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/btf_ids.h>
   6 #include <linux/filter.h>
   7 #include <linux/errno.h>
   8 #include <linux/file.h>
   9 #include <linux/net.h>
  10 #include <linux/workqueue.h>
  11 #include <linux/skmsg.h>
  12 #include <linux/list.h>
  13 #include <linux/jhash.h>
  14 #include <linux/sock_diag.h>
  15 #include <net/udp.h>
  16
  17 struct bpf_stab {
  18         struct bpf_map map;
  19         struct sock **sks;
  20         struct sk_psock_progs progs;
  21         raw_spinlock_t lock;
  22 };
  23
  24 #define SOCK_CREATE_FLAG_MASK                           \
  25         (BPF_F_NUMA_NODE | BPF_F_RDONLY | BPF_F_WRONLY)
  26
  27 static int sock_map_prog_update(struct bpf_map *map, struct bpf_prog *prog,
  28                                 struct bpf_prog *old, u32 which);
  29 static struct sk_psock_progs *sock_map_progs(struct bpf_map *map);
  30
  31 static struct bpf_map *sock_map_alloc(union bpf_attr *attr)
  32 {
  33         struct bpf_stab *stab;
  34
  35         if (!capable(CAP_NET_ADMIN))
  36                 return ERR_PTR(-EPERM);
  37         if (attr->max_entries == 0 ||
  38             attr->key_size    != 4 ||
  39             (attr->value_size != sizeof(u32) &&
  40              attr->value_size != sizeof(u64)) ||
  41             attr->map_flags & ~SOCK_CREATE_FLAG_MASK)
  42                 return ERR_PTR(-EINVAL);
  43
  44         stab = kzalloc(sizeof(*stab), GFP_USER | __GFP_ACCOUNT);
  45         if (!stab)
  46                 return ERR_PTR(-ENOMEM);
  47
  48         bpf_map_init_from_attr(&stab->map, attr);
  49         raw_spin_lock_init(&stab->lock);
  50
  51         stab->sks = bpf_map_area_alloc((u64) stab->map.max_entries *
  52                                        sizeof(struct sock *),
  53                                        stab->map.numa_node);
  54         if (!stab->sks) {
  55                 kfree(stab);
  56                 return ERR_PTR(-ENOMEM);
  57         }
  58
  59         return &stab->map;
  60 }
  61
  62 int sock_map_get_from_fd(const union bpf_attr *attr, struct bpf_prog *prog)
  63 {
  64         u32 ufd = attr->target_fd;
  65         struct bpf_map *map;
  66         struct fd f;
  67         int ret;
  68
  69         if (attr->attach_flags || attr->replace_bpf_fd)
  70                 return -EINVAL;
  71
  72         f = fdget(ufd);
  73         map = __bpf_map_get(f);
  74         if (IS_ERR(map))
  75                 return PTR_ERR(map);
  76         ret = sock_map_prog_update(map, prog, NULL, attr->attach_type);
  77         fdput(f);
  78         return ret;
  79 }
  80
  81 int sock_map_prog_detach(const union bpf_attr *attr, enum bpf_prog_type ptype)
  82 {
  83         u32 ufd = attr->target_fd;
  84         struct bpf_prog *prog;
  85         struct bpf_map *map;
  86         struct fd f;
  87         int ret;
  88
  89         if (attr->attach_flags || attr->replace_bpf_fd)
  90                 return -EINVAL;
  91
  92         f = fdget(ufd);
  93         map = __bpf_map_get(f);
  94         if (IS_ERR(map))
  95                 return PTR_ERR(map);
  96
  97         prog = bpf_prog_get(attr->attach_bpf_fd);
  98         if (IS_ERR(prog)) {
  99                 ret = PTR_ERR(prog);
 100                 goto put_map;
 101         }
 102
 103         if (prog->type != ptype) {
 104                 ret = -EINVAL;
 105                 goto put_prog;
 106         }
 107
 108         ret = sock_map_prog_update(map, NULL, prog, attr->attach_type);
 109 put_prog:
 110         bpf_prog_put(prog);
 111 put_map:
 112         fdput(f);
 113         return ret;
 114 }
 115
 116 static void sock_map_sk_acquire(struct sock *sk)
 117         __acquires(&sk->sk_lock.slock)
 118 {
 119         lock_sock(sk);
 120         preempt_disable();
 121         rcu_read_lock();
 122 }
 123
 124 static void sock_map_sk_release(struct sock *sk)
 125         __releases(&sk->sk_lock.slock)
 126 {
 127         rcu_read_unlock();
 128         preempt_enable();
 129         release_sock(sk);
 130 }
 131
 132 static void sock_map_add_link(struct sk_psock *psock,
 133                               struct sk_psock_link *link,
 134                               struct bpf_map *map, void *link_raw)
 135 {
 136         link->link_raw = link_raw;
 137         link->map = map;
 138         spin_lock_bh(&psock->link_lock);
 139         list_add_tail(&link->list, &psock->link);
 140         spin_unlock_bh(&psock->link_lock);
 141 }
 142
 143 static void sock_map_del_link(struct sock *sk,
 144                               struct sk_psock *psock, void *link_raw)
 145 {
 146         bool strp_stop = false, verdict_stop = false;
 147         struct sk_psock_link *link, *tmp;
 148
 149         spin_lock_bh(&psock->link_lock);
 150         list_for_each_entry_safe(link, tmp, &psock->link, list) {
 151                 if (link->link_raw == link_raw) {
 152                         struct bpf_map *map = link->map;
 153                         struct bpf_stab *stab = container_of(map, struct bpf_stab,
 154                                                              map);
 155                         if (psock->saved_data_ready && stab->progs.stream_parser)
 156                                 strp_stop = true;
 157                         if (psock->saved_data_ready && stab->progs.stream_verdict)
 158                                 verdict_stop = true;
 159                         if (psock->saved_data_ready && stab->progs.skb_verdict)
 160                                 verdict_stop = true;
 161                         list_del(&link->list);
 162                         sk_psock_free_link(link);
 163                 }
 164         }
 165         spin_unlock_bh(&psock->link_lock);
 166         if (strp_stop || verdict_stop) {
 167                 write_lock_bh(&sk->sk_callback_lock);
 168                 if (strp_stop)
 169                         sk_psock_stop_strp(sk, psock);
 170                 if (verdict_stop)
 171                         sk_psock_stop_verdict(sk, psock);
 172
 173                 if (psock->psock_update_sk_prot)
 174                         psock->psock_update_sk_prot(sk, psock, false);
 175                 write_unlock_bh(&sk->sk_callback_lock);
 176         }
 177 }
 178
 179 static void sock_map_unref(struct sock *sk, void *link_raw)
 180 {
 181         struct sk_psock *psock = sk_psock(sk);
 182
 183         if (likely(psock)) {
 184                 sock_map_del_link(sk, psock, link_raw);
 185                 sk_psock_put(sk, psock);
 186         }
 187 }
 188
 189 static int sock_map_init_proto(struct sock *sk, struct sk_psock *psock)
 190 {
 191         if (!sk->sk_prot->psock_update_sk_prot)
 192                 return -EINVAL;
 193         psock->psock_update_sk_prot = sk->sk_prot->psock_update_sk_prot;
 194         return sk->sk_prot->psock_update_sk_prot(sk, psock, false);
 195 }
 196
 197 static struct sk_psock *sock_map_psock_get_checked(struct sock *sk)
 198 {
 199         struct sk_psock *psock;
 200
 201         rcu_read_lock();
 202         psock = sk_psock(sk);
 203         if (psock) {
 204                 if (sk->sk_prot->close != sock_map_close) {
 205                         psock = ERR_PTR(-EBUSY);
 206                         goto out;
 207                 }
 208
 209                 if (!refcount_inc_not_zero(&psock->refcnt))
 210                         psock = ERR_PTR(-EBUSY);
 211         }
 212 out:
 213         rcu_read_unlock();
 214         return psock;
 215 }
 216
 217 static int sock_map_link(struct bpf_map *map, struct sock *sk)
 218 {
 219         struct sk_psock_progs *progs = sock_map_progs(map);
 220         struct bpf_prog *stream_verdict = NULL;
 221         struct bpf_prog *stream_parser = NULL;
 222         struct bpf_prog *skb_verdict = NULL;
 223         struct bpf_prog *msg_parser = NULL;
 224         struct sk_psock *psock;
 225         int ret;
 226
 227         stream_verdict = READ_ONCE(progs->stream_verdict);
 228         if (stream_verdict) {
 229                 stream_verdict = bpf_prog_inc_not_zero(stream_verdict);
 230                 if (IS_ERR(stream_verdict))
 231                         return PTR_ERR(stream_verdict);
 232         }
 233
 234         stream_parser = READ_ONCE(progs->stream_parser);
 235         if (stream_parser) {
 236                 stream_parser = bpf_prog_inc_not_zero(stream_parser);
 237                 if (IS_ERR(stream_parser)) {
 238                         ret = PTR_ERR(stream_parser);
 239                         goto out_put_stream_verdict;
 240                 }
 241         }
 242
 243         msg_parser = READ_ONCE(progs->msg_parser);
 244         if (msg_parser) {
 245                 msg_parser = bpf_prog_inc_not_zero(msg_parser);
 246                 if (IS_ERR(msg_parser)) {
 247                         ret = PTR_ERR(msg_parser);
 248                         goto out_put_stream_parser;
 249                 }
 250         }
 251
 252         skb_verdict = READ_ONCE(progs->skb_verdict);
 253         if (skb_verdict) {
 254                 skb_verdict = bpf_prog_inc_not_zero(skb_verdict);
 255                 if (IS_ERR(skb_verdict)) {
 256                         ret = PTR_ERR(skb_verdict);
 257                         goto out_put_msg_parser;
 258                 }
 259         }
 260
 261         psock = sock_map_psock_get_checked(sk);
 262         if (IS_ERR(psock)) {
 263                 ret = PTR_ERR(psock);
 264                 goto out_progs;
 265         }
 266
 267         if (psock) {
 268                 if ((msg_parser && READ_ONCE(psock->progs.msg_parser)) ||
 269                     (stream_parser  && READ_ONCE(psock->progs.stream_parser)) ||
 270                     (skb_verdict && READ_ONCE(psock->progs.skb_verdict)) ||
 271                     (skb_verdict && READ_ONCE(psock->progs.stream_verdict)) ||
 272                     (stream_verdict && READ_ONCE(psock->progs.skb_verdict)) ||
 273                     (stream_verdict && READ_ONCE(psock->progs.stream_verdict))) {
 274                         sk_psock_put(sk, psock);
 275                         ret = -EBUSY;
 276                         goto out_progs;
 277                 }
 278         } else {
 279                 psock = sk_psock_init(sk, map->numa_node);
 280                 if (IS_ERR(psock)) {
 281                         ret = PTR_ERR(psock);
 282                         goto out_progs;
 283                 }
 284         }
 285
 286         if (msg_parser)
 287                 psock_set_prog(&psock->progs.msg_parser, msg_parser);
 288         if (stream_parser)
 289                 psock_set_prog(&psock->progs.stream_parser, stream_parser);
 290         if (stream_verdict)
 291                 psock_set_prog(&psock->progs.stream_verdict, stream_verdict);
 292         if (skb_verdict)
 293                 psock_set_prog(&psock->progs.skb_verdict, skb_verdict);
 294
 295         ret = sock_map_init_proto(sk, psock);
 296         if (ret < 0)
 297                 goto out_drop;
 298
 299         write_lock_bh(&sk->sk_callback_lock);
 300         if (stream_parser && stream_verdict && !psock->saved_data_ready) {
 301                 ret = sk_psock_init_strp(sk, psock);
 302                 if (ret)
 303                         goto out_unlock_drop;
 304                 sk_psock_start_strp(sk, psock);
 305         } else if (!stream_parser && stream_verdict && !psock->saved_data_ready) {
 306                 sk_psock_start_verdict(sk,psock);
 307         } else if (!stream_verdict && skb_verdict && !psock->saved_data_ready) {
 308                 sk_psock_start_verdict(sk, psock);
 309         }
 310         write_unlock_bh(&sk->sk_callback_lock);
 311         return 0;
 312 out_unlock_drop:
 313         write_unlock_bh(&sk->sk_callback_lock);
 314 out_drop:
 315         sk_psock_put(sk, psock);
 316 out_progs:
 317         if (skb_verdict)
 318                 bpf_prog_put(skb_verdict);
 319 out_put_msg_parser:
 320         if (msg_parser)
 321                 bpf_prog_put(msg_parser);
 322 out_put_stream_parser:
 323         if (stream_parser)
 324                 bpf_prog_put(stream_parser);
 325 out_put_stream_verdict:
 326         if (stream_verdict)
 327                 bpf_prog_put(stream_verdict);
 328         return ret;
 329 }
 330
 331 static void sock_map_free(struct bpf_map *map)
 332 {
 333         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 334         int i;
 335
 336         /* After the sync no updates or deletes will be in-flight so it
 337          * is safe to walk map and remove entries without risking a race
 338          * in EEXIST update case.
 339          */
 340         synchronize_rcu();
 341         for (i = 0; i < stab->map.max_entries; i++) {
 342                 struct sock **psk = &stab->sks[i];
 343                 struct sock *sk;
 344
 345                 sk = xchg(psk, NULL);
 346                 if (sk) {
 347                         lock_sock(sk);
 348                         rcu_read_lock();
 349                         sock_map_unref(sk, psk);
 350                         rcu_read_unlock();
 351                         release_sock(sk);
 352                 }
 353         }
 354
 355         /* wait for psock readers accessing its map link */
 356         synchronize_rcu();
 357
 358         bpf_map_area_free(stab->sks);
 359         kfree(stab);
 360 }
 361
 362 static void sock_map_release_progs(struct bpf_map *map)
 363 {
 364         psock_progs_drop(&container_of(map, struct bpf_stab, map)->progs);
 365 }
 366
 367 static struct sock *__sock_map_lookup_elem(struct bpf_map *map, u32 key)
 368 {
 369         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 370
 371         WARN_ON_ONCE(!rcu_read_lock_held());
 372
 373         if (unlikely(key >= map->max_entries))
 374                 return NULL;
 375         return READ_ONCE(stab->sks[key]);
 376 }
 377
 378 static void *sock_map_lookup(struct bpf_map *map, void *key)
 379 {
 380         struct sock *sk;
 381
 382         sk = __sock_map_lookup_elem(map, *(u32 *)key);
 383         if (!sk)
 384                 return NULL;
 385         if (sk_is_refcounted(sk) && !refcount_inc_not_zero(&sk->sk_refcnt))
 386                 return NULL;
 387         return sk;
 388 }
 389
 390 static void *sock_map_lookup_sys(struct bpf_map *map, void *key)
 391 {
 392         struct sock *sk;
 393
 394         if (map->value_size != sizeof(u64))
 395                 return ERR_PTR(-ENOSPC);
 396
 397         sk = __sock_map_lookup_elem(map, *(u32 *)key);
 398         if (!sk)
 399                 return ERR_PTR(-ENOENT);
 400
 401         __sock_gen_cookie(sk);
 402         return &sk->sk_cookie;
 403 }
 404
 405 static int __sock_map_delete(struct bpf_stab *stab, struct sock *sk_test,
 406                              struct sock **psk)
 407 {
 408         struct sock *sk;
 409         int err = 0;
 410
 411         raw_spin_lock_bh(&stab->lock);
 412         sk = *psk;
 413         if (!sk_test || sk_test == sk)
 414                 sk = xchg(psk, NULL);
 415
 416         if (likely(sk))
 417                 sock_map_unref(sk, psk);
 418         else
 419                 err = -EINVAL;
 420
 421         raw_spin_unlock_bh(&stab->lock);
 422         return err;
 423 }
 424
 425 static void sock_map_delete_from_link(struct bpf_map *map, struct sock *sk,
 426                                       void *link_raw)
 427 {
 428         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 429
 430         __sock_map_delete(stab, sk, link_raw);
 431 }
 432
 433 static int sock_map_delete_elem(struct bpf_map *map, void *key)
 434 {
 435         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 436         u32 i = *(u32 *)key;
 437         struct sock **psk;
 438
 439         if (unlikely(i >= map->max_entries))
 440                 return -EINVAL;
 441
 442         psk = &stab->sks[i];
 443         return __sock_map_delete(stab, NULL, psk);
 444 }
 445
 446 static int sock_map_get_next_key(struct bpf_map *map, void *key, void *next)
 447 {
 448         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 449         u32 i = key ? *(u32 *)key : U32_MAX;
 450         u32 *key_next = next;
 451
 452         if (i == stab->map.max_entries - 1)
 453                 return -ENOENT;
 454         if (i >= stab->map.max_entries)
 455                 *key_next = 0;
 456         else
 457                 *key_next = i + 1;
 458         return 0;
 459 }
 460
 461 static int sock_map_update_common(struct bpf_map *map, u32 idx,
 462                                   struct sock *sk, u64 flags)
 463 {
 464         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 465         struct sk_psock_link *link;
 466         struct sk_psock *psock;
 467         struct sock *osk;
 468         int ret;
 469
 470         WARN_ON_ONCE(!rcu_read_lock_held());
 471         if (unlikely(flags > BPF_EXIST))
 472                 return -EINVAL;
 473         if (unlikely(idx >= map->max_entries))
 474                 return -E2BIG;
 475
 476         link = sk_psock_init_link();
 477         if (!link)
 478                 return -ENOMEM;
 479
 480         ret = sock_map_link(map, sk);
 481         if (ret < 0)
 482                 goto out_free;
 483
 484         psock = sk_psock(sk);
 485         WARN_ON_ONCE(!psock);
 486
 487         raw_spin_lock_bh(&stab->lock);
 488         osk = stab->sks[idx];
 489         if (osk && flags == BPF_NOEXIST) {
 490                 ret = -EEXIST;
 491                 goto out_unlock;
 492         } else if (!osk && flags == BPF_EXIST) {
 493                 ret = -ENOENT;
 494                 goto out_unlock;
 495         }
 496
 497         sock_map_add_link(psock, link, map, &stab->sks[idx]);
 498         stab->sks[idx] = sk;
 499         if (osk)
 500                 sock_map_unref(osk, &stab->sks[idx]);
 501         raw_spin_unlock_bh(&stab->lock);
 502         return 0;
 503 out_unlock:
 504         raw_spin_unlock_bh(&stab->lock);
 505         if (psock)
 506                 sk_psock_put(sk, psock);
 507 out_free:
 508         sk_psock_free_link(link);
 509         return ret;
 510 }
 511
 512 static bool sock_map_op_okay(const struct bpf_sock_ops_kern *ops)
 513 {
 514         return ops->op == BPF_SOCK_OPS_PASSIVE_ESTABLISHED_CB ||
 515                ops->op == BPF_SOCK_OPS_ACTIVE_ESTABLISHED_CB ||
 516                ops->op == BPF_SOCK_OPS_TCP_LISTEN_CB;
 517 }
 518
 519 static bool sock_map_redirect_allowed(const struct sock *sk)
 520 {
 521         if (sk_is_tcp(sk))
 522                 return sk->sk_state != TCP_LISTEN;
 523         else
 524                 return sk->sk_state == TCP_ESTABLISHED;
 525 }
 526
 527 static bool sock_map_sk_is_suitable(const struct sock *sk)
 528 {
 529         return !!sk->sk_prot->psock_update_sk_prot;
 530 }
 531
 532 static bool sock_map_sk_state_allowed(const struct sock *sk)
 533 {
 534         if (sk_is_tcp(sk))
 535                 return (1 << sk->sk_state) & (TCPF_ESTABLISHED | TCPF_LISTEN);
 536         return true;
 537 }
 538
 539 static int sock_hash_update_common(struct bpf_map *map, void *key,
 540                                    struct sock *sk, u64 flags);
 541
 542 int sock_map_update_elem_sys(struct bpf_map *map, void *key, void *value,
 543                              u64 flags)
 544 {
 545         struct socket *sock;
 546         struct sock *sk;
 547         int ret;
 548         u64 ufd;
 549
 550         if (map->value_size == sizeof(u64))
 551                 ufd = *(u64 *)value;
 552         else
 553                 ufd = *(u32 *)value;
 554         if (ufd > S32_MAX)
 555                 return -EINVAL;
 556
 557         sock = sockfd_lookup(ufd, &ret);
 558         if (!sock)
 559                 return ret;
 560         sk = sock->sk;
 561         if (!sk) {
 562                 ret = -EINVAL;
 563                 goto out;
 564         }
 565         if (!sock_map_sk_is_suitable(sk)) {
 566                 ret = -EOPNOTSUPP;
 567                 goto out;
 568         }
 569
 570         sock_map_sk_acquire(sk);
 571         if (!sock_map_sk_state_allowed(sk))
 572                 ret = -EOPNOTSUPP;
 573         else if (map->map_type == BPF_MAP_TYPE_SOCKMAP)
 574                 ret = sock_map_update_common(map, *(u32 *)key, sk, flags);
 575         else
 576                 ret = sock_hash_update_common(map, key, sk, flags);
 577         sock_map_sk_release(sk);
 578 out:
 579         sockfd_put(sock);
 580         return ret;
 581 }
 582
 583 static int sock_map_update_elem(struct bpf_map *map, void *key,
 584                                 void *value, u64 flags)
 585 {
 586         struct sock *sk = (struct sock *)value;
 587         int ret;
 588
 589         if (unlikely(!sk || !sk_fullsock(sk)))
 590                 return -EINVAL;
 591
 592         if (!sock_map_sk_is_suitable(sk))
 593                 return -EOPNOTSUPP;
 594
 595         local_bh_disable();
 596         bh_lock_sock(sk);
 597         if (!sock_map_sk_state_allowed(sk))
 598                 ret = -EOPNOTSUPP;
 599         else if (map->map_type == BPF_MAP_TYPE_SOCKMAP)
 600                 ret = sock_map_update_common(map, *(u32 *)key, sk, flags);
 601         else
 602                 ret = sock_hash_update_common(map, key, sk, flags);
 603         bh_unlock_sock(sk);
 604         local_bh_enable();
 605         return ret;
 606 }
 607
 608 BPF_CALL_4(bpf_sock_map_update, struct bpf_sock_ops_kern *, sops,
 609            struct bpf_map *, map, void *, key, u64, flags)
 610 {
 611         WARN_ON_ONCE(!rcu_read_lock_held());
 612
 613         if (likely(sock_map_sk_is_suitable(sops->sk) &&
 614                    sock_map_op_okay(sops)))
 615                 return sock_map_update_common(map, *(u32 *)key, sops->sk,
 616                                               flags);
 617         return -EOPNOTSUPP;
 618 }
 619
 620 const struct bpf_func_proto bpf_sock_map_update_proto = {
 621         .func           = bpf_sock_map_update,
 622         .gpl_only       = false,
 623         .pkt_access     = true,
 624         .ret_type       = RET_INTEGER,
 625         .arg1_type      = ARG_PTR_TO_CTX,
 626         .arg2_type      = ARG_CONST_MAP_PTR,
 627         .arg3_type      = ARG_PTR_TO_MAP_KEY,
 628         .arg4_type      = ARG_ANYTHING,
 629 };
 630
 631 BPF_CALL_4(bpf_sk_redirect_map, struct sk_buff *, skb,
 632            struct bpf_map *, map, u32, key, u64, flags)
 633 {
 634         struct sock *sk;
 635
 636         if (unlikely(flags & ~(BPF_F_INGRESS)))
 637                 return SK_DROP;
 638
 639         sk = __sock_map_lookup_elem(map, key);
 640         if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
 641                 return SK_DROP;
 642
 643         skb_bpf_set_redir(skb, sk, flags & BPF_F_INGRESS);
 644         return SK_PASS;
 645 }
 646
 647 const struct bpf_func_proto bpf_sk_redirect_map_proto = {
 648         .func           = bpf_sk_redirect_map,
 649         .gpl_only       = false,
 650         .ret_type       = RET_INTEGER,
 651         .arg1_type      = ARG_PTR_TO_CTX,
 652         .arg2_type      = ARG_CONST_MAP_PTR,
 653         .arg3_type      = ARG_ANYTHING,
 654         .arg4_type      = ARG_ANYTHING,
 655 };
 656
 657 BPF_CALL_4(bpf_msg_redirect_map, struct sk_msg *, msg,
 658            struct bpf_map *, map, u32, key, u64, flags)
 659 {
 660         struct sock *sk;
 661
 662         if (unlikely(flags & ~(BPF_F_INGRESS)))
 663                 return SK_DROP;
 664
 665         sk = __sock_map_lookup_elem(map, key);
 666         if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
 667                 return SK_DROP;
 668
 669         msg->flags = flags;
 670         msg->sk_redir = sk;
 671         return SK_PASS;
 672 }
 673
 674 const struct bpf_func_proto bpf_msg_redirect_map_proto = {
 675         .func           = bpf_msg_redirect_map,
 676         .gpl_only       = false,
 677         .ret_type       = RET_INTEGER,
 678         .arg1_type      = ARG_PTR_TO_CTX,
 679         .arg2_type      = ARG_CONST_MAP_PTR,
 680         .arg3_type      = ARG_ANYTHING,
 681         .arg4_type      = ARG_ANYTHING,
 682 };
 683
 684 struct sock_map_seq_info {
 685         struct bpf_map *map;
 686         struct sock *sk;
 687         u32 index;
 688 };
 689
 690 struct bpf_iter__sockmap {
 691         __bpf_md_ptr(struct bpf_iter_meta *, meta);
 692         __bpf_md_ptr(struct bpf_map *, map);
 693         __bpf_md_ptr(void *, key);
 694         __bpf_md_ptr(struct sock *, sk);
 695 };
 696
 697 DEFINE_BPF_ITER_FUNC(sockmap, struct bpf_iter_meta *meta,
 698                      struct bpf_map *map, void *key,
 699                      struct sock *sk)
 700
 701 static void *sock_map_seq_lookup_elem(struct sock_map_seq_info *info)
 702 {
 703         if (unlikely(info->index >= info->map->max_entries))
 704                 return NULL;
 705
 706         info->sk = __sock_map_lookup_elem(info->map, info->index);
 707
 708         /* can't return sk directly, since that might be NULL */
 709         return info;
 710 }
 711
 712 static void *sock_map_seq_start(struct seq_file *seq, loff_t *pos)
 713         __acquires(rcu)
 714 {
 715         struct sock_map_seq_info *info = seq->private;
 716
 717         if (*pos == 0)
 718                 ++*pos;
 719
 720         /* pairs with sock_map_seq_stop */
 721         rcu_read_lock();
 722         return sock_map_seq_lookup_elem(info);
 723 }
 724
 725 static void *sock_map_seq_next(struct seq_file *seq, void *v, loff_t *pos)
 726         __must_hold(rcu)
 727 {
 728         struct sock_map_seq_info *info = seq->private;
 729
 730         ++*pos;
 731         ++info->index;
 732
 733         return sock_map_seq_lookup_elem(info);
 734 }
 735
 736 static int sock_map_seq_show(struct seq_file *seq, void *v)
 737         __must_hold(rcu)
 738 {
 739         struct sock_map_seq_info *info = seq->private;
 740         struct bpf_iter__sockmap ctx = {};
 741         struct bpf_iter_meta meta;
 742         struct bpf_prog *prog;
 743
 744         meta.seq = seq;
 745         prog = bpf_iter_get_info(&meta, !v);
 746         if (!prog)
 747                 return 0;
 748
 749         ctx.meta = &meta;
 750         ctx.map = info->map;
 751         if (v) {
 752                 ctx.key = &info->index;
 753                 ctx.sk = info->sk;
 754         }
 755
 756         return bpf_iter_run_prog(prog, &ctx);
 757 }
 758
 759 static void sock_map_seq_stop(struct seq_file *seq, void *v)
 760         __releases(rcu)
 761 {
 762         if (!v)
 763                 (void)sock_map_seq_show(seq, NULL);
 764
 765         /* pairs with sock_map_seq_start */
 766         rcu_read_unlock();
 767 }
 768
 769 static const struct seq_operations sock_map_seq_ops = {
 770         .start  = sock_map_seq_start,
 771         .next   = sock_map_seq_next,
 772         .stop   = sock_map_seq_stop,
 773         .show   = sock_map_seq_show,
 774 };
 775
 776 static int sock_map_init_seq_private(void *priv_data,
 777                                      struct bpf_iter_aux_info *aux)
 778 {
 779         struct sock_map_seq_info *info = priv_data;
 780
 781         info->map = aux->map;
 782         return 0;
 783 }
 784
 785 static const struct bpf_iter_seq_info sock_map_iter_seq_info = {
 786         .seq_ops                = &sock_map_seq_ops,
 787         .init_seq_private       = sock_map_init_seq_private,
 788         .seq_priv_size          = sizeof(struct sock_map_seq_info),
 789 };
 790
 791 static int sock_map_btf_id;
 792 const struct bpf_map_ops sock_map_ops = {
 793         .map_meta_equal         = bpf_map_meta_equal,
 794         .map_alloc              = sock_map_alloc,
 795         .map_free               = sock_map_free,
 796         .map_get_next_key       = sock_map_get_next_key,
 797         .map_lookup_elem_sys_only = sock_map_lookup_sys,
 798         .map_update_elem        = sock_map_update_elem,
 799         .map_delete_elem        = sock_map_delete_elem,
 800         .map_lookup_elem        = sock_map_lookup,
 801         .map_release_uref       = sock_map_release_progs,
 802         .map_check_btf          = map_check_no_btf,
 803         .map_btf_name           = "bpf_stab",
 804         .map_btf_id             = &sock_map_btf_id,
 805         .iter_seq_info          = &sock_map_iter_seq_info,
 806 };
 807
 808 struct bpf_shtab_elem {
 809         struct rcu_head rcu;
 810         u32 hash;
 811         struct sock *sk;
 812         struct hlist_node node;
 813         u8 key[];
 814 };
 815
 816 struct bpf_shtab_bucket {
 817         struct hlist_head head;
 818         raw_spinlock_t lock;
 819 };
 820
 821 struct bpf_shtab {
 822         struct bpf_map map;
 823         struct bpf_shtab_bucket *buckets;
 824         u32 buckets_num;
 825         u32 elem_size;
 826         struct sk_psock_progs progs;
 827         atomic_t count;
 828 };
 829
 830 static inline u32 sock_hash_bucket_hash(const void *key, u32 len)
 831 {
 832         return jhash(key, len, 0);
 833 }
 834
 835 static struct bpf_shtab_bucket *sock_hash_select_bucket(struct bpf_shtab *htab,
 836                                                         u32 hash)
 837 {
 838         return &htab->buckets[hash & (htab->buckets_num - 1)];
 839 }
 840
 841 static struct bpf_shtab_elem *
 842 sock_hash_lookup_elem_raw(struct hlist_head *head, u32 hash, void *key,
 843                           u32 key_size)
 844 {
 845         struct bpf_shtab_elem *elem;
 846
 847         hlist_for_each_entry_rcu(elem, head, node) {
 848                 if (elem->hash == hash &&
 849                     !memcmp(&elem->key, key, key_size))
 850                         return elem;
 851         }
 852
 853         return NULL;
 854 }
 855
 856 static struct sock *__sock_hash_lookup_elem(struct bpf_map *map, void *key)
 857 {
 858         struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
 859         u32 key_size = map->key_size, hash;
 860         struct bpf_shtab_bucket *bucket;
 861         struct bpf_shtab_elem *elem;
 862
 863         WARN_ON_ONCE(!rcu_read_lock_held());
 864
 865         hash = sock_hash_bucket_hash(key, key_size);
 866         bucket = sock_hash_select_bucket(htab, hash);
 867         elem = sock_hash_lookup_elem_raw(&bucket->head, hash, key, key_size);
 868
 869         return elem ? elem->sk : NULL;
 870 }
 871
 872 static void sock_hash_free_elem(struct bpf_shtab *htab,
 873                                 struct bpf_shtab_elem *elem)
 874 {
 875         atomic_dec(&htab->count);
 876         kfree_rcu(elem, rcu);
 877 }
 878
 879 static void sock_hash_delete_from_link(struct bpf_map *map, struct sock *sk,
 880                                        void *link_raw)
 881 {
 882         struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
 883         struct bpf_shtab_elem *elem_probe, *elem = link_raw;
 884         struct bpf_shtab_bucket *bucket;
 885
 886         WARN_ON_ONCE(!rcu_read_lock_held());
 887         bucket = sock_hash_select_bucket(htab, elem->hash);
 888
 889         /* elem may be deleted in parallel from the map, but access here
 890          * is okay since it's going away only after RCU grace period.
 891          * However, we need to check whether it's still present.
 892          */
 893         raw_spin_lock_bh(&bucket->lock);
 894         elem_probe = sock_hash_lookup_elem_raw(&bucket->head, elem->hash,
 895                                                elem->key, map->key_size);
 896         if (elem_probe && elem_probe == elem) {
 897                 hlist_del_rcu(&elem->node);
 898                 sock_map_unref(elem->sk, elem);
 899                 sock_hash_free_elem(htab, elem);
 900         }
 901         raw_spin_unlock_bh(&bucket->lock);
 902 }
 903
 904 static int sock_hash_delete_elem(struct bpf_map *map, void *key)
 905 {
 906         struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
 907         u32 hash, key_size = map->key_size;
 908         struct bpf_shtab_bucket *bucket;
 909         struct bpf_shtab_elem *elem;
 910         int ret = -ENOENT;
 911
 912         hash = sock_hash_bucket_hash(key, key_size);
 913         bucket = sock_hash_select_bucket(htab, hash);
 914
 915         raw_spin_lock_bh(&bucket->lock);
 916         elem = sock_hash_lookup_elem_raw(&bucket->head, hash, key, key_size);
 917         if (elem) {
 918                 hlist_del_rcu(&elem->node);
 919                 sock_map_unref(elem->sk, elem);
 920                 sock_hash_free_elem(htab, elem);
 921                 ret = 0;
 922         }
 923         raw_spin_unlock_bh(&bucket->lock);
 924         return ret;
 925 }
 926
 927 static struct bpf_shtab_elem *sock_hash_alloc_elem(struct bpf_shtab *htab,
 928                                                    void *key, u32 key_size,
 929                                                    u32 hash, struct sock *sk,
 930                                                    struct bpf_shtab_elem *old)
 931 {
 932         struct bpf_shtab_elem *new;
 933
 934         if (atomic_inc_return(&htab->count) > htab->map.max_entries) {
 935                 if (!old) {
 936                         atomic_dec(&htab->count);
 937                         return ERR_PTR(-E2BIG);
 938                 }
 939         }
 940
 941         new = bpf_map_kmalloc_node(&htab->map, htab->elem_size,
 942                                    GFP_ATOMIC | __GFP_NOWARN,
 943                                    htab->map.numa_node);
 944         if (!new) {
 945                 atomic_dec(&htab->count);
 946                 return ERR_PTR(-ENOMEM);
 947         }
 948         memcpy(new->key, key, key_size);
 949         new->sk = sk;
 950         new->hash = hash;
 951         return new;
 952 }
 953
 954 static int sock_hash_update_common(struct bpf_map *map, void *key,
 955                                    struct sock *sk, u64 flags)
 956 {
 957         struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
 958         u32 key_size = map->key_size, hash;
 959         struct bpf_shtab_elem *elem, *elem_new;
 960         struct bpf_shtab_bucket *bucket;
 961         struct sk_psock_link *link;
 962         struct sk_psock *psock;
 963         int ret;
 964
 965         WARN_ON_ONCE(!rcu_read_lock_held());
 966         if (unlikely(flags > BPF_EXIST))
 967                 return -EINVAL;
 968
 969         link = sk_psock_init_link();
 970         if (!link)
 971                 return -ENOMEM;
 972
 973         ret = sock_map_link(map, sk);
 974         if (ret < 0)
 975                 goto out_free;
 976
 977         psock = sk_psock(sk);
 978         WARN_ON_ONCE(!psock);
 979
 980         hash = sock_hash_bucket_hash(key, key_size);
 981         bucket = sock_hash_select_bucket(htab, hash);
 982
 983         raw_spin_lock_bh(&bucket->lock);
 984         elem = sock_hash_lookup_elem_raw(&bucket->head, hash, key, key_size);
 985         if (elem && flags == BPF_NOEXIST) {
 986                 ret = -EEXIST;
 987                 goto out_unlock;
 988         } else if (!elem && flags == BPF_EXIST) {
 989                 ret = -ENOENT;
 990                 goto out_unlock;
 991         }
 992
 993         elem_new = sock_hash_alloc_elem(htab, key, key_size, hash, sk, elem);
 994         if (IS_ERR(elem_new)) {
 995                 ret = PTR_ERR(elem_new);
 996                 goto out_unlock;
 997         }
 998
 999         sock_map_add_link(psock, link, map, elem_new);
1000         /* Add new element to the head of the list, so that
1001          * concurrent search will find it before old elem.
1002          */
1003         hlist_add_head_rcu(&elem_new->node, &bucket->head);
1004         if (elem) {
1005                 hlist_del_rcu(&elem->node);
1006                 sock_map_unref(elem->sk, elem);
1007                 sock_hash_free_elem(htab, elem);
1008         }
1009         raw_spin_unlock_bh(&bucket->lock);
1010         return 0;
1011 out_unlock:
1012         raw_spin_unlock_bh(&bucket->lock);
1013         sk_psock_put(sk, psock);
1014 out_free:
1015         sk_psock_free_link(link);
1016         return ret;
1017 }
1018
1019 static int sock_hash_get_next_key(struct bpf_map *map, void *key,
1020                                   void *key_next)
1021 {
1022         struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
1023         struct bpf_shtab_elem *elem, *elem_next;
1024         u32 hash, key_size = map->key_size;
1025         struct hlist_head *head;
1026         int i = 0;
1027
1028         if (!key)
1029                 goto find_first_elem;
1030         hash = sock_hash_bucket_hash(key, key_size);
1031         head = &sock_hash_select_bucket(htab, hash)->head;
1032         elem = sock_hash_lookup_elem_raw(head, hash, key, key_size);
1033         if (!elem)
1034                 goto find_first_elem;
1035
1036         elem_next = hlist_entry_safe(rcu_dereference(hlist_next_rcu(&elem->node)),
1037                                      struct bpf_shtab_elem, node);
1038         if (elem_next) {
1039                 memcpy(key_next, elem_next->key, key_size);
1040                 return 0;
1041         }
1042
1043         i = hash & (htab->buckets_num - 1);
1044         i++;
1045 find_first_elem:
1046         for (; i < htab->buckets_num; i++) {
1047                 head = &sock_hash_select_bucket(htab, i)->head;
1048                 elem_next = hlist_entry_safe(rcu_dereference(hlist_first_rcu(head)),
1049                                              struct bpf_shtab_elem, node);
1050                 if (elem_next) {
1051                         memcpy(key_next, elem_next->key, key_size);
1052                         return 0;
1053                 }
1054         }
1055
1056         return -ENOENT;
1057 }
1058
1059 static struct bpf_map *sock_hash_alloc(union bpf_attr *attr)
1060 {
1061         struct bpf_shtab *htab;
1062         int i, err;
1063
1064         if (!capable(CAP_NET_ADMIN))
1065                 return ERR_PTR(-EPERM);
1066         if (attr->max_entries == 0 ||
1067             attr->key_size    == 0 ||
1068             (attr->value_size != sizeof(u32) &&
1069              attr->value_size != sizeof(u64)) ||
1070             attr->map_flags & ~SOCK_CREATE_FLAG_MASK)
1071                 return ERR_PTR(-EINVAL);
1072         if (attr->key_size > MAX_BPF_STACK)
1073                 return ERR_PTR(-E2BIG);
1074
1075         htab = kzalloc(sizeof(*htab), GFP_USER | __GFP_ACCOUNT);
1076         if (!htab)
1077                 return ERR_PTR(-ENOMEM);
1078
1079         bpf_map_init_from_attr(&htab->map, attr);
1080
1081         htab->buckets_num = roundup_pow_of_two(htab->map.max_entries);
1082         htab->elem_size = sizeof(struct bpf_shtab_elem) +
1083                           round_up(htab->map.key_size, 8);
1084         if (htab->buckets_num == 0 ||
1085             htab->buckets_num > U32_MAX / sizeof(struct bpf_shtab_bucket)) {
1086                 err = -EINVAL;
1087                 goto free_htab;
1088         }
1089
1090         htab->buckets = bpf_map_area_alloc(htab->buckets_num *
1091                                            sizeof(struct bpf_shtab_bucket),
1092                                            htab->map.numa_node);
1093         if (!htab->buckets) {
1094                 err = -ENOMEM;
1095                 goto free_htab;
1096         }
1097
1098         for (i = 0; i < htab->buckets_num; i++) {
1099                 INIT_HLIST_HEAD(&htab->buckets[i].head);
1100                 raw_spin_lock_init(&htab->buckets[i].lock);
1101         }
1102
1103         return &htab->map;
1104 free_htab:
1105         kfree(htab);
1106         return ERR_PTR(err);
1107 }
1108
1109 static void sock_hash_free(struct bpf_map *map)
1110 {
1111         struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
1112         struct bpf_shtab_bucket *bucket;
1113         struct hlist_head unlink_list;
1114         struct bpf_shtab_elem *elem;
1115         struct hlist_node *node;
1116         int i;
1117
1118         /* After the sync no updates or deletes will be in-flight so it
1119          * is safe to walk map and remove entries without risking a race
1120          * in EEXIST update case.
1121          */
1122         synchronize_rcu();
1123         for (i = 0; i < htab->buckets_num; i++) {
1124                 bucket = sock_hash_select_bucket(htab, i);
1125
1126                 /* We are racing with sock_hash_delete_from_link to
1127                  * enter the spin-lock critical section. Every socket on
1128                  * the list is still linked to sockhash. Since link
1129                  * exists, psock exists and holds a ref to socket. That
1130                  * lets us to grab a socket ref too.
1131                  */
1132                 raw_spin_lock_bh(&bucket->lock);
1133                 hlist_for_each_entry(elem, &bucket->head, node)
1134                         sock_hold(elem->sk);
1135                 hlist_move_list(&bucket->head, &unlink_list);
1136                 raw_spin_unlock_bh(&bucket->lock);
1137
1138                 /* Process removed entries out of atomic context to
1139                  * block for socket lock before deleting the psock's
1140                  * link to sockhash.
1141                  */
1142                 hlist_for_each_entry_safe(elem, node, &unlink_list, node) {
1143                         hlist_del(&elem->node);
1144                         lock_sock(elem->sk);
1145                         rcu_read_lock();
1146                         sock_map_unref(elem->sk, elem);
1147                         rcu_read_unlock();
1148                         release_sock(elem->sk);
1149                         sock_put(elem->sk);
1150                         sock_hash_free_elem(htab, elem);
1151                 }
1152         }
1153
1154         /* wait for psock readers accessing its map link */
1155         synchronize_rcu();
1156
1157         bpf_map_area_free(htab->buckets);
1158         kfree(htab);
1159 }
1160
1161 static void *sock_hash_lookup_sys(struct bpf_map *map, void *key)
1162 {
1163         struct sock *sk;
1164
1165         if (map->value_size != sizeof(u64))
1166                 return ERR_PTR(-ENOSPC);
1167
1168         sk = __sock_hash_lookup_elem(map, key);
1169         if (!sk)
1170                 return ERR_PTR(-ENOENT);
1171
1172         __sock_gen_cookie(sk);
1173         return &sk->sk_cookie;
1174 }
1175
1176 static void *sock_hash_lookup(struct bpf_map *map, void *key)
1177 {
1178         struct sock *sk;
1179
1180         sk = __sock_hash_lookup_elem(map, key);
1181         if (!sk)
1182                 return NULL;
1183         if (sk_is_refcounted(sk) && !refcount_inc_not_zero(&sk->sk_refcnt))
1184                 return NULL;
1185         return sk;
1186 }
1187
1188 static void sock_hash_release_progs(struct bpf_map *map)
1189 {
1190         psock_progs_drop(&container_of(map, struct bpf_shtab, map)->progs);
1191 }
1192
1193 BPF_CALL_4(bpf_sock_hash_update, struct bpf_sock_ops_kern *, sops,
1194            struct bpf_map *, map, void *, key, u64, flags)
1195 {
1196         WARN_ON_ONCE(!rcu_read_lock_held());
1197
1198         if (likely(sock_map_sk_is_suitable(sops->sk) &&
1199                    sock_map_op_okay(sops)))
1200                 return sock_hash_update_common(map, key, sops->sk, flags);
1201         return -EOPNOTSUPP;
1202 }
1203
1204 const struct bpf_func_proto bpf_sock_hash_update_proto = {
1205         .func           = bpf_sock_hash_update,
1206         .gpl_only       = false,
1207         .pkt_access     = true,
1208         .ret_type       = RET_INTEGER,
1209         .arg1_type      = ARG_PTR_TO_CTX,
1210         .arg2_type      = ARG_CONST_MAP_PTR,
1211         .arg3_type      = ARG_PTR_TO_MAP_KEY,
1212         .arg4_type      = ARG_ANYTHING,
1213 };
1214
1215 BPF_CALL_4(bpf_sk_redirect_hash, struct sk_buff *, skb,
1216            struct bpf_map *, map, void *, key, u64, flags)
1217 {
1218         struct sock *sk;
1219
1220         if (unlikely(flags & ~(BPF_F_INGRESS)))
1221                 return SK_DROP;
1222
1223         sk = __sock_hash_lookup_elem(map, key);
1224         if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
1225                 return SK_DROP;
1226
1227         skb_bpf_set_redir(skb, sk, flags & BPF_F_INGRESS);
1228         return SK_PASS;
1229 }
1230
1231 const struct bpf_func_proto bpf_sk_redirect_hash_proto = {
1232         .func           = bpf_sk_redirect_hash,
1233         .gpl_only       = false,
1234         .ret_type       = RET_INTEGER,
1235         .arg1_type      = ARG_PTR_TO_CTX,
1236         .arg2_type      = ARG_CONST_MAP_PTR,
1237         .arg3_type      = ARG_PTR_TO_MAP_KEY,
1238         .arg4_type      = ARG_ANYTHING,
1239 };
1240
1241 BPF_CALL_4(bpf_msg_redirect_hash, struct sk_msg *, msg,
1242            struct bpf_map *, map, void *, key, u64, flags)
1243 {
1244         struct sock *sk;
1245
1246         if (unlikely(flags & ~(BPF_F_INGRESS)))
1247                 return SK_DROP;
1248
1249         sk = __sock_hash_lookup_elem(map, key);
1250         if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
1251                 return SK_DROP;
1252
1253         msg->flags = flags;
1254         msg->sk_redir = sk;
1255         return SK_PASS;
1256 }
1257
1258 const struct bpf_func_proto bpf_msg_redirect_hash_proto = {
1259         .func           = bpf_msg_redirect_hash,
1260         .gpl_only       = false,
1261         .ret_type       = RET_INTEGER,
1262         .arg1_type      = ARG_PTR_TO_CTX,
1263         .arg2_type      = ARG_CONST_MAP_PTR,
1264         .arg3_type      = ARG_PTR_TO_MAP_KEY,
1265         .arg4_type      = ARG_ANYTHING,
1266 };
1267
1268 struct sock_hash_seq_info {
1269         struct bpf_map *map;
1270         struct bpf_shtab *htab;
1271         u32 bucket_id;
1272 };
1273
1274 static void *sock_hash_seq_find_next(struct sock_hash_seq_info *info,
1275                                      struct bpf_shtab_elem *prev_elem)
1276 {
1277         const struct bpf_shtab *htab = info->htab;
1278         struct bpf_shtab_bucket *bucket;
1279         struct bpf_shtab_elem *elem;
1280         struct hlist_node *node;
1281
1282         /* try to find next elem in the same bucket */
1283         if (prev_elem) {
1284                 node = rcu_dereference(hlist_next_rcu(&prev_elem->node));
1285                 elem = hlist_entry_safe(node, struct bpf_shtab_elem, node);
1286                 if (elem)
1287                         return elem;
1288
1289                 /* no more elements, continue in the next bucket */
1290                 info->bucket_id++;
1291         }
1292
1293         for (; info->bucket_id < htab->buckets_num; info->bucket_id++) {
1294                 bucket = &htab->buckets[info->bucket_id];
1295                 node = rcu_dereference(hlist_first_rcu(&bucket->head));
1296                 elem = hlist_entry_safe(node, struct bpf_shtab_elem, node);
1297                 if (elem)
1298                         return elem;
1299         }
1300
1301         return NULL;
1302 }
1303
1304 static void *sock_hash_seq_start(struct seq_file *seq, loff_t *pos)
1305         __acquires(rcu)
1306 {
1307         struct sock_hash_seq_info *info = seq->private;
1308
1309         if (*pos == 0)
1310                 ++*pos;
1311
1312         /* pairs with sock_hash_seq_stop */
1313         rcu_read_lock();
1314         return sock_hash_seq_find_next(info, NULL);
1315 }
1316
1317 static void *sock_hash_seq_next(struct seq_file *seq, void *v, loff_t *pos)
1318         __must_hold(rcu)
1319 {
1320         struct sock_hash_seq_info *info = seq->private;
1321
1322         ++*pos;
1323         return sock_hash_seq_find_next(info, v);
1324 }
1325
1326 static int sock_hash_seq_show(struct seq_file *seq, void *v)
1327         __must_hold(rcu)
1328 {
1329         struct sock_hash_seq_info *info = seq->private;
1330         struct bpf_iter__sockmap ctx = {};
1331         struct bpf_shtab_elem *elem = v;
1332         struct bpf_iter_meta meta;
1333         struct bpf_prog *prog;
1334
1335         meta.seq = seq;
1336         prog = bpf_iter_get_info(&meta, !elem);
1337         if (!prog)
1338                 return 0;
1339
1340         ctx.meta = &meta;
1341         ctx.map = info->map;
1342         if (elem) {
1343                 ctx.key = elem->key;
1344                 ctx.sk = elem->sk;
1345         }
1346
1347         return bpf_iter_run_prog(prog, &ctx);
1348 }
1349
1350 static void sock_hash_seq_stop(struct seq_file *seq, void *v)
1351         __releases(rcu)
1352 {
1353         if (!v)
1354                 (void)sock_hash_seq_show(seq, NULL);
1355
1356         /* pairs with sock_hash_seq_start */
1357         rcu_read_unlock();
1358 }
1359
1360 static const struct seq_operations sock_hash_seq_ops = {
1361         .start  = sock_hash_seq_start,
1362         .next   = sock_hash_seq_next,
1363         .stop   = sock_hash_seq_stop,
1364         .show   = sock_hash_seq_show,
1365 };
1366
1367 static int sock_hash_init_seq_private(void *priv_data,
1368                                      struct bpf_iter_aux_info *aux)
1369 {
1370         struct sock_hash_seq_info *info = priv_data;
1371
1372         info->map = aux->map;
1373         info->htab = container_of(aux->map, struct bpf_shtab, map);
1374         return 0;
1375 }
1376
1377 static const struct bpf_iter_seq_info sock_hash_iter_seq_info = {
1378         .seq_ops                = &sock_hash_seq_ops,
1379         .init_seq_private       = sock_hash_init_seq_private,
1380         .seq_priv_size          = sizeof(struct sock_hash_seq_info),
1381 };
1382
1383 static int sock_hash_map_btf_id;
1384 const struct bpf_map_ops sock_hash_ops = {
1385         .map_meta_equal         = bpf_map_meta_equal,
1386         .map_alloc              = sock_hash_alloc,
1387         .map_free               = sock_hash_free,
1388         .map_get_next_key       = sock_hash_get_next_key,
1389         .map_update_elem        = sock_map_update_elem,
1390         .map_delete_elem        = sock_hash_delete_elem,
1391         .map_lookup_elem        = sock_hash_lookup,
1392         .map_lookup_elem_sys_only = sock_hash_lookup_sys,
1393         .map_release_uref       = sock_hash_release_progs,
1394         .map_check_btf          = map_check_no_btf,
1395         .map_btf_name           = "bpf_shtab",
1396         .map_btf_id             = &sock_hash_map_btf_id,
1397         .iter_seq_info          = &sock_hash_iter_seq_info,
1398 };
1399
1400 static struct sk_psock_progs *sock_map_progs(struct bpf_map *map)
1401 {
1402         switch (map->map_type) {
1403         case BPF_MAP_TYPE_SOCKMAP:
1404                 return &container_of(map, struct bpf_stab, map)->progs;
1405         case BPF_MAP_TYPE_SOCKHASH:
1406                 return &container_of(map, struct bpf_shtab, map)->progs;
1407         default:
1408                 break;
1409         }
1410
1411         return NULL;
1412 }
1413
1414 static int sock_map_prog_update(struct bpf_map *map, struct bpf_prog *prog,
1415                                 struct bpf_prog *old, u32 which)
1416 {
1417         struct sk_psock_progs *progs = sock_map_progs(map);
1418         struct bpf_prog **pprog;
1419
1420         if (!progs)
1421                 return -EOPNOTSUPP;
1422
1423         switch (which) {
1424         case BPF_SK_MSG_VERDICT:
1425                 pprog = &progs->msg_parser;
1426                 break;
1427 #if IS_ENABLED(CONFIG_BPF_STREAM_PARSER)
1428         case BPF_SK_SKB_STREAM_PARSER:
1429                 pprog = &progs->stream_parser;
1430                 break;
1431 #endif
1432         case BPF_SK_SKB_STREAM_VERDICT:
1433                 if (progs->skb_verdict)
1434                         return -EBUSY;
1435                 pprog = &progs->stream_verdict;
1436                 break;
1437         case BPF_SK_SKB_VERDICT:
1438                 if (progs->stream_verdict)
1439                         return -EBUSY;
1440                 pprog = &progs->skb_verdict;
1441                 break;
1442         default:
1443                 return -EOPNOTSUPP;
1444         }
1445
1446         if (old)
1447                 return psock_replace_prog(pprog, prog, old);
1448
1449         psock_set_prog(pprog, prog);
1450         return 0;
1451 }
1452
1453 static void sock_map_unlink(struct sock *sk, struct sk_psock_link *link)
1454 {
1455         switch (link->map->map_type) {
1456         case BPF_MAP_TYPE_SOCKMAP:
1457                 return sock_map_delete_from_link(link->map, sk,
1458                                                  link->link_raw);
1459         case BPF_MAP_TYPE_SOCKHASH:
1460                 return sock_hash_delete_from_link(link->map, sk,
1461                                                   link->link_raw);
1462         default:
1463                 break;
1464         }
1465 }
1466
1467 static void sock_map_remove_links(struct sock *sk, struct sk_psock *psock)
1468 {
1469         struct sk_psock_link *link;
1470
1471         while ((link = sk_psock_link_pop(psock))) {
1472                 sock_map_unlink(sk, link);
1473                 sk_psock_free_link(link);
1474         }
1475 }
1476
1477 void sock_map_unhash(struct sock *sk)
1478 {
1479         void (*saved_unhash)(struct sock *sk);
1480         struct sk_psock *psock;
1481
1482         rcu_read_lock();
1483         psock = sk_psock(sk);
1484         if (unlikely(!psock)) {
1485                 rcu_read_unlock();
1486                 if (sk->sk_prot->unhash)
1487                         sk->sk_prot->unhash(sk);
1488                 return;
1489         }
1490
1491         saved_unhash = psock->saved_unhash;
1492         sock_map_remove_links(sk, psock);
1493         rcu_read_unlock();
1494         saved_unhash(sk);
1495 }
1496 EXPORT_SYMBOL_GPL(sock_map_unhash);
1497
1498 void sock_map_close(struct sock *sk, long timeout)
1499 {
1500         void (*saved_close)(struct sock *sk, long timeout);
1501         struct sk_psock *psock;
1502
1503         lock_sock(sk);
1504         rcu_read_lock();
1505         psock = sk_psock_get(sk);
1506         if (unlikely(!psock)) {
1507                 rcu_read_unlock();
1508                 release_sock(sk);
1509                 return sk->sk_prot->close(sk, timeout);
1510         }
1511
1512         saved_close = psock->saved_close;
1513         sock_map_remove_links(sk, psock);
1514         rcu_read_unlock();
1515         sk_psock_stop(psock, true);
1516         sk_psock_put(sk, psock);
1517         release_sock(sk);
1518         saved_close(sk, timeout);
1519 }
1520 EXPORT_SYMBOL_GPL(sock_map_close);
1521
1522 static int sock_map_iter_attach_target(struct bpf_prog *prog,
1523                                        union bpf_iter_link_info *linfo,
1524                                        struct bpf_iter_aux_info *aux)
1525 {
1526         struct bpf_map *map;
1527         int err = -EINVAL;
1528
1529         if (!linfo->map.map_fd)
1530                 return -EBADF;
1531
1532         map = bpf_map_get_with_uref(linfo->map.map_fd);
1533         if (IS_ERR(map))
1534                 return PTR_ERR(map);
1535
1536         if (map->map_type != BPF_MAP_TYPE_SOCKMAP &&
1537             map->map_type != BPF_MAP_TYPE_SOCKHASH)
1538                 goto put_map;
1539
1540         if (prog->aux->max_rdonly_access > map->key_size) {
1541                 err = -EACCES;
1542                 goto put_map;
1543         }
1544
1545         aux->map = map;
1546         return 0;
1547
1548 put_map:
1549         bpf_map_put_with_uref(map);
1550         return err;
1551 }
1552
1553 static void sock_map_iter_detach_target(struct bpf_iter_aux_info *aux)
1554 {
1555         bpf_map_put_with_uref(aux->map);
1556 }
1557
1558 static struct bpf_iter_reg sock_map_iter_reg = {
1559         .target                 = "sockmap",
1560         .attach_target          = sock_map_iter_attach_target,
1561         .detach_target          = sock_map_iter_detach_target,
1562         .show_fdinfo            = bpf_iter_map_show_fdinfo,
1563         .fill_link_info         = bpf_iter_map_fill_link_info,
1564         .ctx_arg_info_size      = 2,
1565         .ctx_arg_info           = {
1566                 { offsetof(struct bpf_iter__sockmap, key),
1567                   PTR_TO_RDONLY_BUF_OR_NULL },
1568                 { offsetof(struct bpf_iter__sockmap, sk),
1569                   PTR_TO_BTF_ID_OR_NULL },
1570         },
1571 };
1572
1573 static int __init bpf_sockmap_iter_init(void)
1574 {
1575         sock_map_iter_reg.ctx_arg_info[1].btf_id =
1576                 btf_sock_ids[BTF_SOCK_TYPE_SOCK];
1577         return bpf_iter_reg_target(&sock_map_iter_reg);
1578 }
1579 late_initcall(bpf_sockmap_iter_init);