ip6_output.c 40.8 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>
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);
}

135
int ip6_output(struct sock *sk, struct sk_buff *skb)
L
Linus Torvalds 已提交
136
{
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

208 209
	ip6_flow_hdr(hdr, tclass, ip6_make_flowlabel(net, skb, fl6->flowlabel,
						     np->autoflowlabel));
210

L
Linus Torvalds 已提交
211 212 213 214
	hdr->payload_len = htons(seg_len);
	hdr->nexthdr = proto;
	hdr->hop_limit = hlimit;

A
Alexey Dobriyan 已提交
215 216
	hdr->saddr = fl6->saddr;
	hdr->daddr = *first_hop;
L
Linus Torvalds 已提交
217

218
	skb->protocol = htons(ETH_P_IPV6);
219
	skb->priority = sk->sk_priority;
220
	skb->mark = sk->sk_mark;
221

L
Linus Torvalds 已提交
222
	mtu = dst_mtu(dst);
W
WANG Cong 已提交
223
	if ((skb->len <= mtu) || skb->ignore_df || skb_is_gso(skb)) {
E
Eric Dumazet 已提交
224
		IP6_UPD_PO_STATS(net, ip6_dst_idev(skb_dst(skb)),
225
			      IPSTATS_MIB_OUT, skb->len);
226 227
		return NF_HOOK(NFPROTO_IPV6, NF_INET_LOCAL_OUT, skb, NULL,
			       dst->dev, dst_output);
L
Linus Torvalds 已提交
228 229 230
	}

	skb->dev = dst->dev;
231
	ipv6_local_error(sk, EMSGSIZE, fl6, mtu);
E
Eric Dumazet 已提交
232
	IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)), IPSTATS_MIB_FRAGFAILS);
L
Linus Torvalds 已提交
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
static bool ip6_pkt_too_big(const struct sk_buff *skb, unsigned int mtu)
{
347
	if (skb->len <= mtu)
348 349
		return false;

W
WANG Cong 已提交
350
	/* ipv6 conntrack defrag sets max_frag_size + ignore_df */
351 352 353
	if (IP6CB(skb)->frag_max_size && IP6CB(skb)->frag_max_size > mtu)
		return true;

W
WANG Cong 已提交
354
	if (skb->ignore_df)
355 356
		return false;

357 358 359 360 361 362
	if (skb_is_gso(skb) && skb_gso_network_seglen(skb) <= mtu)
		return false;

	return true;
}

L
Linus Torvalds 已提交
363 364
int ip6_forward(struct sk_buff *skb)
{
E
Eric Dumazet 已提交
365
	struct dst_entry *dst = skb_dst(skb);
366
	struct ipv6hdr *hdr = ipv6_hdr(skb);
L
Linus Torvalds 已提交
367
	struct inet6_skb_parm *opt = IP6CB(skb);
368
	struct net *net = dev_net(dst->dev);
369
	u32 mtu;
370

371
	if (net->ipv6.devconf_all->forwarding == 0)
L
Linus Torvalds 已提交
372 373
		goto error;

374 375 376
	if (skb->pkt_type != PACKET_HOST)
		goto drop;

377 378 379
	if (skb_warn_if_lro(skb))
		goto drop;

L
Linus Torvalds 已提交
380
	if (!xfrm6_policy_check(NULL, XFRM_POLICY_FWD, skb)) {
381 382
		IP6_INC_STATS_BH(net, ip6_dst_idev(dst),
				 IPSTATS_MIB_INDISCARDS);
L
Linus Torvalds 已提交
383 384 385
		goto drop;
	}

386
	skb_forward_csum(skb);
L
Linus Torvalds 已提交
387 388 389 390 391 392 393 394 395 396 397 398 399 400

	/*
	 *	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
	 */
401 402
	if (unlikely(opt->flags & IP6SKB_ROUTERALERT)) {
		if (ip6_call_ra_chain(skb, ntohs(opt->ra)))
L
Linus Torvalds 已提交
403 404 405 406 407 408 409 410 411
			return 0;
	}

	/*
	 *	check and decrement ttl
	 */
	if (hdr->hop_limit <= 1) {
		/* Force OUTPUT device used as source address */
		skb->dev = dst->dev;
412
		icmpv6_send(skb, ICMPV6_TIME_EXCEED, ICMPV6_EXC_HOPLIMIT, 0);
413 414
		IP6_INC_STATS_BH(net, ip6_dst_idev(dst),
				 IPSTATS_MIB_INHDRERRORS);
L
Linus Torvalds 已提交
415 416 417 418 419

		kfree_skb(skb);
		return -ETIMEDOUT;
	}

420
	/* XXX: idev->cnf.proxy_ndp? */
421
	if (net->ipv6.devconf_all->proxy_ndp &&
422
	    pneigh_lookup(&nd_tbl, net, &hdr->daddr, skb->dev, 0)) {
423 424
		int proxied = ip6_forward_proxy_check(skb);
		if (proxied > 0)
425
			return ip6_input(skb);
426
		else if (proxied < 0) {
427 428
			IP6_INC_STATS_BH(net, ip6_dst_idev(dst),
					 IPSTATS_MIB_INDISCARDS);
429 430
			goto drop;
		}
431 432
	}

L
Linus Torvalds 已提交
433
	if (!xfrm6_route_forward(skb)) {
434 435
		IP6_INC_STATS_BH(net, ip6_dst_idev(dst),
				 IPSTATS_MIB_INDISCARDS);
L
Linus Torvalds 已提交
436 437
		goto drop;
	}
E
Eric Dumazet 已提交
438
	dst = skb_dst(skb);
L
Linus Torvalds 已提交
439 440 441

	/* IPv6 specs say nothing about it, but it is clear that we cannot
	   send redirects to source routed frames.
442
	   We don't send redirects to frames decapsulated from IPsec.
L
Linus Torvalds 已提交
443
	 */
444
	if (skb->dev == dst->dev && opt->srcrt == 0 && !skb_sec_path(skb)) {
L
Linus Torvalds 已提交
445
		struct in6_addr *target = NULL;
446
		struct inet_peer *peer;
L
Linus Torvalds 已提交
447 448 449 450 451 452 453 454
		struct rt6_info *rt;

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

		rt = (struct rt6_info *) dst;
455 456
		if (rt->rt6i_flags & RTF_GATEWAY)
			target = &rt->rt6i_gateway;
L
Linus Torvalds 已提交
457 458 459
		else
			target = &hdr->daddr;

460
		peer = inet_getpeer_v6(net->ipv6.peers, &rt->rt6i_dst.addr, 1);
461

L
Linus Torvalds 已提交
462 463 464
		/* Limit redirects both by destination (here)
		   and by source (inside ndisc_send_redirect)
		 */
465
		if (inet_peer_xrlim_allow(peer, 1*HZ))
466
			ndisc_send_redirect(skb, target);
467 468
		if (peer)
			inet_putpeer(peer);
469 470 471
	} else {
		int addrtype = ipv6_addr_type(&hdr->saddr);

L
Linus Torvalds 已提交
472
		/* This check is security critical. */
473 474
		if (addrtype == IPV6_ADDR_ANY ||
		    addrtype & (IPV6_ADDR_MULTICAST | IPV6_ADDR_LOOPBACK))
475 476 477
			goto error;
		if (addrtype & IPV6_ADDR_LINKLOCAL) {
			icmpv6_send(skb, ICMPV6_DEST_UNREACH,
478
				    ICMPV6_NOT_NEIGHBOUR, 0);
479 480
			goto error;
		}
L
Linus Torvalds 已提交
481 482
	}

483
	mtu = ip6_dst_mtu_forward(dst);
484 485 486
	if (mtu < IPV6_MIN_MTU)
		mtu = IPV6_MIN_MTU;

487
	if (ip6_pkt_too_big(skb, mtu)) {
L
Linus Torvalds 已提交
488 489
		/* Again, force OUTPUT device used as source address */
		skb->dev = dst->dev;
490
		icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
491 492 493 494
		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
495 496 497 498 499
		kfree_skb(skb);
		return -EMSGSIZE;
	}

	if (skb_cow(skb, dst->dev->hard_header_len)) {
500 501
		IP6_INC_STATS_BH(net, ip6_dst_idev(dst),
				 IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
502 503 504
		goto drop;
	}

505
	hdr = ipv6_hdr(skb);
L
Linus Torvalds 已提交
506 507

	/* Mangling hops number delayed to point after skb COW */
508

L
Linus Torvalds 已提交
509 510
	hdr->hop_limit--;

511
	IP6_INC_STATS_BH(net, ip6_dst_idev(dst), IPSTATS_MIB_OUTFORWDATAGRAMS);
512
	IP6_ADD_STATS_BH(net, ip6_dst_idev(dst), IPSTATS_MIB_OUTOCTETS, skb->len);
513
	return NF_HOOK(NFPROTO_IPV6, NF_INET_FORWARD, skb, skb->dev, dst->dev,
514
		       ip6_forward_finish);
L
Linus Torvalds 已提交
515 516

error:
517
	IP6_INC_STATS_BH(net, ip6_dst_idev(dst), IPSTATS_MIB_INADDRERRORS);
L
Linus Torvalds 已提交
518 519 520 521 522 523 524 525 526 527
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
528 529
	skb_dst_drop(to);
	skb_dst_set(to, dst_clone(skb_dst(from)));
L
Linus Torvalds 已提交
530
	to->dev = from->dev;
T
Thomas Graf 已提交
531
	to->mark = from->mark;
L
Linus Torvalds 已提交
532 533 534 535

#ifdef CONFIG_NET_SCHED
	to->tc_index = from->tc_index;
#endif
536
	nf_copy(to, from);
537
	skb_copy_secmark(to, from);
L
Linus Torvalds 已提交
538 539
}

E
Eric Dumazet 已提交
540 541 542 543 544 545 546 547
static void ipv6_select_ident(struct frag_hdr *fhdr, struct rt6_info *rt)
{
	static u32 ip6_idents_hashrnd __read_mostly;
	u32 hash, id;

	net_get_random_once(&ip6_idents_hashrnd, sizeof(ip6_idents_hashrnd));

	hash = __ipv6_addr_jhash(&rt->rt6i_dst.addr, ip6_idents_hashrnd);
548 549
	hash = __ipv6_addr_jhash(&rt->rt6i_src.addr, hash);

E
Eric Dumazet 已提交
550 551 552 553
	id = ip_idents_reserve(hash, 1);
	fhdr->identification = htonl(id);
}

554
int ip6_fragment(struct sk_buff *skb, int (*output)(struct sk_buff *))
L
Linus Torvalds 已提交
555 556
{
	struct sk_buff *frag;
557
	struct rt6_info *rt = (struct rt6_info *)skb_dst(skb);
558
	struct ipv6_pinfo *np = skb->sk ? inet6_sk(skb->sk) : NULL;
L
Linus Torvalds 已提交
559 560 561
	struct ipv6hdr *tmp_hdr;
	struct frag_hdr *fh;
	unsigned int mtu, hlen, left, len;
562
	int hroom, troom;
A
Al Viro 已提交
563
	__be32 frag_id = 0;
564
	int ptr, offset = 0, err = 0;
L
Linus Torvalds 已提交
565
	u8 *prevhdr, nexthdr = 0;
E
Eric Dumazet 已提交
566
	struct net *net = dev_net(skb_dst(skb)->dev);
L
Linus Torvalds 已提交
567 568 569 570

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

J
John Heffner 已提交
571
	mtu = ip6_skb_dst_mtu(skb);
572 573

	/* We must not fragment if the socket is set to force MTU discovery
574
	 * or if the skb it not generated by a local socket.
575
	 */
W
WANG Cong 已提交
576
	if (unlikely(!skb->ignore_df && skb->len > mtu) ||
577 578
		     (IP6CB(skb)->frag_max_size &&
		      IP6CB(skb)->frag_max_size > mtu)) {
579 580 581
		if (skb->sk && dst_allfrag(skb_dst(skb)))
			sk_nocaps_add(skb->sk, NETIF_F_GSO_MASK);

E
Eric Dumazet 已提交
582
		skb->dev = skb_dst(skb)->dev;
583
		icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
E
Eric Dumazet 已提交
584
		IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
585
			      IPSTATS_MIB_FRAGFAILS);
586 587 588 589
		kfree_skb(skb);
		return -EMSGSIZE;
	}

590 591 592 593 594
	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
595

596
	if (skb_has_frag_list(skb)) {
L
Linus Torvalds 已提交
597
		int first_len = skb_pagelen(skb);
598
		struct sk_buff *frag2;
L
Linus Torvalds 已提交
599 600 601 602 603 604

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

605
		skb_walk_frags(skb, frag) {
L
Linus Torvalds 已提交
606 607 608 609
			/* Correct geometry. */
			if (frag->len > mtu ||
			    ((frag->len & 7) && frag->next) ||
			    skb_headroom(frag) < hlen)
610
				goto slow_path_clean;
L
Linus Torvalds 已提交
611 612 613

			/* Partially cloned skb? */
			if (skb_shared(frag))
614
				goto slow_path_clean;
615 616 617 618 619 620

			BUG_ON(frag->sk);
			if (skb->sk) {
				frag->sk = skb->sk;
				frag->destructor = sock_wfree;
			}
621
			skb->truesize -= frag->truesize;
L
Linus Torvalds 已提交
622 623 624 625 626
		}

		err = 0;
		offset = 0;
		frag = skb_shinfo(skb)->frag_list;
627
		skb_frag_list_init(skb);
L
Linus Torvalds 已提交
628 629
		/* BUILD HEADER */

Y
YOSHIFUJI Hideaki 已提交
630
		*prevhdr = NEXTHDR_FRAGMENT;
631
		tmp_hdr = kmemdup(skb_network_header(skb), hlen, GFP_ATOMIC);
L
Linus Torvalds 已提交
632
		if (!tmp_hdr) {
E
Eric Dumazet 已提交
633
			IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
634
				      IPSTATS_MIB_FRAGFAILS);
L
Linus Torvalds 已提交
635 636 637 638
			return -ENOMEM;
		}

		__skb_pull(skb, hlen);
639
		fh = (struct frag_hdr *)__skb_push(skb, sizeof(struct frag_hdr));
640 641
		__skb_push(skb, hlen);
		skb_reset_network_header(skb);
642
		memcpy(skb_network_header(skb), tmp_hdr, hlen);
L
Linus Torvalds 已提交
643

644
		ipv6_select_ident(fh, rt);
L
Linus Torvalds 已提交
645 646 647 648 649 650 651 652
		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;
653 654
		ipv6_hdr(skb)->payload_len = htons(first_len -
						   sizeof(struct ipv6hdr));
655

656
		dst_hold(&rt->dst);
L
Linus Torvalds 已提交
657 658 659 660 661 662

		for (;;) {
			/* Prepare header of the next frame,
			 * before previous one went down. */
			if (frag) {
				frag->ip_summed = CHECKSUM_NONE;
663
				skb_reset_transport_header(frag);
664
				fh = (struct frag_hdr *)__skb_push(frag, sizeof(struct frag_hdr));
665 666
				__skb_push(frag, hlen);
				skb_reset_network_header(frag);
667 668
				memcpy(skb_network_header(frag), tmp_hdr,
				       hlen);
L
Linus Torvalds 已提交
669 670 671 672 673 674 675
				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;
676 677 678
				ipv6_hdr(frag)->payload_len =
						htons(frag->len -
						      sizeof(struct ipv6hdr));
L
Linus Torvalds 已提交
679 680
				ip6_copy_metadata(frag, skb);
			}
681

L
Linus Torvalds 已提交
682
			err = output(skb);
683
			if (!err)
684
				IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
685
					      IPSTATS_MIB_FRAGCREATES);
686

L
Linus Torvalds 已提交
687 688 689 690 691 692 693 694
			if (err || !frag)
				break;

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

J
Jesper Juhl 已提交
695
		kfree(tmp_hdr);
L
Linus Torvalds 已提交
696 697

		if (err == 0) {
698
			IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
699
				      IPSTATS_MIB_FRAGOKS);
A
Amerigo Wang 已提交
700
			ip6_rt_put(rt);
L
Linus Torvalds 已提交
701 702 703 704 705 706 707 708 709
			return 0;
		}

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

710
		IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
711
			      IPSTATS_MIB_FRAGFAILS);
A
Amerigo Wang 已提交
712
		ip6_rt_put(rt);
L
Linus Torvalds 已提交
713
		return err;
714 715 716 717 718 719 720 721 722

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
723 724 725
	}

slow_path:
726 727 728 729
	if ((skb->ip_summed == CHECKSUM_PARTIAL) &&
	    skb_checksum_help(skb))
		goto fail;

L
Linus Torvalds 已提交
730 731 732 733 734 735 736 737
	left = skb->len - hlen;		/* Space per frame */
	ptr = hlen;			/* Where to start from */

	/*
	 *	Fragment the datagram.
	 */

	*prevhdr = NEXTHDR_FRAGMENT;
738 739
	hroom = LL_RESERVED_SPACE(rt->dst.dev);
	troom = rt->dst.dev->needed_tailroom;
L
Linus Torvalds 已提交
740 741 742 743

	/*
	 *	Keep copying data until we run out.
	 */
744
	while (left > 0)	{
L
Linus Torvalds 已提交
745 746 747 748
		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
749
		/* IF: we are not sending up to and including the packet end
L
Linus Torvalds 已提交
750 751 752 753 754 755 756 757
		   then align the next start on an eight byte boundary */
		if (len < left)	{
			len &= ~7;
		}
		/*
		 *	Allocate buffer.
		 */

758 759
		if ((frag = alloc_skb(len + hlen + sizeof(struct frag_hdr) +
				      hroom + troom, GFP_ATOMIC)) == NULL) {
760
			NETDEBUG(KERN_INFO "IPv6: frag: no memory for new fragment!\n");
E
Eric Dumazet 已提交
761
			IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
762
				      IPSTATS_MIB_FRAGFAILS);
L
Linus Torvalds 已提交
763 764 765 766 767 768 769 770 771
			err = -ENOMEM;
			goto fail;
		}

		/*
		 *	Set up data on packet
		 */

		ip6_copy_metadata(frag, skb);
772
		skb_reserve(frag, hroom);
L
Linus Torvalds 已提交
773
		skb_put(frag, len + hlen + sizeof(struct frag_hdr));
774
		skb_reset_network_header(frag);
775
		fh = (struct frag_hdr *)(skb_network_header(frag) + hlen);
776 777
		frag->transport_header = (frag->network_header + hlen +
					  sizeof(struct frag_hdr));
L
Linus Torvalds 已提交
778 779 780 781 782 783 784 785 786 787 788

		/*
		 *	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.
		 */
789
		skb_copy_from_linear_data(skb, skb_network_header(frag), hlen);
L
Linus Torvalds 已提交
790 791 792 793 794 795

		/*
		 *	Build fragment header.
		 */
		fh->nexthdr = nexthdr;
		fh->reserved = 0;
796
		if (!frag_id) {
797
			ipv6_select_ident(fh, rt);
L
Linus Torvalds 已提交
798 799 800 801 802 803 804
			frag_id = fh->identification;
		} else
			fh->identification = frag_id;

		/*
		 *	Copy a block of the IP datagram.
		 */
H
Himangi Saraogi 已提交
805 806
		BUG_ON(skb_copy_bits(skb, ptr, skb_transport_header(frag),
				     len));
L
Linus Torvalds 已提交
807 808 809 810 811
		left -= len;

		fh->frag_off = htons(offset);
		if (left > 0)
			fh->frag_off |= htons(IP6_MF);
812 813
		ipv6_hdr(frag)->payload_len = htons(frag->len -
						    sizeof(struct ipv6hdr));
L
Linus Torvalds 已提交
814 815 816 817 818 819 820 821 822 823

		ptr += len;
		offset += len;

		/*
		 *	Put this fragment into the sending queue.
		 */
		err = output(frag);
		if (err)
			goto fail;
824

E
Eric Dumazet 已提交
825
		IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
826
			      IPSTATS_MIB_FRAGCREATES);
L
Linus Torvalds 已提交
827
	}
E
Eric Dumazet 已提交
828
	IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
829
		      IPSTATS_MIB_FRAGOKS);
830
	consume_skb(skb);
L
Linus Torvalds 已提交
831 832 833
	return err;

fail:
E
Eric Dumazet 已提交
834
	IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
835
		      IPSTATS_MIB_FRAGFAILS);
836
	kfree_skb(skb);
L
Linus Torvalds 已提交
837 838 839
	return err;
}

840 841 842
static inline int ip6_rt_check(const struct rt6key *rt_key,
			       const struct in6_addr *fl_addr,
			       const struct in6_addr *addr_cache)
843
{
E
Eric Dumazet 已提交
844 845
	return (rt_key->plen != 128 || !ipv6_addr_equal(fl_addr, &rt_key->addr)) &&
		(addr_cache == NULL || !ipv6_addr_equal(fl_addr, addr_cache));
846 847
}

848 849
static struct dst_entry *ip6_sk_dst_check(struct sock *sk,
					  struct dst_entry *dst,
850
					  const struct flowi6 *fl6)
L
Linus Torvalds 已提交
851
{
852
	struct ipv6_pinfo *np = inet6_sk(sk);
853
	struct rt6_info *rt;
L
Linus Torvalds 已提交
854

855 856 857
	if (!dst)
		goto out;

858 859 860 861 862 863
	if (dst->ops->family != AF_INET6) {
		dst_release(dst);
		return NULL;
	}

	rt = (struct rt6_info *)dst;
864 865 866
	/* Yes, checking route validity in not connected
	 * case is not very simple. Take into account,
	 * that we do not support routing by source, TOS,
867
	 * and MSG_DONTROUTE		--ANK (980726)
868
	 *
869 870
	 * 1. ip6_rt_check(): If route was host route,
	 *    check that cached destination is current.
871 872 873 874 875 876 877 878 879 880
	 *    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.
	 */
881
	if (ip6_rt_check(&rt->rt6i_dst, &fl6->daddr, np->daddr_cache) ||
882
#ifdef CONFIG_IPV6_SUBTREES
883
	    ip6_rt_check(&rt->rt6i_src, &fl6->saddr, np->saddr_cache) ||
884
#endif
885
	    (fl6->flowi6_oif && fl6->flowi6_oif != dst->dev->ifindex)) {
886 887
		dst_release(dst);
		dst = NULL;
L
Linus Torvalds 已提交
888 889
	}

890 891 892 893 894
out:
	return dst;
}

static int ip6_dst_lookup_tail(struct sock *sk,
895
			       struct dst_entry **dst, struct flowi6 *fl6)
896
{
897
	struct net *net = sock_net(sk);
898 899
#ifdef CONFIG_IPV6_OPTIMISTIC_DAD
	struct neighbour *n;
900
	struct rt6_info *rt;
901 902
#endif
	int err;
903

L
Linus Torvalds 已提交
904
	if (*dst == NULL)
905
		*dst = ip6_route_output(net, sk, fl6);
L
Linus Torvalds 已提交
906 907 908 909

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

910
	if (ipv6_addr_any(&fl6->saddr)) {
911 912 913 914
		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);
915
		if (err)
L
Linus Torvalds 已提交
916 917 918
			goto out_err_release;
	}

919
#ifdef CONFIG_IPV6_OPTIMISTIC_DAD
920 921 922 923 924 925 926 927
	/*
	 * 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
	 */
928
	rt = (struct rt6_info *) *dst;
929
	rcu_read_lock_bh();
930
	n = __ipv6_neigh_lookup_noref(rt->dst.dev, rt6_nexthop(rt));
931 932 933 934
	err = n && !(n->nud_state & NUD_VALID) ? -EINVAL : 0;
	rcu_read_unlock_bh();

	if (err) {
935
		struct inet6_ifaddr *ifp;
936
		struct flowi6 fl_gw6;
937 938
		int redirect;

939
		ifp = ipv6_get_ifaddr(net, &fl6->saddr,
940 941 942 943 944 945 946 947 948 949 950 951
				      (*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);
952 953 954
			memcpy(&fl_gw6, fl6, sizeof(struct flowi6));
			memset(&fl_gw6.daddr, 0, sizeof(struct in6_addr));
			*dst = ip6_route_output(net, sk, &fl_gw6);
955 956
			if ((err = (*dst)->error))
				goto out_err_release;
957
		}
958
	}
959 960
#endif

L
Linus Torvalds 已提交
961 962 963
	return 0;

out_err_release:
964
	if (err == -ENETUNREACH)
965
		IP6_INC_STATS(net, NULL, IPSTATS_MIB_OUTNOROUTES);
L
Linus Torvalds 已提交
966 967 968 969
	dst_release(*dst);
	*dst = NULL;
	return err;
}
A
Adrian Bunk 已提交
970

971 972 973 974
/**
 *	ip6_dst_lookup - perform route lookup on flow
 *	@sk: socket which provides route info
 *	@dst: pointer to dst_entry * for result
975
 *	@fl6: flow to lookup
976 977 978 979 980
 *
 *	This function performs a route lookup on the given flow.
 *
 *	It returns zero on success, or a standard errno code on error.
 */
981
int ip6_dst_lookup(struct sock *sk, struct dst_entry **dst, struct flowi6 *fl6)
982 983
{
	*dst = NULL;
984
	return ip6_dst_lookup_tail(sk, dst, fl6);
985
}
986 987
EXPORT_SYMBOL_GPL(ip6_dst_lookup);

988
/**
989 990
 *	ip6_dst_lookup_flow - perform route lookup on flow with ipsec
 *	@sk: socket which provides route info
991
 *	@fl6: flow to lookup
992 993 994 995 996 997 998
 *	@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.
 */
999
struct dst_entry *ip6_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
1000
				      const struct in6_addr *final_dst)
1001 1002 1003 1004
{
	struct dst_entry *dst = NULL;
	int err;

1005
	err = ip6_dst_lookup_tail(sk, &dst, fl6);
1006 1007 1008
	if (err)
		return ERR_PTR(err);
	if (final_dst)
A
Alexey Dobriyan 已提交
1009
		fl6->daddr = *final_dst;
1010

1011
	return xfrm_lookup(sock_net(sk), dst, flowi6_to_flowi(fl6), sk, 0);
1012 1013 1014 1015 1016
}
EXPORT_SYMBOL_GPL(ip6_dst_lookup_flow);

/**
 *	ip6_sk_dst_lookup_flow - perform socket cached route lookup on flow
1017
 *	@sk: socket which provides the dst cache and route info
1018
 *	@fl6: flow to lookup
1019
 *	@final_dst: final destination address for ipsec lookup
1020 1021 1022 1023 1024 1025
 *
 *	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.
 *
1026 1027
 *	It returns a valid dst pointer on success, or a pointer encoded
 *	error code.
1028
 */
1029
struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
1030
					 const struct in6_addr *final_dst)
1031
{
1032 1033
	struct dst_entry *dst = sk_dst_check(sk, inet6_sk(sk)->dst_cookie);
	int err;
1034

1035
	dst = ip6_sk_dst_check(sk, dst, fl6);
1036

1037
	err = ip6_dst_lookup_tail(sk, &dst, fl6);
1038 1039 1040
	if (err)
		return ERR_PTR(err);
	if (final_dst)
A
Alexey Dobriyan 已提交
1041
		fl6->daddr = *final_dst;
1042

1043
	return xfrm_lookup(sock_net(sk), dst, flowi6_to_flowi(fl6), sk, 0);
1044
}
1045
EXPORT_SYMBOL_GPL(ip6_sk_dst_lookup_flow);
1046

A
Adrian Bunk 已提交
1047
static inline int ip6_ufo_append_data(struct sock *sk,
1048 1049 1050
			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,
1051
			int transhdrlen, int mtu, unsigned int flags,
1052
			struct rt6_info *rt)
1053 1054 1055

{
	struct sk_buff *skb;
1056
	struct frag_hdr fhdr;
1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067
	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)
1068
			return err;
1069 1070 1071 1072 1073

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

		/* create space for UDP/IP header */
1074
		skb_put(skb, fragheaderlen + transhdrlen);
1075 1076

		/* initialize network header pointer */
1077
		skb_reset_network_header(skb);
1078 1079

		/* initialize protocol header pointer */
1080
		skb->transport_header = skb->network_header + fragheaderlen;
1081

1082
		skb->protocol = htons(ETH_P_IPV6);
1083 1084 1085
		skb->csum = 0;

		__skb_queue_tail(&sk->sk_write_queue, skb);
1086 1087
	} else if (skb_is_gso(skb)) {
		goto append;
1088 1089
	}

1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100
	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:
1101 1102
	return skb_append_datato_frags(sk, skb, getfrag, from,
				       (length - transhdrlen));
1103
}
L
Linus Torvalds 已提交
1104

1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116
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;
}

1117
static void ip6_append_data_mtu(unsigned int *mtu,
G
Gao feng 已提交
1118 1119 1120
				int *maxfraglen,
				unsigned int fragheaderlen,
				struct sk_buff *skb,
1121
				struct rt6_info *rt,
1122
				unsigned int orig_mtu)
G
Gao feng 已提交
1123 1124 1125 1126
{
	if (!(rt->dst.flags & DST_XFRM_TUNNEL)) {
		if (skb == NULL) {
			/* first fragment, reserve header_len */
1127
			*mtu = orig_mtu - rt->dst.header_len;
G
Gao feng 已提交
1128 1129 1130 1131 1132 1133

		} else {
			/*
			 * this fragment is not first, the headers
			 * space is regarded as data space.
			 */
1134
			*mtu = orig_mtu;
G
Gao feng 已提交
1135 1136 1137 1138 1139 1140
		}
		*maxfraglen = ((*mtu - fragheaderlen) & ~7)
			      + fragheaderlen - sizeof(struct frag_hdr);
	}
}

1141 1142 1143
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,
1144
	int hlimit, int tclass, struct ipv6_txoptions *opt, struct flowi6 *fl6,
1145
	struct rt6_info *rt, unsigned int flags, int dontfrag)
L
Linus Torvalds 已提交
1146 1147 1148
{
	struct inet_sock *inet = inet_sk(sk);
	struct ipv6_pinfo *np = inet6_sk(sk);
1149
	struct inet_cork *cork;
G
Gao feng 已提交
1150
	struct sk_buff *skb, *skb_prev = NULL;
1151
	unsigned int maxfraglen, fragheaderlen, mtu, orig_mtu;
L
Linus Torvalds 已提交
1152
	int exthdrlen;
1153
	int dst_exthdrlen;
L
Linus Torvalds 已提交
1154 1155 1156 1157
	int hh_len;
	int copy;
	int err;
	int offset = 0;
1158
	__u8 tx_flags = 0;
1159
	u32 tskey = 0;
L
Linus Torvalds 已提交
1160 1161 1162

	if (flags&MSG_PROBE)
		return 0;
1163
	cork = &inet->cork.base;
L
Linus Torvalds 已提交
1164 1165 1166 1167 1168
	if (skb_queue_empty(&sk->sk_write_queue)) {
		/*
		 * setup for corking
		 */
		if (opt) {
1169
			if (WARN_ON(np->cork.opt))
L
Linus Torvalds 已提交
1170
				return -EINVAL;
1171

1172
			np->cork.opt = kzalloc(opt->tot_len, sk->sk_allocation);
1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199
			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
1200 1201
			/* need source address above miyazawa*/
		}
1202
		dst_hold(&rt->dst);
1203
		cork->dst = &rt->dst;
1204
		inet->cork.fl.u.ip6 = *fl6;
L
Linus Torvalds 已提交
1205
		np->cork.hop_limit = hlimit;
1206
		np->cork.tclass = tclass;
G
Gao feng 已提交
1207
		if (rt->dst.flags & DST_XFRM_TUNNEL)
1208
			mtu = np->pmtudisc >= IPV6_PMTUDISC_PROBE ?
G
Gao feng 已提交
1209 1210
			      rt->dst.dev->mtu : dst_mtu(&rt->dst);
		else
1211
			mtu = np->pmtudisc >= IPV6_PMTUDISC_PROBE ?
G
Gao feng 已提交
1212
			      rt->dst.dev->mtu : dst_mtu(rt->dst.path);
1213
		if (np->frag_size < mtu) {
1214 1215 1216
			if (np->frag_size)
				mtu = np->frag_size;
		}
1217
		cork->fragsize = mtu;
1218
		if (dst_allfrag(rt->dst.path))
1219 1220
			cork->flags |= IPCORK_ALLFRAG;
		cork->length = 0;
1221
		exthdrlen = (opt ? opt->opt_flen : 0);
L
Linus Torvalds 已提交
1222 1223
		length += exthdrlen;
		transhdrlen += exthdrlen;
1224
		dst_exthdrlen = rt->dst.header_len - rt->rt6i_nfheader_len;
L
Linus Torvalds 已提交
1225
	} else {
1226
		rt = (struct rt6_info *)cork->dst;
1227
		fl6 = &inet->cork.fl.u.ip6;
1228
		opt = np->cork.opt;
L
Linus Torvalds 已提交
1229 1230
		transhdrlen = 0;
		exthdrlen = 0;
1231
		dst_exthdrlen = 0;
1232
		mtu = cork->fragsize;
L
Linus Torvalds 已提交
1233
	}
1234
	orig_mtu = mtu;
L
Linus Torvalds 已提交
1235

1236
	hh_len = LL_RESERVED_SPACE(rt->dst.dev);
L
Linus Torvalds 已提交
1237

1238
	fragheaderlen = sizeof(struct ipv6hdr) + rt->rt6i_nfheader_len +
1239
			(opt ? opt->opt_nflen : 0);
1240 1241
	maxfraglen = ((mtu - fragheaderlen) & ~7) + fragheaderlen -
		     sizeof(struct frag_hdr);
L
Linus Torvalds 已提交
1242 1243

	if (mtu <= sizeof(struct ipv6hdr) + IPV6_MAXPLEN) {
1244 1245 1246
		unsigned int maxnonfragsize, headersize;

		headersize = sizeof(struct ipv6hdr) +
1247
			     (opt ? opt->opt_flen + opt->opt_nflen : 0) +
1248 1249 1250 1251
			     (dst_allfrag(&rt->dst) ?
			      sizeof(struct frag_hdr) : 0) +
			     rt->rt6i_nfheader_len;

W
WANG Cong 已提交
1252
		if (ip6_sk_ignore_df(sk))
1253 1254 1255
			maxnonfragsize = sizeof(struct ipv6hdr) + IPV6_MAXPLEN;
		else
			maxnonfragsize = mtu;
1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270

		/* 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
1271 1272 1273 1274
			return -EMSGSIZE;
		}
	}

1275
	if (sk->sk_type == SOCK_DGRAM || sk->sk_type == SOCK_RAW) {
1276
		sock_tx_timestamp(sk, &tx_flags);
1277 1278 1279 1280
		if (tx_flags & SKBTX_ANY_SW_TSTAMP &&
		    sk->sk_tsflags & SOF_TIMESTAMPING_OPT_ID)
			tskey = sk->sk_tskey++;
	}
1281

L
Linus Torvalds 已提交
1282 1283 1284 1285 1286 1287 1288
	/*
	 * 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
1289
	 * of the buffer to the new fragment when we split
L
Linus Torvalds 已提交
1290 1291
	 * the message.
	 *
1292
	 * FIXME: It may be fragmented into multiple chunks
L
Linus Torvalds 已提交
1293 1294
	 *        at once if non-fragmentable extension headers
	 *        are too large.
1295
	 * --yoshfuji
L
Linus Torvalds 已提交
1296 1297
	 */

1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309
	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;
1310
	}
L
Linus Torvalds 已提交
1311

1312
	if (!skb)
L
Linus Torvalds 已提交
1313 1314 1315 1316
		goto alloc_new_skb;

	while (length > 0) {
		/* Check if the remaining data fits into current packet. */
1317
		copy = (cork->length <= mtu && !(cork->flags & IPCORK_ALLFRAG) ? mtu : maxfraglen) - skb->len;
L
Linus Torvalds 已提交
1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328
		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
1329 1330
			if (skb)
				fraggap = skb->len - maxfraglen;
L
Linus Torvalds 已提交
1331 1332
			else
				fraggap = 0;
G
Gao feng 已提交
1333 1334 1335
			/* update mtu and maxfraglen if necessary */
			if (skb == NULL || skb_prev == NULL)
				ip6_append_data_mtu(&mtu, &maxfraglen,
1336
						    fragheaderlen, skb, rt,
1337
						    orig_mtu);
G
Gao feng 已提交
1338 1339

			skb_prev = skb;
L
Linus Torvalds 已提交
1340 1341 1342 1343 1344 1345 1346

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

G
Gao feng 已提交
1347 1348
			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
1349
			if ((flags & MSG_MORE) &&
1350
			    !(rt->dst.dev->features&NETIF_F_SG))
L
Linus Torvalds 已提交
1351 1352 1353 1354
				alloclen = mtu;
			else
				alloclen = datalen + fragheaderlen;

1355 1356
			alloclen += dst_exthdrlen;

G
Gao feng 已提交
1357 1358 1359 1360 1361 1362 1363 1364 1365 1366
			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
1367 1368 1369

			/*
			 * We just reserve space for fragment header.
1370
			 * Note: this may be overallocation if the message
L
Linus Torvalds 已提交
1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393
			 * (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;
			}
			if (skb == NULL)
				goto error;
			/*
			 *	Fill in the control structures
			 */
1394
			skb->protocol = htons(ETH_P_IPV6);
1395
			skb->ip_summed = CHECKSUM_NONE;
L
Linus Torvalds 已提交
1396
			skb->csum = 0;
1397 1398 1399
			/* reserve for fragmentation and ipsec header */
			skb_reserve(skb, hh_len + sizeof(struct frag_hdr) +
				    dst_exthdrlen);
L
Linus Torvalds 已提交
1400

1401 1402 1403
			/* Only the initial fragment is time stamped */
			skb_shinfo(skb)->tx_flags = tx_flags;
			tx_flags = 0;
1404 1405
			skb_shinfo(skb)->tskey = tskey;
			tskey = 0;
1406

L
Linus Torvalds 已提交
1407 1408 1409
			/*
			 *	Find where to start putting bytes
			 */
1410 1411 1412
			data = skb_put(skb, fraglen);
			skb_set_network_header(skb, exthdrlen);
			data += fragheaderlen;
1413 1414
			skb->transport_header = (skb->network_header +
						 fragheaderlen);
L
Linus Torvalds 已提交
1415 1416 1417 1418 1419 1420 1421
			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;
1422
				pskb_trim_unique(skb_prev, maxfraglen);
L
Linus Torvalds 已提交
1423 1424
			}
			copy = datalen - transhdrlen - fraggap;
1425

L
Linus Torvalds 已提交
1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439
			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;
1440
			dst_exthdrlen = 0;
L
Linus Torvalds 已提交
1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451

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

		if (copy > length)
			copy = length;

1452
		if (!(rt->dst.dev->features&NETIF_F_SG)) {
L
Linus Torvalds 已提交
1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463
			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;
1464
			struct page_frag *pfrag = sk_page_frag(sk);
L
Linus Torvalds 已提交
1465

1466 1467
			err = -ENOMEM;
			if (!sk_page_frag_refill(sk, pfrag))
L
Linus Torvalds 已提交
1468
				goto error;
1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479

			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
1480
			}
1481
			copy = min_t(int, copy, pfrag->size - pfrag->offset);
E
Eric Dumazet 已提交
1482
			if (getfrag(from,
1483 1484 1485 1486 1487 1488
				    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
1489 1490
			skb->len += copy;
			skb->data_len += copy;
1491 1492
			skb->truesize += copy;
			atomic_add(copy, &sk->sk_wmem_alloc);
L
Linus Torvalds 已提交
1493 1494 1495 1496
		}
		offset += copy;
		length -= copy;
	}
1497

L
Linus Torvalds 已提交
1498
	return 0;
1499 1500 1501

error_efault:
	err = -EFAULT;
L
Linus Torvalds 已提交
1502
error:
1503
	cork->length -= length;
1504
	IP6_INC_STATS(sock_net(sk), rt->rt6i_idev, IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
1505 1506
	return err;
}
1507
EXPORT_SYMBOL_GPL(ip6_append_data);
L
Linus Torvalds 已提交
1508

1509 1510
static void ip6_cork_release(struct inet_sock *inet, struct ipv6_pinfo *np)
{
1511 1512 1513 1514 1515 1516 1517 1518 1519
	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;
	}

1520 1521 1522 1523
	if (inet->cork.base.dst) {
		dst_release(inet->cork.base.dst);
		inet->cork.base.dst = NULL;
		inet->cork.base.flags &= ~IPCORK_ALLFRAG;
1524 1525 1526 1527
	}
	memset(&inet->cork.fl, 0, sizeof(inet->cork.fl));
}

L
Linus Torvalds 已提交
1528 1529 1530 1531 1532 1533 1534
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);
1535
	struct net *net = sock_net(sk);
L
Linus Torvalds 已提交
1536 1537
	struct ipv6hdr *hdr;
	struct ipv6_txoptions *opt = np->cork.opt;
1538
	struct rt6_info *rt = (struct rt6_info *)inet->cork.base.dst;
1539 1540
	struct flowi6 *fl6 = &inet->cork.fl.u.ip6;
	unsigned char proto = fl6->flowi6_proto;
L
Linus Torvalds 已提交
1541 1542 1543 1544 1545 1546 1547
	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 */
1548
	if (skb->data < skb_network_header(skb))
1549
		__skb_pull(skb, skb_network_offset(skb));
L
Linus Torvalds 已提交
1550
	while ((tmp_skb = __skb_dequeue(&sk->sk_write_queue)) != NULL) {
1551
		__skb_pull(tmp_skb, skb_network_header_len(skb));
L
Linus Torvalds 已提交
1552 1553 1554 1555 1556 1557 1558 1559 1560
		*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;
	}

1561
	/* Allow local fragmentation. */
W
WANG Cong 已提交
1562
	skb->ignore_df = ip6_sk_ignore_df(sk);
1563

A
Alexey Dobriyan 已提交
1564
	*final_dst = fl6->daddr;
1565
	__skb_pull(skb, skb_network_header_len(skb));
L
Linus Torvalds 已提交
1566 1567 1568 1569 1570
	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);

1571 1572
	skb_push(skb, sizeof(struct ipv6hdr));
	skb_reset_network_header(skb);
1573
	hdr = ipv6_hdr(skb);
1574

1575 1576 1577
	ip6_flow_hdr(hdr, np->cork.tclass,
		     ip6_make_flowlabel(net, skb, fl6->flowlabel,
					np->autoflowlabel));
L
Linus Torvalds 已提交
1578 1579
	hdr->hop_limit = np->cork.hop_limit;
	hdr->nexthdr = proto;
A
Alexey Dobriyan 已提交
1580 1581
	hdr->saddr = fl6->saddr;
	hdr->daddr = *final_dst;
L
Linus Torvalds 已提交
1582

1583
	skb->priority = sk->sk_priority;
1584
	skb->mark = sk->sk_mark;
1585

1586
	skb_dst_set(skb, dst_clone(&rt->dst));
1587
	IP6_UPD_PO_STATS(net, rt->rt6i_idev, IPSTATS_MIB_OUT, skb->len);
1588
	if (proto == IPPROTO_ICMPV6) {
E
Eric Dumazet 已提交
1589
		struct inet6_dev *idev = ip6_dst_idev(skb_dst(skb));
1590

1591 1592
		ICMP6MSGOUT_INC_STATS(net, idev, icmp6_hdr(skb)->icmp6_type);
		ICMP6_INC_STATS(net, idev, ICMP6_MIB_OUTMSGS);
1593 1594
	}

H
Herbert Xu 已提交
1595
	err = ip6_local_out(skb);
L
Linus Torvalds 已提交
1596 1597
	if (err) {
		if (err > 0)
E
Eric Dumazet 已提交
1598
			err = net_xmit_errno(err);
L
Linus Torvalds 已提交
1599 1600 1601 1602 1603
		if (err)
			goto error;
	}

out:
1604
	ip6_cork_release(inet, np);
L
Linus Torvalds 已提交
1605 1606
	return err;
error:
1607
	IP6_INC_STATS(net, rt->rt6i_idev, IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
1608 1609
	goto out;
}
1610
EXPORT_SYMBOL_GPL(ip6_push_pending_frames);
L
Linus Torvalds 已提交
1611 1612 1613 1614 1615 1616

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
1617 1618
		if (skb_dst(skb))
			IP6_INC_STATS(sock_net(sk), ip6_dst_idev(skb_dst(skb)),
1619
				      IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
1620 1621 1622
		kfree_skb(skb);
	}

1623
	ip6_cork_release(inet_sk(sk), inet6_sk(sk));
L
Linus Torvalds 已提交
1624
}
1625
EXPORT_SYMBOL_GPL(ip6_flush_pending_frames);