ip6_output.c 40.2 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 23 24 25 26 27 28 29
 *
 *	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>
 *	Imran Patel	: 	frag id should be in NBO
 *      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>
L
Linus Torvalds 已提交
58

59
static int ip6_finish_output2(struct sk_buff *skb)
L
Linus Torvalds 已提交
60
{
E
Eric Dumazet 已提交
61
	struct dst_entry *dst = skb_dst(skb);
L
Linus Torvalds 已提交
62
	struct net_device *dev = dst->dev;
63
	struct neighbour *neigh;
64 65
	struct in6_addr *nexthop;
	int ret;
L
Linus Torvalds 已提交
66 67 68 69

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

70
	if (ipv6_addr_is_multicast(&ipv6_hdr(skb)->daddr)) {
E
Eric Dumazet 已提交
71
		struct inet6_dev *idev = ip6_dst_idev(skb_dst(skb));
L
Linus Torvalds 已提交
72

73
		if (!(dev->flags & IFF_LOOPBACK) && sk_mc_loop(skb->sk) &&
74
		    ((mroute6_socket(dev_net(dev), skb) &&
75
		     !(IP6CB(skb)->flags & IP6SKB_FORWARDED)) ||
76 77
		     ipv6_chk_mcast_addr(dev, &ipv6_hdr(skb)->daddr,
					 &ipv6_hdr(skb)->saddr))) {
L
Linus Torvalds 已提交
78 79 80 81 82 83
			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)
84 85
				NF_HOOK(NFPROTO_IPV6, NF_INET_POST_ROUTING,
					newskb, NULL, newskb->dev,
86
					dev_loopback_xmit);
L
Linus Torvalds 已提交
87

88
			if (ipv6_hdr(skb)->hop_limit == 0) {
89 90
				IP6_INC_STATS(dev_net(dev), idev,
					      IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
91 92 93 94 95
				kfree_skb(skb);
				return 0;
			}
		}

96 97
		IP6_UPD_PO_STATS(dev_net(dev), idev, IPSTATS_MIB_OUTMCAST,
				skb->len);
98 99 100 101 102 103 104

		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
105 106
	}

107
	rcu_read_lock_bh();
108
	nexthop = rt6_nexthop((struct rt6_info *)dst);
109 110 111 112 113 114 115 116 117
	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();
118

119 120
	IP6_INC_STATS(dev_net(dst->dev),
		      ip6_dst_idev(dst), IPSTATS_MIB_OUTNOROUTES);
121 122
	kfree_skb(skb);
	return -EINVAL;
L
Linus Torvalds 已提交
123 124
}

125 126 127
static int ip6_finish_output(struct sk_buff *skb)
{
	if ((skb->len > ip6_skb_dst_mtu(skb) && !skb_is_gso(skb)) ||
128 129
	    dst_allfrag(skb_dst(skb)) ||
	    (IP6CB(skb)->frag_max_size && skb->len > IP6CB(skb)->frag_max_size))
130 131 132 133 134
		return ip6_fragment(skb, ip6_finish_output2);
	else
		return ip6_finish_output2(skb);
}

L
Linus Torvalds 已提交
135 136
int ip6_output(struct sk_buff *skb)
{
137
	struct net_device *dev = skb_dst(skb)->dev;
E
Eric Dumazet 已提交
138
	struct inet6_dev *idev = ip6_dst_idev(skb_dst(skb));
139
	if (unlikely(idev->cnf.disable_ipv6)) {
140
		IP6_INC_STATS(dev_net(dev), idev,
141
			      IPSTATS_MIB_OUTDISCARDS);
142 143 144 145
		kfree_skb(skb);
		return 0;
	}

146 147 148
	return NF_HOOK_COND(NFPROTO_IPV6, NF_INET_POST_ROUTING, skb, NULL, dev,
			    ip6_finish_output,
			    !(IP6CB(skb)->flags & IP6SKB_REROUTED));
L
Linus Torvalds 已提交
149 150 151
}

/*
S
Shan Wei 已提交
152
 *	xmit an sk_buff (used by TCP, SCTP and DCCP)
L
Linus Torvalds 已提交
153 154
 */

155
int ip6_xmit(struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6,
156
	     struct ipv6_txoptions *opt, int tclass)
L
Linus Torvalds 已提交
157
{
158
	struct net *net = sock_net(sk);
159
	struct ipv6_pinfo *np = inet6_sk(sk);
160
	struct in6_addr *first_hop = &fl6->daddr;
E
Eric Dumazet 已提交
161
	struct dst_entry *dst = skb_dst(skb);
L
Linus Torvalds 已提交
162
	struct ipv6hdr *hdr;
163
	u8  proto = fl6->flowi6_proto;
L
Linus Torvalds 已提交
164
	int seg_len = skb->len;
G
Gerrit Renker 已提交
165
	int hlimit = -1;
L
Linus Torvalds 已提交
166 167 168
	u32 mtu;

	if (opt) {
169
		unsigned int head_room;
L
Linus Torvalds 已提交
170 171 172 173 174 175 176 177 178 179

		/* 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);
180
			if (skb2 == NULL) {
E
Eric Dumazet 已提交
181
				IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
182 183
					      IPSTATS_MIB_OUTDISCARDS);
				kfree_skb(skb);
L
Linus Torvalds 已提交
184 185
				return -ENOBUFS;
			}
186
			consume_skb(skb);
187
			skb = skb2;
188
			skb_set_owner_w(skb, sk);
L
Linus Torvalds 已提交
189 190 191 192 193 194 195
		}
		if (opt->opt_flen)
			ipv6_push_frag_opts(skb, opt, &proto);
		if (opt->opt_nflen)
			ipv6_push_nfrag_opts(skb, opt, &proto, &first_hop);
	}

196 197
	skb_push(skb, sizeof(struct ipv6hdr));
	skb_reset_network_header(skb);
198
	hdr = ipv6_hdr(skb);
L
Linus Torvalds 已提交
199 200 201 202

	/*
	 *	Fill in the IPv6 header
	 */
203
	if (np)
L
Linus Torvalds 已提交
204 205
		hlimit = np->hop_limit;
	if (hlimit < 0)
206
		hlimit = ip6_dst_hoplimit(dst);
L
Linus Torvalds 已提交
207

208
	ip6_flow_hdr(hdr, tclass, fl6->flowlabel);
209

L
Linus Torvalds 已提交
210 211 212 213
	hdr->payload_len = htons(seg_len);
	hdr->nexthdr = proto;
	hdr->hop_limit = hlimit;

A
Alexey Dobriyan 已提交
214 215
	hdr->saddr = fl6->saddr;
	hdr->daddr = *first_hop;
L
Linus Torvalds 已提交
216

217
	skb->protocol = htons(ETH_P_IPV6);
218
	skb->priority = sk->sk_priority;
219
	skb->mark = sk->sk_mark;
220

L
Linus Torvalds 已提交
221
	mtu = dst_mtu(dst);
222
	if ((skb->len <= mtu) || skb->local_df || skb_is_gso(skb)) {
E
Eric Dumazet 已提交
223
		IP6_UPD_PO_STATS(net, ip6_dst_idev(skb_dst(skb)),
224
			      IPSTATS_MIB_OUT, skb->len);
225 226
		return NF_HOOK(NFPROTO_IPV6, NF_INET_LOCAL_OUT, skb, NULL,
			       dst->dev, dst_output);
L
Linus Torvalds 已提交
227 228 229
	}

	skb->dev = dst->dev;
230
	ipv6_local_error(sk, EMSGSIZE, fl6, mtu);
E
Eric Dumazet 已提交
231
	IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)), IPSTATS_MIB_FRAGFAILS);
L
Linus Torvalds 已提交
232 233 234 235
	kfree_skb(skb);
	return -EMSGSIZE;
}

236 237
EXPORT_SYMBOL(ip6_xmit);

L
Linus Torvalds 已提交
238 239 240 241 242 243 244 245
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;
246 247 248
		if (sk && ra->sel == sel &&
		    (!sk->sk_bound_dev_if ||
		     sk->sk_bound_dev_if == skb->dev->ifindex)) {
L
Linus Torvalds 已提交
249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266
			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;
}

267 268
static int ip6_forward_proxy_check(struct sk_buff *skb)
{
269
	struct ipv6hdr *hdr = ipv6_hdr(skb);
270
	u8 nexthdr = hdr->nexthdr;
271
	__be16 frag_off;
272 273 274
	int offset;

	if (ipv6_ext_hdr(nexthdr)) {
275
		offset = ipv6_skip_exthdr(skb, sizeof(*hdr), &nexthdr, &frag_off);
276 277 278 279 280 281 282 283
		if (offset < 0)
			return 0;
	} else
		offset = sizeof(struct ipv6hdr);

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

284 285
		if (!pskb_may_pull(skb, (skb_network_header(skb) +
					 offset + 1 - skb->data)))
286 287
			return 0;

288
		icmp6 = (struct icmp6hdr *)(skb_network_header(skb) + offset);
289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305

		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;
		}
	}

306 307 308 309 310 311 312 313 314 315
	/*
	 * 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;
	}

316 317 318
	return 0;
}

L
Linus Torvalds 已提交
319 320 321 322 323
static inline int ip6_forward_finish(struct sk_buff *skb)
{
	return dst_output(skb);
}

324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344
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;
}

345 346 347 348 349 350 351 352 353 354 355 356 357 358
static bool ip6_pkt_too_big(const struct sk_buff *skb, unsigned int mtu)
{
	if (skb->len <= mtu || skb->local_df)
		return false;

	if (IP6CB(skb)->frag_max_size && IP6CB(skb)->frag_max_size > mtu)
		return true;

	if (skb_is_gso(skb) && skb_gso_network_seglen(skb) <= mtu)
		return false;

	return true;
}

L
Linus Torvalds 已提交
359 360
int ip6_forward(struct sk_buff *skb)
{
E
Eric Dumazet 已提交
361
	struct dst_entry *dst = skb_dst(skb);
362
	struct ipv6hdr *hdr = ipv6_hdr(skb);
L
Linus Torvalds 已提交
363
	struct inet6_skb_parm *opt = IP6CB(skb);
364
	struct net *net = dev_net(dst->dev);
365
	u32 mtu;
366

367
	if (net->ipv6.devconf_all->forwarding == 0)
L
Linus Torvalds 已提交
368 369
		goto error;

370 371 372
	if (skb_warn_if_lro(skb))
		goto drop;

L
Linus Torvalds 已提交
373
	if (!xfrm6_policy_check(NULL, XFRM_POLICY_FWD, skb)) {
374 375
		IP6_INC_STATS_BH(net, ip6_dst_idev(dst),
				 IPSTATS_MIB_INDISCARDS);
L
Linus Torvalds 已提交
376 377 378
		goto drop;
	}

379 380 381
	if (skb->pkt_type != PACKET_HOST)
		goto drop;

382
	skb_forward_csum(skb);
L
Linus Torvalds 已提交
383 384 385 386 387 388 389 390 391 392 393 394 395 396

	/*
	 *	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
	 */
397 398
	if (unlikely(opt->flags & IP6SKB_ROUTERALERT)) {
		if (ip6_call_ra_chain(skb, ntohs(opt->ra)))
L
Linus Torvalds 已提交
399 400 401 402 403 404 405 406 407
			return 0;
	}

	/*
	 *	check and decrement ttl
	 */
	if (hdr->hop_limit <= 1) {
		/* Force OUTPUT device used as source address */
		skb->dev = dst->dev;
408
		icmpv6_send(skb, ICMPV6_TIME_EXCEED, ICMPV6_EXC_HOPLIMIT, 0);
409 410
		IP6_INC_STATS_BH(net, ip6_dst_idev(dst),
				 IPSTATS_MIB_INHDRERRORS);
L
Linus Torvalds 已提交
411 412 413 414 415

		kfree_skb(skb);
		return -ETIMEDOUT;
	}

416
	/* XXX: idev->cnf.proxy_ndp? */
417
	if (net->ipv6.devconf_all->proxy_ndp &&
418
	    pneigh_lookup(&nd_tbl, net, &hdr->daddr, skb->dev, 0)) {
419 420
		int proxied = ip6_forward_proxy_check(skb);
		if (proxied > 0)
421
			return ip6_input(skb);
422
		else if (proxied < 0) {
423 424
			IP6_INC_STATS_BH(net, ip6_dst_idev(dst),
					 IPSTATS_MIB_INDISCARDS);
425 426
			goto drop;
		}
427 428
	}

L
Linus Torvalds 已提交
429
	if (!xfrm6_route_forward(skb)) {
430 431
		IP6_INC_STATS_BH(net, ip6_dst_idev(dst),
				 IPSTATS_MIB_INDISCARDS);
L
Linus Torvalds 已提交
432 433
		goto drop;
	}
E
Eric Dumazet 已提交
434
	dst = skb_dst(skb);
L
Linus Torvalds 已提交
435 436 437

	/* IPv6 specs say nothing about it, but it is clear that we cannot
	   send redirects to source routed frames.
438
	   We don't send redirects to frames decapsulated from IPsec.
L
Linus Torvalds 已提交
439
	 */
440
	if (skb->dev == dst->dev && opt->srcrt == 0 && !skb_sec_path(skb)) {
L
Linus Torvalds 已提交
441
		struct in6_addr *target = NULL;
442
		struct inet_peer *peer;
L
Linus Torvalds 已提交
443 444 445 446 447 448 449 450
		struct rt6_info *rt;

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

		rt = (struct rt6_info *) dst;
451 452
		if (rt->rt6i_flags & RTF_GATEWAY)
			target = &rt->rt6i_gateway;
L
Linus Torvalds 已提交
453 454 455
		else
			target = &hdr->daddr;

456
		peer = inet_getpeer_v6(net->ipv6.peers, &rt->rt6i_dst.addr, 1);
457

L
Linus Torvalds 已提交
458 459 460
		/* Limit redirects both by destination (here)
		   and by source (inside ndisc_send_redirect)
		 */
461
		if (inet_peer_xrlim_allow(peer, 1*HZ))
462
			ndisc_send_redirect(skb, target);
463 464
		if (peer)
			inet_putpeer(peer);
465 466 467
	} else {
		int addrtype = ipv6_addr_type(&hdr->saddr);

L
Linus Torvalds 已提交
468
		/* This check is security critical. */
469 470
		if (addrtype == IPV6_ADDR_ANY ||
		    addrtype & (IPV6_ADDR_MULTICAST | IPV6_ADDR_LOOPBACK))
471 472 473
			goto error;
		if (addrtype & IPV6_ADDR_LINKLOCAL) {
			icmpv6_send(skb, ICMPV6_DEST_UNREACH,
474
				    ICMPV6_NOT_NEIGHBOUR, 0);
475 476
			goto error;
		}
L
Linus Torvalds 已提交
477 478
	}

479
	mtu = ip6_dst_mtu_forward(dst);
480 481 482
	if (mtu < IPV6_MIN_MTU)
		mtu = IPV6_MIN_MTU;

483
	if (ip6_pkt_too_big(skb, mtu)) {
L
Linus Torvalds 已提交
484 485
		/* Again, force OUTPUT device used as source address */
		skb->dev = dst->dev;
486
		icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
487 488 489 490
		IP6_INC_STATS_BH(net, ip6_dst_idev(dst),
				 IPSTATS_MIB_INTOOBIGERRORS);
		IP6_INC_STATS_BH(net, ip6_dst_idev(dst),
				 IPSTATS_MIB_FRAGFAILS);
L
Linus Torvalds 已提交
491 492 493 494 495
		kfree_skb(skb);
		return -EMSGSIZE;
	}

	if (skb_cow(skb, dst->dev->hard_header_len)) {
496 497
		IP6_INC_STATS_BH(net, ip6_dst_idev(dst),
				 IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
498 499 500
		goto drop;
	}

501
	hdr = ipv6_hdr(skb);
L
Linus Torvalds 已提交
502 503

	/* Mangling hops number delayed to point after skb COW */
504

L
Linus Torvalds 已提交
505 506
	hdr->hop_limit--;

507
	IP6_INC_STATS_BH(net, ip6_dst_idev(dst), IPSTATS_MIB_OUTFORWDATAGRAMS);
508
	IP6_ADD_STATS_BH(net, ip6_dst_idev(dst), IPSTATS_MIB_OUTOCTETS, skb->len);
509
	return NF_HOOK(NFPROTO_IPV6, NF_INET_FORWARD, skb, skb->dev, dst->dev,
510
		       ip6_forward_finish);
L
Linus Torvalds 已提交
511 512

error:
513
	IP6_INC_STATS_BH(net, ip6_dst_idev(dst), IPSTATS_MIB_INADDRERRORS);
L
Linus Torvalds 已提交
514 515 516 517 518 519 520 521 522 523
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
524 525
	skb_dst_drop(to);
	skb_dst_set(to, dst_clone(skb_dst(from)));
L
Linus Torvalds 已提交
526
	to->dev = from->dev;
T
Thomas Graf 已提交
527
	to->mark = from->mark;
L
Linus Torvalds 已提交
528 529 530 531

#ifdef CONFIG_NET_SCHED
	to->tc_index = from->tc_index;
#endif
532
	nf_copy(to, from);
533
	skb_copy_secmark(to, from);
L
Linus Torvalds 已提交
534 535
}

536
int ip6_fragment(struct sk_buff *skb, int (*output)(struct sk_buff *))
L
Linus Torvalds 已提交
537 538
{
	struct sk_buff *frag;
E
Eric Dumazet 已提交
539
	struct rt6_info *rt = (struct rt6_info*)skb_dst(skb);
540
	struct ipv6_pinfo *np = skb->sk ? inet6_sk(skb->sk) : NULL;
L
Linus Torvalds 已提交
541 542 543
	struct ipv6hdr *tmp_hdr;
	struct frag_hdr *fh;
	unsigned int mtu, hlen, left, len;
544
	int hroom, troom;
A
Al Viro 已提交
545
	__be32 frag_id = 0;
L
Linus Torvalds 已提交
546 547
	int ptr, offset = 0, err=0;
	u8 *prevhdr, nexthdr = 0;
E
Eric Dumazet 已提交
548
	struct net *net = dev_net(skb_dst(skb)->dev);
L
Linus Torvalds 已提交
549 550 551 552

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

J
John Heffner 已提交
553
	mtu = ip6_skb_dst_mtu(skb);
554 555

	/* We must not fragment if the socket is set to force MTU discovery
556
	 * or if the skb it not generated by a local socket.
557
	 */
558 559 560
	if (unlikely(!skb->local_df && skb->len > mtu) ||
		     (IP6CB(skb)->frag_max_size &&
		      IP6CB(skb)->frag_max_size > mtu)) {
561 562 563
		if (skb->sk && dst_allfrag(skb_dst(skb)))
			sk_nocaps_add(skb->sk, NETIF_F_GSO_MASK);

E
Eric Dumazet 已提交
564
		skb->dev = skb_dst(skb)->dev;
565
		icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
E
Eric Dumazet 已提交
566
		IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
567
			      IPSTATS_MIB_FRAGFAILS);
568 569 570 571
		kfree_skb(skb);
		return -EMSGSIZE;
	}

572 573 574 575 576
	if (np && np->frag_size < mtu) {
		if (np->frag_size)
			mtu = np->frag_size;
	}
	mtu -= hlen + sizeof(struct frag_hdr);
L
Linus Torvalds 已提交
577

578
	if (skb_has_frag_list(skb)) {
L
Linus Torvalds 已提交
579
		int first_len = skb_pagelen(skb);
580
		struct sk_buff *frag2;
L
Linus Torvalds 已提交
581 582 583 584 585 586

		if (first_len - hlen > mtu ||
		    ((first_len - hlen) & 7) ||
		    skb_cloned(skb))
			goto slow_path;

587
		skb_walk_frags(skb, frag) {
L
Linus Torvalds 已提交
588 589 590 591
			/* Correct geometry. */
			if (frag->len > mtu ||
			    ((frag->len & 7) && frag->next) ||
			    skb_headroom(frag) < hlen)
592
				goto slow_path_clean;
L
Linus Torvalds 已提交
593 594 595

			/* Partially cloned skb? */
			if (skb_shared(frag))
596
				goto slow_path_clean;
597 598 599 600 601 602

			BUG_ON(frag->sk);
			if (skb->sk) {
				frag->sk = skb->sk;
				frag->destructor = sock_wfree;
			}
603
			skb->truesize -= frag->truesize;
L
Linus Torvalds 已提交
604 605 606 607 608
		}

		err = 0;
		offset = 0;
		frag = skb_shinfo(skb)->frag_list;
609
		skb_frag_list_init(skb);
L
Linus Torvalds 已提交
610 611
		/* BUILD HEADER */

Y
YOSHIFUJI Hideaki 已提交
612
		*prevhdr = NEXTHDR_FRAGMENT;
613
		tmp_hdr = kmemdup(skb_network_header(skb), hlen, GFP_ATOMIC);
L
Linus Torvalds 已提交
614
		if (!tmp_hdr) {
E
Eric Dumazet 已提交
615
			IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
616
				      IPSTATS_MIB_FRAGFAILS);
L
Linus Torvalds 已提交
617 618 619 620 621
			return -ENOMEM;
		}

		__skb_pull(skb, hlen);
		fh = (struct frag_hdr*)__skb_push(skb, sizeof(struct frag_hdr));
622 623
		__skb_push(skb, hlen);
		skb_reset_network_header(skb);
624
		memcpy(skb_network_header(skb), tmp_hdr, hlen);
L
Linus Torvalds 已提交
625

626
		ipv6_select_ident(fh, rt);
L
Linus Torvalds 已提交
627 628 629 630 631 632 633 634
		fh->nexthdr = nexthdr;
		fh->reserved = 0;
		fh->frag_off = htons(IP6_MF);
		frag_id = fh->identification;

		first_len = skb_pagelen(skb);
		skb->data_len = first_len - skb_headlen(skb);
		skb->len = first_len;
635 636
		ipv6_hdr(skb)->payload_len = htons(first_len -
						   sizeof(struct ipv6hdr));
637

638
		dst_hold(&rt->dst);
L
Linus Torvalds 已提交
639 640 641 642 643 644

		for (;;) {
			/* Prepare header of the next frame,
			 * before previous one went down. */
			if (frag) {
				frag->ip_summed = CHECKSUM_NONE;
645
				skb_reset_transport_header(frag);
L
Linus Torvalds 已提交
646
				fh = (struct frag_hdr*)__skb_push(frag, sizeof(struct frag_hdr));
647 648
				__skb_push(frag, hlen);
				skb_reset_network_header(frag);
649 650
				memcpy(skb_network_header(frag), tmp_hdr,
				       hlen);
L
Linus Torvalds 已提交
651 652 653 654 655 656 657
				offset += skb->len - hlen - sizeof(struct frag_hdr);
				fh->nexthdr = nexthdr;
				fh->reserved = 0;
				fh->frag_off = htons(offset);
				if (frag->next != NULL)
					fh->frag_off |= htons(IP6_MF);
				fh->identification = frag_id;
658 659 660
				ipv6_hdr(frag)->payload_len =
						htons(frag->len -
						      sizeof(struct ipv6hdr));
L
Linus Torvalds 已提交
661 662
				ip6_copy_metadata(frag, skb);
			}
663

L
Linus Torvalds 已提交
664
			err = output(skb);
665
			if(!err)
666
				IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
667
					      IPSTATS_MIB_FRAGCREATES);
668

L
Linus Torvalds 已提交
669 670 671 672 673 674 675 676
			if (err || !frag)
				break;

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

J
Jesper Juhl 已提交
677
		kfree(tmp_hdr);
L
Linus Torvalds 已提交
678 679

		if (err == 0) {
680
			IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
681
				      IPSTATS_MIB_FRAGOKS);
A
Amerigo Wang 已提交
682
			ip6_rt_put(rt);
L
Linus Torvalds 已提交
683 684 685 686 687 688 689 690 691
			return 0;
		}

		while (frag) {
			skb = frag->next;
			kfree_skb(frag);
			frag = skb;
		}

692
		IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
693
			      IPSTATS_MIB_FRAGFAILS);
A
Amerigo Wang 已提交
694
		ip6_rt_put(rt);
L
Linus Torvalds 已提交
695
		return err;
696 697 698 699 700 701 702 703 704

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
705 706 707
	}

slow_path:
708 709 710 711
	if ((skb->ip_summed == CHECKSUM_PARTIAL) &&
	    skb_checksum_help(skb))
		goto fail;

L
Linus Torvalds 已提交
712 713 714 715 716 717 718 719
	left = skb->len - hlen;		/* Space per frame */
	ptr = hlen;			/* Where to start from */

	/*
	 *	Fragment the datagram.
	 */

	*prevhdr = NEXTHDR_FRAGMENT;
720 721
	hroom = LL_RESERVED_SPACE(rt->dst.dev);
	troom = rt->dst.dev->needed_tailroom;
L
Linus Torvalds 已提交
722 723 724 725 726 727 728 729 730

	/*
	 *	Keep copying data until we run out.
	 */
	while(left > 0)	{
		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
731
		/* IF: we are not sending up to and including the packet end
L
Linus Torvalds 已提交
732 733 734 735 736 737 738 739
		   then align the next start on an eight byte boundary */
		if (len < left)	{
			len &= ~7;
		}
		/*
		 *	Allocate buffer.
		 */

740 741
		if ((frag = alloc_skb(len + hlen + sizeof(struct frag_hdr) +
				      hroom + troom, GFP_ATOMIC)) == NULL) {
742
			NETDEBUG(KERN_INFO "IPv6: frag: no memory for new fragment!\n");
E
Eric Dumazet 已提交
743
			IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
744
				      IPSTATS_MIB_FRAGFAILS);
L
Linus Torvalds 已提交
745 746 747 748 749 750 751 752 753
			err = -ENOMEM;
			goto fail;
		}

		/*
		 *	Set up data on packet
		 */

		ip6_copy_metadata(frag, skb);
754
		skb_reserve(frag, hroom);
L
Linus Torvalds 已提交
755
		skb_put(frag, len + hlen + sizeof(struct frag_hdr));
756
		skb_reset_network_header(frag);
757
		fh = (struct frag_hdr *)(skb_network_header(frag) + hlen);
758 759
		frag->transport_header = (frag->network_header + hlen +
					  sizeof(struct frag_hdr));
L
Linus Torvalds 已提交
760 761 762 763 764 765 766 767 768 769 770

		/*
		 *	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.
		 */
771
		skb_copy_from_linear_data(skb, skb_network_header(frag), hlen);
L
Linus Torvalds 已提交
772 773 774 775 776 777

		/*
		 *	Build fragment header.
		 */
		fh->nexthdr = nexthdr;
		fh->reserved = 0;
778
		if (!frag_id) {
779
			ipv6_select_ident(fh, rt);
L
Linus Torvalds 已提交
780 781 782 783 784 785 786
			frag_id = fh->identification;
		} else
			fh->identification = frag_id;

		/*
		 *	Copy a block of the IP datagram.
		 */
787
		if (skb_copy_bits(skb, ptr, skb_transport_header(frag), len))
L
Linus Torvalds 已提交
788 789 790 791 792 793
			BUG();
		left -= len;

		fh->frag_off = htons(offset);
		if (left > 0)
			fh->frag_off |= htons(IP6_MF);
794 795
		ipv6_hdr(frag)->payload_len = htons(frag->len -
						    sizeof(struct ipv6hdr));
L
Linus Torvalds 已提交
796 797 798 799 800 801 802 803 804 805

		ptr += len;
		offset += len;

		/*
		 *	Put this fragment into the sending queue.
		 */
		err = output(frag);
		if (err)
			goto fail;
806

E
Eric Dumazet 已提交
807
		IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
808
			      IPSTATS_MIB_FRAGCREATES);
L
Linus Torvalds 已提交
809
	}
E
Eric Dumazet 已提交
810
	IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
811
		      IPSTATS_MIB_FRAGOKS);
812
	consume_skb(skb);
L
Linus Torvalds 已提交
813 814 815
	return err;

fail:
E
Eric Dumazet 已提交
816
	IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
817
		      IPSTATS_MIB_FRAGFAILS);
818
	kfree_skb(skb);
L
Linus Torvalds 已提交
819 820 821
	return err;
}

822 823 824
static inline int ip6_rt_check(const struct rt6key *rt_key,
			       const struct in6_addr *fl_addr,
			       const struct in6_addr *addr_cache)
825
{
E
Eric Dumazet 已提交
826 827
	return (rt_key->plen != 128 || !ipv6_addr_equal(fl_addr, &rt_key->addr)) &&
		(addr_cache == NULL || !ipv6_addr_equal(fl_addr, addr_cache));
828 829
}

830 831
static struct dst_entry *ip6_sk_dst_check(struct sock *sk,
					  struct dst_entry *dst,
832
					  const struct flowi6 *fl6)
L
Linus Torvalds 已提交
833
{
834
	struct ipv6_pinfo *np = inet6_sk(sk);
835
	struct rt6_info *rt;
L
Linus Torvalds 已提交
836

837 838 839
	if (!dst)
		goto out;

840 841 842 843 844 845
	if (dst->ops->family != AF_INET6) {
		dst_release(dst);
		return NULL;
	}

	rt = (struct rt6_info *)dst;
846 847 848 849 850
	/* Yes, checking route validity in not connected
	 * case is not very simple. Take into account,
	 * that we do not support routing by source, TOS,
	 * and MSG_DONTROUTE 		--ANK (980726)
	 *
851 852
	 * 1. ip6_rt_check(): If route was host route,
	 *    check that cached destination is current.
853 854 855 856 857 858 859 860 861 862
	 *    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.
	 */
863
	if (ip6_rt_check(&rt->rt6i_dst, &fl6->daddr, np->daddr_cache) ||
864
#ifdef CONFIG_IPV6_SUBTREES
865
	    ip6_rt_check(&rt->rt6i_src, &fl6->saddr, np->saddr_cache) ||
866
#endif
867
	    (fl6->flowi6_oif && fl6->flowi6_oif != dst->dev->ifindex)) {
868 869
		dst_release(dst);
		dst = NULL;
L
Linus Torvalds 已提交
870 871
	}

872 873 874 875 876
out:
	return dst;
}

static int ip6_dst_lookup_tail(struct sock *sk,
877
			       struct dst_entry **dst, struct flowi6 *fl6)
878
{
879
	struct net *net = sock_net(sk);
880 881
#ifdef CONFIG_IPV6_OPTIMISTIC_DAD
	struct neighbour *n;
882
	struct rt6_info *rt;
883 884
#endif
	int err;
885

L
Linus Torvalds 已提交
886
	if (*dst == NULL)
887
		*dst = ip6_route_output(net, sk, fl6);
L
Linus Torvalds 已提交
888 889 890 891

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

892
	if (ipv6_addr_any(&fl6->saddr)) {
893 894 895 896
		struct rt6_info *rt = (struct rt6_info *) *dst;
		err = ip6_route_get_saddr(net, rt, &fl6->daddr,
					  sk ? inet6_sk(sk)->srcprefs : 0,
					  &fl6->saddr);
897
		if (err)
L
Linus Torvalds 已提交
898 899 900
			goto out_err_release;
	}

901
#ifdef CONFIG_IPV6_OPTIMISTIC_DAD
902 903 904 905 906 907 908 909
	/*
	 * 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
	 */
910
	rt = (struct rt6_info *) *dst;
911
	rcu_read_lock_bh();
912
	n = __ipv6_neigh_lookup_noref(rt->dst.dev, rt6_nexthop(rt));
913 914 915 916
	err = n && !(n->nud_state & NUD_VALID) ? -EINVAL : 0;
	rcu_read_unlock_bh();

	if (err) {
917
		struct inet6_ifaddr *ifp;
918
		struct flowi6 fl_gw6;
919 920
		int redirect;

921
		ifp = ipv6_get_ifaddr(net, &fl6->saddr,
922 923 924 925 926 927 928 929 930 931 932 933
				      (*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);
934 935 936
			memcpy(&fl_gw6, fl6, sizeof(struct flowi6));
			memset(&fl_gw6.daddr, 0, sizeof(struct in6_addr));
			*dst = ip6_route_output(net, sk, &fl_gw6);
937 938
			if ((err = (*dst)->error))
				goto out_err_release;
939
		}
940
	}
941 942
#endif

L
Linus Torvalds 已提交
943 944 945
	return 0;

out_err_release:
946
	if (err == -ENETUNREACH)
947
		IP6_INC_STATS(net, NULL, IPSTATS_MIB_OUTNOROUTES);
L
Linus Torvalds 已提交
948 949 950 951
	dst_release(*dst);
	*dst = NULL;
	return err;
}
A
Adrian Bunk 已提交
952

953 954 955 956
/**
 *	ip6_dst_lookup - perform route lookup on flow
 *	@sk: socket which provides route info
 *	@dst: pointer to dst_entry * for result
957
 *	@fl6: flow to lookup
958 959 960 961 962
 *
 *	This function performs a route lookup on the given flow.
 *
 *	It returns zero on success, or a standard errno code on error.
 */
963
int ip6_dst_lookup(struct sock *sk, struct dst_entry **dst, struct flowi6 *fl6)
964 965
{
	*dst = NULL;
966
	return ip6_dst_lookup_tail(sk, dst, fl6);
967
}
968 969
EXPORT_SYMBOL_GPL(ip6_dst_lookup);

970
/**
971 972
 *	ip6_dst_lookup_flow - perform route lookup on flow with ipsec
 *	@sk: socket which provides route info
973
 *	@fl6: flow to lookup
974 975 976 977 978 979 980
 *	@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.
 */
981
struct dst_entry *ip6_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
982
				      const struct in6_addr *final_dst)
983 984 985 986
{
	struct dst_entry *dst = NULL;
	int err;

987
	err = ip6_dst_lookup_tail(sk, &dst, fl6);
988 989 990
	if (err)
		return ERR_PTR(err);
	if (final_dst)
A
Alexey Dobriyan 已提交
991
		fl6->daddr = *final_dst;
992

993
	return xfrm_lookup(sock_net(sk), dst, flowi6_to_flowi(fl6), sk, 0);
994 995 996 997 998
}
EXPORT_SYMBOL_GPL(ip6_dst_lookup_flow);

/**
 *	ip6_sk_dst_lookup_flow - perform socket cached route lookup on flow
999
 *	@sk: socket which provides the dst cache and route info
1000
 *	@fl6: flow to lookup
1001
 *	@final_dst: final destination address for ipsec lookup
1002 1003 1004 1005 1006 1007
 *
 *	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.
 *
1008 1009
 *	It returns a valid dst pointer on success, or a pointer encoded
 *	error code.
1010
 */
1011
struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
1012
					 const struct in6_addr *final_dst)
1013
{
1014 1015
	struct dst_entry *dst = sk_dst_check(sk, inet6_sk(sk)->dst_cookie);
	int err;
1016

1017
	dst = ip6_sk_dst_check(sk, dst, fl6);
1018

1019
	err = ip6_dst_lookup_tail(sk, &dst, fl6);
1020 1021 1022
	if (err)
		return ERR_PTR(err);
	if (final_dst)
A
Alexey Dobriyan 已提交
1023
		fl6->daddr = *final_dst;
1024

1025
	return xfrm_lookup(sock_net(sk), dst, flowi6_to_flowi(fl6), sk, 0);
1026
}
1027
EXPORT_SYMBOL_GPL(ip6_sk_dst_lookup_flow);
1028

A
Adrian Bunk 已提交
1029
static inline int ip6_ufo_append_data(struct sock *sk,
1030 1031 1032
			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,
1033 1034
			int transhdrlen, int mtu,unsigned int flags,
			struct rt6_info *rt)
1035 1036 1037

{
	struct sk_buff *skb;
1038
	struct frag_hdr fhdr;
1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049
	int err;

	/* There is support for UDP large send offload by network
	 * device, so create one single skb packet containing complete
	 * udp datagram
	 */
	if ((skb = skb_peek_tail(&sk->sk_write_queue)) == NULL) {
		skb = sock_alloc_send_skb(sk,
			hh_len + fragheaderlen + transhdrlen + 20,
			(flags & MSG_DONTWAIT), &err);
		if (skb == NULL)
1050
			return err;
1051 1052 1053 1054 1055 1056 1057 1058

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

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

		/* initialize network header pointer */
1059
		skb_reset_network_header(skb);
1060 1061

		/* initialize protocol header pointer */
1062
		skb->transport_header = skb->network_header + fragheaderlen;
1063

1064
		skb->protocol = htons(ETH_P_IPV6);
1065 1066 1067
		skb->csum = 0;

		__skb_queue_tail(&sk->sk_write_queue, skb);
1068 1069
	} else if (skb_is_gso(skb)) {
		goto append;
1070 1071
	}

1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082
	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;
	ipv6_select_ident(&fhdr, rt);
	skb_shinfo(skb)->ip6_frag_id = fhdr.identification;

append:
1083 1084
	return skb_append_datato_frags(sk, skb, getfrag, from,
				       (length - transhdrlen));
1085
}
L
Linus Torvalds 已提交
1086

1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098
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;
}

1099
static void ip6_append_data_mtu(unsigned int *mtu,
G
Gao feng 已提交
1100 1101 1102
				int *maxfraglen,
				unsigned int fragheaderlen,
				struct sk_buff *skb,
1103 1104
				struct rt6_info *rt,
				bool pmtuprobe)
G
Gao feng 已提交
1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115
{
	if (!(rt->dst.flags & DST_XFRM_TUNNEL)) {
		if (skb == NULL) {
			/* first fragment, reserve header_len */
			*mtu = *mtu - rt->dst.header_len;

		} else {
			/*
			 * this fragment is not first, the headers
			 * space is regarded as data space.
			 */
1116 1117 1118
			*mtu = min(*mtu, pmtuprobe ?
				   rt->dst.dev->mtu :
				   dst_mtu(rt->dst.path));
G
Gao feng 已提交
1119 1120 1121 1122 1123 1124
		}
		*maxfraglen = ((*mtu - fragheaderlen) & ~7)
			      + fragheaderlen - sizeof(struct frag_hdr);
	}
}

1125 1126 1127
int ip6_append_data(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,
1128
	int hlimit, int tclass, struct ipv6_txoptions *opt, struct flowi6 *fl6,
1129
	struct rt6_info *rt, unsigned int flags, int dontfrag)
L
Linus Torvalds 已提交
1130 1131 1132
{
	struct inet_sock *inet = inet_sk(sk);
	struct ipv6_pinfo *np = inet6_sk(sk);
1133
	struct inet_cork *cork;
G
Gao feng 已提交
1134
	struct sk_buff *skb, *skb_prev = NULL;
1135
	unsigned int maxfraglen, fragheaderlen, mtu;
L
Linus Torvalds 已提交
1136
	int exthdrlen;
1137
	int dst_exthdrlen;
L
Linus Torvalds 已提交
1138 1139 1140 1141
	int hh_len;
	int copy;
	int err;
	int offset = 0;
1142
	__u8 tx_flags = 0;
L
Linus Torvalds 已提交
1143 1144 1145

	if (flags&MSG_PROBE)
		return 0;
1146
	cork = &inet->cork.base;
L
Linus Torvalds 已提交
1147 1148 1149 1150 1151
	if (skb_queue_empty(&sk->sk_write_queue)) {
		/*
		 * setup for corking
		 */
		if (opt) {
1152
			if (WARN_ON(np->cork.opt))
L
Linus Torvalds 已提交
1153
				return -EINVAL;
1154

1155
			np->cork.opt = kzalloc(opt->tot_len, sk->sk_allocation);
1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182
			if (unlikely(np->cork.opt == NULL))
				return -ENOBUFS;

			np->cork.opt->tot_len = opt->tot_len;
			np->cork.opt->opt_flen = opt->opt_flen;
			np->cork.opt->opt_nflen = opt->opt_nflen;

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

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

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

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

L
Linus Torvalds 已提交
1183 1184
			/* need source address above miyazawa*/
		}
1185
		dst_hold(&rt->dst);
1186
		cork->dst = &rt->dst;
1187
		inet->cork.fl.u.ip6 = *fl6;
L
Linus Torvalds 已提交
1188
		np->cork.hop_limit = hlimit;
1189
		np->cork.tclass = tclass;
G
Gao feng 已提交
1190
		if (rt->dst.flags & DST_XFRM_TUNNEL)
1191
			mtu = np->pmtudisc >= IPV6_PMTUDISC_PROBE ?
G
Gao feng 已提交
1192 1193
			      rt->dst.dev->mtu : dst_mtu(&rt->dst);
		else
1194
			mtu = np->pmtudisc >= IPV6_PMTUDISC_PROBE ?
G
Gao feng 已提交
1195
			      rt->dst.dev->mtu : dst_mtu(rt->dst.path);
1196
		if (np->frag_size < mtu) {
1197 1198 1199
			if (np->frag_size)
				mtu = np->frag_size;
		}
1200
		cork->fragsize = mtu;
1201
		if (dst_allfrag(rt->dst.path))
1202 1203
			cork->flags |= IPCORK_ALLFRAG;
		cork->length = 0;
1204
		exthdrlen = (opt ? opt->opt_flen : 0);
L
Linus Torvalds 已提交
1205 1206
		length += exthdrlen;
		transhdrlen += exthdrlen;
1207
		dst_exthdrlen = rt->dst.header_len - rt->rt6i_nfheader_len;
L
Linus Torvalds 已提交
1208
	} else {
1209
		rt = (struct rt6_info *)cork->dst;
1210
		fl6 = &inet->cork.fl.u.ip6;
1211
		opt = np->cork.opt;
L
Linus Torvalds 已提交
1212 1213
		transhdrlen = 0;
		exthdrlen = 0;
1214
		dst_exthdrlen = 0;
1215
		mtu = cork->fragsize;
L
Linus Torvalds 已提交
1216 1217
	}

1218
	hh_len = LL_RESERVED_SPACE(rt->dst.dev);
L
Linus Torvalds 已提交
1219

1220
	fragheaderlen = sizeof(struct ipv6hdr) + rt->rt6i_nfheader_len +
1221
			(opt ? opt->opt_nflen : 0);
1222 1223
	maxfraglen = ((mtu - fragheaderlen) & ~7) + fragheaderlen -
		     sizeof(struct frag_hdr);
L
Linus Torvalds 已提交
1224 1225

	if (mtu <= sizeof(struct ipv6hdr) + IPV6_MAXPLEN) {
1226 1227 1228 1229 1230 1231 1232 1233
		unsigned int maxnonfragsize, headersize;

		headersize = sizeof(struct ipv6hdr) +
			     (opt ? opt->tot_len : 0) +
			     (dst_allfrag(&rt->dst) ?
			      sizeof(struct frag_hdr) : 0) +
			     rt->rt6i_nfheader_len;

1234 1235 1236 1237
		if (ip6_sk_local_df(sk))
			maxnonfragsize = sizeof(struct ipv6hdr) + IPV6_MAXPLEN;
		else
			maxnonfragsize = mtu;
1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252

		/* dontfrag active */
		if ((cork->length + length > mtu - headersize) && dontfrag &&
		    (sk->sk_protocol == IPPROTO_UDP ||
		     sk->sk_protocol == IPPROTO_RAW)) {
			ipv6_local_rxpmtu(sk, fl6, mtu - headersize +
						   sizeof(struct ipv6hdr));
			goto emsgsize;
		}

		if (cork->length + length > maxnonfragsize - headersize) {
emsgsize:
			ipv6_local_error(sk, EMSGSIZE, fl6,
					 mtu - headersize +
					 sizeof(struct ipv6hdr));
L
Linus Torvalds 已提交
1253 1254 1255 1256
			return -EMSGSIZE;
		}
	}

1257
	/* For UDP, check if TX timestamp is enabled */
1258 1259
	if (sk->sk_type == SOCK_DGRAM)
		sock_tx_timestamp(sk, &tx_flags);
1260

L
Linus Torvalds 已提交
1261 1262 1263 1264 1265 1266 1267
	/*
	 * 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
1268
	 * of the buffer to the new fragment when we split
L
Linus Torvalds 已提交
1269 1270
	 * the message.
	 *
1271
	 * FIXME: It may be fragmented into multiple chunks
L
Linus Torvalds 已提交
1272 1273
	 *        at once if non-fragmentable extension headers
	 *        are too large.
1274
	 * --yoshfuji
L
Linus Torvalds 已提交
1275 1276
	 */

1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288
	skb = skb_peek_tail(&sk->sk_write_queue);
	cork->length += length;
	if (((length > mtu) ||
	     (skb && skb_is_gso(skb))) &&
	    (sk->sk_protocol == IPPROTO_UDP) &&
	    (rt->dst.dev->features & NETIF_F_UFO)) {
		err = ip6_ufo_append_data(sk, getfrag, from, length,
					  hh_len, fragheaderlen,
					  transhdrlen, mtu, flags, rt);
		if (err)
			goto error;
		return 0;
1289
	}
L
Linus Torvalds 已提交
1290

1291
	if (!skb)
L
Linus Torvalds 已提交
1292 1293 1294 1295
		goto alloc_new_skb;

	while (length > 0) {
		/* Check if the remaining data fits into current packet. */
1296
		copy = (cork->length <= mtu && !(cork->flags & IPCORK_ALLFRAG) ? mtu : maxfraglen) - skb->len;
L
Linus Torvalds 已提交
1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307
		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
1308 1309
			if (skb)
				fraggap = skb->len - maxfraglen;
L
Linus Torvalds 已提交
1310 1311
			else
				fraggap = 0;
G
Gao feng 已提交
1312 1313 1314
			/* update mtu and maxfraglen if necessary */
			if (skb == NULL || skb_prev == NULL)
				ip6_append_data_mtu(&mtu, &maxfraglen,
1315
						    fragheaderlen, skb, rt,
1316
						    np->pmtudisc >=
1317
						    IPV6_PMTUDISC_PROBE);
G
Gao feng 已提交
1318 1319

			skb_prev = skb;
L
Linus Torvalds 已提交
1320 1321 1322 1323 1324 1325 1326

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

G
Gao feng 已提交
1327 1328
			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
1329
			if ((flags & MSG_MORE) &&
1330
			    !(rt->dst.dev->features&NETIF_F_SG))
L
Linus Torvalds 已提交
1331 1332 1333 1334
				alloclen = mtu;
			else
				alloclen = datalen + fragheaderlen;

1335 1336
			alloclen += dst_exthdrlen;

G
Gao feng 已提交
1337 1338 1339 1340 1341 1342 1343 1344 1345 1346
			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
1347 1348 1349

			/*
			 * We just reserve space for fragment header.
1350
			 * Note: this may be overallocation if the message
L
Linus Torvalds 已提交
1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367
			 * (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);
				if (unlikely(skb == NULL))
					err = -ENOBUFS;
1368 1369 1370 1371 1372 1373
				else {
					/* Only the initial fragment
					 * is time stamped.
					 */
					tx_flags = 0;
				}
L
Linus Torvalds 已提交
1374 1375 1376 1377 1378 1379
			}
			if (skb == NULL)
				goto error;
			/*
			 *	Fill in the control structures
			 */
1380
			skb->protocol = htons(ETH_P_IPV6);
1381
			skb->ip_summed = CHECKSUM_NONE;
L
Linus Torvalds 已提交
1382
			skb->csum = 0;
1383 1384 1385
			/* reserve for fragmentation and ipsec header */
			skb_reserve(skb, hh_len + sizeof(struct frag_hdr) +
				    dst_exthdrlen);
L
Linus Torvalds 已提交
1386

1387 1388 1389
			if (sk->sk_type == SOCK_DGRAM)
				skb_shinfo(skb)->tx_flags = tx_flags;

L
Linus Torvalds 已提交
1390 1391 1392
			/*
			 *	Find where to start putting bytes
			 */
1393 1394 1395
			data = skb_put(skb, fraglen);
			skb_set_network_header(skb, exthdrlen);
			data += fragheaderlen;
1396 1397
			skb->transport_header = (skb->network_header +
						 fragheaderlen);
L
Linus Torvalds 已提交
1398 1399 1400 1401 1402 1403 1404
			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;
1405
				pskb_trim_unique(skb_prev, maxfraglen);
L
Linus Torvalds 已提交
1406 1407
			}
			copy = datalen - transhdrlen - fraggap;
1408

L
Linus Torvalds 已提交
1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422
			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;
1423
			dst_exthdrlen = 0;
L
Linus Torvalds 已提交
1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434

			/*
			 * Put the packet on the pending queue
			 */
			__skb_queue_tail(&sk->sk_write_queue, skb);
			continue;
		}

		if (copy > length)
			copy = length;

1435
		if (!(rt->dst.dev->features&NETIF_F_SG)) {
L
Linus Torvalds 已提交
1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446
			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;
1447
			struct page_frag *pfrag = sk_page_frag(sk);
L
Linus Torvalds 已提交
1448

1449 1450
			err = -ENOMEM;
			if (!sk_page_frag_refill(sk, pfrag))
L
Linus Torvalds 已提交
1451
				goto error;
1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462

			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
1463
			}
1464
			copy = min_t(int, copy, pfrag->size - pfrag->offset);
E
Eric Dumazet 已提交
1465
			if (getfrag(from,
1466 1467 1468 1469 1470 1471
				    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
1472 1473
			skb->len += copy;
			skb->data_len += copy;
1474 1475
			skb->truesize += copy;
			atomic_add(copy, &sk->sk_wmem_alloc);
L
Linus Torvalds 已提交
1476 1477 1478 1479
		}
		offset += copy;
		length -= copy;
	}
1480

L
Linus Torvalds 已提交
1481
	return 0;
1482 1483 1484

error_efault:
	err = -EFAULT;
L
Linus Torvalds 已提交
1485
error:
1486
	cork->length -= length;
1487
	IP6_INC_STATS(sock_net(sk), rt->rt6i_idev, IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
1488 1489
	return err;
}
1490
EXPORT_SYMBOL_GPL(ip6_append_data);
L
Linus Torvalds 已提交
1491

1492 1493
static void ip6_cork_release(struct inet_sock *inet, struct ipv6_pinfo *np)
{
1494 1495 1496 1497 1498 1499 1500 1501 1502
	if (np->cork.opt) {
		kfree(np->cork.opt->dst0opt);
		kfree(np->cork.opt->dst1opt);
		kfree(np->cork.opt->hopopt);
		kfree(np->cork.opt->srcrt);
		kfree(np->cork.opt);
		np->cork.opt = NULL;
	}

1503 1504 1505 1506
	if (inet->cork.base.dst) {
		dst_release(inet->cork.base.dst);
		inet->cork.base.dst = NULL;
		inet->cork.base.flags &= ~IPCORK_ALLFRAG;
1507 1508 1509 1510
	}
	memset(&inet->cork.fl, 0, sizeof(inet->cork.fl));
}

L
Linus Torvalds 已提交
1511 1512 1513 1514 1515 1516 1517
int ip6_push_pending_frames(struct sock *sk)
{
	struct sk_buff *skb, *tmp_skb;
	struct sk_buff **tail_skb;
	struct in6_addr final_dst_buf, *final_dst = &final_dst_buf;
	struct inet_sock *inet = inet_sk(sk);
	struct ipv6_pinfo *np = inet6_sk(sk);
1518
	struct net *net = sock_net(sk);
L
Linus Torvalds 已提交
1519 1520
	struct ipv6hdr *hdr;
	struct ipv6_txoptions *opt = np->cork.opt;
1521
	struct rt6_info *rt = (struct rt6_info *)inet->cork.base.dst;
1522 1523
	struct flowi6 *fl6 = &inet->cork.fl.u.ip6;
	unsigned char proto = fl6->flowi6_proto;
L
Linus Torvalds 已提交
1524 1525 1526 1527 1528 1529 1530
	int err = 0;

	if ((skb = __skb_dequeue(&sk->sk_write_queue)) == NULL)
		goto out;
	tail_skb = &(skb_shinfo(skb)->frag_list);

	/* move skb->data to ip header from ext header */
1531
	if (skb->data < skb_network_header(skb))
1532
		__skb_pull(skb, skb_network_offset(skb));
L
Linus Torvalds 已提交
1533
	while ((tmp_skb = __skb_dequeue(&sk->sk_write_queue)) != NULL) {
1534
		__skb_pull(tmp_skb, skb_network_header_len(skb));
L
Linus Torvalds 已提交
1535 1536 1537 1538 1539 1540 1541 1542 1543
		*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;
	}

1544
	/* Allow local fragmentation. */
1545
	skb->local_df = ip6_sk_local_df(sk);
1546

A
Alexey Dobriyan 已提交
1547
	*final_dst = fl6->daddr;
1548
	__skb_pull(skb, skb_network_header_len(skb));
L
Linus Torvalds 已提交
1549 1550 1551 1552 1553
	if (opt && opt->opt_flen)
		ipv6_push_frag_opts(skb, opt, &proto);
	if (opt && opt->opt_nflen)
		ipv6_push_nfrag_opts(skb, opt, &proto, &final_dst);

1554 1555
	skb_push(skb, sizeof(struct ipv6hdr));
	skb_reset_network_header(skb);
1556
	hdr = ipv6_hdr(skb);
1557

1558
	ip6_flow_hdr(hdr, np->cork.tclass, fl6->flowlabel);
L
Linus Torvalds 已提交
1559 1560
	hdr->hop_limit = np->cork.hop_limit;
	hdr->nexthdr = proto;
A
Alexey Dobriyan 已提交
1561 1562
	hdr->saddr = fl6->saddr;
	hdr->daddr = *final_dst;
L
Linus Torvalds 已提交
1563

1564
	skb->priority = sk->sk_priority;
1565
	skb->mark = sk->sk_mark;
1566

1567
	skb_dst_set(skb, dst_clone(&rt->dst));
1568
	IP6_UPD_PO_STATS(net, rt->rt6i_idev, IPSTATS_MIB_OUT, skb->len);
1569
	if (proto == IPPROTO_ICMPV6) {
E
Eric Dumazet 已提交
1570
		struct inet6_dev *idev = ip6_dst_idev(skb_dst(skb));
1571

1572
		ICMP6MSGOUT_INC_STATS_BH(net, idev, icmp6_hdr(skb)->icmp6_type);
1573
		ICMP6_INC_STATS_BH(net, idev, ICMP6_MIB_OUTMSGS);
1574 1575
	}

H
Herbert Xu 已提交
1576
	err = ip6_local_out(skb);
L
Linus Torvalds 已提交
1577 1578
	if (err) {
		if (err > 0)
E
Eric Dumazet 已提交
1579
			err = net_xmit_errno(err);
L
Linus Torvalds 已提交
1580 1581 1582 1583 1584
		if (err)
			goto error;
	}

out:
1585
	ip6_cork_release(inet, np);
L
Linus Torvalds 已提交
1586 1587
	return err;
error:
1588
	IP6_INC_STATS(net, rt->rt6i_idev, IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
1589 1590
	goto out;
}
1591
EXPORT_SYMBOL_GPL(ip6_push_pending_frames);
L
Linus Torvalds 已提交
1592 1593 1594 1595 1596 1597

void ip6_flush_pending_frames(struct sock *sk)
{
	struct sk_buff *skb;

	while ((skb = __skb_dequeue_tail(&sk->sk_write_queue)) != NULL) {
E
Eric Dumazet 已提交
1598 1599
		if (skb_dst(skb))
			IP6_INC_STATS(sock_net(sk), ip6_dst_idev(skb_dst(skb)),
1600
				      IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
1601 1602 1603
		kfree_skb(skb);
	}

1604
	ip6_cork_release(inet_sk(sk), inet6_sk(sk));
L
Linus Torvalds 已提交
1605
}
1606
EXPORT_SYMBOL_GPL(ip6_flush_pending_frames);