ip6_output.c 41.1 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
int ip6_fragment(struct sk_buff *skb, int (*output)(struct sk_buff *));
L
Linus Torvalds 已提交
60

H
Herbert Xu 已提交
61 62 63 64 65 66 67 68 69
int __ip6_local_out(struct sk_buff *skb)
{
	int len;

	len = skb->len - sizeof(struct ipv6hdr);
	if (len > IPV6_MAXPLEN)
		len = 0;
	ipv6_hdr(skb)->payload_len = htons(len);

70 71
	return nf_hook(NFPROTO_IPV6, NF_INET_LOCAL_OUT, skb, NULL,
		       skb_dst(skb)->dev, dst_output);
H
Herbert Xu 已提交
72 73 74 75 76 77 78 79 80 81 82 83 84 85
}

int ip6_local_out(struct sk_buff *skb)
{
	int err;

	err = __ip6_local_out(skb);
	if (likely(err == 1))
		err = dst_output(skb);

	return err;
}
EXPORT_SYMBOL_GPL(ip6_local_out);

L
Linus Torvalds 已提交
86 87 88
/* dev_loopback_xmit for use with netfilter. */
static int ip6_dev_loopback_xmit(struct sk_buff *newskb)
{
89
	skb_reset_mac_header(newskb);
90
	__skb_pull(newskb, skb_network_offset(newskb));
L
Linus Torvalds 已提交
91 92
	newskb->pkt_type = PACKET_LOOPBACK;
	newskb->ip_summed = CHECKSUM_UNNECESSARY;
E
Eric Dumazet 已提交
93
	WARN_ON(!skb_dst(newskb));
L
Linus Torvalds 已提交
94

E
Eric Dumazet 已提交
95
	netif_rx_ni(newskb);
L
Linus Torvalds 已提交
96 97 98
	return 0;
}

99
static int ip6_finish_output2(struct sk_buff *skb)
L
Linus Torvalds 已提交
100
{
E
Eric Dumazet 已提交
101
	struct dst_entry *dst = skb_dst(skb);
L
Linus Torvalds 已提交
102
	struct net_device *dev = dst->dev;
103
	struct neighbour *neigh;
L
Linus Torvalds 已提交
104 105 106 107

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

108
	if (ipv6_addr_is_multicast(&ipv6_hdr(skb)->daddr)) {
E
Eric Dumazet 已提交
109
		struct inet6_dev *idev = ip6_dst_idev(skb_dst(skb));
L
Linus Torvalds 已提交
110

111
		if (!(dev->flags & IFF_LOOPBACK) && sk_mc_loop(skb->sk) &&
112
		    ((mroute6_socket(dev_net(dev), skb) &&
113
		     !(IP6CB(skb)->flags & IP6SKB_FORWARDED)) ||
114 115
		     ipv6_chk_mcast_addr(dev, &ipv6_hdr(skb)->daddr,
					 &ipv6_hdr(skb)->saddr))) {
L
Linus Torvalds 已提交
116 117 118 119 120 121
			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)
122 123
				NF_HOOK(NFPROTO_IPV6, NF_INET_POST_ROUTING,
					newskb, NULL, newskb->dev,
L
Linus Torvalds 已提交
124 125
					ip6_dev_loopback_xmit);

126
			if (ipv6_hdr(skb)->hop_limit == 0) {
127 128
				IP6_INC_STATS(dev_net(dev), idev,
					      IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
129 130 131 132 133
				kfree_skb(skb);
				return 0;
			}
		}

134 135
		IP6_UPD_PO_STATS(dev_net(dev), idev, IPSTATS_MIB_OUTMCAST,
				skb->len);
L
Linus Torvalds 已提交
136 137
	}

138
	rcu_read_lock();
139
	neigh = dst_get_neighbour(dst);
140 141
	if (neigh) {
		int res = neigh_output(neigh, skb);
142

143 144 145 146
		rcu_read_unlock();
		return res;
	}
	rcu_read_unlock();
147 148 149 150
	IP6_INC_STATS_BH(dev_net(dst->dev),
			 ip6_dst_idev(dst), IPSTATS_MIB_OUTNOROUTES);
	kfree_skb(skb);
	return -EINVAL;
L
Linus Torvalds 已提交
151 152
}

153 154 155 156 157 158 159 160 161
static int ip6_finish_output(struct sk_buff *skb)
{
	if ((skb->len > ip6_skb_dst_mtu(skb) && !skb_is_gso(skb)) ||
	    dst_allfrag(skb_dst(skb)))
		return ip6_fragment(skb, ip6_finish_output2);
	else
		return ip6_finish_output2(skb);
}

L
Linus Torvalds 已提交
162 163
int ip6_output(struct sk_buff *skb)
{
164
	struct net_device *dev = skb_dst(skb)->dev;
E
Eric Dumazet 已提交
165
	struct inet6_dev *idev = ip6_dst_idev(skb_dst(skb));
166
	if (unlikely(idev->cnf.disable_ipv6)) {
167
		IP6_INC_STATS(dev_net(dev), idev,
168
			      IPSTATS_MIB_OUTDISCARDS);
169 170 171 172
		kfree_skb(skb);
		return 0;
	}

173 174 175
	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
176 177 178
}

/*
S
Shan Wei 已提交
179
 *	xmit an sk_buff (used by TCP, SCTP and DCCP)
L
Linus Torvalds 已提交
180 181
 */

182
int ip6_xmit(struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6,
183
	     struct ipv6_txoptions *opt)
L
Linus Torvalds 已提交
184
{
185
	struct net *net = sock_net(sk);
186
	struct ipv6_pinfo *np = inet6_sk(sk);
187
	struct in6_addr *first_hop = &fl6->daddr;
E
Eric Dumazet 已提交
188
	struct dst_entry *dst = skb_dst(skb);
L
Linus Torvalds 已提交
189
	struct ipv6hdr *hdr;
190
	u8  proto = fl6->flowi6_proto;
L
Linus Torvalds 已提交
191
	int seg_len = skb->len;
G
Gerrit Renker 已提交
192 193
	int hlimit = -1;
	int tclass = 0;
L
Linus Torvalds 已提交
194 195 196
	u32 mtu;

	if (opt) {
197
		unsigned int head_room;
L
Linus Torvalds 已提交
198 199 200 201 202 203 204 205 206 207

		/* 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);
208
			if (skb2 == NULL) {
E
Eric Dumazet 已提交
209
				IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
210 211
					      IPSTATS_MIB_OUTDISCARDS);
				kfree_skb(skb);
L
Linus Torvalds 已提交
212 213
				return -ENOBUFS;
			}
214 215
			kfree_skb(skb);
			skb = skb2;
216
			skb_set_owner_w(skb, sk);
L
Linus Torvalds 已提交
217 218 219 220 221 222 223
		}
		if (opt->opt_flen)
			ipv6_push_frag_opts(skb, opt, &proto);
		if (opt->opt_nflen)
			ipv6_push_nfrag_opts(skb, opt, &proto, &first_hop);
	}

224 225
	skb_push(skb, sizeof(struct ipv6hdr));
	skb_reset_network_header(skb);
226
	hdr = ipv6_hdr(skb);
L
Linus Torvalds 已提交
227 228 229 230

	/*
	 *	Fill in the IPv6 header
	 */
G
Gerrit Renker 已提交
231 232
	if (np) {
		tclass = np->tclass;
L
Linus Torvalds 已提交
233
		hlimit = np->hop_limit;
G
Gerrit Renker 已提交
234
	}
L
Linus Torvalds 已提交
235
	if (hlimit < 0)
236
		hlimit = ip6_dst_hoplimit(dst);
L
Linus Torvalds 已提交
237

238
	*(__be32 *)hdr = htonl(0x60000000 | (tclass << 20)) | fl6->flowlabel;
239

L
Linus Torvalds 已提交
240 241 242 243
	hdr->payload_len = htons(seg_len);
	hdr->nexthdr = proto;
	hdr->hop_limit = hlimit;

244
	ipv6_addr_copy(&hdr->saddr, &fl6->saddr);
L
Linus Torvalds 已提交
245 246
	ipv6_addr_copy(&hdr->daddr, first_hop);

247
	skb->priority = sk->sk_priority;
248
	skb->mark = sk->sk_mark;
249

L
Linus Torvalds 已提交
250
	mtu = dst_mtu(dst);
251
	if ((skb->len <= mtu) || skb->local_df || skb_is_gso(skb)) {
E
Eric Dumazet 已提交
252
		IP6_UPD_PO_STATS(net, ip6_dst_idev(skb_dst(skb)),
253
			      IPSTATS_MIB_OUT, skb->len);
254 255
		return NF_HOOK(NFPROTO_IPV6, NF_INET_LOCAL_OUT, skb, NULL,
			       dst->dev, dst_output);
L
Linus Torvalds 已提交
256 257 258 259 260
	}

	if (net_ratelimit())
		printk(KERN_DEBUG "IPv6: sending pkt_too_big to self\n");
	skb->dev = dst->dev;
261
	icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
E
Eric Dumazet 已提交
262
	IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)), IPSTATS_MIB_FRAGFAILS);
L
Linus Torvalds 已提交
263 264 265 266
	kfree_skb(skb);
	return -EMSGSIZE;
}

267 268
EXPORT_SYMBOL(ip6_xmit);

L
Linus Torvalds 已提交
269 270 271 272 273 274 275 276
/*
 *	To avoid extra problems ND packets are send through this
 *	routine. It's code duplication but I really want to avoid
 *	extra checks since ipv6_build_header is used by TCP (which
 *	is for us performance critical)
 */

int ip6_nd_hdr(struct sock *sk, struct sk_buff *skb, struct net_device *dev,
277
	       const struct in6_addr *saddr, const struct in6_addr *daddr,
L
Linus Torvalds 已提交
278 279 280 281 282 283 284 285
	       int proto, int len)
{
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct ipv6hdr *hdr;

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

286 287
	skb_reset_network_header(skb);
	skb_put(skb, sizeof(struct ipv6hdr));
288
	hdr = ipv6_hdr(skb);
L
Linus Torvalds 已提交
289

A
Al Viro 已提交
290
	*(__be32*)hdr = htonl(0x60000000);
L
Linus Torvalds 已提交
291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309

	hdr->payload_len = htons(len);
	hdr->nexthdr = proto;
	hdr->hop_limit = np->hop_limit;

	ipv6_addr_copy(&hdr->saddr, saddr);
	ipv6_addr_copy(&hdr->daddr, daddr);

	return 0;
}

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;
310 311 312
		if (sk && ra->sel == sel &&
		    (!sk->sk_bound_dev_if ||
		     sk->sk_bound_dev_if == skb->dev->ifindex)) {
L
Linus Torvalds 已提交
313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330
			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;
}

331 332
static int ip6_forward_proxy_check(struct sk_buff *skb)
{
333
	struct ipv6hdr *hdr = ipv6_hdr(skb);
334 335 336 337 338 339 340 341 342 343 344 345 346
	u8 nexthdr = hdr->nexthdr;
	int offset;

	if (ipv6_ext_hdr(nexthdr)) {
		offset = ipv6_skip_exthdr(skb, sizeof(*hdr), &nexthdr);
		if (offset < 0)
			return 0;
	} else
		offset = sizeof(struct ipv6hdr);

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

347 348
		if (!pskb_may_pull(skb, (skb_network_header(skb) +
					 offset + 1 - skb->data)))
349 350
			return 0;

351
		icmp6 = (struct icmp6hdr *)(skb_network_header(skb) + offset);
352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368

		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;
		}
	}

369 370 371 372 373 374 375 376 377 378
	/*
	 * 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;
	}

379 380 381
	return 0;
}

L
Linus Torvalds 已提交
382 383 384 385 386 387 388
static inline int ip6_forward_finish(struct sk_buff *skb)
{
	return dst_output(skb);
}

int ip6_forward(struct sk_buff *skb)
{
E
Eric Dumazet 已提交
389
	struct dst_entry *dst = skb_dst(skb);
390
	struct ipv6hdr *hdr = ipv6_hdr(skb);
L
Linus Torvalds 已提交
391
	struct inet6_skb_parm *opt = IP6CB(skb);
392
	struct net *net = dev_net(dst->dev);
393
	struct neighbour *n;
394
	u32 mtu;
395

396
	if (net->ipv6.devconf_all->forwarding == 0)
L
Linus Torvalds 已提交
397 398
		goto error;

399 400 401
	if (skb_warn_if_lro(skb))
		goto drop;

L
Linus Torvalds 已提交
402
	if (!xfrm6_policy_check(NULL, XFRM_POLICY_FWD, skb)) {
403
		IP6_INC_STATS(net, ip6_dst_idev(dst), IPSTATS_MIB_INDISCARDS);
L
Linus Torvalds 已提交
404 405 406
		goto drop;
	}

407 408 409
	if (skb->pkt_type != PACKET_HOST)
		goto drop;

410
	skb_forward_csum(skb);
L
Linus Torvalds 已提交
411 412 413 414 415 416 417 418 419 420 421 422 423 424 425

	/*
	 *	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
	 */
	if (opt->ra) {
426
		u8 *ptr = skb_network_header(skb) + opt->ra;
L
Linus Torvalds 已提交
427 428 429 430 431 432 433 434 435 436
		if (ip6_call_ra_chain(skb, (ptr[2]<<8) + ptr[3]))
			return 0;
	}

	/*
	 *	check and decrement ttl
	 */
	if (hdr->hop_limit <= 1) {
		/* Force OUTPUT device used as source address */
		skb->dev = dst->dev;
437
		icmpv6_send(skb, ICMPV6_TIME_EXCEED, ICMPV6_EXC_HOPLIMIT, 0);
438 439
		IP6_INC_STATS_BH(net,
				 ip6_dst_idev(dst), IPSTATS_MIB_INHDRERRORS);
L
Linus Torvalds 已提交
440 441 442 443 444

		kfree_skb(skb);
		return -ETIMEDOUT;
	}

445
	/* XXX: idev->cnf.proxy_ndp? */
446
	if (net->ipv6.devconf_all->proxy_ndp &&
447
	    pneigh_lookup(&nd_tbl, net, &hdr->daddr, skb->dev, 0)) {
448 449
		int proxied = ip6_forward_proxy_check(skb);
		if (proxied > 0)
450
			return ip6_input(skb);
451
		else if (proxied < 0) {
452 453
			IP6_INC_STATS(net, ip6_dst_idev(dst),
				      IPSTATS_MIB_INDISCARDS);
454 455
			goto drop;
		}
456 457
	}

L
Linus Torvalds 已提交
458
	if (!xfrm6_route_forward(skb)) {
459
		IP6_INC_STATS(net, ip6_dst_idev(dst), IPSTATS_MIB_INDISCARDS);
L
Linus Torvalds 已提交
460 461
		goto drop;
	}
E
Eric Dumazet 已提交
462
	dst = skb_dst(skb);
L
Linus Torvalds 已提交
463 464 465

	/* IPv6 specs say nothing about it, but it is clear that we cannot
	   send redirects to source routed frames.
466
	   We don't send redirects to frames decapsulated from IPsec.
L
Linus Torvalds 已提交
467
	 */
468 469
	n = dst_get_neighbour(dst);
	if (skb->dev == dst->dev && n && opt->srcrt == 0 && !skb_sec_path(skb)) {
L
Linus Torvalds 已提交
470 471 472 473 474 475 476 477 478 479 480 481 482 483
		struct in6_addr *target = NULL;
		struct rt6_info *rt;

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

		rt = (struct rt6_info *) dst;
		if ((rt->rt6i_flags & RTF_GATEWAY))
			target = (struct in6_addr*)&n->primary_key;
		else
			target = &hdr->daddr;

484 485 486
		if (!rt->rt6i_peer)
			rt6_bind_peer(rt, 1);

L
Linus Torvalds 已提交
487 488 489
		/* Limit redirects both by destination (here)
		   and by source (inside ndisc_send_redirect)
		 */
490
		if (inet_peer_xrlim_allow(rt->rt6i_peer, 1*HZ))
L
Linus Torvalds 已提交
491
			ndisc_send_redirect(skb, n, target);
492 493 494
	} else {
		int addrtype = ipv6_addr_type(&hdr->saddr);

L
Linus Torvalds 已提交
495
		/* This check is security critical. */
496 497
		if (addrtype == IPV6_ADDR_ANY ||
		    addrtype & (IPV6_ADDR_MULTICAST | IPV6_ADDR_LOOPBACK))
498 499 500
			goto error;
		if (addrtype & IPV6_ADDR_LINKLOCAL) {
			icmpv6_send(skb, ICMPV6_DEST_UNREACH,
501
				    ICMPV6_NOT_NEIGHBOUR, 0);
502 503
			goto error;
		}
L
Linus Torvalds 已提交
504 505
	}

506 507 508 509
	mtu = dst_mtu(dst);
	if (mtu < IPV6_MIN_MTU)
		mtu = IPV6_MIN_MTU;

510
	if (skb->len > mtu && !skb_is_gso(skb)) {
L
Linus Torvalds 已提交
511 512
		/* Again, force OUTPUT device used as source address */
		skb->dev = dst->dev;
513
		icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
514 515 516 517
		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
518 519 520 521 522
		kfree_skb(skb);
		return -EMSGSIZE;
	}

	if (skb_cow(skb, dst->dev->hard_header_len)) {
523
		IP6_INC_STATS(net, ip6_dst_idev(dst), IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
524 525 526
		goto drop;
	}

527
	hdr = ipv6_hdr(skb);
L
Linus Torvalds 已提交
528 529

	/* Mangling hops number delayed to point after skb COW */
530

L
Linus Torvalds 已提交
531 532
	hdr->hop_limit--;

533
	IP6_INC_STATS_BH(net, ip6_dst_idev(dst), IPSTATS_MIB_OUTFORWDATAGRAMS);
534
	return NF_HOOK(NFPROTO_IPV6, NF_INET_FORWARD, skb, skb->dev, dst->dev,
535
		       ip6_forward_finish);
L
Linus Torvalds 已提交
536 537

error:
538
	IP6_INC_STATS_BH(net, ip6_dst_idev(dst), IPSTATS_MIB_INADDRERRORS);
L
Linus Torvalds 已提交
539 540 541 542 543 544 545 546 547 548
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
549 550
	skb_dst_drop(to);
	skb_dst_set(to, dst_clone(skb_dst(from)));
L
Linus Torvalds 已提交
551
	to->dev = from->dev;
T
Thomas Graf 已提交
552
	to->mark = from->mark;
L
Linus Torvalds 已提交
553 554 555 556

#ifdef CONFIG_NET_SCHED
	to->tc_index = from->tc_index;
#endif
557
	nf_copy(to, from);
558 559 560 561
#if defined(CONFIG_NETFILTER_XT_TARGET_TRACE) || \
    defined(CONFIG_NETFILTER_XT_TARGET_TRACE_MODULE)
	to->nf_trace = from->nf_trace;
#endif
562
	skb_copy_secmark(to, from);
L
Linus Torvalds 已提交
563 564 565 566 567
}

int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr)
{
	u16 offset = sizeof(struct ipv6hdr);
568 569
	struct ipv6_opt_hdr *exthdr =
				(struct ipv6_opt_hdr *)(ipv6_hdr(skb) + 1);
570
	unsigned int packet_len = skb->tail - skb->network_header;
L
Linus Torvalds 已提交
571
	int found_rhdr = 0;
572
	*nexthdr = &ipv6_hdr(skb)->nexthdr;
L
Linus Torvalds 已提交
573 574 575 576 577 578

	while (offset + 1 <= packet_len) {

		switch (**nexthdr) {

		case NEXTHDR_HOP:
579
			break;
L
Linus Torvalds 已提交
580
		case NEXTHDR_ROUTING:
581 582
			found_rhdr = 1;
			break;
L
Linus Torvalds 已提交
583
		case NEXTHDR_DEST:
584
#if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
585 586 587 588 589
			if (ipv6_find_tlv(skb, offset, IPV6_TLV_HAO) >= 0)
				break;
#endif
			if (found_rhdr)
				return offset;
L
Linus Torvalds 已提交
590 591 592 593
			break;
		default :
			return offset;
		}
594 595 596

		offset += ipv6_optlen(exthdr);
		*nexthdr = &exthdr->nexthdr;
597 598
		exthdr = (struct ipv6_opt_hdr *)(skb_network_header(skb) +
						 offset);
L
Linus Torvalds 已提交
599 600 601 602 603
	}

	return offset;
}

604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628
void ipv6_select_ident(struct frag_hdr *fhdr, struct rt6_info *rt)
{
	static atomic_t ipv6_fragmentation_id;
	int old, new;

	if (rt) {
		struct inet_peer *peer;

		if (!rt->rt6i_peer)
			rt6_bind_peer(rt, 1);
		peer = rt->rt6i_peer;
		if (peer) {
			fhdr->identification = htonl(inet_getid(peer, 0));
			return;
		}
	}
	do {
		old = atomic_read(&ipv6_fragmentation_id);
		new = old + 1;
		if (!new)
			new = 1;
	} while (atomic_cmpxchg(&ipv6_fragmentation_id, old, new) != old);
	fhdr->identification = htonl(new);
}

629
int ip6_fragment(struct sk_buff *skb, int (*output)(struct sk_buff *))
L
Linus Torvalds 已提交
630 631
{
	struct sk_buff *frag;
E
Eric Dumazet 已提交
632
	struct rt6_info *rt = (struct rt6_info*)skb_dst(skb);
633
	struct ipv6_pinfo *np = skb->sk ? inet6_sk(skb->sk) : NULL;
L
Linus Torvalds 已提交
634 635 636
	struct ipv6hdr *tmp_hdr;
	struct frag_hdr *fh;
	unsigned int mtu, hlen, left, len;
A
Al Viro 已提交
637
	__be32 frag_id = 0;
L
Linus Torvalds 已提交
638 639
	int ptr, offset = 0, err=0;
	u8 *prevhdr, nexthdr = 0;
E
Eric Dumazet 已提交
640
	struct net *net = dev_net(skb_dst(skb)->dev);
L
Linus Torvalds 已提交
641 642 643 644

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

J
John Heffner 已提交
645
	mtu = ip6_skb_dst_mtu(skb);
646 647

	/* We must not fragment if the socket is set to force MTU discovery
648
	 * or if the skb it not generated by a local socket.
649
	 */
650
	if (!skb->local_df && skb->len > mtu) {
E
Eric Dumazet 已提交
651
		skb->dev = skb_dst(skb)->dev;
652
		icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
E
Eric Dumazet 已提交
653
		IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
654
			      IPSTATS_MIB_FRAGFAILS);
655 656 657 658
		kfree_skb(skb);
		return -EMSGSIZE;
	}

659 660 661 662 663
	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
664

665
	if (skb_has_frag_list(skb)) {
L
Linus Torvalds 已提交
666
		int first_len = skb_pagelen(skb);
667
		struct sk_buff *frag2;
L
Linus Torvalds 已提交
668 669 670 671 672 673

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

674
		skb_walk_frags(skb, frag) {
L
Linus Torvalds 已提交
675 676 677 678
			/* Correct geometry. */
			if (frag->len > mtu ||
			    ((frag->len & 7) && frag->next) ||
			    skb_headroom(frag) < hlen)
679
				goto slow_path_clean;
L
Linus Torvalds 已提交
680 681 682

			/* Partially cloned skb? */
			if (skb_shared(frag))
683
				goto slow_path_clean;
684 685 686 687 688 689

			BUG_ON(frag->sk);
			if (skb->sk) {
				frag->sk = skb->sk;
				frag->destructor = sock_wfree;
			}
690
			skb->truesize -= frag->truesize;
L
Linus Torvalds 已提交
691 692 693 694 695
		}

		err = 0;
		offset = 0;
		frag = skb_shinfo(skb)->frag_list;
696
		skb_frag_list_init(skb);
L
Linus Torvalds 已提交
697 698
		/* BUILD HEADER */

Y
YOSHIFUJI Hideaki 已提交
699
		*prevhdr = NEXTHDR_FRAGMENT;
700
		tmp_hdr = kmemdup(skb_network_header(skb), hlen, GFP_ATOMIC);
L
Linus Torvalds 已提交
701
		if (!tmp_hdr) {
E
Eric Dumazet 已提交
702
			IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
703
				      IPSTATS_MIB_FRAGFAILS);
L
Linus Torvalds 已提交
704 705 706 707 708
			return -ENOMEM;
		}

		__skb_pull(skb, hlen);
		fh = (struct frag_hdr*)__skb_push(skb, sizeof(struct frag_hdr));
709 710
		__skb_push(skb, hlen);
		skb_reset_network_header(skb);
711
		memcpy(skb_network_header(skb), tmp_hdr, hlen);
L
Linus Torvalds 已提交
712

713
		ipv6_select_ident(fh, rt);
L
Linus Torvalds 已提交
714 715 716 717 718 719 720 721
		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;
722 723
		ipv6_hdr(skb)->payload_len = htons(first_len -
						   sizeof(struct ipv6hdr));
724

725
		dst_hold(&rt->dst);
L
Linus Torvalds 已提交
726 727 728 729 730 731

		for (;;) {
			/* Prepare header of the next frame,
			 * before previous one went down. */
			if (frag) {
				frag->ip_summed = CHECKSUM_NONE;
732
				skb_reset_transport_header(frag);
L
Linus Torvalds 已提交
733
				fh = (struct frag_hdr*)__skb_push(frag, sizeof(struct frag_hdr));
734 735
				__skb_push(frag, hlen);
				skb_reset_network_header(frag);
736 737
				memcpy(skb_network_header(frag), tmp_hdr,
				       hlen);
L
Linus Torvalds 已提交
738 739 740 741 742 743 744
				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;
745 746 747
				ipv6_hdr(frag)->payload_len =
						htons(frag->len -
						      sizeof(struct ipv6hdr));
L
Linus Torvalds 已提交
748 749
				ip6_copy_metadata(frag, skb);
			}
750

L
Linus Torvalds 已提交
751
			err = output(skb);
752
			if(!err)
753
				IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
754
					      IPSTATS_MIB_FRAGCREATES);
755

L
Linus Torvalds 已提交
756 757 758 759 760 761 762 763
			if (err || !frag)
				break;

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

J
Jesper Juhl 已提交
764
		kfree(tmp_hdr);
L
Linus Torvalds 已提交
765 766

		if (err == 0) {
767
			IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
768
				      IPSTATS_MIB_FRAGOKS);
769
			dst_release(&rt->dst);
L
Linus Torvalds 已提交
770 771 772 773 774 775 776 777 778
			return 0;
		}

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

779
		IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
780
			      IPSTATS_MIB_FRAGFAILS);
781
		dst_release(&rt->dst);
L
Linus Torvalds 已提交
782
		return err;
783 784 785 786 787 788 789 790 791

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
792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811
	}

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

	/*
	 *	Fragment the datagram.
	 */

	*prevhdr = NEXTHDR_FRAGMENT;

	/*
	 *	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
812
		/* IF: we are not sending up to and including the packet end
L
Linus Torvalds 已提交
813 814 815 816 817 818 819 820
		   then align the next start on an eight byte boundary */
		if (len < left)	{
			len &= ~7;
		}
		/*
		 *	Allocate buffer.
		 */

821
		if ((frag = alloc_skb(len+hlen+sizeof(struct frag_hdr)+LL_ALLOCATED_SPACE(rt->dst.dev), GFP_ATOMIC)) == NULL) {
822
			NETDEBUG(KERN_INFO "IPv6: frag: no memory for new fragment!\n");
E
Eric Dumazet 已提交
823
			IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
824
				      IPSTATS_MIB_FRAGFAILS);
L
Linus Torvalds 已提交
825 826 827 828 829 830 831 832 833
			err = -ENOMEM;
			goto fail;
		}

		/*
		 *	Set up data on packet
		 */

		ip6_copy_metadata(frag, skb);
834
		skb_reserve(frag, LL_RESERVED_SPACE(rt->dst.dev));
L
Linus Torvalds 已提交
835
		skb_put(frag, len + hlen + sizeof(struct frag_hdr));
836
		skb_reset_network_header(frag);
837
		fh = (struct frag_hdr *)(skb_network_header(frag) + hlen);
838 839
		frag->transport_header = (frag->network_header + hlen +
					  sizeof(struct frag_hdr));
L
Linus Torvalds 已提交
840 841 842 843 844 845 846 847 848 849 850

		/*
		 *	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.
		 */
851
		skb_copy_from_linear_data(skb, skb_network_header(frag), hlen);
L
Linus Torvalds 已提交
852 853 854 855 856 857

		/*
		 *	Build fragment header.
		 */
		fh->nexthdr = nexthdr;
		fh->reserved = 0;
858
		if (!frag_id) {
859
			ipv6_select_ident(fh, rt);
L
Linus Torvalds 已提交
860 861 862 863 864 865 866
			frag_id = fh->identification;
		} else
			fh->identification = frag_id;

		/*
		 *	Copy a block of the IP datagram.
		 */
867
		if (skb_copy_bits(skb, ptr, skb_transport_header(frag), len))
L
Linus Torvalds 已提交
868 869 870 871 872 873
			BUG();
		left -= len;

		fh->frag_off = htons(offset);
		if (left > 0)
			fh->frag_off |= htons(IP6_MF);
874 875
		ipv6_hdr(frag)->payload_len = htons(frag->len -
						    sizeof(struct ipv6hdr));
L
Linus Torvalds 已提交
876 877 878 879 880 881 882 883 884 885

		ptr += len;
		offset += len;

		/*
		 *	Put this fragment into the sending queue.
		 */
		err = output(frag);
		if (err)
			goto fail;
886

E
Eric Dumazet 已提交
887
		IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
888
			      IPSTATS_MIB_FRAGCREATES);
L
Linus Torvalds 已提交
889
	}
E
Eric Dumazet 已提交
890
	IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
891
		      IPSTATS_MIB_FRAGOKS);
L
Linus Torvalds 已提交
892 893 894 895
	kfree_skb(skb);
	return err;

fail:
E
Eric Dumazet 已提交
896
	IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
897
		      IPSTATS_MIB_FRAGFAILS);
898
	kfree_skb(skb);
L
Linus Torvalds 已提交
899 900 901
	return err;
}

902 903 904
static inline int ip6_rt_check(const struct rt6key *rt_key,
			       const struct in6_addr *fl_addr,
			       const struct in6_addr *addr_cache)
905
{
E
Eric Dumazet 已提交
906 907
	return (rt_key->plen != 128 || !ipv6_addr_equal(fl_addr, &rt_key->addr)) &&
		(addr_cache == NULL || !ipv6_addr_equal(fl_addr, addr_cache));
908 909
}

910 911
static struct dst_entry *ip6_sk_dst_check(struct sock *sk,
					  struct dst_entry *dst,
912
					  const struct flowi6 *fl6)
L
Linus Torvalds 已提交
913
{
914 915
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct rt6_info *rt = (struct rt6_info *)dst;
L
Linus Torvalds 已提交
916

917 918 919 920 921 922 923 924
	if (!dst)
		goto out;

	/* 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)
	 *
925 926
	 * 1. ip6_rt_check(): If route was host route,
	 *    check that cached destination is current.
927 928 929 930 931 932 933 934 935 936
	 *    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.
	 */
937
	if (ip6_rt_check(&rt->rt6i_dst, &fl6->daddr, np->daddr_cache) ||
938
#ifdef CONFIG_IPV6_SUBTREES
939
	    ip6_rt_check(&rt->rt6i_src, &fl6->saddr, np->saddr_cache) ||
940
#endif
941
	    (fl6->flowi6_oif && fl6->flowi6_oif != dst->dev->ifindex)) {
942 943
		dst_release(dst);
		dst = NULL;
L
Linus Torvalds 已提交
944 945
	}

946 947 948 949 950
out:
	return dst;
}

static int ip6_dst_lookup_tail(struct sock *sk,
951
			       struct dst_entry **dst, struct flowi6 *fl6)
952
{
953
	struct net *net = sock_net(sk);
954 955 956 957
#ifdef CONFIG_IPV6_OPTIMISTIC_DAD
	struct neighbour *n;
#endif
	int err;
958

L
Linus Torvalds 已提交
959
	if (*dst == NULL)
960
		*dst = ip6_route_output(net, sk, fl6);
L
Linus Torvalds 已提交
961 962 963 964

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

965
	if (ipv6_addr_any(&fl6->saddr)) {
966 967 968 969
		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);
970
		if (err)
L
Linus Torvalds 已提交
971 972 973
			goto out_err_release;
	}

974
#ifdef CONFIG_IPV6_OPTIMISTIC_DAD
975 976 977 978 979 980 981 982
	/*
	 * 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
	 */
983
	rcu_read_lock();
984 985
	n = dst_get_neighbour(*dst);
	if (n && !(n->nud_state & NUD_VALID)) {
986
		struct inet6_ifaddr *ifp;
987
		struct flowi6 fl_gw6;
988 989
		int redirect;

990
		rcu_read_unlock();
991
		ifp = ipv6_get_ifaddr(net, &fl6->saddr,
992 993 994 995 996 997 998 999 1000 1001 1002 1003
				      (*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);
1004 1005 1006
			memcpy(&fl_gw6, fl6, sizeof(struct flowi6));
			memset(&fl_gw6.daddr, 0, sizeof(struct in6_addr));
			*dst = ip6_route_output(net, sk, &fl_gw6);
1007 1008
			if ((err = (*dst)->error))
				goto out_err_release;
1009
		}
1010 1011
	} else {
		rcu_read_unlock();
1012
	}
1013 1014
#endif

L
Linus Torvalds 已提交
1015 1016 1017
	return 0;

out_err_release:
1018
	if (err == -ENETUNREACH)
1019
		IP6_INC_STATS_BH(net, NULL, IPSTATS_MIB_OUTNOROUTES);
L
Linus Torvalds 已提交
1020 1021 1022 1023
	dst_release(*dst);
	*dst = NULL;
	return err;
}
A
Adrian Bunk 已提交
1024

1025 1026 1027 1028
/**
 *	ip6_dst_lookup - perform route lookup on flow
 *	@sk: socket which provides route info
 *	@dst: pointer to dst_entry * for result
1029
 *	@fl6: flow to lookup
1030 1031 1032 1033 1034
 *
 *	This function performs a route lookup on the given flow.
 *
 *	It returns zero on success, or a standard errno code on error.
 */
1035
int ip6_dst_lookup(struct sock *sk, struct dst_entry **dst, struct flowi6 *fl6)
1036 1037
{
	*dst = NULL;
1038
	return ip6_dst_lookup_tail(sk, dst, fl6);
1039
}
1040 1041
EXPORT_SYMBOL_GPL(ip6_dst_lookup);

1042
/**
1043 1044
 *	ip6_dst_lookup_flow - perform route lookup on flow with ipsec
 *	@sk: socket which provides route info
1045
 *	@fl6: flow to lookup
1046
 *	@final_dst: final destination address for ipsec lookup
1047
 *	@can_sleep: we are in a sleepable context
1048 1049 1050 1051 1052 1053
 *
 *	This function performs a route lookup on the given flow.
 *
 *	It returns a valid dst pointer on success, or a pointer encoded
 *	error code.
 */
1054
struct dst_entry *ip6_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
1055
				      const struct in6_addr *final_dst,
1056
				      bool can_sleep)
1057 1058 1059 1060
{
	struct dst_entry *dst = NULL;
	int err;

1061
	err = ip6_dst_lookup_tail(sk, &dst, fl6);
1062 1063 1064
	if (err)
		return ERR_PTR(err);
	if (final_dst)
1065
		ipv6_addr_copy(&fl6->daddr, final_dst);
1066
	if (can_sleep)
1067
		fl6->flowi6_flags |= FLOWI_FLAG_CAN_SLEEP;
1068

1069
	return xfrm_lookup(sock_net(sk), dst, flowi6_to_flowi(fl6), sk, 0);
1070 1071 1072 1073 1074
}
EXPORT_SYMBOL_GPL(ip6_dst_lookup_flow);

/**
 *	ip6_sk_dst_lookup_flow - perform socket cached route lookup on flow
1075
 *	@sk: socket which provides the dst cache and route info
1076
 *	@fl6: flow to lookup
1077
 *	@final_dst: final destination address for ipsec lookup
1078
 *	@can_sleep: we are in a sleepable context
1079 1080 1081 1082 1083 1084
 *
 *	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.
 *
1085 1086
 *	It returns a valid dst pointer on success, or a pointer encoded
 *	error code.
1087
 */
1088
struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
1089
					 const struct in6_addr *final_dst,
1090
					 bool can_sleep)
1091
{
1092 1093
	struct dst_entry *dst = sk_dst_check(sk, inet6_sk(sk)->dst_cookie);
	int err;
1094

1095
	dst = ip6_sk_dst_check(sk, dst, fl6);
1096

1097
	err = ip6_dst_lookup_tail(sk, &dst, fl6);
1098 1099 1100
	if (err)
		return ERR_PTR(err);
	if (final_dst)
1101
		ipv6_addr_copy(&fl6->daddr, final_dst);
1102
	if (can_sleep)
1103
		fl6->flowi6_flags |= FLOWI_FLAG_CAN_SLEEP;
1104

1105
	return xfrm_lookup(sock_net(sk), dst, flowi6_to_flowi(fl6), sk, 0);
1106
}
1107
EXPORT_SYMBOL_GPL(ip6_sk_dst_lookup_flow);
1108

A
Adrian Bunk 已提交
1109
static inline int ip6_ufo_append_data(struct sock *sk,
1110 1111 1112
			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,
1113 1114
			int transhdrlen, int mtu,unsigned int flags,
			struct rt6_info *rt)
1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137

{
	struct sk_buff *skb;
	int err;

	/* There is support for UDP large send offload by network
	 * device, so create one single skb packet containing complete
	 * udp datagram
	 */
	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)
			return -ENOMEM;

		/* 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 */
1138
		skb_reset_network_header(skb);
1139 1140

		/* initialize protocol header pointer */
1141
		skb->transport_header = skb->network_header + fragheaderlen;
1142

1143
		skb->ip_summed = CHECKSUM_PARTIAL;
1144 1145 1146 1147 1148 1149 1150 1151
		skb->csum = 0;
	}

	err = skb_append_datato_frags(sk,skb, getfrag, from,
				      (length - transhdrlen));
	if (!err) {
		struct frag_hdr fhdr;

1152 1153 1154 1155 1156
		/* 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;
H
Herbert Xu 已提交
1157
		skb_shinfo(skb)->gso_type = SKB_GSO_UDP;
1158
		ipv6_select_ident(&fhdr, rt);
1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170
		skb_shinfo(skb)->ip6_frag_id = fhdr.identification;
		__skb_queue_tail(&sk->sk_write_queue, skb);

		return 0;
	}
	/* There is not enough support do UPD LSO,
	 * so follow normal path
	 */
	kfree_skb(skb);

	return err;
}
L
Linus Torvalds 已提交
1171

1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183
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;
}

1184 1185 1186
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,
1187
	int hlimit, int tclass, struct ipv6_txoptions *opt, struct flowi6 *fl6,
1188
	struct rt6_info *rt, unsigned int flags, int dontfrag)
L
Linus Torvalds 已提交
1189 1190 1191
{
	struct inet_sock *inet = inet_sk(sk);
	struct ipv6_pinfo *np = inet6_sk(sk);
1192
	struct inet_cork *cork;
L
Linus Torvalds 已提交
1193 1194 1195
	struct sk_buff *skb;
	unsigned int maxfraglen, fragheaderlen;
	int exthdrlen;
1196
	int dst_exthdrlen;
L
Linus Torvalds 已提交
1197 1198 1199 1200 1201 1202
	int hh_len;
	int mtu;
	int copy;
	int err;
	int offset = 0;
	int csummode = CHECKSUM_NONE;
1203
	__u8 tx_flags = 0;
L
Linus Torvalds 已提交
1204 1205 1206

	if (flags&MSG_PROBE)
		return 0;
1207
	cork = &inet->cork.base;
L
Linus Torvalds 已提交
1208 1209 1210 1211 1212
	if (skb_queue_empty(&sk->sk_write_queue)) {
		/*
		 * setup for corking
		 */
		if (opt) {
1213
			if (WARN_ON(np->cork.opt))
L
Linus Torvalds 已提交
1214
				return -EINVAL;
1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243

			np->cork.opt = kmalloc(opt->tot_len, sk->sk_allocation);
			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
1244 1245
			/* need source address above miyazawa*/
		}
1246
		dst_hold(&rt->dst);
1247
		cork->dst = &rt->dst;
1248
		inet->cork.fl.u.ip6 = *fl6;
L
Linus Torvalds 已提交
1249
		np->cork.hop_limit = hlimit;
1250
		np->cork.tclass = tclass;
J
John Heffner 已提交
1251
		mtu = np->pmtudisc == IPV6_PMTUDISC_PROBE ?
1252
		      rt->dst.dev->mtu : dst_mtu(&rt->dst);
1253
		if (np->frag_size < mtu) {
1254 1255 1256
			if (np->frag_size)
				mtu = np->frag_size;
		}
1257
		cork->fragsize = mtu;
1258
		if (dst_allfrag(rt->dst.path))
1259 1260
			cork->flags |= IPCORK_ALLFRAG;
		cork->length = 0;
L
Linus Torvalds 已提交
1261 1262
		sk->sk_sndmsg_page = NULL;
		sk->sk_sndmsg_off = 0;
1263
		exthdrlen = (opt ? opt->opt_flen : 0) - rt->rt6i_nfheader_len;
L
Linus Torvalds 已提交
1264 1265
		length += exthdrlen;
		transhdrlen += exthdrlen;
1266
		dst_exthdrlen = rt->dst.header_len;
L
Linus Torvalds 已提交
1267
	} else {
1268
		rt = (struct rt6_info *)cork->dst;
1269
		fl6 = &inet->cork.fl.u.ip6;
1270
		opt = np->cork.opt;
L
Linus Torvalds 已提交
1271 1272
		transhdrlen = 0;
		exthdrlen = 0;
1273
		dst_exthdrlen = 0;
1274
		mtu = cork->fragsize;
L
Linus Torvalds 已提交
1275 1276
	}

1277
	hh_len = LL_RESERVED_SPACE(rt->dst.dev);
L
Linus Torvalds 已提交
1278

1279
	fragheaderlen = sizeof(struct ipv6hdr) + rt->rt6i_nfheader_len +
1280
			(opt ? opt->opt_nflen : 0);
L
Linus Torvalds 已提交
1281 1282 1283
	maxfraglen = ((mtu - fragheaderlen) & ~7) + fragheaderlen - sizeof(struct frag_hdr);

	if (mtu <= sizeof(struct ipv6hdr) + IPV6_MAXPLEN) {
1284
		if (cork->length + length > sizeof(struct ipv6hdr) + IPV6_MAXPLEN - fragheaderlen) {
1285
			ipv6_local_error(sk, EMSGSIZE, fl6, mtu-exthdrlen);
L
Linus Torvalds 已提交
1286 1287 1288 1289
			return -EMSGSIZE;
		}
	}

1290 1291 1292 1293 1294 1295 1296
	/* For UDP, check if TX timestamp is enabled */
	if (sk->sk_type == SOCK_DGRAM) {
		err = sock_tx_timestamp(sk, &tx_flags);
		if (err)
			goto error;
	}

L
Linus Torvalds 已提交
1297 1298 1299 1300 1301 1302 1303
	/*
	 * 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
1304
	 * of the buffer to the new fragment when we split
L
Linus Torvalds 已提交
1305 1306
	 * the message.
	 *
1307
	 * FIXME: It may be fragmented into multiple chunks
L
Linus Torvalds 已提交
1308 1309
	 *        at once if non-fragmentable extension headers
	 *        are too large.
1310
	 * --yoshfuji
L
Linus Torvalds 已提交
1311 1312
	 */

1313
	cork->length += length;
1314 1315 1316
	if (length > mtu) {
		int proto = sk->sk_protocol;
		if (dontfrag && (proto == IPPROTO_UDP || proto == IPPROTO_RAW)){
1317
			ipv6_local_rxpmtu(sk, fl6, mtu-exthdrlen);
1318 1319
			return -EMSGSIZE;
		}
1320

1321
		if (proto == IPPROTO_UDP &&
1322
		    (rt->dst.dev->features & NETIF_F_UFO)) {
1323 1324 1325

			err = ip6_ufo_append_data(sk, getfrag, from, length,
						  hh_len, fragheaderlen,
1326
						  transhdrlen, mtu, flags, rt);
1327 1328 1329 1330
			if (err)
				goto error;
			return 0;
		}
1331
	}
L
Linus Torvalds 已提交
1332 1333 1334 1335 1336 1337

	if ((skb = skb_peek_tail(&sk->sk_write_queue)) == NULL)
		goto alloc_new_skb;

	while (length > 0) {
		/* Check if the remaining data fits into current packet. */
1338
		copy = (cork->length <= mtu && !(cork->flags & IPCORK_ALLFRAG) ? mtu : maxfraglen) - skb->len;
L
Linus Torvalds 已提交
1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362
		if (copy < length)
			copy = maxfraglen - skb->len;

		if (copy <= 0) {
			char *data;
			unsigned int datalen;
			unsigned int fraglen;
			unsigned int fraggap;
			unsigned int alloclen;
			struct sk_buff *skb_prev;
alloc_new_skb:
			skb_prev = skb;

			/* There's no room in the current skb */
			if (skb_prev)
				fraggap = skb_prev->len - maxfraglen;
			else
				fraggap = 0;

			/*
			 * If remaining data exceeds the mtu,
			 * we know we need more fragment(s).
			 */
			datalen = length + fraggap;
1363
			if (datalen > (cork->length <= mtu && !(cork->flags & IPCORK_ALLFRAG) ? mtu : maxfraglen) - fragheaderlen)
L
Linus Torvalds 已提交
1364 1365 1366 1367
				datalen = maxfraglen - fragheaderlen;

			fraglen = datalen + fragheaderlen;
			if ((flags & MSG_MORE) &&
1368
			    !(rt->dst.dev->features&NETIF_F_SG))
L
Linus Torvalds 已提交
1369 1370 1371 1372
				alloclen = mtu;
			else
				alloclen = datalen + fragheaderlen;

1373 1374
			alloclen += dst_exthdrlen;

L
Linus Torvalds 已提交
1375 1376 1377 1378 1379 1380
			/*
			 * The last fragment gets additional space at tail.
			 * Note: we overallocate on fragments with MSG_MODE
			 * because we have no idea if we're the last one.
			 */
			if (datalen == length + fraggap)
1381
				alloclen += rt->dst.trailer_len;
L
Linus Torvalds 已提交
1382 1383 1384

			/*
			 * We just reserve space for fragment header.
1385
			 * Note: this may be overallocation if the message
L
Linus Torvalds 已提交
1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402
			 * (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;
1403 1404 1405 1406 1407 1408
				else {
					/* Only the initial fragment
					 * is time stamped.
					 */
					tx_flags = 0;
				}
L
Linus Torvalds 已提交
1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419
			}
			if (skb == NULL)
				goto error;
			/*
			 *	Fill in the control structures
			 */
			skb->ip_summed = csummode;
			skb->csum = 0;
			/* reserve for fragmentation */
			skb_reserve(skb, hh_len+sizeof(struct frag_hdr));

1420 1421 1422
			if (sk->sk_type == SOCK_DGRAM)
				skb_shinfo(skb)->tx_flags = tx_flags;

L
Linus Torvalds 已提交
1423 1424 1425
			/*
			 *	Find where to start putting bytes
			 */
1426 1427 1428
			data = skb_put(skb, fraglen + dst_exthdrlen);
			skb_set_network_header(skb, exthdrlen + dst_exthdrlen);
			data += fragheaderlen + dst_exthdrlen;
1429 1430
			skb->transport_header = (skb->network_header +
						 fragheaderlen);
L
Linus Torvalds 已提交
1431 1432 1433 1434 1435 1436 1437
			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;
1438
				pskb_trim_unique(skb_prev, maxfraglen);
L
Linus Torvalds 已提交
1439 1440
			}
			copy = datalen - transhdrlen - fraggap;
1441

L
Linus Torvalds 已提交
1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455
			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;
1456
			dst_exthdrlen = 0;
L
Linus Torvalds 已提交
1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468
			csummode = CHECKSUM_NONE;

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

		if (copy > length)
			copy = length;

1469
		if (!(rt->dst.dev->features&NETIF_F_SG)) {
L
Linus Torvalds 已提交
1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488
			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;
			skb_frag_t *frag = &skb_shinfo(skb)->frags[i-1];
			struct page *page = sk->sk_sndmsg_page;
			int off = sk->sk_sndmsg_off;
			unsigned int left;

			if (page && (left = PAGE_SIZE - off) > 0) {
				if (copy >= left)
					copy = left;
1489
				if (page != skb_frag_page(frag)) {
L
Linus Torvalds 已提交
1490 1491 1492 1493 1494
					if (i == MAX_SKB_FRAGS) {
						err = -EMSGSIZE;
						goto error;
					}
					skb_fill_page_desc(skb, i, page, sk->sk_sndmsg_off, 0);
1495
					skb_frag_ref(skb, i);
L
Linus Torvalds 已提交
1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514
					frag = &skb_shinfo(skb)->frags[i];
				}
			} else if(i < MAX_SKB_FRAGS) {
				if (copy > PAGE_SIZE)
					copy = PAGE_SIZE;
				page = alloc_pages(sk->sk_allocation, 0);
				if (page == NULL) {
					err = -ENOMEM;
					goto error;
				}
				sk->sk_sndmsg_page = page;
				sk->sk_sndmsg_off = 0;

				skb_fill_page_desc(skb, i, page, 0, 0);
				frag = &skb_shinfo(skb)->frags[i];
			} else {
				err = -EMSGSIZE;
				goto error;
			}
1515 1516
			if (getfrag(from, skb_frag_address(frag)+frag->size,
				    offset, copy, skb->len, skb) < 0) {
L
Linus Torvalds 已提交
1517 1518 1519 1520 1521 1522 1523
				err = -EFAULT;
				goto error;
			}
			sk->sk_sndmsg_off += copy;
			frag->size += copy;
			skb->len += copy;
			skb->data_len += copy;
1524 1525
			skb->truesize += copy;
			atomic_add(copy, &sk->sk_wmem_alloc);
L
Linus Torvalds 已提交
1526 1527 1528 1529 1530 1531
		}
		offset += copy;
		length -= copy;
	}
	return 0;
error:
1532
	cork->length -= length;
1533
	IP6_INC_STATS(sock_net(sk), rt->rt6i_idev, IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
1534 1535 1536
	return err;
}

1537 1538
static void ip6_cork_release(struct inet_sock *inet, struct ipv6_pinfo *np)
{
1539 1540 1541 1542 1543 1544 1545 1546 1547
	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;
	}

1548 1549 1550 1551
	if (inet->cork.base.dst) {
		dst_release(inet->cork.base.dst);
		inet->cork.base.dst = NULL;
		inet->cork.base.flags &= ~IPCORK_ALLFRAG;
1552 1553 1554 1555
	}
	memset(&inet->cork.fl, 0, sizeof(inet->cork.fl));
}

L
Linus Torvalds 已提交
1556 1557 1558 1559 1560 1561 1562
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);
1563
	struct net *net = sock_net(sk);
L
Linus Torvalds 已提交
1564 1565
	struct ipv6hdr *hdr;
	struct ipv6_txoptions *opt = np->cork.opt;
1566
	struct rt6_info *rt = (struct rt6_info *)inet->cork.base.dst;
1567 1568
	struct flowi6 *fl6 = &inet->cork.fl.u.ip6;
	unsigned char proto = fl6->flowi6_proto;
L
Linus Torvalds 已提交
1569 1570 1571 1572 1573 1574 1575
	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 */
1576
	if (skb->data < skb_network_header(skb))
1577
		__skb_pull(skb, skb_network_offset(skb));
L
Linus Torvalds 已提交
1578
	while ((tmp_skb = __skb_dequeue(&sk->sk_write_queue)) != NULL) {
1579
		__skb_pull(tmp_skb, skb_network_header_len(skb));
L
Linus Torvalds 已提交
1580 1581 1582 1583 1584 1585 1586 1587 1588
		*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;
	}

1589
	/* Allow local fragmentation. */
1590
	if (np->pmtudisc < IPV6_PMTUDISC_DO)
1591 1592
		skb->local_df = 1;

1593
	ipv6_addr_copy(final_dst, &fl6->daddr);
1594
	__skb_pull(skb, skb_network_header_len(skb));
L
Linus Torvalds 已提交
1595 1596 1597 1598 1599
	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);

1600 1601
	skb_push(skb, sizeof(struct ipv6hdr));
	skb_reset_network_header(skb);
1602
	hdr = ipv6_hdr(skb);
1603

1604
	*(__be32*)hdr = fl6->flowlabel |
1605
		     htonl(0x60000000 | ((int)np->cork.tclass << 20));
L
Linus Torvalds 已提交
1606 1607 1608

	hdr->hop_limit = np->cork.hop_limit;
	hdr->nexthdr = proto;
1609
	ipv6_addr_copy(&hdr->saddr, &fl6->saddr);
L
Linus Torvalds 已提交
1610 1611
	ipv6_addr_copy(&hdr->daddr, final_dst);

1612
	skb->priority = sk->sk_priority;
1613
	skb->mark = sk->sk_mark;
1614

1615
	skb_dst_set(skb, dst_clone(&rt->dst));
1616
	IP6_UPD_PO_STATS(net, rt->rt6i_idev, IPSTATS_MIB_OUT, skb->len);
1617
	if (proto == IPPROTO_ICMPV6) {
E
Eric Dumazet 已提交
1618
		struct inet6_dev *idev = ip6_dst_idev(skb_dst(skb));
1619

1620
		ICMP6MSGOUT_INC_STATS_BH(net, idev, icmp6_hdr(skb)->icmp6_type);
1621
		ICMP6_INC_STATS_BH(net, idev, ICMP6_MIB_OUTMSGS);
1622 1623
	}

H
Herbert Xu 已提交
1624
	err = ip6_local_out(skb);
L
Linus Torvalds 已提交
1625 1626
	if (err) {
		if (err > 0)
E
Eric Dumazet 已提交
1627
			err = net_xmit_errno(err);
L
Linus Torvalds 已提交
1628 1629 1630 1631 1632
		if (err)
			goto error;
	}

out:
1633
	ip6_cork_release(inet, np);
L
Linus Torvalds 已提交
1634 1635
	return err;
error:
1636
	IP6_INC_STATS(net, rt->rt6i_idev, IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
1637 1638 1639 1640 1641 1642 1643 1644
	goto out;
}

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
1645 1646
		if (skb_dst(skb))
			IP6_INC_STATS(sock_net(sk), ip6_dst_idev(skb_dst(skb)),
1647
				      IPSTATS_MIB_OUTDISCARDS);
L
Linus Torvalds 已提交
1648 1649 1650
		kfree_skb(skb);
	}

1651
	ip6_cork_release(inet_sk(sk), inet6_sk(sk));
L
Linus Torvalds 已提交
1652
}