sock.c 80.7 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.
 */

J
Joe Perches 已提交
92 93
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

94
#include <linux/capability.h>
L
Linus Torvalds 已提交
95
#include <linux/errno.h>
96
#include <linux/errqueue.h>
L
Linus Torvalds 已提交
97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114
#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
115
#include <linux/highmem.h>
116
#include <linux/user_namespace.h>
117
#include <linux/static_key.h>
118
#include <linux/memcontrol.h>
119
#include <linux/prefetch.h>
L
Linus Torvalds 已提交
120

121
#include <linux/uaccess.h>
L
Linus Torvalds 已提交
122 123 124 125

#include <linux/netdevice.h>
#include <net/protocol.h>
#include <linux/skbuff.h>
126
#include <net/net_namespace.h>
127
#include <net/request_sock.h>
L
Linus Torvalds 已提交
128
#include <net/sock.h>
129
#include <linux/net_tstamp.h>
L
Linus Torvalds 已提交
130 131
#include <net/xfrm.h>
#include <linux/ipsec.h>
132
#include <net/cls_cgroup.h>
133
#include <net/netprio_cgroup.h>
134
#include <linux/sock_diag.h>
L
Linus Torvalds 已提交
135 136

#include <linux/filter.h>
137
#include <net/sock_reuseport.h>
L
Linus Torvalds 已提交
138

139 140
#include <trace/events/sock.h>

L
Linus Torvalds 已提交
141 142 143 144
#ifdef CONFIG_INET
#include <net/tcp.h>
#endif

145
#include <net/busy_poll.h>
E
Eliezer Tamir 已提交
146

147
static DEFINE_MUTEX(proto_list_mutex);
G
Glauber Costa 已提交
148 149
static LIST_HEAD(proto_list);

150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170
/**
 * sk_ns_capable - General socket capability test
 * @sk: Socket to use a capability on or through
 * @user_ns: The user namespace of the capability to use
 * @cap: The capability to use
 *
 * Test to see if the opener of the socket had when the socket was
 * created and the current process has the capability @cap in the user
 * namespace @user_ns.
 */
bool sk_ns_capable(const struct sock *sk,
		   struct user_namespace *user_ns, int cap)
{
	return file_ns_capable(sk->sk_socket->file, user_ns, cap) &&
		ns_capable(user_ns, cap);
}
EXPORT_SYMBOL(sk_ns_capable);

/**
 * sk_capable - Socket global capability test
 * @sk: Socket to use a capability on or through
171
 * @cap: The global capability to use
172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187
 *
 * Test to see if the opener of the socket had when the socket was
 * created and the current process has the capability @cap in all user
 * namespaces.
 */
bool sk_capable(const struct sock *sk, int cap)
{
	return sk_ns_capable(sk, &init_user_ns, cap);
}
EXPORT_SYMBOL(sk_capable);

/**
 * sk_net_capable - Network namespace socket capability test
 * @sk: Socket to use a capability on or through
 * @cap: The capability to use
 *
188
 * Test to see if the opener of the socket had when the socket was created
189 190 191 192 193 194 195 196 197
 * and the current process has the capability @cap over the network namespace
 * the socket is a member of.
 */
bool sk_net_capable(const struct sock *sk, int cap)
{
	return sk_ns_capable(sk, sock_net(sk)->user_ns, cap);
}
EXPORT_SYMBOL(sk_net_capable);

198 199
/*
 * Each address family might have different locking rules, so we have
200 201
 * one slock key per address family and separate keys for internal and
 * userspace sockets.
202
 */
I
Ingo Molnar 已提交
203
static struct lock_class_key af_family_keys[AF_MAX];
204
static struct lock_class_key af_family_kern_keys[AF_MAX];
I
Ingo Molnar 已提交
205
static struct lock_class_key af_family_slock_keys[AF_MAX];
206
static struct lock_class_key af_family_kern_slock_keys[AF_MAX];
I
Ingo Molnar 已提交
207 208 209 210 211 212

/*
 * Make lock validator output more readable. (we pre-construct these
 * strings build-time, so that runtime initialization of socket
 * locks is fast):
 */
213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230

#define _sock_locks(x)						  \
  x "AF_UNSPEC",	x "AF_UNIX"     ,	x "AF_INET"     , \
  x "AF_AX25"  ,	x "AF_IPX"      ,	x "AF_APPLETALK", \
  x "AF_NETROM",	x "AF_BRIDGE"   ,	x "AF_ATMPVC"   , \
  x "AF_X25"   ,	x "AF_INET6"    ,	x "AF_ROSE"     , \
  x "AF_DECnet",	x "AF_NETBEUI"  ,	x "AF_SECURITY" , \
  x "AF_KEY"   ,	x "AF_NETLINK"  ,	x "AF_PACKET"   , \
  x "AF_ASH"   ,	x "AF_ECONET"   ,	x "AF_ATMSVC"   , \
  x "AF_RDS"   ,	x "AF_SNA"      ,	x "AF_IRDA"     , \
  x "AF_PPPOX" ,	x "AF_WANPIPE"  ,	x "AF_LLC"      , \
  x "27"       ,	x "28"          ,	x "AF_CAN"      , \
  x "AF_TIPC"  ,	x "AF_BLUETOOTH",	x "IUCV"        , \
  x "AF_RXRPC" ,	x "AF_ISDN"     ,	x "AF_PHONET"   , \
  x "AF_IEEE802154",	x "AF_CAIF"	,	x "AF_ALG"      , \
  x "AF_NFC"   ,	x "AF_VSOCK"    ,	x "AF_KCM"      , \
  x "AF_QIPCRTR",	x "AF_SMC"	,	x "AF_MAX"

231
static const char *const af_family_key_strings[AF_MAX+1] = {
232
	_sock_locks("sk_lock-")
I
Ingo Molnar 已提交
233
};
234
static const char *const af_family_slock_key_strings[AF_MAX+1] = {
235
	_sock_locks("slock-")
I
Ingo Molnar 已提交
236
};
237
static const char *const af_family_clock_key_strings[AF_MAX+1] = {
238 239 240 241 242 243 244 245 246 247 248
	_sock_locks("clock-")
};

static const char *const af_family_kern_key_strings[AF_MAX+1] = {
	_sock_locks("k-sk_lock-")
};
static const char *const af_family_kern_slock_key_strings[AF_MAX+1] = {
	_sock_locks("k-slock-")
};
static const char *const af_family_kern_clock_key_strings[AF_MAX+1] = {
	_sock_locks("k-clock-")
249
};
250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300
static const char *const af_family_rlock_key_strings[AF_MAX+1] = {
  "rlock-AF_UNSPEC", "rlock-AF_UNIX"     , "rlock-AF_INET"     ,
  "rlock-AF_AX25"  , "rlock-AF_IPX"      , "rlock-AF_APPLETALK",
  "rlock-AF_NETROM", "rlock-AF_BRIDGE"   , "rlock-AF_ATMPVC"   ,
  "rlock-AF_X25"   , "rlock-AF_INET6"    , "rlock-AF_ROSE"     ,
  "rlock-AF_DECnet", "rlock-AF_NETBEUI"  , "rlock-AF_SECURITY" ,
  "rlock-AF_KEY"   , "rlock-AF_NETLINK"  , "rlock-AF_PACKET"   ,
  "rlock-AF_ASH"   , "rlock-AF_ECONET"   , "rlock-AF_ATMSVC"   ,
  "rlock-AF_RDS"   , "rlock-AF_SNA"      , "rlock-AF_IRDA"     ,
  "rlock-AF_PPPOX" , "rlock-AF_WANPIPE"  , "rlock-AF_LLC"      ,
  "rlock-27"       , "rlock-28"          , "rlock-AF_CAN"      ,
  "rlock-AF_TIPC"  , "rlock-AF_BLUETOOTH", "rlock-AF_IUCV"     ,
  "rlock-AF_RXRPC" , "rlock-AF_ISDN"     , "rlock-AF_PHONET"   ,
  "rlock-AF_IEEE802154", "rlock-AF_CAIF" , "rlock-AF_ALG"      ,
  "rlock-AF_NFC"   , "rlock-AF_VSOCK"    , "rlock-AF_KCM"      ,
  "rlock-AF_QIPCRTR", "rlock-AF_SMC"     , "rlock-AF_MAX"
};
static const char *const af_family_wlock_key_strings[AF_MAX+1] = {
  "wlock-AF_UNSPEC", "wlock-AF_UNIX"     , "wlock-AF_INET"     ,
  "wlock-AF_AX25"  , "wlock-AF_IPX"      , "wlock-AF_APPLETALK",
  "wlock-AF_NETROM", "wlock-AF_BRIDGE"   , "wlock-AF_ATMPVC"   ,
  "wlock-AF_X25"   , "wlock-AF_INET6"    , "wlock-AF_ROSE"     ,
  "wlock-AF_DECnet", "wlock-AF_NETBEUI"  , "wlock-AF_SECURITY" ,
  "wlock-AF_KEY"   , "wlock-AF_NETLINK"  , "wlock-AF_PACKET"   ,
  "wlock-AF_ASH"   , "wlock-AF_ECONET"   , "wlock-AF_ATMSVC"   ,
  "wlock-AF_RDS"   , "wlock-AF_SNA"      , "wlock-AF_IRDA"     ,
  "wlock-AF_PPPOX" , "wlock-AF_WANPIPE"  , "wlock-AF_LLC"      ,
  "wlock-27"       , "wlock-28"          , "wlock-AF_CAN"      ,
  "wlock-AF_TIPC"  , "wlock-AF_BLUETOOTH", "wlock-AF_IUCV"     ,
  "wlock-AF_RXRPC" , "wlock-AF_ISDN"     , "wlock-AF_PHONET"   ,
  "wlock-AF_IEEE802154", "wlock-AF_CAIF" , "wlock-AF_ALG"      ,
  "wlock-AF_NFC"   , "wlock-AF_VSOCK"    , "wlock-AF_KCM"      ,
  "wlock-AF_QIPCRTR", "wlock-AF_SMC"     , "wlock-AF_MAX"
};
static const char *const af_family_elock_key_strings[AF_MAX+1] = {
  "elock-AF_UNSPEC", "elock-AF_UNIX"     , "elock-AF_INET"     ,
  "elock-AF_AX25"  , "elock-AF_IPX"      , "elock-AF_APPLETALK",
  "elock-AF_NETROM", "elock-AF_BRIDGE"   , "elock-AF_ATMPVC"   ,
  "elock-AF_X25"   , "elock-AF_INET6"    , "elock-AF_ROSE"     ,
  "elock-AF_DECnet", "elock-AF_NETBEUI"  , "elock-AF_SECURITY" ,
  "elock-AF_KEY"   , "elock-AF_NETLINK"  , "elock-AF_PACKET"   ,
  "elock-AF_ASH"   , "elock-AF_ECONET"   , "elock-AF_ATMSVC"   ,
  "elock-AF_RDS"   , "elock-AF_SNA"      , "elock-AF_IRDA"     ,
  "elock-AF_PPPOX" , "elock-AF_WANPIPE"  , "elock-AF_LLC"      ,
  "elock-27"       , "elock-28"          , "elock-AF_CAN"      ,
  "elock-AF_TIPC"  , "elock-AF_BLUETOOTH", "elock-AF_IUCV"     ,
  "elock-AF_RXRPC" , "elock-AF_ISDN"     , "elock-AF_PHONET"   ,
  "elock-AF_IEEE802154", "elock-AF_CAIF" , "elock-AF_ALG"      ,
  "elock-AF_NFC"   , "elock-AF_VSOCK"    , "elock-AF_KCM"      ,
  "elock-AF_QIPCRTR", "elock-AF_SMC"     , "elock-AF_MAX"
};
301 302

/*
303
 * sk_callback_lock and sk queues locking rules are per-address-family,
304 305 306
 * so split the lock classes by using a per-AF key:
 */
static struct lock_class_key af_callback_keys[AF_MAX];
307 308 309
static struct lock_class_key af_rlock_keys[AF_MAX];
static struct lock_class_key af_wlock_keys[AF_MAX];
static struct lock_class_key af_elock_keys[AF_MAX];
310
static struct lock_class_key af_kern_callback_keys[AF_MAX];
311

L
Linus Torvalds 已提交
312 313 314 315 316 317
/* 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
318
#define _SK_MEM_OVERHEAD	SKB_TRUESIZE(256)
L
Linus Torvalds 已提交
319 320 321 322
#define SK_WMEM_MAX		(_SK_MEM_OVERHEAD * _SK_MEM_PACKETS)
#define SK_RMEM_MAX		(_SK_MEM_OVERHEAD * _SK_MEM_PACKETS)

/* Run time adjustable parameters. */
323
__u32 sysctl_wmem_max __read_mostly = SK_WMEM_MAX;
324
EXPORT_SYMBOL(sysctl_wmem_max);
325
__u32 sysctl_rmem_max __read_mostly = SK_RMEM_MAX;
326
EXPORT_SYMBOL(sysctl_rmem_max);
327 328
__u32 sysctl_wmem_default __read_mostly = SK_WMEM_MAX;
__u32 sysctl_rmem_default __read_mostly = SK_RMEM_MAX;
L
Linus Torvalds 已提交
329

L
Lucas De Marchi 已提交
330
/* Maximal space eaten by iovec or ancillary data plus some space */
331
int sysctl_optmem_max __read_mostly = sizeof(unsigned long)*(2*UIO_MAXIOV+512);
E
Eric Dumazet 已提交
332
EXPORT_SYMBOL(sysctl_optmem_max);
L
Linus Torvalds 已提交
333

334 335
int sysctl_tstamp_allow_data __read_mostly = 1;

336 337 338
struct static_key memalloc_socks = STATIC_KEY_INIT_FALSE;
EXPORT_SYMBOL_GPL(memalloc_socks);

339 340 341 342 343 344 345 346 347 348 349 350
/**
 * sk_set_memalloc - sets %SOCK_MEMALLOC
 * @sk: socket to set it on
 *
 * Set %SOCK_MEMALLOC on a socket for access to emergency reserves.
 * It's the responsibility of the admin to adjust min_free_kbytes
 * to meet the requirements
 */
void sk_set_memalloc(struct sock *sk)
{
	sock_set_flag(sk, SOCK_MEMALLOC);
	sk->sk_allocation |= __GFP_MEMALLOC;
351
	static_key_slow_inc(&memalloc_socks);
352 353 354 355 356 357 358
}
EXPORT_SYMBOL_GPL(sk_set_memalloc);

void sk_clear_memalloc(struct sock *sk)
{
	sock_reset_flag(sk, SOCK_MEMALLOC);
	sk->sk_allocation &= ~__GFP_MEMALLOC;
359
	static_key_slow_dec(&memalloc_socks);
360 361 362

	/*
	 * SOCK_MEMALLOC is allowed to ignore rmem limits to ensure forward
363 364 365 366
	 * progress of swapping. SOCK_MEMALLOC may be cleared while
	 * it has rmem allocations due to the last swapfile being deactivated
	 * but there is a risk that the socket is unusable due to exceeding
	 * the rmem limits. Reclaim the reserves and obey rmem limits again.
367
	 */
368
	sk_mem_reclaim(sk);
369 370 371
}
EXPORT_SYMBOL_GPL(sk_clear_memalloc);

372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387
int __sk_backlog_rcv(struct sock *sk, struct sk_buff *skb)
{
	int ret;
	unsigned long pflags = current->flags;

	/* these should have been dropped before queueing */
	BUG_ON(!sock_flag(sk, SOCK_MEMALLOC));

	current->flags |= PF_MEMALLOC;
	ret = sk->sk_backlog_rcv(sk, skb);
	tsk_restore_flags(current, pflags, PF_MEMALLOC);

	return ret;
}
EXPORT_SYMBOL(__sk_backlog_rcv);

L
Linus Torvalds 已提交
388 389 390 391 392 393 394 395
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;
396 397
	if (tv.tv_usec < 0 || tv.tv_usec >= USEC_PER_SEC)
		return -EDOM;
L
Linus Torvalds 已提交
398

399
	if (tv.tv_sec < 0) {
400 401
		static int warned __read_mostly;

402
		*timeo_p = 0;
403
		if (warned < 10 && net_ratelimit()) {
404
			warned++;
J
Joe Perches 已提交
405 406
			pr_info("%s: `%s' (pid %d) tries to set negative timeout\n",
				__func__, current->comm, task_pid_nr(current));
407
		}
408 409
		return 0;
	}
L
Linus Torvalds 已提交
410 411 412 413
	*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))
414
		*timeo_p = tv.tv_sec * HZ + DIV_ROUND_UP(tv.tv_usec, USEC_PER_SEC / HZ);
L
Linus Torvalds 已提交
415 416 417 418 419 420 421
	return 0;
}

static void sock_warn_obsolete_bsdism(const char *name)
{
	static int warned;
	static char warncomm[TASK_COMM_LEN];
422 423
	if (strcmp(warncomm, current->comm) && warned < 5) {
		strcpy(warncomm,  current->comm);
J
Joe Perches 已提交
424 425
		pr_warn("process `%s' is using obsolete %s SO_BSDCOMPAT\n",
			warncomm, name);
L
Linus Torvalds 已提交
426 427 428 429
		warned++;
	}
}

430 431 432 433 434 435 436 437 438 439 440
static bool sock_needs_netstamp(const struct sock *sk)
{
	switch (sk->sk_family) {
	case AF_UNSPEC:
	case AF_UNIX:
		return false;
	default:
		return true;
	}
}

E
Eric Dumazet 已提交
441
static void sock_disable_timestamp(struct sock *sk, unsigned long flags)
442
{
E
Eric Dumazet 已提交
443 444
	if (sk->sk_flags & flags) {
		sk->sk_flags &= ~flags;
445 446
		if (sock_needs_netstamp(sk) &&
		    !(sk->sk_flags & SK_FLAGS_TIMESTAMP))
447
			net_disable_timestamp();
L
Linus Torvalds 已提交
448 449 450 451
	}
}


452
int __sock_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
453
{
454 455
	unsigned long flags;
	struct sk_buff_head *list = &sk->sk_receive_queue;
456

E
Eric Dumazet 已提交
457
	if (atomic_read(&sk->sk_rmem_alloc) >= sk->sk_rcvbuf) {
E
Eric Dumazet 已提交
458
		atomic_inc(&sk->sk_drops);
459
		trace_sock_rcvqueue_full(sk, skb);
E
Eric Dumazet 已提交
460
		return -ENOMEM;
461 462
	}

463
	if (!sk_rmem_schedule(sk, skb, skb->truesize)) {
E
Eric Dumazet 已提交
464 465
		atomic_inc(&sk->sk_drops);
		return -ENOBUFS;
466 467
	}

468 469
	skb->dev = NULL;
	skb_set_owner_r(skb, sk);
470

E
Eric Dumazet 已提交
471 472 473 474 475
	/* we escape from rcu protected region, make sure we dont leak
	 * a norefcounted dst
	 */
	skb_dst_force(skb);

476
	spin_lock_irqsave(&list->lock, flags);
477
	sock_skb_set_dropcount(sk, skb);
478 479
	__skb_queue_tail(list, skb);
	spin_unlock_irqrestore(&list->lock, flags);
480 481

	if (!sock_flag(sk, SOCK_DEAD))
482
		sk->sk_data_ready(sk);
E
Eric Dumazet 已提交
483
	return 0;
484
}
485 486 487 488 489 490 491 492 493 494 495 496
EXPORT_SYMBOL(__sock_queue_rcv_skb);

int sock_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
{
	int err;

	err = sk_filter(sk, skb);
	if (err)
		return err;

	return __sock_queue_rcv_skb(sk, skb);
}
497 498
EXPORT_SYMBOL(sock_queue_rcv_skb);

499
int __sk_receive_skb(struct sock *sk, struct sk_buff *skb,
500
		     const int nested, unsigned int trim_cap, bool refcounted)
501 502 503
{
	int rc = NET_RX_SUCCESS;

504
	if (sk_filter_trim_cap(sk, skb, trim_cap))
505 506 507 508
		goto discard_and_relse;

	skb->dev = NULL;

509
	if (sk_rcvqueues_full(sk, sk->sk_rcvbuf)) {
510 511 512
		atomic_inc(&sk->sk_drops);
		goto discard_and_relse;
	}
513 514 515 516
	if (nested)
		bh_lock_sock_nested(sk);
	else
		bh_lock_sock(sk);
I
Ingo Molnar 已提交
517 518 519 520 521 522
	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
523
		rc = sk_backlog_rcv(sk, skb);
I
Ingo Molnar 已提交
524 525

		mutex_release(&sk->sk_lock.dep_map, 1, _RET_IP_);
526
	} else if (sk_add_backlog(sk, skb, sk->sk_rcvbuf)) {
Z
Zhu Yi 已提交
527 528 529 530 531
		bh_unlock_sock(sk);
		atomic_inc(&sk->sk_drops);
		goto discard_and_relse;
	}

532 533
	bh_unlock_sock(sk);
out:
534 535
	if (refcounted)
		sock_put(sk);
536 537 538 539 540
	return rc;
discard_and_relse:
	kfree_skb(skb);
	goto out;
}
541
EXPORT_SYMBOL(__sk_receive_skb);
542 543 544

struct dst_entry *__sk_dst_check(struct sock *sk, u32 cookie)
{
E
Eric Dumazet 已提交
545
	struct dst_entry *dst = __sk_dst_get(sk);
546 547

	if (dst && dst->obsolete && dst->ops->check(dst, cookie) == NULL) {
K
Krishna Kumar 已提交
548
		sk_tx_queue_clear(sk);
549
		sk->sk_dst_pending_confirm = 0;
550
		RCU_INIT_POINTER(sk->sk_dst_cache, NULL);
551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572
		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);

573 574
static int sock_setbindtodevice(struct sock *sk, char __user *optval,
				int optlen)
575 576 577
{
	int ret = -ENOPROTOOPT;
#ifdef CONFIG_NETDEVICES
578
	struct net *net = sock_net(sk);
579 580 581 582 583
	char devname[IFNAMSIZ];
	int index;

	/* Sorry... */
	ret = -EPERM;
584
	if (!ns_capable(net->user_ns, CAP_NET_RAW))
585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603
		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;

604 605
	index = 0;
	if (devname[0] != '\0') {
606
		struct net_device *dev;
607

608 609 610 611 612
		rcu_read_lock();
		dev = dev_get_by_name_rcu(net, devname);
		if (dev)
			index = dev->ifindex;
		rcu_read_unlock();
613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630
		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;
}

631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647
static int sock_getbindtodevice(struct sock *sk, char __user *optval,
				int __user *optlen, int len)
{
	int ret = -ENOPROTOOPT;
#ifdef CONFIG_NETDEVICES
	struct net *net = sock_net(sk);
	char devname[IFNAMSIZ];

	if (sk->sk_bound_dev_if == 0) {
		len = 0;
		goto zero;
	}

	ret = -EINVAL;
	if (len < IFNAMSIZ)
		goto out;

648 649
	ret = netdev_get_name(net, devname, sk->sk_bound_dev_if);
	if (ret)
650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670
		goto out;

	len = strlen(devname) + 1;

	ret = -EFAULT;
	if (copy_to_user(optval, devname, len))
		goto out;

zero:
	ret = -EFAULT;
	if (put_user(len, optlen))
		goto out;

	ret = 0;

out:
#endif

	return ret;
}

671 672 673 674 675 676 677 678
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);
}

679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697
bool sk_mc_loop(struct sock *sk)
{
	if (dev_recursion_level())
		return false;
	if (!sk)
		return true;
	switch (sk->sk_family) {
	case AF_INET:
		return inet_sk(sk)->mc_loop;
#if IS_ENABLED(CONFIG_IPV6)
	case AF_INET6:
		return inet6_sk(sk)->mc_loop;
#endif
	}
	WARN_ON(1);
	return true;
}
EXPORT_SYMBOL(sk_mc_loop);

L
Linus Torvalds 已提交
698 699 700 701 702 703
/*
 *	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,
704
		    char __user *optval, unsigned int optlen)
L
Linus Torvalds 已提交
705
{
E
Eric Dumazet 已提交
706
	struct sock *sk = sock->sk;
L
Linus Torvalds 已提交
707 708 709 710
	int val;
	int valbool;
	struct linger ling;
	int ret = 0;
711

L
Linus Torvalds 已提交
712 713 714 715
	/*
	 *	Options without arguments
	 */

716
	if (optname == SO_BINDTODEVICE)
717
		return sock_setbindtodevice(sk, optval, optlen);
718

719 720
	if (optlen < sizeof(int))
		return -EINVAL;
721

L
Linus Torvalds 已提交
722 723
	if (get_user(val, (int __user *)optval))
		return -EFAULT;
724

E
Eric Dumazet 已提交
725
	valbool = val ? 1 : 0;
L
Linus Torvalds 已提交
726 727 728

	lock_sock(sk);

E
Eric Dumazet 已提交
729
	switch (optname) {
730
	case SO_DEBUG:
E
Eric Dumazet 已提交
731
		if (val && !capable(CAP_NET_ADMIN))
732
			ret = -EACCES;
E
Eric Dumazet 已提交
733
		else
734
			sock_valbool_flag(sk, SOCK_DBG, valbool);
735 736
		break;
	case SO_REUSEADDR:
737
		sk->sk_reuse = (valbool ? SK_CAN_REUSE : SK_NO_REUSE);
738
		break;
T
Tom Herbert 已提交
739 740 741
	case SO_REUSEPORT:
		sk->sk_reuseport = valbool;
		break;
742
	case SO_TYPE:
743
	case SO_PROTOCOL:
744
	case SO_DOMAIN:
745 746 747 748
	case SO_ERROR:
		ret = -ENOPROTOOPT;
		break;
	case SO_DONTROUTE:
749
		sock_valbool_flag(sk, SOCK_LOCALROUTE, valbool);
750 751 752 753 754 755
		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
756 757 758 759 760
		 * 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);
761
set_sndbuf:
762
		sk->sk_userlocks |= SOCK_SNDBUF_LOCK;
763
		sk->sk_sndbuf = max_t(int, val * 2, SOCK_MIN_SNDBUF);
764
		/* Wake up sending tasks if we upped the value. */
765 766
		sk->sk_write_space(sk);
		break;
L
Linus Torvalds 已提交
767

768 769 770 771 772 773
	case SO_SNDBUFFORCE:
		if (!capable(CAP_NET_ADMIN)) {
			ret = -EPERM;
			break;
		}
		goto set_sndbuf;
774

775 776
	case SO_RCVBUF:
		/* Don't error on this BSD doesn't and if you think
777 778 779 780 781
		 * 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);
782
set_rcvbuf:
783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798
		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.
		 */
799
		sk->sk_rcvbuf = max_t(int, val * 2, SOCK_MIN_RCVBUF);
800 801 802 803 804
		break;

	case SO_RCVBUFFORCE:
		if (!capable(CAP_NET_ADMIN)) {
			ret = -EPERM;
L
Linus Torvalds 已提交
805
			break;
806 807
		}
		goto set_rcvbuf;
L
Linus Torvalds 已提交
808

809
	case SO_KEEPALIVE:
810 811
		if (sk->sk_prot->keepalive)
			sk->sk_prot->keepalive(sk, valbool);
812 813 814 815 816 817 818 819
		sock_valbool_flag(sk, SOCK_KEEPOPEN, valbool);
		break;

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

	case SO_NO_CHECK:
820
		sk->sk_no_check_tx = valbool;
821 822 823
		break;

	case SO_PRIORITY:
824 825
		if ((val >= 0 && val <= 6) ||
		    ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
826 827 828 829 830 831 832 833
			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
834
			break;
835
		}
E
Eric Dumazet 已提交
836
		if (copy_from_user(&ling, optval, sizeof(ling))) {
837
			ret = -EFAULT;
L
Linus Torvalds 已提交
838
			break;
839 840 841 842
		}
		if (!ling.l_onoff)
			sock_reset_flag(sk, SOCK_LINGER);
		else {
L
Linus Torvalds 已提交
843
#if (BITS_PER_LONG == 32)
844 845
			if ((unsigned int)ling.l_linger >= MAX_SCHEDULE_TIMEOUT/HZ)
				sk->sk_lingertime = MAX_SCHEDULE_TIMEOUT;
L
Linus Torvalds 已提交
846
			else
847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864
#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:
865
	case SO_TIMESTAMPNS:
866
		if (valbool)  {
867 868 869 870
			if (optname == SO_TIMESTAMP)
				sock_reset_flag(sk, SOCK_RCVTSTAMPNS);
			else
				sock_set_flag(sk, SOCK_RCVTSTAMPNS);
871
			sock_set_flag(sk, SOCK_RCVTSTAMP);
872
			sock_enable_timestamp(sk, SOCK_TIMESTAMP);
873
		} else {
874
			sock_reset_flag(sk, SOCK_RCVTSTAMP);
875 876
			sock_reset_flag(sk, SOCK_RCVTSTAMPNS);
		}
877 878
		break;

879 880
	case SO_TIMESTAMPING:
		if (val & ~SOF_TIMESTAMPING_MASK) {
881
			ret = -EINVAL;
882 883
			break;
		}
884

885
		if (val & SOF_TIMESTAMPING_OPT_ID &&
886
		    !(sk->sk_tsflags & SOF_TIMESTAMPING_OPT_ID)) {
887 888
			if (sk->sk_protocol == IPPROTO_TCP &&
			    sk->sk_type == SOCK_STREAM) {
889 890
				if ((1 << sk->sk_state) &
				    (TCPF_CLOSE | TCPF_LISTEN)) {
891 892 893 894 895 896 897 898
					ret = -EINVAL;
					break;
				}
				sk->sk_tskey = tcp_sk(sk)->snd_una;
			} else {
				sk->sk_tskey = 0;
			}
		}
899 900 901 902 903 904 905

		if (val & SOF_TIMESTAMPING_OPT_STATS &&
		    !(val & SOF_TIMESTAMPING_OPT_TSONLY)) {
			ret = -EINVAL;
			break;
		}

906
		sk->sk_tsflags = val;
907 908 909 910 911
		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
912
					       (1UL << SOCK_TIMESTAMPING_RX_SOFTWARE));
913 914
		break;

915 916 917 918 919 920 921 922 923 924 925 926 927
	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
928

929 930 931 932
	case SO_ATTACH_FILTER:
		ret = -EINVAL;
		if (optlen == sizeof(struct sock_fprog)) {
			struct sock_fprog fprog;
L
Linus Torvalds 已提交
933

934 935
			ret = -EFAULT;
			if (copy_from_user(&fprog, optval, sizeof(fprog)))
L
Linus Torvalds 已提交
936
				break;
937 938 939 940 941

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

942 943 944 945 946 947 948 949 950 951 952 953 954
	case SO_ATTACH_BPF:
		ret = -EINVAL;
		if (optlen == sizeof(u32)) {
			u32 ufd;

			ret = -EFAULT;
			if (copy_from_user(&ufd, optval, sizeof(ufd)))
				break;

			ret = sk_attach_bpf(ufd, sk);
		}
		break;

955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980
	case SO_ATTACH_REUSEPORT_CBPF:
		ret = -EINVAL;
		if (optlen == sizeof(struct sock_fprog)) {
			struct sock_fprog fprog;

			ret = -EFAULT;
			if (copy_from_user(&fprog, optval, sizeof(fprog)))
				break;

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

	case SO_ATTACH_REUSEPORT_EBPF:
		ret = -EINVAL;
		if (optlen == sizeof(u32)) {
			u32 ufd;

			ret = -EFAULT;
			if (copy_from_user(&ufd, optval, sizeof(ufd)))
				break;

			ret = sk_reuseport_attach_bpf(ufd, sk);
		}
		break;

981
	case SO_DETACH_FILTER:
982
		ret = sk_detach_filter(sk);
983
		break;
L
Linus Torvalds 已提交
984

985 986 987 988 989 990 991
	case SO_LOCK_FILTER:
		if (sock_flag(sk, SOCK_FILTER_LOCKED) && !valbool)
			ret = -EPERM;
		else
			sock_valbool_flag(sk, SOCK_FILTER_LOCKED, valbool);
		break;

992 993 994 995 996 997
	case SO_PASSSEC:
		if (valbool)
			set_bit(SOCK_PASSSEC, &sock->flags);
		else
			clear_bit(SOCK_PASSSEC, &sock->flags);
		break;
998
	case SO_MARK:
999
		if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
1000
			ret = -EPERM;
E
Eric Dumazet 已提交
1001
		else
1002 1003
			sk->sk_mark = val;
		break;
C
Catherine Zhang 已提交
1004

1005
	case SO_RXQ_OVFL:
1006
		sock_valbool_flag(sk, SOCK_RXQ_OVFL, valbool);
1007
		break;
1008 1009 1010 1011 1012

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

1013 1014
	case SO_PEEK_OFF:
		if (sock->ops->set_peek_off)
1015
			ret = sock->ops->set_peek_off(sk, val);
1016 1017 1018
		else
			ret = -EOPNOTSUPP;
		break;
1019 1020 1021 1022 1023

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

1024 1025 1026 1027
	case SO_SELECT_ERR_QUEUE:
		sock_valbool_flag(sk, SOCK_SELECT_ERR_QUEUE, valbool);
		break;

1028
#ifdef CONFIG_NET_RX_BUSY_POLL
1029
	case SO_BUSY_POLL:
1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040
		/* allow unprivileged users to decrease the value */
		if ((val > sk->sk_ll_usec) && !capable(CAP_NET_ADMIN))
			ret = -EPERM;
		else {
			if (val < 0)
				ret = -EINVAL;
			else
				sk->sk_ll_usec = val;
		}
		break;
#endif
E
Eric Dumazet 已提交
1041 1042 1043 1044 1045 1046 1047

	case SO_MAX_PACING_RATE:
		sk->sk_max_pacing_rate = val;
		sk->sk_pacing_rate = min(sk->sk_pacing_rate,
					 sk->sk_max_pacing_rate);
		break;

1048 1049 1050 1051
	case SO_INCOMING_CPU:
		sk->sk_incoming_cpu = val;
		break;

1052 1053 1054 1055
	case SO_CNX_ADVICE:
		if (val == 1)
			dst_negative_advice(sk);
		break;
1056 1057 1058
	default:
		ret = -ENOPROTOOPT;
		break;
1059
	}
L
Linus Torvalds 已提交
1060 1061 1062
	release_sock(sk);
	return ret;
}
E
Eric Dumazet 已提交
1063
EXPORT_SYMBOL(sock_setsockopt);
L
Linus Torvalds 已提交
1064 1065


S
stephen hemminger 已提交
1066 1067
static void cred_to_ucred(struct pid *pid, const struct cred *cred,
			  struct ucred *ucred)
1068 1069 1070 1071 1072 1073
{
	ucred->pid = pid_vnr(pid);
	ucred->uid = ucred->gid = -1;
	if (cred) {
		struct user_namespace *current_ns = current_user_ns();

1074 1075
		ucred->uid = from_kuid_munged(current_ns, cred->euid);
		ucred->gid = from_kgid_munged(current_ns, cred->egid);
1076 1077 1078
	}
}

L
Linus Torvalds 已提交
1079 1080 1081 1082
int sock_getsockopt(struct socket *sock, int level, int optname,
		    char __user *optval, int __user *optlen)
{
	struct sock *sk = sock->sk;
1083

1084
	union {
1085
		int val;
1086
		u64 val64;
1087
		struct linger ling;
L
Linus Torvalds 已提交
1088 1089
		struct timeval tm;
	} v;
1090

1091
	int lv = sizeof(int);
L
Linus Torvalds 已提交
1092
	int len;
1093

1094
	if (get_user(len, optlen))
1095
		return -EFAULT;
1096
	if (len < 0)
L
Linus Torvalds 已提交
1097
		return -EINVAL;
1098

1099
	memset(&v, 0, sizeof(v));
1100

E
Eric Dumazet 已提交
1101
	switch (optname) {
1102 1103 1104 1105 1106 1107 1108 1109 1110
	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
1111
		v.val = sock_flag(sk, SOCK_BROADCAST);
1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125
		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;

T
Tom Herbert 已提交
1126 1127 1128 1129
	case SO_REUSEPORT:
		v.val = sk->sk_reuseport;
		break;

1130
	case SO_KEEPALIVE:
E
Eric Dumazet 已提交
1131
		v.val = sock_flag(sk, SOCK_KEEPOPEN);
1132 1133 1134 1135 1136 1137
		break;

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

1138 1139 1140 1141
	case SO_PROTOCOL:
		v.val = sk->sk_protocol;
		break;

1142 1143 1144 1145
	case SO_DOMAIN:
		v.val = sk->sk_family;
		break;

1146 1147
	case SO_ERROR:
		v.val = -sock_error(sk);
E
Eric Dumazet 已提交
1148
		if (v.val == 0)
1149 1150 1151 1152
			v.val = xchg(&sk->sk_err_soft, 0);
		break;

	case SO_OOBINLINE:
E
Eric Dumazet 已提交
1153
		v.val = sock_flag(sk, SOCK_URGINLINE);
1154 1155 1156
		break;

	case SO_NO_CHECK:
1157
		v.val = sk->sk_no_check_tx;
1158 1159 1160 1161 1162 1163 1164 1165
		break;

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

	case SO_LINGER:
		lv		= sizeof(v.ling);
E
Eric Dumazet 已提交
1166
		v.ling.l_onoff	= sock_flag(sk, SOCK_LINGER);
1167 1168 1169 1170 1171 1172 1173 1174
		v.ling.l_linger	= sk->sk_lingertime / HZ;
		break;

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

	case SO_TIMESTAMP:
1175 1176 1177 1178 1179 1180
		v.val = sock_flag(sk, SOCK_RCVTSTAMP) &&
				!sock_flag(sk, SOCK_RCVTSTAMPNS);
		break;

	case SO_TIMESTAMPNS:
		v.val = sock_flag(sk, SOCK_RCVTSTAMPNS);
1181 1182
		break;

1183
	case SO_TIMESTAMPING:
1184
		v.val = sk->sk_tsflags;
1185 1186
		break;

1187
	case SO_RCVTIMEO:
E
Eric Dumazet 已提交
1188
		lv = sizeof(struct timeval);
1189 1190 1191 1192 1193
		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;
1194
			v.tm.tv_usec = ((sk->sk_rcvtimeo % HZ) * USEC_PER_SEC) / HZ;
1195 1196 1197 1198
		}
		break;

	case SO_SNDTIMEO:
E
Eric Dumazet 已提交
1199
		lv = sizeof(struct timeval);
1200 1201 1202 1203 1204
		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;
1205
			v.tm.tv_usec = ((sk->sk_sndtimeo % HZ) * USEC_PER_SEC) / HZ;
1206 1207
		}
		break;
L
Linus Torvalds 已提交
1208

1209 1210 1211
	case SO_RCVLOWAT:
		v.val = sk->sk_rcvlowat;
		break;
L
Linus Torvalds 已提交
1212

1213
	case SO_SNDLOWAT:
E
Eric Dumazet 已提交
1214
		v.val = 1;
1215
		break;
L
Linus Torvalds 已提交
1216

1217
	case SO_PASSCRED:
1218
		v.val = !!test_bit(SOCK_PASSCRED, &sock->flags);
1219
		break;
L
Linus Torvalds 已提交
1220

1221
	case SO_PEERCRED:
1222 1223 1224 1225 1226 1227
	{
		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))
1228 1229
			return -EFAULT;
		goto lenout;
1230
	}
L
Linus Torvalds 已提交
1231

1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243
	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
1244

1245 1246 1247 1248 1249 1250
	/* 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
1251

1252
	case SO_PASSSEC:
1253
		v.val = !!test_bit(SOCK_PASSSEC, &sock->flags);
1254
		break;
C
Catherine Zhang 已提交
1255

1256 1257
	case SO_PEERSEC:
		return security_socket_getpeersec_stream(sock, optval, optlen, len);
L
Linus Torvalds 已提交
1258

1259 1260 1261 1262
	case SO_MARK:
		v.val = sk->sk_mark;
		break;

1263
	case SO_RXQ_OVFL:
E
Eric Dumazet 已提交
1264
		v.val = sock_flag(sk, SOCK_RXQ_OVFL);
1265 1266
		break;

1267
	case SO_WIFI_STATUS:
E
Eric Dumazet 已提交
1268
		v.val = sock_flag(sk, SOCK_WIFI_STATUS);
1269 1270
		break;

1271 1272 1273 1274 1275 1276
	case SO_PEEK_OFF:
		if (!sock->ops->set_peek_off)
			return -EOPNOTSUPP;

		v.val = sk->sk_peek_off;
		break;
1277
	case SO_NOFCS:
E
Eric Dumazet 已提交
1278
		v.val = sock_flag(sk, SOCK_NOFCS);
1279
		break;
1280

1281
	case SO_BINDTODEVICE:
1282 1283
		return sock_getbindtodevice(sk, optval, optlen, len);

1284 1285 1286 1287 1288 1289
	case SO_GET_FILTER:
		len = sk_get_filter(sk, (struct sock_filter __user *)optval, len);
		if (len < 0)
			return len;

		goto lenout;
1290

1291 1292 1293 1294
	case SO_LOCK_FILTER:
		v.val = sock_flag(sk, SOCK_FILTER_LOCKED);
		break;

1295 1296 1297 1298
	case SO_BPF_EXTENSIONS:
		v.val = bpf_tell_extensions();
		break;

1299 1300 1301 1302
	case SO_SELECT_ERR_QUEUE:
		v.val = sock_flag(sk, SOCK_SELECT_ERR_QUEUE);
		break;

1303
#ifdef CONFIG_NET_RX_BUSY_POLL
1304
	case SO_BUSY_POLL:
1305 1306 1307 1308
		v.val = sk->sk_ll_usec;
		break;
#endif

E
Eric Dumazet 已提交
1309 1310 1311 1312
	case SO_MAX_PACING_RATE:
		v.val = sk->sk_max_pacing_rate;
		break;

E
Eric Dumazet 已提交
1313 1314 1315 1316
	case SO_INCOMING_CPU:
		v.val = sk->sk_incoming_cpu;
		break;

1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331
	case SO_MEMINFO:
	{
		u32 meminfo[SK_MEMINFO_VARS];

		if (get_user(len, optlen))
			return -EFAULT;

		sk_get_meminfo(sk, meminfo);

		len = min_t(unsigned int, len, sizeof(meminfo));
		if (copy_to_user(optval, &meminfo, len))
			return -EFAULT;

		goto lenout;
	}
1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343

#ifdef CONFIG_NET_RX_BUSY_POLL
	case SO_INCOMING_NAPI_ID:
		v.val = READ_ONCE(sk->sk_napi_id);

		/* aggregate non-NAPI IDs down to 0 */
		if (v.val < MIN_NAPI_ID)
			v.val = 0;

		break;
#endif

1344 1345 1346 1347 1348 1349 1350
	case SO_COOKIE:
		lv = sizeof(u64);
		if (len < lv)
			return -EINVAL;
		v.val64 = sock_gen_cookie(sk);
		break;

1351
	default:
1352 1353 1354
		/* We implement the SO_SNDLOWAT etc to not be settable
		 * (1003.1g 7).
		 */
1355
		return -ENOPROTOOPT;
L
Linus Torvalds 已提交
1356
	}
1357

L
Linus Torvalds 已提交
1358 1359 1360 1361 1362
	if (len > lv)
		len = lv;
	if (copy_to_user(optval, &v, len))
		return -EFAULT;
lenout:
1363 1364 1365
	if (put_user(len, optlen))
		return -EFAULT;
	return 0;
L
Linus Torvalds 已提交
1366 1367
}

I
Ingo Molnar 已提交
1368 1369 1370 1371 1372
/*
 * Initialize an sk_lock.
 *
 * (We also register the sk_lock with the lock validator.)
 */
D
Dave Jones 已提交
1373
static inline void sock_lock_init(struct sock *sk)
I
Ingo Molnar 已提交
1374
{
1375 1376 1377 1378 1379 1380 1381 1382 1383 1384
	if (sk->sk_kern_sock)
		sock_lock_init_class_and_name(
			sk,
			af_family_kern_slock_key_strings[sk->sk_family],
			af_family_kern_slock_keys + sk->sk_family,
			af_family_kern_key_strings[sk->sk_family],
			af_family_kern_keys + sk->sk_family);
	else
		sock_lock_init_class_and_name(
			sk,
1385 1386 1387 1388
			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
1389 1390
}

E
Eric Dumazet 已提交
1391 1392 1393
/*
 * 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.
1394
 * We must not copy fields between sk_dontcopy_begin and sk_dontcopy_end
E
Eric Dumazet 已提交
1395
 */
1396 1397 1398 1399 1400
static void sock_copy(struct sock *nsk, const struct sock *osk)
{
#ifdef CONFIG_SECURITY_NETWORK
	void *sptr = nsk->sk_security;
#endif
1401 1402 1403 1404 1405
	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));

1406 1407 1408 1409 1410 1411
#ifdef CONFIG_SECURITY_NETWORK
	nsk->sk_security = sptr;
	security_sk_clone(osk, nsk);
#endif
}

1412 1413
static struct sock *sk_prot_alloc(struct proto *prot, gfp_t priority,
		int family)
1414 1415 1416 1417 1418
{
	struct sock *sk;
	struct kmem_cache *slab;

	slab = prot->slab;
1419 1420 1421 1422
	if (slab != NULL) {
		sk = kmem_cache_alloc(slab, priority & ~__GFP_ZERO);
		if (!sk)
			return sk;
E
Eric Dumazet 已提交
1423 1424
		if (priority & __GFP_ZERO)
			sk_prot_clear_nulls(sk, prot->obj_size);
1425
	} else
1426 1427
		sk = kmalloc(prot->obj_size, priority);

1428
	if (sk != NULL) {
1429 1430
		kmemcheck_annotate_bitfield(sk, flags);

1431 1432 1433 1434 1435
		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
1436
		sk_tx_queue_clear(sk);
1437 1438
	}

1439
	return sk;
1440 1441 1442 1443 1444 1445 1446 1447 1448

out_free_sec:
	security_sk_free(sk);
out_free:
	if (slab != NULL)
		kmem_cache_free(slab, sk);
	else
		kfree(sk);
	return NULL;
1449 1450 1451 1452 1453
}

static void sk_prot_free(struct proto *prot, struct sock *sk)
{
	struct kmem_cache *slab;
1454
	struct module *owner;
1455

1456
	owner = prot->owner;
1457
	slab = prot->slab;
1458

T
Tejun Heo 已提交
1459
	cgroup_sk_free(&sk->sk_cgrp_data);
1460
	mem_cgroup_sk_free(sk);
1461
	security_sk_free(sk);
1462 1463 1464 1465
	if (slab != NULL)
		kmem_cache_free(slab, sk);
	else
		kfree(sk);
1466
	module_put(owner);
1467 1468
}

L
Linus Torvalds 已提交
1469 1470
/**
 *	sk_alloc - All socket objects are allocated here
1471
 *	@net: the applicable net namespace
1472 1473 1474
 *	@family: protocol family
 *	@priority: for allocation (%GFP_KERNEL, %GFP_ATOMIC, etc)
 *	@prot: struct proto associated with this new sock instance
1475
 *	@kern: is this to be a kernel socket?
L
Linus Torvalds 已提交
1476
 */
1477
struct sock *sk_alloc(struct net *net, int family, gfp_t priority,
1478
		      struct proto *prot, int kern)
L
Linus Torvalds 已提交
1479
{
1480
	struct sock *sk;
L
Linus Torvalds 已提交
1481

1482
	sk = sk_prot_alloc(prot, priority | __GFP_ZERO, family);
L
Linus Torvalds 已提交
1483
	if (sk) {
1484 1485 1486 1487 1488 1489
		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;
1490
		sk->sk_kern_sock = kern;
1491
		sock_lock_init(sk);
1492 1493 1494 1495
		sk->sk_net_refcnt = kern ? 0 : 1;
		if (likely(sk->sk_net_refcnt))
			get_net(net);
		sock_net_set(sk, net);
1496
		atomic_set(&sk->sk_wmem_alloc, 1);
1497

1498
		mem_cgroup_sk_alloc(sk);
1499
		cgroup_sk_alloc(&sk->sk_cgrp_data);
1500 1501
		sock_update_classid(&sk->sk_cgrp_data);
		sock_update_netprioidx(&sk->sk_cgrp_data);
L
Linus Torvalds 已提交
1502
	}
1503

1504
	return sk;
L
Linus Torvalds 已提交
1505
}
E
Eric Dumazet 已提交
1506
EXPORT_SYMBOL(sk_alloc);
L
Linus Torvalds 已提交
1507

1508 1509 1510 1511
/* Sockets having SOCK_RCU_FREE will call this function after one RCU
 * grace period. This is the case for UDP sockets and TCP listeners.
 */
static void __sk_destruct(struct rcu_head *head)
L
Linus Torvalds 已提交
1512
{
1513
	struct sock *sk = container_of(head, struct sock, sk_rcu);
L
Linus Torvalds 已提交
1514 1515 1516 1517 1518
	struct sk_filter *filter;

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

1519 1520
	filter = rcu_dereference_check(sk->sk_filter,
				       atomic_read(&sk->sk_wmem_alloc) == 0);
L
Linus Torvalds 已提交
1521
	if (filter) {
1522
		sk_filter_uncharge(sk, filter);
1523
		RCU_INIT_POINTER(sk->sk_filter, NULL);
L
Linus Torvalds 已提交
1524
	}
1525 1526
	if (rcu_access_pointer(sk->sk_reuseport_cb))
		reuseport_detach_sock(sk);
L
Linus Torvalds 已提交
1527

E
Eric Dumazet 已提交
1528
	sock_disable_timestamp(sk, SK_FLAGS_TIMESTAMP);
L
Linus Torvalds 已提交
1529 1530

	if (atomic_read(&sk->sk_omem_alloc))
J
Joe Perches 已提交
1531 1532
		pr_debug("%s: optmem leakage (%d bytes) detected\n",
			 __func__, atomic_read(&sk->sk_omem_alloc));
L
Linus Torvalds 已提交
1533

1534 1535 1536 1537 1538
	if (sk->sk_frag.page) {
		put_page(sk->sk_frag.page);
		sk->sk_frag.page = NULL;
	}

1539 1540 1541
	if (sk->sk_peer_cred)
		put_cred(sk->sk_peer_cred);
	put_pid(sk->sk_peer_pid);
1542 1543
	if (likely(sk->sk_net_refcnt))
		put_net(sock_net(sk));
1544
	sk_prot_free(sk->sk_prot_creator, sk);
L
Linus Torvalds 已提交
1545
}
1546

1547 1548 1549 1550 1551 1552 1553 1554
void sk_destruct(struct sock *sk)
{
	if (sock_flag(sk, SOCK_RCU_FREE))
		call_rcu(&sk->sk_rcu, __sk_destruct);
	else
		__sk_destruct(&sk->sk_rcu);
}

1555 1556
static void __sk_free(struct sock *sk)
{
1557
	if (unlikely(sock_diag_has_destroy_listeners(sk) && sk->sk_net_refcnt))
1558 1559 1560 1561 1562
		sock_diag_broadcast_destroy(sk);
	else
		sk_destruct(sk);
}

1563 1564 1565
void sk_free(struct sock *sk)
{
	/*
L
Lucas De Marchi 已提交
1566
	 * We subtract one from sk_wmem_alloc and can know if
1567 1568 1569 1570 1571 1572
	 * 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
1573
EXPORT_SYMBOL(sk_free);
L
Linus Torvalds 已提交
1574

1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595
static void sk_init_common(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);

	rwlock_init(&sk->sk_callback_lock);
	lockdep_set_class_and_name(&sk->sk_receive_queue.lock,
			af_rlock_keys + sk->sk_family,
			af_family_rlock_key_strings[sk->sk_family]);
	lockdep_set_class_and_name(&sk->sk_write_queue.lock,
			af_wlock_keys + sk->sk_family,
			af_family_wlock_key_strings[sk->sk_family]);
	lockdep_set_class_and_name(&sk->sk_error_queue.lock,
			af_elock_keys + sk->sk_family,
			af_family_elock_key_strings[sk->sk_family]);
	lockdep_set_class_and_name(&sk->sk_callback_lock,
			af_callback_keys + sk->sk_family,
			af_family_clock_key_strings[sk->sk_family]);
}

1596 1597 1598 1599 1600 1601 1602 1603
/**
 *	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)
1604
{
1605
	struct sock *newsk;
1606
	bool is_charged = true;
1607

1608
	newsk = sk_prot_alloc(sk->sk_prot, priority, sk->sk_family);
1609 1610 1611
	if (newsk != NULL) {
		struct sk_filter *filter;

1612
		sock_copy(newsk, sk);
1613 1614

		/* SANITY */
1615 1616
		if (likely(newsk->sk_net_refcnt))
			get_net(sock_net(newsk));
1617 1618 1619
		sk_node_init(&newsk->sk_node);
		sock_lock_init(newsk);
		bh_lock_sock(newsk);
1620
		newsk->sk_backlog.head	= newsk->sk_backlog.tail = NULL;
Z
Zhu Yi 已提交
1621
		newsk->sk_backlog.len = 0;
1622 1623

		atomic_set(&newsk->sk_rmem_alloc, 0);
1624 1625 1626 1627
		/*
		 * sk_wmem_alloc set to one (see sk_free() and sock_wfree())
		 */
		atomic_set(&newsk->sk_wmem_alloc, 1);
1628
		atomic_set(&newsk->sk_omem_alloc, 0);
1629
		sk_init_common(newsk);
1630 1631

		newsk->sk_dst_cache	= NULL;
1632
		newsk->sk_dst_pending_confirm = 0;
1633 1634
		newsk->sk_wmem_queued	= 0;
		newsk->sk_forward_alloc = 0;
1635
		atomic_set(&newsk->sk_drops, 0);
1636 1637 1638 1639 1640
		newsk->sk_send_head	= NULL;
		newsk->sk_userlocks	= sk->sk_userlocks & ~SOCK_BINDPORT_LOCK;

		sock_reset_flag(newsk, SOCK_DONE);

1641
		filter = rcu_dereference_protected(newsk->sk_filter, 1);
1642
		if (filter != NULL)
1643 1644 1645 1646 1647
			/* though it's an empty new sock, the charging may fail
			 * if sysctl_optmem_max was changed between creation of
			 * original socket and cloning
			 */
			is_charged = sk_filter_charge(newsk, filter);
1648

1649
		if (unlikely(!is_charged || xfrm_sk_clone_policy(newsk, sk))) {
1650 1651 1652 1653 1654 1655
			/* We need to make sure that we don't uncharge the new
			 * socket if we couldn't charge it in the first place
			 * as otherwise we uncharge the parent's filter.
			 */
			if (!is_charged)
				RCU_INIT_POINTER(newsk->sk_filter, NULL);
1656
			sk_free_unlock_clone(newsk);
1657 1658 1659
			newsk = NULL;
			goto out;
		}
1660
		RCU_INIT_POINTER(newsk->sk_reuseport_cb, NULL);
1661 1662

		newsk->sk_err	   = 0;
1663
		newsk->sk_err_soft = 0;
1664
		newsk->sk_priority = 0;
E
Eric Dumazet 已提交
1665
		newsk->sk_incoming_cpu = raw_smp_processor_id();
E
Eric Dumazet 已提交
1666
		atomic64_set(&newsk->sk_cookie, 0);
1667

1668
		mem_cgroup_sk_alloc(newsk);
1669 1670
		cgroup_sk_alloc(&newsk->sk_cgrp_data);

E
Eric Dumazet 已提交
1671 1672 1673 1674 1675
		/*
		 * Before updating sk_refcnt, we must commit prior changes to memory
		 * (Documentation/RCU/rculist_nulls.txt for details)
		 */
		smp_wmb();
1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689
		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);
1690
		sk_set_socket(newsk, NULL);
1691
		newsk->sk_wq = NULL;
1692 1693

		if (newsk->sk_prot->sockets_allocated)
1694
			sk_sockets_allocated_inc(newsk);
1695

1696 1697
		if (sock_needs_netstamp(sk) &&
		    newsk->sk_flags & SK_FLAGS_TIMESTAMP)
1698
			net_enable_timestamp();
1699 1700 1701 1702
	}
out:
	return newsk;
}
1703
EXPORT_SYMBOL_GPL(sk_clone_lock);
1704

1705 1706 1707 1708 1709 1710 1711 1712 1713 1714
void sk_free_unlock_clone(struct sock *sk)
{
	/* It is still raw copy of parent, so invalidate
	 * destructor and make plain sk_free() */
	sk->sk_destruct = NULL;
	bh_unlock_sock(sk);
	sk_free(sk);
}
EXPORT_SYMBOL_GPL(sk_free_unlock_clone);

1715 1716
void sk_setup_caps(struct sock *sk, struct dst_entry *dst)
{
1717 1718
	u32 max_segs = 1;

E
Eric Dumazet 已提交
1719
	sk_dst_set(sk, dst);
1720 1721
	sk->sk_route_caps = dst->dev->features;
	if (sk->sk_route_caps & NETIF_F_GSO)
1722
		sk->sk_route_caps |= NETIF_F_GSO_SOFTWARE;
E
Eric Dumazet 已提交
1723
	sk->sk_route_caps &= ~sk->sk_route_nocaps;
1724
	if (sk_can_gso(sk)) {
1725
		if (dst->header_len) {
1726
			sk->sk_route_caps &= ~NETIF_F_GSO_MASK;
1727
		} else {
1728
			sk->sk_route_caps |= NETIF_F_SG | NETIF_F_HW_CSUM;
1729
			sk->sk_gso_max_size = dst->dev->gso_max_size;
1730
			max_segs = max_t(u32, dst->dev->gso_max_segs, 1);
1731
		}
1732
	}
1733
	sk->sk_gso_max_segs = max_segs;
1734 1735 1736
}
EXPORT_SYMBOL_GPL(sk_setup_caps);

L
Linus Torvalds 已提交
1737 1738 1739 1740 1741
/*
 *	Simple resource managers for sockets.
 */


1742 1743
/*
 * Write buffer destructor automatically called from kfree_skb.
L
Linus Torvalds 已提交
1744 1745 1746 1747
 */
void sock_wfree(struct sk_buff *skb)
{
	struct sock *sk = skb->sk;
E
Eric Dumazet 已提交
1748
	unsigned int len = skb->truesize;
L
Linus Torvalds 已提交
1749

E
Eric Dumazet 已提交
1750 1751 1752 1753 1754 1755
	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
1756
		sk->sk_write_space(sk);
E
Eric Dumazet 已提交
1757 1758
		len = 1;
	}
1759
	/*
E
Eric Dumazet 已提交
1760 1761
	 * if sk_wmem_alloc reaches 0, we must finish what sk_free()
	 * could not do because of in-flight packets
1762
	 */
E
Eric Dumazet 已提交
1763
	if (atomic_sub_and_test(len, &sk->sk_wmem_alloc))
1764
		__sk_free(sk);
L
Linus Torvalds 已提交
1765
}
E
Eric Dumazet 已提交
1766
EXPORT_SYMBOL(sock_wfree);
L
Linus Torvalds 已提交
1767

E
Eric Dumazet 已提交
1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778
/* This variant of sock_wfree() is used by TCP,
 * since it sets SOCK_USE_WRITE_QUEUE.
 */
void __sock_wfree(struct sk_buff *skb)
{
	struct sock *sk = skb->sk;

	if (atomic_sub_and_test(skb->truesize, &sk->sk_wmem_alloc))
		__sk_free(sk);
}

1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800
void skb_set_owner_w(struct sk_buff *skb, struct sock *sk)
{
	skb_orphan(skb);
	skb->sk = sk;
#ifdef CONFIG_INET
	if (unlikely(!sk_fullsock(sk))) {
		skb->destructor = sock_edemux;
		sock_hold(sk);
		return;
	}
#endif
	skb->destructor = sock_wfree;
	skb_set_hash_from_sk(skb, sk);
	/*
	 * We used to take a refcount on sk, but following operation
	 * is enough to guarantee sk_free() wont free this sock until
	 * all in-flight packets are completed
	 */
	atomic_add(skb->truesize, &sk->sk_wmem_alloc);
}
EXPORT_SYMBOL(skb_set_owner_w);

E
Eric Dumazet 已提交
1801 1802 1803 1804 1805 1806 1807
/* This helper is used by netem, as it can hold packets in its
 * delay queue. We want to allow the owner socket to send more
 * packets, as if they were already TX completed by a typical driver.
 * But we also want to keep skb->sk set because some packet schedulers
 * rely on it (sch_fq for example). So we set skb->truesize to a small
 * amount (1) and decrease sk_wmem_alloc accordingly.
 */
1808 1809
void skb_orphan_partial(struct sk_buff *skb)
{
E
Eric Dumazet 已提交
1810 1811 1812 1813 1814 1815
	/* If this skb is a TCP pure ACK or already went here,
	 * we have nothing to do. 2 is already a very small truesize.
	 */
	if (skb->truesize <= 2)
		return;

1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832
	/* TCP stack sets skb->ooo_okay based on sk_wmem_alloc,
	 * so we do not completely orphan skb, but transfert all
	 * accounted bytes but one, to avoid unexpected reorders.
	 */
	if (skb->destructor == sock_wfree
#ifdef CONFIG_INET
	    || skb->destructor == tcp_wfree
#endif
		) {
		atomic_sub(skb->truesize - 1, &skb->sk->sk_wmem_alloc);
		skb->truesize = 1;
	} else {
		skb_orphan(skb);
	}
}
EXPORT_SYMBOL(skb_orphan_partial);

1833 1834
/*
 * Read buffer destructor automatically called from kfree_skb.
L
Linus Torvalds 已提交
1835 1836 1837 1838
 */
void sock_rfree(struct sk_buff *skb)
{
	struct sock *sk = skb->sk;
E
Eric Dumazet 已提交
1839
	unsigned int len = skb->truesize;
L
Linus Torvalds 已提交
1840

E
Eric Dumazet 已提交
1841 1842
	atomic_sub(len, &sk->sk_rmem_alloc);
	sk_mem_uncharge(sk, len);
L
Linus Torvalds 已提交
1843
}
E
Eric Dumazet 已提交
1844
EXPORT_SYMBOL(sock_rfree);
L
Linus Torvalds 已提交
1845

1846 1847 1848 1849
/*
 * Buffer destructor for skbs that are not used directly in read or write
 * path, e.g. for error handler skbs. Automatically called from kfree_skb.
 */
1850 1851 1852 1853 1854 1855
void sock_efree(struct sk_buff *skb)
{
	sock_put(skb->sk);
}
EXPORT_SYMBOL(sock_efree);

1856
kuid_t sock_i_uid(struct sock *sk)
L
Linus Torvalds 已提交
1857
{
1858
	kuid_t uid;
L
Linus Torvalds 已提交
1859

E
Eric Dumazet 已提交
1860
	read_lock_bh(&sk->sk_callback_lock);
1861
	uid = sk->sk_socket ? SOCK_INODE(sk->sk_socket)->i_uid : GLOBAL_ROOT_UID;
E
Eric Dumazet 已提交
1862
	read_unlock_bh(&sk->sk_callback_lock);
L
Linus Torvalds 已提交
1863 1864
	return uid;
}
E
Eric Dumazet 已提交
1865
EXPORT_SYMBOL(sock_i_uid);
L
Linus Torvalds 已提交
1866 1867 1868 1869 1870

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

E
Eric Dumazet 已提交
1871
	read_lock_bh(&sk->sk_callback_lock);
L
Linus Torvalds 已提交
1872
	ino = sk->sk_socket ? SOCK_INODE(sk->sk_socket)->i_ino : 0;
E
Eric Dumazet 已提交
1873
	read_unlock_bh(&sk->sk_callback_lock);
L
Linus Torvalds 已提交
1874 1875
	return ino;
}
E
Eric Dumazet 已提交
1876
EXPORT_SYMBOL(sock_i_ino);
L
Linus Torvalds 已提交
1877 1878 1879 1880

/*
 * Allocate a skb from the socket's send buffer.
 */
V
Victor Fusco 已提交
1881
struct sk_buff *sock_wmalloc(struct sock *sk, unsigned long size, int force,
A
Al Viro 已提交
1882
			     gfp_t priority)
L
Linus Torvalds 已提交
1883 1884
{
	if (force || atomic_read(&sk->sk_wmem_alloc) < sk->sk_sndbuf) {
E
Eric Dumazet 已提交
1885
		struct sk_buff *skb = alloc_skb(size, priority);
L
Linus Torvalds 已提交
1886 1887 1888 1889 1890 1891 1892
		if (skb) {
			skb_set_owner_w(skb, sk);
			return skb;
		}
	}
	return NULL;
}
E
Eric Dumazet 已提交
1893
EXPORT_SYMBOL(sock_wmalloc);
L
Linus Torvalds 已提交
1894

1895
/*
L
Linus Torvalds 已提交
1896
 * Allocate a memory block from the socket's option memory buffer.
1897
 */
A
Al Viro 已提交
1898
void *sock_kmalloc(struct sock *sk, int size, gfp_t priority)
L
Linus Torvalds 已提交
1899
{
1900
	if ((unsigned int)size <= sysctl_optmem_max &&
L
Linus Torvalds 已提交
1901 1902 1903
	    atomic_read(&sk->sk_omem_alloc) + size < sysctl_optmem_max) {
		void *mem;
		/* First do the add, to avoid the race if kmalloc
1904
		 * might sleep.
L
Linus Torvalds 已提交
1905 1906 1907 1908 1909 1910 1911 1912 1913
		 */
		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
1914
EXPORT_SYMBOL(sock_kmalloc);
L
Linus Torvalds 已提交
1915

1916 1917 1918
/* Free an option memory block. Note, we actually want the inline
 * here as this allows gcc to detect the nullify and fold away the
 * condition entirely.
L
Linus Torvalds 已提交
1919
 */
1920 1921
static inline void __sock_kfree_s(struct sock *sk, void *mem, int size,
				  const bool nullify)
L
Linus Torvalds 已提交
1922
{
1923 1924
	if (WARN_ON_ONCE(!mem))
		return;
1925 1926 1927 1928
	if (nullify)
		kzfree(mem);
	else
		kfree(mem);
L
Linus Torvalds 已提交
1929 1930
	atomic_sub(size, &sk->sk_omem_alloc);
}
1931 1932 1933 1934 1935

void sock_kfree_s(struct sock *sk, void *mem, int size)
{
	__sock_kfree_s(sk, mem, size, false);
}
E
Eric Dumazet 已提交
1936
EXPORT_SYMBOL(sock_kfree_s);
L
Linus Torvalds 已提交
1937

1938 1939 1940 1941 1942 1943
void sock_kzfree_s(struct sock *sk, void *mem, int size)
{
	__sock_kfree_s(sk, mem, size, true);
}
EXPORT_SYMBOL(sock_kzfree_s);

L
Linus Torvalds 已提交
1944 1945 1946
/* 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
1947
static long sock_wait_for_wmem(struct sock *sk, long timeo)
L
Linus Torvalds 已提交
1948 1949 1950
{
	DEFINE_WAIT(wait);

1951
	sk_clear_bit(SOCKWQ_ASYNC_NOSPACE, sk);
L
Linus Torvalds 已提交
1952 1953 1954 1955 1956 1957
	for (;;) {
		if (!timeo)
			break;
		if (signal_pending(current))
			break;
		set_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
E
Eric Dumazet 已提交
1958
		prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
L
Linus Torvalds 已提交
1959 1960 1961 1962 1963 1964 1965 1966
		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
1967
	finish_wait(sk_sleep(sk), &wait);
L
Linus Torvalds 已提交
1968 1969 1970 1971 1972 1973 1974 1975
	return timeo;
}


/*
 *	Generic send/receive buffer handlers
 */

H
Herbert Xu 已提交
1976 1977
struct sk_buff *sock_alloc_send_pskb(struct sock *sk, unsigned long header_len,
				     unsigned long data_len, int noblock,
1978
				     int *errcode, int max_page_order)
L
Linus Torvalds 已提交
1979
{
1980
	struct sk_buff *skb;
L
Linus Torvalds 已提交
1981 1982 1983 1984
	long timeo;
	int err;

	timeo = sock_sndtimeo(sk, noblock);
1985
	for (;;) {
L
Linus Torvalds 已提交
1986 1987 1988 1989 1990 1991 1992 1993
		err = sock_error(sk);
		if (err != 0)
			goto failure;

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

1994 1995
		if (sk_wmem_alloc_get(sk) < sk->sk_sndbuf)
			break;
1996

1997
		sk_set_bit(SOCKWQ_ASYNC_NOSPACE, sk);
1998 1999 2000
		set_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
		err = -EAGAIN;
		if (!timeo)
L
Linus Torvalds 已提交
2001
			goto failure;
2002 2003 2004
		if (signal_pending(current))
			goto interrupted;
		timeo = sock_wait_for_wmem(sk, timeo);
L
Linus Torvalds 已提交
2005
	}
2006 2007 2008 2009
	skb = alloc_skb_with_frags(header_len, data_len, max_page_order,
				   errcode, sk->sk_allocation);
	if (skb)
		skb_set_owner_w(skb, sk);
L
Linus Torvalds 已提交
2010 2011 2012 2013 2014 2015 2016 2017
	return skb;

interrupted:
	err = sock_intr_errno(timeo);
failure:
	*errcode = err;
	return NULL;
}
H
Herbert Xu 已提交
2018
EXPORT_SYMBOL(sock_alloc_send_pskb);
L
Linus Torvalds 已提交
2019

2020
struct sk_buff *sock_alloc_send_skb(struct sock *sk, unsigned long size,
L
Linus Torvalds 已提交
2021 2022
				    int noblock, int *errcode)
{
2023
	return sock_alloc_send_pskb(sk, size, 0, noblock, errcode, 0);
L
Linus Torvalds 已提交
2024
}
E
Eric Dumazet 已提交
2025
EXPORT_SYMBOL(sock_alloc_send_skb);
L
Linus Torvalds 已提交
2026

2027 2028 2029
int __sock_cmsg_send(struct sock *sk, struct msghdr *msg, struct cmsghdr *cmsg,
		     struct sockcm_cookie *sockc)
{
2030 2031
	u32 tsflags;

2032 2033 2034 2035 2036 2037 2038 2039
	switch (cmsg->cmsg_type) {
	case SO_MARK:
		if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
			return -EPERM;
		if (cmsg->cmsg_len != CMSG_LEN(sizeof(u32)))
			return -EINVAL;
		sockc->mark = *(u32 *)CMSG_DATA(cmsg);
		break;
2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050
	case SO_TIMESTAMPING:
		if (cmsg->cmsg_len != CMSG_LEN(sizeof(u32)))
			return -EINVAL;

		tsflags = *(u32 *)CMSG_DATA(cmsg);
		if (tsflags & ~SOF_TIMESTAMPING_TX_RECORD_MASK)
			return -EINVAL;

		sockc->tsflags &= ~SOF_TIMESTAMPING_TX_RECORD_MASK;
		sockc->tsflags |= tsflags;
		break;
2051 2052 2053 2054
	/* SCM_RIGHTS and SCM_CREDENTIALS are semantically in SOL_UNIX. */
	case SCM_RIGHTS:
	case SCM_CREDENTIALS:
		break;
2055 2056 2057 2058 2059 2060 2061
	default:
		return -EINVAL;
	}
	return 0;
}
EXPORT_SYMBOL(__sock_cmsg_send);

E
Edward Jee 已提交
2062 2063 2064 2065
int sock_cmsg_send(struct sock *sk, struct msghdr *msg,
		   struct sockcm_cookie *sockc)
{
	struct cmsghdr *cmsg;
2066
	int ret;
E
Edward Jee 已提交
2067 2068 2069 2070 2071 2072

	for_each_cmsghdr(cmsg, msg) {
		if (!CMSG_OK(msg, cmsg))
			return -EINVAL;
		if (cmsg->cmsg_level != SOL_SOCKET)
			continue;
2073 2074 2075
		ret = __sock_cmsg_send(sk, msg, cmsg, sockc);
		if (ret)
			return ret;
E
Edward Jee 已提交
2076 2077 2078 2079 2080
	}
	return 0;
}
EXPORT_SYMBOL(sock_cmsg_send);

2081 2082 2083
/* On 32bit arches, an skb frag is limited to 2^15 */
#define SKB_FRAG_PAGE_ORDER	get_order(32768)

E
Eric Dumazet 已提交
2084 2085 2086 2087
/**
 * skb_page_frag_refill - check that a page_frag contains enough room
 * @sz: minimum size of the fragment we want to get
 * @pfrag: pointer to page_frag
2088
 * @gfp: priority for memory allocation
E
Eric Dumazet 已提交
2089 2090 2091 2092 2093
 *
 * Note: While this allocator tries to use high order pages, there is
 * no guarantee that allocations succeed. Therefore, @sz MUST be
 * less or equal than PAGE_SIZE.
 */
2094
bool skb_page_frag_refill(unsigned int sz, struct page_frag *pfrag, gfp_t gfp)
2095 2096
{
	if (pfrag->page) {
2097
		if (page_ref_count(pfrag->page) == 1) {
2098 2099 2100
			pfrag->offset = 0;
			return true;
		}
E
Eric Dumazet 已提交
2101
		if (pfrag->offset + sz <= pfrag->size)
2102 2103 2104 2105
			return true;
		put_page(pfrag->page);
	}

2106 2107
	pfrag->offset = 0;
	if (SKB_FRAG_PAGE_ORDER) {
2108 2109 2110 2111
		/* Avoid direct reclaim but allow kswapd to wake */
		pfrag->page = alloc_pages((gfp & ~__GFP_DIRECT_RECLAIM) |
					  __GFP_COMP | __GFP_NOWARN |
					  __GFP_NORETRY,
2112
					  SKB_FRAG_PAGE_ORDER);
2113
		if (likely(pfrag->page)) {
2114
			pfrag->size = PAGE_SIZE << SKB_FRAG_PAGE_ORDER;
2115 2116
			return true;
		}
2117 2118 2119 2120 2121 2122
	}
	pfrag->page = alloc_page(gfp);
	if (likely(pfrag->page)) {
		pfrag->size = PAGE_SIZE;
		return true;
	}
E
Eric Dumazet 已提交
2123 2124 2125 2126 2127 2128 2129 2130 2131
	return false;
}
EXPORT_SYMBOL(skb_page_frag_refill);

bool sk_page_frag_refill(struct sock *sk, struct page_frag *pfrag)
{
	if (likely(skb_page_frag_refill(32U, pfrag, sk->sk_allocation)))
		return true;

2132 2133 2134 2135 2136 2137
	sk_enter_memory_pressure(sk);
	sk_stream_moderate_sndbuf(sk);
	return false;
}
EXPORT_SYMBOL(sk_page_frag_refill);

L
Linus Torvalds 已提交
2138
static void __lock_sock(struct sock *sk)
2139 2140
	__releases(&sk->sk_lock.slock)
	__acquires(&sk->sk_lock.slock)
L
Linus Torvalds 已提交
2141 2142 2143
{
	DEFINE_WAIT(wait);

2144
	for (;;) {
L
Linus Torvalds 已提交
2145 2146 2147 2148 2149
		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);
2150
		if (!sock_owned_by_user(sk))
L
Linus Torvalds 已提交
2151 2152 2153 2154 2155 2156
			break;
	}
	finish_wait(&sk->sk_lock.wq, &wait);
}

static void __release_sock(struct sock *sk)
2157 2158
	__releases(&sk->sk_lock.slock)
	__acquires(&sk->sk_lock.slock)
L
Linus Torvalds 已提交
2159
{
2160
	struct sk_buff *skb, *next;
L
Linus Torvalds 已提交
2161

2162
	while ((skb = sk->sk_backlog.head) != NULL) {
L
Linus Torvalds 已提交
2163 2164
		sk->sk_backlog.head = sk->sk_backlog.tail = NULL;

2165
		spin_unlock_bh(&sk->sk_lock.slock);
L
Linus Torvalds 已提交
2166

2167 2168
		do {
			next = skb->next;
2169
			prefetch(next);
E
Eric Dumazet 已提交
2170
			WARN_ON_ONCE(skb_dst_is_noref(skb));
L
Linus Torvalds 已提交
2171
			skb->next = NULL;
P
Peter Zijlstra 已提交
2172
			sk_backlog_rcv(sk, skb);
L
Linus Torvalds 已提交
2173

2174
			cond_resched();
L
Linus Torvalds 已提交
2175 2176 2177 2178

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

2179 2180
		spin_lock_bh(&sk->sk_lock.slock);
	}
Z
Zhu Yi 已提交
2181 2182 2183 2184 2185 2186

	/*
	 * 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
2187 2188
}

2189 2190 2191 2192 2193 2194 2195
void __sk_flush_backlog(struct sock *sk)
{
	spin_lock_bh(&sk->sk_lock.slock);
	__release_sock(sk);
	spin_unlock_bh(&sk->sk_lock.slock);
}

L
Linus Torvalds 已提交
2196 2197
/**
 * sk_wait_data - wait for data to arrive at sk_receive_queue
2198 2199
 * @sk:    sock to wait on
 * @timeo: for how long
2200
 * @skb:   last skb seen on sk_receive_queue
L
Linus Torvalds 已提交
2201 2202 2203 2204 2205 2206
 *
 * 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.
 */
2207
int sk_wait_data(struct sock *sk, long *timeo, const struct sk_buff *skb)
L
Linus Torvalds 已提交
2208
{
W
WANG Cong 已提交
2209
	DEFINE_WAIT_FUNC(wait, woken_wake_function);
L
Linus Torvalds 已提交
2210 2211
	int rc;

W
WANG Cong 已提交
2212
	add_wait_queue(sk_sleep(sk), &wait);
2213
	sk_set_bit(SOCKWQ_ASYNC_WAITDATA, sk);
W
WANG Cong 已提交
2214
	rc = sk_wait_event(sk, timeo, skb_peek_tail(&sk->sk_receive_queue) != skb, &wait);
2215
	sk_clear_bit(SOCKWQ_ASYNC_WAITDATA, sk);
W
WANG Cong 已提交
2216
	remove_wait_queue(sk_sleep(sk), &wait);
L
Linus Torvalds 已提交
2217 2218 2219 2220
	return rc;
}
EXPORT_SYMBOL(sk_wait_data);

2221
/**
2222
 *	__sk_mem_raise_allocated - increase memory_allocated
2223 2224
 *	@sk: socket
 *	@size: memory size to allocate
2225
 *	@amt: pages to allocate
2226 2227
 *	@kind: allocation type
 *
2228
 *	Similar to __sk_mem_schedule(), but does not update sk_forward_alloc
2229
 */
2230
int __sk_mem_raise_allocated(struct sock *sk, int size, int amt, int kind)
2231 2232
{
	struct proto *prot = sk->sk_prot;
2233
	long allocated = sk_memory_allocated_add(sk, amt);
2234

2235 2236
	if (mem_cgroup_sockets_enabled && sk->sk_memcg &&
	    !mem_cgroup_charge_skmem(sk->sk_memcg, amt))
2237
		goto suppress_allocation;
2238 2239

	/* Under limit. */
2240
	if (allocated <= sk_prot_mem_limits(sk, 0)) {
2241
		sk_leave_memory_pressure(sk);
2242 2243 2244
		return 1;
	}

2245 2246
	/* Under pressure. */
	if (allocated > sk_prot_mem_limits(sk, 1))
2247
		sk_enter_memory_pressure(sk);
2248

2249 2250
	/* Over hard limit. */
	if (allocated > sk_prot_mem_limits(sk, 2))
2251 2252 2253 2254 2255 2256
		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;
2257

2258 2259 2260 2261 2262 2263 2264 2265 2266
	} 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;
	}

2267
	if (sk_has_memory_pressure(sk)) {
2268 2269
		int alloc;

2270
		if (!sk_under_memory_pressure(sk))
2271
			return 1;
2272 2273
		alloc = sk_sockets_allocated_read_positive(sk);
		if (sk_prot_mem_limits(sk, 2) > alloc *
2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291
		    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;
	}

2292 2293
	trace_sock_exceed_buf_limit(sk, prot, allocated);

2294
	sk_memory_allocated_sub(sk, amt);
2295

2296 2297
	if (mem_cgroup_sockets_enabled && sk->sk_memcg)
		mem_cgroup_uncharge_skmem(sk->sk_memcg, amt);
2298

2299 2300
	return 0;
}
2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322
EXPORT_SYMBOL(__sk_mem_raise_allocated);

/**
 *	__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)
{
	int ret, amt = sk_mem_pages(size);

	sk->sk_forward_alloc += amt << SK_MEM_QUANTUM_SHIFT;
	ret = __sk_mem_raise_allocated(sk, size, amt, kind);
	if (!ret)
		sk->sk_forward_alloc -= amt << SK_MEM_QUANTUM_SHIFT;
	return ret;
}
2323 2324 2325
EXPORT_SYMBOL(__sk_mem_schedule);

/**
2326
 *	__sk_mem_reduce_allocated - reclaim memory_allocated
2327
 *	@sk: socket
2328 2329 2330
 *	@amount: number of quanta
 *
 *	Similar to __sk_mem_reclaim(), but does not update sk_forward_alloc
2331
 */
2332
void __sk_mem_reduce_allocated(struct sock *sk, int amount)
2333
{
E
Eric Dumazet 已提交
2334
	sk_memory_allocated_sub(sk, amount);
2335

2336 2337
	if (mem_cgroup_sockets_enabled && sk->sk_memcg)
		mem_cgroup_uncharge_skmem(sk->sk_memcg, amount);
2338

2339 2340 2341
	if (sk_under_memory_pressure(sk) &&
	    (sk_memory_allocated(sk) < sk_prot_mem_limits(sk, 0)))
		sk_leave_memory_pressure(sk);
2342
}
2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355
EXPORT_SYMBOL(__sk_mem_reduce_allocated);

/**
 *	__sk_mem_reclaim - reclaim sk_forward_alloc and memory_allocated
 *	@sk: socket
 *	@amount: number of bytes (rounded down to a SK_MEM_QUANTUM multiple)
 */
void __sk_mem_reclaim(struct sock *sk, int amount)
{
	amount >>= SK_MEM_QUANTUM_SHIFT;
	sk->sk_forward_alloc -= amount << SK_MEM_QUANTUM_SHIFT;
	__sk_mem_reduce_allocated(sk, amount);
}
2356 2357
EXPORT_SYMBOL(__sk_mem_reclaim);

2358 2359 2360 2361 2362 2363 2364 2365 2366
int sk_set_peek_off(struct sock *sk, int val)
{
	if (val < 0)
		return -EINVAL;

	sk->sk_peek_off = val;
	return 0;
}
EXPORT_SYMBOL_GPL(sk_set_peek_off);
2367

L
Linus Torvalds 已提交
2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378
/*
 * 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
2379
EXPORT_SYMBOL(sock_no_bind);
L
Linus Torvalds 已提交
2380

2381
int sock_no_connect(struct socket *sock, struct sockaddr *saddr,
L
Linus Torvalds 已提交
2382 2383 2384 2385
		    int len, int flags)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
2386
EXPORT_SYMBOL(sock_no_connect);
L
Linus Torvalds 已提交
2387 2388 2389 2390 2391

int sock_no_socketpair(struct socket *sock1, struct socket *sock2)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
2392
EXPORT_SYMBOL(sock_no_socketpair);
L
Linus Torvalds 已提交
2393

2394 2395
int sock_no_accept(struct socket *sock, struct socket *newsock, int flags,
		   bool kern)
L
Linus Torvalds 已提交
2396 2397 2398
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
2399
EXPORT_SYMBOL(sock_no_accept);
L
Linus Torvalds 已提交
2400

2401
int sock_no_getname(struct socket *sock, struct sockaddr *saddr,
L
Linus Torvalds 已提交
2402 2403 2404 2405
		    int *len, int peer)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
2406
EXPORT_SYMBOL(sock_no_getname);
L
Linus Torvalds 已提交
2407

E
Eric Dumazet 已提交
2408
unsigned int sock_no_poll(struct file *file, struct socket *sock, poll_table *pt)
L
Linus Torvalds 已提交
2409 2410 2411
{
	return 0;
}
E
Eric Dumazet 已提交
2412
EXPORT_SYMBOL(sock_no_poll);
L
Linus Torvalds 已提交
2413 2414 2415 2416 2417

int sock_no_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
2418
EXPORT_SYMBOL(sock_no_ioctl);
L
Linus Torvalds 已提交
2419 2420 2421 2422 2423

int sock_no_listen(struct socket *sock, int backlog)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
2424
EXPORT_SYMBOL(sock_no_listen);
L
Linus Torvalds 已提交
2425 2426 2427 2428 2429

int sock_no_shutdown(struct socket *sock, int how)
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
2430
EXPORT_SYMBOL(sock_no_shutdown);
L
Linus Torvalds 已提交
2431 2432

int sock_no_setsockopt(struct socket *sock, int level, int optname,
2433
		    char __user *optval, unsigned int optlen)
L
Linus Torvalds 已提交
2434 2435 2436
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
2437
EXPORT_SYMBOL(sock_no_setsockopt);
L
Linus Torvalds 已提交
2438 2439 2440 2441 2442 2443

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
2444
EXPORT_SYMBOL(sock_no_getsockopt);
L
Linus Torvalds 已提交
2445

2446
int sock_no_sendmsg(struct socket *sock, struct msghdr *m, size_t len)
L
Linus Torvalds 已提交
2447 2448 2449
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
2450
EXPORT_SYMBOL(sock_no_sendmsg);
L
Linus Torvalds 已提交
2451

2452 2453
int sock_no_recvmsg(struct socket *sock, struct msghdr *m, size_t len,
		    int flags)
L
Linus Torvalds 已提交
2454 2455 2456
{
	return -EOPNOTSUPP;
}
E
Eric Dumazet 已提交
2457
EXPORT_SYMBOL(sock_no_recvmsg);
L
Linus Torvalds 已提交
2458 2459 2460 2461 2462 2463

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
2464
EXPORT_SYMBOL(sock_no_mmap);
L
Linus Torvalds 已提交
2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477

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
2478
EXPORT_SYMBOL(sock_no_sendpage);
L
Linus Torvalds 已提交
2479 2480 2481 2482 2483 2484 2485

/*
 *	Default Socket Callbacks
 */

static void sock_def_wakeup(struct sock *sk)
{
2486 2487 2488 2489
	struct socket_wq *wq;

	rcu_read_lock();
	wq = rcu_dereference(sk->sk_wq);
H
Herbert Xu 已提交
2490
	if (skwq_has_sleeper(wq))
2491 2492
		wake_up_interruptible_all(&wq->wait);
	rcu_read_unlock();
L
Linus Torvalds 已提交
2493 2494 2495 2496
}

static void sock_def_error_report(struct sock *sk)
{
2497 2498 2499 2500
	struct socket_wq *wq;

	rcu_read_lock();
	wq = rcu_dereference(sk->sk_wq);
H
Herbert Xu 已提交
2501
	if (skwq_has_sleeper(wq))
2502
		wake_up_interruptible_poll(&wq->wait, POLLERR);
2503
	sk_wake_async(sk, SOCK_WAKE_IO, POLL_ERR);
2504
	rcu_read_unlock();
L
Linus Torvalds 已提交
2505 2506
}

2507
static void sock_def_readable(struct sock *sk)
L
Linus Torvalds 已提交
2508
{
2509 2510 2511 2512
	struct socket_wq *wq;

	rcu_read_lock();
	wq = rcu_dereference(sk->sk_wq);
H
Herbert Xu 已提交
2513
	if (skwq_has_sleeper(wq))
2514
		wake_up_interruptible_sync_poll(&wq->wait, POLLIN | POLLPRI |
2515
						POLLRDNORM | POLLRDBAND);
2516
	sk_wake_async(sk, SOCK_WAKE_WAITD, POLL_IN);
2517
	rcu_read_unlock();
L
Linus Torvalds 已提交
2518 2519 2520 2521
}

static void sock_def_write_space(struct sock *sk)
{
2522 2523 2524
	struct socket_wq *wq;

	rcu_read_lock();
L
Linus Torvalds 已提交
2525 2526 2527 2528

	/* Do not wake up a writer until he can make "significant"
	 * progress.  --DaveM
	 */
2529
	if ((atomic_read(&sk->sk_wmem_alloc) << 1) <= sk->sk_sndbuf) {
2530
		wq = rcu_dereference(sk->sk_wq);
H
Herbert Xu 已提交
2531
		if (skwq_has_sleeper(wq))
2532
			wake_up_interruptible_sync_poll(&wq->wait, POLLOUT |
2533
						POLLWRNORM | POLLWRBAND);
L
Linus Torvalds 已提交
2534 2535 2536

		/* Should agree with poll, otherwise some programs break */
		if (sock_writeable(sk))
2537
			sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
L
Linus Torvalds 已提交
2538 2539
	}

2540
	rcu_read_unlock();
L
Linus Torvalds 已提交
2541 2542 2543 2544 2545 2546 2547 2548 2549 2550
}

static void sock_def_destruct(struct sock *sk)
{
}

void sk_send_sigurg(struct sock *sk)
{
	if (sk->sk_socket && sk->sk_socket->file)
		if (send_sigurg(&sk->sk_socket->file->f_owner))
2551
			sk_wake_async(sk, SOCK_WAKE_URG, POLL_PRI);
L
Linus Torvalds 已提交
2552
}
E
Eric Dumazet 已提交
2553
EXPORT_SYMBOL(sk_send_sigurg);
L
Linus Torvalds 已提交
2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564

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)
{
2565
	if (del_timer(timer))
L
Linus Torvalds 已提交
2566 2567 2568 2569 2570 2571
		__sock_put(sk);
}
EXPORT_SYMBOL(sk_stop_timer);

void sock_init_data(struct socket *sock, struct sock *sk)
{
2572
	sk_init_common(sk);
L
Linus Torvalds 已提交
2573 2574 2575
	sk->sk_send_head	=	NULL;

	init_timer(&sk->sk_timer);
2576

L
Linus Torvalds 已提交
2577 2578 2579 2580
	sk->sk_allocation	=	GFP_KERNEL;
	sk->sk_rcvbuf		=	sysctl_rmem_default;
	sk->sk_sndbuf		=	sysctl_wmem_default;
	sk->sk_state		=	TCP_CLOSE;
2581
	sk_set_socket(sk, sock);
L
Linus Torvalds 已提交
2582 2583 2584

	sock_set_flag(sk, SOCK_ZAPPED);

2585
	if (sock) {
L
Linus Torvalds 已提交
2586
		sk->sk_type	=	sock->type;
2587
		sk->sk_wq	=	sock->wq;
L
Linus Torvalds 已提交
2588
		sock->sk	=	sk;
2589 2590
		sk->sk_uid	=	SOCK_INODE(sock)->i_uid;
	} else {
2591
		sk->sk_wq	=	NULL;
2592 2593
		sk->sk_uid	=	make_kuid(sock_net(sk)->user_ns, 0);
	}
L
Linus Torvalds 已提交
2594 2595

	rwlock_init(&sk->sk_callback_lock);
2596 2597 2598 2599 2600 2601 2602 2603
	if (sk->sk_kern_sock)
		lockdep_set_class_and_name(
			&sk->sk_callback_lock,
			af_kern_callback_keys + sk->sk_family,
			af_family_kern_clock_key_strings[sk->sk_family]);
	else
		lockdep_set_class_and_name(
			&sk->sk_callback_lock,
2604 2605
			af_callback_keys + sk->sk_family,
			af_family_clock_key_strings[sk->sk_family]);
L
Linus Torvalds 已提交
2606 2607 2608 2609 2610 2611 2612

	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;

2613 2614
	sk->sk_frag.page	=	NULL;
	sk->sk_frag.offset	=	0;
2615
	sk->sk_peek_off		=	-1;
L
Linus Torvalds 已提交
2616

2617 2618
	sk->sk_peer_pid 	=	NULL;
	sk->sk_peer_cred	=	NULL;
L
Linus Torvalds 已提交
2619 2620 2621 2622 2623
	sk->sk_write_pending	=	0;
	sk->sk_rcvlowat		=	1;
	sk->sk_rcvtimeo		=	MAX_SCHEDULE_TIMEOUT;
	sk->sk_sndtimeo		=	MAX_SCHEDULE_TIMEOUT;

2624
	sk->sk_stamp = SK_DEFAULT_STAMP;
L
Linus Torvalds 已提交
2625

2626
#ifdef CONFIG_NET_RX_BUSY_POLL
E
Eliezer Tamir 已提交
2627
	sk->sk_napi_id		=	0;
2628
	sk->sk_ll_usec		=	sysctl_net_busy_read;
E
Eliezer Tamir 已提交
2629 2630
#endif

E
Eric Dumazet 已提交
2631
	sk->sk_max_pacing_rate = ~0U;
2632
	sk->sk_pacing_rate = ~0U;
2633
	sk->sk_incoming_cpu = -1;
E
Eric Dumazet 已提交
2634 2635 2636 2637 2638
	/*
	 * 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
2639
	atomic_set(&sk->sk_refcnt, 1);
W
Wang Chen 已提交
2640
	atomic_set(&sk->sk_drops, 0);
L
Linus Torvalds 已提交
2641
}
E
Eric Dumazet 已提交
2642
EXPORT_SYMBOL(sock_init_data);
L
Linus Torvalds 已提交
2643

H
Harvey Harrison 已提交
2644
void lock_sock_nested(struct sock *sk, int subclass)
L
Linus Torvalds 已提交
2645 2646
{
	might_sleep();
I
Ingo Molnar 已提交
2647
	spin_lock_bh(&sk->sk_lock.slock);
2648
	if (sk->sk_lock.owned)
L
Linus Torvalds 已提交
2649
		__lock_sock(sk);
2650
	sk->sk_lock.owned = 1;
I
Ingo Molnar 已提交
2651 2652 2653 2654
	spin_unlock(&sk->sk_lock.slock);
	/*
	 * The sk_lock has mutex_lock() semantics here:
	 */
2655
	mutex_acquire(&sk->sk_lock.dep_map, subclass, 0, _RET_IP_);
I
Ingo Molnar 已提交
2656
	local_bh_enable();
L
Linus Torvalds 已提交
2657
}
2658
EXPORT_SYMBOL(lock_sock_nested);
L
Linus Torvalds 已提交
2659

H
Harvey Harrison 已提交
2660
void release_sock(struct sock *sk)
L
Linus Torvalds 已提交
2661
{
I
Ingo Molnar 已提交
2662
	spin_lock_bh(&sk->sk_lock.slock);
L
Linus Torvalds 已提交
2663 2664
	if (sk->sk_backlog.tail)
		__release_sock(sk);
E
Eric Dumazet 已提交
2665

2666 2667 2668
	/* Warning : release_cb() might need to release sk ownership,
	 * ie call sock_release_ownership(sk) before us.
	 */
E
Eric Dumazet 已提交
2669 2670 2671
	if (sk->sk_prot->release_cb)
		sk->sk_prot->release_cb(sk);

2672
	sock_release_ownership(sk);
I
Ingo Molnar 已提交
2673 2674 2675
	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
2676 2677 2678
}
EXPORT_SYMBOL(release_sock);

2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711
/**
 * 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
2712
int sock_get_timestamp(struct sock *sk, struct timeval __user *userstamp)
2713
{
2714
	struct timeval tv;
L
Linus Torvalds 已提交
2715
	if (!sock_flag(sk, SOCK_TIMESTAMP))
2716
		sock_enable_timestamp(sk, SOCK_TIMESTAMP);
2717 2718
	tv = ktime_to_timeval(sk->sk_stamp);
	if (tv.tv_sec == -1)
L
Linus Torvalds 已提交
2719
		return -ENOENT;
2720 2721 2722 2723 2724
	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;
2725
}
L
Linus Torvalds 已提交
2726 2727
EXPORT_SYMBOL(sock_get_timestamp);

2728 2729 2730 2731
int sock_get_timestampns(struct sock *sk, struct timespec __user *userstamp)
{
	struct timespec ts;
	if (!sock_flag(sk, SOCK_TIMESTAMP))
2732
		sock_enable_timestamp(sk, SOCK_TIMESTAMP);
2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743
	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);

2744
void sock_enable_timestamp(struct sock *sk, int flag)
2745
{
2746
	if (!sock_flag(sk, flag)) {
E
Eric Dumazet 已提交
2747 2748
		unsigned long previous_flags = sk->sk_flags;

2749 2750 2751 2752 2753 2754
		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
		 */
2755 2756
		if (sock_needs_netstamp(sk) &&
		    !(previous_flags & SK_FLAGS_TIMESTAMP))
2757
			net_enable_timestamp();
L
Linus Torvalds 已提交
2758 2759 2760
	}
}

2761 2762 2763 2764
int sock_recv_errqueue(struct sock *sk, struct msghdr *msg, int len,
		       int level, int type)
{
	struct sock_exterr_skb *serr;
2765
	struct sk_buff *skb;
2766 2767 2768
	int copied, err;

	err = -EAGAIN;
2769
	skb = sock_dequeue_err_skb(sk);
2770 2771 2772 2773 2774 2775 2776 2777
	if (skb == NULL)
		goto out;

	copied = skb->len;
	if (copied > len) {
		msg->msg_flags |= MSG_TRUNC;
		copied = len;
	}
2778
	err = skb_copy_datagram_msg(skb, 0, msg, copied);
2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796
	if (err)
		goto out_free_skb;

	sock_recv_timestamp(msg, sk, skb);

	serr = SKB_EXT_ERR(skb);
	put_cmsg(msg, level, type, sizeof(serr->ee), &serr->ee);

	msg->msg_flags |= MSG_ERRQUEUE;
	err = copied;

out_free_skb:
	kfree_skb(skb);
out:
	return err;
}
EXPORT_SYMBOL(sock_recv_errqueue);

L
Linus Torvalds 已提交
2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812
/*
 *	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);

2813
#ifdef CONFIG_COMPAT
2814 2815
int compat_sock_common_getsockopt(struct socket *sock, int level, int optname,
				  char __user *optval, int __user *optlen)
2816 2817 2818
{
	struct sock *sk = sock->sk;

2819
	if (sk->sk_prot->compat_getsockopt != NULL)
2820 2821
		return sk->sk_prot->compat_getsockopt(sk, level, optname,
						      optval, optlen);
2822 2823 2824 2825 2826
	return sk->sk_prot->getsockopt(sk, level, optname, optval, optlen);
}
EXPORT_SYMBOL(compat_sock_common_getsockopt);
#endif

2827 2828
int sock_common_recvmsg(struct socket *sock, struct msghdr *msg, size_t size,
			int flags)
L
Linus Torvalds 已提交
2829 2830 2831 2832 2833
{
	struct sock *sk = sock->sk;
	int addr_len = 0;
	int err;

2834
	err = sk->sk_prot->recvmsg(sk, msg, size, flags & MSG_DONTWAIT,
L
Linus Torvalds 已提交
2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845
				   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,
2846
			   char __user *optval, unsigned int optlen)
L
Linus Torvalds 已提交
2847 2848 2849 2850 2851 2852 2853
{
	struct sock *sk = sock->sk;

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

2854
#ifdef CONFIG_COMPAT
2855
int compat_sock_common_setsockopt(struct socket *sock, int level, int optname,
2856
				  char __user *optval, unsigned int optlen)
2857 2858 2859
{
	struct sock *sk = sock->sk;

2860 2861 2862
	if (sk->sk_prot->compat_setsockopt != NULL)
		return sk->sk_prot->compat_setsockopt(sk, level, optname,
						      optval, optlen);
2863 2864 2865 2866 2867
	return sk->sk_prot->setsockopt(sk, level, optname, optval, optlen);
}
EXPORT_SYMBOL(compat_sock_common_setsockopt);
#endif

L
Linus Torvalds 已提交
2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898
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);

2899
	sk_refcnt_debug_release(sk);
2900

L
Linus Torvalds 已提交
2901 2902 2903 2904
	sock_put(sk);
}
EXPORT_SYMBOL(sk_common_release);

2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919
void sk_get_meminfo(const struct sock *sk, u32 *mem)
{
	memset(mem, 0, sizeof(*mem) * SK_MEMINFO_VARS);

	mem[SK_MEMINFO_RMEM_ALLOC] = sk_rmem_alloc_get(sk);
	mem[SK_MEMINFO_RCVBUF] = sk->sk_rcvbuf;
	mem[SK_MEMINFO_WMEM_ALLOC] = sk_wmem_alloc_get(sk);
	mem[SK_MEMINFO_SNDBUF] = sk->sk_sndbuf;
	mem[SK_MEMINFO_FWD_ALLOC] = sk->sk_forward_alloc;
	mem[SK_MEMINFO_WMEM_QUEUED] = sk->sk_wmem_queued;
	mem[SK_MEMINFO_OPTMEM] = atomic_read(&sk->sk_omem_alloc);
	mem[SK_MEMINFO_BACKLOG] = sk->sk_backlog.len;
	mem[SK_MEMINFO_DROPS] = atomic_read(&sk->sk_drops);
}

2920 2921
#ifdef CONFIG_PROC_FS
#define PROTO_INUSE_NR	64	/* should be enough for the first time */
2922 2923 2924
struct prot_inuse {
	int val[PROTO_INUSE_NR];
};
2925 2926

static DECLARE_BITMAP(proto_inuse_idx, PROTO_INUSE_NR);
2927 2928 2929 2930

#ifdef CONFIG_NET_NS
void sock_prot_inuse_add(struct net *net, struct proto *prot, int val)
{
E
Eric Dumazet 已提交
2931
	__this_cpu_add(net->core.inuse->val[prot->inuse_idx], val);
2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946
}
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);

2947
static int __net_init sock_inuse_init_net(struct net *net)
2948 2949 2950 2951 2952
{
	net->core.inuse = alloc_percpu(struct prot_inuse);
	return net->core.inuse ? 0 : -ENOMEM;
}

2953
static void __net_exit sock_inuse_exit_net(struct net *net)
2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972
{
	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
2973 2974
static DEFINE_PER_CPU(struct prot_inuse, prot_inuse);

2975
void sock_prot_inuse_add(struct net *net, struct proto *prot, int val)
2976
{
E
Eric Dumazet 已提交
2977
	__this_cpu_add(prot_inuse.val[prot->inuse_idx], val);
2978 2979 2980
}
EXPORT_SYMBOL_GPL(sock_prot_inuse_add);

2981
int sock_prot_inuse_get(struct net *net, struct proto *prot)
2982 2983 2984 2985 2986 2987 2988 2989 2990 2991
{
	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);
2992
#endif
2993 2994 2995 2996 2997 2998

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)) {
J
Joe Perches 已提交
2999
		pr_err("PROTO_INUSE_NR exhausted\n");
3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020
		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

3021 3022 3023 3024 3025 3026
static void req_prot_cleanup(struct request_sock_ops *rsk_prot)
{
	if (!rsk_prot)
		return;
	kfree(rsk_prot->slab_name);
	rsk_prot->slab_name = NULL;
3027 3028
	kmem_cache_destroy(rsk_prot->slab);
	rsk_prot->slab = NULL;
3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044
}

static int req_prot_init(const struct proto *prot)
{
	struct request_sock_ops *rsk_prot = prot->rsk_prot;

	if (!rsk_prot)
		return 0;

	rsk_prot->slab_name = kasprintf(GFP_KERNEL, "request_sock_%s",
					prot->name);
	if (!rsk_prot->slab_name)
		return -ENOMEM;

	rsk_prot->slab = kmem_cache_create(rsk_prot->slab_name,
					   rsk_prot->obj_size, 0,
3045
					   prot->slab_flags, NULL);
3046 3047 3048 3049 3050 3051 3052 3053 3054

	if (!rsk_prot->slab) {
		pr_crit("%s: Can't create request sock SLAB cache!\n",
			prot->name);
		return -ENOMEM;
	}
	return 0;
}

3055 3056
int proto_register(struct proto *prot, int alloc_slab)
{
L
Linus Torvalds 已提交
3057 3058
	if (alloc_slab) {
		prot->slab = kmem_cache_create(prot->name, prot->obj_size, 0,
3059 3060
					SLAB_HWCACHE_ALIGN | prot->slab_flags,
					NULL);
L
Linus Torvalds 已提交
3061 3062

		if (prot->slab == NULL) {
J
Joe Perches 已提交
3063 3064
			pr_crit("%s: Can't create sock SLAB cache!\n",
				prot->name);
3065
			goto out;
L
Linus Torvalds 已提交
3066
		}
3067

3068 3069
		if (req_prot_init(prot))
			goto out_free_request_sock_slab;
3070

3071
		if (prot->twsk_prot != NULL) {
3072
			prot->twsk_prot->twsk_slab_name = kasprintf(GFP_KERNEL, "tw_sock_%s", prot->name);
3073

3074
			if (prot->twsk_prot->twsk_slab_name == NULL)
3075 3076
				goto out_free_request_sock_slab;

3077
			prot->twsk_prot->twsk_slab =
3078
				kmem_cache_create(prot->twsk_prot->twsk_slab_name,
3079
						  prot->twsk_prot->twsk_obj_size,
3080
						  0,
3081
						  prot->slab_flags,
3082
						  NULL);
3083
			if (prot->twsk_prot->twsk_slab == NULL)
3084 3085
				goto out_free_timewait_sock_slab_name;
		}
L
Linus Torvalds 已提交
3086 3087
	}

3088
	mutex_lock(&proto_list_mutex);
L
Linus Torvalds 已提交
3089
	list_add(&prot->node, &proto_list);
3090
	assign_proto_idx(prot);
3091
	mutex_unlock(&proto_list_mutex);
3092 3093
	return 0;

3094
out_free_timewait_sock_slab_name:
3095
	kfree(prot->twsk_prot->twsk_slab_name);
3096
out_free_request_sock_slab:
3097 3098
	req_prot_cleanup(prot->rsk_prot);

3099 3100
	kmem_cache_destroy(prot->slab);
	prot->slab = NULL;
3101 3102
out:
	return -ENOBUFS;
L
Linus Torvalds 已提交
3103 3104 3105 3106 3107
}
EXPORT_SYMBOL(proto_register);

void proto_unregister(struct proto *prot)
{
3108
	mutex_lock(&proto_list_mutex);
3109
	release_proto_idx(prot);
3110
	list_del(&prot->node);
3111
	mutex_unlock(&proto_list_mutex);
L
Linus Torvalds 已提交
3112

3113 3114
	kmem_cache_destroy(prot->slab);
	prot->slab = NULL;
L
Linus Torvalds 已提交
3115

3116
	req_prot_cleanup(prot->rsk_prot);
3117

3118 3119
	if (prot->twsk_prot != NULL && prot->twsk_prot->twsk_slab != NULL) {
		kmem_cache_destroy(prot->twsk_prot->twsk_slab);
3120
		kfree(prot->twsk_prot->twsk_slab_name);
3121
		prot->twsk_prot->twsk_slab = NULL;
3122
	}
L
Linus Torvalds 已提交
3123 3124 3125 3126 3127
}
EXPORT_SYMBOL(proto_unregister);

#ifdef CONFIG_PROC_FS
static void *proto_seq_start(struct seq_file *seq, loff_t *pos)
3128
	__acquires(proto_list_mutex)
L
Linus Torvalds 已提交
3129
{
3130
	mutex_lock(&proto_list_mutex);
3131
	return seq_list_start_head(&proto_list, *pos);
L
Linus Torvalds 已提交
3132 3133 3134 3135
}

static void *proto_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
3136
	return seq_list_next(v, &proto_list, pos);
L
Linus Torvalds 已提交
3137 3138 3139
}

static void proto_seq_stop(struct seq_file *seq, void *v)
3140
	__releases(proto_list_mutex)
L
Linus Torvalds 已提交
3141
{
3142
	mutex_unlock(&proto_list_mutex);
L
Linus Torvalds 已提交
3143 3144 3145 3146 3147 3148
}

static char proto_method_implemented(const void *method)
{
	return method == NULL ? 'n' : 'y';
}
3149 3150
static long sock_prot_memory_allocated(struct proto *proto)
{
3151
	return proto->memory_allocated != NULL ? proto_memory_allocated(proto) : -1L;
3152 3153 3154 3155 3156 3157 3158
}

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
3159 3160 3161

static void proto_seq_printf(struct seq_file *seq, struct proto *proto)
{
3162

E
Eric Dumazet 已提交
3163
	seq_printf(seq, "%-9s %4u %6d  %6ld   %-3s %6u   %-3s  %-10s "
L
Linus Torvalds 已提交
3164 3165 3166
			"%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,
3167
		   sock_prot_inuse_get(seq_file_net(seq), proto),
3168 3169
		   sock_prot_memory_allocated(proto),
		   sock_prot_memory_pressure(proto),
L
Linus Torvalds 已提交
3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195
		   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)
{
3196
	if (v == &proto_list)
L
Linus Torvalds 已提交
3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207
		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
3208
		proto_seq_printf(seq, list_entry(v, struct proto, node));
L
Linus Torvalds 已提交
3209 3210 3211
	return 0;
}

3212
static const struct seq_operations proto_seq_ops = {
L
Linus Torvalds 已提交
3213 3214 3215 3216 3217 3218 3219 3220
	.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)
{
3221 3222
	return seq_open_net(inode, file, &proto_seq_ops,
			    sizeof(struct seq_net_private));
L
Linus Torvalds 已提交
3223 3224
}

3225
static const struct file_operations proto_seq_fops = {
L
Linus Torvalds 已提交
3226 3227 3228 3229
	.owner		= THIS_MODULE,
	.open		= proto_seq_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
3230 3231 3232 3233 3234
	.release	= seq_release_net,
};

static __net_init int proto_init_net(struct net *net)
{
3235
	if (!proc_create("protocols", S_IRUGO, net->proc_net, &proto_seq_fops))
3236 3237 3238 3239 3240 3241 3242
		return -ENOMEM;

	return 0;
}

static __net_exit void proto_exit_net(struct net *net)
{
3243
	remove_proc_entry("protocols", net->proc_net);
3244 3245 3246 3247 3248 3249
}


static __net_initdata struct pernet_operations proto_net_ops = {
	.init = proto_init_net,
	.exit = proto_exit_net,
L
Linus Torvalds 已提交
3250 3251 3252 3253
};

static int __init proto_init(void)
{
3254
	return register_pernet_subsys(&proto_net_ops);
L
Linus Torvalds 已提交
3255 3256 3257 3258 3259
}

subsys_initcall(proto_init);

#endif /* PROC_FS */
3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270

#ifdef CONFIG_NET_RX_BUSY_POLL
bool sk_busy_loop_end(void *p, unsigned long start_time)
{
	struct sock *sk = p;

	return !skb_queue_empty(&sk->sk_receive_queue) ||
	       sk_busy_loop_timeout(sk, start_time);
}
EXPORT_SYMBOL(sk_busy_loop_end);
#endif /* CONFIG_NET_RX_BUSY_POLL */