xref: /openbmc/linux/net/core/sock_map.c (revision abed054f)
1 // SPDX-License-Identifier: GPL-2.0
2 /* Copyright (c) 2017 - 2018 Covalent IO, Inc. http://covalent.io */
3 
4 #include <linux/bpf.h>
5 #include <linux/btf_ids.h>
6 #include <linux/filter.h>
7 #include <linux/errno.h>
8 #include <linux/file.h>
9 #include <linux/net.h>
10 #include <linux/workqueue.h>
11 #include <linux/skmsg.h>
12 #include <linux/list.h>
13 #include <linux/jhash.h>
14 #include <linux/sock_diag.h>
15 #include <net/udp.h>
16 
17 struct bpf_stab {
18 	struct bpf_map map;
19 	struct sock **sks;
20 	struct sk_psock_progs progs;
21 	raw_spinlock_t lock;
22 };
23 
24 #define SOCK_CREATE_FLAG_MASK				\
25 	(BPF_F_NUMA_NODE | BPF_F_RDONLY | BPF_F_WRONLY)
26 
27 static int sock_map_prog_update(struct bpf_map *map, struct bpf_prog *prog,
28 				struct bpf_prog *old, u32 which);
29 static struct sk_psock_progs *sock_map_progs(struct bpf_map *map);
30 
31 static struct bpf_map *sock_map_alloc(union bpf_attr *attr)
32 {
33 	struct bpf_stab *stab;
34 
35 	if (!capable(CAP_NET_ADMIN))
36 		return ERR_PTR(-EPERM);
37 	if (attr->max_entries == 0 ||
38 	    attr->key_size    != 4 ||
39 	    (attr->value_size != sizeof(u32) &&
40 	     attr->value_size != sizeof(u64)) ||
41 	    attr->map_flags & ~SOCK_CREATE_FLAG_MASK)
42 		return ERR_PTR(-EINVAL);
43 
44 	stab = kzalloc(sizeof(*stab), GFP_USER | __GFP_ACCOUNT);
45 	if (!stab)
46 		return ERR_PTR(-ENOMEM);
47 
48 	bpf_map_init_from_attr(&stab->map, attr);
49 	raw_spin_lock_init(&stab->lock);
50 
51 	stab->sks = bpf_map_area_alloc((u64) stab->map.max_entries *
52 				       sizeof(struct sock *),
53 				       stab->map.numa_node);
54 	if (!stab->sks) {
55 		kfree(stab);
56 		return ERR_PTR(-ENOMEM);
57 	}
58 
59 	return &stab->map;
60 }
61 
62 int sock_map_get_from_fd(const union bpf_attr *attr, struct bpf_prog *prog)
63 {
64 	u32 ufd = attr->target_fd;
65 	struct bpf_map *map;
66 	struct fd f;
67 	int ret;
68 
69 	if (attr->attach_flags || attr->replace_bpf_fd)
70 		return -EINVAL;
71 
72 	f = fdget(ufd);
73 	map = __bpf_map_get(f);
74 	if (IS_ERR(map))
75 		return PTR_ERR(map);
76 	ret = sock_map_prog_update(map, prog, NULL, attr->attach_type);
77 	fdput(f);
78 	return ret;
79 }
80 
81 int sock_map_prog_detach(const union bpf_attr *attr, enum bpf_prog_type ptype)
82 {
83 	u32 ufd = attr->target_fd;
84 	struct bpf_prog *prog;
85 	struct bpf_map *map;
86 	struct fd f;
87 	int ret;
88 
89 	if (attr->attach_flags || attr->replace_bpf_fd)
90 		return -EINVAL;
91 
92 	f = fdget(ufd);
93 	map = __bpf_map_get(f);
94 	if (IS_ERR(map))
95 		return PTR_ERR(map);
96 
97 	prog = bpf_prog_get(attr->attach_bpf_fd);
98 	if (IS_ERR(prog)) {
99 		ret = PTR_ERR(prog);
100 		goto put_map;
101 	}
102 
103 	if (prog->type != ptype) {
104 		ret = -EINVAL;
105 		goto put_prog;
106 	}
107 
108 	ret = sock_map_prog_update(map, NULL, prog, attr->attach_type);
109 put_prog:
110 	bpf_prog_put(prog);
111 put_map:
112 	fdput(f);
113 	return ret;
114 }
115 
116 static void sock_map_sk_acquire(struct sock *sk)
117 	__acquires(&sk->sk_lock.slock)
118 {
119 	lock_sock(sk);
120 	preempt_disable();
121 	rcu_read_lock();
122 }
123 
124 static void sock_map_sk_release(struct sock *sk)
125 	__releases(&sk->sk_lock.slock)
126 {
127 	rcu_read_unlock();
128 	preempt_enable();
129 	release_sock(sk);
130 }
131 
132 static void sock_map_add_link(struct sk_psock *psock,
133 			      struct sk_psock_link *link,
134 			      struct bpf_map *map, void *link_raw)
135 {
136 	link->link_raw = link_raw;
137 	link->map = map;
138 	spin_lock_bh(&psock->link_lock);
139 	list_add_tail(&link->list, &psock->link);
140 	spin_unlock_bh(&psock->link_lock);
141 }
142 
143 static void sock_map_del_link(struct sock *sk,
144 			      struct sk_psock *psock, void *link_raw)
145 {
146 	bool strp_stop = false, verdict_stop = false;
147 	struct sk_psock_link *link, *tmp;
148 
149 	spin_lock_bh(&psock->link_lock);
150 	list_for_each_entry_safe(link, tmp, &psock->link, list) {
151 		if (link->link_raw == link_raw) {
152 			struct bpf_map *map = link->map;
153 			struct bpf_stab *stab = container_of(map, struct bpf_stab,
154 							     map);
155 			if (psock->saved_data_ready && stab->progs.stream_parser)
156 				strp_stop = true;
157 			if (psock->saved_data_ready && stab->progs.stream_verdict)
158 				verdict_stop = true;
159 			if (psock->saved_data_ready && stab->progs.skb_verdict)
160 				verdict_stop = true;
161 			list_del(&link->list);
162 			sk_psock_free_link(link);
163 		}
164 	}
165 	spin_unlock_bh(&psock->link_lock);
166 	if (strp_stop || verdict_stop) {
167 		write_lock_bh(&sk->sk_callback_lock);
168 		if (strp_stop)
169 			sk_psock_stop_strp(sk, psock);
170 		else
171 			sk_psock_stop_verdict(sk, psock);
172 		write_unlock_bh(&sk->sk_callback_lock);
173 	}
174 }
175 
176 static void sock_map_unref(struct sock *sk, void *link_raw)
177 {
178 	struct sk_psock *psock = sk_psock(sk);
179 
180 	if (likely(psock)) {
181 		sock_map_del_link(sk, psock, link_raw);
182 		sk_psock_put(sk, psock);
183 	}
184 }
185 
186 static int sock_map_init_proto(struct sock *sk, struct sk_psock *psock)
187 {
188 	if (!sk->sk_prot->psock_update_sk_prot)
189 		return -EINVAL;
190 	psock->psock_update_sk_prot = sk->sk_prot->psock_update_sk_prot;
191 	return sk->sk_prot->psock_update_sk_prot(sk, psock, false);
192 }
193 
194 static struct sk_psock *sock_map_psock_get_checked(struct sock *sk)
195 {
196 	struct sk_psock *psock;
197 
198 	rcu_read_lock();
199 	psock = sk_psock(sk);
200 	if (psock) {
201 		if (sk->sk_prot->close != sock_map_close) {
202 			psock = ERR_PTR(-EBUSY);
203 			goto out;
204 		}
205 
206 		if (!refcount_inc_not_zero(&psock->refcnt))
207 			psock = ERR_PTR(-EBUSY);
208 	}
209 out:
210 	rcu_read_unlock();
211 	return psock;
212 }
213 
214 static int sock_map_link(struct bpf_map *map, struct sock *sk)
215 {
216 	struct sk_psock_progs *progs = sock_map_progs(map);
217 	struct bpf_prog *stream_verdict = NULL;
218 	struct bpf_prog *stream_parser = NULL;
219 	struct bpf_prog *skb_verdict = NULL;
220 	struct bpf_prog *msg_parser = NULL;
221 	struct sk_psock *psock;
222 	int ret;
223 
224 	stream_verdict = READ_ONCE(progs->stream_verdict);
225 	if (stream_verdict) {
226 		stream_verdict = bpf_prog_inc_not_zero(stream_verdict);
227 		if (IS_ERR(stream_verdict))
228 			return PTR_ERR(stream_verdict);
229 	}
230 
231 	stream_parser = READ_ONCE(progs->stream_parser);
232 	if (stream_parser) {
233 		stream_parser = bpf_prog_inc_not_zero(stream_parser);
234 		if (IS_ERR(stream_parser)) {
235 			ret = PTR_ERR(stream_parser);
236 			goto out_put_stream_verdict;
237 		}
238 	}
239 
240 	msg_parser = READ_ONCE(progs->msg_parser);
241 	if (msg_parser) {
242 		msg_parser = bpf_prog_inc_not_zero(msg_parser);
243 		if (IS_ERR(msg_parser)) {
244 			ret = PTR_ERR(msg_parser);
245 			goto out_put_stream_parser;
246 		}
247 	}
248 
249 	skb_verdict = READ_ONCE(progs->skb_verdict);
250 	if (skb_verdict) {
251 		skb_verdict = bpf_prog_inc_not_zero(skb_verdict);
252 		if (IS_ERR(skb_verdict)) {
253 			ret = PTR_ERR(skb_verdict);
254 			goto out_put_msg_parser;
255 		}
256 	}
257 
258 	psock = sock_map_psock_get_checked(sk);
259 	if (IS_ERR(psock)) {
260 		ret = PTR_ERR(psock);
261 		goto out_progs;
262 	}
263 
264 	if (psock) {
265 		if ((msg_parser && READ_ONCE(psock->progs.msg_parser)) ||
266 		    (stream_parser  && READ_ONCE(psock->progs.stream_parser)) ||
267 		    (skb_verdict && READ_ONCE(psock->progs.skb_verdict)) ||
268 		    (skb_verdict && READ_ONCE(psock->progs.stream_verdict)) ||
269 		    (stream_verdict && READ_ONCE(psock->progs.skb_verdict)) ||
270 		    (stream_verdict && READ_ONCE(psock->progs.stream_verdict))) {
271 			sk_psock_put(sk, psock);
272 			ret = -EBUSY;
273 			goto out_progs;
274 		}
275 	} else {
276 		psock = sk_psock_init(sk, map->numa_node);
277 		if (IS_ERR(psock)) {
278 			ret = PTR_ERR(psock);
279 			goto out_progs;
280 		}
281 	}
282 
283 	if (msg_parser)
284 		psock_set_prog(&psock->progs.msg_parser, msg_parser);
285 
286 	ret = sock_map_init_proto(sk, psock);
287 	if (ret < 0)
288 		goto out_drop;
289 
290 	write_lock_bh(&sk->sk_callback_lock);
291 	if (stream_parser && stream_verdict && !psock->saved_data_ready) {
292 		ret = sk_psock_init_strp(sk, psock);
293 		if (ret)
294 			goto out_unlock_drop;
295 		psock_set_prog(&psock->progs.stream_verdict, stream_verdict);
296 		psock_set_prog(&psock->progs.stream_parser, stream_parser);
297 		sk_psock_start_strp(sk, psock);
298 	} else if (!stream_parser && stream_verdict && !psock->saved_data_ready) {
299 		psock_set_prog(&psock->progs.stream_verdict, stream_verdict);
300 		sk_psock_start_verdict(sk,psock);
301 	} else if (!stream_verdict && skb_verdict && !psock->saved_data_ready) {
302 		psock_set_prog(&psock->progs.skb_verdict, skb_verdict);
303 		sk_psock_start_verdict(sk, psock);
304 	}
305 	write_unlock_bh(&sk->sk_callback_lock);
306 	return 0;
307 out_unlock_drop:
308 	write_unlock_bh(&sk->sk_callback_lock);
309 out_drop:
310 	sk_psock_put(sk, psock);
311 out_progs:
312 	if (skb_verdict)
313 		bpf_prog_put(skb_verdict);
314 out_put_msg_parser:
315 	if (msg_parser)
316 		bpf_prog_put(msg_parser);
317 out_put_stream_parser:
318 	if (stream_parser)
319 		bpf_prog_put(stream_parser);
320 out_put_stream_verdict:
321 	if (stream_verdict)
322 		bpf_prog_put(stream_verdict);
323 	return ret;
324 }
325 
326 static void sock_map_free(struct bpf_map *map)
327 {
328 	struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
329 	int i;
330 
331 	/* After the sync no updates or deletes will be in-flight so it
332 	 * is safe to walk map and remove entries without risking a race
333 	 * in EEXIST update case.
334 	 */
335 	synchronize_rcu();
336 	for (i = 0; i < stab->map.max_entries; i++) {
337 		struct sock **psk = &stab->sks[i];
338 		struct sock *sk;
339 
340 		sk = xchg(psk, NULL);
341 		if (sk) {
342 			lock_sock(sk);
343 			rcu_read_lock();
344 			sock_map_unref(sk, psk);
345 			rcu_read_unlock();
346 			release_sock(sk);
347 		}
348 	}
349 
350 	/* wait for psock readers accessing its map link */
351 	synchronize_rcu();
352 
353 	bpf_map_area_free(stab->sks);
354 	kfree(stab);
355 }
356 
357 static void sock_map_release_progs(struct bpf_map *map)
358 {
359 	psock_progs_drop(&container_of(map, struct bpf_stab, map)->progs);
360 }
361 
362 static struct sock *__sock_map_lookup_elem(struct bpf_map *map, u32 key)
363 {
364 	struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
365 
366 	WARN_ON_ONCE(!rcu_read_lock_held());
367 
368 	if (unlikely(key >= map->max_entries))
369 		return NULL;
370 	return READ_ONCE(stab->sks[key]);
371 }
372 
373 static void *sock_map_lookup(struct bpf_map *map, void *key)
374 {
375 	struct sock *sk;
376 
377 	sk = __sock_map_lookup_elem(map, *(u32 *)key);
378 	if (!sk)
379 		return NULL;
380 	if (sk_is_refcounted(sk) && !refcount_inc_not_zero(&sk->sk_refcnt))
381 		return NULL;
382 	return sk;
383 }
384 
385 static void *sock_map_lookup_sys(struct bpf_map *map, void *key)
386 {
387 	struct sock *sk;
388 
389 	if (map->value_size != sizeof(u64))
390 		return ERR_PTR(-ENOSPC);
391 
392 	sk = __sock_map_lookup_elem(map, *(u32 *)key);
393 	if (!sk)
394 		return ERR_PTR(-ENOENT);
395 
396 	__sock_gen_cookie(sk);
397 	return &sk->sk_cookie;
398 }
399 
400 static int __sock_map_delete(struct bpf_stab *stab, struct sock *sk_test,
401 			     struct sock **psk)
402 {
403 	struct sock *sk;
404 	int err = 0;
405 
406 	raw_spin_lock_bh(&stab->lock);
407 	sk = *psk;
408 	if (!sk_test || sk_test == sk)
409 		sk = xchg(psk, NULL);
410 
411 	if (likely(sk))
412 		sock_map_unref(sk, psk);
413 	else
414 		err = -EINVAL;
415 
416 	raw_spin_unlock_bh(&stab->lock);
417 	return err;
418 }
419 
420 static void sock_map_delete_from_link(struct bpf_map *map, struct sock *sk,
421 				      void *link_raw)
422 {
423 	struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
424 
425 	__sock_map_delete(stab, sk, link_raw);
426 }
427 
428 static int sock_map_delete_elem(struct bpf_map *map, void *key)
429 {
430 	struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
431 	u32 i = *(u32 *)key;
432 	struct sock **psk;
433 
434 	if (unlikely(i >= map->max_entries))
435 		return -EINVAL;
436 
437 	psk = &stab->sks[i];
438 	return __sock_map_delete(stab, NULL, psk);
439 }
440 
441 static int sock_map_get_next_key(struct bpf_map *map, void *key, void *next)
442 {
443 	struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
444 	u32 i = key ? *(u32 *)key : U32_MAX;
445 	u32 *key_next = next;
446 
447 	if (i == stab->map.max_entries - 1)
448 		return -ENOENT;
449 	if (i >= stab->map.max_entries)
450 		*key_next = 0;
451 	else
452 		*key_next = i + 1;
453 	return 0;
454 }
455 
456 static int sock_map_update_common(struct bpf_map *map, u32 idx,
457 				  struct sock *sk, u64 flags)
458 {
459 	struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
460 	struct sk_psock_link *link;
461 	struct sk_psock *psock;
462 	struct sock *osk;
463 	int ret;
464 
465 	WARN_ON_ONCE(!rcu_read_lock_held());
466 	if (unlikely(flags > BPF_EXIST))
467 		return -EINVAL;
468 	if (unlikely(idx >= map->max_entries))
469 		return -E2BIG;
470 
471 	link = sk_psock_init_link();
472 	if (!link)
473 		return -ENOMEM;
474 
475 	ret = sock_map_link(map, sk);
476 	if (ret < 0)
477 		goto out_free;
478 
479 	psock = sk_psock(sk);
480 	WARN_ON_ONCE(!psock);
481 
482 	raw_spin_lock_bh(&stab->lock);
483 	osk = stab->sks[idx];
484 	if (osk && flags == BPF_NOEXIST) {
485 		ret = -EEXIST;
486 		goto out_unlock;
487 	} else if (!osk && flags == BPF_EXIST) {
488 		ret = -ENOENT;
489 		goto out_unlock;
490 	}
491 
492 	sock_map_add_link(psock, link, map, &stab->sks[idx]);
493 	stab->sks[idx] = sk;
494 	if (osk)
495 		sock_map_unref(osk, &stab->sks[idx]);
496 	raw_spin_unlock_bh(&stab->lock);
497 	return 0;
498 out_unlock:
499 	raw_spin_unlock_bh(&stab->lock);
500 	if (psock)
501 		sk_psock_put(sk, psock);
502 out_free:
503 	sk_psock_free_link(link);
504 	return ret;
505 }
506 
507 static bool sock_map_op_okay(const struct bpf_sock_ops_kern *ops)
508 {
509 	return ops->op == BPF_SOCK_OPS_PASSIVE_ESTABLISHED_CB ||
510 	       ops->op == BPF_SOCK_OPS_ACTIVE_ESTABLISHED_CB ||
511 	       ops->op == BPF_SOCK_OPS_TCP_LISTEN_CB;
512 }
513 
514 static bool sk_is_tcp(const struct sock *sk)
515 {
516 	return sk->sk_type == SOCK_STREAM &&
517 	       sk->sk_protocol == IPPROTO_TCP;
518 }
519 
520 static bool sock_map_redirect_allowed(const struct sock *sk)
521 {
522 	if (sk_is_tcp(sk))
523 		return sk->sk_state != TCP_LISTEN;
524 	else
525 		return sk->sk_state == TCP_ESTABLISHED;
526 }
527 
528 static bool sock_map_sk_is_suitable(const struct sock *sk)
529 {
530 	return !!sk->sk_prot->psock_update_sk_prot;
531 }
532 
533 static bool sock_map_sk_state_allowed(const struct sock *sk)
534 {
535 	if (sk_is_tcp(sk))
536 		return (1 << sk->sk_state) & (TCPF_ESTABLISHED | TCPF_LISTEN);
537 	return true;
538 }
539 
540 static int sock_hash_update_common(struct bpf_map *map, void *key,
541 				   struct sock *sk, u64 flags);
542 
543 int sock_map_update_elem_sys(struct bpf_map *map, void *key, void *value,
544 			     u64 flags)
545 {
546 	struct socket *sock;
547 	struct sock *sk;
548 	int ret;
549 	u64 ufd;
550 
551 	if (map->value_size == sizeof(u64))
552 		ufd = *(u64 *)value;
553 	else
554 		ufd = *(u32 *)value;
555 	if (ufd > S32_MAX)
556 		return -EINVAL;
557 
558 	sock = sockfd_lookup(ufd, &ret);
559 	if (!sock)
560 		return ret;
561 	sk = sock->sk;
562 	if (!sk) {
563 		ret = -EINVAL;
564 		goto out;
565 	}
566 	if (!sock_map_sk_is_suitable(sk)) {
567 		ret = -EOPNOTSUPP;
568 		goto out;
569 	}
570 
571 	sock_map_sk_acquire(sk);
572 	if (!sock_map_sk_state_allowed(sk))
573 		ret = -EOPNOTSUPP;
574 	else if (map->map_type == BPF_MAP_TYPE_SOCKMAP)
575 		ret = sock_map_update_common(map, *(u32 *)key, sk, flags);
576 	else
577 		ret = sock_hash_update_common(map, key, sk, flags);
578 	sock_map_sk_release(sk);
579 out:
580 	sockfd_put(sock);
581 	return ret;
582 }
583 
584 static int sock_map_update_elem(struct bpf_map *map, void *key,
585 				void *value, u64 flags)
586 {
587 	struct sock *sk = (struct sock *)value;
588 	int ret;
589 
590 	if (unlikely(!sk || !sk_fullsock(sk)))
591 		return -EINVAL;
592 
593 	if (!sock_map_sk_is_suitable(sk))
594 		return -EOPNOTSUPP;
595 
596 	local_bh_disable();
597 	bh_lock_sock(sk);
598 	if (!sock_map_sk_state_allowed(sk))
599 		ret = -EOPNOTSUPP;
600 	else if (map->map_type == BPF_MAP_TYPE_SOCKMAP)
601 		ret = sock_map_update_common(map, *(u32 *)key, sk, flags);
602 	else
603 		ret = sock_hash_update_common(map, key, sk, flags);
604 	bh_unlock_sock(sk);
605 	local_bh_enable();
606 	return ret;
607 }
608 
609 BPF_CALL_4(bpf_sock_map_update, struct bpf_sock_ops_kern *, sops,
610 	   struct bpf_map *, map, void *, key, u64, flags)
611 {
612 	WARN_ON_ONCE(!rcu_read_lock_held());
613 
614 	if (likely(sock_map_sk_is_suitable(sops->sk) &&
615 		   sock_map_op_okay(sops)))
616 		return sock_map_update_common(map, *(u32 *)key, sops->sk,
617 					      flags);
618 	return -EOPNOTSUPP;
619 }
620 
621 const struct bpf_func_proto bpf_sock_map_update_proto = {
622 	.func		= bpf_sock_map_update,
623 	.gpl_only	= false,
624 	.pkt_access	= true,
625 	.ret_type	= RET_INTEGER,
626 	.arg1_type	= ARG_PTR_TO_CTX,
627 	.arg2_type	= ARG_CONST_MAP_PTR,
628 	.arg3_type	= ARG_PTR_TO_MAP_KEY,
629 	.arg4_type	= ARG_ANYTHING,
630 };
631 
632 BPF_CALL_4(bpf_sk_redirect_map, struct sk_buff *, skb,
633 	   struct bpf_map *, map, u32, key, u64, flags)
634 {
635 	struct sock *sk;
636 
637 	if (unlikely(flags & ~(BPF_F_INGRESS)))
638 		return SK_DROP;
639 
640 	sk = __sock_map_lookup_elem(map, key);
641 	if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
642 		return SK_DROP;
643 
644 	skb_bpf_set_redir(skb, sk, flags & BPF_F_INGRESS);
645 	return SK_PASS;
646 }
647 
648 const struct bpf_func_proto bpf_sk_redirect_map_proto = {
649 	.func           = bpf_sk_redirect_map,
650 	.gpl_only       = false,
651 	.ret_type       = RET_INTEGER,
652 	.arg1_type	= ARG_PTR_TO_CTX,
653 	.arg2_type      = ARG_CONST_MAP_PTR,
654 	.arg3_type      = ARG_ANYTHING,
655 	.arg4_type      = ARG_ANYTHING,
656 };
657 
658 BPF_CALL_4(bpf_msg_redirect_map, struct sk_msg *, msg,
659 	   struct bpf_map *, map, u32, key, u64, flags)
660 {
661 	struct sock *sk;
662 
663 	if (unlikely(flags & ~(BPF_F_INGRESS)))
664 		return SK_DROP;
665 
666 	sk = __sock_map_lookup_elem(map, key);
667 	if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
668 		return SK_DROP;
669 
670 	msg->flags = flags;
671 	msg->sk_redir = sk;
672 	return SK_PASS;
673 }
674 
675 const struct bpf_func_proto bpf_msg_redirect_map_proto = {
676 	.func           = bpf_msg_redirect_map,
677 	.gpl_only       = false,
678 	.ret_type       = RET_INTEGER,
679 	.arg1_type	= ARG_PTR_TO_CTX,
680 	.arg2_type      = ARG_CONST_MAP_PTR,
681 	.arg3_type      = ARG_ANYTHING,
682 	.arg4_type      = ARG_ANYTHING,
683 };
684 
685 struct sock_map_seq_info {
686 	struct bpf_map *map;
687 	struct sock *sk;
688 	u32 index;
689 };
690 
691 struct bpf_iter__sockmap {
692 	__bpf_md_ptr(struct bpf_iter_meta *, meta);
693 	__bpf_md_ptr(struct bpf_map *, map);
694 	__bpf_md_ptr(void *, key);
695 	__bpf_md_ptr(struct sock *, sk);
696 };
697 
698 DEFINE_BPF_ITER_FUNC(sockmap, struct bpf_iter_meta *meta,
699 		     struct bpf_map *map, void *key,
700 		     struct sock *sk)
701 
702 static void *sock_map_seq_lookup_elem(struct sock_map_seq_info *info)
703 {
704 	if (unlikely(info->index >= info->map->max_entries))
705 		return NULL;
706 
707 	info->sk = __sock_map_lookup_elem(info->map, info->index);
708 
709 	/* can't return sk directly, since that might be NULL */
710 	return info;
711 }
712 
713 static void *sock_map_seq_start(struct seq_file *seq, loff_t *pos)
714 	__acquires(rcu)
715 {
716 	struct sock_map_seq_info *info = seq->private;
717 
718 	if (*pos == 0)
719 		++*pos;
720 
721 	/* pairs with sock_map_seq_stop */
722 	rcu_read_lock();
723 	return sock_map_seq_lookup_elem(info);
724 }
725 
726 static void *sock_map_seq_next(struct seq_file *seq, void *v, loff_t *pos)
727 	__must_hold(rcu)
728 {
729 	struct sock_map_seq_info *info = seq->private;
730 
731 	++*pos;
732 	++info->index;
733 
734 	return sock_map_seq_lookup_elem(info);
735 }
736 
737 static int sock_map_seq_show(struct seq_file *seq, void *v)
738 	__must_hold(rcu)
739 {
740 	struct sock_map_seq_info *info = seq->private;
741 	struct bpf_iter__sockmap ctx = {};
742 	struct bpf_iter_meta meta;
743 	struct bpf_prog *prog;
744 
745 	meta.seq = seq;
746 	prog = bpf_iter_get_info(&meta, !v);
747 	if (!prog)
748 		return 0;
749 
750 	ctx.meta = &meta;
751 	ctx.map = info->map;
752 	if (v) {
753 		ctx.key = &info->index;
754 		ctx.sk = info->sk;
755 	}
756 
757 	return bpf_iter_run_prog(prog, &ctx);
758 }
759 
760 static void sock_map_seq_stop(struct seq_file *seq, void *v)
761 	__releases(rcu)
762 {
763 	if (!v)
764 		(void)sock_map_seq_show(seq, NULL);
765 
766 	/* pairs with sock_map_seq_start */
767 	rcu_read_unlock();
768 }
769 
770 static const struct seq_operations sock_map_seq_ops = {
771 	.start	= sock_map_seq_start,
772 	.next	= sock_map_seq_next,
773 	.stop	= sock_map_seq_stop,
774 	.show	= sock_map_seq_show,
775 };
776 
777 static int sock_map_init_seq_private(void *priv_data,
778 				     struct bpf_iter_aux_info *aux)
779 {
780 	struct sock_map_seq_info *info = priv_data;
781 
782 	info->map = aux->map;
783 	return 0;
784 }
785 
786 static const struct bpf_iter_seq_info sock_map_iter_seq_info = {
787 	.seq_ops		= &sock_map_seq_ops,
788 	.init_seq_private	= sock_map_init_seq_private,
789 	.seq_priv_size		= sizeof(struct sock_map_seq_info),
790 };
791 
792 static int sock_map_btf_id;
793 const struct bpf_map_ops sock_map_ops = {
794 	.map_meta_equal		= bpf_map_meta_equal,
795 	.map_alloc		= sock_map_alloc,
796 	.map_free		= sock_map_free,
797 	.map_get_next_key	= sock_map_get_next_key,
798 	.map_lookup_elem_sys_only = sock_map_lookup_sys,
799 	.map_update_elem	= sock_map_update_elem,
800 	.map_delete_elem	= sock_map_delete_elem,
801 	.map_lookup_elem	= sock_map_lookup,
802 	.map_release_uref	= sock_map_release_progs,
803 	.map_check_btf		= map_check_no_btf,
804 	.map_btf_name		= "bpf_stab",
805 	.map_btf_id		= &sock_map_btf_id,
806 	.iter_seq_info		= &sock_map_iter_seq_info,
807 };
808 
809 struct bpf_shtab_elem {
810 	struct rcu_head rcu;
811 	u32 hash;
812 	struct sock *sk;
813 	struct hlist_node node;
814 	u8 key[];
815 };
816 
817 struct bpf_shtab_bucket {
818 	struct hlist_head head;
819 	raw_spinlock_t lock;
820 };
821 
822 struct bpf_shtab {
823 	struct bpf_map map;
824 	struct bpf_shtab_bucket *buckets;
825 	u32 buckets_num;
826 	u32 elem_size;
827 	struct sk_psock_progs progs;
828 	atomic_t count;
829 };
830 
831 static inline u32 sock_hash_bucket_hash(const void *key, u32 len)
832 {
833 	return jhash(key, len, 0);
834 }
835 
836 static struct bpf_shtab_bucket *sock_hash_select_bucket(struct bpf_shtab *htab,
837 							u32 hash)
838 {
839 	return &htab->buckets[hash & (htab->buckets_num - 1)];
840 }
841 
842 static struct bpf_shtab_elem *
843 sock_hash_lookup_elem_raw(struct hlist_head *head, u32 hash, void *key,
844 			  u32 key_size)
845 {
846 	struct bpf_shtab_elem *elem;
847 
848 	hlist_for_each_entry_rcu(elem, head, node) {
849 		if (elem->hash == hash &&
850 		    !memcmp(&elem->key, key, key_size))
851 			return elem;
852 	}
853 
854 	return NULL;
855 }
856 
857 static struct sock *__sock_hash_lookup_elem(struct bpf_map *map, void *key)
858 {
859 	struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
860 	u32 key_size = map->key_size, hash;
861 	struct bpf_shtab_bucket *bucket;
862 	struct bpf_shtab_elem *elem;
863 
864 	WARN_ON_ONCE(!rcu_read_lock_held());
865 
866 	hash = sock_hash_bucket_hash(key, key_size);
867 	bucket = sock_hash_select_bucket(htab, hash);
868 	elem = sock_hash_lookup_elem_raw(&bucket->head, hash, key, key_size);
869 
870 	return elem ? elem->sk : NULL;
871 }
872 
873 static void sock_hash_free_elem(struct bpf_shtab *htab,
874 				struct bpf_shtab_elem *elem)
875 {
876 	atomic_dec(&htab->count);
877 	kfree_rcu(elem, rcu);
878 }
879 
880 static void sock_hash_delete_from_link(struct bpf_map *map, struct sock *sk,
881 				       void *link_raw)
882 {
883 	struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
884 	struct bpf_shtab_elem *elem_probe, *elem = link_raw;
885 	struct bpf_shtab_bucket *bucket;
886 
887 	WARN_ON_ONCE(!rcu_read_lock_held());
888 	bucket = sock_hash_select_bucket(htab, elem->hash);
889 
890 	/* elem may be deleted in parallel from the map, but access here
891 	 * is okay since it's going away only after RCU grace period.
892 	 * However, we need to check whether it's still present.
893 	 */
894 	raw_spin_lock_bh(&bucket->lock);
895 	elem_probe = sock_hash_lookup_elem_raw(&bucket->head, elem->hash,
896 					       elem->key, map->key_size);
897 	if (elem_probe && elem_probe == elem) {
898 		hlist_del_rcu(&elem->node);
899 		sock_map_unref(elem->sk, elem);
900 		sock_hash_free_elem(htab, elem);
901 	}
902 	raw_spin_unlock_bh(&bucket->lock);
903 }
904 
905 static int sock_hash_delete_elem(struct bpf_map *map, void *key)
906 {
907 	struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
908 	u32 hash, key_size = map->key_size;
909 	struct bpf_shtab_bucket *bucket;
910 	struct bpf_shtab_elem *elem;
911 	int ret = -ENOENT;
912 
913 	hash = sock_hash_bucket_hash(key, key_size);
914 	bucket = sock_hash_select_bucket(htab, hash);
915 
916 	raw_spin_lock_bh(&bucket->lock);
917 	elem = sock_hash_lookup_elem_raw(&bucket->head, hash, key, key_size);
918 	if (elem) {
919 		hlist_del_rcu(&elem->node);
920 		sock_map_unref(elem->sk, elem);
921 		sock_hash_free_elem(htab, elem);
922 		ret = 0;
923 	}
924 	raw_spin_unlock_bh(&bucket->lock);
925 	return ret;
926 }
927 
928 static struct bpf_shtab_elem *sock_hash_alloc_elem(struct bpf_shtab *htab,
929 						   void *key, u32 key_size,
930 						   u32 hash, struct sock *sk,
931 						   struct bpf_shtab_elem *old)
932 {
933 	struct bpf_shtab_elem *new;
934 
935 	if (atomic_inc_return(&htab->count) > htab->map.max_entries) {
936 		if (!old) {
937 			atomic_dec(&htab->count);
938 			return ERR_PTR(-E2BIG);
939 		}
940 	}
941 
942 	new = bpf_map_kmalloc_node(&htab->map, htab->elem_size,
943 				   GFP_ATOMIC | __GFP_NOWARN,
944 				   htab->map.numa_node);
945 	if (!new) {
946 		atomic_dec(&htab->count);
947 		return ERR_PTR(-ENOMEM);
948 	}
949 	memcpy(new->key, key, key_size);
950 	new->sk = sk;
951 	new->hash = hash;
952 	return new;
953 }
954 
955 static int sock_hash_update_common(struct bpf_map *map, void *key,
956 				   struct sock *sk, u64 flags)
957 {
958 	struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
959 	u32 key_size = map->key_size, hash;
960 	struct bpf_shtab_elem *elem, *elem_new;
961 	struct bpf_shtab_bucket *bucket;
962 	struct sk_psock_link *link;
963 	struct sk_psock *psock;
964 	int ret;
965 
966 	WARN_ON_ONCE(!rcu_read_lock_held());
967 	if (unlikely(flags > BPF_EXIST))
968 		return -EINVAL;
969 
970 	link = sk_psock_init_link();
971 	if (!link)
972 		return -ENOMEM;
973 
974 	ret = sock_map_link(map, sk);
975 	if (ret < 0)
976 		goto out_free;
977 
978 	psock = sk_psock(sk);
979 	WARN_ON_ONCE(!psock);
980 
981 	hash = sock_hash_bucket_hash(key, key_size);
982 	bucket = sock_hash_select_bucket(htab, hash);
983 
984 	raw_spin_lock_bh(&bucket->lock);
985 	elem = sock_hash_lookup_elem_raw(&bucket->head, hash, key, key_size);
986 	if (elem && flags == BPF_NOEXIST) {
987 		ret = -EEXIST;
988 		goto out_unlock;
989 	} else if (!elem && flags == BPF_EXIST) {
990 		ret = -ENOENT;
991 		goto out_unlock;
992 	}
993 
994 	elem_new = sock_hash_alloc_elem(htab, key, key_size, hash, sk, elem);
995 	if (IS_ERR(elem_new)) {
996 		ret = PTR_ERR(elem_new);
997 		goto out_unlock;
998 	}
999 
1000 	sock_map_add_link(psock, link, map, elem_new);
1001 	/* Add new element to the head of the list, so that
1002 	 * concurrent search will find it before old elem.
1003 	 */
1004 	hlist_add_head_rcu(&elem_new->node, &bucket->head);
1005 	if (elem) {
1006 		hlist_del_rcu(&elem->node);
1007 		sock_map_unref(elem->sk, elem);
1008 		sock_hash_free_elem(htab, elem);
1009 	}
1010 	raw_spin_unlock_bh(&bucket->lock);
1011 	return 0;
1012 out_unlock:
1013 	raw_spin_unlock_bh(&bucket->lock);
1014 	sk_psock_put(sk, psock);
1015 out_free:
1016 	sk_psock_free_link(link);
1017 	return ret;
1018 }
1019 
1020 static int sock_hash_get_next_key(struct bpf_map *map, void *key,
1021 				  void *key_next)
1022 {
1023 	struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
1024 	struct bpf_shtab_elem *elem, *elem_next;
1025 	u32 hash, key_size = map->key_size;
1026 	struct hlist_head *head;
1027 	int i = 0;
1028 
1029 	if (!key)
1030 		goto find_first_elem;
1031 	hash = sock_hash_bucket_hash(key, key_size);
1032 	head = &sock_hash_select_bucket(htab, hash)->head;
1033 	elem = sock_hash_lookup_elem_raw(head, hash, key, key_size);
1034 	if (!elem)
1035 		goto find_first_elem;
1036 
1037 	elem_next = hlist_entry_safe(rcu_dereference(hlist_next_rcu(&elem->node)),
1038 				     struct bpf_shtab_elem, node);
1039 	if (elem_next) {
1040 		memcpy(key_next, elem_next->key, key_size);
1041 		return 0;
1042 	}
1043 
1044 	i = hash & (htab->buckets_num - 1);
1045 	i++;
1046 find_first_elem:
1047 	for (; i < htab->buckets_num; i++) {
1048 		head = &sock_hash_select_bucket(htab, i)->head;
1049 		elem_next = hlist_entry_safe(rcu_dereference(hlist_first_rcu(head)),
1050 					     struct bpf_shtab_elem, node);
1051 		if (elem_next) {
1052 			memcpy(key_next, elem_next->key, key_size);
1053 			return 0;
1054 		}
1055 	}
1056 
1057 	return -ENOENT;
1058 }
1059 
1060 static struct bpf_map *sock_hash_alloc(union bpf_attr *attr)
1061 {
1062 	struct bpf_shtab *htab;
1063 	int i, err;
1064 
1065 	if (!capable(CAP_NET_ADMIN))
1066 		return ERR_PTR(-EPERM);
1067 	if (attr->max_entries == 0 ||
1068 	    attr->key_size    == 0 ||
1069 	    (attr->value_size != sizeof(u32) &&
1070 	     attr->value_size != sizeof(u64)) ||
1071 	    attr->map_flags & ~SOCK_CREATE_FLAG_MASK)
1072 		return ERR_PTR(-EINVAL);
1073 	if (attr->key_size > MAX_BPF_STACK)
1074 		return ERR_PTR(-E2BIG);
1075 
1076 	htab = kzalloc(sizeof(*htab), GFP_USER | __GFP_ACCOUNT);
1077 	if (!htab)
1078 		return ERR_PTR(-ENOMEM);
1079 
1080 	bpf_map_init_from_attr(&htab->map, attr);
1081 
1082 	htab->buckets_num = roundup_pow_of_two(htab->map.max_entries);
1083 	htab->elem_size = sizeof(struct bpf_shtab_elem) +
1084 			  round_up(htab->map.key_size, 8);
1085 	if (htab->buckets_num == 0 ||
1086 	    htab->buckets_num > U32_MAX / sizeof(struct bpf_shtab_bucket)) {
1087 		err = -EINVAL;
1088 		goto free_htab;
1089 	}
1090 
1091 	htab->buckets = bpf_map_area_alloc(htab->buckets_num *
1092 					   sizeof(struct bpf_shtab_bucket),
1093 					   htab->map.numa_node);
1094 	if (!htab->buckets) {
1095 		err = -ENOMEM;
1096 		goto free_htab;
1097 	}
1098 
1099 	for (i = 0; i < htab->buckets_num; i++) {
1100 		INIT_HLIST_HEAD(&htab->buckets[i].head);
1101 		raw_spin_lock_init(&htab->buckets[i].lock);
1102 	}
1103 
1104 	return &htab->map;
1105 free_htab:
1106 	kfree(htab);
1107 	return ERR_PTR(err);
1108 }
1109 
1110 static void sock_hash_free(struct bpf_map *map)
1111 {
1112 	struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
1113 	struct bpf_shtab_bucket *bucket;
1114 	struct hlist_head unlink_list;
1115 	struct bpf_shtab_elem *elem;
1116 	struct hlist_node *node;
1117 	int i;
1118 
1119 	/* After the sync no updates or deletes will be in-flight so it
1120 	 * is safe to walk map and remove entries without risking a race
1121 	 * in EEXIST update case.
1122 	 */
1123 	synchronize_rcu();
1124 	for (i = 0; i < htab->buckets_num; i++) {
1125 		bucket = sock_hash_select_bucket(htab, i);
1126 
1127 		/* We are racing with sock_hash_delete_from_link to
1128 		 * enter the spin-lock critical section. Every socket on
1129 		 * the list is still linked to sockhash. Since link
1130 		 * exists, psock exists and holds a ref to socket. That
1131 		 * lets us to grab a socket ref too.
1132 		 */
1133 		raw_spin_lock_bh(&bucket->lock);
1134 		hlist_for_each_entry(elem, &bucket->head, node)
1135 			sock_hold(elem->sk);
1136 		hlist_move_list(&bucket->head, &unlink_list);
1137 		raw_spin_unlock_bh(&bucket->lock);
1138 
1139 		/* Process removed entries out of atomic context to
1140 		 * block for socket lock before deleting the psock's
1141 		 * link to sockhash.
1142 		 */
1143 		hlist_for_each_entry_safe(elem, node, &unlink_list, node) {
1144 			hlist_del(&elem->node);
1145 			lock_sock(elem->sk);
1146 			rcu_read_lock();
1147 			sock_map_unref(elem->sk, elem);
1148 			rcu_read_unlock();
1149 			release_sock(elem->sk);
1150 			sock_put(elem->sk);
1151 			sock_hash_free_elem(htab, elem);
1152 		}
1153 	}
1154 
1155 	/* wait for psock readers accessing its map link */
1156 	synchronize_rcu();
1157 
1158 	bpf_map_area_free(htab->buckets);
1159 	kfree(htab);
1160 }
1161 
1162 static void *sock_hash_lookup_sys(struct bpf_map *map, void *key)
1163 {
1164 	struct sock *sk;
1165 
1166 	if (map->value_size != sizeof(u64))
1167 		return ERR_PTR(-ENOSPC);
1168 
1169 	sk = __sock_hash_lookup_elem(map, key);
1170 	if (!sk)
1171 		return ERR_PTR(-ENOENT);
1172 
1173 	__sock_gen_cookie(sk);
1174 	return &sk->sk_cookie;
1175 }
1176 
1177 static void *sock_hash_lookup(struct bpf_map *map, void *key)
1178 {
1179 	struct sock *sk;
1180 
1181 	sk = __sock_hash_lookup_elem(map, key);
1182 	if (!sk)
1183 		return NULL;
1184 	if (sk_is_refcounted(sk) && !refcount_inc_not_zero(&sk->sk_refcnt))
1185 		return NULL;
1186 	return sk;
1187 }
1188 
1189 static void sock_hash_release_progs(struct bpf_map *map)
1190 {
1191 	psock_progs_drop(&container_of(map, struct bpf_shtab, map)->progs);
1192 }
1193 
1194 BPF_CALL_4(bpf_sock_hash_update, struct bpf_sock_ops_kern *, sops,
1195 	   struct bpf_map *, map, void *, key, u64, flags)
1196 {
1197 	WARN_ON_ONCE(!rcu_read_lock_held());
1198 
1199 	if (likely(sock_map_sk_is_suitable(sops->sk) &&
1200 		   sock_map_op_okay(sops)))
1201 		return sock_hash_update_common(map, key, sops->sk, flags);
1202 	return -EOPNOTSUPP;
1203 }
1204 
1205 const struct bpf_func_proto bpf_sock_hash_update_proto = {
1206 	.func		= bpf_sock_hash_update,
1207 	.gpl_only	= false,
1208 	.pkt_access	= true,
1209 	.ret_type	= RET_INTEGER,
1210 	.arg1_type	= ARG_PTR_TO_CTX,
1211 	.arg2_type	= ARG_CONST_MAP_PTR,
1212 	.arg3_type	= ARG_PTR_TO_MAP_KEY,
1213 	.arg4_type	= ARG_ANYTHING,
1214 };
1215 
1216 BPF_CALL_4(bpf_sk_redirect_hash, struct sk_buff *, skb,
1217 	   struct bpf_map *, map, void *, key, u64, flags)
1218 {
1219 	struct sock *sk;
1220 
1221 	if (unlikely(flags & ~(BPF_F_INGRESS)))
1222 		return SK_DROP;
1223 
1224 	sk = __sock_hash_lookup_elem(map, key);
1225 	if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
1226 		return SK_DROP;
1227 
1228 	skb_bpf_set_redir(skb, sk, flags & BPF_F_INGRESS);
1229 	return SK_PASS;
1230 }
1231 
1232 const struct bpf_func_proto bpf_sk_redirect_hash_proto = {
1233 	.func           = bpf_sk_redirect_hash,
1234 	.gpl_only       = false,
1235 	.ret_type       = RET_INTEGER,
1236 	.arg1_type	= ARG_PTR_TO_CTX,
1237 	.arg2_type      = ARG_CONST_MAP_PTR,
1238 	.arg3_type      = ARG_PTR_TO_MAP_KEY,
1239 	.arg4_type      = ARG_ANYTHING,
1240 };
1241 
1242 BPF_CALL_4(bpf_msg_redirect_hash, struct sk_msg *, msg,
1243 	   struct bpf_map *, map, void *, key, u64, flags)
1244 {
1245 	struct sock *sk;
1246 
1247 	if (unlikely(flags & ~(BPF_F_INGRESS)))
1248 		return SK_DROP;
1249 
1250 	sk = __sock_hash_lookup_elem(map, key);
1251 	if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
1252 		return SK_DROP;
1253 
1254 	msg->flags = flags;
1255 	msg->sk_redir = sk;
1256 	return SK_PASS;
1257 }
1258 
1259 const struct bpf_func_proto bpf_msg_redirect_hash_proto = {
1260 	.func           = bpf_msg_redirect_hash,
1261 	.gpl_only       = false,
1262 	.ret_type       = RET_INTEGER,
1263 	.arg1_type	= ARG_PTR_TO_CTX,
1264 	.arg2_type      = ARG_CONST_MAP_PTR,
1265 	.arg3_type      = ARG_PTR_TO_MAP_KEY,
1266 	.arg4_type      = ARG_ANYTHING,
1267 };
1268 
1269 struct sock_hash_seq_info {
1270 	struct bpf_map *map;
1271 	struct bpf_shtab *htab;
1272 	u32 bucket_id;
1273 };
1274 
1275 static void *sock_hash_seq_find_next(struct sock_hash_seq_info *info,
1276 				     struct bpf_shtab_elem *prev_elem)
1277 {
1278 	const struct bpf_shtab *htab = info->htab;
1279 	struct bpf_shtab_bucket *bucket;
1280 	struct bpf_shtab_elem *elem;
1281 	struct hlist_node *node;
1282 
1283 	/* try to find next elem in the same bucket */
1284 	if (prev_elem) {
1285 		node = rcu_dereference(hlist_next_rcu(&prev_elem->node));
1286 		elem = hlist_entry_safe(node, struct bpf_shtab_elem, node);
1287 		if (elem)
1288 			return elem;
1289 
1290 		/* no more elements, continue in the next bucket */
1291 		info->bucket_id++;
1292 	}
1293 
1294 	for (; info->bucket_id < htab->buckets_num; info->bucket_id++) {
1295 		bucket = &htab->buckets[info->bucket_id];
1296 		node = rcu_dereference(hlist_first_rcu(&bucket->head));
1297 		elem = hlist_entry_safe(node, struct bpf_shtab_elem, node);
1298 		if (elem)
1299 			return elem;
1300 	}
1301 
1302 	return NULL;
1303 }
1304 
1305 static void *sock_hash_seq_start(struct seq_file *seq, loff_t *pos)
1306 	__acquires(rcu)
1307 {
1308 	struct sock_hash_seq_info *info = seq->private;
1309 
1310 	if (*pos == 0)
1311 		++*pos;
1312 
1313 	/* pairs with sock_hash_seq_stop */
1314 	rcu_read_lock();
1315 	return sock_hash_seq_find_next(info, NULL);
1316 }
1317 
1318 static void *sock_hash_seq_next(struct seq_file *seq, void *v, loff_t *pos)
1319 	__must_hold(rcu)
1320 {
1321 	struct sock_hash_seq_info *info = seq->private;
1322 
1323 	++*pos;
1324 	return sock_hash_seq_find_next(info, v);
1325 }
1326 
1327 static int sock_hash_seq_show(struct seq_file *seq, void *v)
1328 	__must_hold(rcu)
1329 {
1330 	struct sock_hash_seq_info *info = seq->private;
1331 	struct bpf_iter__sockmap ctx = {};
1332 	struct bpf_shtab_elem *elem = v;
1333 	struct bpf_iter_meta meta;
1334 	struct bpf_prog *prog;
1335 
1336 	meta.seq = seq;
1337 	prog = bpf_iter_get_info(&meta, !elem);
1338 	if (!prog)
1339 		return 0;
1340 
1341 	ctx.meta = &meta;
1342 	ctx.map = info->map;
1343 	if (elem) {
1344 		ctx.key = elem->key;
1345 		ctx.sk = elem->sk;
1346 	}
1347 
1348 	return bpf_iter_run_prog(prog, &ctx);
1349 }
1350 
1351 static void sock_hash_seq_stop(struct seq_file *seq, void *v)
1352 	__releases(rcu)
1353 {
1354 	if (!v)
1355 		(void)sock_hash_seq_show(seq, NULL);
1356 
1357 	/* pairs with sock_hash_seq_start */
1358 	rcu_read_unlock();
1359 }
1360 
1361 static const struct seq_operations sock_hash_seq_ops = {
1362 	.start	= sock_hash_seq_start,
1363 	.next	= sock_hash_seq_next,
1364 	.stop	= sock_hash_seq_stop,
1365 	.show	= sock_hash_seq_show,
1366 };
1367 
1368 static int sock_hash_init_seq_private(void *priv_data,
1369 				     struct bpf_iter_aux_info *aux)
1370 {
1371 	struct sock_hash_seq_info *info = priv_data;
1372 
1373 	info->map = aux->map;
1374 	info->htab = container_of(aux->map, struct bpf_shtab, map);
1375 	return 0;
1376 }
1377 
1378 static const struct bpf_iter_seq_info sock_hash_iter_seq_info = {
1379 	.seq_ops		= &sock_hash_seq_ops,
1380 	.init_seq_private	= sock_hash_init_seq_private,
1381 	.seq_priv_size		= sizeof(struct sock_hash_seq_info),
1382 };
1383 
1384 static int sock_hash_map_btf_id;
1385 const struct bpf_map_ops sock_hash_ops = {
1386 	.map_meta_equal		= bpf_map_meta_equal,
1387 	.map_alloc		= sock_hash_alloc,
1388 	.map_free		= sock_hash_free,
1389 	.map_get_next_key	= sock_hash_get_next_key,
1390 	.map_update_elem	= sock_map_update_elem,
1391 	.map_delete_elem	= sock_hash_delete_elem,
1392 	.map_lookup_elem	= sock_hash_lookup,
1393 	.map_lookup_elem_sys_only = sock_hash_lookup_sys,
1394 	.map_release_uref	= sock_hash_release_progs,
1395 	.map_check_btf		= map_check_no_btf,
1396 	.map_btf_name		= "bpf_shtab",
1397 	.map_btf_id		= &sock_hash_map_btf_id,
1398 	.iter_seq_info		= &sock_hash_iter_seq_info,
1399 };
1400 
1401 static struct sk_psock_progs *sock_map_progs(struct bpf_map *map)
1402 {
1403 	switch (map->map_type) {
1404 	case BPF_MAP_TYPE_SOCKMAP:
1405 		return &container_of(map, struct bpf_stab, map)->progs;
1406 	case BPF_MAP_TYPE_SOCKHASH:
1407 		return &container_of(map, struct bpf_shtab, map)->progs;
1408 	default:
1409 		break;
1410 	}
1411 
1412 	return NULL;
1413 }
1414 
1415 static int sock_map_prog_update(struct bpf_map *map, struct bpf_prog *prog,
1416 				struct bpf_prog *old, u32 which)
1417 {
1418 	struct sk_psock_progs *progs = sock_map_progs(map);
1419 	struct bpf_prog **pprog;
1420 
1421 	if (!progs)
1422 		return -EOPNOTSUPP;
1423 
1424 	switch (which) {
1425 	case BPF_SK_MSG_VERDICT:
1426 		pprog = &progs->msg_parser;
1427 		break;
1428 #if IS_ENABLED(CONFIG_BPF_STREAM_PARSER)
1429 	case BPF_SK_SKB_STREAM_PARSER:
1430 		pprog = &progs->stream_parser;
1431 		break;
1432 #endif
1433 	case BPF_SK_SKB_STREAM_VERDICT:
1434 		if (progs->skb_verdict)
1435 			return -EBUSY;
1436 		pprog = &progs->stream_verdict;
1437 		break;
1438 	case BPF_SK_SKB_VERDICT:
1439 		if (progs->stream_verdict)
1440 			return -EBUSY;
1441 		pprog = &progs->skb_verdict;
1442 		break;
1443 	default:
1444 		return -EOPNOTSUPP;
1445 	}
1446 
1447 	if (old)
1448 		return psock_replace_prog(pprog, prog, old);
1449 
1450 	psock_set_prog(pprog, prog);
1451 	return 0;
1452 }
1453 
1454 static void sock_map_unlink(struct sock *sk, struct sk_psock_link *link)
1455 {
1456 	switch (link->map->map_type) {
1457 	case BPF_MAP_TYPE_SOCKMAP:
1458 		return sock_map_delete_from_link(link->map, sk,
1459 						 link->link_raw);
1460 	case BPF_MAP_TYPE_SOCKHASH:
1461 		return sock_hash_delete_from_link(link->map, sk,
1462 						  link->link_raw);
1463 	default:
1464 		break;
1465 	}
1466 }
1467 
1468 static void sock_map_remove_links(struct sock *sk, struct sk_psock *psock)
1469 {
1470 	struct sk_psock_link *link;
1471 
1472 	while ((link = sk_psock_link_pop(psock))) {
1473 		sock_map_unlink(sk, link);
1474 		sk_psock_free_link(link);
1475 	}
1476 }
1477 
1478 void sock_map_unhash(struct sock *sk)
1479 {
1480 	void (*saved_unhash)(struct sock *sk);
1481 	struct sk_psock *psock;
1482 
1483 	rcu_read_lock();
1484 	psock = sk_psock(sk);
1485 	if (unlikely(!psock)) {
1486 		rcu_read_unlock();
1487 		if (sk->sk_prot->unhash)
1488 			sk->sk_prot->unhash(sk);
1489 		return;
1490 	}
1491 
1492 	saved_unhash = psock->saved_unhash;
1493 	sock_map_remove_links(sk, psock);
1494 	rcu_read_unlock();
1495 	saved_unhash(sk);
1496 }
1497 EXPORT_SYMBOL_GPL(sock_map_unhash);
1498 
1499 void sock_map_close(struct sock *sk, long timeout)
1500 {
1501 	void (*saved_close)(struct sock *sk, long timeout);
1502 	struct sk_psock *psock;
1503 
1504 	lock_sock(sk);
1505 	rcu_read_lock();
1506 	psock = sk_psock_get(sk);
1507 	if (unlikely(!psock)) {
1508 		rcu_read_unlock();
1509 		release_sock(sk);
1510 		return sk->sk_prot->close(sk, timeout);
1511 	}
1512 
1513 	saved_close = psock->saved_close;
1514 	sock_map_remove_links(sk, psock);
1515 	rcu_read_unlock();
1516 	sk_psock_stop(psock, true);
1517 	sk_psock_put(sk, psock);
1518 	release_sock(sk);
1519 	saved_close(sk, timeout);
1520 }
1521 EXPORT_SYMBOL_GPL(sock_map_close);
1522 
1523 static int sock_map_iter_attach_target(struct bpf_prog *prog,
1524 				       union bpf_iter_link_info *linfo,
1525 				       struct bpf_iter_aux_info *aux)
1526 {
1527 	struct bpf_map *map;
1528 	int err = -EINVAL;
1529 
1530 	if (!linfo->map.map_fd)
1531 		return -EBADF;
1532 
1533 	map = bpf_map_get_with_uref(linfo->map.map_fd);
1534 	if (IS_ERR(map))
1535 		return PTR_ERR(map);
1536 
1537 	if (map->map_type != BPF_MAP_TYPE_SOCKMAP &&
1538 	    map->map_type != BPF_MAP_TYPE_SOCKHASH)
1539 		goto put_map;
1540 
1541 	if (prog->aux->max_rdonly_access > map->key_size) {
1542 		err = -EACCES;
1543 		goto put_map;
1544 	}
1545 
1546 	aux->map = map;
1547 	return 0;
1548 
1549 put_map:
1550 	bpf_map_put_with_uref(map);
1551 	return err;
1552 }
1553 
1554 static void sock_map_iter_detach_target(struct bpf_iter_aux_info *aux)
1555 {
1556 	bpf_map_put_with_uref(aux->map);
1557 }
1558 
1559 static struct bpf_iter_reg sock_map_iter_reg = {
1560 	.target			= "sockmap",
1561 	.attach_target		= sock_map_iter_attach_target,
1562 	.detach_target		= sock_map_iter_detach_target,
1563 	.show_fdinfo		= bpf_iter_map_show_fdinfo,
1564 	.fill_link_info		= bpf_iter_map_fill_link_info,
1565 	.ctx_arg_info_size	= 2,
1566 	.ctx_arg_info		= {
1567 		{ offsetof(struct bpf_iter__sockmap, key),
1568 		  PTR_TO_RDONLY_BUF_OR_NULL },
1569 		{ offsetof(struct bpf_iter__sockmap, sk),
1570 		  PTR_TO_BTF_ID_OR_NULL },
1571 	},
1572 };
1573 
1574 static int __init bpf_sockmap_iter_init(void)
1575 {
1576 	sock_map_iter_reg.ctx_arg_info[1].btf_id =
1577 		btf_sock_ids[BTF_SOCK_TYPE_SOCK];
1578 	return bpf_iter_reg_target(&sock_map_iter_reg);
1579 }
1580 late_initcall(bpf_sockmap_iter_init);
1581