reassembly.c 18.8 KB
Newer Older
L
Linus Torvalds 已提交
1 2
/*
 *	IPv6 fragment reassembly
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
 *
 *	Based on: net/ipv4/ip_fragment.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.
 */

16 17
/*
 *	Fixes:
L
Linus Torvalds 已提交
18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42
 *	Andi Kleen	Make it work with multiple hosts.
 *			More RFC compliance.
 *
 *      Horst von Brand Add missing #include <linux/string.h>
 *	Alexey Kuznetsov	SMP races, threading, cleanup.
 *	Patrick McHardy		LRU queue of frag heads for evictor.
 *	Mitsuru KANDA @USAGI	Register inet6_protocol{}.
 *	David Stevens and
 *	YOSHIFUJI,H. @USAGI	Always remove fragment header to
 *				calculate ICV correctly.
 */
#include <linux/errno.h>
#include <linux/types.h>
#include <linux/string.h>
#include <linux/socket.h>
#include <linux/sockios.h>
#include <linux/jiffies.h>
#include <linux/net.h>
#include <linux/list.h>
#include <linux/netdevice.h>
#include <linux/in6.h>
#include <linux/ipv6.h>
#include <linux/icmpv6.h>
#include <linux/random.h>
#include <linux/jhash.h>
43
#include <linux/skbuff.h>
44
#include <linux/slab.h>
45
#include <linux/export.h>
L
Linus Torvalds 已提交
46 47 48 49 50

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

#include <net/ipv6.h>
51
#include <net/ip6_route.h>
L
Linus Torvalds 已提交
52 53 54 55 56
#include <net/protocol.h>
#include <net/transp_v6.h>
#include <net/rawv6.h>
#include <net/ndisc.h>
#include <net/addrconf.h>
57
#include <net/inet_frag.h>
L
Linus Torvalds 已提交
58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73

struct ip6frag_skb_cb
{
	struct inet6_skb_parm	h;
	int			offset;
};

#define FRAG6_CB(skb)	((struct ip6frag_skb_cb*)((skb)->cb))


/*
 *	Equivalent of ipv4 struct ipq
 */

struct frag_queue
{
74
	struct inet_frag_queue	q;
L
Linus Torvalds 已提交
75

A
Al Viro 已提交
76
	__be32			id;		/* fragment id		*/
77
	u32			user;
L
Linus Torvalds 已提交
78 79 80 81 82 83 84 85
	struct in6_addr		saddr;
	struct in6_addr		daddr;

	int			iif;
	unsigned int		csum;
	__u16			nhoffset;
};

86
static struct inet_frags ip6_frags;
L
Linus Torvalds 已提交
87

88
int ip6_frag_nqueues(struct net *net)
89
{
90
	return net->ipv6.frags.nqueues;
91
}
L
Linus Torvalds 已提交
92

93
int ip6_frag_mem(struct net *net)
94
{
95
	return atomic_read(&net->ipv6.frags.mem);
96
}
L
Linus Torvalds 已提交
97

98 99 100
static int ip6_frag_reasm(struct frag_queue *fq, struct sk_buff *prev,
			  struct net_device *dev);

101 102 103 104
/*
 * callers should be careful not to use the hash value outside the ipfrag_lock
 * as doing so could race with ipfrag_hash_rnd being recalculated.
 */
105 106
unsigned int inet6_hash_frag(__be32 id, const struct in6_addr *saddr,
			     const struct in6_addr *daddr, u32 rnd)
L
Linus Torvalds 已提交
107
{
108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123
	u32 c;

	c = jhash_3words((__force u32)saddr->s6_addr32[0],
			 (__force u32)saddr->s6_addr32[1],
			 (__force u32)saddr->s6_addr32[2],
			 rnd);

	c = jhash_3words((__force u32)saddr->s6_addr32[3],
			 (__force u32)daddr->s6_addr32[0],
			 (__force u32)daddr->s6_addr32[1],
			 c);

	c =  jhash_3words((__force u32)daddr->s6_addr32[2],
			  (__force u32)daddr->s6_addr32[3],
			  (__force u32)id,
			  c);
L
Linus Torvalds 已提交
124

125
	return c & (INETFRAGS_HASHSZ - 1);
L
Linus Torvalds 已提交
126
}
127
EXPORT_SYMBOL_GPL(inet6_hash_frag);
L
Linus Torvalds 已提交
128

129
static unsigned int ip6_hashfn(struct inet_frag_queue *q)
L
Linus Torvalds 已提交
130
{
131
	struct frag_queue *fq;
L
Linus Torvalds 已提交
132

133
	fq = container_of(q, struct frag_queue, q);
134
	return inet6_hash_frag(fq->id, &fq->saddr, &fq->daddr, ip6_frags.rnd);
L
Linus Torvalds 已提交
135 136
}

137 138 139 140 141 142
int ip6_frag_match(struct inet_frag_queue *q, void *a)
{
	struct frag_queue *fq;
	struct ip6_create_arg *arg = a;

	fq = container_of(q, struct frag_queue, q);
143
	return (fq->id == arg->id && fq->user == arg->user &&
144 145 146 147 148
			ipv6_addr_equal(&fq->saddr, arg->src) &&
			ipv6_addr_equal(&fq->daddr, arg->dst));
}
EXPORT_SYMBOL(ip6_frag_match);

149
void ip6_frag_init(struct inet_frag_queue *q, void *a)
L
Linus Torvalds 已提交
150
{
151 152 153 154
	struct frag_queue *fq = container_of(q, struct frag_queue, q);
	struct ip6_create_arg *arg = a;

	fq->id = arg->id;
155
	fq->user = arg->user;
A
Alexey Dobriyan 已提交
156 157
	fq->saddr = *arg->src;
	fq->daddr = *arg->dst;
L
Linus Torvalds 已提交
158
}
159
EXPORT_SYMBOL(ip6_frag_init);
L
Linus Torvalds 已提交
160 161 162

/* Destruction primitives. */

163
static __inline__ void fq_put(struct frag_queue *fq)
L
Linus Torvalds 已提交
164
{
P
Pavel Emelyanov 已提交
165
	inet_frag_put(&fq->q, &ip6_frags);
L
Linus Torvalds 已提交
166 167 168 169 170 171 172
}

/* Kill fq entry. It is not destroyed immediately,
 * because caller (and someone more) holds reference count.
 */
static __inline__ void fq_kill(struct frag_queue *fq)
{
173
	inet_frag_kill(&fq->q, &ip6_frags);
L
Linus Torvalds 已提交
174 175
}

176
static void ip6_evictor(struct net *net, struct inet6_dev *idev)
L
Linus Torvalds 已提交
177
{
178 179
	int evicted;

180
	evicted = inet_frag_evictor(&net->ipv6.frags, &ip6_frags);
181
	if (evicted)
182
		IP6_ADD_STATS_BH(net, idev, IPSTATS_MIB_REASMFAILS, evicted);
L
Linus Torvalds 已提交
183 184 185 186
}

static void ip6_frag_expire(unsigned long data)
{
187
	struct frag_queue *fq;
188
	struct net_device *dev = NULL;
189
	struct net *net;
L
Linus Torvalds 已提交
190

191 192
	fq = container_of((struct inet_frag_queue *)data, struct frag_queue, q);

193
	spin_lock(&fq->q.lock);
L
Linus Torvalds 已提交
194

195
	if (fq->q.last_in & INET_FRAG_COMPLETE)
L
Linus Torvalds 已提交
196 197 198 199
		goto out;

	fq_kill(fq);

200
	net = container_of(fq->q.net, struct net, ipv6.frags);
E
Eric Dumazet 已提交
201 202
	rcu_read_lock();
	dev = dev_get_by_index_rcu(net, fq->iif);
203
	if (!dev)
E
Eric Dumazet 已提交
204
		goto out_rcu_unlock;
205

206 207
	IP6_INC_STATS_BH(net, __in6_dev_get(dev), IPSTATS_MIB_REASMTIMEOUT);
	IP6_INC_STATS_BH(net, __in6_dev_get(dev), IPSTATS_MIB_REASMFAILS);
L
Linus Torvalds 已提交
208

209
	/* Don't send error if the first segment did not arrive. */
210
	if (!(fq->q.last_in & INET_FRAG_FIRST_IN) || !fq->q.fragments)
E
Eric Dumazet 已提交
211
		goto out_rcu_unlock;
212 213 214 215 216 217

	/*
	   But use as source device on which LAST ARRIVED
	   segment was received. And do not use fq->dev
	   pointer directly, device might already disappeared.
	 */
218
	fq->q.fragments->dev = dev;
219
	icmpv6_send(fq->q.fragments, ICMPV6_TIME_EXCEED, ICMPV6_EXC_FRAGTIME, 0);
E
Eric Dumazet 已提交
220 221
out_rcu_unlock:
	rcu_read_unlock();
L
Linus Torvalds 已提交
222
out:
223
	spin_unlock(&fq->q.lock);
224
	fq_put(fq);
L
Linus Torvalds 已提交
225 226
}

227
static __inline__ struct frag_queue *
228
fq_find(struct net *net, __be32 id, const struct in6_addr *src, const struct in6_addr *dst)
L
Linus Torvalds 已提交
229
{
230 231
	struct inet_frag_queue *q;
	struct ip6_create_arg arg;
232
	unsigned int hash;
L
Linus Torvalds 已提交
233

234
	arg.id = id;
235
	arg.user = IP6_DEFRAG_LOCAL_DELIVER;
236 237
	arg.src = src;
	arg.dst = dst;
238 239

	read_lock(&ip6_frags.lock);
240
	hash = inet6_hash_frag(id, src, dst, ip6_frags.rnd);
L
Linus Torvalds 已提交
241

242
	q = inet_frag_find(&net->ipv6.frags, &ip6_frags, &arg, hash);
243
	if (q == NULL)
244
		return NULL;
L
Linus Torvalds 已提交
245

246
	return container_of(q, struct frag_queue, q);
L
Linus Torvalds 已提交
247 248
}

249
static int ip6_frag_queue(struct frag_queue *fq, struct sk_buff *skb,
L
Linus Torvalds 已提交
250 251 252
			   struct frag_hdr *fhdr, int nhoff)
{
	struct sk_buff *prev, *next;
253
	struct net_device *dev;
L
Linus Torvalds 已提交
254
	int offset, end;
E
Eric Dumazet 已提交
255
	struct net *net = dev_net(skb_dst(skb)->dev);
L
Linus Torvalds 已提交
256

257
	if (fq->q.last_in & INET_FRAG_COMPLETE)
L
Linus Torvalds 已提交
258 259 260
		goto err;

	offset = ntohs(fhdr->frag_off) & ~0x7;
261 262
	end = offset + (ntohs(ipv6_hdr(skb)->payload_len) -
			((u8 *)(fhdr + 1) - (u8 *)(ipv6_hdr(skb) + 1)));
L
Linus Torvalds 已提交
263 264

	if ((unsigned int)end > IPV6_MAXPLEN) {
E
Eric Dumazet 已提交
265
		IP6_INC_STATS_BH(net, ip6_dst_idev(skb_dst(skb)),
266
				 IPSTATS_MIB_INHDRERRORS);
267 268 269
		icmpv6_param_prob(skb, ICMPV6_HDR_FIELD,
				  ((u8 *)&fhdr->frag_off -
				   skb_network_header(skb)));
270
		return -1;
L
Linus Torvalds 已提交
271 272
	}

273 274
	if (skb->ip_summed == CHECKSUM_COMPLETE) {
		const unsigned char *nh = skb_network_header(skb);
275
		skb->csum = csum_sub(skb->csum,
276 277 278
				     csum_partial(nh, (u8 *)(fhdr + 1) - nh,
						  0));
	}
L
Linus Torvalds 已提交
279 280 281 282 283 284

	/* Is this the final fragment? */
	if (!(fhdr->frag_off & htons(IP6_MF))) {
		/* If we already have some bits beyond end
		 * or have different end, the segment is corrupted.
		 */
285
		if (end < fq->q.len ||
286
		    ((fq->q.last_in & INET_FRAG_LAST_IN) && end != fq->q.len))
L
Linus Torvalds 已提交
287
			goto err;
288
		fq->q.last_in |= INET_FRAG_LAST_IN;
289
		fq->q.len = end;
L
Linus Torvalds 已提交
290 291 292 293 294 295 296 297
	} else {
		/* Check if the fragment is rounded to 8 bytes.
		 * Required by the RFC.
		 */
		if (end & 0x7) {
			/* RFC2460 says always send parameter problem in
			 * this case. -DaveM
			 */
E
Eric Dumazet 已提交
298
			IP6_INC_STATS_BH(net, ip6_dst_idev(skb_dst(skb)),
299
					 IPSTATS_MIB_INHDRERRORS);
300
			icmpv6_param_prob(skb, ICMPV6_HDR_FIELD,
L
Linus Torvalds 已提交
301
					  offsetof(struct ipv6hdr, payload_len));
302
			return -1;
L
Linus Torvalds 已提交
303
		}
304
		if (end > fq->q.len) {
L
Linus Torvalds 已提交
305
			/* Some bits beyond end -> corruption. */
306
			if (fq->q.last_in & INET_FRAG_LAST_IN)
L
Linus Torvalds 已提交
307
				goto err;
308
			fq->q.len = end;
L
Linus Torvalds 已提交
309 310 311 312 313 314 315 316 317
		}
	}

	if (end == offset)
		goto err;

	/* Point into the IP datagram 'data' part. */
	if (!pskb_pull(skb, (u8 *) (fhdr + 1) - skb->data))
		goto err;
318

319 320
	if (pskb_trim_rcsum(skb, end - offset))
		goto err;
L
Linus Torvalds 已提交
321 322 323 324 325

	/* Find out which fragments are in front and at the back of us
	 * in the chain of fragments so far.  We must know where to put
	 * this fragment, right?
	 */
326 327 328 329 330
	prev = fq->q.fragments_tail;
	if (!prev || FRAG6_CB(prev)->offset < offset) {
		next = NULL;
		goto found;
	}
L
Linus Torvalds 已提交
331
	prev = NULL;
332
	for(next = fq->q.fragments; next != NULL; next = next->next) {
L
Linus Torvalds 已提交
333 334 335 336 337
		if (FRAG6_CB(next)->offset >= offset)
			break;	/* bingo! */
		prev = next;
	}

338
found:
E
Eric Dumazet 已提交
339 340
	/* RFC5722, Section 4, amended by Errata ID : 3089
	 *                          When reassembling an IPv6 datagram, if
341 342
	 *   one or more its constituent fragments is determined to be an
	 *   overlapping fragment, the entire datagram (and any constituent
E
Eric Dumazet 已提交
343
	 *   fragments) MUST be silently discarded.
L
Linus Torvalds 已提交
344 345
	 */

346 347
	/* Check for overlap with preceding fragment. */
	if (prev &&
S
Shan Wei 已提交
348
	    (FRAG6_CB(prev)->offset + prev->len) > offset)
349
		goto discard_fq;
L
Linus Torvalds 已提交
350

351 352 353
	/* Look for overlap with succeeding segment. */
	if (next && FRAG6_CB(next)->offset < end)
		goto discard_fq;
L
Linus Torvalds 已提交
354 355 356 357 358

	FRAG6_CB(skb)->offset = offset;

	/* Insert this fragment in the chain of fragments. */
	skb->next = next;
359 360
	if (!next)
		fq->q.fragments_tail = skb;
L
Linus Torvalds 已提交
361 362 363
	if (prev)
		prev->next = skb;
	else
364
		fq->q.fragments = skb;
L
Linus Torvalds 已提交
365

366 367 368 369 370
	dev = skb->dev;
	if (dev) {
		fq->iif = dev->ifindex;
		skb->dev = NULL;
	}
371 372
	fq->q.stamp = skb->tstamp;
	fq->q.meat += skb->len;
373
	atomic_add(skb->truesize, &fq->q.net->mem);
L
Linus Torvalds 已提交
374 375 376 377 378 379

	/* The first fragment.
	 * nhoffset is obtained from the first fragment, of course.
	 */
	if (offset == 0) {
		fq->nhoffset = nhoff;
380
		fq->q.last_in |= INET_FRAG_FIRST_IN;
L
Linus Torvalds 已提交
381
	}
382

383 384
	if (fq->q.last_in == (INET_FRAG_FIRST_IN | INET_FRAG_LAST_IN) &&
	    fq->q.meat == fq->q.len)
385 386
		return ip6_frag_reasm(fq, prev, dev);

387
	write_lock(&ip6_frags.lock);
388
	list_move_tail(&fq->q.lru_list, &fq->q.net->lru_list);
389
	write_unlock(&ip6_frags.lock);
390
	return -1;
L
Linus Torvalds 已提交
391

392 393
discard_fq:
	fq_kill(fq);
L
Linus Torvalds 已提交
394
err:
E
Eric Dumazet 已提交
395
	IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
396
		      IPSTATS_MIB_REASMFAILS);
L
Linus Torvalds 已提交
397
	kfree_skb(skb);
398
	return -1;
L
Linus Torvalds 已提交
399 400 401 402 403 404 405 406 407 408 409
}

/*
 *	Check if this packet is complete.
 *	Returns NULL on failure by any reason, and pointer
 *	to current nexthdr field in reassembled frame.
 *
 *	It is called with locked fq, and caller must check that
 *	queue is eligible for reassembly i.e. it is not COMPLETE,
 *	the last and the first frames arrived and all the bits are here.
 */
410
static int ip6_frag_reasm(struct frag_queue *fq, struct sk_buff *prev,
L
Linus Torvalds 已提交
411 412
			  struct net_device *dev)
{
413
	struct net *net = container_of(fq->q.net, struct net, ipv6.frags);
414
	struct sk_buff *fp, *head = fq->q.fragments;
L
Linus Torvalds 已提交
415 416 417 418 419
	int    payload_len;
	unsigned int nhoff;

	fq_kill(fq);

420 421 422 423 424 425 426 427 428
	/* Make the one we just received the head. */
	if (prev) {
		head = prev->next;
		fp = skb_clone(head, GFP_ATOMIC);

		if (!fp)
			goto out_oom;

		fp->next = head->next;
429 430
		if (!fp->next)
			fq->q.fragments_tail = fp;
431 432
		prev->next = fp;

433 434
		skb_morph(head, fq->q.fragments);
		head->next = fq->q.fragments->next;
435

436 437
		kfree_skb(fq->q.fragments);
		fq->q.fragments = head;
438 439
	}

440 441
	WARN_ON(head == NULL);
	WARN_ON(FRAG6_CB(head)->offset != 0);
L
Linus Torvalds 已提交
442 443

	/* Unfragmented part is taken from the first segment. */
444
	payload_len = ((head->data - skb_network_header(head)) -
445
		       sizeof(struct ipv6hdr) + fq->q.len -
446
		       sizeof(struct frag_hdr));
L
Linus Torvalds 已提交
447 448 449 450 451 452 453 454 455 456
	if (payload_len > IPV6_MAXPLEN)
		goto out_oversize;

	/* Head of list must not be cloned. */
	if (skb_cloned(head) && pskb_expand_head(head, 0, 0, GFP_ATOMIC))
		goto out_oom;

	/* If the first fragment is fragmented itself, we split
	 * it to two chunks: the first with data and paged part
	 * and the second, holding only fragments. */
457
	if (skb_has_frag_list(head)) {
L
Linus Torvalds 已提交
458 459 460 461 462 463 464 465
		struct sk_buff *clone;
		int i, plen = 0;

		if ((clone = alloc_skb(0, GFP_ATOMIC)) == NULL)
			goto out_oom;
		clone->next = head->next;
		head->next = clone;
		skb_shinfo(clone)->frag_list = skb_shinfo(head)->frag_list;
466
		skb_frag_list_init(head);
E
Eric Dumazet 已提交
467 468
		for (i = 0; i < skb_shinfo(head)->nr_frags; i++)
			plen += skb_frag_size(&skb_shinfo(head)->frags[i]);
L
Linus Torvalds 已提交
469 470 471 472 473
		clone->len = clone->data_len = head->data_len - plen;
		head->data_len -= clone->len;
		head->len -= clone->len;
		clone->csum = 0;
		clone->ip_summed = head->ip_summed;
474
		atomic_add(clone->truesize, &fq->q.net->mem);
L
Linus Torvalds 已提交
475 476 477 478 479
	}

	/* We have to remove fragment header from datagram and to relocate
	 * header in order to calculate ICV correctly. */
	nhoff = fq->nhoffset;
480
	skb_network_header(head)[nhoff] = skb_transport_header(head)[0];
481
	memmove(head->head + sizeof(struct frag_hdr), head->head,
L
Linus Torvalds 已提交
482
		(head->data - head->head) - sizeof(struct frag_hdr));
483 484
	head->mac_header += sizeof(struct frag_hdr);
	head->network_header += sizeof(struct frag_hdr);
L
Linus Torvalds 已提交
485 486

	skb_shinfo(head)->frag_list = head->next;
487
	skb_reset_transport_header(head);
488
	skb_push(head, head->data - skb_network_header(head));
L
Linus Torvalds 已提交
489 490 491 492 493 494

	for (fp=head->next; fp; fp = fp->next) {
		head->data_len += fp->len;
		head->len += fp->len;
		if (head->ip_summed != fp->ip_summed)
			head->ip_summed = CHECKSUM_NONE;
495
		else if (head->ip_summed == CHECKSUM_COMPLETE)
L
Linus Torvalds 已提交
496 497 498
			head->csum = csum_add(head->csum, fp->csum);
		head->truesize += fp->truesize;
	}
E
Eric Dumazet 已提交
499
	atomic_sub(head->truesize, &fq->q.net->mem);
L
Linus Torvalds 已提交
500 501 502

	head->next = NULL;
	head->dev = dev;
503
	head->tstamp = fq->q.stamp;
504
	ipv6_hdr(head)->payload_len = htons(payload_len);
505
	IP6CB(head)->nhoff = nhoff;
L
Linus Torvalds 已提交
506 507

	/* Yes, and fold redundant checksum back. 8) */
508
	if (head->ip_summed == CHECKSUM_COMPLETE)
509
		head->csum = csum_partial(skb_network_header(head),
510
					  skb_network_header_len(head),
511
					  head->csum);
L
Linus Torvalds 已提交
512

513
	rcu_read_lock();
514
	IP6_INC_STATS_BH(net, __in6_dev_get(dev), IPSTATS_MIB_REASMOKS);
515
	rcu_read_unlock();
516
	fq->q.fragments = NULL;
517
	fq->q.fragments_tail = NULL;
L
Linus Torvalds 已提交
518 519 520 521 522 523 524 525 526 527
	return 1;

out_oversize:
	if (net_ratelimit())
		printk(KERN_DEBUG "ip6_frag_reasm: payload len = %d\n", payload_len);
	goto out_fail;
out_oom:
	if (net_ratelimit())
		printk(KERN_DEBUG "ip6_frag_reasm: no memory for reassembly\n");
out_fail:
528
	rcu_read_lock();
529
	IP6_INC_STATS_BH(net, __in6_dev_get(dev), IPSTATS_MIB_REASMFAILS);
530
	rcu_read_unlock();
L
Linus Torvalds 已提交
531 532 533
	return -1;
}

534
static int ipv6_frag_rcv(struct sk_buff *skb)
L
Linus Torvalds 已提交
535 536 537
{
	struct frag_hdr *fhdr;
	struct frag_queue *fq;
538
	const struct ipv6hdr *hdr = ipv6_hdr(skb);
E
Eric Dumazet 已提交
539
	struct net *net = dev_net(skb_dst(skb)->dev);
L
Linus Torvalds 已提交
540

E
Eric Dumazet 已提交
541
	IP6_INC_STATS_BH(net, ip6_dst_idev(skb_dst(skb)), IPSTATS_MIB_REASMREQDS);
L
Linus Torvalds 已提交
542 543

	/* Jumbo payload inhibits frag. header */
544 545 546
	if (hdr->payload_len==0)
		goto fail_hdr;

547
	if (!pskb_may_pull(skb, (skb_transport_offset(skb) +
548 549
				 sizeof(struct frag_hdr))))
		goto fail_hdr;
L
Linus Torvalds 已提交
550

551
	hdr = ipv6_hdr(skb);
552
	fhdr = (struct frag_hdr *)skb_transport_header(skb);
L
Linus Torvalds 已提交
553 554 555

	if (!(fhdr->frag_off & htons(0xFFF9))) {
		/* It is not a fragmented frame */
556
		skb->transport_header += sizeof(struct frag_hdr);
557
		IP6_INC_STATS_BH(net,
E
Eric Dumazet 已提交
558
				 ip6_dst_idev(skb_dst(skb)), IPSTATS_MIB_REASMOKS);
L
Linus Torvalds 已提交
559

560
		IP6CB(skb)->nhoff = (u8 *)fhdr - skb_network_header(skb);
L
Linus Torvalds 已提交
561 562 563
		return 1;
	}

564
	if (atomic_read(&net->ipv6.frags.mem) > net->ipv6.frags.high_thresh)
E
Eric Dumazet 已提交
565
		ip6_evictor(net, ip6_dst_idev(skb_dst(skb)));
L
Linus Torvalds 已提交
566

567 568
	fq = fq_find(net, fhdr->identification, &hdr->saddr, &hdr->daddr);
	if (fq != NULL) {
569
		int ret;
L
Linus Torvalds 已提交
570

571
		spin_lock(&fq->q.lock);
L
Linus Torvalds 已提交
572

573
		ret = ip6_frag_queue(fq, skb, fhdr, IP6CB(skb)->nhoff);
L
Linus Torvalds 已提交
574

575
		spin_unlock(&fq->q.lock);
576
		fq_put(fq);
L
Linus Torvalds 已提交
577 578 579
		return ret;
	}

E
Eric Dumazet 已提交
580
	IP6_INC_STATS_BH(net, ip6_dst_idev(skb_dst(skb)), IPSTATS_MIB_REASMFAILS);
L
Linus Torvalds 已提交
581 582
	kfree_skb(skb);
	return -1;
583 584

fail_hdr:
E
Eric Dumazet 已提交
585
	IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)), IPSTATS_MIB_INHDRERRORS);
586 587
	icmpv6_param_prob(skb, ICMPV6_HDR_FIELD, skb_network_header_len(skb));
	return -1;
L
Linus Torvalds 已提交
588 589
}

590
static const struct inet6_protocol frag_protocol =
L
Linus Torvalds 已提交
591 592 593 594 595
{
	.handler	=	ipv6_frag_rcv,
	.flags		=	INET6_PROTO_NOPOLICY,
};

596
#ifdef CONFIG_SYSCTL
597
static struct ctl_table ip6_frags_ns_ctl_table[] = {
598 599
	{
		.procname	= "ip6frag_high_thresh",
600
		.data		= &init_net.ipv6.frags.high_thresh,
601 602
		.maxlen		= sizeof(int),
		.mode		= 0644,
A
Alexey Dobriyan 已提交
603
		.proc_handler	= proc_dointvec
604 605 606
	},
	{
		.procname	= "ip6frag_low_thresh",
607
		.data		= &init_net.ipv6.frags.low_thresh,
608 609
		.maxlen		= sizeof(int),
		.mode		= 0644,
A
Alexey Dobriyan 已提交
610
		.proc_handler	= proc_dointvec
611 612 613
	},
	{
		.procname	= "ip6frag_time",
614
		.data		= &init_net.ipv6.frags.timeout,
615 616
		.maxlen		= sizeof(int),
		.mode		= 0644,
A
Alexey Dobriyan 已提交
617
		.proc_handler	= proc_dointvec_jiffies,
618
	},
619 620 621 622
	{ }
};

static struct ctl_table ip6_frags_ctl_table[] = {
623 624
	{
		.procname	= "ip6frag_secret_interval",
625
		.data		= &ip6_frags.secret_interval,
626 627
		.maxlen		= sizeof(int),
		.mode		= 0644,
A
Alexey Dobriyan 已提交
628
		.proc_handler	= proc_dointvec_jiffies,
629 630 631 632
	},
	{ }
};

633
static int __net_init ip6_frags_ns_sysctl_register(struct net *net)
634
{
635
	struct ctl_table *table;
636 637
	struct ctl_table_header *hdr;

638
	table = ip6_frags_ns_ctl_table;
O
Octavian Purdila 已提交
639
	if (!net_eq(net, &init_net)) {
640
		table = kmemdup(table, sizeof(ip6_frags_ns_ctl_table), GFP_KERNEL);
641 642 643
		if (table == NULL)
			goto err_alloc;

644 645
		table[0].data = &net->ipv6.frags.high_thresh;
		table[1].data = &net->ipv6.frags.low_thresh;
646
		table[2].data = &net->ipv6.frags.timeout;
647 648 649 650 651 652 653 654 655 656
	}

	hdr = register_net_sysctl_table(net, net_ipv6_ctl_path, table);
	if (hdr == NULL)
		goto err_reg;

	net->ipv6.sysctl.frags_hdr = hdr;
	return 0;

err_reg:
O
Octavian Purdila 已提交
657
	if (!net_eq(net, &init_net))
658 659 660 661 662
		kfree(table);
err_alloc:
	return -ENOMEM;
}

663
static void __net_exit ip6_frags_ns_sysctl_unregister(struct net *net)
664 665 666 667 668
{
	struct ctl_table *table;

	table = net->ipv6.sysctl.frags_hdr->ctl_table_arg;
	unregister_net_sysctl_table(net->ipv6.sysctl.frags_hdr);
669 670
	if (!net_eq(net, &init_net))
		kfree(table);
671
}
672 673 674 675 676 677 678 679 680 681 682 683 684 685

static struct ctl_table_header *ip6_ctl_header;

static int ip6_frags_sysctl_register(void)
{
	ip6_ctl_header = register_net_sysctl_rotable(net_ipv6_ctl_path,
			ip6_frags_ctl_table);
	return ip6_ctl_header == NULL ? -ENOMEM : 0;
}

static void ip6_frags_sysctl_unregister(void)
{
	unregister_net_sysctl_table(ip6_ctl_header);
}
686
#else
687
static inline int ip6_frags_ns_sysctl_register(struct net *net)
688
{
689 690
	return 0;
}
691

692
static inline void ip6_frags_ns_sysctl_unregister(struct net *net)
693 694
{
}
695 696 697 698 699 700 701 702 703

static inline int ip6_frags_sysctl_register(void)
{
	return 0;
}

static inline void ip6_frags_sysctl_unregister(void)
{
}
704
#endif
D
Daniel Lezcano 已提交
705

706
static int __net_init ipv6_frags_init_net(struct net *net)
707
{
708 709
	net->ipv6.frags.high_thresh = IPV6_FRAG_HIGH_THRESH;
	net->ipv6.frags.low_thresh = IPV6_FRAG_LOW_THRESH;
710
	net->ipv6.frags.timeout = IPV6_FRAG_TIMEOUT;
711

712 713
	inet_frags_init_net(&net->ipv6.frags);

714
	return ip6_frags_ns_sysctl_register(net);
715 716
}

717
static void __net_exit ipv6_frags_exit_net(struct net *net)
718
{
719
	ip6_frags_ns_sysctl_unregister(net);
720 721 722 723 724 725 726 727
	inet_frags_exit_net(&net->ipv6.frags, &ip6_frags);
}

static struct pernet_operations ip6_frags_ops = {
	.init = ipv6_frags_init_net,
	.exit = ipv6_frags_exit_net,
};

728
int __init ipv6_frag_init(void)
L
Linus Torvalds 已提交
729
{
730
	int ret;
L
Linus Torvalds 已提交
731

732 733 734
	ret = inet6_add_protocol(&frag_protocol, IPPROTO_FRAGMENT);
	if (ret)
		goto out;
735

736 737 738 739
	ret = ip6_frags_sysctl_register();
	if (ret)
		goto err_sysctl;

740 741 742
	ret = register_pernet_subsys(&ip6_frags_ops);
	if (ret)
		goto err_pernet;
743

744
	ip6_frags.hashfn = ip6_hashfn;
745
	ip6_frags.constructor = ip6_frag_init;
746
	ip6_frags.destructor = NULL;
747 748
	ip6_frags.skb_free = NULL;
	ip6_frags.qsize = sizeof(struct frag_queue);
749
	ip6_frags.match = ip6_frag_match;
750
	ip6_frags.frag_expire = ip6_frag_expire;
751
	ip6_frags.secret_interval = 10 * 60 * HZ;
752
	inet_frags_init(&ip6_frags);
753 754
out:
	return ret;
755 756

err_pernet:
757 758
	ip6_frags_sysctl_unregister();
err_sysctl:
759 760
	inet6_del_protocol(&frag_protocol, IPPROTO_FRAGMENT);
	goto out;
761 762 763 764 765
}

void ipv6_frag_exit(void)
{
	inet_frags_fini(&ip6_frags);
766
	ip6_frags_sysctl_unregister();
767
	unregister_pernet_subsys(&ip6_frags_ops);
768
	inet6_del_protocol(&frag_protocol, IPPROTO_FRAGMENT);
L
Linus Torvalds 已提交
769
}