xref: /kernel/linux/linux-6.6/net/xfrm/xfrm_input.c (revision 62306a36)
1// SPDX-License-Identifier: GPL-2.0
2/*
3 * xfrm_input.c
4 *
5 * Changes:
6 * 	YOSHIFUJI Hideaki @USAGI
7 * 		Split up af-specific portion
8 *
9 */
10
11#include <linux/bottom_half.h>
12#include <linux/cache.h>
13#include <linux/interrupt.h>
14#include <linux/slab.h>
15#include <linux/module.h>
16#include <linux/netdevice.h>
17#include <linux/percpu.h>
18#include <net/dst.h>
19#include <net/ip.h>
20#include <net/xfrm.h>
21#include <net/ip_tunnels.h>
22#include <net/ip6_tunnel.h>
23#include <net/dst_metadata.h>
24
25#include "xfrm_inout.h"
26
27struct xfrm_trans_tasklet {
28	struct work_struct work;
29	spinlock_t queue_lock;
30	struct sk_buff_head queue;
31};
32
33struct xfrm_trans_cb {
34	union {
35		struct inet_skb_parm	h4;
36#if IS_ENABLED(CONFIG_IPV6)
37		struct inet6_skb_parm	h6;
38#endif
39	} header;
40	int (*finish)(struct net *net, struct sock *sk, struct sk_buff *skb);
41	struct net *net;
42};
43
44#define XFRM_TRANS_SKB_CB(__skb) ((struct xfrm_trans_cb *)&((__skb)->cb[0]))
45
46static DEFINE_SPINLOCK(xfrm_input_afinfo_lock);
47static struct xfrm_input_afinfo const __rcu *xfrm_input_afinfo[2][AF_INET6 + 1];
48
49static struct gro_cells gro_cells;
50static struct net_device xfrm_napi_dev;
51
52static DEFINE_PER_CPU(struct xfrm_trans_tasklet, xfrm_trans_tasklet);
53
54int xfrm_input_register_afinfo(const struct xfrm_input_afinfo *afinfo)
55{
56	int err = 0;
57
58	if (WARN_ON(afinfo->family > AF_INET6))
59		return -EAFNOSUPPORT;
60
61	spin_lock_bh(&xfrm_input_afinfo_lock);
62	if (unlikely(xfrm_input_afinfo[afinfo->is_ipip][afinfo->family]))
63		err = -EEXIST;
64	else
65		rcu_assign_pointer(xfrm_input_afinfo[afinfo->is_ipip][afinfo->family], afinfo);
66	spin_unlock_bh(&xfrm_input_afinfo_lock);
67	return err;
68}
69EXPORT_SYMBOL(xfrm_input_register_afinfo);
70
71int xfrm_input_unregister_afinfo(const struct xfrm_input_afinfo *afinfo)
72{
73	int err = 0;
74
75	spin_lock_bh(&xfrm_input_afinfo_lock);
76	if (likely(xfrm_input_afinfo[afinfo->is_ipip][afinfo->family])) {
77		if (unlikely(xfrm_input_afinfo[afinfo->is_ipip][afinfo->family] != afinfo))
78			err = -EINVAL;
79		else
80			RCU_INIT_POINTER(xfrm_input_afinfo[afinfo->is_ipip][afinfo->family], NULL);
81	}
82	spin_unlock_bh(&xfrm_input_afinfo_lock);
83	synchronize_rcu();
84	return err;
85}
86EXPORT_SYMBOL(xfrm_input_unregister_afinfo);
87
88static const struct xfrm_input_afinfo *xfrm_input_get_afinfo(u8 family, bool is_ipip)
89{
90	const struct xfrm_input_afinfo *afinfo;
91
92	if (WARN_ON_ONCE(family > AF_INET6))
93		return NULL;
94
95	rcu_read_lock();
96	afinfo = rcu_dereference(xfrm_input_afinfo[is_ipip][family]);
97	if (unlikely(!afinfo))
98		rcu_read_unlock();
99	return afinfo;
100}
101
102static int xfrm_rcv_cb(struct sk_buff *skb, unsigned int family, u8 protocol,
103		       int err)
104{
105	bool is_ipip = (protocol == IPPROTO_IPIP || protocol == IPPROTO_IPV6);
106	const struct xfrm_input_afinfo *afinfo;
107	int ret;
108
109	afinfo = xfrm_input_get_afinfo(family, is_ipip);
110	if (!afinfo)
111		return -EAFNOSUPPORT;
112
113	ret = afinfo->callback(skb, protocol, err);
114	rcu_read_unlock();
115
116	return ret;
117}
118
119struct sec_path *secpath_set(struct sk_buff *skb)
120{
121	struct sec_path *sp, *tmp = skb_ext_find(skb, SKB_EXT_SEC_PATH);
122
123	sp = skb_ext_add(skb, SKB_EXT_SEC_PATH);
124	if (!sp)
125		return NULL;
126
127	if (tmp) /* reused existing one (was COW'd if needed) */
128		return sp;
129
130	/* allocated new secpath */
131	memset(sp->ovec, 0, sizeof(sp->ovec));
132	sp->olen = 0;
133	sp->len = 0;
134	sp->verified_cnt = 0;
135
136	return sp;
137}
138EXPORT_SYMBOL(secpath_set);
139
140/* Fetch spi and seq from ipsec header */
141
142int xfrm_parse_spi(struct sk_buff *skb, u8 nexthdr, __be32 *spi, __be32 *seq)
143{
144	int offset, offset_seq;
145	int hlen;
146
147	switch (nexthdr) {
148	case IPPROTO_AH:
149		hlen = sizeof(struct ip_auth_hdr);
150		offset = offsetof(struct ip_auth_hdr, spi);
151		offset_seq = offsetof(struct ip_auth_hdr, seq_no);
152		break;
153	case IPPROTO_ESP:
154		hlen = sizeof(struct ip_esp_hdr);
155		offset = offsetof(struct ip_esp_hdr, spi);
156		offset_seq = offsetof(struct ip_esp_hdr, seq_no);
157		break;
158	case IPPROTO_COMP:
159		if (!pskb_may_pull(skb, sizeof(struct ip_comp_hdr)))
160			return -EINVAL;
161		*spi = htonl(ntohs(*(__be16 *)(skb_transport_header(skb) + 2)));
162		*seq = 0;
163		return 0;
164	default:
165		return 1;
166	}
167
168	if (!pskb_may_pull(skb, hlen))
169		return -EINVAL;
170
171	*spi = *(__be32 *)(skb_transport_header(skb) + offset);
172	*seq = *(__be32 *)(skb_transport_header(skb) + offset_seq);
173	return 0;
174}
175EXPORT_SYMBOL(xfrm_parse_spi);
176
177static int xfrm4_remove_beet_encap(struct xfrm_state *x, struct sk_buff *skb)
178{
179	struct iphdr *iph;
180	int optlen = 0;
181	int err = -EINVAL;
182
183	skb->protocol = htons(ETH_P_IP);
184
185	if (unlikely(XFRM_MODE_SKB_CB(skb)->protocol == IPPROTO_BEETPH)) {
186		struct ip_beet_phdr *ph;
187		int phlen;
188
189		if (!pskb_may_pull(skb, sizeof(*ph)))
190			goto out;
191
192		ph = (struct ip_beet_phdr *)skb->data;
193
194		phlen = sizeof(*ph) + ph->padlen;
195		optlen = ph->hdrlen * 8 + (IPV4_BEET_PHMAXLEN - phlen);
196		if (optlen < 0 || optlen & 3 || optlen > 250)
197			goto out;
198
199		XFRM_MODE_SKB_CB(skb)->protocol = ph->nexthdr;
200
201		if (!pskb_may_pull(skb, phlen))
202			goto out;
203		__skb_pull(skb, phlen);
204	}
205
206	skb_push(skb, sizeof(*iph));
207	skb_reset_network_header(skb);
208	skb_mac_header_rebuild(skb);
209
210	xfrm4_beet_make_header(skb);
211
212	iph = ip_hdr(skb);
213
214	iph->ihl += optlen / 4;
215	iph->tot_len = htons(skb->len);
216	iph->daddr = x->sel.daddr.a4;
217	iph->saddr = x->sel.saddr.a4;
218	iph->check = 0;
219	iph->check = ip_fast_csum(skb_network_header(skb), iph->ihl);
220	err = 0;
221out:
222	return err;
223}
224
225static void ipip_ecn_decapsulate(struct sk_buff *skb)
226{
227	struct iphdr *inner_iph = ipip_hdr(skb);
228
229	if (INET_ECN_is_ce(XFRM_MODE_SKB_CB(skb)->tos))
230		IP_ECN_set_ce(inner_iph);
231}
232
233static int xfrm4_remove_tunnel_encap(struct xfrm_state *x, struct sk_buff *skb)
234{
235	int err = -EINVAL;
236
237	skb->protocol = htons(ETH_P_IP);
238
239	if (!pskb_may_pull(skb, sizeof(struct iphdr)))
240		goto out;
241
242	err = skb_unclone(skb, GFP_ATOMIC);
243	if (err)
244		goto out;
245
246	if (x->props.flags & XFRM_STATE_DECAP_DSCP)
247		ipv4_copy_dscp(XFRM_MODE_SKB_CB(skb)->tos, ipip_hdr(skb));
248	if (!(x->props.flags & XFRM_STATE_NOECN))
249		ipip_ecn_decapsulate(skb);
250
251	skb_reset_network_header(skb);
252	skb_mac_header_rebuild(skb);
253	if (skb->mac_len)
254		eth_hdr(skb)->h_proto = skb->protocol;
255
256	err = 0;
257
258out:
259	return err;
260}
261
262static void ipip6_ecn_decapsulate(struct sk_buff *skb)
263{
264	struct ipv6hdr *inner_iph = ipipv6_hdr(skb);
265
266	if (INET_ECN_is_ce(XFRM_MODE_SKB_CB(skb)->tos))
267		IP6_ECN_set_ce(skb, inner_iph);
268}
269
270static int xfrm6_remove_tunnel_encap(struct xfrm_state *x, struct sk_buff *skb)
271{
272	int err = -EINVAL;
273
274	skb->protocol = htons(ETH_P_IPV6);
275
276	if (!pskb_may_pull(skb, sizeof(struct ipv6hdr)))
277		goto out;
278
279	err = skb_unclone(skb, GFP_ATOMIC);
280	if (err)
281		goto out;
282
283	if (x->props.flags & XFRM_STATE_DECAP_DSCP)
284		ipv6_copy_dscp(XFRM_MODE_SKB_CB(skb)->tos, ipipv6_hdr(skb));
285	if (!(x->props.flags & XFRM_STATE_NOECN))
286		ipip6_ecn_decapsulate(skb);
287
288	skb_reset_network_header(skb);
289	skb_mac_header_rebuild(skb);
290	if (skb->mac_len)
291		eth_hdr(skb)->h_proto = skb->protocol;
292
293	err = 0;
294
295out:
296	return err;
297}
298
299static int xfrm6_remove_beet_encap(struct xfrm_state *x, struct sk_buff *skb)
300{
301	struct ipv6hdr *ip6h;
302	int size = sizeof(struct ipv6hdr);
303	int err;
304
305	skb->protocol = htons(ETH_P_IPV6);
306
307	err = skb_cow_head(skb, size + skb->mac_len);
308	if (err)
309		goto out;
310
311	__skb_push(skb, size);
312	skb_reset_network_header(skb);
313	skb_mac_header_rebuild(skb);
314
315	xfrm6_beet_make_header(skb);
316
317	ip6h = ipv6_hdr(skb);
318	ip6h->payload_len = htons(skb->len - size);
319	ip6h->daddr = x->sel.daddr.in6;
320	ip6h->saddr = x->sel.saddr.in6;
321	err = 0;
322out:
323	return err;
324}
325
326/* Remove encapsulation header.
327 *
328 * The IP header will be moved over the top of the encapsulation
329 * header.
330 *
331 * On entry, the transport header shall point to where the IP header
332 * should be and the network header shall be set to where the IP
333 * header currently is.  skb->data shall point to the start of the
334 * payload.
335 */
336static int
337xfrm_inner_mode_encap_remove(struct xfrm_state *x,
338			     struct sk_buff *skb)
339{
340	switch (x->props.mode) {
341	case XFRM_MODE_BEET:
342		switch (x->sel.family) {
343		case AF_INET:
344			return xfrm4_remove_beet_encap(x, skb);
345		case AF_INET6:
346			return xfrm6_remove_beet_encap(x, skb);
347		}
348		break;
349	case XFRM_MODE_TUNNEL:
350		switch (XFRM_MODE_SKB_CB(skb)->protocol) {
351		case IPPROTO_IPIP:
352			return xfrm4_remove_tunnel_encap(x, skb);
353		case IPPROTO_IPV6:
354			return xfrm6_remove_tunnel_encap(x, skb);
355		break;
356		}
357		return -EINVAL;
358	}
359
360	WARN_ON_ONCE(1);
361	return -EOPNOTSUPP;
362}
363
364static int xfrm_prepare_input(struct xfrm_state *x, struct sk_buff *skb)
365{
366	switch (x->props.family) {
367	case AF_INET:
368		xfrm4_extract_header(skb);
369		break;
370	case AF_INET6:
371		xfrm6_extract_header(skb);
372		break;
373	default:
374		WARN_ON_ONCE(1);
375		return -EAFNOSUPPORT;
376	}
377
378	return xfrm_inner_mode_encap_remove(x, skb);
379}
380
381/* Remove encapsulation header.
382 *
383 * The IP header will be moved over the top of the encapsulation header.
384 *
385 * On entry, skb_transport_header() shall point to where the IP header
386 * should be and skb_network_header() shall be set to where the IP header
387 * currently is.  skb->data shall point to the start of the payload.
388 */
389static int xfrm4_transport_input(struct xfrm_state *x, struct sk_buff *skb)
390{
391	int ihl = skb->data - skb_transport_header(skb);
392
393	if (skb->transport_header != skb->network_header) {
394		memmove(skb_transport_header(skb),
395			skb_network_header(skb), ihl);
396		skb->network_header = skb->transport_header;
397	}
398	ip_hdr(skb)->tot_len = htons(skb->len + ihl);
399	skb_reset_transport_header(skb);
400	return 0;
401}
402
403static int xfrm6_transport_input(struct xfrm_state *x, struct sk_buff *skb)
404{
405#if IS_ENABLED(CONFIG_IPV6)
406	int ihl = skb->data - skb_transport_header(skb);
407
408	if (skb->transport_header != skb->network_header) {
409		memmove(skb_transport_header(skb),
410			skb_network_header(skb), ihl);
411		skb->network_header = skb->transport_header;
412	}
413	ipv6_hdr(skb)->payload_len = htons(skb->len + ihl -
414					   sizeof(struct ipv6hdr));
415	skb_reset_transport_header(skb);
416	return 0;
417#else
418	WARN_ON_ONCE(1);
419	return -EAFNOSUPPORT;
420#endif
421}
422
423static int xfrm_inner_mode_input(struct xfrm_state *x,
424				 struct sk_buff *skb)
425{
426	switch (x->props.mode) {
427	case XFRM_MODE_BEET:
428	case XFRM_MODE_TUNNEL:
429		return xfrm_prepare_input(x, skb);
430	case XFRM_MODE_TRANSPORT:
431		if (x->props.family == AF_INET)
432			return xfrm4_transport_input(x, skb);
433		if (x->props.family == AF_INET6)
434			return xfrm6_transport_input(x, skb);
435		break;
436	case XFRM_MODE_ROUTEOPTIMIZATION:
437		WARN_ON_ONCE(1);
438		break;
439	default:
440		WARN_ON_ONCE(1);
441		break;
442	}
443
444	return -EOPNOTSUPP;
445}
446
447int xfrm_input(struct sk_buff *skb, int nexthdr, __be32 spi, int encap_type)
448{
449	const struct xfrm_state_afinfo *afinfo;
450	struct net *net = dev_net(skb->dev);
451	int err;
452	__be32 seq;
453	__be32 seq_hi;
454	struct xfrm_state *x = NULL;
455	xfrm_address_t *daddr;
456	u32 mark = skb->mark;
457	unsigned int family = AF_UNSPEC;
458	int decaps = 0;
459	int async = 0;
460	bool xfrm_gro = false;
461	bool crypto_done = false;
462	struct xfrm_offload *xo = xfrm_offload(skb);
463	struct sec_path *sp;
464
465	if (encap_type < 0) {
466		x = xfrm_input_state(skb);
467
468		if (unlikely(x->km.state != XFRM_STATE_VALID)) {
469			if (x->km.state == XFRM_STATE_ACQ)
470				XFRM_INC_STATS(net, LINUX_MIB_XFRMACQUIREERROR);
471			else
472				XFRM_INC_STATS(net,
473					       LINUX_MIB_XFRMINSTATEINVALID);
474
475			if (encap_type == -1)
476				dev_put(skb->dev);
477			goto drop;
478		}
479
480		family = x->props.family;
481
482		/* An encap_type of -1 indicates async resumption. */
483		if (encap_type == -1) {
484			async = 1;
485			seq = XFRM_SKB_CB(skb)->seq.input.low;
486			goto resume;
487		}
488
489		/* encap_type < -1 indicates a GRO call. */
490		encap_type = 0;
491		seq = XFRM_SPI_SKB_CB(skb)->seq;
492
493		if (xo && (xo->flags & CRYPTO_DONE)) {
494			crypto_done = true;
495			family = XFRM_SPI_SKB_CB(skb)->family;
496
497			if (!(xo->status & CRYPTO_SUCCESS)) {
498				if (xo->status &
499				    (CRYPTO_TRANSPORT_AH_AUTH_FAILED |
500				     CRYPTO_TRANSPORT_ESP_AUTH_FAILED |
501				     CRYPTO_TUNNEL_AH_AUTH_FAILED |
502				     CRYPTO_TUNNEL_ESP_AUTH_FAILED)) {
503
504					xfrm_audit_state_icvfail(x, skb,
505								 x->type->proto);
506					x->stats.integrity_failed++;
507					XFRM_INC_STATS(net, LINUX_MIB_XFRMINSTATEPROTOERROR);
508					goto drop;
509				}
510
511				if (xo->status & CRYPTO_INVALID_PROTOCOL) {
512					XFRM_INC_STATS(net, LINUX_MIB_XFRMINSTATEPROTOERROR);
513					goto drop;
514				}
515
516				XFRM_INC_STATS(net, LINUX_MIB_XFRMINBUFFERERROR);
517				goto drop;
518			}
519
520			if (xfrm_parse_spi(skb, nexthdr, &spi, &seq)) {
521				XFRM_INC_STATS(net, LINUX_MIB_XFRMINHDRERROR);
522				goto drop;
523			}
524		}
525
526		goto lock;
527	}
528
529	family = XFRM_SPI_SKB_CB(skb)->family;
530
531	/* if tunnel is present override skb->mark value with tunnel i_key */
532	switch (family) {
533	case AF_INET:
534		if (XFRM_TUNNEL_SKB_CB(skb)->tunnel.ip4)
535			mark = be32_to_cpu(XFRM_TUNNEL_SKB_CB(skb)->tunnel.ip4->parms.i_key);
536		break;
537	case AF_INET6:
538		if (XFRM_TUNNEL_SKB_CB(skb)->tunnel.ip6)
539			mark = be32_to_cpu(XFRM_TUNNEL_SKB_CB(skb)->tunnel.ip6->parms.i_key);
540		break;
541	}
542
543	sp = secpath_set(skb);
544	if (!sp) {
545		XFRM_INC_STATS(net, LINUX_MIB_XFRMINERROR);
546		goto drop;
547	}
548
549	seq = 0;
550	if (!spi && xfrm_parse_spi(skb, nexthdr, &spi, &seq)) {
551		secpath_reset(skb);
552		XFRM_INC_STATS(net, LINUX_MIB_XFRMINHDRERROR);
553		goto drop;
554	}
555
556	daddr = (xfrm_address_t *)(skb_network_header(skb) +
557				   XFRM_SPI_SKB_CB(skb)->daddroff);
558	do {
559		sp = skb_sec_path(skb);
560
561		if (sp->len == XFRM_MAX_DEPTH) {
562			secpath_reset(skb);
563			XFRM_INC_STATS(net, LINUX_MIB_XFRMINBUFFERERROR);
564			goto drop;
565		}
566
567		x = xfrm_state_lookup(net, mark, daddr, spi, nexthdr, family);
568		if (x == NULL) {
569			secpath_reset(skb);
570			XFRM_INC_STATS(net, LINUX_MIB_XFRMINNOSTATES);
571			xfrm_audit_state_notfound(skb, family, spi, seq);
572			goto drop;
573		}
574
575		skb->mark = xfrm_smark_get(skb->mark, x);
576
577		sp->xvec[sp->len++] = x;
578
579		skb_dst_force(skb);
580		if (!skb_dst(skb)) {
581			XFRM_INC_STATS(net, LINUX_MIB_XFRMINERROR);
582			goto drop;
583		}
584
585lock:
586		spin_lock(&x->lock);
587
588		if (unlikely(x->km.state != XFRM_STATE_VALID)) {
589			if (x->km.state == XFRM_STATE_ACQ)
590				XFRM_INC_STATS(net, LINUX_MIB_XFRMACQUIREERROR);
591			else
592				XFRM_INC_STATS(net,
593					       LINUX_MIB_XFRMINSTATEINVALID);
594			goto drop_unlock;
595		}
596
597		if ((x->encap ? x->encap->encap_type : 0) != encap_type) {
598			XFRM_INC_STATS(net, LINUX_MIB_XFRMINSTATEMISMATCH);
599			goto drop_unlock;
600		}
601
602		if (xfrm_replay_check(x, skb, seq)) {
603			XFRM_INC_STATS(net, LINUX_MIB_XFRMINSTATESEQERROR);
604			goto drop_unlock;
605		}
606
607		if (xfrm_state_check_expire(x)) {
608			XFRM_INC_STATS(net, LINUX_MIB_XFRMINSTATEEXPIRED);
609			goto drop_unlock;
610		}
611
612		spin_unlock(&x->lock);
613
614		if (xfrm_tunnel_check(skb, x, family)) {
615			XFRM_INC_STATS(net, LINUX_MIB_XFRMINSTATEMODEERROR);
616			goto drop;
617		}
618
619		seq_hi = htonl(xfrm_replay_seqhi(x, seq));
620
621		XFRM_SKB_CB(skb)->seq.input.low = seq;
622		XFRM_SKB_CB(skb)->seq.input.hi = seq_hi;
623
624		dev_hold(skb->dev);
625
626		if (crypto_done)
627			nexthdr = x->type_offload->input_tail(x, skb);
628		else
629			nexthdr = x->type->input(x, skb);
630
631		if (nexthdr == -EINPROGRESS)
632			return 0;
633resume:
634		dev_put(skb->dev);
635
636		spin_lock(&x->lock);
637		if (nexthdr < 0) {
638			if (nexthdr == -EBADMSG) {
639				xfrm_audit_state_icvfail(x, skb,
640							 x->type->proto);
641				x->stats.integrity_failed++;
642			}
643			XFRM_INC_STATS(net, LINUX_MIB_XFRMINSTATEPROTOERROR);
644			goto drop_unlock;
645		}
646
647		/* only the first xfrm gets the encap type */
648		encap_type = 0;
649
650		if (xfrm_replay_recheck(x, skb, seq)) {
651			XFRM_INC_STATS(net, LINUX_MIB_XFRMINSTATESEQERROR);
652			goto drop_unlock;
653		}
654
655		xfrm_replay_advance(x, seq);
656
657		x->curlft.bytes += skb->len;
658		x->curlft.packets++;
659		x->lastused = ktime_get_real_seconds();
660
661		spin_unlock(&x->lock);
662
663		XFRM_MODE_SKB_CB(skb)->protocol = nexthdr;
664
665		if (xfrm_inner_mode_input(x, skb)) {
666			XFRM_INC_STATS(net, LINUX_MIB_XFRMINSTATEMODEERROR);
667			goto drop;
668		}
669
670		if (x->outer_mode.flags & XFRM_MODE_FLAG_TUNNEL) {
671			decaps = 1;
672			break;
673		}
674
675		/*
676		 * We need the inner address.  However, we only get here for
677		 * transport mode so the outer address is identical.
678		 */
679		daddr = &x->id.daddr;
680		family = x->props.family;
681
682		err = xfrm_parse_spi(skb, nexthdr, &spi, &seq);
683		if (err < 0) {
684			XFRM_INC_STATS(net, LINUX_MIB_XFRMINHDRERROR);
685			goto drop;
686		}
687		crypto_done = false;
688	} while (!err);
689
690	err = xfrm_rcv_cb(skb, family, x->type->proto, 0);
691	if (err)
692		goto drop;
693
694	nf_reset_ct(skb);
695
696	if (decaps) {
697		sp = skb_sec_path(skb);
698		if (sp)
699			sp->olen = 0;
700		if (skb_valid_dst(skb))
701			skb_dst_drop(skb);
702		gro_cells_receive(&gro_cells, skb);
703		return 0;
704	} else {
705		xo = xfrm_offload(skb);
706		if (xo)
707			xfrm_gro = xo->flags & XFRM_GRO;
708
709		err = -EAFNOSUPPORT;
710		rcu_read_lock();
711		afinfo = xfrm_state_afinfo_get_rcu(x->props.family);
712		if (likely(afinfo))
713			err = afinfo->transport_finish(skb, xfrm_gro || async);
714		rcu_read_unlock();
715		if (xfrm_gro) {
716			sp = skb_sec_path(skb);
717			if (sp)
718				sp->olen = 0;
719			if (skb_valid_dst(skb))
720				skb_dst_drop(skb);
721			gro_cells_receive(&gro_cells, skb);
722			return err;
723		}
724
725		return err;
726	}
727
728drop_unlock:
729	spin_unlock(&x->lock);
730drop:
731	xfrm_rcv_cb(skb, family, x && x->type ? x->type->proto : nexthdr, -1);
732	kfree_skb(skb);
733	return 0;
734}
735EXPORT_SYMBOL(xfrm_input);
736
737int xfrm_input_resume(struct sk_buff *skb, int nexthdr)
738{
739	return xfrm_input(skb, nexthdr, 0, -1);
740}
741EXPORT_SYMBOL(xfrm_input_resume);
742
743static void xfrm_trans_reinject(struct work_struct *work)
744{
745	struct xfrm_trans_tasklet *trans = container_of(work, struct xfrm_trans_tasklet, work);
746	struct sk_buff_head queue;
747	struct sk_buff *skb;
748
749	__skb_queue_head_init(&queue);
750	spin_lock_bh(&trans->queue_lock);
751	skb_queue_splice_init(&trans->queue, &queue);
752	spin_unlock_bh(&trans->queue_lock);
753
754	local_bh_disable();
755	while ((skb = __skb_dequeue(&queue)))
756		XFRM_TRANS_SKB_CB(skb)->finish(XFRM_TRANS_SKB_CB(skb)->net,
757					       NULL, skb);
758	local_bh_enable();
759}
760
761int xfrm_trans_queue_net(struct net *net, struct sk_buff *skb,
762			 int (*finish)(struct net *, struct sock *,
763				       struct sk_buff *))
764{
765	struct xfrm_trans_tasklet *trans;
766
767	trans = this_cpu_ptr(&xfrm_trans_tasklet);
768
769	if (skb_queue_len(&trans->queue) >= READ_ONCE(netdev_max_backlog))
770		return -ENOBUFS;
771
772	BUILD_BUG_ON(sizeof(struct xfrm_trans_cb) > sizeof(skb->cb));
773
774	XFRM_TRANS_SKB_CB(skb)->finish = finish;
775	XFRM_TRANS_SKB_CB(skb)->net = net;
776	spin_lock_bh(&trans->queue_lock);
777	__skb_queue_tail(&trans->queue, skb);
778	spin_unlock_bh(&trans->queue_lock);
779	schedule_work(&trans->work);
780	return 0;
781}
782EXPORT_SYMBOL(xfrm_trans_queue_net);
783
784int xfrm_trans_queue(struct sk_buff *skb,
785		     int (*finish)(struct net *, struct sock *,
786				   struct sk_buff *))
787{
788	return xfrm_trans_queue_net(dev_net(skb->dev), skb, finish);
789}
790EXPORT_SYMBOL(xfrm_trans_queue);
791
792void __init xfrm_input_init(void)
793{
794	int err;
795	int i;
796
797	init_dummy_netdev(&xfrm_napi_dev);
798	err = gro_cells_init(&gro_cells, &xfrm_napi_dev);
799	if (err)
800		gro_cells.cells = NULL;
801
802	for_each_possible_cpu(i) {
803		struct xfrm_trans_tasklet *trans;
804
805		trans = &per_cpu(xfrm_trans_tasklet, i);
806		spin_lock_init(&trans->queue_lock);
807		__skb_queue_head_init(&trans->queue);
808		INIT_WORK(&trans->work, xfrm_trans_reinject);
809	}
810}
811