sock.c 64.5 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9
/*
 * INET		An implementation of the TCP/IP protocol suite for the LINUX
 *		operating system.  INET is implemented using the  BSD Socket
 *		interface as the means of communication with the user level.
 *
 *		Generic socket support routines. Memory allocators, socket lock/release
 *		handler for protocols to use and generic option handler.
 *
 *
10
 * Authors:	Ross Biro
L
Linus Torvalds 已提交
11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
 *		Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
 *		Florian La Roche, <flla@stud.uni-sb.de>
 *		Alan Cox, <A.Cox@swansea.ac.uk>
 *
 * Fixes:
 *		Alan Cox	: 	Numerous verify_area() problems
 *		Alan Cox	:	Connecting on a connecting socket
 *					now returns an error for tcp.
 *		Alan Cox	:	sock->protocol is set correctly.
 *					and is not sometimes left as 0.
 *		Alan Cox	:	connect handles icmp errors on a
 *					connect properly. Unfortunately there
 *					is a restart syscall nasty there. I
 *					can't match BSD without hacking the C
 *					library. Ideas urgently sought!
 *		Alan Cox	:	Disallow bind() to addresses that are
 *					not ours - especially broadcast ones!!
 *		Alan Cox	:	Socket 1024 _IS_ ok for users. (fencepost)
 *		Alan Cox	:	sock_wfree/sock_rfree don't destroy sockets,
 *					instead they leave that for the DESTROY timer.
 *		Alan Cox	:	Clean up error flag in accept
 *		Alan Cox	:	TCP ack handling is buggy, the DESTROY timer
 *					was buggy. Put a remove_sock() in the handler
 *					for memory when we hit 0. Also altered the timer
35
 *					code. The ACK stuff can wait and needs major
L
Linus Torvalds 已提交
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 83 84 85 86 87 88 89 90 91
 *					TCP layer surgery.
 *		Alan Cox	:	Fixed TCP ack bug, removed remove sock
 *					and fixed timer/inet_bh race.
 *		Alan Cox	:	Added zapped flag for TCP
 *		Alan Cox	:	Move kfree_skb into skbuff.c and tidied up surplus code
 *		Alan Cox	:	for new sk_buff allocations wmalloc/rmalloc now call alloc_skb
 *		Alan Cox	:	kfree_s calls now are kfree_skbmem so we can track skb resources
 *		Alan Cox	:	Supports socket option broadcast now as does udp. Packet and raw need fixing.
 *		Alan Cox	:	Added RCVBUF,SNDBUF size setting. It suddenly occurred to me how easy it was so...
 *		Rick Sladkey	:	Relaxed UDP rules for matching packets.
 *		C.E.Hawkins	:	IFF_PROMISC/SIOCGHWADDR support
 *	Pauline Middelink	:	identd support
 *		Alan Cox	:	Fixed connect() taking signals I think.
 *		Alan Cox	:	SO_LINGER supported
 *		Alan Cox	:	Error reporting fixes
 *		Anonymous	:	inet_create tidied up (sk->reuse setting)
 *		Alan Cox	:	inet sockets don't set sk->type!
 *		Alan Cox	:	Split socket option code
 *		Alan Cox	:	Callbacks
 *		Alan Cox	:	Nagle flag for Charles & Johannes stuff
 *		Alex		:	Removed restriction on inet fioctl
 *		Alan Cox	:	Splitting INET from NET core
 *		Alan Cox	:	Fixed bogus SO_TYPE handling in getsockopt()
 *		Adam Caldwell	:	Missing return in SO_DONTROUTE/SO_DEBUG code
 *		Alan Cox	:	Split IP from generic code
 *		Alan Cox	:	New kfree_skbmem()
 *		Alan Cox	:	Make SO_DEBUG superuser only.
 *		Alan Cox	:	Allow anyone to clear SO_DEBUG
 *					(compatibility fix)
 *		Alan Cox	:	Added optimistic memory grabbing for AF_UNIX throughput.
 *		Alan Cox	:	Allocator for a socket is settable.
 *		Alan Cox	:	SO_ERROR includes soft errors.
 *		Alan Cox	:	Allow NULL arguments on some SO_ opts
 *		Alan Cox	: 	Generic socket allocation to make hooks
 *					easier (suggested by Craig Metz).
 *		Michael Pall	:	SO_ERROR returns positive errno again
 *              Steve Whitehouse:       Added default destructor to free
 *                                      protocol private data.
 *              Steve Whitehouse:       Added various other default routines
 *                                      common to several socket families.
 *              Chris Evans     :       Call suser() check last on F_SETOWN
 *		Jay Schulist	:	Added SO_ATTACH_FILTER and SO_DETACH_FILTER.
 *		Andi Kleen	:	Add sock_kmalloc()/sock_kfree_s()
 *		Andi Kleen	:	Fix write_space callback
 *		Chris Evans	:	Security fixes - signedness again
 *		Arnaldo C. Melo :       cleanups, use skb_queue_purge
 *
 * To Fix:
 *
 *
 *		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.
 */

92
#include <linux/capability.h>
L
Linus Torvalds 已提交
93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
#include <linux/errno.h>
#include <linux/types.h>
#include <linux/socket.h>
#include <linux/in.h>
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/sched.h>
#include <linux/timer.h>
#include <linux/string.h>
#include <linux/sockios.h>
#include <linux/net.h>
#include <linux/mm.h>
#include <linux/slab.h>
#include <linux/interrupt.h>
#include <linux/poll.h>
#include <linux/tcp.h>
#include <linux/init.h>
A
Al Viro 已提交
112
#include <linux/highmem.h>
113
#include <linux/user_namespace.h>
L
Linus Torvalds 已提交
114 115 116 117 118 119 120

#include <asm/uaccess.h>
#include <asm/system.h>

#include <linux/netdevice.h>
#include <net/protocol.h>
#include <linux/skbuff.h>
121
#include <net/net_namespace.h>
122
#include <net/request_sock.h>
L
Linus Torvalds 已提交
123
#include <net/sock.h>
124
#include <linux/net_tstamp.h>
L
Linus Torvalds 已提交
125 126
#include <net/xfrm.h>
#include <linux/ipsec.h>
127
#include <net/cls_cgroup.h>
L
Linus Torvalds 已提交
128 129 130

#include <linux/filter.h>

131 132
#include <trace/events/sock.h>

L
Linus Torvalds 已提交
133 134 135 136
#ifdef CONFIG_INET
#include <net/tcp.h>
#endif

137 138 139 140
/*
 * Each address family might have different locking rules, so we have
 * one slock key per address family:
 */
I
Ingo Molnar 已提交
141 142 143 144 145 146 147 148
static struct lock_class_key af_family_keys[AF_MAX];
static struct lock_class_key af_family_slock_keys[AF_MAX];

/*
 * Make lock validator output more readable. (we pre-construct these
 * strings build-time, so that runtime initialization of socket
 * locks is fast):
 */
149
static const char *const af_family_key_strings[AF_MAX+1] = {
I
Ingo Molnar 已提交
150 151 152 153 154 155 156
  "sk_lock-AF_UNSPEC", "sk_lock-AF_UNIX"     , "sk_lock-AF_INET"     ,
  "sk_lock-AF_AX25"  , "sk_lock-AF_IPX"      , "sk_lock-AF_APPLETALK",
  "sk_lock-AF_NETROM", "sk_lock-AF_BRIDGE"   , "sk_lock-AF_ATMPVC"   ,
  "sk_lock-AF_X25"   , "sk_lock-AF_INET6"    , "sk_lock-AF_ROSE"     ,
  "sk_lock-AF_DECnet", "sk_lock-AF_NETBEUI"  , "sk_lock-AF_SECURITY" ,
  "sk_lock-AF_KEY"   , "sk_lock-AF_NETLINK"  , "sk_lock-AF_PACKET"   ,
  "sk_lock-AF_ASH"   , "sk_lock-AF_ECONET"   , "sk_lock-AF_ATMSVC"   ,
A
Andy Grover 已提交
157
  "sk_lock-AF_RDS"   , "sk_lock-AF_SNA"      , "sk_lock-AF_IRDA"     ,
I
Ingo Molnar 已提交
158
  "sk_lock-AF_PPPOX" , "sk_lock-AF_WANPIPE"  , "sk_lock-AF_LLC"      ,
159
  "sk_lock-27"       , "sk_lock-28"          , "sk_lock-AF_CAN"      ,
160
  "sk_lock-AF_TIPC"  , "sk_lock-AF_BLUETOOTH", "sk_lock-IUCV"        ,
R
Remi Denis-Courmont 已提交
161
  "sk_lock-AF_RXRPC" , "sk_lock-AF_ISDN"     , "sk_lock-AF_PHONET"   ,
162
  "sk_lock-AF_IEEE802154", "sk_lock-AF_CAIF" , "sk_lock-AF_ALG"      ,
A
Aloisio Almeida Jr 已提交
163
  "sk_lock-AF_NFC"   , "sk_lock-AF_MAX"
I
Ingo Molnar 已提交
164
};
165
static const char *const af_family_slock_key_strings[AF_MAX+1] = {
I
Ingo Molnar 已提交
166 167 168 169 170 171 172
  "slock-AF_UNSPEC", "slock-AF_UNIX"     , "slock-AF_INET"     ,
  "slock-AF_AX25"  , "slock-AF_IPX"      , "slock-AF_APPLETALK",
  "slock-AF_NETROM", "slock-AF_BRIDGE"   , "slock-AF_ATMPVC"   ,
  "slock-AF_X25"   , "slock-AF_INET6"    , "slock-AF_ROSE"     ,
  "slock-AF_DECnet", "slock-AF_NETBEUI"  , "slock-AF_SECURITY" ,
  "slock-AF_KEY"   , "slock-AF_NETLINK"  , "slock-AF_PACKET"   ,
  "slock-AF_ASH"   , "slock-AF_ECONET"   , "slock-AF_ATMSVC"   ,
A
Andy Grover 已提交
173
  "slock-AF_RDS"   , "slock-AF_SNA"      , "slock-AF_IRDA"     ,
I
Ingo Molnar 已提交
174
  "slock-AF_PPPOX" , "slock-AF_WANPIPE"  , "slock-AF_LLC"      ,
175
  "slock-27"       , "slock-28"          , "slock-AF_CAN"      ,
176
  "slock-AF_TIPC"  , "slock-AF_BLUETOOTH", "slock-AF_IUCV"     ,
R
Remi Denis-Courmont 已提交
177
  "slock-AF_RXRPC" , "slock-AF_ISDN"     , "slock-AF_PHONET"   ,
178
  "slock-AF_IEEE802154", "slock-AF_CAIF" , "slock-AF_ALG"      ,
A
Aloisio Almeida Jr 已提交
179
  "slock-AF_NFC"   , "slock-AF_MAX"
I
Ingo Molnar 已提交
180
};
181
static const char *const af_family_clock_key_strings[AF_MAX+1] = {
182 183 184 185 186 187 188
  "clock-AF_UNSPEC", "clock-AF_UNIX"     , "clock-AF_INET"     ,
  "clock-AF_AX25"  , "clock-AF_IPX"      , "clock-AF_APPLETALK",
  "clock-AF_NETROM", "clock-AF_BRIDGE"   , "clock-AF_ATMPVC"   ,
  "clock-AF_X25"   , "clock-AF_INET6"    , "clock-AF_ROSE"     ,
  "clock-AF_DECnet", "clock-AF_NETBEUI"  , "clock-AF_SECURITY" ,
  "clock-AF_KEY"   , "clock-AF_NETLINK"  , "clock-AF_PACKET"   ,
  "clock-AF_ASH"   , "clock-AF_ECONET"   , "clock-AF_ATMSVC"   ,
A
Andy Grover 已提交
189
  "clock-AF_RDS"   , "clock-AF_SNA"      , "clock-AF_IRDA"     ,
190
  "clock-AF_PPPOX" , "clock-AF_WANPIPE"  , "clock-AF_LLC"      ,
191
  "clock-27"       , "clock-28"          , "clock-AF_CAN"      ,
192
  "clock-AF_TIPC"  , "clock-AF_BLUETOOTH", "clock-AF_IUCV"     ,
R
Remi Denis-Courmont 已提交
193
  "clock-AF_RXRPC" , "clock-AF_ISDN"     , "clock-AF_PHONET"   ,
194
  "clock-AF_IEEE802154", "clock-AF_CAIF" , "clock-AF_ALG"      ,
A
Aloisio Almeida Jr 已提交
195
  "clock-AF_NFC"   , "clock-AF_MAX"
196
};
197 198 199 200 201 202 203

/*
 * sk_callback_lock locking rules are per-address-family,
 * so split the lock classes by using a per-AF key:
 */
static struct lock_class_key af_callback_keys[AF_MAX];

L
Linus Torvalds 已提交
204 205 206 207 208 209 210 211 212 213 214
/* Take into consideration the size of the struct sk_buff overhead in the
 * determination of these values, since that is non-constant across
 * platforms.  This makes socket queueing behavior and performance
 * not depend upon such differences.
 */
#define _SK_MEM_PACKETS		256
#define _SK_MEM_OVERHEAD	(sizeof(struct sk_buff) + 256)
#define SK_WMEM_MAX		(_SK_MEM_OVERHEAD * _SK_MEM_PACKETS)
#define SK_RMEM_MAX		(_SK_MEM_OVERHEAD * _SK_MEM_PACKETS)

/* Run time adjustable parameters. */
215 216 217 218
__u32 sysctl_wmem_max __read_mostly = SK_WMEM_MAX;
__u32 sysctl_rmem_max __read_mostly = SK_RMEM_MAX;
__u32 sysctl_wmem_default __read_mostly = SK_WMEM_MAX;
__u32 sysctl_rmem_default __read_mostly = SK_RMEM_MAX;
L
Linus Torvalds 已提交
219

L
Lucas De Marchi 已提交
220
/* Maximal space eaten by iovec or ancillary data plus some space */
221
int sysctl_optmem_max __read_mostly = sizeof(unsigned long)*(2*UIO_MAXIOV+512);
E
Eric Dumazet 已提交
222
EXPORT_SYMBOL(sysctl_optmem_max);
L
Linus Torvalds 已提交
223

224 225 226 227 228
#if defined(CONFIG_CGROUPS) && !defined(CONFIG_NET_CLS_CGROUP)
int net_cls_subsys_id = -1;
EXPORT_SYMBOL_GPL(net_cls_subsys_id);
#endif

L
Linus Torvalds 已提交
229 230 231 232 233 234 235 236
static int sock_set_timeout(long *timeo_p, char __user *optval, int optlen)
{
	struct timeval tv;

	if (optlen < sizeof(tv))
		return -EINVAL;
	if (copy_from_user(&tv, optval, sizeof(tv)))
		return -EFAULT;
237 238
	if (tv.tv_usec < 0 || tv.tv_usec >= USEC_PER_SEC)
		return -EDOM;
L
Linus Torvalds 已提交
239

240
	if (tv.tv_sec < 0) {
241 242
		static int warned __read_mostly;

243
		*timeo_p = 0;
244
		if (warned < 10 && net_ratelimit()) {
245 246 247
			warned++;
			printk(KERN_INFO "sock_set_timeout: `%s' (pid %d) "
			       "tries to set negative timeout\n",
248
				current->comm, task_pid_nr(current));
249
		}
250 251
		return 0;
	}
L
Linus Torvalds 已提交
252 253 254 255 256 257 258 259 260 261 262 263
	*timeo_p = MAX_SCHEDULE_TIMEOUT;
	if (tv.tv_sec == 0 && tv.tv_usec == 0)
		return 0;
	if (tv.tv_sec < (MAX_SCHEDULE_TIMEOUT/HZ - 1))
		*timeo_p = tv.tv_sec*HZ + (tv.tv_usec+(1000000/HZ-1))/(1000000/HZ);
	return 0;
}

static void sock_warn_obsolete_bsdism(const char *name)
{
	static int warned;
	static char warncomm[TASK_COMM_LEN];
264 265
	if (strcmp(warncomm, current->comm) && warned < 5) {
		strcpy(warncomm,  current->comm);
L
Linus Torvalds 已提交
266 267 268 269 270 271
		printk(KERN_WARNING "process `%s' is using obsolete "
		       "%s SO_BSDCOMPAT\n", warncomm, name);
		warned++;
	}
}

272
static void sock_disable_timestamp(struct sock *sk, int flag)
273
{
274 275 276 277 278 279
	if (sock_flag(sk, flag)) {
		sock_reset_flag(sk, flag);
		if (!sock_flag(sk, SOCK_TIMESTAMP) &&
		    !sock_flag(sk, SOCK_TIMESTAMPING_RX_SOFTWARE)) {
			net_disable_timestamp();
		}
L
Linus Torvalds 已提交
280 281 282 283
	}
}


284 285
int sock_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
{
E
Eric Dumazet 已提交
286
	int err;
287
	int skb_len;
288 289
	unsigned long flags;
	struct sk_buff_head *list = &sk->sk_receive_queue;
290

R
Rami Rosen 已提交
291
	/* Cast sk->rcvbuf to unsigned... It's pointless, but reduces
292 293 294 295
	   number of warnings when compiling with -W --ANK
	 */
	if (atomic_read(&sk->sk_rmem_alloc) + skb->truesize >=
	    (unsigned)sk->sk_rcvbuf) {
E
Eric Dumazet 已提交
296
		atomic_inc(&sk->sk_drops);
297
		trace_sock_rcvqueue_full(sk, skb);
E
Eric Dumazet 已提交
298
		return -ENOMEM;
299 300
	}

301
	err = sk_filter(sk, skb);
302
	if (err)
E
Eric Dumazet 已提交
303
		return err;
304

305
	if (!sk_rmem_schedule(sk, skb->truesize)) {
E
Eric Dumazet 已提交
306 307
		atomic_inc(&sk->sk_drops);
		return -ENOBUFS;
308 309
	}

310 311
	skb->dev = NULL;
	skb_set_owner_r(skb, sk);
312

313 314 315 316 317 318 319
	/* Cache the SKB length before we tack it onto the receive
	 * queue.  Once it is added it no longer belongs to us and
	 * may be freed by other threads of control pulling packets
	 * from the queue.
	 */
	skb_len = skb->len;

E
Eric Dumazet 已提交
320 321 322 323 324
	/* we escape from rcu protected region, make sure we dont leak
	 * a norefcounted dst
	 */
	skb_dst_force(skb);

325 326 327 328
	spin_lock_irqsave(&list->lock, flags);
	skb->dropcount = atomic_read(&sk->sk_drops);
	__skb_queue_tail(list, skb);
	spin_unlock_irqrestore(&list->lock, flags);
329 330 331

	if (!sock_flag(sk, SOCK_DEAD))
		sk->sk_data_ready(sk, skb_len);
E
Eric Dumazet 已提交
332
	return 0;
333 334 335
}
EXPORT_SYMBOL(sock_queue_rcv_skb);

336
int sk_receive_skb(struct sock *sk, struct sk_buff *skb, const int nested)
337 338 339
{
	int rc = NET_RX_SUCCESS;

340
	if (sk_filter(sk, skb))
341 342 343 344
		goto discard_and_relse;

	skb->dev = NULL;

345 346 347 348
	if (sk_rcvqueues_full(sk, skb)) {
		atomic_inc(&sk->sk_drops);
		goto discard_and_relse;
	}
349 350 351 352
	if (nested)
		bh_lock_sock_nested(sk);
	else
		bh_lock_sock(sk);
I
Ingo Molnar 已提交
353 354 355 356 357 358
	if (!sock_owned_by_user(sk)) {
		/*
		 * trylock + unlock semantics:
		 */
		mutex_acquire(&sk->sk_lock.dep_map, 0, 1, _RET_IP_);

P
Peter Zijlstra 已提交
359
		rc = sk_backlog_rcv(sk, skb);
I
Ingo Molnar 已提交
360 361

		mutex_release(&sk->sk_lock.dep_map, 1, _RET_IP_);
Z
Zhu Yi 已提交
362
	} else if (sk_add_backlog(sk, skb)) {
Z
Zhu Yi 已提交
363 364 365 366 367
		bh_unlock_sock(sk);
		atomic_inc(&sk->sk_drops);
		goto discard_and_relse;
	}

368 369 370 371 372 373 374 375 376 377
	bh_unlock_sock(sk);
out:
	sock_put(sk);
	return rc;
discard_and_relse:
	kfree_skb(skb);
	goto out;
}
EXPORT_SYMBOL(sk_receive_skb);

K
Krishna Kumar 已提交
378 379 380 381 382 383
void sk_reset_txq(struct sock *sk)
{
	sk_tx_queue_clear(sk);
}
EXPORT_SYMBOL(sk_reset_txq);

384 385
struct dst_entry *__sk_dst_check(struct sock *sk, u32 cookie)
{
E
Eric Dumazet 已提交
386
	struct dst_entry *dst = __sk_dst_get(sk);
387 388

	if (dst && dst->obsolete && dst->ops->check(dst, cookie) == NULL) {
K
Krishna Kumar 已提交
389
		sk_tx_queue_clear(sk);
390
		RCU_INIT_POINTER(sk->sk_dst_cache, NULL);
391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412
		dst_release(dst);
		return NULL;
	}

	return dst;
}
EXPORT_SYMBOL(__sk_dst_check);

struct dst_entry *sk_dst_check(struct sock *sk, u32 cookie)
{
	struct dst_entry *dst = sk_dst_get(sk);

	if (dst && dst->obsolete && dst->ops->check(dst, cookie) == NULL) {
		sk_dst_reset(sk);
		dst_release(dst);
		return NULL;
	}

	return dst;
}
EXPORT_SYMBOL(sk_dst_check);

413 414 415 416
static int sock_bindtodevice(struct sock *sk, char __user *optval, int optlen)
{
	int ret = -ENOPROTOOPT;
#ifdef CONFIG_NETDEVICES
417
	struct net *net = sock_net(sk);
418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442
	char devname[IFNAMSIZ];
	int index;

	/* Sorry... */
	ret = -EPERM;
	if (!capable(CAP_NET_RAW))
		goto out;

	ret = -EINVAL;
	if (optlen < 0)
		goto out;

	/* Bind this socket to a particular device like "eth0",
	 * as specified in the passed interface name. If the
	 * name is "" or the option length is zero the socket
	 * is not bound.
	 */
	if (optlen > IFNAMSIZ - 1)
		optlen = IFNAMSIZ - 1;
	memset(devname, 0, sizeof(devname));

	ret = -EFAULT;
	if (copy_from_user(devname, optval, optlen))
		goto out;

443 444
	index = 0;
	if (devname[0] != '\0') {
445
		struct net_device *dev;
446

447 448 449 450 451
		rcu_read_lock();
		dev = dev_get_by_name_rcu(net, devname);
		if (dev)
			index = dev->ifindex;
		rcu_read_unlock();
452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469
		ret = -ENODEV;
		if (!dev)
			goto out;
	}

	lock_sock(sk);
	sk->sk_bound_dev_if = index;
	sk_dst_reset(sk);
	release_sock(sk);

	ret = 0;

out:
#endif

	return ret;
}

470 471 472 473 474 475 476 477
static inline void sock_valbool_flag(struct sock *sk, int bit, int valbool)
{
	if (valbool)
		sock_set_flag(sk, bit);
	else
		sock_reset_flag(sk, bit);
}

L
Linus Torvalds 已提交
478 479 480 481 482 483
/*
 *	This is meant for all protocols to use and covers goings on
 *	at the socket level. Everything here is generic.
 */

int sock_setsockopt(struct socket *sock, int level, int optname,
484
		    char __user *optval, unsigned int optlen)
L
Linus Torvalds 已提交
485
{
E
Eric Dumazet 已提交
486
	struct sock *sk = sock->sk;
L
Linus Torvalds 已提交
487 488 489 490
	int val;
	int valbool;
	struct linger ling;
	int ret = 0;
491

L
Linus Torvalds 已提交
492 493 494 495
	/*
	 *	Options without arguments
	 */

496 497 498
	if (optname == SO_BINDTODEVICE)
		return sock_bindtodevice(sk, optval, optlen);

499 500
	if (optlen < sizeof(int))
		return -EINVAL;
501

L
Linus Torvalds 已提交
502 503
	if (get_user(val, (int __user *)optval))
		return -EFAULT;
504

E
Eric Dumazet 已提交
505
	valbool = val ? 1 : 0;
L
Linus Torvalds 已提交
506 507 508

	lock_sock(sk);

E
Eric Dumazet 已提交
509
	switch (optname) {
510
	case SO_DEBUG:
E
Eric Dumazet 已提交
511
		if (val && !capable(CAP_NET_ADMIN))
512
			ret = -EACCES;
E
Eric Dumazet 已提交
513
		else
514
			sock_valbool_flag(sk, SOCK_DBG, valbool);
515 516 517 518 519
		break;
	case SO_REUSEADDR:
		sk->sk_reuse = valbool;
		break;
	case SO_TYPE:
520
	case SO_PROTOCOL:
521
	case SO_DOMAIN:
522 523 524 525
	case SO_ERROR:
		ret = -ENOPROTOOPT;
		break;
	case SO_DONTROUTE:
526
		sock_valbool_flag(sk, SOCK_LOCALROUTE, valbool);
527 528 529 530 531 532 533 534 535 536 537 538
		break;
	case SO_BROADCAST:
		sock_valbool_flag(sk, SOCK_BROADCAST, valbool);
		break;
	case SO_SNDBUF:
		/* Don't error on this BSD doesn't and if you think
		   about it this is right. Otherwise apps have to
		   play 'guess the biggest size' games. RCVBUF/SNDBUF
		   are treated in BSD as hints */

		if (val > sysctl_wmem_max)
			val = sysctl_wmem_max;
539
set_sndbuf:
540 541 542 543 544
		sk->sk_userlocks |= SOCK_SNDBUF_LOCK;
		if ((val * 2) < SOCK_MIN_SNDBUF)
			sk->sk_sndbuf = SOCK_MIN_SNDBUF;
		else
			sk->sk_sndbuf = val * 2;
L
Linus Torvalds 已提交
545

546 547 548 549 550 551
		/*
		 *	Wake up sending tasks if we
		 *	upped the value.
		 */
		sk->sk_write_space(sk);
		break;
L
Linus Torvalds 已提交
552

553 554 555 556 557 558
	case SO_SNDBUFFORCE:
		if (!capable(CAP_NET_ADMIN)) {
			ret = -EPERM;
			break;
		}
		goto set_sndbuf;
559

560 561 562 563 564
	case SO_RCVBUF:
		/* Don't error on this BSD doesn't and if you think
		   about it this is right. Otherwise apps have to
		   play 'guess the biggest size' games. RCVBUF/SNDBUF
		   are treated in BSD as hints */
565

566 567
		if (val > sysctl_rmem_max)
			val = sysctl_rmem_max;
568
set_rcvbuf:
569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593
		sk->sk_userlocks |= SOCK_RCVBUF_LOCK;
		/*
		 * We double it on the way in to account for
		 * "struct sk_buff" etc. overhead.   Applications
		 * assume that the SO_RCVBUF setting they make will
		 * allow that much actual data to be received on that
		 * socket.
		 *
		 * Applications are unaware that "struct sk_buff" and
		 * other overheads allocate from the receive buffer
		 * during socket buffer allocation.
		 *
		 * And after considering the possible alternatives,
		 * returning the value we actually used in getsockopt
		 * is the most desirable behavior.
		 */
		if ((val * 2) < SOCK_MIN_RCVBUF)
			sk->sk_rcvbuf = SOCK_MIN_RCVBUF;
		else
			sk->sk_rcvbuf = val * 2;
		break;

	case SO_RCVBUFFORCE:
		if (!capable(CAP_NET_ADMIN)) {
			ret = -EPERM;
L
Linus Torvalds 已提交
594
			break;
595 596
		}
		goto set_rcvbuf;
L
Linus Torvalds 已提交
597

598
	case SO_KEEPALIVE:
L
Linus Torvalds 已提交
599
#ifdef CONFIG_INET
600 601
		if (sk->sk_protocol == IPPROTO_TCP)
			tcp_set_keepalive(sk, valbool);
L
Linus Torvalds 已提交
602
#endif
603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623
		sock_valbool_flag(sk, SOCK_KEEPOPEN, valbool);
		break;

	case SO_OOBINLINE:
		sock_valbool_flag(sk, SOCK_URGINLINE, valbool);
		break;

	case SO_NO_CHECK:
		sk->sk_no_check = valbool;
		break;

	case SO_PRIORITY:
		if ((val >= 0 && val <= 6) || capable(CAP_NET_ADMIN))
			sk->sk_priority = val;
		else
			ret = -EPERM;
		break;

	case SO_LINGER:
		if (optlen < sizeof(ling)) {
			ret = -EINVAL;	/* 1003.1g */
L
Linus Torvalds 已提交
624
			break;
625
		}
E
Eric Dumazet 已提交
626
		if (copy_from_user(&ling, optval, sizeof(ling))) {
627
			ret = -EFAULT;
L
Linus Torvalds 已提交
628
			break;
629 630 631 632
		}
		if (!ling.l_onoff)
			sock_reset_flag(sk, SOCK_LINGER);
		else {
L
Linus Torvalds 已提交
633
#if (BITS_PER_LONG == 32)
634 635
			if ((unsigned int)ling.l_linger >= MAX_SCHEDULE_TIMEOUT/HZ)
				sk->sk_lingertime = MAX_SCHEDULE_TIMEOUT;
L
Linus Torvalds 已提交
636
			else
637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654
#endif
				sk->sk_lingertime = (unsigned int)ling.l_linger * HZ;
			sock_set_flag(sk, SOCK_LINGER);
		}
		break;

	case SO_BSDCOMPAT:
		sock_warn_obsolete_bsdism("setsockopt");
		break;

	case SO_PASSCRED:
		if (valbool)
			set_bit(SOCK_PASSCRED, &sock->flags);
		else
			clear_bit(SOCK_PASSCRED, &sock->flags);
		break;

	case SO_TIMESTAMP:
655
	case SO_TIMESTAMPNS:
656
		if (valbool)  {
657 658 659 660
			if (optname == SO_TIMESTAMP)
				sock_reset_flag(sk, SOCK_RCVTSTAMPNS);
			else
				sock_set_flag(sk, SOCK_RCVTSTAMPNS);
661
			sock_set_flag(sk, SOCK_RCVTSTAMP);
662
			sock_enable_timestamp(sk, SOCK_TIMESTAMP);
663
		} else {
664
			sock_reset_flag(sk, SOCK_RCVTSTAMP);
665 666
			sock_reset_flag(sk, SOCK_RCVTSTAMPNS);
		}
667 668
		break;

669 670
	case SO_TIMESTAMPING:
		if (val & ~SOF_TIMESTAMPING_MASK) {
671
			ret = -EINVAL;
672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693
			break;
		}
		sock_valbool_flag(sk, SOCK_TIMESTAMPING_TX_HARDWARE,
				  val & SOF_TIMESTAMPING_TX_HARDWARE);
		sock_valbool_flag(sk, SOCK_TIMESTAMPING_TX_SOFTWARE,
				  val & SOF_TIMESTAMPING_TX_SOFTWARE);
		sock_valbool_flag(sk, SOCK_TIMESTAMPING_RX_HARDWARE,
				  val & SOF_TIMESTAMPING_RX_HARDWARE);
		if (val & SOF_TIMESTAMPING_RX_SOFTWARE)
			sock_enable_timestamp(sk,
					      SOCK_TIMESTAMPING_RX_SOFTWARE);
		else
			sock_disable_timestamp(sk,
					       SOCK_TIMESTAMPING_RX_SOFTWARE);
		sock_valbool_flag(sk, SOCK_TIMESTAMPING_SOFTWARE,
				  val & SOF_TIMESTAMPING_SOFTWARE);
		sock_valbool_flag(sk, SOCK_TIMESTAMPING_SYS_HARDWARE,
				  val & SOF_TIMESTAMPING_SYS_HARDWARE);
		sock_valbool_flag(sk, SOCK_TIMESTAMPING_RAW_HARDWARE,
				  val & SOF_TIMESTAMPING_RAW_HARDWARE);
		break;

694 695 696 697 698 699 700 701 702 703 704 705 706
	case SO_RCVLOWAT:
		if (val < 0)
			val = INT_MAX;
		sk->sk_rcvlowat = val ? : 1;
		break;

	case SO_RCVTIMEO:
		ret = sock_set_timeout(&sk->sk_rcvtimeo, optval, optlen);
		break;

	case SO_SNDTIMEO:
		ret = sock_set_timeout(&sk->sk_sndtimeo, optval, optlen);
		break;
L
Linus Torvalds 已提交
707

708 709 710 711
	case SO_ATTACH_FILTER:
		ret = -EINVAL;
		if (optlen == sizeof(struct sock_fprog)) {
			struct sock_fprog fprog;
L
Linus Torvalds 已提交
712

713 714
			ret = -EFAULT;
			if (copy_from_user(&fprog, optval, sizeof(fprog)))
L
Linus Torvalds 已提交
715
				break;
716 717 718 719 720 721

			ret = sk_attach_filter(&fprog, sk);
		}
		break;

	case SO_DETACH_FILTER:
722
		ret = sk_detach_filter(sk);
723
		break;
L
Linus Torvalds 已提交
724

725 726 727 728 729 730
	case SO_PASSSEC:
		if (valbool)
			set_bit(SOCK_PASSSEC, &sock->flags);
		else
			clear_bit(SOCK_PASSSEC, &sock->flags);
		break;
731 732 733
	case SO_MARK:
		if (!capable(CAP_NET_ADMIN))
			ret = -EPERM;
E
Eric Dumazet 已提交
734
		else
735 736
			sk->sk_mark = val;
		break;
C
Catherine Zhang 已提交
737

L
Linus Torvalds 已提交
738 739
		/* We implement the SO_SNDLOWAT etc to
		   not be settable (1003.1g 5.3) */
740
	case SO_RXQ_OVFL:
741
		sock_valbool_flag(sk, SOCK_RXQ_OVFL, valbool);
742
		break;
743 744 745
	default:
		ret = -ENOPROTOOPT;
		break;
746
	}
L
Linus Torvalds 已提交
747 748 749
	release_sock(sk);
	return ret;
}
E
Eric Dumazet 已提交
750
EXPORT_SYMBOL(sock_setsockopt);
L
Linus Torvalds 已提交
751 752


753 754 755 756 757 758 759 760 761 762 763 764
void cred_to_ucred(struct pid *pid, const struct cred *cred,
		   struct ucred *ucred)
{
	ucred->pid = pid_vnr(pid);
	ucred->uid = ucred->gid = -1;
	if (cred) {
		struct user_namespace *current_ns = current_user_ns();

		ucred->uid = user_ns_map_uid(current_ns, cred, cred->euid);
		ucred->gid = user_ns_map_gid(current_ns, cred, cred->egid);
	}
}
765
EXPORT_SYMBOL_GPL(cred_to_ucred);
766

L
Linus Torvalds 已提交
767 768 769 770
int sock_getsockopt(struct socket *sock, int level, int optname,
		    char __user *optval, int __user *optlen)
{
	struct sock *sk = sock->sk;
771

772
	union {
773 774
		int val;
		struct linger ling;
L
Linus Torvalds 已提交
775 776
		struct timeval tm;
	} v;
777

778
	int lv = sizeof(int);
L
Linus Torvalds 已提交
779
	int len;
780

781
	if (get_user(len, optlen))
782
		return -EFAULT;
783
	if (len < 0)
L
Linus Torvalds 已提交
784
		return -EINVAL;
785

786
	memset(&v, 0, sizeof(v));
787

E
Eric Dumazet 已提交
788
	switch (optname) {
789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820
	case SO_DEBUG:
		v.val = sock_flag(sk, SOCK_DBG);
		break;

	case SO_DONTROUTE:
		v.val = sock_flag(sk, SOCK_LOCALROUTE);
		break;

	case SO_BROADCAST:
		v.val = !!sock_flag(sk, SOCK_BROADCAST);
		break;

	case SO_SNDBUF:
		v.val = sk->sk_sndbuf;
		break;

	case SO_RCVBUF:
		v.val = sk->sk_rcvbuf;
		break;

	case SO_REUSEADDR:
		v.val = sk->sk_reuse;
		break;

	case SO_KEEPALIVE:
		v.val = !!sock_flag(sk, SOCK_KEEPOPEN);
		break;

	case SO_TYPE:
		v.val = sk->sk_type;
		break;

821 822 823 824
	case SO_PROTOCOL:
		v.val = sk->sk_protocol;
		break;

825 826 827 828
	case SO_DOMAIN:
		v.val = sk->sk_family;
		break;

829 830
	case SO_ERROR:
		v.val = -sock_error(sk);
E
Eric Dumazet 已提交
831
		if (v.val == 0)
832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857
			v.val = xchg(&sk->sk_err_soft, 0);
		break;

	case SO_OOBINLINE:
		v.val = !!sock_flag(sk, SOCK_URGINLINE);
		break;

	case SO_NO_CHECK:
		v.val = sk->sk_no_check;
		break;

	case SO_PRIORITY:
		v.val = sk->sk_priority;
		break;

	case SO_LINGER:
		lv		= sizeof(v.ling);
		v.ling.l_onoff	= !!sock_flag(sk, SOCK_LINGER);
		v.ling.l_linger	= sk->sk_lingertime / HZ;
		break;

	case SO_BSDCOMPAT:
		sock_warn_obsolete_bsdism("getsockopt");
		break;

	case SO_TIMESTAMP:
858 859 860 861 862 863
		v.val = sock_flag(sk, SOCK_RCVTSTAMP) &&
				!sock_flag(sk, SOCK_RCVTSTAMPNS);
		break;

	case SO_TIMESTAMPNS:
		v.val = sock_flag(sk, SOCK_RCVTSTAMPNS);
864 865
		break;

866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883
	case SO_TIMESTAMPING:
		v.val = 0;
		if (sock_flag(sk, SOCK_TIMESTAMPING_TX_HARDWARE))
			v.val |= SOF_TIMESTAMPING_TX_HARDWARE;
		if (sock_flag(sk, SOCK_TIMESTAMPING_TX_SOFTWARE))
			v.val |= SOF_TIMESTAMPING_TX_SOFTWARE;
		if (sock_flag(sk, SOCK_TIMESTAMPING_RX_HARDWARE))
			v.val |= SOF_TIMESTAMPING_RX_HARDWARE;
		if (sock_flag(sk, SOCK_TIMESTAMPING_RX_SOFTWARE))
			v.val |= SOF_TIMESTAMPING_RX_SOFTWARE;
		if (sock_flag(sk, SOCK_TIMESTAMPING_SOFTWARE))
			v.val |= SOF_TIMESTAMPING_SOFTWARE;
		if (sock_flag(sk, SOCK_TIMESTAMPING_SYS_HARDWARE))
			v.val |= SOF_TIMESTAMPING_SYS_HARDWARE;
		if (sock_flag(sk, SOCK_TIMESTAMPING_RAW_HARDWARE))
			v.val |= SOF_TIMESTAMPING_RAW_HARDWARE;
		break;

884
	case SO_RCVTIMEO:
E
Eric Dumazet 已提交
885
		lv = sizeof(struct timeval);
886 887 888 889 890 891 892 893 894 895
		if (sk->sk_rcvtimeo == MAX_SCHEDULE_TIMEOUT) {
			v.tm.tv_sec = 0;
			v.tm.tv_usec = 0;
		} else {
			v.tm.tv_sec = sk->sk_rcvtimeo / HZ;
			v.tm.tv_usec = ((sk->sk_rcvtimeo % HZ) * 1000000) / HZ;
		}
		break;

	case SO_SNDTIMEO:
E
Eric Dumazet 已提交
896
		lv = sizeof(struct timeval);
897 898 899 900 901 902 903 904
		if (sk->sk_sndtimeo == MAX_SCHEDULE_TIMEOUT) {
			v.tm.tv_sec = 0;
			v.tm.tv_usec = 0;
		} else {
			v.tm.tv_sec = sk->sk_sndtimeo / HZ;
			v.tm.tv_usec = ((sk->sk_sndtimeo % HZ) * 1000000) / HZ;
		}
		break;
L
Linus Torvalds 已提交
905

906 907 908
	case SO_RCVLOWAT:
		v.val = sk->sk_rcvlowat;
		break;
L
Linus Torvalds 已提交
909

910
	case SO_SNDLOWAT:
E
Eric Dumazet 已提交
911
		v.val = 1;
912
		break;
L
Linus Torvalds 已提交
913

914 915 916
	case SO_PASSCRED:
		v.val = test_bit(SOCK_PASSCRED, &sock->flags) ? 1 : 0;
		break;
L
Linus Torvalds 已提交
917

918
	case SO_PEERCRED:
919 920 921 922 923 924
	{
		struct ucred peercred;
		if (len > sizeof(peercred))
			len = sizeof(peercred);
		cred_to_ucred(sk->sk_peer_pid, sk->sk_peer_cred, &peercred);
		if (copy_to_user(optval, &peercred, len))
925 926
			return -EFAULT;
		goto lenout;
927
	}
L
Linus Torvalds 已提交
928

929 930 931 932 933 934 935 936 937 938 939 940
	case SO_PEERNAME:
	{
		char address[128];

		if (sock->ops->getname(sock, (struct sockaddr *)address, &lv, 2))
			return -ENOTCONN;
		if (lv < len)
			return -EINVAL;
		if (copy_to_user(optval, address, len))
			return -EFAULT;
		goto lenout;
	}
L
Linus Torvalds 已提交
941

942 943 944 945 946 947
	/* Dubious BSD thing... Probably nobody even uses it, but
	 * the UNIX standard wants it for whatever reason... -DaveM
	 */
	case SO_ACCEPTCONN:
		v.val = sk->sk_state == TCP_LISTEN;
		break;
L
Linus Torvalds 已提交
948

949 950 951
	case SO_PASSSEC:
		v.val = test_bit(SOCK_PASSSEC, &sock->flags) ? 1 : 0;
		break;
C
Catherine Zhang 已提交
952

953 954
	case SO_PEERSEC:
		return security_socket_getpeersec_stream(sock, optval, optlen, len);
L
Linus Torvalds 已提交
955

956 957 958 959
	case SO_MARK:
		v.val = sk->sk_mark;
		break;

960 961 962 963
	case SO_RXQ_OVFL:
		v.val = !!sock_flag(sk, SOCK_RXQ_OVFL);
		break;

964 965
	default:
		return -ENOPROTOOPT;
L
Linus Torvalds 已提交
966
	}
967

L
Linus Torvalds 已提交
968 969 970 971 972
	if (len > lv)
		len = lv;
	if (copy_to_user(optval, &v, len))
		return -EFAULT;
lenout:
973 974 975
	if (put_user(len, optlen))
		return -EFAULT;
	return 0;
L
Linus Torvalds 已提交
976 977
}

I
Ingo Molnar 已提交
978 979 980 981 982
/*
 * Initialize an sk_lock.
 *
 * (We also register the sk_lock with the lock validator.)
 */
D
Dave Jones 已提交
983
static inline void sock_lock_init(struct sock *sk)
I
Ingo Molnar 已提交
984
{
985 986 987 988 989
	sock_lock_init_class_and_name(sk,
			af_family_slock_key_strings[sk->sk_family],
			af_family_slock_keys + sk->sk_family,
			af_family_key_strings[sk->sk_family],
			af_family_keys + sk->sk_family);
I
Ingo Molnar 已提交
990 991
}

E
Eric Dumazet 已提交
992 993 994
/*
 * Copy all fields from osk to nsk but nsk->sk_refcnt must not change yet,
 * even temporarly, because of RCU lookups. sk_node should also be left as is.
995
 * We must not copy fields between sk_dontcopy_begin and sk_dontcopy_end
E
Eric Dumazet 已提交
996
 */
997 998 999 1000 1001
static void sock_copy(struct sock *nsk, const struct sock *osk)
{
#ifdef CONFIG_SECURITY_NETWORK
	void *sptr = nsk->sk_security;
#endif
1002 1003 1004 1005 1006
	memcpy(nsk, osk, offsetof(struct sock, sk_dontcopy_begin));

	memcpy(&nsk->sk_dontcopy_end, &osk->sk_dontcopy_end,
	       osk->sk_prot->obj_size - offsetof(struct sock, sk_dontcopy_end));

1007 1008 1009 1010 1011 1012
#ifdef CONFIG_SECURITY_NETWORK
	nsk->sk_security = sptr;
	security_sk_clone(osk, nsk);
#endif
}

1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042
/*
 * caches using SLAB_DESTROY_BY_RCU should let .next pointer from nulls nodes
 * un-modified. Special care is taken when initializing object to zero.
 */
static inline void sk_prot_clear_nulls(struct sock *sk, int size)
{
	if (offsetof(struct sock, sk_node.next) != 0)
		memset(sk, 0, offsetof(struct sock, sk_node.next));
	memset(&sk->sk_node.pprev, 0,
	       size - offsetof(struct sock, sk_node.pprev));
}

void sk_prot_clear_portaddr_nulls(struct sock *sk, int size)
{
	unsigned long nulls1, nulls2;

	nulls1 = offsetof(struct sock, __sk_common.skc_node.next);
	nulls2 = offsetof(struct sock, __sk_common.skc_portaddr_node.next);
	if (nulls1 > nulls2)
		swap(nulls1, nulls2);

	if (nulls1 != 0)
		memset((char *)sk, 0, nulls1);
	memset((char *)sk + nulls1 + sizeof(void *), 0,
	       nulls2 - nulls1 - sizeof(void *));
	memset((char *)sk + nulls2 + sizeof(void *), 0,
	       size - nulls2 - sizeof(void *));
}
EXPORT_SYMBOL(sk_prot_clear_portaddr_nulls);

1043 1044
static struct sock *sk_prot_alloc(struct proto *prot, gfp_t priority,
		int family)
1045 1046 1047 1048 1049
{
	struct sock *sk;
	struct kmem_cache *slab;

	slab = prot->slab;
1050 1051 1052 1053 1054
	if (slab != NULL) {
		sk = kmem_cache_alloc(slab, priority & ~__GFP_ZERO);
		if (!sk)
			return sk;
		if (priority & __GFP_ZERO) {
1055 1056 1057 1058
			if (prot->clear_sk)
				prot->clear_sk(sk, prot->obj_size);
			else
				sk_prot_clear_nulls(sk, prot->obj_size);
1059
		}
1060
	} else
1061 1062
		sk = kmalloc(prot->obj_size, priority);

1063
	if (sk != NULL) {
1064 1065
		kmemcheck_annotate_bitfield(sk, flags);

1066 1067 1068 1069 1070
		if (security_sk_alloc(sk, family, priority))
			goto out_free;

		if (!try_module_get(prot->owner))
			goto out_free_sec;
K
Krishna Kumar 已提交
1071
		sk_tx_queue_clear(sk);
1072 1073
	}

1074
	return sk;
1075 1076 1077 1078 1079 1080 1081 1082 1083

out_free_sec:
	security_sk_free(sk);
out_free:
	if (slab != NULL)
		kmem_cache_free(slab, sk);
	else
		kfree(sk);
	return NULL;
1084 1085 1086 1087 1088
}

static void sk_prot_free(struct proto *prot, struct sock *sk)
{
	struct kmem_cache *slab;
1089
	struct module *owner;
1090

1091
	owner = prot->owner;
1092
	slab = prot->slab;
1093 1094

	security_sk_free(sk);
1095 1096 1097 1098
	if (slab != NULL)
		kmem_cache_free(slab, sk);
	else
		kfree(sk);
1099
	module_put(owner);
1100 1101
}

1102 1103 1104
#ifdef CONFIG_CGROUPS
void sock_update_classid(struct sock *sk)
{
1105
	u32 classid;
1106

1107 1108 1109
	rcu_read_lock();  /* doing current task, which cannot vanish. */
	classid = task_cls_classid(current);
	rcu_read_unlock();
1110 1111 1112
	if (classid && classid != sk->sk_classid)
		sk->sk_classid = classid;
}
1113
EXPORT_SYMBOL(sock_update_classid);
1114 1115
#endif

L
Linus Torvalds 已提交
1116 1117
/**
 *	sk_alloc - All socket objects are allocated here
1118
 *	@net: the applicable net namespace
1119 1120 1121
 *	@family: protocol family
 *	@priority: for allocation (%GFP_KERNEL, %GFP_ATOMIC, etc)
 *	@prot: struct proto associated with this new sock instance
L
Linus Torvalds 已提交
1122
 */
1123
struct sock *sk_alloc(struct net *net, int family, gfp_t priority,
1124
		      struct proto *prot)
L
Linus Torvalds 已提交
1125
{
1126
	struct sock *sk;
L
Linus Torvalds 已提交
1127

1128
	sk = sk_prot_alloc(prot, priority | __GFP_ZERO, family);
L
Linus Torvalds 已提交
1129
	if (sk) {
1130 1131 1132 1133 1134 1135 1136
		sk->sk_family = family;
		/*
		 * See comment in struct sock definition to understand
		 * why we need sk_prot_creator -acme
		 */
		sk->sk_prot = sk->sk_prot_creator = prot;
		sock_lock_init(sk);
1137
		sock_net_set(sk, get_net(net));
1138
		atomic_set(&sk->sk_wmem_alloc, 1);
1139 1140

		sock_update_classid(sk);
L
Linus Torvalds 已提交
1141
	}
1142

1143
	return sk;
L
Linus Torvalds 已提交
1144
}
E
Eric Dumazet 已提交
1145
EXPORT_SYMBOL(sk_alloc);
L
Linus Torvalds 已提交
1146

1147
static void __sk_free(struct sock *sk)
L
Linus Torvalds 已提交
1148 1149 1150 1151 1152 1153
{
	struct sk_filter *filter;

	if (sk->sk_destruct)
		sk->sk_destruct(sk);

1154 1155
	filter = rcu_dereference_check(sk->sk_filter,
				       atomic_read(&sk->sk_wmem_alloc) == 0);
L
Linus Torvalds 已提交
1156
	if (filter) {
1157
		sk_filter_uncharge(sk, filter);
1158
		RCU_INIT_POINTER(sk->sk_filter, NULL);
L
Linus Torvalds 已提交
1159 1160
	}

1161 1162
	sock_disable_timestamp(sk, SOCK_TIMESTAMP);
	sock_disable_timestamp(sk, SOCK_TIMESTAMPING_RX_SOFTWARE);
L
Linus Torvalds 已提交
1163 1164 1165

	if (atomic_read(&sk->sk_omem_alloc))
		printk(KERN_DEBUG "%s: optmem leakage (%d bytes) detected.\n",
1166
		       __func__, atomic_read(&sk->sk_omem_alloc));
L
Linus Torvalds 已提交
1167

1168 1169 1170
	if (sk->sk_peer_cred)
		put_cred(sk->sk_peer_cred);
	put_pid(sk->sk_peer_pid);
1171
	put_net(sock_net(sk));
1172
	sk_prot_free(sk->sk_prot_creator, sk);
L
Linus Torvalds 已提交
1173
}
1174 1175 1176 1177

void sk_free(struct sock *sk)
{
	/*
L
Lucas De Marchi 已提交
1178
	 * We subtract one from sk_wmem_alloc and can know if
1179 1180 1181 1182 1183 1184
	 * some packets are still in some tx queue.
	 * If not null, sock_wfree() will call __sk_free(sk) later
	 */
	if (atomic_dec_and_test(&sk->sk_wmem_alloc))
		__sk_free(sk);
}
E
Eric Dumazet 已提交
1185
EXPORT_SYMBOL(sk_free);
L
Linus Torvalds 已提交
1186

1187
/*
L
Lucas De Marchi 已提交
1188 1189
 * Last sock_put should drop reference to sk->sk_net. It has already
 * been dropped in sk_change_net. Taking reference to stopping namespace
1190
 * is not an option.
L
Lucas De Marchi 已提交
1191
 * Take reference to a socket to remove it from hash _alive_ and after that
1192 1193 1194 1195 1196 1197 1198 1199 1200
 * destroy it in the context of init_net.
 */
void sk_release_kernel(struct sock *sk)
{
	if (sk == NULL || sk->sk_socket == NULL)
		return;

	sock_hold(sk);
	sock_release(sk->sk_socket);
1201
	release_net(sock_net(sk));
1202
	sock_net_set(sk, get_net(&init_net));
1203 1204
	sock_put(sk);
}
1205
EXPORT_SYMBOL(sk_release_kernel);
1206

A
Al Viro 已提交
1207
struct sock *sk_clone(const struct sock *sk, const gfp_t priority)
1208
{
1209
	struct sock *newsk;
1210

1211
	newsk = sk_prot_alloc(sk->sk_prot, priority, sk->sk_family);
1212 1213 1214
	if (newsk != NULL) {
		struct sk_filter *filter;

1215
		sock_copy(newsk, sk);
1216 1217

		/* SANITY */
1218
		get_net(sock_net(newsk));
1219 1220 1221
		sk_node_init(&newsk->sk_node);
		sock_lock_init(newsk);
		bh_lock_sock(newsk);
1222
		newsk->sk_backlog.head	= newsk->sk_backlog.tail = NULL;
Z
Zhu Yi 已提交
1223
		newsk->sk_backlog.len = 0;
1224 1225

		atomic_set(&newsk->sk_rmem_alloc, 0);
1226 1227 1228 1229
		/*
		 * sk_wmem_alloc set to one (see sk_free() and sock_wfree())
		 */
		atomic_set(&newsk->sk_wmem_alloc, 1);
1230 1231 1232
		atomic_set(&newsk->sk_omem_alloc, 0);
		skb_queue_head_init(&newsk->sk_receive_queue);
		skb_queue_head_init(&newsk->sk_write_queue);
1233 1234 1235
#ifdef CONFIG_NET_DMA
		skb_queue_head_init(&newsk->sk_async_wait_queue);
#endif
1236

E
Eric Dumazet 已提交
1237
		spin_lock_init(&newsk->sk_dst_lock);
1238
		rwlock_init(&newsk->sk_callback_lock);
1239 1240 1241
		lockdep_set_class_and_name(&newsk->sk_callback_lock,
				af_callback_keys + newsk->sk_family,
				af_family_clock_key_strings[newsk->sk_family]);
1242 1243 1244 1245 1246 1247 1248 1249 1250 1251

		newsk->sk_dst_cache	= NULL;
		newsk->sk_wmem_queued	= 0;
		newsk->sk_forward_alloc = 0;
		newsk->sk_send_head	= NULL;
		newsk->sk_userlocks	= sk->sk_userlocks & ~SOCK_BINDPORT_LOCK;

		sock_reset_flag(newsk, SOCK_DONE);
		skb_queue_head_init(&newsk->sk_error_queue);

1252
		filter = rcu_dereference_protected(newsk->sk_filter, 1);
1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266
		if (filter != NULL)
			sk_filter_charge(newsk, filter);

		if (unlikely(xfrm_sk_clone_policy(newsk))) {
			/* It is still raw copy of parent, so invalidate
			 * destructor and make plain sk_free() */
			newsk->sk_destruct = NULL;
			sk_free(newsk);
			newsk = NULL;
			goto out;
		}

		newsk->sk_err	   = 0;
		newsk->sk_priority = 0;
E
Eric Dumazet 已提交
1267 1268 1269 1270 1271
		/*
		 * Before updating sk_refcnt, we must commit prior changes to memory
		 * (Documentation/RCU/rculist_nulls.txt for details)
		 */
		smp_wmb();
1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285
		atomic_set(&newsk->sk_refcnt, 2);

		/*
		 * Increment the counter in the same struct proto as the master
		 * sock (sk_refcnt_debug_inc uses newsk->sk_prot->socks, that
		 * is the same as sk->sk_prot->socks, as this field was copied
		 * with memcpy).
		 *
		 * This _changes_ the previous behaviour, where
		 * tcp_create_openreq_child always was incrementing the
		 * equivalent to tcp_prot->socks (inet_sock_nr), so this have
		 * to be taken into account in all callers. -acme
		 */
		sk_refcnt_debug_inc(newsk);
1286
		sk_set_socket(newsk, NULL);
1287
		newsk->sk_wq = NULL;
1288 1289

		if (newsk->sk_prot->sockets_allocated)
1290
			percpu_counter_inc(newsk->sk_prot->sockets_allocated);
1291 1292 1293 1294

		if (sock_flag(newsk, SOCK_TIMESTAMP) ||
		    sock_flag(newsk, SOCK_TIMESTAMPING_RX_SOFTWARE))
			net_enable_timestamp();
1295 1296 1297 1298 1299 1300
	}
out:
	return newsk;
}
EXPORT_SYMBOL_GPL(sk_clone);

1301 1302 1303 1304 1305
void sk_setup_caps(struct sock *sk, struct dst_entry *dst)
{
	__sk_dst_set(sk, dst);
	sk->sk_route_caps = dst->dev->features;
	if (sk->sk_route_caps & NETIF_F_GSO)
1306
		sk->sk_route_caps |= NETIF_F_GSO_SOFTWARE;
E
Eric Dumazet 已提交
1307
	sk->sk_route_caps &= ~sk->sk_route_nocaps;
1308
	if (sk_can_gso(sk)) {
1309
		if (dst->header_len) {
1310
			sk->sk_route_caps &= ~NETIF_F_GSO_MASK;
1311
		} else {
1312
			sk->sk_route_caps |= NETIF_F_SG | NETIF_F_HW_CSUM;
1313 1314
			sk->sk_gso_max_size = dst->dev->gso_max_size;
		}
1315 1316 1317 1318
	}
}
EXPORT_SYMBOL_GPL(sk_setup_caps);

L
Linus Torvalds 已提交
1319 1320
void __init sk_init(void)
{
1321
	if (totalram_pages <= 4096) {
L
Linus Torvalds 已提交
1322 1323 1324 1325
		sysctl_wmem_max = 32767;
		sysctl_rmem_max = 32767;
		sysctl_wmem_default = 32767;
		sysctl_rmem_default = 32767;
1326
	} else if (totalram_pages >= 131072) {
L
Linus Torvalds 已提交
1327 1328 1329 1330 1331 1332 1333 1334 1335 1336
		sysctl_wmem_max = 131071;
		sysctl_rmem_max = 131071;
	}
}

/*
 *	Simple resource managers for sockets.
 */


1337 1338
/*
 * Write buffer destructor automatically called from kfree_skb.
L
Linus Torvalds 已提交
1339 1340 1341 1342
 */
void sock_wfree(struct sk_buff *skb)
{
	struct sock *sk = skb->sk;
E
Eric Dumazet 已提交
1343
	unsigned int len = skb->truesize;
L
Linus Torvalds 已提交
1344

E
Eric Dumazet 已提交
1345 1346 1347 1348 1349 1350
	if (!sock_flag(sk, SOCK_USE_WRITE_QUEUE)) {
		/*
		 * Keep a reference on sk_wmem_alloc, this will be released
		 * after sk_write_space() call
		 */
		atomic_sub(len - 1, &sk->sk_wmem_alloc);
L
Linus Torvalds 已提交
1351
		sk->sk_write_space(sk);
E
Eric Dumazet 已提交
1352 1353
		len = 1;
	}
1354
	/*
E
Eric Dumazet 已提交
1355 1356
	 * if sk_wmem_alloc reaches 0, we must finish what sk_free()
	 * could not do because of in-flight packets
1357
	 */
E
Eric Dumazet 已提交
1358
	if (atomic_sub_and_test(len, &sk->sk_wmem_alloc))
1359
		__sk_free(sk);
L
Linus Torvalds 已提交
1360
}
E
Eric Dumazet 已提交
1361
EXPORT_SYMBOL(sock_wfree);
L
Linus Torvalds 已提交
1362

1363 1364
/*
 * Read buffer destructor automatically called from kfree_skb.
L
Linus Torvalds 已提交
1365 1366 1367 1368
 */
void sock_rfree(struct sk_buff *skb)
{
	struct sock *sk = skb->sk;
E
Eric Dumazet 已提交
1369
	unsigned int len = skb->truesize;
L
Linus Torvalds 已提交
1370

E
Eric Dumazet 已提交
1371 1372
	atomic_sub(len, &sk->sk_rmem_alloc);
	sk_mem_uncharge(sk, len);
L
Linus Torvalds 已提交
1373
}
E
Eric Dumazet 已提交
1374
EXPORT_SYMBOL(sock_rfree);
L
Linus Torvalds 已提交
1375 1376 1377 1378 1379 1380


int sock_i_uid(struct sock *sk)
{
	int uid;

E
Eric Dumazet 已提交
1381
	read_lock_bh(&sk->sk_callback_lock);
L
Linus Torvalds 已提交
1382
	uid = sk->sk_socket ? SOCK_INODE(sk->sk_socket)->i_uid : 0;
E
Eric Dumazet 已提交
1383
	read_unlock_bh(&sk->sk_callback_lock);
L
Linus Torvalds 已提交
1384 1385
	return uid;
}
E
Eric Dumazet 已提交
1386
EXPORT_SYMBOL(sock_i_uid);
L
Linus Torvalds 已提交
1387 1388 1389 1390 1391

unsigned long sock_i_ino(struct sock *sk)
{
	unsigned long ino;

E
Eric Dumazet 已提交
1392
	read_lock_bh(&sk->sk_callback_lock);
L
Linus Torvalds 已提交
1393
	ino = sk->sk_socket ? SOCK_INODE(sk->sk_socket)->i_ino : 0;
E
Eric Dumazet 已提交
1394
	read_unlock_bh(&sk->sk_callback_lock);
L
Linus Torvalds 已提交
1395 1396
	return ino;
}
E
Eric Dumazet 已提交
1397
EXPORT_SYMBOL(sock_i_ino);
L
Linus Torvalds 已提交
1398 1399 1400 1401

/*
 * Allocate a skb from the socket's send buffer.
 */
V
Victor Fusco 已提交
1402
struct sk_buff *sock_wmalloc(struct sock *sk, unsigned long size, int force,
A
Al Viro 已提交
1403
			     gfp_t priority)
L
Linus Torvalds 已提交
1404 1405
{
	if (force || atomic_read(&sk->sk_wmem_alloc) < sk->sk_sndbuf) {
E
Eric Dumazet 已提交
1406
		struct sk_buff *skb = alloc_skb(size, priority);
L
Linus Torvalds 已提交
1407 1408 1409 1410 1411 1412 1413
		if (skb) {
			skb_set_owner_w(skb, sk);
			return skb;
		}
	}
	return NULL;
}
E
Eric Dumazet 已提交
1414
EXPORT_SYMBOL(sock_wmalloc);
L
Linus Torvalds 已提交
1415 1416 1417

/*
 * Allocate a skb from the socket's receive buffer.
1418
 */
V
Victor Fusco 已提交
1419
struct sk_buff *sock_rmalloc(struct sock *sk, unsigned long size, int force,
A
Al Viro 已提交
1420
			     gfp_t priority)
L
Linus Torvalds 已提交
1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431
{
	if (force || atomic_read(&sk->sk_rmem_alloc) < sk->sk_rcvbuf) {
		struct sk_buff *skb = alloc_skb(size, priority);
		if (skb) {
			skb_set_owner_r(skb, sk);
			return skb;
		}
	}
	return NULL;
}

1432
/*
L
Linus Torvalds 已提交
1433
 * Allocate a memory block from the socket's option memory buffer.
1434
 */
A
Al Viro 已提交
1435
void *sock_kmalloc(struct sock *sk, int size, gfp_t priority)
L
Linus Torvalds 已提交
1436 1437 1438 1439 1440
{
	if ((unsigned)size <= sysctl_optmem_max &&
	    atomic_read(&sk->sk_omem_alloc) + size < sysctl_optmem_max) {
		void *mem;
		/* First do the add, to avoid the race if kmalloc
1441
		 * might sleep.
L
Linus Torvalds 已提交
1442 1443 1444 1445 1446 1447 1448 1449 1450
		 */
		atomic_add(size, &sk->sk_omem_alloc);
		mem = kmalloc(size, priority);
		if (mem)
			return mem;
		atomic_sub(size, &sk->sk_omem_alloc);
	}
	return NULL;
}
E
Eric Dumazet 已提交
1451
EXPORT_SYMBOL(sock_kmalloc);
L
Linus Torvalds 已提交
1452 1453 1454 1455 1456 1457 1458 1459 1460

/*
 * Free an option memory block.
 */
void sock_kfree_s(struct sock *sk, void *mem, int size)
{
	kfree(mem);
	atomic_sub(size, &sk->sk_omem_alloc);
}
E
Eric Dumazet 已提交
1461
EXPORT_SYMBOL(sock_kfree_s);
L
Linus Torvalds 已提交
1462 1463 1464 1465

/* It is almost wait_for_tcp_memory minus release_sock/lock_sock.
   I think, these locks should be removed for datagram sockets.
 */
E
Eric Dumazet 已提交
1466
static long sock_wait_for_wmem(struct sock *sk, long timeo)
L
Linus Torvalds 已提交
1467 1468 1469 1470 1471 1472 1473 1474 1475 1476
{
	DEFINE_WAIT(wait);

	clear_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags);
	for (;;) {
		if (!timeo)
			break;
		if (signal_pending(current))
			break;
		set_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
E
Eric Dumazet 已提交
1477
		prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
L
Linus Torvalds 已提交
1478 1479 1480 1481 1482 1483 1484 1485
		if (atomic_read(&sk->sk_wmem_alloc) < sk->sk_sndbuf)
			break;
		if (sk->sk_shutdown & SEND_SHUTDOWN)
			break;
		if (sk->sk_err)
			break;
		timeo = schedule_timeout(timeo);
	}
E
Eric Dumazet 已提交
1486
	finish_wait(sk_sleep(sk), &wait);
L
Linus Torvalds 已提交
1487 1488 1489 1490 1491 1492 1493 1494
	return timeo;
}


/*
 *	Generic send/receive buffer handlers
 */

H
Herbert Xu 已提交
1495 1496 1497
struct sk_buff *sock_alloc_send_pskb(struct sock *sk, unsigned long header_len,
				     unsigned long data_len, int noblock,
				     int *errcode)
L
Linus Torvalds 已提交
1498 1499
{
	struct sk_buff *skb;
A
Al Viro 已提交
1500
	gfp_t gfp_mask;
L
Linus Torvalds 已提交
1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518
	long timeo;
	int err;

	gfp_mask = sk->sk_allocation;
	if (gfp_mask & __GFP_WAIT)
		gfp_mask |= __GFP_REPEAT;

	timeo = sock_sndtimeo(sk, noblock);
	while (1) {
		err = sock_error(sk);
		if (err != 0)
			goto failure;

		err = -EPIPE;
		if (sk->sk_shutdown & SEND_SHUTDOWN)
			goto failure;

		if (atomic_read(&sk->sk_wmem_alloc) < sk->sk_sndbuf) {
1519
			skb = alloc_skb(header_len, gfp_mask);
L
Linus Torvalds 已提交
1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541
			if (skb) {
				int npages;
				int i;

				/* No pages, we're done... */
				if (!data_len)
					break;

				npages = (data_len + (PAGE_SIZE - 1)) >> PAGE_SHIFT;
				skb->truesize += data_len;
				skb_shinfo(skb)->nr_frags = npages;
				for (i = 0; i < npages; i++) {
					struct page *page;

					page = alloc_pages(sk->sk_allocation, 0);
					if (!page) {
						err = -ENOBUFS;
						skb_shinfo(skb)->nr_frags = i;
						kfree_skb(skb);
						goto failure;
					}

1542 1543 1544 1545 1546
					__skb_fill_page_desc(skb, i,
							page, 0,
							(data_len >= PAGE_SIZE ?
							 PAGE_SIZE :
							 data_len));
L
Linus Torvalds 已提交
1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574
					data_len -= PAGE_SIZE;
				}

				/* Full success... */
				break;
			}
			err = -ENOBUFS;
			goto failure;
		}
		set_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags);
		set_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
		err = -EAGAIN;
		if (!timeo)
			goto failure;
		if (signal_pending(current))
			goto interrupted;
		timeo = sock_wait_for_wmem(sk, timeo);
	}

	skb_set_owner_w(skb, sk);
	return skb;

interrupted:
	err = sock_intr_errno(timeo);
failure:
	*errcode = err;
	return NULL;
}
H
Herbert Xu 已提交
1575
EXPORT_SYMBOL(sock_alloc_send_pskb);
L
Linus Torvalds 已提交
1576

1577
struct sk_buff *sock_alloc_send_skb(struct sock *sk, unsigned long size,
L
Linus Torvalds 已提交
1578 1579 1580 1581
				    int noblock, int *errcode)
{
	return sock_alloc_send_pskb(sk, size, 0, noblock, errcode);
}
E
Eric Dumazet 已提交
1582
EXPORT_SYMBOL(sock_alloc_send_skb);
L
Linus Torvalds 已提交
1583 1584

static void __lock_sock(struct sock *sk)
1585 1586
	__releases(&sk->sk_lock.slock)
	__acquires(&sk->sk_lock.slock)
L
Linus Torvalds 已提交
1587 1588 1589
{
	DEFINE_WAIT(wait);

1590
	for (;;) {
L
Linus Torvalds 已提交
1591 1592 1593 1594 1595
		prepare_to_wait_exclusive(&sk->sk_lock.wq, &wait,
					TASK_UNINTERRUPTIBLE);
		spin_unlock_bh(&sk->sk_lock.slock);
		schedule();
		spin_lock_bh(&sk->sk_lock.slock);
1596
		if (!sock_owned_by_user(sk))
L
Linus Torvalds 已提交
1597 1598 1599 1600 1601 1602
			break;
	}
	finish_wait(&sk->sk_lock.wq, &wait);
}

static void __release_sock(struct sock *sk)
1603 1604
	__releases(&sk->sk_lock.slock)
	__acquires(&sk->sk_lock.slock)
L
Linus Torvalds 已提交
1605 1606 1607 1608 1609 1610 1611 1612 1613 1614
{
	struct sk_buff *skb = sk->sk_backlog.head;

	do {
		sk->sk_backlog.head = sk->sk_backlog.tail = NULL;
		bh_unlock_sock(sk);

		do {
			struct sk_buff *next = skb->next;

E
Eric Dumazet 已提交
1615
			WARN_ON_ONCE(skb_dst_is_noref(skb));
L
Linus Torvalds 已提交
1616
			skb->next = NULL;
P
Peter Zijlstra 已提交
1617
			sk_backlog_rcv(sk, skb);
L
Linus Torvalds 已提交
1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630

			/*
			 * We are in process context here with softirqs
			 * disabled, use cond_resched_softirq() to preempt.
			 * This is safe to do because we've taken the backlog
			 * queue private:
			 */
			cond_resched_softirq();

			skb = next;
		} while (skb != NULL);

		bh_lock_sock(sk);
1631
	} while ((skb = sk->sk_backlog.head) != NULL);
Z
Zhu Yi 已提交
1632 1633 1634 1635 1636 1637

	/*
	 * Doing the zeroing here guarantee we can not loop forever
	 * while a wild producer attempts to flood us.
	 */
	sk->sk_backlog.len = 0;
L
Linus Torvalds 已提交
1638 1639 1640 1641
}

/**
 * sk_wait_data - wait for data to arrive at sk_receive_queue
1642 1643
 * @sk:    sock to wait on
 * @timeo: for how long
L
Linus Torvalds 已提交
1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654
 *
 * Now socket state including sk->sk_err is changed only under lock,
 * hence we may omit checks after joining wait queue.
 * We check receive queue before schedule() only as optimization;
 * it is very likely that release_sock() added new data.
 */
int sk_wait_data(struct sock *sk, long *timeo)
{
	int rc;
	DEFINE_WAIT(wait);

E
Eric Dumazet 已提交
1655
	prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
L
Linus Torvalds 已提交
1656 1657 1658
	set_bit(SOCK_ASYNC_WAITDATA, &sk->sk_socket->flags);
	rc = sk_wait_event(sk, timeo, !skb_queue_empty(&sk->sk_receive_queue));
	clear_bit(SOCK_ASYNC_WAITDATA, &sk->sk_socket->flags);
E
Eric Dumazet 已提交
1659
	finish_wait(sk_sleep(sk), &wait);
L
Linus Torvalds 已提交
1660 1661 1662 1663
	return rc;
}
EXPORT_SYMBOL(sk_wait_data);

1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677
/**
 *	__sk_mem_schedule - increase sk_forward_alloc and memory_allocated
 *	@sk: socket
 *	@size: memory size to allocate
 *	@kind: allocation type
 *
 *	If kind is SK_MEM_SEND, it means wmem allocation. Otherwise it means
 *	rmem allocation. This function assumes that protocols which have
 *	memory_pressure use sk_wmem_queued as write buffer accounting.
 */
int __sk_mem_schedule(struct sock *sk, int size, int kind)
{
	struct proto *prot = sk->sk_prot;
	int amt = sk_mem_pages(size);
E
Eric Dumazet 已提交
1678
	long allocated;
1679 1680

	sk->sk_forward_alloc += amt * SK_MEM_QUANTUM;
E
Eric Dumazet 已提交
1681
	allocated = atomic_long_add_return(amt, prot->memory_allocated);
1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692

	/* Under limit. */
	if (allocated <= prot->sysctl_mem[0]) {
		if (prot->memory_pressure && *prot->memory_pressure)
			*prot->memory_pressure = 0;
		return 1;
	}

	/* Under pressure. */
	if (allocated > prot->sysctl_mem[1])
		if (prot->enter_memory_pressure)
1693
			prot->enter_memory_pressure(sk);
1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712

	/* Over hard limit. */
	if (allocated > prot->sysctl_mem[2])
		goto suppress_allocation;

	/* guarantee minimum buffer size under pressure */
	if (kind == SK_MEM_RECV) {
		if (atomic_read(&sk->sk_rmem_alloc) < prot->sysctl_rmem[0])
			return 1;
	} else { /* SK_MEM_SEND */
		if (sk->sk_type == SOCK_STREAM) {
			if (sk->sk_wmem_queued < prot->sysctl_wmem[0])
				return 1;
		} else if (atomic_read(&sk->sk_wmem_alloc) <
			   prot->sysctl_wmem[0])
				return 1;
	}

	if (prot->memory_pressure) {
1713 1714 1715 1716 1717 1718
		int alloc;

		if (!*prot->memory_pressure)
			return 1;
		alloc = percpu_counter_read_positive(prot->sockets_allocated);
		if (prot->sysctl_mem[2] > alloc *
1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736
		    sk_mem_pages(sk->sk_wmem_queued +
				 atomic_read(&sk->sk_rmem_alloc) +
				 sk->sk_forward_alloc))
			return 1;
	}

suppress_allocation:

	if (kind == SK_MEM_SEND && sk->sk_type == SOCK_STREAM) {
		sk_stream_moderate_sndbuf(sk);

		/* Fail only if socket is _under_ its sndbuf.
		 * In this case we cannot block, so that we have to fail.
		 */
		if (sk->sk_wmem_queued + size >= sk->sk_sndbuf)
			return 1;
	}

1737 1738
	trace_sock_exceed_buf_limit(sk, prot, allocated);

1739 1740
	/* Alas. Undo changes. */
	sk->sk_forward_alloc -= amt * SK_MEM_QUANTUM;
E
Eric Dumazet 已提交
1741
	atomic_long_sub(amt, prot->memory_allocated);
1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753
	return 0;
}
EXPORT_SYMBOL(__sk_mem_schedule);

/**
 *	__sk_reclaim - reclaim memory_allocated
 *	@sk: socket
 */
void __sk_mem_reclaim(struct sock *sk)
{
	struct proto *prot = sk->sk_prot;

E
Eric Dumazet 已提交
1754
	atomic_long_sub(sk->sk_forward_alloc >> SK_MEM_QUANTUM_SHIFT,
1755 1756 1757 1758
		   prot->memory_allocated);
	sk->sk_forward_alloc &= SK_MEM_QUANTUM - 1;

	if (prot->memory_pressure && *prot->memory_pressure &&
E
Eric Dumazet 已提交
1759
	    (atomic_long_read(prot->memory_allocated) < prot->sysctl_mem[0]))
1760 1761 1762 1763 1764
		*prot->memory_pressure = 0;
}
EXPORT_SYMBOL(__sk_mem_reclaim);


L
Linus Torvalds 已提交
1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775
/*
 * Set of default routines for initialising struct proto_ops when
 * the protocol does not support a particular function. In certain
 * cases where it makes no sense for a protocol to have a "do nothing"
 * function, some default processing is provided.
 */

int sock_no_bind(struct socket *sock, struct sockaddr *saddr, int len)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1776
EXPORT_SYMBOL(sock_no_bind);
L
Linus Torvalds 已提交
1777

1778
int sock_no_connect(struct socket *sock, struct sockaddr *saddr,
L
Linus Torvalds 已提交
1779 1780 1781 1782
		    int len, int flags)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1783
EXPORT_SYMBOL(sock_no_connect);
L
Linus Torvalds 已提交
1784 1785 1786 1787 1788

int sock_no_socketpair(struct socket *sock1, struct socket *sock2)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1789
EXPORT_SYMBOL(sock_no_socketpair);
L
Linus Torvalds 已提交
1790 1791 1792 1793 1794

int sock_no_accept(struct socket *sock, struct socket *newsock, int flags)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1795
EXPORT_SYMBOL(sock_no_accept);
L
Linus Torvalds 已提交
1796

1797
int sock_no_getname(struct socket *sock, struct sockaddr *saddr,
L
Linus Torvalds 已提交
1798 1799 1800 1801
		    int *len, int peer)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1802
EXPORT_SYMBOL(sock_no_getname);
L
Linus Torvalds 已提交
1803

E
Eric Dumazet 已提交
1804
unsigned int sock_no_poll(struct file *file, struct socket *sock, poll_table *pt)
L
Linus Torvalds 已提交
1805 1806 1807
{
	return 0;
}
E
Eric Dumazet 已提交
1808
EXPORT_SYMBOL(sock_no_poll);
L
Linus Torvalds 已提交
1809 1810 1811 1812 1813

int sock_no_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1814
EXPORT_SYMBOL(sock_no_ioctl);
L
Linus Torvalds 已提交
1815 1816 1817 1818 1819

int sock_no_listen(struct socket *sock, int backlog)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1820
EXPORT_SYMBOL(sock_no_listen);
L
Linus Torvalds 已提交
1821 1822 1823 1824 1825

int sock_no_shutdown(struct socket *sock, int how)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1826
EXPORT_SYMBOL(sock_no_shutdown);
L
Linus Torvalds 已提交
1827 1828

int sock_no_setsockopt(struct socket *sock, int level, int optname,
1829
		    char __user *optval, unsigned int optlen)
L
Linus Torvalds 已提交
1830 1831 1832
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1833
EXPORT_SYMBOL(sock_no_setsockopt);
L
Linus Torvalds 已提交
1834 1835 1836 1837 1838 1839

int sock_no_getsockopt(struct socket *sock, int level, int optname,
		    char __user *optval, int __user *optlen)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1840
EXPORT_SYMBOL(sock_no_getsockopt);
L
Linus Torvalds 已提交
1841 1842 1843 1844 1845 1846

int sock_no_sendmsg(struct kiocb *iocb, struct socket *sock, struct msghdr *m,
		    size_t len)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1847
EXPORT_SYMBOL(sock_no_sendmsg);
L
Linus Torvalds 已提交
1848 1849 1850 1851 1852 1853

int sock_no_recvmsg(struct kiocb *iocb, struct socket *sock, struct msghdr *m,
		    size_t len, int flags)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1854
EXPORT_SYMBOL(sock_no_recvmsg);
L
Linus Torvalds 已提交
1855 1856 1857 1858 1859 1860

int sock_no_mmap(struct file *file, struct socket *sock, struct vm_area_struct *vma)
{
	/* Mirror missing mmap method error code */
	return -ENODEV;
}
E
Eric Dumazet 已提交
1861
EXPORT_SYMBOL(sock_no_mmap);
L
Linus Torvalds 已提交
1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874

ssize_t sock_no_sendpage(struct socket *sock, struct page *page, int offset, size_t size, int flags)
{
	ssize_t res;
	struct msghdr msg = {.msg_flags = flags};
	struct kvec iov;
	char *kaddr = kmap(page);
	iov.iov_base = kaddr + offset;
	iov.iov_len = size;
	res = kernel_sendmsg(sock, &msg, &iov, 1, size);
	kunmap(page);
	return res;
}
E
Eric Dumazet 已提交
1875
EXPORT_SYMBOL(sock_no_sendpage);
L
Linus Torvalds 已提交
1876 1877 1878 1879 1880 1881 1882

/*
 *	Default Socket Callbacks
 */

static void sock_def_wakeup(struct sock *sk)
{
1883 1884 1885 1886 1887 1888 1889
	struct socket_wq *wq;

	rcu_read_lock();
	wq = rcu_dereference(sk->sk_wq);
	if (wq_has_sleeper(wq))
		wake_up_interruptible_all(&wq->wait);
	rcu_read_unlock();
L
Linus Torvalds 已提交
1890 1891 1892 1893
}

static void sock_def_error_report(struct sock *sk)
{
1894 1895 1896 1897 1898 1899
	struct socket_wq *wq;

	rcu_read_lock();
	wq = rcu_dereference(sk->sk_wq);
	if (wq_has_sleeper(wq))
		wake_up_interruptible_poll(&wq->wait, POLLERR);
1900
	sk_wake_async(sk, SOCK_WAKE_IO, POLL_ERR);
1901
	rcu_read_unlock();
L
Linus Torvalds 已提交
1902 1903 1904 1905
}

static void sock_def_readable(struct sock *sk, int len)
{
1906 1907 1908 1909 1910
	struct socket_wq *wq;

	rcu_read_lock();
	wq = rcu_dereference(sk->sk_wq);
	if (wq_has_sleeper(wq))
1911
		wake_up_interruptible_sync_poll(&wq->wait, POLLIN | POLLPRI |
1912
						POLLRDNORM | POLLRDBAND);
1913
	sk_wake_async(sk, SOCK_WAKE_WAITD, POLL_IN);
1914
	rcu_read_unlock();
L
Linus Torvalds 已提交
1915 1916 1917 1918
}

static void sock_def_write_space(struct sock *sk)
{
1919 1920 1921
	struct socket_wq *wq;

	rcu_read_lock();
L
Linus Torvalds 已提交
1922 1923 1924 1925

	/* Do not wake up a writer until he can make "significant"
	 * progress.  --DaveM
	 */
1926
	if ((atomic_read(&sk->sk_wmem_alloc) << 1) <= sk->sk_sndbuf) {
1927 1928 1929
		wq = rcu_dereference(sk->sk_wq);
		if (wq_has_sleeper(wq))
			wake_up_interruptible_sync_poll(&wq->wait, POLLOUT |
1930
						POLLWRNORM | POLLWRBAND);
L
Linus Torvalds 已提交
1931 1932 1933

		/* Should agree with poll, otherwise some programs break */
		if (sock_writeable(sk))
1934
			sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
L
Linus Torvalds 已提交
1935 1936
	}

1937
	rcu_read_unlock();
L
Linus Torvalds 已提交
1938 1939 1940 1941
}

static void sock_def_destruct(struct sock *sk)
{
J
Jesper Juhl 已提交
1942
	kfree(sk->sk_protinfo);
L
Linus Torvalds 已提交
1943 1944 1945 1946 1947 1948
}

void sk_send_sigurg(struct sock *sk)
{
	if (sk->sk_socket && sk->sk_socket->file)
		if (send_sigurg(&sk->sk_socket->file->f_owner))
1949
			sk_wake_async(sk, SOCK_WAKE_URG, POLL_PRI);
L
Linus Torvalds 已提交
1950
}
E
Eric Dumazet 已提交
1951
EXPORT_SYMBOL(sk_send_sigurg);
L
Linus Torvalds 已提交
1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972

void sk_reset_timer(struct sock *sk, struct timer_list* timer,
		    unsigned long expires)
{
	if (!mod_timer(timer, expires))
		sock_hold(sk);
}
EXPORT_SYMBOL(sk_reset_timer);

void sk_stop_timer(struct sock *sk, struct timer_list* timer)
{
	if (timer_pending(timer) && del_timer(timer))
		__sock_put(sk);
}
EXPORT_SYMBOL(sk_stop_timer);

void sock_init_data(struct socket *sock, struct sock *sk)
{
	skb_queue_head_init(&sk->sk_receive_queue);
	skb_queue_head_init(&sk->sk_write_queue);
	skb_queue_head_init(&sk->sk_error_queue);
1973 1974 1975
#ifdef CONFIG_NET_DMA
	skb_queue_head_init(&sk->sk_async_wait_queue);
#endif
L
Linus Torvalds 已提交
1976 1977 1978 1979

	sk->sk_send_head	=	NULL;

	init_timer(&sk->sk_timer);
1980

L
Linus Torvalds 已提交
1981 1982 1983 1984
	sk->sk_allocation	=	GFP_KERNEL;
	sk->sk_rcvbuf		=	sysctl_rmem_default;
	sk->sk_sndbuf		=	sysctl_wmem_default;
	sk->sk_state		=	TCP_CLOSE;
1985
	sk_set_socket(sk, sock);
L
Linus Torvalds 已提交
1986 1987 1988

	sock_set_flag(sk, SOCK_ZAPPED);

1989
	if (sock) {
L
Linus Torvalds 已提交
1990
		sk->sk_type	=	sock->type;
1991
		sk->sk_wq	=	sock->wq;
L
Linus Torvalds 已提交
1992 1993
		sock->sk	=	sk;
	} else
1994
		sk->sk_wq	=	NULL;
L
Linus Torvalds 已提交
1995

E
Eric Dumazet 已提交
1996
	spin_lock_init(&sk->sk_dst_lock);
L
Linus Torvalds 已提交
1997
	rwlock_init(&sk->sk_callback_lock);
1998 1999 2000
	lockdep_set_class_and_name(&sk->sk_callback_lock,
			af_callback_keys + sk->sk_family,
			af_family_clock_key_strings[sk->sk_family]);
L
Linus Torvalds 已提交
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

	sk->sk_state_change	=	sock_def_wakeup;
	sk->sk_data_ready	=	sock_def_readable;
	sk->sk_write_space	=	sock_def_write_space;
	sk->sk_error_report	=	sock_def_error_report;
	sk->sk_destruct		=	sock_def_destruct;

	sk->sk_sndmsg_page	=	NULL;
	sk->sk_sndmsg_off	=	0;

2011 2012
	sk->sk_peer_pid 	=	NULL;
	sk->sk_peer_cred	=	NULL;
L
Linus Torvalds 已提交
2013 2014 2015 2016 2017
	sk->sk_write_pending	=	0;
	sk->sk_rcvlowat		=	1;
	sk->sk_rcvtimeo		=	MAX_SCHEDULE_TIMEOUT;
	sk->sk_sndtimeo		=	MAX_SCHEDULE_TIMEOUT;

2018
	sk->sk_stamp = ktime_set(-1L, 0);
L
Linus Torvalds 已提交
2019

E
Eric Dumazet 已提交
2020 2021 2022 2023 2024
	/*
	 * Before updating sk_refcnt, we must commit prior changes to memory
	 * (Documentation/RCU/rculist_nulls.txt for details)
	 */
	smp_wmb();
L
Linus Torvalds 已提交
2025
	atomic_set(&sk->sk_refcnt, 1);
W
Wang Chen 已提交
2026
	atomic_set(&sk->sk_drops, 0);
L
Linus Torvalds 已提交
2027
}
E
Eric Dumazet 已提交
2028
EXPORT_SYMBOL(sock_init_data);
L
Linus Torvalds 已提交
2029

H
Harvey Harrison 已提交
2030
void lock_sock_nested(struct sock *sk, int subclass)
L
Linus Torvalds 已提交
2031 2032
{
	might_sleep();
I
Ingo Molnar 已提交
2033
	spin_lock_bh(&sk->sk_lock.slock);
2034
	if (sk->sk_lock.owned)
L
Linus Torvalds 已提交
2035
		__lock_sock(sk);
2036
	sk->sk_lock.owned = 1;
I
Ingo Molnar 已提交
2037 2038 2039 2040
	spin_unlock(&sk->sk_lock.slock);
	/*
	 * The sk_lock has mutex_lock() semantics here:
	 */
2041
	mutex_acquire(&sk->sk_lock.dep_map, subclass, 0, _RET_IP_);
I
Ingo Molnar 已提交
2042
	local_bh_enable();
L
Linus Torvalds 已提交
2043
}
2044
EXPORT_SYMBOL(lock_sock_nested);
L
Linus Torvalds 已提交
2045

H
Harvey Harrison 已提交
2046
void release_sock(struct sock *sk)
L
Linus Torvalds 已提交
2047
{
I
Ingo Molnar 已提交
2048 2049 2050 2051 2052 2053
	/*
	 * The sk_lock has mutex_unlock() semantics:
	 */
	mutex_release(&sk->sk_lock.dep_map, 1, _RET_IP_);

	spin_lock_bh(&sk->sk_lock.slock);
L
Linus Torvalds 已提交
2054 2055
	if (sk->sk_backlog.tail)
		__release_sock(sk);
2056
	sk->sk_lock.owned = 0;
I
Ingo Molnar 已提交
2057 2058 2059
	if (waitqueue_active(&sk->sk_lock.wq))
		wake_up(&sk->sk_lock.wq);
	spin_unlock_bh(&sk->sk_lock.slock);
L
Linus Torvalds 已提交
2060 2061 2062
}
EXPORT_SYMBOL(release_sock);

2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095
/**
 * lock_sock_fast - fast version of lock_sock
 * @sk: socket
 *
 * This version should be used for very small section, where process wont block
 * return false if fast path is taken
 *   sk_lock.slock locked, owned = 0, BH disabled
 * return true if slow path is taken
 *   sk_lock.slock unlocked, owned = 1, BH enabled
 */
bool lock_sock_fast(struct sock *sk)
{
	might_sleep();
	spin_lock_bh(&sk->sk_lock.slock);

	if (!sk->sk_lock.owned)
		/*
		 * Note : We must disable BH
		 */
		return false;

	__lock_sock(sk);
	sk->sk_lock.owned = 1;
	spin_unlock(&sk->sk_lock.slock);
	/*
	 * The sk_lock has mutex_lock() semantics here:
	 */
	mutex_acquire(&sk->sk_lock.dep_map, 0, 0, _RET_IP_);
	local_bh_enable();
	return true;
}
EXPORT_SYMBOL(lock_sock_fast);

L
Linus Torvalds 已提交
2096
int sock_get_timestamp(struct sock *sk, struct timeval __user *userstamp)
2097
{
2098
	struct timeval tv;
L
Linus Torvalds 已提交
2099
	if (!sock_flag(sk, SOCK_TIMESTAMP))
2100
		sock_enable_timestamp(sk, SOCK_TIMESTAMP);
2101 2102
	tv = ktime_to_timeval(sk->sk_stamp);
	if (tv.tv_sec == -1)
L
Linus Torvalds 已提交
2103
		return -ENOENT;
2104 2105 2106 2107 2108
	if (tv.tv_sec == 0) {
		sk->sk_stamp = ktime_get_real();
		tv = ktime_to_timeval(sk->sk_stamp);
	}
	return copy_to_user(userstamp, &tv, sizeof(tv)) ? -EFAULT : 0;
2109
}
L
Linus Torvalds 已提交
2110 2111
EXPORT_SYMBOL(sock_get_timestamp);

2112 2113 2114 2115
int sock_get_timestampns(struct sock *sk, struct timespec __user *userstamp)
{
	struct timespec ts;
	if (!sock_flag(sk, SOCK_TIMESTAMP))
2116
		sock_enable_timestamp(sk, SOCK_TIMESTAMP);
2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127
	ts = ktime_to_timespec(sk->sk_stamp);
	if (ts.tv_sec == -1)
		return -ENOENT;
	if (ts.tv_sec == 0) {
		sk->sk_stamp = ktime_get_real();
		ts = ktime_to_timespec(sk->sk_stamp);
	}
	return copy_to_user(userstamp, &ts, sizeof(ts)) ? -EFAULT : 0;
}
EXPORT_SYMBOL(sock_get_timestampns);

2128
void sock_enable_timestamp(struct sock *sk, int flag)
2129
{
2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141
	if (!sock_flag(sk, flag)) {
		sock_set_flag(sk, flag);
		/*
		 * we just set one of the two flags which require net
		 * time stamping, but time stamping might have been on
		 * already because of the other one
		 */
		if (!sock_flag(sk,
				flag == SOCK_TIMESTAMP ?
				SOCK_TIMESTAMPING_RX_SOFTWARE :
				SOCK_TIMESTAMP))
			net_enable_timestamp();
L
Linus Torvalds 已提交
2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160
	}
}

/*
 *	Get a socket option on an socket.
 *
 *	FIX: POSIX 1003.1g is very ambiguous here. It states that
 *	asynchronous errors should be reported by getsockopt. We assume
 *	this means if you specify SO_ERROR (otherwise whats the point of it).
 */
int sock_common_getsockopt(struct socket *sock, int level, int optname,
			   char __user *optval, int __user *optlen)
{
	struct sock *sk = sock->sk;

	return sk->sk_prot->getsockopt(sk, level, optname, optval, optlen);
}
EXPORT_SYMBOL(sock_common_getsockopt);

2161
#ifdef CONFIG_COMPAT
2162 2163
int compat_sock_common_getsockopt(struct socket *sock, int level, int optname,
				  char __user *optval, int __user *optlen)
2164 2165 2166
{
	struct sock *sk = sock->sk;

2167
	if (sk->sk_prot->compat_getsockopt != NULL)
2168 2169
		return sk->sk_prot->compat_getsockopt(sk, level, optname,
						      optval, optlen);
2170 2171 2172 2173 2174
	return sk->sk_prot->getsockopt(sk, level, optname, optval, optlen);
}
EXPORT_SYMBOL(compat_sock_common_getsockopt);
#endif

L
Linus Torvalds 已提交
2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193
int sock_common_recvmsg(struct kiocb *iocb, struct socket *sock,
			struct msghdr *msg, size_t size, int flags)
{
	struct sock *sk = sock->sk;
	int addr_len = 0;
	int err;

	err = sk->sk_prot->recvmsg(iocb, sk, msg, size, flags & MSG_DONTWAIT,
				   flags & ~MSG_DONTWAIT, &addr_len);
	if (err >= 0)
		msg->msg_namelen = addr_len;
	return err;
}
EXPORT_SYMBOL(sock_common_recvmsg);

/*
 *	Set socket options on an inet socket.
 */
int sock_common_setsockopt(struct socket *sock, int level, int optname,
2194
			   char __user *optval, unsigned int optlen)
L
Linus Torvalds 已提交
2195 2196 2197 2198 2199 2200 2201
{
	struct sock *sk = sock->sk;

	return sk->sk_prot->setsockopt(sk, level, optname, optval, optlen);
}
EXPORT_SYMBOL(sock_common_setsockopt);

2202
#ifdef CONFIG_COMPAT
2203
int compat_sock_common_setsockopt(struct socket *sock, int level, int optname,
2204
				  char __user *optval, unsigned int optlen)
2205 2206 2207
{
	struct sock *sk = sock->sk;

2208 2209 2210
	if (sk->sk_prot->compat_setsockopt != NULL)
		return sk->sk_prot->compat_setsockopt(sk, level, optname,
						      optval, optlen);
2211 2212 2213 2214 2215
	return sk->sk_prot->setsockopt(sk, level, optname, optval, optlen);
}
EXPORT_SYMBOL(compat_sock_common_setsockopt);
#endif

L
Linus Torvalds 已提交
2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246
void sk_common_release(struct sock *sk)
{
	if (sk->sk_prot->destroy)
		sk->sk_prot->destroy(sk);

	/*
	 * Observation: when sock_common_release is called, processes have
	 * no access to socket. But net still has.
	 * Step one, detach it from networking:
	 *
	 * A. Remove from hash tables.
	 */

	sk->sk_prot->unhash(sk);

	/*
	 * In this point socket cannot receive new packets, but it is possible
	 * that some packets are in flight because some CPU runs receiver and
	 * did hash table lookup before we unhashed socket. They will achieve
	 * receive queue and will be purged by socket destructor.
	 *
	 * Also we still have packets pending on receive queue and probably,
	 * our own packets waiting in device queues. sock_destroy will drain
	 * receive queue, but transmitted packets will delay socket destruction
	 * until the last reference will be released.
	 */

	sock_orphan(sk);

	xfrm_sk_free_policy(sk);

2247
	sk_refcnt_debug_release(sk);
L
Linus Torvalds 已提交
2248 2249 2250 2251 2252 2253 2254
	sock_put(sk);
}
EXPORT_SYMBOL(sk_common_release);

static DEFINE_RWLOCK(proto_list_lock);
static LIST_HEAD(proto_list);

2255 2256
#ifdef CONFIG_PROC_FS
#define PROTO_INUSE_NR	64	/* should be enough for the first time */
2257 2258 2259
struct prot_inuse {
	int val[PROTO_INUSE_NR];
};
2260 2261

static DECLARE_BITMAP(proto_inuse_idx, PROTO_INUSE_NR);
2262 2263 2264 2265

#ifdef CONFIG_NET_NS
void sock_prot_inuse_add(struct net *net, struct proto *prot, int val)
{
E
Eric Dumazet 已提交
2266
	__this_cpu_add(net->core.inuse->val[prot->inuse_idx], val);
2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281
}
EXPORT_SYMBOL_GPL(sock_prot_inuse_add);

int sock_prot_inuse_get(struct net *net, struct proto *prot)
{
	int cpu, idx = prot->inuse_idx;
	int res = 0;

	for_each_possible_cpu(cpu)
		res += per_cpu_ptr(net->core.inuse, cpu)->val[idx];

	return res >= 0 ? res : 0;
}
EXPORT_SYMBOL_GPL(sock_prot_inuse_get);

2282
static int __net_init sock_inuse_init_net(struct net *net)
2283 2284 2285 2286 2287
{
	net->core.inuse = alloc_percpu(struct prot_inuse);
	return net->core.inuse ? 0 : -ENOMEM;
}

2288
static void __net_exit sock_inuse_exit_net(struct net *net)
2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307
{
	free_percpu(net->core.inuse);
}

static struct pernet_operations net_inuse_ops = {
	.init = sock_inuse_init_net,
	.exit = sock_inuse_exit_net,
};

static __init int net_inuse_init(void)
{
	if (register_pernet_subsys(&net_inuse_ops))
		panic("Cannot initialize net inuse counters");

	return 0;
}

core_initcall(net_inuse_init);
#else
2308 2309
static DEFINE_PER_CPU(struct prot_inuse, prot_inuse);

2310
void sock_prot_inuse_add(struct net *net, struct proto *prot, int val)
2311
{
E
Eric Dumazet 已提交
2312
	__this_cpu_add(prot_inuse.val[prot->inuse_idx], val);
2313 2314 2315
}
EXPORT_SYMBOL_GPL(sock_prot_inuse_add);

2316
int sock_prot_inuse_get(struct net *net, struct proto *prot)
2317 2318 2319 2320 2321 2322 2323 2324 2325 2326
{
	int cpu, idx = prot->inuse_idx;
	int res = 0;

	for_each_possible_cpu(cpu)
		res += per_cpu(prot_inuse, cpu).val[idx];

	return res >= 0 ? res : 0;
}
EXPORT_SYMBOL_GPL(sock_prot_inuse_get);
2327
#endif
2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355

static void assign_proto_idx(struct proto *prot)
{
	prot->inuse_idx = find_first_zero_bit(proto_inuse_idx, PROTO_INUSE_NR);

	if (unlikely(prot->inuse_idx == PROTO_INUSE_NR - 1)) {
		printk(KERN_ERR "PROTO_INUSE_NR exhausted\n");
		return;
	}

	set_bit(prot->inuse_idx, proto_inuse_idx);
}

static void release_proto_idx(struct proto *prot)
{
	if (prot->inuse_idx != PROTO_INUSE_NR - 1)
		clear_bit(prot->inuse_idx, proto_inuse_idx);
}
#else
static inline void assign_proto_idx(struct proto *prot)
{
}

static inline void release_proto_idx(struct proto *prot)
{
}
#endif

2356 2357
int proto_register(struct proto *prot, int alloc_slab)
{
L
Linus Torvalds 已提交
2358 2359
	if (alloc_slab) {
		prot->slab = kmem_cache_create(prot->name, prot->obj_size, 0,
2360 2361
					SLAB_HWCACHE_ALIGN | prot->slab_flags,
					NULL);
L
Linus Torvalds 已提交
2362 2363 2364 2365

		if (prot->slab == NULL) {
			printk(KERN_CRIT "%s: Can't create sock SLAB cache!\n",
			       prot->name);
2366
			goto out;
L
Linus Torvalds 已提交
2367
		}
2368 2369

		if (prot->rsk_prot != NULL) {
2370
			prot->rsk_prot->slab_name = kasprintf(GFP_KERNEL, "request_sock_%s", prot->name);
2371
			if (prot->rsk_prot->slab_name == NULL)
2372 2373
				goto out_free_sock_slab;

2374
			prot->rsk_prot->slab = kmem_cache_create(prot->rsk_prot->slab_name,
2375
								 prot->rsk_prot->obj_size, 0,
2376
								 SLAB_HWCACHE_ALIGN, NULL);
2377 2378 2379 2380 2381 2382 2383

			if (prot->rsk_prot->slab == NULL) {
				printk(KERN_CRIT "%s: Can't create request sock SLAB cache!\n",
				       prot->name);
				goto out_free_request_sock_slab_name;
			}
		}
2384

2385
		if (prot->twsk_prot != NULL) {
2386
			prot->twsk_prot->twsk_slab_name = kasprintf(GFP_KERNEL, "tw_sock_%s", prot->name);
2387

2388
			if (prot->twsk_prot->twsk_slab_name == NULL)
2389 2390
				goto out_free_request_sock_slab;

2391
			prot->twsk_prot->twsk_slab =
2392
				kmem_cache_create(prot->twsk_prot->twsk_slab_name,
2393
						  prot->twsk_prot->twsk_obj_size,
2394 2395 2396
						  0,
						  SLAB_HWCACHE_ALIGN |
							prot->slab_flags,
2397
						  NULL);
2398
			if (prot->twsk_prot->twsk_slab == NULL)
2399 2400
				goto out_free_timewait_sock_slab_name;
		}
L
Linus Torvalds 已提交
2401 2402
	}

2403
	write_lock(&proto_list_lock);
L
Linus Torvalds 已提交
2404
	list_add(&prot->node, &proto_list);
2405
	assign_proto_idx(prot);
L
Linus Torvalds 已提交
2406
	write_unlock(&proto_list_lock);
2407 2408
	return 0;

2409
out_free_timewait_sock_slab_name:
2410
	kfree(prot->twsk_prot->twsk_slab_name);
2411 2412 2413 2414 2415
out_free_request_sock_slab:
	if (prot->rsk_prot && prot->rsk_prot->slab) {
		kmem_cache_destroy(prot->rsk_prot->slab);
		prot->rsk_prot->slab = NULL;
	}
2416
out_free_request_sock_slab_name:
2417 2418
	if (prot->rsk_prot)
		kfree(prot->rsk_prot->slab_name);
2419 2420 2421
out_free_sock_slab:
	kmem_cache_destroy(prot->slab);
	prot->slab = NULL;
2422 2423
out:
	return -ENOBUFS;
L
Linus Torvalds 已提交
2424 2425 2426 2427 2428 2429
}
EXPORT_SYMBOL(proto_register);

void proto_unregister(struct proto *prot)
{
	write_lock(&proto_list_lock);
2430
	release_proto_idx(prot);
2431 2432
	list_del(&prot->node);
	write_unlock(&proto_list_lock);
L
Linus Torvalds 已提交
2433 2434 2435 2436 2437 2438

	if (prot->slab != NULL) {
		kmem_cache_destroy(prot->slab);
		prot->slab = NULL;
	}

2439 2440
	if (prot->rsk_prot != NULL && prot->rsk_prot->slab != NULL) {
		kmem_cache_destroy(prot->rsk_prot->slab);
2441
		kfree(prot->rsk_prot->slab_name);
2442 2443 2444
		prot->rsk_prot->slab = NULL;
	}

2445 2446
	if (prot->twsk_prot != NULL && prot->twsk_prot->twsk_slab != NULL) {
		kmem_cache_destroy(prot->twsk_prot->twsk_slab);
2447
		kfree(prot->twsk_prot->twsk_slab_name);
2448
		prot->twsk_prot->twsk_slab = NULL;
2449
	}
L
Linus Torvalds 已提交
2450 2451 2452 2453 2454
}
EXPORT_SYMBOL(proto_unregister);

#ifdef CONFIG_PROC_FS
static void *proto_seq_start(struct seq_file *seq, loff_t *pos)
2455
	__acquires(proto_list_lock)
L
Linus Torvalds 已提交
2456 2457
{
	read_lock(&proto_list_lock);
2458
	return seq_list_start_head(&proto_list, *pos);
L
Linus Torvalds 已提交
2459 2460 2461 2462
}

static void *proto_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
2463
	return seq_list_next(v, &proto_list, pos);
L
Linus Torvalds 已提交
2464 2465 2466
}

static void proto_seq_stop(struct seq_file *seq, void *v)
2467
	__releases(proto_list_lock)
L
Linus Torvalds 已提交
2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478
{
	read_unlock(&proto_list_lock);
}

static char proto_method_implemented(const void *method)
{
	return method == NULL ? 'n' : 'y';
}

static void proto_seq_printf(struct seq_file *seq, struct proto *proto)
{
E
Eric Dumazet 已提交
2479
	seq_printf(seq, "%-9s %4u %6d  %6ld   %-3s %6u   %-3s  %-10s "
L
Linus Torvalds 已提交
2480 2481 2482
			"%2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c\n",
		   proto->name,
		   proto->obj_size,
2483
		   sock_prot_inuse_get(seq_file_net(seq), proto),
E
Eric Dumazet 已提交
2484
		   proto->memory_allocated != NULL ? atomic_long_read(proto->memory_allocated) : -1L,
L
Linus Torvalds 已提交
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
		   proto->memory_pressure != NULL ? *proto->memory_pressure ? "yes" : "no" : "NI",
		   proto->max_header,
		   proto->slab == NULL ? "no" : "yes",
		   module_name(proto->owner),
		   proto_method_implemented(proto->close),
		   proto_method_implemented(proto->connect),
		   proto_method_implemented(proto->disconnect),
		   proto_method_implemented(proto->accept),
		   proto_method_implemented(proto->ioctl),
		   proto_method_implemented(proto->init),
		   proto_method_implemented(proto->destroy),
		   proto_method_implemented(proto->shutdown),
		   proto_method_implemented(proto->setsockopt),
		   proto_method_implemented(proto->getsockopt),
		   proto_method_implemented(proto->sendmsg),
		   proto_method_implemented(proto->recvmsg),
		   proto_method_implemented(proto->sendpage),
		   proto_method_implemented(proto->bind),
		   proto_method_implemented(proto->backlog_rcv),
		   proto_method_implemented(proto->hash),
		   proto_method_implemented(proto->unhash),
		   proto_method_implemented(proto->get_port),
		   proto_method_implemented(proto->enter_memory_pressure));
}

static int proto_seq_show(struct seq_file *seq, void *v)
{
2512
	if (v == &proto_list)
L
Linus Torvalds 已提交
2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523
		seq_printf(seq, "%-9s %-4s %-8s %-6s %-5s %-7s %-4s %-10s %s",
			   "protocol",
			   "size",
			   "sockets",
			   "memory",
			   "press",
			   "maxhdr",
			   "slab",
			   "module",
			   "cl co di ac io in de sh ss gs se re sp bi br ha uh gp em\n");
	else
2524
		proto_seq_printf(seq, list_entry(v, struct proto, node));
L
Linus Torvalds 已提交
2525 2526 2527
	return 0;
}

2528
static const struct seq_operations proto_seq_ops = {
L
Linus Torvalds 已提交
2529 2530 2531 2532 2533 2534 2535 2536
	.start  = proto_seq_start,
	.next   = proto_seq_next,
	.stop   = proto_seq_stop,
	.show   = proto_seq_show,
};

static int proto_seq_open(struct inode *inode, struct file *file)
{
2537 2538
	return seq_open_net(inode, file, &proto_seq_ops,
			    sizeof(struct seq_net_private));
L
Linus Torvalds 已提交
2539 2540
}

2541
static const struct file_operations proto_seq_fops = {
L
Linus Torvalds 已提交
2542 2543 2544 2545
	.owner		= THIS_MODULE,
	.open		= proto_seq_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565
	.release	= seq_release_net,
};

static __net_init int proto_init_net(struct net *net)
{
	if (!proc_net_fops_create(net, "protocols", S_IRUGO, &proto_seq_fops))
		return -ENOMEM;

	return 0;
}

static __net_exit void proto_exit_net(struct net *net)
{
	proc_net_remove(net, "protocols");
}


static __net_initdata struct pernet_operations proto_net_ops = {
	.init = proto_init_net,
	.exit = proto_exit_net,
L
Linus Torvalds 已提交
2566 2567 2568 2569
};

static int __init proto_init(void)
{
2570
	return register_pernet_subsys(&proto_net_ops);
L
Linus Torvalds 已提交
2571 2572 2573 2574 2575
}

subsys_initcall(proto_init);

#endif /* PROC_FS */