udp.c 37.0 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>
L
Linus Torvalds 已提交
38 39
#include <asm/uaccess.h>

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
static u32 udp6_ehashfn(const struct net *net,
			const struct in6_addr *laddr,
			const u16 lport,
			const struct in6_addr *faddr,
			const __be16 fport)
63
{
64 65 66 67 68 69 70 71 72 73 74 75 76
	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);

77
	return __inet6_ehashfn(lhash, lport, fhash, fport,
78
			       udp_ipv6_hash_secret + net_hash_mix(net));
79 80
}

81 82 83
static u32 udp6_portaddr_hash(const struct net *net,
			      const struct in6_addr *addr6,
			      unsigned int port)
84 85 86 87 88
{
	unsigned int hash, mix = net_hash_mix(net);

	if (ipv6_addr_any(addr6))
		hash = jhash_1word(0, mix);
B
Brian Haley 已提交
89
	else if (ipv6_addr_v4mapped(addr6))
90
		hash = jhash_1word((__force u32)addr6->s6_addr32[3], mix);
91
	else
92
		hash = jhash2((__force u32 *)addr6->s6_addr32, 4, mix);
93 94 95 96

	return hash ^ port;
}

97
int udp_v6_get_port(struct sock *sk, unsigned short snum)
L
Linus Torvalds 已提交
98
{
E
Eric Dumazet 已提交
99 100
	unsigned int hash2_nulladdr =
		udp6_portaddr_hash(sock_net(sk), &in6addr_any, snum);
101
	unsigned int hash2_partial =
102
		udp6_portaddr_hash(sock_net(sk), &sk->sk_v6_rcv_saddr, 0);
E
Eric Dumazet 已提交
103

104
	/* precompute partial secondary hash */
E
Eric Dumazet 已提交
105 106
	udp_sk(sk)->udp_portaddr_hash = hash2_partial;
	return udp_lib_get_port(sk, snum, ipv6_rcv_saddr_equal, hash2_nulladdr);
L
Linus Torvalds 已提交
107 108
}

E
Eric Dumazet 已提交
109 110 111
static void udp_v6_rehash(struct sock *sk)
{
	u16 new_hash = udp6_portaddr_hash(sock_net(sk),
112
					  &sk->sk_v6_rcv_saddr,
E
Eric Dumazet 已提交
113 114 115 116 117
					  inet_sk(sk)->inet_num);

	udp_lib_rehash(sk, new_hash);
}

118 119 120 121
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,
			 int dif)
122
{
123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155
	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++;
	}

	if (sk->sk_bound_dev_if) {
		if (sk->sk_bound_dev_if != dif)
			return -1;
		score++;
156
	}
157

158 159 160
	if (sk->sk_incoming_cpu == raw_smp_processor_id())
		score++;

161 162 163
	return score;
}

164
/* called with rcu_read_lock() */
E
Eric Dumazet 已提交
165 166 167
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,
168
		struct udp_hslot *hslot2,
169
		struct sk_buff *skb)
E
Eric Dumazet 已提交
170 171
{
	struct sock *sk, *result;
172 173
	int score, badness, matches = 0, reuseport = 0;
	u32 hash = 0;
E
Eric Dumazet 已提交
174 175 176

	result = NULL;
	badness = -1;
177
	udp_portaddr_for_each_entry_rcu(sk, &hslot2->head) {
178
		score = compute_score(sk, net, saddr, sport,
E
Eric Dumazet 已提交
179 180
				      daddr, hnum, dif);
		if (score > badness) {
181 182
			reuseport = sk->sk_reuseport;
			if (reuseport) {
183 184
				hash = udp6_ehashfn(net, daddr, hnum,
						    saddr, sport);
185

186
				result = reuseport_select_sock(sk, hash, skb,
187
							sizeof(struct udphdr));
188 189
				if (result)
					return result;
190
				matches = 1;
191
			}
192 193
			result = sk;
			badness = score;
194 195
		} else if (score == badness && reuseport) {
			matches++;
196
			if (reciprocal_scale(hash, matches) == 0)
197 198
				result = sk;
			hash = next_pseudo_random32(hash);
E
Eric Dumazet 已提交
199 200 201 202 203
		}
	}
	return result;
}

204
/* rcu_read_lock() must be held */
205
struct sock *__udp6_lib_lookup(struct net *net,
206 207
				      const struct in6_addr *saddr, __be16 sport,
				      const struct in6_addr *daddr, __be16 dport,
208 209
				      int dif, struct udp_table *udptable,
				      struct sk_buff *skb)
L
Linus Torvalds 已提交
210
{
211
	struct sock *sk, *result;
L
Linus Torvalds 已提交
212
	unsigned short hnum = ntohs(dport);
E
Eric Dumazet 已提交
213 214
	unsigned int hash2, slot2, slot = udp_hashfn(net, hnum, udptable->mask);
	struct udp_hslot *hslot2, *hslot = &udptable->hash[slot];
215 216
	int score, badness, matches = 0, reuseport = 0;
	u32 hash = 0;
217

E
Eric Dumazet 已提交
218 219 220 221 222 223 224 225 226
	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,
					  daddr, hnum, dif,
227
					  hslot2, skb);
E
Eric Dumazet 已提交
228
		if (!result) {
229
			unsigned int old_slot2 = slot2;
E
Eric Dumazet 已提交
230 231
			hash2 = udp6_portaddr_hash(net, &in6addr_any, hnum);
			slot2 = hash2 & udptable->mask;
232 233 234 235
			/* avoid searching the same slot again. */
			if (unlikely(slot2 == old_slot2))
				return result;

E
Eric Dumazet 已提交
236 237 238 239
			hslot2 = &udptable->hash2[slot2];
			if (hslot->count < hslot2->count)
				goto begin;

240
			result = udp6_lib_lookup2(net, saddr, sport,
241 242
						  daddr, hnum, dif,
						  hslot2, skb);
E
Eric Dumazet 已提交
243 244 245
		}
		return result;
	}
246 247 248
begin:
	result = NULL;
	badness = -1;
249
	sk_for_each_rcu(sk, &hslot->head) {
250
		score = compute_score(sk, net, saddr, sport, daddr, hnum, dif);
251
		if (score > badness) {
252 253
			reuseport = sk->sk_reuseport;
			if (reuseport) {
254 255
				hash = udp6_ehashfn(net, daddr, hnum,
						    saddr, sport);
256
				result = reuseport_select_sock(sk, hash, skb,
257
							sizeof(struct udphdr));
258 259
				if (result)
					return result;
260 261
				matches = 1;
			}
262 263
			result = sk;
			badness = score;
264 265
		} else if (score == badness && reuseport) {
			matches++;
266
			if (reciprocal_scale(hash, matches) == 0)
267 268
				result = sk;
			hash = next_pseudo_random32(hash);
L
Linus Torvalds 已提交
269 270 271 272
		}
	}
	return result;
}
273
EXPORT_SYMBOL_GPL(__udp6_lib_lookup);
L
Linus Torvalds 已提交
274

275 276
static struct sock *__udp6_lib_lookup_skb(struct sk_buff *skb,
					  __be16 sport, __be16 dport,
277
					  struct udp_table *udptable)
278
{
279
	const struct ipv6hdr *iph = ipv6_hdr(skb);
280
	struct sock *sk;
281

282 283
	sk = skb_steal_sock(skb);
	if (unlikely(sk))
284
		return sk;
285
	return __udp6_lib_lookup(dev_net(skb->dev), &iph->saddr, sport,
E
Eric Dumazet 已提交
286
				 &iph->daddr, dport, inet6_iif(skb),
287
				 udptable, skb);
288 289
}

290 291 292 293 294
struct sock *udp6_lib_lookup_skb(struct sk_buff *skb,
				 __be16 sport, __be16 dport)
{
	const struct ipv6hdr *iph = ipv6_hdr(skb);

295
	return __udp6_lib_lookup(dev_net(skb->dev), &iph->saddr, sport,
296 297 298 299 300
				 &iph->daddr, dport, inet6_iif(skb),
				 &udp_table, skb);
}
EXPORT_SYMBOL_GPL(udp6_lib_lookup_skb);

301 302 303 304 305
/* Must be called under rcu_read_lock().
 * Does increment socket refcount.
 */
#if IS_ENABLED(CONFIG_NETFILTER_XT_MATCH_SOCKET) || \
    IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TPROXY)
306 307 308
struct sock *udp6_lib_lookup(struct net *net, const struct in6_addr *saddr, __be16 sport,
			     const struct in6_addr *daddr, __be16 dport, int dif)
{
309 310 311 312 313 314 315
	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;
316 317
}
EXPORT_SYMBOL_GPL(udp6_lib_lookup);
318
#endif
319

L
Linus Torvalds 已提交
320
/*
321 322
 *	This should be easy, if there is something there we
 *	return it, otherwise we block.
L
Linus Torvalds 已提交
323 324
 */

325
int udpv6_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
L
Linus Torvalds 已提交
326 327 328 329
		  int noblock, int flags, int *addr_len)
{
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct inet_sock *inet = inet_sk(sk);
330
	struct sk_buff *skb;
331
	unsigned int ulen, copied;
332
	int peeked, peeking, off;
333 334
	int err;
	int is_udplite = IS_UDPLITE(sk);
335
	bool checksum_valid = false;
336
	int is_udp4;
337
	bool slow;
L
Linus Torvalds 已提交
338 339

	if (flags & MSG_ERRQUEUE)
340
		return ipv6_recv_error(sk, msg, len, addr_len);
L
Linus Torvalds 已提交
341

342
	if (np->rxpmtu && np->rxopt.bits.rxpmtu)
343
		return ipv6_recv_rxpmtu(sk, msg, len, addr_len);
344

L
Linus Torvalds 已提交
345
try_again:
346
	peeking = off = sk_peek_offset(sk, flags);
347
	skb = __skb_recv_datagram(sk, flags | (noblock ? MSG_DONTWAIT : 0),
348
				  &peeked, &off, &err);
L
Linus Torvalds 已提交
349
	if (!skb)
350
		return err;
L
Linus Torvalds 已提交
351

352
	ulen = skb->len;
353
	copied = len;
354 355
	if (copied > ulen - off)
		copied = ulen - off;
356
	else if (copied < ulen)
357
		msg->msg_flags |= MSG_TRUNC;
L
Linus Torvalds 已提交
358

359 360
	is_udp4 = (skb->protocol == htons(ETH_P_IP));

361
	/*
362 363 364
	 * 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.
365 366
	 */

367
	if (copied < ulen || UDP_SKB_CB(skb)->partial_cov || peeking) {
368 369
		checksum_valid = !udp_lib_checksum_complete(skb);
		if (!checksum_valid)
L
Linus Torvalds 已提交
370
			goto csum_copy_err;
371 372
	}

373
	if (checksum_valid || skb_csum_unnecessary(skb))
374
		err = skb_copy_datagram_msg(skb, off, msg, copied);
375
	else {
376
		err = skb_copy_and_csum_datagram_msg(skb, off, msg);
L
Linus Torvalds 已提交
377 378 379
		if (err == -EINVAL)
			goto csum_copy_err;
	}
380 381
	if (unlikely(err)) {
		trace_kfree_skb(skb, udpv6_recvmsg);
382 383 384
		if (!peeked) {
			atomic_inc(&sk->sk_drops);
			if (is_udp4)
385 386
				UDP_INC_STATS(sock_net(sk), UDP_MIB_INERRORS,
					      is_udplite);
387
			else
388 389
				UDP6_INC_STATS(sock_net(sk), UDP_MIB_INERRORS,
					       is_udplite);
390
		}
391 392
		skb_free_datagram_locked(sk, skb);
		return err;
393
	}
394 395
	if (!peeked) {
		if (is_udp4)
396 397
			UDP_INC_STATS(sock_net(sk), UDP_MIB_INDATAGRAMS,
				      is_udplite);
398
		else
399 400
			UDP6_INC_STATS(sock_net(sk), UDP_MIB_INDATAGRAMS,
				       is_udplite);
401
	}
402

403
	sock_recv_ts_and_drops(msg, sk, skb);
L
Linus Torvalds 已提交
404 405 406

	/* Copy the address. */
	if (msg->msg_name) {
407
		DECLARE_SOCKADDR(struct sockaddr_in6 *, sin6, msg->msg_name);
L
Linus Torvalds 已提交
408
		sin6->sin6_family = AF_INET6;
409
		sin6->sin6_port = udp_hdr(skb)->source;
L
Linus Torvalds 已提交
410 411
		sin6->sin6_flowinfo = 0;

412
		if (is_udp4) {
B
Brian Haley 已提交
413 414
			ipv6_addr_set_v4mapped(ip_hdr(skb)->saddr,
					       &sin6->sin6_addr);
415 416
			sin6->sin6_scope_id = 0;
		} else {
A
Alexey Dobriyan 已提交
417
			sin6->sin6_addr = ipv6_hdr(skb)->saddr;
418 419
			sin6->sin6_scope_id =
				ipv6_iface_scope_id(&sin6->sin6_addr,
420
						    inet6_iif(skb));
L
Linus Torvalds 已提交
421
		}
422
		*addr_len = sizeof(*sin6);
L
Linus Torvalds 已提交
423
	}
424 425 426 427

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

428
	if (is_udp4) {
L
Linus Torvalds 已提交
429
		if (inet->cmsg_flags)
E
Eric Dumazet 已提交
430 431
			ip_cmsg_recv_offset(msg, skb,
					    sizeof(struct udphdr), off);
L
Linus Torvalds 已提交
432 433
	} else {
		if (np->rxopt.all)
434
			ip6_datagram_recv_specific_ctl(sk, msg, skb);
435
	}
L
Linus Torvalds 已提交
436

437
	err = copied;
L
Linus Torvalds 已提交
438
	if (flags & MSG_TRUNC)
439
		err = ulen;
L
Linus Torvalds 已提交
440

441
	__skb_free_datagram_locked(sk, skb, peeking ? -err : err);
L
Linus Torvalds 已提交
442 443 444
	return err;

csum_copy_err:
445
	slow = lock_sock_fast(sk);
446
	if (!skb_kill_datagram(sk, skb, flags)) {
447
		if (is_udp4) {
448 449 450 451
			UDP_INC_STATS(sock_net(sk),
				      UDP_MIB_CSUMERRORS, is_udplite);
			UDP_INC_STATS(sock_net(sk),
				      UDP_MIB_INERRORS, is_udplite);
452
		} else {
453 454 455 456
			UDP6_INC_STATS(sock_net(sk),
				       UDP_MIB_CSUMERRORS, is_udplite);
			UDP6_INC_STATS(sock_net(sk),
				       UDP_MIB_INERRORS, is_udplite);
457
		}
458
	}
459
	unlock_sock_fast(sk, slow);
L
Linus Torvalds 已提交
460

461 462
	/* starting over for a new packet, but check if we need to yield */
	cond_resched();
463
	msg->msg_flags &= ~MSG_TRUNC;
L
Linus Torvalds 已提交
464 465 466
	goto try_again;
}

467
void __udp6_lib_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
468
		    u8 type, u8 code, int offset, __be32 info,
469
		    struct udp_table *udptable)
L
Linus Torvalds 已提交
470 471
{
	struct ipv6_pinfo *np;
472 473 474
	const struct ipv6hdr *hdr = (const struct ipv6hdr *)skb->data;
	const struct in6_addr *saddr = &hdr->saddr;
	const struct in6_addr *daddr = &hdr->daddr;
475
	struct udphdr *uh = (struct udphdr *)(skb->data+offset);
L
Linus Torvalds 已提交
476
	struct sock *sk;
477
	int harderr;
L
Linus Torvalds 已提交
478
	int err;
479
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
480

481
	sk = __udp6_lib_lookup(net, daddr, uh->dest, saddr, uh->source,
482
			       inet6_iif(skb), udptable, skb);
483
	if (!sk) {
E
Eric Dumazet 已提交
484 485
		__ICMP6_INC_STATS(net, __in6_dev_get(skb->dev),
				  ICMP6_MIB_INERRORS);
L
Linus Torvalds 已提交
486
		return;
487
	}
L
Linus Torvalds 已提交
488

489 490 491
	harderr = icmpv6_err_convert(type, code, &err);
	np = inet6_sk(sk);

492 493 494
	if (type == ICMPV6_PKT_TOOBIG) {
		if (!ip6_sk_accept_pmtu(sk))
			goto out;
495
		ip6_sk_update_pmtu(skb, sk, info);
496 497
		if (np->pmtudisc != IPV6_PMTUDISC_DONT)
			harderr = 1;
498
	}
499
	if (type == NDISC_REDIRECT) {
500
		ip6_sk_redirect(skb, sk);
501 502
		goto out;
	}
503

504 505 506 507
	if (!np->recverr) {
		if (!harderr || sk->sk_state != TCP_ESTABLISHED)
			goto out;
	} else {
L
Linus Torvalds 已提交
508
		ipv6_icmp_error(sk, skb, err, uh->dest, ntohl(info), (u8 *)(uh+1));
509
	}
510

L
Linus Torvalds 已提交
511 512 513
	sk->sk_err = err;
	sk->sk_error_report(sk);
out:
514
	return;
L
Linus Torvalds 已提交
515 516
}

517 518 519 520
static int __udpv6_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
{
	int rc;

521
	if (!ipv6_addr_any(&sk->sk_v6_daddr)) {
522
		sock_rps_save_rxhash(sk, skb);
523
		sk_mark_napi_id(sk, skb);
E
Eric Dumazet 已提交
524
		sk_incoming_cpu_update(sk);
525
	}
526

527
	rc = __sock_queue_rcv_skb(sk, skb);
528 529 530 531 532
	if (rc < 0) {
		int is_udplite = IS_UDPLITE(sk);

		/* Note that an ENOMEM error is charged twice */
		if (rc == -ENOMEM)
533
			UDP6_INC_STATS(sock_net(sk),
534
					 UDP_MIB_RCVBUFERRORS, is_udplite);
535
		UDP6_INC_STATS(sock_net(sk), UDP_MIB_INERRORS, is_udplite);
536 537 538 539 540 541
		kfree_skb(skb);
		return -1;
	}
	return 0;
}

542
static __inline__ void udpv6_err(struct sk_buff *skb,
543
				 struct inet6_skb_parm *opt, u8 type,
544
				 u8 code, int offset, __be32 info)
545
{
546
	__udp6_lib_err(skb, opt, type, code, offset, info, &udp_table);
547 548
}

549 550 551 552 553 554 555 556
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);

557
int udpv6_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
L
Linus Torvalds 已提交
558
{
559
	struct udp_sock *up = udp_sk(sk);
560
	int rc;
W
Wang Chen 已提交
561
	int is_udplite = IS_UDPLITE(sk);
562

563 564
	if (!xfrm6_policy_check(sk, XFRM_POLICY_IN, skb))
		goto drop;
L
Linus Torvalds 已提交
565

566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581
	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);
582
		if (encap_rcv) {
583 584
			int ret;

585 586 587 588
			/* Verify checksum before giving to encap */
			if (udp_lib_checksum_complete(skb))
				goto csum_error;

589 590
			ret = encap_rcv(sk, skb);
			if (ret <= 0) {
591 592 593
				__UDP_INC_STATS(sock_net(sk),
						UDP_MIB_INDATAGRAMS,
						is_udplite);
594 595 596 597 598 599 600
				return -ret;
			}
		}

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

601 602 603
	/*
	 * UDP-Lite specific tests, ignored on UDP sockets (see net/ipv4/udp.c).
	 */
W
Wang Chen 已提交
604
	if ((is_udplite & UDPLITE_RECV_CC)  &&  UDP_SKB_CB(skb)->partial_cov) {
605 606

		if (up->pcrlen == 0) {          /* full coverage was set  */
607 608
			net_dbg_ratelimited("UDPLITE6: partial coverage %d while full coverage %d requested\n",
					    UDP_SKB_CB(skb)->cscov, skb->len);
609 610 611
			goto drop;
		}
		if (UDP_SKB_CB(skb)->cscov  <  up->pcrlen) {
612 613
			net_dbg_ratelimited("UDPLITE6: coverage %d too small, need min %d\n",
					    UDP_SKB_CB(skb)->cscov, up->pcrlen);
614 615
			goto drop;
		}
L
Linus Torvalds 已提交
616 617
	}

618 619 620 621
	if (rcu_access_pointer(sk->sk_filter) &&
	    udp_lib_checksum_complete(skb))
		goto csum_error;

622
	if (sk_filter_trim_cap(sk, skb, sizeof(struct udphdr)))
623
		goto drop;
624

625
	udp_csum_pull_header(skb);
626
	if (sk_rcvqueues_full(sk, sk->sk_rcvbuf)) {
627 628
		__UDP6_INC_STATS(sock_net(sk),
				 UDP_MIB_RCVBUFERRORS, is_udplite);
629
		goto drop;
630
	}
631

632
	skb_dst_drop(skb);
633

634 635 636 637 638 639 640 641 642
	bh_lock_sock(sk);
	rc = 0;
	if (!sock_owned_by_user(sk))
		rc = __udpv6_queue_rcv_skb(sk, skb);
	else if (sk_add_backlog(sk, skb, sk->sk_rcvbuf)) {
		bh_unlock_sock(sk);
		goto drop;
	}
	bh_unlock_sock(sk);
643 644

	return rc;
645

646
csum_error:
647
	__UDP6_INC_STATS(sock_net(sk), UDP_MIB_CSUMERRORS, is_udplite);
648
drop:
649
	__UDP6_INC_STATS(sock_net(sk), UDP_MIB_INERRORS, is_udplite);
650
	atomic_inc(&sk->sk_drops);
651 652
	kfree_skb(skb);
	return -1;
L
Linus Torvalds 已提交
653 654
}

655 656 657 658
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
659
{
660
	struct inet_sock *inet = inet_sk(sk);
L
Linus Torvalds 已提交
661

662 663 664 665 666 667 668 669
	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)) ||
670 671 672
	    (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)))
673 674 675 676
		return false;
	if (!inet6_mc_check(sk, loc_addr, rmt_addr))
		return false;
	return true;
L
Linus Torvalds 已提交
677 678
}

679 680 681 682 683
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.
	 */
684 685 686
	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));
687 688
}

L
Linus Torvalds 已提交
689 690 691 692
/*
 * Note: called only from the BH handler context,
 * so we don't need to lock the hashes.
 */
693
static int __udp6_lib_mcast_deliver(struct net *net, struct sk_buff *skb,
694
		const struct in6_addr *saddr, const struct in6_addr *daddr,
695
		struct udp_table *udptable, int proto)
L
Linus Torvalds 已提交
696
{
697
	struct sock *sk, *first = NULL;
698
	const struct udphdr *uh = udp_hdr(skb);
699 700
	unsigned short hnum = ntohs(uh->dest);
	struct udp_hslot *hslot = udp_hashslot(udptable, net, hnum);
701
	unsigned int offset = offsetof(typeof(*sk), sk_node);
702
	unsigned int hash2 = 0, hash2_any = 0, use_hash2 = (hslot->count > 10);
703 704 705
	int dif = inet6_iif(skb);
	struct hlist_node *node;
	struct sk_buff *nskb;
706 707 708

	if (use_hash2) {
		hash2_any = udp6_portaddr_hash(net, &in6addr_any, hnum) &
709 710
			    udptable->mask;
		hash2 = udp6_portaddr_hash(net, daddr, hnum) & udptable->mask;
711
start_lookup:
712
		hslot = &udptable->hash2[hash2];
713 714
		offset = offsetof(typeof(*sk), __sk_common.skc_portaddr_node);
	}
L
Linus Torvalds 已提交
715

716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731
	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);
732 733 734 735
			__UDP6_INC_STATS(net, UDP_MIB_RCVBUFERRORS,
					 IS_UDPLITE(sk));
			__UDP6_INC_STATS(net, UDP_MIB_INERRORS,
					 IS_UDPLITE(sk));
736
			continue;
H
Hideo Aoki 已提交
737
		}
738

739 740 741
		if (udpv6_queue_rcv_skb(sk, nskb) > 0)
			consume_skb(nskb);
	}
742

743 744 745 746 747 748
	/* 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;
	}

749 750 751
	if (first) {
		if (udpv6_queue_rcv_skb(first, skb) > 0)
			consume_skb(skb);
752
	} else {
753
		kfree_skb(skb);
754 755
		__UDP6_INC_STATS(net, UDP_MIB_IGNOREDMULTI,
				 proto == IPPROTO_UDPLITE);
756
	}
757
	return 0;
L
Linus Torvalds 已提交
758 759
}

760
int __udp6_lib_rcv(struct sk_buff *skb, struct udp_table *udptable,
761
		   int proto)
L
Linus Torvalds 已提交
762
{
763
	const struct in6_addr *saddr, *daddr;
764
	struct net *net = dev_net(skb->dev);
765
	struct udphdr *uh;
766
	struct sock *sk;
L
Linus Torvalds 已提交
767 768 769
	u32 ulen = 0;

	if (!pskb_may_pull(skb, sizeof(struct udphdr)))
770
		goto discard;
L
Linus Torvalds 已提交
771

772 773
	saddr = &ipv6_hdr(skb)->saddr;
	daddr = &ipv6_hdr(skb)->daddr;
774
	uh = udp_hdr(skb);
L
Linus Torvalds 已提交
775 776

	ulen = ntohs(uh->len);
777 778
	if (ulen > skb->len)
		goto short_packet;
L
Linus Torvalds 已提交
779

780 781
	if (proto == IPPROTO_UDP) {
		/* UDP validates ulen. */
L
Linus Torvalds 已提交
782

783 784 785
		/* Check for jumbo payload */
		if (ulen == 0)
			ulen = skb->len;
L
Linus Torvalds 已提交
786

787 788
		if (ulen < sizeof(*uh))
			goto short_packet;
L
Linus Torvalds 已提交
789

790 791 792
		if (ulen < skb->len) {
			if (pskb_trim_rcsum(skb, ulen))
				goto short_packet;
793 794
			saddr = &ipv6_hdr(skb)->saddr;
			daddr = &ipv6_hdr(skb)->daddr;
795
			uh = udp_hdr(skb);
796 797
		}
	}
L
Linus Torvalds 已提交
798

799
	if (udp6_csum_init(skb, uh, proto))
800
		goto csum_error;
801

802 803
	/*
	 *	Multicast receive code
L
Linus Torvalds 已提交
804
	 */
805
	if (ipv6_addr_is_multicast(daddr))
806
		return __udp6_lib_mcast_deliver(net, skb,
807
				saddr, daddr, udptable, proto);
L
Linus Torvalds 已提交
808 809 810

	/* Unicast */

811
	/*
L
Linus Torvalds 已提交
812 813 814
	 * check socket cache ... must talk to Alan about his plans
	 * for sock caches... i'll skip this for now.
	 */
815
	sk = __udp6_lib_lookup_skb(skb, uh->source, uh->dest, udptable);
816
	if (sk) {
817 818
		int ret;

819
		if (!uh->check && !udp_sk(sk)->no_check6_rx) {
820 821 822 823
			udp6_csum_zero_error(skb);
			goto csum_error;
		}

824
		if (inet_get_convert_csum(sk) && uh->check && !IS_UDPLITE(sk))
825 826 827
			skb_checksum_try_convert(skb, IPPROTO_UDP, uh->check,
						 ip6_compute_pseudo);

828
		ret = udpv6_queue_rcv_skb(sk, skb);
L
Linus Torvalds 已提交
829

830
		/* a return value > 0 means to resubmit the input */
831
		if (ret > 0)
832
			return ret;
L
Linus Torvalds 已提交
833

S
Stephen Hemminger 已提交
834
		return 0;
L
Linus Torvalds 已提交
835
	}
836

837 838 839 840 841
	if (!uh->check) {
		udp6_csum_zero_error(skb);
		goto csum_error;
	}

842
	if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb))
843
		goto discard;
844 845

	if (udp_lib_checksum_complete(skb))
846
		goto csum_error;
847

848
	__UDP6_INC_STATS(net, UDP_MIB_NOPORTS, proto == IPPROTO_UDPLITE);
849 850 851
	icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_PORT_UNREACH, 0);

	kfree_skb(skb);
S
Stephen Hemminger 已提交
852
	return 0;
L
Linus Torvalds 已提交
853

854
short_packet:
855 856 857 858 859
	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));
860 861
	goto discard;
csum_error:
862
	__UDP6_INC_STATS(net, UDP_MIB_CSUMERRORS, proto == IPPROTO_UDPLITE);
L
Linus Torvalds 已提交
863
discard:
864
	__UDP6_INC_STATS(net, UDP_MIB_INERRORS, proto == IPPROTO_UDPLITE);
L
Linus Torvalds 已提交
865
	kfree_skb(skb);
S
Stephen Hemminger 已提交
866
	return 0;
L
Linus Torvalds 已提交
867
}
868

869
static __inline__ int udpv6_rcv(struct sk_buff *skb)
870
{
871
	return __udp6_lib_rcv(skb, &udp_table, IPPROTO_UDP);
872 873
}

L
Linus Torvalds 已提交
874 875 876 877 878 879 880
/*
 * 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);

881 882 883
	if (up->pending == AF_INET)
		udp_flush_pending_frames(sk);
	else if (up->pending) {
L
Linus Torvalds 已提交
884 885 886
		up->len = 0;
		up->pending = 0;
		ip6_flush_pending_frames(sk);
887
	}
L
Linus Torvalds 已提交
888 889
}

890
/**
891 892 893 894
 *	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)
895 896 897 898 899 900 901
 */
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
902
	struct sk_buff *frags = skb_shinfo(skb)->frag_list;
903 904
	__wsum csum = 0;

V
Vlad Yasevich 已提交
905
	if (!frags) {
906 907 908 909 910 911 912 913 914 915 916 917 918 919 920
		/* 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
921 922 923
		do {
			csum = csum_add(csum, frags->csum);
		} while ((frags = frags->next));
924 925 926 927 928 929 930 931

		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
932 933 934 935
/*
 *	Sending
 */

V
Vlad Yasevich 已提交
936
static int udp_v6_send_skb(struct sk_buff *skb, struct flowi6 *fl6)
L
Linus Torvalds 已提交
937
{
V
Vlad Yasevich 已提交
938
	struct sock *sk = skb->sk;
L
Linus Torvalds 已提交
939 940
	struct udphdr *uh;
	int err = 0;
W
Wang Chen 已提交
941
	int is_udplite = IS_UDPLITE(sk);
942
	__wsum csum = 0;
V
Vlad Yasevich 已提交
943 944
	int offset = skb_transport_offset(skb);
	int len = skb->len - offset;
L
Linus Torvalds 已提交
945 946 947 948

	/*
	 * Create a UDP header
	 */
949
	uh = udp_hdr(skb);
950 951
	uh->source = fl6->fl6_sport;
	uh->dest = fl6->fl6_dport;
V
Vlad Yasevich 已提交
952
	uh->len = htons(len);
L
Linus Torvalds 已提交
953 954
	uh->check = 0;

W
Wang Chen 已提交
955
	if (is_udplite)
V
Vlad Yasevich 已提交
956 957
		csum = udplite_csum(skb);
	else if (udp_sk(sk)->no_check6_tx) {   /* UDP csum disabled */
958 959 960
		skb->ip_summed = CHECKSUM_NONE;
		goto send;
	} else if (skb->ip_summed == CHECKSUM_PARTIAL) { /* UDP hardware csum */
V
Vlad Yasevich 已提交
961
		udp6_hwcsum_outgoing(sk, skb, &fl6->saddr, &fl6->daddr, len);
962 963
		goto send;
	} else
V
Vlad Yasevich 已提交
964
		csum = udp_csum(skb);
L
Linus Torvalds 已提交
965

966
	/* add protocol-dependent pseudo-header */
967
	uh->check = csum_ipv6_magic(&fl6->saddr, &fl6->daddr,
V
Vlad Yasevich 已提交
968
				    len, fl6->flowi6_proto, csum);
L
Linus Torvalds 已提交
969
	if (uh->check == 0)
970
		uh->check = CSUM_MANGLED_0;
L
Linus Torvalds 已提交
971

972
send:
V
Vlad Yasevich 已提交
973
	err = ip6_send_skb(skb);
E
Eric Dumazet 已提交
974 975
	if (err) {
		if (err == -ENOBUFS && !inet6_sk(sk)->recverr) {
976 977
			UDP6_INC_STATS(sock_net(sk),
				       UDP_MIB_SNDBUFERRORS, is_udplite);
E
Eric Dumazet 已提交
978 979
			err = 0;
		}
980 981 982 983
	} else {
		UDP6_INC_STATS(sock_net(sk),
			       UDP_MIB_OUTDATAGRAMS, is_udplite);
	}
V
Vlad Yasevich 已提交
984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007
	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
1008 1009 1010 1011 1012 1013
out:
	up->len = 0;
	up->pending = 0;
	return err;
}

1014
int udpv6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
L
Linus Torvalds 已提交
1015 1016 1017 1018 1019
{
	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);
1020
	DECLARE_SOCKADDR(struct sockaddr_in6 *, sin6, msg->msg_name);
1021
	struct in6_addr *daddr, *final_p, final;
L
Linus Torvalds 已提交
1022
	struct ipv6_txoptions *opt = NULL;
1023
	struct ipv6_txoptions *opt_to_free = NULL;
L
Linus Torvalds 已提交
1024
	struct ip6_flowlabel *flowlabel = NULL;
1025
	struct flowi6 fl6;
L
Linus Torvalds 已提交
1026
	struct dst_entry *dst;
W
Wei Wang 已提交
1027
	struct ipcm6_cookie ipc6;
L
Linus Torvalds 已提交
1028 1029 1030 1031
	int addr_len = msg->msg_namelen;
	int ulen = len;
	int corkreq = up->corkflag || msg->msg_flags&MSG_MORE;
	int err;
1032
	int connected = 0;
W
Wang Chen 已提交
1033
	int is_udplite = IS_UDPLITE(sk);
1034
	int (*getfrag)(void *, char *, int, int, int, struct sk_buff *);
1035
	struct sockcm_cookie sockc;
L
Linus Torvalds 已提交
1036

W
Wei Wang 已提交
1037 1038 1039 1040
	ipc6.hlimit = -1;
	ipc6.tclass = -1;
	ipc6.dontfrag = -1;

L
Linus Torvalds 已提交
1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064
	/* 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;
			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;
1065
		daddr = &sk->sk_v6_daddr;
1066
	} else
L
Linus Torvalds 已提交
1067 1068 1069
		daddr = NULL;

	if (daddr) {
B
Brian Haley 已提交
1070
		if (ipv6_addr_v4mapped(daddr)) {
L
Linus Torvalds 已提交
1071 1072
			struct sockaddr_in sin;
			sin.sin_family = AF_INET;
E
Eric Dumazet 已提交
1073
			sin.sin_port = sin6 ? sin6->sin6_port : inet->inet_dport;
L
Linus Torvalds 已提交
1074 1075 1076 1077 1078 1079
			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;
1080
			return udp_sendmsg(sk, msg, len);
L
Linus Torvalds 已提交
1081 1082 1083 1084
		}
	}

	if (up->pending == AF_INET)
1085
		return udp_sendmsg(sk, msg, len);
L
Linus Torvalds 已提交
1086 1087

	/* Rough check on arithmetic overflow,
1088
	   better check is made in ip6_append_data().
L
Linus Torvalds 已提交
1089 1090 1091
	   */
	if (len > INT_MAX - sizeof(struct udphdr))
		return -EMSGSIZE;
1092

1093
	getfrag  =  is_udplite ?  udplite_getfrag : ip_generic_getfrag;
L
Linus Torvalds 已提交
1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111
	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);

1112
	memset(&fl6, 0, sizeof(fl6));
L
Linus Torvalds 已提交
1113 1114 1115 1116 1117

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

1118
		fl6.fl6_dport = sin6->sin6_port;
L
Linus Torvalds 已提交
1119 1120 1121
		daddr = &sin6->sin6_addr;

		if (np->sndflow) {
1122 1123 1124
			fl6.flowlabel = sin6->sin6_flowinfo&IPV6_FLOWINFO_MASK;
			if (fl6.flowlabel&IPV6_FLOWLABEL_MASK) {
				flowlabel = fl6_sock_lookup(sk, fl6.flowlabel);
1125
				if (!flowlabel)
L
Linus Torvalds 已提交
1126 1127 1128 1129 1130 1131 1132 1133 1134
					return -EINVAL;
			}
		}

		/*
		 * Otherwise it will be difficult to maintain
		 * sk->sk_dst_cache.
		 */
		if (sk->sk_state == TCP_ESTABLISHED &&
1135 1136
		    ipv6_addr_equal(daddr, &sk->sk_v6_daddr))
			daddr = &sk->sk_v6_daddr;
L
Linus Torvalds 已提交
1137 1138 1139

		if (addr_len >= sizeof(struct sockaddr_in6) &&
		    sin6->sin6_scope_id &&
1140
		    __ipv6_addr_needs_scope_id(__ipv6_addr_type(daddr)))
1141
			fl6.flowi6_oif = sin6->sin6_scope_id;
L
Linus Torvalds 已提交
1142 1143 1144 1145
	} else {
		if (sk->sk_state != TCP_ESTABLISHED)
			return -EDESTADDRREQ;

1146
		fl6.fl6_dport = inet->inet_dport;
1147
		daddr = &sk->sk_v6_daddr;
1148
		fl6.flowlabel = np->flow_label;
1149
		connected = 1;
L
Linus Torvalds 已提交
1150 1151
	}

1152 1153
	if (!fl6.flowi6_oif)
		fl6.flowi6_oif = sk->sk_bound_dev_if;
L
Linus Torvalds 已提交
1154

1155 1156
	if (!fl6.flowi6_oif)
		fl6.flowi6_oif = np->sticky_pktinfo.ipi6_ifindex;
1157

1158
	fl6.flowi6_mark = sk->sk_mark;
1159
	sockc.tsflags = sk->sk_tsflags;
B
Brian Haley 已提交
1160

L
Linus Torvalds 已提交
1161 1162 1163 1164
	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
1165
		ipc6.opt = opt;
L
Linus Torvalds 已提交
1166

W
Wei Wang 已提交
1167
		err = ip6_datagram_send_ctl(sock_net(sk), sk, msg, &fl6, &ipc6, &sockc);
L
Linus Torvalds 已提交
1168 1169 1170 1171
		if (err < 0) {
			fl6_sock_release(flowlabel);
			return err;
		}
1172 1173
		if ((fl6.flowlabel&IPV6_FLOWLABEL_MASK) && !flowlabel) {
			flowlabel = fl6_sock_lookup(sk, fl6.flowlabel);
1174
			if (!flowlabel)
L
Linus Torvalds 已提交
1175 1176 1177 1178
				return -EINVAL;
		}
		if (!(opt->opt_nflen|opt->opt_flen))
			opt = NULL;
1179
		connected = 0;
L
Linus Torvalds 已提交
1180
	}
1181 1182 1183 1184
	if (!opt) {
		opt = txopt_get(np);
		opt_to_free = opt;
	}
1185 1186 1187
	if (flowlabel)
		opt = fl6_merge_options(&opt_space, flowlabel, opt);
	opt = ipv6_fixup_options(&opt_space, opt);
W
Wei Wang 已提交
1188
	ipc6.opt = opt;
L
Linus Torvalds 已提交
1189

1190
	fl6.flowi6_proto = sk->sk_protocol;
1191
	if (!ipv6_addr_any(daddr))
A
Alexey Dobriyan 已提交
1192
		fl6.daddr = *daddr;
1193
	else
1194 1195
		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
1196
		fl6.saddr = np->saddr;
1197
	fl6.fl6_sport = inet->inet_sport;
1198

1199
	final_p = fl6_update_dst(&fl6, opt, &final);
1200
	if (final_p)
1201
		connected = 0;
L
Linus Torvalds 已提交
1202

1203 1204
	if (!fl6.flowi6_oif && ipv6_addr_is_multicast(&fl6.daddr)) {
		fl6.flowi6_oif = np->mcast_oif;
1205
		connected = 0;
1206 1207
	} else if (!fl6.flowi6_oif)
		fl6.flowi6_oif = np->ucast_oif;
L
Linus Torvalds 已提交
1208

1209
	security_sk_classify_flow(sk, flowi6_to_flowi(&fl6));
V
Venkat Yekkirala 已提交
1210

1211 1212 1213 1214 1215
	if (ipc6.tclass < 0)
		ipc6.tclass = np->tclass;

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

1216
	dst = ip6_sk_dst_lookup_flow(sk, &fl6, final_p);
1217 1218 1219
	if (IS_ERR(dst)) {
		err = PTR_ERR(dst);
		dst = NULL;
L
Linus Torvalds 已提交
1220
		goto out;
1221
	}
L
Linus Torvalds 已提交
1222

W
Wei Wang 已提交
1223 1224
	if (ipc6.hlimit < 0)
		ipc6.hlimit = ip6_sk_dst_hoplimit(np, &fl6, dst);
L
Linus Torvalds 已提交
1225 1226 1227 1228 1229

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

1230 1231 1232 1233 1234
	/* 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
1235
				   sizeof(struct udphdr), &ipc6,
1236
				   &fl6, (struct rt6_info *)dst,
W
Wei Wang 已提交
1237
				   msg->msg_flags, &sockc);
1238 1239 1240 1241 1242 1243
		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
1244 1245 1246 1247 1248 1249
	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);

1250
		net_dbg_ratelimited("udp cork app bug 2\n");
L
Linus Torvalds 已提交
1251 1252 1253 1254 1255 1256 1257
		err = -EINVAL;
		goto out;
	}

	up->pending = AF_INET6;

do_append_data:
W
Wei Wang 已提交
1258 1259
	if (ipc6.dontfrag < 0)
		ipc6.dontfrag = np->dontfrag;
L
Linus Torvalds 已提交
1260
	up->len += ulen;
W
Wei Wang 已提交
1261 1262 1263
	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
1264 1265 1266
	if (err)
		udp_v6_flush_pending_frames(sk);
	else if (!corkreq)
1267
		err = udp_v6_push_pending_frames(sk);
H
Herbert Xu 已提交
1268 1269
	else if (unlikely(skb_queue_empty(&sk->sk_write_queue)))
		up->pending = 0;
L
Linus Torvalds 已提交
1270

1271 1272 1273 1274 1275
	if (err > 0)
		err = np->recverr ? net_xmit_errno(err) : 0;
	release_sock(sk);

release_dst:
1276 1277 1278
	if (dst) {
		if (connected) {
			ip6_dst_store(sk, dst,
1279 1280
				      ipv6_addr_equal(&fl6.daddr, &sk->sk_v6_daddr) ?
				      &sk->sk_v6_daddr : NULL,
1281
#ifdef CONFIG_IPV6_SUBTREES
1282
				      ipv6_addr_equal(&fl6.saddr, &np->saddr) ?
1283 1284 1285
				      &np->saddr :
#endif
				      NULL);
1286 1287 1288
		} else {
			dst_release(dst);
		}
1289
		dst = NULL;
1290 1291
	}

L
Linus Torvalds 已提交
1292
out:
1293
	dst_release(dst);
L
Linus Torvalds 已提交
1294
	fl6_sock_release(flowlabel);
1295
	txopt_put(opt_to_free);
1296
	if (!err)
L
Linus Torvalds 已提交
1297
		return len;
1298 1299 1300 1301 1302 1303 1304 1305
	/*
	 * 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)) {
1306 1307
		UDP6_INC_STATS(sock_net(sk),
			       UDP_MIB_SNDBUFERRORS, is_udplite);
1308
	}
L
Linus Torvalds 已提交
1309 1310 1311 1312 1313 1314 1315 1316 1317 1318
	return err;

do_confirm:
	dst_confirm(dst);
	if (!(msg->msg_flags&MSG_PROBE) || len)
		goto back_from_confirm;
	err = 0;
	goto out;
}

1319
void udpv6_destroy_sock(struct sock *sk)
L
Linus Torvalds 已提交
1320
{
T
Tom Parkin 已提交
1321
	struct udp_sock *up = udp_sk(sk);
L
Linus Torvalds 已提交
1322 1323 1324 1325
	lock_sock(sk);
	udp_v6_flush_pending_frames(sk);
	release_sock(sk);

T
Tom Parkin 已提交
1326 1327 1328 1329 1330 1331 1332
	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
1333 1334 1335 1336 1337 1338
	inet6_destroy_sock(sk);
}

/*
 *	Socket option code for UDP
 */
1339
int udpv6_setsockopt(struct sock *sk, int level, int optname,
1340
		     char __user *optval, unsigned int optlen)
1341
{
1342
	if (level == SOL_UDP  ||  level == SOL_UDPLITE)
1343 1344
		return udp_lib_setsockopt(sk, level, optname, optval, optlen,
					  udp_v6_push_pending_frames);
1345
	return ipv6_setsockopt(sk, level, optname, optval, optlen);
1346 1347 1348
}

#ifdef CONFIG_COMPAT
1349
int compat_udpv6_setsockopt(struct sock *sk, int level, int optname,
1350
			    char __user *optval, unsigned int optlen)
1351
{
1352
	if (level == SOL_UDP  ||  level == SOL_UDPLITE)
1353 1354
		return udp_lib_setsockopt(sk, level, optname, optval, optlen,
					  udp_v6_push_pending_frames);
1355
	return compat_ipv6_setsockopt(sk, level, optname, optval, optlen);
1356 1357 1358
}
#endif

1359 1360
int udpv6_getsockopt(struct sock *sk, int level, int optname,
		     char __user *optval, int __user *optlen)
1361
{
1362
	if (level == SOL_UDP  ||  level == SOL_UDPLITE)
1363
		return udp_lib_getsockopt(sk, level, optname, optval, optlen);
1364
	return ipv6_getsockopt(sk, level, optname, optval, optlen);
1365 1366 1367
}

#ifdef CONFIG_COMPAT
1368 1369
int compat_udpv6_getsockopt(struct sock *sk, int level, int optname,
			    char __user *optval, int __user *optlen)
1370
{
1371
	if (level == SOL_UDP  ||  level == SOL_UDPLITE)
1372
		return udp_lib_getsockopt(sk, level, optname, optval, optlen);
1373
	return compat_ipv6_getsockopt(sk, level, optname, optval, optlen);
1374 1375 1376
}
#endif

1377
static const struct inet6_protocol udpv6_protocol = {
L
Linus Torvalds 已提交
1378 1379 1380 1381 1382 1383 1384
	.handler	=	udpv6_rcv,
	.err_handler	=	udpv6_err,
	.flags		=	INET6_PROTO_NOPOLICY|INET6_PROTO_FINAL,
};

/* ------------------------------------------------------------------------ */
#ifdef CONFIG_PROC_FS
1385
int udp6_seq_show(struct seq_file *seq, void *v)
L
Linus Torvalds 已提交
1386
{
1387 1388 1389 1390 1391 1392 1393 1394 1395
	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
1396 1397 1398
	return 0;
}

1399 1400 1401 1402 1403 1404 1405 1406
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
1407 1408 1409
static struct udp_seq_afinfo udp6_seq_afinfo = {
	.name		= "udp6",
	.family		= AF_INET6,
1410
	.udp_table	= &udp_table,
1411
	.seq_fops	= &udp6_afinfo_seq_fops,
1412 1413 1414
	.seq_ops	= {
		.show		= udp6_seq_show,
	},
L
Linus Torvalds 已提交
1415 1416
};

1417
int __net_init udp6_proc_init(struct net *net)
L
Linus Torvalds 已提交
1418
{
1419
	return udp_proc_register(net, &udp6_seq_afinfo);
L
Linus Torvalds 已提交
1420 1421
}

I
Ian Morris 已提交
1422 1423
void udp6_proc_exit(struct net *net)
{
1424
	udp_proc_unregister(net, &udp6_seq_afinfo);
L
Linus Torvalds 已提交
1425 1426 1427 1428 1429 1430
}
#endif /* CONFIG_PROC_FS */

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

struct proto udpv6_prot = {
1431 1432
	.name		   = "UDPv6",
	.owner		   = THIS_MODULE,
1433
	.close		   = udp_lib_close,
1434 1435 1436 1437 1438 1439 1440 1441
	.connect	   = ip6_datagram_connect,
	.disconnect	   = udp_disconnect,
	.ioctl		   = udp_ioctl,
	.destroy	   = udpv6_destroy_sock,
	.setsockopt	   = udpv6_setsockopt,
	.getsockopt	   = udpv6_getsockopt,
	.sendmsg	   = udpv6_sendmsg,
	.recvmsg	   = udpv6_recvmsg,
1442
	.backlog_rcv	   = __udpv6_queue_rcv_skb,
1443
	.release_cb	   = ip6_datagram_release_cb,
1444 1445
	.hash		   = udp_lib_hash,
	.unhash		   = udp_lib_unhash,
E
Eric Dumazet 已提交
1446
	.rehash		   = udp_v6_rehash,
1447
	.get_port	   = udp_v6_get_port,
H
Hideo Aoki 已提交
1448 1449 1450 1451
	.memory_allocated  = &udp_memory_allocated,
	.sysctl_mem	   = sysctl_udp_mem,
	.sysctl_wmem	   = &sysctl_udp_wmem_min,
	.sysctl_rmem	   = &sysctl_udp_rmem_min,
1452
	.obj_size	   = sizeof(struct udp6_sock),
1453
	.h.udp_table	   = &udp_table,
1454
#ifdef CONFIG_COMPAT
1455 1456
	.compat_setsockopt = compat_udpv6_setsockopt,
	.compat_getsockopt = compat_udpv6_getsockopt,
1457
#endif
1458
	.diag_destroy      = udp_abort,
L
Linus Torvalds 已提交
1459 1460 1461 1462 1463 1464 1465 1466 1467 1468
};

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

1469
int __init udpv6_init(void)
L
Linus Torvalds 已提交
1470
{
1471 1472
	int ret;

1473 1474
	ret = inet6_add_protocol(&udpv6_protocol, IPPROTO_UDP);
	if (ret)
1475
		goto out;
1476

1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487
	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;
}

1488
void udpv6_exit(void)
1489 1490 1491
{
	inet6_unregister_protosw(&udpv6_protosw);
	inet6_del_protocol(&udpv6_protocol, IPPROTO_UDP);
L
Linus Torvalds 已提交
1492
}