817 818 819 820 821 822 823
	case SOCK_DGRAM:
		sock->ops = &unix_dgram_ops;
		break;
	case SOCK_SEQPACKET:
		sock->ops = &unix_seqpacket_ops;
		break;
	default:
		return -ESOCKTNOSUPPORT;
	}

824
	return unix_create1(net, sock, kern) ? 0 : -ENOMEM;
L
Linus Torvalds 已提交
825 826 827 828 829 830 831 832 833
}

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

	if (!sk)
		return 0;

834
	unix_release_sock(sk, 0);
L
Linus Torvalds 已提交
835 836
	sock->sk = NULL;

837
	return 0;
L
Linus Torvalds 已提交
838 839 840 841 842
}

static int unix_autobind(struct socket *sock)
{
	struct sock *sk = sock->sk;
843
	struct net *net = sock_net(sk);
L
Linus Torvalds 已提交
844 845
	struct unix_sock *u = unix_sk(sk);
	static u32 ordernum = 1;
E
Eric Dumazet 已提交
846
	struct unix_address *addr;
L
Linus Torvalds 已提交
847
	int err;
848
	unsigned int retries = 0;
L
Linus Torvalds 已提交
849

850 851 852
	err = mutex_lock_interruptible(&u->readlock);
	if (err)
		return err;
L
Linus Torvalds 已提交
853 854 855 856 857 858

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

	err = -ENOMEM;
859
	addr = kzalloc(sizeof(*addr) + sizeof(short) + 16, GFP_KERNEL);
L
Linus Torvalds 已提交
860 861 862 863 864 865 866 867
	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);
868
	addr->hash = unix_hash_fold(csum_partial(addr->name, addr->len, 0));
L
Linus Torvalds 已提交
869

870
	spin_lock(&unix_table_lock);
L
Linus Torvalds 已提交
871 872
	ordernum = (ordernum+1)&0xFFFFF;

873
	if (__unix_find_socket_byname(net, addr->name, addr->len, sock->type,
L
Linus Torvalds 已提交
874
				      addr->hash)) {
875
		spin_unlock(&unix_table_lock);
876 877 878 879 880 881 882 883 884 885 886
		/*
		 * __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
887 888 889 890 891 892 893
		goto retry;
	}
	addr->hash ^= sk->sk_type;

	__unix_remove_socket(sk);
	u->addr = addr;
	__unix_insert_socket(&unix_socket_table[addr->hash], sk);
894
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
895 896
	err = 0;

I
Ingo Molnar 已提交
897
out:	mutex_unlock(&u->readlock);
L
Linus Torvalds 已提交
898 899 900
	return err;
}

901 902
static struct sock *unix_find_other(struct net *net,
				    struct sockaddr_un *sunname, int len,
903
				    int type, unsigned int hash, int *error)
L
Linus Torvalds 已提交
904 905
{
	struct sock *u;
906
	struct path path;
L
Linus Torvalds 已提交
907
	int err = 0;
908

L
Linus Torvalds 已提交
909
	if (sunname->sun_path[0]) {
910 911
		struct inode *inode;
		err = kern_path(sunname->sun_path, LOOKUP_FOLLOW, &path);
L
Linus Torvalds 已提交
912 913
		if (err)
			goto fail;
914
		inode = d_backing_inode(path.dentry);
915
		err = inode_permission(inode, MAY_WRITE);
L
Linus Torvalds 已提交
916 917 918 919
		if (err)
			goto put_fail;

		err = -ECONNREFUSED;
920
		if (!S_ISSOCK(inode->i_mode))
L
Linus Torvalds 已提交
921
			goto put_fail;
922
		u = unix_find_socket_byinode(inode);
L
Linus Torvalds 已提交
923 924 925 926
		if (!u)
			goto put_fail;

		if (u->sk_type == type)
A
Al Viro 已提交
927
			touch_atime(&path);
L
Linus Torvalds 已提交
928

929
		path_put(&path);
L
Linus Torvalds 已提交
930

931
		err = -EPROTOTYPE;
L
Linus Torvalds 已提交
932 933 934 935 936 937
		if (u->sk_type != type) {
			sock_put(u);
			goto fail;
		}
	} else {
		err = -ECONNREFUSED;
938
		u = unix_find_socket_byname(net, sunname, len, type, hash);
L
Linus Torvalds 已提交
939 940
		if (u) {
			struct dentry *dentry;
A
Al Viro 已提交
941
			dentry = unix_sk(u)->path.dentry;
L
Linus Torvalds 已提交
942
			if (dentry)
A
Al Viro 已提交
943
				touch_atime(&unix_sk(u)->path);
L
Linus Torvalds 已提交
944 945 946 947 948 949
		} else
			goto fail;
	}
	return u;

put_fail:
950
	path_put(&path);
L
Linus Torvalds 已提交
951
fail:
952
	*error = err;
L
Linus Torvalds 已提交
953 954 955
	return NULL;
}

956 957
static int unix_mknod(struct dentry *dentry, struct path *path, umode_t mode,
		      struct path *res)
A
Al Viro 已提交
958
{
959
	int err;
A
Al Viro 已提交
960

961
	err = security_path_mknod(path, dentry, mode, 0);
A
Al Viro 已提交
962
	if (!err) {
963
		err = vfs_mknod(d_inode(path->dentry), dentry, mode, 0);
A
Al Viro 已提交
964
		if (!err) {
965
			res->mnt = mntget(path->mnt);
A
Al Viro 已提交
966 967 968
			res->dentry = dget(dentry);
		}
	}
969

A
Al Viro 已提交
970 971
	return err;
}
L
Linus Torvalds 已提交
972 973 974 975

static int unix_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
{
	struct sock *sk = sock->sk;
976
	struct net *net = sock_net(sk);
L
Linus Torvalds 已提交
977
	struct unix_sock *u = unix_sk(sk);
978
	struct sockaddr_un *sunaddr = (struct sockaddr_un *)uaddr;
979
	char *sun_path = sunaddr->sun_path;
980
	int err, name_err;
981
	unsigned int hash;
L
Linus Torvalds 已提交
982 983
	struct unix_address *addr;
	struct hlist_head *list;
984 985
	struct path path;
	struct dentry *dentry;
L
Linus Torvalds 已提交
986 987 988 989 990

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

991
	if (addr_len == sizeof(short)) {
L
Linus Torvalds 已提交
992 993 994 995 996 997 998 999 1000
		err = unix_autobind(sock);
		goto out;
	}

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

1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015
	name_err = 0;
	dentry = NULL;
	if (sun_path[0]) {
		/* Get the parent directory, calculate the hash for last
		 * component.
		 */
		dentry = kern_path_create(AT_FDCWD, sun_path, &path, 0);

		if (IS_ERR(dentry)) {
			/* delay report until after 'already bound' check */
			name_err = PTR_ERR(dentry);
			dentry = NULL;
		}
	}

1016 1017
	err = mutex_lock_interruptible(&u->readlock);
	if (err)
1018
		goto out_path;
L
Linus Torvalds 已提交
1019 1020 1021 1022 1023

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

1024 1025 1026 1027 1028
	if (name_err) {
		err = name_err == -EEXIST ? -EADDRINUSE : name_err;
		goto out_up;
	}

L
Linus Torvalds 已提交
1029 1030 1031 1032 1033 1034 1035 1036 1037 1038
	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);

1039 1040
	if (dentry) {
		struct path u_path;
A
Al Viro 已提交
1041
		umode_t mode = S_IFSOCK |
A
Al Viro 已提交
1042
		       (SOCK_INODE(sock)->i_mode & ~current_umask());
1043
		err = unix_mknod(dentry, &path, mode, &u_path);
A
Al Viro 已提交
1044 1045 1046 1047 1048 1049
		if (err) {
			if (err == -EEXIST)
				err = -EADDRINUSE;
			unix_release_addr(addr);
			goto out_up;
		}
L
Linus Torvalds 已提交
1050
		addr->hash = UNIX_HASH_SIZE;
1051
		hash = d_backing_inode(dentry)->i_ino & (UNIX_HASH_SIZE - 1);
A
Al Viro 已提交
1052
		spin_lock(&unix_table_lock);
1053
		u->path = u_path;
A
Al Viro 已提交
1054 1055 1056
		list = &unix_socket_table[hash];
	} else {
		spin_lock(&unix_table_lock);
L
Linus Torvalds 已提交
1057
		err = -EADDRINUSE;
1058
		if (__unix_find_socket_byname(net, sunaddr, addr_len,
L
Linus Torvalds 已提交
1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072
					      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:
1073
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
1074
out_up:
I
Ingo Molnar 已提交
1075
	mutex_unlock(&u->readlock);
1076 1077 1078 1079
out_path:
	if (dentry)
		done_path_create(&path, dentry);

L
Linus Torvalds 已提交
1080 1081 1082 1083
out:
	return err;
}

1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108
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
1109 1110 1111 1112
static int unix_dgram_connect(struct socket *sock, struct sockaddr *addr,
			      int alen, int flags)
{
	struct sock *sk = sock->sk;
1113
	struct net *net = sock_net(sk);
1114
	struct sockaddr_un *sunaddr = (struct sockaddr_un *)addr;
L
Linus Torvalds 已提交
1115
	struct sock *other;
1116
	unsigned int hash;
L
Linus Torvalds 已提交
1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128
	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;

1129
restart:
1130
		other = unix_find_other(net, sunaddr, alen, sock->type, hash, &err);
L
Linus Torvalds 已提交
1131 1132 1133
		if (!other)
			goto out;

1134 1135 1136 1137 1138 1139 1140 1141
		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
1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155

		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;
1156
		unix_state_double_lock(sk, other);
L
Linus Torvalds 已提交
1157 1158 1159 1160 1161 1162 1163
	}

	/*
	 * If it was connected, reconnect.
	 */
	if (unix_peer(sk)) {
		struct sock *old_peer = unix_peer(sk);
1164
		unix_peer(sk) = other;
1165 1166
		unix_dgram_peer_wake_disconnect_wakeup(sk, old_peer);

1167
		unix_state_double_unlock(sk, other);
L
Linus Torvalds 已提交
1168 1169 1170 1171 1172

		if (other != old_peer)
			unix_dgram_disconnected(sk, old_peer);
		sock_put(old_peer);
	} else {
1173
		unix_peer(sk) = other;
1174
		unix_state_double_unlock(sk, other);
L
Linus Torvalds 已提交
1175
	}
1176
	return 0;
L
Linus Torvalds 已提交
1177 1178

out_unlock:
1179
	unix_state_double_unlock(sk, other);
L
Linus Torvalds 已提交
1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194
	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) &&
1195
		unix_recvq_full(other);
L
Linus Torvalds 已提交
1196

1197
	unix_state_unlock(other);
L
Linus Torvalds 已提交
1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208

	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)
{
1209
	struct sockaddr_un *sunaddr = (struct sockaddr_un *)uaddr;
L
Linus Torvalds 已提交
1210
	struct sock *sk = sock->sk;
1211
	struct net *net = sock_net(sk);
L
Linus Torvalds 已提交
1212 1213 1214 1215
	struct unix_sock *u = unix_sk(sk), *newu, *otheru;
	struct sock *newsk = NULL;
	struct sock *other = NULL;
	struct sk_buff *skb = NULL;
1216
	unsigned int hash;
L
Linus Torvalds 已提交
1217 1218 1219 1220 1221 1222 1223 1224 1225
	int st;
	int err;
	long timeo;

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

1226 1227
	if (test_bit(SOCK_PASSCRED, &sock->flags) && !u->addr &&
	    (err = unix_autobind(sock)) != 0)
L
Linus Torvalds 已提交
1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239
		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 */
1240
	newsk = unix_create1(sock_net(sk), NULL, 0);
L
Linus Torvalds 已提交
1241 1242 1243 1244 1245 1246 1247 1248 1249 1250
	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. */
1251
	other = unix_find_other(net, sunaddr, addr_len, sk->sk_type, hash, &err);
L
Linus Torvalds 已提交
1252 1253 1254 1255
	if (!other)
		goto out;

	/* Latch state of peer */
1256
	unix_state_lock(other);
L
Linus Torvalds 已提交
1257 1258 1259

	/* Apparently VFS overslept socket death. Retry. */
	if (sock_flag(other, SOCK_DEAD)) {
1260
		unix_state_unlock(other);
L
Linus Torvalds 已提交
1261 1262 1263 1264 1265 1266 1267
		sock_put(other);
		goto restart;
	}

	err = -ECONNREFUSED;
	if (other->sk_state != TCP_LISTEN)
		goto out_unlock;
1268 1269
	if (other->sk_shutdown & RCV_SHUTDOWN)
		goto out_unlock;
L
Linus Torvalds 已提交
1270

1271
	if (unix_recvq_full(other)) {
L
Linus Torvalds 已提交
1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282
		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;
1283
	}
L
Linus Torvalds 已提交
1284 1285 1286

	/* Latch our state.

D
Daniel Baluta 已提交
1287
	   It is tricky place. We need to grab our state lock and cannot
L
Linus Torvalds 已提交
1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310
	   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;
	}

1311
	unix_state_lock_nested(sk);
L
Linus Torvalds 已提交
1312 1313

	if (sk->sk_state != st) {
1314 1315
		unix_state_unlock(sk);
		unix_state_unlock(other);
L
Linus Torvalds 已提交
1316 1317 1318 1319
		sock_put(other);
		goto restart;
	}

1320
	err = security_unix_stream_connect(sk, other, newsk);
L
Linus Torvalds 已提交
1321
	if (err) {
1322
		unix_state_unlock(sk);
L
Linus Torvalds 已提交
1323 1324 1325 1326 1327 1328 1329 1330 1331
		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;
1332
	init_peercred(newsk);
L
Linus Torvalds 已提交
1333
	newu = unix_sk(newsk);
1334
	RCU_INIT_POINTER(newsk->sk_wq, &newu->peer_wq);
L
Linus Torvalds 已提交
1335 1336 1337 1338 1339 1340 1341
	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
1342 1343 1344
	if (otheru->path.dentry) {
		path_get(&otheru->path);
		newu->path = otheru->path;
L
Linus Torvalds 已提交
1345 1346 1347
	}

	/* Set credentials */
1348
	copy_peercred(sk, other);
L
Linus Torvalds 已提交
1349 1350 1351

	sock->state	= SS_CONNECTED;
	sk->sk_state	= TCP_ESTABLISHED;
1352 1353
	sock_hold(newsk);

1354
	smp_mb__after_atomic();	/* sock_hold() does an atomic_inc() */
1355
	unix_peer(sk)	= newsk;
L
Linus Torvalds 已提交
1356

1357
	unix_state_unlock(sk);
L
Linus Torvalds 已提交
1358 1359 1360 1361 1362

	/* 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);
1363
	unix_state_unlock(other);
1364
	other->sk_data_ready(other);
L
Linus Torvalds 已提交
1365 1366 1367 1368 1369
	sock_put(other);
	return 0;

out_unlock:
	if (other)
1370
		unix_state_unlock(other);
L
Linus Torvalds 已提交
1371 1372

out:
1373
	kfree_skb(skb);
L
Linus Torvalds 已提交
1374 1375 1376 1377 1378 1379 1380 1381 1382
	if (newsk)
		unix_release_sock(newsk, 0);
	if (other)
		sock_put(other);
	return err;
}

static int unix_socketpair(struct socket *socka, struct socket *sockb)
{
1383
	struct sock *ska = socka->sk, *skb = sockb->sk;
L
Linus Torvalds 已提交
1384 1385 1386 1387

	/* Join our sockets back to back */
	sock_hold(ska);
	sock_hold(skb);
1388 1389
	unix_peer(ska) = skb;
	unix_peer(skb) = ska;
1390 1391
	init_peercred(ska);
	init_peercred(skb);
L
Linus Torvalds 已提交
1392 1393 1394 1395 1396 1397 1398 1399 1400 1401

	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;
}

1402 1403 1404 1405 1406 1407 1408 1409 1410
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
1411 1412 1413 1414 1415 1416 1417 1418
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
1419
	if (sock->type != SOCK_STREAM && sock->type != SOCK_SEQPACKET)
L
Linus Torvalds 已提交
1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442
		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 */
1443
	unix_state_lock(tsk);
L
Linus Torvalds 已提交
1444
	newsock->state = SS_CONNECTED;
1445
	unix_sock_inherit_flags(sock, newsock);
L
Linus Torvalds 已提交
1446
	sock_graft(tsk, newsock);
1447
	unix_state_unlock(tsk);
L
Linus Torvalds 已提交
1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458
	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;
1459
	DECLARE_SOCKADDR(struct sockaddr_un *, sunaddr, uaddr);
L
Linus Torvalds 已提交
1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473
	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);
1474
	unix_state_lock(sk);
L
Linus Torvalds 已提交
1475 1476 1477 1478 1479 1480 1481 1482 1483 1484
	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);
	}
1485
	unix_state_unlock(sk);
L
Linus Torvalds 已提交
1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497
	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
1498
	for (i = scm->fp->count-1; i >= 0; i--)
1499
		unix_notinflight(scm->fp->user, scm->fp->fp[i]);
L
Linus Torvalds 已提交
1500 1501
}

1502
static void unix_destruct_scm(struct sk_buff *skb)
L
Linus Torvalds 已提交
1503 1504 1505
{
	struct scm_cookie scm;
	memset(&scm, 0, sizeof(scm));
1506 1507 1508
	scm.pid  = UNIXCB(skb).pid;
	if (UNIXCB(skb).fp)
		unix_detach_fds(&scm, skb);
L
Linus Torvalds 已提交
1509 1510 1511 1512 1513 1514 1515

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

1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530
/*
 * 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;
}

E
Eric Dumazet 已提交
1531 1532
#define MAX_RECURSION_LEVEL 4

1533
static int unix_attach_fds(struct scm_cookie *scm, struct sk_buff *skb)
L
Linus Torvalds 已提交
1534 1535
{
	int i;
E
Eric Dumazet 已提交
1536 1537 1538
	unsigned char max_level = 0;
	int unix_sock_count = 0;

1539 1540 1541
	if (too_many_unix_fds(current))
		return -ETOOMANYREFS;

E
Eric Dumazet 已提交
1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552
	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;
1553 1554 1555 1556 1557 1558 1559 1560 1561 1562

	/*
	 * 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;

1563
	for (i = scm->fp->count - 1; i >= 0; i--)
1564
		unix_inflight(scm->fp->user, scm->fp->fp[i]);
E
Eric Dumazet 已提交
1565
	return max_level;
L
Linus Torvalds 已提交
1566 1567
}

1568
static int unix_scm_to_skb(struct scm_cookie *scm, struct sk_buff *skb, bool send_fds)
1569 1570
{
	int err = 0;
1571

1572
	UNIXCB(skb).pid  = get_pid(scm->pid);
1573 1574
	UNIXCB(skb).uid = scm->creds.uid;
	UNIXCB(skb).gid = scm->creds.gid;
1575
	UNIXCB(skb).fp = NULL;
1576
	unix_get_secdata(scm, skb);
1577 1578 1579 1580 1581 1582 1583
	if (scm->fp && send_fds)
		err = unix_attach_fds(scm, skb);

	skb->destructor = unix_destruct_scm;
	return err;
}

1584 1585 1586 1587 1588 1589 1590 1591
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);
}

1592 1593 1594 1595 1596 1597 1598 1599
/*
 * 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)
{
1600
	if (UNIXCB(skb).pid)
1601
		return;
1602
	if (unix_passcred_enabled(sock, other)) {
1603
		UNIXCB(skb).pid  = get_pid(task_tgid(current));
1604
		current_uid_gid(&UNIXCB(skb).uid, &UNIXCB(skb).gid);
1605 1606 1607
	}
}

1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636
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
1637 1638 1639 1640
/*
 *	Send AF_UNIX data.
 */

1641 1642
static int unix_dgram_sendmsg(struct socket *sock, struct msghdr *msg,
			      size_t len)
L
Linus Torvalds 已提交
1643 1644
{
	struct sock *sk = sock->sk;
1645
	struct net *net = sock_net(sk);
L
Linus Torvalds 已提交
1646
	struct unix_sock *u = unix_sk(sk);
1647
	DECLARE_SOCKADDR(struct sockaddr_un *, sunaddr, msg->msg_name);
L
Linus Torvalds 已提交
1648 1649 1650
	struct sock *other = NULL;
	int namelen = 0; /* fake GCC */
	int err;
1651
	unsigned int hash;
1652
	struct sk_buff *skb;
L
Linus Torvalds 已提交
1653
	long timeo;
C
Christoph Hellwig 已提交
1654
	struct scm_cookie scm;
E
Eric Dumazet 已提交
1655
	int max_level;
1656
	int data_len = 0;
1657
	int sk_locked;
L
Linus Torvalds 已提交
1658

1659
	wait_for_unix_gc();
C
Christoph Hellwig 已提交
1660
	err = scm_send(sock, msg, &scm, false);
L
Linus Torvalds 已提交
1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680
	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;
	}

1681 1682
	if (test_bit(SOCK_PASSCRED, &sock->flags) && !u->addr
	    && (err = unix_autobind(sock)) != 0)
L
Linus Torvalds 已提交
1683 1684 1685 1686 1687 1688
		goto out;

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

1689
	if (len > SKB_MAX_ALLOC) {
1690 1691 1692
		data_len = min_t(size_t,
				 len - SKB_MAX_ALLOC,
				 MAX_SKB_FRAGS * PAGE_SIZE);
1693 1694 1695 1696
		data_len = PAGE_ALIGN(data_len);

		BUILD_BUG_ON(SKB_MAX_ALLOC < PAGE_SIZE);
	}
1697 1698

	skb = sock_alloc_send_pskb(sk, len - data_len, data_len,
1699 1700
				   msg->msg_flags & MSG_DONTWAIT, &err,
				   PAGE_ALLOC_COSTLY_ORDER);
1701
	if (skb == NULL)
L
Linus Torvalds 已提交
1702 1703
		goto out;

C
Christoph Hellwig 已提交
1704
	err = unix_scm_to_skb(&scm, skb, true);
E
Eric Dumazet 已提交
1705
	if (err < 0)
1706
		goto out_free;
E
Eric Dumazet 已提交
1707
	max_level = err + 1;
C
Catherine Zhang 已提交
1708

1709 1710 1711
	skb_put(skb, len - data_len);
	skb->data_len = data_len;
	skb->len = len;
A
Al Viro 已提交
1712
	err = skb_copy_datagram_from_iter(skb, 0, &msg->msg_iter, len);
L
Linus Torvalds 已提交
1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723
	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;

1724
		other = unix_find_other(net, sunaddr, namelen, sk->sk_type,
L
Linus Torvalds 已提交
1725
					hash, &err);
1726
		if (other == NULL)
L
Linus Torvalds 已提交
1727 1728 1729
			goto out_free;
	}

A
Alban Crequy 已提交
1730 1731 1732 1733 1734 1735
	if (sk_filter(other, skb) < 0) {
		/* Toss the packet but do not return any error to the sender */
		err = len;
		goto out_free;
	}

1736
	sk_locked = 0;
1737
	unix_state_lock(other);
1738
restart_locked:
L
Linus Torvalds 已提交
1739 1740 1741 1742
	err = -EPERM;
	if (!unix_may_send(sk, other))
		goto out_unlock;

1743
	if (unlikely(sock_flag(other, SOCK_DEAD))) {
L
Linus Torvalds 已提交
1744 1745 1746 1747
		/*
		 *	Check with 1003.1g - what should
		 *	datagram error
		 */
1748
		unix_state_unlock(other);
L
Linus Torvalds 已提交
1749 1750
		sock_put(other);

1751 1752 1753
		if (!sk_locked)
			unix_state_lock(sk);

L
Linus Torvalds 已提交
1754 1755
		err = 0;
		if (unix_peer(sk) == other) {
1756
			unix_peer(sk) = NULL;
1757 1758
			unix_dgram_peer_wake_disconnect_wakeup(sk, other);

1759
			unix_state_unlock(sk);
L
Linus Torvalds 已提交
1760 1761 1762 1763 1764

			unix_dgram_disconnected(sk, other);
			sock_put(other);
			err = -ECONNREFUSED;
		} else {
1765
			unix_state_unlock(sk);
L
Linus Torvalds 已提交
1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783
		}

		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;
	}

1784 1785 1786 1787 1788 1789
	/* 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))) {
1790 1791 1792 1793 1794 1795 1796 1797
		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
1798 1799
		}

1800 1801 1802 1803
		if (!sk_locked) {
			unix_state_unlock(other);
			unix_state_double_lock(sk, other);
		}
L
Linus Torvalds 已提交
1804

1805 1806 1807 1808 1809 1810
		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
1811

1812 1813 1814 1815
		if (!sk_locked) {
			sk_locked = 1;
			goto restart_locked;
		}
L
Linus Torvalds 已提交
1816 1817
	}

1818 1819 1820
	if (unlikely(sk_locked))
		unix_state_unlock(sk);

1821 1822
	if (sock_flag(other, SOCK_RCVTSTAMP))
		__net_timestamp(skb);
1823
	maybe_add_creds(skb, sock, other);
L
Linus Torvalds 已提交
1824
	skb_queue_tail(&other->sk_receive_queue, skb);
E
Eric Dumazet 已提交
1825 1826
	if (max_level > unix_sk(other)->recursion_level)
		unix_sk(other)->recursion_level = max_level;
1827
	unix_state_unlock(other);
1828
	other->sk_data_ready(other);
L
Linus Torvalds 已提交
1829
	sock_put(other);
C
Christoph Hellwig 已提交
1830
	scm_destroy(&scm);
L
Linus Torvalds 已提交
1831 1832 1833
	return len;

out_unlock:
1834 1835
	if (sk_locked)
		unix_state_unlock(sk);
1836
	unix_state_unlock(other);
L
Linus Torvalds 已提交
1837 1838 1839 1840 1841
out_free:
	kfree_skb(skb);
out:
	if (other)
		sock_put(other);
C
Christoph Hellwig 已提交
1842
	scm_destroy(&scm);
L
Linus Torvalds 已提交
1843 1844 1845
	return err;
}

1846 1847 1848 1849
/* 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))
1850

1851 1852
static int unix_stream_sendmsg(struct socket *sock, struct msghdr *msg,
			       size_t len)
L
Linus Torvalds 已提交
1853 1854 1855
{
	struct sock *sk = sock->sk;
	struct sock *other = NULL;
E
Eric Dumazet 已提交
1856
	int err, size;
1857
	struct sk_buff *skb;
1858
	int sent = 0;
C
Christoph Hellwig 已提交
1859
	struct scm_cookie scm;
1860
	bool fds_sent = false;
E
Eric Dumazet 已提交
1861
	int max_level;
1862
	int data_len;
L
Linus Torvalds 已提交
1863

1864
	wait_for_unix_gc();
C
Christoph Hellwig 已提交
1865
	err = scm_send(sock, msg, &scm, false);
L
Linus Torvalds 已提交
1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877
	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;
1878
		other = unix_peer(sk);
L
Linus Torvalds 已提交
1879 1880 1881 1882 1883 1884 1885
		if (!other)
			goto out_err;
	}

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

E
Eric Dumazet 已提交
1886
	while (sent < len) {
1887
		size = len - sent;
L
Linus Torvalds 已提交
1888 1889

		/* Keep two messages in the pipe so it schedules better */
1890
		size = min_t(int, size, (sk->sk_sndbuf >> 1) - 64);
L
Linus Torvalds 已提交
1891

1892 1893
		/* allow fallback to order-0 allocations */
		size = min_t(int, size, SKB_MAX_HEAD(0) + UNIX_SKB_FRAGS_SZ);
1894

1895
		data_len = max_t(int, 0, size - SKB_MAX_HEAD(0));
L
Linus Torvalds 已提交
1896

1897 1898
		data_len = min_t(size_t, size, PAGE_ALIGN(data_len));

1899
		skb = sock_alloc_send_pskb(sk, size - data_len, data_len,
1900 1901
					   msg->msg_flags & MSG_DONTWAIT, &err,
					   get_order(UNIX_SKB_FRAGS_SZ));
1902
		if (!skb)
L
Linus Torvalds 已提交
1903 1904
			goto out_err;

1905
		/* Only send the fds in the first buffer */
C
Christoph Hellwig 已提交
1906
		err = unix_scm_to_skb(&scm, skb, !fds_sent);
E
Eric Dumazet 已提交
1907
		if (err < 0) {
1908
			kfree_skb(skb);
1909
			goto out_err;
1910
		}
E
Eric Dumazet 已提交
1911
		max_level = err + 1;
1912
		fds_sent = true;
L
Linus Torvalds 已提交
1913

1914 1915 1916
		skb_put(skb, size - data_len);
		skb->data_len = data_len;
		skb->len = size;
A
Al Viro 已提交
1917
		err = skb_copy_datagram_from_iter(skb, 0, &msg->msg_iter, size);
E
Eric Dumazet 已提交
1918
		if (err) {
L
Linus Torvalds 已提交
1919
			kfree_skb(skb);
1920
			goto out_err;
L
Linus Torvalds 已提交
1921 1922
		}

1923
		unix_state_lock(other);
L
Linus Torvalds 已提交
1924 1925 1926 1927 1928

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

1929
		maybe_add_creds(skb, sock, other);
L
Linus Torvalds 已提交
1930
		skb_queue_tail(&other->sk_receive_queue, skb);
E
Eric Dumazet 已提交
1931 1932
		if (max_level > unix_sk(other)->recursion_level)
			unix_sk(other)->recursion_level = max_level;
1933
		unix_state_unlock(other);
1934
		other->sk_data_ready(other);
1935
		sent += size;
L
Linus Torvalds 已提交
1936 1937
	}

C
Christoph Hellwig 已提交
1938
	scm_destroy(&scm);
L
Linus Torvalds 已提交
1939 1940 1941 1942

	return sent;

pipe_err_free:
1943
	unix_state_unlock(other);
L
Linus Torvalds 已提交
1944 1945
	kfree_skb(skb);
pipe_err:
E
Eric Dumazet 已提交
1946 1947
	if (sent == 0 && !(msg->msg_flags&MSG_NOSIGNAL))
		send_sig(SIGPIPE, current, 0);
L
Linus Torvalds 已提交
1948 1949
	err = -EPIPE;
out_err:
C
Christoph Hellwig 已提交
1950
	scm_destroy(&scm);
L
Linus Torvalds 已提交
1951 1952 1953
	return sent ? : err;
}

1954 1955 1956
static ssize_t unix_stream_sendpage(struct socket *socket, struct page *page,
				    int offset, size_t size, int flags)
{
1957 1958 1959 1960
	int err;
	bool send_sigpipe = false;
	bool init_scm = true;
	struct scm_cookie scm;
1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977
	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)
1978
			goto err;
1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991
	}

	/* 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;
		goto err;
	}

	if (sk->sk_shutdown & SEND_SHUTDOWN) {
		err = -EPIPE;
1992
		send_sigpipe = true;
1993 1994 1995 1996 1997 1998 1999 2000
		goto err_unlock;
	}

	unix_state_lock(other);

	if (sock_flag(other, SOCK_DEAD) ||
	    other->sk_shutdown & RCV_SHUTDOWN) {
		err = -EPIPE;
2001
		send_sigpipe = true;
2002 2003 2004
		goto err_state_unlock;
	}

2005 2006 2007 2008 2009 2010 2011
	if (init_scm) {
		err = maybe_init_creds(&scm, socket, other);
		if (err)
			goto err_state_unlock;
		init_scm = false;
	}

2012 2013 2014
	skb = skb_peek_tail(&other->sk_receive_queue);
	if (tail && tail == skb) {
		skb = newskb;
2015 2016
	} else if (!skb || !unix_skb_scm_eq(skb, &scm)) {
		if (newskb) {
2017
			skb = newskb;
2018 2019
		} else {
			tail = skb;
2020
			goto alloc_skb;
2021
		}
2022 2023 2024 2025 2026 2027
	} 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);
2028
		newskb = NULL;
2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040
	}

	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);

2041
	if (newskb) {
2042 2043 2044
		err = unix_scm_to_skb(&scm, skb, false);
		if (err)
			goto err_state_unlock;
2045
		spin_lock(&other->sk_receive_queue.lock);
2046
		__skb_queue_tail(&other->sk_receive_queue, newskb);
2047 2048
		spin_unlock(&other->sk_receive_queue.lock);
	}
2049 2050 2051 2052 2053

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

	other->sk_data_ready(other);
2054
	scm_destroy(&scm);
2055 2056 2057 2058 2059 2060 2061 2062 2063 2064
	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);
2065 2066
	if (!init_scm)
		scm_destroy(&scm);
2067 2068 2069
	return err;
}

2070 2071
static int unix_seqpacket_sendmsg(struct socket *sock, struct msghdr *msg,
				  size_t len)
L
Linus Torvalds 已提交
2072 2073 2074
{
	int err;
	struct sock *sk = sock->sk;
2075

L
Linus Torvalds 已提交
2076 2077 2078 2079 2080 2081 2082 2083 2084 2085
	err = sock_error(sk);
	if (err)
		return err;

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

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

2086
	return unix_dgram_sendmsg(sock, msg, len);
L
Linus Torvalds 已提交
2087
}
2088

2089 2090
static int unix_seqpacket_recvmsg(struct socket *sock, struct msghdr *msg,
				  size_t size, int flags)
2091 2092 2093 2094 2095 2096
{
	struct sock *sk = sock->sk;

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

2097
	return unix_dgram_recvmsg(sock, msg, size, flags);
2098 2099
}

L
Linus Torvalds 已提交
2100 2101 2102 2103 2104 2105 2106 2107 2108 2109
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);
	}
}

2110 2111
static int unix_dgram_recvmsg(struct socket *sock, struct msghdr *msg,
			      size_t size, int flags)
L
Linus Torvalds 已提交
2112
{
C
Christoph Hellwig 已提交
2113
	struct scm_cookie scm;
L
Linus Torvalds 已提交
2114 2115
	struct sock *sk = sock->sk;
	struct unix_sock *u = unix_sk(sk);
2116 2117
	struct sk_buff *skb, *last;
	long timeo;
L
Linus Torvalds 已提交
2118
	int err;
2119
	int peeked, skip;
L
Linus Torvalds 已提交
2120 2121 2122 2123 2124

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

2125
	timeo = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
L
Linus Torvalds 已提交
2126

2127 2128
	do {
		mutex_lock(&u->readlock);
2129

2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143
		skip = sk_peek_offset(sk, flags);
		skb = __skb_try_recv_datagram(sk, flags, &peeked, &skip, &err,
					      &last);
		if (skb)
			break;

		mutex_unlock(&u->readlock);

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

	if (!skb) { /* implies readlock unlocked */
2144 2145 2146 2147 2148 2149
		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);
2150
		goto out;
2151
	}
L
Linus Torvalds 已提交
2152

2153 2154 2155 2156
	if (wq_has_sleeper(&u->peer_wait))
		wake_up_interruptible_sync_poll(&u->peer_wait,
						POLLOUT | POLLWRNORM |
						POLLWRBAND);
L
Linus Torvalds 已提交
2157 2158 2159 2160

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

2161 2162 2163
	if (size > skb->len - skip)
		size = skb->len - skip;
	else if (size < skb->len - skip)
L
Linus Torvalds 已提交
2164 2165
		msg->msg_flags |= MSG_TRUNC;

2166
	err = skb_copy_datagram_msg(skb, skip, msg, size);
L
Linus Torvalds 已提交
2167 2168 2169
	if (err)
		goto out_free;

2170 2171 2172
	if (sock_flag(sk, SOCK_RCVTSTAMP))
		__sock_recv_timestamp(msg, sk, skb);

C
Christoph Hellwig 已提交
2173 2174 2175 2176
	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
2177

E
Eric Dumazet 已提交
2178
	if (!(flags & MSG_PEEK)) {
L
Linus Torvalds 已提交
2179
		if (UNIXCB(skb).fp)
C
Christoph Hellwig 已提交
2180
			unix_detach_fds(&scm, skb);
2181 2182

		sk_peek_offset_bwd(sk, skb->len);
E
Eric Dumazet 已提交
2183
	} else {
L
Linus Torvalds 已提交
2184 2185 2186 2187 2188 2189
		/* 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)
2190 2191 2192 2193 2194

		   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
2195
		*/
2196 2197 2198

		sk_peek_offset_fwd(sk, size);

L
Linus Torvalds 已提交
2199
		if (UNIXCB(skb).fp)
C
Christoph Hellwig 已提交
2200
			scm.fp = scm_fp_dup(UNIXCB(skb).fp);
L
Linus Torvalds 已提交
2201
	}
2202
	err = (flags & MSG_TRUNC) ? skb->len - skip : size;
L
Linus Torvalds 已提交
2203

C
Christoph Hellwig 已提交
2204
	scm_recv(sock, msg, &scm, flags);
L
Linus Torvalds 已提交
2205 2206

out_free:
E
Eric Dumazet 已提交
2207
	skb_free_datagram(sk, skb);
I
Ingo Molnar 已提交
2208
	mutex_unlock(&u->readlock);
L
Linus Torvalds 已提交
2209 2210 2211 2212 2213
out:
	return err;
}

/*
2214
 *	Sleep until more data has arrived. But check for races..
L
Linus Torvalds 已提交
2215
 */
2216
static long unix_stream_data_wait(struct sock *sk, long timeo,
2217
				  struct sk_buff *last, unsigned int last_len)
L
Linus Torvalds 已提交
2218
{
2219
	struct sk_buff *tail;
L
Linus Torvalds 已提交
2220 2221
	DEFINE_WAIT(wait);

2222
	unix_state_lock(sk);
L
Linus Torvalds 已提交
2223 2224

	for (;;) {
E
Eric Dumazet 已提交
2225
		prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
L
Linus Torvalds 已提交
2226

2227 2228 2229
		tail = skb_peek_tail(&sk->sk_receive_queue);
		if (tail != last ||
		    (tail && tail->len != last_len) ||
L
Linus Torvalds 已提交
2230 2231 2232 2233 2234 2235
		    sk->sk_err ||
		    (sk->sk_shutdown & RCV_SHUTDOWN) ||
		    signal_pending(current) ||
		    !timeo)
			break;

2236
		sk_set_bit(SOCKWQ_ASYNC_WAITDATA, sk);
2237
		unix_state_unlock(sk);
2238
		timeo = freezable_schedule_timeout(timeo);
2239
		unix_state_lock(sk);
2240 2241 2242 2243

		if (sock_flag(sk, SOCK_DEAD))
			break;

2244
		sk_clear_bit(SOCKWQ_ASYNC_WAITDATA, sk);
L
Linus Torvalds 已提交
2245 2246
	}

E
Eric Dumazet 已提交
2247
	finish_wait(sk_sleep(sk), &wait);
2248
	unix_state_unlock(sk);
L
Linus Torvalds 已提交
2249 2250 2251
	return timeo;
}

2252 2253 2254 2255 2256
static unsigned int unix_skb_len(const struct sk_buff *skb)
{
	return skb->len - UNIXCB(skb).consumed;
}

2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268
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;
};

static int unix_stream_read_generic(struct unix_stream_read_state *state)
L
Linus Torvalds 已提交
2269
{
C
Christoph Hellwig 已提交
2270
	struct scm_cookie scm;
2271
	struct socket *sock = state->socket;
L
Linus Torvalds 已提交
2272 2273 2274
	struct sock *sk = sock->sk;
	struct unix_sock *u = unix_sk(sk);
	int copied = 0;
2275
	int flags = state->flags;
2276
	int noblock = flags & MSG_DONTWAIT;
2277
	bool check_creds = false;
L
Linus Torvalds 已提交
2278 2279 2280
	int target;
	int err = 0;
	long timeo;
2281
	int skip;
2282 2283
	size_t size = state->size;
	unsigned int last_len;
L
Linus Torvalds 已提交
2284

2285 2286
	if (unlikely(sk->sk_state != TCP_ESTABLISHED)) {
		err = -EINVAL;
L
Linus Torvalds 已提交
2287
		goto out;
2288
	}
L
Linus Torvalds 已提交
2289

2290 2291
	if (unlikely(flags & MSG_OOB)) {
		err = -EOPNOTSUPP;
L
Linus Torvalds 已提交
2292
		goto out;
2293
	}
L
Linus Torvalds 已提交
2294

2295
	target = sock_rcvlowat(sk, flags & MSG_WAITALL, size);
2296
	timeo = sock_rcvtimeo(sk, noblock);
L
Linus Torvalds 已提交
2297

2298 2299
	memset(&scm, 0, sizeof(scm));

L
Linus Torvalds 已提交
2300 2301 2302
	/* Lock the socket to prevent queue disordering
	 * while sleeps in memcpy_tomsg
	 */
2303
	mutex_lock(&u->readlock);
L
Linus Torvalds 已提交
2304

2305 2306 2307 2308 2309
	if (flags & MSG_PEEK)
		skip = sk_peek_offset(sk, flags);
	else
		skip = 0;

E
Eric Dumazet 已提交
2310
	do {
L
Linus Torvalds 已提交
2311
		int chunk;
2312
		bool drop_skb;
2313
		struct sk_buff *skb, *last;
L
Linus Torvalds 已提交
2314

2315
		unix_state_lock(sk);
2316 2317 2318 2319
		if (sock_flag(sk, SOCK_DEAD)) {
			err = -ECONNRESET;
			goto unlock;
		}
2320
		last = skb = skb_peek(&sk->sk_receive_queue);
2321
		last_len = last ? last->len : 0;
2322
again:
E
Eric Dumazet 已提交
2323
		if (skb == NULL) {
E
Eric Dumazet 已提交
2324
			unix_sk(sk)->recursion_level = 0;
L
Linus Torvalds 已提交
2325
			if (copied >= target)
2326
				goto unlock;
L
Linus Torvalds 已提交
2327 2328 2329 2330

			/*
			 *	POSIX 1003.1g mandates this order.
			 */
2331

E
Eric Dumazet 已提交
2332 2333
			err = sock_error(sk);
			if (err)
2334
				goto unlock;
L
Linus Torvalds 已提交
2335
			if (sk->sk_shutdown & RCV_SHUTDOWN)
2336 2337 2338
				goto unlock;

			unix_state_unlock(sk);
2339 2340
			if (!timeo) {
				err = -EAGAIN;
L
Linus Torvalds 已提交
2341
				break;
2342 2343
			}

I
Ingo Molnar 已提交
2344
			mutex_unlock(&u->readlock);
L
Linus Torvalds 已提交
2345

2346 2347
			timeo = unix_stream_data_wait(sk, timeo, last,
						      last_len);
L
Linus Torvalds 已提交
2348

2349
			if (signal_pending(current)) {
L
Linus Torvalds 已提交
2350
				err = sock_intr_errno(timeo);
2351
				scm_destroy(&scm);
L
Linus Torvalds 已提交
2352 2353
				goto out;
			}
2354

2355
			mutex_lock(&u->readlock);
L
Linus Torvalds 已提交
2356
			continue;
2357
unlock:
2358 2359
			unix_state_unlock(sk);
			break;
L
Linus Torvalds 已提交
2360
		}
2361

2362 2363
		while (skip >= unix_skb_len(skb)) {
			skip -= unix_skb_len(skb);
2364
			last = skb;
2365
			last_len = skb->len;
2366
			skb = skb_peek_next(skb, &sk->sk_receive_queue);
2367 2368
			if (!skb)
				goto again;
2369 2370
		}

2371
		unix_state_unlock(sk);
L
Linus Torvalds 已提交
2372 2373 2374

		if (check_creds) {
			/* Never glue messages from different writers */
2375
			if (!unix_skb_scm_eq(skb, &scm))
L
Linus Torvalds 已提交
2376
				break;
2377
		} else if (test_bit(SOCK_PASSCRED, &sock->flags)) {
L
Linus Torvalds 已提交
2378
			/* Copy credentials */
C
Christoph Hellwig 已提交
2379
			scm_set_cred(&scm, UNIXCB(skb).pid, UNIXCB(skb).uid, UNIXCB(skb).gid);
2380
			unix_set_secdata(&scm, skb);
2381
			check_creds = true;
L
Linus Torvalds 已提交
2382 2383 2384
		}

		/* Copy address just once */
2385 2386 2387 2388
		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
2389 2390 2391
			sunaddr = NULL;
		}

2392
		chunk = min_t(unsigned int, unix_skb_len(skb) - skip, size);
2393
		skb_get(skb);
2394
		chunk = state->recv_actor(skb, skip, chunk, state);
2395 2396 2397
		drop_skb = !unix_skb_len(skb);
		/* skb is only safe to use if !drop_skb */
		consume_skb(skb);
2398
		if (chunk < 0) {
L
Linus Torvalds 已提交
2399 2400 2401 2402 2403 2404 2405
			if (copied == 0)
				copied = -EFAULT;
			break;
		}
		copied += chunk;
		size -= chunk;

2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417
		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
2418
		/* Mark read part of skb as used */
E
Eric Dumazet 已提交
2419
		if (!(flags & MSG_PEEK)) {
2420
			UNIXCB(skb).consumed += chunk;
L
Linus Torvalds 已提交
2421

2422 2423
			sk_peek_offset_bwd(sk, chunk);

L
Linus Torvalds 已提交
2424
			if (UNIXCB(skb).fp)
C
Christoph Hellwig 已提交
2425
				unix_detach_fds(&scm, skb);
L
Linus Torvalds 已提交
2426

2427
			if (unix_skb_len(skb))
L
Linus Torvalds 已提交
2428 2429
				break;

2430
			skb_unlink(skb, &sk->sk_receive_queue);
2431
			consume_skb(skb);
L
Linus Torvalds 已提交
2432

C
Christoph Hellwig 已提交
2433
			if (scm.fp)
L
Linus Torvalds 已提交
2434
				break;
E
Eric Dumazet 已提交
2435
		} else {
L
Linus Torvalds 已提交
2436 2437 2438
			/* It is questionable, see note in unix_dgram_recvmsg.
			 */
			if (UNIXCB(skb).fp)
C
Christoph Hellwig 已提交
2439
				scm.fp = scm_fp_dup(UNIXCB(skb).fp);
L
Linus Torvalds 已提交
2440

2441
			sk_peek_offset_fwd(sk, chunk);
2442

2443 2444 2445
			if (UNIXCB(skb).fp)
				break;

2446
			skip = 0;
2447 2448 2449 2450 2451 2452 2453
			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
2454 2455 2456 2457
			break;
		}
	} while (size);

I
Ingo Molnar 已提交
2458
	mutex_unlock(&u->readlock);
2459 2460 2461 2462
	if (state->msg)
		scm_recv(sock, state->msg, &scm, flags);
	else
		scm_destroy(&scm);
L
Linus Torvalds 已提交
2463 2464 2465 2466
out:
	return copied ? : err;
}

2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 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 2530 2531 2532 2533 2534 2535 2536 2537
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
	};

	return unix_stream_read_generic(&state);
}

static ssize_t skb_unix_socket_splice(struct sock *sk,
				      struct pipe_inode_info *pipe,
				      struct splice_pipe_desc *spd)
{
	int ret;
	struct unix_sock *u = unix_sk(sk);

	mutex_unlock(&u->readlock);
	ret = splice_to_pipe(pipe, spd);
	mutex_lock(&u->readlock);

	return ret;
}

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,
			       state->pipe, chunk, state->splice_flags,
			       skb_unix_socket_splice);
}

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;

	return unix_stream_read_generic(&state);
}

L
Linus Torvalds 已提交
2538 2539 2540 2541 2542
static int unix_shutdown(struct socket *sock, int mode)
{
	struct sock *sk = sock->sk;
	struct sock *other;

2543 2544 2545 2546 2547 2548 2549 2550
	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;
2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576

	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
2577
	}
2578 2579 2580
	if (other)
		sock_put(other);

L
Linus Torvalds 已提交
2581 2582 2583
	return 0;
}

2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595
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)
2596
			amount += unix_skb_len(skb);
2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613
	} 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
2614 2615 2616
static int unix_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
{
	struct sock *sk = sock->sk;
2617
	long amount = 0;
L
Linus Torvalds 已提交
2618 2619
	int err;

E
Eric Dumazet 已提交
2620 2621
	switch (cmd) {
	case SIOCOUTQ:
2622
		amount = unix_outq_len(sk);
E
Eric Dumazet 已提交
2623 2624 2625
		err = put_user(amount, (int __user *)arg);
		break;
	case SIOCINQ:
2626 2627 2628 2629
		amount = unix_inq_len(sk);
		if (amount < 0)
			err = amount;
		else
L
Linus Torvalds 已提交
2630
			err = put_user(amount, (int __user *)arg);
2631
		break;
E
Eric Dumazet 已提交
2632 2633 2634
	default:
		err = -ENOIOCTLCMD;
		break;
L
Linus Torvalds 已提交
2635 2636 2637 2638
	}
	return err;
}

E
Eric Dumazet 已提交
2639
static unsigned int unix_poll(struct file *file, struct socket *sock, poll_table *wait)
L
Linus Torvalds 已提交
2640 2641 2642 2643
{
	struct sock *sk = sock->sk;
	unsigned int mask;

E
Eric Dumazet 已提交
2644
	sock_poll_wait(file, sk_sleep(sk), wait);
L
Linus Torvalds 已提交
2645 2646 2647 2648 2649 2650 2651
	mask = 0;

	/* exceptional events? */
	if (sk->sk_err)
		mask |= POLLERR;
	if (sk->sk_shutdown == SHUTDOWN_MASK)
		mask |= POLLHUP;
2652
	if (sk->sk_shutdown & RCV_SHUTDOWN)
E
Eric Dumazet 已提交
2653
		mask |= POLLRDHUP | POLLIN | POLLRDNORM;
L
Linus Torvalds 已提交
2654 2655

	/* readable? */
E
Eric Dumazet 已提交
2656
	if (!skb_queue_empty(&sk->sk_receive_queue))
L
Linus Torvalds 已提交
2657 2658 2659
		mask |= POLLIN | POLLRDNORM;

	/* Connection-based need to check for termination and startup */
E
Eric Dumazet 已提交
2660 2661
	if ((sk->sk_type == SOCK_STREAM || sk->sk_type == SOCK_SEQPACKET) &&
	    sk->sk_state == TCP_CLOSE)
L
Linus Torvalds 已提交
2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673
		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;
}

2674 2675
static unsigned int unix_dgram_poll(struct file *file, struct socket *sock,
				    poll_table *wait)
2676
{
2677 2678
	struct sock *sk = sock->sk, *other;
	unsigned int mask, writable;
2679

E
Eric Dumazet 已提交
2680
	sock_poll_wait(file, sk_sleep(sk), wait);
2681 2682 2683 2684
	mask = 0;

	/* exceptional events? */
	if (sk->sk_err || !skb_queue_empty(&sk->sk_error_queue))
2685
		mask |= POLLERR |
2686
			(sock_flag(sk, SOCK_SELECT_ERR_QUEUE) ? POLLPRI : 0);
2687

2688
	if (sk->sk_shutdown & RCV_SHUTDOWN)
2689
		mask |= POLLRDHUP | POLLIN | POLLRDNORM;
2690 2691 2692 2693
	if (sk->sk_shutdown == SHUTDOWN_MASK)
		mask |= POLLHUP;

	/* readable? */
2694
	if (!skb_queue_empty(&sk->sk_receive_queue))
2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705
		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
2706
	/* No write status requested, avoid expensive OUT tests. */
2707
	if (!(poll_requested_events(wait) & (POLLWRBAND|POLLWRNORM|POLLOUT)))
E
Eric Dumazet 已提交
2708 2709
		return mask;

2710
	writable = unix_writable(sk);
2711 2712 2713 2714 2715 2716 2717 2718 2719 2720
	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);
2721 2722 2723
	}

	if (writable)
2724 2725
		mask |= POLLOUT | POLLWRNORM | POLLWRBAND;
	else
2726
		sk_set_bit(SOCKWQ_ASYNC_NOSPACE, sk);
2727 2728 2729

	return mask;
}
L
Linus Torvalds 已提交
2730 2731

#ifdef CONFIG_PROC_FS
2732

E
Eric Dumazet 已提交
2733 2734 2735 2736 2737
#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))
2738

E
Eric Dumazet 已提交
2739
static struct sock *unix_from_bucket(struct seq_file *seq, loff_t *pos)
L
Linus Torvalds 已提交
2740
{
E
Eric Dumazet 已提交
2741 2742 2743 2744
	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
2745

E
Eric Dumazet 已提交
2746 2747
	for (sk = sk_head(&unix_socket_table[bucket]); sk; sk = sk_next(sk)) {
		if (sock_net(sk) != seq_file_net(seq))
2748
			continue;
E
Eric Dumazet 已提交
2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767
		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
2768
	}
E
Eric Dumazet 已提交
2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779

	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
2780 2781 2782 2783
	return NULL;
}

static void *unix_seq_start(struct seq_file *seq, loff_t *pos)
2784
	__acquires(unix_table_lock)
L
Linus Torvalds 已提交
2785
{
2786
	spin_lock(&unix_table_lock);
E
Eric Dumazet 已提交
2787 2788 2789 2790 2791 2792 2793 2794

	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
2795 2796 2797 2798 2799
}

static void *unix_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
	++*pos;
E
Eric Dumazet 已提交
2800
	return unix_next_socket(seq, v, pos);
L
Linus Torvalds 已提交
2801 2802 2803
}

static void unix_seq_stop(struct seq_file *seq, void *v)
2804
	__releases(unix_table_lock)
L
Linus Torvalds 已提交
2805
{
2806
	spin_unlock(&unix_table_lock);
L
Linus Torvalds 已提交
2807 2808 2809 2810
}

static int unix_seq_show(struct seq_file *seq, void *v)
{
2811

J
Joe Perches 已提交
2812
	if (v == SEQ_START_TOKEN)
L
Linus Torvalds 已提交
2813 2814 2815 2816 2817
		seq_puts(seq, "Num       RefCount Protocol Flags    Type St "
			 "Inode Path\n");
	else {
		struct sock *s = v;
		struct unix_sock *u = unix_sk(s);
2818
		unix_state_lock(s);
L
Linus Torvalds 已提交
2819

D
Dan Rosenberg 已提交
2820
		seq_printf(seq, "%pK: %08X %08X %08X %04X %02X %5lu",
L
Linus Torvalds 已提交
2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845
			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]);
		}
2846
		unix_state_unlock(s);
L
Linus Torvalds 已提交
2847 2848 2849 2850 2851 2852
		seq_putc(seq, '\n');
	}

	return 0;
}

2853
static const struct seq_operations unix_seq_ops = {
L
Linus Torvalds 已提交
2854 2855 2856 2857 2858 2859 2860 2861
	.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)
{
2862
	return seq_open_net(inode, file, &unix_seq_ops,
E
Eric Dumazet 已提交
2863
			    sizeof(struct seq_net_private));
L
Linus Torvalds 已提交
2864 2865
}

2866
static const struct file_operations unix_seq_fops = {
L
Linus Torvalds 已提交
2867 2868 2869 2870
	.owner		= THIS_MODULE,
	.open		= unix_seq_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
2871
	.release	= seq_release_net,
L
Linus Torvalds 已提交
2872 2873 2874 2875
};

#endif

2876
static const struct net_proto_family unix_family_ops = {
L
Linus Torvalds 已提交
2877 2878 2879 2880 2881
	.family = PF_UNIX,
	.create = unix_create,
	.owner	= THIS_MODULE,
};

2882

2883
static int __net_init unix_net_init(struct net *net)
2884 2885 2886
{
	int error = -ENOMEM;

2887
	net->unx.sysctl_max_dgram_qlen = 10;
2888 2889
	if (unix_sysctl_register(net))
		goto out;
2890

2891
#ifdef CONFIG_PROC_FS
2892
	if (!proc_create("unix", 0, net->proc_net, &unix_seq_fops)) {
2893
		unix_sysctl_unregister(net);
2894
		goto out;
2895
	}
2896 2897 2898
#endif
	error = 0;
out:
2899
	return error;
2900 2901
}

2902
static void __net_exit unix_net_exit(struct net *net)
2903
{
2904
	unix_sysctl_unregister(net);
2905
	remove_proc_entry("unix", net->proc_net);
2906 2907 2908 2909 2910 2911 2912
}

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

L
Linus Torvalds 已提交
2913 2914 2915 2916
static int __init af_unix_init(void)
{
	int rc = -1;

2917
	BUILD_BUG_ON(sizeof(struct unix_skb_parms) > FIELD_SIZEOF(struct sk_buff, cb));
L
Linus Torvalds 已提交
2918 2919

	rc = proto_register(&unix_proto, 1);
2920
	if (rc != 0) {
W
wangweidong 已提交
2921
		pr_crit("%s: Cannot create unix_sock SLAB cache!\n", __func__);
L
Linus Torvalds 已提交
2922 2923 2924 2925
		goto out;
	}

	sock_register(&unix_family_ops);
2926
	register_pernet_subsys(&unix_net_ops);
L
Linus Torvalds 已提交
2927 2928 2929 2930 2931 2932 2933 2934
out:
	return rc;
}

static void __exit af_unix_exit(void)
{
	sock_unregister(PF_UNIX);
	proto_unregister(&unix_proto);
2935
	unregister_pernet_subsys(&unix_net_ops);
L
Linus Torvalds 已提交
2936 2937
}

2938 2939 2940 2941 2942
/* 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
2943 2944 2945 2946
module_exit(af_unix_exit);

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