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 struct bpf_map *sock_map_alloc(union bpf_attr *attr)
  28 {
  29         struct bpf_stab *stab;
  30         u64 cost;
  31         int err;
  32
  33         if (!capable(CAP_NET_ADMIN))
  34                 return ERR_PTR(-EPERM);
  35         if (attr->max_entries == 0 ||
  36             attr->key_size    != 4 ||
  37             (attr->value_size != sizeof(u32) &&
  38              attr->value_size != sizeof(u64)) ||
  39             attr->map_flags & ~SOCK_CREATE_FLAG_MASK)
  40                 return ERR_PTR(-EINVAL);
  41
  42         stab = kzalloc(sizeof(*stab), GFP_USER);
  43         if (!stab)
  44                 return ERR_PTR(-ENOMEM);
  45
  46         bpf_map_init_from_attr(&stab->map, attr);
  47         raw_spin_lock_init(&stab->lock);
  48
  49         /* Make sure page count doesn't overflow. */
  50         cost = (u64) stab->map.max_entries * sizeof(struct sock *);
  51         err = bpf_map_charge_init(&stab->map.memory, cost);
  52         if (err)
  53                 goto free_stab;
  54
  55         stab->sks = bpf_map_area_alloc(stab->map.max_entries *
  56                                        sizeof(struct sock *),
  57                                        stab->map.numa_node);
  58         if (stab->sks)
  59                 return &stab->map;
  60         err = -ENOMEM;
  61         bpf_map_charge_finish(&stab->map.memory);
  62 free_stab:
  63         kfree(stab);
  64         return ERR_PTR(err);
  65 }
  66
  67 int sock_map_get_from_fd(const union bpf_attr *attr, struct bpf_prog *prog)
  68 {
  69         u32 ufd = attr->target_fd;
  70         struct bpf_map *map;
  71         struct fd f;
  72         int ret;
  73
  74         if (attr->attach_flags || attr->replace_bpf_fd)
  75                 return -EINVAL;
  76
  77         f = fdget(ufd);
  78         map = __bpf_map_get(f);
  79         if (IS_ERR(map))
  80                 return PTR_ERR(map);
  81         ret = sock_map_prog_update(map, prog, NULL, attr->attach_type);
  82         fdput(f);
  83         return ret;
  84 }
  85
  86 int sock_map_prog_detach(const union bpf_attr *attr, enum bpf_prog_type ptype)
  87 {
  88         u32 ufd = attr->target_fd;
  89         struct bpf_prog *prog;
  90         struct bpf_map *map;
  91         struct fd f;
  92         int ret;
  93
  94         if (attr->attach_flags || attr->replace_bpf_fd)
  95                 return -EINVAL;
  96
  97         f = fdget(ufd);
  98         map = __bpf_map_get(f);
  99         if (IS_ERR(map))
 100                 return PTR_ERR(map);
 101
 102         prog = bpf_prog_get(attr->attach_bpf_fd);
 103         if (IS_ERR(prog)) {
 104                 ret = PTR_ERR(prog);
 105                 goto put_map;
 106         }
 107
 108         if (prog->type != ptype) {
 109                 ret = -EINVAL;
 110                 goto put_prog;
 111         }
 112
 113         ret = sock_map_prog_update(map, NULL, prog, attr->attach_type);
 114 put_prog:
 115         bpf_prog_put(prog);
 116 put_map:
 117         fdput(f);
 118         return ret;
 119 }
 120
 121 static void sock_map_sk_acquire(struct sock *sk)
 122         __acquires(&sk->sk_lock.slock)
 123 {
 124         lock_sock(sk);
 125         preempt_disable();
 126         rcu_read_lock();
 127 }
 128
 129 static void sock_map_sk_release(struct sock *sk)
 130         __releases(&sk->sk_lock.slock)
 131 {
 132         rcu_read_unlock();
 133         preempt_enable();
 134         release_sock(sk);
 135 }
 136
 137 static void sock_map_add_link(struct sk_psock *psock,
 138                               struct sk_psock_link *link,
 139                               struct bpf_map *map, void *link_raw)
 140 {
 141         link->link_raw = link_raw;
 142         link->map = map;
 143         spin_lock_bh(&psock->link_lock);
 144         list_add_tail(&link->list, &psock->link);
 145         spin_unlock_bh(&psock->link_lock);
 146 }
 147
 148 static void sock_map_del_link(struct sock *sk,
 149                               struct sk_psock *psock, void *link_raw)
 150 {
 151         bool strp_stop = false, verdict_stop = false;
 152         struct sk_psock_link *link, *tmp;
 153
 154         spin_lock_bh(&psock->link_lock);
 155         list_for_each_entry_safe(link, tmp, &psock->link, list) {
 156                 if (link->link_raw == link_raw) {
 157                         struct bpf_map *map = link->map;
 158                         struct bpf_stab *stab = container_of(map, struct bpf_stab,
 159                                                              map);
 160                         if (psock->parser.enabled && stab->progs.skb_parser)
 161                                 strp_stop = true;
 162                         if (psock->parser.enabled && stab->progs.skb_verdict)
 163                                 verdict_stop = true;
 164                         list_del(&link->list);
 165                         sk_psock_free_link(link);
 166                 }
 167         }
 168         spin_unlock_bh(&psock->link_lock);
 169         if (strp_stop || verdict_stop) {
 170                 write_lock_bh(&sk->sk_callback_lock);
 171                 if (strp_stop)
 172                         sk_psock_stop_strp(sk, psock);
 173                 else
 174                         sk_psock_stop_verdict(sk, psock);
 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         struct proto *prot;
 192
 193         switch (sk->sk_type) {
 194         case SOCK_STREAM:
 195                 prot = tcp_bpf_get_proto(sk, psock);
 196                 break;
 197
 198         case SOCK_DGRAM:
 199                 prot = udp_bpf_get_proto(sk, psock);
 200                 break;
 201
 202         default:
 203                 return -EINVAL;
 204         }
 205
 206         if (IS_ERR(prot))
 207                 return PTR_ERR(prot);
 208
 209         sk_psock_update_proto(sk, psock, prot);
 210         return 0;
 211 }
 212
 213 static struct sk_psock *sock_map_psock_get_checked(struct sock *sk)
 214 {
 215         struct sk_psock *psock;
 216
 217         rcu_read_lock();
 218         psock = sk_psock(sk);
 219         if (psock) {
 220                 if (sk->sk_prot->close != sock_map_close) {
 221                         psock = ERR_PTR(-EBUSY);
 222                         goto out;
 223                 }
 224
 225                 if (!refcount_inc_not_zero(&psock->refcnt))
 226                         psock = ERR_PTR(-EBUSY);
 227         }
 228 out:
 229         rcu_read_unlock();
 230         return psock;
 231 }
 232
 233 static int sock_map_link(struct bpf_map *map, struct sk_psock_progs *progs,
 234                          struct sock *sk)
 235 {
 236         struct bpf_prog *msg_parser, *skb_parser, *skb_verdict;
 237         struct sk_psock *psock;
 238         int ret;
 239
 240         skb_verdict = READ_ONCE(progs->skb_verdict);
 241         skb_parser = READ_ONCE(progs->skb_parser);
 242         if (skb_verdict) {
 243                 skb_verdict = bpf_prog_inc_not_zero(skb_verdict);
 244                 if (IS_ERR(skb_verdict))
 245                         return PTR_ERR(skb_verdict);
 246         }
 247         if (skb_parser) {
 248                 skb_parser = bpf_prog_inc_not_zero(skb_parser);
 249                 if (IS_ERR(skb_parser)) {
 250                         bpf_prog_put(skb_verdict);
 251                         return PTR_ERR(skb_parser);
 252                 }
 253         }
 254
 255         msg_parser = READ_ONCE(progs->msg_parser);
 256         if (msg_parser) {
 257                 msg_parser = bpf_prog_inc_not_zero(msg_parser);
 258                 if (IS_ERR(msg_parser)) {
 259                         ret = PTR_ERR(msg_parser);
 260                         goto out;
 261                 }
 262         }
 263
 264         psock = sock_map_psock_get_checked(sk);
 265         if (IS_ERR(psock)) {
 266                 ret = PTR_ERR(psock);
 267                 goto out_progs;
 268         }
 269
 270         if (psock) {
 271                 if ((msg_parser && READ_ONCE(psock->progs.msg_parser)) ||
 272                     (skb_parser  && READ_ONCE(psock->progs.skb_parser)) ||
 273                     (skb_verdict && READ_ONCE(psock->progs.skb_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
 289         ret = sock_map_init_proto(sk, psock);
 290         if (ret < 0)
 291                 goto out_drop;
 292
 293         write_lock_bh(&sk->sk_callback_lock);
 294         if (skb_parser && skb_verdict && !psock->parser.enabled) {
 295                 ret = sk_psock_init_strp(sk, psock);
 296                 if (ret)
 297                         goto out_unlock_drop;
 298                 psock_set_prog(&psock->progs.skb_verdict, skb_verdict);
 299                 psock_set_prog(&psock->progs.skb_parser, skb_parser);
 300                 sk_psock_start_strp(sk, psock);
 301         } else if (!skb_parser && skb_verdict && !psock->parser.enabled) {
 302                 psock_set_prog(&psock->progs.skb_verdict, skb_verdict);
 303                 sk_psock_start_verdict(sk,psock);
 304         }
 305         write_unlock_bh(&sk->sk_callback_lock);
 306         return 0;
 307 out_unlock_drop:
 308         write_unlock_bh(&sk->sk_callback_lock);
 309 out_drop:
 310         sk_psock_put(sk, psock);
 311 out_progs:
 312         if (msg_parser)
 313                 bpf_prog_put(msg_parser);
 314 out:
 315         if (skb_verdict)
 316                 bpf_prog_put(skb_verdict);
 317         if (skb_parser)
 318                 bpf_prog_put(skb_parser);
 319         return ret;
 320 }
 321
 322 static int sock_map_link_no_progs(struct bpf_map *map, struct sock *sk)
 323 {
 324         struct sk_psock *psock;
 325         int ret;
 326
 327         psock = sock_map_psock_get_checked(sk);
 328         if (IS_ERR(psock))
 329                 return PTR_ERR(psock);
 330
 331         if (!psock) {
 332                 psock = sk_psock_init(sk, map->numa_node);
 333                 if (IS_ERR(psock))
 334                         return PTR_ERR(psock);
 335         }
 336
 337         ret = sock_map_init_proto(sk, psock);
 338         if (ret < 0)
 339                 sk_psock_put(sk, psock);
 340         return ret;
 341 }
 342
 343 static void sock_map_free(struct bpf_map *map)
 344 {
 345         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 346         int i;
 347
 348         /* After the sync no updates or deletes will be in-flight so it
 349          * is safe to walk map and remove entries without risking a race
 350          * in EEXIST update case.
 351          */
 352         synchronize_rcu();
 353         for (i = 0; i < stab->map.max_entries; i++) {
 354                 struct sock **psk = &stab->sks[i];
 355                 struct sock *sk;
 356
 357                 sk = xchg(psk, NULL);
 358                 if (sk) {
 359                         lock_sock(sk);
 360                         rcu_read_lock();
 361                         sock_map_unref(sk, psk);
 362                         rcu_read_unlock();
 363                         release_sock(sk);
 364                 }
 365         }
 366
 367         /* wait for psock readers accessing its map link */
 368         synchronize_rcu();
 369
 370         bpf_map_area_free(stab->sks);
 371         kfree(stab);
 372 }
 373
 374 static void sock_map_release_progs(struct bpf_map *map)
 375 {
 376         psock_progs_drop(&container_of(map, struct bpf_stab, map)->progs);
 377 }
 378
 379 static struct sock *__sock_map_lookup_elem(struct bpf_map *map, u32 key)
 380 {
 381         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 382
 383         WARN_ON_ONCE(!rcu_read_lock_held());
 384
 385         if (unlikely(key >= map->max_entries))
 386                 return NULL;
 387         return READ_ONCE(stab->sks[key]);
 388 }
 389
 390 static void *sock_map_lookup(struct bpf_map *map, void *key)
 391 {
 392         struct sock *sk;
 393
 394         sk = __sock_map_lookup_elem(map, *(u32 *)key);
 395         if (!sk)
 396                 return NULL;
 397         if (sk_is_refcounted(sk) && !refcount_inc_not_zero(&sk->sk_refcnt))
 398                 return NULL;
 399         return sk;
 400 }
 401
 402 static void *sock_map_lookup_sys(struct bpf_map *map, void *key)
 403 {
 404         struct sock *sk;
 405
 406         if (map->value_size != sizeof(u64))
 407                 return ERR_PTR(-ENOSPC);
 408
 409         sk = __sock_map_lookup_elem(map, *(u32 *)key);
 410         if (!sk)
 411                 return ERR_PTR(-ENOENT);
 412
 413         __sock_gen_cookie(sk);
 414         return &sk->sk_cookie;
 415 }
 416
 417 static int __sock_map_delete(struct bpf_stab *stab, struct sock *sk_test,
 418                              struct sock **psk)
 419 {
 420         struct sock *sk;
 421         int err = 0;
 422
 423         raw_spin_lock_bh(&stab->lock);
 424         sk = *psk;
 425         if (!sk_test || sk_test == sk)
 426                 sk = xchg(psk, NULL);
 427
 428         if (likely(sk))
 429                 sock_map_unref(sk, psk);
 430         else
 431                 err = -EINVAL;
 432
 433         raw_spin_unlock_bh(&stab->lock);
 434         return err;
 435 }
 436
 437 static void sock_map_delete_from_link(struct bpf_map *map, struct sock *sk,
 438                                       void *link_raw)
 439 {
 440         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 441
 442         __sock_map_delete(stab, sk, link_raw);
 443 }
 444
 445 static int sock_map_delete_elem(struct bpf_map *map, void *key)
 446 {
 447         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 448         u32 i = *(u32 *)key;
 449         struct sock **psk;
 450
 451         if (unlikely(i >= map->max_entries))
 452                 return -EINVAL;
 453
 454         psk = &stab->sks[i];
 455         return __sock_map_delete(stab, NULL, psk);
 456 }
 457
 458 static int sock_map_get_next_key(struct bpf_map *map, void *key, void *next)
 459 {
 460         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 461         u32 i = key ? *(u32 *)key : U32_MAX;
 462         u32 *key_next = next;
 463
 464         if (i == stab->map.max_entries - 1)
 465                 return -ENOENT;
 466         if (i >= stab->map.max_entries)
 467                 *key_next = 0;
 468         else
 469                 *key_next = i + 1;
 470         return 0;
 471 }
 472
 473 static bool sock_map_redirect_allowed(const struct sock *sk);
 474
 475 static int sock_map_update_common(struct bpf_map *map, u32 idx,
 476                                   struct sock *sk, u64 flags)
 477 {
 478         struct bpf_stab *stab = container_of(map, struct bpf_stab, map);
 479         struct sk_psock_link *link;
 480         struct sk_psock *psock;
 481         struct sock *osk;
 482         int ret;
 483
 484         WARN_ON_ONCE(!rcu_read_lock_held());
 485         if (unlikely(flags > BPF_EXIST))
 486                 return -EINVAL;
 487         if (unlikely(idx >= map->max_entries))
 488                 return -E2BIG;
 489
 490         link = sk_psock_init_link();
 491         if (!link)
 492                 return -ENOMEM;
 493
 494         /* Only sockets we can redirect into/from in BPF need to hold
 495          * refs to parser/verdict progs and have their sk_data_ready
 496          * and sk_write_space callbacks overridden.
 497          */
 498         if (sock_map_redirect_allowed(sk))
 499                 ret = sock_map_link(map, &stab->progs, sk);
 500         else
 501                 ret = sock_map_link_no_progs(map, sk);
 502         if (ret < 0)
 503                 goto out_free;
 504
 505         psock = sk_psock(sk);
 506         WARN_ON_ONCE(!psock);
 507
 508         raw_spin_lock_bh(&stab->lock);
 509         osk = stab->sks[idx];
 510         if (osk && flags == BPF_NOEXIST) {
 511                 ret = -EEXIST;
 512                 goto out_unlock;
 513         } else if (!osk && flags == BPF_EXIST) {
 514                 ret = -ENOENT;
 515                 goto out_unlock;
 516         }
 517
 518         sock_map_add_link(psock, link, map, &stab->sks[idx]);
 519         stab->sks[idx] = sk;
 520         if (osk)
 521                 sock_map_unref(osk, &stab->sks[idx]);
 522         raw_spin_unlock_bh(&stab->lock);
 523         return 0;
 524 out_unlock:
 525         raw_spin_unlock_bh(&stab->lock);
 526         if (psock)
 527                 sk_psock_put(sk, psock);
 528 out_free:
 529         sk_psock_free_link(link);
 530         return ret;
 531 }
 532
 533 static bool sock_map_op_okay(const struct bpf_sock_ops_kern *ops)
 534 {
 535         return ops->op == BPF_SOCK_OPS_PASSIVE_ESTABLISHED_CB ||
 536                ops->op == BPF_SOCK_OPS_ACTIVE_ESTABLISHED_CB ||
 537                ops->op == BPF_SOCK_OPS_TCP_LISTEN_CB;
 538 }
 539
 540 static bool sk_is_tcp(const struct sock *sk)
 541 {
 542         return sk->sk_type == SOCK_STREAM &&
 543                sk->sk_protocol == IPPROTO_TCP;
 544 }
 545
 546 static bool sk_is_udp(const struct sock *sk)
 547 {
 548         return sk->sk_type == SOCK_DGRAM &&
 549                sk->sk_protocol == IPPROTO_UDP;
 550 }
 551
 552 static bool sock_map_redirect_allowed(const struct sock *sk)
 553 {
 554         return sk_is_tcp(sk) && sk->sk_state != TCP_LISTEN;
 555 }
 556
 557 static bool sock_map_sk_is_suitable(const struct sock *sk)
 558 {
 559         return sk_is_tcp(sk) || sk_is_udp(sk);
 560 }
 561
 562 static bool sock_map_sk_state_allowed(const struct sock *sk)
 563 {
 564         if (sk_is_tcp(sk))
 565                 return (1 << sk->sk_state) & (TCPF_ESTABLISHED | TCPF_LISTEN);
 566         else if (sk_is_udp(sk))
 567                 return sk_hashed(sk);
 568
 569         return false;
 570 }
 571
 572 static int sock_hash_update_common(struct bpf_map *map, void *key,
 573                                    struct sock *sk, u64 flags);
 574
 575 int sock_map_update_elem_sys(struct bpf_map *map, void *key, void *value,
 576                              u64 flags)
 577 {
 578         struct socket *sock;
 579         struct sock *sk;
 580         int ret;
 581         u64 ufd;
 582
 583         if (map->value_size == sizeof(u64))
 584                 ufd = *(u64 *)value;
 585         else
 586                 ufd = *(u32 *)value;
 587         if (ufd > S32_MAX)
 588                 return -EINVAL;
 589
 590         sock = sockfd_lookup(ufd, &ret);
 591         if (!sock)
 592                 return ret;
 593         sk = sock->sk;
 594         if (!sk) {
 595                 ret = -EINVAL;
 596                 goto out;
 597         }
 598         if (!sock_map_sk_is_suitable(sk)) {
 599                 ret = -EOPNOTSUPP;
 600                 goto out;
 601         }
 602
 603         sock_map_sk_acquire(sk);
 604         if (!sock_map_sk_state_allowed(sk))
 605                 ret = -EOPNOTSUPP;
 606         else if (map->map_type == BPF_MAP_TYPE_SOCKMAP)
 607                 ret = sock_map_update_common(map, *(u32 *)key, sk, flags);
 608         else
 609                 ret = sock_hash_update_common(map, key, sk, flags);
 610         sock_map_sk_release(sk);
 611 out:
 612         fput(sock->file);
 613         return ret;
 614 }
 615
 616 static int sock_map_update_elem(struct bpf_map *map, void *key,
 617                                 void *value, u64 flags)
 618 {
 619         struct sock *sk = (struct sock *)value;
 620         int ret;
 621
 622         if (unlikely(!sk || !sk_fullsock(sk)))
 623                 return -EINVAL;
 624
 625         if (!sock_map_sk_is_suitable(sk))
 626                 return -EOPNOTSUPP;
 627
 628         local_bh_disable();
 629         bh_lock_sock(sk);
 630         if (!sock_map_sk_state_allowed(sk))
 631                 ret = -EOPNOTSUPP;
 632         else if (map->map_type == BPF_MAP_TYPE_SOCKMAP)
 633                 ret = sock_map_update_common(map, *(u32 *)key, sk, flags);
 634         else
 635                 ret = sock_hash_update_common(map, key, sk, flags);
 636         bh_unlock_sock(sk);
 637         local_bh_enable();
 638         return ret;
 639 }
 640
 641 BPF_CALL_4(bpf_sock_map_update, struct bpf_sock_ops_kern *, sops,
 642            struct bpf_map *, map, void *, key, u64, flags)
 643 {
 644         WARN_ON_ONCE(!rcu_read_lock_held());
 645
 646         if (likely(sock_map_sk_is_suitable(sops->sk) &&
 647                    sock_map_op_okay(sops)))
 648                 return sock_map_update_common(map, *(u32 *)key, sops->sk,
 649                                               flags);
 650         return -EOPNOTSUPP;
 651 }
 652
 653 const struct bpf_func_proto bpf_sock_map_update_proto = {
 654         .func           = bpf_sock_map_update,
 655         .gpl_only       = false,
 656         .pkt_access     = true,
 657         .ret_type       = RET_INTEGER,
 658         .arg1_type      = ARG_PTR_TO_CTX,
 659         .arg2_type      = ARG_CONST_MAP_PTR,
 660         .arg3_type      = ARG_PTR_TO_MAP_KEY,
 661         .arg4_type      = ARG_ANYTHING,
 662 };
 663
 664 BPF_CALL_4(bpf_sk_redirect_map, struct sk_buff *, skb,
 665            struct bpf_map *, map, u32, key, u64, flags)
 666 {
 667         struct tcp_skb_cb *tcb = TCP_SKB_CB(skb);
 668         struct sock *sk;
 669
 670         if (unlikely(flags & ~(BPF_F_INGRESS)))
 671                 return SK_DROP;
 672
 673         sk = __sock_map_lookup_elem(map, key);
 674         if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
 675                 return SK_DROP;
 676
 677         tcb->bpf.flags = flags;
 678         tcb->bpf.sk_redir = sk;
 679         return SK_PASS;
 680 }
 681
 682 const struct bpf_func_proto bpf_sk_redirect_map_proto = {
 683         .func           = bpf_sk_redirect_map,
 684         .gpl_only       = false,
 685         .ret_type       = RET_INTEGER,
 686         .arg1_type      = ARG_PTR_TO_CTX,
 687         .arg2_type      = ARG_CONST_MAP_PTR,
 688         .arg3_type      = ARG_ANYTHING,
 689         .arg4_type      = ARG_ANYTHING,
 690 };
 691
 692 BPF_CALL_4(bpf_msg_redirect_map, struct sk_msg *, msg,
 693            struct bpf_map *, map, u32, key, u64, flags)
 694 {
 695         struct sock *sk;
 696
 697         if (unlikely(flags & ~(BPF_F_INGRESS)))
 698                 return SK_DROP;
 699
 700         sk = __sock_map_lookup_elem(map, key);
 701         if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
 702                 return SK_DROP;
 703
 704         msg->flags = flags;
 705         msg->sk_redir = sk;
 706         return SK_PASS;
 707 }
 708
 709 const struct bpf_func_proto bpf_msg_redirect_map_proto = {
 710         .func           = bpf_msg_redirect_map,
 711         .gpl_only       = false,
 712         .ret_type       = RET_INTEGER,
 713         .arg1_type      = ARG_PTR_TO_CTX,
 714         .arg2_type      = ARG_CONST_MAP_PTR,
 715         .arg3_type      = ARG_ANYTHING,
 716         .arg4_type      = ARG_ANYTHING,
 717 };
 718
 719 struct sock_map_seq_info {
 720         struct bpf_map *map;
 721         struct sock *sk;
 722         u32 index;
 723 };
 724
 725 struct bpf_iter__sockmap {
 726         __bpf_md_ptr(struct bpf_iter_meta *, meta);
 727         __bpf_md_ptr(struct bpf_map *, map);
 728         __bpf_md_ptr(void *, key);
 729         __bpf_md_ptr(struct sock *, sk);
 730 };
 731
 732 DEFINE_BPF_ITER_FUNC(sockmap, struct bpf_iter_meta *meta,
 733                      struct bpf_map *map, void *key,
 734                      struct sock *sk)
 735
 736 static void *sock_map_seq_lookup_elem(struct sock_map_seq_info *info)
 737 {
 738         if (unlikely(info->index >= info->map->max_entries))
 739                 return NULL;
 740
 741         info->sk = __sock_map_lookup_elem(info->map, info->index);
 742
 743         /* can't return sk directly, since that might be NULL */
 744         return info;
 745 }
 746
 747 static void *sock_map_seq_start(struct seq_file *seq, loff_t *pos)
 748 {
 749         struct sock_map_seq_info *info = seq->private;
 750
 751         if (*pos == 0)
 752                 ++*pos;
 753
 754         /* pairs with sock_map_seq_stop */
 755         rcu_read_lock();
 756         return sock_map_seq_lookup_elem(info);
 757 }
 758
 759 static void *sock_map_seq_next(struct seq_file *seq, void *v, loff_t *pos)
 760 {
 761         struct sock_map_seq_info *info = seq->private;
 762
 763         ++*pos;
 764         ++info->index;
 765
 766         return sock_map_seq_lookup_elem(info);
 767 }
 768
 769 static int sock_map_seq_show(struct seq_file *seq, void *v)
 770 {
 771         struct sock_map_seq_info *info = seq->private;
 772         struct bpf_iter__sockmap ctx = {};
 773         struct bpf_iter_meta meta;
 774         struct bpf_prog *prog;
 775
 776         meta.seq = seq;
 777         prog = bpf_iter_get_info(&meta, !v);
 778         if (!prog)
 779                 return 0;
 780
 781         ctx.meta = &meta;
 782         ctx.map = info->map;
 783         if (v) {
 784                 ctx.key = &info->index;
 785                 ctx.sk = info->sk;
 786         }
 787
 788         return bpf_iter_run_prog(prog, &ctx);
 789 }
 790
 791 static void sock_map_seq_stop(struct seq_file *seq, void *v)
 792 {
 793         if (!v)
 794                 (void)sock_map_seq_show(seq, NULL);
 795
 796         /* pairs with sock_map_seq_start */
 797         rcu_read_unlock();
 798 }
 799
 800 static const struct seq_operations sock_map_seq_ops = {
 801         .start  = sock_map_seq_start,
 802         .next   = sock_map_seq_next,
 803         .stop   = sock_map_seq_stop,
 804         .show   = sock_map_seq_show,
 805 };
 806
 807 static int sock_map_init_seq_private(void *priv_data,
 808                                      struct bpf_iter_aux_info *aux)
 809 {
 810         struct sock_map_seq_info *info = priv_data;
 811
 812         info->map = aux->map;
 813         return 0;
 814 }
 815
 816 static const struct bpf_iter_seq_info sock_map_iter_seq_info = {
 817         .seq_ops                = &sock_map_seq_ops,
 818         .init_seq_private       = sock_map_init_seq_private,
 819         .seq_priv_size          = sizeof(struct sock_map_seq_info),
 820 };
 821
 822 static int sock_map_btf_id;
 823 const struct bpf_map_ops sock_map_ops = {
 824         .map_meta_equal         = bpf_map_meta_equal,
 825         .map_alloc              = sock_map_alloc,
 826         .map_free               = sock_map_free,
 827         .map_get_next_key       = sock_map_get_next_key,
 828         .map_lookup_elem_sys_only = sock_map_lookup_sys,
 829         .map_update_elem        = sock_map_update_elem,
 830         .map_delete_elem        = sock_map_delete_elem,
 831         .map_lookup_elem        = sock_map_lookup,
 832         .map_release_uref       = sock_map_release_progs,
 833         .map_check_btf          = map_check_no_btf,
 834         .map_btf_name           = "bpf_stab",
 835         .map_btf_id             = &sock_map_btf_id,
 836         .iter_seq_info          = &sock_map_iter_seq_info,
 837 };
 838
 839 struct bpf_shtab_elem {
 840         struct rcu_head rcu;
 841         u32 hash;
 842         struct sock *sk;
 843         struct hlist_node node;
 844         u8 key[];
 845 };
 846
 847 struct bpf_shtab_bucket {
 848         struct hlist_head head;
 849         raw_spinlock_t lock;
 850 };
 851
 852 struct bpf_shtab {
 853         struct bpf_map map;
 854         struct bpf_shtab_bucket *buckets;
 855         u32 buckets_num;
 856         u32 elem_size;
 857         struct sk_psock_progs progs;
 858         atomic_t count;
 859 };
 860
 861 static inline u32 sock_hash_bucket_hash(const void *key, u32 len)
 862 {
 863         return jhash(key, len, 0);
 864 }
 865
 866 static struct bpf_shtab_bucket *sock_hash_select_bucket(struct bpf_shtab *htab,
 867                                                         u32 hash)
 868 {
 869         return &htab->buckets[hash & (htab->buckets_num - 1)];
 870 }
 871
 872 static struct bpf_shtab_elem *
 873 sock_hash_lookup_elem_raw(struct hlist_head *head, u32 hash, void *key,
 874                           u32 key_size)
 875 {
 876         struct bpf_shtab_elem *elem;
 877
 878         hlist_for_each_entry_rcu(elem, head, node) {
 879                 if (elem->hash == hash &&
 880                     !memcmp(&elem->key, key, key_size))
 881                         return elem;
 882         }
 883
 884         return NULL;
 885 }
 886
 887 static struct sock *__sock_hash_lookup_elem(struct bpf_map *map, void *key)
 888 {
 889         struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
 890         u32 key_size = map->key_size, hash;
 891         struct bpf_shtab_bucket *bucket;
 892         struct bpf_shtab_elem *elem;
 893
 894         WARN_ON_ONCE(!rcu_read_lock_held());
 895
 896         hash = sock_hash_bucket_hash(key, key_size);
 897         bucket = sock_hash_select_bucket(htab, hash);
 898         elem = sock_hash_lookup_elem_raw(&bucket->head, hash, key, key_size);
 899
 900         return elem ? elem->sk : NULL;
 901 }
 902
 903 static void sock_hash_free_elem(struct bpf_shtab *htab,
 904                                 struct bpf_shtab_elem *elem)
 905 {
 906         atomic_dec(&htab->count);
 907         kfree_rcu(elem, rcu);
 908 }
 909
 910 static void sock_hash_delete_from_link(struct bpf_map *map, struct sock *sk,
 911                                        void *link_raw)
 912 {
 913         struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
 914         struct bpf_shtab_elem *elem_probe, *elem = link_raw;
 915         struct bpf_shtab_bucket *bucket;
 916
 917         WARN_ON_ONCE(!rcu_read_lock_held());
 918         bucket = sock_hash_select_bucket(htab, elem->hash);
 919
 920         /* elem may be deleted in parallel from the map, but access here
 921          * is okay since it's going away only after RCU grace period.
 922          * However, we need to check whether it's still present.
 923          */
 924         raw_spin_lock_bh(&bucket->lock);
 925         elem_probe = sock_hash_lookup_elem_raw(&bucket->head, elem->hash,
 926                                                elem->key, map->key_size);
 927         if (elem_probe && elem_probe == elem) {
 928                 hlist_del_rcu(&elem->node);
 929                 sock_map_unref(elem->sk, elem);
 930                 sock_hash_free_elem(htab, elem);
 931         }
 932         raw_spin_unlock_bh(&bucket->lock);
 933 }
 934
 935 static int sock_hash_delete_elem(struct bpf_map *map, void *key)
 936 {
 937         struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
 938         u32 hash, key_size = map->key_size;
 939         struct bpf_shtab_bucket *bucket;
 940         struct bpf_shtab_elem *elem;
 941         int ret = -ENOENT;
 942
 943         hash = sock_hash_bucket_hash(key, key_size);
 944         bucket = sock_hash_select_bucket(htab, hash);
 945
 946         raw_spin_lock_bh(&bucket->lock);
 947         elem = sock_hash_lookup_elem_raw(&bucket->head, hash, key, key_size);
 948         if (elem) {
 949                 hlist_del_rcu(&elem->node);
 950                 sock_map_unref(elem->sk, elem);
 951                 sock_hash_free_elem(htab, elem);
 952                 ret = 0;
 953         }
 954         raw_spin_unlock_bh(&bucket->lock);
 955         return ret;
 956 }
 957
 958 static struct bpf_shtab_elem *sock_hash_alloc_elem(struct bpf_shtab *htab,
 959                                                    void *key, u32 key_size,
 960                                                    u32 hash, struct sock *sk,
 961                                                    struct bpf_shtab_elem *old)
 962 {
 963         struct bpf_shtab_elem *new;
 964
 965         if (atomic_inc_return(&htab->count) > htab->map.max_entries) {
 966                 if (!old) {
 967                         atomic_dec(&htab->count);
 968                         return ERR_PTR(-E2BIG);
 969                 }
 970         }
 971
 972         new = kmalloc_node(htab->elem_size, GFP_ATOMIC | __GFP_NOWARN,
 973                            htab->map.numa_node);
 974         if (!new) {
 975                 atomic_dec(&htab->count);
 976                 return ERR_PTR(-ENOMEM);
 977         }
 978         memcpy(new->key, key, key_size);
 979         new->sk = sk;
 980         new->hash = hash;
 981         return new;
 982 }
 983
 984 static int sock_hash_update_common(struct bpf_map *map, void *key,
 985                                    struct sock *sk, u64 flags)
 986 {
 987         struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
 988         u32 key_size = map->key_size, hash;
 989         struct bpf_shtab_elem *elem, *elem_new;
 990         struct bpf_shtab_bucket *bucket;
 991         struct sk_psock_link *link;
 992         struct sk_psock *psock;
 993         int ret;
 994
 995         WARN_ON_ONCE(!rcu_read_lock_held());
 996         if (unlikely(flags > BPF_EXIST))
 997                 return -EINVAL;
 998
 999         link = sk_psock_init_link();
1000         if (!link)
1001                 return -ENOMEM;
1002
1003         /* Only sockets we can redirect into/from in BPF need to hold
1004          * refs to parser/verdict progs and have their sk_data_ready
1005          * and sk_write_space callbacks overridden.
1006          */
1007         if (sock_map_redirect_allowed(sk))
1008                 ret = sock_map_link(map, &htab->progs, sk);
1009         else
1010                 ret = sock_map_link_no_progs(map, sk);
1011         if (ret < 0)
1012                 goto out_free;
1013
1014         psock = sk_psock(sk);
1015         WARN_ON_ONCE(!psock);
1016
1017         hash = sock_hash_bucket_hash(key, key_size);
1018         bucket = sock_hash_select_bucket(htab, hash);
1019
1020         raw_spin_lock_bh(&bucket->lock);
1021         elem = sock_hash_lookup_elem_raw(&bucket->head, hash, key, key_size);
1022         if (elem && flags == BPF_NOEXIST) {
1023                 ret = -EEXIST;
1024                 goto out_unlock;
1025         } else if (!elem && flags == BPF_EXIST) {
1026                 ret = -ENOENT;
1027                 goto out_unlock;
1028         }
1029
1030         elem_new = sock_hash_alloc_elem(htab, key, key_size, hash, sk, elem);
1031         if (IS_ERR(elem_new)) {
1032                 ret = PTR_ERR(elem_new);
1033                 goto out_unlock;
1034         }
1035
1036         sock_map_add_link(psock, link, map, elem_new);
1037         /* Add new element to the head of the list, so that
1038          * concurrent search will find it before old elem.
1039          */
1040         hlist_add_head_rcu(&elem_new->node, &bucket->head);
1041         if (elem) {
1042                 hlist_del_rcu(&elem->node);
1043                 sock_map_unref(elem->sk, elem);
1044                 sock_hash_free_elem(htab, elem);
1045         }
1046         raw_spin_unlock_bh(&bucket->lock);
1047         return 0;
1048 out_unlock:
1049         raw_spin_unlock_bh(&bucket->lock);
1050         sk_psock_put(sk, psock);
1051 out_free:
1052         sk_psock_free_link(link);
1053         return ret;
1054 }
1055
1056 static int sock_hash_get_next_key(struct bpf_map *map, void *key,
1057                                   void *key_next)
1058 {
1059         struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
1060         struct bpf_shtab_elem *elem, *elem_next;
1061         u32 hash, key_size = map->key_size;
1062         struct hlist_head *head;
1063         int i = 0;
1064
1065         if (!key)
1066                 goto find_first_elem;
1067         hash = sock_hash_bucket_hash(key, key_size);
1068         head = &sock_hash_select_bucket(htab, hash)->head;
1069         elem = sock_hash_lookup_elem_raw(head, hash, key, key_size);
1070         if (!elem)
1071                 goto find_first_elem;
1072
1073         elem_next = hlist_entry_safe(rcu_dereference(hlist_next_rcu(&elem->node)),
1074                                      struct bpf_shtab_elem, node);
1075         if (elem_next) {
1076                 memcpy(key_next, elem_next->key, key_size);
1077                 return 0;
1078         }
1079
1080         i = hash & (htab->buckets_num - 1);
1081         i++;
1082 find_first_elem:
1083         for (; i < htab->buckets_num; i++) {
1084                 head = &sock_hash_select_bucket(htab, i)->head;
1085                 elem_next = hlist_entry_safe(rcu_dereference(hlist_first_rcu(head)),
1086                                              struct bpf_shtab_elem, node);
1087                 if (elem_next) {
1088                         memcpy(key_next, elem_next->key, key_size);
1089                         return 0;
1090                 }
1091         }
1092
1093         return -ENOENT;
1094 }
1095
1096 static struct bpf_map *sock_hash_alloc(union bpf_attr *attr)
1097 {
1098         struct bpf_shtab *htab;
1099         int i, err;
1100         u64 cost;
1101
1102         if (!capable(CAP_NET_ADMIN))
1103                 return ERR_PTR(-EPERM);
1104         if (attr->max_entries == 0 ||
1105             attr->key_size    == 0 ||
1106             (attr->value_size != sizeof(u32) &&
1107              attr->value_size != sizeof(u64)) ||
1108             attr->map_flags & ~SOCK_CREATE_FLAG_MASK)
1109                 return ERR_PTR(-EINVAL);
1110         if (attr->key_size > MAX_BPF_STACK)
1111                 return ERR_PTR(-E2BIG);
1112
1113         htab = kzalloc(sizeof(*htab), GFP_USER);
1114         if (!htab)
1115                 return ERR_PTR(-ENOMEM);
1116
1117         bpf_map_init_from_attr(&htab->map, attr);
1118
1119         htab->buckets_num = roundup_pow_of_two(htab->map.max_entries);
1120         htab->elem_size = sizeof(struct bpf_shtab_elem) +
1121                           round_up(htab->map.key_size, 8);
1122         if (htab->buckets_num == 0 ||
1123             htab->buckets_num > U32_MAX / sizeof(struct bpf_shtab_bucket)) {
1124                 err = -EINVAL;
1125                 goto free_htab;
1126         }
1127
1128         cost = (u64) htab->buckets_num * sizeof(struct bpf_shtab_bucket) +
1129                (u64) htab->elem_size * htab->map.max_entries;
1130         if (cost >= U32_MAX - PAGE_SIZE) {
1131                 err = -EINVAL;
1132                 goto free_htab;
1133         }
1134         err = bpf_map_charge_init(&htab->map.memory, cost);
1135         if (err)
1136                 goto free_htab;
1137
1138         htab->buckets = bpf_map_area_alloc(htab->buckets_num *
1139                                            sizeof(struct bpf_shtab_bucket),
1140                                            htab->map.numa_node);
1141         if (!htab->buckets) {
1142                 bpf_map_charge_finish(&htab->map.memory);
1143                 err = -ENOMEM;
1144                 goto free_htab;
1145         }
1146
1147         for (i = 0; i < htab->buckets_num; i++) {
1148                 INIT_HLIST_HEAD(&htab->buckets[i].head);
1149                 raw_spin_lock_init(&htab->buckets[i].lock);
1150         }
1151
1152         return &htab->map;
1153 free_htab:
1154         kfree(htab);
1155         return ERR_PTR(err);
1156 }
1157
1158 static void sock_hash_free(struct bpf_map *map)
1159 {
1160         struct bpf_shtab *htab = container_of(map, struct bpf_shtab, map);
1161         struct bpf_shtab_bucket *bucket;
1162         struct hlist_head unlink_list;
1163         struct bpf_shtab_elem *elem;
1164         struct hlist_node *node;
1165         int i;
1166
1167         /* After the sync no updates or deletes will be in-flight so it
1168          * is safe to walk map and remove entries without risking a race
1169          * in EEXIST update case.
1170          */
1171         synchronize_rcu();
1172         for (i = 0; i < htab->buckets_num; i++) {
1173                 bucket = sock_hash_select_bucket(htab, i);
1174
1175                 /* We are racing with sock_hash_delete_from_link to
1176                  * enter the spin-lock critical section. Every socket on
1177                  * the list is still linked to sockhash. Since link
1178                  * exists, psock exists and holds a ref to socket. That
1179                  * lets us to grab a socket ref too.
1180                  */
1181                 raw_spin_lock_bh(&bucket->lock);
1182                 hlist_for_each_entry(elem, &bucket->head, node)
1183                         sock_hold(elem->sk);
1184                 hlist_move_list(&bucket->head, &unlink_list);
1185                 raw_spin_unlock_bh(&bucket->lock);
1186
1187                 /* Process removed entries out of atomic context to
1188                  * block for socket lock before deleting the psock's
1189                  * link to sockhash.
1190                  */
1191                 hlist_for_each_entry_safe(elem, node, &unlink_list, node) {
1192                         hlist_del(&elem->node);
1193                         lock_sock(elem->sk);
1194                         rcu_read_lock();
1195                         sock_map_unref(elem->sk, elem);
1196                         rcu_read_unlock();
1197                         release_sock(elem->sk);
1198                         sock_put(elem->sk);
1199                         sock_hash_free_elem(htab, elem);
1200                 }
1201         }
1202
1203         /* wait for psock readers accessing its map link */
1204         synchronize_rcu();
1205
1206         bpf_map_area_free(htab->buckets);
1207         kfree(htab);
1208 }
1209
1210 static void *sock_hash_lookup_sys(struct bpf_map *map, void *key)
1211 {
1212         struct sock *sk;
1213
1214         if (map->value_size != sizeof(u64))
1215                 return ERR_PTR(-ENOSPC);
1216
1217         sk = __sock_hash_lookup_elem(map, key);
1218         if (!sk)
1219                 return ERR_PTR(-ENOENT);
1220
1221         __sock_gen_cookie(sk);
1222         return &sk->sk_cookie;
1223 }
1224
1225 static void *sock_hash_lookup(struct bpf_map *map, void *key)
1226 {
1227         struct sock *sk;
1228
1229         sk = __sock_hash_lookup_elem(map, key);
1230         if (!sk)
1231                 return NULL;
1232         if (sk_is_refcounted(sk) && !refcount_inc_not_zero(&sk->sk_refcnt))
1233                 return NULL;
1234         return sk;
1235 }
1236
1237 static void sock_hash_release_progs(struct bpf_map *map)
1238 {
1239         psock_progs_drop(&container_of(map, struct bpf_shtab, map)->progs);
1240 }
1241
1242 BPF_CALL_4(bpf_sock_hash_update, struct bpf_sock_ops_kern *, sops,
1243            struct bpf_map *, map, void *, key, u64, flags)
1244 {
1245         WARN_ON_ONCE(!rcu_read_lock_held());
1246
1247         if (likely(sock_map_sk_is_suitable(sops->sk) &&
1248                    sock_map_op_okay(sops)))
1249                 return sock_hash_update_common(map, key, sops->sk, flags);
1250         return -EOPNOTSUPP;
1251 }
1252
1253 const struct bpf_func_proto bpf_sock_hash_update_proto = {
1254         .func           = bpf_sock_hash_update,
1255         .gpl_only       = false,
1256         .pkt_access     = true,
1257         .ret_type       = RET_INTEGER,
1258         .arg1_type      = ARG_PTR_TO_CTX,
1259         .arg2_type      = ARG_CONST_MAP_PTR,
1260         .arg3_type      = ARG_PTR_TO_MAP_KEY,
1261         .arg4_type      = ARG_ANYTHING,
1262 };
1263
1264 BPF_CALL_4(bpf_sk_redirect_hash, struct sk_buff *, skb,
1265            struct bpf_map *, map, void *, key, u64, flags)
1266 {
1267         struct tcp_skb_cb *tcb = TCP_SKB_CB(skb);
1268         struct sock *sk;
1269
1270         if (unlikely(flags & ~(BPF_F_INGRESS)))
1271                 return SK_DROP;
1272
1273         sk = __sock_hash_lookup_elem(map, key);
1274         if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
1275                 return SK_DROP;
1276
1277         tcb->bpf.flags = flags;
1278         tcb->bpf.sk_redir = sk;
1279         return SK_PASS;
1280 }
1281
1282 const struct bpf_func_proto bpf_sk_redirect_hash_proto = {
1283         .func           = bpf_sk_redirect_hash,
1284         .gpl_only       = false,
1285         .ret_type       = RET_INTEGER,
1286         .arg1_type      = ARG_PTR_TO_CTX,
1287         .arg2_type      = ARG_CONST_MAP_PTR,
1288         .arg3_type      = ARG_PTR_TO_MAP_KEY,
1289         .arg4_type      = ARG_ANYTHING,
1290 };
1291
1292 BPF_CALL_4(bpf_msg_redirect_hash, struct sk_msg *, msg,
1293            struct bpf_map *, map, void *, key, u64, flags)
1294 {
1295         struct sock *sk;
1296
1297         if (unlikely(flags & ~(BPF_F_INGRESS)))
1298                 return SK_DROP;
1299
1300         sk = __sock_hash_lookup_elem(map, key);
1301         if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
1302                 return SK_DROP;
1303
1304         msg->flags = flags;
1305         msg->sk_redir = sk;
1306         return SK_PASS;
1307 }
1308
1309 const struct bpf_func_proto bpf_msg_redirect_hash_proto = {
1310         .func           = bpf_msg_redirect_hash,
1311         .gpl_only       = false,
1312         .ret_type       = RET_INTEGER,
1313         .arg1_type      = ARG_PTR_TO_CTX,
1314         .arg2_type      = ARG_CONST_MAP_PTR,
1315         .arg3_type      = ARG_PTR_TO_MAP_KEY,
1316         .arg4_type      = ARG_ANYTHING,
1317 };
1318
1319 struct sock_hash_seq_info {
1320         struct bpf_map *map;
1321         struct bpf_shtab *htab;
1322         u32 bucket_id;
1323 };
1324
1325 static void *sock_hash_seq_find_next(struct sock_hash_seq_info *info,
1326                                      struct bpf_shtab_elem *prev_elem)
1327 {
1328         const struct bpf_shtab *htab = info->htab;
1329         struct bpf_shtab_bucket *bucket;
1330         struct bpf_shtab_elem *elem;
1331         struct hlist_node *node;
1332
1333         /* try to find next elem in the same bucket */
1334         if (prev_elem) {
1335                 node = rcu_dereference(hlist_next_rcu(&prev_elem->node));
1336                 elem = hlist_entry_safe(node, struct bpf_shtab_elem, node);
1337                 if (elem)
1338                         return elem;
1339
1340                 /* no more elements, continue in the next bucket */
1341                 info->bucket_id++;
1342         }
1343
1344         for (; info->bucket_id < htab->buckets_num; info->bucket_id++) {
1345                 bucket = &htab->buckets[info->bucket_id];
1346                 node = rcu_dereference(hlist_first_rcu(&bucket->head));
1347                 elem = hlist_entry_safe(node, struct bpf_shtab_elem, node);
1348                 if (elem)
1349                         return elem;
1350         }
1351
1352         return NULL;
1353 }
1354
1355 static void *sock_hash_seq_start(struct seq_file *seq, loff_t *pos)
1356 {
1357         struct sock_hash_seq_info *info = seq->private;
1358
1359         if (*pos == 0)
1360                 ++*pos;
1361
1362         /* pairs with sock_hash_seq_stop */
1363         rcu_read_lock();
1364         return sock_hash_seq_find_next(info, NULL);
1365 }
1366
1367 static void *sock_hash_seq_next(struct seq_file *seq, void *v, loff_t *pos)
1368 {
1369         struct sock_hash_seq_info *info = seq->private;
1370
1371         ++*pos;
1372         return sock_hash_seq_find_next(info, v);
1373 }
1374
1375 static int sock_hash_seq_show(struct seq_file *seq, void *v)
1376 {
1377         struct sock_hash_seq_info *info = seq->private;
1378         struct bpf_iter__sockmap ctx = {};
1379         struct bpf_shtab_elem *elem = v;
1380         struct bpf_iter_meta meta;
1381         struct bpf_prog *prog;
1382
1383         meta.seq = seq;
1384         prog = bpf_iter_get_info(&meta, !elem);
1385         if (!prog)
1386                 return 0;
1387
1388         ctx.meta = &meta;
1389         ctx.map = info->map;
1390         if (elem) {
1391                 ctx.key = elem->key;
1392                 ctx.sk = elem->sk;
1393         }
1394
1395         return bpf_iter_run_prog(prog, &ctx);
1396 }
1397
1398 static void sock_hash_seq_stop(struct seq_file *seq, void *v)
1399 {
1400         if (!v)
1401                 (void)sock_hash_seq_show(seq, NULL);
1402
1403         /* pairs with sock_hash_seq_start */
1404         rcu_read_unlock();
1405 }
1406
1407 static const struct seq_operations sock_hash_seq_ops = {
1408         .start  = sock_hash_seq_start,
1409         .next   = sock_hash_seq_next,
1410         .stop   = sock_hash_seq_stop,
1411         .show   = sock_hash_seq_show,
1412 };
1413
1414 static int sock_hash_init_seq_private(void *priv_data,
1415                                      struct bpf_iter_aux_info *aux)
1416 {
1417         struct sock_hash_seq_info *info = priv_data;
1418
1419         info->map = aux->map;
1420         info->htab = container_of(aux->map, struct bpf_shtab, map);
1421         return 0;
1422 }
1423
1424 static const struct bpf_iter_seq_info sock_hash_iter_seq_info = {
1425         .seq_ops                = &sock_hash_seq_ops,
1426         .init_seq_private       = sock_hash_init_seq_private,
1427         .seq_priv_size          = sizeof(struct sock_hash_seq_info),
1428 };
1429
1430 static int sock_hash_map_btf_id;
1431 const struct bpf_map_ops sock_hash_ops = {
1432         .map_meta_equal         = bpf_map_meta_equal,
1433         .map_alloc              = sock_hash_alloc,
1434         .map_free               = sock_hash_free,
1435         .map_get_next_key       = sock_hash_get_next_key,
1436         .map_update_elem        = sock_map_update_elem,
1437         .map_delete_elem        = sock_hash_delete_elem,
1438         .map_lookup_elem        = sock_hash_lookup,
1439         .map_lookup_elem_sys_only = sock_hash_lookup_sys,
1440         .map_release_uref       = sock_hash_release_progs,
1441         .map_check_btf          = map_check_no_btf,
1442         .map_btf_name           = "bpf_shtab",
1443         .map_btf_id             = &sock_hash_map_btf_id,
1444         .iter_seq_info          = &sock_hash_iter_seq_info,
1445 };
1446
1447 static struct sk_psock_progs *sock_map_progs(struct bpf_map *map)
1448 {
1449         switch (map->map_type) {
1450         case BPF_MAP_TYPE_SOCKMAP:
1451                 return &container_of(map, struct bpf_stab, map)->progs;
1452         case BPF_MAP_TYPE_SOCKHASH:
1453                 return &container_of(map, struct bpf_shtab, map)->progs;
1454         default:
1455                 break;
1456         }
1457
1458         return NULL;
1459 }
1460
1461 int sock_map_prog_update(struct bpf_map *map, struct bpf_prog *prog,
1462                          struct bpf_prog *old, u32 which)
1463 {
1464         struct sk_psock_progs *progs = sock_map_progs(map);
1465         struct bpf_prog **pprog;
1466
1467         if (!progs)
1468                 return -EOPNOTSUPP;
1469
1470         switch (which) {
1471         case BPF_SK_MSG_VERDICT:
1472                 pprog = &progs->msg_parser;
1473                 break;
1474         case BPF_SK_SKB_STREAM_PARSER:
1475                 pprog = &progs->skb_parser;
1476                 break;
1477         case BPF_SK_SKB_STREAM_VERDICT:
1478                 pprog = &progs->skb_verdict;
1479                 break;
1480         default:
1481                 return -EOPNOTSUPP;
1482         }
1483
1484         if (old)
1485                 return psock_replace_prog(pprog, prog, old);
1486
1487         psock_set_prog(pprog, prog);
1488         return 0;
1489 }
1490
1491 static void sock_map_unlink(struct sock *sk, struct sk_psock_link *link)
1492 {
1493         switch (link->map->map_type) {
1494         case BPF_MAP_TYPE_SOCKMAP:
1495                 return sock_map_delete_from_link(link->map, sk,
1496                                                  link->link_raw);
1497         case BPF_MAP_TYPE_SOCKHASH:
1498                 return sock_hash_delete_from_link(link->map, sk,
1499                                                   link->link_raw);
1500         default:
1501                 break;
1502         }
1503 }
1504
1505 static void sock_map_remove_links(struct sock *sk, struct sk_psock *psock)
1506 {
1507         struct sk_psock_link *link;
1508
1509         while ((link = sk_psock_link_pop(psock))) {
1510                 sock_map_unlink(sk, link);
1511                 sk_psock_free_link(link);
1512         }
1513 }
1514
1515 void sock_map_unhash(struct sock *sk)
1516 {
1517         void (*saved_unhash)(struct sock *sk);
1518         struct sk_psock *psock;
1519
1520         rcu_read_lock();
1521         psock = sk_psock(sk);
1522         if (unlikely(!psock)) {
1523                 rcu_read_unlock();
1524                 if (sk->sk_prot->unhash)
1525                         sk->sk_prot->unhash(sk);
1526                 return;
1527         }
1528
1529         saved_unhash = psock->saved_unhash;
1530         sock_map_remove_links(sk, psock);
1531         rcu_read_unlock();
1532         saved_unhash(sk);
1533 }
1534
1535 void sock_map_close(struct sock *sk, long timeout)
1536 {
1537         void (*saved_close)(struct sock *sk, long timeout);
1538         struct sk_psock *psock;
1539
1540         lock_sock(sk);
1541         rcu_read_lock();
1542         psock = sk_psock(sk);
1543         if (unlikely(!psock)) {
1544                 rcu_read_unlock();
1545                 release_sock(sk);
1546                 return sk->sk_prot->close(sk, timeout);
1547         }
1548
1549         saved_close = psock->saved_close;
1550         sock_map_remove_links(sk, psock);
1551         rcu_read_unlock();
1552         release_sock(sk);
1553         saved_close(sk, timeout);
1554 }
1555
1556 static int sock_map_iter_attach_target(struct bpf_prog *prog,
1557                                        union bpf_iter_link_info *linfo,
1558                                        struct bpf_iter_aux_info *aux)
1559 {
1560         struct bpf_map *map;
1561         int err = -EINVAL;
1562
1563         if (!linfo->map.map_fd)
1564                 return -EBADF;
1565
1566         map = bpf_map_get_with_uref(linfo->map.map_fd);
1567         if (IS_ERR(map))
1568                 return PTR_ERR(map);
1569
1570         if (map->map_type != BPF_MAP_TYPE_SOCKMAP &&
1571             map->map_type != BPF_MAP_TYPE_SOCKHASH)
1572                 goto put_map;
1573
1574         if (prog->aux->max_rdonly_access > map->key_size) {
1575                 err = -EACCES;
1576                 goto put_map;
1577         }
1578
1579         aux->map = map;
1580         return 0;
1581
1582 put_map:
1583         bpf_map_put_with_uref(map);
1584         return err;
1585 }
1586
1587 static void sock_map_iter_detach_target(struct bpf_iter_aux_info *aux)
1588 {
1589         bpf_map_put_with_uref(aux->map);
1590 }
1591
1592 static struct bpf_iter_reg sock_map_iter_reg = {
1593         .target                 = "sockmap",
1594         .attach_target          = sock_map_iter_attach_target,
1595         .detach_target          = sock_map_iter_detach_target,
1596         .show_fdinfo            = bpf_iter_map_show_fdinfo,
1597         .fill_link_info         = bpf_iter_map_fill_link_info,
1598         .ctx_arg_info_size      = 2,
1599         .ctx_arg_info           = {
1600                 { offsetof(struct bpf_iter__sockmap, key),
1601                   PTR_TO_RDONLY_BUF_OR_NULL },
1602                 { offsetof(struct bpf_iter__sockmap, sk),
1603                   PTR_TO_BTF_ID_OR_NULL },
1604         },
1605 };
1606
1607 static int __init bpf_sockmap_iter_init(void)
1608 {
1609         sock_map_iter_reg.ctx_arg_info[1].btf_id =
1610                 btf_sock_ids[BTF_SOCK_TYPE_SOCK];
1611         return bpf_iter_reg_target(&sock_map_iter_reg);
1612 }
1613 late_initcall(bpf_sockmap_iter_init);