net/mptcp/pm_netlink.c

   1 // SPDX-License-Identifier: GPL-2.0
   2 /* Multipath TCP
   3  *
   4  * Copyright (c) 2020, Red Hat, Inc.
   5  */
   6
   7 #define pr_fmt(fmt) "MPTCP: " fmt
   8
   9 #include <linux/inet.h>
  10 #include <linux/kernel.h>
  11 #include <net/tcp.h>
  12 #include <net/netns/generic.h>
  13 #include <net/mptcp.h>
  14 #include <net/genetlink.h>
  15 #include <uapi/linux/mptcp.h>
  16
  17 #include "protocol.h"
  18 #include "mib.h"
  19
  20 /* forward declaration */
  21 static struct genl_family mptcp_genl_family;
  22
  23 static int pm_nl_pernet_id;
  24
  25 struct mptcp_pm_add_entry {
  26         struct list_head        list;
  27         struct mptcp_addr_info  addr;
  28         struct timer_list       add_timer;
  29         struct mptcp_sock       *sock;
  30         u8                      retrans_times;
  31 };
  32
  33 struct pm_nl_pernet {
  34         /* protects pernet updates */
  35         spinlock_t              lock;
  36         struct list_head        local_addr_list;
  37         unsigned int            addrs;
  38         unsigned int            stale_loss_cnt;
  39         unsigned int            add_addr_signal_max;
  40         unsigned int            add_addr_accept_max;
  41         unsigned int            local_addr_max;
  42         unsigned int            subflows_max;
  43         unsigned int            next_id;
  44         DECLARE_BITMAP(id_bitmap, MPTCP_PM_MAX_ADDR_ID + 1);
  45 };
  46
  47 #define MPTCP_PM_ADDR_MAX       8
  48 #define ADD_ADDR_RETRANS_MAX    3
  49
  50 static struct pm_nl_pernet *pm_nl_get_pernet(const struct net *net)
  51 {
  52         return net_generic(net, pm_nl_pernet_id);
  53 }
  54
  55 static struct pm_nl_pernet *
  56 pm_nl_get_pernet_from_msk(const struct mptcp_sock *msk)
  57 {
  58         return pm_nl_get_pernet(sock_net((struct sock *)msk));
  59 }
  60
  61 bool mptcp_addresses_equal(const struct mptcp_addr_info *a,
  62                            const struct mptcp_addr_info *b, bool use_port)
  63 {
  64         bool addr_equals = false;
  65
  66         if (a->family == b->family) {
  67                 if (a->family == AF_INET)
  68                         addr_equals = a->addr.s_addr == b->addr.s_addr;
  69 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
  70                 else
  71                         addr_equals = !ipv6_addr_cmp(&a->addr6, &b->addr6);
  72         } else if (a->family == AF_INET) {
  73                 if (ipv6_addr_v4mapped(&b->addr6))
  74                         addr_equals = a->addr.s_addr == b->addr6.s6_addr32[3];
  75         } else if (b->family == AF_INET) {
  76                 if (ipv6_addr_v4mapped(&a->addr6))
  77                         addr_equals = a->addr6.s6_addr32[3] == b->addr.s_addr;
  78 #endif
  79         }
  80
  81         if (!addr_equals)
  82                 return false;
  83         if (!use_port)
  84                 return true;
  85
  86         return a->port == b->port;
  87 }
  88
  89 static void local_address(const struct sock_common *skc,
  90                           struct mptcp_addr_info *addr)
  91 {
  92         addr->family = skc->skc_family;
  93         addr->port = htons(skc->skc_num);
  94         if (addr->family == AF_INET)
  95                 addr->addr.s_addr = skc->skc_rcv_saddr;
  96 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
  97         else if (addr->family == AF_INET6)
  98                 addr->addr6 = skc->skc_v6_rcv_saddr;
  99 #endif
 100 }
 101
 102 static void remote_address(const struct sock_common *skc,
 103                            struct mptcp_addr_info *addr)
 104 {
 105         addr->family = skc->skc_family;
 106         addr->port = skc->skc_dport;
 107         if (addr->family == AF_INET)
 108                 addr->addr.s_addr = skc->skc_daddr;
 109 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
 110         else if (addr->family == AF_INET6)
 111                 addr->addr6 = skc->skc_v6_daddr;
 112 #endif
 113 }
 114
 115 static bool lookup_subflow_by_saddr(const struct list_head *list,
 116                                     const struct mptcp_addr_info *saddr)
 117 {
 118         struct mptcp_subflow_context *subflow;
 119         struct mptcp_addr_info cur;
 120         struct sock_common *skc;
 121
 122         list_for_each_entry(subflow, list, node) {
 123                 skc = (struct sock_common *)mptcp_subflow_tcp_sock(subflow);
 124
 125                 local_address(skc, &cur);
 126                 if (mptcp_addresses_equal(&cur, saddr, saddr->port))
 127                         return true;
 128         }
 129
 130         return false;
 131 }
 132
 133 static bool lookup_subflow_by_daddr(const struct list_head *list,
 134                                     const struct mptcp_addr_info *daddr)
 135 {
 136         struct mptcp_subflow_context *subflow;
 137         struct mptcp_addr_info cur;
 138         struct sock_common *skc;
 139
 140         list_for_each_entry(subflow, list, node) {
 141                 skc = (struct sock_common *)mptcp_subflow_tcp_sock(subflow);
 142
 143                 remote_address(skc, &cur);
 144                 if (mptcp_addresses_equal(&cur, daddr, daddr->port))
 145                         return true;
 146         }
 147
 148         return false;
 149 }
 150
 151 static struct mptcp_pm_addr_entry *
 152 select_local_address(const struct pm_nl_pernet *pernet,
 153                      const struct mptcp_sock *msk)
 154 {
 155         const struct sock *sk = (const struct sock *)msk;
 156         struct mptcp_pm_addr_entry *entry, *ret = NULL;
 157
 158         msk_owned_by_me(msk);
 159
 160         rcu_read_lock();
 161         list_for_each_entry_rcu(entry, &pernet->local_addr_list, list) {
 162                 if (!(entry->flags & MPTCP_PM_ADDR_FLAG_SUBFLOW))
 163                         continue;
 164
 165                 if (!test_bit(entry->addr.id, msk->pm.id_avail_bitmap))
 166                         continue;
 167
 168                 if (entry->addr.family != sk->sk_family) {
 169 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
 170                         if ((entry->addr.family == AF_INET &&
 171                              !ipv6_addr_v4mapped(&sk->sk_v6_daddr)) ||
 172                             (sk->sk_family == AF_INET &&
 173                              !ipv6_addr_v4mapped(&entry->addr.addr6)))
 174 #endif
 175                                 continue;
 176                 }
 177
 178                 ret = entry;
 179                 break;
 180         }
 181         rcu_read_unlock();
 182         return ret;
 183 }
 184
 185 static struct mptcp_pm_addr_entry *
 186 select_signal_address(struct pm_nl_pernet *pernet, const struct mptcp_sock *msk)
 187 {
 188         struct mptcp_pm_addr_entry *entry, *ret = NULL;
 189
 190         rcu_read_lock();
 191         /* do not keep any additional per socket state, just signal
 192          * the address list in order.
 193          * Note: removal from the local address list during the msk life-cycle
 194          * can lead to additional addresses not being announced.
 195          */
 196         list_for_each_entry_rcu(entry, &pernet->local_addr_list, list) {
 197                 if (!test_bit(entry->addr.id, msk->pm.id_avail_bitmap))
 198                         continue;
 199
 200                 if (!(entry->flags & MPTCP_PM_ADDR_FLAG_SIGNAL))
 201                         continue;
 202
 203                 ret = entry;
 204                 break;
 205         }
 206         rcu_read_unlock();
 207         return ret;
 208 }
 209
 210 unsigned int mptcp_pm_get_add_addr_signal_max(const struct mptcp_sock *msk)
 211 {
 212         const struct pm_nl_pernet *pernet = pm_nl_get_pernet_from_msk(msk);
 213
 214         return READ_ONCE(pernet->add_addr_signal_max);
 215 }
 216 EXPORT_SYMBOL_GPL(mptcp_pm_get_add_addr_signal_max);
 217
 218 unsigned int mptcp_pm_get_add_addr_accept_max(const struct mptcp_sock *msk)
 219 {
 220         struct pm_nl_pernet *pernet = pm_nl_get_pernet_from_msk(msk);
 221
 222         return READ_ONCE(pernet->add_addr_accept_max);
 223 }
 224 EXPORT_SYMBOL_GPL(mptcp_pm_get_add_addr_accept_max);
 225
 226 unsigned int mptcp_pm_get_subflows_max(const struct mptcp_sock *msk)
 227 {
 228         struct pm_nl_pernet *pernet = pm_nl_get_pernet_from_msk(msk);
 229
 230         return READ_ONCE(pernet->subflows_max);
 231 }
 232 EXPORT_SYMBOL_GPL(mptcp_pm_get_subflows_max);
 233
 234 unsigned int mptcp_pm_get_local_addr_max(const struct mptcp_sock *msk)
 235 {
 236         struct pm_nl_pernet *pernet = pm_nl_get_pernet_from_msk(msk);
 237
 238         return READ_ONCE(pernet->local_addr_max);
 239 }
 240 EXPORT_SYMBOL_GPL(mptcp_pm_get_local_addr_max);
 241
 242 bool mptcp_pm_nl_check_work_pending(struct mptcp_sock *msk)
 243 {
 244         struct pm_nl_pernet *pernet = pm_nl_get_pernet_from_msk(msk);
 245
 246         if (msk->pm.subflows == mptcp_pm_get_subflows_max(msk) ||
 247             (find_next_and_bit(pernet->id_bitmap, msk->pm.id_avail_bitmap,
 248                                MPTCP_PM_MAX_ADDR_ID + 1, 0) == MPTCP_PM_MAX_ADDR_ID + 1)) {
 249                 WRITE_ONCE(msk->pm.work_pending, false);
 250                 return false;
 251         }
 252         return true;
 253 }
 254
 255 struct mptcp_pm_add_entry *
 256 mptcp_lookup_anno_list_by_saddr(const struct mptcp_sock *msk,
 257                                 const struct mptcp_addr_info *addr)
 258 {
 259         struct mptcp_pm_add_entry *entry;
 260
 261         lockdep_assert_held(&msk->pm.lock);
 262
 263         list_for_each_entry(entry, &msk->pm.anno_list, list) {
 264                 if (mptcp_addresses_equal(&entry->addr, addr, true))
 265                         return entry;
 266         }
 267
 268         return NULL;
 269 }
 270
 271 bool mptcp_pm_sport_in_anno_list(struct mptcp_sock *msk, const struct sock *sk)
 272 {
 273         struct mptcp_pm_add_entry *entry;
 274         struct mptcp_addr_info saddr;
 275         bool ret = false;
 276
 277         local_address((struct sock_common *)sk, &saddr);
 278
 279         spin_lock_bh(&msk->pm.lock);
 280         list_for_each_entry(entry, &msk->pm.anno_list, list) {
 281                 if (mptcp_addresses_equal(&entry->addr, &saddr, true)) {
 282                         ret = true;
 283                         goto out;
 284                 }
 285         }
 286
 287 out:
 288         spin_unlock_bh(&msk->pm.lock);
 289         return ret;
 290 }
 291
 292 static void mptcp_pm_add_timer(struct timer_list *timer)
 293 {
 294         struct mptcp_pm_add_entry *entry = from_timer(entry, timer, add_timer);
 295         struct mptcp_sock *msk = entry->sock;
 296         struct sock *sk = (struct sock *)msk;
 297
 298         pr_debug("msk=%p", msk);
 299
 300         if (!msk)
 301                 return;
 302
 303         if (inet_sk_state_load(sk) == TCP_CLOSE)
 304                 return;
 305
 306         if (!entry->addr.id)
 307                 return;
 308
 309         if (mptcp_pm_should_add_signal_addr(msk)) {
 310                 sk_reset_timer(sk, timer, jiffies + TCP_RTO_MAX / 8);
 311                 goto out;
 312         }
 313
 314         spin_lock_bh(&msk->pm.lock);
 315
 316         if (!mptcp_pm_should_add_signal_addr(msk)) {
 317                 pr_debug("retransmit ADD_ADDR id=%d", entry->addr.id);
 318                 mptcp_pm_announce_addr(msk, &entry->addr, false);
 319                 mptcp_pm_add_addr_send_ack(msk);
 320                 entry->retrans_times++;
 321         }
 322
 323         if (entry->retrans_times < ADD_ADDR_RETRANS_MAX)
 324                 sk_reset_timer(sk, timer,
 325                                jiffies + mptcp_get_add_addr_timeout(sock_net(sk)));
 326
 327         spin_unlock_bh(&msk->pm.lock);
 328
 329         if (entry->retrans_times == ADD_ADDR_RETRANS_MAX)
 330                 mptcp_pm_subflow_established(msk);
 331
 332 out:
 333         __sock_put(sk);
 334 }
 335
 336 struct mptcp_pm_add_entry *
 337 mptcp_pm_del_add_timer(struct mptcp_sock *msk,
 338                        const struct mptcp_addr_info *addr, bool check_id)
 339 {
 340         struct mptcp_pm_add_entry *entry;
 341         struct sock *sk = (struct sock *)msk;
 342
 343         spin_lock_bh(&msk->pm.lock);
 344         entry = mptcp_lookup_anno_list_by_saddr(msk, addr);
 345         if (entry && (!check_id || entry->addr.id == addr->id))
 346                 entry->retrans_times = ADD_ADDR_RETRANS_MAX;
 347         spin_unlock_bh(&msk->pm.lock);
 348
 349         if (entry && (!check_id || entry->addr.id == addr->id))
 350                 sk_stop_timer_sync(sk, &entry->add_timer);
 351
 352         return entry;
 353 }
 354
 355 bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk,
 356                               const struct mptcp_pm_addr_entry *entry)
 357 {
 358         struct mptcp_pm_add_entry *add_entry = NULL;
 359         struct sock *sk = (struct sock *)msk;
 360         struct net *net = sock_net(sk);
 361
 362         lockdep_assert_held(&msk->pm.lock);
 363
 364         add_entry = mptcp_lookup_anno_list_by_saddr(msk, &entry->addr);
 365
 366         if (add_entry) {
 367                 if (mptcp_pm_is_kernel(msk))
 368                         return false;
 369
 370                 sk_reset_timer(sk, &add_entry->add_timer,
 371                                jiffies + mptcp_get_add_addr_timeout(net));
 372                 return true;
 373         }
 374
 375         add_entry = kmalloc(sizeof(*add_entry), GFP_ATOMIC);
 376         if (!add_entry)
 377                 return false;
 378
 379         list_add(&add_entry->list, &msk->pm.anno_list);
 380
 381         add_entry->addr = entry->addr;
 382         add_entry->sock = msk;
 383         add_entry->retrans_times = 0;
 384
 385         timer_setup(&add_entry->add_timer, mptcp_pm_add_timer, 0);
 386         sk_reset_timer(sk, &add_entry->add_timer,
 387                        jiffies + mptcp_get_add_addr_timeout(net));
 388
 389         return true;
 390 }
 391
 392 void mptcp_pm_free_anno_list(struct mptcp_sock *msk)
 393 {
 394         struct mptcp_pm_add_entry *entry, *tmp;
 395         struct sock *sk = (struct sock *)msk;
 396         LIST_HEAD(free_list);
 397
 398         pr_debug("msk=%p", msk);
 399
 400         spin_lock_bh(&msk->pm.lock);
 401         list_splice_init(&msk->pm.anno_list, &free_list);
 402         spin_unlock_bh(&msk->pm.lock);
 403
 404         list_for_each_entry_safe(entry, tmp, &free_list, list) {
 405                 sk_stop_timer_sync(sk, &entry->add_timer);
 406                 kfree(entry);
 407         }
 408 }
 409
 410 static bool lookup_address_in_vec(const struct mptcp_addr_info *addrs, unsigned int nr,
 411                                   const struct mptcp_addr_info *addr)
 412 {
 413         int i;
 414
 415         for (i = 0; i < nr; i++) {
 416                 if (mptcp_addresses_equal(&addrs[i], addr, addr->port))
 417                         return true;
 418         }
 419
 420         return false;
 421 }
 422
 423 /* Fill all the remote addresses into the array addrs[],
 424  * and return the array size.
 425  */
 426 static unsigned int fill_remote_addresses_vec(struct mptcp_sock *msk, bool fullmesh,
 427                                               struct mptcp_addr_info *addrs)
 428 {
 429         bool deny_id0 = READ_ONCE(msk->pm.remote_deny_join_id0);
 430         struct sock *sk = (struct sock *)msk, *ssk;
 431         struct mptcp_subflow_context *subflow;
 432         struct mptcp_addr_info remote = { 0 };
 433         unsigned int subflows_max;
 434         int i = 0;
 435
 436         subflows_max = mptcp_pm_get_subflows_max(msk);
 437         remote_address((struct sock_common *)sk, &remote);
 438
 439         /* Non-fullmesh endpoint, fill in the single entry
 440          * corresponding to the primary MPC subflow remote address
 441          */
 442         if (!fullmesh) {
 443                 if (deny_id0)
 444                         return 0;
 445
 446                 msk->pm.subflows++;
 447                 addrs[i++] = remote;
 448         } else {
 449                 mptcp_for_each_subflow(msk, subflow) {
 450                         ssk = mptcp_subflow_tcp_sock(subflow);
 451                         remote_address((struct sock_common *)ssk, &addrs[i]);
 452                         if (deny_id0 && mptcp_addresses_equal(&addrs[i], &remote, false))
 453                                 continue;
 454
 455                         if (!lookup_address_in_vec(addrs, i, &addrs[i]) &&
 456                             msk->pm.subflows < subflows_max) {
 457                                 msk->pm.subflows++;
 458                                 i++;
 459                         }
 460                 }
 461         }
 462
 463         return i;
 464 }
 465
 466 static struct mptcp_pm_addr_entry *
 467 __lookup_addr_by_id(struct pm_nl_pernet *pernet, unsigned int id)
 468 {
 469         struct mptcp_pm_addr_entry *entry;
 470
 471         list_for_each_entry(entry, &pernet->local_addr_list, list) {
 472                 if (entry->addr.id == id)
 473                         return entry;
 474         }
 475         return NULL;
 476 }
 477
 478 static struct mptcp_pm_addr_entry *
 479 __lookup_addr(struct pm_nl_pernet *pernet, const struct mptcp_addr_info *info,
 480               bool lookup_by_id)
 481 {
 482         struct mptcp_pm_addr_entry *entry;
 483
 484         list_for_each_entry(entry, &pernet->local_addr_list, list) {
 485                 if ((!lookup_by_id && mptcp_addresses_equal(&entry->addr, info, true)) ||
 486                     (lookup_by_id && entry->addr.id == info->id))
 487                         return entry;
 488         }
 489         return NULL;
 490 }
 491
 492 static int
 493 lookup_id_by_addr(const struct pm_nl_pernet *pernet, const struct mptcp_addr_info *addr)
 494 {
 495         const struct mptcp_pm_addr_entry *entry;
 496         int ret = -1;
 497
 498         rcu_read_lock();
 499         list_for_each_entry(entry, &pernet->local_addr_list, list) {
 500                 if (mptcp_addresses_equal(&entry->addr, addr, entry->addr.port)) {
 501                         ret = entry->addr.id;
 502                         break;
 503                 }
 504         }
 505         rcu_read_unlock();
 506         return ret;
 507 }
 508
 509 static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk)
 510 {
 511         struct sock *sk = (struct sock *)msk;
 512         struct mptcp_pm_addr_entry *local;
 513         unsigned int add_addr_signal_max;
 514         unsigned int local_addr_max;
 515         struct pm_nl_pernet *pernet;
 516         unsigned int subflows_max;
 517
 518         pernet = pm_nl_get_pernet(sock_net(sk));
 519
 520         add_addr_signal_max = mptcp_pm_get_add_addr_signal_max(msk);
 521         local_addr_max = mptcp_pm_get_local_addr_max(msk);
 522         subflows_max = mptcp_pm_get_subflows_max(msk);
 523
 524         /* do lazy endpoint usage accounting for the MPC subflows */
 525         if (unlikely(!(msk->pm.status & BIT(MPTCP_PM_MPC_ENDPOINT_ACCOUNTED))) && msk->first) {
 526                 struct mptcp_addr_info mpc_addr;
 527                 int mpc_id;
 528
 529                 local_address((struct sock_common *)msk->first, &mpc_addr);
 530                 mpc_id = lookup_id_by_addr(pernet, &mpc_addr);
 531                 if (mpc_id >= 0)
 532                         __clear_bit(mpc_id, msk->pm.id_avail_bitmap);
 533
 534                 msk->pm.status |= BIT(MPTCP_PM_MPC_ENDPOINT_ACCOUNTED);
 535         }
 536
 537         pr_debug("local %d:%d signal %d:%d subflows %d:%d\n",
 538                  msk->pm.local_addr_used, local_addr_max,
 539                  msk->pm.add_addr_signaled, add_addr_signal_max,
 540                  msk->pm.subflows, subflows_max);
 541
 542         /* check first for announce */
 543         if (msk->pm.add_addr_signaled < add_addr_signal_max) {
 544                 local = select_signal_address(pernet, msk);
 545
 546                 /* due to racing events on both ends we can reach here while
 547                  * previous add address is still running: if we invoke now
 548                  * mptcp_pm_announce_addr(), that will fail and the
 549                  * corresponding id will be marked as used.
 550                  * Instead let the PM machinery reschedule us when the
 551                  * current address announce will be completed.
 552                  */
 553                 if (msk->pm.addr_signal & BIT(MPTCP_ADD_ADDR_SIGNAL))
 554                         return;
 555
 556                 if (local) {
 557                         if (mptcp_pm_alloc_anno_list(msk, local)) {
 558                                 __clear_bit(local->addr.id, msk->pm.id_avail_bitmap);
 559                                 msk->pm.add_addr_signaled++;
 560                                 mptcp_pm_announce_addr(msk, &local->addr, false);
 561                                 mptcp_pm_nl_addr_send_ack(msk);
 562                         }
 563                 }
 564         }
 565
 566         /* check if should create a new subflow */
 567         while (msk->pm.local_addr_used < local_addr_max &&
 568                msk->pm.subflows < subflows_max) {
 569                 struct mptcp_addr_info addrs[MPTCP_PM_ADDR_MAX];
 570                 bool fullmesh;
 571                 int i, nr;
 572
 573                 local = select_local_address(pernet, msk);
 574                 if (!local)
 575                         break;
 576
 577                 fullmesh = !!(local->flags & MPTCP_PM_ADDR_FLAG_FULLMESH);
 578
 579                 msk->pm.local_addr_used++;
 580                 nr = fill_remote_addresses_vec(msk, fullmesh, addrs);
 581                 if (nr)
 582                         __clear_bit(local->addr.id, msk->pm.id_avail_bitmap);
 583                 spin_unlock_bh(&msk->pm.lock);
 584                 for (i = 0; i < nr; i++)
 585                         __mptcp_subflow_connect(sk, &local->addr, &addrs[i]);
 586                 spin_lock_bh(&msk->pm.lock);
 587         }
 588         mptcp_pm_nl_check_work_pending(msk);
 589 }
 590
 591 static void mptcp_pm_nl_fully_established(struct mptcp_sock *msk)
 592 {
 593         mptcp_pm_create_subflow_or_signal_addr(msk);
 594 }
 595
 596 static void mptcp_pm_nl_subflow_established(struct mptcp_sock *msk)
 597 {
 598         mptcp_pm_create_subflow_or_signal_addr(msk);
 599 }
 600
 601 /* Fill all the local addresses into the array addrs[],
 602  * and return the array size.
 603  */
 604 static unsigned int fill_local_addresses_vec(struct mptcp_sock *msk,
 605                                              struct mptcp_addr_info *addrs)
 606 {
 607         struct sock *sk = (struct sock *)msk;
 608         struct mptcp_pm_addr_entry *entry;
 609         struct mptcp_addr_info local;
 610         struct pm_nl_pernet *pernet;
 611         unsigned int subflows_max;
 612         int i = 0;
 613
 614         pernet = pm_nl_get_pernet_from_msk(msk);
 615         subflows_max = mptcp_pm_get_subflows_max(msk);
 616
 617         rcu_read_lock();
 618         list_for_each_entry_rcu(entry, &pernet->local_addr_list, list) {
 619                 if (!(entry->flags & MPTCP_PM_ADDR_FLAG_FULLMESH))
 620                         continue;
 621
 622                 if (entry->addr.family != sk->sk_family) {
 623 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
 624                         if ((entry->addr.family == AF_INET &&
 625                              !ipv6_addr_v4mapped(&sk->sk_v6_daddr)) ||
 626                             (sk->sk_family == AF_INET &&
 627                              !ipv6_addr_v4mapped(&entry->addr.addr6)))
 628 #endif
 629                                 continue;
 630                 }
 631
 632                 if (msk->pm.subflows < subflows_max) {
 633                         msk->pm.subflows++;
 634                         addrs[i++] = entry->addr;
 635                 }
 636         }
 637         rcu_read_unlock();
 638
 639         /* If the array is empty, fill in the single
 640          * 'IPADDRANY' local address
 641          */
 642         if (!i) {
 643                 memset(&local, 0, sizeof(local));
 644                 local.family = msk->pm.remote.family;
 645
 646                 msk->pm.subflows++;
 647                 addrs[i++] = local;
 648         }
 649
 650         return i;
 651 }
 652
 653 static void mptcp_pm_nl_add_addr_received(struct mptcp_sock *msk)
 654 {
 655         struct mptcp_addr_info addrs[MPTCP_PM_ADDR_MAX];
 656         struct sock *sk = (struct sock *)msk;
 657         unsigned int add_addr_accept_max;
 658         struct mptcp_addr_info remote;
 659         unsigned int subflows_max;
 660         int i, nr;
 661
 662         add_addr_accept_max = mptcp_pm_get_add_addr_accept_max(msk);
 663         subflows_max = mptcp_pm_get_subflows_max(msk);
 664
 665         pr_debug("accepted %d:%d remote family %d",
 666                  msk->pm.add_addr_accepted, add_addr_accept_max,
 667                  msk->pm.remote.family);
 668
 669         remote = msk->pm.remote;
 670         mptcp_pm_announce_addr(msk, &remote, true);
 671         mptcp_pm_nl_addr_send_ack(msk);
 672
 673         if (lookup_subflow_by_daddr(&msk->conn_list, &remote))
 674                 return;
 675
 676         /* pick id 0 port, if none is provided the remote address */
 677         if (!remote.port)
 678                 remote.port = sk->sk_dport;
 679
 680         /* connect to the specified remote address, using whatever
 681          * local address the routing configuration will pick.
 682          */
 683         nr = fill_local_addresses_vec(msk, addrs);
 684
 685         msk->pm.add_addr_accepted++;
 686         if (msk->pm.add_addr_accepted >= add_addr_accept_max ||
 687             msk->pm.subflows >= subflows_max)
 688                 WRITE_ONCE(msk->pm.accept_addr, false);
 689
 690         spin_unlock_bh(&msk->pm.lock);
 691         for (i = 0; i < nr; i++)
 692                 __mptcp_subflow_connect(sk, &addrs[i], &remote);
 693         spin_lock_bh(&msk->pm.lock);
 694 }
 695
 696 void mptcp_pm_nl_addr_send_ack(struct mptcp_sock *msk)
 697 {
 698         struct mptcp_subflow_context *subflow;
 699
 700         msk_owned_by_me(msk);
 701         lockdep_assert_held(&msk->pm.lock);
 702
 703         if (!mptcp_pm_should_add_signal(msk) &&
 704             !mptcp_pm_should_rm_signal(msk))
 705                 return;
 706
 707         subflow = list_first_entry_or_null(&msk->conn_list, typeof(*subflow), node);
 708         if (subflow) {
 709                 struct sock *ssk = mptcp_subflow_tcp_sock(subflow);
 710
 711                 spin_unlock_bh(&msk->pm.lock);
 712                 pr_debug("send ack for %s",
 713                          mptcp_pm_should_add_signal(msk) ? "add_addr" : "rm_addr");
 714
 715                 mptcp_subflow_send_ack(ssk);
 716                 spin_lock_bh(&msk->pm.lock);
 717         }
 718 }
 719
 720 int mptcp_pm_nl_mp_prio_send_ack(struct mptcp_sock *msk,
 721                                  struct mptcp_addr_info *addr,
 722                                  struct mptcp_addr_info *rem,
 723                                  u8 bkup)
 724 {
 725         struct mptcp_subflow_context *subflow;
 726
 727         pr_debug("bkup=%d", bkup);
 728
 729         mptcp_for_each_subflow(msk, subflow) {
 730                 struct sock *ssk = mptcp_subflow_tcp_sock(subflow);
 731                 struct mptcp_addr_info local, remote;
 732                 bool slow;
 733
 734                 local_address((struct sock_common *)ssk, &local);
 735                 if (!mptcp_addresses_equal(&local, addr, addr->port))
 736                         continue;
 737
 738                 if (rem && rem->family != AF_UNSPEC) {
 739                         remote_address((struct sock_common *)ssk, &remote);
 740                         if (!mptcp_addresses_equal(&remote, rem, rem->port))
 741                                 continue;
 742                 }
 743
 744                 slow = lock_sock_fast(ssk);
 745                 if (subflow->backup != bkup)
 746                         msk->last_snd = NULL;
 747                 subflow->backup = bkup;
 748                 subflow->send_mp_prio = 1;
 749                 subflow->request_bkup = bkup;
 750
 751                 pr_debug("send ack for mp_prio");
 752                 __mptcp_subflow_send_ack(ssk);
 753                 unlock_sock_fast(ssk, slow);
 754
 755                 return 0;
 756         }
 757
 758         return -EINVAL;
 759 }
 760
 761 static void mptcp_pm_nl_rm_addr_or_subflow(struct mptcp_sock *msk,
 762                                            const struct mptcp_rm_list *rm_list,
 763                                            enum linux_mptcp_mib_field rm_type)
 764 {
 765         struct mptcp_subflow_context *subflow, *tmp;
 766         struct sock *sk = (struct sock *)msk;
 767         u8 i;
 768
 769         pr_debug("%s rm_list_nr %d",
 770                  rm_type == MPTCP_MIB_RMADDR ? "address" : "subflow", rm_list->nr);
 771
 772         msk_owned_by_me(msk);
 773
 774         if (sk->sk_state == TCP_LISTEN)
 775                 return;
 776
 777         if (!rm_list->nr)
 778                 return;
 779
 780         if (list_empty(&msk->conn_list))
 781                 return;
 782
 783         for (i = 0; i < rm_list->nr; i++) {
 784                 bool removed = false;
 785
 786                 list_for_each_entry_safe(subflow, tmp, &msk->conn_list, node) {
 787                         struct sock *ssk = mptcp_subflow_tcp_sock(subflow);
 788                         int how = RCV_SHUTDOWN | SEND_SHUTDOWN;
 789                         u8 id = subflow->local_id;
 790
 791                         if (rm_type == MPTCP_MIB_RMADDR)
 792                                 id = subflow->remote_id;
 793
 794                         if (rm_list->ids[i] != id)
 795                                 continue;
 796
 797                         pr_debug(" -> %s rm_list_ids[%d]=%u local_id=%u remote_id=%u",
 798                                  rm_type == MPTCP_MIB_RMADDR ? "address" : "subflow",
 799                                  i, rm_list->ids[i], subflow->local_id, subflow->remote_id);
 800                         spin_unlock_bh(&msk->pm.lock);
 801                         mptcp_subflow_shutdown(sk, ssk, how);
 802
 803                         /* the following takes care of updating the subflows counter */
 804                         mptcp_close_ssk(sk, ssk, subflow);
 805                         spin_lock_bh(&msk->pm.lock);
 806
 807                         removed = true;
 808                         __MPTCP_INC_STATS(sock_net(sk), rm_type);
 809                 }
 810                 if (rm_type == MPTCP_MIB_RMSUBFLOW)
 811                         __set_bit(rm_list->ids[i], msk->pm.id_avail_bitmap);
 812                 if (!removed)
 813                         continue;
 814
 815                 if (!mptcp_pm_is_kernel(msk))
 816                         continue;
 817
 818                 if (rm_type == MPTCP_MIB_RMADDR) {
 819                         msk->pm.add_addr_accepted--;
 820                         WRITE_ONCE(msk->pm.accept_addr, true);
 821                 } else if (rm_type == MPTCP_MIB_RMSUBFLOW) {
 822                         msk->pm.local_addr_used--;
 823                 }
 824         }
 825 }
 826
 827 static void mptcp_pm_nl_rm_addr_received(struct mptcp_sock *msk)
 828 {
 829         mptcp_pm_nl_rm_addr_or_subflow(msk, &msk->pm.rm_list_rx, MPTCP_MIB_RMADDR);
 830 }
 831
 832 void mptcp_pm_nl_rm_subflow_received(struct mptcp_sock *msk,
 833                                      const struct mptcp_rm_list *rm_list)
 834 {
 835         mptcp_pm_nl_rm_addr_or_subflow(msk, rm_list, MPTCP_MIB_RMSUBFLOW);
 836 }
 837
 838 void mptcp_pm_nl_work(struct mptcp_sock *msk)
 839 {
 840         struct mptcp_pm_data *pm = &msk->pm;
 841
 842         msk_owned_by_me(msk);
 843
 844         if (!(pm->status & MPTCP_PM_WORK_MASK))
 845                 return;
 846
 847         spin_lock_bh(&msk->pm.lock);
 848
 849         pr_debug("msk=%p status=%x", msk, pm->status);
 850         if (pm->status & BIT(MPTCP_PM_ADD_ADDR_RECEIVED)) {
 851                 pm->status &= ~BIT(MPTCP_PM_ADD_ADDR_RECEIVED);
 852                 mptcp_pm_nl_add_addr_received(msk);
 853         }
 854         if (pm->status & BIT(MPTCP_PM_ADD_ADDR_SEND_ACK)) {
 855                 pm->status &= ~BIT(MPTCP_PM_ADD_ADDR_SEND_ACK);
 856                 mptcp_pm_nl_addr_send_ack(msk);
 857         }
 858         if (pm->status & BIT(MPTCP_PM_RM_ADDR_RECEIVED)) {
 859                 pm->status &= ~BIT(MPTCP_PM_RM_ADDR_RECEIVED);
 860                 mptcp_pm_nl_rm_addr_received(msk);
 861         }
 862         if (pm->status & BIT(MPTCP_PM_ESTABLISHED)) {
 863                 pm->status &= ~BIT(MPTCP_PM_ESTABLISHED);
 864                 mptcp_pm_nl_fully_established(msk);
 865         }
 866         if (pm->status & BIT(MPTCP_PM_SUBFLOW_ESTABLISHED)) {
 867                 pm->status &= ~BIT(MPTCP_PM_SUBFLOW_ESTABLISHED);
 868                 mptcp_pm_nl_subflow_established(msk);
 869         }
 870
 871         spin_unlock_bh(&msk->pm.lock);
 872 }
 873
 874 static bool address_use_port(struct mptcp_pm_addr_entry *entry)
 875 {
 876         return (entry->flags &
 877                 (MPTCP_PM_ADDR_FLAG_SIGNAL | MPTCP_PM_ADDR_FLAG_SUBFLOW)) ==
 878                 MPTCP_PM_ADDR_FLAG_SIGNAL;
 879 }
 880
 881 /* caller must ensure the RCU grace period is already elapsed */
 882 static void __mptcp_pm_release_addr_entry(struct mptcp_pm_addr_entry *entry)
 883 {
 884         if (entry->lsk)
 885                 sock_release(entry->lsk);
 886         kfree(entry);
 887 }
 888
 889 static int mptcp_pm_nl_append_new_local_addr(struct pm_nl_pernet *pernet,
 890                                              struct mptcp_pm_addr_entry *entry)
 891 {
 892         struct mptcp_pm_addr_entry *cur, *del_entry = NULL;
 893         unsigned int addr_max;
 894         int ret = -EINVAL;
 895
 896         spin_lock_bh(&pernet->lock);
 897         /* to keep the code simple, don't do IDR-like allocation for address ID,
 898          * just bail when we exceed limits
 899          */
 900         if (pernet->next_id == MPTCP_PM_MAX_ADDR_ID)
 901                 pernet->next_id = 1;
 902         if (pernet->addrs >= MPTCP_PM_ADDR_MAX)
 903                 goto out;
 904         if (test_bit(entry->addr.id, pernet->id_bitmap))
 905                 goto out;
 906
 907         /* do not insert duplicate address, differentiate on port only
 908          * singled addresses
 909          */
 910         list_for_each_entry(cur, &pernet->local_addr_list, list) {
 911                 if (mptcp_addresses_equal(&cur->addr, &entry->addr,
 912                                           address_use_port(entry) &&
 913                                           address_use_port(cur))) {
 914                         /* allow replacing the exiting endpoint only if such
 915                          * endpoint is an implicit one and the user-space
 916                          * did not provide an endpoint id
 917                          */
 918                         if (!(cur->flags & MPTCP_PM_ADDR_FLAG_IMPLICIT))
 919                                 goto out;
 920                         if (entry->addr.id)
 921                                 goto out;
 922
 923                         pernet->addrs--;
 924                         entry->addr.id = cur->addr.id;
 925                         list_del_rcu(&cur->list);
 926                         del_entry = cur;
 927                         break;
 928                 }
 929         }
 930
 931         if (!entry->addr.id) {
 932 find_next:
 933                 entry->addr.id = find_next_zero_bit(pernet->id_bitmap,
 934                                                     MPTCP_PM_MAX_ADDR_ID + 1,
 935                                                     pernet->next_id);
 936                 if (!entry->addr.id && pernet->next_id != 1) {
 937                         pernet->next_id = 1;
 938                         goto find_next;
 939                 }
 940         }
 941
 942         if (!entry->addr.id)
 943                 goto out;
 944
 945         __set_bit(entry->addr.id, pernet->id_bitmap);
 946         if (entry->addr.id > pernet->next_id)
 947                 pernet->next_id = entry->addr.id;
 948
 949         if (entry->flags & MPTCP_PM_ADDR_FLAG_SIGNAL) {
 950                 addr_max = pernet->add_addr_signal_max;
 951                 WRITE_ONCE(pernet->add_addr_signal_max, addr_max + 1);
 952         }
 953         if (entry->flags & MPTCP_PM_ADDR_FLAG_SUBFLOW) {
 954                 addr_max = pernet->local_addr_max;
 955                 WRITE_ONCE(pernet->local_addr_max, addr_max + 1);
 956         }
 957
 958         pernet->addrs++;
 959         list_add_tail_rcu(&entry->list, &pernet->local_addr_list);
 960         ret = entry->addr.id;
 961
 962 out:
 963         spin_unlock_bh(&pernet->lock);
 964
 965         /* just replaced an existing entry, free it */
 966         if (del_entry) {
 967                 synchronize_rcu();
 968                 __mptcp_pm_release_addr_entry(del_entry);
 969         }
 970         return ret;
 971 }
 972
 973 static int mptcp_pm_nl_create_listen_socket(struct sock *sk,
 974                                             struct mptcp_pm_addr_entry *entry)
 975 {
 976         int addrlen = sizeof(struct sockaddr_in);
 977         struct sockaddr_storage addr;
 978         struct mptcp_sock *msk;
 979         struct socket *ssock;
 980         int backlog = 1024;
 981         int err;
 982
 983         err = sock_create_kern(sock_net(sk), entry->addr.family,
 984                                SOCK_STREAM, IPPROTO_MPTCP, &entry->lsk);
 985         if (err)
 986                 return err;
 987
 988         msk = mptcp_sk(entry->lsk->sk);
 989         if (!msk) {
 990                 err = -EINVAL;
 991                 goto out;
 992         }
 993
 994         ssock = __mptcp_nmpc_socket(msk);
 995         if (!ssock) {
 996                 err = -EINVAL;
 997                 goto out;
 998         }
 999
1000         mptcp_info2sockaddr(&entry->addr, &addr, entry->addr.family);
1001 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
1002         if (entry->addr.family == AF_INET6)
1003                 addrlen = sizeof(struct sockaddr_in6);
1004 #endif
1005         err = kernel_bind(ssock, (struct sockaddr *)&addr, addrlen);
1006         if (err) {
1007                 pr_warn("kernel_bind error, err=%d", err);
1008                 goto out;
1009         }
1010
1011         err = kernel_listen(ssock, backlog);
1012         if (err) {
1013                 pr_warn("kernel_listen error, err=%d", err);
1014                 goto out;
1015         }
1016
1017         return 0;
1018
1019 out:
1020         sock_release(entry->lsk);
1021         return err;
1022 }
1023
1024 int mptcp_pm_nl_get_local_id(struct mptcp_sock *msk, struct sock_common *skc)
1025 {
1026         struct mptcp_pm_addr_entry *entry;
1027         struct mptcp_addr_info skc_local;
1028         struct mptcp_addr_info msk_local;
1029         struct pm_nl_pernet *pernet;
1030         int ret = -1;
1031
1032         if (WARN_ON_ONCE(!msk))
1033                 return -1;
1034
1035         /* The 0 ID mapping is defined by the first subflow, copied into the msk
1036          * addr
1037          */
1038         local_address((struct sock_common *)msk, &msk_local);
1039         local_address((struct sock_common *)skc, &skc_local);
1040         if (mptcp_addresses_equal(&msk_local, &skc_local, false))
1041                 return 0;
1042
1043         if (mptcp_pm_is_userspace(msk))
1044                 return mptcp_userspace_pm_get_local_id(msk, &skc_local);
1045
1046         pernet = pm_nl_get_pernet_from_msk(msk);
1047
1048         rcu_read_lock();
1049         list_for_each_entry_rcu(entry, &pernet->local_addr_list, list) {
1050                 if (mptcp_addresses_equal(&entry->addr, &skc_local, entry->addr.port)) {
1051                         ret = entry->addr.id;
1052                         break;
1053                 }
1054         }
1055         rcu_read_unlock();
1056         if (ret >= 0)
1057                 return ret;
1058
1059         /* address not found, add to local list */
1060         entry = kmalloc(sizeof(*entry), GFP_ATOMIC);
1061         if (!entry)
1062                 return -ENOMEM;
1063
1064         entry->addr = skc_local;
1065         entry->addr.id = 0;
1066         entry->addr.port = 0;
1067         entry->ifindex = 0;
1068         entry->flags = MPTCP_PM_ADDR_FLAG_IMPLICIT;
1069         entry->lsk = NULL;
1070         ret = mptcp_pm_nl_append_new_local_addr(pernet, entry);
1071         if (ret < 0)
1072                 kfree(entry);
1073
1074         return ret;
1075 }
1076
1077 #define MPTCP_PM_CMD_GRP_OFFSET       0
1078 #define MPTCP_PM_EV_GRP_OFFSET        1
1079
1080 static const struct genl_multicast_group mptcp_pm_mcgrps[] = {
1081         [MPTCP_PM_CMD_GRP_OFFSET]       = { .name = MPTCP_PM_CMD_GRP_NAME, },
1082         [MPTCP_PM_EV_GRP_OFFSET]        = { .name = MPTCP_PM_EV_GRP_NAME,
1083                                             .flags = GENL_UNS_ADMIN_PERM,
1084                                           },
1085 };
1086
1087 static const struct nla_policy
1088 mptcp_pm_addr_policy[MPTCP_PM_ADDR_ATTR_MAX + 1] = {
1089         [MPTCP_PM_ADDR_ATTR_FAMILY]     = { .type       = NLA_U16,      },
1090         [MPTCP_PM_ADDR_ATTR_ID]         = { .type       = NLA_U8,       },
1091         [MPTCP_PM_ADDR_ATTR_ADDR4]      = { .type       = NLA_U32,      },
1092         [MPTCP_PM_ADDR_ATTR_ADDR6]      =
1093                 NLA_POLICY_EXACT_LEN(sizeof(struct in6_addr)),
1094         [MPTCP_PM_ADDR_ATTR_PORT]       = { .type       = NLA_U16       },
1095         [MPTCP_PM_ADDR_ATTR_FLAGS]      = { .type       = NLA_U32       },
1096         [MPTCP_PM_ADDR_ATTR_IF_IDX]     = { .type       = NLA_S32       },
1097 };
1098
1099 static const struct nla_policy mptcp_pm_policy[MPTCP_PM_ATTR_MAX + 1] = {
1100         [MPTCP_PM_ATTR_ADDR]            =
1101                                         NLA_POLICY_NESTED(mptcp_pm_addr_policy),
1102         [MPTCP_PM_ATTR_RCV_ADD_ADDRS]   = { .type       = NLA_U32,      },
1103         [MPTCP_PM_ATTR_SUBFLOWS]        = { .type       = NLA_U32,      },
1104         [MPTCP_PM_ATTR_TOKEN]           = { .type       = NLA_U32,      },
1105         [MPTCP_PM_ATTR_LOC_ID]          = { .type       = NLA_U8,       },
1106         [MPTCP_PM_ATTR_ADDR_REMOTE]     =
1107                                         NLA_POLICY_NESTED(mptcp_pm_addr_policy),
1108 };
1109
1110 void mptcp_pm_nl_subflow_chk_stale(const struct mptcp_sock *msk, struct sock *ssk)
1111 {
1112         struct mptcp_subflow_context *iter, *subflow = mptcp_subflow_ctx(ssk);
1113         struct sock *sk = (struct sock *)msk;
1114         unsigned int active_max_loss_cnt;
1115         struct net *net = sock_net(sk);
1116         unsigned int stale_loss_cnt;
1117         bool slow;
1118
1119         stale_loss_cnt = mptcp_stale_loss_cnt(net);
1120         if (subflow->stale || !stale_loss_cnt || subflow->stale_count <= stale_loss_cnt)
1121                 return;
1122
1123         /* look for another available subflow not in loss state */
1124         active_max_loss_cnt = max_t(int, stale_loss_cnt - 1, 1);
1125         mptcp_for_each_subflow(msk, iter) {
1126                 if (iter != subflow && mptcp_subflow_active(iter) &&
1127                     iter->stale_count < active_max_loss_cnt) {
1128                         /* we have some alternatives, try to mark this subflow as idle ...*/
1129                         slow = lock_sock_fast(ssk);
1130                         if (!tcp_rtx_and_write_queues_empty(ssk)) {
1131                                 subflow->stale = 1;
1132                                 __mptcp_retransmit_pending_data(sk);
1133                                 MPTCP_INC_STATS(sock_net(sk), MPTCP_MIB_SUBFLOWSTALE);
1134                         }
1135                         unlock_sock_fast(ssk, slow);
1136
1137                         /* always try to push the pending data regarless of re-injections:
1138                          * we can possibly use backup subflows now, and subflow selection
1139                          * is cheap under the msk socket lock
1140                          */
1141                         __mptcp_push_pending(sk, 0);
1142                         return;
1143                 }
1144         }
1145 }
1146
1147 static int mptcp_pm_family_to_addr(int family)
1148 {
1149 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
1150         if (family == AF_INET6)
1151                 return MPTCP_PM_ADDR_ATTR_ADDR6;
1152 #endif
1153         return MPTCP_PM_ADDR_ATTR_ADDR4;
1154 }
1155
1156 static int mptcp_pm_parse_pm_addr_attr(struct nlattr *tb[],
1157                                        const struct nlattr *attr,
1158                                        struct genl_info *info,
1159                                        struct mptcp_addr_info *addr,
1160                                        bool require_family)
1161 {
1162         int err, addr_addr;
1163
1164         if (!attr) {
1165                 GENL_SET_ERR_MSG(info, "missing address info");
1166                 return -EINVAL;
1167         }
1168
1169         /* no validation needed - was already done via nested policy */
1170         err = nla_parse_nested_deprecated(tb, MPTCP_PM_ADDR_ATTR_MAX, attr,
1171                                           mptcp_pm_addr_policy, info->extack);
1172         if (err)
1173                 return err;
1174
1175         if (tb[MPTCP_PM_ADDR_ATTR_ID])
1176                 addr->id = nla_get_u8(tb[MPTCP_PM_ADDR_ATTR_ID]);
1177
1178         if (!tb[MPTCP_PM_ADDR_ATTR_FAMILY]) {
1179                 if (!require_family)
1180                         return err;
1181
1182                 NL_SET_ERR_MSG_ATTR(info->extack, attr,
1183                                     "missing family");
1184                 return -EINVAL;
1185         }
1186
1187         addr->family = nla_get_u16(tb[MPTCP_PM_ADDR_ATTR_FAMILY]);
1188         if (addr->family != AF_INET
1189 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
1190             && addr->family != AF_INET6
1191 #endif
1192             ) {
1193                 NL_SET_ERR_MSG_ATTR(info->extack, attr,
1194                                     "unknown address family");
1195                 return -EINVAL;
1196         }
1197         addr_addr = mptcp_pm_family_to_addr(addr->family);
1198         if (!tb[addr_addr]) {
1199                 NL_SET_ERR_MSG_ATTR(info->extack, attr,
1200                                     "missing address data");
1201                 return -EINVAL;
1202         }
1203
1204 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
1205         if (addr->family == AF_INET6)
1206                 addr->addr6 = nla_get_in6_addr(tb[addr_addr]);
1207         else
1208 #endif
1209                 addr->addr.s_addr = nla_get_in_addr(tb[addr_addr]);
1210
1211         if (tb[MPTCP_PM_ADDR_ATTR_PORT])
1212                 addr->port = htons(nla_get_u16(tb[MPTCP_PM_ADDR_ATTR_PORT]));
1213
1214         return err;
1215 }
1216
1217 int mptcp_pm_parse_addr(struct nlattr *attr, struct genl_info *info,
1218                         struct mptcp_addr_info *addr)
1219 {
1220         struct nlattr *tb[MPTCP_PM_ADDR_ATTR_MAX + 1];
1221
1222         memset(addr, 0, sizeof(*addr));
1223
1224         return mptcp_pm_parse_pm_addr_attr(tb, attr, info, addr, true);
1225 }
1226
1227 int mptcp_pm_parse_entry(struct nlattr *attr, struct genl_info *info,
1228                          bool require_family,
1229                          struct mptcp_pm_addr_entry *entry)
1230 {
1231         struct nlattr *tb[MPTCP_PM_ADDR_ATTR_MAX + 1];
1232         int err;
1233
1234         memset(entry, 0, sizeof(*entry));
1235
1236         err = mptcp_pm_parse_pm_addr_attr(tb, attr, info, &entry->addr, require_family);
1237         if (err)
1238                 return err;
1239
1240         if (tb[MPTCP_PM_ADDR_ATTR_IF_IDX]) {
1241                 u32 val = nla_get_s32(tb[MPTCP_PM_ADDR_ATTR_IF_IDX]);
1242
1243                 entry->ifindex = val;
1244         }
1245
1246         if (tb[MPTCP_PM_ADDR_ATTR_FLAGS])
1247                 entry->flags = nla_get_u32(tb[MPTCP_PM_ADDR_ATTR_FLAGS]);
1248
1249         if (tb[MPTCP_PM_ADDR_ATTR_PORT])
1250                 entry->addr.port = htons(nla_get_u16(tb[MPTCP_PM_ADDR_ATTR_PORT]));
1251
1252         return 0;
1253 }
1254
1255 static struct pm_nl_pernet *genl_info_pm_nl(struct genl_info *info)
1256 {
1257         return pm_nl_get_pernet(genl_info_net(info));
1258 }
1259
1260 static int mptcp_nl_add_subflow_or_signal_addr(struct net *net)
1261 {
1262         struct mptcp_sock *msk;
1263         long s_slot = 0, s_num = 0;
1264
1265         while ((msk = mptcp_token_iter_next(net, &s_slot, &s_num)) != NULL) {
1266                 struct sock *sk = (struct sock *)msk;
1267
1268                 if (!READ_ONCE(msk->fully_established) ||
1269                     mptcp_pm_is_userspace(msk))
1270                         goto next;
1271
1272                 lock_sock(sk);
1273                 spin_lock_bh(&msk->pm.lock);
1274                 mptcp_pm_create_subflow_or_signal_addr(msk);
1275                 spin_unlock_bh(&msk->pm.lock);
1276                 release_sock(sk);
1277
1278 next:
1279                 sock_put(sk);
1280                 cond_resched();
1281         }
1282
1283         return 0;
1284 }
1285
1286 static int mptcp_nl_cmd_add_addr(struct sk_buff *skb, struct genl_info *info)
1287 {
1288         struct nlattr *attr = info->attrs[MPTCP_PM_ATTR_ADDR];
1289         struct pm_nl_pernet *pernet = genl_info_pm_nl(info);
1290         struct mptcp_pm_addr_entry addr, *entry;
1291         int ret;
1292
1293         ret = mptcp_pm_parse_entry(attr, info, true, &addr);
1294         if (ret < 0)
1295                 return ret;
1296
1297         if (addr.addr.port && !(addr.flags & MPTCP_PM_ADDR_FLAG_SIGNAL)) {
1298                 GENL_SET_ERR_MSG(info, "flags must have signal when using port");
1299                 return -EINVAL;
1300         }
1301
1302         if (addr.flags & MPTCP_PM_ADDR_FLAG_SIGNAL &&
1303             addr.flags & MPTCP_PM_ADDR_FLAG_FULLMESH) {
1304                 GENL_SET_ERR_MSG(info, "flags mustn't have both signal and fullmesh");
1305                 return -EINVAL;
1306         }
1307
1308         if (addr.flags & MPTCP_PM_ADDR_FLAG_IMPLICIT) {
1309                 GENL_SET_ERR_MSG(info, "can't create IMPLICIT endpoint");
1310                 return -EINVAL;
1311         }
1312
1313         entry = kmalloc(sizeof(*entry), GFP_KERNEL);
1314         if (!entry) {
1315                 GENL_SET_ERR_MSG(info, "can't allocate addr");
1316                 return -ENOMEM;
1317         }
1318
1319         *entry = addr;
1320         if (entry->addr.port) {
1321                 ret = mptcp_pm_nl_create_listen_socket(skb->sk, entry);
1322                 if (ret) {
1323                         GENL_SET_ERR_MSG(info, "create listen socket error");
1324                         kfree(entry);
1325                         return ret;
1326                 }
1327         }
1328         ret = mptcp_pm_nl_append_new_local_addr(pernet, entry);
1329         if (ret < 0) {
1330                 GENL_SET_ERR_MSG(info, "too many addresses or duplicate one");
1331                 if (entry->lsk)
1332                         sock_release(entry->lsk);
1333                 kfree(entry);
1334                 return ret;
1335         }
1336
1337         mptcp_nl_add_subflow_or_signal_addr(sock_net(skb->sk));
1338
1339         return 0;
1340 }
1341
1342 int mptcp_pm_get_flags_and_ifindex_by_id(struct mptcp_sock *msk, unsigned int id,
1343                                          u8 *flags, int *ifindex)
1344 {
1345         struct mptcp_pm_addr_entry *entry;
1346         struct sock *sk = (struct sock *)msk;
1347         struct net *net = sock_net(sk);
1348
1349         *flags = 0;
1350         *ifindex = 0;
1351
1352         if (id) {
1353                 if (mptcp_pm_is_userspace(msk))
1354                         return mptcp_userspace_pm_get_flags_and_ifindex_by_id(msk,
1355                                                                               id,
1356                                                                               flags,
1357                                                                               ifindex);
1358
1359                 rcu_read_lock();
1360                 entry = __lookup_addr_by_id(pm_nl_get_pernet(net), id);
1361                 if (entry) {
1362                         *flags = entry->flags;
1363                         *ifindex = entry->ifindex;
1364                 }
1365                 rcu_read_unlock();
1366         }
1367
1368         return 0;
1369 }
1370
1371 static bool remove_anno_list_by_saddr(struct mptcp_sock *msk,
1372                                       const struct mptcp_addr_info *addr)
1373 {
1374         struct mptcp_pm_add_entry *entry;
1375
1376         entry = mptcp_pm_del_add_timer(msk, addr, false);
1377         if (entry) {
1378                 list_del(&entry->list);
1379                 kfree(entry);
1380                 return true;
1381         }
1382
1383         return false;
1384 }
1385
1386 static bool mptcp_pm_remove_anno_addr(struct mptcp_sock *msk,
1387                                       const struct mptcp_addr_info *addr,
1388                                       bool force)
1389 {
1390         struct mptcp_rm_list list = { .nr = 0 };
1391         bool ret;
1392
1393         list.ids[list.nr++] = addr->id;
1394
1395         ret = remove_anno_list_by_saddr(msk, addr);
1396         if (ret || force) {
1397                 spin_lock_bh(&msk->pm.lock);
1398                 mptcp_pm_remove_addr(msk, &list);
1399                 spin_unlock_bh(&msk->pm.lock);
1400         }
1401         return ret;
1402 }
1403
1404 static int mptcp_nl_remove_subflow_and_signal_addr(struct net *net,
1405                                                    const struct mptcp_pm_addr_entry *entry)
1406 {
1407         const struct mptcp_addr_info *addr = &entry->addr;
1408         struct mptcp_rm_list list = { .nr = 0 };
1409         long s_slot = 0, s_num = 0;
1410         struct mptcp_sock *msk;
1411
1412         pr_debug("remove_id=%d", addr->id);
1413
1414         list.ids[list.nr++] = addr->id;
1415
1416         while ((msk = mptcp_token_iter_next(net, &s_slot, &s_num)) != NULL) {
1417                 struct sock *sk = (struct sock *)msk;
1418                 bool remove_subflow;
1419
1420                 if (mptcp_pm_is_userspace(msk))
1421                         goto next;
1422
1423                 if (list_empty(&msk->conn_list)) {
1424                         mptcp_pm_remove_anno_addr(msk, addr, false);
1425                         goto next;
1426                 }
1427
1428                 lock_sock(sk);
1429                 remove_subflow = lookup_subflow_by_saddr(&msk->conn_list, addr);
1430                 mptcp_pm_remove_anno_addr(msk, addr, remove_subflow &&
1431                                           !(entry->flags & MPTCP_PM_ADDR_FLAG_IMPLICIT));
1432                 if (remove_subflow)
1433                         mptcp_pm_remove_subflow(msk, &list);
1434                 release_sock(sk);
1435
1436 next:
1437                 sock_put(sk);
1438                 cond_resched();
1439         }
1440
1441         return 0;
1442 }
1443
1444 static int mptcp_nl_remove_id_zero_address(struct net *net,
1445                                            struct mptcp_addr_info *addr)
1446 {
1447         struct mptcp_rm_list list = { .nr = 0 };
1448         long s_slot = 0, s_num = 0;
1449         struct mptcp_sock *msk;
1450
1451         list.ids[list.nr++] = 0;
1452
1453         while ((msk = mptcp_token_iter_next(net, &s_slot, &s_num)) != NULL) {
1454                 struct sock *sk = (struct sock *)msk;
1455                 struct mptcp_addr_info msk_local;
1456
1457                 if (list_empty(&msk->conn_list) || mptcp_pm_is_userspace(msk))
1458                         goto next;
1459
1460                 local_address((struct sock_common *)msk, &msk_local);
1461                 if (!mptcp_addresses_equal(&msk_local, addr, addr->port))
1462                         goto next;
1463
1464                 lock_sock(sk);
1465                 spin_lock_bh(&msk->pm.lock);
1466                 mptcp_pm_remove_addr(msk, &list);
1467                 mptcp_pm_nl_rm_subflow_received(msk, &list);
1468                 spin_unlock_bh(&msk->pm.lock);
1469                 release_sock(sk);
1470
1471 next:
1472                 sock_put(sk);
1473                 cond_resched();
1474         }
1475
1476         return 0;
1477 }
1478
1479 static int mptcp_nl_cmd_del_addr(struct sk_buff *skb, struct genl_info *info)
1480 {
1481         struct nlattr *attr = info->attrs[MPTCP_PM_ATTR_ADDR];
1482         struct pm_nl_pernet *pernet = genl_info_pm_nl(info);
1483         struct mptcp_pm_addr_entry addr, *entry;
1484         unsigned int addr_max;
1485         int ret;
1486
1487         ret = mptcp_pm_parse_entry(attr, info, false, &addr);
1488         if (ret < 0)
1489                 return ret;
1490
1491         /* the zero id address is special: the first address used by the msk
1492          * always gets such an id, so different subflows can have different zero
1493          * id addresses. Additionally zero id is not accounted for in id_bitmap.
1494          * Let's use an 'mptcp_rm_list' instead of the common remove code.
1495          */
1496         if (addr.addr.id == 0)
1497                 return mptcp_nl_remove_id_zero_address(sock_net(skb->sk), &addr.addr);
1498
1499         spin_lock_bh(&pernet->lock);
1500         entry = __lookup_addr_by_id(pernet, addr.addr.id);
1501         if (!entry) {
1502                 GENL_SET_ERR_MSG(info, "address not found");
1503                 spin_unlock_bh(&pernet->lock);
1504                 return -EINVAL;
1505         }
1506         if (entry->flags & MPTCP_PM_ADDR_FLAG_SIGNAL) {
1507                 addr_max = pernet->add_addr_signal_max;
1508                 WRITE_ONCE(pernet->add_addr_signal_max, addr_max - 1);
1509         }
1510         if (entry->flags & MPTCP_PM_ADDR_FLAG_SUBFLOW) {
1511                 addr_max = pernet->local_addr_max;
1512                 WRITE_ONCE(pernet->local_addr_max, addr_max - 1);
1513         }
1514
1515         pernet->addrs--;
1516         list_del_rcu(&entry->list);
1517         __clear_bit(entry->addr.id, pernet->id_bitmap);
1518         spin_unlock_bh(&pernet->lock);
1519
1520         mptcp_nl_remove_subflow_and_signal_addr(sock_net(skb->sk), entry);
1521         synchronize_rcu();
1522         __mptcp_pm_release_addr_entry(entry);
1523
1524         return ret;
1525 }
1526
1527 void mptcp_pm_remove_addrs_and_subflows(struct mptcp_sock *msk,
1528                                         struct list_head *rm_list)
1529 {
1530         struct mptcp_rm_list alist = { .nr = 0 }, slist = { .nr = 0 };
1531         struct mptcp_pm_addr_entry *entry;
1532
1533         list_for_each_entry(entry, rm_list, list) {
1534                 if (lookup_subflow_by_saddr(&msk->conn_list, &entry->addr) &&
1535                     slist.nr < MPTCP_RM_IDS_MAX)
1536                         slist.ids[slist.nr++] = entry->addr.id;
1537
1538                 if (remove_anno_list_by_saddr(msk, &entry->addr) &&
1539                     alist.nr < MPTCP_RM_IDS_MAX)
1540                         alist.ids[alist.nr++] = entry->addr.id;
1541         }
1542
1543         if (alist.nr) {
1544                 spin_lock_bh(&msk->pm.lock);
1545                 mptcp_pm_remove_addr(msk, &alist);
1546                 spin_unlock_bh(&msk->pm.lock);
1547         }
1548         if (slist.nr)
1549                 mptcp_pm_remove_subflow(msk, &slist);
1550 }
1551
1552 static void mptcp_nl_remove_addrs_list(struct net *net,
1553                                        struct list_head *rm_list)
1554 {
1555         long s_slot = 0, s_num = 0;
1556         struct mptcp_sock *msk;
1557
1558         if (list_empty(rm_list))
1559                 return;
1560
1561         while ((msk = mptcp_token_iter_next(net, &s_slot, &s_num)) != NULL) {
1562                 struct sock *sk = (struct sock *)msk;
1563
1564                 if (!mptcp_pm_is_userspace(msk)) {
1565                         lock_sock(sk);
1566                         mptcp_pm_remove_addrs_and_subflows(msk, rm_list);
1567                         release_sock(sk);
1568                 }
1569
1570                 sock_put(sk);
1571                 cond_resched();
1572         }
1573 }
1574
1575 /* caller must ensure the RCU grace period is already elapsed */
1576 static void __flush_addrs(struct list_head *list)
1577 {
1578         while (!list_empty(list)) {
1579                 struct mptcp_pm_addr_entry *cur;
1580
1581                 cur = list_entry(list->next,
1582                                  struct mptcp_pm_addr_entry, list);
1583                 list_del_rcu(&cur->list);
1584                 __mptcp_pm_release_addr_entry(cur);
1585         }
1586 }
1587
1588 static void __reset_counters(struct pm_nl_pernet *pernet)
1589 {
1590         WRITE_ONCE(pernet->add_addr_signal_max, 0);
1591         WRITE_ONCE(pernet->add_addr_accept_max, 0);
1592         WRITE_ONCE(pernet->local_addr_max, 0);
1593         pernet->addrs = 0;
1594 }
1595
1596 static int mptcp_nl_cmd_flush_addrs(struct sk_buff *skb, struct genl_info *info)
1597 {
1598         struct pm_nl_pernet *pernet = genl_info_pm_nl(info);
1599         LIST_HEAD(free_list);
1600
1601         spin_lock_bh(&pernet->lock);
1602         list_splice_init(&pernet->local_addr_list, &free_list);
1603         __reset_counters(pernet);
1604         pernet->next_id = 1;
1605         bitmap_zero(pernet->id_bitmap, MPTCP_PM_MAX_ADDR_ID + 1);
1606         spin_unlock_bh(&pernet->lock);
1607         mptcp_nl_remove_addrs_list(sock_net(skb->sk), &free_list);
1608         synchronize_rcu();
1609         __flush_addrs(&free_list);
1610         return 0;
1611 }
1612
1613 static int mptcp_nl_fill_addr(struct sk_buff *skb,
1614                               struct mptcp_pm_addr_entry *entry)
1615 {
1616         struct mptcp_addr_info *addr = &entry->addr;
1617         struct nlattr *attr;
1618
1619         attr = nla_nest_start(skb, MPTCP_PM_ATTR_ADDR);
1620         if (!attr)
1621                 return -EMSGSIZE;
1622
1623         if (nla_put_u16(skb, MPTCP_PM_ADDR_ATTR_FAMILY, addr->family))
1624                 goto nla_put_failure;
1625         if (nla_put_u16(skb, MPTCP_PM_ADDR_ATTR_PORT, ntohs(addr->port)))
1626                 goto nla_put_failure;
1627         if (nla_put_u8(skb, MPTCP_PM_ADDR_ATTR_ID, addr->id))
1628                 goto nla_put_failure;
1629         if (nla_put_u32(skb, MPTCP_PM_ADDR_ATTR_FLAGS, entry->flags))
1630                 goto nla_put_failure;
1631         if (entry->ifindex &&
1632             nla_put_s32(skb, MPTCP_PM_ADDR_ATTR_IF_IDX, entry->ifindex))
1633                 goto nla_put_failure;
1634
1635         if (addr->family == AF_INET &&
1636             nla_put_in_addr(skb, MPTCP_PM_ADDR_ATTR_ADDR4,
1637                             addr->addr.s_addr))
1638                 goto nla_put_failure;
1639 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
1640         else if (addr->family == AF_INET6 &&
1641                  nla_put_in6_addr(skb, MPTCP_PM_ADDR_ATTR_ADDR6, &addr->addr6))
1642                 goto nla_put_failure;
1643 #endif
1644         nla_nest_end(skb, attr);
1645         return 0;
1646
1647 nla_put_failure:
1648         nla_nest_cancel(skb, attr);
1649         return -EMSGSIZE;
1650 }
1651
1652 static int mptcp_nl_cmd_get_addr(struct sk_buff *skb, struct genl_info *info)
1653 {
1654         struct nlattr *attr = info->attrs[MPTCP_PM_ATTR_ADDR];
1655         struct pm_nl_pernet *pernet = genl_info_pm_nl(info);
1656         struct mptcp_pm_addr_entry addr, *entry;
1657         struct sk_buff *msg;
1658         void *reply;
1659         int ret;
1660
1661         ret = mptcp_pm_parse_entry(attr, info, false, &addr);
1662         if (ret < 0)
1663                 return ret;
1664
1665         msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
1666         if (!msg)
1667                 return -ENOMEM;
1668
1669         reply = genlmsg_put_reply(msg, info, &mptcp_genl_family, 0,
1670                                   info->genlhdr->cmd);
1671         if (!reply) {
1672                 GENL_SET_ERR_MSG(info, "not enough space in Netlink message");
1673                 ret = -EMSGSIZE;
1674                 goto fail;
1675         }
1676
1677         spin_lock_bh(&pernet->lock);
1678         entry = __lookup_addr_by_id(pernet, addr.addr.id);
1679         if (!entry) {
1680                 GENL_SET_ERR_MSG(info, "address not found");
1681                 ret = -EINVAL;
1682                 goto unlock_fail;
1683         }
1684
1685         ret = mptcp_nl_fill_addr(msg, entry);
1686         if (ret)
1687                 goto unlock_fail;
1688
1689         genlmsg_end(msg, reply);
1690         ret = genlmsg_reply(msg, info);
1691         spin_unlock_bh(&pernet->lock);
1692         return ret;
1693
1694 unlock_fail:
1695         spin_unlock_bh(&pernet->lock);
1696
1697 fail:
1698         nlmsg_free(msg);
1699         return ret;
1700 }
1701
1702 static int mptcp_nl_cmd_dump_addrs(struct sk_buff *msg,
1703                                    struct netlink_callback *cb)
1704 {
1705         struct net *net = sock_net(msg->sk);
1706         struct mptcp_pm_addr_entry *entry;
1707         struct pm_nl_pernet *pernet;
1708         int id = cb->args[0];
1709         void *hdr;
1710         int i;
1711
1712         pernet = pm_nl_get_pernet(net);
1713
1714         spin_lock_bh(&pernet->lock);
1715         for (i = id; i < MPTCP_PM_MAX_ADDR_ID + 1; i++) {
1716                 if (test_bit(i, pernet->id_bitmap)) {
1717                         entry = __lookup_addr_by_id(pernet, i);
1718                         if (!entry)
1719                                 break;
1720
1721                         if (entry->addr.id <= id)
1722                                 continue;
1723
1724                         hdr = genlmsg_put(msg, NETLINK_CB(cb->skb).portid,
1725                                           cb->nlh->nlmsg_seq, &mptcp_genl_family,
1726                                           NLM_F_MULTI, MPTCP_PM_CMD_GET_ADDR);
1727                         if (!hdr)
1728                                 break;
1729
1730                         if (mptcp_nl_fill_addr(msg, entry) < 0) {
1731                                 genlmsg_cancel(msg, hdr);
1732                                 break;
1733                         }
1734
1735                         id = entry->addr.id;
1736                         genlmsg_end(msg, hdr);
1737                 }
1738         }
1739         spin_unlock_bh(&pernet->lock);
1740
1741         cb->args[0] = id;
1742         return msg->len;
1743 }
1744
1745 static int parse_limit(struct genl_info *info, int id, unsigned int *limit)
1746 {
1747         struct nlattr *attr = info->attrs[id];
1748
1749         if (!attr)
1750                 return 0;
1751
1752         *limit = nla_get_u32(attr);
1753         if (*limit > MPTCP_PM_ADDR_MAX) {
1754                 GENL_SET_ERR_MSG(info, "limit greater than maximum");
1755                 return -EINVAL;
1756         }
1757         return 0;
1758 }
1759
1760 static int
1761 mptcp_nl_cmd_set_limits(struct sk_buff *skb, struct genl_info *info)
1762 {
1763         struct pm_nl_pernet *pernet = genl_info_pm_nl(info);
1764         unsigned int rcv_addrs, subflows;
1765         int ret;
1766
1767         spin_lock_bh(&pernet->lock);
1768         rcv_addrs = pernet->add_addr_accept_max;
1769         ret = parse_limit(info, MPTCP_PM_ATTR_RCV_ADD_ADDRS, &rcv_addrs);
1770         if (ret)
1771                 goto unlock;
1772
1773         subflows = pernet->subflows_max;
1774         ret = parse_limit(info, MPTCP_PM_ATTR_SUBFLOWS, &subflows);
1775         if (ret)
1776                 goto unlock;
1777
1778         WRITE_ONCE(pernet->add_addr_accept_max, rcv_addrs);
1779         WRITE_ONCE(pernet->subflows_max, subflows);
1780
1781 unlock:
1782         spin_unlock_bh(&pernet->lock);
1783         return ret;
1784 }
1785
1786 static int
1787 mptcp_nl_cmd_get_limits(struct sk_buff *skb, struct genl_info *info)
1788 {
1789         struct pm_nl_pernet *pernet = genl_info_pm_nl(info);
1790         struct sk_buff *msg;
1791         void *reply;
1792
1793         msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
1794         if (!msg)
1795                 return -ENOMEM;
1796
1797         reply = genlmsg_put_reply(msg, info, &mptcp_genl_family, 0,
1798                                   MPTCP_PM_CMD_GET_LIMITS);
1799         if (!reply)
1800                 goto fail;
1801
1802         if (nla_put_u32(msg, MPTCP_PM_ATTR_RCV_ADD_ADDRS,
1803                         READ_ONCE(pernet->add_addr_accept_max)))
1804                 goto fail;
1805
1806         if (nla_put_u32(msg, MPTCP_PM_ATTR_SUBFLOWS,
1807                         READ_ONCE(pernet->subflows_max)))
1808                 goto fail;
1809
1810         genlmsg_end(msg, reply);
1811         return genlmsg_reply(msg, info);
1812
1813 fail:
1814         GENL_SET_ERR_MSG(info, "not enough space in Netlink message");
1815         nlmsg_free(msg);
1816         return -EMSGSIZE;
1817 }
1818
1819 static void mptcp_pm_nl_fullmesh(struct mptcp_sock *msk,
1820                                  struct mptcp_addr_info *addr)
1821 {
1822         struct mptcp_rm_list list = { .nr = 0 };
1823
1824         list.ids[list.nr++] = addr->id;
1825
1826         spin_lock_bh(&msk->pm.lock);
1827         mptcp_pm_nl_rm_subflow_received(msk, &list);
1828         mptcp_pm_create_subflow_or_signal_addr(msk);
1829         spin_unlock_bh(&msk->pm.lock);
1830 }
1831
1832 static int mptcp_nl_set_flags(struct net *net,
1833                               struct mptcp_addr_info *addr,
1834                               u8 bkup, u8 changed)
1835 {
1836         long s_slot = 0, s_num = 0;
1837         struct mptcp_sock *msk;
1838         int ret = -EINVAL;
1839
1840         while ((msk = mptcp_token_iter_next(net, &s_slot, &s_num)) != NULL) {
1841                 struct sock *sk = (struct sock *)msk;
1842
1843                 if (list_empty(&msk->conn_list) || mptcp_pm_is_userspace(msk))
1844                         goto next;
1845
1846                 lock_sock(sk);
1847                 if (changed & MPTCP_PM_ADDR_FLAG_BACKUP)
1848                         ret = mptcp_pm_nl_mp_prio_send_ack(msk, addr, NULL, bkup);
1849                 if (changed & MPTCP_PM_ADDR_FLAG_FULLMESH)
1850                         mptcp_pm_nl_fullmesh(msk, addr);
1851                 release_sock(sk);
1852
1853 next:
1854                 sock_put(sk);
1855                 cond_resched();
1856         }
1857
1858         return ret;
1859 }
1860
1861 static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info)
1862 {
1863         struct mptcp_pm_addr_entry addr = { .addr = { .family = AF_UNSPEC }, }, *entry;
1864         struct mptcp_pm_addr_entry remote = { .addr = { .family = AF_UNSPEC }, };
1865         struct nlattr *attr_rem = info->attrs[MPTCP_PM_ATTR_ADDR_REMOTE];
1866         struct nlattr *token = info->attrs[MPTCP_PM_ATTR_TOKEN];
1867         struct nlattr *attr = info->attrs[MPTCP_PM_ATTR_ADDR];
1868         struct pm_nl_pernet *pernet = genl_info_pm_nl(info);
1869         u8 changed, mask = MPTCP_PM_ADDR_FLAG_BACKUP |
1870                            MPTCP_PM_ADDR_FLAG_FULLMESH;
1871         struct net *net = sock_net(skb->sk);
1872         u8 bkup = 0, lookup_by_id = 0;
1873         int ret;
1874
1875         ret = mptcp_pm_parse_entry(attr, info, false, &addr);
1876         if (ret < 0)
1877                 return ret;
1878
1879         if (attr_rem) {
1880                 ret = mptcp_pm_parse_entry(attr_rem, info, false, &remote);
1881                 if (ret < 0)
1882                         return ret;
1883         }
1884
1885         if (addr.flags & MPTCP_PM_ADDR_FLAG_BACKUP)
1886                 bkup = 1;
1887         if (addr.addr.family == AF_UNSPEC) {
1888                 lookup_by_id = 1;
1889                 if (!addr.addr.id)
1890                         return -EOPNOTSUPP;
1891         }
1892
1893         if (token)
1894                 return mptcp_userspace_pm_set_flags(sock_net(skb->sk),
1895                                                     token, &addr, &remote, bkup);
1896
1897         spin_lock_bh(&pernet->lock);
1898         entry = __lookup_addr(pernet, &addr.addr, lookup_by_id);
1899         if (!entry) {
1900                 spin_unlock_bh(&pernet->lock);
1901                 return -EINVAL;
1902         }
1903         if ((addr.flags & MPTCP_PM_ADDR_FLAG_FULLMESH) &&
1904             (entry->flags & MPTCP_PM_ADDR_FLAG_SIGNAL)) {
1905                 spin_unlock_bh(&pernet->lock);
1906                 return -EINVAL;
1907         }
1908
1909         changed = (addr.flags ^ entry->flags) & mask;
1910         entry->flags = (entry->flags & ~mask) | (addr.flags & mask);
1911         addr = *entry;
1912         spin_unlock_bh(&pernet->lock);
1913
1914         mptcp_nl_set_flags(net, &addr.addr, bkup, changed);
1915         return 0;
1916 }
1917
1918 static void mptcp_nl_mcast_send(struct net *net, struct sk_buff *nlskb, gfp_t gfp)
1919 {
1920         genlmsg_multicast_netns(&mptcp_genl_family, net,
1921                                 nlskb, 0, MPTCP_PM_EV_GRP_OFFSET, gfp);
1922 }
1923
1924 bool mptcp_userspace_pm_active(const struct mptcp_sock *msk)
1925 {
1926         return genl_has_listeners(&mptcp_genl_family,
1927                                   sock_net((const struct sock *)msk),
1928                                   MPTCP_PM_EV_GRP_OFFSET);
1929 }
1930
1931 static int mptcp_event_add_subflow(struct sk_buff *skb, const struct sock *ssk)
1932 {
1933         const struct inet_sock *issk = inet_sk(ssk);
1934         const struct mptcp_subflow_context *sf;
1935
1936         if (nla_put_u16(skb, MPTCP_ATTR_FAMILY, ssk->sk_family))
1937                 return -EMSGSIZE;
1938
1939         switch (ssk->sk_family) {
1940         case AF_INET:
1941                 if (nla_put_in_addr(skb, MPTCP_ATTR_SADDR4, issk->inet_saddr))
1942                         return -EMSGSIZE;
1943                 if (nla_put_in_addr(skb, MPTCP_ATTR_DADDR4, issk->inet_daddr))
1944                         return -EMSGSIZE;
1945                 break;
1946 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
1947         case AF_INET6: {
1948                 const struct ipv6_pinfo *np = inet6_sk(ssk);
1949
1950                 if (nla_put_in6_addr(skb, MPTCP_ATTR_SADDR6, &np->saddr))
1951                         return -EMSGSIZE;
1952                 if (nla_put_in6_addr(skb, MPTCP_ATTR_DADDR6, &ssk->sk_v6_daddr))
1953                         return -EMSGSIZE;
1954                 break;
1955         }
1956 #endif
1957         default:
1958                 WARN_ON_ONCE(1);
1959                 return -EMSGSIZE;
1960         }
1961
1962         if (nla_put_be16(skb, MPTCP_ATTR_SPORT, issk->inet_sport))
1963                 return -EMSGSIZE;
1964         if (nla_put_be16(skb, MPTCP_ATTR_DPORT, issk->inet_dport))
1965                 return -EMSGSIZE;
1966
1967         sf = mptcp_subflow_ctx(ssk);
1968         if (WARN_ON_ONCE(!sf))
1969                 return -EINVAL;
1970
1971         if (nla_put_u8(skb, MPTCP_ATTR_LOC_ID, sf->local_id))
1972                 return -EMSGSIZE;
1973
1974         if (nla_put_u8(skb, MPTCP_ATTR_REM_ID, sf->remote_id))
1975                 return -EMSGSIZE;
1976
1977         return 0;
1978 }
1979
1980 static int mptcp_event_put_token_and_ssk(struct sk_buff *skb,
1981                                          const struct mptcp_sock *msk,
1982                                          const struct sock *ssk)
1983 {
1984         const struct sock *sk = (const struct sock *)msk;
1985         const struct mptcp_subflow_context *sf;
1986         u8 sk_err;
1987
1988         if (nla_put_u32(skb, MPTCP_ATTR_TOKEN, msk->token))
1989                 return -EMSGSIZE;
1990
1991         if (mptcp_event_add_subflow(skb, ssk))
1992                 return -EMSGSIZE;
1993
1994         sf = mptcp_subflow_ctx(ssk);
1995         if (WARN_ON_ONCE(!sf))
1996                 return -EINVAL;
1997
1998         if (nla_put_u8(skb, MPTCP_ATTR_BACKUP, sf->backup))
1999                 return -EMSGSIZE;
2000
2001         if (ssk->sk_bound_dev_if &&
2002             nla_put_s32(skb, MPTCP_ATTR_IF_IDX, ssk->sk_bound_dev_if))
2003                 return -EMSGSIZE;
2004
2005         sk_err = ssk->sk_err;
2006         if (sk_err && sk->sk_state == TCP_ESTABLISHED &&
2007             nla_put_u8(skb, MPTCP_ATTR_ERROR, sk_err))
2008                 return -EMSGSIZE;
2009
2010         return 0;
2011 }
2012
2013 static int mptcp_event_sub_established(struct sk_buff *skb,
2014                                        const struct mptcp_sock *msk,
2015                                        const struct sock *ssk)
2016 {
2017         return mptcp_event_put_token_and_ssk(skb, msk, ssk);
2018 }
2019
2020 static int mptcp_event_sub_closed(struct sk_buff *skb,
2021                                   const struct mptcp_sock *msk,
2022                                   const struct sock *ssk)
2023 {
2024         const struct mptcp_subflow_context *sf;
2025
2026         if (mptcp_event_put_token_and_ssk(skb, msk, ssk))
2027                 return -EMSGSIZE;
2028
2029         sf = mptcp_subflow_ctx(ssk);
2030         if (!sf->reset_seen)
2031                 return 0;
2032
2033         if (nla_put_u32(skb, MPTCP_ATTR_RESET_REASON, sf->reset_reason))
2034                 return -EMSGSIZE;
2035
2036         if (nla_put_u32(skb, MPTCP_ATTR_RESET_FLAGS, sf->reset_transient))
2037                 return -EMSGSIZE;
2038
2039         return 0;
2040 }
2041
2042 static int mptcp_event_created(struct sk_buff *skb,
2043                                const struct mptcp_sock *msk,
2044                                const struct sock *ssk)
2045 {
2046         int err = nla_put_u32(skb, MPTCP_ATTR_TOKEN, msk->token);
2047
2048         if (err)
2049                 return err;
2050
2051         if (nla_put_u8(skb, MPTCP_ATTR_SERVER_SIDE, READ_ONCE(msk->pm.server_side)))
2052                 return -EMSGSIZE;
2053
2054         return mptcp_event_add_subflow(skb, ssk);
2055 }
2056
2057 void mptcp_event_addr_removed(const struct mptcp_sock *msk, uint8_t id)
2058 {
2059         struct net *net = sock_net((const struct sock *)msk);
2060         struct nlmsghdr *nlh;
2061         struct sk_buff *skb;
2062
2063         if (!genl_has_listeners(&mptcp_genl_family, net, MPTCP_PM_EV_GRP_OFFSET))
2064                 return;
2065
2066         skb = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_ATOMIC);
2067         if (!skb)
2068                 return;
2069
2070         nlh = genlmsg_put(skb, 0, 0, &mptcp_genl_family, 0, MPTCP_EVENT_REMOVED);
2071         if (!nlh)
2072                 goto nla_put_failure;
2073
2074         if (nla_put_u32(skb, MPTCP_ATTR_TOKEN, msk->token))
2075                 goto nla_put_failure;
2076
2077         if (nla_put_u8(skb, MPTCP_ATTR_REM_ID, id))
2078                 goto nla_put_failure;
2079
2080         genlmsg_end(skb, nlh);
2081         mptcp_nl_mcast_send(net, skb, GFP_ATOMIC);
2082         return;
2083
2084 nla_put_failure:
2085         kfree_skb(skb);
2086 }
2087
2088 void mptcp_event_addr_announced(const struct sock *ssk,
2089                                 const struct mptcp_addr_info *info)
2090 {
2091         struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(ssk);
2092         struct mptcp_sock *msk = mptcp_sk(subflow->conn);
2093         struct net *net = sock_net(ssk);
2094         struct nlmsghdr *nlh;
2095         struct sk_buff *skb;
2096
2097         if (!genl_has_listeners(&mptcp_genl_family, net, MPTCP_PM_EV_GRP_OFFSET))
2098                 return;
2099
2100         skb = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_ATOMIC);
2101         if (!skb)
2102                 return;
2103
2104         nlh = genlmsg_put(skb, 0, 0, &mptcp_genl_family, 0,
2105                           MPTCP_EVENT_ANNOUNCED);
2106         if (!nlh)
2107                 goto nla_put_failure;
2108
2109         if (nla_put_u32(skb, MPTCP_ATTR_TOKEN, msk->token))
2110                 goto nla_put_failure;
2111
2112         if (nla_put_u8(skb, MPTCP_ATTR_REM_ID, info->id))
2113                 goto nla_put_failure;
2114
2115         if (nla_put_be16(skb, MPTCP_ATTR_DPORT,
2116                          info->port == 0 ?
2117                          inet_sk(ssk)->inet_dport :
2118                          info->port))
2119                 goto nla_put_failure;
2120
2121         switch (info->family) {
2122         case AF_INET:
2123                 if (nla_put_in_addr(skb, MPTCP_ATTR_DADDR4, info->addr.s_addr))
2124                         goto nla_put_failure;
2125                 break;
2126 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
2127         case AF_INET6:
2128                 if (nla_put_in6_addr(skb, MPTCP_ATTR_DADDR6, &info->addr6))
2129                         goto nla_put_failure;
2130                 break;
2131 #endif
2132         default:
2133                 WARN_ON_ONCE(1);
2134                 goto nla_put_failure;
2135         }
2136
2137         genlmsg_end(skb, nlh);
2138         mptcp_nl_mcast_send(net, skb, GFP_ATOMIC);
2139         return;
2140
2141 nla_put_failure:
2142         kfree_skb(skb);
2143 }
2144
2145 void mptcp_event(enum mptcp_event_type type, const struct mptcp_sock *msk,
2146                  const struct sock *ssk, gfp_t gfp)
2147 {
2148         struct net *net = sock_net((const struct sock *)msk);
2149         struct nlmsghdr *nlh;
2150         struct sk_buff *skb;
2151
2152         if (!genl_has_listeners(&mptcp_genl_family, net, MPTCP_PM_EV_GRP_OFFSET))
2153                 return;
2154
2155         skb = nlmsg_new(NLMSG_DEFAULT_SIZE, gfp);
2156         if (!skb)
2157                 return;
2158
2159         nlh = genlmsg_put(skb, 0, 0, &mptcp_genl_family, 0, type);
2160         if (!nlh)
2161                 goto nla_put_failure;
2162
2163         switch (type) {
2164         case MPTCP_EVENT_UNSPEC:
2165                 WARN_ON_ONCE(1);
2166                 break;
2167         case MPTCP_EVENT_CREATED:
2168         case MPTCP_EVENT_ESTABLISHED:
2169                 if (mptcp_event_created(skb, msk, ssk) < 0)
2170                         goto nla_put_failure;
2171                 break;
2172         case MPTCP_EVENT_CLOSED:
2173                 if (nla_put_u32(skb, MPTCP_ATTR_TOKEN, msk->token) < 0)
2174                         goto nla_put_failure;
2175                 break;
2176         case MPTCP_EVENT_ANNOUNCED:
2177         case MPTCP_EVENT_REMOVED:
2178                 /* call mptcp_event_addr_announced()/removed instead */
2179                 WARN_ON_ONCE(1);
2180                 break;
2181         case MPTCP_EVENT_SUB_ESTABLISHED:
2182         case MPTCP_EVENT_SUB_PRIORITY:
2183                 if (mptcp_event_sub_established(skb, msk, ssk) < 0)
2184                         goto nla_put_failure;
2185                 break;
2186         case MPTCP_EVENT_SUB_CLOSED:
2187                 if (mptcp_event_sub_closed(skb, msk, ssk) < 0)
2188                         goto nla_put_failure;
2189                 break;
2190         }
2191
2192         genlmsg_end(skb, nlh);
2193         mptcp_nl_mcast_send(net, skb, gfp);
2194         return;
2195
2196 nla_put_failure:
2197         kfree_skb(skb);
2198 }
2199
2200 static const struct genl_small_ops mptcp_pm_ops[] = {
2201         {
2202                 .cmd    = MPTCP_PM_CMD_ADD_ADDR,
2203                 .doit   = mptcp_nl_cmd_add_addr,
2204                 .flags  = GENL_ADMIN_PERM,
2205         },
2206         {
2207                 .cmd    = MPTCP_PM_CMD_DEL_ADDR,
2208                 .doit   = mptcp_nl_cmd_del_addr,
2209                 .flags  = GENL_ADMIN_PERM,
2210         },
2211         {
2212                 .cmd    = MPTCP_PM_CMD_FLUSH_ADDRS,
2213                 .doit   = mptcp_nl_cmd_flush_addrs,
2214                 .flags  = GENL_ADMIN_PERM,
2215         },
2216         {
2217                 .cmd    = MPTCP_PM_CMD_GET_ADDR,
2218                 .doit   = mptcp_nl_cmd_get_addr,
2219                 .dumpit   = mptcp_nl_cmd_dump_addrs,
2220         },
2221         {
2222                 .cmd    = MPTCP_PM_CMD_SET_LIMITS,
2223                 .doit   = mptcp_nl_cmd_set_limits,
2224                 .flags  = GENL_ADMIN_PERM,
2225         },
2226         {
2227                 .cmd    = MPTCP_PM_CMD_GET_LIMITS,
2228                 .doit   = mptcp_nl_cmd_get_limits,
2229         },
2230         {
2231                 .cmd    = MPTCP_PM_CMD_SET_FLAGS,
2232                 .doit   = mptcp_nl_cmd_set_flags,
2233                 .flags  = GENL_ADMIN_PERM,
2234         },
2235         {
2236                 .cmd    = MPTCP_PM_CMD_ANNOUNCE,
2237                 .doit   = mptcp_nl_cmd_announce,
2238                 .flags  = GENL_ADMIN_PERM,
2239         },
2240         {
2241                 .cmd    = MPTCP_PM_CMD_REMOVE,
2242                 .doit   = mptcp_nl_cmd_remove,
2243                 .flags  = GENL_ADMIN_PERM,
2244         },
2245         {
2246                 .cmd    = MPTCP_PM_CMD_SUBFLOW_CREATE,
2247                 .doit   = mptcp_nl_cmd_sf_create,
2248                 .flags  = GENL_ADMIN_PERM,
2249         },
2250         {
2251                 .cmd    = MPTCP_PM_CMD_SUBFLOW_DESTROY,
2252                 .doit   = mptcp_nl_cmd_sf_destroy,
2253                 .flags  = GENL_ADMIN_PERM,
2254         },
2255 };
2256
2257 static struct genl_family mptcp_genl_family __ro_after_init = {
2258         .name           = MPTCP_PM_NAME,
2259         .version        = MPTCP_PM_VER,
2260         .maxattr        = MPTCP_PM_ATTR_MAX,
2261         .policy         = mptcp_pm_policy,
2262         .netnsok        = true,
2263         .module         = THIS_MODULE,
2264         .small_ops      = mptcp_pm_ops,
2265         .n_small_ops    = ARRAY_SIZE(mptcp_pm_ops),
2266         .mcgrps         = mptcp_pm_mcgrps,
2267         .n_mcgrps       = ARRAY_SIZE(mptcp_pm_mcgrps),
2268 };
2269
2270 static int __net_init pm_nl_init_net(struct net *net)
2271 {
2272         struct pm_nl_pernet *pernet = pm_nl_get_pernet(net);
2273
2274         INIT_LIST_HEAD_RCU(&pernet->local_addr_list);
2275
2276         /* Cit. 2 subflows ought to be enough for anybody. */
2277         pernet->subflows_max = 2;
2278         pernet->next_id = 1;
2279         pernet->stale_loss_cnt = 4;
2280         spin_lock_init(&pernet->lock);
2281
2282         /* No need to initialize other pernet fields, the struct is zeroed at
2283          * allocation time.
2284          */
2285
2286         return 0;
2287 }
2288
2289 static void __net_exit pm_nl_exit_net(struct list_head *net_list)
2290 {
2291         struct net *net;
2292
2293         list_for_each_entry(net, net_list, exit_list) {
2294                 struct pm_nl_pernet *pernet = pm_nl_get_pernet(net);
2295
2296                 /* net is removed from namespace list, can't race with
2297                  * other modifiers, also netns core already waited for a
2298                  * RCU grace period.
2299                  */
2300                 __flush_addrs(&pernet->local_addr_list);
2301         }
2302 }
2303
2304 static struct pernet_operations mptcp_pm_pernet_ops = {
2305         .init = pm_nl_init_net,
2306         .exit_batch = pm_nl_exit_net,
2307         .id = &pm_nl_pernet_id,
2308         .size = sizeof(struct pm_nl_pernet),
2309 };
2310
2311 void __init mptcp_pm_nl_init(void)
2312 {
2313         if (register_pernet_subsys(&mptcp_pm_pernet_ops) < 0)
2314                 panic("Failed to register MPTCP PM pernet subsystem.\n");
2315
2316         if (genl_register_family(&mptcp_genl_family))
2317                 panic("Failed to register MPTCP PM netlink family\n");
2318 }