xref: /openbmc/linux/net/core/sock_map.c (revision bb26cfd9)
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 		if (verdict_stop)
171 			sk_psock_stop_verdict(sk, psock);
172 
173 		if (psock->psock_update_sk_prot)
174 			psock->psock_update_sk_prot(sk, psock, false);
175 		write_unlock_bh(&sk->sk_callback_lock);
176 	}
177 }
178 
179 static void sock_map_unref(struct sock *sk, void *link_raw)
180 {
181 	struct sk_psock *psock = sk_psock(sk);
182 
183 	if (likely(psock)) {
184 		sock_map_del_link(sk, psock, link_raw);
185 		sk_psock_put(sk, psock);
186 	}
187 }
188 
189 static int sock_map_init_proto(struct sock *sk, struct sk_psock *psock)
190 {
191 	if (!sk->sk_prot->psock_update_sk_prot)
192 		return -EINVAL;
193 	psock->psock_update_sk_prot = sk->sk_prot->psock_update_sk_prot;
194 	return sk->sk_prot->psock_update_sk_prot(sk, psock, false);
195 }
196 
197 static struct sk_psock *sock_map_psock_get_checked(struct sock *sk)
198 {
199 	struct sk_psock *psock;
200 
201 	rcu_read_lock();
202 	psock = sk_psock(sk);
203 	if (psock) {
204 		if (sk->sk_prot->close != sock_map_close) {
205 			psock = ERR_PTR(-EBUSY);
206 			goto out;
207 		}
208 
209 		if (!refcount_inc_not_zero(&psock->refcnt))
210 			psock = ERR_PTR(-EBUSY);
211 	}
212 out:
213 	rcu_read_unlock();
214 	return psock;
215 }
216 
217 static int sock_map_link(struct bpf_map *map, struct sock *sk)
218 {
219 	struct sk_psock_progs *progs = sock_map_progs(map);
220 	struct bpf_prog *stream_verdict = NULL;
221 	struct bpf_prog *stream_parser = NULL;
222 	struct bpf_prog *skb_verdict = NULL;
223 	struct bpf_prog *msg_parser = NULL;
224 	struct sk_psock *psock;
225 	int ret;
226 
227 	stream_verdict = READ_ONCE(progs->stream_verdict);
228 	if (stream_verdict) {
229 		stream_verdict = bpf_prog_inc_not_zero(stream_verdict);
230 		if (IS_ERR(stream_verdict))
231 			return PTR_ERR(stream_verdict);
232 	}
233 
234 	stream_parser = READ_ONCE(progs->stream_parser);
235 	if (stream_parser) {
236 		stream_parser = bpf_prog_inc_not_zero(stream_parser);
237 		if (IS_ERR(stream_parser)) {
238 			ret = PTR_ERR(stream_parser);
239 			goto out_put_stream_verdict;
240 		}
241 	}
242 
243 	msg_parser = READ_ONCE(progs->msg_parser);
244 	if (msg_parser) {
245 		msg_parser = bpf_prog_inc_not_zero(msg_parser);
246 		if (IS_ERR(msg_parser)) {
247 			ret = PTR_ERR(msg_parser);
248 			goto out_put_stream_parser;
249 		}
250 	}
251 
252 	skb_verdict = READ_ONCE(progs->skb_verdict);
253 	if (skb_verdict) {
254 		skb_verdict = bpf_prog_inc_not_zero(skb_verdict);
255 		if (IS_ERR(skb_verdict)) {
256 			ret = PTR_ERR(skb_verdict);
257 			goto out_put_msg_parser;
258 		}
259 	}
260 
261 	psock = sock_map_psock_get_checked(sk);
262 	if (IS_ERR(psock)) {
263 		ret = PTR_ERR(psock);
264 		goto out_progs;
265 	}
266 
267 	if (psock) {
268 		if ((msg_parser && READ_ONCE(psock->progs.msg_parser)) ||
269 		    (stream_parser  && READ_ONCE(psock->progs.stream_parser)) ||
270 		    (skb_verdict && READ_ONCE(psock->progs.skb_verdict)) ||
271 		    (skb_verdict && READ_ONCE(psock->progs.stream_verdict)) ||
272 		    (stream_verdict && READ_ONCE(psock->progs.skb_verdict)) ||
273 		    (stream_verdict && READ_ONCE(psock->progs.stream_verdict))) {
274 			sk_psock_put(sk, psock);
275 			ret = -EBUSY;
276 			goto out_progs;
277 		}
278 	} else {
279 		psock = sk_psock_init(sk, map->numa_node);
280 		if (IS_ERR(psock)) {
281 			ret = PTR_ERR(psock);
282 			goto out_progs;
283 		}
284 	}
285 
286 	if (msg_parser)
287 		psock_set_prog(&psock->progs.msg_parser, msg_parser);
288 	if (stream_parser)
289 		psock_set_prog(&psock->progs.stream_parser, stream_parser);
290 	if (stream_verdict)
291 		psock_set_prog(&psock->progs.stream_verdict, stream_verdict);
292 	if (skb_verdict)
293 		psock_set_prog(&psock->progs.skb_verdict, skb_verdict);
294 
295 	/* msg_* and stream_* programs references tracked in psock after this
296 	 * point. Reference dec and cleanup will occur through psock destructor
297 	 */
298 	ret = sock_map_init_proto(sk, psock);
299 	if (ret < 0) {
300 		sk_psock_put(sk, psock);
301 		goto out;
302 	}
303 
304 	write_lock_bh(&sk->sk_callback_lock);
305 	if (stream_parser && stream_verdict && !psock->saved_data_ready) {
306 		ret = sk_psock_init_strp(sk, psock);
307 		if (ret) {
308 			write_unlock_bh(&sk->sk_callback_lock);
309 			sk_psock_put(sk, psock);
310 			goto out;
311 		}
312 		sk_psock_start_strp(sk, psock);
313 	} else if (!stream_parser && stream_verdict && !psock->saved_data_ready) {
314 		sk_psock_start_verdict(sk,psock);
315 	} else if (!stream_verdict && skb_verdict && !psock->saved_data_ready) {
316 		sk_psock_start_verdict(sk, psock);
317 	}
318 	write_unlock_bh(&sk->sk_callback_lock);
319 	return 0;
320 out_progs:
321 	if (skb_verdict)
322 		bpf_prog_put(skb_verdict);
323 out_put_msg_parser:
324 	if (msg_parser)
325 		bpf_prog_put(msg_parser);
326 out_put_stream_parser:
327 	if (stream_parser)
328 		bpf_prog_put(stream_parser);
329 out_put_stream_verdict:
330 	if (stream_verdict)
331 		bpf_prog_put(stream_verdict);
332 out:
333 	return ret;
334 }
335 
336 static void sock_map_free(struct bpf_map *map)
337 {
338 	struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
339 	int i;
340 
341 	/* After the sync no updates or deletes will be in-flight so it
342 	 * is safe to walk map and remove entries without risking a race
343 	 * in EEXIST update case.
344 	 */
345 	synchronize_rcu();
346 	for (i = 0; i < stab->map.max_entries; i++) {
347 		struct sock **psk = &stab->sks[i];
348 		struct sock *sk;
349 
350 		sk = xchg(psk, NULL);
351 		if (sk) {
352 			lock_sock(sk);
353 			rcu_read_lock();
354 			sock_map_unref(sk, psk);
355 			rcu_read_unlock();
356 			release_sock(sk);
357 		}
358 	}
359 
360 	/* wait for psock readers accessing its map link */
361 	synchronize_rcu();
362 
363 	bpf_map_area_free(stab->sks);
364 	kfree(stab);
365 }
366 
367 static void sock_map_release_progs(struct bpf_map *map)
368 {
369 	psock_progs_drop(&container_of(map, struct bpf_stab, map)->progs);
370 }
371 
372 static struct sock *__sock_map_lookup_elem(struct bpf_map *map, u32 key)
373 {
374 	struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
375 
376 	WARN_ON_ONCE(!rcu_read_lock_held());
377 
378 	if (unlikely(key >= map->max_entries))
379 		return NULL;
380 	return READ_ONCE(stab->sks[key]);
381 }
382 
383 static void *sock_map_lookup(struct bpf_map *map, void *key)
384 {
385 	struct sock *sk;
386 
387 	sk = __sock_map_lookup_elem(map, *(u32 *)key);
388 	if (!sk)
389 		return NULL;
390 	if (sk_is_refcounted(sk) && !refcount_inc_not_zero(&sk->sk_refcnt))
391 		return NULL;
392 	return sk;
393 }
394 
395 static void *sock_map_lookup_sys(struct bpf_map *map, void *key)
396 {
397 	struct sock *sk;
398 
399 	if (map->value_size != sizeof(u64))
400 		return ERR_PTR(-ENOSPC);
401 
402 	sk = __sock_map_lookup_elem(map, *(u32 *)key);
403 	if (!sk)
404 		return ERR_PTR(-ENOENT);
405 
406 	__sock_gen_cookie(sk);
407 	return &sk->sk_cookie;
408 }
409 
410 static int __sock_map_delete(struct bpf_stab *stab, struct sock *sk_test,
411 			     struct sock **psk)
412 {
413 	struct sock *sk;
414 	int err = 0;
415 
416 	raw_spin_lock_bh(&stab->lock);
417 	sk = *psk;
418 	if (!sk_test || sk_test == sk)
419 		sk = xchg(psk, NULL);
420 
421 	if (likely(sk))
422 		sock_map_unref(sk, psk);
423 	else
424 		err = -EINVAL;
425 
426 	raw_spin_unlock_bh(&stab->lock);
427 	return err;
428 }
429 
430 static void sock_map_delete_from_link(struct bpf_map *map, struct sock *sk,
431 				      void *link_raw)
432 {
433 	struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
434 
435 	__sock_map_delete(stab, sk, link_raw);
436 }
437 
438 static int sock_map_delete_elem(struct bpf_map *map, void *key)
439 {
440 	struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
441 	u32 i = *(u32 *)key;
442 	struct sock **psk;
443 
444 	if (unlikely(i >= map->max_entries))
445 		return -EINVAL;
446 
447 	psk = &stab->sks[i];
448 	return __sock_map_delete(stab, NULL, psk);
449 }
450 
451 static int sock_map_get_next_key(struct bpf_map *map, void *key, void *next)
452 {
453 	struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
454 	u32 i = key ? *(u32 *)key : U32_MAX;
455 	u32 *key_next = next;
456 
457 	if (i == stab->map.max_entries - 1)
458 		return -ENOENT;
459 	if (i >= stab->map.max_entries)
460 		*key_next = 0;
461 	else
462 		*key_next = i + 1;
463 	return 0;
464 }
465 
466 static int sock_map_update_common(struct bpf_map *map, u32 idx,
467 				  struct sock *sk, u64 flags)
468 {
469 	struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
470 	struct sk_psock_link *link;
471 	struct sk_psock *psock;
472 	struct sock *osk;
473 	int ret;
474 
475 	WARN_ON_ONCE(!rcu_read_lock_held());
476 	if (unlikely(flags > BPF_EXIST))
477 		return -EINVAL;
478 	if (unlikely(idx >= map->max_entries))
479 		return -E2BIG;
480 
481 	link = sk_psock_init_link();
482 	if (!link)
483 		return -ENOMEM;
484 
485 	ret = sock_map_link(map, sk);
486 	if (ret < 0)
487 		goto out_free;
488 
489 	psock = sk_psock(sk);
490 	WARN_ON_ONCE(!psock);
491 
492 	raw_spin_lock_bh(&stab->lock);
493 	osk = stab->sks[idx];
494 	if (osk && flags == BPF_NOEXIST) {
495 		ret = -EEXIST;
496 		goto out_unlock;
497 	} else if (!osk && flags == BPF_EXIST) {
498 		ret = -ENOENT;
499 		goto out_unlock;
500 	}
501 
502 	sock_map_add_link(psock, link, map, &stab->sks[idx]);
503 	stab->sks[idx] = sk;
504 	if (osk)
505 		sock_map_unref(osk, &stab->sks[idx]);
506 	raw_spin_unlock_bh(&stab->lock);
507 	return 0;
508 out_unlock:
509 	raw_spin_unlock_bh(&stab->lock);
510 	if (psock)
511 		sk_psock_put(sk, psock);
512 out_free:
513 	sk_psock_free_link(link);
514 	return ret;
515 }
516 
517 static bool sock_map_op_okay(const struct bpf_sock_ops_kern *ops)
518 {
519 	return ops->op == BPF_SOCK_OPS_PASSIVE_ESTABLISHED_CB ||
520 	       ops->op == BPF_SOCK_OPS_ACTIVE_ESTABLISHED_CB ||
521 	       ops->op == BPF_SOCK_OPS_TCP_LISTEN_CB;
522 }
523 
524 static bool sock_map_redirect_allowed(const struct sock *sk)
525 {
526 	if (sk_is_tcp(sk))
527 		return sk->sk_state != TCP_LISTEN;
528 	else
529 		return sk->sk_state == TCP_ESTABLISHED;
530 }
531 
532 static bool sock_map_sk_is_suitable(const struct sock *sk)
533 {
534 	return !!sk->sk_prot->psock_update_sk_prot;
535 }
536 
537 static bool sock_map_sk_state_allowed(const struct sock *sk)
538 {
539 	if (sk_is_tcp(sk))
540 		return (1 << sk->sk_state) & (TCPF_ESTABLISHED | TCPF_LISTEN);
541 	return true;
542 }
543 
544 static int sock_hash_update_common(struct bpf_map *map, void *key,
545 				   struct sock *sk, u64 flags);
546 
547 int sock_map_update_elem_sys(struct bpf_map *map, void *key, void *value,
548 			     u64 flags)
549 {
550 	struct socket *sock;
551 	struct sock *sk;
552 	int ret;
553 	u64 ufd;
554 
555 	if (map->value_size == sizeof(u64))
556 		ufd = *(u64 *)value;
557 	else
558 		ufd = *(u32 *)value;
559 	if (ufd > S32_MAX)
560 		return -EINVAL;
561 
562 	sock = sockfd_lookup(ufd, &ret);
563 	if (!sock)
564 		return ret;
565 	sk = sock->sk;
566 	if (!sk) {
567 		ret = -EINVAL;
568 		goto out;
569 	}
570 	if (!sock_map_sk_is_suitable(sk)) {
571 		ret = -EOPNOTSUPP;
572 		goto out;
573 	}
574 
575 	sock_map_sk_acquire(sk);
576 	if (!sock_map_sk_state_allowed(sk))
577 		ret = -EOPNOTSUPP;
578 	else if (map->map_type == BPF_MAP_TYPE_SOCKMAP)
579 		ret = sock_map_update_common(map, *(u32 *)key, sk, flags);
580 	else
581 		ret = sock_hash_update_common(map, key, sk, flags);
582 	sock_map_sk_release(sk);
583 out:
584 	sockfd_put(sock);
585 	return ret;
586 }
587 
588 static int sock_map_update_elem(struct bpf_map *map, void *key,
589 				void *value, u64 flags)
590 {
591 	struct sock *sk = (struct sock *)value;
592 	int ret;
593 
594 	if (unlikely(!sk || !sk_fullsock(sk)))
595 		return -EINVAL;
596 
597 	if (!sock_map_sk_is_suitable(sk))
598 		return -EOPNOTSUPP;
599 
600 	local_bh_disable();
601 	bh_lock_sock(sk);
602 	if (!sock_map_sk_state_allowed(sk))
603 		ret = -EOPNOTSUPP;
604 	else if (map->map_type == BPF_MAP_TYPE_SOCKMAP)
605 		ret = sock_map_update_common(map, *(u32 *)key, sk, flags);
606 	else
607 		ret = sock_hash_update_common(map, key, sk, flags);
608 	bh_unlock_sock(sk);
609 	local_bh_enable();
610 	return ret;
611 }
612 
613 BPF_CALL_4(bpf_sock_map_update, struct bpf_sock_ops_kern *, sops,
614 	   struct bpf_map *, map, void *, key, u64, flags)
615 {
616 	WARN_ON_ONCE(!rcu_read_lock_held());
617 
618 	if (likely(sock_map_sk_is_suitable(sops->sk) &&
619 		   sock_map_op_okay(sops)))
620 		return sock_map_update_common(map, *(u32 *)key, sops->sk,
621 					      flags);
622 	return -EOPNOTSUPP;
623 }
624 
625 const struct bpf_func_proto bpf_sock_map_update_proto = {
626 	.func		= bpf_sock_map_update,
627 	.gpl_only	= false,
628 	.pkt_access	= true,
629 	.ret_type	= RET_INTEGER,
630 	.arg1_type	= ARG_PTR_TO_CTX,
631 	.arg2_type	= ARG_CONST_MAP_PTR,
632 	.arg3_type	= ARG_PTR_TO_MAP_KEY,
633 	.arg4_type	= ARG_ANYTHING,
634 };
635 
636 BPF_CALL_4(bpf_sk_redirect_map, struct sk_buff *, skb,
637 	   struct bpf_map *, map, u32, key, u64, flags)
638 {
639 	struct sock *sk;
640 
641 	if (unlikely(flags & ~(BPF_F_INGRESS)))
642 		return SK_DROP;
643 
644 	sk = __sock_map_lookup_elem(map, key);
645 	if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
646 		return SK_DROP;
647 
648 	skb_bpf_set_redir(skb, sk, flags & BPF_F_INGRESS);
649 	return SK_PASS;
650 }
651 
652 const struct bpf_func_proto bpf_sk_redirect_map_proto = {
653 	.func           = bpf_sk_redirect_map,
654 	.gpl_only       = false,
655 	.ret_type       = RET_INTEGER,
656 	.arg1_type	= ARG_PTR_TO_CTX,
657 	.arg2_type      = ARG_CONST_MAP_PTR,
658 	.arg3_type      = ARG_ANYTHING,
659 	.arg4_type      = ARG_ANYTHING,
660 };
661 
662 BPF_CALL_4(bpf_msg_redirect_map, struct sk_msg *, msg,
663 	   struct bpf_map *, map, u32, key, u64, flags)
664 {
665 	struct sock *sk;
666 
667 	if (unlikely(flags & ~(BPF_F_INGRESS)))
668 		return SK_DROP;
669 
670 	sk = __sock_map_lookup_elem(map, key);
671 	if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
672 		return SK_DROP;
673 
674 	msg->flags = flags;
675 	msg->sk_redir = sk;
676 	return SK_PASS;
677 }
678 
679 const struct bpf_func_proto bpf_msg_redirect_map_proto = {
680 	.func           = bpf_msg_redirect_map,
681 	.gpl_only       = false,
682 	.ret_type       = RET_INTEGER,
683 	.arg1_type	= ARG_PTR_TO_CTX,
684 	.arg2_type      = ARG_CONST_MAP_PTR,
685 	.arg3_type      = ARG_ANYTHING,
686 	.arg4_type      = ARG_ANYTHING,
687 };
688 
689 struct sock_map_seq_info {
690 	struct bpf_map *map;
691 	struct sock *sk;
692 	u32 index;
693 };
694 
695 struct bpf_iter__sockmap {
696 	__bpf_md_ptr(struct bpf_iter_meta *, meta);
697 	__bpf_md_ptr(struct bpf_map *, map);
698 	__bpf_md_ptr(void *, key);
699 	__bpf_md_ptr(struct sock *, sk);
700 };
701 
702 DEFINE_BPF_ITER_FUNC(sockmap, struct bpf_iter_meta *meta,
703 		     struct bpf_map *map, void *key,
704 		     struct sock *sk)
705 
706 static void *sock_map_seq_lookup_elem(struct sock_map_seq_info *info)
707 {
708 	if (unlikely(info->index >= info->map->max_entries))
709 		return NULL;
710 
711 	info->sk = __sock_map_lookup_elem(info->map, info->index);
712 
713 	/* can't return sk directly, since that might be NULL */
714 	return info;
715 }
716 
717 static void *sock_map_seq_start(struct seq_file *seq, loff_t *pos)
718 	__acquires(rcu)
719 {
720 	struct sock_map_seq_info *info = seq->private;
721 
722 	if (*pos == 0)
723 		++*pos;
724 
725 	/* pairs with sock_map_seq_stop */
726 	rcu_read_lock();
727 	return sock_map_seq_lookup_elem(info);
728 }
729 
730 static void *sock_map_seq_next(struct seq_file *seq, void *v, loff_t *pos)
731 	__must_hold(rcu)
732 {
733 	struct sock_map_seq_info *info = seq->private;
734 
735 	++*pos;
736 	++info->index;
737 
738 	return sock_map_seq_lookup_elem(info);
739 }
740 
741 static int sock_map_seq_show(struct seq_file *seq, void *v)
742 	__must_hold(rcu)
743 {
744 	struct sock_map_seq_info *info = seq->private;
745 	struct bpf_iter__sockmap ctx = {};
746 	struct bpf_iter_meta meta;
747 	struct bpf_prog *prog;
748 
749 	meta.seq = seq;
750 	prog = bpf_iter_get_info(&meta, !v);
751 	if (!prog)
752 		return 0;
753 
754 	ctx.meta = &meta;
755 	ctx.map = info->map;
756 	if (v) {
757 		ctx.key = &info->index;
758 		ctx.sk = info->sk;
759 	}
760 
761 	return bpf_iter_run_prog(prog, &ctx);
762 }
763 
764 static void sock_map_seq_stop(struct seq_file *seq, void *v)
765 	__releases(rcu)
766 {
767 	if (!v)
768 		(void)sock_map_seq_show(seq, NULL);
769 
770 	/* pairs with sock_map_seq_start */
771 	rcu_read_unlock();
772 }
773 
774 static const struct seq_operations sock_map_seq_ops = {
775 	.start	= sock_map_seq_start,
776 	.next	= sock_map_seq_next,
777 	.stop	= sock_map_seq_stop,
778 	.show	= sock_map_seq_show,
779 };
780 
781 static int sock_map_init_seq_private(void *priv_data,
782 				     struct bpf_iter_aux_info *aux)
783 {
784 	struct sock_map_seq_info *info = priv_data;
785 
786 	info->map = aux->map;
787 	return 0;
788 }
789 
790 static const struct bpf_iter_seq_info sock_map_iter_seq_info = {
791 	.seq_ops		= &sock_map_seq_ops,
792 	.init_seq_private	= sock_map_init_seq_private,
793 	.seq_priv_size		= sizeof(struct sock_map_seq_info),
794 };
795 
796 BTF_ID_LIST_SINGLE(sock_map_btf_ids, struct, bpf_stab)
797 const struct bpf_map_ops sock_map_ops = {
798 	.map_meta_equal		= bpf_map_meta_equal,
799 	.map_alloc		= sock_map_alloc,
800 	.map_free		= sock_map_free,
801 	.map_get_next_key	= sock_map_get_next_key,
802 	.map_lookup_elem_sys_only = sock_map_lookup_sys,
803 	.map_update_elem	= sock_map_update_elem,
804 	.map_delete_elem	= sock_map_delete_elem,
805 	.map_lookup_elem	= sock_map_lookup,
806 	.map_release_uref	= sock_map_release_progs,
807 	.map_check_btf		= map_check_no_btf,
808 	.map_btf_id		= &sock_map_btf_ids[0],
809 	.iter_seq_info		= &sock_map_iter_seq_info,
810 };
811 
812 struct bpf_shtab_elem {
813 	struct rcu_head rcu;
814 	u32 hash;
815 	struct sock *sk;
816 	struct hlist_node node;
817 	u8 key[];
818 };
819 
820 struct bpf_shtab_bucket {
821 	struct hlist_head head;
822 	raw_spinlock_t lock;
823 };
824 
825 struct bpf_shtab {
826 	struct bpf_map map;
827 	struct bpf_shtab_bucket *buckets;
828 	u32 buckets_num;
829 	u32 elem_size;
830 	struct sk_psock_progs progs;
831 	atomic_t count;
832 };
833 
834 static inline u32 sock_hash_bucket_hash(const void *key, u32 len)
835 {
836 	return jhash(key, len, 0);
837 }
838 
839 static struct bpf_shtab_bucket *sock_hash_select_bucket(struct bpf_shtab *htab,
840 							u32 hash)
841 {
842 	return &htab->buckets[hash & (htab->buckets_num - 1)];
843 }
844 
845 static struct bpf_shtab_elem *
846 sock_hash_lookup_elem_raw(struct hlist_head *head, u32 hash, void *key,
847 			  u32 key_size)
848 {
849 	struct bpf_shtab_elem *elem;
850 
851 	hlist_for_each_entry_rcu(elem, head, node) {
852 		if (elem->hash == hash &&
853 		    !memcmp(&elem->key, key, key_size))
854 			return elem;
855 	}
856 
857 	return NULL;
858 }
859 
860 static struct sock *__sock_hash_lookup_elem(struct bpf_map *map, void *key)
861 {
862 	struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
863 	u32 key_size = map->key_size, hash;
864 	struct bpf_shtab_bucket *bucket;
865 	struct bpf_shtab_elem *elem;
866 
867 	WARN_ON_ONCE(!rcu_read_lock_held());
868 
869 	hash = sock_hash_bucket_hash(key, key_size);
870 	bucket = sock_hash_select_bucket(htab, hash);
871 	elem = sock_hash_lookup_elem_raw(&bucket->head, hash, key, key_size);
872 
873 	return elem ? elem->sk : NULL;
874 }
875 
876 static void sock_hash_free_elem(struct bpf_shtab *htab,
877 				struct bpf_shtab_elem *elem)
878 {
879 	atomic_dec(&htab->count);
880 	kfree_rcu(elem, rcu);
881 }
882 
883 static void sock_hash_delete_from_link(struct bpf_map *map, struct sock *sk,
884 				       void *link_raw)
885 {
886 	struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
887 	struct bpf_shtab_elem *elem_probe, *elem = link_raw;
888 	struct bpf_shtab_bucket *bucket;
889 
890 	WARN_ON_ONCE(!rcu_read_lock_held());
891 	bucket = sock_hash_select_bucket(htab, elem->hash);
892 
893 	/* elem may be deleted in parallel from the map, but access here
894 	 * is okay since it's going away only after RCU grace period.
895 	 * However, we need to check whether it's still present.
896 	 */
897 	raw_spin_lock_bh(&bucket->lock);
898 	elem_probe = sock_hash_lookup_elem_raw(&bucket->head, elem->hash,
899 					       elem->key, map->key_size);
900 	if (elem_probe && elem_probe == elem) {
901 		hlist_del_rcu(&elem->node);
902 		sock_map_unref(elem->sk, elem);
903 		sock_hash_free_elem(htab, elem);
904 	}
905 	raw_spin_unlock_bh(&bucket->lock);
906 }
907 
908 static int sock_hash_delete_elem(struct bpf_map *map, void *key)
909 {
910 	struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
911 	u32 hash, key_size = map->key_size;
912 	struct bpf_shtab_bucket *bucket;
913 	struct bpf_shtab_elem *elem;
914 	int ret = -ENOENT;
915 
916 	hash = sock_hash_bucket_hash(key, key_size);
917 	bucket = sock_hash_select_bucket(htab, hash);
918 
919 	raw_spin_lock_bh(&bucket->lock);
920 	elem = sock_hash_lookup_elem_raw(&bucket->head, hash, key, key_size);
921 	if (elem) {
922 		hlist_del_rcu(&elem->node);
923 		sock_map_unref(elem->sk, elem);
924 		sock_hash_free_elem(htab, elem);
925 		ret = 0;
926 	}
927 	raw_spin_unlock_bh(&bucket->lock);
928 	return ret;
929 }
930 
931 static struct bpf_shtab_elem *sock_hash_alloc_elem(struct bpf_shtab *htab,
932 						   void *key, u32 key_size,
933 						   u32 hash, struct sock *sk,
934 						   struct bpf_shtab_elem *old)
935 {
936 	struct bpf_shtab_elem *new;
937 
938 	if (atomic_inc_return(&htab->count) > htab->map.max_entries) {
939 		if (!old) {
940 			atomic_dec(&htab->count);
941 			return ERR_PTR(-E2BIG);
942 		}
943 	}
944 
945 	new = bpf_map_kmalloc_node(&htab->map, htab->elem_size,
946 				   GFP_ATOMIC | __GFP_NOWARN,
947 				   htab->map.numa_node);
948 	if (!new) {
949 		atomic_dec(&htab->count);
950 		return ERR_PTR(-ENOMEM);
951 	}
952 	memcpy(new->key, key, key_size);
953 	new->sk = sk;
954 	new->hash = hash;
955 	return new;
956 }
957 
958 static int sock_hash_update_common(struct bpf_map *map, void *key,
959 				   struct sock *sk, u64 flags)
960 {
961 	struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
962 	u32 key_size = map->key_size, hash;
963 	struct bpf_shtab_elem *elem, *elem_new;
964 	struct bpf_shtab_bucket *bucket;
965 	struct sk_psock_link *link;
966 	struct sk_psock *psock;
967 	int ret;
968 
969 	WARN_ON_ONCE(!rcu_read_lock_held());
970 	if (unlikely(flags > BPF_EXIST))
971 		return -EINVAL;
972 
973 	link = sk_psock_init_link();
974 	if (!link)
975 		return -ENOMEM;
976 
977 	ret = sock_map_link(map, sk);
978 	if (ret < 0)
979 		goto out_free;
980 
981 	psock = sk_psock(sk);
982 	WARN_ON_ONCE(!psock);
983 
984 	hash = sock_hash_bucket_hash(key, key_size);
985 	bucket = sock_hash_select_bucket(htab, hash);
986 
987 	raw_spin_lock_bh(&bucket->lock);
988 	elem = sock_hash_lookup_elem_raw(&bucket->head, hash, key, key_size);
989 	if (elem && flags == BPF_NOEXIST) {
990 		ret = -EEXIST;
991 		goto out_unlock;
992 	} else if (!elem && flags == BPF_EXIST) {
993 		ret = -ENOENT;
994 		goto out_unlock;
995 	}
996 
997 	elem_new = sock_hash_alloc_elem(htab, key, key_size, hash, sk, elem);
998 	if (IS_ERR(elem_new)) {
999 		ret = PTR_ERR(elem_new);
1000 		goto out_unlock;
1001 	}
1002 
1003 	sock_map_add_link(psock, link, map, elem_new);
1004 	/* Add new element to the head of the list, so that
1005 	 * concurrent search will find it before old elem.
1006 	 */
1007 	hlist_add_head_rcu(&elem_new->node, &bucket->head);
1008 	if (elem) {
1009 		hlist_del_rcu(&elem->node);
1010 		sock_map_unref(elem->sk, elem);
1011 		sock_hash_free_elem(htab, elem);
1012 	}
1013 	raw_spin_unlock_bh(&bucket->lock);
1014 	return 0;
1015 out_unlock:
1016 	raw_spin_unlock_bh(&bucket->lock);
1017 	sk_psock_put(sk, psock);
1018 out_free:
1019 	sk_psock_free_link(link);
1020 	return ret;
1021 }
1022 
1023 static int sock_hash_get_next_key(struct bpf_map *map, void *key,
1024 				  void *key_next)
1025 {
1026 	struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
1027 	struct bpf_shtab_elem *elem, *elem_next;
1028 	u32 hash, key_size = map->key_size;
1029 	struct hlist_head *head;
1030 	int i = 0;
1031 
1032 	if (!key)
1033 		goto find_first_elem;
1034 	hash = sock_hash_bucket_hash(key, key_size);
1035 	head = &sock_hash_select_bucket(htab, hash)->head;
1036 	elem = sock_hash_lookup_elem_raw(head, hash, key, key_size);
1037 	if (!elem)
1038 		goto find_first_elem;
1039 
1040 	elem_next = hlist_entry_safe(rcu_dereference(hlist_next_rcu(&elem->node)),
1041 				     struct bpf_shtab_elem, node);
1042 	if (elem_next) {
1043 		memcpy(key_next, elem_next->key, key_size);
1044 		return 0;
1045 	}
1046 
1047 	i = hash & (htab->buckets_num - 1);
1048 	i++;
1049 find_first_elem:
1050 	for (; i < htab->buckets_num; i++) {
1051 		head = &sock_hash_select_bucket(htab, i)->head;
1052 		elem_next = hlist_entry_safe(rcu_dereference(hlist_first_rcu(head)),
1053 					     struct bpf_shtab_elem, node);
1054 		if (elem_next) {
1055 			memcpy(key_next, elem_next->key, key_size);
1056 			return 0;
1057 		}
1058 	}
1059 
1060 	return -ENOENT;
1061 }
1062 
1063 static struct bpf_map *sock_hash_alloc(union bpf_attr *attr)
1064 {
1065 	struct bpf_shtab *htab;
1066 	int i, err;
1067 
1068 	if (!capable(CAP_NET_ADMIN))
1069 		return ERR_PTR(-EPERM);
1070 	if (attr->max_entries == 0 ||
1071 	    attr->key_size    == 0 ||
1072 	    (attr->value_size != sizeof(u32) &&
1073 	     attr->value_size != sizeof(u64)) ||
1074 	    attr->map_flags & ~SOCK_CREATE_FLAG_MASK)
1075 		return ERR_PTR(-EINVAL);
1076 	if (attr->key_size > MAX_BPF_STACK)
1077 		return ERR_PTR(-E2BIG);
1078 
1079 	htab = kzalloc(sizeof(*htab), GFP_USER | __GFP_ACCOUNT);
1080 	if (!htab)
1081 		return ERR_PTR(-ENOMEM);
1082 
1083 	bpf_map_init_from_attr(&htab->map, attr);
1084 
1085 	htab->buckets_num = roundup_pow_of_two(htab->map.max_entries);
1086 	htab->elem_size = sizeof(struct bpf_shtab_elem) +
1087 			  round_up(htab->map.key_size, 8);
1088 	if (htab->buckets_num == 0 ||
1089 	    htab->buckets_num > U32_MAX / sizeof(struct bpf_shtab_bucket)) {
1090 		err = -EINVAL;
1091 		goto free_htab;
1092 	}
1093 
1094 	htab->buckets = bpf_map_area_alloc(htab->buckets_num *
1095 					   sizeof(struct bpf_shtab_bucket),
1096 					   htab->map.numa_node);
1097 	if (!htab->buckets) {
1098 		err = -ENOMEM;
1099 		goto free_htab;
1100 	}
1101 
1102 	for (i = 0; i < htab->buckets_num; i++) {
1103 		INIT_HLIST_HEAD(&htab->buckets[i].head);
1104 		raw_spin_lock_init(&htab->buckets[i].lock);
1105 	}
1106 
1107 	return &htab->map;
1108 free_htab:
1109 	kfree(htab);
1110 	return ERR_PTR(err);
1111 }
1112 
1113 static void sock_hash_free(struct bpf_map *map)
1114 {
1115 	struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
1116 	struct bpf_shtab_bucket *bucket;
1117 	struct hlist_head unlink_list;
1118 	struct bpf_shtab_elem *elem;
1119 	struct hlist_node *node;
1120 	int i;
1121 
1122 	/* After the sync no updates or deletes will be in-flight so it
1123 	 * is safe to walk map and remove entries without risking a race
1124 	 * in EEXIST update case.
1125 	 */
1126 	synchronize_rcu();
1127 	for (i = 0; i < htab->buckets_num; i++) {
1128 		bucket = sock_hash_select_bucket(htab, i);
1129 
1130 		/* We are racing with sock_hash_delete_from_link to
1131 		 * enter the spin-lock critical section. Every socket on
1132 		 * the list is still linked to sockhash. Since link
1133 		 * exists, psock exists and holds a ref to socket. That
1134 		 * lets us to grab a socket ref too.
1135 		 */
1136 		raw_spin_lock_bh(&bucket->lock);
1137 		hlist_for_each_entry(elem, &bucket->head, node)
1138 			sock_hold(elem->sk);
1139 		hlist_move_list(&bucket->head, &unlink_list);
1140 		raw_spin_unlock_bh(&bucket->lock);
1141 
1142 		/* Process removed entries out of atomic context to
1143 		 * block for socket lock before deleting the psock's
1144 		 * link to sockhash.
1145 		 */
1146 		hlist_for_each_entry_safe(elem, node, &unlink_list, node) {
1147 			hlist_del(&elem->node);
1148 			lock_sock(elem->sk);
1149 			rcu_read_lock();
1150 			sock_map_unref(elem->sk, elem);
1151 			rcu_read_unlock();
1152 			release_sock(elem->sk);
1153 			sock_put(elem->sk);
1154 			sock_hash_free_elem(htab, elem);
1155 		}
1156 	}
1157 
1158 	/* wait for psock readers accessing its map link */
1159 	synchronize_rcu();
1160 
1161 	bpf_map_area_free(htab->buckets);
1162 	kfree(htab);
1163 }
1164 
1165 static void *sock_hash_lookup_sys(struct bpf_map *map, void *key)
1166 {
1167 	struct sock *sk;
1168 
1169 	if (map->value_size != sizeof(u64))
1170 		return ERR_PTR(-ENOSPC);
1171 
1172 	sk = __sock_hash_lookup_elem(map, key);
1173 	if (!sk)
1174 		return ERR_PTR(-ENOENT);
1175 
1176 	__sock_gen_cookie(sk);
1177 	return &sk->sk_cookie;
1178 }
1179 
1180 static void *sock_hash_lookup(struct bpf_map *map, void *key)
1181 {
1182 	struct sock *sk;
1183 
1184 	sk = __sock_hash_lookup_elem(map, key);
1185 	if (!sk)
1186 		return NULL;
1187 	if (sk_is_refcounted(sk) && !refcount_inc_not_zero(&sk->sk_refcnt))
1188 		return NULL;
1189 	return sk;
1190 }
1191 
1192 static void sock_hash_release_progs(struct bpf_map *map)
1193 {
1194 	psock_progs_drop(&container_of(map, struct bpf_shtab, map)->progs);
1195 }
1196 
1197 BPF_CALL_4(bpf_sock_hash_update, struct bpf_sock_ops_kern *, sops,
1198 	   struct bpf_map *, map, void *, key, u64, flags)
1199 {
1200 	WARN_ON_ONCE(!rcu_read_lock_held());
1201 
1202 	if (likely(sock_map_sk_is_suitable(sops->sk) &&
1203 		   sock_map_op_okay(sops)))
1204 		return sock_hash_update_common(map, key, sops->sk, flags);
1205 	return -EOPNOTSUPP;
1206 }
1207 
1208 const struct bpf_func_proto bpf_sock_hash_update_proto = {
1209 	.func		= bpf_sock_hash_update,
1210 	.gpl_only	= false,
1211 	.pkt_access	= true,
1212 	.ret_type	= RET_INTEGER,
1213 	.arg1_type	= ARG_PTR_TO_CTX,
1214 	.arg2_type	= ARG_CONST_MAP_PTR,
1215 	.arg3_type	= ARG_PTR_TO_MAP_KEY,
1216 	.arg4_type	= ARG_ANYTHING,
1217 };
1218 
1219 BPF_CALL_4(bpf_sk_redirect_hash, struct sk_buff *, skb,
1220 	   struct bpf_map *, map, void *, key, u64, flags)
1221 {
1222 	struct sock *sk;
1223 
1224 	if (unlikely(flags & ~(BPF_F_INGRESS)))
1225 		return SK_DROP;
1226 
1227 	sk = __sock_hash_lookup_elem(map, key);
1228 	if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
1229 		return SK_DROP;
1230 
1231 	skb_bpf_set_redir(skb, sk, flags & BPF_F_INGRESS);
1232 	return SK_PASS;
1233 }
1234 
1235 const struct bpf_func_proto bpf_sk_redirect_hash_proto = {
1236 	.func           = bpf_sk_redirect_hash,
1237 	.gpl_only       = false,
1238 	.ret_type       = RET_INTEGER,
1239 	.arg1_type	= ARG_PTR_TO_CTX,
1240 	.arg2_type      = ARG_CONST_MAP_PTR,
1241 	.arg3_type      = ARG_PTR_TO_MAP_KEY,
1242 	.arg4_type      = ARG_ANYTHING,
1243 };
1244 
1245 BPF_CALL_4(bpf_msg_redirect_hash, struct sk_msg *, msg,
1246 	   struct bpf_map *, map, void *, key, u64, flags)
1247 {
1248 	struct sock *sk;
1249 
1250 	if (unlikely(flags & ~(BPF_F_INGRESS)))
1251 		return SK_DROP;
1252 
1253 	sk = __sock_hash_lookup_elem(map, key);
1254 	if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
1255 		return SK_DROP;
1256 
1257 	msg->flags = flags;
1258 	msg->sk_redir = sk;
1259 	return SK_PASS;
1260 }
1261 
1262 const struct bpf_func_proto bpf_msg_redirect_hash_proto = {
1263 	.func           = bpf_msg_redirect_hash,
1264 	.gpl_only       = false,
1265 	.ret_type       = RET_INTEGER,
1266 	.arg1_type	= ARG_PTR_TO_CTX,
1267 	.arg2_type      = ARG_CONST_MAP_PTR,
1268 	.arg3_type      = ARG_PTR_TO_MAP_KEY,
1269 	.arg4_type      = ARG_ANYTHING,
1270 };
1271 
1272 struct sock_hash_seq_info {
1273 	struct bpf_map *map;
1274 	struct bpf_shtab *htab;
1275 	u32 bucket_id;
1276 };
1277 
1278 static void *sock_hash_seq_find_next(struct sock_hash_seq_info *info,
1279 				     struct bpf_shtab_elem *prev_elem)
1280 {
1281 	const struct bpf_shtab *htab = info->htab;
1282 	struct bpf_shtab_bucket *bucket;
1283 	struct bpf_shtab_elem *elem;
1284 	struct hlist_node *node;
1285 
1286 	/* try to find next elem in the same bucket */
1287 	if (prev_elem) {
1288 		node = rcu_dereference(hlist_next_rcu(&prev_elem->node));
1289 		elem = hlist_entry_safe(node, struct bpf_shtab_elem, node);
1290 		if (elem)
1291 			return elem;
1292 
1293 		/* no more elements, continue in the next bucket */
1294 		info->bucket_id++;
1295 	}
1296 
1297 	for (; info->bucket_id < htab->buckets_num; info->bucket_id++) {
1298 		bucket = &htab->buckets[info->bucket_id];
1299 		node = rcu_dereference(hlist_first_rcu(&bucket->head));
1300 		elem = hlist_entry_safe(node, struct bpf_shtab_elem, node);
1301 		if (elem)
1302 			return elem;
1303 	}
1304 
1305 	return NULL;
1306 }
1307 
1308 static void *sock_hash_seq_start(struct seq_file *seq, loff_t *pos)
1309 	__acquires(rcu)
1310 {
1311 	struct sock_hash_seq_info *info = seq->private;
1312 
1313 	if (*pos == 0)
1314 		++*pos;
1315 
1316 	/* pairs with sock_hash_seq_stop */
1317 	rcu_read_lock();
1318 	return sock_hash_seq_find_next(info, NULL);
1319 }
1320 
1321 static void *sock_hash_seq_next(struct seq_file *seq, void *v, loff_t *pos)
1322 	__must_hold(rcu)
1323 {
1324 	struct sock_hash_seq_info *info = seq->private;
1325 
1326 	++*pos;
1327 	return sock_hash_seq_find_next(info, v);
1328 }
1329 
1330 static int sock_hash_seq_show(struct seq_file *seq, void *v)
1331 	__must_hold(rcu)
1332 {
1333 	struct sock_hash_seq_info *info = seq->private;
1334 	struct bpf_iter__sockmap ctx = {};
1335 	struct bpf_shtab_elem *elem = v;
1336 	struct bpf_iter_meta meta;
1337 	struct bpf_prog *prog;
1338 
1339 	meta.seq = seq;
1340 	prog = bpf_iter_get_info(&meta, !elem);
1341 	if (!prog)
1342 		return 0;
1343 
1344 	ctx.meta = &meta;
1345 	ctx.map = info->map;
1346 	if (elem) {
1347 		ctx.key = elem->key;
1348 		ctx.sk = elem->sk;
1349 	}
1350 
1351 	return bpf_iter_run_prog(prog, &ctx);
1352 }
1353 
1354 static void sock_hash_seq_stop(struct seq_file *seq, void *v)
1355 	__releases(rcu)
1356 {
1357 	if (!v)
1358 		(void)sock_hash_seq_show(seq, NULL);
1359 
1360 	/* pairs with sock_hash_seq_start */
1361 	rcu_read_unlock();
1362 }
1363 
1364 static const struct seq_operations sock_hash_seq_ops = {
1365 	.start	= sock_hash_seq_start,
1366 	.next	= sock_hash_seq_next,
1367 	.stop	= sock_hash_seq_stop,
1368 	.show	= sock_hash_seq_show,
1369 };
1370 
1371 static int sock_hash_init_seq_private(void *priv_data,
1372 				     struct bpf_iter_aux_info *aux)
1373 {
1374 	struct sock_hash_seq_info *info = priv_data;
1375 
1376 	info->map = aux->map;
1377 	info->htab = container_of(aux->map, struct bpf_shtab, map);
1378 	return 0;
1379 }
1380 
1381 static const struct bpf_iter_seq_info sock_hash_iter_seq_info = {
1382 	.seq_ops		= &sock_hash_seq_ops,
1383 	.init_seq_private	= sock_hash_init_seq_private,
1384 	.seq_priv_size		= sizeof(struct sock_hash_seq_info),
1385 };
1386 
1387 BTF_ID_LIST_SINGLE(sock_hash_map_btf_ids, struct, bpf_shtab)
1388 const struct bpf_map_ops sock_hash_ops = {
1389 	.map_meta_equal		= bpf_map_meta_equal,
1390 	.map_alloc		= sock_hash_alloc,
1391 	.map_free		= sock_hash_free,
1392 	.map_get_next_key	= sock_hash_get_next_key,
1393 	.map_update_elem	= sock_map_update_elem,
1394 	.map_delete_elem	= sock_hash_delete_elem,
1395 	.map_lookup_elem	= sock_hash_lookup,
1396 	.map_lookup_elem_sys_only = sock_hash_lookup_sys,
1397 	.map_release_uref	= sock_hash_release_progs,
1398 	.map_check_btf		= map_check_no_btf,
1399 	.map_btf_id		= &sock_hash_map_btf_ids[0],
1400 	.iter_seq_info		= &sock_hash_iter_seq_info,
1401 };
1402 
1403 static struct sk_psock_progs *sock_map_progs(struct bpf_map *map)
1404 {
1405 	switch (map->map_type) {
1406 	case BPF_MAP_TYPE_SOCKMAP:
1407 		return &container_of(map, struct bpf_stab, map)->progs;
1408 	case BPF_MAP_TYPE_SOCKHASH:
1409 		return &container_of(map, struct bpf_shtab, map)->progs;
1410 	default:
1411 		break;
1412 	}
1413 
1414 	return NULL;
1415 }
1416 
1417 static int sock_map_prog_lookup(struct bpf_map *map, struct bpf_prog ***pprog,
1418 				u32 which)
1419 {
1420 	struct sk_psock_progs *progs = sock_map_progs(map);
1421 
1422 	if (!progs)
1423 		return -EOPNOTSUPP;
1424 
1425 	switch (which) {
1426 	case BPF_SK_MSG_VERDICT:
1427 		*pprog = &progs->msg_parser;
1428 		break;
1429 #if IS_ENABLED(CONFIG_BPF_STREAM_PARSER)
1430 	case BPF_SK_SKB_STREAM_PARSER:
1431 		*pprog = &progs->stream_parser;
1432 		break;
1433 #endif
1434 	case BPF_SK_SKB_STREAM_VERDICT:
1435 		if (progs->skb_verdict)
1436 			return -EBUSY;
1437 		*pprog = &progs->stream_verdict;
1438 		break;
1439 	case BPF_SK_SKB_VERDICT:
1440 		if (progs->stream_verdict)
1441 			return -EBUSY;
1442 		*pprog = &progs->skb_verdict;
1443 		break;
1444 	default:
1445 		return -EOPNOTSUPP;
1446 	}
1447 
1448 	return 0;
1449 }
1450 
1451 static int sock_map_prog_update(struct bpf_map *map, struct bpf_prog *prog,
1452 				struct bpf_prog *old, u32 which)
1453 {
1454 	struct bpf_prog **pprog;
1455 	int ret;
1456 
1457 	ret = sock_map_prog_lookup(map, &pprog, which);
1458 	if (ret)
1459 		return ret;
1460 
1461 	if (old)
1462 		return psock_replace_prog(pprog, prog, old);
1463 
1464 	psock_set_prog(pprog, prog);
1465 	return 0;
1466 }
1467 
1468 int sock_map_bpf_prog_query(const union bpf_attr *attr,
1469 			    union bpf_attr __user *uattr)
1470 {
1471 	__u32 __user *prog_ids = u64_to_user_ptr(attr->query.prog_ids);
1472 	u32 prog_cnt = 0, flags = 0, ufd = attr->target_fd;
1473 	struct bpf_prog **pprog;
1474 	struct bpf_prog *prog;
1475 	struct bpf_map *map;
1476 	struct fd f;
1477 	u32 id = 0;
1478 	int ret;
1479 
1480 	if (attr->query.query_flags)
1481 		return -EINVAL;
1482 
1483 	f = fdget(ufd);
1484 	map = __bpf_map_get(f);
1485 	if (IS_ERR(map))
1486 		return PTR_ERR(map);
1487 
1488 	rcu_read_lock();
1489 
1490 	ret = sock_map_prog_lookup(map, &pprog, attr->query.attach_type);
1491 	if (ret)
1492 		goto end;
1493 
1494 	prog = *pprog;
1495 	prog_cnt = !prog ? 0 : 1;
1496 
1497 	if (!attr->query.prog_cnt || !prog_ids || !prog_cnt)
1498 		goto end;
1499 
1500 	/* we do not hold the refcnt, the bpf prog may be released
1501 	 * asynchronously and the id would be set to 0.
1502 	 */
1503 	id = data_race(prog->aux->id);
1504 	if (id == 0)
1505 		prog_cnt = 0;
1506 
1507 end:
1508 	rcu_read_unlock();
1509 
1510 	if (copy_to_user(&uattr->query.attach_flags, &flags, sizeof(flags)) ||
1511 	    (id != 0 && copy_to_user(prog_ids, &id, sizeof(u32))) ||
1512 	    copy_to_user(&uattr->query.prog_cnt, &prog_cnt, sizeof(prog_cnt)))
1513 		ret = -EFAULT;
1514 
1515 	fdput(f);
1516 	return ret;
1517 }
1518 
1519 static void sock_map_unlink(struct sock *sk, struct sk_psock_link *link)
1520 {
1521 	switch (link->map->map_type) {
1522 	case BPF_MAP_TYPE_SOCKMAP:
1523 		return sock_map_delete_from_link(link->map, sk,
1524 						 link->link_raw);
1525 	case BPF_MAP_TYPE_SOCKHASH:
1526 		return sock_hash_delete_from_link(link->map, sk,
1527 						  link->link_raw);
1528 	default:
1529 		break;
1530 	}
1531 }
1532 
1533 static void sock_map_remove_links(struct sock *sk, struct sk_psock *psock)
1534 {
1535 	struct sk_psock_link *link;
1536 
1537 	while ((link = sk_psock_link_pop(psock))) {
1538 		sock_map_unlink(sk, link);
1539 		sk_psock_free_link(link);
1540 	}
1541 }
1542 
1543 void sock_map_unhash(struct sock *sk)
1544 {
1545 	void (*saved_unhash)(struct sock *sk);
1546 	struct sk_psock *psock;
1547 
1548 	rcu_read_lock();
1549 	psock = sk_psock(sk);
1550 	if (unlikely(!psock)) {
1551 		rcu_read_unlock();
1552 		if (sk->sk_prot->unhash)
1553 			sk->sk_prot->unhash(sk);
1554 		return;
1555 	}
1556 
1557 	saved_unhash = psock->saved_unhash;
1558 	sock_map_remove_links(sk, psock);
1559 	rcu_read_unlock();
1560 	saved_unhash(sk);
1561 }
1562 EXPORT_SYMBOL_GPL(sock_map_unhash);
1563 
1564 void sock_map_destroy(struct sock *sk)
1565 {
1566 	void (*saved_destroy)(struct sock *sk);
1567 	struct sk_psock *psock;
1568 
1569 	rcu_read_lock();
1570 	psock = sk_psock_get(sk);
1571 	if (unlikely(!psock)) {
1572 		rcu_read_unlock();
1573 		if (sk->sk_prot->destroy)
1574 			sk->sk_prot->destroy(sk);
1575 		return;
1576 	}
1577 
1578 	saved_destroy = psock->saved_destroy;
1579 	sock_map_remove_links(sk, psock);
1580 	rcu_read_unlock();
1581 	sk_psock_stop(psock, false);
1582 	sk_psock_put(sk, psock);
1583 	saved_destroy(sk);
1584 }
1585 EXPORT_SYMBOL_GPL(sock_map_destroy);
1586 
1587 void sock_map_close(struct sock *sk, long timeout)
1588 {
1589 	void (*saved_close)(struct sock *sk, long timeout);
1590 	struct sk_psock *psock;
1591 
1592 	lock_sock(sk);
1593 	rcu_read_lock();
1594 	psock = sk_psock_get(sk);
1595 	if (unlikely(!psock)) {
1596 		rcu_read_unlock();
1597 		release_sock(sk);
1598 		return sk->sk_prot->close(sk, timeout);
1599 	}
1600 
1601 	saved_close = psock->saved_close;
1602 	sock_map_remove_links(sk, psock);
1603 	rcu_read_unlock();
1604 	sk_psock_stop(psock, true);
1605 	sk_psock_put(sk, psock);
1606 	release_sock(sk);
1607 	saved_close(sk, timeout);
1608 }
1609 EXPORT_SYMBOL_GPL(sock_map_close);
1610 
1611 static int sock_map_iter_attach_target(struct bpf_prog *prog,
1612 				       union bpf_iter_link_info *linfo,
1613 				       struct bpf_iter_aux_info *aux)
1614 {
1615 	struct bpf_map *map;
1616 	int err = -EINVAL;
1617 
1618 	if (!linfo->map.map_fd)
1619 		return -EBADF;
1620 
1621 	map = bpf_map_get_with_uref(linfo->map.map_fd);
1622 	if (IS_ERR(map))
1623 		return PTR_ERR(map);
1624 
1625 	if (map->map_type != BPF_MAP_TYPE_SOCKMAP &&
1626 	    map->map_type != BPF_MAP_TYPE_SOCKHASH)
1627 		goto put_map;
1628 
1629 	if (prog->aux->max_rdonly_access > map->key_size) {
1630 		err = -EACCES;
1631 		goto put_map;
1632 	}
1633 
1634 	aux->map = map;
1635 	return 0;
1636 
1637 put_map:
1638 	bpf_map_put_with_uref(map);
1639 	return err;
1640 }
1641 
1642 static void sock_map_iter_detach_target(struct bpf_iter_aux_info *aux)
1643 {
1644 	bpf_map_put_with_uref(aux->map);
1645 }
1646 
1647 static struct bpf_iter_reg sock_map_iter_reg = {
1648 	.target			= "sockmap",
1649 	.attach_target		= sock_map_iter_attach_target,
1650 	.detach_target		= sock_map_iter_detach_target,
1651 	.show_fdinfo		= bpf_iter_map_show_fdinfo,
1652 	.fill_link_info		= bpf_iter_map_fill_link_info,
1653 	.ctx_arg_info_size	= 2,
1654 	.ctx_arg_info		= {
1655 		{ offsetof(struct bpf_iter__sockmap, key),
1656 		  PTR_TO_BUF | PTR_MAYBE_NULL | MEM_RDONLY },
1657 		{ offsetof(struct bpf_iter__sockmap, sk),
1658 		  PTR_TO_BTF_ID_OR_NULL },
1659 	},
1660 };
1661 
1662 static int __init bpf_sockmap_iter_init(void)
1663 {
1664 	sock_map_iter_reg.ctx_arg_info[1].btf_id =
1665 		btf_sock_ids[BTF_SOCK_TYPE_SOCK];
1666 	return bpf_iter_reg_target(&sock_map_iter_reg);
1667 }
1668 late_initcall(bpf_sockmap_iter_init);
1669