xref: /linux/net/ipv6/ip6_output.c (revision de2fe5e07d58424bc286fff3fd3c1b0bf933cd58)
1 /*
2  *	IPv6 output functions
3  *	Linux INET6 implementation
4  *
5  *	Authors:
6  *	Pedro Roque		<roque@di.fc.ul.pt>
7  *
8  *	$Id: ip6_output.c,v 1.34 2002/02/01 22:01:04 davem Exp $
9  *
10  *	Based on linux/net/ipv4/ip_output.c
11  *
12  *	This program is free software; you can redistribute it and/or
13  *      modify it under the terms of the GNU General Public License
14  *      as published by the Free Software Foundation; either version
15  *      2 of the License, or (at your option) any later version.
16  *
17  *	Changes:
18  *	A.N.Kuznetsov	:	airthmetics in fragmentation.
19  *				extension headers are implemented.
20  *				route changes now work.
21  *				ip6_forward does not confuse sniffers.
22  *				etc.
23  *
24  *      H. von Brand    :       Added missing #include <linux/string.h>
25  *	Imran Patel	: 	frag id should be in NBO
26  *      Kazunori MIYAZAWA @USAGI
27  *			:       add ip6_append_data and related functions
28  *				for datagram xmit
29  */
30 
31 #include <linux/config.h>
32 #include <linux/errno.h>
33 #include <linux/types.h>
34 #include <linux/string.h>
35 #include <linux/socket.h>
36 #include <linux/net.h>
37 #include <linux/netdevice.h>
38 #include <linux/if_arp.h>
39 #include <linux/in6.h>
40 #include <linux/tcp.h>
41 #include <linux/route.h>
42 
43 #include <linux/netfilter.h>
44 #include <linux/netfilter_ipv6.h>
45 
46 #include <net/sock.h>
47 #include <net/snmp.h>
48 
49 #include <net/ipv6.h>
50 #include <net/ndisc.h>
51 #include <net/protocol.h>
52 #include <net/ip6_route.h>
53 #include <net/addrconf.h>
54 #include <net/rawv6.h>
55 #include <net/icmp.h>
56 #include <net/xfrm.h>
57 #include <net/checksum.h>
58 
59 static int ip6_fragment(struct sk_buff *skb, int (*output)(struct sk_buff *));
60 
61 static __inline__ void ipv6_select_ident(struct sk_buff *skb, struct frag_hdr *fhdr)
62 {
63 	static u32 ipv6_fragmentation_id = 1;
64 	static DEFINE_SPINLOCK(ip6_id_lock);
65 
66 	spin_lock_bh(&ip6_id_lock);
67 	fhdr->identification = htonl(ipv6_fragmentation_id);
68 	if (++ipv6_fragmentation_id == 0)
69 		ipv6_fragmentation_id = 1;
70 	spin_unlock_bh(&ip6_id_lock);
71 }
72 
73 static inline int ip6_output_finish(struct sk_buff *skb)
74 {
75 
76 	struct dst_entry *dst = skb->dst;
77 	struct hh_cache *hh = dst->hh;
78 
79 	if (hh) {
80 		int hh_alen;
81 
82 		read_lock_bh(&hh->hh_lock);
83 		hh_alen = HH_DATA_ALIGN(hh->hh_len);
84 		memcpy(skb->data - hh_alen, hh->hh_data, hh_alen);
85 		read_unlock_bh(&hh->hh_lock);
86 	        skb_push(skb, hh->hh_len);
87 		return hh->hh_output(skb);
88 	} else if (dst->neighbour)
89 		return dst->neighbour->output(skb);
90 
91 	IP6_INC_STATS_BH(IPSTATS_MIB_OUTNOROUTES);
92 	kfree_skb(skb);
93 	return -EINVAL;
94 
95 }
96 
97 /* dev_loopback_xmit for use with netfilter. */
98 static int ip6_dev_loopback_xmit(struct sk_buff *newskb)
99 {
100 	newskb->mac.raw = newskb->data;
101 	__skb_pull(newskb, newskb->nh.raw - newskb->data);
102 	newskb->pkt_type = PACKET_LOOPBACK;
103 	newskb->ip_summed = CHECKSUM_UNNECESSARY;
104 	BUG_TRAP(newskb->dst);
105 
106 	netif_rx(newskb);
107 	return 0;
108 }
109 
110 
111 static int ip6_output2(struct sk_buff *skb)
112 {
113 	struct dst_entry *dst = skb->dst;
114 	struct net_device *dev = dst->dev;
115 
116 	skb->protocol = htons(ETH_P_IPV6);
117 	skb->dev = dev;
118 
119 	if (ipv6_addr_is_multicast(&skb->nh.ipv6h->daddr)) {
120 		struct ipv6_pinfo* np = skb->sk ? inet6_sk(skb->sk) : NULL;
121 
122 		if (!(dev->flags & IFF_LOOPBACK) && (!np || np->mc_loop) &&
123 		    ipv6_chk_mcast_addr(dev, &skb->nh.ipv6h->daddr,
124 				&skb->nh.ipv6h->saddr)) {
125 			struct sk_buff *newskb = skb_clone(skb, GFP_ATOMIC);
126 
127 			/* Do not check for IFF_ALLMULTI; multicast routing
128 			   is not supported in any case.
129 			 */
130 			if (newskb)
131 				NF_HOOK(PF_INET6, NF_IP6_POST_ROUTING, newskb, NULL,
132 					newskb->dev,
133 					ip6_dev_loopback_xmit);
134 
135 			if (skb->nh.ipv6h->hop_limit == 0) {
136 				IP6_INC_STATS(IPSTATS_MIB_OUTDISCARDS);
137 				kfree_skb(skb);
138 				return 0;
139 			}
140 		}
141 
142 		IP6_INC_STATS(IPSTATS_MIB_OUTMCASTPKTS);
143 	}
144 
145 	return NF_HOOK(PF_INET6, NF_IP6_POST_ROUTING, skb,NULL, skb->dev,ip6_output_finish);
146 }
147 
148 int ip6_output(struct sk_buff *skb)
149 {
150 	if ((skb->len > dst_mtu(skb->dst) && !skb_shinfo(skb)->ufo_size) ||
151 				dst_allfrag(skb->dst))
152 		return ip6_fragment(skb, ip6_output2);
153 	else
154 		return ip6_output2(skb);
155 }
156 
157 /*
158  *	xmit an sk_buff (used by TCP)
159  */
160 
161 int ip6_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl,
162 	     struct ipv6_txoptions *opt, int ipfragok)
163 {
164 	struct ipv6_pinfo *np = inet6_sk(sk);
165 	struct in6_addr *first_hop = &fl->fl6_dst;
166 	struct dst_entry *dst = skb->dst;
167 	struct ipv6hdr *hdr;
168 	u8  proto = fl->proto;
169 	int seg_len = skb->len;
170 	int hlimit, tclass;
171 	u32 mtu;
172 
173 	if (opt) {
174 		int head_room;
175 
176 		/* First: exthdrs may take lots of space (~8K for now)
177 		   MAX_HEADER is not enough.
178 		 */
179 		head_room = opt->opt_nflen + opt->opt_flen;
180 		seg_len += head_room;
181 		head_room += sizeof(struct ipv6hdr) + LL_RESERVED_SPACE(dst->dev);
182 
183 		if (skb_headroom(skb) < head_room) {
184 			struct sk_buff *skb2 = skb_realloc_headroom(skb, head_room);
185 			kfree_skb(skb);
186 			skb = skb2;
187 			if (skb == NULL) {
188 				IP6_INC_STATS(IPSTATS_MIB_OUTDISCARDS);
189 				return -ENOBUFS;
190 			}
191 			if (sk)
192 				skb_set_owner_w(skb, sk);
193 		}
194 		if (opt->opt_flen)
195 			ipv6_push_frag_opts(skb, opt, &proto);
196 		if (opt->opt_nflen)
197 			ipv6_push_nfrag_opts(skb, opt, &proto, &first_hop);
198 	}
199 
200 	hdr = skb->nh.ipv6h = (struct ipv6hdr*)skb_push(skb, sizeof(struct ipv6hdr));
201 
202 	/*
203 	 *	Fill in the IPv6 header
204 	 */
205 
206 	hlimit = -1;
207 	if (np)
208 		hlimit = np->hop_limit;
209 	if (hlimit < 0)
210 		hlimit = dst_metric(dst, RTAX_HOPLIMIT);
211 	if (hlimit < 0)
212 		hlimit = ipv6_get_hoplimit(dst->dev);
213 
214 	tclass = -1;
215 	if (np)
216 		tclass = np->tclass;
217 	if (tclass < 0)
218 		tclass = 0;
219 
220 	*(u32 *)hdr = htonl(0x60000000 | (tclass << 20)) | fl->fl6_flowlabel;
221 
222 	hdr->payload_len = htons(seg_len);
223 	hdr->nexthdr = proto;
224 	hdr->hop_limit = hlimit;
225 
226 	ipv6_addr_copy(&hdr->saddr, &fl->fl6_src);
227 	ipv6_addr_copy(&hdr->daddr, first_hop);
228 
229 	skb->priority = sk->sk_priority;
230 
231 	mtu = dst_mtu(dst);
232 	if ((skb->len <= mtu) || ipfragok) {
233 		IP6_INC_STATS(IPSTATS_MIB_OUTREQUESTS);
234 		return NF_HOOK(PF_INET6, NF_IP6_LOCAL_OUT, skb, NULL, dst->dev,
235 				dst_output);
236 	}
237 
238 	if (net_ratelimit())
239 		printk(KERN_DEBUG "IPv6: sending pkt_too_big to self\n");
240 	skb->dev = dst->dev;
241 	icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu, skb->dev);
242 	IP6_INC_STATS(IPSTATS_MIB_FRAGFAILS);
243 	kfree_skb(skb);
244 	return -EMSGSIZE;
245 }
246 
247 /*
248  *	To avoid extra problems ND packets are send through this
249  *	routine. It's code duplication but I really want to avoid
250  *	extra checks since ipv6_build_header is used by TCP (which
251  *	is for us performance critical)
252  */
253 
254 int ip6_nd_hdr(struct sock *sk, struct sk_buff *skb, struct net_device *dev,
255 	       struct in6_addr *saddr, struct in6_addr *daddr,
256 	       int proto, int len)
257 {
258 	struct ipv6_pinfo *np = inet6_sk(sk);
259 	struct ipv6hdr *hdr;
260 	int totlen;
261 
262 	skb->protocol = htons(ETH_P_IPV6);
263 	skb->dev = dev;
264 
265 	totlen = len + sizeof(struct ipv6hdr);
266 
267 	hdr = (struct ipv6hdr *) skb_put(skb, sizeof(struct ipv6hdr));
268 	skb->nh.ipv6h = hdr;
269 
270 	*(u32*)hdr = htonl(0x60000000);
271 
272 	hdr->payload_len = htons(len);
273 	hdr->nexthdr = proto;
274 	hdr->hop_limit = np->hop_limit;
275 
276 	ipv6_addr_copy(&hdr->saddr, saddr);
277 	ipv6_addr_copy(&hdr->daddr, daddr);
278 
279 	return 0;
280 }
281 
282 static int ip6_call_ra_chain(struct sk_buff *skb, int sel)
283 {
284 	struct ip6_ra_chain *ra;
285 	struct sock *last = NULL;
286 
287 	read_lock(&ip6_ra_lock);
288 	for (ra = ip6_ra_chain; ra; ra = ra->next) {
289 		struct sock *sk = ra->sk;
290 		if (sk && ra->sel == sel &&
291 		    (!sk->sk_bound_dev_if ||
292 		     sk->sk_bound_dev_if == skb->dev->ifindex)) {
293 			if (last) {
294 				struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
295 				if (skb2)
296 					rawv6_rcv(last, skb2);
297 			}
298 			last = sk;
299 		}
300 	}
301 
302 	if (last) {
303 		rawv6_rcv(last, skb);
304 		read_unlock(&ip6_ra_lock);
305 		return 1;
306 	}
307 	read_unlock(&ip6_ra_lock);
308 	return 0;
309 }
310 
311 static inline int ip6_forward_finish(struct sk_buff *skb)
312 {
313 	return dst_output(skb);
314 }
315 
316 int ip6_forward(struct sk_buff *skb)
317 {
318 	struct dst_entry *dst = skb->dst;
319 	struct ipv6hdr *hdr = skb->nh.ipv6h;
320 	struct inet6_skb_parm *opt = IP6CB(skb);
321 
322 	if (ipv6_devconf.forwarding == 0)
323 		goto error;
324 
325 	if (!xfrm6_policy_check(NULL, XFRM_POLICY_FWD, skb)) {
326 		IP6_INC_STATS(IPSTATS_MIB_INDISCARDS);
327 		goto drop;
328 	}
329 
330 	skb->ip_summed = CHECKSUM_NONE;
331 
332 	/*
333 	 *	We DO NOT make any processing on
334 	 *	RA packets, pushing them to user level AS IS
335 	 *	without ane WARRANTY that application will be able
336 	 *	to interpret them. The reason is that we
337 	 *	cannot make anything clever here.
338 	 *
339 	 *	We are not end-node, so that if packet contains
340 	 *	AH/ESP, we cannot make anything.
341 	 *	Defragmentation also would be mistake, RA packets
342 	 *	cannot be fragmented, because there is no warranty
343 	 *	that different fragments will go along one path. --ANK
344 	 */
345 	if (opt->ra) {
346 		u8 *ptr = skb->nh.raw + opt->ra;
347 		if (ip6_call_ra_chain(skb, (ptr[2]<<8) + ptr[3]))
348 			return 0;
349 	}
350 
351 	/*
352 	 *	check and decrement ttl
353 	 */
354 	if (hdr->hop_limit <= 1) {
355 		/* Force OUTPUT device used as source address */
356 		skb->dev = dst->dev;
357 		icmpv6_send(skb, ICMPV6_TIME_EXCEED, ICMPV6_EXC_HOPLIMIT,
358 			    0, skb->dev);
359 
360 		kfree_skb(skb);
361 		return -ETIMEDOUT;
362 	}
363 
364 	if (!xfrm6_route_forward(skb)) {
365 		IP6_INC_STATS(IPSTATS_MIB_INDISCARDS);
366 		goto drop;
367 	}
368 	dst = skb->dst;
369 
370 	/* IPv6 specs say nothing about it, but it is clear that we cannot
371 	   send redirects to source routed frames.
372 	 */
373 	if (skb->dev == dst->dev && dst->neighbour && opt->srcrt == 0) {
374 		struct in6_addr *target = NULL;
375 		struct rt6_info *rt;
376 		struct neighbour *n = dst->neighbour;
377 
378 		/*
379 		 *	incoming and outgoing devices are the same
380 		 *	send a redirect.
381 		 */
382 
383 		rt = (struct rt6_info *) dst;
384 		if ((rt->rt6i_flags & RTF_GATEWAY))
385 			target = (struct in6_addr*)&n->primary_key;
386 		else
387 			target = &hdr->daddr;
388 
389 		/* Limit redirects both by destination (here)
390 		   and by source (inside ndisc_send_redirect)
391 		 */
392 		if (xrlim_allow(dst, 1*HZ))
393 			ndisc_send_redirect(skb, n, target);
394 	} else if (ipv6_addr_type(&hdr->saddr)&(IPV6_ADDR_MULTICAST|IPV6_ADDR_LOOPBACK
395 						|IPV6_ADDR_LINKLOCAL)) {
396 		/* This check is security critical. */
397 		goto error;
398 	}
399 
400 	if (skb->len > dst_mtu(dst)) {
401 		/* Again, force OUTPUT device used as source address */
402 		skb->dev = dst->dev;
403 		icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, dst_mtu(dst), skb->dev);
404 		IP6_INC_STATS_BH(IPSTATS_MIB_INTOOBIGERRORS);
405 		IP6_INC_STATS_BH(IPSTATS_MIB_FRAGFAILS);
406 		kfree_skb(skb);
407 		return -EMSGSIZE;
408 	}
409 
410 	if (skb_cow(skb, dst->dev->hard_header_len)) {
411 		IP6_INC_STATS(IPSTATS_MIB_OUTDISCARDS);
412 		goto drop;
413 	}
414 
415 	hdr = skb->nh.ipv6h;
416 
417 	/* Mangling hops number delayed to point after skb COW */
418 
419 	hdr->hop_limit--;
420 
421 	IP6_INC_STATS_BH(IPSTATS_MIB_OUTFORWDATAGRAMS);
422 	return NF_HOOK(PF_INET6,NF_IP6_FORWARD, skb, skb->dev, dst->dev, ip6_forward_finish);
423 
424 error:
425 	IP6_INC_STATS_BH(IPSTATS_MIB_INADDRERRORS);
426 drop:
427 	kfree_skb(skb);
428 	return -EINVAL;
429 }
430 
431 static void ip6_copy_metadata(struct sk_buff *to, struct sk_buff *from)
432 {
433 	to->pkt_type = from->pkt_type;
434 	to->priority = from->priority;
435 	to->protocol = from->protocol;
436 	dst_release(to->dst);
437 	to->dst = dst_clone(from->dst);
438 	to->dev = from->dev;
439 
440 #ifdef CONFIG_NET_SCHED
441 	to->tc_index = from->tc_index;
442 #endif
443 #ifdef CONFIG_NETFILTER
444 	to->nfmark = from->nfmark;
445 	/* Connection association is same as pre-frag packet */
446 	nf_conntrack_put(to->nfct);
447 	to->nfct = from->nfct;
448 	nf_conntrack_get(to->nfct);
449 	to->nfctinfo = from->nfctinfo;
450 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
451 	nf_conntrack_put_reasm(to->nfct_reasm);
452 	to->nfct_reasm = from->nfct_reasm;
453 	nf_conntrack_get_reasm(to->nfct_reasm);
454 #endif
455 #ifdef CONFIG_BRIDGE_NETFILTER
456 	nf_bridge_put(to->nf_bridge);
457 	to->nf_bridge = from->nf_bridge;
458 	nf_bridge_get(to->nf_bridge);
459 #endif
460 #endif
461 }
462 
463 int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr)
464 {
465 	u16 offset = sizeof(struct ipv6hdr);
466 	struct ipv6_opt_hdr *exthdr = (struct ipv6_opt_hdr*)(skb->nh.ipv6h + 1);
467 	unsigned int packet_len = skb->tail - skb->nh.raw;
468 	int found_rhdr = 0;
469 	*nexthdr = &skb->nh.ipv6h->nexthdr;
470 
471 	while (offset + 1 <= packet_len) {
472 
473 		switch (**nexthdr) {
474 
475 		case NEXTHDR_HOP:
476 		case NEXTHDR_ROUTING:
477 		case NEXTHDR_DEST:
478 			if (**nexthdr == NEXTHDR_ROUTING) found_rhdr = 1;
479 			if (**nexthdr == NEXTHDR_DEST && found_rhdr) return offset;
480 			offset += ipv6_optlen(exthdr);
481 			*nexthdr = &exthdr->nexthdr;
482 			exthdr = (struct ipv6_opt_hdr*)(skb->nh.raw + offset);
483 			break;
484 		default :
485 			return offset;
486 		}
487 	}
488 
489 	return offset;
490 }
491 
492 static int ip6_fragment(struct sk_buff *skb, int (*output)(struct sk_buff *))
493 {
494 	struct net_device *dev;
495 	struct sk_buff *frag;
496 	struct rt6_info *rt = (struct rt6_info*)skb->dst;
497 	struct ipv6_pinfo *np = skb->sk ? inet6_sk(skb->sk) : NULL;
498 	struct ipv6hdr *tmp_hdr;
499 	struct frag_hdr *fh;
500 	unsigned int mtu, hlen, left, len;
501 	u32 frag_id = 0;
502 	int ptr, offset = 0, err=0;
503 	u8 *prevhdr, nexthdr = 0;
504 
505 	dev = rt->u.dst.dev;
506 	hlen = ip6_find_1stfragopt(skb, &prevhdr);
507 	nexthdr = *prevhdr;
508 
509 	mtu = dst_mtu(&rt->u.dst);
510 	if (np && np->frag_size < mtu) {
511 		if (np->frag_size)
512 			mtu = np->frag_size;
513 	}
514 	mtu -= hlen + sizeof(struct frag_hdr);
515 
516 	if (skb_shinfo(skb)->frag_list) {
517 		int first_len = skb_pagelen(skb);
518 
519 		if (first_len - hlen > mtu ||
520 		    ((first_len - hlen) & 7) ||
521 		    skb_cloned(skb))
522 			goto slow_path;
523 
524 		for (frag = skb_shinfo(skb)->frag_list; frag; frag = frag->next) {
525 			/* Correct geometry. */
526 			if (frag->len > mtu ||
527 			    ((frag->len & 7) && frag->next) ||
528 			    skb_headroom(frag) < hlen)
529 			    goto slow_path;
530 
531 			/* Partially cloned skb? */
532 			if (skb_shared(frag))
533 				goto slow_path;
534 
535 			BUG_ON(frag->sk);
536 			if (skb->sk) {
537 				sock_hold(skb->sk);
538 				frag->sk = skb->sk;
539 				frag->destructor = sock_wfree;
540 				skb->truesize -= frag->truesize;
541 			}
542 		}
543 
544 		err = 0;
545 		offset = 0;
546 		frag = skb_shinfo(skb)->frag_list;
547 		skb_shinfo(skb)->frag_list = NULL;
548 		/* BUILD HEADER */
549 
550 		tmp_hdr = kmalloc(hlen, GFP_ATOMIC);
551 		if (!tmp_hdr) {
552 			IP6_INC_STATS(IPSTATS_MIB_FRAGFAILS);
553 			return -ENOMEM;
554 		}
555 
556 		*prevhdr = NEXTHDR_FRAGMENT;
557 		memcpy(tmp_hdr, skb->nh.raw, hlen);
558 		__skb_pull(skb, hlen);
559 		fh = (struct frag_hdr*)__skb_push(skb, sizeof(struct frag_hdr));
560 		skb->nh.raw = __skb_push(skb, hlen);
561 		memcpy(skb->nh.raw, tmp_hdr, hlen);
562 
563 		ipv6_select_ident(skb, fh);
564 		fh->nexthdr = nexthdr;
565 		fh->reserved = 0;
566 		fh->frag_off = htons(IP6_MF);
567 		frag_id = fh->identification;
568 
569 		first_len = skb_pagelen(skb);
570 		skb->data_len = first_len - skb_headlen(skb);
571 		skb->len = first_len;
572 		skb->nh.ipv6h->payload_len = htons(first_len - sizeof(struct ipv6hdr));
573 
574 
575 		for (;;) {
576 			/* Prepare header of the next frame,
577 			 * before previous one went down. */
578 			if (frag) {
579 				frag->ip_summed = CHECKSUM_NONE;
580 				frag->h.raw = frag->data;
581 				fh = (struct frag_hdr*)__skb_push(frag, sizeof(struct frag_hdr));
582 				frag->nh.raw = __skb_push(frag, hlen);
583 				memcpy(frag->nh.raw, tmp_hdr, hlen);
584 				offset += skb->len - hlen - sizeof(struct frag_hdr);
585 				fh->nexthdr = nexthdr;
586 				fh->reserved = 0;
587 				fh->frag_off = htons(offset);
588 				if (frag->next != NULL)
589 					fh->frag_off |= htons(IP6_MF);
590 				fh->identification = frag_id;
591 				frag->nh.ipv6h->payload_len = htons(frag->len - sizeof(struct ipv6hdr));
592 				ip6_copy_metadata(frag, skb);
593 			}
594 
595 			err = output(skb);
596 			if (err || !frag)
597 				break;
598 
599 			skb = frag;
600 			frag = skb->next;
601 			skb->next = NULL;
602 		}
603 
604 		kfree(tmp_hdr);
605 
606 		if (err == 0) {
607 			IP6_INC_STATS(IPSTATS_MIB_FRAGOKS);
608 			return 0;
609 		}
610 
611 		while (frag) {
612 			skb = frag->next;
613 			kfree_skb(frag);
614 			frag = skb;
615 		}
616 
617 		IP6_INC_STATS(IPSTATS_MIB_FRAGFAILS);
618 		return err;
619 	}
620 
621 slow_path:
622 	left = skb->len - hlen;		/* Space per frame */
623 	ptr = hlen;			/* Where to start from */
624 
625 	/*
626 	 *	Fragment the datagram.
627 	 */
628 
629 	*prevhdr = NEXTHDR_FRAGMENT;
630 
631 	/*
632 	 *	Keep copying data until we run out.
633 	 */
634 	while(left > 0)	{
635 		len = left;
636 		/* IF: it doesn't fit, use 'mtu' - the data space left */
637 		if (len > mtu)
638 			len = mtu;
639 		/* IF: we are not sending upto and including the packet end
640 		   then align the next start on an eight byte boundary */
641 		if (len < left)	{
642 			len &= ~7;
643 		}
644 		/*
645 		 *	Allocate buffer.
646 		 */
647 
648 		if ((frag = alloc_skb(len+hlen+sizeof(struct frag_hdr)+LL_RESERVED_SPACE(rt->u.dst.dev), GFP_ATOMIC)) == NULL) {
649 			NETDEBUG(KERN_INFO "IPv6: frag: no memory for new fragment!\n");
650 			IP6_INC_STATS(IPSTATS_MIB_FRAGFAILS);
651 			err = -ENOMEM;
652 			goto fail;
653 		}
654 
655 		/*
656 		 *	Set up data on packet
657 		 */
658 
659 		ip6_copy_metadata(frag, skb);
660 		skb_reserve(frag, LL_RESERVED_SPACE(rt->u.dst.dev));
661 		skb_put(frag, len + hlen + sizeof(struct frag_hdr));
662 		frag->nh.raw = frag->data;
663 		fh = (struct frag_hdr*)(frag->data + hlen);
664 		frag->h.raw = frag->data + hlen + sizeof(struct frag_hdr);
665 
666 		/*
667 		 *	Charge the memory for the fragment to any owner
668 		 *	it might possess
669 		 */
670 		if (skb->sk)
671 			skb_set_owner_w(frag, skb->sk);
672 
673 		/*
674 		 *	Copy the packet header into the new buffer.
675 		 */
676 		memcpy(frag->nh.raw, skb->data, hlen);
677 
678 		/*
679 		 *	Build fragment header.
680 		 */
681 		fh->nexthdr = nexthdr;
682 		fh->reserved = 0;
683 		if (!frag_id) {
684 			ipv6_select_ident(skb, fh);
685 			frag_id = fh->identification;
686 		} else
687 			fh->identification = frag_id;
688 
689 		/*
690 		 *	Copy a block of the IP datagram.
691 		 */
692 		if (skb_copy_bits(skb, ptr, frag->h.raw, len))
693 			BUG();
694 		left -= len;
695 
696 		fh->frag_off = htons(offset);
697 		if (left > 0)
698 			fh->frag_off |= htons(IP6_MF);
699 		frag->nh.ipv6h->payload_len = htons(frag->len - sizeof(struct ipv6hdr));
700 
701 		ptr += len;
702 		offset += len;
703 
704 		/*
705 		 *	Put this fragment into the sending queue.
706 		 */
707 
708 		IP6_INC_STATS(IPSTATS_MIB_FRAGCREATES);
709 
710 		err = output(frag);
711 		if (err)
712 			goto fail;
713 	}
714 	kfree_skb(skb);
715 	IP6_INC_STATS(IPSTATS_MIB_FRAGOKS);
716 	return err;
717 
718 fail:
719 	kfree_skb(skb);
720 	IP6_INC_STATS(IPSTATS_MIB_FRAGFAILS);
721 	return err;
722 }
723 
724 int ip6_dst_lookup(struct sock *sk, struct dst_entry **dst, struct flowi *fl)
725 {
726 	int err = 0;
727 
728 	*dst = NULL;
729 	if (sk) {
730 		struct ipv6_pinfo *np = inet6_sk(sk);
731 
732 		*dst = sk_dst_check(sk, np->dst_cookie);
733 		if (*dst) {
734 			struct rt6_info *rt = (struct rt6_info*)*dst;
735 
736 			/* Yes, checking route validity in not connected
737 			 * case is not very simple. Take into account,
738 			 * that we do not support routing by source, TOS,
739 			 * and MSG_DONTROUTE 		--ANK (980726)
740 			 *
741 			 * 1. If route was host route, check that
742 			 *    cached destination is current.
743 			 *    If it is network route, we still may
744 			 *    check its validity using saved pointer
745 			 *    to the last used address: daddr_cache.
746 			 *    We do not want to save whole address now,
747 			 *    (because main consumer of this service
748 			 *    is tcp, which has not this problem),
749 			 *    so that the last trick works only on connected
750 			 *    sockets.
751 			 * 2. oif also should be the same.
752 			 */
753 			if (((rt->rt6i_dst.plen != 128 ||
754 			      !ipv6_addr_equal(&fl->fl6_dst,
755 					       &rt->rt6i_dst.addr))
756 			     && (np->daddr_cache == NULL ||
757 				 !ipv6_addr_equal(&fl->fl6_dst,
758 						  np->daddr_cache)))
759 			    || (fl->oif && fl->oif != (*dst)->dev->ifindex)) {
760 				dst_release(*dst);
761 				*dst = NULL;
762 			}
763 		}
764 	}
765 
766 	if (*dst == NULL)
767 		*dst = ip6_route_output(sk, fl);
768 
769 	if ((err = (*dst)->error))
770 		goto out_err_release;
771 
772 	if (ipv6_addr_any(&fl->fl6_src)) {
773 		err = ipv6_get_saddr(*dst, &fl->fl6_dst, &fl->fl6_src);
774 
775 		if (err)
776 			goto out_err_release;
777 	}
778 
779 	return 0;
780 
781 out_err_release:
782 	dst_release(*dst);
783 	*dst = NULL;
784 	return err;
785 }
786 
787 EXPORT_SYMBOL_GPL(ip6_dst_lookup);
788 
789 static inline int ip6_ufo_append_data(struct sock *sk,
790 			int getfrag(void *from, char *to, int offset, int len,
791 			int odd, struct sk_buff *skb),
792 			void *from, int length, int hh_len, int fragheaderlen,
793 			int transhdrlen, int mtu,unsigned int flags)
794 
795 {
796 	struct sk_buff *skb;
797 	int err;
798 
799 	/* There is support for UDP large send offload by network
800 	 * device, so create one single skb packet containing complete
801 	 * udp datagram
802 	 */
803 	if ((skb = skb_peek_tail(&sk->sk_write_queue)) == NULL) {
804 		skb = sock_alloc_send_skb(sk,
805 			hh_len + fragheaderlen + transhdrlen + 20,
806 			(flags & MSG_DONTWAIT), &err);
807 		if (skb == NULL)
808 			return -ENOMEM;
809 
810 		/* reserve space for Hardware header */
811 		skb_reserve(skb, hh_len);
812 
813 		/* create space for UDP/IP header */
814 		skb_put(skb,fragheaderlen + transhdrlen);
815 
816 		/* initialize network header pointer */
817 		skb->nh.raw = skb->data;
818 
819 		/* initialize protocol header pointer */
820 		skb->h.raw = skb->data + fragheaderlen;
821 
822 		skb->ip_summed = CHECKSUM_HW;
823 		skb->csum = 0;
824 		sk->sk_sndmsg_off = 0;
825 	}
826 
827 	err = skb_append_datato_frags(sk,skb, getfrag, from,
828 				      (length - transhdrlen));
829 	if (!err) {
830 		struct frag_hdr fhdr;
831 
832 		/* specify the length of each IP datagram fragment*/
833 		skb_shinfo(skb)->ufo_size = (mtu - fragheaderlen) -
834 						sizeof(struct frag_hdr);
835 		ipv6_select_ident(skb, &fhdr);
836 		skb_shinfo(skb)->ip6_frag_id = fhdr.identification;
837 		__skb_queue_tail(&sk->sk_write_queue, skb);
838 
839 		return 0;
840 	}
841 	/* There is not enough support do UPD LSO,
842 	 * so follow normal path
843 	 */
844 	kfree_skb(skb);
845 
846 	return err;
847 }
848 
849 int ip6_append_data(struct sock *sk, int getfrag(void *from, char *to,
850 	int offset, int len, int odd, struct sk_buff *skb),
851 	void *from, int length, int transhdrlen,
852 	int hlimit, int tclass, struct ipv6_txoptions *opt, struct flowi *fl,
853 	struct rt6_info *rt, unsigned int flags)
854 {
855 	struct inet_sock *inet = inet_sk(sk);
856 	struct ipv6_pinfo *np = inet6_sk(sk);
857 	struct sk_buff *skb;
858 	unsigned int maxfraglen, fragheaderlen;
859 	int exthdrlen;
860 	int hh_len;
861 	int mtu;
862 	int copy;
863 	int err;
864 	int offset = 0;
865 	int csummode = CHECKSUM_NONE;
866 
867 	if (flags&MSG_PROBE)
868 		return 0;
869 	if (skb_queue_empty(&sk->sk_write_queue)) {
870 		/*
871 		 * setup for corking
872 		 */
873 		if (opt) {
874 			if (np->cork.opt == NULL) {
875 				np->cork.opt = kmalloc(opt->tot_len,
876 						       sk->sk_allocation);
877 				if (unlikely(np->cork.opt == NULL))
878 					return -ENOBUFS;
879 			} else if (np->cork.opt->tot_len < opt->tot_len) {
880 				printk(KERN_DEBUG "ip6_append_data: invalid option length\n");
881 				return -EINVAL;
882 			}
883 			memcpy(np->cork.opt, opt, opt->tot_len);
884 			inet->cork.flags |= IPCORK_OPT;
885 			/* need source address above miyazawa*/
886 		}
887 		dst_hold(&rt->u.dst);
888 		np->cork.rt = rt;
889 		inet->cork.fl = *fl;
890 		np->cork.hop_limit = hlimit;
891 		np->cork.tclass = tclass;
892 		mtu = dst_mtu(rt->u.dst.path);
893 		if (np->frag_size < mtu) {
894 			if (np->frag_size)
895 				mtu = np->frag_size;
896 		}
897 		inet->cork.fragsize = mtu;
898 		if (dst_allfrag(rt->u.dst.path))
899 			inet->cork.flags |= IPCORK_ALLFRAG;
900 		inet->cork.length = 0;
901 		sk->sk_sndmsg_page = NULL;
902 		sk->sk_sndmsg_off = 0;
903 		exthdrlen = rt->u.dst.header_len + (opt ? opt->opt_flen : 0);
904 		length += exthdrlen;
905 		transhdrlen += exthdrlen;
906 	} else {
907 		rt = np->cork.rt;
908 		fl = &inet->cork.fl;
909 		if (inet->cork.flags & IPCORK_OPT)
910 			opt = np->cork.opt;
911 		transhdrlen = 0;
912 		exthdrlen = 0;
913 		mtu = inet->cork.fragsize;
914 	}
915 
916 	hh_len = LL_RESERVED_SPACE(rt->u.dst.dev);
917 
918 	fragheaderlen = sizeof(struct ipv6hdr) + (opt ? opt->opt_nflen : 0);
919 	maxfraglen = ((mtu - fragheaderlen) & ~7) + fragheaderlen - sizeof(struct frag_hdr);
920 
921 	if (mtu <= sizeof(struct ipv6hdr) + IPV6_MAXPLEN) {
922 		if (inet->cork.length + length > sizeof(struct ipv6hdr) + IPV6_MAXPLEN - fragheaderlen) {
923 			ipv6_local_error(sk, EMSGSIZE, fl, mtu-exthdrlen);
924 			return -EMSGSIZE;
925 		}
926 	}
927 
928 	/*
929 	 * Let's try using as much space as possible.
930 	 * Use MTU if total length of the message fits into the MTU.
931 	 * Otherwise, we need to reserve fragment header and
932 	 * fragment alignment (= 8-15 octects, in total).
933 	 *
934 	 * Note that we may need to "move" the data from the tail of
935 	 * of the buffer to the new fragment when we split
936 	 * the message.
937 	 *
938 	 * FIXME: It may be fragmented into multiple chunks
939 	 *        at once if non-fragmentable extension headers
940 	 *        are too large.
941 	 * --yoshfuji
942 	 */
943 
944 	inet->cork.length += length;
945 	if (((length > mtu) && (sk->sk_protocol == IPPROTO_UDP)) &&
946 	    (rt->u.dst.dev->features & NETIF_F_UFO)) {
947 
948 		err = ip6_ufo_append_data(sk, getfrag, from, length, hh_len,
949 					  fragheaderlen, transhdrlen, mtu,
950 					  flags);
951 		if (err)
952 			goto error;
953 		return 0;
954 	}
955 
956 	if ((skb = skb_peek_tail(&sk->sk_write_queue)) == NULL)
957 		goto alloc_new_skb;
958 
959 	while (length > 0) {
960 		/* Check if the remaining data fits into current packet. */
961 		copy = (inet->cork.length <= mtu && !(inet->cork.flags & IPCORK_ALLFRAG) ? mtu : maxfraglen) - skb->len;
962 		if (copy < length)
963 			copy = maxfraglen - skb->len;
964 
965 		if (copy <= 0) {
966 			char *data;
967 			unsigned int datalen;
968 			unsigned int fraglen;
969 			unsigned int fraggap;
970 			unsigned int alloclen;
971 			struct sk_buff *skb_prev;
972 alloc_new_skb:
973 			skb_prev = skb;
974 
975 			/* There's no room in the current skb */
976 			if (skb_prev)
977 				fraggap = skb_prev->len - maxfraglen;
978 			else
979 				fraggap = 0;
980 
981 			/*
982 			 * If remaining data exceeds the mtu,
983 			 * we know we need more fragment(s).
984 			 */
985 			datalen = length + fraggap;
986 			if (datalen > (inet->cork.length <= mtu && !(inet->cork.flags & IPCORK_ALLFRAG) ? mtu : maxfraglen) - fragheaderlen)
987 				datalen = maxfraglen - fragheaderlen;
988 
989 			fraglen = datalen + fragheaderlen;
990 			if ((flags & MSG_MORE) &&
991 			    !(rt->u.dst.dev->features&NETIF_F_SG))
992 				alloclen = mtu;
993 			else
994 				alloclen = datalen + fragheaderlen;
995 
996 			/*
997 			 * The last fragment gets additional space at tail.
998 			 * Note: we overallocate on fragments with MSG_MODE
999 			 * because we have no idea if we're the last one.
1000 			 */
1001 			if (datalen == length + fraggap)
1002 				alloclen += rt->u.dst.trailer_len;
1003 
1004 			/*
1005 			 * We just reserve space for fragment header.
1006 			 * Note: this may be overallocation if the message
1007 			 * (without MSG_MORE) fits into the MTU.
1008 			 */
1009 			alloclen += sizeof(struct frag_hdr);
1010 
1011 			if (transhdrlen) {
1012 				skb = sock_alloc_send_skb(sk,
1013 						alloclen + hh_len,
1014 						(flags & MSG_DONTWAIT), &err);
1015 			} else {
1016 				skb = NULL;
1017 				if (atomic_read(&sk->sk_wmem_alloc) <=
1018 				    2 * sk->sk_sndbuf)
1019 					skb = sock_wmalloc(sk,
1020 							   alloclen + hh_len, 1,
1021 							   sk->sk_allocation);
1022 				if (unlikely(skb == NULL))
1023 					err = -ENOBUFS;
1024 			}
1025 			if (skb == NULL)
1026 				goto error;
1027 			/*
1028 			 *	Fill in the control structures
1029 			 */
1030 			skb->ip_summed = csummode;
1031 			skb->csum = 0;
1032 			/* reserve for fragmentation */
1033 			skb_reserve(skb, hh_len+sizeof(struct frag_hdr));
1034 
1035 			/*
1036 			 *	Find where to start putting bytes
1037 			 */
1038 			data = skb_put(skb, fraglen);
1039 			skb->nh.raw = data + exthdrlen;
1040 			data += fragheaderlen;
1041 			skb->h.raw = data + exthdrlen;
1042 
1043 			if (fraggap) {
1044 				skb->csum = skb_copy_and_csum_bits(
1045 					skb_prev, maxfraglen,
1046 					data + transhdrlen, fraggap, 0);
1047 				skb_prev->csum = csum_sub(skb_prev->csum,
1048 							  skb->csum);
1049 				data += fraggap;
1050 				skb_trim(skb_prev, maxfraglen);
1051 			}
1052 			copy = datalen - transhdrlen - fraggap;
1053 			if (copy < 0) {
1054 				err = -EINVAL;
1055 				kfree_skb(skb);
1056 				goto error;
1057 			} else if (copy > 0 && getfrag(from, data + transhdrlen, offset, copy, fraggap, skb) < 0) {
1058 				err = -EFAULT;
1059 				kfree_skb(skb);
1060 				goto error;
1061 			}
1062 
1063 			offset += copy;
1064 			length -= datalen - fraggap;
1065 			transhdrlen = 0;
1066 			exthdrlen = 0;
1067 			csummode = CHECKSUM_NONE;
1068 
1069 			/*
1070 			 * Put the packet on the pending queue
1071 			 */
1072 			__skb_queue_tail(&sk->sk_write_queue, skb);
1073 			continue;
1074 		}
1075 
1076 		if (copy > length)
1077 			copy = length;
1078 
1079 		if (!(rt->u.dst.dev->features&NETIF_F_SG)) {
1080 			unsigned int off;
1081 
1082 			off = skb->len;
1083 			if (getfrag(from, skb_put(skb, copy),
1084 						offset, copy, off, skb) < 0) {
1085 				__skb_trim(skb, off);
1086 				err = -EFAULT;
1087 				goto error;
1088 			}
1089 		} else {
1090 			int i = skb_shinfo(skb)->nr_frags;
1091 			skb_frag_t *frag = &skb_shinfo(skb)->frags[i-1];
1092 			struct page *page = sk->sk_sndmsg_page;
1093 			int off = sk->sk_sndmsg_off;
1094 			unsigned int left;
1095 
1096 			if (page && (left = PAGE_SIZE - off) > 0) {
1097 				if (copy >= left)
1098 					copy = left;
1099 				if (page != frag->page) {
1100 					if (i == MAX_SKB_FRAGS) {
1101 						err = -EMSGSIZE;
1102 						goto error;
1103 					}
1104 					get_page(page);
1105 					skb_fill_page_desc(skb, i, page, sk->sk_sndmsg_off, 0);
1106 					frag = &skb_shinfo(skb)->frags[i];
1107 				}
1108 			} else if(i < MAX_SKB_FRAGS) {
1109 				if (copy > PAGE_SIZE)
1110 					copy = PAGE_SIZE;
1111 				page = alloc_pages(sk->sk_allocation, 0);
1112 				if (page == NULL) {
1113 					err = -ENOMEM;
1114 					goto error;
1115 				}
1116 				sk->sk_sndmsg_page = page;
1117 				sk->sk_sndmsg_off = 0;
1118 
1119 				skb_fill_page_desc(skb, i, page, 0, 0);
1120 				frag = &skb_shinfo(skb)->frags[i];
1121 				skb->truesize += PAGE_SIZE;
1122 				atomic_add(PAGE_SIZE, &sk->sk_wmem_alloc);
1123 			} else {
1124 				err = -EMSGSIZE;
1125 				goto error;
1126 			}
1127 			if (getfrag(from, page_address(frag->page)+frag->page_offset+frag->size, offset, copy, skb->len, skb) < 0) {
1128 				err = -EFAULT;
1129 				goto error;
1130 			}
1131 			sk->sk_sndmsg_off += copy;
1132 			frag->size += copy;
1133 			skb->len += copy;
1134 			skb->data_len += copy;
1135 		}
1136 		offset += copy;
1137 		length -= copy;
1138 	}
1139 	return 0;
1140 error:
1141 	inet->cork.length -= length;
1142 	IP6_INC_STATS(IPSTATS_MIB_OUTDISCARDS);
1143 	return err;
1144 }
1145 
1146 int ip6_push_pending_frames(struct sock *sk)
1147 {
1148 	struct sk_buff *skb, *tmp_skb;
1149 	struct sk_buff **tail_skb;
1150 	struct in6_addr final_dst_buf, *final_dst = &final_dst_buf;
1151 	struct inet_sock *inet = inet_sk(sk);
1152 	struct ipv6_pinfo *np = inet6_sk(sk);
1153 	struct ipv6hdr *hdr;
1154 	struct ipv6_txoptions *opt = np->cork.opt;
1155 	struct rt6_info *rt = np->cork.rt;
1156 	struct flowi *fl = &inet->cork.fl;
1157 	unsigned char proto = fl->proto;
1158 	int err = 0;
1159 
1160 	if ((skb = __skb_dequeue(&sk->sk_write_queue)) == NULL)
1161 		goto out;
1162 	tail_skb = &(skb_shinfo(skb)->frag_list);
1163 
1164 	/* move skb->data to ip header from ext header */
1165 	if (skb->data < skb->nh.raw)
1166 		__skb_pull(skb, skb->nh.raw - skb->data);
1167 	while ((tmp_skb = __skb_dequeue(&sk->sk_write_queue)) != NULL) {
1168 		__skb_pull(tmp_skb, skb->h.raw - skb->nh.raw);
1169 		*tail_skb = tmp_skb;
1170 		tail_skb = &(tmp_skb->next);
1171 		skb->len += tmp_skb->len;
1172 		skb->data_len += tmp_skb->len;
1173 		skb->truesize += tmp_skb->truesize;
1174 		__sock_put(tmp_skb->sk);
1175 		tmp_skb->destructor = NULL;
1176 		tmp_skb->sk = NULL;
1177 	}
1178 
1179 	ipv6_addr_copy(final_dst, &fl->fl6_dst);
1180 	__skb_pull(skb, skb->h.raw - skb->nh.raw);
1181 	if (opt && opt->opt_flen)
1182 		ipv6_push_frag_opts(skb, opt, &proto);
1183 	if (opt && opt->opt_nflen)
1184 		ipv6_push_nfrag_opts(skb, opt, &proto, &final_dst);
1185 
1186 	skb->nh.ipv6h = hdr = (struct ipv6hdr*) skb_push(skb, sizeof(struct ipv6hdr));
1187 
1188 	*(u32*)hdr = fl->fl6_flowlabel |
1189 		     htonl(0x60000000 | ((int)np->cork.tclass << 20));
1190 
1191 	if (skb->len <= sizeof(struct ipv6hdr) + IPV6_MAXPLEN)
1192 		hdr->payload_len = htons(skb->len - sizeof(struct ipv6hdr));
1193 	else
1194 		hdr->payload_len = 0;
1195 	hdr->hop_limit = np->cork.hop_limit;
1196 	hdr->nexthdr = proto;
1197 	ipv6_addr_copy(&hdr->saddr, &fl->fl6_src);
1198 	ipv6_addr_copy(&hdr->daddr, final_dst);
1199 
1200 	skb->priority = sk->sk_priority;
1201 
1202 	skb->dst = dst_clone(&rt->u.dst);
1203 	IP6_INC_STATS(IPSTATS_MIB_OUTREQUESTS);
1204 	err = NF_HOOK(PF_INET6, NF_IP6_LOCAL_OUT, skb, NULL, skb->dst->dev, dst_output);
1205 	if (err) {
1206 		if (err > 0)
1207 			err = np->recverr ? net_xmit_errno(err) : 0;
1208 		if (err)
1209 			goto error;
1210 	}
1211 
1212 out:
1213 	inet->cork.flags &= ~IPCORK_OPT;
1214 	kfree(np->cork.opt);
1215 	np->cork.opt = NULL;
1216 	if (np->cork.rt) {
1217 		dst_release(&np->cork.rt->u.dst);
1218 		np->cork.rt = NULL;
1219 		inet->cork.flags &= ~IPCORK_ALLFRAG;
1220 	}
1221 	memset(&inet->cork.fl, 0, sizeof(inet->cork.fl));
1222 	return err;
1223 error:
1224 	goto out;
1225 }
1226 
1227 void ip6_flush_pending_frames(struct sock *sk)
1228 {
1229 	struct inet_sock *inet = inet_sk(sk);
1230 	struct ipv6_pinfo *np = inet6_sk(sk);
1231 	struct sk_buff *skb;
1232 
1233 	while ((skb = __skb_dequeue_tail(&sk->sk_write_queue)) != NULL) {
1234 		IP6_INC_STATS(IPSTATS_MIB_OUTDISCARDS);
1235 		kfree_skb(skb);
1236 	}
1237 
1238 	inet->cork.flags &= ~IPCORK_OPT;
1239 
1240 	kfree(np->cork.opt);
1241 	np->cork.opt = NULL;
1242 	if (np->cork.rt) {
1243 		dst_release(&np->cork.rt->u.dst);
1244 		np->cork.rt = NULL;
1245 		inet->cork.flags &= ~IPCORK_ALLFRAG;
1246 	}
1247 	memset(&inet->cork.fl, 0, sizeof(inet->cork.fl));
1248 }
1249