ip6_output.c 45.6 KB
Newer Older
L
Linus Torvalds 已提交
1 2
/*
 *	IPv6 output functions
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
 *
 *	Based on linux/net/ipv4/ip_output.c
 *
 *	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.
 *
 *	Changes:
 *	A.N.Kuznetsov	:	airthmetics in fragmentation.
 *				extension headers are implemented.
 *				route changes now work.
 *				ip6_forward does not confuse sniffers.
 *				etc.
 *
 *      H. von Brand    :       Added missing #include <linux/string.h>
23
 *	Imran Patel	:	frag id should be in NBO
L
Linus Torvalds 已提交
24 25 26 27 28 29
 *      Kazunori MIYAZAWA @USAGI
 *			:       add ip6_append_data and related functions
 *				for datagram xmit
 */

#include <linux/errno.h>
H
Herbert Xu 已提交
30
#include <linux/kernel.h>
L
Linus Torvalds 已提交
31 32 33 34 35 36 37 38
#include <linux/string.h>
#include <linux/socket.h>
#include <linux/net.h>
#include <linux/netdevice.h>
#include <linux/if_arp.h>
#include <linux/in6.h>
#include <linux/tcp.h>
#include <linux/route.h>
39
#include <linux/module.h>
40
#include <linux/slab.h>
L
Linus Torvalds 已提交
41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56

#include <linux/netfilter.h>
#include <linux/netfilter_ipv6.h>

#include <net/sock.h>
#include <net/snmp.h>

#include <net/ipv6.h>
#include <net/ndisc.h>
#include <net/protocol.h>
#include <net/ip6_route.h>
#include <net/addrconf.h>
#include <net/rawv6.h>
#include <net/icmp.h>
#include <net/xfrm.h>
#include <net/checksum.h>
57
#include <linux/mroute6.h>
D
David Ahern 已提交
58
#include <net/l3mdev.h>
59
#include <net/lwtunnel.h>
L
Linus Torvalds 已提交
60

61
static int ip6_finish_output2(struct net *net, struct sock *sk, struct sk_buff *skb)
L
Linus Torvalds 已提交
62
{
E
Eric Dumazet 已提交
63
	struct dst_entry *dst = skb_dst(skb);
L
Linus Torvalds 已提交
64
	struct net_device *dev = dst->dev;
65
	struct neighbour *neigh;
66 67
	struct in6_addr *nexthop;
	int ret;
L
Linus Torvalds 已提交
68 69 70 71

	skb->protocol = htons(ETH_P_IPV6);
	skb->dev = dev;

72
	if (ipv6_addr_is_multicast(&ipv6_hdr(skb)->daddr)) {
E
Eric Dumazet 已提交
73
		struct inet6_dev *idev = ip6_dst_idev(skb_dst(skb));
L
Linus Torvalds 已提交
74

75
		if (!(dev->flags & IFF_LOOPBACK) && sk_mc_loop(sk) &&
76
		    ((mroute6_socket(net, skb) &&
77
		     !(IP6CB(skb)->flags & IP6SKB_FORWARDED)) ||
78 79
		     ipv6_chk_mcast_addr(dev, &ipv6_hdr(skb)->daddr,
					 &ipv6_hdr(skb)->saddr))) {
L
Linus Torvalds 已提交
80 81 82 83 84 85
			struct sk_buff *newskb = skb_clone(skb, GFP_ATOMIC);

			/* Do not check for IFF_ALLMULTI; multicast routing
			   is not supported in any case.
			 */
			if (newskb)
86
				NF_HOOK(NFPROTO_IPV6, NF_INET_POST_ROUTING,
87
					net, sk, newskb, NULL, newskb->dev,
88
					dev_loopback_xmit);
L
Linus Torvalds 已提交
89

90
			if (ipv6_hdr(skb)->hop_limit == 0) {
91
				IP6_INC_STATS(net, idev,
92
					      IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
93 94 95 96 97
				kfree_skb(skb);
				return 0;
			}
		}

98
		IP6_UPD_PO_STATS(net, idev, IPSTATS_MIB_OUTMCAST, skb->len);
99 100 101 102 103 104 105

		if (IPV6_ADDR_MC_SCOPE(&ipv6_hdr(skb)->daddr) <=
		    IPV6_ADDR_SCOPE_NODELOCAL &&
		    !(dev->flags & IFF_LOOPBACK)) {
			kfree_skb(skb);
			return 0;
		}
L
Linus Torvalds 已提交
106 107
	}

108 109 110 111 112 113 114
	if (lwtunnel_xmit_redirect(dst->lwtstate)) {
		int res = lwtunnel_xmit(skb);

		if (res < 0 || res == LWTUNNEL_XMIT_DONE)
			return res;
	}

115
	rcu_read_lock_bh();
116
	nexthop = rt6_nexthop((struct rt6_info *)dst, &ipv6_hdr(skb)->daddr);
117 118 119 120 121 122 123 124 125
	neigh = __ipv6_neigh_lookup_noref(dst->dev, nexthop);
	if (unlikely(!neigh))
		neigh = __neigh_create(&nd_tbl, nexthop, dst->dev, false);
	if (!IS_ERR(neigh)) {
		ret = dst_neigh_output(dst, neigh, skb);
		rcu_read_unlock_bh();
		return ret;
	}
	rcu_read_unlock_bh();
126

127
	IP6_INC_STATS(net, ip6_dst_idev(dst), IPSTATS_MIB_OUTNOROUTES);
128 129
	kfree_skb(skb);
	return -EINVAL;
L
Linus Torvalds 已提交
130 131
}

132
static int ip6_finish_output(struct net *net, struct sock *sk, struct sk_buff *skb)
133 134
{
	if ((skb->len > ip6_skb_dst_mtu(skb) && !skb_is_gso(skb)) ||
135 136
	    dst_allfrag(skb_dst(skb)) ||
	    (IP6CB(skb)->frag_max_size && skb->len > IP6CB(skb)->frag_max_size))
137
		return ip6_fragment(net, sk, skb, ip6_finish_output2);
138
	else
139
		return ip6_finish_output2(net, sk, skb);
140 141
}

E
Eric W. Biederman 已提交
142
int ip6_output(struct net *net, struct sock *sk, struct sk_buff *skb)
L
Linus Torvalds 已提交
143
{
144
	struct net_device *dev = skb_dst(skb)->dev;
E
Eric Dumazet 已提交
145
	struct inet6_dev *idev = ip6_dst_idev(skb_dst(skb));
146

147
	if (unlikely(idev->cnf.disable_ipv6)) {
148
		IP6_INC_STATS(net, idev, IPSTATS_MIB_OUTDISCARDS);
149 150 151 152
		kfree_skb(skb);
		return 0;
	}

153 154
	return NF_HOOK_COND(NFPROTO_IPV6, NF_INET_POST_ROUTING,
			    net, sk, skb, NULL, dev,
155 156
			    ip6_finish_output,
			    !(IP6CB(skb)->flags & IP6SKB_REROUTED));
L
Linus Torvalds 已提交
157 158 159
}

/*
160 161 162 163
 * xmit an sk_buff (used by TCP, SCTP and DCCP)
 * Note : socket lock is not held for SYNACK packets, but might be modified
 * by calls to skb_set_owner_w() and ipv6_local_error(),
 * which are using proper atomic operations or spinlocks.
L
Linus Torvalds 已提交
164
 */
165
int ip6_xmit(const struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6,
166
	     struct ipv6_txoptions *opt, int tclass)
L
Linus Torvalds 已提交
167
{
168
	struct net *net = sock_net(sk);
169
	const struct ipv6_pinfo *np = inet6_sk(sk);
170
	struct in6_addr *first_hop = &fl6->daddr;
E
Eric Dumazet 已提交
171
	struct dst_entry *dst = skb_dst(skb);
L
Linus Torvalds 已提交
172
	struct ipv6hdr *hdr;
173
	u8  proto = fl6->flowi6_proto;
L
Linus Torvalds 已提交
174
	int seg_len = skb->len;
G
Gerrit Renker 已提交
175
	int hlimit = -1;
L
Linus Torvalds 已提交
176 177 178
	u32 mtu;

	if (opt) {
179
		unsigned int head_room;
L
Linus Torvalds 已提交
180 181 182 183 184 185 186 187 188 189

		/* First: exthdrs may take lots of space (~8K for now)
		   MAX_HEADER is not enough.
		 */
		head_room = opt->opt_nflen + opt->opt_flen;
		seg_len += head_room;
		head_room += sizeof(struct ipv6hdr) + LL_RESERVED_SPACE(dst->dev);

		if (skb_headroom(skb) < head_room) {
			struct sk_buff *skb2 = skb_realloc_headroom(skb, head_room);
190
			if (!skb2) {
E
Eric Dumazet 已提交
191
				IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
192 193
					      IPSTATS_MIB_OUTDISCARDS);
				kfree_skb(skb);
L
Linus Torvalds 已提交
194 195
				return -ENOBUFS;
			}
196
			consume_skb(skb);
197
			skb = skb2;
198 199 200 201
			/* skb_set_owner_w() changes sk->sk_wmem_alloc atomically,
			 * it is safe to call in our context (socket lock not held)
			 */
			skb_set_owner_w(skb, (struct sock *)sk);
L
Linus Torvalds 已提交
202 203 204 205
		}
		if (opt->opt_flen)
			ipv6_push_frag_opts(skb, opt, &proto);
		if (opt->opt_nflen)
206 207
			ipv6_push_nfrag_opts(skb, opt, &proto, &first_hop,
					     &fl6->saddr);
L
Linus Torvalds 已提交
208 209
	}

210 211
	skb_push(skb, sizeof(struct ipv6hdr));
	skb_reset_network_header(skb);
212
	hdr = ipv6_hdr(skb);
L
Linus Torvalds 已提交
213 214 215 216

	/*
	 *	Fill in the IPv6 header
	 */
217
	if (np)
L
Linus Torvalds 已提交
218 219
		hlimit = np->hop_limit;
	if (hlimit < 0)
220
		hlimit = ip6_dst_hoplimit(dst);
L
Linus Torvalds 已提交
221

222
	ip6_flow_hdr(hdr, tclass, ip6_make_flowlabel(net, skb, fl6->flowlabel,
223
						     np->autoflowlabel, fl6));
224

L
Linus Torvalds 已提交
225 226 227 228
	hdr->payload_len = htons(seg_len);
	hdr->nexthdr = proto;
	hdr->hop_limit = hlimit;

A
Alexey Dobriyan 已提交
229 230
	hdr->saddr = fl6->saddr;
	hdr->daddr = *first_hop;
L
Linus Torvalds 已提交
231

232
	skb->protocol = htons(ETH_P_IPV6);
233
	skb->priority = sk->sk_priority;
234
	skb->mark = sk->sk_mark;
235

L
Linus Torvalds 已提交
236
	mtu = dst_mtu(dst);
W
WANG Cong 已提交
237
	if ((skb->len <= mtu) || skb->ignore_df || skb_is_gso(skb)) {
E
Eric Dumazet 已提交
238
		IP6_UPD_PO_STATS(net, ip6_dst_idev(skb_dst(skb)),
239
			      IPSTATS_MIB_OUT, skb->len);
240 241 242 243 244 245 246 247

		/* if egress device is enslaved to an L3 master device pass the
		 * skb to its handler for processing
		 */
		skb = l3mdev_ip6_out((struct sock *)sk, skb);
		if (unlikely(!skb))
			return 0;

248 249 250
		/* hooks should never assume socket lock is held.
		 * we promote our socket to non const
		 */
251
		return NF_HOOK(NFPROTO_IPV6, NF_INET_LOCAL_OUT,
252
			       net, (struct sock *)sk, skb, NULL, dst->dev,
253
			       dst_output);
L
Linus Torvalds 已提交
254 255 256
	}

	skb->dev = dst->dev;
257 258 259 260 261
	/* ipv6_local_error() does not require socket lock,
	 * we promote our socket to non const
	 */
	ipv6_local_error((struct sock *)sk, EMSGSIZE, fl6, mtu);

E
Eric Dumazet 已提交
262
	IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)), IPSTATS_MIB_FRAGFAILS);
L
Linus Torvalds 已提交
263 264 265
	kfree_skb(skb);
	return -EMSGSIZE;
}
266 267
EXPORT_SYMBOL(ip6_xmit);

L
Linus Torvalds 已提交
268 269 270 271 272 273 274 275
static int ip6_call_ra_chain(struct sk_buff *skb, int sel)
{
	struct ip6_ra_chain *ra;
	struct sock *last = NULL;

	read_lock(&ip6_ra_lock);
	for (ra = ip6_ra_chain; ra; ra = ra->next) {
		struct sock *sk = ra->sk;
276 277 278
		if (sk && ra->sel == sel &&
		    (!sk->sk_bound_dev_if ||
		     sk->sk_bound_dev_if == skb->dev->ifindex)) {
L
Linus Torvalds 已提交
279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296
			if (last) {
				struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
				if (skb2)
					rawv6_rcv(last, skb2);
			}
			last = sk;
		}
	}

	if (last) {
		rawv6_rcv(last, skb);
		read_unlock(&ip6_ra_lock);
		return 1;
	}
	read_unlock(&ip6_ra_lock);
	return 0;
}

297 298
static int ip6_forward_proxy_check(struct sk_buff *skb)
{
299
	struct ipv6hdr *hdr = ipv6_hdr(skb);
300
	u8 nexthdr = hdr->nexthdr;
301
	__be16 frag_off;
302 303 304
	int offset;

	if (ipv6_ext_hdr(nexthdr)) {
305
		offset = ipv6_skip_exthdr(skb, sizeof(*hdr), &nexthdr, &frag_off);
306 307 308 309 310 311 312 313
		if (offset < 0)
			return 0;
	} else
		offset = sizeof(struct ipv6hdr);

	if (nexthdr == IPPROTO_ICMPV6) {
		struct icmp6hdr *icmp6;

314 315
		if (!pskb_may_pull(skb, (skb_network_header(skb) +
					 offset + 1 - skb->data)))
316 317
			return 0;

318
		icmp6 = (struct icmp6hdr *)(skb_network_header(skb) + offset);
319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335

		switch (icmp6->icmp6_type) {
		case NDISC_ROUTER_SOLICITATION:
		case NDISC_ROUTER_ADVERTISEMENT:
		case NDISC_NEIGHBOUR_SOLICITATION:
		case NDISC_NEIGHBOUR_ADVERTISEMENT:
		case NDISC_REDIRECT:
			/* For reaction involving unicast neighbor discovery
			 * message destined to the proxied address, pass it to
			 * input function.
			 */
			return 1;
		default:
			break;
		}
	}

336 337 338 339 340 341 342 343 344 345
	/*
	 * The proxying router can't forward traffic sent to a link-local
	 * address, so signal the sender and discard the packet. This
	 * behavior is clarified by the MIPv6 specification.
	 */
	if (ipv6_addr_type(&hdr->daddr) & IPV6_ADDR_LINKLOCAL) {
		dst_link_failure(skb);
		return -1;
	}

346 347 348
	return 0;
}

349 350
static inline int ip6_forward_finish(struct net *net, struct sock *sk,
				     struct sk_buff *skb)
L
Linus Torvalds 已提交
351
{
352
	return dst_output(net, sk, skb);
L
Linus Torvalds 已提交
353 354
}

355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375
static unsigned int ip6_dst_mtu_forward(const struct dst_entry *dst)
{
	unsigned int mtu;
	struct inet6_dev *idev;

	if (dst_metric_locked(dst, RTAX_MTU)) {
		mtu = dst_metric_raw(dst, RTAX_MTU);
		if (mtu)
			return mtu;
	}

	mtu = IPV6_MIN_MTU;
	rcu_read_lock();
	idev = __in6_dev_get(dst->dev);
	if (idev)
		mtu = idev->cnf.mtu6;
	rcu_read_unlock();

	return mtu;
}

376 377
static bool ip6_pkt_too_big(const struct sk_buff *skb, unsigned int mtu)
{
378
	if (skb->len <= mtu)
379 380
		return false;

W
WANG Cong 已提交
381
	/* ipv6 conntrack defrag sets max_frag_size + ignore_df */
382 383 384
	if (IP6CB(skb)->frag_max_size && IP6CB(skb)->frag_max_size > mtu)
		return true;

W
WANG Cong 已提交
385
	if (skb->ignore_df)
386 387
		return false;

388
	if (skb_is_gso(skb) && skb_gso_validate_mtu(skb, mtu))
389 390 391 392 393
		return false;

	return true;
}

L
Linus Torvalds 已提交
394 395
int ip6_forward(struct sk_buff *skb)
{
E
Eric Dumazet 已提交
396
	struct dst_entry *dst = skb_dst(skb);
397
	struct ipv6hdr *hdr = ipv6_hdr(skb);
L
Linus Torvalds 已提交
398
	struct inet6_skb_parm *opt = IP6CB(skb);
399
	struct net *net = dev_net(dst->dev);
400
	u32 mtu;
401

402
	if (net->ipv6.devconf_all->forwarding == 0)
L
Linus Torvalds 已提交
403 404
		goto error;

405 406 407
	if (skb->pkt_type != PACKET_HOST)
		goto drop;

408 409 410
	if (unlikely(skb->sk))
		goto drop;

411 412 413
	if (skb_warn_if_lro(skb))
		goto drop;

L
Linus Torvalds 已提交
414
	if (!xfrm6_policy_check(NULL, XFRM_POLICY_FWD, skb)) {
E
Eric Dumazet 已提交
415 416
		__IP6_INC_STATS(net, ip6_dst_idev(dst),
				IPSTATS_MIB_INDISCARDS);
L
Linus Torvalds 已提交
417 418 419
		goto drop;
	}

420
	skb_forward_csum(skb);
L
Linus Torvalds 已提交
421 422 423 424 425 426 427 428 429 430 431 432 433 434

	/*
	 *	We DO NOT make any processing on
	 *	RA packets, pushing them to user level AS IS
	 *	without ane WARRANTY that application will be able
	 *	to interpret them. The reason is that we
	 *	cannot make anything clever here.
	 *
	 *	We are not end-node, so that if packet contains
	 *	AH/ESP, we cannot make anything.
	 *	Defragmentation also would be mistake, RA packets
	 *	cannot be fragmented, because there is no warranty
	 *	that different fragments will go along one path. --ANK
	 */
435 436
	if (unlikely(opt->flags & IP6SKB_ROUTERALERT)) {
		if (ip6_call_ra_chain(skb, ntohs(opt->ra)))
L
Linus Torvalds 已提交
437 438 439 440 441 442 443 444 445
			return 0;
	}

	/*
	 *	check and decrement ttl
	 */
	if (hdr->hop_limit <= 1) {
		/* Force OUTPUT device used as source address */
		skb->dev = dst->dev;
446
		icmpv6_send(skb, ICMPV6_TIME_EXCEED, ICMPV6_EXC_HOPLIMIT, 0);
E
Eric Dumazet 已提交
447 448
		__IP6_INC_STATS(net, ip6_dst_idev(dst),
				IPSTATS_MIB_INHDRERRORS);
L
Linus Torvalds 已提交
449 450 451 452 453

		kfree_skb(skb);
		return -ETIMEDOUT;
	}

454
	/* XXX: idev->cnf.proxy_ndp? */
455
	if (net->ipv6.devconf_all->proxy_ndp &&
456
	    pneigh_lookup(&nd_tbl, net, &hdr->daddr, skb->dev, 0)) {
457 458
		int proxied = ip6_forward_proxy_check(skb);
		if (proxied > 0)
459
			return ip6_input(skb);
460
		else if (proxied < 0) {
E
Eric Dumazet 已提交
461 462
			__IP6_INC_STATS(net, ip6_dst_idev(dst),
					IPSTATS_MIB_INDISCARDS);
463 464
			goto drop;
		}
465 466
	}

L
Linus Torvalds 已提交
467
	if (!xfrm6_route_forward(skb)) {
E
Eric Dumazet 已提交
468 469
		__IP6_INC_STATS(net, ip6_dst_idev(dst),
				IPSTATS_MIB_INDISCARDS);
L
Linus Torvalds 已提交
470 471
		goto drop;
	}
E
Eric Dumazet 已提交
472
	dst = skb_dst(skb);
L
Linus Torvalds 已提交
473 474 475

	/* IPv6 specs say nothing about it, but it is clear that we cannot
	   send redirects to source routed frames.
476
	   We don't send redirects to frames decapsulated from IPsec.
L
Linus Torvalds 已提交
477
	 */
478
	if (skb->dev == dst->dev && opt->srcrt == 0 && !skb_sec_path(skb)) {
L
Linus Torvalds 已提交
479
		struct in6_addr *target = NULL;
480
		struct inet_peer *peer;
L
Linus Torvalds 已提交
481 482 483 484 485 486 487 488
		struct rt6_info *rt;

		/*
		 *	incoming and outgoing devices are the same
		 *	send a redirect.
		 */

		rt = (struct rt6_info *) dst;
489 490
		if (rt->rt6i_flags & RTF_GATEWAY)
			target = &rt->rt6i_gateway;
L
Linus Torvalds 已提交
491 492 493
		else
			target = &hdr->daddr;

494
		peer = inet_getpeer_v6(net->ipv6.peers, &hdr->daddr, 1);
495

L
Linus Torvalds 已提交
496 497 498
		/* Limit redirects both by destination (here)
		   and by source (inside ndisc_send_redirect)
		 */
499
		if (inet_peer_xrlim_allow(peer, 1*HZ))
500
			ndisc_send_redirect(skb, target);
501 502
		if (peer)
			inet_putpeer(peer);
503 504 505
	} else {
		int addrtype = ipv6_addr_type(&hdr->saddr);

L
Linus Torvalds 已提交
506
		/* This check is security critical. */
507 508
		if (addrtype == IPV6_ADDR_ANY ||
		    addrtype & (IPV6_ADDR_MULTICAST | IPV6_ADDR_LOOPBACK))
509 510 511
			goto error;
		if (addrtype & IPV6_ADDR_LINKLOCAL) {
			icmpv6_send(skb, ICMPV6_DEST_UNREACH,
512
				    ICMPV6_NOT_NEIGHBOUR, 0);
513 514
			goto error;
		}
L
Linus Torvalds 已提交
515 516
	}

517
	mtu = ip6_dst_mtu_forward(dst);
518 519 520
	if (mtu < IPV6_MIN_MTU)
		mtu = IPV6_MIN_MTU;

521
	if (ip6_pkt_too_big(skb, mtu)) {
L
Linus Torvalds 已提交
522 523
		/* Again, force OUTPUT device used as source address */
		skb->dev = dst->dev;
524
		icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
E
Eric Dumazet 已提交
525 526 527 528
		__IP6_INC_STATS(net, ip6_dst_idev(dst),
				IPSTATS_MIB_INTOOBIGERRORS);
		__IP6_INC_STATS(net, ip6_dst_idev(dst),
				IPSTATS_MIB_FRAGFAILS);
L
Linus Torvalds 已提交
529 530 531 532 533
		kfree_skb(skb);
		return -EMSGSIZE;
	}

	if (skb_cow(skb, dst->dev->hard_header_len)) {
E
Eric Dumazet 已提交
534 535
		__IP6_INC_STATS(net, ip6_dst_idev(dst),
				IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
536 537 538
		goto drop;
	}

539
	hdr = ipv6_hdr(skb);
L
Linus Torvalds 已提交
540 541

	/* Mangling hops number delayed to point after skb COW */
542

L
Linus Torvalds 已提交
543 544
	hdr->hop_limit--;

E
Eric Dumazet 已提交
545 546
	__IP6_INC_STATS(net, ip6_dst_idev(dst), IPSTATS_MIB_OUTFORWDATAGRAMS);
	__IP6_ADD_STATS(net, ip6_dst_idev(dst), IPSTATS_MIB_OUTOCTETS, skb->len);
547 548
	return NF_HOOK(NFPROTO_IPV6, NF_INET_FORWARD,
		       net, NULL, skb, skb->dev, dst->dev,
549
		       ip6_forward_finish);
L
Linus Torvalds 已提交
550 551

error:
E
Eric Dumazet 已提交
552
	__IP6_INC_STATS(net, ip6_dst_idev(dst), IPSTATS_MIB_INADDRERRORS);
L
Linus Torvalds 已提交
553 554 555 556 557 558 559 560 561 562
drop:
	kfree_skb(skb);
	return -EINVAL;
}

static void ip6_copy_metadata(struct sk_buff *to, struct sk_buff *from)
{
	to->pkt_type = from->pkt_type;
	to->priority = from->priority;
	to->protocol = from->protocol;
E
Eric Dumazet 已提交
563 564
	skb_dst_drop(to);
	skb_dst_set(to, dst_clone(skb_dst(from)));
L
Linus Torvalds 已提交
565
	to->dev = from->dev;
T
Thomas Graf 已提交
566
	to->mark = from->mark;
L
Linus Torvalds 已提交
567 568 569 570

#ifdef CONFIG_NET_SCHED
	to->tc_index = from->tc_index;
#endif
571
	nf_copy(to, from);
572
	skb_copy_secmark(to, from);
L
Linus Torvalds 已提交
573 574
}

575 576
int ip6_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
		 int (*output)(struct net *, struct sock *, struct sk_buff *))
L
Linus Torvalds 已提交
577 578
{
	struct sk_buff *frag;
579
	struct rt6_info *rt = (struct rt6_info *)skb_dst(skb);
580 581
	struct ipv6_pinfo *np = skb->sk && !dev_recursion_level() ?
				inet6_sk(skb->sk) : NULL;
L
Linus Torvalds 已提交
582 583 584
	struct ipv6hdr *tmp_hdr;
	struct frag_hdr *fh;
	unsigned int mtu, hlen, left, len;
585
	int hroom, troom;
586
	__be32 frag_id;
587
	int ptr, offset = 0, err = 0;
L
Linus Torvalds 已提交
588 589 590 591 592
	u8 *prevhdr, nexthdr = 0;

	hlen = ip6_find_1stfragopt(skb, &prevhdr);
	nexthdr = *prevhdr;

J
John Heffner 已提交
593
	mtu = ip6_skb_dst_mtu(skb);
594 595

	/* We must not fragment if the socket is set to force MTU discovery
596
	 * or if the skb it not generated by a local socket.
597
	 */
598 599
	if (unlikely(!skb->ignore_df && skb->len > mtu))
		goto fail_toobig;
600

601 602 603 604 605 606 607 608
	if (IP6CB(skb)->frag_max_size) {
		if (IP6CB(skb)->frag_max_size > mtu)
			goto fail_toobig;

		/* don't send fragments larger than what we received */
		mtu = IP6CB(skb)->frag_max_size;
		if (mtu < IPV6_MIN_MTU)
			mtu = IPV6_MIN_MTU;
609 610
	}

611 612 613 614
	if (np && np->frag_size < mtu) {
		if (np->frag_size)
			mtu = np->frag_size;
	}
615
	if (mtu < hlen + sizeof(struct frag_hdr) + 8)
616
		goto fail_toobig;
617
	mtu -= hlen + sizeof(struct frag_hdr);
L
Linus Torvalds 已提交
618

619 620
	frag_id = ipv6_select_ident(net, &ipv6_hdr(skb)->daddr,
				    &ipv6_hdr(skb)->saddr);
621

622 623 624 625
	if (skb->ip_summed == CHECKSUM_PARTIAL &&
	    (err = skb_checksum_help(skb)))
		goto fail;

626
	hroom = LL_RESERVED_SPACE(rt->dst.dev);
627
	if (skb_has_frag_list(skb)) {
628
		unsigned int first_len = skb_pagelen(skb);
629
		struct sk_buff *frag2;
L
Linus Torvalds 已提交
630 631 632

		if (first_len - hlen > mtu ||
		    ((first_len - hlen) & 7) ||
633 634
		    skb_cloned(skb) ||
		    skb_headroom(skb) < (hroom + sizeof(struct frag_hdr)))
L
Linus Torvalds 已提交
635 636
			goto slow_path;

637
		skb_walk_frags(skb, frag) {
L
Linus Torvalds 已提交
638 639 640
			/* Correct geometry. */
			if (frag->len > mtu ||
			    ((frag->len & 7) && frag->next) ||
641
			    skb_headroom(frag) < (hlen + hroom + sizeof(struct frag_hdr)))
642
				goto slow_path_clean;
L
Linus Torvalds 已提交
643 644 645

			/* Partially cloned skb? */
			if (skb_shared(frag))
646
				goto slow_path_clean;
647 648 649 650 651 652

			BUG_ON(frag->sk);
			if (skb->sk) {
				frag->sk = skb->sk;
				frag->destructor = sock_wfree;
			}
653
			skb->truesize -= frag->truesize;
L
Linus Torvalds 已提交
654 655 656 657 658 659
		}

		err = 0;
		offset = 0;
		/* BUILD HEADER */

Y
YOSHIFUJI Hideaki 已提交
660
		*prevhdr = NEXTHDR_FRAGMENT;
661
		tmp_hdr = kmemdup(skb_network_header(skb), hlen, GFP_ATOMIC);
L
Linus Torvalds 已提交
662
		if (!tmp_hdr) {
E
Eric Dumazet 已提交
663
			IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
664
				      IPSTATS_MIB_FRAGFAILS);
665 666
			err = -ENOMEM;
			goto fail;
L
Linus Torvalds 已提交
667
		}
668 669
		frag = skb_shinfo(skb)->frag_list;
		skb_frag_list_init(skb);
L
Linus Torvalds 已提交
670 671

		__skb_pull(skb, hlen);
672
		fh = (struct frag_hdr *)__skb_push(skb, sizeof(struct frag_hdr));
673 674
		__skb_push(skb, hlen);
		skb_reset_network_header(skb);
675
		memcpy(skb_network_header(skb), tmp_hdr, hlen);
L
Linus Torvalds 已提交
676 677 678 679

		fh->nexthdr = nexthdr;
		fh->reserved = 0;
		fh->frag_off = htons(IP6_MF);
680
		fh->identification = frag_id;
L
Linus Torvalds 已提交
681 682 683 684

		first_len = skb_pagelen(skb);
		skb->data_len = first_len - skb_headlen(skb);
		skb->len = first_len;
685 686
		ipv6_hdr(skb)->payload_len = htons(first_len -
						   sizeof(struct ipv6hdr));
687

688
		dst_hold(&rt->dst);
L
Linus Torvalds 已提交
689 690 691 692 693 694

		for (;;) {
			/* Prepare header of the next frame,
			 * before previous one went down. */
			if (frag) {
				frag->ip_summed = CHECKSUM_NONE;
695
				skb_reset_transport_header(frag);
696
				fh = (struct frag_hdr *)__skb_push(frag, sizeof(struct frag_hdr));
697 698
				__skb_push(frag, hlen);
				skb_reset_network_header(frag);
699 700
				memcpy(skb_network_header(frag), tmp_hdr,
				       hlen);
L
Linus Torvalds 已提交
701 702 703 704
				offset += skb->len - hlen - sizeof(struct frag_hdr);
				fh->nexthdr = nexthdr;
				fh->reserved = 0;
				fh->frag_off = htons(offset);
705
				if (frag->next)
L
Linus Torvalds 已提交
706 707
					fh->frag_off |= htons(IP6_MF);
				fh->identification = frag_id;
708 709 710
				ipv6_hdr(frag)->payload_len =
						htons(frag->len -
						      sizeof(struct ipv6hdr));
L
Linus Torvalds 已提交
711 712
				ip6_copy_metadata(frag, skb);
			}
713

714
			err = output(net, sk, skb);
715
			if (!err)
716
				IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
717
					      IPSTATS_MIB_FRAGCREATES);
718

L
Linus Torvalds 已提交
719 720 721 722 723 724 725 726
			if (err || !frag)
				break;

			skb = frag;
			frag = skb->next;
			skb->next = NULL;
		}

J
Jesper Juhl 已提交
727
		kfree(tmp_hdr);
L
Linus Torvalds 已提交
728 729

		if (err == 0) {
730
			IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
731
				      IPSTATS_MIB_FRAGOKS);
A
Amerigo Wang 已提交
732
			ip6_rt_put(rt);
L
Linus Torvalds 已提交
733 734 735
			return 0;
		}

736
		kfree_skb_list(frag);
L
Linus Torvalds 已提交
737

738
		IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
739
			      IPSTATS_MIB_FRAGFAILS);
A
Amerigo Wang 已提交
740
		ip6_rt_put(rt);
L
Linus Torvalds 已提交
741
		return err;
742 743 744 745 746 747 748 749 750

slow_path_clean:
		skb_walk_frags(skb, frag2) {
			if (frag2 == frag)
				break;
			frag2->sk = NULL;
			frag2->destructor = NULL;
			skb->truesize += frag2->truesize;
		}
L
Linus Torvalds 已提交
751 752 753 754 755 756 757 758 759 760 761
	}

slow_path:
	left = skb->len - hlen;		/* Space per frame */
	ptr = hlen;			/* Where to start from */

	/*
	 *	Fragment the datagram.
	 */

	*prevhdr = NEXTHDR_FRAGMENT;
762
	troom = rt->dst.dev->needed_tailroom;
L
Linus Torvalds 已提交
763 764 765 766

	/*
	 *	Keep copying data until we run out.
	 */
767
	while (left > 0)	{
L
Linus Torvalds 已提交
768 769 770 771
		len = left;
		/* IF: it doesn't fit, use 'mtu' - the data space left */
		if (len > mtu)
			len = mtu;
L
Lucas De Marchi 已提交
772
		/* IF: we are not sending up to and including the packet end
L
Linus Torvalds 已提交
773 774 775 776 777
		   then align the next start on an eight byte boundary */
		if (len < left)	{
			len &= ~7;
		}

778 779 780 781
		/* Allocate buffer */
		frag = alloc_skb(len + hlen + sizeof(struct frag_hdr) +
				 hroom + troom, GFP_ATOMIC);
		if (!frag) {
E
Eric Dumazet 已提交
782
			IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
783
				      IPSTATS_MIB_FRAGFAILS);
L
Linus Torvalds 已提交
784 785 786 787 788 789 790 791 792
			err = -ENOMEM;
			goto fail;
		}

		/*
		 *	Set up data on packet
		 */

		ip6_copy_metadata(frag, skb);
793
		skb_reserve(frag, hroom);
L
Linus Torvalds 已提交
794
		skb_put(frag, len + hlen + sizeof(struct frag_hdr));
795
		skb_reset_network_header(frag);
796
		fh = (struct frag_hdr *)(skb_network_header(frag) + hlen);
797 798
		frag->transport_header = (frag->network_header + hlen +
					  sizeof(struct frag_hdr));
L
Linus Torvalds 已提交
799 800 801 802 803 804 805 806 807 808 809

		/*
		 *	Charge the memory for the fragment to any owner
		 *	it might possess
		 */
		if (skb->sk)
			skb_set_owner_w(frag, skb->sk);

		/*
		 *	Copy the packet header into the new buffer.
		 */
810
		skb_copy_from_linear_data(skb, skb_network_header(frag), hlen);
L
Linus Torvalds 已提交
811 812 813 814 815 816

		/*
		 *	Build fragment header.
		 */
		fh->nexthdr = nexthdr;
		fh->reserved = 0;
817
		fh->identification = frag_id;
L
Linus Torvalds 已提交
818 819 820 821

		/*
		 *	Copy a block of the IP datagram.
		 */
H
Himangi Saraogi 已提交
822 823
		BUG_ON(skb_copy_bits(skb, ptr, skb_transport_header(frag),
				     len));
L
Linus Torvalds 已提交
824 825 826 827 828
		left -= len;

		fh->frag_off = htons(offset);
		if (left > 0)
			fh->frag_off |= htons(IP6_MF);
829 830
		ipv6_hdr(frag)->payload_len = htons(frag->len -
						    sizeof(struct ipv6hdr));
L
Linus Torvalds 已提交
831 832 833 834 835 836 837

		ptr += len;
		offset += len;

		/*
		 *	Put this fragment into the sending queue.
		 */
838
		err = output(net, sk, frag);
L
Linus Torvalds 已提交
839 840
		if (err)
			goto fail;
841

E
Eric Dumazet 已提交
842
		IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
843
			      IPSTATS_MIB_FRAGCREATES);
L
Linus Torvalds 已提交
844
	}
E
Eric Dumazet 已提交
845
	IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
846
		      IPSTATS_MIB_FRAGOKS);
847
	consume_skb(skb);
L
Linus Torvalds 已提交
848 849
	return err;

850 851 852 853 854 855 856 857
fail_toobig:
	if (skb->sk && dst_allfrag(skb_dst(skb)))
		sk_nocaps_add(skb->sk, NETIF_F_GSO_MASK);

	skb->dev = skb_dst(skb)->dev;
	icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
	err = -EMSGSIZE;

L
Linus Torvalds 已提交
858
fail:
E
Eric Dumazet 已提交
859
	IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
860
		      IPSTATS_MIB_FRAGFAILS);
861
	kfree_skb(skb);
L
Linus Torvalds 已提交
862 863 864
	return err;
}

865 866 867
static inline int ip6_rt_check(const struct rt6key *rt_key,
			       const struct in6_addr *fl_addr,
			       const struct in6_addr *addr_cache)
868
{
E
Eric Dumazet 已提交
869
	return (rt_key->plen != 128 || !ipv6_addr_equal(fl_addr, &rt_key->addr)) &&
870
		(!addr_cache || !ipv6_addr_equal(fl_addr, addr_cache));
871 872
}

873 874
static struct dst_entry *ip6_sk_dst_check(struct sock *sk,
					  struct dst_entry *dst,
875
					  const struct flowi6 *fl6)
L
Linus Torvalds 已提交
876
{
877
	struct ipv6_pinfo *np = inet6_sk(sk);
878
	struct rt6_info *rt;
L
Linus Torvalds 已提交
879

880 881 882
	if (!dst)
		goto out;

883 884 885 886 887 888
	if (dst->ops->family != AF_INET6) {
		dst_release(dst);
		return NULL;
	}

	rt = (struct rt6_info *)dst;
889 890 891
	/* Yes, checking route validity in not connected
	 * case is not very simple. Take into account,
	 * that we do not support routing by source, TOS,
892
	 * and MSG_DONTROUTE		--ANK (980726)
893
	 *
894 895
	 * 1. ip6_rt_check(): If route was host route,
	 *    check that cached destination is current.
896 897 898 899 900 901 902 903 904 905
	 *    If it is network route, we still may
	 *    check its validity using saved pointer
	 *    to the last used address: daddr_cache.
	 *    We do not want to save whole address now,
	 *    (because main consumer of this service
	 *    is tcp, which has not this problem),
	 *    so that the last trick works only on connected
	 *    sockets.
	 * 2. oif also should be the same.
	 */
906
	if (ip6_rt_check(&rt->rt6i_dst, &fl6->daddr, np->daddr_cache) ||
907
#ifdef CONFIG_IPV6_SUBTREES
908
	    ip6_rt_check(&rt->rt6i_src, &fl6->saddr, np->saddr_cache) ||
909
#endif
D
David Ahern 已提交
910 911
	   (!(fl6->flowi6_flags & FLOWI_FLAG_SKIP_NH_OIF) &&
	      (fl6->flowi6_oif && fl6->flowi6_oif != dst->dev->ifindex))) {
912 913
		dst_release(dst);
		dst = NULL;
L
Linus Torvalds 已提交
914 915
	}

916 917 918 919
out:
	return dst;
}

920
static int ip6_dst_lookup_tail(struct net *net, const struct sock *sk,
921
			       struct dst_entry **dst, struct flowi6 *fl6)
922
{
923 924
#ifdef CONFIG_IPV6_OPTIMISTIC_DAD
	struct neighbour *n;
925
	struct rt6_info *rt;
926 927
#endif
	int err;
928
	int flags = 0;
929

930 931 932 933 934 935 936 937 938 939 940 941
	/* The correct way to handle this would be to do
	 * ip6_route_get_saddr, and then ip6_route_output; however,
	 * the route-specific preferred source forces the
	 * ip6_route_output call _before_ ip6_route_get_saddr.
	 *
	 * In source specific routing (no src=any default route),
	 * ip6_route_output will fail given src=any saddr, though, so
	 * that's why we try it again later.
	 */
	if (ipv6_addr_any(&fl6->saddr) && (!*dst || !(*dst)->error)) {
		struct rt6_info *rt;
		bool had_dst = *dst != NULL;
L
Linus Torvalds 已提交
942

943 944 945
		if (!had_dst)
			*dst = ip6_route_output(net, sk, fl6);
		rt = (*dst)->error ? NULL : (struct rt6_info *)*dst;
946 947 948
		err = ip6_route_get_saddr(net, rt, &fl6->daddr,
					  sk ? inet6_sk(sk)->srcprefs : 0,
					  &fl6->saddr);
949
		if (err)
L
Linus Torvalds 已提交
950
			goto out_err_release;
951 952 953 954 955 956 957 958 959

		/* If we had an erroneous initial result, pretend it
		 * never existed and let the SA-enabled version take
		 * over.
		 */
		if (!had_dst && (*dst)->error) {
			dst_release(*dst);
			*dst = NULL;
		}
960 961 962

		if (fl6->flowi6_oif)
			flags |= RT6_LOOKUP_F_IFACE;
L
Linus Torvalds 已提交
963 964
	}

965
	if (!*dst)
966
		*dst = ip6_route_output_flags(net, sk, fl6, flags);
967 968 969 970 971

	err = (*dst)->error;
	if (err)
		goto out_err_release;

972
#ifdef CONFIG_IPV6_OPTIMISTIC_DAD
973 974 975 976 977 978 979 980
	/*
	 * Here if the dst entry we've looked up
	 * has a neighbour entry that is in the INCOMPLETE
	 * state and the src address from the flow is
	 * marked as OPTIMISTIC, we release the found
	 * dst entry and replace it instead with the
	 * dst entry of the nexthop router
	 */
981
	rt = (struct rt6_info *) *dst;
982
	rcu_read_lock_bh();
983 984
	n = __ipv6_neigh_lookup_noref(rt->dst.dev,
				      rt6_nexthop(rt, &fl6->daddr));
985 986 987 988
	err = n && !(n->nud_state & NUD_VALID) ? -EINVAL : 0;
	rcu_read_unlock_bh();

	if (err) {
989
		struct inet6_ifaddr *ifp;
990
		struct flowi6 fl_gw6;
991 992
		int redirect;

993
		ifp = ipv6_get_ifaddr(net, &fl6->saddr,
994 995 996 997 998 999 1000 1001 1002 1003 1004 1005
				      (*dst)->dev, 1);

		redirect = (ifp && ifp->flags & IFA_F_OPTIMISTIC);
		if (ifp)
			in6_ifa_put(ifp);

		if (redirect) {
			/*
			 * We need to get the dst entry for the
			 * default router instead
			 */
			dst_release(*dst);
1006 1007 1008
			memcpy(&fl_gw6, fl6, sizeof(struct flowi6));
			memset(&fl_gw6.daddr, 0, sizeof(struct in6_addr));
			*dst = ip6_route_output(net, sk, &fl_gw6);
1009 1010
			err = (*dst)->error;
			if (err)
1011
				goto out_err_release;
1012
		}
1013
	}
1014 1015
#endif

L
Linus Torvalds 已提交
1016 1017 1018 1019 1020
	return 0;

out_err_release:
	dst_release(*dst);
	*dst = NULL;
1021

1022 1023
	if (err == -ENETUNREACH)
		IP6_INC_STATS(net, NULL, IPSTATS_MIB_OUTNOROUTES);
L
Linus Torvalds 已提交
1024 1025
	return err;
}
A
Adrian Bunk 已提交
1026

1027 1028 1029 1030
/**
 *	ip6_dst_lookup - perform route lookup on flow
 *	@sk: socket which provides route info
 *	@dst: pointer to dst_entry * for result
1031
 *	@fl6: flow to lookup
1032 1033 1034 1035 1036
 *
 *	This function performs a route lookup on the given flow.
 *
 *	It returns zero on success, or a standard errno code on error.
 */
1037 1038
int ip6_dst_lookup(struct net *net, struct sock *sk, struct dst_entry **dst,
		   struct flowi6 *fl6)
1039 1040
{
	*dst = NULL;
1041
	return ip6_dst_lookup_tail(net, sk, dst, fl6);
1042
}
1043 1044
EXPORT_SYMBOL_GPL(ip6_dst_lookup);

1045
/**
1046 1047
 *	ip6_dst_lookup_flow - perform route lookup on flow with ipsec
 *	@sk: socket which provides route info
1048
 *	@fl6: flow to lookup
1049 1050 1051 1052 1053 1054 1055
 *	@final_dst: final destination address for ipsec lookup
 *
 *	This function performs a route lookup on the given flow.
 *
 *	It returns a valid dst pointer on success, or a pointer encoded
 *	error code.
 */
1056
struct dst_entry *ip6_dst_lookup_flow(const struct sock *sk, struct flowi6 *fl6,
1057
				      const struct in6_addr *final_dst)
1058 1059 1060 1061
{
	struct dst_entry *dst = NULL;
	int err;

1062
	err = ip6_dst_lookup_tail(sock_net(sk), sk, &dst, fl6);
1063 1064 1065
	if (err)
		return ERR_PTR(err);
	if (final_dst)
A
Alexey Dobriyan 已提交
1066
		fl6->daddr = *final_dst;
1067

1068
	return xfrm_lookup_route(sock_net(sk), dst, flowi6_to_flowi(fl6), sk, 0);
1069 1070 1071 1072 1073
}
EXPORT_SYMBOL_GPL(ip6_dst_lookup_flow);

/**
 *	ip6_sk_dst_lookup_flow - perform socket cached route lookup on flow
1074
 *	@sk: socket which provides the dst cache and route info
1075
 *	@fl6: flow to lookup
1076
 *	@final_dst: final destination address for ipsec lookup
1077 1078 1079 1080 1081 1082
 *
 *	This function performs a route lookup on the given flow with the
 *	possibility of using the cached route in the socket if it is valid.
 *	It will take the socket dst lock when operating on the dst cache.
 *	As a result, this function can only be used in process context.
 *
1083 1084
 *	It returns a valid dst pointer on success, or a pointer encoded
 *	error code.
1085
 */
1086
struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
1087
					 const struct in6_addr *final_dst)
1088
{
1089
	struct dst_entry *dst = sk_dst_check(sk, inet6_sk(sk)->dst_cookie);
1090

1091
	dst = ip6_sk_dst_check(sk, dst, fl6);
1092 1093
	if (!dst)
		dst = ip6_dst_lookup_flow(sk, fl6, final_dst);
1094

1095
	return dst;
1096
}
1097
EXPORT_SYMBOL_GPL(ip6_sk_dst_lookup_flow);
1098

A
Adrian Bunk 已提交
1099
static inline int ip6_ufo_append_data(struct sock *sk,
1100
			struct sk_buff_head *queue,
1101 1102 1103
			int getfrag(void *from, char *to, int offset, int len,
			int odd, struct sk_buff *skb),
			void *from, int length, int hh_len, int fragheaderlen,
1104 1105
			int exthdrlen, int transhdrlen, int mtu,
			unsigned int flags, const struct flowi6 *fl6)
1106 1107 1108 1109 1110 1111 1112 1113 1114

{
	struct sk_buff *skb;
	int err;

	/* There is support for UDP large send offload by network
	 * device, so create one single skb packet containing complete
	 * udp datagram
	 */
1115
	skb = skb_peek_tail(queue);
1116
	if (!skb) {
1117 1118 1119
		skb = sock_alloc_send_skb(sk,
			hh_len + fragheaderlen + transhdrlen + 20,
			(flags & MSG_DONTWAIT), &err);
1120
		if (!skb)
1121
			return err;
1122 1123 1124 1125 1126

		/* reserve space for Hardware header */
		skb_reserve(skb, hh_len);

		/* create space for UDP/IP header */
1127
		skb_put(skb, fragheaderlen + transhdrlen);
1128 1129

		/* initialize network header pointer */
1130
		skb_set_network_header(skb, exthdrlen);
1131 1132

		/* initialize protocol header pointer */
1133
		skb->transport_header = skb->network_header + fragheaderlen;
1134

1135
		skb->protocol = htons(ETH_P_IPV6);
1136 1137
		skb->csum = 0;

1138
		__skb_queue_tail(queue, skb);
1139 1140
	} else if (skb_is_gso(skb)) {
		goto append;
1141 1142
	}

1143 1144 1145 1146 1147 1148 1149
	skb->ip_summed = CHECKSUM_PARTIAL;
	/* Specify the length of each IPv6 datagram fragment.
	 * It has to be a multiple of 8.
	 */
	skb_shinfo(skb)->gso_size = (mtu - fragheaderlen -
				     sizeof(struct frag_hdr)) & ~7;
	skb_shinfo(skb)->gso_type = SKB_GSO_UDP;
1150 1151 1152
	skb_shinfo(skb)->ip6_frag_id = ipv6_select_ident(sock_net(sk),
							 &fl6->daddr,
							 &fl6->saddr);
1153 1154

append:
1155 1156
	return skb_append_datato_frags(sk, skb, getfrag, from,
				       (length - transhdrlen));
1157
}
L
Linus Torvalds 已提交
1158

1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170
static inline struct ipv6_opt_hdr *ip6_opt_dup(struct ipv6_opt_hdr *src,
					       gfp_t gfp)
{
	return src ? kmemdup(src, (src->hdrlen + 1) * 8, gfp) : NULL;
}

static inline struct ipv6_rt_hdr *ip6_rthdr_dup(struct ipv6_rt_hdr *src,
						gfp_t gfp)
{
	return src ? kmemdup(src, (src->hdrlen + 1) * 8, gfp) : NULL;
}

1171
static void ip6_append_data_mtu(unsigned int *mtu,
G
Gao feng 已提交
1172 1173 1174
				int *maxfraglen,
				unsigned int fragheaderlen,
				struct sk_buff *skb,
1175
				struct rt6_info *rt,
1176
				unsigned int orig_mtu)
G
Gao feng 已提交
1177 1178
{
	if (!(rt->dst.flags & DST_XFRM_TUNNEL)) {
1179
		if (!skb) {
G
Gao feng 已提交
1180
			/* first fragment, reserve header_len */
1181
			*mtu = orig_mtu - rt->dst.header_len;
G
Gao feng 已提交
1182 1183 1184 1185 1186 1187

		} else {
			/*
			 * this fragment is not first, the headers
			 * space is regarded as data space.
			 */
1188
			*mtu = orig_mtu;
G
Gao feng 已提交
1189 1190 1191 1192 1193 1194
		}
		*maxfraglen = ((*mtu - fragheaderlen) & ~7)
			      + fragheaderlen - sizeof(struct frag_hdr);
	}
}

1195
static int ip6_setup_cork(struct sock *sk, struct inet_cork_full *cork,
W
Wei Wang 已提交
1196
			  struct inet6_cork *v6_cork, struct ipcm6_cookie *ipc6,
1197 1198 1199 1200
			  struct rt6_info *rt, struct flowi6 *fl6)
{
	struct ipv6_pinfo *np = inet6_sk(sk);
	unsigned int mtu;
W
Wei Wang 已提交
1201
	struct ipv6_txoptions *opt = ipc6->opt;
1202 1203 1204 1205 1206 1207 1208 1209 1210

	/*
	 * setup for corking
	 */
	if (opt) {
		if (WARN_ON(v6_cork->opt))
			return -EINVAL;

		v6_cork->opt = kzalloc(opt->tot_len, sk->sk_allocation);
1211
		if (unlikely(!v6_cork->opt))
1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242
			return -ENOBUFS;

		v6_cork->opt->tot_len = opt->tot_len;
		v6_cork->opt->opt_flen = opt->opt_flen;
		v6_cork->opt->opt_nflen = opt->opt_nflen;

		v6_cork->opt->dst0opt = ip6_opt_dup(opt->dst0opt,
						    sk->sk_allocation);
		if (opt->dst0opt && !v6_cork->opt->dst0opt)
			return -ENOBUFS;

		v6_cork->opt->dst1opt = ip6_opt_dup(opt->dst1opt,
						    sk->sk_allocation);
		if (opt->dst1opt && !v6_cork->opt->dst1opt)
			return -ENOBUFS;

		v6_cork->opt->hopopt = ip6_opt_dup(opt->hopopt,
						   sk->sk_allocation);
		if (opt->hopopt && !v6_cork->opt->hopopt)
			return -ENOBUFS;

		v6_cork->opt->srcrt = ip6_rthdr_dup(opt->srcrt,
						    sk->sk_allocation);
		if (opt->srcrt && !v6_cork->opt->srcrt)
			return -ENOBUFS;

		/* need source address above miyazawa*/
	}
	dst_hold(&rt->dst);
	cork->base.dst = &rt->dst;
	cork->fl.u.ip6 = *fl6;
W
Wei Wang 已提交
1243 1244
	v6_cork->hop_limit = ipc6->hlimit;
	v6_cork->tclass = ipc6->tclass;
1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262
	if (rt->dst.flags & DST_XFRM_TUNNEL)
		mtu = np->pmtudisc >= IPV6_PMTUDISC_PROBE ?
		      rt->dst.dev->mtu : dst_mtu(&rt->dst);
	else
		mtu = np->pmtudisc >= IPV6_PMTUDISC_PROBE ?
		      rt->dst.dev->mtu : dst_mtu(rt->dst.path);
	if (np->frag_size < mtu) {
		if (np->frag_size)
			mtu = np->frag_size;
	}
	cork->base.fragsize = mtu;
	if (dst_allfrag(rt->dst.path))
		cork->base.flags |= IPCORK_ALLFRAG;
	cork->base.length = 0;

	return 0;
}

1263 1264 1265 1266 1267 1268 1269 1270 1271
static int __ip6_append_data(struct sock *sk,
			     struct flowi6 *fl6,
			     struct sk_buff_head *queue,
			     struct inet_cork *cork,
			     struct inet6_cork *v6_cork,
			     struct page_frag *pfrag,
			     int getfrag(void *from, char *to, int offset,
					 int len, int odd, struct sk_buff *skb),
			     void *from, int length, int transhdrlen,
W
Wei Wang 已提交
1272
			     unsigned int flags, struct ipcm6_cookie *ipc6,
1273
			     const struct sockcm_cookie *sockc)
L
Linus Torvalds 已提交
1274
{
G
Gao feng 已提交
1275
	struct sk_buff *skb, *skb_prev = NULL;
1276
	unsigned int maxfraglen, fragheaderlen, mtu, orig_mtu;
1277 1278
	int exthdrlen = 0;
	int dst_exthdrlen = 0;
L
Linus Torvalds 已提交
1279 1280 1281 1282
	int hh_len;
	int copy;
	int err;
	int offset = 0;
1283
	__u8 tx_flags = 0;
1284
	u32 tskey = 0;
1285 1286
	struct rt6_info *rt = (struct rt6_info *)cork->dst;
	struct ipv6_txoptions *opt = v6_cork->opt;
1287
	int csummode = CHECKSUM_NONE;
1288
	unsigned int maxnonfragsize, headersize;
L
Linus Torvalds 已提交
1289

1290 1291 1292
	skb = skb_peek_tail(queue);
	if (!skb) {
		exthdrlen = opt ? opt->opt_flen : 0;
1293
		dst_exthdrlen = rt->dst.header_len - rt->rt6i_nfheader_len;
L
Linus Torvalds 已提交
1294
	}
1295

1296
	mtu = cork->fragsize;
1297
	orig_mtu = mtu;
L
Linus Torvalds 已提交
1298

1299
	hh_len = LL_RESERVED_SPACE(rt->dst.dev);
L
Linus Torvalds 已提交
1300

1301
	fragheaderlen = sizeof(struct ipv6hdr) + rt->rt6i_nfheader_len +
1302
			(opt ? opt->opt_nflen : 0);
1303 1304
	maxfraglen = ((mtu - fragheaderlen) & ~7) + fragheaderlen -
		     sizeof(struct frag_hdr);
L
Linus Torvalds 已提交
1305

1306 1307 1308 1309 1310 1311
	headersize = sizeof(struct ipv6hdr) +
		     (opt ? opt->opt_flen + opt->opt_nflen : 0) +
		     (dst_allfrag(&rt->dst) ?
		      sizeof(struct frag_hdr) : 0) +
		     rt->rt6i_nfheader_len;

W
Wei Wang 已提交
1312
	if (cork->length + length > mtu - headersize && ipc6->dontfrag &&
1313 1314 1315 1316 1317 1318
	    (sk->sk_protocol == IPPROTO_UDP ||
	     sk->sk_protocol == IPPROTO_RAW)) {
		ipv6_local_rxpmtu(sk, fl6, mtu - headersize +
				sizeof(struct ipv6hdr));
		goto emsgsize;
	}
1319

1320 1321 1322 1323
	if (ip6_sk_ignore_df(sk))
		maxnonfragsize = sizeof(struct ipv6hdr) + IPV6_MAXPLEN;
	else
		maxnonfragsize = mtu;
1324

1325
	if (cork->length + length > maxnonfragsize - headersize) {
1326
emsgsize:
1327 1328 1329 1330
		ipv6_local_error(sk, EMSGSIZE, fl6,
				 mtu - headersize +
				 sizeof(struct ipv6hdr));
		return -EMSGSIZE;
L
Linus Torvalds 已提交
1331 1332
	}

1333 1334 1335 1336 1337 1338 1339
	/* CHECKSUM_PARTIAL only with no extension headers and when
	 * we are not going to fragment
	 */
	if (transhdrlen && sk->sk_protocol == IPPROTO_UDP &&
	    headersize == sizeof(struct ipv6hdr) &&
	    length < mtu - headersize &&
	    !(flags & MSG_MORE) &&
1340
	    rt->dst.dev->features & (NETIF_F_IPV6_CSUM | NETIF_F_HW_CSUM))
1341 1342
		csummode = CHECKSUM_PARTIAL;

1343
	if (sk->sk_type == SOCK_DGRAM || sk->sk_type == SOCK_RAW) {
1344
		sock_tx_timestamp(sk, sockc->tsflags, &tx_flags);
1345 1346 1347 1348
		if (tx_flags & SKBTX_ANY_SW_TSTAMP &&
		    sk->sk_tsflags & SOF_TIMESTAMPING_OPT_ID)
			tskey = sk->sk_tskey++;
	}
1349

L
Linus Torvalds 已提交
1350 1351 1352 1353 1354 1355 1356
	/*
	 * Let's try using as much space as possible.
	 * Use MTU if total length of the message fits into the MTU.
	 * Otherwise, we need to reserve fragment header and
	 * fragment alignment (= 8-15 octects, in total).
	 *
	 * Note that we may need to "move" the data from the tail of
1357
	 * of the buffer to the new fragment when we split
L
Linus Torvalds 已提交
1358 1359
	 * the message.
	 *
1360
	 * FIXME: It may be fragmented into multiple chunks
L
Linus Torvalds 已提交
1361 1362
	 *        at once if non-fragmentable extension headers
	 *        are too large.
1363
	 * --yoshfuji
L
Linus Torvalds 已提交
1364 1365
	 */

1366 1367 1368 1369
	cork->length += length;
	if (((length > mtu) ||
	     (skb && skb_is_gso(skb))) &&
	    (sk->sk_protocol == IPPROTO_UDP) &&
1370
	    (rt->dst.dev->features & NETIF_F_UFO) && !rt->dst.header_len &&
1371
	    (sk->sk_type == SOCK_DGRAM) && !udp_get_no_check6_tx(sk)) {
1372
		err = ip6_ufo_append_data(sk, queue, getfrag, from, length,
1373
					  hh_len, fragheaderlen, exthdrlen,
1374
					  transhdrlen, mtu, flags, fl6);
1375 1376 1377
		if (err)
			goto error;
		return 0;
1378
	}
L
Linus Torvalds 已提交
1379

1380
	if (!skb)
L
Linus Torvalds 已提交
1381 1382 1383 1384
		goto alloc_new_skb;

	while (length > 0) {
		/* Check if the remaining data fits into current packet. */
1385
		copy = (cork->length <= mtu && !(cork->flags & IPCORK_ALLFRAG) ? mtu : maxfraglen) - skb->len;
L
Linus Torvalds 已提交
1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396
		if (copy < length)
			copy = maxfraglen - skb->len;

		if (copy <= 0) {
			char *data;
			unsigned int datalen;
			unsigned int fraglen;
			unsigned int fraggap;
			unsigned int alloclen;
alloc_new_skb:
			/* There's no room in the current skb */
G
Gao feng 已提交
1397 1398
			if (skb)
				fraggap = skb->len - maxfraglen;
L
Linus Torvalds 已提交
1399 1400
			else
				fraggap = 0;
G
Gao feng 已提交
1401
			/* update mtu and maxfraglen if necessary */
1402
			if (!skb || !skb_prev)
G
Gao feng 已提交
1403
				ip6_append_data_mtu(&mtu, &maxfraglen,
1404
						    fragheaderlen, skb, rt,
1405
						    orig_mtu);
G
Gao feng 已提交
1406 1407

			skb_prev = skb;
L
Linus Torvalds 已提交
1408 1409 1410 1411 1412 1413 1414

			/*
			 * If remaining data exceeds the mtu,
			 * we know we need more fragment(s).
			 */
			datalen = length + fraggap;

G
Gao feng 已提交
1415 1416
			if (datalen > (cork->length <= mtu && !(cork->flags & IPCORK_ALLFRAG) ? mtu : maxfraglen) - fragheaderlen)
				datalen = maxfraglen - fragheaderlen - rt->dst.trailer_len;
L
Linus Torvalds 已提交
1417
			if ((flags & MSG_MORE) &&
1418
			    !(rt->dst.dev->features&NETIF_F_SG))
L
Linus Torvalds 已提交
1419 1420 1421 1422
				alloclen = mtu;
			else
				alloclen = datalen + fragheaderlen;

1423 1424
			alloclen += dst_exthdrlen;

G
Gao feng 已提交
1425 1426 1427 1428 1429 1430 1431 1432 1433 1434
			if (datalen != length + fraggap) {
				/*
				 * this is not the last fragment, the trailer
				 * space is regarded as data space.
				 */
				datalen += rt->dst.trailer_len;
			}

			alloclen += rt->dst.trailer_len;
			fraglen = datalen + fragheaderlen;
L
Linus Torvalds 已提交
1435 1436 1437

			/*
			 * We just reserve space for fragment header.
1438
			 * Note: this may be overallocation if the message
L
Linus Torvalds 已提交
1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453
			 * (without MSG_MORE) fits into the MTU.
			 */
			alloclen += sizeof(struct frag_hdr);

			if (transhdrlen) {
				skb = sock_alloc_send_skb(sk,
						alloclen + hh_len,
						(flags & MSG_DONTWAIT), &err);
			} else {
				skb = NULL;
				if (atomic_read(&sk->sk_wmem_alloc) <=
				    2 * sk->sk_sndbuf)
					skb = sock_wmalloc(sk,
							   alloclen + hh_len, 1,
							   sk->sk_allocation);
1454
				if (unlikely(!skb))
L
Linus Torvalds 已提交
1455 1456
					err = -ENOBUFS;
			}
1457
			if (!skb)
L
Linus Torvalds 已提交
1458 1459 1460 1461
				goto error;
			/*
			 *	Fill in the control structures
			 */
1462
			skb->protocol = htons(ETH_P_IPV6);
1463
			skb->ip_summed = csummode;
L
Linus Torvalds 已提交
1464
			skb->csum = 0;
1465 1466 1467
			/* reserve for fragmentation and ipsec header */
			skb_reserve(skb, hh_len + sizeof(struct frag_hdr) +
				    dst_exthdrlen);
L
Linus Torvalds 已提交
1468

1469 1470 1471
			/* Only the initial fragment is time stamped */
			skb_shinfo(skb)->tx_flags = tx_flags;
			tx_flags = 0;
1472 1473
			skb_shinfo(skb)->tskey = tskey;
			tskey = 0;
1474

L
Linus Torvalds 已提交
1475 1476 1477
			/*
			 *	Find where to start putting bytes
			 */
1478 1479 1480
			data = skb_put(skb, fraglen);
			skb_set_network_header(skb, exthdrlen);
			data += fragheaderlen;
1481 1482
			skb->transport_header = (skb->network_header +
						 fragheaderlen);
L
Linus Torvalds 已提交
1483 1484 1485 1486 1487 1488 1489
			if (fraggap) {
				skb->csum = skb_copy_and_csum_bits(
					skb_prev, maxfraglen,
					data + transhdrlen, fraggap, 0);
				skb_prev->csum = csum_sub(skb_prev->csum,
							  skb->csum);
				data += fraggap;
1490
				pskb_trim_unique(skb_prev, maxfraglen);
L
Linus Torvalds 已提交
1491 1492
			}
			copy = datalen - transhdrlen - fraggap;
1493

L
Linus Torvalds 已提交
1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507
			if (copy < 0) {
				err = -EINVAL;
				kfree_skb(skb);
				goto error;
			} else if (copy > 0 && getfrag(from, data + transhdrlen, offset, copy, fraggap, skb) < 0) {
				err = -EFAULT;
				kfree_skb(skb);
				goto error;
			}

			offset += copy;
			length -= datalen - fraggap;
			transhdrlen = 0;
			exthdrlen = 0;
1508
			dst_exthdrlen = 0;
L
Linus Torvalds 已提交
1509 1510 1511 1512

			/*
			 * Put the packet on the pending queue
			 */
1513
			__skb_queue_tail(queue, skb);
L
Linus Torvalds 已提交
1514 1515 1516 1517 1518 1519
			continue;
		}

		if (copy > length)
			copy = length;

1520
		if (!(rt->dst.dev->features&NETIF_F_SG)) {
L
Linus Torvalds 已提交
1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532
			unsigned int off;

			off = skb->len;
			if (getfrag(from, skb_put(skb, copy),
						offset, copy, off, skb) < 0) {
				__skb_trim(skb, off);
				err = -EFAULT;
				goto error;
			}
		} else {
			int i = skb_shinfo(skb)->nr_frags;

1533 1534
			err = -ENOMEM;
			if (!sk_page_frag_refill(sk, pfrag))
L
Linus Torvalds 已提交
1535
				goto error;
1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546

			if (!skb_can_coalesce(skb, i, pfrag->page,
					      pfrag->offset)) {
				err = -EMSGSIZE;
				if (i == MAX_SKB_FRAGS)
					goto error;

				__skb_fill_page_desc(skb, i, pfrag->page,
						     pfrag->offset, 0);
				skb_shinfo(skb)->nr_frags = ++i;
				get_page(pfrag->page);
L
Linus Torvalds 已提交
1547
			}
1548
			copy = min_t(int, copy, pfrag->size - pfrag->offset);
E
Eric Dumazet 已提交
1549
			if (getfrag(from,
1550 1551 1552 1553 1554 1555
				    page_address(pfrag->page) + pfrag->offset,
				    offset, copy, skb->len, skb) < 0)
				goto error_efault;

			pfrag->offset += copy;
			skb_frag_size_add(&skb_shinfo(skb)->frags[i - 1], copy);
L
Linus Torvalds 已提交
1556 1557
			skb->len += copy;
			skb->data_len += copy;
1558 1559
			skb->truesize += copy;
			atomic_add(copy, &sk->sk_wmem_alloc);
L
Linus Torvalds 已提交
1560 1561 1562 1563
		}
		offset += copy;
		length -= copy;
	}
1564

L
Linus Torvalds 已提交
1565
	return 0;
1566 1567 1568

error_efault:
	err = -EFAULT;
L
Linus Torvalds 已提交
1569
error:
1570
	cork->length -= length;
1571
	IP6_INC_STATS(sock_net(sk), rt->rt6i_idev, IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
1572 1573
	return err;
}
1574 1575 1576 1577

int ip6_append_data(struct sock *sk,
		    int getfrag(void *from, char *to, int offset, int len,
				int odd, struct sk_buff *skb),
W
Wei Wang 已提交
1578 1579 1580
		    void *from, int length, int transhdrlen,
		    struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
		    struct rt6_info *rt, unsigned int flags,
1581
		    const struct sockcm_cookie *sockc)
1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593
{
	struct inet_sock *inet = inet_sk(sk);
	struct ipv6_pinfo *np = inet6_sk(sk);
	int exthdrlen;
	int err;

	if (flags&MSG_PROBE)
		return 0;
	if (skb_queue_empty(&sk->sk_write_queue)) {
		/*
		 * setup for corking
		 */
W
Wei Wang 已提交
1594 1595
		err = ip6_setup_cork(sk, &inet->cork, &np->cork,
				     ipc6, rt, fl6);
1596 1597 1598
		if (err)
			return err;

W
Wei Wang 已提交
1599
		exthdrlen = (ipc6->opt ? ipc6->opt->opt_flen : 0);
1600 1601 1602 1603 1604 1605 1606 1607 1608
		length += exthdrlen;
		transhdrlen += exthdrlen;
	} else {
		fl6 = &inet->cork.fl.u.ip6;
		transhdrlen = 0;
	}

	return __ip6_append_data(sk, fl6, &sk->sk_write_queue, &inet->cork.base,
				 &np->cork, sk_page_frag(sk), getfrag,
W
Wei Wang 已提交
1609
				 from, length, transhdrlen, flags, ipc6, sockc);
1610
}
1611
EXPORT_SYMBOL_GPL(ip6_append_data);
L
Linus Torvalds 已提交
1612

1613 1614
static void ip6_cork_release(struct inet_cork_full *cork,
			     struct inet6_cork *v6_cork)
1615
{
1616 1617 1618 1619 1620 1621 1622
	if (v6_cork->opt) {
		kfree(v6_cork->opt->dst0opt);
		kfree(v6_cork->opt->dst1opt);
		kfree(v6_cork->opt->hopopt);
		kfree(v6_cork->opt->srcrt);
		kfree(v6_cork->opt);
		v6_cork->opt = NULL;
1623 1624
	}

1625 1626 1627 1628
	if (cork->base.dst) {
		dst_release(cork->base.dst);
		cork->base.dst = NULL;
		cork->base.flags &= ~IPCORK_ALLFRAG;
1629
	}
1630
	memset(&cork->fl, 0, sizeof(cork->fl));
1631 1632
}

V
Vlad Yasevich 已提交
1633 1634 1635 1636
struct sk_buff *__ip6_make_skb(struct sock *sk,
			       struct sk_buff_head *queue,
			       struct inet_cork_full *cork,
			       struct inet6_cork *v6_cork)
L
Linus Torvalds 已提交
1637 1638 1639 1640 1641
{
	struct sk_buff *skb, *tmp_skb;
	struct sk_buff **tail_skb;
	struct in6_addr final_dst_buf, *final_dst = &final_dst_buf;
	struct ipv6_pinfo *np = inet6_sk(sk);
1642
	struct net *net = sock_net(sk);
L
Linus Torvalds 已提交
1643
	struct ipv6hdr *hdr;
V
Vlad Yasevich 已提交
1644 1645 1646
	struct ipv6_txoptions *opt = v6_cork->opt;
	struct rt6_info *rt = (struct rt6_info *)cork->base.dst;
	struct flowi6 *fl6 = &cork->fl.u.ip6;
1647
	unsigned char proto = fl6->flowi6_proto;
L
Linus Torvalds 已提交
1648

V
Vlad Yasevich 已提交
1649
	skb = __skb_dequeue(queue);
1650
	if (!skb)
L
Linus Torvalds 已提交
1651 1652 1653 1654
		goto out;
	tail_skb = &(skb_shinfo(skb)->frag_list);

	/* move skb->data to ip header from ext header */
1655
	if (skb->data < skb_network_header(skb))
1656
		__skb_pull(skb, skb_network_offset(skb));
V
Vlad Yasevich 已提交
1657
	while ((tmp_skb = __skb_dequeue(queue)) != NULL) {
1658
		__skb_pull(tmp_skb, skb_network_header_len(skb));
L
Linus Torvalds 已提交
1659 1660 1661 1662 1663 1664 1665 1666 1667
		*tail_skb = tmp_skb;
		tail_skb = &(tmp_skb->next);
		skb->len += tmp_skb->len;
		skb->data_len += tmp_skb->len;
		skb->truesize += tmp_skb->truesize;
		tmp_skb->destructor = NULL;
		tmp_skb->sk = NULL;
	}

1668
	/* Allow local fragmentation. */
W
WANG Cong 已提交
1669
	skb->ignore_df = ip6_sk_ignore_df(sk);
1670

A
Alexey Dobriyan 已提交
1671
	*final_dst = fl6->daddr;
1672
	__skb_pull(skb, skb_network_header_len(skb));
L
Linus Torvalds 已提交
1673 1674 1675
	if (opt && opt->opt_flen)
		ipv6_push_frag_opts(skb, opt, &proto);
	if (opt && opt->opt_nflen)
1676
		ipv6_push_nfrag_opts(skb, opt, &proto, &final_dst, &fl6->saddr);
L
Linus Torvalds 已提交
1677

1678 1679
	skb_push(skb, sizeof(struct ipv6hdr));
	skb_reset_network_header(skb);
1680
	hdr = ipv6_hdr(skb);
1681

V
Vlad Yasevich 已提交
1682
	ip6_flow_hdr(hdr, v6_cork->tclass,
1683
		     ip6_make_flowlabel(net, skb, fl6->flowlabel,
1684
					np->autoflowlabel, fl6));
V
Vlad Yasevich 已提交
1685
	hdr->hop_limit = v6_cork->hop_limit;
L
Linus Torvalds 已提交
1686
	hdr->nexthdr = proto;
A
Alexey Dobriyan 已提交
1687 1688
	hdr->saddr = fl6->saddr;
	hdr->daddr = *final_dst;
L
Linus Torvalds 已提交
1689

1690
	skb->priority = sk->sk_priority;
1691
	skb->mark = sk->sk_mark;
1692

1693
	skb_dst_set(skb, dst_clone(&rt->dst));
1694
	IP6_UPD_PO_STATS(net, rt->rt6i_idev, IPSTATS_MIB_OUT, skb->len);
1695
	if (proto == IPPROTO_ICMPV6) {
E
Eric Dumazet 已提交
1696
		struct inet6_dev *idev = ip6_dst_idev(skb_dst(skb));
1697

1698 1699
		ICMP6MSGOUT_INC_STATS(net, idev, icmp6_hdr(skb)->icmp6_type);
		ICMP6_INC_STATS(net, idev, ICMP6_MIB_OUTMSGS);
1700 1701
	}

V
Vlad Yasevich 已提交
1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712
	ip6_cork_release(cork, v6_cork);
out:
	return skb;
}

int ip6_send_skb(struct sk_buff *skb)
{
	struct net *net = sock_net(skb->sk);
	struct rt6_info *rt = (struct rt6_info *)skb_dst(skb);
	int err;

1713
	err = ip6_local_out(net, skb->sk, skb);
L
Linus Torvalds 已提交
1714 1715
	if (err) {
		if (err > 0)
E
Eric Dumazet 已提交
1716
			err = net_xmit_errno(err);
L
Linus Torvalds 已提交
1717
		if (err)
V
Vlad Yasevich 已提交
1718 1719
			IP6_INC_STATS(net, rt->rt6i_idev,
				      IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
1720 1721 1722
	}

	return err;
V
Vlad Yasevich 已提交
1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733
}

int ip6_push_pending_frames(struct sock *sk)
{
	struct sk_buff *skb;

	skb = ip6_finish_skb(sk);
	if (!skb)
		return 0;

	return ip6_send_skb(skb);
L
Linus Torvalds 已提交
1734
}
1735
EXPORT_SYMBOL_GPL(ip6_push_pending_frames);
L
Linus Torvalds 已提交
1736

1737
static void __ip6_flush_pending_frames(struct sock *sk,
V
Vlad Yasevich 已提交
1738 1739 1740
				       struct sk_buff_head *queue,
				       struct inet_cork_full *cork,
				       struct inet6_cork *v6_cork)
L
Linus Torvalds 已提交
1741 1742 1743
{
	struct sk_buff *skb;

1744
	while ((skb = __skb_dequeue_tail(queue)) != NULL) {
E
Eric Dumazet 已提交
1745 1746
		if (skb_dst(skb))
			IP6_INC_STATS(sock_net(sk), ip6_dst_idev(skb_dst(skb)),
1747
				      IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
1748 1749 1750
		kfree_skb(skb);
	}

V
Vlad Yasevich 已提交
1751
	ip6_cork_release(cork, v6_cork);
L
Linus Torvalds 已提交
1752
}
1753 1754 1755

void ip6_flush_pending_frames(struct sock *sk)
{
V
Vlad Yasevich 已提交
1756 1757
	__ip6_flush_pending_frames(sk, &sk->sk_write_queue,
				   &inet_sk(sk)->cork, &inet6_sk(sk)->cork);
1758
}
1759
EXPORT_SYMBOL_GPL(ip6_flush_pending_frames);
V
Vlad Yasevich 已提交
1760 1761 1762 1763 1764

struct sk_buff *ip6_make_skb(struct sock *sk,
			     int getfrag(void *from, char *to, int offset,
					 int len, int odd, struct sk_buff *skb),
			     void *from, int length, int transhdrlen,
W
Wei Wang 已提交
1765
			     struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
V
Vlad Yasevich 已提交
1766
			     struct rt6_info *rt, unsigned int flags,
W
Wei Wang 已提交
1767
			     const struct sockcm_cookie *sockc)
V
Vlad Yasevich 已提交
1768 1769 1770 1771
{
	struct inet_cork_full cork;
	struct inet6_cork v6_cork;
	struct sk_buff_head queue;
W
Wei Wang 已提交
1772
	int exthdrlen = (ipc6->opt ? ipc6->opt->opt_flen : 0);
V
Vlad Yasevich 已提交
1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783
	int err;

	if (flags & MSG_PROBE)
		return NULL;

	__skb_queue_head_init(&queue);

	cork.base.flags = 0;
	cork.base.addr = 0;
	cork.base.opt = NULL;
	v6_cork.opt = NULL;
W
Wei Wang 已提交
1784
	err = ip6_setup_cork(sk, &cork, &v6_cork, ipc6, rt, fl6);
V
Vlad Yasevich 已提交
1785 1786 1787
	if (err)
		return ERR_PTR(err);

W
Wei Wang 已提交
1788 1789
	if (ipc6->dontfrag < 0)
		ipc6->dontfrag = inet6_sk(sk)->dontfrag;
V
Vlad Yasevich 已提交
1790 1791 1792 1793

	err = __ip6_append_data(sk, fl6, &queue, &cork.base, &v6_cork,
				&current->task_frag, getfrag, from,
				length + exthdrlen, transhdrlen + exthdrlen,
W
Wei Wang 已提交
1794
				flags, ipc6, sockc);
V
Vlad Yasevich 已提交
1795 1796 1797 1798 1799 1800 1801
	if (err) {
		__ip6_flush_pending_frames(sk, &queue, &cork, &v6_cork);
		return ERR_PTR(err);
	}

	return __ip6_make_skb(sk, &queue, &cork, &v6_cork);
}