udp.c 38.9 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/inet_hashtables.h>
50
#include <net/inet6_hashtables.h>
51
#include <net/busy_poll.h>
52
#include <net/sock_reuseport.h>
L
Linus Torvalds 已提交
53 54 55

#include <linux/proc_fs.h>
#include <linux/seq_file.h>
56
#include <trace/events/skb.h>
57
#include "udp_impl.h"
L
Linus Torvalds 已提交
58

59 60 61 62 63 64 65 66 67 68
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;
}

69 70 71 72 73
static u32 udp6_ehashfn(const struct net *net,
			const struct in6_addr *laddr,
			const u16 lport,
			const struct in6_addr *faddr,
			const __be16 fport)
74
{
75 76 77 78 79 80 81 82 83 84 85 86 87
	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);

88
	return __inet6_ehashfn(lhash, lport, fhash, fport,
89
			       udp_ipv6_hash_secret + net_hash_mix(net));
90 91
}

92 93 94
static u32 udp6_portaddr_hash(const struct net *net,
			      const struct in6_addr *addr6,
			      unsigned int port)
95 96 97 98 99
{
	unsigned int hash, mix = net_hash_mix(net);

	if (ipv6_addr_any(addr6))
		hash = jhash_1word(0, mix);
B
Brian Haley 已提交
100
	else if (ipv6_addr_v4mapped(addr6))
101
		hash = jhash_1word((__force u32)addr6->s6_addr32[3], mix);
102
	else
103
		hash = jhash2((__force u32 *)addr6->s6_addr32, 4, mix);
104 105 106 107

	return hash ^ port;
}

108
int udp_v6_get_port(struct sock *sk, unsigned short snum)
L
Linus Torvalds 已提交
109
{
E
Eric Dumazet 已提交
110 111
	unsigned int hash2_nulladdr =
		udp6_portaddr_hash(sock_net(sk), &in6addr_any, snum);
112
	unsigned int hash2_partial =
113
		udp6_portaddr_hash(sock_net(sk), &sk->sk_v6_rcv_saddr, 0);
E
Eric Dumazet 已提交
114

115
	/* precompute partial secondary hash */
E
Eric Dumazet 已提交
116
	udp_sk(sk)->udp_portaddr_hash = hash2_partial;
117
	return udp_lib_get_port(sk, snum, hash2_nulladdr);
L
Linus Torvalds 已提交
118 119
}

E
Eric Dumazet 已提交
120 121 122
static void udp_v6_rehash(struct sock *sk)
{
	u16 new_hash = udp6_portaddr_hash(sock_net(sk),
123
					  &sk->sk_v6_rcv_saddr,
E
Eric Dumazet 已提交
124 125 126 127 128
					  inet_sk(sk)->inet_num);

	udp_lib_rehash(sk, new_hash);
}

129 130 131
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,
132
			 int dif, bool exact_dif)
133
{
134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162
	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++;
	}

163
	if (sk->sk_bound_dev_if || exact_dif) {
164 165 166
		if (sk->sk_bound_dev_if != dif)
			return -1;
		score++;
167
	}
168

169 170 171
	if (sk->sk_incoming_cpu == raw_smp_processor_id())
		score++;

172 173 174
	return score;
}

175
/* called with rcu_read_lock() */
E
Eric Dumazet 已提交
176 177 178
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,
179
		bool exact_dif, struct udp_hslot *hslot2,
180
		struct sk_buff *skb)
E
Eric Dumazet 已提交
181 182
{
	struct sock *sk, *result;
183 184
	int score, badness, matches = 0, reuseport = 0;
	u32 hash = 0;
E
Eric Dumazet 已提交
185 186 187

	result = NULL;
	badness = -1;
188
	udp_portaddr_for_each_entry_rcu(sk, &hslot2->head) {
189
		score = compute_score(sk, net, saddr, sport,
190
				      daddr, hnum, dif, exact_dif);
E
Eric Dumazet 已提交
191
		if (score > badness) {
192 193
			reuseport = sk->sk_reuseport;
			if (reuseport) {
194 195
				hash = udp6_ehashfn(net, daddr, hnum,
						    saddr, sport);
196

197
				result = reuseport_select_sock(sk, hash, skb,
198
							sizeof(struct udphdr));
199 200
				if (result)
					return result;
201
				matches = 1;
202
			}
203 204
			result = sk;
			badness = score;
205 206
		} else if (score == badness && reuseport) {
			matches++;
207
			if (reciprocal_scale(hash, matches) == 0)
208 209
				result = sk;
			hash = next_pseudo_random32(hash);
E
Eric Dumazet 已提交
210 211 212 213 214
		}
	}
	return result;
}

215
/* rcu_read_lock() must be held */
216
struct sock *__udp6_lib_lookup(struct net *net,
217 218
				      const struct in6_addr *saddr, __be16 sport,
				      const struct in6_addr *daddr, __be16 dport,
219 220
				      int dif, struct udp_table *udptable,
				      struct sk_buff *skb)
L
Linus Torvalds 已提交
221
{
222
	struct sock *sk, *result;
L
Linus Torvalds 已提交
223
	unsigned short hnum = ntohs(dport);
E
Eric Dumazet 已提交
224 225
	unsigned int hash2, slot2, slot = udp_hashfn(net, hnum, udptable->mask);
	struct udp_hslot *hslot2, *hslot = &udptable->hash[slot];
226
	bool exact_dif = udp6_lib_exact_dif_match(net, skb);
227 228
	int score, badness, matches = 0, reuseport = 0;
	u32 hash = 0;
229

E
Eric Dumazet 已提交
230 231 232 233 234 235 236 237
	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,
238
					  daddr, hnum, dif, exact_dif,
239
					  hslot2, skb);
E
Eric Dumazet 已提交
240
		if (!result) {
241
			unsigned int old_slot2 = slot2;
E
Eric Dumazet 已提交
242 243
			hash2 = udp6_portaddr_hash(net, &in6addr_any, hnum);
			slot2 = hash2 & udptable->mask;
244 245 246 247
			/* avoid searching the same slot again. */
			if (unlikely(slot2 == old_slot2))
				return result;

E
Eric Dumazet 已提交
248 249 250 251
			hslot2 = &udptable->hash2[slot2];
			if (hslot->count < hslot2->count)
				goto begin;

252
			result = udp6_lib_lookup2(net, saddr, sport,
253
						  daddr, hnum, dif,
254 255
						  exact_dif, hslot2,
						  skb);
E
Eric Dumazet 已提交
256 257 258
		}
		return result;
	}
259 260 261
begin:
	result = NULL;
	badness = -1;
262
	sk_for_each_rcu(sk, &hslot->head) {
263 264
		score = compute_score(sk, net, saddr, sport, daddr, hnum, dif,
				      exact_dif);
265
		if (score > badness) {
266 267
			reuseport = sk->sk_reuseport;
			if (reuseport) {
268 269
				hash = udp6_ehashfn(net, daddr, hnum,
						    saddr, sport);
270
				result = reuseport_select_sock(sk, hash, skb,
271
							sizeof(struct udphdr));
272 273
				if (result)
					return result;
274 275
				matches = 1;
			}
276 277
			result = sk;
			badness = score;
278 279
		} else if (score == badness && reuseport) {
			matches++;
280
			if (reciprocal_scale(hash, matches) == 0)
281 282
				result = sk;
			hash = next_pseudo_random32(hash);
L
Linus Torvalds 已提交
283 284 285 286
		}
	}
	return result;
}
287
EXPORT_SYMBOL_GPL(__udp6_lib_lookup);
L
Linus Torvalds 已提交
288

289 290
static struct sock *__udp6_lib_lookup_skb(struct sk_buff *skb,
					  __be16 sport, __be16 dport,
291
					  struct udp_table *udptable)
292
{
293
	const struct ipv6hdr *iph = ipv6_hdr(skb);
294
	struct sock *sk;
295

296 297
	sk = skb_steal_sock(skb);
	if (unlikely(sk))
298
		return sk;
299
	return __udp6_lib_lookup(dev_net(skb->dev), &iph->saddr, sport,
E
Eric Dumazet 已提交
300
				 &iph->daddr, dport, inet6_iif(skb),
301
				 udptable, skb);
302 303
}

304 305 306 307 308
struct sock *udp6_lib_lookup_skb(struct sk_buff *skb,
				 __be16 sport, __be16 dport)
{
	const struct ipv6hdr *iph = ipv6_hdr(skb);

309
	return __udp6_lib_lookup(dev_net(skb->dev), &iph->saddr, sport,
310 311 312 313 314
				 &iph->daddr, dport, inet6_iif(skb),
				 &udp_table, skb);
}
EXPORT_SYMBOL_GPL(udp6_lib_lookup_skb);

315 316 317 318
/* Must be called under rcu_read_lock().
 * Does increment socket refcount.
 */
#if IS_ENABLED(CONFIG_NETFILTER_XT_MATCH_SOCKET) || \
319 320
    IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TPROXY) || \
    IS_ENABLED(CONFIG_NF_SOCKET_IPV6)
321 322 323
struct sock *udp6_lib_lookup(struct net *net, const struct in6_addr *saddr, __be16 sport,
			     const struct in6_addr *daddr, __be16 dport, int dif)
{
324 325 326 327 328 329 330
	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;
331 332
}
EXPORT_SYMBOL_GPL(udp6_lib_lookup);
333
#endif
334

L
Linus Torvalds 已提交
335
/*
336 337
 *	This should be easy, if there is something there we
 *	return it, otherwise we block.
L
Linus Torvalds 已提交
338 339
 */

340
int udpv6_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
L
Linus Torvalds 已提交
341 342 343 344
		  int noblock, int flags, int *addr_len)
{
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct inet_sock *inet = inet_sk(sk);
345
	struct sk_buff *skb;
346
	unsigned int ulen, copied;
347
	int peeked, peeking, off;
348 349
	int err;
	int is_udplite = IS_UDPLITE(sk);
350
	bool checksum_valid = false;
351
	int is_udp4;
L
Linus Torvalds 已提交
352 353

	if (flags & MSG_ERRQUEUE)
354
		return ipv6_recv_error(sk, msg, len, addr_len);
L
Linus Torvalds 已提交
355

356
	if (np->rxpmtu && np->rxopt.bits.rxpmtu)
357
		return ipv6_recv_rxpmtu(sk, msg, len, addr_len);
358

L
Linus Torvalds 已提交
359
try_again:
360
	peeking = off = sk_peek_offset(sk, flags);
361
	skb = __skb_recv_udp(sk, flags, noblock, &peeked, &off, &err);
L
Linus Torvalds 已提交
362
	if (!skb)
363
		return err;
L
Linus Torvalds 已提交
364

365
	ulen = skb->len;
366
	copied = len;
367 368
	if (copied > ulen - off)
		copied = ulen - off;
369
	else if (copied < ulen)
370
		msg->msg_flags |= MSG_TRUNC;
L
Linus Torvalds 已提交
371

372 373
	is_udp4 = (skb->protocol == htons(ETH_P_IP));

374
	/*
375 376 377
	 * 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.
378 379
	 */

380 381
	if (copied < ulen || peeking ||
	    (is_udplite && UDP_SKB_CB(skb)->partial_cov)) {
382 383
		checksum_valid = !udp_lib_checksum_complete(skb);
		if (!checksum_valid)
L
Linus Torvalds 已提交
384
			goto csum_copy_err;
385 386
	}

387
	if (checksum_valid || skb_csum_unnecessary(skb))
388
		err = skb_copy_datagram_msg(skb, off, msg, copied);
389
	else {
390
		err = skb_copy_and_csum_datagram_msg(skb, off, msg);
L
Linus Torvalds 已提交
391 392 393
		if (err == -EINVAL)
			goto csum_copy_err;
	}
394
	if (unlikely(err)) {
395 396 397
		if (!peeked) {
			atomic_inc(&sk->sk_drops);
			if (is_udp4)
398 399
				UDP_INC_STATS(sock_net(sk), UDP_MIB_INERRORS,
					      is_udplite);
400
			else
401 402
				UDP6_INC_STATS(sock_net(sk), UDP_MIB_INERRORS,
					       is_udplite);
403
		}
404
		kfree_skb(skb);
405
		return err;
406
	}
407 408
	if (!peeked) {
		if (is_udp4)
409 410
			UDP_INC_STATS(sock_net(sk), UDP_MIB_INDATAGRAMS,
				      is_udplite);
411
		else
412 413
			UDP6_INC_STATS(sock_net(sk), UDP_MIB_INDATAGRAMS,
				       is_udplite);
414
	}
415

416
	sock_recv_ts_and_drops(msg, sk, skb);
L
Linus Torvalds 已提交
417 418 419

	/* Copy the address. */
	if (msg->msg_name) {
420
		DECLARE_SOCKADDR(struct sockaddr_in6 *, sin6, msg->msg_name);
L
Linus Torvalds 已提交
421
		sin6->sin6_family = AF_INET6;
422
		sin6->sin6_port = udp_hdr(skb)->source;
L
Linus Torvalds 已提交
423 424
		sin6->sin6_flowinfo = 0;

425
		if (is_udp4) {
B
Brian Haley 已提交
426 427
			ipv6_addr_set_v4mapped(ip_hdr(skb)->saddr,
					       &sin6->sin6_addr);
428 429
			sin6->sin6_scope_id = 0;
		} else {
A
Alexey Dobriyan 已提交
430
			sin6->sin6_addr = ipv6_hdr(skb)->saddr;
431 432
			sin6->sin6_scope_id =
				ipv6_iface_scope_id(&sin6->sin6_addr,
433
						    inet6_iif(skb));
L
Linus Torvalds 已提交
434
		}
435
		*addr_len = sizeof(*sin6);
L
Linus Torvalds 已提交
436
	}
437 438 439 440

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

441
	if (is_udp4) {
L
Linus Torvalds 已提交
442
		if (inet->cmsg_flags)
443
			ip_cmsg_recv_offset(msg, sk, skb,
E
Eric Dumazet 已提交
444
					    sizeof(struct udphdr), off);
L
Linus Torvalds 已提交
445 446
	} else {
		if (np->rxopt.all)
447
			ip6_datagram_recv_specific_ctl(sk, msg, skb);
448
	}
L
Linus Torvalds 已提交
449

450
	err = copied;
L
Linus Torvalds 已提交
451
	if (flags & MSG_TRUNC)
452
		err = ulen;
L
Linus Torvalds 已提交
453

454
	skb_consume_udp(sk, skb, peeking ? -err : err);
L
Linus Torvalds 已提交
455 456 457
	return err;

csum_copy_err:
458
	if (!__sk_queue_drop_skb(sk, skb, flags, udp_skb_destructor)) {
459
		if (is_udp4) {
460 461 462 463
			UDP_INC_STATS(sock_net(sk),
				      UDP_MIB_CSUMERRORS, is_udplite);
			UDP_INC_STATS(sock_net(sk),
				      UDP_MIB_INERRORS, is_udplite);
464
		} else {
465 466 467 468
			UDP6_INC_STATS(sock_net(sk),
				       UDP_MIB_CSUMERRORS, is_udplite);
			UDP6_INC_STATS(sock_net(sk),
				       UDP_MIB_INERRORS, is_udplite);
469
		}
470
	}
471
	kfree_skb(skb);
L
Linus Torvalds 已提交
472

473 474
	/* starting over for a new packet, but check if we need to yield */
	cond_resched();
475
	msg->msg_flags &= ~MSG_TRUNC;
L
Linus Torvalds 已提交
476 477 478
	goto try_again;
}

479
void __udp6_lib_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
480
		    u8 type, u8 code, int offset, __be32 info,
481
		    struct udp_table *udptable)
L
Linus Torvalds 已提交
482 483
{
	struct ipv6_pinfo *np;
484 485 486
	const struct ipv6hdr *hdr = (const struct ipv6hdr *)skb->data;
	const struct in6_addr *saddr = &hdr->saddr;
	const struct in6_addr *daddr = &hdr->daddr;
487
	struct udphdr *uh = (struct udphdr *)(skb->data+offset);
L
Linus Torvalds 已提交
488
	struct sock *sk;
489
	int harderr;
L
Linus Torvalds 已提交
490
	int err;
491
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
492

493
	sk = __udp6_lib_lookup(net, daddr, uh->dest, saddr, uh->source,
494
			       inet6_iif(skb), udptable, skb);
495
	if (!sk) {
E
Eric Dumazet 已提交
496 497
		__ICMP6_INC_STATS(net, __in6_dev_get(skb->dev),
				  ICMP6_MIB_INERRORS);
L
Linus Torvalds 已提交
498
		return;
499
	}
L
Linus Torvalds 已提交
500

501 502 503
	harderr = icmpv6_err_convert(type, code, &err);
	np = inet6_sk(sk);

504 505 506
	if (type == ICMPV6_PKT_TOOBIG) {
		if (!ip6_sk_accept_pmtu(sk))
			goto out;
507
		ip6_sk_update_pmtu(skb, sk, info);
508 509
		if (np->pmtudisc != IPV6_PMTUDISC_DONT)
			harderr = 1;
510
	}
511
	if (type == NDISC_REDIRECT) {
512
		ip6_sk_redirect(skb, sk);
513 514
		goto out;
	}
515

516 517 518 519
	if (!np->recverr) {
		if (!harderr || sk->sk_state != TCP_ESTABLISHED)
			goto out;
	} else {
L
Linus Torvalds 已提交
520
		ipv6_icmp_error(sk, skb, err, uh->dest, ntohl(info), (u8 *)(uh+1));
521
	}
522

L
Linus Torvalds 已提交
523 524 525
	sk->sk_err = err;
	sk->sk_error_report(sk);
out:
526
	return;
L
Linus Torvalds 已提交
527 528
}

529
static int __udpv6_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
530 531 532
{
	int rc;

533
	if (!ipv6_addr_any(&sk->sk_v6_daddr)) {
534
		sock_rps_save_rxhash(sk, skb);
535
		sk_mark_napi_id(sk, skb);
E
Eric Dumazet 已提交
536
		sk_incoming_cpu_update(sk);
537 538
	} else {
		sk_mark_napi_id_once(sk, skb);
539
	}
540

541
	rc = __udp_enqueue_schedule_skb(sk, skb);
542 543 544 545 546
	if (rc < 0) {
		int is_udplite = IS_UDPLITE(sk);

		/* Note that an ENOMEM error is charged twice */
		if (rc == -ENOMEM)
547
			UDP6_INC_STATS(sock_net(sk),
548
					 UDP_MIB_RCVBUFERRORS, is_udplite);
549
		UDP6_INC_STATS(sock_net(sk), UDP_MIB_INERRORS, is_udplite);
550 551 552
		kfree_skb(skb);
		return -1;
	}
553

554 555 556
	return 0;
}

557
static __inline__ void udpv6_err(struct sk_buff *skb,
558
				 struct inet6_skb_parm *opt, u8 type,
559
				 u8 code, int offset, __be32 info)
560
{
561
	__udp6_lib_err(skb, opt, type, code, offset, info, &udp_table);
562 563
}

564 565 566 567 568 569 570 571
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);

572
static int udpv6_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
L
Linus Torvalds 已提交
573
{
574
	struct udp_sock *up = udp_sk(sk);
W
Wang Chen 已提交
575
	int is_udplite = IS_UDPLITE(sk);
576

577 578
	if (!xfrm6_policy_check(sk, XFRM_POLICY_IN, skb))
		goto drop;
L
Linus Torvalds 已提交
579

580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595
	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);
596
		if (encap_rcv) {
597 598
			int ret;

599 600 601 602
			/* Verify checksum before giving to encap */
			if (udp_lib_checksum_complete(skb))
				goto csum_error;

603 604
			ret = encap_rcv(sk, skb);
			if (ret <= 0) {
605 606 607
				__UDP_INC_STATS(sock_net(sk),
						UDP_MIB_INDATAGRAMS,
						is_udplite);
608 609 610 611 612 613 614
				return -ret;
			}
		}

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

615 616 617
	/*
	 * UDP-Lite specific tests, ignored on UDP sockets (see net/ipv4/udp.c).
	 */
W
Wang Chen 已提交
618
	if ((is_udplite & UDPLITE_RECV_CC)  &&  UDP_SKB_CB(skb)->partial_cov) {
619 620

		if (up->pcrlen == 0) {          /* full coverage was set  */
621 622
			net_dbg_ratelimited("UDPLITE6: partial coverage %d while full coverage %d requested\n",
					    UDP_SKB_CB(skb)->cscov, skb->len);
623 624 625
			goto drop;
		}
		if (UDP_SKB_CB(skb)->cscov  <  up->pcrlen) {
626 627
			net_dbg_ratelimited("UDPLITE6: coverage %d too small, need min %d\n",
					    UDP_SKB_CB(skb)->cscov, up->pcrlen);
628 629
			goto drop;
		}
L
Linus Torvalds 已提交
630 631
	}

632 633 634 635
	if (rcu_access_pointer(sk->sk_filter) &&
	    udp_lib_checksum_complete(skb))
		goto csum_error;

636
	if (sk_filter_trim_cap(sk, skb, sizeof(struct udphdr)))
637
		goto drop;
638

639
	udp_csum_pull_header(skb);
640

641
	skb_dst_drop(skb);
642

643
	return __udpv6_queue_rcv_skb(sk, skb);
644

645
csum_error:
646
	__UDP6_INC_STATS(sock_net(sk), UDP_MIB_CSUMERRORS, is_udplite);
647
drop:
648
	__UDP6_INC_STATS(sock_net(sk), UDP_MIB_INERRORS, is_udplite);
649
	atomic_inc(&sk->sk_drops);
650 651
	kfree_skb(skb);
	return -1;
L
Linus Torvalds 已提交
652 653
}

654 655 656 657
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
658
{
659
	struct inet_sock *inet = inet_sk(sk);
L
Linus Torvalds 已提交
660

661 662 663 664 665 666 667 668
	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)) ||
669 670 671
	    (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)))
672 673 674 675
		return false;
	if (!inet6_mc_check(sk, loc_addr, rmt_addr))
		return false;
	return true;
L
Linus Torvalds 已提交
676 677
}

678 679 680 681 682
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.
	 */
683 684 685
	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));
686 687
}

L
Linus Torvalds 已提交
688 689 690 691
/*
 * Note: called only from the BH handler context,
 * so we don't need to lock the hashes.
 */
692
static int __udp6_lib_mcast_deliver(struct net *net, struct sk_buff *skb,
693
		const struct in6_addr *saddr, const struct in6_addr *daddr,
694
		struct udp_table *udptable, int proto)
L
Linus Torvalds 已提交
695
{
696
	struct sock *sk, *first = NULL;
697
	const struct udphdr *uh = udp_hdr(skb);
698 699
	unsigned short hnum = ntohs(uh->dest);
	struct udp_hslot *hslot = udp_hashslot(udptable, net, hnum);
700
	unsigned int offset = offsetof(typeof(*sk), sk_node);
701
	unsigned int hash2 = 0, hash2_any = 0, use_hash2 = (hslot->count > 10);
702 703 704
	int dif = inet6_iif(skb);
	struct hlist_node *node;
	struct sk_buff *nskb;
705 706 707

	if (use_hash2) {
		hash2_any = udp6_portaddr_hash(net, &in6addr_any, hnum) &
708 709
			    udptable->mask;
		hash2 = udp6_portaddr_hash(net, daddr, hnum) & udptable->mask;
710
start_lookup:
711
		hslot = &udptable->hash2[hash2];
712 713
		offset = offsetof(typeof(*sk), __sk_common.skc_portaddr_node);
	}
L
Linus Torvalds 已提交
714

715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730
	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);
731 732 733 734
			__UDP6_INC_STATS(net, UDP_MIB_RCVBUFERRORS,
					 IS_UDPLITE(sk));
			__UDP6_INC_STATS(net, UDP_MIB_INERRORS,
					 IS_UDPLITE(sk));
735
			continue;
H
Hideo Aoki 已提交
736
		}
737

738 739 740
		if (udpv6_queue_rcv_skb(sk, nskb) > 0)
			consume_skb(nskb);
	}
741

742 743 744 745 746 747
	/* 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;
	}

748 749 750
	if (first) {
		if (udpv6_queue_rcv_skb(first, skb) > 0)
			consume_skb(skb);
751
	} else {
752
		kfree_skb(skb);
753 754
		__UDP6_INC_STATS(net, UDP_MIB_IGNOREDMULTI,
				 proto == IPPROTO_UDPLITE);
755
	}
756
	return 0;
L
Linus Torvalds 已提交
757 758
}

759
int __udp6_lib_rcv(struct sk_buff *skb, struct udp_table *udptable,
760
		   int proto)
L
Linus Torvalds 已提交
761
{
762
	const struct in6_addr *saddr, *daddr;
763
	struct net *net = dev_net(skb->dev);
764
	struct udphdr *uh;
765
	struct sock *sk;
L
Linus Torvalds 已提交
766 767 768
	u32 ulen = 0;

	if (!pskb_may_pull(skb, sizeof(struct udphdr)))
769
		goto discard;
L
Linus Torvalds 已提交
770

771 772
	saddr = &ipv6_hdr(skb)->saddr;
	daddr = &ipv6_hdr(skb)->daddr;
773
	uh = udp_hdr(skb);
L
Linus Torvalds 已提交
774 775

	ulen = ntohs(uh->len);
776 777
	if (ulen > skb->len)
		goto short_packet;
L
Linus Torvalds 已提交
778

779 780
	if (proto == IPPROTO_UDP) {
		/* UDP validates ulen. */
L
Linus Torvalds 已提交
781

782 783 784
		/* Check for jumbo payload */
		if (ulen == 0)
			ulen = skb->len;
L
Linus Torvalds 已提交
785

786 787
		if (ulen < sizeof(*uh))
			goto short_packet;
L
Linus Torvalds 已提交
788

789 790 791
		if (ulen < skb->len) {
			if (pskb_trim_rcsum(skb, ulen))
				goto short_packet;
792 793
			saddr = &ipv6_hdr(skb)->saddr;
			daddr = &ipv6_hdr(skb)->daddr;
794
			uh = udp_hdr(skb);
795 796
		}
	}
L
Linus Torvalds 已提交
797

798
	if (udp6_csum_init(skb, uh, proto))
799
		goto csum_error;
800

801 802
	/*
	 *	Multicast receive code
L
Linus Torvalds 已提交
803
	 */
804
	if (ipv6_addr_is_multicast(daddr))
805
		return __udp6_lib_mcast_deliver(net, skb,
806
				saddr, daddr, udptable, proto);
L
Linus Torvalds 已提交
807 808 809

	/* Unicast */

810
	/*
L
Linus Torvalds 已提交
811 812 813
	 * check socket cache ... must talk to Alan about his plans
	 * for sock caches... i'll skip this for now.
	 */
814
	sk = __udp6_lib_lookup_skb(skb, uh->source, uh->dest, udptable);
815
	if (sk) {
816 817
		int ret;

818
		if (!uh->check && !udp_sk(sk)->no_check6_rx) {
819 820 821 822
			udp6_csum_zero_error(skb);
			goto csum_error;
		}

823
		if (inet_get_convert_csum(sk) && uh->check && !IS_UDPLITE(sk))
824 825 826
			skb_checksum_try_convert(skb, IPPROTO_UDP, uh->check,
						 ip6_compute_pseudo);

827
		ret = udpv6_queue_rcv_skb(sk, skb);
L
Linus Torvalds 已提交
828

829
		/* a return value > 0 means to resubmit the input */
830
		if (ret > 0)
831
			return ret;
L
Linus Torvalds 已提交
832

S
Stephen Hemminger 已提交
833
		return 0;
L
Linus Torvalds 已提交
834
	}
835

836 837 838 839 840
	if (!uh->check) {
		udp6_csum_zero_error(skb);
		goto csum_error;
	}

841
	if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb))
842
		goto discard;
843 844

	if (udp_lib_checksum_complete(skb))
845
		goto csum_error;
846

847
	__UDP6_INC_STATS(net, UDP_MIB_NOPORTS, proto == IPPROTO_UDPLITE);
848 849 850
	icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_PORT_UNREACH, 0);

	kfree_skb(skb);
S
Stephen Hemminger 已提交
851
	return 0;
L
Linus Torvalds 已提交
852

853
short_packet:
854 855 856 857 858
	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));
859 860
	goto discard;
csum_error:
861
	__UDP6_INC_STATS(net, UDP_MIB_CSUMERRORS, proto == IPPROTO_UDPLITE);
L
Linus Torvalds 已提交
862
discard:
863
	__UDP6_INC_STATS(net, UDP_MIB_INERRORS, proto == IPPROTO_UDPLITE);
L
Linus Torvalds 已提交
864
	kfree_skb(skb);
S
Stephen Hemminger 已提交
865
	return 0;
L
Linus Torvalds 已提交
866
}
867

868

869 870 871 872 873
static struct sock *__udp6_lib_demux_lookup(struct net *net,
			__be16 loc_port, const struct in6_addr *loc_addr,
			__be16 rmt_port, const struct in6_addr *rmt_addr,
			int dif)
{
874 875 876 877 878
	unsigned short hnum = ntohs(loc_port);
	unsigned int hash2 = udp6_portaddr_hash(net, loc_addr, hnum);
	unsigned int slot2 = hash2 & udp_table.mask;
	struct udp_hslot *hslot2 = &udp_table.hash2[slot2];
	const __portpair ports = INET_COMBINED_PORTS(rmt_port, hnum);
879 880
	struct sock *sk;

881 882 883 884 885 886 887
	udp_portaddr_for_each_entry_rcu(sk, &hslot2->head) {
		if (INET6_MATCH(sk, net, rmt_addr, loc_addr, ports, dif))
			return sk;
		/* Only check first socket in chain */
		break;
	}
	return NULL;
888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911
}

static void udp_v6_early_demux(struct sk_buff *skb)
{
	struct net *net = dev_net(skb->dev);
	const struct udphdr *uh;
	struct sock *sk;
	struct dst_entry *dst;
	int dif = skb->dev->ifindex;

	if (!pskb_may_pull(skb, skb_transport_offset(skb) +
	    sizeof(struct udphdr)))
		return;

	uh = udp_hdr(skb);

	if (skb->pkt_type == PACKET_HOST)
		sk = __udp6_lib_demux_lookup(net, uh->dest,
					     &ipv6_hdr(skb)->daddr,
					     uh->source, &ipv6_hdr(skb)->saddr,
					     dif);
	else
		return;

912
	if (!sk || !atomic_inc_not_zero_hint(&sk->sk_refcnt, 2))
913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930
		return;

	skb->sk = sk;
	skb->destructor = sock_efree;
	dst = READ_ONCE(sk->sk_rx_dst);

	if (dst)
		dst = dst_check(dst, inet6_sk(sk)->rx_dst_cookie);
	if (dst) {
		if (dst->flags & DST_NOCACHE) {
			if (likely(atomic_inc_not_zero(&dst->__refcnt)))
				skb_dst_set(skb, dst);
		} else {
			skb_dst_set_noref(skb, dst);
		}
	}
}

931
static __inline__ int udpv6_rcv(struct sk_buff *skb)
932
{
933
	return __udp6_lib_rcv(skb, &udp_table, IPPROTO_UDP);
934 935
}

L
Linus Torvalds 已提交
936 937 938 939 940 941 942
/*
 * 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);

943 944 945
	if (up->pending == AF_INET)
		udp_flush_pending_frames(sk);
	else if (up->pending) {
L
Linus Torvalds 已提交
946 947 948
		up->len = 0;
		up->pending = 0;
		ip6_flush_pending_frames(sk);
949
	}
L
Linus Torvalds 已提交
950 951
}

952
/**
953 954 955 956
 *	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)
957 958 959 960 961 962 963
 */
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
964
	struct sk_buff *frags = skb_shinfo(skb)->frag_list;
965 966
	__wsum csum = 0;

V
Vlad Yasevich 已提交
967
	if (!frags) {
968 969 970 971 972 973 974 975 976 977 978 979 980 981 982
		/* 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
983 984 985
		do {
			csum = csum_add(csum, frags->csum);
		} while ((frags = frags->next));
986 987 988 989 990 991 992 993

		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
994 995 996 997
/*
 *	Sending
 */

V
Vlad Yasevich 已提交
998
static int udp_v6_send_skb(struct sk_buff *skb, struct flowi6 *fl6)
L
Linus Torvalds 已提交
999
{
V
Vlad Yasevich 已提交
1000
	struct sock *sk = skb->sk;
L
Linus Torvalds 已提交
1001 1002
	struct udphdr *uh;
	int err = 0;
W
Wang Chen 已提交
1003
	int is_udplite = IS_UDPLITE(sk);
1004
	__wsum csum = 0;
V
Vlad Yasevich 已提交
1005 1006
	int offset = skb_transport_offset(skb);
	int len = skb->len - offset;
L
Linus Torvalds 已提交
1007 1008 1009 1010

	/*
	 * Create a UDP header
	 */
1011
	uh = udp_hdr(skb);
1012 1013
	uh->source = fl6->fl6_sport;
	uh->dest = fl6->fl6_dport;
V
Vlad Yasevich 已提交
1014
	uh->len = htons(len);
L
Linus Torvalds 已提交
1015 1016
	uh->check = 0;

W
Wang Chen 已提交
1017
	if (is_udplite)
V
Vlad Yasevich 已提交
1018 1019
		csum = udplite_csum(skb);
	else if (udp_sk(sk)->no_check6_tx) {   /* UDP csum disabled */
1020 1021 1022
		skb->ip_summed = CHECKSUM_NONE;
		goto send;
	} else if (skb->ip_summed == CHECKSUM_PARTIAL) { /* UDP hardware csum */
V
Vlad Yasevich 已提交
1023
		udp6_hwcsum_outgoing(sk, skb, &fl6->saddr, &fl6->daddr, len);
1024 1025
		goto send;
	} else
V
Vlad Yasevich 已提交
1026
		csum = udp_csum(skb);
L
Linus Torvalds 已提交
1027

1028
	/* add protocol-dependent pseudo-header */
1029
	uh->check = csum_ipv6_magic(&fl6->saddr, &fl6->daddr,
V
Vlad Yasevich 已提交
1030
				    len, fl6->flowi6_proto, csum);
L
Linus Torvalds 已提交
1031
	if (uh->check == 0)
1032
		uh->check = CSUM_MANGLED_0;
L
Linus Torvalds 已提交
1033

1034
send:
V
Vlad Yasevich 已提交
1035
	err = ip6_send_skb(skb);
E
Eric Dumazet 已提交
1036 1037
	if (err) {
		if (err == -ENOBUFS && !inet6_sk(sk)->recverr) {
1038 1039
			UDP6_INC_STATS(sock_net(sk),
				       UDP_MIB_SNDBUFERRORS, is_udplite);
E
Eric Dumazet 已提交
1040 1041
			err = 0;
		}
1042 1043 1044 1045
	} else {
		UDP6_INC_STATS(sock_net(sk),
			       UDP_MIB_OUTDATAGRAMS, is_udplite);
	}
V
Vlad Yasevich 已提交
1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069
	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
1070 1071 1072 1073 1074 1075
out:
	up->len = 0;
	up->pending = 0;
	return err;
}

1076
int udpv6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
L
Linus Torvalds 已提交
1077 1078 1079 1080 1081
{
	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);
1082
	DECLARE_SOCKADDR(struct sockaddr_in6 *, sin6, msg->msg_name);
1083
	struct in6_addr *daddr, *final_p, final;
L
Linus Torvalds 已提交
1084
	struct ipv6_txoptions *opt = NULL;
1085
	struct ipv6_txoptions *opt_to_free = NULL;
L
Linus Torvalds 已提交
1086
	struct ip6_flowlabel *flowlabel = NULL;
1087
	struct flowi6 fl6;
L
Linus Torvalds 已提交
1088
	struct dst_entry *dst;
W
Wei Wang 已提交
1089
	struct ipcm6_cookie ipc6;
L
Linus Torvalds 已提交
1090 1091 1092 1093
	int addr_len = msg->msg_namelen;
	int ulen = len;
	int corkreq = up->corkflag || msg->msg_flags&MSG_MORE;
	int err;
1094
	int connected = 0;
W
Wang Chen 已提交
1095
	int is_udplite = IS_UDPLITE(sk);
1096
	int (*getfrag)(void *, char *, int, int, int, struct sk_buff *);
1097
	struct sockcm_cookie sockc;
L
Linus Torvalds 已提交
1098

W
Wei Wang 已提交
1099 1100 1101
	ipc6.hlimit = -1;
	ipc6.tclass = -1;
	ipc6.dontfrag = -1;
1102
	sockc.tsflags = sk->sk_tsflags;
W
Wei Wang 已提交
1103

L
Linus Torvalds 已提交
1104 1105 1106 1107 1108 1109 1110 1111 1112 1113
	/* 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;
1114 1115 1116 1117
			if (ipv6_addr_any(daddr) &&
			    ipv6_addr_v4mapped(&np->saddr))
				ipv6_addr_set_v4mapped(htonl(INADDR_LOOPBACK),
						       daddr);
L
Linus Torvalds 已提交
1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131
			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;
1132
		daddr = &sk->sk_v6_daddr;
1133
	} else
L
Linus Torvalds 已提交
1134 1135 1136
		daddr = NULL;

	if (daddr) {
B
Brian Haley 已提交
1137
		if (ipv6_addr_v4mapped(daddr)) {
L
Linus Torvalds 已提交
1138 1139
			struct sockaddr_in sin;
			sin.sin_family = AF_INET;
E
Eric Dumazet 已提交
1140
			sin.sin_port = sin6 ? sin6->sin6_port : inet->inet_dport;
L
Linus Torvalds 已提交
1141 1142 1143 1144 1145 1146
			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;
1147
			return udp_sendmsg(sk, msg, len);
L
Linus Torvalds 已提交
1148 1149 1150 1151
		}
	}

	if (up->pending == AF_INET)
1152
		return udp_sendmsg(sk, msg, len);
L
Linus Torvalds 已提交
1153 1154

	/* Rough check on arithmetic overflow,
1155
	   better check is made in ip6_append_data().
L
Linus Torvalds 已提交
1156 1157 1158
	   */
	if (len > INT_MAX - sizeof(struct udphdr))
		return -EMSGSIZE;
1159

1160
	getfrag  =  is_udplite ?  udplite_getfrag : ip_generic_getfrag;
L
Linus Torvalds 已提交
1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178
	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);

1179
	memset(&fl6, 0, sizeof(fl6));
L
Linus Torvalds 已提交
1180 1181 1182 1183 1184

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

1185
		fl6.fl6_dport = sin6->sin6_port;
L
Linus Torvalds 已提交
1186 1187 1188
		daddr = &sin6->sin6_addr;

		if (np->sndflow) {
1189 1190 1191
			fl6.flowlabel = sin6->sin6_flowinfo&IPV6_FLOWINFO_MASK;
			if (fl6.flowlabel&IPV6_FLOWLABEL_MASK) {
				flowlabel = fl6_sock_lookup(sk, fl6.flowlabel);
1192
				if (!flowlabel)
L
Linus Torvalds 已提交
1193 1194 1195 1196 1197 1198 1199 1200 1201
					return -EINVAL;
			}
		}

		/*
		 * Otherwise it will be difficult to maintain
		 * sk->sk_dst_cache.
		 */
		if (sk->sk_state == TCP_ESTABLISHED &&
1202 1203
		    ipv6_addr_equal(daddr, &sk->sk_v6_daddr))
			daddr = &sk->sk_v6_daddr;
L
Linus Torvalds 已提交
1204 1205 1206

		if (addr_len >= sizeof(struct sockaddr_in6) &&
		    sin6->sin6_scope_id &&
1207
		    __ipv6_addr_needs_scope_id(__ipv6_addr_type(daddr)))
1208
			fl6.flowi6_oif = sin6->sin6_scope_id;
L
Linus Torvalds 已提交
1209 1210 1211 1212
	} else {
		if (sk->sk_state != TCP_ESTABLISHED)
			return -EDESTADDRREQ;

1213
		fl6.fl6_dport = inet->inet_dport;
1214
		daddr = &sk->sk_v6_daddr;
1215
		fl6.flowlabel = np->flow_label;
1216
		connected = 1;
L
Linus Torvalds 已提交
1217 1218
	}

1219 1220
	if (!fl6.flowi6_oif)
		fl6.flowi6_oif = sk->sk_bound_dev_if;
L
Linus Torvalds 已提交
1221

1222 1223
	if (!fl6.flowi6_oif)
		fl6.flowi6_oif = np->sticky_pktinfo.ipi6_ifindex;
1224

1225
	fl6.flowi6_mark = sk->sk_mark;
1226
	fl6.flowi6_uid = sk->sk_uid;
B
Brian Haley 已提交
1227

L
Linus Torvalds 已提交
1228 1229 1230 1231
	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
1232
		ipc6.opt = opt;
L
Linus Torvalds 已提交
1233

W
Wei Wang 已提交
1234
		err = ip6_datagram_send_ctl(sock_net(sk), sk, msg, &fl6, &ipc6, &sockc);
L
Linus Torvalds 已提交
1235 1236 1237 1238
		if (err < 0) {
			fl6_sock_release(flowlabel);
			return err;
		}
1239 1240
		if ((fl6.flowlabel&IPV6_FLOWLABEL_MASK) && !flowlabel) {
			flowlabel = fl6_sock_lookup(sk, fl6.flowlabel);
1241
			if (!flowlabel)
L
Linus Torvalds 已提交
1242 1243 1244 1245
				return -EINVAL;
		}
		if (!(opt->opt_nflen|opt->opt_flen))
			opt = NULL;
1246
		connected = 0;
L
Linus Torvalds 已提交
1247
	}
1248 1249 1250 1251
	if (!opt) {
		opt = txopt_get(np);
		opt_to_free = opt;
	}
1252 1253 1254
	if (flowlabel)
		opt = fl6_merge_options(&opt_space, flowlabel, opt);
	opt = ipv6_fixup_options(&opt_space, opt);
W
Wei Wang 已提交
1255
	ipc6.opt = opt;
L
Linus Torvalds 已提交
1256

1257
	fl6.flowi6_proto = sk->sk_protocol;
1258
	if (!ipv6_addr_any(daddr))
A
Alexey Dobriyan 已提交
1259
		fl6.daddr = *daddr;
1260
	else
1261 1262
		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
1263
		fl6.saddr = np->saddr;
1264
	fl6.fl6_sport = inet->inet_sport;
1265

1266
	final_p = fl6_update_dst(&fl6, opt, &final);
1267
	if (final_p)
1268
		connected = 0;
L
Linus Torvalds 已提交
1269

1270 1271
	if (!fl6.flowi6_oif && ipv6_addr_is_multicast(&fl6.daddr)) {
		fl6.flowi6_oif = np->mcast_oif;
1272
		connected = 0;
1273 1274
	} else if (!fl6.flowi6_oif)
		fl6.flowi6_oif = np->ucast_oif;
L
Linus Torvalds 已提交
1275

1276
	security_sk_classify_flow(sk, flowi6_to_flowi(&fl6));
V
Venkat Yekkirala 已提交
1277

1278 1279 1280 1281 1282
	if (ipc6.tclass < 0)
		ipc6.tclass = np->tclass;

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

1283
	dst = ip6_sk_dst_lookup_flow(sk, &fl6, final_p);
1284 1285 1286
	if (IS_ERR(dst)) {
		err = PTR_ERR(dst);
		dst = NULL;
L
Linus Torvalds 已提交
1287
		goto out;
1288
	}
L
Linus Torvalds 已提交
1289

W
Wei Wang 已提交
1290 1291
	if (ipc6.hlimit < 0)
		ipc6.hlimit = ip6_sk_dst_hoplimit(np, &fl6, dst);
L
Linus Torvalds 已提交
1292 1293 1294 1295 1296

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

1297 1298 1299 1300 1301
	/* 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
1302
				   sizeof(struct udphdr), &ipc6,
1303
				   &fl6, (struct rt6_info *)dst,
W
Wei Wang 已提交
1304
				   msg->msg_flags, &sockc);
1305 1306 1307 1308 1309 1310
		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
1311 1312 1313 1314 1315 1316
	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);

1317
		net_dbg_ratelimited("udp cork app bug 2\n");
L
Linus Torvalds 已提交
1318 1319 1320 1321 1322 1323 1324
		err = -EINVAL;
		goto out;
	}

	up->pending = AF_INET6;

do_append_data:
W
Wei Wang 已提交
1325 1326
	if (ipc6.dontfrag < 0)
		ipc6.dontfrag = np->dontfrag;
L
Linus Torvalds 已提交
1327
	up->len += ulen;
W
Wei Wang 已提交
1328 1329 1330
	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
1331 1332 1333
	if (err)
		udp_v6_flush_pending_frames(sk);
	else if (!corkreq)
1334
		err = udp_v6_push_pending_frames(sk);
H
Herbert Xu 已提交
1335 1336
	else if (unlikely(skb_queue_empty(&sk->sk_write_queue)))
		up->pending = 0;
L
Linus Torvalds 已提交
1337

1338 1339 1340 1341 1342
	if (err > 0)
		err = np->recverr ? net_xmit_errno(err) : 0;
	release_sock(sk);

release_dst:
1343 1344 1345
	if (dst) {
		if (connected) {
			ip6_dst_store(sk, dst,
1346 1347
				      ipv6_addr_equal(&fl6.daddr, &sk->sk_v6_daddr) ?
				      &sk->sk_v6_daddr : NULL,
1348
#ifdef CONFIG_IPV6_SUBTREES
1349
				      ipv6_addr_equal(&fl6.saddr, &np->saddr) ?
1350 1351 1352
				      &np->saddr :
#endif
				      NULL);
1353 1354 1355
		} else {
			dst_release(dst);
		}
1356
		dst = NULL;
1357 1358
	}

L
Linus Torvalds 已提交
1359
out:
1360
	dst_release(dst);
L
Linus Torvalds 已提交
1361
	fl6_sock_release(flowlabel);
1362
	txopt_put(opt_to_free);
1363
	if (!err)
L
Linus Torvalds 已提交
1364
		return len;
1365 1366 1367 1368 1369 1370 1371 1372
	/*
	 * 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)) {
1373 1374
		UDP6_INC_STATS(sock_net(sk),
			       UDP_MIB_SNDBUFERRORS, is_udplite);
1375
	}
L
Linus Torvalds 已提交
1376 1377 1378
	return err;

do_confirm:
1379 1380
	if (msg->msg_flags & MSG_PROBE)
		dst_confirm_neigh(dst, &fl6.daddr);
L
Linus Torvalds 已提交
1381 1382 1383 1384 1385 1386
	if (!(msg->msg_flags&MSG_PROBE) || len)
		goto back_from_confirm;
	err = 0;
	goto out;
}

1387
void udpv6_destroy_sock(struct sock *sk)
L
Linus Torvalds 已提交
1388
{
T
Tom Parkin 已提交
1389
	struct udp_sock *up = udp_sk(sk);
L
Linus Torvalds 已提交
1390 1391 1392 1393
	lock_sock(sk);
	udp_v6_flush_pending_frames(sk);
	release_sock(sk);

T
Tom Parkin 已提交
1394 1395 1396 1397 1398 1399 1400
	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
1401 1402 1403 1404 1405 1406
	inet6_destroy_sock(sk);
}

/*
 *	Socket option code for UDP
 */
1407
int udpv6_setsockopt(struct sock *sk, int level, int optname,
1408
		     char __user *optval, unsigned int optlen)
1409
{
1410
	if (level == SOL_UDP  ||  level == SOL_UDPLITE)
1411 1412
		return udp_lib_setsockopt(sk, level, optname, optval, optlen,
					  udp_v6_push_pending_frames);
1413
	return ipv6_setsockopt(sk, level, optname, optval, optlen);
1414 1415 1416
}

#ifdef CONFIG_COMPAT
1417
int compat_udpv6_setsockopt(struct sock *sk, int level, int optname,
1418
			    char __user *optval, unsigned int optlen)
1419
{
1420
	if (level == SOL_UDP  ||  level == SOL_UDPLITE)
1421 1422
		return udp_lib_setsockopt(sk, level, optname, optval, optlen,
					  udp_v6_push_pending_frames);
1423
	return compat_ipv6_setsockopt(sk, level, optname, optval, optlen);
1424 1425 1426
}
#endif

1427 1428
int udpv6_getsockopt(struct sock *sk, int level, int optname,
		     char __user *optval, int __user *optlen)
1429
{
1430
	if (level == SOL_UDP  ||  level == SOL_UDPLITE)
1431
		return udp_lib_getsockopt(sk, level, optname, optval, optlen);
1432
	return ipv6_getsockopt(sk, level, optname, optval, optlen);
1433 1434 1435
}

#ifdef CONFIG_COMPAT
1436 1437
int compat_udpv6_getsockopt(struct sock *sk, int level, int optname,
			    char __user *optval, int __user *optlen)
1438
{
1439
	if (level == SOL_UDP  ||  level == SOL_UDPLITE)
1440
		return udp_lib_getsockopt(sk, level, optname, optval, optlen);
1441
	return compat_ipv6_getsockopt(sk, level, optname, optval, optlen);
1442 1443 1444
}
#endif

1445
static struct inet6_protocol udpv6_protocol = {
1446
	.early_demux	=	udp_v6_early_demux,
1447
	.early_demux_handler =  udp_v6_early_demux,
L
Linus Torvalds 已提交
1448 1449 1450 1451 1452 1453 1454
	.handler	=	udpv6_rcv,
	.err_handler	=	udpv6_err,
	.flags		=	INET6_PROTO_NOPOLICY|INET6_PROTO_FINAL,
};

/* ------------------------------------------------------------------------ */
#ifdef CONFIG_PROC_FS
1455
int udp6_seq_show(struct seq_file *seq, void *v)
L
Linus Torvalds 已提交
1456
{
1457 1458 1459 1460 1461 1462 1463 1464 1465
	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
1466 1467 1468
	return 0;
}

1469 1470 1471 1472 1473 1474 1475 1476
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
1477 1478 1479
static struct udp_seq_afinfo udp6_seq_afinfo = {
	.name		= "udp6",
	.family		= AF_INET6,
1480
	.udp_table	= &udp_table,
1481
	.seq_fops	= &udp6_afinfo_seq_fops,
1482 1483 1484
	.seq_ops	= {
		.show		= udp6_seq_show,
	},
L
Linus Torvalds 已提交
1485 1486
};

1487
int __net_init udp6_proc_init(struct net *net)
L
Linus Torvalds 已提交
1488
{
1489
	return udp_proc_register(net, &udp6_seq_afinfo);
L
Linus Torvalds 已提交
1490 1491
}

I
Ian Morris 已提交
1492 1493
void udp6_proc_exit(struct net *net)
{
1494
	udp_proc_unregister(net, &udp6_seq_afinfo);
L
Linus Torvalds 已提交
1495 1496 1497 1498 1499 1500
}
#endif /* CONFIG_PROC_FS */

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

struct proto udpv6_prot = {
1501 1502
	.name		   = "UDPv6",
	.owner		   = THIS_MODULE,
1503
	.close		   = udp_lib_close,
1504 1505 1506
	.connect	   = ip6_datagram_connect,
	.disconnect	   = udp_disconnect,
	.ioctl		   = udp_ioctl,
1507
	.init		   = udp_init_sock,
1508 1509 1510 1511 1512
	.destroy	   = udpv6_destroy_sock,
	.setsockopt	   = udpv6_setsockopt,
	.getsockopt	   = udpv6_getsockopt,
	.sendmsg	   = udpv6_sendmsg,
	.recvmsg	   = udpv6_recvmsg,
1513
	.release_cb	   = ip6_datagram_release_cb,
1514 1515
	.hash		   = udp_lib_hash,
	.unhash		   = udp_lib_unhash,
E
Eric Dumazet 已提交
1516
	.rehash		   = udp_v6_rehash,
1517
	.get_port	   = udp_v6_get_port,
H
Hideo Aoki 已提交
1518 1519 1520 1521
	.memory_allocated  = &udp_memory_allocated,
	.sysctl_mem	   = sysctl_udp_mem,
	.sysctl_wmem	   = &sysctl_udp_wmem_min,
	.sysctl_rmem	   = &sysctl_udp_rmem_min,
1522
	.obj_size	   = sizeof(struct udp6_sock),
1523
	.h.udp_table	   = &udp_table,
1524
#ifdef CONFIG_COMPAT
1525 1526
	.compat_setsockopt = compat_udpv6_setsockopt,
	.compat_getsockopt = compat_udpv6_getsockopt,
1527
#endif
1528
	.diag_destroy      = udp_abort,
L
Linus Torvalds 已提交
1529 1530 1531 1532 1533 1534 1535 1536 1537 1538
};

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

1539
int __init udpv6_init(void)
L
Linus Torvalds 已提交
1540
{
1541 1542
	int ret;

1543 1544
	ret = inet6_add_protocol(&udpv6_protocol, IPPROTO_UDP);
	if (ret)
1545
		goto out;
1546

1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557
	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;
}

1558
void udpv6_exit(void)
1559 1560 1561
{
	inet6_unregister_protosw(&udpv6_protosw);
	inet6_del_protocol(&udpv6_protocol, IPPROTO_UDP);
L
Linus Torvalds 已提交
1562
}