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 #include <linux/skbuff_ref.h> 41 42 #include "tls.h" 43 #include "trace.h" 44 45 /* device_offload_lock is used to synchronize tls_dev_add 46 * against NETDEV_DOWN notifications. 47 */ 48 static DECLARE_RWSEM(device_offload_lock); 49 50 static struct workqueue_struct *destruct_wq __read_mostly; 51 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 struct page *dummy_page; 57 58 static void tls_device_free_ctx(struct tls_context *ctx) 59 { 60 if (ctx->tx_conf == TLS_HW) 61 kfree(tls_offload_ctx_tx(ctx)); 62 63 if (ctx->rx_conf == TLS_HW) 64 kfree(tls_offload_ctx_rx(ctx)); 65 66 tls_ctx_free(NULL, ctx); 67 } 68 69 static void tls_device_tx_del_task(struct work_struct *work) 70 { 71 struct tls_offload_context_tx *offload_ctx = 72 container_of(work, struct tls_offload_context_tx, destruct_work); 73 struct tls_context *ctx = offload_ctx->ctx; 74 struct net_device *netdev; 75 76 /* Safe, because this is the destroy flow, refcount is 0, so 77 * tls_device_down can't store this field in parallel. 78 */ 79 netdev = rcu_dereference_protected(ctx->netdev, 80 !refcount_read(&ctx->refcount)); 81 82 netdev->tlsdev_ops->tls_dev_del(netdev, ctx, TLS_OFFLOAD_CTX_DIR_TX); 83 dev_put(netdev); 84 ctx->netdev = NULL; 85 tls_device_free_ctx(ctx); 86 } 87 88 static void tls_device_queue_ctx_destruction(struct tls_context *ctx) 89 { 90 struct net_device *netdev; 91 unsigned long flags; 92 bool async_cleanup; 93 94 spin_lock_irqsave(&tls_device_lock, flags); 95 if (unlikely(!refcount_dec_and_test(&ctx->refcount))) { 96 spin_unlock_irqrestore(&tls_device_lock, flags); 97 return; 98 } 99 100 list_del(&ctx->list); /* Remove from tls_device_list / tls_device_down_list */ 101 102 /* Safe, because this is the destroy flow, refcount is 0, so 103 * tls_device_down can't store this field in parallel. 104 */ 105 netdev = rcu_dereference_protected(ctx->netdev, 106 !refcount_read(&ctx->refcount)); 107 108 async_cleanup = netdev && ctx->tx_conf == TLS_HW; 109 if (async_cleanup) { 110 struct tls_offload_context_tx *offload_ctx = tls_offload_ctx_tx(ctx); 111 112 /* queue_work inside the spinlock 113 * to make sure tls_device_down waits for that work. 114 */ 115 queue_work(destruct_wq, &offload_ctx->destruct_work); 116 } 117 spin_unlock_irqrestore(&tls_device_lock, flags); 118 119 if (!async_cleanup) 120 tls_device_free_ctx(ctx); 121 } 122 123 /* We assume that the socket is already connected */ 124 static struct net_device *get_netdev_for_sock(struct sock *sk) 125 { 126 struct dst_entry *dst = sk_dst_get(sk); 127 struct net_device *netdev = NULL; 128 129 if (likely(dst)) { 130 netdev = netdev_sk_get_lowest_dev(dst->dev, sk); 131 dev_hold(netdev); 132 } 133 134 dst_release(dst); 135 136 return netdev; 137 } 138 139 static void destroy_record(struct tls_record_info *record) 140 { 141 int i; 142 143 for (i = 0; i < record->num_frags; i++) 144 __skb_frag_unref(&record->frags[i], false); 145 kfree(record); 146 } 147 148 static void delete_all_records(struct tls_offload_context_tx *offload_ctx) 149 { 150 struct tls_record_info *info, *temp; 151 152 list_for_each_entry_safe(info, temp, &offload_ctx->records_list, list) { 153 list_del(&info->list); 154 destroy_record(info); 155 } 156 157 offload_ctx->retransmit_hint = NULL; 158 } 159 160 static void tls_icsk_clean_acked(struct sock *sk, u32 acked_seq) 161 { 162 struct tls_context *tls_ctx = tls_get_ctx(sk); 163 struct tls_record_info *info, *temp; 164 struct tls_offload_context_tx *ctx; 165 u64 deleted_records = 0; 166 unsigned long flags; 167 168 if (!tls_ctx) 169 return; 170 171 ctx = tls_offload_ctx_tx(tls_ctx); 172 173 spin_lock_irqsave(&ctx->lock, flags); 174 info = ctx->retransmit_hint; 175 if (info && !before(acked_seq, info->end_seq)) 176 ctx->retransmit_hint = NULL; 177 178 list_for_each_entry_safe(info, temp, &ctx->records_list, list) { 179 if (before(acked_seq, info->end_seq)) 180 break; 181 list_del(&info->list); 182 183 destroy_record(info); 184 deleted_records++; 185 } 186 187 ctx->unacked_record_sn += deleted_records; 188 spin_unlock_irqrestore(&ctx->lock, flags); 189 } 190 191 /* At this point, there should be no references on this 192 * socket and no in-flight SKBs associated with this 193 * socket, so it is safe to free all the resources. 194 */ 195 void tls_device_sk_destruct(struct sock *sk) 196 { 197 struct tls_context *tls_ctx = tls_get_ctx(sk); 198 struct tls_offload_context_tx *ctx = tls_offload_ctx_tx(tls_ctx); 199 200 tls_ctx->sk_destruct(sk); 201 202 if (tls_ctx->tx_conf == TLS_HW) { 203 if (ctx->open_record) 204 destroy_record(ctx->open_record); 205 delete_all_records(ctx); 206 crypto_free_aead(ctx->aead_send); 207 clean_acked_data_disable(inet_csk(sk)); 208 } 209 210 tls_device_queue_ctx_destruction(tls_ctx); 211 } 212 EXPORT_SYMBOL_GPL(tls_device_sk_destruct); 213 214 void tls_device_free_resources_tx(struct sock *sk) 215 { 216 struct tls_context *tls_ctx = tls_get_ctx(sk); 217 218 tls_free_partial_record(sk, tls_ctx); 219 } 220 221 void tls_offload_tx_resync_request(struct sock *sk, u32 got_seq, u32 exp_seq) 222 { 223 struct tls_context *tls_ctx = tls_get_ctx(sk); 224 225 trace_tls_device_tx_resync_req(sk, got_seq, exp_seq); 226 WARN_ON(test_and_set_bit(TLS_TX_SYNC_SCHED, &tls_ctx->flags)); 227 } 228 EXPORT_SYMBOL_GPL(tls_offload_tx_resync_request); 229 230 static void tls_device_resync_tx(struct sock *sk, struct tls_context *tls_ctx, 231 u32 seq) 232 { 233 struct net_device *netdev; 234 int err = 0; 235 u8 *rcd_sn; 236 237 tcp_write_collapse_fence(sk); 238 rcd_sn = tls_ctx->tx.rec_seq; 239 240 trace_tls_device_tx_resync_send(sk, seq, rcd_sn); 241 down_read(&device_offload_lock); 242 netdev = rcu_dereference_protected(tls_ctx->netdev, 243 lockdep_is_held(&device_offload_lock)); 244 if (netdev) 245 err = netdev->tlsdev_ops->tls_dev_resync(netdev, sk, seq, 246 rcd_sn, 247 TLS_OFFLOAD_CTX_DIR_TX); 248 up_read(&device_offload_lock); 249 if (err) 250 return; 251 252 clear_bit_unlock(TLS_TX_SYNC_SCHED, &tls_ctx->flags); 253 } 254 255 static void tls_append_frag(struct tls_record_info *record, 256 struct page_frag *pfrag, 257 int size) 258 { 259 skb_frag_t *frag; 260 261 frag = &record->frags[record->num_frags - 1]; 262 if (skb_frag_page(frag) == pfrag->page && 263 skb_frag_off(frag) + skb_frag_size(frag) == pfrag->offset) { 264 skb_frag_size_add(frag, size); 265 } else { 266 ++frag; 267 skb_frag_fill_page_desc(frag, pfrag->page, pfrag->offset, 268 size); 269 ++record->num_frags; 270 get_page(pfrag->page); 271 } 272 273 pfrag->offset += size; 274 record->len += size; 275 } 276 277 static int tls_push_record(struct sock *sk, 278 struct tls_context *ctx, 279 struct tls_offload_context_tx *offload_ctx, 280 struct tls_record_info *record, 281 int flags) 282 { 283 struct tls_prot_info *prot = &ctx->prot_info; 284 struct tcp_sock *tp = tcp_sk(sk); 285 skb_frag_t *frag; 286 int i; 287 288 record->end_seq = tp->write_seq + record->len; 289 list_add_tail_rcu(&record->list, &offload_ctx->records_list); 290 offload_ctx->open_record = NULL; 291 292 if (test_bit(TLS_TX_SYNC_SCHED, &ctx->flags)) 293 tls_device_resync_tx(sk, ctx, tp->write_seq); 294 295 tls_advance_record_sn(sk, prot, &ctx->tx); 296 297 for (i = 0; i < record->num_frags; i++) { 298 frag = &record->frags[i]; 299 sg_unmark_end(&offload_ctx->sg_tx_data[i]); 300 sg_set_page(&offload_ctx->sg_tx_data[i], skb_frag_page(frag), 301 skb_frag_size(frag), skb_frag_off(frag)); 302 sk_mem_charge(sk, skb_frag_size(frag)); 303 get_page(skb_frag_page(frag)); 304 } 305 sg_mark_end(&offload_ctx->sg_tx_data[record->num_frags - 1]); 306 307 /* all ready, send */ 308 return tls_push_sg(sk, ctx, offload_ctx->sg_tx_data, 0, flags); 309 } 310 311 static void tls_device_record_close(struct sock *sk, 312 struct tls_context *ctx, 313 struct tls_record_info *record, 314 struct page_frag *pfrag, 315 unsigned char record_type) 316 { 317 struct tls_prot_info *prot = &ctx->prot_info; 318 struct page_frag dummy_tag_frag; 319 320 /* append tag 321 * device will fill in the tag, we just need to append a placeholder 322 * use socket memory to improve coalescing (re-using a single buffer 323 * increases frag count) 324 * if we can't allocate memory now use the dummy page 325 */ 326 if (unlikely(pfrag->size - pfrag->offset < prot->tag_size) && 327 !skb_page_frag_refill(prot->tag_size, pfrag, sk->sk_allocation)) { 328 dummy_tag_frag.page = dummy_page; 329 dummy_tag_frag.offset = 0; 330 pfrag = &dummy_tag_frag; 331 } 332 tls_append_frag(record, pfrag, prot->tag_size); 333 334 /* fill prepend */ 335 tls_fill_prepend(ctx, skb_frag_address(&record->frags[0]), 336 record->len - prot->overhead_size, 337 record_type); 338 } 339 340 static int tls_create_new_record(struct tls_offload_context_tx *offload_ctx, 341 struct page_frag *pfrag, 342 size_t prepend_size) 343 { 344 struct tls_record_info *record; 345 skb_frag_t *frag; 346 347 record = kmalloc(sizeof(*record), GFP_KERNEL); 348 if (!record) 349 return -ENOMEM; 350 351 frag = &record->frags[0]; 352 skb_frag_fill_page_desc(frag, pfrag->page, pfrag->offset, 353 prepend_size); 354 355 get_page(pfrag->page); 356 pfrag->offset += prepend_size; 357 358 record->num_frags = 1; 359 record->len = prepend_size; 360 offload_ctx->open_record = record; 361 return 0; 362 } 363 364 static int tls_do_allocation(struct sock *sk, 365 struct tls_offload_context_tx *offload_ctx, 366 struct page_frag *pfrag, 367 size_t prepend_size) 368 { 369 int ret; 370 371 if (!offload_ctx->open_record) { 372 if (unlikely(!skb_page_frag_refill(prepend_size, pfrag, 373 sk->sk_allocation))) { 374 READ_ONCE(sk->sk_prot)->enter_memory_pressure(sk); 375 sk_stream_moderate_sndbuf(sk); 376 return -ENOMEM; 377 } 378 379 ret = tls_create_new_record(offload_ctx, pfrag, prepend_size); 380 if (ret) 381 return ret; 382 383 if (pfrag->size > pfrag->offset) 384 return 0; 385 } 386 387 if (!sk_page_frag_refill(sk, pfrag)) 388 return -ENOMEM; 389 390 return 0; 391 } 392 393 static int tls_device_copy_data(void *addr, size_t bytes, struct iov_iter *i) 394 { 395 size_t pre_copy, nocache; 396 397 pre_copy = ~((unsigned long)addr - 1) & (SMP_CACHE_BYTES - 1); 398 if (pre_copy) { 399 pre_copy = min(pre_copy, bytes); 400 if (copy_from_iter(addr, pre_copy, i) != pre_copy) 401 return -EFAULT; 402 bytes -= pre_copy; 403 addr += pre_copy; 404 } 405 406 nocache = round_down(bytes, SMP_CACHE_BYTES); 407 if (copy_from_iter_nocache(addr, nocache, i) != nocache) 408 return -EFAULT; 409 bytes -= nocache; 410 addr += nocache; 411 412 if (bytes && copy_from_iter(addr, bytes, i) != bytes) 413 return -EFAULT; 414 415 return 0; 416 } 417 418 static int tls_push_data(struct sock *sk, 419 struct iov_iter *iter, 420 size_t size, int flags, 421 unsigned char record_type) 422 { 423 struct tls_context *tls_ctx = tls_get_ctx(sk); 424 struct tls_prot_info *prot = &tls_ctx->prot_info; 425 struct tls_offload_context_tx *ctx = tls_offload_ctx_tx(tls_ctx); 426 struct tls_record_info *record; 427 int tls_push_record_flags; 428 struct page_frag *pfrag; 429 size_t orig_size = size; 430 u32 max_open_record_len; 431 bool more = false; 432 bool done = false; 433 int copy, rc = 0; 434 long timeo; 435 436 if (flags & 437 ~(MSG_MORE | MSG_DONTWAIT | MSG_NOSIGNAL | 438 MSG_SPLICE_PAGES | MSG_EOR)) 439 return -EOPNOTSUPP; 440 441 if ((flags & (MSG_MORE | MSG_EOR)) == (MSG_MORE | MSG_EOR)) 442 return -EINVAL; 443 444 if (unlikely(sk->sk_err)) 445 return -sk->sk_err; 446 447 flags |= MSG_SENDPAGE_DECRYPTED; 448 tls_push_record_flags = flags | MSG_MORE; 449 450 timeo = sock_sndtimeo(sk, flags & MSG_DONTWAIT); 451 if (tls_is_partially_sent_record(tls_ctx)) { 452 rc = tls_push_partial_record(sk, tls_ctx, flags); 453 if (rc < 0) 454 return rc; 455 } 456 457 pfrag = sk_page_frag(sk); 458 459 /* TLS_HEADER_SIZE is not counted as part of the TLS record, and 460 * we need to leave room for an authentication tag. 461 */ 462 max_open_record_len = TLS_MAX_PAYLOAD_SIZE + 463 prot->prepend_size; 464 do { 465 rc = tls_do_allocation(sk, ctx, pfrag, prot->prepend_size); 466 if (unlikely(rc)) { 467 rc = sk_stream_wait_memory(sk, &timeo); 468 if (!rc) 469 continue; 470 471 record = ctx->open_record; 472 if (!record) 473 break; 474 handle_error: 475 if (record_type != TLS_RECORD_TYPE_DATA) { 476 /* avoid sending partial 477 * record with type != 478 * application_data 479 */ 480 size = orig_size; 481 destroy_record(record); 482 ctx->open_record = NULL; 483 } else if (record->len > prot->prepend_size) { 484 goto last_record; 485 } 486 487 break; 488 } 489 490 record = ctx->open_record; 491 492 copy = min_t(size_t, size, max_open_record_len - record->len); 493 if (copy && (flags & MSG_SPLICE_PAGES)) { 494 struct page_frag zc_pfrag; 495 struct page **pages = &zc_pfrag.page; 496 size_t off; 497 498 rc = iov_iter_extract_pages(iter, &pages, 499 copy, 1, 0, &off); 500 if (rc <= 0) { 501 if (rc == 0) 502 rc = -EIO; 503 goto handle_error; 504 } 505 copy = rc; 506 507 if (WARN_ON_ONCE(!sendpage_ok(zc_pfrag.page))) { 508 iov_iter_revert(iter, copy); 509 rc = -EIO; 510 goto handle_error; 511 } 512 513 zc_pfrag.offset = off; 514 zc_pfrag.size = copy; 515 tls_append_frag(record, &zc_pfrag, copy); 516 } else if (copy) { 517 copy = min_t(size_t, copy, pfrag->size - pfrag->offset); 518 519 rc = tls_device_copy_data(page_address(pfrag->page) + 520 pfrag->offset, copy, 521 iter); 522 if (rc) 523 goto handle_error; 524 tls_append_frag(record, pfrag, copy); 525 } 526 527 size -= copy; 528 if (!size) { 529 last_record: 530 tls_push_record_flags = flags; 531 if (flags & MSG_MORE) { 532 more = true; 533 break; 534 } 535 536 done = true; 537 } 538 539 if (done || record->len >= max_open_record_len || 540 (record->num_frags >= MAX_SKB_FRAGS - 1)) { 541 tls_device_record_close(sk, tls_ctx, record, 542 pfrag, record_type); 543 544 rc = tls_push_record(sk, 545 tls_ctx, 546 ctx, 547 record, 548 tls_push_record_flags); 549 if (rc < 0) 550 break; 551 } 552 } while (!done); 553 554 tls_ctx->pending_open_record_frags = more; 555 556 if (orig_size - size > 0) 557 rc = orig_size - size; 558 559 return rc; 560 } 561 562 int tls_device_sendmsg(struct sock *sk, struct msghdr *msg, size_t size) 563 { 564 unsigned char record_type = TLS_RECORD_TYPE_DATA; 565 struct tls_context *tls_ctx = tls_get_ctx(sk); 566 int rc; 567 568 if (!tls_ctx->zerocopy_sendfile) 569 msg->msg_flags &= ~MSG_SPLICE_PAGES; 570 571 mutex_lock(&tls_ctx->tx_lock); 572 lock_sock(sk); 573 574 if (unlikely(msg->msg_controllen)) { 575 rc = tls_process_cmsg(sk, msg, &record_type); 576 if (rc) 577 goto out; 578 } 579 580 rc = tls_push_data(sk, &msg->msg_iter, size, msg->msg_flags, 581 record_type); 582 583 out: 584 release_sock(sk); 585 mutex_unlock(&tls_ctx->tx_lock); 586 return rc; 587 } 588 589 void tls_device_splice_eof(struct socket *sock) 590 { 591 struct sock *sk = sock->sk; 592 struct tls_context *tls_ctx = tls_get_ctx(sk); 593 struct iov_iter iter = {}; 594 595 if (!tls_is_partially_sent_record(tls_ctx)) 596 return; 597 598 mutex_lock(&tls_ctx->tx_lock); 599 lock_sock(sk); 600 601 if (tls_is_partially_sent_record(tls_ctx)) { 602 iov_iter_bvec(&iter, ITER_SOURCE, NULL, 0, 0); 603 tls_push_data(sk, &iter, 0, 0, TLS_RECORD_TYPE_DATA); 604 } 605 606 release_sock(sk); 607 mutex_unlock(&tls_ctx->tx_lock); 608 } 609 610 struct tls_record_info *tls_get_record(struct tls_offload_context_tx *context, 611 u32 seq, u64 *p_record_sn) 612 { 613 u64 record_sn = context->hint_record_sn; 614 struct tls_record_info *info, *last; 615 616 info = context->retransmit_hint; 617 if (!info || 618 before(seq, info->end_seq - info->len)) { 619 /* if retransmit_hint is irrelevant start 620 * from the beginning of the list 621 */ 622 info = list_first_entry_or_null(&context->records_list, 623 struct tls_record_info, list); 624 if (!info) 625 return NULL; 626 /* send the start_marker record if seq number is before the 627 * tls offload start marker sequence number. This record is 628 * required to handle TCP packets which are before TLS offload 629 * started. 630 * And if it's not start marker, look if this seq number 631 * belongs to the list. 632 */ 633 if (likely(!tls_record_is_start_marker(info))) { 634 /* we have the first record, get the last record to see 635 * if this seq number belongs to the list. 636 */ 637 last = list_last_entry(&context->records_list, 638 struct tls_record_info, list); 639 640 if (!between(seq, tls_record_start_seq(info), 641 last->end_seq)) 642 return NULL; 643 } 644 record_sn = context->unacked_record_sn; 645 } 646 647 /* We just need the _rcu for the READ_ONCE() */ 648 rcu_read_lock(); 649 list_for_each_entry_from_rcu(info, &context->records_list, list) { 650 if (before(seq, info->end_seq)) { 651 if (!context->retransmit_hint || 652 after(info->end_seq, 653 context->retransmit_hint->end_seq)) { 654 context->hint_record_sn = record_sn; 655 context->retransmit_hint = info; 656 } 657 *p_record_sn = record_sn; 658 goto exit_rcu_unlock; 659 } 660 record_sn++; 661 } 662 info = NULL; 663 664 exit_rcu_unlock: 665 rcu_read_unlock(); 666 return info; 667 } 668 EXPORT_SYMBOL(tls_get_record); 669 670 static int tls_device_push_pending_record(struct sock *sk, int flags) 671 { 672 struct iov_iter iter; 673 674 iov_iter_kvec(&iter, ITER_SOURCE, NULL, 0, 0); 675 return tls_push_data(sk, &iter, 0, flags, TLS_RECORD_TYPE_DATA); 676 } 677 678 void tls_device_write_space(struct sock *sk, struct tls_context *ctx) 679 { 680 if (tls_is_partially_sent_record(ctx)) { 681 gfp_t sk_allocation = sk->sk_allocation; 682 683 WARN_ON_ONCE(sk->sk_write_pending); 684 685 sk->sk_allocation = GFP_ATOMIC; 686 tls_push_partial_record(sk, ctx, 687 MSG_DONTWAIT | MSG_NOSIGNAL | 688 MSG_SENDPAGE_DECRYPTED); 689 sk->sk_allocation = sk_allocation; 690 } 691 } 692 693 static void tls_device_resync_rx(struct tls_context *tls_ctx, 694 struct sock *sk, u32 seq, u8 *rcd_sn) 695 { 696 struct tls_offload_context_rx *rx_ctx = tls_offload_ctx_rx(tls_ctx); 697 struct net_device *netdev; 698 699 trace_tls_device_rx_resync_send(sk, seq, rcd_sn, rx_ctx->resync_type); 700 rcu_read_lock(); 701 netdev = rcu_dereference(tls_ctx->netdev); 702 if (netdev) 703 netdev->tlsdev_ops->tls_dev_resync(netdev, sk, seq, rcd_sn, 704 TLS_OFFLOAD_CTX_DIR_RX); 705 rcu_read_unlock(); 706 TLS_INC_STATS(sock_net(sk), LINUX_MIB_TLSRXDEVICERESYNC); 707 } 708 709 static bool 710 tls_device_rx_resync_async(struct tls_offload_resync_async *resync_async, 711 s64 resync_req, u32 *seq, u16 *rcd_delta) 712 { 713 u32 is_async = resync_req & RESYNC_REQ_ASYNC; 714 u32 req_seq = resync_req >> 32; 715 u32 req_end = req_seq + ((resync_req >> 16) & 0xffff); 716 u16 i; 717 718 *rcd_delta = 0; 719 720 if (is_async) { 721 /* shouldn't get to wraparound: 722 * too long in async stage, something bad happened 723 */ 724 if (WARN_ON_ONCE(resync_async->rcd_delta == USHRT_MAX)) 725 return false; 726 727 /* asynchronous stage: log all headers seq such that 728 * req_seq <= seq <= end_seq, and wait for real resync request 729 */ 730 if (before(*seq, req_seq)) 731 return false; 732 if (!after(*seq, req_end) && 733 resync_async->loglen < TLS_DEVICE_RESYNC_ASYNC_LOGMAX) 734 resync_async->log[resync_async->loglen++] = *seq; 735 736 resync_async->rcd_delta++; 737 738 return false; 739 } 740 741 /* synchronous stage: check against the logged entries and 742 * proceed to check the next entries if no match was found 743 */ 744 for (i = 0; i < resync_async->loglen; i++) 745 if (req_seq == resync_async->log[i] && 746 atomic64_try_cmpxchg(&resync_async->req, &resync_req, 0)) { 747 *rcd_delta = resync_async->rcd_delta - i; 748 *seq = req_seq; 749 resync_async->loglen = 0; 750 resync_async->rcd_delta = 0; 751 return true; 752 } 753 754 resync_async->loglen = 0; 755 resync_async->rcd_delta = 0; 756 757 if (req_seq == *seq && 758 atomic64_try_cmpxchg(&resync_async->req, 759 &resync_req, 0)) 760 return true; 761 762 return false; 763 } 764 765 void tls_device_rx_resync_new_rec(struct sock *sk, u32 rcd_len, u32 seq) 766 { 767 struct tls_context *tls_ctx = tls_get_ctx(sk); 768 struct tls_offload_context_rx *rx_ctx; 769 u8 rcd_sn[TLS_MAX_REC_SEQ_SIZE]; 770 u32 sock_data, is_req_pending; 771 struct tls_prot_info *prot; 772 s64 resync_req; 773 u16 rcd_delta; 774 u32 req_seq; 775 776 if (tls_ctx->rx_conf != TLS_HW) 777 return; 778 if (unlikely(test_bit(TLS_RX_DEV_DEGRADED, &tls_ctx->flags))) 779 return; 780 781 prot = &tls_ctx->prot_info; 782 rx_ctx = tls_offload_ctx_rx(tls_ctx); 783 memcpy(rcd_sn, tls_ctx->rx.rec_seq, prot->rec_seq_size); 784 785 switch (rx_ctx->resync_type) { 786 case TLS_OFFLOAD_SYNC_TYPE_DRIVER_REQ: 787 resync_req = atomic64_read(&rx_ctx->resync_req); 788 req_seq = resync_req >> 32; 789 seq += TLS_HEADER_SIZE - 1; 790 is_req_pending = resync_req; 791 792 if (likely(!is_req_pending) || req_seq != seq || 793 !atomic64_try_cmpxchg(&rx_ctx->resync_req, &resync_req, 0)) 794 return; 795 break; 796 case TLS_OFFLOAD_SYNC_TYPE_CORE_NEXT_HINT: 797 if (likely(!rx_ctx->resync_nh_do_now)) 798 return; 799 800 /* head of next rec is already in, note that the sock_inq will 801 * include the currently parsed message when called from parser 802 */ 803 sock_data = tcp_inq(sk); 804 if (sock_data > rcd_len) { 805 trace_tls_device_rx_resync_nh_delay(sk, sock_data, 806 rcd_len); 807 return; 808 } 809 810 rx_ctx->resync_nh_do_now = 0; 811 seq += rcd_len; 812 tls_bigint_increment(rcd_sn, prot->rec_seq_size); 813 break; 814 case TLS_OFFLOAD_SYNC_TYPE_DRIVER_REQ_ASYNC: 815 resync_req = atomic64_read(&rx_ctx->resync_async->req); 816 is_req_pending = resync_req; 817 if (likely(!is_req_pending)) 818 return; 819 820 if (!tls_device_rx_resync_async(rx_ctx->resync_async, 821 resync_req, &seq, &rcd_delta)) 822 return; 823 tls_bigint_subtract(rcd_sn, rcd_delta); 824 break; 825 } 826 827 tls_device_resync_rx(tls_ctx, sk, seq, rcd_sn); 828 } 829 830 static void tls_device_core_ctrl_rx_resync(struct tls_context *tls_ctx, 831 struct tls_offload_context_rx *ctx, 832 struct sock *sk, struct sk_buff *skb) 833 { 834 struct strp_msg *rxm; 835 836 /* device will request resyncs by itself based on stream scan */ 837 if (ctx->resync_type != TLS_OFFLOAD_SYNC_TYPE_CORE_NEXT_HINT) 838 return; 839 /* already scheduled */ 840 if (ctx->resync_nh_do_now) 841 return; 842 /* seen decrypted fragments since last fully-failed record */ 843 if (ctx->resync_nh_reset) { 844 ctx->resync_nh_reset = 0; 845 ctx->resync_nh.decrypted_failed = 1; 846 ctx->resync_nh.decrypted_tgt = TLS_DEVICE_RESYNC_NH_START_IVAL; 847 return; 848 } 849 850 if (++ctx->resync_nh.decrypted_failed <= ctx->resync_nh.decrypted_tgt) 851 return; 852 853 /* doing resync, bump the next target in case it fails */ 854 if (ctx->resync_nh.decrypted_tgt < TLS_DEVICE_RESYNC_NH_MAX_IVAL) 855 ctx->resync_nh.decrypted_tgt *= 2; 856 else 857 ctx->resync_nh.decrypted_tgt += TLS_DEVICE_RESYNC_NH_MAX_IVAL; 858 859 rxm = strp_msg(skb); 860 861 /* head of next rec is already in, parser will sync for us */ 862 if (tcp_inq(sk) > rxm->full_len) { 863 trace_tls_device_rx_resync_nh_schedule(sk); 864 ctx->resync_nh_do_now = 1; 865 } else { 866 struct tls_prot_info *prot = &tls_ctx->prot_info; 867 u8 rcd_sn[TLS_MAX_REC_SEQ_SIZE]; 868 869 memcpy(rcd_sn, tls_ctx->rx.rec_seq, prot->rec_seq_size); 870 tls_bigint_increment(rcd_sn, prot->rec_seq_size); 871 872 tls_device_resync_rx(tls_ctx, sk, tcp_sk(sk)->copied_seq, 873 rcd_sn); 874 } 875 } 876 877 static int 878 tls_device_reencrypt(struct sock *sk, struct tls_context *tls_ctx) 879 { 880 struct tls_sw_context_rx *sw_ctx = tls_sw_ctx_rx(tls_ctx); 881 const struct tls_cipher_desc *cipher_desc; 882 int err, offset, copy, data_len, pos; 883 struct sk_buff *skb, *skb_iter; 884 struct scatterlist sg[1]; 885 struct strp_msg *rxm; 886 char *orig_buf, *buf; 887 888 cipher_desc = get_cipher_desc(tls_ctx->crypto_recv.info.cipher_type); 889 DEBUG_NET_WARN_ON_ONCE(!cipher_desc || !cipher_desc->offloadable); 890 891 rxm = strp_msg(tls_strp_msg(sw_ctx)); 892 orig_buf = kmalloc(rxm->full_len + TLS_HEADER_SIZE + cipher_desc->iv, 893 sk->sk_allocation); 894 if (!orig_buf) 895 return -ENOMEM; 896 buf = orig_buf; 897 898 err = tls_strp_msg_cow(sw_ctx); 899 if (unlikely(err)) 900 goto free_buf; 901 902 skb = tls_strp_msg(sw_ctx); 903 rxm = strp_msg(skb); 904 offset = rxm->offset; 905 906 sg_init_table(sg, 1); 907 sg_set_buf(&sg[0], buf, 908 rxm->full_len + TLS_HEADER_SIZE + cipher_desc->iv); 909 err = skb_copy_bits(skb, offset, buf, TLS_HEADER_SIZE + cipher_desc->iv); 910 if (err) 911 goto free_buf; 912 913 /* We are interested only in the decrypted data not the auth */ 914 err = decrypt_skb(sk, sg); 915 if (err != -EBADMSG) 916 goto free_buf; 917 else 918 err = 0; 919 920 data_len = rxm->full_len - cipher_desc->tag; 921 922 if (skb_pagelen(skb) > offset) { 923 copy = min_t(int, skb_pagelen(skb) - offset, data_len); 924 925 if (skb->decrypted) { 926 err = skb_store_bits(skb, offset, buf, copy); 927 if (err) 928 goto free_buf; 929 } 930 931 offset += copy; 932 buf += copy; 933 } 934 935 pos = skb_pagelen(skb); 936 skb_walk_frags(skb, skb_iter) { 937 int frag_pos; 938 939 /* Practically all frags must belong to msg if reencrypt 940 * is needed with current strparser and coalescing logic, 941 * but strparser may "get optimized", so let's be safe. 942 */ 943 if (pos + skb_iter->len <= offset) 944 goto done_with_frag; 945 if (pos >= data_len + rxm->offset) 946 break; 947 948 frag_pos = offset - pos; 949 copy = min_t(int, skb_iter->len - frag_pos, 950 data_len + rxm->offset - offset); 951 952 if (skb_iter->decrypted) { 953 err = skb_store_bits(skb_iter, frag_pos, buf, copy); 954 if (err) 955 goto free_buf; 956 } 957 958 offset += copy; 959 buf += copy; 960 done_with_frag: 961 pos += skb_iter->len; 962 } 963 964 free_buf: 965 kfree(orig_buf); 966 return err; 967 } 968 969 int tls_device_decrypted(struct sock *sk, struct tls_context *tls_ctx) 970 { 971 struct tls_offload_context_rx *ctx = tls_offload_ctx_rx(tls_ctx); 972 struct tls_sw_context_rx *sw_ctx = tls_sw_ctx_rx(tls_ctx); 973 struct sk_buff *skb = tls_strp_msg(sw_ctx); 974 struct strp_msg *rxm = strp_msg(skb); 975 int is_decrypted, is_encrypted; 976 977 if (!tls_strp_msg_mixed_decrypted(sw_ctx)) { 978 is_decrypted = skb->decrypted; 979 is_encrypted = !is_decrypted; 980 } else { 981 is_decrypted = 0; 982 is_encrypted = 0; 983 } 984 985 trace_tls_device_decrypted(sk, tcp_sk(sk)->copied_seq - rxm->full_len, 986 tls_ctx->rx.rec_seq, rxm->full_len, 987 is_encrypted, is_decrypted); 988 989 if (unlikely(test_bit(TLS_RX_DEV_DEGRADED, &tls_ctx->flags))) { 990 if (likely(is_encrypted || is_decrypted)) 991 return is_decrypted; 992 993 /* After tls_device_down disables the offload, the next SKB will 994 * likely have initial fragments decrypted, and final ones not 995 * decrypted. We need to reencrypt that single SKB. 996 */ 997 return tls_device_reencrypt(sk, tls_ctx); 998 } 999 1000 /* Return immediately if the record is either entirely plaintext or 1001 * entirely ciphertext. Otherwise handle reencrypt partially decrypted 1002 * record. 1003 */ 1004 if (is_decrypted) { 1005 ctx->resync_nh_reset = 1; 1006 return is_decrypted; 1007 } 1008 if (is_encrypted) { 1009 tls_device_core_ctrl_rx_resync(tls_ctx, ctx, sk, skb); 1010 return 0; 1011 } 1012 1013 ctx->resync_nh_reset = 1; 1014 return tls_device_reencrypt(sk, tls_ctx); 1015 } 1016 1017 static void tls_device_attach(struct tls_context *ctx, struct sock *sk, 1018 struct net_device *netdev) 1019 { 1020 if (sk->sk_destruct != tls_device_sk_destruct) { 1021 refcount_set(&ctx->refcount, 1); 1022 dev_hold(netdev); 1023 RCU_INIT_POINTER(ctx->netdev, netdev); 1024 spin_lock_irq(&tls_device_lock); 1025 list_add_tail(&ctx->list, &tls_device_list); 1026 spin_unlock_irq(&tls_device_lock); 1027 1028 ctx->sk_destruct = sk->sk_destruct; 1029 smp_store_release(&sk->sk_destruct, tls_device_sk_destruct); 1030 } 1031 } 1032 1033 static struct tls_offload_context_tx *alloc_offload_ctx_tx(struct tls_context *ctx) 1034 { 1035 struct tls_offload_context_tx *offload_ctx; 1036 __be64 rcd_sn; 1037 1038 offload_ctx = kzalloc(sizeof(*offload_ctx), GFP_KERNEL); 1039 if (!offload_ctx) 1040 return NULL; 1041 1042 INIT_WORK(&offload_ctx->destruct_work, tls_device_tx_del_task); 1043 INIT_LIST_HEAD(&offload_ctx->records_list); 1044 spin_lock_init(&offload_ctx->lock); 1045 sg_init_table(offload_ctx->sg_tx_data, 1046 ARRAY_SIZE(offload_ctx->sg_tx_data)); 1047 1048 /* start at rec_seq - 1 to account for the start marker record */ 1049 memcpy(&rcd_sn, ctx->tx.rec_seq, sizeof(rcd_sn)); 1050 offload_ctx->unacked_record_sn = be64_to_cpu(rcd_sn) - 1; 1051 1052 offload_ctx->ctx = ctx; 1053 1054 return offload_ctx; 1055 } 1056 1057 int tls_set_device_offload(struct sock *sk) 1058 { 1059 struct tls_record_info *start_marker_record; 1060 struct tls_offload_context_tx *offload_ctx; 1061 const struct tls_cipher_desc *cipher_desc; 1062 struct tls_crypto_info *crypto_info; 1063 struct tls_prot_info *prot; 1064 struct net_device *netdev; 1065 struct tls_context *ctx; 1066 char *iv, *rec_seq; 1067 int rc; 1068 1069 ctx = tls_get_ctx(sk); 1070 prot = &ctx->prot_info; 1071 1072 if (ctx->priv_ctx_tx) 1073 return -EEXIST; 1074 1075 netdev = get_netdev_for_sock(sk); 1076 if (!netdev) { 1077 pr_err_ratelimited("%s: netdev not found\n", __func__); 1078 return -EINVAL; 1079 } 1080 1081 if (!(netdev->features & NETIF_F_HW_TLS_TX)) { 1082 rc = -EOPNOTSUPP; 1083 goto release_netdev; 1084 } 1085 1086 crypto_info = &ctx->crypto_send.info; 1087 if (crypto_info->version != TLS_1_2_VERSION) { 1088 rc = -EOPNOTSUPP; 1089 goto release_netdev; 1090 } 1091 1092 cipher_desc = get_cipher_desc(crypto_info->cipher_type); 1093 if (!cipher_desc || !cipher_desc->offloadable) { 1094 rc = -EINVAL; 1095 goto release_netdev; 1096 } 1097 1098 rc = init_prot_info(prot, crypto_info, cipher_desc); 1099 if (rc) 1100 goto release_netdev; 1101 1102 iv = crypto_info_iv(crypto_info, cipher_desc); 1103 rec_seq = crypto_info_rec_seq(crypto_info, cipher_desc); 1104 1105 memcpy(ctx->tx.iv + cipher_desc->salt, iv, cipher_desc->iv); 1106 memcpy(ctx->tx.rec_seq, rec_seq, cipher_desc->rec_seq); 1107 1108 start_marker_record = kmalloc(sizeof(*start_marker_record), GFP_KERNEL); 1109 if (!start_marker_record) { 1110 rc = -ENOMEM; 1111 goto release_netdev; 1112 } 1113 1114 offload_ctx = alloc_offload_ctx_tx(ctx); 1115 if (!offload_ctx) { 1116 rc = -ENOMEM; 1117 goto free_marker_record; 1118 } 1119 1120 rc = tls_sw_fallback_init(sk, offload_ctx, crypto_info); 1121 if (rc) 1122 goto free_offload_ctx; 1123 1124 start_marker_record->end_seq = tcp_sk(sk)->write_seq; 1125 start_marker_record->len = 0; 1126 start_marker_record->num_frags = 0; 1127 list_add_tail(&start_marker_record->list, &offload_ctx->records_list); 1128 1129 clean_acked_data_enable(inet_csk(sk), &tls_icsk_clean_acked); 1130 ctx->push_pending_record = tls_device_push_pending_record; 1131 1132 /* TLS offload is greatly simplified if we don't send 1133 * SKBs where only part of the payload needs to be encrypted. 1134 * So mark the last skb in the write queue as end of record. 1135 */ 1136 tcp_write_collapse_fence(sk); 1137 1138 /* Avoid offloading if the device is down 1139 * We don't want to offload new flows after 1140 * the NETDEV_DOWN event 1141 * 1142 * device_offload_lock is taken in tls_devices's NETDEV_DOWN 1143 * handler thus protecting from the device going down before 1144 * ctx was added to tls_device_list. 1145 */ 1146 down_read(&device_offload_lock); 1147 if (!(netdev->flags & IFF_UP)) { 1148 rc = -EINVAL; 1149 goto release_lock; 1150 } 1151 1152 ctx->priv_ctx_tx = offload_ctx; 1153 rc = netdev->tlsdev_ops->tls_dev_add(netdev, sk, TLS_OFFLOAD_CTX_DIR_TX, 1154 &ctx->crypto_send.info, 1155 tcp_sk(sk)->write_seq); 1156 trace_tls_device_offload_set(sk, TLS_OFFLOAD_CTX_DIR_TX, 1157 tcp_sk(sk)->write_seq, rec_seq, rc); 1158 if (rc) 1159 goto release_lock; 1160 1161 tls_device_attach(ctx, sk, netdev); 1162 up_read(&device_offload_lock); 1163 1164 /* following this assignment tls_is_skb_tx_device_offloaded 1165 * will return true and the context might be accessed 1166 * by the netdev's xmit function. 1167 */ 1168 smp_store_release(&sk->sk_validate_xmit_skb, tls_validate_xmit_skb); 1169 dev_put(netdev); 1170 1171 return 0; 1172 1173 release_lock: 1174 up_read(&device_offload_lock); 1175 clean_acked_data_disable(inet_csk(sk)); 1176 crypto_free_aead(offload_ctx->aead_send); 1177 free_offload_ctx: 1178 kfree(offload_ctx); 1179 ctx->priv_ctx_tx = NULL; 1180 free_marker_record: 1181 kfree(start_marker_record); 1182 release_netdev: 1183 dev_put(netdev); 1184 return rc; 1185 } 1186 1187 int tls_set_device_offload_rx(struct sock *sk, struct tls_context *ctx) 1188 { 1189 struct tls12_crypto_info_aes_gcm_128 *info; 1190 struct tls_offload_context_rx *context; 1191 struct net_device *netdev; 1192 int rc = 0; 1193 1194 if (ctx->crypto_recv.info.version != TLS_1_2_VERSION) 1195 return -EOPNOTSUPP; 1196 1197 netdev = get_netdev_for_sock(sk); 1198 if (!netdev) { 1199 pr_err_ratelimited("%s: netdev not found\n", __func__); 1200 return -EINVAL; 1201 } 1202 1203 if (!(netdev->features & NETIF_F_HW_TLS_RX)) { 1204 rc = -EOPNOTSUPP; 1205 goto release_netdev; 1206 } 1207 1208 /* Avoid offloading if the device is down 1209 * We don't want to offload new flows after 1210 * the NETDEV_DOWN event 1211 * 1212 * device_offload_lock is taken in tls_devices's NETDEV_DOWN 1213 * handler thus protecting from the device going down before 1214 * ctx was added to tls_device_list. 1215 */ 1216 down_read(&device_offload_lock); 1217 if (!(netdev->flags & IFF_UP)) { 1218 rc = -EINVAL; 1219 goto release_lock; 1220 } 1221 1222 context = kzalloc(sizeof(*context), GFP_KERNEL); 1223 if (!context) { 1224 rc = -ENOMEM; 1225 goto release_lock; 1226 } 1227 context->resync_nh_reset = 1; 1228 1229 ctx->priv_ctx_rx = context; 1230 rc = tls_set_sw_offload(sk, 0); 1231 if (rc) 1232 goto release_ctx; 1233 1234 rc = netdev->tlsdev_ops->tls_dev_add(netdev, sk, TLS_OFFLOAD_CTX_DIR_RX, 1235 &ctx->crypto_recv.info, 1236 tcp_sk(sk)->copied_seq); 1237 info = (void *)&ctx->crypto_recv.info; 1238 trace_tls_device_offload_set(sk, TLS_OFFLOAD_CTX_DIR_RX, 1239 tcp_sk(sk)->copied_seq, info->rec_seq, rc); 1240 if (rc) 1241 goto free_sw_resources; 1242 1243 tls_device_attach(ctx, sk, netdev); 1244 up_read(&device_offload_lock); 1245 1246 dev_put(netdev); 1247 1248 return 0; 1249 1250 free_sw_resources: 1251 up_read(&device_offload_lock); 1252 tls_sw_free_resources_rx(sk); 1253 down_read(&device_offload_lock); 1254 release_ctx: 1255 ctx->priv_ctx_rx = NULL; 1256 release_lock: 1257 up_read(&device_offload_lock); 1258 release_netdev: 1259 dev_put(netdev); 1260 return rc; 1261 } 1262 1263 void tls_device_offload_cleanup_rx(struct sock *sk) 1264 { 1265 struct tls_context *tls_ctx = tls_get_ctx(sk); 1266 struct net_device *netdev; 1267 1268 down_read(&device_offload_lock); 1269 netdev = rcu_dereference_protected(tls_ctx->netdev, 1270 lockdep_is_held(&device_offload_lock)); 1271 if (!netdev) 1272 goto out; 1273 1274 netdev->tlsdev_ops->tls_dev_del(netdev, tls_ctx, 1275 TLS_OFFLOAD_CTX_DIR_RX); 1276 1277 if (tls_ctx->tx_conf != TLS_HW) { 1278 dev_put(netdev); 1279 rcu_assign_pointer(tls_ctx->netdev, NULL); 1280 } else { 1281 set_bit(TLS_RX_DEV_CLOSED, &tls_ctx->flags); 1282 } 1283 out: 1284 up_read(&device_offload_lock); 1285 tls_sw_release_resources_rx(sk); 1286 } 1287 1288 static int tls_device_down(struct net_device *netdev) 1289 { 1290 struct tls_context *ctx, *tmp; 1291 unsigned long flags; 1292 LIST_HEAD(list); 1293 1294 /* Request a write lock to block new offload attempts */ 1295 down_write(&device_offload_lock); 1296 1297 spin_lock_irqsave(&tls_device_lock, flags); 1298 list_for_each_entry_safe(ctx, tmp, &tls_device_list, list) { 1299 struct net_device *ctx_netdev = 1300 rcu_dereference_protected(ctx->netdev, 1301 lockdep_is_held(&device_offload_lock)); 1302 1303 if (ctx_netdev != netdev || 1304 !refcount_inc_not_zero(&ctx->refcount)) 1305 continue; 1306 1307 list_move(&ctx->list, &list); 1308 } 1309 spin_unlock_irqrestore(&tls_device_lock, flags); 1310 1311 list_for_each_entry_safe(ctx, tmp, &list, list) { 1312 /* Stop offloaded TX and switch to the fallback. 1313 * tls_is_skb_tx_device_offloaded will return false. 1314 */ 1315 WRITE_ONCE(ctx->sk->sk_validate_xmit_skb, tls_validate_xmit_skb_sw); 1316 1317 /* Stop the RX and TX resync. 1318 * tls_dev_resync must not be called after tls_dev_del. 1319 */ 1320 rcu_assign_pointer(ctx->netdev, NULL); 1321 1322 /* Start skipping the RX resync logic completely. */ 1323 set_bit(TLS_RX_DEV_DEGRADED, &ctx->flags); 1324 1325 /* Sync with inflight packets. After this point: 1326 * TX: no non-encrypted packets will be passed to the driver. 1327 * RX: resync requests from the driver will be ignored. 1328 */ 1329 synchronize_net(); 1330 1331 /* Release the offload context on the driver side. */ 1332 if (ctx->tx_conf == TLS_HW) 1333 netdev->tlsdev_ops->tls_dev_del(netdev, ctx, 1334 TLS_OFFLOAD_CTX_DIR_TX); 1335 if (ctx->rx_conf == TLS_HW && 1336 !test_bit(TLS_RX_DEV_CLOSED, &ctx->flags)) 1337 netdev->tlsdev_ops->tls_dev_del(netdev, ctx, 1338 TLS_OFFLOAD_CTX_DIR_RX); 1339 1340 dev_put(netdev); 1341 1342 /* Move the context to a separate list for two reasons: 1343 * 1. When the context is deallocated, list_del is called. 1344 * 2. It's no longer an offloaded context, so we don't want to 1345 * run offload-specific code on this context. 1346 */ 1347 spin_lock_irqsave(&tls_device_lock, flags); 1348 list_move_tail(&ctx->list, &tls_device_down_list); 1349 spin_unlock_irqrestore(&tls_device_lock, flags); 1350 1351 /* Device contexts for RX and TX will be freed in on sk_destruct 1352 * by tls_device_free_ctx. rx_conf and tx_conf stay in TLS_HW. 1353 * Now release the ref taken above. 1354 */ 1355 if (refcount_dec_and_test(&ctx->refcount)) { 1356 /* sk_destruct ran after tls_device_down took a ref, and 1357 * it returned early. Complete the destruction here. 1358 */ 1359 list_del(&ctx->list); 1360 tls_device_free_ctx(ctx); 1361 } 1362 } 1363 1364 up_write(&device_offload_lock); 1365 1366 flush_workqueue(destruct_wq); 1367 1368 return NOTIFY_DONE; 1369 } 1370 1371 static int tls_dev_event(struct notifier_block *this, unsigned long event, 1372 void *ptr) 1373 { 1374 struct net_device *dev = netdev_notifier_info_to_dev(ptr); 1375 1376 if (!dev->tlsdev_ops && 1377 !(dev->features & (NETIF_F_HW_TLS_RX | NETIF_F_HW_TLS_TX))) 1378 return NOTIFY_DONE; 1379 1380 switch (event) { 1381 case NETDEV_REGISTER: 1382 case NETDEV_FEAT_CHANGE: 1383 if (netif_is_bond_master(dev)) 1384 return NOTIFY_DONE; 1385 if ((dev->features & NETIF_F_HW_TLS_RX) && 1386 !dev->tlsdev_ops->tls_dev_resync) 1387 return NOTIFY_BAD; 1388 1389 if (dev->tlsdev_ops && 1390 dev->tlsdev_ops->tls_dev_add && 1391 dev->tlsdev_ops->tls_dev_del) 1392 return NOTIFY_DONE; 1393 else 1394 return NOTIFY_BAD; 1395 case NETDEV_DOWN: 1396 return tls_device_down(dev); 1397 } 1398 return NOTIFY_DONE; 1399 } 1400 1401 static struct notifier_block tls_dev_notifier = { 1402 .notifier_call = tls_dev_event, 1403 }; 1404 1405 int __init tls_device_init(void) 1406 { 1407 int err; 1408 1409 dummy_page = alloc_page(GFP_KERNEL); 1410 if (!dummy_page) 1411 return -ENOMEM; 1412 1413 destruct_wq = alloc_workqueue("ktls_device_destruct", 0, 0); 1414 if (!destruct_wq) { 1415 err = -ENOMEM; 1416 goto err_free_dummy; 1417 } 1418 1419 err = register_netdevice_notifier(&tls_dev_notifier); 1420 if (err) 1421 goto err_destroy_wq; 1422 1423 return 0; 1424 1425 err_destroy_wq: 1426 destroy_workqueue(destruct_wq); 1427 err_free_dummy: 1428 put_page(dummy_page); 1429 return err; 1430 } 1431 1432 void __exit tls_device_cleanup(void) 1433 { 1434 unregister_netdevice_notifier(&tls_dev_notifier); 1435 destroy_workqueue(destruct_wq); 1436 clean_acked_data_flush(); 1437 put_page(dummy_page); 1438 } 1439