net/core/sock_map.c

   1 // SPDX-License-Identifier: GPL-2.0
   2 /* Copyright (c) 2017 - 2018 Covalent IO, Inc. http://covalent.io */
   3
   4 #include <linux/bpf.h>
   5 #include <linux/filter.h>
   6 #include <linux/errno.h>
   7 #include <linux/file.h>
   8 #include <linux/net.h>
   9 #include <linux/workqueue.h>
  10 #include <linux/skmsg.h>
  11 #include <linux/list.h>
  12 #include <linux/jhash.h>
  13 #include <linux/sock_diag.h>
  14 #include <net/udp.h>
  15
  16 struct bpf_stab {
  17         struct bpf_map map;
  18         struct sock **sks;
  19         struct sk_psock_progs progs;
  20         raw_spinlock_t lock;
  21 };
  22
  23 #define SOCK_CREATE_FLAG_MASK                           \
  24         (BPF_F_NUMA_NODE | BPF_F_RDONLY | BPF_F_WRONLY)
  25
  26 static struct bpf_map *sock_map_alloc(union bpf_attr *attr)
  27 {
  28         struct bpf_stab *stab;
  29         u64 cost;
  30         int err;
  31
  32         if (!capable(CAP_NET_ADMIN))
  33                 return ERR_PTR(-EPERM);
  34         if (attr->max_entries == 0 ||
  35             attr->key_size    != 4 ||
  36             (attr->value_size != sizeof(u32) &&
  37              attr->value_size != sizeof(u64)) ||
  38             attr->map_flags & ~SOCK_CREATE_FLAG_MASK)
  39                 return ERR_PTR(-EINVAL);
  40
  41         stab = kzalloc(sizeof(*stab), GFP_USER);
  42         if (!stab)
  43                 return ERR_PTR(-ENOMEM);
  44
  45         bpf_map_init_from_attr(&stab->map, attr);
  46         raw_spin_lock_init(&stab->lock);
  47
  48         /* Make sure page count doesn't overflow. */
  49         cost = (u64) stab->map.max_entries * sizeof(struct sock *);
  50         err = bpf_map_charge_init(&stab->map.memory, cost);
  51         if (err)
  52                 goto free_stab;
  53
  54         stab->sks = bpf_map_area_alloc(stab->map.max_entries *
  55                                        sizeof(struct sock *),
  56                                        stab->map.numa_node);
  57         if (stab->sks)
  58                 return &stab->map;
  59         err = -ENOMEM;
  60         bpf_map_charge_finish(&stab->map.memory);
  61 free_stab:
  62         kfree(stab);
  63         return ERR_PTR(err);
  64 }
  65
  66 int sock_map_get_from_fd(const union bpf_attr *attr, struct bpf_prog *prog)
  67 {
  68         u32 ufd = attr->target_fd;
  69         struct bpf_map *map;
  70         struct fd f;
  71         int ret;
  72
  73         f = fdget(ufd);
  74         map = __bpf_map_get(f);
  75         if (IS_ERR(map))
  76                 return PTR_ERR(map);
  77         ret = sock_map_prog_update(map, prog, attr->attach_type);
  78         fdput(f);
  79         return ret;
  80 }
  81
  82 static void sock_map_sk_acquire(struct sock *sk)
  83         __acquires(&sk->sk_lock.slock)
  84 {
  85         lock_sock(sk);
  86         preempt_disable();
  87         rcu_read_lock();
  88 }
  89
  90 static void sock_map_sk_release(struct sock *sk)
  91         __releases(&sk->sk_lock.slock)
  92 {
  93         rcu_read_unlock();
  94         preempt_enable();
  95         release_sock(sk);
  96 }
  97
  98 static void sock_map_add_link(struct sk_psock *psock,
  99                               struct sk_psock_link *link,
 100                               struct bpf_map *map, void *link_raw)
 101 {
 102         link->link_raw = link_raw;
 103         link->map = map;
 104         spin_lock_bh(&psock->link_lock);
 105         list_add_tail(&link->list, &psock->link);
 106         spin_unlock_bh(&psock->link_lock);
 107 }
 108
 109 static void sock_map_del_link(struct sock *sk,
 110                               struct sk_psock *psock, void *link_raw)
 111 {
 112         struct sk_psock_link *link, *tmp;
 113         bool strp_stop = false;
 114
 115         spin_lock_bh(&psock->link_lock);
 116         list_for_each_entry_safe(link, tmp, &psock->link, list) {
 117                 if (link->link_raw == link_raw) {
 118                         struct bpf_map *map = link->map;
 119                         struct bpf_stab *stab = container_of(map, struct bpf_stab,
 120                                                              map);
 121                         if (psock->parser.enabled && stab->progs.skb_parser)
 122                                 strp_stop = true;
 123                         list_del(&link->list);
 124                         sk_psock_free_link(link);
 125                 }
 126         }
 127         spin_unlock_bh(&psock->link_lock);
 128         if (strp_stop) {
 129                 write_lock_bh(&sk->sk_callback_lock);
 130                 sk_psock_stop_strp(sk, psock);
 131                 write_unlock_bh(&sk->sk_callback_lock);
 132         }
 133 }
 134
 135 static void sock_map_unref(struct sock *sk, void *link_raw)
 136 {
 137         struct sk_psock *psock = sk_psock(sk);
 138
 139         if (likely(psock)) {
 140                 sock_map_del_link(sk, psock, link_raw);
 141                 sk_psock_put(sk, psock);
 142         }
 143 }
 144
 145 static int sock_map_init_proto(struct sock *sk, struct sk_psock *psock)
 146 {
 147         struct proto *prot;
 148
 149         sock_owned_by_me(sk);
 150
 151         switch (sk->sk_type) {
 152         case SOCK_STREAM:
 153                 prot = tcp_bpf_get_proto(sk, psock);
 154                 break;
 155
 156         case SOCK_DGRAM:
 157                 prot = udp_bpf_get_proto(sk, psock);
 158                 break;
 159
 160         default:
 161                 return -EINVAL;
 162         }
 163
 164         if (IS_ERR(prot))
 165                 return PTR_ERR(prot);
 166
 167         sk_psock_update_proto(sk, psock, prot);
 168         return 0;
 169 }
 170
 171 static struct sk_psock *sock_map_psock_get_checked(struct sock *sk)
 172 {
 173         struct sk_psock *psock;
 174
 175         rcu_read_lock();
 176         psock = sk_psock(sk);
 177         if (psock) {
 178                 if (sk->sk_prot->close != sock_map_close) {
 179                         psock = ERR_PTR(-EBUSY);
 180                         goto out;
 181                 }
 182
 183                 if (!refcount_inc_not_zero(&psock->refcnt))
 184                         psock = ERR_PTR(-EBUSY);
 185         }
 186 out:
 187         rcu_read_unlock();
 188         return psock;
 189 }
 190
 191 static int sock_map_link(struct bpf_map *map, struct sk_psock_progs *progs,
 192                          struct sock *sk)
 193 {
 194         struct bpf_prog *msg_parser, *skb_parser, *skb_verdict;
 195         struct sk_psock *psock;
 196         bool skb_progs;
 197         int ret;
 198
 199         skb_verdict = READ_ONCE(progs->skb_verdict);
 200         skb_parser = READ_ONCE(progs->skb_parser);
 201         skb_progs = skb_parser && skb_verdict;
 202         if (skb_progs) {
 203                 skb_verdict = bpf_prog_inc_not_zero(skb_verdict);
 204                 if (IS_ERR(skb_verdict))
 205                         return PTR_ERR(skb_verdict);
 206                 skb_parser = bpf_prog_inc_not_zero(skb_parser);
 207                 if (IS_ERR(skb_parser)) {
 208                         bpf_prog_put(skb_verdict);
 209                         return PTR_ERR(skb_parser);
 210                 }
 211         }
 212
 213         msg_parser = READ_ONCE(progs->msg_parser);
 214         if (msg_parser) {
 215                 msg_parser = bpf_prog_inc_not_zero(msg_parser);
 216                 if (IS_ERR(msg_parser)) {
 217                         ret = PTR_ERR(msg_parser);
 218                         goto out;
 219                 }
 220         }
 221
 222         psock = sock_map_psock_get_checked(sk);
 223         if (IS_ERR(psock)) {
 224                 ret = PTR_ERR(psock);
 225                 goto out_progs;
 226         }
 227
 228         if (psock) {
 229                 if ((msg_parser && READ_ONCE(psock->progs.msg_parser)) ||
 230                     (skb_progs  && READ_ONCE(psock->progs.skb_parser))) {
 231                         sk_psock_put(sk, psock);
 232                         ret = -EBUSY;
 233                         goto out_progs;
 234                 }
 235         } else {
 236                 psock = sk_psock_init(sk, map->numa_node);
 237                 if (!psock) {
 238                         ret = -ENOMEM;
 239                         goto out_progs;
 240                 }
 241         }
 242
 243         if (msg_parser)
 244                 psock_set_prog(&psock->progs.msg_parser, msg_parser);
 245
 246         ret = sock_map_init_proto(sk, psock);
 247         if (ret < 0)
 248                 goto out_drop;
 249
 250         write_lock_bh(&sk->sk_callback_lock);
 251         if (skb_progs && !psock->parser.enabled) {
 252                 ret = sk_psock_init_strp(sk, psock);
 253                 if (ret) {
 254                         write_unlock_bh(&sk->sk_callback_lock);
 255                         goto out_drop;
 256                 }
 257                 psock_set_prog(&psock->progs.skb_verdict, skb_verdict);
 258                 psock_set_prog(&psock->progs.skb_parser, skb_parser);
 259                 sk_psock_start_strp(sk, psock);
 260         }
 261         write_unlock_bh(&sk->sk_callback_lock);
 262         return 0;
 263 out_drop:
 264         sk_psock_put(sk, psock);
 265 out_progs:
 266         if (msg_parser)
 267                 bpf_prog_put(msg_parser);
 268 out:
 269         if (skb_progs) {
 270                 bpf_prog_put(skb_verdict);
 271                 bpf_prog_put(skb_parser);
 272         }
 273         return ret;
 274 }
 275
 276 static int sock_map_link_no_progs(struct bpf_map *map, struct sock *sk)
 277 {
 278         struct sk_psock *psock;
 279         int ret;
 280
 281         psock = sock_map_psock_get_checked(sk);
 282         if (IS_ERR(psock))
 283                 return PTR_ERR(psock);
 284
 285         if (!psock) {
 286                 psock = sk_psock_init(sk, map->numa_node);
 287                 if (!psock)
 288                         return -ENOMEM;
 289         }
 290
 291         ret = sock_map_init_proto(sk, psock);
 292         if (ret < 0)
 293                 sk_psock_put(sk, psock);
 294         return ret;
 295 }
 296
 297 static void sock_map_free(struct bpf_map *map)
 298 {
 299         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 300         int i;
 301
 302         /* After the sync no updates or deletes will be in-flight so it
 303          * is safe to walk map and remove entries without risking a race
 304          * in EEXIST update case.
 305          */
 306         synchronize_rcu();
 307         for (i = 0; i < stab->map.max_entries; i++) {
 308                 struct sock **psk = &stab->sks[i];
 309                 struct sock *sk;
 310
 311                 sk = xchg(psk, NULL);
 312                 if (sk) {
 313                         lock_sock(sk);
 314                         rcu_read_lock();
 315                         sock_map_unref(sk, psk);
 316                         rcu_read_unlock();
 317                         release_sock(sk);
 318                 }
 319         }
 320
 321         /* wait for psock readers accessing its map link */
 322         synchronize_rcu();
 323
 324         bpf_map_area_free(stab->sks);
 325         kfree(stab);
 326 }
 327
 328 static void sock_map_release_progs(struct bpf_map *map)
 329 {
 330         psock_progs_drop(&container_of(map, struct bpf_stab, map)->progs);
 331 }
 332
 333 static struct sock *__sock_map_lookup_elem(struct bpf_map *map, u32 key)
 334 {
 335         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 336
 337         WARN_ON_ONCE(!rcu_read_lock_held());
 338
 339         if (unlikely(key >= map->max_entries))
 340                 return NULL;
 341         return READ_ONCE(stab->sks[key]);
 342 }
 343
 344 static void *sock_map_lookup(struct bpf_map *map, void *key)
 345 {
 346         return __sock_map_lookup_elem(map, *(u32 *)key);
 347 }
 348
 349 static void *sock_map_lookup_sys(struct bpf_map *map, void *key)
 350 {
 351         struct sock *sk;
 352
 353         if (map->value_size != sizeof(u64))
 354                 return ERR_PTR(-ENOSPC);
 355
 356         sk = __sock_map_lookup_elem(map, *(u32 *)key);
 357         if (!sk)
 358                 return ERR_PTR(-ENOENT);
 359
 360         sock_gen_cookie(sk);
 361         return &sk->sk_cookie;
 362 }
 363
 364 static int __sock_map_delete(struct bpf_stab *stab, struct sock *sk_test,
 365                              struct sock **psk)
 366 {
 367         struct sock *sk;
 368         int err = 0;
 369
 370         raw_spin_lock_bh(&stab->lock);
 371         sk = *psk;
 372         if (!sk_test || sk_test == sk)
 373                 sk = xchg(psk, NULL);
 374
 375         if (likely(sk))
 376                 sock_map_unref(sk, psk);
 377         else
 378                 err = -EINVAL;
 379
 380         raw_spin_unlock_bh(&stab->lock);
 381         return err;
 382 }
 383
 384 static void sock_map_delete_from_link(struct bpf_map *map, struct sock *sk,
 385                                       void *link_raw)
 386 {
 387         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 388
 389         __sock_map_delete(stab, sk, link_raw);
 390 }
 391
 392 static int sock_map_delete_elem(struct bpf_map *map, void *key)
 393 {
 394         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 395         u32 i = *(u32 *)key;
 396         struct sock **psk;
 397
 398         if (unlikely(i >= map->max_entries))
 399                 return -EINVAL;
 400
 401         psk = &stab->sks[i];
 402         return __sock_map_delete(stab, NULL, psk);
 403 }
 404
 405 static int sock_map_get_next_key(struct bpf_map *map, void *key, void *next)
 406 {
 407         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 408         u32 i = key ? *(u32 *)key : U32_MAX;
 409         u32 *key_next = next;
 410
 411         if (i == stab->map.max_entries - 1)
 412                 return -ENOENT;
 413         if (i >= stab->map.max_entries)
 414                 *key_next = 0;
 415         else
 416                 *key_next = i + 1;
 417         return 0;
 418 }
 419
 420 static bool sock_map_redirect_allowed(const struct sock *sk)
 421 {
 422         return sk->sk_state != TCP_LISTEN;
 423 }
 424
 425 static int sock_map_update_common(struct bpf_map *map, u32 idx,
 426                                   struct sock *sk, u64 flags)
 427 {
 428         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 429         struct sk_psock_link *link;
 430         struct sk_psock *psock;
 431         struct sock *osk;
 432         int ret;
 433
 434         WARN_ON_ONCE(!rcu_read_lock_held());
 435         if (unlikely(flags > BPF_EXIST))
 436                 return -EINVAL;
 437         if (unlikely(idx >= map->max_entries))
 438                 return -E2BIG;
 439         if (inet_csk_has_ulp(sk))
 440                 return -EINVAL;
 441
 442         link = sk_psock_init_link();
 443         if (!link)
 444                 return -ENOMEM;
 445
 446         /* Only sockets we can redirect into/from in BPF need to hold
 447          * refs to parser/verdict progs and have their sk_data_ready
 448          * and sk_write_space callbacks overridden.
 449          */
 450         if (sock_map_redirect_allowed(sk))
 451                 ret = sock_map_link(map, &stab->progs, sk);
 452         else
 453                 ret = sock_map_link_no_progs(map, sk);
 454         if (ret < 0)
 455                 goto out_free;
 456
 457         psock = sk_psock(sk);
 458         WARN_ON_ONCE(!psock);
 459
 460         raw_spin_lock_bh(&stab->lock);
 461         osk = stab->sks[idx];
 462         if (osk && flags == BPF_NOEXIST) {
 463                 ret = -EEXIST;
 464                 goto out_unlock;
 465         } else if (!osk && flags == BPF_EXIST) {
 466                 ret = -ENOENT;
 467                 goto out_unlock;
 468         }
 469
 470         sock_map_add_link(psock, link, map, &stab->sks[idx]);
 471         stab->sks[idx] = sk;
 472         if (osk)
 473                 sock_map_unref(osk, &stab->sks[idx]);
 474         raw_spin_unlock_bh(&stab->lock);
 475         return 0;
 476 out_unlock:
 477         raw_spin_unlock_bh(&stab->lock);
 478         if (psock)
 479                 sk_psock_put(sk, psock);
 480 out_free:
 481         sk_psock_free_link(link);
 482         return ret;
 483 }
 484
 485 static bool sock_map_op_okay(const struct bpf_sock_ops_kern *ops)
 486 {
 487         return ops->op == BPF_SOCK_OPS_PASSIVE_ESTABLISHED_CB ||
 488                ops->op == BPF_SOCK_OPS_ACTIVE_ESTABLISHED_CB ||
 489                ops->op == BPF_SOCK_OPS_TCP_LISTEN_CB;
 490 }
 491
 492 static bool sk_is_tcp(const struct sock *sk)
 493 {
 494         return sk->sk_type == SOCK_STREAM &&
 495                sk->sk_protocol == IPPROTO_TCP;
 496 }
 497
 498 static bool sk_is_udp(const struct sock *sk)
 499 {
 500         return sk->sk_type == SOCK_DGRAM &&
 501                sk->sk_protocol == IPPROTO_UDP;
 502 }
 503
 504 static bool sock_map_sk_is_suitable(const struct sock *sk)
 505 {
 506         return sk_is_tcp(sk) || sk_is_udp(sk);
 507 }
 508
 509 static bool sock_map_sk_state_allowed(const struct sock *sk)
 510 {
 511         if (sk_is_tcp(sk))
 512                 return (1 << sk->sk_state) & (TCPF_ESTABLISHED | TCPF_LISTEN);
 513         else if (sk_is_udp(sk))
 514                 return sk_hashed(sk);
 515
 516         return false;
 517 }
 518
 519 static int sock_map_update_elem(struct bpf_map *map, void *key,
 520                                 void *value, u64 flags)
 521 {
 522         u32 idx = *(u32 *)key;
 523         struct socket *sock;
 524         struct sock *sk;
 525         int ret;
 526         u64 ufd;
 527
 528         if (map->value_size == sizeof(u64))
 529                 ufd = *(u64 *)value;
 530         else
 531                 ufd = *(u32 *)value;
 532         if (ufd > S32_MAX)
 533                 return -EINVAL;
 534
 535         sock = sockfd_lookup(ufd, &ret);
 536         if (!sock)
 537                 return ret;
 538         sk = sock->sk;
 539         if (!sk) {
 540                 ret = -EINVAL;
 541                 goto out;
 542         }
 543         if (!sock_map_sk_is_suitable(sk)) {
 544                 ret = -EOPNOTSUPP;
 545                 goto out;
 546         }
 547
 548         sock_map_sk_acquire(sk);
 549         if (!sock_map_sk_state_allowed(sk))
 550                 ret = -EOPNOTSUPP;
 551         else
 552                 ret = sock_map_update_common(map, idx, sk, flags);
 553         sock_map_sk_release(sk);
 554 out:
 555         fput(sock->file);
 556         return ret;
 557 }
 558
 559 BPF_CALL_4(bpf_sock_map_update, struct bpf_sock_ops_kern *, sops,
 560            struct bpf_map *, map, void *, key, u64, flags)
 561 {
 562         WARN_ON_ONCE(!rcu_read_lock_held());
 563
 564         if (likely(sock_map_sk_is_suitable(sops->sk) &&
 565                    sock_map_op_okay(sops)))
 566                 return sock_map_update_common(map, *(u32 *)key, sops->sk,
 567                                               flags);
 568         return -EOPNOTSUPP;
 569 }
 570
 571 const struct bpf_func_proto bpf_sock_map_update_proto = {
 572         .func           = bpf_sock_map_update,
 573         .gpl_only       = false,
 574         .pkt_access     = true,
 575         .ret_type       = RET_INTEGER,
 576         .arg1_type      = ARG_PTR_TO_CTX,
 577         .arg2_type      = ARG_CONST_MAP_PTR,
 578         .arg3_type      = ARG_PTR_TO_MAP_KEY,
 579         .arg4_type      = ARG_ANYTHING,
 580 };
 581
 582 BPF_CALL_4(bpf_sk_redirect_map, struct sk_buff *, skb,
 583            struct bpf_map *, map, u32, key, u64, flags)
 584 {
 585         struct tcp_skb_cb *tcb = TCP_SKB_CB(skb);
 586         struct sock *sk;
 587
 588         if (unlikely(flags & ~(BPF_F_INGRESS)))
 589                 return SK_DROP;
 590
 591         sk = __sock_map_lookup_elem(map, key);
 592         if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
 593                 return SK_DROP;
 594
 595         tcb->bpf.flags = flags;
 596         tcb->bpf.sk_redir = sk;
 597         return SK_PASS;
 598 }
 599
 600 const struct bpf_func_proto bpf_sk_redirect_map_proto = {
 601         .func           = bpf_sk_redirect_map,
 602         .gpl_only       = false,
 603         .ret_type       = RET_INTEGER,
 604         .arg1_type      = ARG_PTR_TO_CTX,
 605         .arg2_type      = ARG_CONST_MAP_PTR,
 606         .arg3_type      = ARG_ANYTHING,
 607         .arg4_type      = ARG_ANYTHING,
 608 };
 609
 610 BPF_CALL_4(bpf_msg_redirect_map, struct sk_msg *, msg,
 611            struct bpf_map *, map, u32, key, u64, flags)
 612 {
 613         struct sock *sk;
 614
 615         if (unlikely(flags & ~(BPF_F_INGRESS)))
 616                 return SK_DROP;
 617
 618         sk = __sock_map_lookup_elem(map, key);
 619         if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
 620                 return SK_DROP;
 621
 622         msg->flags = flags;
 623         msg->sk_redir = sk;
 624         return SK_PASS;
 625 }
 626
 627 const struct bpf_func_proto bpf_msg_redirect_map_proto = {
 628         .func           = bpf_msg_redirect_map,
 629         .gpl_only       = false,
 630         .ret_type       = RET_INTEGER,
 631         .arg1_type      = ARG_PTR_TO_CTX,
 632         .arg2_type      = ARG_CONST_MAP_PTR,
 633         .arg3_type      = ARG_ANYTHING,
 634         .arg4_type      = ARG_ANYTHING,
 635 };
 636
 637 const struct bpf_map_ops sock_map_ops = {
 638         .map_alloc              = sock_map_alloc,
 639         .map_free               = sock_map_free,
 640         .map_get_next_key       = sock_map_get_next_key,
 641         .map_lookup_elem_sys_only = sock_map_lookup_sys,
 642         .map_update_elem        = sock_map_update_elem,
 643         .map_delete_elem        = sock_map_delete_elem,
 644         .map_lookup_elem        = sock_map_lookup,
 645         .map_release_uref       = sock_map_release_progs,
 646         .map_check_btf          = map_check_no_btf,
 647 };
 648
 649 struct bpf_htab_elem {
 650         struct rcu_head rcu;
 651         u32 hash;
 652         struct sock *sk;
 653         struct hlist_node node;
 654         u8 key[];
 655 };
 656
 657 struct bpf_htab_bucket {
 658         struct hlist_head head;
 659         raw_spinlock_t lock;
 660 };
 661
 662 struct bpf_htab {
 663         struct bpf_map map;
 664         struct bpf_htab_bucket *buckets;
 665         u32 buckets_num;
 666         u32 elem_size;
 667         struct sk_psock_progs progs;
 668         atomic_t count;
 669 };
 670
 671 static inline u32 sock_hash_bucket_hash(const void *key, u32 len)
 672 {
 673         return jhash(key, len, 0);
 674 }
 675
 676 static struct bpf_htab_bucket *sock_hash_select_bucket(struct bpf_htab *htab,
 677                                                        u32 hash)
 678 {
 679         return &htab->buckets[hash & (htab->buckets_num - 1)];
 680 }
 681
 682 static struct bpf_htab_elem *
 683 sock_hash_lookup_elem_raw(struct hlist_head *head, u32 hash, void *key,
 684                           u32 key_size)
 685 {
 686         struct bpf_htab_elem *elem;
 687
 688         hlist_for_each_entry_rcu(elem, head, node) {
 689                 if (elem->hash == hash &&
 690                     !memcmp(&elem->key, key, key_size))
 691                         return elem;
 692         }
 693
 694         return NULL;
 695 }
 696
 697 static struct sock *__sock_hash_lookup_elem(struct bpf_map *map, void *key)
 698 {
 699         struct bpf_htab *htab = container_of(map, struct bpf_htab, map);
 700         u32 key_size = map->key_size, hash;
 701         struct bpf_htab_bucket *bucket;
 702         struct bpf_htab_elem *elem;
 703
 704         WARN_ON_ONCE(!rcu_read_lock_held());
 705
 706         hash = sock_hash_bucket_hash(key, key_size);
 707         bucket = sock_hash_select_bucket(htab, hash);
 708         elem = sock_hash_lookup_elem_raw(&bucket->head, hash, key, key_size);
 709
 710         return elem ? elem->sk : NULL;
 711 }
 712
 713 static void sock_hash_free_elem(struct bpf_htab *htab,
 714                                 struct bpf_htab_elem *elem)
 715 {
 716         atomic_dec(&htab->count);
 717         kfree_rcu(elem, rcu);
 718 }
 719
 720 static void sock_hash_delete_from_link(struct bpf_map *map, struct sock *sk,
 721                                        void *link_raw)
 722 {
 723         struct bpf_htab *htab = container_of(map, struct bpf_htab, map);
 724         struct bpf_htab_elem *elem_probe, *elem = link_raw;
 725         struct bpf_htab_bucket *bucket;
 726
 727         WARN_ON_ONCE(!rcu_read_lock_held());
 728         bucket = sock_hash_select_bucket(htab, elem->hash);
 729
 730         /* elem may be deleted in parallel from the map, but access here
 731          * is okay since it's going away only after RCU grace period.
 732          * However, we need to check whether it's still present.
 733          */
 734         raw_spin_lock_bh(&bucket->lock);
 735         elem_probe = sock_hash_lookup_elem_raw(&bucket->head, elem->hash,
 736                                                elem->key, map->key_size);
 737         if (elem_probe && elem_probe == elem) {
 738                 hlist_del_rcu(&elem->node);
 739                 sock_map_unref(elem->sk, elem);
 740                 sock_hash_free_elem(htab, elem);
 741         }
 742         raw_spin_unlock_bh(&bucket->lock);
 743 }
 744
 745 static int sock_hash_delete_elem(struct bpf_map *map, void *key)
 746 {
 747         struct bpf_htab *htab = container_of(map, struct bpf_htab, map);
 748         u32 hash, key_size = map->key_size;
 749         struct bpf_htab_bucket *bucket;
 750         struct bpf_htab_elem *elem;
 751         int ret = -ENOENT;
 752
 753         hash = sock_hash_bucket_hash(key, key_size);
 754         bucket = sock_hash_select_bucket(htab, hash);
 755
 756         raw_spin_lock_bh(&bucket->lock);
 757         elem = sock_hash_lookup_elem_raw(&bucket->head, hash, key, key_size);
 758         if (elem) {
 759                 hlist_del_rcu(&elem->node);
 760                 sock_map_unref(elem->sk, elem);
 761                 sock_hash_free_elem(htab, elem);
 762                 ret = 0;
 763         }
 764         raw_spin_unlock_bh(&bucket->lock);
 765         return ret;
 766 }
 767
 768 static struct bpf_htab_elem *sock_hash_alloc_elem(struct bpf_htab *htab,
 769                                                   void *key, u32 key_size,
 770                                                   u32 hash, struct sock *sk,
 771                                                   struct bpf_htab_elem *old)
 772 {
 773         struct bpf_htab_elem *new;
 774
 775         if (atomic_inc_return(&htab->count) > htab->map.max_entries) {
 776                 if (!old) {
 777                         atomic_dec(&htab->count);
 778                         return ERR_PTR(-E2BIG);
 779                 }
 780         }
 781
 782         new = kmalloc_node(htab->elem_size, GFP_ATOMIC | __GFP_NOWARN,
 783                            htab->map.numa_node);
 784         if (!new) {
 785                 atomic_dec(&htab->count);
 786                 return ERR_PTR(-ENOMEM);
 787         }
 788         memcpy(new->key, key, key_size);
 789         new->sk = sk;
 790         new->hash = hash;
 791         return new;
 792 }
 793
 794 static int sock_hash_update_common(struct bpf_map *map, void *key,
 795                                    struct sock *sk, u64 flags)
 796 {
 797         struct bpf_htab *htab = container_of(map, struct bpf_htab, map);
 798         u32 key_size = map->key_size, hash;
 799         struct bpf_htab_elem *elem, *elem_new;
 800         struct bpf_htab_bucket *bucket;
 801         struct sk_psock_link *link;
 802         struct sk_psock *psock;
 803         int ret;
 804
 805         WARN_ON_ONCE(!rcu_read_lock_held());
 806         if (unlikely(flags > BPF_EXIST))
 807                 return -EINVAL;
 808         if (inet_csk_has_ulp(sk))
 809                 return -EINVAL;
 810
 811         link = sk_psock_init_link();
 812         if (!link)
 813                 return -ENOMEM;
 814
 815         /* Only sockets we can redirect into/from in BPF need to hold
 816          * refs to parser/verdict progs and have their sk_data_ready
 817          * and sk_write_space callbacks overridden.
 818          */
 819         if (sock_map_redirect_allowed(sk))
 820                 ret = sock_map_link(map, &htab->progs, sk);
 821         else
 822                 ret = sock_map_link_no_progs(map, sk);
 823         if (ret < 0)
 824                 goto out_free;
 825
 826         psock = sk_psock(sk);
 827         WARN_ON_ONCE(!psock);
 828
 829         hash = sock_hash_bucket_hash(key, key_size);
 830         bucket = sock_hash_select_bucket(htab, hash);
 831
 832         raw_spin_lock_bh(&bucket->lock);
 833         elem = sock_hash_lookup_elem_raw(&bucket->head, hash, key, key_size);
 834         if (elem && flags == BPF_NOEXIST) {
 835                 ret = -EEXIST;
 836                 goto out_unlock;
 837         } else if (!elem && flags == BPF_EXIST) {
 838                 ret = -ENOENT;
 839                 goto out_unlock;
 840         }
 841
 842         elem_new = sock_hash_alloc_elem(htab, key, key_size, hash, sk, elem);
 843         if (IS_ERR(elem_new)) {
 844                 ret = PTR_ERR(elem_new);
 845                 goto out_unlock;
 846         }
 847
 848         sock_map_add_link(psock, link, map, elem_new);
 849         /* Add new element to the head of the list, so that
 850          * concurrent search will find it before old elem.
 851          */
 852         hlist_add_head_rcu(&elem_new->node, &bucket->head);
 853         if (elem) {
 854                 hlist_del_rcu(&elem->node);
 855                 sock_map_unref(elem->sk, elem);
 856                 sock_hash_free_elem(htab, elem);
 857         }
 858         raw_spin_unlock_bh(&bucket->lock);
 859         return 0;
 860 out_unlock:
 861         raw_spin_unlock_bh(&bucket->lock);
 862         sk_psock_put(sk, psock);
 863 out_free:
 864         sk_psock_free_link(link);
 865         return ret;
 866 }
 867
 868 static int sock_hash_update_elem(struct bpf_map *map, void *key,
 869                                  void *value, u64 flags)
 870 {
 871         struct socket *sock;
 872         struct sock *sk;
 873         int ret;
 874         u64 ufd;
 875
 876         if (map->value_size == sizeof(u64))
 877                 ufd = *(u64 *)value;
 878         else
 879                 ufd = *(u32 *)value;
 880         if (ufd > S32_MAX)
 881                 return -EINVAL;
 882
 883         sock = sockfd_lookup(ufd, &ret);
 884         if (!sock)
 885                 return ret;
 886         sk = sock->sk;
 887         if (!sk) {
 888                 ret = -EINVAL;
 889                 goto out;
 890         }
 891         if (!sock_map_sk_is_suitable(sk)) {
 892                 ret = -EOPNOTSUPP;
 893                 goto out;
 894         }
 895
 896         sock_map_sk_acquire(sk);
 897         if (!sock_map_sk_state_allowed(sk))
 898                 ret = -EOPNOTSUPP;
 899         else
 900                 ret = sock_hash_update_common(map, key, sk, flags);
 901         sock_map_sk_release(sk);
 902 out:
 903         fput(sock->file);
 904         return ret;
 905 }
 906
 907 static int sock_hash_get_next_key(struct bpf_map *map, void *key,
 908                                   void *key_next)
 909 {
 910         struct bpf_htab *htab = container_of(map, struct bpf_htab, map);
 911         struct bpf_htab_elem *elem, *elem_next;
 912         u32 hash, key_size = map->key_size;
 913         struct hlist_head *head;
 914         int i = 0;
 915
 916         if (!key)
 917                 goto find_first_elem;
 918         hash = sock_hash_bucket_hash(key, key_size);
 919         head = &sock_hash_select_bucket(htab, hash)->head;
 920         elem = sock_hash_lookup_elem_raw(head, hash, key, key_size);
 921         if (!elem)
 922                 goto find_first_elem;
 923
 924         elem_next = hlist_entry_safe(rcu_dereference_raw(hlist_next_rcu(&elem->node)),
 925                                      struct bpf_htab_elem, node);
 926         if (elem_next) {
 927                 memcpy(key_next, elem_next->key, key_size);
 928                 return 0;
 929         }
 930
 931         i = hash & (htab->buckets_num - 1);
 932         i++;
 933 find_first_elem:
 934         for (; i < htab->buckets_num; i++) {
 935                 head = &sock_hash_select_bucket(htab, i)->head;
 936                 elem_next = hlist_entry_safe(rcu_dereference_raw(hlist_first_rcu(head)),
 937                                              struct bpf_htab_elem, node);
 938                 if (elem_next) {
 939                         memcpy(key_next, elem_next->key, key_size);
 940                         return 0;
 941                 }
 942         }
 943
 944         return -ENOENT;
 945 }
 946
 947 static struct bpf_map *sock_hash_alloc(union bpf_attr *attr)
 948 {
 949         struct bpf_htab *htab;
 950         int i, err;
 951         u64 cost;
 952
 953         if (!capable(CAP_NET_ADMIN))
 954                 return ERR_PTR(-EPERM);
 955         if (attr->max_entries == 0 ||
 956             attr->key_size    == 0 ||
 957             (attr->value_size != sizeof(u32) &&
 958              attr->value_size != sizeof(u64)) ||
 959             attr->map_flags & ~SOCK_CREATE_FLAG_MASK)
 960                 return ERR_PTR(-EINVAL);
 961         if (attr->key_size > MAX_BPF_STACK)
 962                 return ERR_PTR(-E2BIG);
 963
 964         htab = kzalloc(sizeof(*htab), GFP_USER);
 965         if (!htab)
 966                 return ERR_PTR(-ENOMEM);
 967
 968         bpf_map_init_from_attr(&htab->map, attr);
 969
 970         htab->buckets_num = roundup_pow_of_two(htab->map.max_entries);
 971         htab->elem_size = sizeof(struct bpf_htab_elem) +
 972                           round_up(htab->map.key_size, 8);
 973         if (htab->buckets_num == 0 ||
 974             htab->buckets_num > U32_MAX / sizeof(struct bpf_htab_bucket)) {
 975                 err = -EINVAL;
 976                 goto free_htab;
 977         }
 978
 979         cost = (u64) htab->buckets_num * sizeof(struct bpf_htab_bucket) +
 980                (u64) htab->elem_size * htab->map.max_entries;
 981         if (cost >= U32_MAX - PAGE_SIZE) {
 982                 err = -EINVAL;
 983                 goto free_htab;
 984         }
 985
 986         htab->buckets = bpf_map_area_alloc(htab->buckets_num *
 987                                            sizeof(struct bpf_htab_bucket),
 988                                            htab->map.numa_node);
 989         if (!htab->buckets) {
 990                 err = -ENOMEM;
 991                 goto free_htab;
 992         }
 993
 994         for (i = 0; i < htab->buckets_num; i++) {
 995                 INIT_HLIST_HEAD(&htab->buckets[i].head);
 996                 raw_spin_lock_init(&htab->buckets[i].lock);
 997         }
 998
 999         return &htab->map;
1000 free_htab:
1001         kfree(htab);
1002         return ERR_PTR(err);
1003 }
1004
1005 static void sock_hash_free(struct bpf_map *map)
1006 {
1007         struct bpf_htab *htab = container_of(map, struct bpf_htab, map);
1008         struct bpf_htab_bucket *bucket;
1009         struct bpf_htab_elem *elem;
1010         struct hlist_node *node;
1011         int i;
1012
1013         /* After the sync no updates or deletes will be in-flight so it
1014          * is safe to walk map and remove entries without risking a race
1015          * in EEXIST update case.
1016          */
1017         synchronize_rcu();
1018         for (i = 0; i < htab->buckets_num; i++) {
1019                 bucket = sock_hash_select_bucket(htab, i);
1020                 hlist_for_each_entry_safe(elem, node, &bucket->head, node) {
1021                         hlist_del_rcu(&elem->node);
1022                         lock_sock(elem->sk);
1023                         rcu_read_lock();
1024                         sock_map_unref(elem->sk, elem);
1025                         rcu_read_unlock();
1026                         release_sock(elem->sk);
1027                 }
1028         }
1029
1030         /* wait for psock readers accessing its map link */
1031         synchronize_rcu();
1032
1033         bpf_map_area_free(htab->buckets);
1034         kfree(htab);
1035 }
1036
1037 static void *sock_hash_lookup_sys(struct bpf_map *map, void *key)
1038 {
1039         struct sock *sk;
1040
1041         if (map->value_size != sizeof(u64))
1042                 return ERR_PTR(-ENOSPC);
1043
1044         sk = __sock_hash_lookup_elem(map, key);
1045         if (!sk)
1046                 return ERR_PTR(-ENOENT);
1047
1048         sock_gen_cookie(sk);
1049         return &sk->sk_cookie;
1050 }
1051
1052 static void *sock_hash_lookup(struct bpf_map *map, void *key)
1053 {
1054         return __sock_hash_lookup_elem(map, key);
1055 }
1056
1057 static void sock_hash_release_progs(struct bpf_map *map)
1058 {
1059         psock_progs_drop(&container_of(map, struct bpf_htab, map)->progs);
1060 }
1061
1062 BPF_CALL_4(bpf_sock_hash_update, struct bpf_sock_ops_kern *, sops,
1063            struct bpf_map *, map, void *, key, u64, flags)
1064 {
1065         WARN_ON_ONCE(!rcu_read_lock_held());
1066
1067         if (likely(sock_map_sk_is_suitable(sops->sk) &&
1068                    sock_map_op_okay(sops)))
1069                 return sock_hash_update_common(map, key, sops->sk, flags);
1070         return -EOPNOTSUPP;
1071 }
1072
1073 const struct bpf_func_proto bpf_sock_hash_update_proto = {
1074         .func           = bpf_sock_hash_update,
1075         .gpl_only       = false,
1076         .pkt_access     = true,
1077         .ret_type       = RET_INTEGER,
1078         .arg1_type      = ARG_PTR_TO_CTX,
1079         .arg2_type      = ARG_CONST_MAP_PTR,
1080         .arg3_type      = ARG_PTR_TO_MAP_KEY,
1081         .arg4_type      = ARG_ANYTHING,
1082 };
1083
1084 BPF_CALL_4(bpf_sk_redirect_hash, struct sk_buff *, skb,
1085            struct bpf_map *, map, void *, key, u64, flags)
1086 {
1087         struct tcp_skb_cb *tcb = TCP_SKB_CB(skb);
1088         struct sock *sk;
1089
1090         if (unlikely(flags & ~(BPF_F_INGRESS)))
1091                 return SK_DROP;
1092
1093         sk = __sock_hash_lookup_elem(map, key);
1094         if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
1095                 return SK_DROP;
1096
1097         tcb->bpf.flags = flags;
1098         tcb->bpf.sk_redir = sk;
1099         return SK_PASS;
1100 }
1101
1102 const struct bpf_func_proto bpf_sk_redirect_hash_proto = {
1103         .func           = bpf_sk_redirect_hash,
1104         .gpl_only       = false,
1105         .ret_type       = RET_INTEGER,
1106         .arg1_type      = ARG_PTR_TO_CTX,
1107         .arg2_type      = ARG_CONST_MAP_PTR,
1108         .arg3_type      = ARG_PTR_TO_MAP_KEY,
1109         .arg4_type      = ARG_ANYTHING,
1110 };
1111
1112 BPF_CALL_4(bpf_msg_redirect_hash, struct sk_msg *, msg,
1113            struct bpf_map *, map, void *, key, u64, flags)
1114 {
1115         struct sock *sk;
1116
1117         if (unlikely(flags & ~(BPF_F_INGRESS)))
1118                 return SK_DROP;
1119
1120         sk = __sock_hash_lookup_elem(map, key);
1121         if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
1122                 return SK_DROP;
1123
1124         msg->flags = flags;
1125         msg->sk_redir = sk;
1126         return SK_PASS;
1127 }
1128
1129 const struct bpf_func_proto bpf_msg_redirect_hash_proto = {
1130         .func           = bpf_msg_redirect_hash,
1131         .gpl_only       = false,
1132         .ret_type       = RET_INTEGER,
1133         .arg1_type      = ARG_PTR_TO_CTX,
1134         .arg2_type      = ARG_CONST_MAP_PTR,
1135         .arg3_type      = ARG_PTR_TO_MAP_KEY,
1136         .arg4_type      = ARG_ANYTHING,
1137 };
1138
1139 const struct bpf_map_ops sock_hash_ops = {
1140         .map_alloc              = sock_hash_alloc,
1141         .map_free               = sock_hash_free,
1142         .map_get_next_key       = sock_hash_get_next_key,
1143         .map_update_elem        = sock_hash_update_elem,
1144         .map_delete_elem        = sock_hash_delete_elem,
1145         .map_lookup_elem        = sock_hash_lookup,
1146         .map_lookup_elem_sys_only = sock_hash_lookup_sys,
1147         .map_release_uref       = sock_hash_release_progs,
1148         .map_check_btf          = map_check_no_btf,
1149 };
1150
1151 static struct sk_psock_progs *sock_map_progs(struct bpf_map *map)
1152 {
1153         switch (map->map_type) {
1154         case BPF_MAP_TYPE_SOCKMAP:
1155                 return &container_of(map, struct bpf_stab, map)->progs;
1156         case BPF_MAP_TYPE_SOCKHASH:
1157                 return &container_of(map, struct bpf_htab, map)->progs;
1158         default:
1159                 break;
1160         }
1161
1162         return NULL;
1163 }
1164
1165 int sock_map_prog_update(struct bpf_map *map, struct bpf_prog *prog,
1166                          u32 which)
1167 {
1168         struct sk_psock_progs *progs = sock_map_progs(map);
1169
1170         if (!progs)
1171                 return -EOPNOTSUPP;
1172
1173         switch (which) {
1174         case BPF_SK_MSG_VERDICT:
1175                 psock_set_prog(&progs->msg_parser, prog);
1176                 break;
1177         case BPF_SK_SKB_STREAM_PARSER:
1178                 psock_set_prog(&progs->skb_parser, prog);
1179                 break;
1180         case BPF_SK_SKB_STREAM_VERDICT:
1181                 psock_set_prog(&progs->skb_verdict, prog);
1182                 break;
1183         default:
1184                 return -EOPNOTSUPP;
1185         }
1186
1187         return 0;
1188 }
1189
1190 static void sock_map_unlink(struct sock *sk, struct sk_psock_link *link)
1191 {
1192         switch (link->map->map_type) {
1193         case BPF_MAP_TYPE_SOCKMAP:
1194                 return sock_map_delete_from_link(link->map, sk,
1195                                                  link->link_raw);
1196         case BPF_MAP_TYPE_SOCKHASH:
1197                 return sock_hash_delete_from_link(link->map, sk,
1198                                                   link->link_raw);
1199         default:
1200                 break;
1201         }
1202 }
1203
1204 static void sock_map_remove_links(struct sock *sk, struct sk_psock *psock)
1205 {
1206         struct sk_psock_link *link;
1207
1208         while ((link = sk_psock_link_pop(psock))) {
1209                 sock_map_unlink(sk, link);
1210                 sk_psock_free_link(link);
1211         }
1212 }
1213
1214 void sock_map_unhash(struct sock *sk)
1215 {
1216         void (*saved_unhash)(struct sock *sk);
1217         struct sk_psock *psock;
1218
1219         rcu_read_lock();
1220         psock = sk_psock(sk);
1221         if (unlikely(!psock)) {
1222                 rcu_read_unlock();
1223                 if (sk->sk_prot->unhash)
1224                         sk->sk_prot->unhash(sk);
1225                 return;
1226         }
1227
1228         saved_unhash = psock->saved_unhash;
1229         sock_map_remove_links(sk, psock);
1230         rcu_read_unlock();
1231         saved_unhash(sk);
1232 }
1233
1234 void sock_map_close(struct sock *sk, long timeout)
1235 {
1236         void (*saved_close)(struct sock *sk, long timeout);
1237         struct sk_psock *psock;
1238
1239         lock_sock(sk);
1240         rcu_read_lock();
1241         psock = sk_psock(sk);
1242         if (unlikely(!psock)) {
1243                 rcu_read_unlock();
1244                 release_sock(sk);
1245                 return sk->sk_prot->close(sk, timeout);
1246         }
1247
1248         saved_close = psock->saved_close;
1249         sock_map_remove_links(sk, psock);
1250         rcu_read_unlock();
1251         release_sock(sk);
1252         saved_close(sk, timeout);
1253 }