xref: /kernel/linux/linux-5.10/net/ipv6/reassembly.c (revision 8c2ecf20) 
1// SPDX-License-Identifier: GPL-2.0-or-later
2/*
3 *	IPv6 fragment reassembly
4 *	Linux INET6 implementation
5 *
6 *	Authors:
7 *	Pedro Roque		<roque@di.fc.ul.pt>
8 *
9 *	Based on: net/ipv4/ip_fragment.c
10 */
11
12/*
13 *	Fixes:
14 *	Andi Kleen	Make it work with multiple hosts.
15 *			More RFC compliance.
16 *
17 *      Horst von Brand Add missing #include <linux/string.h>
18 *	Alexey Kuznetsov	SMP races, threading, cleanup.
19 *	Patrick McHardy		LRU queue of frag heads for evictor.
20 *	Mitsuru KANDA @USAGI	Register inet6_protocol{}.
21 *	David Stevens and
22 *	YOSHIFUJI,H. @USAGI	Always remove fragment header to
23 *				calculate ICV correctly.
24 */
25
26#define pr_fmt(fmt) "IPv6: " fmt
27
28#include <linux/errno.h>
29#include <linux/types.h>
30#include <linux/string.h>
31#include <linux/socket.h>
32#include <linux/sockios.h>
33#include <linux/jiffies.h>
34#include <linux/net.h>
35#include <linux/list.h>
36#include <linux/netdevice.h>
37#include <linux/in6.h>
38#include <linux/ipv6.h>
39#include <linux/icmpv6.h>
40#include <linux/random.h>
41#include <linux/jhash.h>
42#include <linux/skbuff.h>
43#include <linux/slab.h>
44#include <linux/export.h>
45#include <linux/tcp.h>
46#include <linux/udp.h>
47
48#include <net/sock.h>
49#include <net/snmp.h>
50
51#include <net/ipv6.h>
52#include <net/ip6_route.h>
53#include <net/protocol.h>
54#include <net/transp_v6.h>
55#include <net/rawv6.h>
56#include <net/ndisc.h>
57#include <net/addrconf.h>
58#include <net/ipv6_frag.h>
59#include <net/inet_ecn.h>
60
61static const char ip6_frag_cache_name[] = "ip6-frags";
62
63static u8 ip6_frag_ecn(const struct ipv6hdr *ipv6h)
64{
65	return 1 << (ipv6_get_dsfield(ipv6h) & INET_ECN_MASK);
66}
67
68static struct inet_frags ip6_frags;
69
70static int ip6_frag_reasm(struct frag_queue *fq, struct sk_buff *skb,
71			  struct sk_buff *prev_tail, struct net_device *dev);
72
73static void ip6_frag_expire(struct timer_list *t)
74{
75	struct inet_frag_queue *frag = from_timer(frag, t, timer);
76	struct frag_queue *fq;
77
78	fq = container_of(frag, struct frag_queue, q);
79
80	ip6frag_expire_frag_queue(fq->q.fqdir->net, fq);
81}
82
83static struct frag_queue *
84fq_find(struct net *net, __be32 id, const struct ipv6hdr *hdr, int iif)
85{
86	struct frag_v6_compare_key key = {
87		.id = id,
88		.saddr = hdr->saddr,
89		.daddr = hdr->daddr,
90		.user = IP6_DEFRAG_LOCAL_DELIVER,
91		.iif = iif,
92	};
93	struct inet_frag_queue *q;
94
95	if (!(ipv6_addr_type(&hdr->daddr) & (IPV6_ADDR_MULTICAST |
96					    IPV6_ADDR_LINKLOCAL)))
97		key.iif = 0;
98
99	q = inet_frag_find(net->ipv6.fqdir, &key);
100	if (!q)
101		return NULL;
102
103	return container_of(q, struct frag_queue, q);
104}
105
106static int ip6_frag_queue(struct frag_queue *fq, struct sk_buff *skb,
107			  struct frag_hdr *fhdr, int nhoff,
108			  u32 *prob_offset)
109{
110	struct net *net = dev_net(skb_dst(skb)->dev);
111	int offset, end, fragsize;
112	struct sk_buff *prev_tail;
113	struct net_device *dev;
114	int err = -ENOENT;
115	u8 ecn;
116
117	if (fq->q.flags & INET_FRAG_COMPLETE)
118		goto err;
119
120	err = -EINVAL;
121	offset = ntohs(fhdr->frag_off) & ~0x7;
122	end = offset + (ntohs(ipv6_hdr(skb)->payload_len) -
123			((u8 *)(fhdr + 1) - (u8 *)(ipv6_hdr(skb) + 1)));
124
125	if ((unsigned int)end > IPV6_MAXPLEN) {
126		*prob_offset = (u8 *)&fhdr->frag_off - skb_network_header(skb);
127		/* note that if prob_offset is set, the skb is freed elsewhere,
128		 * we do not free it here.
129		 */
130		return -1;
131	}
132
133	ecn = ip6_frag_ecn(ipv6_hdr(skb));
134
135	if (skb->ip_summed == CHECKSUM_COMPLETE) {
136		const unsigned char *nh = skb_network_header(skb);
137		skb->csum = csum_sub(skb->csum,
138				     csum_partial(nh, (u8 *)(fhdr + 1) - nh,
139						  0));
140	}
141
142	/* Is this the final fragment? */
143	if (!(fhdr->frag_off & htons(IP6_MF))) {
144		/* If we already have some bits beyond end
145		 * or have different end, the segment is corrupted.
146		 */
147		if (end < fq->q.len ||
148		    ((fq->q.flags & INET_FRAG_LAST_IN) && end != fq->q.len))
149			goto discard_fq;
150		fq->q.flags |= INET_FRAG_LAST_IN;
151		fq->q.len = end;
152	} else {
153		/* Check if the fragment is rounded to 8 bytes.
154		 * Required by the RFC.
155		 */
156		if (end & 0x7) {
157			/* RFC2460 says always send parameter problem in
158			 * this case. -DaveM
159			 */
160			*prob_offset = offsetof(struct ipv6hdr, payload_len);
161			return -1;
162		}
163		if (end > fq->q.len) {
164			/* Some bits beyond end -> corruption. */
165			if (fq->q.flags & INET_FRAG_LAST_IN)
166				goto discard_fq;
167			fq->q.len = end;
168		}
169	}
170
171	if (end == offset)
172		goto discard_fq;
173
174	err = -ENOMEM;
175	/* Point into the IP datagram 'data' part. */
176	if (!pskb_pull(skb, (u8 *) (fhdr + 1) - skb->data))
177		goto discard_fq;
178
179	err = pskb_trim_rcsum(skb, end - offset);
180	if (err)
181		goto discard_fq;
182
183	/* Note : skb->rbnode and skb->dev share the same location. */
184	dev = skb->dev;
185	/* Makes sure compiler wont do silly aliasing games */
186	barrier();
187
188	prev_tail = fq->q.fragments_tail;
189	err = inet_frag_queue_insert(&fq->q, skb, offset, end);
190	if (err)
191		goto insert_error;
192
193	if (dev)
194		fq->iif = dev->ifindex;
195
196	fq->q.stamp = skb->tstamp;
197	fq->q.meat += skb->len;
198	fq->ecn |= ecn;
199	add_frag_mem_limit(fq->q.fqdir, skb->truesize);
200
201	fragsize = -skb_network_offset(skb) + skb->len;
202	if (fragsize > fq->q.max_size)
203		fq->q.max_size = fragsize;
204
205	/* The first fragment.
206	 * nhoffset is obtained from the first fragment, of course.
207	 */
208	if (offset == 0) {
209		fq->nhoffset = nhoff;
210		fq->q.flags |= INET_FRAG_FIRST_IN;
211	}
212
213	if (fq->q.flags == (INET_FRAG_FIRST_IN | INET_FRAG_LAST_IN) &&
214	    fq->q.meat == fq->q.len) {
215		unsigned long orefdst = skb->_skb_refdst;
216
217		skb->_skb_refdst = 0UL;
218		err = ip6_frag_reasm(fq, skb, prev_tail, dev);
219		skb->_skb_refdst = orefdst;
220		return err;
221	}
222
223	skb_dst_drop(skb);
224	return -EINPROGRESS;
225
226insert_error:
227	if (err == IPFRAG_DUP) {
228		kfree_skb(skb);
229		return -EINVAL;
230	}
231	err = -EINVAL;
232	__IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
233			IPSTATS_MIB_REASM_OVERLAPS);
234discard_fq:
235	inet_frag_kill(&fq->q);
236	__IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
237			IPSTATS_MIB_REASMFAILS);
238err:
239	kfree_skb(skb);
240	return err;
241}
242
243/*
244 *	Check if this packet is complete.
245 *
246 *	It is called with locked fq, and caller must check that
247 *	queue is eligible for reassembly i.e. it is not COMPLETE,
248 *	the last and the first frames arrived and all the bits are here.
249 */
250static int ip6_frag_reasm(struct frag_queue *fq, struct sk_buff *skb,
251			  struct sk_buff *prev_tail, struct net_device *dev)
252{
253	struct net *net = fq->q.fqdir->net;
254	unsigned int nhoff;
255	void *reasm_data;
256	int payload_len;
257	u8 ecn;
258
259	inet_frag_kill(&fq->q);
260
261	ecn = ip_frag_ecn_table[fq->ecn];
262	if (unlikely(ecn == 0xff))
263		goto out_fail;
264
265	reasm_data = inet_frag_reasm_prepare(&fq->q, skb, prev_tail);
266	if (!reasm_data)
267		goto out_oom;
268
269	payload_len = ((skb->data - skb_network_header(skb)) -
270		       sizeof(struct ipv6hdr) + fq->q.len -
271		       sizeof(struct frag_hdr));
272	if (payload_len > IPV6_MAXPLEN)
273		goto out_oversize;
274
275	/* We have to remove fragment header from datagram and to relocate
276	 * header in order to calculate ICV correctly. */
277	nhoff = fq->nhoffset;
278	skb_network_header(skb)[nhoff] = skb_transport_header(skb)[0];
279	memmove(skb->head + sizeof(struct frag_hdr), skb->head,
280		(skb->data - skb->head) - sizeof(struct frag_hdr));
281	if (skb_mac_header_was_set(skb))
282		skb->mac_header += sizeof(struct frag_hdr);
283	skb->network_header += sizeof(struct frag_hdr);
284
285	skb_reset_transport_header(skb);
286
287	inet_frag_reasm_finish(&fq->q, skb, reasm_data, true);
288
289	skb->dev = dev;
290	ipv6_hdr(skb)->payload_len = htons(payload_len);
291	ipv6_change_dsfield(ipv6_hdr(skb), 0xff, ecn);
292	IP6CB(skb)->nhoff = nhoff;
293	IP6CB(skb)->flags |= IP6SKB_FRAGMENTED;
294	IP6CB(skb)->frag_max_size = fq->q.max_size;
295
296	/* Yes, and fold redundant checksum back. 8) */
297	skb_postpush_rcsum(skb, skb_network_header(skb),
298			   skb_network_header_len(skb));
299
300	rcu_read_lock();
301	__IP6_INC_STATS(net, __in6_dev_stats_get(dev, skb), IPSTATS_MIB_REASMOKS);
302	rcu_read_unlock();
303	fq->q.rb_fragments = RB_ROOT;
304	fq->q.fragments_tail = NULL;
305	fq->q.last_run_head = NULL;
306	return 1;
307
308out_oversize:
309	net_dbg_ratelimited("ip6_frag_reasm: payload len = %d\n", payload_len);
310	goto out_fail;
311out_oom:
312	net_dbg_ratelimited("ip6_frag_reasm: no memory for reassembly\n");
313out_fail:
314	rcu_read_lock();
315	__IP6_INC_STATS(net, __in6_dev_stats_get(dev, skb), IPSTATS_MIB_REASMFAILS);
316	rcu_read_unlock();
317	inet_frag_kill(&fq->q);
318	return -1;
319}
320
321static int ipv6_frag_rcv(struct sk_buff *skb)
322{
323	struct frag_hdr *fhdr;
324	struct frag_queue *fq;
325	const struct ipv6hdr *hdr = ipv6_hdr(skb);
326	struct net *net = dev_net(skb_dst(skb)->dev);
327	u8 nexthdr;
328	int iif;
329
330	if (IP6CB(skb)->flags & IP6SKB_FRAGMENTED)
331		goto fail_hdr;
332
333	__IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)), IPSTATS_MIB_REASMREQDS);
334
335	/* Jumbo payload inhibits frag. header */
336	if (hdr->payload_len == 0)
337		goto fail_hdr;
338
339	if (!pskb_may_pull(skb, (skb_transport_offset(skb) +
340				 sizeof(struct frag_hdr))))
341		goto fail_hdr;
342
343	hdr = ipv6_hdr(skb);
344	fhdr = (struct frag_hdr *)skb_transport_header(skb);
345
346	if (!(fhdr->frag_off & htons(IP6_OFFSET | IP6_MF))) {
347		/* It is not a fragmented frame */
348		skb->transport_header += sizeof(struct frag_hdr);
349		__IP6_INC_STATS(net,
350				ip6_dst_idev(skb_dst(skb)), IPSTATS_MIB_REASMOKS);
351
352		IP6CB(skb)->nhoff = (u8 *)fhdr - skb_network_header(skb);
353		IP6CB(skb)->flags |= IP6SKB_FRAGMENTED;
354		IP6CB(skb)->frag_max_size = ntohs(hdr->payload_len) +
355					    sizeof(struct ipv6hdr);
356		return 1;
357	}
358
359	/* RFC 8200, Section 4.5 Fragment Header:
360	 * If the first fragment does not include all headers through an
361	 * Upper-Layer header, then that fragment should be discarded and
362	 * an ICMP Parameter Problem, Code 3, message should be sent to
363	 * the source of the fragment, with the Pointer field set to zero.
364	 */
365	nexthdr = hdr->nexthdr;
366	if (ipv6frag_thdr_truncated(skb, skb_transport_offset(skb), &nexthdr)) {
367		__IP6_INC_STATS(net, __in6_dev_get_safely(skb->dev),
368				IPSTATS_MIB_INHDRERRORS);
369		icmpv6_param_prob(skb, ICMPV6_HDR_INCOMP, 0);
370		return -1;
371	}
372
373	iif = skb->dev ? skb->dev->ifindex : 0;
374	fq = fq_find(net, fhdr->identification, hdr, iif);
375	if (fq) {
376		u32 prob_offset = 0;
377		int ret;
378
379		spin_lock(&fq->q.lock);
380
381		fq->iif = iif;
382		ret = ip6_frag_queue(fq, skb, fhdr, IP6CB(skb)->nhoff,
383				     &prob_offset);
384
385		spin_unlock(&fq->q.lock);
386		inet_frag_put(&fq->q);
387		if (prob_offset) {
388			__IP6_INC_STATS(net, __in6_dev_get_safely(skb->dev),
389					IPSTATS_MIB_INHDRERRORS);
390			/* icmpv6_param_prob() calls kfree_skb(skb) */
391			icmpv6_param_prob(skb, ICMPV6_HDR_FIELD, prob_offset);
392		}
393		return ret;
394	}
395
396	__IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)), IPSTATS_MIB_REASMFAILS);
397	kfree_skb(skb);
398	return -1;
399
400fail_hdr:
401	__IP6_INC_STATS(net, __in6_dev_get_safely(skb->dev),
402			IPSTATS_MIB_INHDRERRORS);
403	icmpv6_param_prob(skb, ICMPV6_HDR_FIELD, skb_network_header_len(skb));
404	return -1;
405}
406
407static const struct inet6_protocol frag_protocol = {
408	.handler	=	ipv6_frag_rcv,
409	.flags		=	INET6_PROTO_NOPOLICY,
410};
411
412#ifdef CONFIG_SYSCTL
413
414static struct ctl_table ip6_frags_ns_ctl_table[] = {
415	{
416		.procname	= "ip6frag_high_thresh",
417		.maxlen		= sizeof(unsigned long),
418		.mode		= 0644,
419		.proc_handler	= proc_doulongvec_minmax,
420	},
421	{
422		.procname	= "ip6frag_low_thresh",
423		.maxlen		= sizeof(unsigned long),
424		.mode		= 0644,
425		.proc_handler	= proc_doulongvec_minmax,
426	},
427	{
428		.procname	= "ip6frag_time",
429		.maxlen		= sizeof(int),
430		.mode		= 0644,
431		.proc_handler	= proc_dointvec_jiffies,
432	},
433	{ }
434};
435
436/* secret interval has been deprecated */
437static int ip6_frags_secret_interval_unused;
438static struct ctl_table ip6_frags_ctl_table[] = {
439	{
440		.procname	= "ip6frag_secret_interval",
441		.data		= &ip6_frags_secret_interval_unused,
442		.maxlen		= sizeof(int),
443		.mode		= 0644,
444		.proc_handler	= proc_dointvec_jiffies,
445	},
446	{ }
447};
448
449static int __net_init ip6_frags_ns_sysctl_register(struct net *net)
450{
451	struct ctl_table *table;
452	struct ctl_table_header *hdr;
453
454	table = ip6_frags_ns_ctl_table;
455	if (!net_eq(net, &init_net)) {
456		table = kmemdup(table, sizeof(ip6_frags_ns_ctl_table), GFP_KERNEL);
457		if (!table)
458			goto err_alloc;
459
460	}
461	table[0].data	= &net->ipv6.fqdir->high_thresh;
462	table[0].extra1	= &net->ipv6.fqdir->low_thresh;
463	table[1].data	= &net->ipv6.fqdir->low_thresh;
464	table[1].extra2	= &net->ipv6.fqdir->high_thresh;
465	table[2].data	= &net->ipv6.fqdir->timeout;
466
467	hdr = register_net_sysctl(net, "net/ipv6", table);
468	if (!hdr)
469		goto err_reg;
470
471	net->ipv6.sysctl.frags_hdr = hdr;
472	return 0;
473
474err_reg:
475	if (!net_eq(net, &init_net))
476		kfree(table);
477err_alloc:
478	return -ENOMEM;
479}
480
481static void __net_exit ip6_frags_ns_sysctl_unregister(struct net *net)
482{
483	struct ctl_table *table;
484
485	table = net->ipv6.sysctl.frags_hdr->ctl_table_arg;
486	unregister_net_sysctl_table(net->ipv6.sysctl.frags_hdr);
487	if (!net_eq(net, &init_net))
488		kfree(table);
489}
490
491static struct ctl_table_header *ip6_ctl_header;
492
493static int ip6_frags_sysctl_register(void)
494{
495	ip6_ctl_header = register_net_sysctl(&init_net, "net/ipv6",
496			ip6_frags_ctl_table);
497	return ip6_ctl_header == NULL ? -ENOMEM : 0;
498}
499
500static void ip6_frags_sysctl_unregister(void)
501{
502	unregister_net_sysctl_table(ip6_ctl_header);
503}
504#else
505static int ip6_frags_ns_sysctl_register(struct net *net)
506{
507	return 0;
508}
509
510static void ip6_frags_ns_sysctl_unregister(struct net *net)
511{
512}
513
514static int ip6_frags_sysctl_register(void)
515{
516	return 0;
517}
518
519static void ip6_frags_sysctl_unregister(void)
520{
521}
522#endif
523
524static int __net_init ipv6_frags_init_net(struct net *net)
525{
526	int res;
527
528	res = fqdir_init(&net->ipv6.fqdir, &ip6_frags, net);
529	if (res < 0)
530		return res;
531
532	net->ipv6.fqdir->high_thresh = IPV6_FRAG_HIGH_THRESH;
533	net->ipv6.fqdir->low_thresh = IPV6_FRAG_LOW_THRESH;
534	net->ipv6.fqdir->timeout = IPV6_FRAG_TIMEOUT;
535
536	res = ip6_frags_ns_sysctl_register(net);
537	if (res < 0)
538		fqdir_exit(net->ipv6.fqdir);
539	return res;
540}
541
542static void __net_exit ipv6_frags_pre_exit_net(struct net *net)
543{
544	fqdir_pre_exit(net->ipv6.fqdir);
545}
546
547static void __net_exit ipv6_frags_exit_net(struct net *net)
548{
549	ip6_frags_ns_sysctl_unregister(net);
550	fqdir_exit(net->ipv6.fqdir);
551}
552
553static struct pernet_operations ip6_frags_ops = {
554	.init		= ipv6_frags_init_net,
555	.pre_exit	= ipv6_frags_pre_exit_net,
556	.exit		= ipv6_frags_exit_net,
557};
558
559static const struct rhashtable_params ip6_rhash_params = {
560	.head_offset		= offsetof(struct inet_frag_queue, node),
561	.hashfn			= ip6frag_key_hashfn,
562	.obj_hashfn		= ip6frag_obj_hashfn,
563	.obj_cmpfn		= ip6frag_obj_cmpfn,
564	.automatic_shrinking	= true,
565};
566
567int __init ipv6_frag_init(void)
568{
569	int ret;
570
571	ip6_frags.constructor = ip6frag_init;
572	ip6_frags.destructor = NULL;
573	ip6_frags.qsize = sizeof(struct frag_queue);
574	ip6_frags.frag_expire = ip6_frag_expire;
575	ip6_frags.frags_cache_name = ip6_frag_cache_name;
576	ip6_frags.rhash_params = ip6_rhash_params;
577	ret = inet_frags_init(&ip6_frags);
578	if (ret)
579		goto out;
580
581	ret = inet6_add_protocol(&frag_protocol, IPPROTO_FRAGMENT);
582	if (ret)
583		goto err_protocol;
584
585	ret = ip6_frags_sysctl_register();
586	if (ret)
587		goto err_sysctl;
588
589	ret = register_pernet_subsys(&ip6_frags_ops);
590	if (ret)
591		goto err_pernet;
592
593out:
594	return ret;
595
596err_pernet:
597	ip6_frags_sysctl_unregister();
598err_sysctl:
599	inet6_del_protocol(&frag_protocol, IPPROTO_FRAGMENT);
600err_protocol:
601	inet_frags_fini(&ip6_frags);
602	goto out;
603}
604
605void ipv6_frag_exit(void)
606{
607	ip6_frags_sysctl_unregister();
608	unregister_pernet_subsys(&ip6_frags_ops);
609	inet6_del_protocol(&frag_protocol, IPPROTO_FRAGMENT);
610	inet_frags_fini(&ip6_frags);
611}
612