sock.c 66.8 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>
114
#include <linux/static_key.h>
115
#include <linux/memcontrol.h>
116
#include <linux/prefetch.h>
L
Linus Torvalds 已提交
117 118 119 120 121 122

#include <asm/uaccess.h>

#include <linux/netdevice.h>
#include <net/protocol.h>
#include <linux/skbuff.h>
123
#include <net/net_namespace.h>
124
#include <net/request_sock.h>
L
Linus Torvalds 已提交
125
#include <net/sock.h>
126
#include <linux/net_tstamp.h>
L
Linus Torvalds 已提交
127 128
#include <net/xfrm.h>
#include <linux/ipsec.h>
129
#include <net/cls_cgroup.h>
130
#include <net/netprio_cgroup.h>
L
Linus Torvalds 已提交
131 132 133

#include <linux/filter.h>

134 135
#include <trace/events/sock.h>

L
Linus Torvalds 已提交
136 137 138 139
#ifdef CONFIG_INET
#include <net/tcp.h>
#endif

140
static DEFINE_MUTEX(proto_list_mutex);
G
Glauber Costa 已提交
141 142 143 144 145 146 147 148
static LIST_HEAD(proto_list);

#ifdef CONFIG_CGROUP_MEM_RES_CTLR_KMEM
int mem_cgroup_sockets_init(struct cgroup *cgrp, struct cgroup_subsys *ss)
{
	struct proto *proto;
	int ret = 0;

149
	mutex_lock(&proto_list_mutex);
G
Glauber Costa 已提交
150 151 152 153 154 155 156 157
	list_for_each_entry(proto, &proto_list, node) {
		if (proto->init_cgroup) {
			ret = proto->init_cgroup(cgrp, ss);
			if (ret)
				goto out;
		}
	}

158
	mutex_unlock(&proto_list_mutex);
G
Glauber Costa 已提交
159 160 161 162
	return ret;
out:
	list_for_each_entry_continue_reverse(proto, &proto_list, node)
		if (proto->destroy_cgroup)
163
			proto->destroy_cgroup(cgrp);
164
	mutex_unlock(&proto_list_mutex);
G
Glauber Costa 已提交
165 166 167
	return ret;
}

168
void mem_cgroup_sockets_destroy(struct cgroup *cgrp)
G
Glauber Costa 已提交
169 170 171
{
	struct proto *proto;

172
	mutex_lock(&proto_list_mutex);
G
Glauber Costa 已提交
173 174
	list_for_each_entry_reverse(proto, &proto_list, node)
		if (proto->destroy_cgroup)
175
			proto->destroy_cgroup(cgrp);
176
	mutex_unlock(&proto_list_mutex);
G
Glauber Costa 已提交
177 178 179
}
#endif

180 181 182 183
/*
 * Each address family might have different locking rules, so we have
 * one slock key per address family:
 */
I
Ingo Molnar 已提交
184 185 186
static struct lock_class_key af_family_keys[AF_MAX];
static struct lock_class_key af_family_slock_keys[AF_MAX];

187
struct static_key memcg_socket_limit_enabled;
G
Glauber Costa 已提交
188 189
EXPORT_SYMBOL(memcg_socket_limit_enabled);

I
Ingo Molnar 已提交
190 191 192 193 194
/*
 * Make lock validator output more readable. (we pre-construct these
 * strings build-time, so that runtime initialization of socket
 * locks is fast):
 */
195
static const char *const af_family_key_strings[AF_MAX+1] = {
I
Ingo Molnar 已提交
196 197 198 199 200 201 202
  "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
203
  "sk_lock-AF_RDS"   , "sk_lock-AF_SNA"      , "sk_lock-AF_IRDA"     ,
I
Ingo Molnar 已提交
204
  "sk_lock-AF_PPPOX" , "sk_lock-AF_WANPIPE"  , "sk_lock-AF_LLC"      ,
205
  "sk_lock-27"       , "sk_lock-28"          , "sk_lock-AF_CAN"      ,
206
  "sk_lock-AF_TIPC"  , "sk_lock-AF_BLUETOOTH", "sk_lock-IUCV"        ,
R
Remi Denis-Courmont 已提交
207
  "sk_lock-AF_RXRPC" , "sk_lock-AF_ISDN"     , "sk_lock-AF_PHONET"   ,
208
  "sk_lock-AF_IEEE802154", "sk_lock-AF_CAIF" , "sk_lock-AF_ALG"      ,
A
Aloisio Almeida Jr 已提交
209
  "sk_lock-AF_NFC"   , "sk_lock-AF_MAX"
I
Ingo Molnar 已提交
210
};
211
static const char *const af_family_slock_key_strings[AF_MAX+1] = {
I
Ingo Molnar 已提交
212 213 214 215 216 217 218
  "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
219
  "slock-AF_RDS"   , "slock-AF_SNA"      , "slock-AF_IRDA"     ,
I
Ingo Molnar 已提交
220
  "slock-AF_PPPOX" , "slock-AF_WANPIPE"  , "slock-AF_LLC"      ,
221
  "slock-27"       , "slock-28"          , "slock-AF_CAN"      ,
222
  "slock-AF_TIPC"  , "slock-AF_BLUETOOTH", "slock-AF_IUCV"     ,
R
Remi Denis-Courmont 已提交
223
  "slock-AF_RXRPC" , "slock-AF_ISDN"     , "slock-AF_PHONET"   ,
224
  "slock-AF_IEEE802154", "slock-AF_CAIF" , "slock-AF_ALG"      ,
A
Aloisio Almeida Jr 已提交
225
  "slock-AF_NFC"   , "slock-AF_MAX"
I
Ingo Molnar 已提交
226
};
227
static const char *const af_family_clock_key_strings[AF_MAX+1] = {
228 229 230 231 232 233 234
  "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
235
  "clock-AF_RDS"   , "clock-AF_SNA"      , "clock-AF_IRDA"     ,
236
  "clock-AF_PPPOX" , "clock-AF_WANPIPE"  , "clock-AF_LLC"      ,
237
  "clock-27"       , "clock-28"          , "clock-AF_CAN"      ,
238
  "clock-AF_TIPC"  , "clock-AF_BLUETOOTH", "clock-AF_IUCV"     ,
R
Remi Denis-Courmont 已提交
239
  "clock-AF_RXRPC" , "clock-AF_ISDN"     , "clock-AF_PHONET"   ,
240
  "clock-AF_IEEE802154", "clock-AF_CAIF" , "clock-AF_ALG"      ,
A
Aloisio Almeida Jr 已提交
241
  "clock-AF_NFC"   , "clock-AF_MAX"
242
};
243 244 245 246 247 248 249

/*
 * 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
250 251 252 253 254 255
/* 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
E
Eric Dumazet 已提交
256
#define _SK_MEM_OVERHEAD	SKB_TRUESIZE(256)
L
Linus Torvalds 已提交
257 258 259 260
#define SK_WMEM_MAX		(_SK_MEM_OVERHEAD * _SK_MEM_PACKETS)
#define SK_RMEM_MAX		(_SK_MEM_OVERHEAD * _SK_MEM_PACKETS)

/* Run time adjustable parameters. */
261
__u32 sysctl_wmem_max __read_mostly = SK_WMEM_MAX;
262
EXPORT_SYMBOL(sysctl_wmem_max);
263
__u32 sysctl_rmem_max __read_mostly = SK_RMEM_MAX;
264
EXPORT_SYMBOL(sysctl_rmem_max);
265 266
__u32 sysctl_wmem_default __read_mostly = SK_WMEM_MAX;
__u32 sysctl_rmem_default __read_mostly = SK_RMEM_MAX;
L
Linus Torvalds 已提交
267

L
Lucas De Marchi 已提交
268
/* Maximal space eaten by iovec or ancillary data plus some space */
269
int sysctl_optmem_max __read_mostly = sizeof(unsigned long)*(2*UIO_MAXIOV+512);
E
Eric Dumazet 已提交
270
EXPORT_SYMBOL(sysctl_optmem_max);
L
Linus Torvalds 已提交
271

272 273
#if defined(CONFIG_CGROUPS)
#if !defined(CONFIG_NET_CLS_CGROUP)
274 275 276
int net_cls_subsys_id = -1;
EXPORT_SYMBOL_GPL(net_cls_subsys_id);
#endif
277 278 279 280 281
#if !defined(CONFIG_NETPRIO_CGROUP)
int net_prio_subsys_id = -1;
EXPORT_SYMBOL_GPL(net_prio_subsys_id);
#endif
#endif
282

L
Linus Torvalds 已提交
283 284 285 286 287 288 289 290
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;
291 292
	if (tv.tv_usec < 0 || tv.tv_usec >= USEC_PER_SEC)
		return -EDOM;
L
Linus Torvalds 已提交
293

294
	if (tv.tv_sec < 0) {
295 296
		static int warned __read_mostly;

297
		*timeo_p = 0;
298
		if (warned < 10 && net_ratelimit()) {
299 300 301
			warned++;
			printk(KERN_INFO "sock_set_timeout: `%s' (pid %d) "
			       "tries to set negative timeout\n",
302
				current->comm, task_pid_nr(current));
303
		}
304 305
		return 0;
	}
L
Linus Torvalds 已提交
306 307 308 309 310 311 312 313 314 315 316 317
	*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];
318 319
	if (strcmp(warncomm, current->comm) && warned < 5) {
		strcpy(warncomm,  current->comm);
L
Linus Torvalds 已提交
320 321 322 323 324 325
		printk(KERN_WARNING "process `%s' is using obsolete "
		       "%s SO_BSDCOMPAT\n", warncomm, name);
		warned++;
	}
}

E
Eric Dumazet 已提交
326 327 328
#define SK_FLAGS_TIMESTAMP ((1UL << SOCK_TIMESTAMP) | (1UL << SOCK_TIMESTAMPING_RX_SOFTWARE))

static void sock_disable_timestamp(struct sock *sk, unsigned long flags)
329
{
E
Eric Dumazet 已提交
330 331 332
	if (sk->sk_flags & flags) {
		sk->sk_flags &= ~flags;
		if (!(sk->sk_flags & SK_FLAGS_TIMESTAMP))
333
			net_disable_timestamp();
L
Linus Torvalds 已提交
334 335 336 337
	}
}


338 339
int sock_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
{
E
Eric Dumazet 已提交
340
	int err;
341
	int skb_len;
342 343
	unsigned long flags;
	struct sk_buff_head *list = &sk->sk_receive_queue;
344

E
Eric Dumazet 已提交
345
	if (atomic_read(&sk->sk_rmem_alloc) >= sk->sk_rcvbuf) {
E
Eric Dumazet 已提交
346
		atomic_inc(&sk->sk_drops);
347
		trace_sock_rcvqueue_full(sk, skb);
E
Eric Dumazet 已提交
348
		return -ENOMEM;
349 350
	}

351
	err = sk_filter(sk, skb);
352
	if (err)
E
Eric Dumazet 已提交
353
		return err;
354

355
	if (!sk_rmem_schedule(sk, skb->truesize)) {
E
Eric Dumazet 已提交
356 357
		atomic_inc(&sk->sk_drops);
		return -ENOBUFS;
358 359
	}

360 361
	skb->dev = NULL;
	skb_set_owner_r(skb, sk);
362

363 364 365 366 367 368 369
	/* 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
370 371 372 373 374
	/* we escape from rcu protected region, make sure we dont leak
	 * a norefcounted dst
	 */
	skb_dst_force(skb);

375 376 377 378
	spin_lock_irqsave(&list->lock, flags);
	skb->dropcount = atomic_read(&sk->sk_drops);
	__skb_queue_tail(list, skb);
	spin_unlock_irqrestore(&list->lock, flags);
379 380 381

	if (!sock_flag(sk, SOCK_DEAD))
		sk->sk_data_ready(sk, skb_len);
E
Eric Dumazet 已提交
382
	return 0;
383 384 385
}
EXPORT_SYMBOL(sock_queue_rcv_skb);

386
int sk_receive_skb(struct sock *sk, struct sk_buff *skb, const int nested)
387 388 389
{
	int rc = NET_RX_SUCCESS;

390
	if (sk_filter(sk, skb))
391 392 393 394
		goto discard_and_relse;

	skb->dev = NULL;

395
	if (sk_rcvqueues_full(sk, skb, sk->sk_rcvbuf)) {
396 397 398
		atomic_inc(&sk->sk_drops);
		goto discard_and_relse;
	}
399 400 401 402
	if (nested)
		bh_lock_sock_nested(sk);
	else
		bh_lock_sock(sk);
I
Ingo Molnar 已提交
403 404 405 406 407 408
	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
409
		rc = sk_backlog_rcv(sk, skb);
I
Ingo Molnar 已提交
410 411

		mutex_release(&sk->sk_lock.dep_map, 1, _RET_IP_);
412
	} else if (sk_add_backlog(sk, skb, sk->sk_rcvbuf)) {
Z
Zhu Yi 已提交
413 414 415 416 417
		bh_unlock_sock(sk);
		atomic_inc(&sk->sk_drops);
		goto discard_and_relse;
	}

418 419 420 421 422 423 424 425 426 427
	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
428 429 430 431 432 433
void sk_reset_txq(struct sock *sk)
{
	sk_tx_queue_clear(sk);
}
EXPORT_SYMBOL(sk_reset_txq);

434 435
struct dst_entry *__sk_dst_check(struct sock *sk, u32 cookie)
{
E
Eric Dumazet 已提交
436
	struct dst_entry *dst = __sk_dst_get(sk);
437 438

	if (dst && dst->obsolete && dst->ops->check(dst, cookie) == NULL) {
K
Krishna Kumar 已提交
439
		sk_tx_queue_clear(sk);
440
		RCU_INIT_POINTER(sk->sk_dst_cache, NULL);
441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462
		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);

463 464 465 466
static int sock_bindtodevice(struct sock *sk, char __user *optval, int optlen)
{
	int ret = -ENOPROTOOPT;
#ifdef CONFIG_NETDEVICES
467
	struct net *net = sock_net(sk);
468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492
	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;

493 494
	index = 0;
	if (devname[0] != '\0') {
495
		struct net_device *dev;
496

497 498 499 500 501
		rcu_read_lock();
		dev = dev_get_by_name_rcu(net, devname);
		if (dev)
			index = dev->ifindex;
		rcu_read_unlock();
502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519
		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;
}

520 521 522 523 524 525 526 527
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
528 529 530 531 532 533
/*
 *	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,
534
		    char __user *optval, unsigned int optlen)
L
Linus Torvalds 已提交
535
{
E
Eric Dumazet 已提交
536
	struct sock *sk = sock->sk;
L
Linus Torvalds 已提交
537 538 539 540
	int val;
	int valbool;
	struct linger ling;
	int ret = 0;
541

L
Linus Torvalds 已提交
542 543 544 545
	/*
	 *	Options without arguments
	 */

546 547 548
	if (optname == SO_BINDTODEVICE)
		return sock_bindtodevice(sk, optval, optlen);

549 550
	if (optlen < sizeof(int))
		return -EINVAL;
551

L
Linus Torvalds 已提交
552 553
	if (get_user(val, (int __user *)optval))
		return -EFAULT;
554

E
Eric Dumazet 已提交
555
	valbool = val ? 1 : 0;
L
Linus Torvalds 已提交
556 557 558

	lock_sock(sk);

E
Eric Dumazet 已提交
559
	switch (optname) {
560
	case SO_DEBUG:
E
Eric Dumazet 已提交
561
		if (val && !capable(CAP_NET_ADMIN))
562
			ret = -EACCES;
E
Eric Dumazet 已提交
563
		else
564
			sock_valbool_flag(sk, SOCK_DBG, valbool);
565 566
		break;
	case SO_REUSEADDR:
567
		sk->sk_reuse = (valbool ? SK_CAN_REUSE : SK_NO_REUSE);
568 569
		break;
	case SO_TYPE:
570
	case SO_PROTOCOL:
571
	case SO_DOMAIN:
572 573 574 575
	case SO_ERROR:
		ret = -ENOPROTOOPT;
		break;
	case SO_DONTROUTE:
576
		sock_valbool_flag(sk, SOCK_LOCALROUTE, valbool);
577 578 579 580 581 582
		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
583 584 585 586 587
		 * about it this is right. Otherwise apps have to
		 * play 'guess the biggest size' games. RCVBUF/SNDBUF
		 * are treated in BSD as hints
		 */
		val = min_t(u32, val, sysctl_wmem_max);
588
set_sndbuf:
589
		sk->sk_userlocks |= SOCK_SNDBUF_LOCK;
590 591
		sk->sk_sndbuf = max_t(u32, val * 2, SOCK_MIN_SNDBUF);
		/* Wake up sending tasks if we upped the value. */
592 593
		sk->sk_write_space(sk);
		break;
L
Linus Torvalds 已提交
594

595 596 597 598 599 600
	case SO_SNDBUFFORCE:
		if (!capable(CAP_NET_ADMIN)) {
			ret = -EPERM;
			break;
		}
		goto set_sndbuf;
601

602 603
	case SO_RCVBUF:
		/* Don't error on this BSD doesn't and if you think
604 605 606 607 608
		 * about it this is right. Otherwise apps have to
		 * play 'guess the biggest size' games. RCVBUF/SNDBUF
		 * are treated in BSD as hints
		 */
		val = min_t(u32, val, sysctl_rmem_max);
609
set_rcvbuf:
610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625
		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.
		 */
626
		sk->sk_rcvbuf = max_t(u32, val * 2, SOCK_MIN_RCVBUF);
627 628 629 630 631
		break;

	case SO_RCVBUFFORCE:
		if (!capable(CAP_NET_ADMIN)) {
			ret = -EPERM;
L
Linus Torvalds 已提交
632
			break;
633 634
		}
		goto set_rcvbuf;
L
Linus Torvalds 已提交
635

636
	case SO_KEEPALIVE:
L
Linus Torvalds 已提交
637
#ifdef CONFIG_INET
638 639
		if (sk->sk_protocol == IPPROTO_TCP)
			tcp_set_keepalive(sk, valbool);
L
Linus Torvalds 已提交
640
#endif
641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661
		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
662
			break;
663
		}
E
Eric Dumazet 已提交
664
		if (copy_from_user(&ling, optval, sizeof(ling))) {
665
			ret = -EFAULT;
L
Linus Torvalds 已提交
666
			break;
667 668 669 670
		}
		if (!ling.l_onoff)
			sock_reset_flag(sk, SOCK_LINGER);
		else {
L
Linus Torvalds 已提交
671
#if (BITS_PER_LONG == 32)
672 673
			if ((unsigned int)ling.l_linger >= MAX_SCHEDULE_TIMEOUT/HZ)
				sk->sk_lingertime = MAX_SCHEDULE_TIMEOUT;
L
Linus Torvalds 已提交
674
			else
675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692
#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:
693
	case SO_TIMESTAMPNS:
694
		if (valbool)  {
695 696 697 698
			if (optname == SO_TIMESTAMP)
				sock_reset_flag(sk, SOCK_RCVTSTAMPNS);
			else
				sock_set_flag(sk, SOCK_RCVTSTAMPNS);
699
			sock_set_flag(sk, SOCK_RCVTSTAMP);
700
			sock_enable_timestamp(sk, SOCK_TIMESTAMP);
701
		} else {
702
			sock_reset_flag(sk, SOCK_RCVTSTAMP);
703 704
			sock_reset_flag(sk, SOCK_RCVTSTAMPNS);
		}
705 706
		break;

707 708
	case SO_TIMESTAMPING:
		if (val & ~SOF_TIMESTAMPING_MASK) {
709
			ret = -EINVAL;
710 711 712 713 714 715 716 717 718 719 720 721 722
			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,
E
Eric Dumazet 已提交
723
					       (1UL << SOCK_TIMESTAMPING_RX_SOFTWARE));
724 725 726 727 728 729 730 731
		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;

732 733 734 735 736 737 738 739 740 741 742 743 744
	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
745

746 747 748 749
	case SO_ATTACH_FILTER:
		ret = -EINVAL;
		if (optlen == sizeof(struct sock_fprog)) {
			struct sock_fprog fprog;
L
Linus Torvalds 已提交
750

751 752
			ret = -EFAULT;
			if (copy_from_user(&fprog, optval, sizeof(fprog)))
L
Linus Torvalds 已提交
753
				break;
754 755 756 757 758 759

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

	case SO_DETACH_FILTER:
760
		ret = sk_detach_filter(sk);
761
		break;
L
Linus Torvalds 已提交
762

763 764 765 766 767 768
	case SO_PASSSEC:
		if (valbool)
			set_bit(SOCK_PASSSEC, &sock->flags);
		else
			clear_bit(SOCK_PASSSEC, &sock->flags);
		break;
769 770 771
	case SO_MARK:
		if (!capable(CAP_NET_ADMIN))
			ret = -EPERM;
E
Eric Dumazet 已提交
772
		else
773 774
			sk->sk_mark = val;
		break;
C
Catherine Zhang 已提交
775

L
Linus Torvalds 已提交
776 777
		/* We implement the SO_SNDLOWAT etc to
		   not be settable (1003.1g 5.3) */
778
	case SO_RXQ_OVFL:
779
		sock_valbool_flag(sk, SOCK_RXQ_OVFL, valbool);
780
		break;
781 782 783 784 785

	case SO_WIFI_STATUS:
		sock_valbool_flag(sk, SOCK_WIFI_STATUS, valbool);
		break;

786 787 788 789 790 791
	case SO_PEEK_OFF:
		if (sock->ops->set_peek_off)
			sock->ops->set_peek_off(sk, val);
		else
			ret = -EOPNOTSUPP;
		break;
792 793 794 795 796

	case SO_NOFCS:
		sock_valbool_flag(sk, SOCK_NOFCS, valbool);
		break;

797 798 799
	default:
		ret = -ENOPROTOOPT;
		break;
800
	}
L
Linus Torvalds 已提交
801 802 803
	release_sock(sk);
	return ret;
}
E
Eric Dumazet 已提交
804
EXPORT_SYMBOL(sock_setsockopt);
L
Linus Torvalds 已提交
805 806


807 808 809 810 811 812 813 814 815 816 817 818
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);
	}
}
819
EXPORT_SYMBOL_GPL(cred_to_ucred);
820

L
Linus Torvalds 已提交
821 822 823 824
int sock_getsockopt(struct socket *sock, int level, int optname,
		    char __user *optval, int __user *optlen)
{
	struct sock *sk = sock->sk;
825

826
	union {
827 828
		int val;
		struct linger ling;
L
Linus Torvalds 已提交
829 830
		struct timeval tm;
	} v;
831

832
	int lv = sizeof(int);
L
Linus Torvalds 已提交
833
	int len;
834

835
	if (get_user(len, optlen))
836
		return -EFAULT;
837
	if (len < 0)
L
Linus Torvalds 已提交
838
		return -EINVAL;
839

840
	memset(&v, 0, sizeof(v));
841

E
Eric Dumazet 已提交
842
	switch (optname) {
843 844 845 846 847 848 849 850 851
	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:
E
Eric Dumazet 已提交
852
		v.val = sock_flag(sk, SOCK_BROADCAST);
853 854 855 856 857 858 859 860 861 862 863 864 865 866 867
		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:
E
Eric Dumazet 已提交
868
		v.val = sock_flag(sk, SOCK_KEEPOPEN);
869 870 871 872 873 874
		break;

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

875 876 877 878
	case SO_PROTOCOL:
		v.val = sk->sk_protocol;
		break;

879 880 881 882
	case SO_DOMAIN:
		v.val = sk->sk_family;
		break;

883 884
	case SO_ERROR:
		v.val = -sock_error(sk);
E
Eric Dumazet 已提交
885
		if (v.val == 0)
886 887 888 889
			v.val = xchg(&sk->sk_err_soft, 0);
		break;

	case SO_OOBINLINE:
E
Eric Dumazet 已提交
890
		v.val = sock_flag(sk, SOCK_URGINLINE);
891 892 893 894 895 896 897 898 899 900 901 902
		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);
E
Eric Dumazet 已提交
903
		v.ling.l_onoff	= sock_flag(sk, SOCK_LINGER);
904 905 906 907 908 909 910 911
		v.ling.l_linger	= sk->sk_lingertime / HZ;
		break;

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

	case SO_TIMESTAMP:
912 913 914 915 916 917
		v.val = sock_flag(sk, SOCK_RCVTSTAMP) &&
				!sock_flag(sk, SOCK_RCVTSTAMPNS);
		break;

	case SO_TIMESTAMPNS:
		v.val = sock_flag(sk, SOCK_RCVTSTAMPNS);
918 919
		break;

920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937
	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;

938
	case SO_RCVTIMEO:
E
Eric Dumazet 已提交
939
		lv = sizeof(struct timeval);
940 941 942 943 944 945 946 947 948 949
		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
950
		lv = sizeof(struct timeval);
951 952 953 954 955 956 957 958
		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
959

960 961 962
	case SO_RCVLOWAT:
		v.val = sk->sk_rcvlowat;
		break;
L
Linus Torvalds 已提交
963

964
	case SO_SNDLOWAT:
E
Eric Dumazet 已提交
965
		v.val = 1;
966
		break;
L
Linus Torvalds 已提交
967

968
	case SO_PASSCRED:
969
		v.val = !!test_bit(SOCK_PASSCRED, &sock->flags);
970
		break;
L
Linus Torvalds 已提交
971

972
	case SO_PEERCRED:
973 974 975 976 977 978
	{
		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))
979 980
			return -EFAULT;
		goto lenout;
981
	}
L
Linus Torvalds 已提交
982

983 984 985 986 987 988 989 990 991 992 993 994
	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
995

996 997 998 999 1000 1001
	/* 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
1002

1003
	case SO_PASSSEC:
1004
		v.val = !!test_bit(SOCK_PASSSEC, &sock->flags);
1005
		break;
C
Catherine Zhang 已提交
1006

1007 1008
	case SO_PEERSEC:
		return security_socket_getpeersec_stream(sock, optval, optlen, len);
L
Linus Torvalds 已提交
1009

1010 1011 1012 1013
	case SO_MARK:
		v.val = sk->sk_mark;
		break;

1014
	case SO_RXQ_OVFL:
E
Eric Dumazet 已提交
1015
		v.val = sock_flag(sk, SOCK_RXQ_OVFL);
1016 1017
		break;

1018
	case SO_WIFI_STATUS:
E
Eric Dumazet 已提交
1019
		v.val = sock_flag(sk, SOCK_WIFI_STATUS);
1020 1021
		break;

1022 1023 1024 1025 1026 1027
	case SO_PEEK_OFF:
		if (!sock->ops->set_peek_off)
			return -EOPNOTSUPP;

		v.val = sk->sk_peek_off;
		break;
1028
	case SO_NOFCS:
E
Eric Dumazet 已提交
1029
		v.val = sock_flag(sk, SOCK_NOFCS);
1030
		break;
1031 1032
	default:
		return -ENOPROTOOPT;
L
Linus Torvalds 已提交
1033
	}
1034

L
Linus Torvalds 已提交
1035 1036 1037 1038 1039
	if (len > lv)
		len = lv;
	if (copy_to_user(optval, &v, len))
		return -EFAULT;
lenout:
1040 1041 1042
	if (put_user(len, optlen))
		return -EFAULT;
	return 0;
L
Linus Torvalds 已提交
1043 1044
}

I
Ingo Molnar 已提交
1045 1046 1047 1048 1049
/*
 * Initialize an sk_lock.
 *
 * (We also register the sk_lock with the lock validator.)
 */
D
Dave Jones 已提交
1050
static inline void sock_lock_init(struct sock *sk)
I
Ingo Molnar 已提交
1051
{
1052 1053 1054 1055 1056
	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
1057 1058
}

E
Eric Dumazet 已提交
1059 1060 1061
/*
 * 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.
1062
 * We must not copy fields between sk_dontcopy_begin and sk_dontcopy_end
E
Eric Dumazet 已提交
1063
 */
1064 1065 1066 1067 1068
static void sock_copy(struct sock *nsk, const struct sock *osk)
{
#ifdef CONFIG_SECURITY_NETWORK
	void *sptr = nsk->sk_security;
#endif
1069 1070 1071 1072 1073
	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));

1074 1075 1076 1077 1078 1079
#ifdef CONFIG_SECURITY_NETWORK
	nsk->sk_security = sptr;
	security_sk_clone(osk, nsk);
#endif
}

1080 1081 1082 1083 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 1109
/*
 * 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);

1110 1111
static struct sock *sk_prot_alloc(struct proto *prot, gfp_t priority,
		int family)
1112 1113 1114 1115 1116
{
	struct sock *sk;
	struct kmem_cache *slab;

	slab = prot->slab;
1117 1118 1119 1120 1121
	if (slab != NULL) {
		sk = kmem_cache_alloc(slab, priority & ~__GFP_ZERO);
		if (!sk)
			return sk;
		if (priority & __GFP_ZERO) {
1122 1123 1124 1125
			if (prot->clear_sk)
				prot->clear_sk(sk, prot->obj_size);
			else
				sk_prot_clear_nulls(sk, prot->obj_size);
1126
		}
1127
	} else
1128 1129
		sk = kmalloc(prot->obj_size, priority);

1130
	if (sk != NULL) {
1131 1132
		kmemcheck_annotate_bitfield(sk, flags);

1133 1134 1135 1136 1137
		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
1138
		sk_tx_queue_clear(sk);
1139 1140
	}

1141
	return sk;
1142 1143 1144 1145 1146 1147 1148 1149 1150

out_free_sec:
	security_sk_free(sk);
out_free:
	if (slab != NULL)
		kmem_cache_free(slab, sk);
	else
		kfree(sk);
	return NULL;
1151 1152 1153 1154 1155
}

static void sk_prot_free(struct proto *prot, struct sock *sk)
{
	struct kmem_cache *slab;
1156
	struct module *owner;
1157

1158
	owner = prot->owner;
1159
	slab = prot->slab;
1160 1161

	security_sk_free(sk);
1162 1163 1164 1165
	if (slab != NULL)
		kmem_cache_free(slab, sk);
	else
		kfree(sk);
1166
	module_put(owner);
1167 1168
}

1169 1170 1171
#ifdef CONFIG_CGROUPS
void sock_update_classid(struct sock *sk)
{
1172
	u32 classid;
1173

1174 1175 1176
	rcu_read_lock();  /* doing current task, which cannot vanish. */
	classid = task_cls_classid(current);
	rcu_read_unlock();
1177 1178 1179
	if (classid && classid != sk->sk_classid)
		sk->sk_classid = classid;
}
1180
EXPORT_SYMBOL(sock_update_classid);
1181 1182 1183 1184 1185

void sock_update_netprioidx(struct sock *sk)
{
	if (in_interrupt())
		return;
1186 1187

	sk->sk_cgrp_prioidx = task_netprioidx(current);
1188 1189
}
EXPORT_SYMBOL_GPL(sock_update_netprioidx);
1190 1191
#endif

L
Linus Torvalds 已提交
1192 1193
/**
 *	sk_alloc - All socket objects are allocated here
1194
 *	@net: the applicable net namespace
1195 1196 1197
 *	@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
1198
 */
1199
struct sock *sk_alloc(struct net *net, int family, gfp_t priority,
1200
		      struct proto *prot)
L
Linus Torvalds 已提交
1201
{
1202
	struct sock *sk;
L
Linus Torvalds 已提交
1203

1204
	sk = sk_prot_alloc(prot, priority | __GFP_ZERO, family);
L
Linus Torvalds 已提交
1205
	if (sk) {
1206 1207 1208 1209 1210 1211 1212
		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);
1213
		sock_net_set(sk, get_net(net));
1214
		atomic_set(&sk->sk_wmem_alloc, 1);
1215 1216

		sock_update_classid(sk);
1217
		sock_update_netprioidx(sk);
L
Linus Torvalds 已提交
1218
	}
1219

1220
	return sk;
L
Linus Torvalds 已提交
1221
}
E
Eric Dumazet 已提交
1222
EXPORT_SYMBOL(sk_alloc);
L
Linus Torvalds 已提交
1223

1224
static void __sk_free(struct sock *sk)
L
Linus Torvalds 已提交
1225 1226 1227 1228 1229 1230
{
	struct sk_filter *filter;

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

1231 1232
	filter = rcu_dereference_check(sk->sk_filter,
				       atomic_read(&sk->sk_wmem_alloc) == 0);
L
Linus Torvalds 已提交
1233
	if (filter) {
1234
		sk_filter_uncharge(sk, filter);
1235
		RCU_INIT_POINTER(sk->sk_filter, NULL);
L
Linus Torvalds 已提交
1236 1237
	}

E
Eric Dumazet 已提交
1238
	sock_disable_timestamp(sk, SK_FLAGS_TIMESTAMP);
L
Linus Torvalds 已提交
1239 1240 1241

	if (atomic_read(&sk->sk_omem_alloc))
		printk(KERN_DEBUG "%s: optmem leakage (%d bytes) detected.\n",
1242
		       __func__, atomic_read(&sk->sk_omem_alloc));
L
Linus Torvalds 已提交
1243

1244 1245 1246
	if (sk->sk_peer_cred)
		put_cred(sk->sk_peer_cred);
	put_pid(sk->sk_peer_pid);
1247
	put_net(sock_net(sk));
1248
	sk_prot_free(sk->sk_prot_creator, sk);
L
Linus Torvalds 已提交
1249
}
1250 1251 1252 1253

void sk_free(struct sock *sk)
{
	/*
L
Lucas De Marchi 已提交
1254
	 * We subtract one from sk_wmem_alloc and can know if
1255 1256 1257 1258 1259 1260
	 * 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
1261
EXPORT_SYMBOL(sk_free);
L
Linus Torvalds 已提交
1262

1263
/*
L
Lucas De Marchi 已提交
1264 1265
 * Last sock_put should drop reference to sk->sk_net. It has already
 * been dropped in sk_change_net. Taking reference to stopping namespace
1266
 * is not an option.
L
Lucas De Marchi 已提交
1267
 * Take reference to a socket to remove it from hash _alive_ and after that
1268 1269 1270 1271 1272 1273 1274 1275 1276
 * 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);
1277
	release_net(sock_net(sk));
1278
	sock_net_set(sk, get_net(&init_net));
1279 1280
	sock_put(sk);
}
1281
EXPORT_SYMBOL(sk_release_kernel);
1282

1283 1284 1285 1286 1287 1288
static void sk_update_clone(const struct sock *sk, struct sock *newsk)
{
	if (mem_cgroup_sockets_enabled && sk->sk_cgrp)
		sock_update_memcg(newsk);
}

1289 1290 1291 1292 1293 1294 1295 1296
/**
 *	sk_clone_lock - clone a socket, and lock its clone
 *	@sk: the socket to clone
 *	@priority: for allocation (%GFP_KERNEL, %GFP_ATOMIC, etc)
 *
 *	Caller must unlock socket even in error path (bh_unlock_sock(newsk))
 */
struct sock *sk_clone_lock(const struct sock *sk, const gfp_t priority)
1297
{
1298
	struct sock *newsk;
1299

1300
	newsk = sk_prot_alloc(sk->sk_prot, priority, sk->sk_family);
1301 1302 1303
	if (newsk != NULL) {
		struct sk_filter *filter;

1304
		sock_copy(newsk, sk);
1305 1306

		/* SANITY */
1307
		get_net(sock_net(newsk));
1308 1309 1310
		sk_node_init(&newsk->sk_node);
		sock_lock_init(newsk);
		bh_lock_sock(newsk);
1311
		newsk->sk_backlog.head	= newsk->sk_backlog.tail = NULL;
Z
Zhu Yi 已提交
1312
		newsk->sk_backlog.len = 0;
1313 1314

		atomic_set(&newsk->sk_rmem_alloc, 0);
1315 1316 1317 1318
		/*
		 * sk_wmem_alloc set to one (see sk_free() and sock_wfree())
		 */
		atomic_set(&newsk->sk_wmem_alloc, 1);
1319 1320 1321
		atomic_set(&newsk->sk_omem_alloc, 0);
		skb_queue_head_init(&newsk->sk_receive_queue);
		skb_queue_head_init(&newsk->sk_write_queue);
1322 1323 1324
#ifdef CONFIG_NET_DMA
		skb_queue_head_init(&newsk->sk_async_wait_queue);
#endif
1325

E
Eric Dumazet 已提交
1326
		spin_lock_init(&newsk->sk_dst_lock);
1327
		rwlock_init(&newsk->sk_callback_lock);
1328 1329 1330
		lockdep_set_class_and_name(&newsk->sk_callback_lock,
				af_callback_keys + newsk->sk_family,
				af_family_clock_key_strings[newsk->sk_family]);
1331 1332 1333 1334 1335 1336 1337 1338 1339 1340

		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);

1341
		filter = rcu_dereference_protected(newsk->sk_filter, 1);
1342 1343 1344 1345 1346 1347 1348
		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;
1349
			bh_unlock_sock(newsk);
1350 1351 1352 1353 1354 1355 1356
			sk_free(newsk);
			newsk = NULL;
			goto out;
		}

		newsk->sk_err	   = 0;
		newsk->sk_priority = 0;
E
Eric Dumazet 已提交
1357 1358 1359 1360 1361
		/*
		 * Before updating sk_refcnt, we must commit prior changes to memory
		 * (Documentation/RCU/rculist_nulls.txt for details)
		 */
		smp_wmb();
1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375
		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);
1376
		sk_set_socket(newsk, NULL);
1377
		newsk->sk_wq = NULL;
1378

1379 1380
		sk_update_clone(sk, newsk);

1381
		if (newsk->sk_prot->sockets_allocated)
1382
			sk_sockets_allocated_inc(newsk);
1383

E
Eric Dumazet 已提交
1384
		if (newsk->sk_flags & SK_FLAGS_TIMESTAMP)
1385
			net_enable_timestamp();
1386 1387 1388 1389
	}
out:
	return newsk;
}
1390
EXPORT_SYMBOL_GPL(sk_clone_lock);
1391

1392 1393 1394 1395 1396
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)
1397
		sk->sk_route_caps |= NETIF_F_GSO_SOFTWARE;
E
Eric Dumazet 已提交
1398
	sk->sk_route_caps &= ~sk->sk_route_nocaps;
1399
	if (sk_can_gso(sk)) {
1400
		if (dst->header_len) {
1401
			sk->sk_route_caps &= ~NETIF_F_GSO_MASK;
1402
		} else {
1403
			sk->sk_route_caps |= NETIF_F_SG | NETIF_F_HW_CSUM;
1404 1405
			sk->sk_gso_max_size = dst->dev->gso_max_size;
		}
1406 1407 1408 1409
	}
}
EXPORT_SYMBOL_GPL(sk_setup_caps);

L
Linus Torvalds 已提交
1410 1411
void __init sk_init(void)
{
1412
	if (totalram_pages <= 4096) {
L
Linus Torvalds 已提交
1413 1414 1415 1416
		sysctl_wmem_max = 32767;
		sysctl_rmem_max = 32767;
		sysctl_wmem_default = 32767;
		sysctl_rmem_default = 32767;
1417
	} else if (totalram_pages >= 131072) {
L
Linus Torvalds 已提交
1418 1419 1420 1421 1422 1423 1424 1425 1426 1427
		sysctl_wmem_max = 131071;
		sysctl_rmem_max = 131071;
	}
}

/*
 *	Simple resource managers for sockets.
 */


1428 1429
/*
 * Write buffer destructor automatically called from kfree_skb.
L
Linus Torvalds 已提交
1430 1431 1432 1433
 */
void sock_wfree(struct sk_buff *skb)
{
	struct sock *sk = skb->sk;
E
Eric Dumazet 已提交
1434
	unsigned int len = skb->truesize;
L
Linus Torvalds 已提交
1435

E
Eric Dumazet 已提交
1436 1437 1438 1439 1440 1441
	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
1442
		sk->sk_write_space(sk);
E
Eric Dumazet 已提交
1443 1444
		len = 1;
	}
1445
	/*
E
Eric Dumazet 已提交
1446 1447
	 * if sk_wmem_alloc reaches 0, we must finish what sk_free()
	 * could not do because of in-flight packets
1448
	 */
E
Eric Dumazet 已提交
1449
	if (atomic_sub_and_test(len, &sk->sk_wmem_alloc))
1450
		__sk_free(sk);
L
Linus Torvalds 已提交
1451
}
E
Eric Dumazet 已提交
1452
EXPORT_SYMBOL(sock_wfree);
L
Linus Torvalds 已提交
1453

1454 1455
/*
 * Read buffer destructor automatically called from kfree_skb.
L
Linus Torvalds 已提交
1456 1457 1458 1459
 */
void sock_rfree(struct sk_buff *skb)
{
	struct sock *sk = skb->sk;
E
Eric Dumazet 已提交
1460
	unsigned int len = skb->truesize;
L
Linus Torvalds 已提交
1461

E
Eric Dumazet 已提交
1462 1463
	atomic_sub(len, &sk->sk_rmem_alloc);
	sk_mem_uncharge(sk, len);
L
Linus Torvalds 已提交
1464
}
E
Eric Dumazet 已提交
1465
EXPORT_SYMBOL(sock_rfree);
L
Linus Torvalds 已提交
1466 1467 1468 1469 1470 1471


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

E
Eric Dumazet 已提交
1472
	read_lock_bh(&sk->sk_callback_lock);
L
Linus Torvalds 已提交
1473
	uid = sk->sk_socket ? SOCK_INODE(sk->sk_socket)->i_uid : 0;
E
Eric Dumazet 已提交
1474
	read_unlock_bh(&sk->sk_callback_lock);
L
Linus Torvalds 已提交
1475 1476
	return uid;
}
E
Eric Dumazet 已提交
1477
EXPORT_SYMBOL(sock_i_uid);
L
Linus Torvalds 已提交
1478 1479 1480 1481 1482

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

E
Eric Dumazet 已提交
1483
	read_lock_bh(&sk->sk_callback_lock);
L
Linus Torvalds 已提交
1484
	ino = sk->sk_socket ? SOCK_INODE(sk->sk_socket)->i_ino : 0;
E
Eric Dumazet 已提交
1485
	read_unlock_bh(&sk->sk_callback_lock);
L
Linus Torvalds 已提交
1486 1487
	return ino;
}
E
Eric Dumazet 已提交
1488
EXPORT_SYMBOL(sock_i_ino);
L
Linus Torvalds 已提交
1489 1490 1491 1492

/*
 * Allocate a skb from the socket's send buffer.
 */
V
Victor Fusco 已提交
1493
struct sk_buff *sock_wmalloc(struct sock *sk, unsigned long size, int force,
A
Al Viro 已提交
1494
			     gfp_t priority)
L
Linus Torvalds 已提交
1495 1496
{
	if (force || atomic_read(&sk->sk_wmem_alloc) < sk->sk_sndbuf) {
E
Eric Dumazet 已提交
1497
		struct sk_buff *skb = alloc_skb(size, priority);
L
Linus Torvalds 已提交
1498 1499 1500 1501 1502 1503 1504
		if (skb) {
			skb_set_owner_w(skb, sk);
			return skb;
		}
	}
	return NULL;
}
E
Eric Dumazet 已提交
1505
EXPORT_SYMBOL(sock_wmalloc);
L
Linus Torvalds 已提交
1506 1507 1508

/*
 * Allocate a skb from the socket's receive buffer.
1509
 */
V
Victor Fusco 已提交
1510
struct sk_buff *sock_rmalloc(struct sock *sk, unsigned long size, int force,
A
Al Viro 已提交
1511
			     gfp_t priority)
L
Linus Torvalds 已提交
1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522
{
	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;
}

1523
/*
L
Linus Torvalds 已提交
1524
 * Allocate a memory block from the socket's option memory buffer.
1525
 */
A
Al Viro 已提交
1526
void *sock_kmalloc(struct sock *sk, int size, gfp_t priority)
L
Linus Torvalds 已提交
1527
{
1528
	if ((unsigned int)size <= sysctl_optmem_max &&
L
Linus Torvalds 已提交
1529 1530 1531
	    atomic_read(&sk->sk_omem_alloc) + size < sysctl_optmem_max) {
		void *mem;
		/* First do the add, to avoid the race if kmalloc
1532
		 * might sleep.
L
Linus Torvalds 已提交
1533 1534 1535 1536 1537 1538 1539 1540 1541
		 */
		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
1542
EXPORT_SYMBOL(sock_kmalloc);
L
Linus Torvalds 已提交
1543 1544 1545 1546 1547 1548 1549 1550 1551

/*
 * 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
1552
EXPORT_SYMBOL(sock_kfree_s);
L
Linus Torvalds 已提交
1553 1554 1555 1556

/* 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
1557
static long sock_wait_for_wmem(struct sock *sk, long timeo)
L
Linus Torvalds 已提交
1558 1559 1560 1561 1562 1563 1564 1565 1566 1567
{
	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
1568
		prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
L
Linus Torvalds 已提交
1569 1570 1571 1572 1573 1574 1575 1576
		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
1577
	finish_wait(sk_sleep(sk), &wait);
L
Linus Torvalds 已提交
1578 1579 1580 1581 1582 1583 1584 1585
	return timeo;
}


/*
 *	Generic send/receive buffer handlers
 */

H
Herbert Xu 已提交
1586 1587 1588
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
1589 1590
{
	struct sk_buff *skb;
A
Al Viro 已提交
1591
	gfp_t gfp_mask;
L
Linus Torvalds 已提交
1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609
	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) {
1610
			skb = alloc_skb(header_len, gfp_mask);
L
Linus Torvalds 已提交
1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632
			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;
					}

1633 1634 1635 1636 1637
					__skb_fill_page_desc(skb, i,
							page, 0,
							(data_len >= PAGE_SIZE ?
							 PAGE_SIZE :
							 data_len));
L
Linus Torvalds 已提交
1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665
					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
1666
EXPORT_SYMBOL(sock_alloc_send_pskb);
L
Linus Torvalds 已提交
1667

1668
struct sk_buff *sock_alloc_send_skb(struct sock *sk, unsigned long size,
L
Linus Torvalds 已提交
1669 1670 1671 1672
				    int noblock, int *errcode)
{
	return sock_alloc_send_pskb(sk, size, 0, noblock, errcode);
}
E
Eric Dumazet 已提交
1673
EXPORT_SYMBOL(sock_alloc_send_skb);
L
Linus Torvalds 已提交
1674 1675

static void __lock_sock(struct sock *sk)
1676 1677
	__releases(&sk->sk_lock.slock)
	__acquires(&sk->sk_lock.slock)
L
Linus Torvalds 已提交
1678 1679 1680
{
	DEFINE_WAIT(wait);

1681
	for (;;) {
L
Linus Torvalds 已提交
1682 1683 1684 1685 1686
		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);
1687
		if (!sock_owned_by_user(sk))
L
Linus Torvalds 已提交
1688 1689 1690 1691 1692 1693
			break;
	}
	finish_wait(&sk->sk_lock.wq, &wait);
}

static void __release_sock(struct sock *sk)
1694 1695
	__releases(&sk->sk_lock.slock)
	__acquires(&sk->sk_lock.slock)
L
Linus Torvalds 已提交
1696 1697 1698 1699 1700 1701 1702 1703 1704 1705
{
	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;

1706
			prefetch(next);
E
Eric Dumazet 已提交
1707
			WARN_ON_ONCE(skb_dst_is_noref(skb));
L
Linus Torvalds 已提交
1708
			skb->next = NULL;
P
Peter Zijlstra 已提交
1709
			sk_backlog_rcv(sk, skb);
L
Linus Torvalds 已提交
1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722

			/*
			 * 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);
1723
	} while ((skb = sk->sk_backlog.head) != NULL);
Z
Zhu Yi 已提交
1724 1725 1726 1727 1728 1729

	/*
	 * 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
1730 1731 1732 1733
}

/**
 * sk_wait_data - wait for data to arrive at sk_receive_queue
1734 1735
 * @sk:    sock to wait on
 * @timeo: for how long
L
Linus Torvalds 已提交
1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746
 *
 * 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
1747
	prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
L
Linus Torvalds 已提交
1748 1749 1750
	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
1751
	finish_wait(sk_sleep(sk), &wait);
L
Linus Torvalds 已提交
1752 1753 1754 1755
	return rc;
}
EXPORT_SYMBOL(sk_wait_data);

1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769
/**
 *	__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
1770
	long allocated;
G
Glauber Costa 已提交
1771
	int parent_status = UNDER_LIMIT;
1772 1773

	sk->sk_forward_alloc += amt * SK_MEM_QUANTUM;
1774

G
Glauber Costa 已提交
1775
	allocated = sk_memory_allocated_add(sk, amt, &parent_status);
1776 1777

	/* Under limit. */
G
Glauber Costa 已提交
1778 1779
	if (parent_status == UNDER_LIMIT &&
			allocated <= sk_prot_mem_limits(sk, 0)) {
1780
		sk_leave_memory_pressure(sk);
1781 1782 1783
		return 1;
	}

G
Glauber Costa 已提交
1784 1785 1786
	/* Under pressure. (we or our parents) */
	if ((parent_status > SOFT_LIMIT) ||
			allocated > sk_prot_mem_limits(sk, 1))
1787
		sk_enter_memory_pressure(sk);
1788

G
Glauber Costa 已提交
1789 1790 1791
	/* Over hard limit (we or our parents) */
	if ((parent_status == OVER_LIMIT) ||
			(allocated > sk_prot_mem_limits(sk, 2)))
1792 1793 1794 1795 1796 1797
		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;
1798

1799 1800 1801 1802 1803 1804 1805 1806 1807
	} 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;
	}

1808
	if (sk_has_memory_pressure(sk)) {
1809 1810
		int alloc;

1811
		if (!sk_under_memory_pressure(sk))
1812
			return 1;
1813 1814
		alloc = sk_sockets_allocated_read_positive(sk);
		if (sk_prot_mem_limits(sk, 2) > alloc *
1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832
		    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;
	}

1833 1834
	trace_sock_exceed_buf_limit(sk, prot, allocated);

1835 1836
	/* Alas. Undo changes. */
	sk->sk_forward_alloc -= amt * SK_MEM_QUANTUM;
1837

1838
	sk_memory_allocated_sub(sk, amt);
1839

1840 1841 1842 1843 1844 1845 1846 1847 1848 1849
	return 0;
}
EXPORT_SYMBOL(__sk_mem_schedule);

/**
 *	__sk_reclaim - reclaim memory_allocated
 *	@sk: socket
 */
void __sk_mem_reclaim(struct sock *sk)
{
1850
	sk_memory_allocated_sub(sk,
1851
				sk->sk_forward_alloc >> SK_MEM_QUANTUM_SHIFT);
1852 1853
	sk->sk_forward_alloc &= SK_MEM_QUANTUM - 1;

1854 1855 1856
	if (sk_under_memory_pressure(sk) &&
	    (sk_memory_allocated(sk) < sk_prot_mem_limits(sk, 0)))
		sk_leave_memory_pressure(sk);
1857 1858 1859 1860
}
EXPORT_SYMBOL(__sk_mem_reclaim);


L
Linus Torvalds 已提交
1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871
/*
 * 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
1872
EXPORT_SYMBOL(sock_no_bind);
L
Linus Torvalds 已提交
1873

1874
int sock_no_connect(struct socket *sock, struct sockaddr *saddr,
L
Linus Torvalds 已提交
1875 1876 1877 1878
		    int len, int flags)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1879
EXPORT_SYMBOL(sock_no_connect);
L
Linus Torvalds 已提交
1880 1881 1882 1883 1884

int sock_no_socketpair(struct socket *sock1, struct socket *sock2)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1885
EXPORT_SYMBOL(sock_no_socketpair);
L
Linus Torvalds 已提交
1886 1887 1888 1889 1890

int sock_no_accept(struct socket *sock, struct socket *newsock, int flags)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1891
EXPORT_SYMBOL(sock_no_accept);
L
Linus Torvalds 已提交
1892

1893
int sock_no_getname(struct socket *sock, struct sockaddr *saddr,
L
Linus Torvalds 已提交
1894 1895 1896 1897
		    int *len, int peer)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1898
EXPORT_SYMBOL(sock_no_getname);
L
Linus Torvalds 已提交
1899

E
Eric Dumazet 已提交
1900
unsigned int sock_no_poll(struct file *file, struct socket *sock, poll_table *pt)
L
Linus Torvalds 已提交
1901 1902 1903
{
	return 0;
}
E
Eric Dumazet 已提交
1904
EXPORT_SYMBOL(sock_no_poll);
L
Linus Torvalds 已提交
1905 1906 1907 1908 1909

int sock_no_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1910
EXPORT_SYMBOL(sock_no_ioctl);
L
Linus Torvalds 已提交
1911 1912 1913 1914 1915

int sock_no_listen(struct socket *sock, int backlog)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1916
EXPORT_SYMBOL(sock_no_listen);
L
Linus Torvalds 已提交
1917 1918 1919 1920 1921

int sock_no_shutdown(struct socket *sock, int how)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1922
EXPORT_SYMBOL(sock_no_shutdown);
L
Linus Torvalds 已提交
1923 1924

int sock_no_setsockopt(struct socket *sock, int level, int optname,
1925
		    char __user *optval, unsigned int optlen)
L
Linus Torvalds 已提交
1926 1927 1928
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1929
EXPORT_SYMBOL(sock_no_setsockopt);
L
Linus Torvalds 已提交
1930 1931 1932 1933 1934 1935

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
1936
EXPORT_SYMBOL(sock_no_getsockopt);
L
Linus Torvalds 已提交
1937 1938 1939 1940 1941 1942

int sock_no_sendmsg(struct kiocb *iocb, struct socket *sock, struct msghdr *m,
		    size_t len)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
1943
EXPORT_SYMBOL(sock_no_sendmsg);
L
Linus Torvalds 已提交
1944 1945 1946 1947 1948 1949

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
1950
EXPORT_SYMBOL(sock_no_recvmsg);
L
Linus Torvalds 已提交
1951 1952 1953 1954 1955 1956

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
1957
EXPORT_SYMBOL(sock_no_mmap);
L
Linus Torvalds 已提交
1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970

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
1971
EXPORT_SYMBOL(sock_no_sendpage);
L
Linus Torvalds 已提交
1972 1973 1974 1975 1976 1977 1978

/*
 *	Default Socket Callbacks
 */

static void sock_def_wakeup(struct sock *sk)
{
1979 1980 1981 1982 1983 1984 1985
	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
1986 1987 1988 1989
}

static void sock_def_error_report(struct sock *sk)
{
1990 1991 1992 1993 1994 1995
	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);
1996
	sk_wake_async(sk, SOCK_WAKE_IO, POLL_ERR);
1997
	rcu_read_unlock();
L
Linus Torvalds 已提交
1998 1999 2000 2001
}

static void sock_def_readable(struct sock *sk, int len)
{
2002 2003 2004 2005 2006
	struct socket_wq *wq;

	rcu_read_lock();
	wq = rcu_dereference(sk->sk_wq);
	if (wq_has_sleeper(wq))
2007
		wake_up_interruptible_sync_poll(&wq->wait, POLLIN | POLLPRI |
2008
						POLLRDNORM | POLLRDBAND);
2009
	sk_wake_async(sk, SOCK_WAKE_WAITD, POLL_IN);
2010
	rcu_read_unlock();
L
Linus Torvalds 已提交
2011 2012 2013 2014
}

static void sock_def_write_space(struct sock *sk)
{
2015 2016 2017
	struct socket_wq *wq;

	rcu_read_lock();
L
Linus Torvalds 已提交
2018 2019 2020 2021

	/* Do not wake up a writer until he can make "significant"
	 * progress.  --DaveM
	 */
2022
	if ((atomic_read(&sk->sk_wmem_alloc) << 1) <= sk->sk_sndbuf) {
2023 2024 2025
		wq = rcu_dereference(sk->sk_wq);
		if (wq_has_sleeper(wq))
			wake_up_interruptible_sync_poll(&wq->wait, POLLOUT |
2026
						POLLWRNORM | POLLWRBAND);
L
Linus Torvalds 已提交
2027 2028 2029

		/* Should agree with poll, otherwise some programs break */
		if (sock_writeable(sk))
2030
			sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
L
Linus Torvalds 已提交
2031 2032
	}

2033
	rcu_read_unlock();
L
Linus Torvalds 已提交
2034 2035 2036 2037
}

static void sock_def_destruct(struct sock *sk)
{
J
Jesper Juhl 已提交
2038
	kfree(sk->sk_protinfo);
L
Linus Torvalds 已提交
2039 2040 2041 2042 2043 2044
}

void sk_send_sigurg(struct sock *sk)
{
	if (sk->sk_socket && sk->sk_socket->file)
		if (send_sigurg(&sk->sk_socket->file->f_owner))
2045
			sk_wake_async(sk, SOCK_WAKE_URG, POLL_PRI);
L
Linus Torvalds 已提交
2046
}
E
Eric Dumazet 已提交
2047
EXPORT_SYMBOL(sk_send_sigurg);
L
Linus Torvalds 已提交
2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068

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);
2069 2070 2071
#ifdef CONFIG_NET_DMA
	skb_queue_head_init(&sk->sk_async_wait_queue);
#endif
L
Linus Torvalds 已提交
2072 2073 2074 2075

	sk->sk_send_head	=	NULL;

	init_timer(&sk->sk_timer);
2076

L
Linus Torvalds 已提交
2077 2078 2079 2080
	sk->sk_allocation	=	GFP_KERNEL;
	sk->sk_rcvbuf		=	sysctl_rmem_default;
	sk->sk_sndbuf		=	sysctl_wmem_default;
	sk->sk_state		=	TCP_CLOSE;
2081
	sk_set_socket(sk, sock);
L
Linus Torvalds 已提交
2082 2083 2084

	sock_set_flag(sk, SOCK_ZAPPED);

2085
	if (sock) {
L
Linus Torvalds 已提交
2086
		sk->sk_type	=	sock->type;
2087
		sk->sk_wq	=	sock->wq;
L
Linus Torvalds 已提交
2088 2089
		sock->sk	=	sk;
	} else
2090
		sk->sk_wq	=	NULL;
L
Linus Torvalds 已提交
2091

E
Eric Dumazet 已提交
2092
	spin_lock_init(&sk->sk_dst_lock);
L
Linus Torvalds 已提交
2093
	rwlock_init(&sk->sk_callback_lock);
2094 2095 2096
	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
2097 2098 2099 2100 2101 2102 2103 2104 2105

	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;
2106
	sk->sk_peek_off		=	-1;
L
Linus Torvalds 已提交
2107

2108 2109
	sk->sk_peer_pid 	=	NULL;
	sk->sk_peer_cred	=	NULL;
L
Linus Torvalds 已提交
2110 2111 2112 2113 2114
	sk->sk_write_pending	=	0;
	sk->sk_rcvlowat		=	1;
	sk->sk_rcvtimeo		=	MAX_SCHEDULE_TIMEOUT;
	sk->sk_sndtimeo		=	MAX_SCHEDULE_TIMEOUT;

2115
	sk->sk_stamp = ktime_set(-1L, 0);
L
Linus Torvalds 已提交
2116

E
Eric Dumazet 已提交
2117 2118 2119 2120 2121
	/*
	 * 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
2122
	atomic_set(&sk->sk_refcnt, 1);
W
Wang Chen 已提交
2123
	atomic_set(&sk->sk_drops, 0);
L
Linus Torvalds 已提交
2124
}
E
Eric Dumazet 已提交
2125
EXPORT_SYMBOL(sock_init_data);
L
Linus Torvalds 已提交
2126

H
Harvey Harrison 已提交
2127
void lock_sock_nested(struct sock *sk, int subclass)
L
Linus Torvalds 已提交
2128 2129
{
	might_sleep();
I
Ingo Molnar 已提交
2130
	spin_lock_bh(&sk->sk_lock.slock);
2131
	if (sk->sk_lock.owned)
L
Linus Torvalds 已提交
2132
		__lock_sock(sk);
2133
	sk->sk_lock.owned = 1;
I
Ingo Molnar 已提交
2134 2135 2136 2137
	spin_unlock(&sk->sk_lock.slock);
	/*
	 * The sk_lock has mutex_lock() semantics here:
	 */
2138
	mutex_acquire(&sk->sk_lock.dep_map, subclass, 0, _RET_IP_);
I
Ingo Molnar 已提交
2139
	local_bh_enable();
L
Linus Torvalds 已提交
2140
}
2141
EXPORT_SYMBOL(lock_sock_nested);
L
Linus Torvalds 已提交
2142

H
Harvey Harrison 已提交
2143
void release_sock(struct sock *sk)
L
Linus Torvalds 已提交
2144
{
I
Ingo Molnar 已提交
2145 2146 2147 2148 2149 2150
	/*
	 * 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
2151 2152
	if (sk->sk_backlog.tail)
		__release_sock(sk);
2153
	sk->sk_lock.owned = 0;
I
Ingo Molnar 已提交
2154 2155 2156
	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
2157 2158 2159
}
EXPORT_SYMBOL(release_sock);

2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192
/**
 * 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
2193
int sock_get_timestamp(struct sock *sk, struct timeval __user *userstamp)
2194
{
2195
	struct timeval tv;
L
Linus Torvalds 已提交
2196
	if (!sock_flag(sk, SOCK_TIMESTAMP))
2197
		sock_enable_timestamp(sk, SOCK_TIMESTAMP);
2198 2199
	tv = ktime_to_timeval(sk->sk_stamp);
	if (tv.tv_sec == -1)
L
Linus Torvalds 已提交
2200
		return -ENOENT;
2201 2202 2203 2204 2205
	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;
2206
}
L
Linus Torvalds 已提交
2207 2208
EXPORT_SYMBOL(sock_get_timestamp);

2209 2210 2211 2212
int sock_get_timestampns(struct sock *sk, struct timespec __user *userstamp)
{
	struct timespec ts;
	if (!sock_flag(sk, SOCK_TIMESTAMP))
2213
		sock_enable_timestamp(sk, SOCK_TIMESTAMP);
2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224
	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);

2225
void sock_enable_timestamp(struct sock *sk, int flag)
2226
{
2227
	if (!sock_flag(sk, flag)) {
E
Eric Dumazet 已提交
2228 2229
		unsigned long previous_flags = sk->sk_flags;

2230 2231 2232 2233 2234 2235
		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
		 */
E
Eric Dumazet 已提交
2236
		if (!(previous_flags & SK_FLAGS_TIMESTAMP))
2237
			net_enable_timestamp();
L
Linus Torvalds 已提交
2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256
	}
}

/*
 *	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);

2257
#ifdef CONFIG_COMPAT
2258 2259
int compat_sock_common_getsockopt(struct socket *sock, int level, int optname,
				  char __user *optval, int __user *optlen)
2260 2261 2262
{
	struct sock *sk = sock->sk;

2263
	if (sk->sk_prot->compat_getsockopt != NULL)
2264 2265
		return sk->sk_prot->compat_getsockopt(sk, level, optname,
						      optval, optlen);
2266 2267 2268 2269 2270
	return sk->sk_prot->getsockopt(sk, level, optname, optval, optlen);
}
EXPORT_SYMBOL(compat_sock_common_getsockopt);
#endif

L
Linus Torvalds 已提交
2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289
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,
2290
			   char __user *optval, unsigned int optlen)
L
Linus Torvalds 已提交
2291 2292 2293 2294 2295 2296 2297
{
	struct sock *sk = sock->sk;

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

2298
#ifdef CONFIG_COMPAT
2299
int compat_sock_common_setsockopt(struct socket *sock, int level, int optname,
2300
				  char __user *optval, unsigned int optlen)
2301 2302 2303
{
	struct sock *sk = sock->sk;

2304 2305 2306
	if (sk->sk_prot->compat_setsockopt != NULL)
		return sk->sk_prot->compat_setsockopt(sk, level, optname,
						      optval, optlen);
2307 2308 2309 2310 2311
	return sk->sk_prot->setsockopt(sk, level, optname, optval, optlen);
}
EXPORT_SYMBOL(compat_sock_common_setsockopt);
#endif

L
Linus Torvalds 已提交
2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342
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);

2343
	sk_refcnt_debug_release(sk);
L
Linus Torvalds 已提交
2344 2345 2346 2347
	sock_put(sk);
}
EXPORT_SYMBOL(sk_common_release);

2348 2349
#ifdef CONFIG_PROC_FS
#define PROTO_INUSE_NR	64	/* should be enough for the first time */
2350 2351 2352
struct prot_inuse {
	int val[PROTO_INUSE_NR];
};
2353 2354

static DECLARE_BITMAP(proto_inuse_idx, PROTO_INUSE_NR);
2355 2356 2357 2358

#ifdef CONFIG_NET_NS
void sock_prot_inuse_add(struct net *net, struct proto *prot, int val)
{
E
Eric Dumazet 已提交
2359
	__this_cpu_add(net->core.inuse->val[prot->inuse_idx], val);
2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374
}
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);

2375
static int __net_init sock_inuse_init_net(struct net *net)
2376 2377 2378 2379 2380
{
	net->core.inuse = alloc_percpu(struct prot_inuse);
	return net->core.inuse ? 0 : -ENOMEM;
}

2381
static void __net_exit sock_inuse_exit_net(struct net *net)
2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400
{
	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
2401 2402
static DEFINE_PER_CPU(struct prot_inuse, prot_inuse);

2403
void sock_prot_inuse_add(struct net *net, struct proto *prot, int val)
2404
{
E
Eric Dumazet 已提交
2405
	__this_cpu_add(prot_inuse.val[prot->inuse_idx], val);
2406 2407 2408
}
EXPORT_SYMBOL_GPL(sock_prot_inuse_add);

2409
int sock_prot_inuse_get(struct net *net, struct proto *prot)
2410 2411 2412 2413 2414 2415 2416 2417 2418 2419
{
	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);
2420
#endif
2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448

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

2449 2450
int proto_register(struct proto *prot, int alloc_slab)
{
L
Linus Torvalds 已提交
2451 2452
	if (alloc_slab) {
		prot->slab = kmem_cache_create(prot->name, prot->obj_size, 0,
2453 2454
					SLAB_HWCACHE_ALIGN | prot->slab_flags,
					NULL);
L
Linus Torvalds 已提交
2455 2456 2457 2458

		if (prot->slab == NULL) {
			printk(KERN_CRIT "%s: Can't create sock SLAB cache!\n",
			       prot->name);
2459
			goto out;
L
Linus Torvalds 已提交
2460
		}
2461 2462

		if (prot->rsk_prot != NULL) {
2463
			prot->rsk_prot->slab_name = kasprintf(GFP_KERNEL, "request_sock_%s", prot->name);
2464
			if (prot->rsk_prot->slab_name == NULL)
2465 2466
				goto out_free_sock_slab;

2467
			prot->rsk_prot->slab = kmem_cache_create(prot->rsk_prot->slab_name,
2468
								 prot->rsk_prot->obj_size, 0,
2469
								 SLAB_HWCACHE_ALIGN, NULL);
2470 2471 2472 2473 2474 2475 2476

			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;
			}
		}
2477

2478
		if (prot->twsk_prot != NULL) {
2479
			prot->twsk_prot->twsk_slab_name = kasprintf(GFP_KERNEL, "tw_sock_%s", prot->name);
2480

2481
			if (prot->twsk_prot->twsk_slab_name == NULL)
2482 2483
				goto out_free_request_sock_slab;

2484
			prot->twsk_prot->twsk_slab =
2485
				kmem_cache_create(prot->twsk_prot->twsk_slab_name,
2486
						  prot->twsk_prot->twsk_obj_size,
2487 2488 2489
						  0,
						  SLAB_HWCACHE_ALIGN |
							prot->slab_flags,
2490
						  NULL);
2491
			if (prot->twsk_prot->twsk_slab == NULL)
2492 2493
				goto out_free_timewait_sock_slab_name;
		}
L
Linus Torvalds 已提交
2494 2495
	}

2496
	mutex_lock(&proto_list_mutex);
L
Linus Torvalds 已提交
2497
	list_add(&prot->node, &proto_list);
2498
	assign_proto_idx(prot);
2499
	mutex_unlock(&proto_list_mutex);
2500 2501
	return 0;

2502
out_free_timewait_sock_slab_name:
2503
	kfree(prot->twsk_prot->twsk_slab_name);
2504 2505 2506 2507 2508
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;
	}
2509
out_free_request_sock_slab_name:
2510 2511
	if (prot->rsk_prot)
		kfree(prot->rsk_prot->slab_name);
2512 2513 2514
out_free_sock_slab:
	kmem_cache_destroy(prot->slab);
	prot->slab = NULL;
2515 2516
out:
	return -ENOBUFS;
L
Linus Torvalds 已提交
2517 2518 2519 2520 2521
}
EXPORT_SYMBOL(proto_register);

void proto_unregister(struct proto *prot)
{
2522
	mutex_lock(&proto_list_mutex);
2523
	release_proto_idx(prot);
2524
	list_del(&prot->node);
2525
	mutex_unlock(&proto_list_mutex);
L
Linus Torvalds 已提交
2526 2527 2528 2529 2530 2531

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

2532 2533
	if (prot->rsk_prot != NULL && prot->rsk_prot->slab != NULL) {
		kmem_cache_destroy(prot->rsk_prot->slab);
2534
		kfree(prot->rsk_prot->slab_name);
2535 2536 2537
		prot->rsk_prot->slab = NULL;
	}

2538 2539
	if (prot->twsk_prot != NULL && prot->twsk_prot->twsk_slab != NULL) {
		kmem_cache_destroy(prot->twsk_prot->twsk_slab);
2540
		kfree(prot->twsk_prot->twsk_slab_name);
2541
		prot->twsk_prot->twsk_slab = NULL;
2542
	}
L
Linus Torvalds 已提交
2543 2544 2545 2546 2547
}
EXPORT_SYMBOL(proto_unregister);

#ifdef CONFIG_PROC_FS
static void *proto_seq_start(struct seq_file *seq, loff_t *pos)
2548
	__acquires(proto_list_mutex)
L
Linus Torvalds 已提交
2549
{
2550
	mutex_lock(&proto_list_mutex);
2551
	return seq_list_start_head(&proto_list, *pos);
L
Linus Torvalds 已提交
2552 2553 2554 2555
}

static void *proto_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
2556
	return seq_list_next(v, &proto_list, pos);
L
Linus Torvalds 已提交
2557 2558 2559
}

static void proto_seq_stop(struct seq_file *seq, void *v)
2560
	__releases(proto_list_mutex)
L
Linus Torvalds 已提交
2561
{
2562
	mutex_unlock(&proto_list_mutex);
L
Linus Torvalds 已提交
2563 2564 2565 2566 2567 2568
}

static char proto_method_implemented(const void *method)
{
	return method == NULL ? 'n' : 'y';
}
2569 2570
static long sock_prot_memory_allocated(struct proto *proto)
{
2571
	return proto->memory_allocated != NULL ? proto_memory_allocated(proto) : -1L;
2572 2573 2574 2575 2576 2577 2578
}

static char *sock_prot_memory_pressure(struct proto *proto)
{
	return proto->memory_pressure != NULL ?
	proto_memory_pressure(proto) ? "yes" : "no" : "NI";
}
L
Linus Torvalds 已提交
2579 2580 2581

static void proto_seq_printf(struct seq_file *seq, struct proto *proto)
{
2582

E
Eric Dumazet 已提交
2583
	seq_printf(seq, "%-9s %4u %6d  %6ld   %-3s %6u   %-3s  %-10s "
L
Linus Torvalds 已提交
2584 2585 2586
			"%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,
2587
		   sock_prot_inuse_get(seq_file_net(seq), proto),
2588 2589
		   sock_prot_memory_allocated(proto),
		   sock_prot_memory_pressure(proto),
L
Linus Torvalds 已提交
2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615
		   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)
{
2616
	if (v == &proto_list)
L
Linus Torvalds 已提交
2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627
		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
2628
		proto_seq_printf(seq, list_entry(v, struct proto, node));
L
Linus Torvalds 已提交
2629 2630 2631
	return 0;
}

2632
static const struct seq_operations proto_seq_ops = {
L
Linus Torvalds 已提交
2633 2634 2635 2636 2637 2638 2639 2640
	.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)
{
2641 2642
	return seq_open_net(inode, file, &proto_seq_ops,
			    sizeof(struct seq_net_private));
L
Linus Torvalds 已提交
2643 2644
}

2645
static const struct file_operations proto_seq_fops = {
L
Linus Torvalds 已提交
2646 2647 2648 2649
	.owner		= THIS_MODULE,
	.open		= proto_seq_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669
	.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
2670 2671 2672 2673
};

static int __init proto_init(void)
{
2674
	return register_pernet_subsys(&proto_net_ops);
L
Linus Torvalds 已提交
2675 2676 2677 2678 2679
}

subsys_initcall(proto_init);

#endif /* PROC_FS */