xref: /openbmc/linux/net/ipv4/inet_diag.c (revision dbf563ee) 
1 // SPDX-License-Identifier: GPL-2.0-or-later
2 /*
3  * inet_diag.c	Module for monitoring INET transport protocols sockets.
4  *
5  * Authors:	Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
6  */
7 
8 #include <linux/kernel.h>
9 #include <linux/module.h>
10 #include <linux/types.h>
11 #include <linux/fcntl.h>
12 #include <linux/random.h>
13 #include <linux/slab.h>
14 #include <linux/cache.h>
15 #include <linux/init.h>
16 #include <linux/time.h>
17 
18 #include <net/icmp.h>
19 #include <net/tcp.h>
20 #include <net/ipv6.h>
21 #include <net/inet_common.h>
22 #include <net/inet_connection_sock.h>
23 #include <net/inet_hashtables.h>
24 #include <net/inet_timewait_sock.h>
25 #include <net/inet6_hashtables.h>
26 #include <net/bpf_sk_storage.h>
27 #include <net/netlink.h>
28 
29 #include <linux/inet.h>
30 #include <linux/stddef.h>
31 
32 #include <linux/inet_diag.h>
33 #include <linux/sock_diag.h>
34 
35 static const struct inet_diag_handler **inet_diag_table;
36 
37 struct inet_diag_entry {
38 	const __be32 *saddr;
39 	const __be32 *daddr;
40 	u16 sport;
41 	u16 dport;
42 	u16 family;
43 	u16 userlocks;
44 	u32 ifindex;
45 	u32 mark;
46 #ifdef CONFIG_SOCK_CGROUP_DATA
47 	u64 cgroup_id;
48 #endif
49 };
50 
51 static DEFINE_MUTEX(inet_diag_table_mutex);
52 
53 static const struct inet_diag_handler *inet_diag_lock_handler(int proto)
54 {
55 	if (proto < 0 || proto >= IPPROTO_MAX) {
56 		mutex_lock(&inet_diag_table_mutex);
57 		return ERR_PTR(-ENOENT);
58 	}
59 
60 	if (!inet_diag_table[proto])
61 		sock_load_diag_module(AF_INET, proto);
62 
63 	mutex_lock(&inet_diag_table_mutex);
64 	if (!inet_diag_table[proto])
65 		return ERR_PTR(-ENOENT);
66 
67 	return inet_diag_table[proto];
68 }
69 
70 static void inet_diag_unlock_handler(const struct inet_diag_handler *handler)
71 {
72 	mutex_unlock(&inet_diag_table_mutex);
73 }
74 
75 void inet_diag_msg_common_fill(struct inet_diag_msg *r, struct sock *sk)
76 {
77 	r->idiag_family = sk->sk_family;
78 
79 	r->id.idiag_sport = htons(sk->sk_num);
80 	r->id.idiag_dport = sk->sk_dport;
81 	r->id.idiag_if = sk->sk_bound_dev_if;
82 	sock_diag_save_cookie(sk, r->id.idiag_cookie);
83 
84 #if IS_ENABLED(CONFIG_IPV6)
85 	if (sk->sk_family == AF_INET6) {
86 		*(struct in6_addr *)r->id.idiag_src = sk->sk_v6_rcv_saddr;
87 		*(struct in6_addr *)r->id.idiag_dst = sk->sk_v6_daddr;
88 	} else
89 #endif
90 	{
91 	memset(&r->id.idiag_src, 0, sizeof(r->id.idiag_src));
92 	memset(&r->id.idiag_dst, 0, sizeof(r->id.idiag_dst));
93 
94 	r->id.idiag_src[0] = sk->sk_rcv_saddr;
95 	r->id.idiag_dst[0] = sk->sk_daddr;
96 	}
97 }
98 EXPORT_SYMBOL_GPL(inet_diag_msg_common_fill);
99 
100 static size_t inet_sk_attr_size(struct sock *sk,
101 				const struct inet_diag_req_v2 *req,
102 				bool net_admin)
103 {
104 	const struct inet_diag_handler *handler;
105 	size_t aux = 0;
106 
107 	handler = inet_diag_table[req->sdiag_protocol];
108 	if (handler && handler->idiag_get_aux_size)
109 		aux = handler->idiag_get_aux_size(sk, net_admin);
110 
111 	return	  nla_total_size(sizeof(struct tcp_info))
112 		+ nla_total_size(sizeof(struct inet_diag_msg))
113 		+ inet_diag_msg_attrs_size()
114 		+ nla_total_size(sizeof(struct inet_diag_meminfo))
115 		+ nla_total_size(SK_MEMINFO_VARS * sizeof(u32))
116 		+ nla_total_size(TCP_CA_NAME_MAX)
117 		+ nla_total_size(sizeof(struct tcpvegas_info))
118 		+ aux
119 		+ 64;
120 }
121 
122 int inet_diag_msg_attrs_fill(struct sock *sk, struct sk_buff *skb,
123 			     struct inet_diag_msg *r, int ext,
124 			     struct user_namespace *user_ns,
125 			     bool net_admin)
126 {
127 	const struct inet_sock *inet = inet_sk(sk);
128 
129 	if (nla_put_u8(skb, INET_DIAG_SHUTDOWN, sk->sk_shutdown))
130 		goto errout;
131 
132 	/* IPv6 dual-stack sockets use inet->tos for IPv4 connections,
133 	 * hence this needs to be included regardless of socket family.
134 	 */
135 	if (ext & (1 << (INET_DIAG_TOS - 1)))
136 		if (nla_put_u8(skb, INET_DIAG_TOS, inet->tos) < 0)
137 			goto errout;
138 
139 #if IS_ENABLED(CONFIG_IPV6)
140 	if (r->idiag_family == AF_INET6) {
141 		if (ext & (1 << (INET_DIAG_TCLASS - 1)))
142 			if (nla_put_u8(skb, INET_DIAG_TCLASS,
143 				       inet6_sk(sk)->tclass) < 0)
144 				goto errout;
145 
146 		if (((1 << sk->sk_state) & (TCPF_LISTEN | TCPF_CLOSE)) &&
147 		    nla_put_u8(skb, INET_DIAG_SKV6ONLY, ipv6_only_sock(sk)))
148 			goto errout;
149 	}
150 #endif
151 
152 	if (net_admin && nla_put_u32(skb, INET_DIAG_MARK, sk->sk_mark))
153 		goto errout;
154 
155 	if (ext & (1 << (INET_DIAG_CLASS_ID - 1)) ||
156 	    ext & (1 << (INET_DIAG_TCLASS - 1))) {
157 		u32 classid = 0;
158 
159 #ifdef CONFIG_SOCK_CGROUP_DATA
160 		classid = sock_cgroup_classid(&sk->sk_cgrp_data);
161 #endif
162 		/* Fallback to socket priority if class id isn't set.
163 		 * Classful qdiscs use it as direct reference to class.
164 		 * For cgroup2 classid is always zero.
165 		 */
166 		if (!classid)
167 			classid = sk->sk_priority;
168 
169 		if (nla_put_u32(skb, INET_DIAG_CLASS_ID, classid))
170 			goto errout;
171 	}
172 
173 #ifdef CONFIG_SOCK_CGROUP_DATA
174 	if (nla_put_u64_64bit(skb, INET_DIAG_CGROUP_ID,
175 			      cgroup_id(sock_cgroup_ptr(&sk->sk_cgrp_data)),
176 			      INET_DIAG_PAD))
177 		goto errout;
178 #endif
179 
180 	r->idiag_uid = from_kuid_munged(user_ns, sock_i_uid(sk));
181 	r->idiag_inode = sock_i_ino(sk);
182 
183 	return 0;
184 errout:
185 	return 1;
186 }
187 EXPORT_SYMBOL_GPL(inet_diag_msg_attrs_fill);
188 
189 static int inet_diag_parse_attrs(const struct nlmsghdr *nlh, int hdrlen,
190 				 struct nlattr **req_nlas)
191 {
192 	struct nlattr *nla;
193 	int remaining;
194 
195 	nlmsg_for_each_attr(nla, nlh, hdrlen, remaining) {
196 		int type = nla_type(nla);
197 
198 		if (type == INET_DIAG_REQ_PROTOCOL && nla_len(nla) != sizeof(u32))
199 			return -EINVAL;
200 
201 		if (type < __INET_DIAG_REQ_MAX)
202 			req_nlas[type] = nla;
203 	}
204 	return 0;
205 }
206 
207 static int inet_diag_get_protocol(const struct inet_diag_req_v2 *req,
208 				  const struct inet_diag_dump_data *data)
209 {
210 	if (data->req_nlas[INET_DIAG_REQ_PROTOCOL])
211 		return nla_get_u32(data->req_nlas[INET_DIAG_REQ_PROTOCOL]);
212 	return req->sdiag_protocol;
213 }
214 
215 #define MAX_DUMP_ALLOC_SIZE (KMALLOC_MAX_SIZE - SKB_DATA_ALIGN(sizeof(struct skb_shared_info)))
216 
217 int inet_sk_diag_fill(struct sock *sk, struct inet_connection_sock *icsk,
218 		      struct sk_buff *skb, struct netlink_callback *cb,
219 		      const struct inet_diag_req_v2 *req,
220 		      u16 nlmsg_flags, bool net_admin)
221 {
222 	const struct tcp_congestion_ops *ca_ops;
223 	const struct inet_diag_handler *handler;
224 	struct inet_diag_dump_data *cb_data;
225 	int ext = req->idiag_ext;
226 	struct inet_diag_msg *r;
227 	struct nlmsghdr  *nlh;
228 	struct nlattr *attr;
229 	void *info = NULL;
230 
231 	cb_data = cb->data;
232 	handler = inet_diag_table[inet_diag_get_protocol(req, cb_data)];
233 	BUG_ON(!handler);
234 
235 	nlh = nlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
236 			cb->nlh->nlmsg_type, sizeof(*r), nlmsg_flags);
237 	if (!nlh)
238 		return -EMSGSIZE;
239 
240 	r = nlmsg_data(nlh);
241 	BUG_ON(!sk_fullsock(sk));
242 
243 	inet_diag_msg_common_fill(r, sk);
244 	r->idiag_state = sk->sk_state;
245 	r->idiag_timer = 0;
246 	r->idiag_retrans = 0;
247 
248 	if (inet_diag_msg_attrs_fill(sk, skb, r, ext,
249 				     sk_user_ns(NETLINK_CB(cb->skb).sk),
250 				     net_admin))
251 		goto errout;
252 
253 	if (ext & (1 << (INET_DIAG_MEMINFO - 1))) {
254 		struct inet_diag_meminfo minfo = {
255 			.idiag_rmem = sk_rmem_alloc_get(sk),
256 			.idiag_wmem = READ_ONCE(sk->sk_wmem_queued),
257 			.idiag_fmem = sk->sk_forward_alloc,
258 			.idiag_tmem = sk_wmem_alloc_get(sk),
259 		};
260 
261 		if (nla_put(skb, INET_DIAG_MEMINFO, sizeof(minfo), &minfo) < 0)
262 			goto errout;
263 	}
264 
265 	if (ext & (1 << (INET_DIAG_SKMEMINFO - 1)))
266 		if (sock_diag_put_meminfo(sk, skb, INET_DIAG_SKMEMINFO))
267 			goto errout;
268 
269 	/*
270 	 * RAW sockets might have user-defined protocols assigned,
271 	 * so report the one supplied on socket creation.
272 	 */
273 	if (sk->sk_type == SOCK_RAW) {
274 		if (nla_put_u8(skb, INET_DIAG_PROTOCOL, sk->sk_protocol))
275 			goto errout;
276 	}
277 
278 	if (!icsk) {
279 		handler->idiag_get_info(sk, r, NULL);
280 		goto out;
281 	}
282 
283 	if (icsk->icsk_pending == ICSK_TIME_RETRANS ||
284 	    icsk->icsk_pending == ICSK_TIME_REO_TIMEOUT ||
285 	    icsk->icsk_pending == ICSK_TIME_LOSS_PROBE) {
286 		r->idiag_timer = 1;
287 		r->idiag_retrans = icsk->icsk_retransmits;
288 		r->idiag_expires =
289 			jiffies_delta_to_msecs(icsk->icsk_timeout - jiffies);
290 	} else if (icsk->icsk_pending == ICSK_TIME_PROBE0) {
291 		r->idiag_timer = 4;
292 		r->idiag_retrans = icsk->icsk_probes_out;
293 		r->idiag_expires =
294 			jiffies_delta_to_msecs(icsk->icsk_timeout - jiffies);
295 	} else if (timer_pending(&sk->sk_timer)) {
296 		r->idiag_timer = 2;
297 		r->idiag_retrans = icsk->icsk_probes_out;
298 		r->idiag_expires =
299 			jiffies_delta_to_msecs(sk->sk_timer.expires - jiffies);
300 	} else {
301 		r->idiag_timer = 0;
302 		r->idiag_expires = 0;
303 	}
304 
305 	if ((ext & (1 << (INET_DIAG_INFO - 1))) && handler->idiag_info_size) {
306 		attr = nla_reserve_64bit(skb, INET_DIAG_INFO,
307 					 handler->idiag_info_size,
308 					 INET_DIAG_PAD);
309 		if (!attr)
310 			goto errout;
311 
312 		info = nla_data(attr);
313 	}
314 
315 	if (ext & (1 << (INET_DIAG_CONG - 1))) {
316 		int err = 0;
317 
318 		rcu_read_lock();
319 		ca_ops = READ_ONCE(icsk->icsk_ca_ops);
320 		if (ca_ops)
321 			err = nla_put_string(skb, INET_DIAG_CONG, ca_ops->name);
322 		rcu_read_unlock();
323 		if (err < 0)
324 			goto errout;
325 	}
326 
327 	handler->idiag_get_info(sk, r, info);
328 
329 	if (ext & (1 << (INET_DIAG_INFO - 1)) && handler->idiag_get_aux)
330 		if (handler->idiag_get_aux(sk, net_admin, skb) < 0)
331 			goto errout;
332 
333 	if (sk->sk_state < TCP_TIME_WAIT) {
334 		union tcp_cc_info info;
335 		size_t sz = 0;
336 		int attr;
337 
338 		rcu_read_lock();
339 		ca_ops = READ_ONCE(icsk->icsk_ca_ops);
340 		if (ca_ops && ca_ops->get_info)
341 			sz = ca_ops->get_info(sk, ext, &attr, &info);
342 		rcu_read_unlock();
343 		if (sz && nla_put(skb, attr, sz, &info) < 0)
344 			goto errout;
345 	}
346 
347 	/* Keep it at the end for potential retry with a larger skb,
348 	 * or else do best-effort fitting, which is only done for the
349 	 * first_nlmsg.
350 	 */
351 	if (cb_data->bpf_stg_diag) {
352 		bool first_nlmsg = ((unsigned char *)nlh == skb->data);
353 		unsigned int prev_min_dump_alloc;
354 		unsigned int total_nla_size = 0;
355 		unsigned int msg_len;
356 		int err;
357 
358 		msg_len = skb_tail_pointer(skb) - (unsigned char *)nlh;
359 		err = bpf_sk_storage_diag_put(cb_data->bpf_stg_diag, sk, skb,
360 					      INET_DIAG_SK_BPF_STORAGES,
361 					      &total_nla_size);
362 
363 		if (!err)
364 			goto out;
365 
366 		total_nla_size += msg_len;
367 		prev_min_dump_alloc = cb->min_dump_alloc;
368 		if (total_nla_size > prev_min_dump_alloc)
369 			cb->min_dump_alloc = min_t(u32, total_nla_size,
370 						   MAX_DUMP_ALLOC_SIZE);
371 
372 		if (!first_nlmsg)
373 			goto errout;
374 
375 		if (cb->min_dump_alloc > prev_min_dump_alloc)
376 			/* Retry with pskb_expand_head() with
377 			 * __GFP_DIRECT_RECLAIM
378 			 */
379 			goto errout;
380 
381 		WARN_ON_ONCE(total_nla_size <= prev_min_dump_alloc);
382 
383 		/* Send what we have for this sk
384 		 * and move on to the next sk in the following
385 		 * dump()
386 		 */
387 	}
388 
389 out:
390 	nlmsg_end(skb, nlh);
391 	return 0;
392 
393 errout:
394 	nlmsg_cancel(skb, nlh);
395 	return -EMSGSIZE;
396 }
397 EXPORT_SYMBOL_GPL(inet_sk_diag_fill);
398 
399 static int inet_twsk_diag_fill(struct sock *sk,
400 			       struct sk_buff *skb,
401 			       struct netlink_callback *cb,
402 			       u16 nlmsg_flags)
403 {
404 	struct inet_timewait_sock *tw = inet_twsk(sk);
405 	struct inet_diag_msg *r;
406 	struct nlmsghdr *nlh;
407 	long tmo;
408 
409 	nlh = nlmsg_put(skb, NETLINK_CB(cb->skb).portid,
410 			cb->nlh->nlmsg_seq, cb->nlh->nlmsg_type,
411 			sizeof(*r), nlmsg_flags);
412 	if (!nlh)
413 		return -EMSGSIZE;
414 
415 	r = nlmsg_data(nlh);
416 	BUG_ON(tw->tw_state != TCP_TIME_WAIT);
417 
418 	inet_diag_msg_common_fill(r, sk);
419 	r->idiag_retrans      = 0;
420 
421 	r->idiag_state	      = tw->tw_substate;
422 	r->idiag_timer	      = 3;
423 	tmo = tw->tw_timer.expires - jiffies;
424 	r->idiag_expires      = jiffies_delta_to_msecs(tmo);
425 	r->idiag_rqueue	      = 0;
426 	r->idiag_wqueue	      = 0;
427 	r->idiag_uid	      = 0;
428 	r->idiag_inode	      = 0;
429 
430 	nlmsg_end(skb, nlh);
431 	return 0;
432 }
433 
434 static int inet_req_diag_fill(struct sock *sk, struct sk_buff *skb,
435 			      struct netlink_callback *cb,
436 			      u16 nlmsg_flags, bool net_admin)
437 {
438 	struct request_sock *reqsk = inet_reqsk(sk);
439 	struct inet_diag_msg *r;
440 	struct nlmsghdr *nlh;
441 	long tmo;
442 
443 	nlh = nlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
444 			cb->nlh->nlmsg_type, sizeof(*r), nlmsg_flags);
445 	if (!nlh)
446 		return -EMSGSIZE;
447 
448 	r = nlmsg_data(nlh);
449 	inet_diag_msg_common_fill(r, sk);
450 	r->idiag_state = TCP_SYN_RECV;
451 	r->idiag_timer = 1;
452 	r->idiag_retrans = reqsk->num_retrans;
453 
454 	BUILD_BUG_ON(offsetof(struct inet_request_sock, ir_cookie) !=
455 		     offsetof(struct sock, sk_cookie));
456 
457 	tmo = inet_reqsk(sk)->rsk_timer.expires - jiffies;
458 	r->idiag_expires = jiffies_delta_to_msecs(tmo);
459 	r->idiag_rqueue	= 0;
460 	r->idiag_wqueue	= 0;
461 	r->idiag_uid	= 0;
462 	r->idiag_inode	= 0;
463 
464 	if (net_admin && nla_put_u32(skb, INET_DIAG_MARK,
465 				     inet_rsk(reqsk)->ir_mark))
466 		return -EMSGSIZE;
467 
468 	nlmsg_end(skb, nlh);
469 	return 0;
470 }
471 
472 static int sk_diag_fill(struct sock *sk, struct sk_buff *skb,
473 			struct netlink_callback *cb,
474 			const struct inet_diag_req_v2 *r,
475 			u16 nlmsg_flags, bool net_admin)
476 {
477 	if (sk->sk_state == TCP_TIME_WAIT)
478 		return inet_twsk_diag_fill(sk, skb, cb, nlmsg_flags);
479 
480 	if (sk->sk_state == TCP_NEW_SYN_RECV)
481 		return inet_req_diag_fill(sk, skb, cb, nlmsg_flags, net_admin);
482 
483 	return inet_sk_diag_fill(sk, inet_csk(sk), skb, cb, r, nlmsg_flags,
484 				 net_admin);
485 }
486 
487 struct sock *inet_diag_find_one_icsk(struct net *net,
488 				     struct inet_hashinfo *hashinfo,
489 				     const struct inet_diag_req_v2 *req)
490 {
491 	struct sock *sk;
492 
493 	rcu_read_lock();
494 	if (req->sdiag_family == AF_INET)
495 		sk = inet_lookup(net, hashinfo, NULL, 0, req->id.idiag_dst[0],
496 				 req->id.idiag_dport, req->id.idiag_src[0],
497 				 req->id.idiag_sport, req->id.idiag_if);
498 #if IS_ENABLED(CONFIG_IPV6)
499 	else if (req->sdiag_family == AF_INET6) {
500 		if (ipv6_addr_v4mapped((struct in6_addr *)req->id.idiag_dst) &&
501 		    ipv6_addr_v4mapped((struct in6_addr *)req->id.idiag_src))
502 			sk = inet_lookup(net, hashinfo, NULL, 0, req->id.idiag_dst[3],
503 					 req->id.idiag_dport, req->id.idiag_src[3],
504 					 req->id.idiag_sport, req->id.idiag_if);
505 		else
506 			sk = inet6_lookup(net, hashinfo, NULL, 0,
507 					  (struct in6_addr *)req->id.idiag_dst,
508 					  req->id.idiag_dport,
509 					  (struct in6_addr *)req->id.idiag_src,
510 					  req->id.idiag_sport,
511 					  req->id.idiag_if);
512 	}
513 #endif
514 	else {
515 		rcu_read_unlock();
516 		return ERR_PTR(-EINVAL);
517 	}
518 	rcu_read_unlock();
519 	if (!sk)
520 		return ERR_PTR(-ENOENT);
521 
522 	if (sock_diag_check_cookie(sk, req->id.idiag_cookie)) {
523 		sock_gen_put(sk);
524 		return ERR_PTR(-ENOENT);
525 	}
526 
527 	return sk;
528 }
529 EXPORT_SYMBOL_GPL(inet_diag_find_one_icsk);
530 
531 int inet_diag_dump_one_icsk(struct inet_hashinfo *hashinfo,
532 			    struct netlink_callback *cb,
533 			    const struct inet_diag_req_v2 *req)
534 {
535 	struct sk_buff *in_skb = cb->skb;
536 	bool net_admin = netlink_net_capable(in_skb, CAP_NET_ADMIN);
537 	struct net *net = sock_net(in_skb->sk);
538 	struct sk_buff *rep;
539 	struct sock *sk;
540 	int err;
541 
542 	sk = inet_diag_find_one_icsk(net, hashinfo, req);
543 	if (IS_ERR(sk))
544 		return PTR_ERR(sk);
545 
546 	rep = nlmsg_new(inet_sk_attr_size(sk, req, net_admin), GFP_KERNEL);
547 	if (!rep) {
548 		err = -ENOMEM;
549 		goto out;
550 	}
551 
552 	err = sk_diag_fill(sk, rep, cb, req, 0, net_admin);
553 	if (err < 0) {
554 		WARN_ON(err == -EMSGSIZE);
555 		nlmsg_free(rep);
556 		goto out;
557 	}
558 	err = netlink_unicast(net->diag_nlsk, rep, NETLINK_CB(in_skb).portid,
559 			      MSG_DONTWAIT);
560 	if (err > 0)
561 		err = 0;
562 
563 out:
564 	if (sk)
565 		sock_gen_put(sk);
566 
567 	return err;
568 }
569 EXPORT_SYMBOL_GPL(inet_diag_dump_one_icsk);
570 
571 static int inet_diag_cmd_exact(int cmd, struct sk_buff *in_skb,
572 			       const struct nlmsghdr *nlh,
573 			       int hdrlen,
574 			       const struct inet_diag_req_v2 *req)
575 {
576 	const struct inet_diag_handler *handler;
577 	struct inet_diag_dump_data dump_data;
578 	int err, protocol;
579 
580 	memset(&dump_data, 0, sizeof(dump_data));
581 	err = inet_diag_parse_attrs(nlh, hdrlen, dump_data.req_nlas);
582 	if (err)
583 		return err;
584 
585 	protocol = inet_diag_get_protocol(req, &dump_data);
586 
587 	handler = inet_diag_lock_handler(protocol);
588 	if (IS_ERR(handler)) {
589 		err = PTR_ERR(handler);
590 	} else if (cmd == SOCK_DIAG_BY_FAMILY) {
591 		struct netlink_callback cb = {
592 			.nlh = nlh,
593 			.skb = in_skb,
594 			.data = &dump_data,
595 		};
596 		err = handler->dump_one(&cb, req);
597 	} else if (cmd == SOCK_DESTROY && handler->destroy) {
598 		err = handler->destroy(in_skb, req);
599 	} else {
600 		err = -EOPNOTSUPP;
601 	}
602 	inet_diag_unlock_handler(handler);
603 
604 	return err;
605 }
606 
607 static int bitstring_match(const __be32 *a1, const __be32 *a2, int bits)
608 {
609 	int words = bits >> 5;
610 
611 	bits &= 0x1f;
612 
613 	if (words) {
614 		if (memcmp(a1, a2, words << 2))
615 			return 0;
616 	}
617 	if (bits) {
618 		__be32 w1, w2;
619 		__be32 mask;
620 
621 		w1 = a1[words];
622 		w2 = a2[words];
623 
624 		mask = htonl((0xffffffff) << (32 - bits));
625 
626 		if ((w1 ^ w2) & mask)
627 			return 0;
628 	}
629 
630 	return 1;
631 }
632 
633 static int inet_diag_bc_run(const struct nlattr *_bc,
634 			    const struct inet_diag_entry *entry)
635 {
636 	const void *bc = nla_data(_bc);
637 	int len = nla_len(_bc);
638 
639 	while (len > 0) {
640 		int yes = 1;
641 		const struct inet_diag_bc_op *op = bc;
642 
643 		switch (op->code) {
644 		case INET_DIAG_BC_NOP:
645 			break;
646 		case INET_DIAG_BC_JMP:
647 			yes = 0;
648 			break;
649 		case INET_DIAG_BC_S_EQ:
650 			yes = entry->sport == op[1].no;
651 			break;
652 		case INET_DIAG_BC_S_GE:
653 			yes = entry->sport >= op[1].no;
654 			break;
655 		case INET_DIAG_BC_S_LE:
656 			yes = entry->sport <= op[1].no;
657 			break;
658 		case INET_DIAG_BC_D_EQ:
659 			yes = entry->dport == op[1].no;
660 			break;
661 		case INET_DIAG_BC_D_GE:
662 			yes = entry->dport >= op[1].no;
663 			break;
664 		case INET_DIAG_BC_D_LE:
665 			yes = entry->dport <= op[1].no;
666 			break;
667 		case INET_DIAG_BC_AUTO:
668 			yes = !(entry->userlocks & SOCK_BINDPORT_LOCK);
669 			break;
670 		case INET_DIAG_BC_S_COND:
671 		case INET_DIAG_BC_D_COND: {
672 			const struct inet_diag_hostcond *cond;
673 			const __be32 *addr;
674 
675 			cond = (const struct inet_diag_hostcond *)(op + 1);
676 			if (cond->port != -1 &&
677 			    cond->port != (op->code == INET_DIAG_BC_S_COND ?
678 					     entry->sport : entry->dport)) {
679 				yes = 0;
680 				break;
681 			}
682 
683 			if (op->code == INET_DIAG_BC_S_COND)
684 				addr = entry->saddr;
685 			else
686 				addr = entry->daddr;
687 
688 			if (cond->family != AF_UNSPEC &&
689 			    cond->family != entry->family) {
690 				if (entry->family == AF_INET6 &&
691 				    cond->family == AF_INET) {
692 					if (addr[0] == 0 && addr[1] == 0 &&
693 					    addr[2] == htonl(0xffff) &&
694 					    bitstring_match(addr + 3,
695 							    cond->addr,
696 							    cond->prefix_len))
697 						break;
698 				}
699 				yes = 0;
700 				break;
701 			}
702 
703 			if (cond->prefix_len == 0)
704 				break;
705 			if (bitstring_match(addr, cond->addr,
706 					    cond->prefix_len))
707 				break;
708 			yes = 0;
709 			break;
710 		}
711 		case INET_DIAG_BC_DEV_COND: {
712 			u32 ifindex;
713 
714 			ifindex = *((const u32 *)(op + 1));
715 			if (ifindex != entry->ifindex)
716 				yes = 0;
717 			break;
718 		}
719 		case INET_DIAG_BC_MARK_COND: {
720 			struct inet_diag_markcond *cond;
721 
722 			cond = (struct inet_diag_markcond *)(op + 1);
723 			if ((entry->mark & cond->mask) != cond->mark)
724 				yes = 0;
725 			break;
726 		}
727 #ifdef CONFIG_SOCK_CGROUP_DATA
728 		case INET_DIAG_BC_CGROUP_COND: {
729 			u64 cgroup_id;
730 
731 			cgroup_id = get_unaligned((const u64 *)(op + 1));
732 			if (cgroup_id != entry->cgroup_id)
733 				yes = 0;
734 			break;
735 		}
736 #endif
737 		}
738 
739 		if (yes) {
740 			len -= op->yes;
741 			bc += op->yes;
742 		} else {
743 			len -= op->no;
744 			bc += op->no;
745 		}
746 	}
747 	return len == 0;
748 }
749 
750 /* This helper is available for all sockets (ESTABLISH, TIMEWAIT, SYN_RECV)
751  */
752 static void entry_fill_addrs(struct inet_diag_entry *entry,
753 			     const struct sock *sk)
754 {
755 #if IS_ENABLED(CONFIG_IPV6)
756 	if (sk->sk_family == AF_INET6) {
757 		entry->saddr = sk->sk_v6_rcv_saddr.s6_addr32;
758 		entry->daddr = sk->sk_v6_daddr.s6_addr32;
759 	} else
760 #endif
761 	{
762 		entry->saddr = &sk->sk_rcv_saddr;
763 		entry->daddr = &sk->sk_daddr;
764 	}
765 }
766 
767 int inet_diag_bc_sk(const struct nlattr *bc, struct sock *sk)
768 {
769 	struct inet_sock *inet = inet_sk(sk);
770 	struct inet_diag_entry entry;
771 
772 	if (!bc)
773 		return 1;
774 
775 	entry.family = sk->sk_family;
776 	entry_fill_addrs(&entry, sk);
777 	entry.sport = inet->inet_num;
778 	entry.dport = ntohs(inet->inet_dport);
779 	entry.ifindex = sk->sk_bound_dev_if;
780 	entry.userlocks = sk_fullsock(sk) ? sk->sk_userlocks : 0;
781 	if (sk_fullsock(sk))
782 		entry.mark = sk->sk_mark;
783 	else if (sk->sk_state == TCP_NEW_SYN_RECV)
784 		entry.mark = inet_rsk(inet_reqsk(sk))->ir_mark;
785 	else
786 		entry.mark = 0;
787 #ifdef CONFIG_SOCK_CGROUP_DATA
788 	entry.cgroup_id = sk_fullsock(sk) ?
789 		cgroup_id(sock_cgroup_ptr(&sk->sk_cgrp_data)) : 0;
790 #endif
791 
792 	return inet_diag_bc_run(bc, &entry);
793 }
794 EXPORT_SYMBOL_GPL(inet_diag_bc_sk);
795 
796 static int valid_cc(const void *bc, int len, int cc)
797 {
798 	while (len >= 0) {
799 		const struct inet_diag_bc_op *op = bc;
800 
801 		if (cc > len)
802 			return 0;
803 		if (cc == len)
804 			return 1;
805 		if (op->yes < 4 || op->yes & 3)
806 			return 0;
807 		len -= op->yes;
808 		bc  += op->yes;
809 	}
810 	return 0;
811 }
812 
813 /* data is u32 ifindex */
814 static bool valid_devcond(const struct inet_diag_bc_op *op, int len,
815 			  int *min_len)
816 {
817 	/* Check ifindex space. */
818 	*min_len += sizeof(u32);
819 	if (len < *min_len)
820 		return false;
821 
822 	return true;
823 }
824 /* Validate an inet_diag_hostcond. */
825 static bool valid_hostcond(const struct inet_diag_bc_op *op, int len,
826 			   int *min_len)
827 {
828 	struct inet_diag_hostcond *cond;
829 	int addr_len;
830 
831 	/* Check hostcond space. */
832 	*min_len += sizeof(struct inet_diag_hostcond);
833 	if (len < *min_len)
834 		return false;
835 	cond = (struct inet_diag_hostcond *)(op + 1);
836 
837 	/* Check address family and address length. */
838 	switch (cond->family) {
839 	case AF_UNSPEC:
840 		addr_len = 0;
841 		break;
842 	case AF_INET:
843 		addr_len = sizeof(struct in_addr);
844 		break;
845 	case AF_INET6:
846 		addr_len = sizeof(struct in6_addr);
847 		break;
848 	default:
849 		return false;
850 	}
851 	*min_len += addr_len;
852 	if (len < *min_len)
853 		return false;
854 
855 	/* Check prefix length (in bits) vs address length (in bytes). */
856 	if (cond->prefix_len > 8 * addr_len)
857 		return false;
858 
859 	return true;
860 }
861 
862 /* Validate a port comparison operator. */
863 static bool valid_port_comparison(const struct inet_diag_bc_op *op,
864 				  int len, int *min_len)
865 {
866 	/* Port comparisons put the port in a follow-on inet_diag_bc_op. */
867 	*min_len += sizeof(struct inet_diag_bc_op);
868 	if (len < *min_len)
869 		return false;
870 	return true;
871 }
872 
873 static bool valid_markcond(const struct inet_diag_bc_op *op, int len,
874 			   int *min_len)
875 {
876 	*min_len += sizeof(struct inet_diag_markcond);
877 	return len >= *min_len;
878 }
879 
880 #ifdef CONFIG_SOCK_CGROUP_DATA
881 static bool valid_cgroupcond(const struct inet_diag_bc_op *op, int len,
882 			     int *min_len)
883 {
884 	*min_len += sizeof(u64);
885 	return len >= *min_len;
886 }
887 #endif
888 
889 static int inet_diag_bc_audit(const struct nlattr *attr,
890 			      const struct sk_buff *skb)
891 {
892 	bool net_admin = netlink_net_capable(skb, CAP_NET_ADMIN);
893 	const void *bytecode, *bc;
894 	int bytecode_len, len;
895 
896 	if (!attr || nla_len(attr) < sizeof(struct inet_diag_bc_op))
897 		return -EINVAL;
898 
899 	bytecode = bc = nla_data(attr);
900 	len = bytecode_len = nla_len(attr);
901 
902 	while (len > 0) {
903 		int min_len = sizeof(struct inet_diag_bc_op);
904 		const struct inet_diag_bc_op *op = bc;
905 
906 		switch (op->code) {
907 		case INET_DIAG_BC_S_COND:
908 		case INET_DIAG_BC_D_COND:
909 			if (!valid_hostcond(bc, len, &min_len))
910 				return -EINVAL;
911 			break;
912 		case INET_DIAG_BC_DEV_COND:
913 			if (!valid_devcond(bc, len, &min_len))
914 				return -EINVAL;
915 			break;
916 		case INET_DIAG_BC_S_EQ:
917 		case INET_DIAG_BC_S_GE:
918 		case INET_DIAG_BC_S_LE:
919 		case INET_DIAG_BC_D_EQ:
920 		case INET_DIAG_BC_D_GE:
921 		case INET_DIAG_BC_D_LE:
922 			if (!valid_port_comparison(bc, len, &min_len))
923 				return -EINVAL;
924 			break;
925 		case INET_DIAG_BC_MARK_COND:
926 			if (!net_admin)
927 				return -EPERM;
928 			if (!valid_markcond(bc, len, &min_len))
929 				return -EINVAL;
930 			break;
931 #ifdef CONFIG_SOCK_CGROUP_DATA
932 		case INET_DIAG_BC_CGROUP_COND:
933 			if (!valid_cgroupcond(bc, len, &min_len))
934 				return -EINVAL;
935 			break;
936 #endif
937 		case INET_DIAG_BC_AUTO:
938 		case INET_DIAG_BC_JMP:
939 		case INET_DIAG_BC_NOP:
940 			break;
941 		default:
942 			return -EINVAL;
943 		}
944 
945 		if (op->code != INET_DIAG_BC_NOP) {
946 			if (op->no < min_len || op->no > len + 4 || op->no & 3)
947 				return -EINVAL;
948 			if (op->no < len &&
949 			    !valid_cc(bytecode, bytecode_len, len - op->no))
950 				return -EINVAL;
951 		}
952 
953 		if (op->yes < min_len || op->yes > len + 4 || op->yes & 3)
954 			return -EINVAL;
955 		bc  += op->yes;
956 		len -= op->yes;
957 	}
958 	return len == 0 ? 0 : -EINVAL;
959 }
960 
961 static void twsk_build_assert(void)
962 {
963 	BUILD_BUG_ON(offsetof(struct inet_timewait_sock, tw_family) !=
964 		     offsetof(struct sock, sk_family));
965 
966 	BUILD_BUG_ON(offsetof(struct inet_timewait_sock, tw_num) !=
967 		     offsetof(struct inet_sock, inet_num));
968 
969 	BUILD_BUG_ON(offsetof(struct inet_timewait_sock, tw_dport) !=
970 		     offsetof(struct inet_sock, inet_dport));
971 
972 	BUILD_BUG_ON(offsetof(struct inet_timewait_sock, tw_rcv_saddr) !=
973 		     offsetof(struct inet_sock, inet_rcv_saddr));
974 
975 	BUILD_BUG_ON(offsetof(struct inet_timewait_sock, tw_daddr) !=
976 		     offsetof(struct inet_sock, inet_daddr));
977 
978 #if IS_ENABLED(CONFIG_IPV6)
979 	BUILD_BUG_ON(offsetof(struct inet_timewait_sock, tw_v6_rcv_saddr) !=
980 		     offsetof(struct sock, sk_v6_rcv_saddr));
981 
982 	BUILD_BUG_ON(offsetof(struct inet_timewait_sock, tw_v6_daddr) !=
983 		     offsetof(struct sock, sk_v6_daddr));
984 #endif
985 }
986 
987 void inet_diag_dump_icsk(struct inet_hashinfo *hashinfo, struct sk_buff *skb,
988 			 struct netlink_callback *cb,
989 			 const struct inet_diag_req_v2 *r)
990 {
991 	bool net_admin = netlink_net_capable(cb->skb, CAP_NET_ADMIN);
992 	struct inet_diag_dump_data *cb_data = cb->data;
993 	struct net *net = sock_net(skb->sk);
994 	u32 idiag_states = r->idiag_states;
995 	int i, num, s_i, s_num;
996 	struct nlattr *bc;
997 	struct sock *sk;
998 
999 	bc = cb_data->inet_diag_nla_bc;
1000 	if (idiag_states & TCPF_SYN_RECV)
1001 		idiag_states |= TCPF_NEW_SYN_RECV;
1002 	s_i = cb->args[1];
1003 	s_num = num = cb->args[2];
1004 
1005 	if (cb->args[0] == 0) {
1006 		if (!(idiag_states & TCPF_LISTEN) || r->id.idiag_dport)
1007 			goto skip_listen_ht;
1008 
1009 		for (i = s_i; i < INET_LHTABLE_SIZE; i++) {
1010 			struct inet_listen_hashbucket *ilb;
1011 			struct hlist_nulls_node *node;
1012 
1013 			num = 0;
1014 			ilb = &hashinfo->listening_hash[i];
1015 			spin_lock(&ilb->lock);
1016 			sk_nulls_for_each(sk, node, &ilb->nulls_head) {
1017 				struct inet_sock *inet = inet_sk(sk);
1018 
1019 				if (!net_eq(sock_net(sk), net))
1020 					continue;
1021 
1022 				if (num < s_num) {
1023 					num++;
1024 					continue;
1025 				}
1026 
1027 				if (r->sdiag_family != AF_UNSPEC &&
1028 				    sk->sk_family != r->sdiag_family)
1029 					goto next_listen;
1030 
1031 				if (r->id.idiag_sport != inet->inet_sport &&
1032 				    r->id.idiag_sport)
1033 					goto next_listen;
1034 
1035 				if (!inet_diag_bc_sk(bc, sk))
1036 					goto next_listen;
1037 
1038 				if (inet_sk_diag_fill(sk, inet_csk(sk), skb,
1039 						      cb, r, NLM_F_MULTI,
1040 						      net_admin) < 0) {
1041 					spin_unlock(&ilb->lock);
1042 					goto done;
1043 				}
1044 
1045 next_listen:
1046 				++num;
1047 			}
1048 			spin_unlock(&ilb->lock);
1049 
1050 			s_num = 0;
1051 		}
1052 skip_listen_ht:
1053 		cb->args[0] = 1;
1054 		s_i = num = s_num = 0;
1055 	}
1056 
1057 	if (!(idiag_states & ~TCPF_LISTEN))
1058 		goto out;
1059 
1060 #define SKARR_SZ 16
1061 	for (i = s_i; i <= hashinfo->ehash_mask; i++) {
1062 		struct inet_ehash_bucket *head = &hashinfo->ehash[i];
1063 		spinlock_t *lock = inet_ehash_lockp(hashinfo, i);
1064 		struct hlist_nulls_node *node;
1065 		struct sock *sk_arr[SKARR_SZ];
1066 		int num_arr[SKARR_SZ];
1067 		int idx, accum, res;
1068 
1069 		if (hlist_nulls_empty(&head->chain))
1070 			continue;
1071 
1072 		if (i > s_i)
1073 			s_num = 0;
1074 
1075 next_chunk:
1076 		num = 0;
1077 		accum = 0;
1078 		spin_lock_bh(lock);
1079 		sk_nulls_for_each(sk, node, &head->chain) {
1080 			int state;
1081 
1082 			if (!net_eq(sock_net(sk), net))
1083 				continue;
1084 			if (num < s_num)
1085 				goto next_normal;
1086 			state = (sk->sk_state == TCP_TIME_WAIT) ?
1087 				inet_twsk(sk)->tw_substate : sk->sk_state;
1088 			if (!(idiag_states & (1 << state)))
1089 				goto next_normal;
1090 			if (r->sdiag_family != AF_UNSPEC &&
1091 			    sk->sk_family != r->sdiag_family)
1092 				goto next_normal;
1093 			if (r->id.idiag_sport != htons(sk->sk_num) &&
1094 			    r->id.idiag_sport)
1095 				goto next_normal;
1096 			if (r->id.idiag_dport != sk->sk_dport &&
1097 			    r->id.idiag_dport)
1098 				goto next_normal;
1099 			twsk_build_assert();
1100 
1101 			if (!inet_diag_bc_sk(bc, sk))
1102 				goto next_normal;
1103 
1104 			if (!refcount_inc_not_zero(&sk->sk_refcnt))
1105 				goto next_normal;
1106 
1107 			num_arr[accum] = num;
1108 			sk_arr[accum] = sk;
1109 			if (++accum == SKARR_SZ)
1110 				break;
1111 next_normal:
1112 			++num;
1113 		}
1114 		spin_unlock_bh(lock);
1115 		res = 0;
1116 		for (idx = 0; idx < accum; idx++) {
1117 			if (res >= 0) {
1118 				res = sk_diag_fill(sk_arr[idx], skb, cb, r,
1119 						   NLM_F_MULTI, net_admin);
1120 				if (res < 0)
1121 					num = num_arr[idx];
1122 			}
1123 			sock_gen_put(sk_arr[idx]);
1124 		}
1125 		if (res < 0)
1126 			break;
1127 		cond_resched();
1128 		if (accum == SKARR_SZ) {
1129 			s_num = num + 1;
1130 			goto next_chunk;
1131 		}
1132 	}
1133 
1134 done:
1135 	cb->args[1] = i;
1136 	cb->args[2] = num;
1137 out:
1138 	;
1139 }
1140 EXPORT_SYMBOL_GPL(inet_diag_dump_icsk);
1141 
1142 static int __inet_diag_dump(struct sk_buff *skb, struct netlink_callback *cb,
1143 			    const struct inet_diag_req_v2 *r)
1144 {
1145 	struct inet_diag_dump_data *cb_data = cb->data;
1146 	const struct inet_diag_handler *handler;
1147 	u32 prev_min_dump_alloc;
1148 	int protocol, err = 0;
1149 
1150 	protocol = inet_diag_get_protocol(r, cb_data);
1151 
1152 again:
1153 	prev_min_dump_alloc = cb->min_dump_alloc;
1154 	handler = inet_diag_lock_handler(protocol);
1155 	if (!IS_ERR(handler))
1156 		handler->dump(skb, cb, r);
1157 	else
1158 		err = PTR_ERR(handler);
1159 	inet_diag_unlock_handler(handler);
1160 
1161 	/* The skb is not large enough to fit one sk info and
1162 	 * inet_sk_diag_fill() has requested for a larger skb.
1163 	 */
1164 	if (!skb->len && cb->min_dump_alloc > prev_min_dump_alloc) {
1165 		err = pskb_expand_head(skb, 0, cb->min_dump_alloc, GFP_KERNEL);
1166 		if (!err)
1167 			goto again;
1168 	}
1169 
1170 	return err ? : skb->len;
1171 }
1172 
1173 static int inet_diag_dump(struct sk_buff *skb, struct netlink_callback *cb)
1174 {
1175 	return __inet_diag_dump(skb, cb, nlmsg_data(cb->nlh));
1176 }
1177 
1178 static int __inet_diag_dump_start(struct netlink_callback *cb, int hdrlen)
1179 {
1180 	const struct nlmsghdr *nlh = cb->nlh;
1181 	struct inet_diag_dump_data *cb_data;
1182 	struct sk_buff *skb = cb->skb;
1183 	struct nlattr *nla;
1184 	int err;
1185 
1186 	cb_data = kzalloc(sizeof(*cb_data), GFP_KERNEL);
1187 	if (!cb_data)
1188 		return -ENOMEM;
1189 
1190 	err = inet_diag_parse_attrs(nlh, hdrlen, cb_data->req_nlas);
1191 	if (err) {
1192 		kfree(cb_data);
1193 		return err;
1194 	}
1195 	nla = cb_data->inet_diag_nla_bc;
1196 	if (nla) {
1197 		err = inet_diag_bc_audit(nla, skb);
1198 		if (err) {
1199 			kfree(cb_data);
1200 			return err;
1201 		}
1202 	}
1203 
1204 	nla = cb_data->inet_diag_nla_bpf_stgs;
1205 	if (nla) {
1206 		struct bpf_sk_storage_diag *bpf_stg_diag;
1207 
1208 		bpf_stg_diag = bpf_sk_storage_diag_alloc(nla);
1209 		if (IS_ERR(bpf_stg_diag)) {
1210 			kfree(cb_data);
1211 			return PTR_ERR(bpf_stg_diag);
1212 		}
1213 		cb_data->bpf_stg_diag = bpf_stg_diag;
1214 	}
1215 
1216 	cb->data = cb_data;
1217 	return 0;
1218 }
1219 
1220 static int inet_diag_dump_start(struct netlink_callback *cb)
1221 {
1222 	return __inet_diag_dump_start(cb, sizeof(struct inet_diag_req_v2));
1223 }
1224 
1225 static int inet_diag_dump_start_compat(struct netlink_callback *cb)
1226 {
1227 	return __inet_diag_dump_start(cb, sizeof(struct inet_diag_req));
1228 }
1229 
1230 static int inet_diag_dump_done(struct netlink_callback *cb)
1231 {
1232 	struct inet_diag_dump_data *cb_data = cb->data;
1233 
1234 	bpf_sk_storage_diag_free(cb_data->bpf_stg_diag);
1235 	kfree(cb->data);
1236 
1237 	return 0;
1238 }
1239 
1240 static int inet_diag_type2proto(int type)
1241 {
1242 	switch (type) {
1243 	case TCPDIAG_GETSOCK:
1244 		return IPPROTO_TCP;
1245 	case DCCPDIAG_GETSOCK:
1246 		return IPPROTO_DCCP;
1247 	default:
1248 		return 0;
1249 	}
1250 }
1251 
1252 static int inet_diag_dump_compat(struct sk_buff *skb,
1253 				 struct netlink_callback *cb)
1254 {
1255 	struct inet_diag_req *rc = nlmsg_data(cb->nlh);
1256 	struct inet_diag_req_v2 req;
1257 
1258 	req.sdiag_family = AF_UNSPEC; /* compatibility */
1259 	req.sdiag_protocol = inet_diag_type2proto(cb->nlh->nlmsg_type);
1260 	req.idiag_ext = rc->idiag_ext;
1261 	req.idiag_states = rc->idiag_states;
1262 	req.id = rc->id;
1263 
1264 	return __inet_diag_dump(skb, cb, &req);
1265 }
1266 
1267 static int inet_diag_get_exact_compat(struct sk_buff *in_skb,
1268 				      const struct nlmsghdr *nlh)
1269 {
1270 	struct inet_diag_req *rc = nlmsg_data(nlh);
1271 	struct inet_diag_req_v2 req;
1272 
1273 	req.sdiag_family = rc->idiag_family;
1274 	req.sdiag_protocol = inet_diag_type2proto(nlh->nlmsg_type);
1275 	req.idiag_ext = rc->idiag_ext;
1276 	req.idiag_states = rc->idiag_states;
1277 	req.id = rc->id;
1278 
1279 	return inet_diag_cmd_exact(SOCK_DIAG_BY_FAMILY, in_skb, nlh,
1280 				   sizeof(struct inet_diag_req), &req);
1281 }
1282 
1283 static int inet_diag_rcv_msg_compat(struct sk_buff *skb, struct nlmsghdr *nlh)
1284 {
1285 	int hdrlen = sizeof(struct inet_diag_req);
1286 	struct net *net = sock_net(skb->sk);
1287 
1288 	if (nlh->nlmsg_type >= INET_DIAG_GETSOCK_MAX ||
1289 	    nlmsg_len(nlh) < hdrlen)
1290 		return -EINVAL;
1291 
1292 	if (nlh->nlmsg_flags & NLM_F_DUMP) {
1293 		struct netlink_dump_control c = {
1294 			.start = inet_diag_dump_start_compat,
1295 			.done = inet_diag_dump_done,
1296 			.dump = inet_diag_dump_compat,
1297 		};
1298 		return netlink_dump_start(net->diag_nlsk, skb, nlh, &c);
1299 	}
1300 
1301 	return inet_diag_get_exact_compat(skb, nlh);
1302 }
1303 
1304 static int inet_diag_handler_cmd(struct sk_buff *skb, struct nlmsghdr *h)
1305 {
1306 	int hdrlen = sizeof(struct inet_diag_req_v2);
1307 	struct net *net = sock_net(skb->sk);
1308 
1309 	if (nlmsg_len(h) < hdrlen)
1310 		return -EINVAL;
1311 
1312 	if (h->nlmsg_type == SOCK_DIAG_BY_FAMILY &&
1313 	    h->nlmsg_flags & NLM_F_DUMP) {
1314 		struct netlink_dump_control c = {
1315 			.start = inet_diag_dump_start,
1316 			.done = inet_diag_dump_done,
1317 			.dump = inet_diag_dump,
1318 		};
1319 		return netlink_dump_start(net->diag_nlsk, skb, h, &c);
1320 	}
1321 
1322 	return inet_diag_cmd_exact(h->nlmsg_type, skb, h, hdrlen,
1323 				   nlmsg_data(h));
1324 }
1325 
1326 static
1327 int inet_diag_handler_get_info(struct sk_buff *skb, struct sock *sk)
1328 {
1329 	const struct inet_diag_handler *handler;
1330 	struct nlmsghdr *nlh;
1331 	struct nlattr *attr;
1332 	struct inet_diag_msg *r;
1333 	void *info = NULL;
1334 	int err = 0;
1335 
1336 	nlh = nlmsg_put(skb, 0, 0, SOCK_DIAG_BY_FAMILY, sizeof(*r), 0);
1337 	if (!nlh)
1338 		return -ENOMEM;
1339 
1340 	r = nlmsg_data(nlh);
1341 	memset(r, 0, sizeof(*r));
1342 	inet_diag_msg_common_fill(r, sk);
1343 	if (sk->sk_type == SOCK_DGRAM || sk->sk_type == SOCK_STREAM)
1344 		r->id.idiag_sport = inet_sk(sk)->inet_sport;
1345 	r->idiag_state = sk->sk_state;
1346 
1347 	if ((err = nla_put_u8(skb, INET_DIAG_PROTOCOL, sk->sk_protocol))) {
1348 		nlmsg_cancel(skb, nlh);
1349 		return err;
1350 	}
1351 
1352 	handler = inet_diag_lock_handler(sk->sk_protocol);
1353 	if (IS_ERR(handler)) {
1354 		inet_diag_unlock_handler(handler);
1355 		nlmsg_cancel(skb, nlh);
1356 		return PTR_ERR(handler);
1357 	}
1358 
1359 	attr = handler->idiag_info_size
1360 		? nla_reserve_64bit(skb, INET_DIAG_INFO,
1361 				    handler->idiag_info_size,
1362 				    INET_DIAG_PAD)
1363 		: NULL;
1364 	if (attr)
1365 		info = nla_data(attr);
1366 
1367 	handler->idiag_get_info(sk, r, info);
1368 	inet_diag_unlock_handler(handler);
1369 
1370 	nlmsg_end(skb, nlh);
1371 	return 0;
1372 }
1373 
1374 static const struct sock_diag_handler inet_diag_handler = {
1375 	.family = AF_INET,
1376 	.dump = inet_diag_handler_cmd,
1377 	.get_info = inet_diag_handler_get_info,
1378 	.destroy = inet_diag_handler_cmd,
1379 };
1380 
1381 static const struct sock_diag_handler inet6_diag_handler = {
1382 	.family = AF_INET6,
1383 	.dump = inet_diag_handler_cmd,
1384 	.get_info = inet_diag_handler_get_info,
1385 	.destroy = inet_diag_handler_cmd,
1386 };
1387 
1388 int inet_diag_register(const struct inet_diag_handler *h)
1389 {
1390 	const __u16 type = h->idiag_type;
1391 	int err = -EINVAL;
1392 
1393 	if (type >= IPPROTO_MAX)
1394 		goto out;
1395 
1396 	mutex_lock(&inet_diag_table_mutex);
1397 	err = -EEXIST;
1398 	if (!inet_diag_table[type]) {
1399 		inet_diag_table[type] = h;
1400 		err = 0;
1401 	}
1402 	mutex_unlock(&inet_diag_table_mutex);
1403 out:
1404 	return err;
1405 }
1406 EXPORT_SYMBOL_GPL(inet_diag_register);
1407 
1408 void inet_diag_unregister(const struct inet_diag_handler *h)
1409 {
1410 	const __u16 type = h->idiag_type;
1411 
1412 	if (type >= IPPROTO_MAX)
1413 		return;
1414 
1415 	mutex_lock(&inet_diag_table_mutex);
1416 	inet_diag_table[type] = NULL;
1417 	mutex_unlock(&inet_diag_table_mutex);
1418 }
1419 EXPORT_SYMBOL_GPL(inet_diag_unregister);
1420 
1421 static int __init inet_diag_init(void)
1422 {
1423 	const int inet_diag_table_size = (IPPROTO_MAX *
1424 					  sizeof(struct inet_diag_handler *));
1425 	int err = -ENOMEM;
1426 
1427 	inet_diag_table = kzalloc(inet_diag_table_size, GFP_KERNEL);
1428 	if (!inet_diag_table)
1429 		goto out;
1430 
1431 	err = sock_diag_register(&inet_diag_handler);
1432 	if (err)
1433 		goto out_free_nl;
1434 
1435 	err = sock_diag_register(&inet6_diag_handler);
1436 	if (err)
1437 		goto out_free_inet;
1438 
1439 	sock_diag_register_inet_compat(inet_diag_rcv_msg_compat);
1440 out:
1441 	return err;
1442 
1443 out_free_inet:
1444 	sock_diag_unregister(&inet_diag_handler);
1445 out_free_nl:
1446 	kfree(inet_diag_table);
1447 	goto out;
1448 }
1449 
1450 static void __exit inet_diag_exit(void)
1451 {
1452 	sock_diag_unregister(&inet6_diag_handler);
1453 	sock_diag_unregister(&inet_diag_handler);
1454 	sock_diag_unregister_inet_compat(inet_diag_rcv_msg_compat);
1455 	kfree(inet_diag_table);
1456 }
1457 
1458 module_init(inet_diag_init);
1459 module_exit(inet_diag_exit);
1460 MODULE_LICENSE("GPL");
1461 MODULE_ALIAS_NET_PF_PROTO_TYPE(PF_NETLINK, NETLINK_SOCK_DIAG, 2 /* AF_INET */);
1462 MODULE_ALIAS_NET_PF_PROTO_TYPE(PF_NETLINK, NETLINK_SOCK_DIAG, 10 /* AF_INET6 */);
1463