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 /* device_offload_lock is used to synchronize tls_dev_add 42 * against NETDEV_DOWN notifications. 43 */ 44 static DECLARE_RWSEM(device_offload_lock); 45 46 static void tls_device_gc_task(struct work_struct *work); 47 48 static DECLARE_WORK(tls_device_gc_work, tls_device_gc_task); 49 static LIST_HEAD(tls_device_gc_list); 50 static LIST_HEAD(tls_device_list); 51 static DEFINE_SPINLOCK(tls_device_lock); 52 53 static void tls_device_free_ctx(struct tls_context *ctx) 54 { 55 if (ctx->tx_conf == TLS_HW) 56 kfree(tls_offload_ctx_tx(ctx)); 57 58 if (ctx->rx_conf == TLS_HW) 59 kfree(tls_offload_ctx_rx(ctx)); 60 61 kfree(ctx); 62 } 63 64 static void tls_device_gc_task(struct work_struct *work) 65 { 66 struct tls_context *ctx, *tmp; 67 unsigned long flags; 68 LIST_HEAD(gc_list); 69 70 spin_lock_irqsave(&tls_device_lock, flags); 71 list_splice_init(&tls_device_gc_list, &gc_list); 72 spin_unlock_irqrestore(&tls_device_lock, flags); 73 74 list_for_each_entry_safe(ctx, tmp, &gc_list, list) { 75 struct net_device *netdev = ctx->netdev; 76 77 if (netdev && ctx->tx_conf == TLS_HW) { 78 netdev->tlsdev_ops->tls_dev_del(netdev, ctx, 79 TLS_OFFLOAD_CTX_DIR_TX); 80 dev_put(netdev); 81 ctx->netdev = NULL; 82 } 83 84 list_del(&ctx->list); 85 tls_device_free_ctx(ctx); 86 } 87 } 88 89 static void tls_device_attach(struct tls_context *ctx, struct sock *sk, 90 struct net_device *netdev) 91 { 92 if (sk->sk_destruct != tls_device_sk_destruct) { 93 refcount_set(&ctx->refcount, 1); 94 dev_hold(netdev); 95 ctx->netdev = netdev; 96 spin_lock_irq(&tls_device_lock); 97 list_add_tail(&ctx->list, &tls_device_list); 98 spin_unlock_irq(&tls_device_lock); 99 100 ctx->sk_destruct = sk->sk_destruct; 101 sk->sk_destruct = tls_device_sk_destruct; 102 } 103 } 104 105 static void tls_device_queue_ctx_destruction(struct tls_context *ctx) 106 { 107 unsigned long flags; 108 109 spin_lock_irqsave(&tls_device_lock, flags); 110 list_move_tail(&ctx->list, &tls_device_gc_list); 111 112 /* schedule_work inside the spinlock 113 * to make sure tls_device_down waits for that work. 114 */ 115 schedule_work(&tls_device_gc_work); 116 117 spin_unlock_irqrestore(&tls_device_lock, flags); 118 } 119 120 /* We assume that the socket is already connected */ 121 static struct net_device *get_netdev_for_sock(struct sock *sk) 122 { 123 struct dst_entry *dst = sk_dst_get(sk); 124 struct net_device *netdev = NULL; 125 126 if (likely(dst)) { 127 netdev = dst->dev; 128 dev_hold(netdev); 129 } 130 131 dst_release(dst); 132 133 return netdev; 134 } 135 136 static void destroy_record(struct tls_record_info *record) 137 { 138 int nr_frags = record->num_frags; 139 skb_frag_t *frag; 140 141 while (nr_frags-- > 0) { 142 frag = &record->frags[nr_frags]; 143 __skb_frag_unref(frag); 144 } 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 list_del(&info->list); 178 destroy_record(info); 179 deleted_records++; 180 } 181 182 list_for_each_entry_safe(info, temp, &ctx->records_list, list) { 183 if (before(acked_seq, info->end_seq)) 184 break; 185 list_del(&info->list); 186 187 destroy_record(info); 188 deleted_records++; 189 } 190 191 ctx->unacked_record_sn += deleted_records; 192 spin_unlock_irqrestore(&ctx->lock, flags); 193 } 194 195 /* At this point, there should be no references on this 196 * socket and no in-flight SKBs associated with this 197 * socket, so it is safe to free all the resources. 198 */ 199 void tls_device_sk_destruct(struct sock *sk) 200 { 201 struct tls_context *tls_ctx = tls_get_ctx(sk); 202 struct tls_offload_context_tx *ctx = tls_offload_ctx_tx(tls_ctx); 203 204 tls_ctx->sk_destruct(sk); 205 206 if (tls_ctx->tx_conf == TLS_HW) { 207 if (ctx->open_record) 208 destroy_record(ctx->open_record); 209 delete_all_records(ctx); 210 crypto_free_aead(ctx->aead_send); 211 clean_acked_data_disable(inet_csk(sk)); 212 } 213 214 if (refcount_dec_and_test(&tls_ctx->refcount)) 215 tls_device_queue_ctx_destruction(tls_ctx); 216 } 217 EXPORT_SYMBOL(tls_device_sk_destruct); 218 219 static void tls_append_frag(struct tls_record_info *record, 220 struct page_frag *pfrag, 221 int size) 222 { 223 skb_frag_t *frag; 224 225 frag = &record->frags[record->num_frags - 1]; 226 if (frag->page.p == pfrag->page && 227 frag->page_offset + frag->size == pfrag->offset) { 228 frag->size += size; 229 } else { 230 ++frag; 231 frag->page.p = pfrag->page; 232 frag->page_offset = pfrag->offset; 233 frag->size = size; 234 ++record->num_frags; 235 get_page(pfrag->page); 236 } 237 238 pfrag->offset += size; 239 record->len += size; 240 } 241 242 static int tls_push_record(struct sock *sk, 243 struct tls_context *ctx, 244 struct tls_offload_context_tx *offload_ctx, 245 struct tls_record_info *record, 246 struct page_frag *pfrag, 247 int flags, 248 unsigned char record_type) 249 { 250 struct tcp_sock *tp = tcp_sk(sk); 251 struct page_frag dummy_tag_frag; 252 skb_frag_t *frag; 253 int i; 254 255 /* fill prepend */ 256 frag = &record->frags[0]; 257 tls_fill_prepend(ctx, 258 skb_frag_address(frag), 259 record->len - ctx->tx.prepend_size, 260 record_type); 261 262 /* HW doesn't care about the data in the tag, because it fills it. */ 263 dummy_tag_frag.page = skb_frag_page(frag); 264 dummy_tag_frag.offset = 0; 265 266 tls_append_frag(record, &dummy_tag_frag, ctx->tx.tag_size); 267 record->end_seq = tp->write_seq + record->len; 268 spin_lock_irq(&offload_ctx->lock); 269 list_add_tail(&record->list, &offload_ctx->records_list); 270 spin_unlock_irq(&offload_ctx->lock); 271 offload_ctx->open_record = NULL; 272 set_bit(TLS_PENDING_CLOSED_RECORD, &ctx->flags); 273 tls_advance_record_sn(sk, &ctx->tx); 274 275 for (i = 0; i < record->num_frags; i++) { 276 frag = &record->frags[i]; 277 sg_unmark_end(&offload_ctx->sg_tx_data[i]); 278 sg_set_page(&offload_ctx->sg_tx_data[i], skb_frag_page(frag), 279 frag->size, frag->page_offset); 280 sk_mem_charge(sk, frag->size); 281 get_page(skb_frag_page(frag)); 282 } 283 sg_mark_end(&offload_ctx->sg_tx_data[record->num_frags - 1]); 284 285 /* all ready, send */ 286 return tls_push_sg(sk, ctx, offload_ctx->sg_tx_data, 0, flags); 287 } 288 289 static int tls_create_new_record(struct tls_offload_context_tx *offload_ctx, 290 struct page_frag *pfrag, 291 size_t prepend_size) 292 { 293 struct tls_record_info *record; 294 skb_frag_t *frag; 295 296 record = kmalloc(sizeof(*record), GFP_KERNEL); 297 if (!record) 298 return -ENOMEM; 299 300 frag = &record->frags[0]; 301 __skb_frag_set_page(frag, pfrag->page); 302 frag->page_offset = pfrag->offset; 303 skb_frag_size_set(frag, prepend_size); 304 305 get_page(pfrag->page); 306 pfrag->offset += prepend_size; 307 308 record->num_frags = 1; 309 record->len = prepend_size; 310 offload_ctx->open_record = record; 311 return 0; 312 } 313 314 static int tls_do_allocation(struct sock *sk, 315 struct tls_offload_context_tx *offload_ctx, 316 struct page_frag *pfrag, 317 size_t prepend_size) 318 { 319 int ret; 320 321 if (!offload_ctx->open_record) { 322 if (unlikely(!skb_page_frag_refill(prepend_size, pfrag, 323 sk->sk_allocation))) { 324 sk->sk_prot->enter_memory_pressure(sk); 325 sk_stream_moderate_sndbuf(sk); 326 return -ENOMEM; 327 } 328 329 ret = tls_create_new_record(offload_ctx, pfrag, prepend_size); 330 if (ret) 331 return ret; 332 333 if (pfrag->size > pfrag->offset) 334 return 0; 335 } 336 337 if (!sk_page_frag_refill(sk, pfrag)) 338 return -ENOMEM; 339 340 return 0; 341 } 342 343 static int tls_push_data(struct sock *sk, 344 struct iov_iter *msg_iter, 345 size_t size, int flags, 346 unsigned char record_type) 347 { 348 struct tls_context *tls_ctx = tls_get_ctx(sk); 349 struct tls_offload_context_tx *ctx = tls_offload_ctx_tx(tls_ctx); 350 int tls_push_record_flags = flags | MSG_SENDPAGE_NOTLAST; 351 int more = flags & (MSG_SENDPAGE_NOTLAST | MSG_MORE); 352 struct tls_record_info *record = ctx->open_record; 353 struct page_frag *pfrag; 354 size_t orig_size = size; 355 u32 max_open_record_len; 356 int copy, rc = 0; 357 bool done = false; 358 long timeo; 359 360 if (flags & 361 ~(MSG_MORE | MSG_DONTWAIT | MSG_NOSIGNAL | MSG_SENDPAGE_NOTLAST)) 362 return -ENOTSUPP; 363 364 if (sk->sk_err) 365 return -sk->sk_err; 366 367 timeo = sock_sndtimeo(sk, flags & MSG_DONTWAIT); 368 rc = tls_complete_pending_work(sk, tls_ctx, flags, &timeo); 369 if (rc < 0) 370 return rc; 371 372 pfrag = sk_page_frag(sk); 373 374 /* TLS_HEADER_SIZE is not counted as part of the TLS record, and 375 * we need to leave room for an authentication tag. 376 */ 377 max_open_record_len = TLS_MAX_PAYLOAD_SIZE + 378 tls_ctx->tx.prepend_size; 379 do { 380 rc = tls_do_allocation(sk, ctx, pfrag, 381 tls_ctx->tx.prepend_size); 382 if (rc) { 383 rc = sk_stream_wait_memory(sk, &timeo); 384 if (!rc) 385 continue; 386 387 record = ctx->open_record; 388 if (!record) 389 break; 390 handle_error: 391 if (record_type != TLS_RECORD_TYPE_DATA) { 392 /* avoid sending partial 393 * record with type != 394 * application_data 395 */ 396 size = orig_size; 397 destroy_record(record); 398 ctx->open_record = NULL; 399 } else if (record->len > tls_ctx->tx.prepend_size) { 400 goto last_record; 401 } 402 403 break; 404 } 405 406 record = ctx->open_record; 407 copy = min_t(size_t, size, (pfrag->size - pfrag->offset)); 408 copy = min_t(size_t, copy, (max_open_record_len - record->len)); 409 410 if (copy_from_iter_nocache(page_address(pfrag->page) + 411 pfrag->offset, 412 copy, msg_iter) != copy) { 413 rc = -EFAULT; 414 goto handle_error; 415 } 416 tls_append_frag(record, pfrag, copy); 417 418 size -= copy; 419 if (!size) { 420 last_record: 421 tls_push_record_flags = flags; 422 if (more) { 423 tls_ctx->pending_open_record_frags = 424 record->num_frags; 425 break; 426 } 427 428 done = true; 429 } 430 431 if (done || record->len >= max_open_record_len || 432 (record->num_frags >= MAX_SKB_FRAGS - 1)) { 433 rc = tls_push_record(sk, 434 tls_ctx, 435 ctx, 436 record, 437 pfrag, 438 tls_push_record_flags, 439 record_type); 440 if (rc < 0) 441 break; 442 } 443 } while (!done); 444 445 if (orig_size - size > 0) 446 rc = orig_size - size; 447 448 return rc; 449 } 450 451 int tls_device_sendmsg(struct sock *sk, struct msghdr *msg, size_t size) 452 { 453 unsigned char record_type = TLS_RECORD_TYPE_DATA; 454 int rc; 455 456 lock_sock(sk); 457 458 if (unlikely(msg->msg_controllen)) { 459 rc = tls_proccess_cmsg(sk, msg, &record_type); 460 if (rc) 461 goto out; 462 } 463 464 rc = tls_push_data(sk, &msg->msg_iter, size, 465 msg->msg_flags, record_type); 466 467 out: 468 release_sock(sk); 469 return rc; 470 } 471 472 int tls_device_sendpage(struct sock *sk, struct page *page, 473 int offset, size_t size, int flags) 474 { 475 struct iov_iter msg_iter; 476 char *kaddr = kmap(page); 477 struct kvec iov; 478 int rc; 479 480 if (flags & MSG_SENDPAGE_NOTLAST) 481 flags |= MSG_MORE; 482 483 lock_sock(sk); 484 485 if (flags & MSG_OOB) { 486 rc = -ENOTSUPP; 487 goto out; 488 } 489 490 iov.iov_base = kaddr + offset; 491 iov.iov_len = size; 492 iov_iter_kvec(&msg_iter, WRITE | ITER_KVEC, &iov, 1, size); 493 rc = tls_push_data(sk, &msg_iter, size, 494 flags, TLS_RECORD_TYPE_DATA); 495 kunmap(page); 496 497 out: 498 release_sock(sk); 499 return rc; 500 } 501 502 struct tls_record_info *tls_get_record(struct tls_offload_context_tx *context, 503 u32 seq, u64 *p_record_sn) 504 { 505 u64 record_sn = context->hint_record_sn; 506 struct tls_record_info *info; 507 508 info = context->retransmit_hint; 509 if (!info || 510 before(seq, info->end_seq - info->len)) { 511 /* if retransmit_hint is irrelevant start 512 * from the beggining of the list 513 */ 514 info = list_first_entry(&context->records_list, 515 struct tls_record_info, list); 516 record_sn = context->unacked_record_sn; 517 } 518 519 list_for_each_entry_from(info, &context->records_list, list) { 520 if (before(seq, info->end_seq)) { 521 if (!context->retransmit_hint || 522 after(info->end_seq, 523 context->retransmit_hint->end_seq)) { 524 context->hint_record_sn = record_sn; 525 context->retransmit_hint = info; 526 } 527 *p_record_sn = record_sn; 528 return info; 529 } 530 record_sn++; 531 } 532 533 return NULL; 534 } 535 EXPORT_SYMBOL(tls_get_record); 536 537 static int tls_device_push_pending_record(struct sock *sk, int flags) 538 { 539 struct iov_iter msg_iter; 540 541 iov_iter_kvec(&msg_iter, WRITE | ITER_KVEC, NULL, 0, 0); 542 return tls_push_data(sk, &msg_iter, 0, flags, TLS_RECORD_TYPE_DATA); 543 } 544 545 void handle_device_resync(struct sock *sk, u32 seq, u64 rcd_sn) 546 { 547 struct tls_context *tls_ctx = tls_get_ctx(sk); 548 struct net_device *netdev = tls_ctx->netdev; 549 struct tls_offload_context_rx *rx_ctx; 550 u32 is_req_pending; 551 s64 resync_req; 552 u32 req_seq; 553 554 if (tls_ctx->rx_conf != TLS_HW) 555 return; 556 557 rx_ctx = tls_offload_ctx_rx(tls_ctx); 558 resync_req = atomic64_read(&rx_ctx->resync_req); 559 req_seq = ntohl(resync_req >> 32) - ((u32)TLS_HEADER_SIZE - 1); 560 is_req_pending = resync_req; 561 562 if (unlikely(is_req_pending) && req_seq == seq && 563 atomic64_try_cmpxchg(&rx_ctx->resync_req, &resync_req, 0)) 564 netdev->tlsdev_ops->tls_dev_resync_rx(netdev, sk, 565 seq + TLS_HEADER_SIZE - 1, 566 rcd_sn); 567 } 568 569 static int tls_device_reencrypt(struct sock *sk, struct sk_buff *skb) 570 { 571 struct strp_msg *rxm = strp_msg(skb); 572 int err = 0, offset = rxm->offset, copy, nsg; 573 struct sk_buff *skb_iter, *unused; 574 struct scatterlist sg[1]; 575 char *orig_buf, *buf; 576 577 orig_buf = kmalloc(rxm->full_len + TLS_HEADER_SIZE + 578 TLS_CIPHER_AES_GCM_128_IV_SIZE, sk->sk_allocation); 579 if (!orig_buf) 580 return -ENOMEM; 581 buf = orig_buf; 582 583 nsg = skb_cow_data(skb, 0, &unused); 584 if (unlikely(nsg < 0)) { 585 err = nsg; 586 goto free_buf; 587 } 588 589 sg_init_table(sg, 1); 590 sg_set_buf(&sg[0], buf, 591 rxm->full_len + TLS_HEADER_SIZE + 592 TLS_CIPHER_AES_GCM_128_IV_SIZE); 593 skb_copy_bits(skb, offset, buf, 594 TLS_HEADER_SIZE + TLS_CIPHER_AES_GCM_128_IV_SIZE); 595 596 /* We are interested only in the decrypted data not the auth */ 597 err = decrypt_skb(sk, skb, sg); 598 if (err != -EBADMSG) 599 goto free_buf; 600 else 601 err = 0; 602 603 copy = min_t(int, skb_pagelen(skb) - offset, 604 rxm->full_len - TLS_CIPHER_AES_GCM_128_TAG_SIZE); 605 606 if (skb->decrypted) 607 skb_store_bits(skb, offset, buf, copy); 608 609 offset += copy; 610 buf += copy; 611 612 skb_walk_frags(skb, skb_iter) { 613 copy = min_t(int, skb_iter->len, 614 rxm->full_len - offset + rxm->offset - 615 TLS_CIPHER_AES_GCM_128_TAG_SIZE); 616 617 if (skb_iter->decrypted) 618 skb_store_bits(skb_iter, offset, buf, copy); 619 620 offset += copy; 621 buf += copy; 622 } 623 624 free_buf: 625 kfree(orig_buf); 626 return err; 627 } 628 629 int tls_device_decrypted(struct sock *sk, struct sk_buff *skb) 630 { 631 struct tls_context *tls_ctx = tls_get_ctx(sk); 632 struct tls_offload_context_rx *ctx = tls_offload_ctx_rx(tls_ctx); 633 int is_decrypted = skb->decrypted; 634 int is_encrypted = !is_decrypted; 635 struct sk_buff *skb_iter; 636 637 /* Skip if it is already decrypted */ 638 if (ctx->sw.decrypted) 639 return 0; 640 641 /* Check if all the data is decrypted already */ 642 skb_walk_frags(skb, skb_iter) { 643 is_decrypted &= skb_iter->decrypted; 644 is_encrypted &= !skb_iter->decrypted; 645 } 646 647 ctx->sw.decrypted |= is_decrypted; 648 649 /* Return immedeatly if the record is either entirely plaintext or 650 * entirely ciphertext. Otherwise handle reencrypt partially decrypted 651 * record. 652 */ 653 return (is_encrypted || is_decrypted) ? 0 : 654 tls_device_reencrypt(sk, skb); 655 } 656 657 int tls_set_device_offload(struct sock *sk, struct tls_context *ctx) 658 { 659 u16 nonce_size, tag_size, iv_size, rec_seq_size; 660 struct tls_record_info *start_marker_record; 661 struct tls_offload_context_tx *offload_ctx; 662 struct tls_crypto_info *crypto_info; 663 struct net_device *netdev; 664 char *iv, *rec_seq; 665 struct sk_buff *skb; 666 int rc = -EINVAL; 667 __be64 rcd_sn; 668 669 if (!ctx) 670 goto out; 671 672 if (ctx->priv_ctx_tx) { 673 rc = -EEXIST; 674 goto out; 675 } 676 677 start_marker_record = kmalloc(sizeof(*start_marker_record), GFP_KERNEL); 678 if (!start_marker_record) { 679 rc = -ENOMEM; 680 goto out; 681 } 682 683 offload_ctx = kzalloc(TLS_OFFLOAD_CONTEXT_SIZE_TX, GFP_KERNEL); 684 if (!offload_ctx) { 685 rc = -ENOMEM; 686 goto free_marker_record; 687 } 688 689 crypto_info = &ctx->crypto_send; 690 switch (crypto_info->cipher_type) { 691 case TLS_CIPHER_AES_GCM_128: 692 nonce_size = TLS_CIPHER_AES_GCM_128_IV_SIZE; 693 tag_size = TLS_CIPHER_AES_GCM_128_TAG_SIZE; 694 iv_size = TLS_CIPHER_AES_GCM_128_IV_SIZE; 695 iv = ((struct tls12_crypto_info_aes_gcm_128 *)crypto_info)->iv; 696 rec_seq_size = TLS_CIPHER_AES_GCM_128_REC_SEQ_SIZE; 697 rec_seq = 698 ((struct tls12_crypto_info_aes_gcm_128 *)crypto_info)->rec_seq; 699 break; 700 default: 701 rc = -EINVAL; 702 goto free_offload_ctx; 703 } 704 705 ctx->tx.prepend_size = TLS_HEADER_SIZE + nonce_size; 706 ctx->tx.tag_size = tag_size; 707 ctx->tx.overhead_size = ctx->tx.prepend_size + ctx->tx.tag_size; 708 ctx->tx.iv_size = iv_size; 709 ctx->tx.iv = kmalloc(iv_size + TLS_CIPHER_AES_GCM_128_SALT_SIZE, 710 GFP_KERNEL); 711 if (!ctx->tx.iv) { 712 rc = -ENOMEM; 713 goto free_offload_ctx; 714 } 715 716 memcpy(ctx->tx.iv + TLS_CIPHER_AES_GCM_128_SALT_SIZE, iv, iv_size); 717 718 ctx->tx.rec_seq_size = rec_seq_size; 719 ctx->tx.rec_seq = kmemdup(rec_seq, rec_seq_size, GFP_KERNEL); 720 if (!ctx->tx.rec_seq) { 721 rc = -ENOMEM; 722 goto free_iv; 723 } 724 725 rc = tls_sw_fallback_init(sk, offload_ctx, crypto_info); 726 if (rc) 727 goto free_rec_seq; 728 729 /* start at rec_seq - 1 to account for the start marker record */ 730 memcpy(&rcd_sn, ctx->tx.rec_seq, sizeof(rcd_sn)); 731 offload_ctx->unacked_record_sn = be64_to_cpu(rcd_sn) - 1; 732 733 start_marker_record->end_seq = tcp_sk(sk)->write_seq; 734 start_marker_record->len = 0; 735 start_marker_record->num_frags = 0; 736 737 INIT_LIST_HEAD(&offload_ctx->records_list); 738 list_add_tail(&start_marker_record->list, &offload_ctx->records_list); 739 spin_lock_init(&offload_ctx->lock); 740 sg_init_table(offload_ctx->sg_tx_data, 741 ARRAY_SIZE(offload_ctx->sg_tx_data)); 742 743 clean_acked_data_enable(inet_csk(sk), &tls_icsk_clean_acked); 744 ctx->push_pending_record = tls_device_push_pending_record; 745 746 /* TLS offload is greatly simplified if we don't send 747 * SKBs where only part of the payload needs to be encrypted. 748 * So mark the last skb in the write queue as end of record. 749 */ 750 skb = tcp_write_queue_tail(sk); 751 if (skb) 752 TCP_SKB_CB(skb)->eor = 1; 753 754 /* We support starting offload on multiple sockets 755 * concurrently, so we only need a read lock here. 756 * This lock must precede get_netdev_for_sock to prevent races between 757 * NETDEV_DOWN and setsockopt. 758 */ 759 down_read(&device_offload_lock); 760 netdev = get_netdev_for_sock(sk); 761 if (!netdev) { 762 pr_err_ratelimited("%s: netdev not found\n", __func__); 763 rc = -EINVAL; 764 goto release_lock; 765 } 766 767 if (!(netdev->features & NETIF_F_HW_TLS_TX)) { 768 rc = -ENOTSUPP; 769 goto release_netdev; 770 } 771 772 /* Avoid offloading if the device is down 773 * We don't want to offload new flows after 774 * the NETDEV_DOWN event 775 */ 776 if (!(netdev->flags & IFF_UP)) { 777 rc = -EINVAL; 778 goto release_netdev; 779 } 780 781 ctx->priv_ctx_tx = offload_ctx; 782 rc = netdev->tlsdev_ops->tls_dev_add(netdev, sk, TLS_OFFLOAD_CTX_DIR_TX, 783 &ctx->crypto_send, 784 tcp_sk(sk)->write_seq); 785 if (rc) 786 goto release_netdev; 787 788 tls_device_attach(ctx, sk, netdev); 789 790 /* following this assignment tls_is_sk_tx_device_offloaded 791 * will return true and the context might be accessed 792 * by the netdev's xmit function. 793 */ 794 smp_store_release(&sk->sk_validate_xmit_skb, tls_validate_xmit_skb); 795 dev_put(netdev); 796 up_read(&device_offload_lock); 797 goto out; 798 799 release_netdev: 800 dev_put(netdev); 801 release_lock: 802 up_read(&device_offload_lock); 803 clean_acked_data_disable(inet_csk(sk)); 804 crypto_free_aead(offload_ctx->aead_send); 805 free_rec_seq: 806 kfree(ctx->tx.rec_seq); 807 free_iv: 808 kfree(ctx->tx.iv); 809 free_offload_ctx: 810 kfree(offload_ctx); 811 ctx->priv_ctx_tx = NULL; 812 free_marker_record: 813 kfree(start_marker_record); 814 out: 815 return rc; 816 } 817 818 int tls_set_device_offload_rx(struct sock *sk, struct tls_context *ctx) 819 { 820 struct tls_offload_context_rx *context; 821 struct net_device *netdev; 822 int rc = 0; 823 824 /* We support starting offload on multiple sockets 825 * concurrently, so we only need a read lock here. 826 * This lock must precede get_netdev_for_sock to prevent races between 827 * NETDEV_DOWN and setsockopt. 828 */ 829 down_read(&device_offload_lock); 830 netdev = get_netdev_for_sock(sk); 831 if (!netdev) { 832 pr_err_ratelimited("%s: netdev not found\n", __func__); 833 rc = -EINVAL; 834 goto release_lock; 835 } 836 837 if (!(netdev->features & NETIF_F_HW_TLS_RX)) { 838 pr_err_ratelimited("%s: netdev %s with no TLS offload\n", 839 __func__, netdev->name); 840 rc = -ENOTSUPP; 841 goto release_netdev; 842 } 843 844 /* Avoid offloading if the device is down 845 * We don't want to offload new flows after 846 * the NETDEV_DOWN event 847 */ 848 if (!(netdev->flags & IFF_UP)) { 849 rc = -EINVAL; 850 goto release_netdev; 851 } 852 853 context = kzalloc(TLS_OFFLOAD_CONTEXT_SIZE_RX, GFP_KERNEL); 854 if (!context) { 855 rc = -ENOMEM; 856 goto release_netdev; 857 } 858 859 ctx->priv_ctx_rx = context; 860 rc = tls_set_sw_offload(sk, ctx, 0); 861 if (rc) 862 goto release_ctx; 863 864 rc = netdev->tlsdev_ops->tls_dev_add(netdev, sk, TLS_OFFLOAD_CTX_DIR_RX, 865 &ctx->crypto_recv, 866 tcp_sk(sk)->copied_seq); 867 if (rc) { 868 pr_err_ratelimited("%s: The netdev has refused to offload this socket\n", 869 __func__); 870 goto free_sw_resources; 871 } 872 873 tls_device_attach(ctx, sk, netdev); 874 goto release_netdev; 875 876 free_sw_resources: 877 tls_sw_free_resources_rx(sk); 878 release_ctx: 879 ctx->priv_ctx_rx = NULL; 880 release_netdev: 881 dev_put(netdev); 882 release_lock: 883 up_read(&device_offload_lock); 884 return rc; 885 } 886 887 void tls_device_offload_cleanup_rx(struct sock *sk) 888 { 889 struct tls_context *tls_ctx = tls_get_ctx(sk); 890 struct net_device *netdev; 891 892 down_read(&device_offload_lock); 893 netdev = tls_ctx->netdev; 894 if (!netdev) 895 goto out; 896 897 if (!(netdev->features & NETIF_F_HW_TLS_RX)) { 898 pr_err_ratelimited("%s: device is missing NETIF_F_HW_TLS_RX cap\n", 899 __func__); 900 goto out; 901 } 902 903 netdev->tlsdev_ops->tls_dev_del(netdev, tls_ctx, 904 TLS_OFFLOAD_CTX_DIR_RX); 905 906 if (tls_ctx->tx_conf != TLS_HW) { 907 dev_put(netdev); 908 tls_ctx->netdev = NULL; 909 } 910 out: 911 up_read(&device_offload_lock); 912 kfree(tls_ctx->rx.rec_seq); 913 kfree(tls_ctx->rx.iv); 914 tls_sw_release_resources_rx(sk); 915 } 916 917 static int tls_device_down(struct net_device *netdev) 918 { 919 struct tls_context *ctx, *tmp; 920 unsigned long flags; 921 LIST_HEAD(list); 922 923 /* Request a write lock to block new offload attempts */ 924 down_write(&device_offload_lock); 925 926 spin_lock_irqsave(&tls_device_lock, flags); 927 list_for_each_entry_safe(ctx, tmp, &tls_device_list, list) { 928 if (ctx->netdev != netdev || 929 !refcount_inc_not_zero(&ctx->refcount)) 930 continue; 931 932 list_move(&ctx->list, &list); 933 } 934 spin_unlock_irqrestore(&tls_device_lock, flags); 935 936 list_for_each_entry_safe(ctx, tmp, &list, list) { 937 if (ctx->tx_conf == TLS_HW) 938 netdev->tlsdev_ops->tls_dev_del(netdev, ctx, 939 TLS_OFFLOAD_CTX_DIR_TX); 940 if (ctx->rx_conf == TLS_HW) 941 netdev->tlsdev_ops->tls_dev_del(netdev, ctx, 942 TLS_OFFLOAD_CTX_DIR_RX); 943 ctx->netdev = NULL; 944 dev_put(netdev); 945 list_del_init(&ctx->list); 946 947 if (refcount_dec_and_test(&ctx->refcount)) 948 tls_device_free_ctx(ctx); 949 } 950 951 up_write(&device_offload_lock); 952 953 flush_work(&tls_device_gc_work); 954 955 return NOTIFY_DONE; 956 } 957 958 static int tls_dev_event(struct notifier_block *this, unsigned long event, 959 void *ptr) 960 { 961 struct net_device *dev = netdev_notifier_info_to_dev(ptr); 962 963 if (!(dev->features & (NETIF_F_HW_TLS_RX | NETIF_F_HW_TLS_TX))) 964 return NOTIFY_DONE; 965 966 switch (event) { 967 case NETDEV_REGISTER: 968 case NETDEV_FEAT_CHANGE: 969 if ((dev->features & NETIF_F_HW_TLS_RX) && 970 !dev->tlsdev_ops->tls_dev_resync_rx) 971 return NOTIFY_BAD; 972 973 if (dev->tlsdev_ops && 974 dev->tlsdev_ops->tls_dev_add && 975 dev->tlsdev_ops->tls_dev_del) 976 return NOTIFY_DONE; 977 else 978 return NOTIFY_BAD; 979 case NETDEV_DOWN: 980 return tls_device_down(dev); 981 } 982 return NOTIFY_DONE; 983 } 984 985 static struct notifier_block tls_dev_notifier = { 986 .notifier_call = tls_dev_event, 987 }; 988 989 void __init tls_device_init(void) 990 { 991 register_netdevice_notifier(&tls_dev_notifier); 992 } 993 994 void __exit tls_device_cleanup(void) 995 { 996 unregister_netdevice_notifier(&tls_dev_notifier); 997 flush_work(&tls_device_gc_work); 998 } 999