udp.c 37.3 KB
Newer Older
L
Linus Torvalds 已提交
1 2
/*
 *	UDP over IPv6
3
 *	Linux INET6 implementation
L
Linus Torvalds 已提交
4 5
 *
 *	Authors:
6
 *	Pedro Roque		<roque@di.fc.ul.pt>
L
Linus Torvalds 已提交
7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
 *
 *	Based on linux/ipv4/udp.c
 *
 *	Fixes:
 *	Hideaki YOSHIFUJI	:	sin6_scope_id support
 *	YOSHIFUJI Hideaki @USAGI and:	Support IPV6_V6ONLY socket option, which
 *	Alexey Kuznetsov		allow both IPv4 and IPv6 sockets to bind
 *					a single port at the same time.
 *      Kazunori MIYAZAWA @USAGI:       change process style to use ip6_append_data
 *      YOSHIFUJI Hideaki @USAGI:	convert /proc/net/udp6 to seq_file.
 *
 *	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.
 */

#include <linux/errno.h>
#include <linux/types.h>
#include <linux/socket.h>
#include <linux/sockios.h>
#include <linux/net.h>
#include <linux/in6.h>
#include <linux/netdevice.h>
#include <linux/if_arp.h>
#include <linux/ipv6.h>
#include <linux/icmpv6.h>
#include <linux/init.h>
35
#include <linux/module.h>
36
#include <linux/skbuff.h>
37
#include <linux/slab.h>
38
#include <linux/uaccess.h>
L
Linus Torvalds 已提交
39

40
#include <net/addrconf.h>
L
Linus Torvalds 已提交
41 42 43 44 45
#include <net/ndisc.h>
#include <net/protocol.h>
#include <net/transp_v6.h>
#include <net/ip6_route.h>
#include <net/raw.h>
46
#include <net/tcp_states.h>
L
Linus Torvalds 已提交
47 48
#include <net/ip6_checksum.h>
#include <net/xfrm.h>
49
#include <net/inet6_hashtables.h>
50
#include <net/busy_poll.h>
51
#include <net/sock_reuseport.h>
L
Linus Torvalds 已提交
52 53 54

#include <linux/proc_fs.h>
#include <linux/seq_file.h>
55
#include <trace/events/skb.h>
56
#include "udp_impl.h"
L
Linus Torvalds 已提交
57

58 59 60 61 62 63 64 65 66 67
static bool udp6_lib_exact_dif_match(struct net *net, struct sk_buff *skb)
{
#if defined(CONFIG_NET_L3_MASTER_DEV)
	if (!net->ipv4.sysctl_udp_l3mdev_accept &&
	    skb && ipv6_l3mdev_skb(IP6CB(skb)->flags))
		return true;
#endif
	return false;
}

68 69 70 71 72
static u32 udp6_ehashfn(const struct net *net,
			const struct in6_addr *laddr,
			const u16 lport,
			const struct in6_addr *faddr,
			const __be16 fport)
73
{
74 75 76 77 78 79 80 81 82 83 84 85 86
	static u32 udp6_ehash_secret __read_mostly;
	static u32 udp_ipv6_hash_secret __read_mostly;

	u32 lhash, fhash;

	net_get_random_once(&udp6_ehash_secret,
			    sizeof(udp6_ehash_secret));
	net_get_random_once(&udp_ipv6_hash_secret,
			    sizeof(udp_ipv6_hash_secret));

	lhash = (__force u32)laddr->s6_addr32[3];
	fhash = __ipv6_addr_jhash(faddr, udp_ipv6_hash_secret);

87
	return __inet6_ehashfn(lhash, lport, fhash, fport,
88
			       udp_ipv6_hash_secret + net_hash_mix(net));
89 90
}

91 92 93
static u32 udp6_portaddr_hash(const struct net *net,
			      const struct in6_addr *addr6,
			      unsigned int port)
94 95 96 97 98
{
	unsigned int hash, mix = net_hash_mix(net);

	if (ipv6_addr_any(addr6))
		hash = jhash_1word(0, mix);
B
Brian Haley 已提交
99
	else if (ipv6_addr_v4mapped(addr6))
100
		hash = jhash_1word((__force u32)addr6->s6_addr32[3], mix);
101
	else
102
		hash = jhash2((__force u32 *)addr6->s6_addr32, 4, mix);
103 104 105 106

	return hash ^ port;
}

107
int udp_v6_get_port(struct sock *sk, unsigned short snum)
L
Linus Torvalds 已提交
108
{
E
Eric Dumazet 已提交
109 110
	unsigned int hash2_nulladdr =
		udp6_portaddr_hash(sock_net(sk), &in6addr_any, snum);
111
	unsigned int hash2_partial =
112
		udp6_portaddr_hash(sock_net(sk), &sk->sk_v6_rcv_saddr, 0);
E
Eric Dumazet 已提交
113

114
	/* precompute partial secondary hash */
E
Eric Dumazet 已提交
115
	udp_sk(sk)->udp_portaddr_hash = hash2_partial;
116
	return udp_lib_get_port(sk, snum, hash2_nulladdr);
L
Linus Torvalds 已提交
117 118
}

E
Eric Dumazet 已提交
119 120 121
static void udp_v6_rehash(struct sock *sk)
{
	u16 new_hash = udp6_portaddr_hash(sock_net(sk),
122
					  &sk->sk_v6_rcv_saddr,
E
Eric Dumazet 已提交
123 124 125 126 127
					  inet_sk(sk)->inet_num);

	udp_lib_rehash(sk, new_hash);
}

128 129 130
static int compute_score(struct sock *sk, struct net *net,
			 const struct in6_addr *saddr, __be16 sport,
			 const struct in6_addr *daddr, unsigned short hnum,
131
			 int dif, bool exact_dif)
132
{
133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161
	int score;
	struct inet_sock *inet;

	if (!net_eq(sock_net(sk), net) ||
	    udp_sk(sk)->udp_port_hash != hnum ||
	    sk->sk_family != PF_INET6)
		return -1;

	score = 0;
	inet = inet_sk(sk);

	if (inet->inet_dport) {
		if (inet->inet_dport != sport)
			return -1;
		score++;
	}

	if (!ipv6_addr_any(&sk->sk_v6_rcv_saddr)) {
		if (!ipv6_addr_equal(&sk->sk_v6_rcv_saddr, daddr))
			return -1;
		score++;
	}

	if (!ipv6_addr_any(&sk->sk_v6_daddr)) {
		if (!ipv6_addr_equal(&sk->sk_v6_daddr, saddr))
			return -1;
		score++;
	}

162
	if (sk->sk_bound_dev_if || exact_dif) {
163 164 165
		if (sk->sk_bound_dev_if != dif)
			return -1;
		score++;
166
	}
167

168 169 170
	if (sk->sk_incoming_cpu == raw_smp_processor_id())
		score++;

171 172 173
	return score;
}

174
/* called with rcu_read_lock() */
E
Eric Dumazet 已提交
175 176 177
static struct sock *udp6_lib_lookup2(struct net *net,
		const struct in6_addr *saddr, __be16 sport,
		const struct in6_addr *daddr, unsigned int hnum, int dif,
178
		bool exact_dif, struct udp_hslot *hslot2,
179
		struct sk_buff *skb)
E
Eric Dumazet 已提交
180 181
{
	struct sock *sk, *result;
182 183
	int score, badness, matches = 0, reuseport = 0;
	u32 hash = 0;
E
Eric Dumazet 已提交
184 185 186

	result = NULL;
	badness = -1;
187
	udp_portaddr_for_each_entry_rcu(sk, &hslot2->head) {
188
		score = compute_score(sk, net, saddr, sport,
189
				      daddr, hnum, dif, exact_dif);
E
Eric Dumazet 已提交
190
		if (score > badness) {
191 192
			reuseport = sk->sk_reuseport;
			if (reuseport) {
193 194
				hash = udp6_ehashfn(net, daddr, hnum,
						    saddr, sport);
195

196
				result = reuseport_select_sock(sk, hash, skb,
197
							sizeof(struct udphdr));
198 199
				if (result)
					return result;
200
				matches = 1;
201
			}
202 203
			result = sk;
			badness = score;
204 205
		} else if (score == badness && reuseport) {
			matches++;
206
			if (reciprocal_scale(hash, matches) == 0)
207 208
				result = sk;
			hash = next_pseudo_random32(hash);
E
Eric Dumazet 已提交
209 210 211 212 213
		}
	}
	return result;
}

214
/* rcu_read_lock() must be held */
215
struct sock *__udp6_lib_lookup(struct net *net,
216 217
				      const struct in6_addr *saddr, __be16 sport,
				      const struct in6_addr *daddr, __be16 dport,
218 219
				      int dif, struct udp_table *udptable,
				      struct sk_buff *skb)
L
Linus Torvalds 已提交
220
{
221
	struct sock *sk, *result;
L
Linus Torvalds 已提交
222
	unsigned short hnum = ntohs(dport);
E
Eric Dumazet 已提交
223 224
	unsigned int hash2, slot2, slot = udp_hashfn(net, hnum, udptable->mask);
	struct udp_hslot *hslot2, *hslot = &udptable->hash[slot];
225
	bool exact_dif = udp6_lib_exact_dif_match(net, skb);
226 227
	int score, badness, matches = 0, reuseport = 0;
	u32 hash = 0;
228

E
Eric Dumazet 已提交
229 230 231 232 233 234 235 236
	if (hslot->count > 10) {
		hash2 = udp6_portaddr_hash(net, daddr, hnum);
		slot2 = hash2 & udptable->mask;
		hslot2 = &udptable->hash2[slot2];
		if (hslot->count < hslot2->count)
			goto begin;

		result = udp6_lib_lookup2(net, saddr, sport,
237
					  daddr, hnum, dif, exact_dif,
238
					  hslot2, skb);
E
Eric Dumazet 已提交
239
		if (!result) {
240
			unsigned int old_slot2 = slot2;
E
Eric Dumazet 已提交
241 242
			hash2 = udp6_portaddr_hash(net, &in6addr_any, hnum);
			slot2 = hash2 & udptable->mask;
243 244 245 246
			/* avoid searching the same slot again. */
			if (unlikely(slot2 == old_slot2))
				return result;

E
Eric Dumazet 已提交
247 248 249 250
			hslot2 = &udptable->hash2[slot2];
			if (hslot->count < hslot2->count)
				goto begin;

251
			result = udp6_lib_lookup2(net, saddr, sport,
252
						  daddr, hnum, dif,
253 254
						  exact_dif, hslot2,
						  skb);
E
Eric Dumazet 已提交
255 256 257
		}
		return result;
	}
258 259 260
begin:
	result = NULL;
	badness = -1;
261
	sk_for_each_rcu(sk, &hslot->head) {
262 263
		score = compute_score(sk, net, saddr, sport, daddr, hnum, dif,
				      exact_dif);
264
		if (score > badness) {
265 266
			reuseport = sk->sk_reuseport;
			if (reuseport) {
267 268
				hash = udp6_ehashfn(net, daddr, hnum,
						    saddr, sport);
269
				result = reuseport_select_sock(sk, hash, skb,
270
							sizeof(struct udphdr));
271 272
				if (result)
					return result;
273 274
				matches = 1;
			}
275 276
			result = sk;
			badness = score;
277 278
		} else if (score == badness && reuseport) {
			matches++;
279
			if (reciprocal_scale(hash, matches) == 0)
280 281
				result = sk;
			hash = next_pseudo_random32(hash);
L
Linus Torvalds 已提交
282 283 284 285
		}
	}
	return result;
}
286
EXPORT_SYMBOL_GPL(__udp6_lib_lookup);
L
Linus Torvalds 已提交
287

288 289
static struct sock *__udp6_lib_lookup_skb(struct sk_buff *skb,
					  __be16 sport, __be16 dport,
290
					  struct udp_table *udptable)
291
{
292
	const struct ipv6hdr *iph = ipv6_hdr(skb);
293
	struct sock *sk;
294

295 296
	sk = skb_steal_sock(skb);
	if (unlikely(sk))
297
		return sk;
298
	return __udp6_lib_lookup(dev_net(skb->dev), &iph->saddr, sport,
E
Eric Dumazet 已提交
299
				 &iph->daddr, dport, inet6_iif(skb),
300
				 udptable, skb);
301 302
}

303 304 305 306 307
struct sock *udp6_lib_lookup_skb(struct sk_buff *skb,
				 __be16 sport, __be16 dport)
{
	const struct ipv6hdr *iph = ipv6_hdr(skb);

308
	return __udp6_lib_lookup(dev_net(skb->dev), &iph->saddr, sport,
309 310 311 312 313
				 &iph->daddr, dport, inet6_iif(skb),
				 &udp_table, skb);
}
EXPORT_SYMBOL_GPL(udp6_lib_lookup_skb);

314 315 316 317
/* Must be called under rcu_read_lock().
 * Does increment socket refcount.
 */
#if IS_ENABLED(CONFIG_NETFILTER_XT_MATCH_SOCKET) || \
318 319
    IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TPROXY) || \
    IS_ENABLED(CONFIG_NF_SOCKET_IPV6)
320 321 322
struct sock *udp6_lib_lookup(struct net *net, const struct in6_addr *saddr, __be16 sport,
			     const struct in6_addr *daddr, __be16 dport, int dif)
{
323 324 325 326 327 328 329
	struct sock *sk;

	sk =  __udp6_lib_lookup(net, saddr, sport, daddr, dport,
				dif, &udp_table, NULL);
	if (sk && !atomic_inc_not_zero(&sk->sk_refcnt))
		sk = NULL;
	return sk;
330 331
}
EXPORT_SYMBOL_GPL(udp6_lib_lookup);
332
#endif
333

L
Linus Torvalds 已提交
334
/*
335 336
 *	This should be easy, if there is something there we
 *	return it, otherwise we block.
L
Linus Torvalds 已提交
337 338
 */

339
int udpv6_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
L
Linus Torvalds 已提交
340 341 342 343
		  int noblock, int flags, int *addr_len)
{
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct inet_sock *inet = inet_sk(sk);
344
	struct sk_buff *skb;
345
	unsigned int ulen, copied;
346
	int peeked, peeking, off;
347 348
	int err;
	int is_udplite = IS_UDPLITE(sk);
349
	bool checksum_valid = false;
350
	int is_udp4;
L
Linus Torvalds 已提交
351 352

	if (flags & MSG_ERRQUEUE)
353
		return ipv6_recv_error(sk, msg, len, addr_len);
L
Linus Torvalds 已提交
354

355
	if (np->rxpmtu && np->rxopt.bits.rxpmtu)
356
		return ipv6_recv_rxpmtu(sk, msg, len, addr_len);
357

L
Linus Torvalds 已提交
358
try_again:
359
	peeking = off = sk_peek_offset(sk, flags);
360
	skb = __skb_recv_udp(sk, flags, noblock, &peeked, &off, &err);
L
Linus Torvalds 已提交
361
	if (!skb)
362
		return err;
L
Linus Torvalds 已提交
363

364
	ulen = skb->len;
365
	copied = len;
366 367
	if (copied > ulen - off)
		copied = ulen - off;
368
	else if (copied < ulen)
369
		msg->msg_flags |= MSG_TRUNC;
L
Linus Torvalds 已提交
370

371 372
	is_udp4 = (skb->protocol == htons(ETH_P_IP));

373
	/*
374 375 376
	 * If checksum is needed at all, try to do it while copying the
	 * data.  If the data is truncated, or if we only want a partial
	 * coverage checksum (UDP-Lite), do it before the copy.
377 378
	 */

379 380
	if (copied < ulen || peeking ||
	    (is_udplite && UDP_SKB_CB(skb)->partial_cov)) {
381 382
		checksum_valid = !udp_lib_checksum_complete(skb);
		if (!checksum_valid)
L
Linus Torvalds 已提交
383
			goto csum_copy_err;
384 385
	}

386
	if (checksum_valid || skb_csum_unnecessary(skb))
387
		err = skb_copy_datagram_msg(skb, off, msg, copied);
388
	else {
389
		err = skb_copy_and_csum_datagram_msg(skb, off, msg);
L
Linus Torvalds 已提交
390 391 392
		if (err == -EINVAL)
			goto csum_copy_err;
	}
393
	if (unlikely(err)) {
394 395 396
		if (!peeked) {
			atomic_inc(&sk->sk_drops);
			if (is_udp4)
397 398
				UDP_INC_STATS(sock_net(sk), UDP_MIB_INERRORS,
					      is_udplite);
399
			else
400 401
				UDP6_INC_STATS(sock_net(sk), UDP_MIB_INERRORS,
					       is_udplite);
402
		}
403
		kfree_skb(skb);
404
		return err;
405
	}
406 407
	if (!peeked) {
		if (is_udp4)
408 409
			UDP_INC_STATS(sock_net(sk), UDP_MIB_INDATAGRAMS,
				      is_udplite);
410
		else
411 412
			UDP6_INC_STATS(sock_net(sk), UDP_MIB_INDATAGRAMS,
				       is_udplite);
413
	}
414

415
	sock_recv_ts_and_drops(msg, sk, skb);
L
Linus Torvalds 已提交
416 417 418

	/* Copy the address. */
	if (msg->msg_name) {
419
		DECLARE_SOCKADDR(struct sockaddr_in6 *, sin6, msg->msg_name);
L
Linus Torvalds 已提交
420
		sin6->sin6_family = AF_INET6;
421
		sin6->sin6_port = udp_hdr(skb)->source;
L
Linus Torvalds 已提交
422 423
		sin6->sin6_flowinfo = 0;

424
		if (is_udp4) {
B
Brian Haley 已提交
425 426
			ipv6_addr_set_v4mapped(ip_hdr(skb)->saddr,
					       &sin6->sin6_addr);
427 428
			sin6->sin6_scope_id = 0;
		} else {
A
Alexey Dobriyan 已提交
429
			sin6->sin6_addr = ipv6_hdr(skb)->saddr;
430 431
			sin6->sin6_scope_id =
				ipv6_iface_scope_id(&sin6->sin6_addr,
432
						    inet6_iif(skb));
L
Linus Torvalds 已提交
433
		}
434
		*addr_len = sizeof(*sin6);
L
Linus Torvalds 已提交
435
	}
436 437 438 439

	if (np->rxopt.all)
		ip6_datagram_recv_common_ctl(sk, msg, skb);

440
	if (is_udp4) {
L
Linus Torvalds 已提交
441
		if (inet->cmsg_flags)
442
			ip_cmsg_recv_offset(msg, sk, skb,
E
Eric Dumazet 已提交
443
					    sizeof(struct udphdr), off);
L
Linus Torvalds 已提交
444 445
	} else {
		if (np->rxopt.all)
446
			ip6_datagram_recv_specific_ctl(sk, msg, skb);
447
	}
L
Linus Torvalds 已提交
448

449
	err = copied;
L
Linus Torvalds 已提交
450
	if (flags & MSG_TRUNC)
451
		err = ulen;
L
Linus Torvalds 已提交
452

453
	skb_consume_udp(sk, skb, peeking ? -err : err);
L
Linus Torvalds 已提交
454 455 456
	return err;

csum_copy_err:
457
	if (!__sk_queue_drop_skb(sk, skb, flags, udp_skb_destructor)) {
458
		if (is_udp4) {
459 460 461 462
			UDP_INC_STATS(sock_net(sk),
				      UDP_MIB_CSUMERRORS, is_udplite);
			UDP_INC_STATS(sock_net(sk),
				      UDP_MIB_INERRORS, is_udplite);
463
		} else {
464 465 466 467
			UDP6_INC_STATS(sock_net(sk),
				       UDP_MIB_CSUMERRORS, is_udplite);
			UDP6_INC_STATS(sock_net(sk),
				       UDP_MIB_INERRORS, is_udplite);
468
		}
469
	}
470
	kfree_skb(skb);
L
Linus Torvalds 已提交
471

472 473
	/* starting over for a new packet, but check if we need to yield */
	cond_resched();
474
	msg->msg_flags &= ~MSG_TRUNC;
L
Linus Torvalds 已提交
475 476 477
	goto try_again;
}

478
void __udp6_lib_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
479
		    u8 type, u8 code, int offset, __be32 info,
480
		    struct udp_table *udptable)
L
Linus Torvalds 已提交
481 482
{
	struct ipv6_pinfo *np;
483 484 485
	const struct ipv6hdr *hdr = (const struct ipv6hdr *)skb->data;
	const struct in6_addr *saddr = &hdr->saddr;
	const struct in6_addr *daddr = &hdr->daddr;
486
	struct udphdr *uh = (struct udphdr *)(skb->data+offset);
L
Linus Torvalds 已提交
487
	struct sock *sk;
488
	int harderr;
L
Linus Torvalds 已提交
489
	int err;
490
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
491

492
	sk = __udp6_lib_lookup(net, daddr, uh->dest, saddr, uh->source,
493
			       inet6_iif(skb), udptable, skb);
494
	if (!sk) {
E
Eric Dumazet 已提交
495 496
		__ICMP6_INC_STATS(net, __in6_dev_get(skb->dev),
				  ICMP6_MIB_INERRORS);
L
Linus Torvalds 已提交
497
		return;
498
	}
L
Linus Torvalds 已提交
499

500 501 502
	harderr = icmpv6_err_convert(type, code, &err);
	np = inet6_sk(sk);

503 504 505
	if (type == ICMPV6_PKT_TOOBIG) {
		if (!ip6_sk_accept_pmtu(sk))
			goto out;
506
		ip6_sk_update_pmtu(skb, sk, info);
507 508
		if (np->pmtudisc != IPV6_PMTUDISC_DONT)
			harderr = 1;
509
	}
510
	if (type == NDISC_REDIRECT) {
511
		ip6_sk_redirect(skb, sk);
512 513
		goto out;
	}
514

515 516 517 518
	if (!np->recverr) {
		if (!harderr || sk->sk_state != TCP_ESTABLISHED)
			goto out;
	} else {
L
Linus Torvalds 已提交
519
		ipv6_icmp_error(sk, skb, err, uh->dest, ntohl(info), (u8 *)(uh+1));
520
	}
521

L
Linus Torvalds 已提交
522 523 524
	sk->sk_err = err;
	sk->sk_error_report(sk);
out:
525
	return;
L
Linus Torvalds 已提交
526 527
}

528
int __udpv6_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
529 530 531
{
	int rc;

532
	if (!ipv6_addr_any(&sk->sk_v6_daddr)) {
533
		sock_rps_save_rxhash(sk, skb);
534
		sk_mark_napi_id(sk, skb);
E
Eric Dumazet 已提交
535
		sk_incoming_cpu_update(sk);
536 537
	} else {
		sk_mark_napi_id_once(sk, skb);
538
	}
539

540
	rc = __udp_enqueue_schedule_skb(sk, skb);
541 542 543 544 545
	if (rc < 0) {
		int is_udplite = IS_UDPLITE(sk);

		/* Note that an ENOMEM error is charged twice */
		if (rc == -ENOMEM)
546
			UDP6_INC_STATS(sock_net(sk),
547
					 UDP_MIB_RCVBUFERRORS, is_udplite);
548
		UDP6_INC_STATS(sock_net(sk), UDP_MIB_INERRORS, is_udplite);
549 550 551
		kfree_skb(skb);
		return -1;
	}
552

553 554 555
	return 0;
}

556
static __inline__ void udpv6_err(struct sk_buff *skb,
557
				 struct inet6_skb_parm *opt, u8 type,
558
				 u8 code, int offset, __be32 info)
559
{
560
	__udp6_lib_err(skb, opt, type, code, offset, info, &udp_table);
561 562
}

563 564 565 566 567 568 569 570
static struct static_key udpv6_encap_needed __read_mostly;
void udpv6_encap_enable(void)
{
	if (!static_key_enabled(&udpv6_encap_needed))
		static_key_slow_inc(&udpv6_encap_needed);
}
EXPORT_SYMBOL(udpv6_encap_enable);

571
int udpv6_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
L
Linus Torvalds 已提交
572
{
573
	struct udp_sock *up = udp_sk(sk);
W
Wang Chen 已提交
574
	int is_udplite = IS_UDPLITE(sk);
575

576 577
	if (!xfrm6_policy_check(sk, XFRM_POLICY_IN, skb))
		goto drop;
L
Linus Torvalds 已提交
578

579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594
	if (static_key_false(&udpv6_encap_needed) && up->encap_type) {
		int (*encap_rcv)(struct sock *sk, struct sk_buff *skb);

		/*
		 * This is an encapsulation socket so pass the skb to
		 * the socket's udp_encap_rcv() hook. Otherwise, just
		 * fall through and pass this up the UDP socket.
		 * up->encap_rcv() returns the following value:
		 * =0 if skb was successfully passed to the encap
		 *    handler or was discarded by it.
		 * >0 if skb should be passed on to UDP.
		 * <0 if skb should be resubmitted as proto -N
		 */

		/* if we're overly short, let UDP handle it */
		encap_rcv = ACCESS_ONCE(up->encap_rcv);
595
		if (encap_rcv) {
596 597
			int ret;

598 599 600 601
			/* Verify checksum before giving to encap */
			if (udp_lib_checksum_complete(skb))
				goto csum_error;

602 603
			ret = encap_rcv(sk, skb);
			if (ret <= 0) {
604 605 606
				__UDP_INC_STATS(sock_net(sk),
						UDP_MIB_INDATAGRAMS,
						is_udplite);
607 608 609 610 611 612 613
				return -ret;
			}
		}

		/* FALLTHROUGH -- it's a UDP Packet */
	}

614 615 616
	/*
	 * UDP-Lite specific tests, ignored on UDP sockets (see net/ipv4/udp.c).
	 */
W
Wang Chen 已提交
617
	if ((is_udplite & UDPLITE_RECV_CC)  &&  UDP_SKB_CB(skb)->partial_cov) {
618 619

		if (up->pcrlen == 0) {          /* full coverage was set  */
620 621
			net_dbg_ratelimited("UDPLITE6: partial coverage %d while full coverage %d requested\n",
					    UDP_SKB_CB(skb)->cscov, skb->len);
622 623 624
			goto drop;
		}
		if (UDP_SKB_CB(skb)->cscov  <  up->pcrlen) {
625 626
			net_dbg_ratelimited("UDPLITE6: coverage %d too small, need min %d\n",
					    UDP_SKB_CB(skb)->cscov, up->pcrlen);
627 628
			goto drop;
		}
L
Linus Torvalds 已提交
629 630
	}

631 632 633 634
	if (rcu_access_pointer(sk->sk_filter) &&
	    udp_lib_checksum_complete(skb))
		goto csum_error;

635
	if (sk_filter_trim_cap(sk, skb, sizeof(struct udphdr)))
636
		goto drop;
637

638
	udp_csum_pull_header(skb);
639

640
	skb_dst_drop(skb);
641

642
	return __udpv6_queue_rcv_skb(sk, skb);
643

644
csum_error:
645
	__UDP6_INC_STATS(sock_net(sk), UDP_MIB_CSUMERRORS, is_udplite);
646
drop:
647
	__UDP6_INC_STATS(sock_net(sk), UDP_MIB_INERRORS, is_udplite);
648
	atomic_inc(&sk->sk_drops);
649 650
	kfree_skb(skb);
	return -1;
L
Linus Torvalds 已提交
651 652
}

653 654 655 656
static bool __udp_v6_is_mcast_sock(struct net *net, struct sock *sk,
				   __be16 loc_port, const struct in6_addr *loc_addr,
				   __be16 rmt_port, const struct in6_addr *rmt_addr,
				   int dif, unsigned short hnum)
L
Linus Torvalds 已提交
657
{
658
	struct inet_sock *inet = inet_sk(sk);
L
Linus Torvalds 已提交
659

660 661 662 663 664 665 666 667
	if (!net_eq(sock_net(sk), net))
		return false;

	if (udp_sk(sk)->udp_port_hash != hnum ||
	    sk->sk_family != PF_INET6 ||
	    (inet->inet_dport && inet->inet_dport != rmt_port) ||
	    (!ipv6_addr_any(&sk->sk_v6_daddr) &&
		    !ipv6_addr_equal(&sk->sk_v6_daddr, rmt_addr)) ||
668 669 670
	    (sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif) ||
	    (!ipv6_addr_any(&sk->sk_v6_rcv_saddr) &&
		    !ipv6_addr_equal(&sk->sk_v6_rcv_saddr, loc_addr)))
671 672 673 674
		return false;
	if (!inet6_mc_check(sk, loc_addr, rmt_addr))
		return false;
	return true;
L
Linus Torvalds 已提交
675 676
}

677 678 679 680 681
static void udp6_csum_zero_error(struct sk_buff *skb)
{
	/* RFC 2460 section 8.1 says that we SHOULD log
	 * this error. Well, it is reasonable.
	 */
682 683 684
	net_dbg_ratelimited("IPv6: udp checksum is 0 for [%pI6c]:%u->[%pI6c]:%u\n",
			    &ipv6_hdr(skb)->saddr, ntohs(udp_hdr(skb)->source),
			    &ipv6_hdr(skb)->daddr, ntohs(udp_hdr(skb)->dest));
685 686
}

L
Linus Torvalds 已提交
687 688 689 690
/*
 * Note: called only from the BH handler context,
 * so we don't need to lock the hashes.
 */
691
static int __udp6_lib_mcast_deliver(struct net *net, struct sk_buff *skb,
692
		const struct in6_addr *saddr, const struct in6_addr *daddr,
693
		struct udp_table *udptable, int proto)
L
Linus Torvalds 已提交
694
{
695
	struct sock *sk, *first = NULL;
696
	const struct udphdr *uh = udp_hdr(skb);
697 698
	unsigned short hnum = ntohs(uh->dest);
	struct udp_hslot *hslot = udp_hashslot(udptable, net, hnum);
699
	unsigned int offset = offsetof(typeof(*sk), sk_node);
700
	unsigned int hash2 = 0, hash2_any = 0, use_hash2 = (hslot->count > 10);
701 702 703
	int dif = inet6_iif(skb);
	struct hlist_node *node;
	struct sk_buff *nskb;
704 705 706

	if (use_hash2) {
		hash2_any = udp6_portaddr_hash(net, &in6addr_any, hnum) &
707 708
			    udptable->mask;
		hash2 = udp6_portaddr_hash(net, daddr, hnum) & udptable->mask;
709
start_lookup:
710
		hslot = &udptable->hash2[hash2];
711 712
		offset = offsetof(typeof(*sk), __sk_common.skc_portaddr_node);
	}
L
Linus Torvalds 已提交
713

714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729
	sk_for_each_entry_offset_rcu(sk, node, &hslot->head, offset) {
		if (!__udp_v6_is_mcast_sock(net, sk, uh->dest, daddr,
					    uh->source, saddr, dif, hnum))
			continue;
		/* If zero checksum and no_check is not on for
		 * the socket then skip it.
		 */
		if (!uh->check && !udp_sk(sk)->no_check6_rx)
			continue;
		if (!first) {
			first = sk;
			continue;
		}
		nskb = skb_clone(skb, GFP_ATOMIC);
		if (unlikely(!nskb)) {
			atomic_inc(&sk->sk_drops);
730 731 732 733
			__UDP6_INC_STATS(net, UDP_MIB_RCVBUFERRORS,
					 IS_UDPLITE(sk));
			__UDP6_INC_STATS(net, UDP_MIB_INERRORS,
					 IS_UDPLITE(sk));
734
			continue;
H
Hideo Aoki 已提交
735
		}
736

737 738 739
		if (udpv6_queue_rcv_skb(sk, nskb) > 0)
			consume_skb(nskb);
	}
740

741 742 743 744 745 746
	/* Also lookup *:port if we are using hash2 and haven't done so yet. */
	if (use_hash2 && hash2 != hash2_any) {
		hash2 = hash2_any;
		goto start_lookup;
	}

747 748 749
	if (first) {
		if (udpv6_queue_rcv_skb(first, skb) > 0)
			consume_skb(skb);
750
	} else {
751
		kfree_skb(skb);
752 753
		__UDP6_INC_STATS(net, UDP_MIB_IGNOREDMULTI,
				 proto == IPPROTO_UDPLITE);
754
	}
755
	return 0;
L
Linus Torvalds 已提交
756 757
}

758
int __udp6_lib_rcv(struct sk_buff *skb, struct udp_table *udptable,
759
		   int proto)
L
Linus Torvalds 已提交
760
{
761
	const struct in6_addr *saddr, *daddr;
762
	struct net *net = dev_net(skb->dev);
763
	struct udphdr *uh;
764
	struct sock *sk;
L
Linus Torvalds 已提交
765 766 767
	u32 ulen = 0;

	if (!pskb_may_pull(skb, sizeof(struct udphdr)))
768
		goto discard;
L
Linus Torvalds 已提交
769

770 771
	saddr = &ipv6_hdr(skb)->saddr;
	daddr = &ipv6_hdr(skb)->daddr;
772
	uh = udp_hdr(skb);
L
Linus Torvalds 已提交
773 774

	ulen = ntohs(uh->len);
775 776
	if (ulen > skb->len)
		goto short_packet;
L
Linus Torvalds 已提交
777

778 779
	if (proto == IPPROTO_UDP) {
		/* UDP validates ulen. */
L
Linus Torvalds 已提交
780

781 782 783
		/* Check for jumbo payload */
		if (ulen == 0)
			ulen = skb->len;
L
Linus Torvalds 已提交
784

785 786
		if (ulen < sizeof(*uh))
			goto short_packet;
L
Linus Torvalds 已提交
787

788 789 790
		if (ulen < skb->len) {
			if (pskb_trim_rcsum(skb, ulen))
				goto short_packet;
791 792
			saddr = &ipv6_hdr(skb)->saddr;
			daddr = &ipv6_hdr(skb)->daddr;
793
			uh = udp_hdr(skb);
794 795
		}
	}
L
Linus Torvalds 已提交
796

797
	if (udp6_csum_init(skb, uh, proto))
798
		goto csum_error;
799

800 801
	/*
	 *	Multicast receive code
L
Linus Torvalds 已提交
802
	 */
803
	if (ipv6_addr_is_multicast(daddr))
804
		return __udp6_lib_mcast_deliver(net, skb,
805
				saddr, daddr, udptable, proto);
L
Linus Torvalds 已提交
806 807 808

	/* Unicast */

809
	/*
L
Linus Torvalds 已提交
810 811 812
	 * check socket cache ... must talk to Alan about his plans
	 * for sock caches... i'll skip this for now.
	 */
813
	sk = __udp6_lib_lookup_skb(skb, uh->source, uh->dest, udptable);
814
	if (sk) {
815 816
		int ret;

817
		if (!uh->check && !udp_sk(sk)->no_check6_rx) {
818 819 820 821
			udp6_csum_zero_error(skb);
			goto csum_error;
		}

822
		if (inet_get_convert_csum(sk) && uh->check && !IS_UDPLITE(sk))
823 824 825
			skb_checksum_try_convert(skb, IPPROTO_UDP, uh->check,
						 ip6_compute_pseudo);

826
		ret = udpv6_queue_rcv_skb(sk, skb);
L
Linus Torvalds 已提交
827

828
		/* a return value > 0 means to resubmit the input */
829
		if (ret > 0)
830
			return ret;
L
Linus Torvalds 已提交
831

S
Stephen Hemminger 已提交
832
		return 0;
L
Linus Torvalds 已提交
833
	}
834

835 836 837 838 839
	if (!uh->check) {
		udp6_csum_zero_error(skb);
		goto csum_error;
	}

840
	if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb))
841
		goto discard;
842 843

	if (udp_lib_checksum_complete(skb))
844
		goto csum_error;
845

846
	__UDP6_INC_STATS(net, UDP_MIB_NOPORTS, proto == IPPROTO_UDPLITE);
847 848 849
	icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_PORT_UNREACH, 0);

	kfree_skb(skb);
S
Stephen Hemminger 已提交
850
	return 0;
L
Linus Torvalds 已提交
851

852
short_packet:
853 854 855 856 857
	net_dbg_ratelimited("UDP%sv6: short packet: From [%pI6c]:%u %d/%d to [%pI6c]:%u\n",
			    proto == IPPROTO_UDPLITE ? "-Lite" : "",
			    saddr, ntohs(uh->source),
			    ulen, skb->len,
			    daddr, ntohs(uh->dest));
858 859
	goto discard;
csum_error:
860
	__UDP6_INC_STATS(net, UDP_MIB_CSUMERRORS, proto == IPPROTO_UDPLITE);
L
Linus Torvalds 已提交
861
discard:
862
	__UDP6_INC_STATS(net, UDP_MIB_INERRORS, proto == IPPROTO_UDPLITE);
L
Linus Torvalds 已提交
863
	kfree_skb(skb);
S
Stephen Hemminger 已提交
864
	return 0;
L
Linus Torvalds 已提交
865
}
866

867
static __inline__ int udpv6_rcv(struct sk_buff *skb)
868
{
869
	return __udp6_lib_rcv(skb, &udp_table, IPPROTO_UDP);
870 871
}

L
Linus Torvalds 已提交
872 873 874 875 876 877 878
/*
 * Throw away all pending data and cancel the corking. Socket is locked.
 */
static void udp_v6_flush_pending_frames(struct sock *sk)
{
	struct udp_sock *up = udp_sk(sk);

879 880 881
	if (up->pending == AF_INET)
		udp_flush_pending_frames(sk);
	else if (up->pending) {
L
Linus Torvalds 已提交
882 883 884
		up->len = 0;
		up->pending = 0;
		ip6_flush_pending_frames(sk);
885
	}
L
Linus Torvalds 已提交
886 887
}

888
/**
889 890 891 892
 *	udp6_hwcsum_outgoing  -  handle outgoing HW checksumming
 *	@sk:	socket we are sending on
 *	@skb:	sk_buff containing the filled-in UDP header
 *		(checksum field must be zeroed out)
893 894 895 896 897 898 899
 */
static void udp6_hwcsum_outgoing(struct sock *sk, struct sk_buff *skb,
				 const struct in6_addr *saddr,
				 const struct in6_addr *daddr, int len)
{
	unsigned int offset;
	struct udphdr *uh = udp_hdr(skb);
V
Vlad Yasevich 已提交
900
	struct sk_buff *frags = skb_shinfo(skb)->frag_list;
901 902
	__wsum csum = 0;

V
Vlad Yasevich 已提交
903
	if (!frags) {
904 905 906 907 908 909 910 911 912 913 914 915 916 917 918
		/* Only one fragment on the socket.  */
		skb->csum_start = skb_transport_header(skb) - skb->head;
		skb->csum_offset = offsetof(struct udphdr, check);
		uh->check = ~csum_ipv6_magic(saddr, daddr, len, IPPROTO_UDP, 0);
	} else {
		/*
		 * HW-checksum won't work as there are two or more
		 * fragments on the socket so that all csums of sk_buffs
		 * should be together
		 */
		offset = skb_transport_offset(skb);
		skb->csum = skb_checksum(skb, offset, skb->len - offset, 0);

		skb->ip_summed = CHECKSUM_NONE;

V
Vlad Yasevich 已提交
919 920 921
		do {
			csum = csum_add(csum, frags->csum);
		} while ((frags = frags->next));
922 923 924 925 926 927 928 929

		uh->check = csum_ipv6_magic(saddr, daddr, len, IPPROTO_UDP,
					    csum);
		if (uh->check == 0)
			uh->check = CSUM_MANGLED_0;
	}
}

L
Linus Torvalds 已提交
930 931 932 933
/*
 *	Sending
 */

V
Vlad Yasevich 已提交
934
static int udp_v6_send_skb(struct sk_buff *skb, struct flowi6 *fl6)
L
Linus Torvalds 已提交
935
{
V
Vlad Yasevich 已提交
936
	struct sock *sk = skb->sk;
L
Linus Torvalds 已提交
937 938
	struct udphdr *uh;
	int err = 0;
W
Wang Chen 已提交
939
	int is_udplite = IS_UDPLITE(sk);
940
	__wsum csum = 0;
V
Vlad Yasevich 已提交
941 942
	int offset = skb_transport_offset(skb);
	int len = skb->len - offset;
L
Linus Torvalds 已提交
943 944 945 946

	/*
	 * Create a UDP header
	 */
947
	uh = udp_hdr(skb);
948 949
	uh->source = fl6->fl6_sport;
	uh->dest = fl6->fl6_dport;
V
Vlad Yasevich 已提交
950
	uh->len = htons(len);
L
Linus Torvalds 已提交
951 952
	uh->check = 0;

W
Wang Chen 已提交
953
	if (is_udplite)
V
Vlad Yasevich 已提交
954 955
		csum = udplite_csum(skb);
	else if (udp_sk(sk)->no_check6_tx) {   /* UDP csum disabled */
956 957 958
		skb->ip_summed = CHECKSUM_NONE;
		goto send;
	} else if (skb->ip_summed == CHECKSUM_PARTIAL) { /* UDP hardware csum */
V
Vlad Yasevich 已提交
959
		udp6_hwcsum_outgoing(sk, skb, &fl6->saddr, &fl6->daddr, len);
960 961
		goto send;
	} else
V
Vlad Yasevich 已提交
962
		csum = udp_csum(skb);
L
Linus Torvalds 已提交
963

964
	/* add protocol-dependent pseudo-header */
965
	uh->check = csum_ipv6_magic(&fl6->saddr, &fl6->daddr,
V
Vlad Yasevich 已提交
966
				    len, fl6->flowi6_proto, csum);
L
Linus Torvalds 已提交
967
	if (uh->check == 0)
968
		uh->check = CSUM_MANGLED_0;
L
Linus Torvalds 已提交
969

970
send:
V
Vlad Yasevich 已提交
971
	err = ip6_send_skb(skb);
E
Eric Dumazet 已提交
972 973
	if (err) {
		if (err == -ENOBUFS && !inet6_sk(sk)->recverr) {
974 975
			UDP6_INC_STATS(sock_net(sk),
				       UDP_MIB_SNDBUFERRORS, is_udplite);
E
Eric Dumazet 已提交
976 977
			err = 0;
		}
978 979 980 981
	} else {
		UDP6_INC_STATS(sock_net(sk),
			       UDP_MIB_OUTDATAGRAMS, is_udplite);
	}
V
Vlad Yasevich 已提交
982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005
	return err;
}

static int udp_v6_push_pending_frames(struct sock *sk)
{
	struct sk_buff *skb;
	struct udp_sock  *up = udp_sk(sk);
	struct flowi6 fl6;
	int err = 0;

	if (up->pending == AF_INET)
		return udp_push_pending_frames(sk);

	/* ip6_finish_skb will release the cork, so make a copy of
	 * fl6 here.
	 */
	fl6 = inet_sk(sk)->cork.fl.u.ip6;

	skb = ip6_finish_skb(sk);
	if (!skb)
		goto out;

	err = udp_v6_send_skb(skb, &fl6);

L
Linus Torvalds 已提交
1006 1007 1008 1009 1010 1011
out:
	up->len = 0;
	up->pending = 0;
	return err;
}

1012
int udpv6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
L
Linus Torvalds 已提交
1013 1014 1015 1016 1017
{
	struct ipv6_txoptions opt_space;
	struct udp_sock *up = udp_sk(sk);
	struct inet_sock *inet = inet_sk(sk);
	struct ipv6_pinfo *np = inet6_sk(sk);
1018
	DECLARE_SOCKADDR(struct sockaddr_in6 *, sin6, msg->msg_name);
1019
	struct in6_addr *daddr, *final_p, final;
L
Linus Torvalds 已提交
1020
	struct ipv6_txoptions *opt = NULL;
1021
	struct ipv6_txoptions *opt_to_free = NULL;
L
Linus Torvalds 已提交
1022
	struct ip6_flowlabel *flowlabel = NULL;
1023
	struct flowi6 fl6;
L
Linus Torvalds 已提交
1024
	struct dst_entry *dst;
W
Wei Wang 已提交
1025
	struct ipcm6_cookie ipc6;
L
Linus Torvalds 已提交
1026 1027 1028 1029
	int addr_len = msg->msg_namelen;
	int ulen = len;
	int corkreq = up->corkflag || msg->msg_flags&MSG_MORE;
	int err;
1030
	int connected = 0;
W
Wang Chen 已提交
1031
	int is_udplite = IS_UDPLITE(sk);
1032
	int (*getfrag)(void *, char *, int, int, int, struct sk_buff *);
1033
	struct sockcm_cookie sockc;
L
Linus Torvalds 已提交
1034

W
Wei Wang 已提交
1035 1036 1037 1038
	ipc6.hlimit = -1;
	ipc6.tclass = -1;
	ipc6.dontfrag = -1;

L
Linus Torvalds 已提交
1039 1040 1041 1042 1043 1044 1045 1046 1047 1048
	/* destination address check */
	if (sin6) {
		if (addr_len < offsetof(struct sockaddr, sa_data))
			return -EINVAL;

		switch (sin6->sin6_family) {
		case AF_INET6:
			if (addr_len < SIN6_LEN_RFC2133)
				return -EINVAL;
			daddr = &sin6->sin6_addr;
1049 1050 1051 1052
			if (ipv6_addr_any(daddr) &&
			    ipv6_addr_v4mapped(&np->saddr))
				ipv6_addr_set_v4mapped(htonl(INADDR_LOOPBACK),
						       daddr);
L
Linus Torvalds 已提交
1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066
			break;
		case AF_INET:
			goto do_udp_sendmsg;
		case AF_UNSPEC:
			msg->msg_name = sin6 = NULL;
			msg->msg_namelen = addr_len = 0;
			daddr = NULL;
			break;
		default:
			return -EINVAL;
		}
	} else if (!up->pending) {
		if (sk->sk_state != TCP_ESTABLISHED)
			return -EDESTADDRREQ;
1067
		daddr = &sk->sk_v6_daddr;
1068
	} else
L
Linus Torvalds 已提交
1069 1070 1071
		daddr = NULL;

	if (daddr) {
B
Brian Haley 已提交
1072
		if (ipv6_addr_v4mapped(daddr)) {
L
Linus Torvalds 已提交
1073 1074
			struct sockaddr_in sin;
			sin.sin_family = AF_INET;
E
Eric Dumazet 已提交
1075
			sin.sin_port = sin6 ? sin6->sin6_port : inet->inet_dport;
L
Linus Torvalds 已提交
1076 1077 1078 1079 1080 1081
			sin.sin_addr.s_addr = daddr->s6_addr32[3];
			msg->msg_name = &sin;
			msg->msg_namelen = sizeof(sin);
do_udp_sendmsg:
			if (__ipv6_only_sock(sk))
				return -ENETUNREACH;
1082
			return udp_sendmsg(sk, msg, len);
L
Linus Torvalds 已提交
1083 1084 1085 1086
		}
	}

	if (up->pending == AF_INET)
1087
		return udp_sendmsg(sk, msg, len);
L
Linus Torvalds 已提交
1088 1089

	/* Rough check on arithmetic overflow,
1090
	   better check is made in ip6_append_data().
L
Linus Torvalds 已提交
1091 1092 1093
	   */
	if (len > INT_MAX - sizeof(struct udphdr))
		return -EMSGSIZE;
1094

1095
	getfrag  =  is_udplite ?  udplite_getfrag : ip_generic_getfrag;
L
Linus Torvalds 已提交
1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113
	if (up->pending) {
		/*
		 * There are pending frames.
		 * The socket lock must be held while it's corked.
		 */
		lock_sock(sk);
		if (likely(up->pending)) {
			if (unlikely(up->pending != AF_INET6)) {
				release_sock(sk);
				return -EAFNOSUPPORT;
			}
			dst = NULL;
			goto do_append_data;
		}
		release_sock(sk);
	}
	ulen += sizeof(struct udphdr);

1114
	memset(&fl6, 0, sizeof(fl6));
L
Linus Torvalds 已提交
1115 1116 1117 1118 1119

	if (sin6) {
		if (sin6->sin6_port == 0)
			return -EINVAL;

1120
		fl6.fl6_dport = sin6->sin6_port;
L
Linus Torvalds 已提交
1121 1122 1123
		daddr = &sin6->sin6_addr;

		if (np->sndflow) {
1124 1125 1126
			fl6.flowlabel = sin6->sin6_flowinfo&IPV6_FLOWINFO_MASK;
			if (fl6.flowlabel&IPV6_FLOWLABEL_MASK) {
				flowlabel = fl6_sock_lookup(sk, fl6.flowlabel);
1127
				if (!flowlabel)
L
Linus Torvalds 已提交
1128 1129 1130 1131 1132 1133 1134 1135 1136
					return -EINVAL;
			}
		}

		/*
		 * Otherwise it will be difficult to maintain
		 * sk->sk_dst_cache.
		 */
		if (sk->sk_state == TCP_ESTABLISHED &&
1137 1138
		    ipv6_addr_equal(daddr, &sk->sk_v6_daddr))
			daddr = &sk->sk_v6_daddr;
L
Linus Torvalds 已提交
1139 1140 1141

		if (addr_len >= sizeof(struct sockaddr_in6) &&
		    sin6->sin6_scope_id &&
1142
		    __ipv6_addr_needs_scope_id(__ipv6_addr_type(daddr)))
1143
			fl6.flowi6_oif = sin6->sin6_scope_id;
L
Linus Torvalds 已提交
1144 1145 1146 1147
	} else {
		if (sk->sk_state != TCP_ESTABLISHED)
			return -EDESTADDRREQ;

1148
		fl6.fl6_dport = inet->inet_dport;
1149
		daddr = &sk->sk_v6_daddr;
1150
		fl6.flowlabel = np->flow_label;
1151
		connected = 1;
L
Linus Torvalds 已提交
1152 1153
	}

1154 1155
	if (!fl6.flowi6_oif)
		fl6.flowi6_oif = sk->sk_bound_dev_if;
L
Linus Torvalds 已提交
1156

1157 1158
	if (!fl6.flowi6_oif)
		fl6.flowi6_oif = np->sticky_pktinfo.ipi6_ifindex;
1159

1160
	fl6.flowi6_mark = sk->sk_mark;
1161
	fl6.flowi6_uid = sk->sk_uid;
1162
	sockc.tsflags = sk->sk_tsflags;
B
Brian Haley 已提交
1163

L
Linus Torvalds 已提交
1164 1165 1166 1167
	if (msg->msg_controllen) {
		opt = &opt_space;
		memset(opt, 0, sizeof(struct ipv6_txoptions));
		opt->tot_len = sizeof(*opt);
W
Wei Wang 已提交
1168
		ipc6.opt = opt;
L
Linus Torvalds 已提交
1169

W
Wei Wang 已提交
1170
		err = ip6_datagram_send_ctl(sock_net(sk), sk, msg, &fl6, &ipc6, &sockc);
L
Linus Torvalds 已提交
1171 1172 1173 1174
		if (err < 0) {
			fl6_sock_release(flowlabel);
			return err;
		}
1175 1176
		if ((fl6.flowlabel&IPV6_FLOWLABEL_MASK) && !flowlabel) {
			flowlabel = fl6_sock_lookup(sk, fl6.flowlabel);
1177
			if (!flowlabel)
L
Linus Torvalds 已提交
1178 1179 1180 1181
				return -EINVAL;
		}
		if (!(opt->opt_nflen|opt->opt_flen))
			opt = NULL;
1182
		connected = 0;
L
Linus Torvalds 已提交
1183
	}
1184 1185 1186 1187
	if (!opt) {
		opt = txopt_get(np);
		opt_to_free = opt;
	}
1188 1189 1190
	if (flowlabel)
		opt = fl6_merge_options(&opt_space, flowlabel, opt);
	opt = ipv6_fixup_options(&opt_space, opt);
W
Wei Wang 已提交
1191
	ipc6.opt = opt;
L
Linus Torvalds 已提交
1192

1193
	fl6.flowi6_proto = sk->sk_protocol;
1194
	if (!ipv6_addr_any(daddr))
A
Alexey Dobriyan 已提交
1195
		fl6.daddr = *daddr;
1196
	else
1197 1198
		fl6.daddr.s6_addr[15] = 0x1; /* :: means loopback (BSD'ism) */
	if (ipv6_addr_any(&fl6.saddr) && !ipv6_addr_any(&np->saddr))
A
Alexey Dobriyan 已提交
1199
		fl6.saddr = np->saddr;
1200
	fl6.fl6_sport = inet->inet_sport;
1201

1202
	final_p = fl6_update_dst(&fl6, opt, &final);
1203
	if (final_p)
1204
		connected = 0;
L
Linus Torvalds 已提交
1205

1206 1207
	if (!fl6.flowi6_oif && ipv6_addr_is_multicast(&fl6.daddr)) {
		fl6.flowi6_oif = np->mcast_oif;
1208
		connected = 0;
1209 1210
	} else if (!fl6.flowi6_oif)
		fl6.flowi6_oif = np->ucast_oif;
L
Linus Torvalds 已提交
1211

1212
	security_sk_classify_flow(sk, flowi6_to_flowi(&fl6));
V
Venkat Yekkirala 已提交
1213

1214 1215 1216 1217 1218
	if (ipc6.tclass < 0)
		ipc6.tclass = np->tclass;

	fl6.flowlabel = ip6_make_flowinfo(ipc6.tclass, fl6.flowlabel);

1219
	dst = ip6_sk_dst_lookup_flow(sk, &fl6, final_p);
1220 1221 1222
	if (IS_ERR(dst)) {
		err = PTR_ERR(dst);
		dst = NULL;
L
Linus Torvalds 已提交
1223
		goto out;
1224
	}
L
Linus Torvalds 已提交
1225

W
Wei Wang 已提交
1226 1227
	if (ipc6.hlimit < 0)
		ipc6.hlimit = ip6_sk_dst_hoplimit(np, &fl6, dst);
L
Linus Torvalds 已提交
1228 1229 1230 1231 1232

	if (msg->msg_flags&MSG_CONFIRM)
		goto do_confirm;
back_from_confirm:

1233 1234 1235 1236 1237
	/* Lockless fast path for the non-corking case */
	if (!corkreq) {
		struct sk_buff *skb;

		skb = ip6_make_skb(sk, getfrag, msg, ulen,
W
Wei Wang 已提交
1238
				   sizeof(struct udphdr), &ipc6,
1239
				   &fl6, (struct rt6_info *)dst,
W
Wei Wang 已提交
1240
				   msg->msg_flags, &sockc);
1241 1242 1243 1244 1245 1246
		err = PTR_ERR(skb);
		if (!IS_ERR_OR_NULL(skb))
			err = udp_v6_send_skb(skb, &fl6);
		goto release_dst;
	}

L
Linus Torvalds 已提交
1247 1248 1249 1250 1251 1252
	lock_sock(sk);
	if (unlikely(up->pending)) {
		/* The socket is already corked while preparing it. */
		/* ... which is an evident application bug. --ANK */
		release_sock(sk);

1253
		net_dbg_ratelimited("udp cork app bug 2\n");
L
Linus Torvalds 已提交
1254 1255 1256 1257 1258 1259 1260
		err = -EINVAL;
		goto out;
	}

	up->pending = AF_INET6;

do_append_data:
W
Wei Wang 已提交
1261 1262
	if (ipc6.dontfrag < 0)
		ipc6.dontfrag = np->dontfrag;
L
Linus Torvalds 已提交
1263
	up->len += ulen;
W
Wei Wang 已提交
1264 1265 1266
	err = ip6_append_data(sk, getfrag, msg, ulen, sizeof(struct udphdr),
			      &ipc6, &fl6, (struct rt6_info *)dst,
			      corkreq ? msg->msg_flags|MSG_MORE : msg->msg_flags, &sockc);
L
Linus Torvalds 已提交
1267 1268 1269
	if (err)
		udp_v6_flush_pending_frames(sk);
	else if (!corkreq)
1270
		err = udp_v6_push_pending_frames(sk);
H
Herbert Xu 已提交
1271 1272
	else if (unlikely(skb_queue_empty(&sk->sk_write_queue)))
		up->pending = 0;
L
Linus Torvalds 已提交
1273

1274 1275 1276 1277 1278
	if (err > 0)
		err = np->recverr ? net_xmit_errno(err) : 0;
	release_sock(sk);

release_dst:
1279 1280 1281
	if (dst) {
		if (connected) {
			ip6_dst_store(sk, dst,
1282 1283
				      ipv6_addr_equal(&fl6.daddr, &sk->sk_v6_daddr) ?
				      &sk->sk_v6_daddr : NULL,
1284
#ifdef CONFIG_IPV6_SUBTREES
1285
				      ipv6_addr_equal(&fl6.saddr, &np->saddr) ?
1286 1287 1288
				      &np->saddr :
#endif
				      NULL);
1289 1290 1291
		} else {
			dst_release(dst);
		}
1292
		dst = NULL;
1293 1294
	}

L
Linus Torvalds 已提交
1295
out:
1296
	dst_release(dst);
L
Linus Torvalds 已提交
1297
	fl6_sock_release(flowlabel);
1298
	txopt_put(opt_to_free);
1299
	if (!err)
L
Linus Torvalds 已提交
1300
		return len;
1301 1302 1303 1304 1305 1306 1307 1308
	/*
	 * ENOBUFS = no kernel mem, SOCK_NOSPACE = no sndbuf space.  Reporting
	 * ENOBUFS might not be good (it's not tunable per se), but otherwise
	 * we don't have a good statistic (IpOutDiscards but it can be too many
	 * things).  We could add another new stat but at least for now that
	 * seems like overkill.
	 */
	if (err == -ENOBUFS || test_bit(SOCK_NOSPACE, &sk->sk_socket->flags)) {
1309 1310
		UDP6_INC_STATS(sock_net(sk),
			       UDP_MIB_SNDBUFERRORS, is_udplite);
1311
	}
L
Linus Torvalds 已提交
1312 1313 1314
	return err;

do_confirm:
1315 1316
	if (msg->msg_flags & MSG_PROBE)
		dst_confirm_neigh(dst, &fl6.daddr);
L
Linus Torvalds 已提交
1317 1318 1319 1320 1321 1322
	if (!(msg->msg_flags&MSG_PROBE) || len)
		goto back_from_confirm;
	err = 0;
	goto out;
}

1323
void udpv6_destroy_sock(struct sock *sk)
L
Linus Torvalds 已提交
1324
{
T
Tom Parkin 已提交
1325
	struct udp_sock *up = udp_sk(sk);
L
Linus Torvalds 已提交
1326 1327 1328 1329
	lock_sock(sk);
	udp_v6_flush_pending_frames(sk);
	release_sock(sk);

T
Tom Parkin 已提交
1330 1331 1332 1333 1334 1335 1336
	if (static_key_false(&udpv6_encap_needed) && up->encap_type) {
		void (*encap_destroy)(struct sock *sk);
		encap_destroy = ACCESS_ONCE(up->encap_destroy);
		if (encap_destroy)
			encap_destroy(sk);
	}

L
Linus Torvalds 已提交
1337 1338 1339 1340 1341 1342
	inet6_destroy_sock(sk);
}

/*
 *	Socket option code for UDP
 */
1343
int udpv6_setsockopt(struct sock *sk, int level, int optname,
1344
		     char __user *optval, unsigned int optlen)
1345
{
1346
	if (level == SOL_UDP  ||  level == SOL_UDPLITE)
1347 1348
		return udp_lib_setsockopt(sk, level, optname, optval, optlen,
					  udp_v6_push_pending_frames);
1349
	return ipv6_setsockopt(sk, level, optname, optval, optlen);
1350 1351 1352
}

#ifdef CONFIG_COMPAT
1353
int compat_udpv6_setsockopt(struct sock *sk, int level, int optname,
1354
			    char __user *optval, unsigned int optlen)
1355
{
1356
	if (level == SOL_UDP  ||  level == SOL_UDPLITE)
1357 1358
		return udp_lib_setsockopt(sk, level, optname, optval, optlen,
					  udp_v6_push_pending_frames);
1359
	return compat_ipv6_setsockopt(sk, level, optname, optval, optlen);
1360 1361 1362
}
#endif

1363 1364
int udpv6_getsockopt(struct sock *sk, int level, int optname,
		     char __user *optval, int __user *optlen)
1365
{
1366
	if (level == SOL_UDP  ||  level == SOL_UDPLITE)
1367
		return udp_lib_getsockopt(sk, level, optname, optval, optlen);
1368
	return ipv6_getsockopt(sk, level, optname, optval, optlen);
1369 1370 1371
}

#ifdef CONFIG_COMPAT
1372 1373
int compat_udpv6_getsockopt(struct sock *sk, int level, int optname,
			    char __user *optval, int __user *optlen)
1374
{
1375
	if (level == SOL_UDP  ||  level == SOL_UDPLITE)
1376
		return udp_lib_getsockopt(sk, level, optname, optval, optlen);
1377
	return compat_ipv6_getsockopt(sk, level, optname, optval, optlen);
1378 1379 1380
}
#endif

1381
static const struct inet6_protocol udpv6_protocol = {
L
Linus Torvalds 已提交
1382 1383 1384 1385 1386 1387 1388
	.handler	=	udpv6_rcv,
	.err_handler	=	udpv6_err,
	.flags		=	INET6_PROTO_NOPOLICY|INET6_PROTO_FINAL,
};

/* ------------------------------------------------------------------------ */
#ifdef CONFIG_PROC_FS
1389
int udp6_seq_show(struct seq_file *seq, void *v)
L
Linus Torvalds 已提交
1390
{
1391 1392 1393 1394 1395 1396 1397 1398 1399
	if (v == SEQ_START_TOKEN) {
		seq_puts(seq, IPV6_SEQ_DGRAM_HEADER);
	} else {
		int bucket = ((struct udp_iter_state *)seq->private)->bucket;
		struct inet_sock *inet = inet_sk(v);
		__u16 srcp = ntohs(inet->inet_sport);
		__u16 destp = ntohs(inet->inet_dport);
		ip6_dgram_sock_seq_show(seq, v, srcp, destp, bucket);
	}
L
Linus Torvalds 已提交
1400 1401 1402
	return 0;
}

1403 1404 1405 1406 1407 1408 1409 1410
static const struct file_operations udp6_afinfo_seq_fops = {
	.owner    = THIS_MODULE,
	.open     = udp_seq_open,
	.read     = seq_read,
	.llseek   = seq_lseek,
	.release  = seq_release_net
};

L
Linus Torvalds 已提交
1411 1412 1413
static struct udp_seq_afinfo udp6_seq_afinfo = {
	.name		= "udp6",
	.family		= AF_INET6,
1414
	.udp_table	= &udp_table,
1415
	.seq_fops	= &udp6_afinfo_seq_fops,
1416 1417 1418
	.seq_ops	= {
		.show		= udp6_seq_show,
	},
L
Linus Torvalds 已提交
1419 1420
};

1421
int __net_init udp6_proc_init(struct net *net)
L
Linus Torvalds 已提交
1422
{
1423
	return udp_proc_register(net, &udp6_seq_afinfo);
L
Linus Torvalds 已提交
1424 1425
}

I
Ian Morris 已提交
1426 1427
void udp6_proc_exit(struct net *net)
{
1428
	udp_proc_unregister(net, &udp6_seq_afinfo);
L
Linus Torvalds 已提交
1429 1430 1431 1432 1433 1434
}
#endif /* CONFIG_PROC_FS */

/* ------------------------------------------------------------------------ */

struct proto udpv6_prot = {
1435 1436
	.name		   = "UDPv6",
	.owner		   = THIS_MODULE,
1437
	.close		   = udp_lib_close,
1438 1439 1440
	.connect	   = ip6_datagram_connect,
	.disconnect	   = udp_disconnect,
	.ioctl		   = udp_ioctl,
1441
	.init		   = udp_init_sock,
1442 1443 1444 1445 1446
	.destroy	   = udpv6_destroy_sock,
	.setsockopt	   = udpv6_setsockopt,
	.getsockopt	   = udpv6_getsockopt,
	.sendmsg	   = udpv6_sendmsg,
	.recvmsg	   = udpv6_recvmsg,
1447
	.release_cb	   = ip6_datagram_release_cb,
1448 1449
	.hash		   = udp_lib_hash,
	.unhash		   = udp_lib_unhash,
E
Eric Dumazet 已提交
1450
	.rehash		   = udp_v6_rehash,
1451
	.get_port	   = udp_v6_get_port,
H
Hideo Aoki 已提交
1452 1453 1454 1455
	.memory_allocated  = &udp_memory_allocated,
	.sysctl_mem	   = sysctl_udp_mem,
	.sysctl_wmem	   = &sysctl_udp_wmem_min,
	.sysctl_rmem	   = &sysctl_udp_rmem_min,
1456
	.obj_size	   = sizeof(struct udp6_sock),
1457
	.h.udp_table	   = &udp_table,
1458
#ifdef CONFIG_COMPAT
1459 1460
	.compat_setsockopt = compat_udpv6_setsockopt,
	.compat_getsockopt = compat_udpv6_getsockopt,
1461
#endif
1462
	.diag_destroy      = udp_abort,
L
Linus Torvalds 已提交
1463 1464 1465 1466 1467 1468 1469 1470 1471 1472
};

static struct inet_protosw udpv6_protosw = {
	.type =      SOCK_DGRAM,
	.protocol =  IPPROTO_UDP,
	.prot =      &udpv6_prot,
	.ops =       &inet6_dgram_ops,
	.flags =     INET_PROTOSW_PERMANENT,
};

1473
int __init udpv6_init(void)
L
Linus Torvalds 已提交
1474
{
1475 1476
	int ret;

1477 1478
	ret = inet6_add_protocol(&udpv6_protocol, IPPROTO_UDP);
	if (ret)
1479
		goto out;
1480

1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491
	ret = inet6_register_protosw(&udpv6_protosw);
	if (ret)
		goto out_udpv6_protocol;
out:
	return ret;

out_udpv6_protocol:
	inet6_del_protocol(&udpv6_protocol, IPPROTO_UDP);
	goto out;
}

1492
void udpv6_exit(void)
1493 1494 1495
{
	inet6_unregister_protosw(&udpv6_protosw);
	inet6_del_protocol(&udpv6_protocol, IPPROTO_UDP);
L
Linus Torvalds 已提交
1496
}