xref: /openbmc/linux/net/tls/tls_device.c (revision 2cc39179)
1 /* Copyright (c) 2018, Mellanox Technologies All rights reserved.
2  *
3  * This software is available to you under a choice of one of two
4  * licenses.  You may choose to be licensed under the terms of the GNU
5  * General Public License (GPL) Version 2, available from the file
6  * COPYING in the main directory of this source tree, or the
7  * OpenIB.org BSD license below:
8  *
9  *     Redistribution and use in source and binary forms, with or
10  *     without modification, are permitted provided that the following
11  *     conditions are met:
12  *
13  *      - Redistributions of source code must retain the above
14  *        copyright notice, this list of conditions and the following
15  *        disclaimer.
16  *
17  *      - Redistributions in binary form must reproduce the above
18  *        copyright notice, this list of conditions and the following
19  *        disclaimer in the documentation and/or other materials
20  *        provided with the distribution.
21  *
22  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
23  * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
24  * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
25  * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
26  * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
27  * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
28  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
29  * SOFTWARE.
30  */
31 
32 #include <crypto/aead.h>
33 #include <linux/highmem.h>
34 #include <linux/module.h>
35 #include <linux/netdevice.h>
36 #include <net/dst.h>
37 #include <net/inet_connection_sock.h>
38 #include <net/tcp.h>
39 #include <net/tls.h>
40 
41 #include "trace.h"
42 
43 /* device_offload_lock is used to synchronize tls_dev_add
44  * against NETDEV_DOWN notifications.
45  */
46 static DECLARE_RWSEM(device_offload_lock);
47 
48 static void tls_device_gc_task(struct work_struct *work);
49 
50 static DECLARE_WORK(tls_device_gc_work, tls_device_gc_task);
51 static LIST_HEAD(tls_device_gc_list);
52 static LIST_HEAD(tls_device_list);
53 static LIST_HEAD(tls_device_down_list);
54 static DEFINE_SPINLOCK(tls_device_lock);
55 
56 static void tls_device_free_ctx(struct tls_context *ctx)
57 {
58 	if (ctx->tx_conf == TLS_HW) {
59 		kfree(tls_offload_ctx_tx(ctx));
60 		kfree(ctx->tx.rec_seq);
61 		kfree(ctx->tx.iv);
62 	}
63 
64 	if (ctx->rx_conf == TLS_HW)
65 		kfree(tls_offload_ctx_rx(ctx));
66 
67 	tls_ctx_free(NULL, ctx);
68 }
69 
70 static void tls_device_gc_task(struct work_struct *work)
71 {
72 	struct tls_context *ctx, *tmp;
73 	unsigned long flags;
74 	LIST_HEAD(gc_list);
75 
76 	spin_lock_irqsave(&tls_device_lock, flags);
77 	list_splice_init(&tls_device_gc_list, &gc_list);
78 	spin_unlock_irqrestore(&tls_device_lock, flags);
79 
80 	list_for_each_entry_safe(ctx, tmp, &gc_list, list) {
81 		struct net_device *netdev = ctx->netdev;
82 
83 		if (netdev && ctx->tx_conf == TLS_HW) {
84 			netdev->tlsdev_ops->tls_dev_del(netdev, ctx,
85 							TLS_OFFLOAD_CTX_DIR_TX);
86 			dev_put(netdev);
87 			ctx->netdev = NULL;
88 		}
89 
90 		list_del(&ctx->list);
91 		tls_device_free_ctx(ctx);
92 	}
93 }
94 
95 static void tls_device_queue_ctx_destruction(struct tls_context *ctx)
96 {
97 	unsigned long flags;
98 
99 	spin_lock_irqsave(&tls_device_lock, flags);
100 	list_move_tail(&ctx->list, &tls_device_gc_list);
101 
102 	/* schedule_work inside the spinlock
103 	 * to make sure tls_device_down waits for that work.
104 	 */
105 	schedule_work(&tls_device_gc_work);
106 
107 	spin_unlock_irqrestore(&tls_device_lock, flags);
108 }
109 
110 /* We assume that the socket is already connected */
111 static struct net_device *get_netdev_for_sock(struct sock *sk)
112 {
113 	struct dst_entry *dst = sk_dst_get(sk);
114 	struct net_device *netdev = NULL;
115 
116 	if (likely(dst)) {
117 		netdev = netdev_sk_get_lowest_dev(dst->dev, sk);
118 		dev_hold(netdev);
119 	}
120 
121 	dst_release(dst);
122 
123 	return netdev;
124 }
125 
126 static void destroy_record(struct tls_record_info *record)
127 {
128 	int i;
129 
130 	for (i = 0; i < record->num_frags; i++)
131 		__skb_frag_unref(&record->frags[i], false);
132 	kfree(record);
133 }
134 
135 static void delete_all_records(struct tls_offload_context_tx *offload_ctx)
136 {
137 	struct tls_record_info *info, *temp;
138 
139 	list_for_each_entry_safe(info, temp, &offload_ctx->records_list, list) {
140 		list_del(&info->list);
141 		destroy_record(info);
142 	}
143 
144 	offload_ctx->retransmit_hint = NULL;
145 }
146 
147 static void tls_icsk_clean_acked(struct sock *sk, u32 acked_seq)
148 {
149 	struct tls_context *tls_ctx = tls_get_ctx(sk);
150 	struct tls_record_info *info, *temp;
151 	struct tls_offload_context_tx *ctx;
152 	u64 deleted_records = 0;
153 	unsigned long flags;
154 
155 	if (!tls_ctx)
156 		return;
157 
158 	ctx = tls_offload_ctx_tx(tls_ctx);
159 
160 	spin_lock_irqsave(&ctx->lock, flags);
161 	info = ctx->retransmit_hint;
162 	if (info && !before(acked_seq, info->end_seq))
163 		ctx->retransmit_hint = NULL;
164 
165 	list_for_each_entry_safe(info, temp, &ctx->records_list, list) {
166 		if (before(acked_seq, info->end_seq))
167 			break;
168 		list_del(&info->list);
169 
170 		destroy_record(info);
171 		deleted_records++;
172 	}
173 
174 	ctx->unacked_record_sn += deleted_records;
175 	spin_unlock_irqrestore(&ctx->lock, flags);
176 }
177 
178 /* At this point, there should be no references on this
179  * socket and no in-flight SKBs associated with this
180  * socket, so it is safe to free all the resources.
181  */
182 void tls_device_sk_destruct(struct sock *sk)
183 {
184 	struct tls_context *tls_ctx = tls_get_ctx(sk);
185 	struct tls_offload_context_tx *ctx = tls_offload_ctx_tx(tls_ctx);
186 
187 	tls_ctx->sk_destruct(sk);
188 
189 	if (tls_ctx->tx_conf == TLS_HW) {
190 		if (ctx->open_record)
191 			destroy_record(ctx->open_record);
192 		delete_all_records(ctx);
193 		crypto_free_aead(ctx->aead_send);
194 		clean_acked_data_disable(inet_csk(sk));
195 	}
196 
197 	if (refcount_dec_and_test(&tls_ctx->refcount))
198 		tls_device_queue_ctx_destruction(tls_ctx);
199 }
200 EXPORT_SYMBOL_GPL(tls_device_sk_destruct);
201 
202 void tls_device_free_resources_tx(struct sock *sk)
203 {
204 	struct tls_context *tls_ctx = tls_get_ctx(sk);
205 
206 	tls_free_partial_record(sk, tls_ctx);
207 }
208 
209 void tls_offload_tx_resync_request(struct sock *sk, u32 got_seq, u32 exp_seq)
210 {
211 	struct tls_context *tls_ctx = tls_get_ctx(sk);
212 
213 	trace_tls_device_tx_resync_req(sk, got_seq, exp_seq);
214 	WARN_ON(test_and_set_bit(TLS_TX_SYNC_SCHED, &tls_ctx->flags));
215 }
216 EXPORT_SYMBOL_GPL(tls_offload_tx_resync_request);
217 
218 static void tls_device_resync_tx(struct sock *sk, struct tls_context *tls_ctx,
219 				 u32 seq)
220 {
221 	struct net_device *netdev;
222 	struct sk_buff *skb;
223 	int err = 0;
224 	u8 *rcd_sn;
225 
226 	skb = tcp_write_queue_tail(sk);
227 	if (skb)
228 		TCP_SKB_CB(skb)->eor = 1;
229 
230 	rcd_sn = tls_ctx->tx.rec_seq;
231 
232 	trace_tls_device_tx_resync_send(sk, seq, rcd_sn);
233 	down_read(&device_offload_lock);
234 	netdev = tls_ctx->netdev;
235 	if (netdev)
236 		err = netdev->tlsdev_ops->tls_dev_resync(netdev, sk, seq,
237 							 rcd_sn,
238 							 TLS_OFFLOAD_CTX_DIR_TX);
239 	up_read(&device_offload_lock);
240 	if (err)
241 		return;
242 
243 	clear_bit_unlock(TLS_TX_SYNC_SCHED, &tls_ctx->flags);
244 }
245 
246 static void tls_append_frag(struct tls_record_info *record,
247 			    struct page_frag *pfrag,
248 			    int size)
249 {
250 	skb_frag_t *frag;
251 
252 	frag = &record->frags[record->num_frags - 1];
253 	if (skb_frag_page(frag) == pfrag->page &&
254 	    skb_frag_off(frag) + skb_frag_size(frag) == pfrag->offset) {
255 		skb_frag_size_add(frag, size);
256 	} else {
257 		++frag;
258 		__skb_frag_set_page(frag, pfrag->page);
259 		skb_frag_off_set(frag, pfrag->offset);
260 		skb_frag_size_set(frag, size);
261 		++record->num_frags;
262 		get_page(pfrag->page);
263 	}
264 
265 	pfrag->offset += size;
266 	record->len += size;
267 }
268 
269 static int tls_push_record(struct sock *sk,
270 			   struct tls_context *ctx,
271 			   struct tls_offload_context_tx *offload_ctx,
272 			   struct tls_record_info *record,
273 			   int flags)
274 {
275 	struct tls_prot_info *prot = &ctx->prot_info;
276 	struct tcp_sock *tp = tcp_sk(sk);
277 	skb_frag_t *frag;
278 	int i;
279 
280 	record->end_seq = tp->write_seq + record->len;
281 	list_add_tail_rcu(&record->list, &offload_ctx->records_list);
282 	offload_ctx->open_record = NULL;
283 
284 	if (test_bit(TLS_TX_SYNC_SCHED, &ctx->flags))
285 		tls_device_resync_tx(sk, ctx, tp->write_seq);
286 
287 	tls_advance_record_sn(sk, prot, &ctx->tx);
288 
289 	for (i = 0; i < record->num_frags; i++) {
290 		frag = &record->frags[i];
291 		sg_unmark_end(&offload_ctx->sg_tx_data[i]);
292 		sg_set_page(&offload_ctx->sg_tx_data[i], skb_frag_page(frag),
293 			    skb_frag_size(frag), skb_frag_off(frag));
294 		sk_mem_charge(sk, skb_frag_size(frag));
295 		get_page(skb_frag_page(frag));
296 	}
297 	sg_mark_end(&offload_ctx->sg_tx_data[record->num_frags - 1]);
298 
299 	/* all ready, send */
300 	return tls_push_sg(sk, ctx, offload_ctx->sg_tx_data, 0, flags);
301 }
302 
303 static int tls_device_record_close(struct sock *sk,
304 				   struct tls_context *ctx,
305 				   struct tls_record_info *record,
306 				   struct page_frag *pfrag,
307 				   unsigned char record_type)
308 {
309 	struct tls_prot_info *prot = &ctx->prot_info;
310 	int ret;
311 
312 	/* append tag
313 	 * device will fill in the tag, we just need to append a placeholder
314 	 * use socket memory to improve coalescing (re-using a single buffer
315 	 * increases frag count)
316 	 * if we can't allocate memory now, steal some back from data
317 	 */
318 	if (likely(skb_page_frag_refill(prot->tag_size, pfrag,
319 					sk->sk_allocation))) {
320 		ret = 0;
321 		tls_append_frag(record, pfrag, prot->tag_size);
322 	} else {
323 		ret = prot->tag_size;
324 		if (record->len <= prot->overhead_size)
325 			return -ENOMEM;
326 	}
327 
328 	/* fill prepend */
329 	tls_fill_prepend(ctx, skb_frag_address(&record->frags[0]),
330 			 record->len - prot->overhead_size,
331 			 record_type);
332 	return ret;
333 }
334 
335 static int tls_create_new_record(struct tls_offload_context_tx *offload_ctx,
336 				 struct page_frag *pfrag,
337 				 size_t prepend_size)
338 {
339 	struct tls_record_info *record;
340 	skb_frag_t *frag;
341 
342 	record = kmalloc(sizeof(*record), GFP_KERNEL);
343 	if (!record)
344 		return -ENOMEM;
345 
346 	frag = &record->frags[0];
347 	__skb_frag_set_page(frag, pfrag->page);
348 	skb_frag_off_set(frag, pfrag->offset);
349 	skb_frag_size_set(frag, prepend_size);
350 
351 	get_page(pfrag->page);
352 	pfrag->offset += prepend_size;
353 
354 	record->num_frags = 1;
355 	record->len = prepend_size;
356 	offload_ctx->open_record = record;
357 	return 0;
358 }
359 
360 static int tls_do_allocation(struct sock *sk,
361 			     struct tls_offload_context_tx *offload_ctx,
362 			     struct page_frag *pfrag,
363 			     size_t prepend_size)
364 {
365 	int ret;
366 
367 	if (!offload_ctx->open_record) {
368 		if (unlikely(!skb_page_frag_refill(prepend_size, pfrag,
369 						   sk->sk_allocation))) {
370 			READ_ONCE(sk->sk_prot)->enter_memory_pressure(sk);
371 			sk_stream_moderate_sndbuf(sk);
372 			return -ENOMEM;
373 		}
374 
375 		ret = tls_create_new_record(offload_ctx, pfrag, prepend_size);
376 		if (ret)
377 			return ret;
378 
379 		if (pfrag->size > pfrag->offset)
380 			return 0;
381 	}
382 
383 	if (!sk_page_frag_refill(sk, pfrag))
384 		return -ENOMEM;
385 
386 	return 0;
387 }
388 
389 static int tls_device_copy_data(void *addr, size_t bytes, struct iov_iter *i)
390 {
391 	size_t pre_copy, nocache;
392 
393 	pre_copy = ~((unsigned long)addr - 1) & (SMP_CACHE_BYTES - 1);
394 	if (pre_copy) {
395 		pre_copy = min(pre_copy, bytes);
396 		if (copy_from_iter(addr, pre_copy, i) != pre_copy)
397 			return -EFAULT;
398 		bytes -= pre_copy;
399 		addr += pre_copy;
400 	}
401 
402 	nocache = round_down(bytes, SMP_CACHE_BYTES);
403 	if (copy_from_iter_nocache(addr, nocache, i) != nocache)
404 		return -EFAULT;
405 	bytes -= nocache;
406 	addr += nocache;
407 
408 	if (bytes && copy_from_iter(addr, bytes, i) != bytes)
409 		return -EFAULT;
410 
411 	return 0;
412 }
413 
414 union tls_iter_offset {
415 	struct iov_iter *msg_iter;
416 	int offset;
417 };
418 
419 static int tls_push_data(struct sock *sk,
420 			 union tls_iter_offset iter_offset,
421 			 size_t size, int flags,
422 			 unsigned char record_type,
423 			 struct page *zc_page)
424 {
425 	struct tls_context *tls_ctx = tls_get_ctx(sk);
426 	struct tls_prot_info *prot = &tls_ctx->prot_info;
427 	struct tls_offload_context_tx *ctx = tls_offload_ctx_tx(tls_ctx);
428 	struct tls_record_info *record;
429 	int tls_push_record_flags;
430 	struct page_frag *pfrag;
431 	size_t orig_size = size;
432 	u32 max_open_record_len;
433 	bool more = false;
434 	bool done = false;
435 	int copy, rc = 0;
436 	long timeo;
437 
438 	if (flags &
439 	    ~(MSG_MORE | MSG_DONTWAIT | MSG_NOSIGNAL | MSG_SENDPAGE_NOTLAST))
440 		return -EOPNOTSUPP;
441 
442 	if (unlikely(sk->sk_err))
443 		return -sk->sk_err;
444 
445 	flags |= MSG_SENDPAGE_DECRYPTED;
446 	tls_push_record_flags = flags | MSG_SENDPAGE_NOTLAST;
447 
448 	timeo = sock_sndtimeo(sk, flags & MSG_DONTWAIT);
449 	if (tls_is_partially_sent_record(tls_ctx)) {
450 		rc = tls_push_partial_record(sk, tls_ctx, flags);
451 		if (rc < 0)
452 			return rc;
453 	}
454 
455 	pfrag = sk_page_frag(sk);
456 
457 	/* TLS_HEADER_SIZE is not counted as part of the TLS record, and
458 	 * we need to leave room for an authentication tag.
459 	 */
460 	max_open_record_len = TLS_MAX_PAYLOAD_SIZE +
461 			      prot->prepend_size;
462 	do {
463 		rc = tls_do_allocation(sk, ctx, pfrag, prot->prepend_size);
464 		if (unlikely(rc)) {
465 			rc = sk_stream_wait_memory(sk, &timeo);
466 			if (!rc)
467 				continue;
468 
469 			record = ctx->open_record;
470 			if (!record)
471 				break;
472 handle_error:
473 			if (record_type != TLS_RECORD_TYPE_DATA) {
474 				/* avoid sending partial
475 				 * record with type !=
476 				 * application_data
477 				 */
478 				size = orig_size;
479 				destroy_record(record);
480 				ctx->open_record = NULL;
481 			} else if (record->len > prot->prepend_size) {
482 				goto last_record;
483 			}
484 
485 			break;
486 		}
487 
488 		record = ctx->open_record;
489 
490 		copy = min_t(size_t, size, max_open_record_len - record->len);
491 		if (copy && zc_page) {
492 			struct page_frag zc_pfrag;
493 
494 			zc_pfrag.page = zc_page;
495 			zc_pfrag.offset = iter_offset.offset;
496 			zc_pfrag.size = copy;
497 			tls_append_frag(record, &zc_pfrag, copy);
498 		} else if (copy) {
499 			copy = min_t(size_t, copy, pfrag->size - pfrag->offset);
500 
501 			rc = tls_device_copy_data(page_address(pfrag->page) +
502 						  pfrag->offset, copy,
503 						  iter_offset.msg_iter);
504 			if (rc)
505 				goto handle_error;
506 			tls_append_frag(record, pfrag, copy);
507 		}
508 
509 		size -= copy;
510 		if (!size) {
511 last_record:
512 			tls_push_record_flags = flags;
513 			if (flags & (MSG_SENDPAGE_NOTLAST | MSG_MORE)) {
514 				more = true;
515 				break;
516 			}
517 
518 			done = true;
519 		}
520 
521 		if (done || record->len >= max_open_record_len ||
522 		    (record->num_frags >= MAX_SKB_FRAGS - 1)) {
523 			rc = tls_device_record_close(sk, tls_ctx, record,
524 						     pfrag, record_type);
525 			if (rc) {
526 				if (rc > 0) {
527 					size += rc;
528 				} else {
529 					size = orig_size;
530 					destroy_record(record);
531 					ctx->open_record = NULL;
532 					break;
533 				}
534 			}
535 
536 			rc = tls_push_record(sk,
537 					     tls_ctx,
538 					     ctx,
539 					     record,
540 					     tls_push_record_flags);
541 			if (rc < 0)
542 				break;
543 		}
544 	} while (!done);
545 
546 	tls_ctx->pending_open_record_frags = more;
547 
548 	if (orig_size - size > 0)
549 		rc = orig_size - size;
550 
551 	return rc;
552 }
553 
554 int tls_device_sendmsg(struct sock *sk, struct msghdr *msg, size_t size)
555 {
556 	unsigned char record_type = TLS_RECORD_TYPE_DATA;
557 	struct tls_context *tls_ctx = tls_get_ctx(sk);
558 	union tls_iter_offset iter;
559 	int rc;
560 
561 	mutex_lock(&tls_ctx->tx_lock);
562 	lock_sock(sk);
563 
564 	if (unlikely(msg->msg_controllen)) {
565 		rc = tls_proccess_cmsg(sk, msg, &record_type);
566 		if (rc)
567 			goto out;
568 	}
569 
570 	iter.msg_iter = &msg->msg_iter;
571 	rc = tls_push_data(sk, iter, size, msg->msg_flags, record_type, NULL);
572 
573 out:
574 	release_sock(sk);
575 	mutex_unlock(&tls_ctx->tx_lock);
576 	return rc;
577 }
578 
579 int tls_device_sendpage(struct sock *sk, struct page *page,
580 			int offset, size_t size, int flags)
581 {
582 	struct tls_context *tls_ctx = tls_get_ctx(sk);
583 	union tls_iter_offset iter_offset;
584 	struct iov_iter msg_iter;
585 	char *kaddr;
586 	struct kvec iov;
587 	int rc;
588 
589 	if (flags & MSG_SENDPAGE_NOTLAST)
590 		flags |= MSG_MORE;
591 
592 	mutex_lock(&tls_ctx->tx_lock);
593 	lock_sock(sk);
594 
595 	if (flags & MSG_OOB) {
596 		rc = -EOPNOTSUPP;
597 		goto out;
598 	}
599 
600 	if (tls_ctx->zerocopy_sendfile) {
601 		iter_offset.offset = offset;
602 		rc = tls_push_data(sk, iter_offset, size,
603 				   flags, TLS_RECORD_TYPE_DATA, page);
604 		goto out;
605 	}
606 
607 	kaddr = kmap(page);
608 	iov.iov_base = kaddr + offset;
609 	iov.iov_len = size;
610 	iov_iter_kvec(&msg_iter, WRITE, &iov, 1, size);
611 	iter_offset.msg_iter = &msg_iter;
612 	rc = tls_push_data(sk, iter_offset, size, flags, TLS_RECORD_TYPE_DATA,
613 			   NULL);
614 	kunmap(page);
615 
616 out:
617 	release_sock(sk);
618 	mutex_unlock(&tls_ctx->tx_lock);
619 	return rc;
620 }
621 
622 struct tls_record_info *tls_get_record(struct tls_offload_context_tx *context,
623 				       u32 seq, u64 *p_record_sn)
624 {
625 	u64 record_sn = context->hint_record_sn;
626 	struct tls_record_info *info, *last;
627 
628 	info = context->retransmit_hint;
629 	if (!info ||
630 	    before(seq, info->end_seq - info->len)) {
631 		/* if retransmit_hint is irrelevant start
632 		 * from the beginning of the list
633 		 */
634 		info = list_first_entry_or_null(&context->records_list,
635 						struct tls_record_info, list);
636 		if (!info)
637 			return NULL;
638 		/* send the start_marker record if seq number is before the
639 		 * tls offload start marker sequence number. This record is
640 		 * required to handle TCP packets which are before TLS offload
641 		 * started.
642 		 *  And if it's not start marker, look if this seq number
643 		 * belongs to the list.
644 		 */
645 		if (likely(!tls_record_is_start_marker(info))) {
646 			/* we have the first record, get the last record to see
647 			 * if this seq number belongs to the list.
648 			 */
649 			last = list_last_entry(&context->records_list,
650 					       struct tls_record_info, list);
651 
652 			if (!between(seq, tls_record_start_seq(info),
653 				     last->end_seq))
654 				return NULL;
655 		}
656 		record_sn = context->unacked_record_sn;
657 	}
658 
659 	/* We just need the _rcu for the READ_ONCE() */
660 	rcu_read_lock();
661 	list_for_each_entry_from_rcu(info, &context->records_list, list) {
662 		if (before(seq, info->end_seq)) {
663 			if (!context->retransmit_hint ||
664 			    after(info->end_seq,
665 				  context->retransmit_hint->end_seq)) {
666 				context->hint_record_sn = record_sn;
667 				context->retransmit_hint = info;
668 			}
669 			*p_record_sn = record_sn;
670 			goto exit_rcu_unlock;
671 		}
672 		record_sn++;
673 	}
674 	info = NULL;
675 
676 exit_rcu_unlock:
677 	rcu_read_unlock();
678 	return info;
679 }
680 EXPORT_SYMBOL(tls_get_record);
681 
682 static int tls_device_push_pending_record(struct sock *sk, int flags)
683 {
684 	union tls_iter_offset iter;
685 	struct iov_iter msg_iter;
686 
687 	iov_iter_kvec(&msg_iter, WRITE, NULL, 0, 0);
688 	iter.msg_iter = &msg_iter;
689 	return tls_push_data(sk, iter, 0, flags, TLS_RECORD_TYPE_DATA, NULL);
690 }
691 
692 void tls_device_write_space(struct sock *sk, struct tls_context *ctx)
693 {
694 	if (tls_is_partially_sent_record(ctx)) {
695 		gfp_t sk_allocation = sk->sk_allocation;
696 
697 		WARN_ON_ONCE(sk->sk_write_pending);
698 
699 		sk->sk_allocation = GFP_ATOMIC;
700 		tls_push_partial_record(sk, ctx,
701 					MSG_DONTWAIT | MSG_NOSIGNAL |
702 					MSG_SENDPAGE_DECRYPTED);
703 		sk->sk_allocation = sk_allocation;
704 	}
705 }
706 
707 static void tls_device_resync_rx(struct tls_context *tls_ctx,
708 				 struct sock *sk, u32 seq, u8 *rcd_sn)
709 {
710 	struct tls_offload_context_rx *rx_ctx = tls_offload_ctx_rx(tls_ctx);
711 	struct net_device *netdev;
712 
713 	trace_tls_device_rx_resync_send(sk, seq, rcd_sn, rx_ctx->resync_type);
714 	rcu_read_lock();
715 	netdev = READ_ONCE(tls_ctx->netdev);
716 	if (netdev)
717 		netdev->tlsdev_ops->tls_dev_resync(netdev, sk, seq, rcd_sn,
718 						   TLS_OFFLOAD_CTX_DIR_RX);
719 	rcu_read_unlock();
720 	TLS_INC_STATS(sock_net(sk), LINUX_MIB_TLSRXDEVICERESYNC);
721 }
722 
723 static bool
724 tls_device_rx_resync_async(struct tls_offload_resync_async *resync_async,
725 			   s64 resync_req, u32 *seq, u16 *rcd_delta)
726 {
727 	u32 is_async = resync_req & RESYNC_REQ_ASYNC;
728 	u32 req_seq = resync_req >> 32;
729 	u32 req_end = req_seq + ((resync_req >> 16) & 0xffff);
730 	u16 i;
731 
732 	*rcd_delta = 0;
733 
734 	if (is_async) {
735 		/* shouldn't get to wraparound:
736 		 * too long in async stage, something bad happened
737 		 */
738 		if (WARN_ON_ONCE(resync_async->rcd_delta == USHRT_MAX))
739 			return false;
740 
741 		/* asynchronous stage: log all headers seq such that
742 		 * req_seq <= seq <= end_seq, and wait for real resync request
743 		 */
744 		if (before(*seq, req_seq))
745 			return false;
746 		if (!after(*seq, req_end) &&
747 		    resync_async->loglen < TLS_DEVICE_RESYNC_ASYNC_LOGMAX)
748 			resync_async->log[resync_async->loglen++] = *seq;
749 
750 		resync_async->rcd_delta++;
751 
752 		return false;
753 	}
754 
755 	/* synchronous stage: check against the logged entries and
756 	 * proceed to check the next entries if no match was found
757 	 */
758 	for (i = 0; i < resync_async->loglen; i++)
759 		if (req_seq == resync_async->log[i] &&
760 		    atomic64_try_cmpxchg(&resync_async->req, &resync_req, 0)) {
761 			*rcd_delta = resync_async->rcd_delta - i;
762 			*seq = req_seq;
763 			resync_async->loglen = 0;
764 			resync_async->rcd_delta = 0;
765 			return true;
766 		}
767 
768 	resync_async->loglen = 0;
769 	resync_async->rcd_delta = 0;
770 
771 	if (req_seq == *seq &&
772 	    atomic64_try_cmpxchg(&resync_async->req,
773 				 &resync_req, 0))
774 		return true;
775 
776 	return false;
777 }
778 
779 void tls_device_rx_resync_new_rec(struct sock *sk, u32 rcd_len, u32 seq)
780 {
781 	struct tls_context *tls_ctx = tls_get_ctx(sk);
782 	struct tls_offload_context_rx *rx_ctx;
783 	u8 rcd_sn[TLS_MAX_REC_SEQ_SIZE];
784 	u32 sock_data, is_req_pending;
785 	struct tls_prot_info *prot;
786 	s64 resync_req;
787 	u16 rcd_delta;
788 	u32 req_seq;
789 
790 	if (tls_ctx->rx_conf != TLS_HW)
791 		return;
792 	if (unlikely(test_bit(TLS_RX_DEV_DEGRADED, &tls_ctx->flags)))
793 		return;
794 
795 	prot = &tls_ctx->prot_info;
796 	rx_ctx = tls_offload_ctx_rx(tls_ctx);
797 	memcpy(rcd_sn, tls_ctx->rx.rec_seq, prot->rec_seq_size);
798 
799 	switch (rx_ctx->resync_type) {
800 	case TLS_OFFLOAD_SYNC_TYPE_DRIVER_REQ:
801 		resync_req = atomic64_read(&rx_ctx->resync_req);
802 		req_seq = resync_req >> 32;
803 		seq += TLS_HEADER_SIZE - 1;
804 		is_req_pending = resync_req;
805 
806 		if (likely(!is_req_pending) || req_seq != seq ||
807 		    !atomic64_try_cmpxchg(&rx_ctx->resync_req, &resync_req, 0))
808 			return;
809 		break;
810 	case TLS_OFFLOAD_SYNC_TYPE_CORE_NEXT_HINT:
811 		if (likely(!rx_ctx->resync_nh_do_now))
812 			return;
813 
814 		/* head of next rec is already in, note that the sock_inq will
815 		 * include the currently parsed message when called from parser
816 		 */
817 		sock_data = tcp_inq(sk);
818 		if (sock_data > rcd_len) {
819 			trace_tls_device_rx_resync_nh_delay(sk, sock_data,
820 							    rcd_len);
821 			return;
822 		}
823 
824 		rx_ctx->resync_nh_do_now = 0;
825 		seq += rcd_len;
826 		tls_bigint_increment(rcd_sn, prot->rec_seq_size);
827 		break;
828 	case TLS_OFFLOAD_SYNC_TYPE_DRIVER_REQ_ASYNC:
829 		resync_req = atomic64_read(&rx_ctx->resync_async->req);
830 		is_req_pending = resync_req;
831 		if (likely(!is_req_pending))
832 			return;
833 
834 		if (!tls_device_rx_resync_async(rx_ctx->resync_async,
835 						resync_req, &seq, &rcd_delta))
836 			return;
837 		tls_bigint_subtract(rcd_sn, rcd_delta);
838 		break;
839 	}
840 
841 	tls_device_resync_rx(tls_ctx, sk, seq, rcd_sn);
842 }
843 
844 static void tls_device_core_ctrl_rx_resync(struct tls_context *tls_ctx,
845 					   struct tls_offload_context_rx *ctx,
846 					   struct sock *sk, struct sk_buff *skb)
847 {
848 	struct strp_msg *rxm;
849 
850 	/* device will request resyncs by itself based on stream scan */
851 	if (ctx->resync_type != TLS_OFFLOAD_SYNC_TYPE_CORE_NEXT_HINT)
852 		return;
853 	/* already scheduled */
854 	if (ctx->resync_nh_do_now)
855 		return;
856 	/* seen decrypted fragments since last fully-failed record */
857 	if (ctx->resync_nh_reset) {
858 		ctx->resync_nh_reset = 0;
859 		ctx->resync_nh.decrypted_failed = 1;
860 		ctx->resync_nh.decrypted_tgt = TLS_DEVICE_RESYNC_NH_START_IVAL;
861 		return;
862 	}
863 
864 	if (++ctx->resync_nh.decrypted_failed <= ctx->resync_nh.decrypted_tgt)
865 		return;
866 
867 	/* doing resync, bump the next target in case it fails */
868 	if (ctx->resync_nh.decrypted_tgt < TLS_DEVICE_RESYNC_NH_MAX_IVAL)
869 		ctx->resync_nh.decrypted_tgt *= 2;
870 	else
871 		ctx->resync_nh.decrypted_tgt += TLS_DEVICE_RESYNC_NH_MAX_IVAL;
872 
873 	rxm = strp_msg(skb);
874 
875 	/* head of next rec is already in, parser will sync for us */
876 	if (tcp_inq(sk) > rxm->full_len) {
877 		trace_tls_device_rx_resync_nh_schedule(sk);
878 		ctx->resync_nh_do_now = 1;
879 	} else {
880 		struct tls_prot_info *prot = &tls_ctx->prot_info;
881 		u8 rcd_sn[TLS_MAX_REC_SEQ_SIZE];
882 
883 		memcpy(rcd_sn, tls_ctx->rx.rec_seq, prot->rec_seq_size);
884 		tls_bigint_increment(rcd_sn, prot->rec_seq_size);
885 
886 		tls_device_resync_rx(tls_ctx, sk, tcp_sk(sk)->copied_seq,
887 				     rcd_sn);
888 	}
889 }
890 
891 static int tls_device_reencrypt(struct sock *sk, struct sk_buff *skb)
892 {
893 	struct strp_msg *rxm = strp_msg(skb);
894 	int err = 0, offset = rxm->offset, copy, nsg, data_len, pos;
895 	struct sk_buff *skb_iter, *unused;
896 	struct scatterlist sg[1];
897 	char *orig_buf, *buf;
898 
899 	orig_buf = kmalloc(rxm->full_len + TLS_HEADER_SIZE +
900 			   TLS_CIPHER_AES_GCM_128_IV_SIZE, sk->sk_allocation);
901 	if (!orig_buf)
902 		return -ENOMEM;
903 	buf = orig_buf;
904 
905 	nsg = skb_cow_data(skb, 0, &unused);
906 	if (unlikely(nsg < 0)) {
907 		err = nsg;
908 		goto free_buf;
909 	}
910 
911 	sg_init_table(sg, 1);
912 	sg_set_buf(&sg[0], buf,
913 		   rxm->full_len + TLS_HEADER_SIZE +
914 		   TLS_CIPHER_AES_GCM_128_IV_SIZE);
915 	err = skb_copy_bits(skb, offset, buf,
916 			    TLS_HEADER_SIZE + TLS_CIPHER_AES_GCM_128_IV_SIZE);
917 	if (err)
918 		goto free_buf;
919 
920 	/* We are interested only in the decrypted data not the auth */
921 	err = decrypt_skb(sk, skb, sg);
922 	if (err != -EBADMSG)
923 		goto free_buf;
924 	else
925 		err = 0;
926 
927 	data_len = rxm->full_len - TLS_CIPHER_AES_GCM_128_TAG_SIZE;
928 
929 	if (skb_pagelen(skb) > offset) {
930 		copy = min_t(int, skb_pagelen(skb) - offset, data_len);
931 
932 		if (skb->decrypted) {
933 			err = skb_store_bits(skb, offset, buf, copy);
934 			if (err)
935 				goto free_buf;
936 		}
937 
938 		offset += copy;
939 		buf += copy;
940 	}
941 
942 	pos = skb_pagelen(skb);
943 	skb_walk_frags(skb, skb_iter) {
944 		int frag_pos;
945 
946 		/* Practically all frags must belong to msg if reencrypt
947 		 * is needed with current strparser and coalescing logic,
948 		 * but strparser may "get optimized", so let's be safe.
949 		 */
950 		if (pos + skb_iter->len <= offset)
951 			goto done_with_frag;
952 		if (pos >= data_len + rxm->offset)
953 			break;
954 
955 		frag_pos = offset - pos;
956 		copy = min_t(int, skb_iter->len - frag_pos,
957 			     data_len + rxm->offset - offset);
958 
959 		if (skb_iter->decrypted) {
960 			err = skb_store_bits(skb_iter, frag_pos, buf, copy);
961 			if (err)
962 				goto free_buf;
963 		}
964 
965 		offset += copy;
966 		buf += copy;
967 done_with_frag:
968 		pos += skb_iter->len;
969 	}
970 
971 free_buf:
972 	kfree(orig_buf);
973 	return err;
974 }
975 
976 int tls_device_decrypted(struct sock *sk, struct tls_context *tls_ctx,
977 			 struct sk_buff *skb, struct strp_msg *rxm)
978 {
979 	struct tls_offload_context_rx *ctx = tls_offload_ctx_rx(tls_ctx);
980 	int is_decrypted = skb->decrypted;
981 	int is_encrypted = !is_decrypted;
982 	struct sk_buff *skb_iter;
983 
984 	/* Check if all the data is decrypted already */
985 	skb_walk_frags(skb, skb_iter) {
986 		is_decrypted &= skb_iter->decrypted;
987 		is_encrypted &= !skb_iter->decrypted;
988 	}
989 
990 	trace_tls_device_decrypted(sk, tcp_sk(sk)->copied_seq - rxm->full_len,
991 				   tls_ctx->rx.rec_seq, rxm->full_len,
992 				   is_encrypted, is_decrypted);
993 
994 	if (unlikely(test_bit(TLS_RX_DEV_DEGRADED, &tls_ctx->flags))) {
995 		if (likely(is_encrypted || is_decrypted))
996 			return is_decrypted;
997 
998 		/* After tls_device_down disables the offload, the next SKB will
999 		 * likely have initial fragments decrypted, and final ones not
1000 		 * decrypted. We need to reencrypt that single SKB.
1001 		 */
1002 		return tls_device_reencrypt(sk, skb);
1003 	}
1004 
1005 	/* Return immediately if the record is either entirely plaintext or
1006 	 * entirely ciphertext. Otherwise handle reencrypt partially decrypted
1007 	 * record.
1008 	 */
1009 	if (is_decrypted) {
1010 		ctx->resync_nh_reset = 1;
1011 		return is_decrypted;
1012 	}
1013 	if (is_encrypted) {
1014 		tls_device_core_ctrl_rx_resync(tls_ctx, ctx, sk, skb);
1015 		return 0;
1016 	}
1017 
1018 	ctx->resync_nh_reset = 1;
1019 	return tls_device_reencrypt(sk, skb);
1020 }
1021 
1022 static void tls_device_attach(struct tls_context *ctx, struct sock *sk,
1023 			      struct net_device *netdev)
1024 {
1025 	if (sk->sk_destruct != tls_device_sk_destruct) {
1026 		refcount_set(&ctx->refcount, 1);
1027 		dev_hold(netdev);
1028 		ctx->netdev = netdev;
1029 		spin_lock_irq(&tls_device_lock);
1030 		list_add_tail(&ctx->list, &tls_device_list);
1031 		spin_unlock_irq(&tls_device_lock);
1032 
1033 		ctx->sk_destruct = sk->sk_destruct;
1034 		smp_store_release(&sk->sk_destruct, tls_device_sk_destruct);
1035 	}
1036 }
1037 
1038 int tls_set_device_offload(struct sock *sk, struct tls_context *ctx)
1039 {
1040 	u16 nonce_size, tag_size, iv_size, rec_seq_size, salt_size;
1041 	struct tls_context *tls_ctx = tls_get_ctx(sk);
1042 	struct tls_prot_info *prot = &tls_ctx->prot_info;
1043 	struct tls_record_info *start_marker_record;
1044 	struct tls_offload_context_tx *offload_ctx;
1045 	struct tls_crypto_info *crypto_info;
1046 	struct net_device *netdev;
1047 	char *iv, *rec_seq;
1048 	struct sk_buff *skb;
1049 	__be64 rcd_sn;
1050 	int rc;
1051 
1052 	if (!ctx)
1053 		return -EINVAL;
1054 
1055 	if (ctx->priv_ctx_tx)
1056 		return -EEXIST;
1057 
1058 	netdev = get_netdev_for_sock(sk);
1059 	if (!netdev) {
1060 		pr_err_ratelimited("%s: netdev not found\n", __func__);
1061 		return -EINVAL;
1062 	}
1063 
1064 	if (!(netdev->features & NETIF_F_HW_TLS_TX)) {
1065 		rc = -EOPNOTSUPP;
1066 		goto release_netdev;
1067 	}
1068 
1069 	crypto_info = &ctx->crypto_send.info;
1070 	if (crypto_info->version != TLS_1_2_VERSION) {
1071 		rc = -EOPNOTSUPP;
1072 		goto release_netdev;
1073 	}
1074 
1075 	switch (crypto_info->cipher_type) {
1076 	case TLS_CIPHER_AES_GCM_128:
1077 		nonce_size = TLS_CIPHER_AES_GCM_128_IV_SIZE;
1078 		tag_size = TLS_CIPHER_AES_GCM_128_TAG_SIZE;
1079 		iv_size = TLS_CIPHER_AES_GCM_128_IV_SIZE;
1080 		iv = ((struct tls12_crypto_info_aes_gcm_128 *)crypto_info)->iv;
1081 		rec_seq_size = TLS_CIPHER_AES_GCM_128_REC_SEQ_SIZE;
1082 		salt_size = TLS_CIPHER_AES_GCM_128_SALT_SIZE;
1083 		rec_seq =
1084 		 ((struct tls12_crypto_info_aes_gcm_128 *)crypto_info)->rec_seq;
1085 		break;
1086 	default:
1087 		rc = -EINVAL;
1088 		goto release_netdev;
1089 	}
1090 
1091 	/* Sanity-check the rec_seq_size for stack allocations */
1092 	if (rec_seq_size > TLS_MAX_REC_SEQ_SIZE) {
1093 		rc = -EINVAL;
1094 		goto release_netdev;
1095 	}
1096 
1097 	prot->version = crypto_info->version;
1098 	prot->cipher_type = crypto_info->cipher_type;
1099 	prot->prepend_size = TLS_HEADER_SIZE + nonce_size;
1100 	prot->tag_size = tag_size;
1101 	prot->overhead_size = prot->prepend_size + prot->tag_size;
1102 	prot->iv_size = iv_size;
1103 	prot->salt_size = salt_size;
1104 	ctx->tx.iv = kmalloc(iv_size + TLS_CIPHER_AES_GCM_128_SALT_SIZE,
1105 			     GFP_KERNEL);
1106 	if (!ctx->tx.iv) {
1107 		rc = -ENOMEM;
1108 		goto release_netdev;
1109 	}
1110 
1111 	memcpy(ctx->tx.iv + TLS_CIPHER_AES_GCM_128_SALT_SIZE, iv, iv_size);
1112 
1113 	prot->rec_seq_size = rec_seq_size;
1114 	ctx->tx.rec_seq = kmemdup(rec_seq, rec_seq_size, GFP_KERNEL);
1115 	if (!ctx->tx.rec_seq) {
1116 		rc = -ENOMEM;
1117 		goto free_iv;
1118 	}
1119 
1120 	start_marker_record = kmalloc(sizeof(*start_marker_record), GFP_KERNEL);
1121 	if (!start_marker_record) {
1122 		rc = -ENOMEM;
1123 		goto free_rec_seq;
1124 	}
1125 
1126 	offload_ctx = kzalloc(TLS_OFFLOAD_CONTEXT_SIZE_TX, GFP_KERNEL);
1127 	if (!offload_ctx) {
1128 		rc = -ENOMEM;
1129 		goto free_marker_record;
1130 	}
1131 
1132 	rc = tls_sw_fallback_init(sk, offload_ctx, crypto_info);
1133 	if (rc)
1134 		goto free_offload_ctx;
1135 
1136 	/* start at rec_seq - 1 to account for the start marker record */
1137 	memcpy(&rcd_sn, ctx->tx.rec_seq, sizeof(rcd_sn));
1138 	offload_ctx->unacked_record_sn = be64_to_cpu(rcd_sn) - 1;
1139 
1140 	start_marker_record->end_seq = tcp_sk(sk)->write_seq;
1141 	start_marker_record->len = 0;
1142 	start_marker_record->num_frags = 0;
1143 
1144 	INIT_LIST_HEAD(&offload_ctx->records_list);
1145 	list_add_tail(&start_marker_record->list, &offload_ctx->records_list);
1146 	spin_lock_init(&offload_ctx->lock);
1147 	sg_init_table(offload_ctx->sg_tx_data,
1148 		      ARRAY_SIZE(offload_ctx->sg_tx_data));
1149 
1150 	clean_acked_data_enable(inet_csk(sk), &tls_icsk_clean_acked);
1151 	ctx->push_pending_record = tls_device_push_pending_record;
1152 
1153 	/* TLS offload is greatly simplified if we don't send
1154 	 * SKBs where only part of the payload needs to be encrypted.
1155 	 * So mark the last skb in the write queue as end of record.
1156 	 */
1157 	skb = tcp_write_queue_tail(sk);
1158 	if (skb)
1159 		TCP_SKB_CB(skb)->eor = 1;
1160 
1161 	/* Avoid offloading if the device is down
1162 	 * We don't want to offload new flows after
1163 	 * the NETDEV_DOWN event
1164 	 *
1165 	 * device_offload_lock is taken in tls_devices's NETDEV_DOWN
1166 	 * handler thus protecting from the device going down before
1167 	 * ctx was added to tls_device_list.
1168 	 */
1169 	down_read(&device_offload_lock);
1170 	if (!(netdev->flags & IFF_UP)) {
1171 		rc = -EINVAL;
1172 		goto release_lock;
1173 	}
1174 
1175 	ctx->priv_ctx_tx = offload_ctx;
1176 	rc = netdev->tlsdev_ops->tls_dev_add(netdev, sk, TLS_OFFLOAD_CTX_DIR_TX,
1177 					     &ctx->crypto_send.info,
1178 					     tcp_sk(sk)->write_seq);
1179 	trace_tls_device_offload_set(sk, TLS_OFFLOAD_CTX_DIR_TX,
1180 				     tcp_sk(sk)->write_seq, rec_seq, rc);
1181 	if (rc)
1182 		goto release_lock;
1183 
1184 	tls_device_attach(ctx, sk, netdev);
1185 	up_read(&device_offload_lock);
1186 
1187 	/* following this assignment tls_is_sk_tx_device_offloaded
1188 	 * will return true and the context might be accessed
1189 	 * by the netdev's xmit function.
1190 	 */
1191 	smp_store_release(&sk->sk_validate_xmit_skb, tls_validate_xmit_skb);
1192 	dev_put(netdev);
1193 
1194 	return 0;
1195 
1196 release_lock:
1197 	up_read(&device_offload_lock);
1198 	clean_acked_data_disable(inet_csk(sk));
1199 	crypto_free_aead(offload_ctx->aead_send);
1200 free_offload_ctx:
1201 	kfree(offload_ctx);
1202 	ctx->priv_ctx_tx = NULL;
1203 free_marker_record:
1204 	kfree(start_marker_record);
1205 free_rec_seq:
1206 	kfree(ctx->tx.rec_seq);
1207 free_iv:
1208 	kfree(ctx->tx.iv);
1209 release_netdev:
1210 	dev_put(netdev);
1211 	return rc;
1212 }
1213 
1214 int tls_set_device_offload_rx(struct sock *sk, struct tls_context *ctx)
1215 {
1216 	struct tls12_crypto_info_aes_gcm_128 *info;
1217 	struct tls_offload_context_rx *context;
1218 	struct net_device *netdev;
1219 	int rc = 0;
1220 
1221 	if (ctx->crypto_recv.info.version != TLS_1_2_VERSION)
1222 		return -EOPNOTSUPP;
1223 
1224 	netdev = get_netdev_for_sock(sk);
1225 	if (!netdev) {
1226 		pr_err_ratelimited("%s: netdev not found\n", __func__);
1227 		return -EINVAL;
1228 	}
1229 
1230 	if (!(netdev->features & NETIF_F_HW_TLS_RX)) {
1231 		rc = -EOPNOTSUPP;
1232 		goto release_netdev;
1233 	}
1234 
1235 	/* Avoid offloading if the device is down
1236 	 * We don't want to offload new flows after
1237 	 * the NETDEV_DOWN event
1238 	 *
1239 	 * device_offload_lock is taken in tls_devices's NETDEV_DOWN
1240 	 * handler thus protecting from the device going down before
1241 	 * ctx was added to tls_device_list.
1242 	 */
1243 	down_read(&device_offload_lock);
1244 	if (!(netdev->flags & IFF_UP)) {
1245 		rc = -EINVAL;
1246 		goto release_lock;
1247 	}
1248 
1249 	context = kzalloc(TLS_OFFLOAD_CONTEXT_SIZE_RX, GFP_KERNEL);
1250 	if (!context) {
1251 		rc = -ENOMEM;
1252 		goto release_lock;
1253 	}
1254 	context->resync_nh_reset = 1;
1255 
1256 	ctx->priv_ctx_rx = context;
1257 	rc = tls_set_sw_offload(sk, ctx, 0);
1258 	if (rc)
1259 		goto release_ctx;
1260 
1261 	rc = netdev->tlsdev_ops->tls_dev_add(netdev, sk, TLS_OFFLOAD_CTX_DIR_RX,
1262 					     &ctx->crypto_recv.info,
1263 					     tcp_sk(sk)->copied_seq);
1264 	info = (void *)&ctx->crypto_recv.info;
1265 	trace_tls_device_offload_set(sk, TLS_OFFLOAD_CTX_DIR_RX,
1266 				     tcp_sk(sk)->copied_seq, info->rec_seq, rc);
1267 	if (rc)
1268 		goto free_sw_resources;
1269 
1270 	tls_device_attach(ctx, sk, netdev);
1271 	up_read(&device_offload_lock);
1272 
1273 	dev_put(netdev);
1274 
1275 	return 0;
1276 
1277 free_sw_resources:
1278 	up_read(&device_offload_lock);
1279 	tls_sw_free_resources_rx(sk);
1280 	down_read(&device_offload_lock);
1281 release_ctx:
1282 	ctx->priv_ctx_rx = NULL;
1283 release_lock:
1284 	up_read(&device_offload_lock);
1285 release_netdev:
1286 	dev_put(netdev);
1287 	return rc;
1288 }
1289 
1290 void tls_device_offload_cleanup_rx(struct sock *sk)
1291 {
1292 	struct tls_context *tls_ctx = tls_get_ctx(sk);
1293 	struct net_device *netdev;
1294 
1295 	down_read(&device_offload_lock);
1296 	netdev = tls_ctx->netdev;
1297 	if (!netdev)
1298 		goto out;
1299 
1300 	netdev->tlsdev_ops->tls_dev_del(netdev, tls_ctx,
1301 					TLS_OFFLOAD_CTX_DIR_RX);
1302 
1303 	if (tls_ctx->tx_conf != TLS_HW) {
1304 		dev_put(netdev);
1305 		tls_ctx->netdev = NULL;
1306 	} else {
1307 		set_bit(TLS_RX_DEV_CLOSED, &tls_ctx->flags);
1308 	}
1309 out:
1310 	up_read(&device_offload_lock);
1311 	tls_sw_release_resources_rx(sk);
1312 }
1313 
1314 static int tls_device_down(struct net_device *netdev)
1315 {
1316 	struct tls_context *ctx, *tmp;
1317 	unsigned long flags;
1318 	LIST_HEAD(list);
1319 
1320 	/* Request a write lock to block new offload attempts */
1321 	down_write(&device_offload_lock);
1322 
1323 	spin_lock_irqsave(&tls_device_lock, flags);
1324 	list_for_each_entry_safe(ctx, tmp, &tls_device_list, list) {
1325 		if (ctx->netdev != netdev ||
1326 		    !refcount_inc_not_zero(&ctx->refcount))
1327 			continue;
1328 
1329 		list_move(&ctx->list, &list);
1330 	}
1331 	spin_unlock_irqrestore(&tls_device_lock, flags);
1332 
1333 	list_for_each_entry_safe(ctx, tmp, &list, list)	{
1334 		/* Stop offloaded TX and switch to the fallback.
1335 		 * tls_is_sk_tx_device_offloaded will return false.
1336 		 */
1337 		WRITE_ONCE(ctx->sk->sk_validate_xmit_skb, tls_validate_xmit_skb_sw);
1338 
1339 		/* Stop the RX and TX resync.
1340 		 * tls_dev_resync must not be called after tls_dev_del.
1341 		 */
1342 		WRITE_ONCE(ctx->netdev, NULL);
1343 
1344 		/* Start skipping the RX resync logic completely. */
1345 		set_bit(TLS_RX_DEV_DEGRADED, &ctx->flags);
1346 
1347 		/* Sync with inflight packets. After this point:
1348 		 * TX: no non-encrypted packets will be passed to the driver.
1349 		 * RX: resync requests from the driver will be ignored.
1350 		 */
1351 		synchronize_net();
1352 
1353 		/* Release the offload context on the driver side. */
1354 		if (ctx->tx_conf == TLS_HW)
1355 			netdev->tlsdev_ops->tls_dev_del(netdev, ctx,
1356 							TLS_OFFLOAD_CTX_DIR_TX);
1357 		if (ctx->rx_conf == TLS_HW &&
1358 		    !test_bit(TLS_RX_DEV_CLOSED, &ctx->flags))
1359 			netdev->tlsdev_ops->tls_dev_del(netdev, ctx,
1360 							TLS_OFFLOAD_CTX_DIR_RX);
1361 
1362 		dev_put(netdev);
1363 
1364 		/* Move the context to a separate list for two reasons:
1365 		 * 1. When the context is deallocated, list_del is called.
1366 		 * 2. It's no longer an offloaded context, so we don't want to
1367 		 *    run offload-specific code on this context.
1368 		 */
1369 		spin_lock_irqsave(&tls_device_lock, flags);
1370 		list_move_tail(&ctx->list, &tls_device_down_list);
1371 		spin_unlock_irqrestore(&tls_device_lock, flags);
1372 
1373 		/* Device contexts for RX and TX will be freed in on sk_destruct
1374 		 * by tls_device_free_ctx. rx_conf and tx_conf stay in TLS_HW.
1375 		 * Now release the ref taken above.
1376 		 */
1377 		if (refcount_dec_and_test(&ctx->refcount))
1378 			tls_device_free_ctx(ctx);
1379 	}
1380 
1381 	up_write(&device_offload_lock);
1382 
1383 	flush_work(&tls_device_gc_work);
1384 
1385 	return NOTIFY_DONE;
1386 }
1387 
1388 static int tls_dev_event(struct notifier_block *this, unsigned long event,
1389 			 void *ptr)
1390 {
1391 	struct net_device *dev = netdev_notifier_info_to_dev(ptr);
1392 
1393 	if (!dev->tlsdev_ops &&
1394 	    !(dev->features & (NETIF_F_HW_TLS_RX | NETIF_F_HW_TLS_TX)))
1395 		return NOTIFY_DONE;
1396 
1397 	switch (event) {
1398 	case NETDEV_REGISTER:
1399 	case NETDEV_FEAT_CHANGE:
1400 		if (netif_is_bond_master(dev))
1401 			return NOTIFY_DONE;
1402 		if ((dev->features & NETIF_F_HW_TLS_RX) &&
1403 		    !dev->tlsdev_ops->tls_dev_resync)
1404 			return NOTIFY_BAD;
1405 
1406 		if  (dev->tlsdev_ops &&
1407 		     dev->tlsdev_ops->tls_dev_add &&
1408 		     dev->tlsdev_ops->tls_dev_del)
1409 			return NOTIFY_DONE;
1410 		else
1411 			return NOTIFY_BAD;
1412 	case NETDEV_DOWN:
1413 		return tls_device_down(dev);
1414 	}
1415 	return NOTIFY_DONE;
1416 }
1417 
1418 static struct notifier_block tls_dev_notifier = {
1419 	.notifier_call	= tls_dev_event,
1420 };
1421 
1422 void __init tls_device_init(void)
1423 {
1424 	register_netdevice_notifier(&tls_dev_notifier);
1425 }
1426 
1427 void __exit tls_device_cleanup(void)
1428 {
1429 	unregister_netdevice_notifier(&tls_dev_notifier);
1430 	flush_work(&tls_device_gc_work);
1431 	clean_acked_data_flush();
1432 }
1433