udp.c 39.3 KB
Newer Older
L
Linus Torvalds 已提交
1 2
/*
 *	UDP over IPv6
3
 *	Linux INET6 implementation
L
Linus Torvalds 已提交
4 5
 *
 *	Authors:
6
 *	Pedro Roque		<roque@di.fc.ul.pt>
L
Linus Torvalds 已提交
7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
 *
 *	Based on linux/ipv4/udp.c
 *
 *	Fixes:
 *	Hideaki YOSHIFUJI	:	sin6_scope_id support
 *	YOSHIFUJI Hideaki @USAGI and:	Support IPV6_V6ONLY socket option, which
 *	Alexey Kuznetsov		allow both IPv4 and IPv6 sockets to bind
 *					a single port at the same time.
 *      Kazunori MIYAZAWA @USAGI:       change process style to use ip6_append_data
 *      YOSHIFUJI Hideaki @USAGI:	convert /proc/net/udp6 to seq_file.
 *
 *	This program is free software; you can redistribute it and/or
 *      modify it under the terms of the GNU General Public License
 *      as published by the Free Software Foundation; either version
 *      2 of the License, or (at your option) any later version.
 */

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

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

295
	return __udp6_lib_lookup(dev_net(skb->dev), &iph->saddr, sport,
E
Eric Dumazet 已提交
296
				 &iph->daddr, dport, inet6_iif(skb),
297
				 udptable, skb);
298 299
}

300 301 302 303 304
struct sock *udp6_lib_lookup_skb(struct sk_buff *skb,
				 __be16 sport, __be16 dport)
{
	const struct ipv6hdr *iph = ipv6_hdr(skb);

305
	return __udp6_lib_lookup(dev_net(skb->dev), &iph->saddr, sport,
306 307 308 309 310
				 &iph->daddr, dport, inet6_iif(skb),
				 &udp_table, skb);
}
EXPORT_SYMBOL_GPL(udp6_lib_lookup_skb);

311 312 313 314
/* Must be called under rcu_read_lock().
 * Does increment socket refcount.
 */
#if IS_ENABLED(CONFIG_NETFILTER_XT_MATCH_SOCKET) || \
315 316
    IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TPROXY) || \
    IS_ENABLED(CONFIG_NF_SOCKET_IPV6)
317 318 319
struct sock *udp6_lib_lookup(struct net *net, const struct in6_addr *saddr, __be16 sport,
			     const struct in6_addr *daddr, __be16 dport, int dif)
{
320 321 322 323
	struct sock *sk;

	sk =  __udp6_lib_lookup(net, saddr, sport, daddr, dport,
				dif, &udp_table, NULL);
324
	if (sk && !refcount_inc_not_zero(&sk->sk_refcnt))
325 326
		sk = NULL;
	return sk;
327 328
}
EXPORT_SYMBOL_GPL(udp6_lib_lookup);
329
#endif
330

L
Linus Torvalds 已提交
331
/*
332 333
 *	This should be easy, if there is something there we
 *	return it, otherwise we block.
L
Linus Torvalds 已提交
334 335
 */

336
int udpv6_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
L
Linus Torvalds 已提交
337 338 339 340
		  int noblock, int flags, int *addr_len)
{
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct inet_sock *inet = inet_sk(sk);
341
	struct sk_buff *skb;
342
	unsigned int ulen, copied;
343
	int peeked, peeking, off;
344 345
	int err;
	int is_udplite = IS_UDPLITE(sk);
346
	bool checksum_valid = false;
347
	int is_udp4;
L
Linus Torvalds 已提交
348 349

	if (flags & MSG_ERRQUEUE)
350
		return ipv6_recv_error(sk, msg, len, addr_len);
L
Linus Torvalds 已提交
351

352
	if (np->rxpmtu && np->rxopt.bits.rxpmtu)
353
		return ipv6_recv_rxpmtu(sk, msg, len, addr_len);
354

L
Linus Torvalds 已提交
355
try_again:
356
	peeking = off = sk_peek_offset(sk, flags);
357
	skb = __skb_recv_udp(sk, flags, noblock, &peeked, &off, &err);
L
Linus Torvalds 已提交
358
	if (!skb)
359
		return err;
L
Linus Torvalds 已提交
360

361
	ulen = udp_skb_len(skb);
362
	copied = len;
363 364
	if (copied > ulen - off)
		copied = ulen - off;
365
	else if (copied < ulen)
366
		msg->msg_flags |= MSG_TRUNC;
L
Linus Torvalds 已提交
367

368 369
	is_udp4 = (skb->protocol == htons(ETH_P_IP));

370
	/*
371 372 373
	 * 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.
374 375
	 */

376 377
	if (copied < ulen || peeking ||
	    (is_udplite && UDP_SKB_CB(skb)->partial_cov)) {
378 379
		checksum_valid = udp_skb_csum_unnecessary(skb) ||
				!__udp_lib_checksum_complete(skb);
380
		if (!checksum_valid)
L
Linus Torvalds 已提交
381
			goto csum_copy_err;
382 383
	}

384 385 386 387 388 389
	if (checksum_valid || udp_skb_csum_unnecessary(skb)) {
		if (udp_skb_is_linear(skb))
			err = copy_linear_skb(skb, copied, off, &msg->msg_iter);
		else
			err = skb_copy_datagram_msg(skb, off, msg, copied);
	} 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:
458 459
	if (!__sk_queue_drop_skb(sk, &udp_sk(sk)->reader_queue, skb, flags,
				 udp_skb_destructor)) {
460
		if (is_udp4) {
461 462 463 464
			UDP_INC_STATS(sock_net(sk),
				      UDP_MIB_CSUMERRORS, is_udplite);
			UDP_INC_STATS(sock_net(sk),
				      UDP_MIB_INERRORS, is_udplite);
465
		} else {
466 467 468 469
			UDP6_INC_STATS(sock_net(sk),
				       UDP_MIB_CSUMERRORS, is_udplite);
			UDP6_INC_STATS(sock_net(sk),
				       UDP_MIB_INERRORS, is_udplite);
470
		}
471
	}
472
	kfree_skb(skb);
L
Linus Torvalds 已提交
473

474 475
	/* starting over for a new packet, but check if we need to yield */
	cond_resched();
476
	msg->msg_flags &= ~MSG_TRUNC;
L
Linus Torvalds 已提交
477 478 479
	goto try_again;
}

480
void __udp6_lib_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
481
		    u8 type, u8 code, int offset, __be32 info,
482
		    struct udp_table *udptable)
L
Linus Torvalds 已提交
483 484
{
	struct ipv6_pinfo *np;
485 486 487
	const struct ipv6hdr *hdr = (const struct ipv6hdr *)skb->data;
	const struct in6_addr *saddr = &hdr->saddr;
	const struct in6_addr *daddr = &hdr->daddr;
488
	struct udphdr *uh = (struct udphdr *)(skb->data+offset);
L
Linus Torvalds 已提交
489
	struct sock *sk;
490
	int harderr;
L
Linus Torvalds 已提交
491
	int err;
492
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
493

494
	sk = __udp6_lib_lookup(net, daddr, uh->dest, saddr, uh->source,
495
			       inet6_iif(skb), udptable, skb);
496
	if (!sk) {
E
Eric Dumazet 已提交
497 498
		__ICMP6_INC_STATS(net, __in6_dev_get(skb->dev),
				  ICMP6_MIB_INERRORS);
L
Linus Torvalds 已提交
499
		return;
500
	}
L
Linus Torvalds 已提交
501

502 503 504
	harderr = icmpv6_err_convert(type, code, &err);
	np = inet6_sk(sk);

505 506 507
	if (type == ICMPV6_PKT_TOOBIG) {
		if (!ip6_sk_accept_pmtu(sk))
			goto out;
508
		ip6_sk_update_pmtu(skb, sk, info);
509 510
		if (np->pmtudisc != IPV6_PMTUDISC_DONT)
			harderr = 1;
511
	}
512
	if (type == NDISC_REDIRECT) {
513
		ip6_sk_redirect(skb, sk);
514 515
		goto out;
	}
516

517 518 519 520
	if (!np->recverr) {
		if (!harderr || sk->sk_state != TCP_ESTABLISHED)
			goto out;
	} else {
L
Linus Torvalds 已提交
521
		ipv6_icmp_error(sk, skb, err, uh->dest, ntohl(info), (u8 *)(uh+1));
522
	}
523

L
Linus Torvalds 已提交
524 525 526
	sk->sk_err = err;
	sk->sk_error_report(sk);
out:
527
	return;
L
Linus Torvalds 已提交
528 529
}

530
static int __udpv6_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
531 532 533
{
	int rc;

534
	if (!ipv6_addr_any(&sk->sk_v6_daddr)) {
535
		sock_rps_save_rxhash(sk, skb);
536
		sk_mark_napi_id(sk, skb);
E
Eric Dumazet 已提交
537
		sk_incoming_cpu_update(sk);
538 539
	} else {
		sk_mark_napi_id_once(sk, skb);
540
	}
541

542
	rc = __udp_enqueue_schedule_skb(sk, skb);
543 544 545 546 547
	if (rc < 0) {
		int is_udplite = IS_UDPLITE(sk);

		/* Note that an ENOMEM error is charged twice */
		if (rc == -ENOMEM)
548
			UDP6_INC_STATS(sock_net(sk),
549
					 UDP_MIB_RCVBUFERRORS, is_udplite);
550
		UDP6_INC_STATS(sock_net(sk), UDP_MIB_INERRORS, is_udplite);
551 552 553
		kfree_skb(skb);
		return -1;
	}
554

555 556 557
	return 0;
}

558
static __inline__ void udpv6_err(struct sk_buff *skb,
559
				 struct inet6_skb_parm *opt, u8 type,
560
				 u8 code, int offset, __be32 info)
561
{
562
	__udp6_lib_err(skb, opt, type, code, offset, info, &udp_table);
563 564
}

565 566 567 568 569 570 571 572
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);

573
static int udpv6_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
L
Linus Torvalds 已提交
574
{
575
	struct udp_sock *up = udp_sk(sk);
W
Wang Chen 已提交
576
	int is_udplite = IS_UDPLITE(sk);
577

578 579
	if (!xfrm6_policy_check(sk, XFRM_POLICY_IN, skb))
		goto drop;
L
Linus Torvalds 已提交
580

581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596
	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);
597
		if (encap_rcv) {
598 599
			int ret;

600 601 602 603
			/* Verify checksum before giving to encap */
			if (udp_lib_checksum_complete(skb))
				goto csum_error;

604 605
			ret = encap_rcv(sk, skb);
			if (ret <= 0) {
606 607 608
				__UDP_INC_STATS(sock_net(sk),
						UDP_MIB_INDATAGRAMS,
						is_udplite);
609 610 611 612 613 614 615
				return -ret;
			}
		}

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

616 617 618
	/*
	 * UDP-Lite specific tests, ignored on UDP sockets (see net/ipv4/udp.c).
	 */
W
Wang Chen 已提交
619
	if ((is_udplite & UDPLITE_RECV_CC)  &&  UDP_SKB_CB(skb)->partial_cov) {
620 621

		if (up->pcrlen == 0) {          /* full coverage was set  */
622 623
			net_dbg_ratelimited("UDPLITE6: partial coverage %d while full coverage %d requested\n",
					    UDP_SKB_CB(skb)->cscov, skb->len);
624 625 626
			goto drop;
		}
		if (UDP_SKB_CB(skb)->cscov  <  up->pcrlen) {
627 628
			net_dbg_ratelimited("UDPLITE6: coverage %d too small, need min %d\n",
					    UDP_SKB_CB(skb)->cscov, up->pcrlen);
629 630
			goto drop;
		}
L
Linus Torvalds 已提交
631 632
	}

633
	prefetch(&sk->sk_rmem_alloc);
634 635 636 637
	if (rcu_access_pointer(sk->sk_filter) &&
	    udp_lib_checksum_complete(skb))
		goto csum_error;

638
	if (sk_filter_trim_cap(sk, skb, sizeof(struct udphdr)))
639
		goto drop;
640

641
	udp_csum_pull_header(skb);
642

643
	skb_dst_drop(skb);
644

645
	return __udpv6_queue_rcv_skb(sk, skb);
646

647
csum_error:
648
	__UDP6_INC_STATS(sock_net(sk), UDP_MIB_CSUMERRORS, is_udplite);
649
drop:
650
	__UDP6_INC_STATS(sock_net(sk), UDP_MIB_INERRORS, is_udplite);
651
	atomic_inc(&sk->sk_drops);
652 653
	kfree_skb(skb);
	return -1;
L
Linus Torvalds 已提交
654 655
}

656 657 658 659
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
660
{
661
	struct inet_sock *inet = inet_sk(sk);
L
Linus Torvalds 已提交
662

663 664 665 666 667 668 669 670
	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)) ||
671 672 673
	    (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)))
674 675 676 677
		return false;
	if (!inet6_mc_check(sk, loc_addr, rmt_addr))
		return false;
	return true;
L
Linus Torvalds 已提交
678 679
}

680 681 682 683 684
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.
	 */
685 686 687
	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));
688 689
}

L
Linus Torvalds 已提交
690 691 692 693
/*
 * Note: called only from the BH handler context,
 * so we don't need to lock the hashes.
 */
694
static int __udp6_lib_mcast_deliver(struct net *net, struct sk_buff *skb,
695
		const struct in6_addr *saddr, const struct in6_addr *daddr,
696
		struct udp_table *udptable, int proto)
L
Linus Torvalds 已提交
697
{
698
	struct sock *sk, *first = NULL;
699
	const struct udphdr *uh = udp_hdr(skb);
700 701
	unsigned short hnum = ntohs(uh->dest);
	struct udp_hslot *hslot = udp_hashslot(udptable, net, hnum);
702
	unsigned int offset = offsetof(typeof(*sk), sk_node);
703
	unsigned int hash2 = 0, hash2_any = 0, use_hash2 = (hslot->count > 10);
704 705 706
	int dif = inet6_iif(skb);
	struct hlist_node *node;
	struct sk_buff *nskb;
707 708 709

	if (use_hash2) {
		hash2_any = udp6_portaddr_hash(net, &in6addr_any, hnum) &
710 711
			    udptable->mask;
		hash2 = udp6_portaddr_hash(net, daddr, hnum) & udptable->mask;
712
start_lookup:
713
		hslot = &udptable->hash2[hash2];
714 715
		offset = offsetof(typeof(*sk), __sk_common.skc_portaddr_node);
	}
L
Linus Torvalds 已提交
716

717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732
	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);
733 734 735 736
			__UDP6_INC_STATS(net, UDP_MIB_RCVBUFERRORS,
					 IS_UDPLITE(sk));
			__UDP6_INC_STATS(net, UDP_MIB_INERRORS,
					 IS_UDPLITE(sk));
737
			continue;
H
Hideo Aoki 已提交
738
		}
739

740 741 742
		if (udpv6_queue_rcv_skb(sk, nskb) > 0)
			consume_skb(nskb);
	}
743

744 745 746 747 748 749
	/* 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;
	}

750 751 752
	if (first) {
		if (udpv6_queue_rcv_skb(first, skb) > 0)
			consume_skb(skb);
753
	} else {
754
		kfree_skb(skb);
755 756
		__UDP6_INC_STATS(net, UDP_MIB_IGNOREDMULTI,
				 proto == IPPROTO_UDPLITE);
757
	}
758
	return 0;
L
Linus Torvalds 已提交
759 760
}

761
int __udp6_lib_rcv(struct sk_buff *skb, struct udp_table *udptable,
762
		   int proto)
L
Linus Torvalds 已提交
763
{
764
	const struct in6_addr *saddr, *daddr;
765
	struct net *net = dev_net(skb->dev);
766
	struct udphdr *uh;
767
	struct sock *sk;
L
Linus Torvalds 已提交
768 769 770
	u32 ulen = 0;

	if (!pskb_may_pull(skb, sizeof(struct udphdr)))
771
		goto discard;
L
Linus Torvalds 已提交
772

773 774
	saddr = &ipv6_hdr(skb)->saddr;
	daddr = &ipv6_hdr(skb)->daddr;
775
	uh = udp_hdr(skb);
L
Linus Torvalds 已提交
776 777

	ulen = ntohs(uh->len);
778 779
	if (ulen > skb->len)
		goto short_packet;
L
Linus Torvalds 已提交
780

781 782
	if (proto == IPPROTO_UDP) {
		/* UDP validates ulen. */
L
Linus Torvalds 已提交
783

784 785 786
		/* Check for jumbo payload */
		if (ulen == 0)
			ulen = skb->len;
L
Linus Torvalds 已提交
787

788 789
		if (ulen < sizeof(*uh))
			goto short_packet;
L
Linus Torvalds 已提交
790

791 792 793
		if (ulen < skb->len) {
			if (pskb_trim_rcsum(skb, ulen))
				goto short_packet;
794 795
			saddr = &ipv6_hdr(skb)->saddr;
			daddr = &ipv6_hdr(skb)->daddr;
796
			uh = udp_hdr(skb);
797 798
		}
	}
L
Linus Torvalds 已提交
799

800
	if (udp6_csum_init(skb, uh, proto))
801
		goto csum_error;
802

803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820
	/* Check if the socket is already available, e.g. due to early demux */
	sk = skb_steal_sock(skb);
	if (sk) {
		struct dst_entry *dst = skb_dst(skb);
		int ret;

		if (unlikely(sk->sk_rx_dst != dst))
			udp_sk_rx_dst_set(sk, dst);

		ret = udpv6_queue_rcv_skb(sk, skb);
		sock_put(sk);

		/* a return value > 0 means to resubmit the input */
		if (ret > 0)
			return ret;
		return 0;
	}

821 822
	/*
	 *	Multicast receive code
L
Linus Torvalds 已提交
823
	 */
824
	if (ipv6_addr_is_multicast(daddr))
825
		return __udp6_lib_mcast_deliver(net, skb,
826
				saddr, daddr, udptable, proto);
L
Linus Torvalds 已提交
827 828

	/* Unicast */
829
	sk = __udp6_lib_lookup_skb(skb, uh->source, uh->dest, udptable);
830
	if (sk) {
831 832
		int ret;

833
		if (!uh->check && !udp_sk(sk)->no_check6_rx) {
834 835 836 837
			udp6_csum_zero_error(skb);
			goto csum_error;
		}

838
		if (inet_get_convert_csum(sk) && uh->check && !IS_UDPLITE(sk))
839 840 841
			skb_checksum_try_convert(skb, IPPROTO_UDP, uh->check,
						 ip6_compute_pseudo);

842
		ret = udpv6_queue_rcv_skb(sk, skb);
L
Linus Torvalds 已提交
843

844
		/* a return value > 0 means to resubmit the input */
845
		if (ret > 0)
846
			return ret;
L
Linus Torvalds 已提交
847

S
Stephen Hemminger 已提交
848
		return 0;
L
Linus Torvalds 已提交
849
	}
850

851 852 853 854 855
	if (!uh->check) {
		udp6_csum_zero_error(skb);
		goto csum_error;
	}

856
	if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb))
857
		goto discard;
858 859

	if (udp_lib_checksum_complete(skb))
860
		goto csum_error;
861

862
	__UDP6_INC_STATS(net, UDP_MIB_NOPORTS, proto == IPPROTO_UDPLITE);
863 864 865
	icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_PORT_UNREACH, 0);

	kfree_skb(skb);
S
Stephen Hemminger 已提交
866
	return 0;
L
Linus Torvalds 已提交
867

868
short_packet:
869 870 871 872 873
	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));
874 875
	goto discard;
csum_error:
876
	__UDP6_INC_STATS(net, UDP_MIB_CSUMERRORS, proto == IPPROTO_UDPLITE);
L
Linus Torvalds 已提交
877
discard:
878
	__UDP6_INC_STATS(net, UDP_MIB_INERRORS, proto == IPPROTO_UDPLITE);
L
Linus Torvalds 已提交
879
	kfree_skb(skb);
S
Stephen Hemminger 已提交
880
	return 0;
L
Linus Torvalds 已提交
881
}
882

883

884 885 886 887 888
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)
{
889 890 891 892 893
	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);
894 895
	struct sock *sk;

896
	udp_portaddr_for_each_entry_rcu(sk, &hslot2->head) {
897 898
		if (sk->sk_state == TCP_ESTABLISHED &&
		    INET6_MATCH(sk, net, rmt_addr, loc_addr, ports, dif))
899 900 901 902 903
			return sk;
		/* Only check first socket in chain */
		break;
	}
	return NULL;
904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927
}

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;

928
	if (!sk || !refcount_inc_not_zero(&sk->sk_refcnt))
929 930 931 932 933 934 935 936 937
		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) {
938 939 940 941 942
		/* set noref for now.
		 * any place which wants to hold dst has to call
		 * dst_hold_safe()
		 */
		skb_dst_set_noref(skb, dst);
943 944 945
	}
}

946
static __inline__ int udpv6_rcv(struct sk_buff *skb)
947
{
948
	return __udp6_lib_rcv(skb, &udp_table, IPPROTO_UDP);
949 950
}

L
Linus Torvalds 已提交
951 952 953 954 955 956 957
/*
 * 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);

958 959 960
	if (up->pending == AF_INET)
		udp_flush_pending_frames(sk);
	else if (up->pending) {
L
Linus Torvalds 已提交
961 962 963
		up->len = 0;
		up->pending = 0;
		ip6_flush_pending_frames(sk);
964
	}
L
Linus Torvalds 已提交
965 966
}

967
/**
968 969 970 971
 *	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)
972 973 974 975 976 977 978
 */
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
979
	struct sk_buff *frags = skb_shinfo(skb)->frag_list;
980 981
	__wsum csum = 0;

V
Vlad Yasevich 已提交
982
	if (!frags) {
983 984 985 986 987 988 989 990 991 992 993 994 995 996 997
		/* 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
998 999 1000
		do {
			csum = csum_add(csum, frags->csum);
		} while ((frags = frags->next));
1001 1002 1003 1004 1005 1006 1007 1008

		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
1009 1010 1011 1012
/*
 *	Sending
 */

V
Vlad Yasevich 已提交
1013
static int udp_v6_send_skb(struct sk_buff *skb, struct flowi6 *fl6)
L
Linus Torvalds 已提交
1014
{
V
Vlad Yasevich 已提交
1015
	struct sock *sk = skb->sk;
L
Linus Torvalds 已提交
1016 1017
	struct udphdr *uh;
	int err = 0;
W
Wang Chen 已提交
1018
	int is_udplite = IS_UDPLITE(sk);
1019
	__wsum csum = 0;
V
Vlad Yasevich 已提交
1020 1021
	int offset = skb_transport_offset(skb);
	int len = skb->len - offset;
L
Linus Torvalds 已提交
1022 1023 1024 1025

	/*
	 * Create a UDP header
	 */
1026
	uh = udp_hdr(skb);
1027 1028
	uh->source = fl6->fl6_sport;
	uh->dest = fl6->fl6_dport;
V
Vlad Yasevich 已提交
1029
	uh->len = htons(len);
L
Linus Torvalds 已提交
1030 1031
	uh->check = 0;

W
Wang Chen 已提交
1032
	if (is_udplite)
V
Vlad Yasevich 已提交
1033 1034
		csum = udplite_csum(skb);
	else if (udp_sk(sk)->no_check6_tx) {   /* UDP csum disabled */
1035 1036 1037
		skb->ip_summed = CHECKSUM_NONE;
		goto send;
	} else if (skb->ip_summed == CHECKSUM_PARTIAL) { /* UDP hardware csum */
V
Vlad Yasevich 已提交
1038
		udp6_hwcsum_outgoing(sk, skb, &fl6->saddr, &fl6->daddr, len);
1039 1040
		goto send;
	} else
V
Vlad Yasevich 已提交
1041
		csum = udp_csum(skb);
L
Linus Torvalds 已提交
1042

1043
	/* add protocol-dependent pseudo-header */
1044
	uh->check = csum_ipv6_magic(&fl6->saddr, &fl6->daddr,
V
Vlad Yasevich 已提交
1045
				    len, fl6->flowi6_proto, csum);
L
Linus Torvalds 已提交
1046
	if (uh->check == 0)
1047
		uh->check = CSUM_MANGLED_0;
L
Linus Torvalds 已提交
1048

1049
send:
V
Vlad Yasevich 已提交
1050
	err = ip6_send_skb(skb);
E
Eric Dumazet 已提交
1051 1052
	if (err) {
		if (err == -ENOBUFS && !inet6_sk(sk)->recverr) {
1053 1054
			UDP6_INC_STATS(sock_net(sk),
				       UDP_MIB_SNDBUFERRORS, is_udplite);
E
Eric Dumazet 已提交
1055 1056
			err = 0;
		}
1057 1058 1059 1060
	} else {
		UDP6_INC_STATS(sock_net(sk),
			       UDP_MIB_OUTDATAGRAMS, is_udplite);
	}
V
Vlad Yasevich 已提交
1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084
	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
1085 1086 1087 1088 1089 1090
out:
	up->len = 0;
	up->pending = 0;
	return err;
}

1091
int udpv6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
L
Linus Torvalds 已提交
1092 1093 1094 1095 1096
{
	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);
1097
	DECLARE_SOCKADDR(struct sockaddr_in6 *, sin6, msg->msg_name);
1098
	struct in6_addr *daddr, *final_p, final;
L
Linus Torvalds 已提交
1099
	struct ipv6_txoptions *opt = NULL;
1100
	struct ipv6_txoptions *opt_to_free = NULL;
L
Linus Torvalds 已提交
1101
	struct ip6_flowlabel *flowlabel = NULL;
1102
	struct flowi6 fl6;
L
Linus Torvalds 已提交
1103
	struct dst_entry *dst;
W
Wei Wang 已提交
1104
	struct ipcm6_cookie ipc6;
L
Linus Torvalds 已提交
1105 1106 1107 1108
	int addr_len = msg->msg_namelen;
	int ulen = len;
	int corkreq = up->corkflag || msg->msg_flags&MSG_MORE;
	int err;
1109
	int connected = 0;
W
Wang Chen 已提交
1110
	int is_udplite = IS_UDPLITE(sk);
1111
	int (*getfrag)(void *, char *, int, int, int, struct sk_buff *);
1112
	struct sockcm_cookie sockc;
L
Linus Torvalds 已提交
1113

W
Wei Wang 已提交
1114 1115 1116
	ipc6.hlimit = -1;
	ipc6.tclass = -1;
	ipc6.dontfrag = -1;
1117
	sockc.tsflags = sk->sk_tsflags;
W
Wei Wang 已提交
1118

L
Linus Torvalds 已提交
1119 1120 1121 1122 1123 1124 1125 1126 1127 1128
	/* 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;
1129 1130 1131 1132
			if (ipv6_addr_any(daddr) &&
			    ipv6_addr_v4mapped(&np->saddr))
				ipv6_addr_set_v4mapped(htonl(INADDR_LOOPBACK),
						       daddr);
L
Linus Torvalds 已提交
1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146
			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;
1147
		daddr = &sk->sk_v6_daddr;
1148
	} else
L
Linus Torvalds 已提交
1149 1150 1151
		daddr = NULL;

	if (daddr) {
B
Brian Haley 已提交
1152
		if (ipv6_addr_v4mapped(daddr)) {
L
Linus Torvalds 已提交
1153 1154
			struct sockaddr_in sin;
			sin.sin_family = AF_INET;
E
Eric Dumazet 已提交
1155
			sin.sin_port = sin6 ? sin6->sin6_port : inet->inet_dport;
L
Linus Torvalds 已提交
1156 1157 1158 1159 1160 1161
			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;
1162
			return udp_sendmsg(sk, msg, len);
L
Linus Torvalds 已提交
1163 1164 1165 1166
		}
	}

	if (up->pending == AF_INET)
1167
		return udp_sendmsg(sk, msg, len);
L
Linus Torvalds 已提交
1168 1169

	/* Rough check on arithmetic overflow,
1170
	   better check is made in ip6_append_data().
L
Linus Torvalds 已提交
1171 1172 1173
	   */
	if (len > INT_MAX - sizeof(struct udphdr))
		return -EMSGSIZE;
1174

1175
	getfrag  =  is_udplite ?  udplite_getfrag : ip_generic_getfrag;
L
Linus Torvalds 已提交
1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193
	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);

1194
	memset(&fl6, 0, sizeof(fl6));
L
Linus Torvalds 已提交
1195 1196 1197 1198 1199

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

1200
		fl6.fl6_dport = sin6->sin6_port;
L
Linus Torvalds 已提交
1201 1202 1203
		daddr = &sin6->sin6_addr;

		if (np->sndflow) {
1204 1205 1206
			fl6.flowlabel = sin6->sin6_flowinfo&IPV6_FLOWINFO_MASK;
			if (fl6.flowlabel&IPV6_FLOWLABEL_MASK) {
				flowlabel = fl6_sock_lookup(sk, fl6.flowlabel);
1207
				if (!flowlabel)
L
Linus Torvalds 已提交
1208 1209 1210 1211 1212 1213 1214 1215 1216
					return -EINVAL;
			}
		}

		/*
		 * Otherwise it will be difficult to maintain
		 * sk->sk_dst_cache.
		 */
		if (sk->sk_state == TCP_ESTABLISHED &&
1217 1218
		    ipv6_addr_equal(daddr, &sk->sk_v6_daddr))
			daddr = &sk->sk_v6_daddr;
L
Linus Torvalds 已提交
1219 1220 1221

		if (addr_len >= sizeof(struct sockaddr_in6) &&
		    sin6->sin6_scope_id &&
1222
		    __ipv6_addr_needs_scope_id(__ipv6_addr_type(daddr)))
1223
			fl6.flowi6_oif = sin6->sin6_scope_id;
L
Linus Torvalds 已提交
1224 1225 1226 1227
	} else {
		if (sk->sk_state != TCP_ESTABLISHED)
			return -EDESTADDRREQ;

1228
		fl6.fl6_dport = inet->inet_dport;
1229
		daddr = &sk->sk_v6_daddr;
1230
		fl6.flowlabel = np->flow_label;
1231
		connected = 1;
L
Linus Torvalds 已提交
1232 1233
	}

1234 1235
	if (!fl6.flowi6_oif)
		fl6.flowi6_oif = sk->sk_bound_dev_if;
L
Linus Torvalds 已提交
1236

1237 1238
	if (!fl6.flowi6_oif)
		fl6.flowi6_oif = np->sticky_pktinfo.ipi6_ifindex;
1239

1240
	fl6.flowi6_mark = sk->sk_mark;
1241
	fl6.flowi6_uid = sk->sk_uid;
B
Brian Haley 已提交
1242

L
Linus Torvalds 已提交
1243 1244 1245 1246
	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
1247
		ipc6.opt = opt;
L
Linus Torvalds 已提交
1248

W
Wei Wang 已提交
1249
		err = ip6_datagram_send_ctl(sock_net(sk), sk, msg, &fl6, &ipc6, &sockc);
L
Linus Torvalds 已提交
1250 1251 1252 1253
		if (err < 0) {
			fl6_sock_release(flowlabel);
			return err;
		}
1254 1255
		if ((fl6.flowlabel&IPV6_FLOWLABEL_MASK) && !flowlabel) {
			flowlabel = fl6_sock_lookup(sk, fl6.flowlabel);
1256
			if (!flowlabel)
L
Linus Torvalds 已提交
1257 1258 1259 1260
				return -EINVAL;
		}
		if (!(opt->opt_nflen|opt->opt_flen))
			opt = NULL;
1261
		connected = 0;
L
Linus Torvalds 已提交
1262
	}
1263 1264 1265 1266
	if (!opt) {
		opt = txopt_get(np);
		opt_to_free = opt;
	}
1267 1268 1269
	if (flowlabel)
		opt = fl6_merge_options(&opt_space, flowlabel, opt);
	opt = ipv6_fixup_options(&opt_space, opt);
W
Wei Wang 已提交
1270
	ipc6.opt = opt;
L
Linus Torvalds 已提交
1271

1272
	fl6.flowi6_proto = sk->sk_protocol;
1273
	if (!ipv6_addr_any(daddr))
A
Alexey Dobriyan 已提交
1274
		fl6.daddr = *daddr;
1275
	else
1276 1277
		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
1278
		fl6.saddr = np->saddr;
1279
	fl6.fl6_sport = inet->inet_sport;
1280

1281
	final_p = fl6_update_dst(&fl6, opt, &final);
1282
	if (final_p)
1283
		connected = 0;
L
Linus Torvalds 已提交
1284

1285 1286
	if (!fl6.flowi6_oif && ipv6_addr_is_multicast(&fl6.daddr)) {
		fl6.flowi6_oif = np->mcast_oif;
1287
		connected = 0;
1288 1289
	} else if (!fl6.flowi6_oif)
		fl6.flowi6_oif = np->ucast_oif;
L
Linus Torvalds 已提交
1290

1291
	security_sk_classify_flow(sk, flowi6_to_flowi(&fl6));
V
Venkat Yekkirala 已提交
1292

1293 1294 1295 1296 1297
	if (ipc6.tclass < 0)
		ipc6.tclass = np->tclass;

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

1298
	dst = ip6_sk_dst_lookup_flow(sk, &fl6, final_p);
1299 1300 1301
	if (IS_ERR(dst)) {
		err = PTR_ERR(dst);
		dst = NULL;
L
Linus Torvalds 已提交
1302
		goto out;
1303
	}
L
Linus Torvalds 已提交
1304

W
Wei Wang 已提交
1305 1306
	if (ipc6.hlimit < 0)
		ipc6.hlimit = ip6_sk_dst_hoplimit(np, &fl6, dst);
L
Linus Torvalds 已提交
1307 1308 1309 1310 1311

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

1312 1313 1314 1315 1316
	/* 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
1317
				   sizeof(struct udphdr), &ipc6,
1318
				   &fl6, (struct rt6_info *)dst,
W
Wei Wang 已提交
1319
				   msg->msg_flags, &sockc);
1320 1321 1322 1323 1324 1325
		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
1326 1327 1328 1329 1330 1331
	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);

1332
		net_dbg_ratelimited("udp cork app bug 2\n");
L
Linus Torvalds 已提交
1333 1334 1335 1336 1337 1338 1339
		err = -EINVAL;
		goto out;
	}

	up->pending = AF_INET6;

do_append_data:
W
Wei Wang 已提交
1340 1341
	if (ipc6.dontfrag < 0)
		ipc6.dontfrag = np->dontfrag;
L
Linus Torvalds 已提交
1342
	up->len += ulen;
W
Wei Wang 已提交
1343 1344 1345
	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
1346 1347 1348
	if (err)
		udp_v6_flush_pending_frames(sk);
	else if (!corkreq)
1349
		err = udp_v6_push_pending_frames(sk);
H
Herbert Xu 已提交
1350 1351
	else if (unlikely(skb_queue_empty(&sk->sk_write_queue)))
		up->pending = 0;
L
Linus Torvalds 已提交
1352

1353 1354 1355 1356 1357
	if (err > 0)
		err = np->recverr ? net_xmit_errno(err) : 0;
	release_sock(sk);

release_dst:
1358 1359 1360
	if (dst) {
		if (connected) {
			ip6_dst_store(sk, dst,
1361 1362
				      ipv6_addr_equal(&fl6.daddr, &sk->sk_v6_daddr) ?
				      &sk->sk_v6_daddr : NULL,
1363
#ifdef CONFIG_IPV6_SUBTREES
1364
				      ipv6_addr_equal(&fl6.saddr, &np->saddr) ?
1365 1366 1367
				      &np->saddr :
#endif
				      NULL);
1368 1369 1370
		} else {
			dst_release(dst);
		}
1371
		dst = NULL;
1372 1373
	}

L
Linus Torvalds 已提交
1374
out:
1375
	dst_release(dst);
L
Linus Torvalds 已提交
1376
	fl6_sock_release(flowlabel);
1377
	txopt_put(opt_to_free);
1378
	if (!err)
L
Linus Torvalds 已提交
1379
		return len;
1380 1381 1382 1383 1384 1385 1386 1387
	/*
	 * 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)) {
1388 1389
		UDP6_INC_STATS(sock_net(sk),
			       UDP_MIB_SNDBUFERRORS, is_udplite);
1390
	}
L
Linus Torvalds 已提交
1391 1392 1393
	return err;

do_confirm:
1394 1395
	if (msg->msg_flags & MSG_PROBE)
		dst_confirm_neigh(dst, &fl6.daddr);
L
Linus Torvalds 已提交
1396 1397 1398 1399 1400 1401
	if (!(msg->msg_flags&MSG_PROBE) || len)
		goto back_from_confirm;
	err = 0;
	goto out;
}

1402
void udpv6_destroy_sock(struct sock *sk)
L
Linus Torvalds 已提交
1403
{
T
Tom Parkin 已提交
1404
	struct udp_sock *up = udp_sk(sk);
L
Linus Torvalds 已提交
1405 1406 1407 1408
	lock_sock(sk);
	udp_v6_flush_pending_frames(sk);
	release_sock(sk);

T
Tom Parkin 已提交
1409 1410 1411 1412 1413 1414 1415
	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
1416 1417 1418 1419 1420 1421
	inet6_destroy_sock(sk);
}

/*
 *	Socket option code for UDP
 */
1422
int udpv6_setsockopt(struct sock *sk, int level, int optname,
1423
		     char __user *optval, unsigned int optlen)
1424
{
1425
	if (level == SOL_UDP  ||  level == SOL_UDPLITE)
1426 1427
		return udp_lib_setsockopt(sk, level, optname, optval, optlen,
					  udp_v6_push_pending_frames);
1428
	return ipv6_setsockopt(sk, level, optname, optval, optlen);
1429 1430 1431
}

#ifdef CONFIG_COMPAT
1432
int compat_udpv6_setsockopt(struct sock *sk, int level, int optname,
1433
			    char __user *optval, unsigned int optlen)
1434
{
1435
	if (level == SOL_UDP  ||  level == SOL_UDPLITE)
1436 1437
		return udp_lib_setsockopt(sk, level, optname, optval, optlen,
					  udp_v6_push_pending_frames);
1438
	return compat_ipv6_setsockopt(sk, level, optname, optval, optlen);
1439 1440 1441
}
#endif

1442 1443
int udpv6_getsockopt(struct sock *sk, int level, int optname,
		     char __user *optval, int __user *optlen)
1444
{
1445
	if (level == SOL_UDP  ||  level == SOL_UDPLITE)
1446
		return udp_lib_getsockopt(sk, level, optname, optval, optlen);
1447
	return ipv6_getsockopt(sk, level, optname, optval, optlen);
1448 1449 1450
}

#ifdef CONFIG_COMPAT
1451 1452
int compat_udpv6_getsockopt(struct sock *sk, int level, int optname,
			    char __user *optval, int __user *optlen)
1453
{
1454
	if (level == SOL_UDP  ||  level == SOL_UDPLITE)
1455
		return udp_lib_getsockopt(sk, level, optname, optval, optlen);
1456
	return compat_ipv6_getsockopt(sk, level, optname, optval, optlen);
1457 1458 1459
}
#endif

1460
static struct inet6_protocol udpv6_protocol = {
1461
	.early_demux	=	udp_v6_early_demux,
1462
	.early_demux_handler =  udp_v6_early_demux,
L
Linus Torvalds 已提交
1463 1464 1465 1466 1467 1468 1469
	.handler	=	udpv6_rcv,
	.err_handler	=	udpv6_err,
	.flags		=	INET6_PROTO_NOPOLICY|INET6_PROTO_FINAL,
};

/* ------------------------------------------------------------------------ */
#ifdef CONFIG_PROC_FS
1470
int udp6_seq_show(struct seq_file *seq, void *v)
L
Linus Torvalds 已提交
1471
{
1472 1473 1474 1475 1476 1477 1478 1479 1480
	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
1481 1482 1483
	return 0;
}

1484 1485 1486 1487 1488 1489 1490 1491
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
1492 1493 1494
static struct udp_seq_afinfo udp6_seq_afinfo = {
	.name		= "udp6",
	.family		= AF_INET6,
1495
	.udp_table	= &udp_table,
1496
	.seq_fops	= &udp6_afinfo_seq_fops,
1497 1498 1499
	.seq_ops	= {
		.show		= udp6_seq_show,
	},
L
Linus Torvalds 已提交
1500 1501
};

1502
int __net_init udp6_proc_init(struct net *net)
L
Linus Torvalds 已提交
1503
{
1504
	return udp_proc_register(net, &udp6_seq_afinfo);
L
Linus Torvalds 已提交
1505 1506
}

I
Ian Morris 已提交
1507 1508
void udp6_proc_exit(struct net *net)
{
1509
	udp_proc_unregister(net, &udp6_seq_afinfo);
L
Linus Torvalds 已提交
1510 1511 1512 1513 1514 1515
}
#endif /* CONFIG_PROC_FS */

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

struct proto udpv6_prot = {
1516 1517
	.name		   = "UDPv6",
	.owner		   = THIS_MODULE,
1518
	.close		   = udp_lib_close,
1519 1520 1521
	.connect	   = ip6_datagram_connect,
	.disconnect	   = udp_disconnect,
	.ioctl		   = udp_ioctl,
1522
	.init		   = udp_init_sock,
1523 1524 1525 1526 1527
	.destroy	   = udpv6_destroy_sock,
	.setsockopt	   = udpv6_setsockopt,
	.getsockopt	   = udpv6_getsockopt,
	.sendmsg	   = udpv6_sendmsg,
	.recvmsg	   = udpv6_recvmsg,
1528
	.release_cb	   = ip6_datagram_release_cb,
1529 1530
	.hash		   = udp_lib_hash,
	.unhash		   = udp_lib_unhash,
E
Eric Dumazet 已提交
1531
	.rehash		   = udp_v6_rehash,
1532
	.get_port	   = udp_v6_get_port,
H
Hideo Aoki 已提交
1533 1534 1535 1536
	.memory_allocated  = &udp_memory_allocated,
	.sysctl_mem	   = sysctl_udp_mem,
	.sysctl_wmem	   = &sysctl_udp_wmem_min,
	.sysctl_rmem	   = &sysctl_udp_rmem_min,
1537
	.obj_size	   = sizeof(struct udp6_sock),
1538
	.h.udp_table	   = &udp_table,
1539
#ifdef CONFIG_COMPAT
1540 1541
	.compat_setsockopt = compat_udpv6_setsockopt,
	.compat_getsockopt = compat_udpv6_getsockopt,
1542
#endif
1543
	.diag_destroy      = udp_abort,
L
Linus Torvalds 已提交
1544 1545 1546 1547 1548 1549 1550 1551 1552 1553
};

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

1554
int __init udpv6_init(void)
L
Linus Torvalds 已提交
1555
{
1556 1557
	int ret;

1558 1559
	ret = inet6_add_protocol(&udpv6_protocol, IPPROTO_UDP);
	if (ret)
1560
		goto out;
1561

1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572
	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;
}

1573
void udpv6_exit(void)
1574 1575 1576
{
	inet6_unregister_protosw(&udpv6_protosw);
	inet6_del_protocol(&udpv6_protocol, IPPROTO_UDP);
L
Linus Torvalds 已提交
1577
}