Paolo Abeni | 01cacb0 | 2020-03-27 14:48:51 -0700 | [diff] [blame] | 1 | // SPDX-License-Identifier: GPL-2.0 |
| 2 | /* Multipath TCP |
| 3 | * |
| 4 | * Copyright (c) 2020, Red Hat, Inc. |
| 5 | */ |
| 6 | |
Geliang Tang | c85adce | 2020-04-03 17:14:08 +0800 | [diff] [blame] | 7 | #define pr_fmt(fmt) "MPTCP: " fmt |
| 8 | |
Paolo Abeni | 01cacb0 | 2020-03-27 14:48:51 -0700 | [diff] [blame] | 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 | |
| 19 | /* forward declaration */ |
| 20 | static struct genl_family mptcp_genl_family; |
| 21 | |
| 22 | static int pm_nl_pernet_id; |
| 23 | |
| 24 | struct mptcp_pm_addr_entry { |
| 25 | struct list_head list; |
| 26 | unsigned int flags; |
| 27 | int ifindex; |
| 28 | struct mptcp_addr_info addr; |
| 29 | struct rcu_head rcu; |
| 30 | }; |
| 31 | |
| 32 | struct pm_nl_pernet { |
| 33 | /* protects pernet updates */ |
| 34 | spinlock_t lock; |
| 35 | struct list_head local_addr_list; |
| 36 | unsigned int addrs; |
| 37 | unsigned int add_addr_signal_max; |
| 38 | unsigned int add_addr_accept_max; |
| 39 | unsigned int local_addr_max; |
| 40 | unsigned int subflows_max; |
| 41 | unsigned int next_id; |
| 42 | }; |
| 43 | |
| 44 | #define MPTCP_PM_ADDR_MAX 8 |
| 45 | |
| 46 | static bool addresses_equal(const struct mptcp_addr_info *a, |
| 47 | struct mptcp_addr_info *b, bool use_port) |
| 48 | { |
| 49 | bool addr_equals = false; |
| 50 | |
| 51 | if (a->family != b->family) |
| 52 | return false; |
| 53 | |
| 54 | if (a->family == AF_INET) |
| 55 | addr_equals = a->addr.s_addr == b->addr.s_addr; |
| 56 | #if IS_ENABLED(CONFIG_MPTCP_IPV6) |
| 57 | else |
| 58 | addr_equals = !ipv6_addr_cmp(&a->addr6, &b->addr6); |
| 59 | #endif |
| 60 | |
| 61 | if (!addr_equals) |
| 62 | return false; |
| 63 | if (!use_port) |
| 64 | return true; |
| 65 | |
| 66 | return a->port == b->port; |
| 67 | } |
| 68 | |
| 69 | static void local_address(const struct sock_common *skc, |
| 70 | struct mptcp_addr_info *addr) |
| 71 | { |
| 72 | addr->port = 0; |
| 73 | addr->family = skc->skc_family; |
| 74 | if (addr->family == AF_INET) |
| 75 | addr->addr.s_addr = skc->skc_rcv_saddr; |
| 76 | #if IS_ENABLED(CONFIG_MPTCP_IPV6) |
| 77 | else if (addr->family == AF_INET6) |
| 78 | addr->addr6 = skc->skc_v6_rcv_saddr; |
| 79 | #endif |
| 80 | } |
| 81 | |
| 82 | static void remote_address(const struct sock_common *skc, |
| 83 | struct mptcp_addr_info *addr) |
| 84 | { |
| 85 | addr->family = skc->skc_family; |
| 86 | addr->port = skc->skc_dport; |
| 87 | if (addr->family == AF_INET) |
| 88 | addr->addr.s_addr = skc->skc_daddr; |
| 89 | #if IS_ENABLED(CONFIG_MPTCP_IPV6) |
| 90 | else if (addr->family == AF_INET6) |
| 91 | addr->addr6 = skc->skc_v6_daddr; |
| 92 | #endif |
| 93 | } |
| 94 | |
| 95 | static bool lookup_subflow_by_saddr(const struct list_head *list, |
| 96 | struct mptcp_addr_info *saddr) |
| 97 | { |
| 98 | struct mptcp_subflow_context *subflow; |
| 99 | struct mptcp_addr_info cur; |
| 100 | struct sock_common *skc; |
| 101 | |
| 102 | list_for_each_entry(subflow, list, node) { |
| 103 | skc = (struct sock_common *)mptcp_subflow_tcp_sock(subflow); |
| 104 | |
| 105 | local_address(skc, &cur); |
| 106 | if (addresses_equal(&cur, saddr, false)) |
| 107 | return true; |
| 108 | } |
| 109 | |
| 110 | return false; |
| 111 | } |
| 112 | |
| 113 | static struct mptcp_pm_addr_entry * |
| 114 | select_local_address(const struct pm_nl_pernet *pernet, |
| 115 | struct mptcp_sock *msk) |
| 116 | { |
| 117 | struct mptcp_pm_addr_entry *entry, *ret = NULL; |
| 118 | |
| 119 | rcu_read_lock(); |
| 120 | spin_lock_bh(&msk->join_list_lock); |
| 121 | list_for_each_entry_rcu(entry, &pernet->local_addr_list, list) { |
| 122 | if (!(entry->flags & MPTCP_PM_ADDR_FLAG_SUBFLOW)) |
| 123 | continue; |
| 124 | |
| 125 | /* avoid any address already in use by subflows and |
| 126 | * pending join |
| 127 | */ |
| 128 | if (entry->addr.family == ((struct sock *)msk)->sk_family && |
| 129 | !lookup_subflow_by_saddr(&msk->conn_list, &entry->addr) && |
| 130 | !lookup_subflow_by_saddr(&msk->join_list, &entry->addr)) { |
| 131 | ret = entry; |
| 132 | break; |
| 133 | } |
| 134 | } |
| 135 | spin_unlock_bh(&msk->join_list_lock); |
| 136 | rcu_read_unlock(); |
| 137 | return ret; |
| 138 | } |
| 139 | |
| 140 | static struct mptcp_pm_addr_entry * |
| 141 | select_signal_address(struct pm_nl_pernet *pernet, unsigned int pos) |
| 142 | { |
| 143 | struct mptcp_pm_addr_entry *entry, *ret = NULL; |
| 144 | int i = 0; |
| 145 | |
| 146 | rcu_read_lock(); |
| 147 | /* do not keep any additional per socket state, just signal |
| 148 | * the address list in order. |
| 149 | * Note: removal from the local address list during the msk life-cycle |
| 150 | * can lead to additional addresses not being announced. |
| 151 | */ |
| 152 | list_for_each_entry_rcu(entry, &pernet->local_addr_list, list) { |
| 153 | if (!(entry->flags & MPTCP_PM_ADDR_FLAG_SIGNAL)) |
| 154 | continue; |
| 155 | if (i++ == pos) { |
| 156 | ret = entry; |
| 157 | break; |
| 158 | } |
| 159 | } |
| 160 | rcu_read_unlock(); |
| 161 | return ret; |
| 162 | } |
| 163 | |
| 164 | static void check_work_pending(struct mptcp_sock *msk) |
| 165 | { |
| 166 | if (msk->pm.add_addr_signaled == msk->pm.add_addr_signal_max && |
| 167 | (msk->pm.local_addr_used == msk->pm.local_addr_max || |
| 168 | msk->pm.subflows == msk->pm.subflows_max)) |
| 169 | WRITE_ONCE(msk->pm.work_pending, false); |
| 170 | } |
| 171 | |
| 172 | static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk) |
| 173 | { |
| 174 | struct sock *sk = (struct sock *)msk; |
| 175 | struct mptcp_pm_addr_entry *local; |
| 176 | struct mptcp_addr_info remote; |
| 177 | struct pm_nl_pernet *pernet; |
| 178 | |
| 179 | pernet = net_generic(sock_net((struct sock *)msk), pm_nl_pernet_id); |
| 180 | |
| 181 | pr_debug("local %d:%d signal %d:%d subflows %d:%d\n", |
| 182 | msk->pm.local_addr_used, msk->pm.local_addr_max, |
| 183 | msk->pm.add_addr_signaled, msk->pm.add_addr_signal_max, |
| 184 | msk->pm.subflows, msk->pm.subflows_max); |
| 185 | |
| 186 | /* check first for announce */ |
| 187 | if (msk->pm.add_addr_signaled < msk->pm.add_addr_signal_max) { |
| 188 | local = select_signal_address(pernet, |
| 189 | msk->pm.add_addr_signaled); |
| 190 | |
| 191 | if (local) { |
| 192 | msk->pm.add_addr_signaled++; |
| 193 | mptcp_pm_announce_addr(msk, &local->addr); |
| 194 | } else { |
| 195 | /* pick failed, avoid fourther attempts later */ |
| 196 | msk->pm.local_addr_used = msk->pm.add_addr_signal_max; |
| 197 | } |
| 198 | |
| 199 | check_work_pending(msk); |
| 200 | } |
| 201 | |
| 202 | /* check if should create a new subflow */ |
| 203 | if (msk->pm.local_addr_used < msk->pm.local_addr_max && |
| 204 | msk->pm.subflows < msk->pm.subflows_max) { |
| 205 | remote_address((struct sock_common *)sk, &remote); |
| 206 | |
| 207 | local = select_local_address(pernet, msk); |
| 208 | if (local) { |
| 209 | msk->pm.local_addr_used++; |
| 210 | msk->pm.subflows++; |
| 211 | check_work_pending(msk); |
| 212 | spin_unlock_bh(&msk->pm.lock); |
| 213 | __mptcp_subflow_connect(sk, local->ifindex, |
| 214 | &local->addr, &remote); |
| 215 | spin_lock_bh(&msk->pm.lock); |
| 216 | return; |
| 217 | } |
| 218 | |
| 219 | /* lookup failed, avoid fourther attempts later */ |
| 220 | msk->pm.local_addr_used = msk->pm.local_addr_max; |
| 221 | check_work_pending(msk); |
| 222 | } |
| 223 | } |
| 224 | |
| 225 | void mptcp_pm_nl_fully_established(struct mptcp_sock *msk) |
| 226 | { |
| 227 | mptcp_pm_create_subflow_or_signal_addr(msk); |
| 228 | } |
| 229 | |
| 230 | void mptcp_pm_nl_subflow_established(struct mptcp_sock *msk) |
| 231 | { |
| 232 | mptcp_pm_create_subflow_or_signal_addr(msk); |
| 233 | } |
| 234 | |
| 235 | void mptcp_pm_nl_add_addr_received(struct mptcp_sock *msk) |
| 236 | { |
| 237 | struct sock *sk = (struct sock *)msk; |
| 238 | struct mptcp_addr_info remote; |
| 239 | struct mptcp_addr_info local; |
| 240 | |
| 241 | pr_debug("accepted %d:%d remote family %d", |
| 242 | msk->pm.add_addr_accepted, msk->pm.add_addr_accept_max, |
| 243 | msk->pm.remote.family); |
| 244 | msk->pm.add_addr_accepted++; |
| 245 | msk->pm.subflows++; |
| 246 | if (msk->pm.add_addr_accepted >= msk->pm.add_addr_accept_max || |
| 247 | msk->pm.subflows >= msk->pm.subflows_max) |
| 248 | WRITE_ONCE(msk->pm.accept_addr, false); |
| 249 | |
| 250 | /* connect to the specified remote address, using whatever |
| 251 | * local address the routing configuration will pick. |
| 252 | */ |
| 253 | remote = msk->pm.remote; |
| 254 | if (!remote.port) |
| 255 | remote.port = sk->sk_dport; |
| 256 | memset(&local, 0, sizeof(local)); |
| 257 | local.family = remote.family; |
| 258 | |
| 259 | spin_unlock_bh(&msk->pm.lock); |
| 260 | __mptcp_subflow_connect((struct sock *)msk, 0, &local, &remote); |
| 261 | spin_lock_bh(&msk->pm.lock); |
| 262 | } |
| 263 | |
| 264 | static bool address_use_port(struct mptcp_pm_addr_entry *entry) |
| 265 | { |
| 266 | return (entry->flags & |
| 267 | (MPTCP_PM_ADDR_FLAG_SIGNAL | MPTCP_PM_ADDR_FLAG_SUBFLOW)) == |
| 268 | MPTCP_PM_ADDR_FLAG_SIGNAL; |
| 269 | } |
| 270 | |
| 271 | static int mptcp_pm_nl_append_new_local_addr(struct pm_nl_pernet *pernet, |
| 272 | struct mptcp_pm_addr_entry *entry) |
| 273 | { |
| 274 | struct mptcp_pm_addr_entry *cur; |
| 275 | int ret = -EINVAL; |
| 276 | |
| 277 | spin_lock_bh(&pernet->lock); |
| 278 | /* to keep the code simple, don't do IDR-like allocation for address ID, |
| 279 | * just bail when we exceed limits |
| 280 | */ |
| 281 | if (pernet->next_id > 255) |
| 282 | goto out; |
| 283 | if (pernet->addrs >= MPTCP_PM_ADDR_MAX) |
| 284 | goto out; |
| 285 | |
| 286 | /* do not insert duplicate address, differentiate on port only |
| 287 | * singled addresses |
| 288 | */ |
| 289 | list_for_each_entry(cur, &pernet->local_addr_list, list) { |
| 290 | if (addresses_equal(&cur->addr, &entry->addr, |
| 291 | address_use_port(entry) && |
| 292 | address_use_port(cur))) |
| 293 | goto out; |
| 294 | } |
| 295 | |
| 296 | if (entry->flags & MPTCP_PM_ADDR_FLAG_SIGNAL) |
| 297 | pernet->add_addr_signal_max++; |
| 298 | if (entry->flags & MPTCP_PM_ADDR_FLAG_SUBFLOW) |
| 299 | pernet->local_addr_max++; |
| 300 | |
| 301 | entry->addr.id = pernet->next_id++; |
| 302 | pernet->addrs++; |
| 303 | list_add_tail_rcu(&entry->list, &pernet->local_addr_list); |
| 304 | ret = entry->addr.id; |
| 305 | |
| 306 | out: |
| 307 | spin_unlock_bh(&pernet->lock); |
| 308 | return ret; |
| 309 | } |
| 310 | |
| 311 | int mptcp_pm_nl_get_local_id(struct mptcp_sock *msk, struct sock_common *skc) |
| 312 | { |
| 313 | struct mptcp_pm_addr_entry *entry; |
| 314 | struct mptcp_addr_info skc_local; |
| 315 | struct mptcp_addr_info msk_local; |
| 316 | struct pm_nl_pernet *pernet; |
| 317 | int ret = -1; |
| 318 | |
| 319 | if (WARN_ON_ONCE(!msk)) |
| 320 | return -1; |
| 321 | |
| 322 | /* The 0 ID mapping is defined by the first subflow, copied into the msk |
| 323 | * addr |
| 324 | */ |
| 325 | local_address((struct sock_common *)msk, &msk_local); |
| 326 | local_address((struct sock_common *)msk, &skc_local); |
| 327 | if (addresses_equal(&msk_local, &skc_local, false)) |
| 328 | return 0; |
| 329 | |
| 330 | pernet = net_generic(sock_net((struct sock *)msk), pm_nl_pernet_id); |
| 331 | |
| 332 | rcu_read_lock(); |
| 333 | list_for_each_entry_rcu(entry, &pernet->local_addr_list, list) { |
| 334 | if (addresses_equal(&entry->addr, &skc_local, false)) { |
| 335 | ret = entry->addr.id; |
| 336 | break; |
| 337 | } |
| 338 | } |
| 339 | rcu_read_unlock(); |
| 340 | if (ret >= 0) |
| 341 | return ret; |
| 342 | |
| 343 | /* address not found, add to local list */ |
| 344 | entry = kmalloc(sizeof(*entry), GFP_KERNEL); |
| 345 | if (!entry) |
| 346 | return -ENOMEM; |
| 347 | |
| 348 | entry->flags = 0; |
| 349 | entry->addr = skc_local; |
| 350 | ret = mptcp_pm_nl_append_new_local_addr(pernet, entry); |
| 351 | if (ret < 0) |
| 352 | kfree(entry); |
| 353 | |
| 354 | return ret; |
| 355 | } |
| 356 | |
| 357 | void mptcp_pm_nl_data_init(struct mptcp_sock *msk) |
| 358 | { |
| 359 | struct mptcp_pm_data *pm = &msk->pm; |
| 360 | struct pm_nl_pernet *pernet; |
| 361 | bool subflows; |
| 362 | |
| 363 | pernet = net_generic(sock_net((struct sock *)msk), pm_nl_pernet_id); |
| 364 | |
| 365 | pm->add_addr_signal_max = READ_ONCE(pernet->add_addr_signal_max); |
| 366 | pm->add_addr_accept_max = READ_ONCE(pernet->add_addr_accept_max); |
| 367 | pm->local_addr_max = READ_ONCE(pernet->local_addr_max); |
| 368 | pm->subflows_max = READ_ONCE(pernet->subflows_max); |
| 369 | subflows = !!pm->subflows_max; |
| 370 | WRITE_ONCE(pm->work_pending, (!!pm->local_addr_max && subflows) || |
| 371 | !!pm->add_addr_signal_max); |
| 372 | WRITE_ONCE(pm->accept_addr, !!pm->add_addr_accept_max && subflows); |
| 373 | WRITE_ONCE(pm->accept_subflow, subflows); |
| 374 | } |
| 375 | |
| 376 | #define MPTCP_PM_CMD_GRP_OFFSET 0 |
| 377 | |
| 378 | static const struct genl_multicast_group mptcp_pm_mcgrps[] = { |
| 379 | [MPTCP_PM_CMD_GRP_OFFSET] = { .name = MPTCP_PM_CMD_GRP_NAME, }, |
| 380 | }; |
| 381 | |
| 382 | static const struct nla_policy |
| 383 | mptcp_pm_addr_policy[MPTCP_PM_ADDR_ATTR_MAX + 1] = { |
| 384 | [MPTCP_PM_ADDR_ATTR_FAMILY] = { .type = NLA_U16, }, |
| 385 | [MPTCP_PM_ADDR_ATTR_ID] = { .type = NLA_U8, }, |
| 386 | [MPTCP_PM_ADDR_ATTR_ADDR4] = { .type = NLA_U32, }, |
| 387 | [MPTCP_PM_ADDR_ATTR_ADDR6] = { .type = NLA_EXACT_LEN, |
| 388 | .len = sizeof(struct in6_addr), }, |
| 389 | [MPTCP_PM_ADDR_ATTR_PORT] = { .type = NLA_U16 }, |
| 390 | [MPTCP_PM_ADDR_ATTR_FLAGS] = { .type = NLA_U32 }, |
| 391 | [MPTCP_PM_ADDR_ATTR_IF_IDX] = { .type = NLA_S32 }, |
| 392 | }; |
| 393 | |
| 394 | static const struct nla_policy mptcp_pm_policy[MPTCP_PM_ATTR_MAX + 1] = { |
| 395 | [MPTCP_PM_ATTR_ADDR] = |
| 396 | NLA_POLICY_NESTED(mptcp_pm_addr_policy), |
| 397 | [MPTCP_PM_ATTR_RCV_ADD_ADDRS] = { .type = NLA_U32, }, |
| 398 | [MPTCP_PM_ATTR_SUBFLOWS] = { .type = NLA_U32, }, |
| 399 | }; |
| 400 | |
| 401 | static int mptcp_pm_family_to_addr(int family) |
| 402 | { |
| 403 | #if IS_ENABLED(CONFIG_MPTCP_IPV6) |
| 404 | if (family == AF_INET6) |
| 405 | return MPTCP_PM_ADDR_ATTR_ADDR6; |
| 406 | #endif |
| 407 | return MPTCP_PM_ADDR_ATTR_ADDR4; |
| 408 | } |
| 409 | |
| 410 | static int mptcp_pm_parse_addr(struct nlattr *attr, struct genl_info *info, |
| 411 | bool require_family, |
| 412 | struct mptcp_pm_addr_entry *entry) |
| 413 | { |
| 414 | struct nlattr *tb[MPTCP_PM_ADDR_ATTR_MAX + 1]; |
| 415 | int err, addr_addr; |
| 416 | |
| 417 | if (!attr) { |
| 418 | GENL_SET_ERR_MSG(info, "missing address info"); |
| 419 | return -EINVAL; |
| 420 | } |
| 421 | |
| 422 | /* no validation needed - was already done via nested policy */ |
| 423 | err = nla_parse_nested_deprecated(tb, MPTCP_PM_ADDR_ATTR_MAX, attr, |
| 424 | mptcp_pm_addr_policy, info->extack); |
| 425 | if (err) |
| 426 | return err; |
| 427 | |
| 428 | memset(entry, 0, sizeof(*entry)); |
| 429 | if (!tb[MPTCP_PM_ADDR_ATTR_FAMILY]) { |
| 430 | if (!require_family) |
| 431 | goto skip_family; |
| 432 | |
| 433 | NL_SET_ERR_MSG_ATTR(info->extack, attr, |
| 434 | "missing family"); |
| 435 | return -EINVAL; |
| 436 | } |
| 437 | |
| 438 | entry->addr.family = nla_get_u16(tb[MPTCP_PM_ADDR_ATTR_FAMILY]); |
| 439 | if (entry->addr.family != AF_INET |
| 440 | #if IS_ENABLED(CONFIG_MPTCP_IPV6) |
| 441 | && entry->addr.family != AF_INET6 |
| 442 | #endif |
| 443 | ) { |
| 444 | NL_SET_ERR_MSG_ATTR(info->extack, attr, |
| 445 | "unknown address family"); |
| 446 | return -EINVAL; |
| 447 | } |
| 448 | addr_addr = mptcp_pm_family_to_addr(entry->addr.family); |
| 449 | if (!tb[addr_addr]) { |
| 450 | NL_SET_ERR_MSG_ATTR(info->extack, attr, |
| 451 | "missing address data"); |
| 452 | return -EINVAL; |
| 453 | } |
| 454 | |
| 455 | #if IS_ENABLED(CONFIG_MPTCP_IPV6) |
| 456 | if (entry->addr.family == AF_INET6) |
| 457 | entry->addr.addr6 = nla_get_in6_addr(tb[addr_addr]); |
| 458 | else |
| 459 | #endif |
| 460 | entry->addr.addr.s_addr = nla_get_in_addr(tb[addr_addr]); |
| 461 | |
| 462 | skip_family: |
| 463 | if (tb[MPTCP_PM_ADDR_ATTR_IF_IDX]) |
| 464 | entry->ifindex = nla_get_s32(tb[MPTCP_PM_ADDR_ATTR_IF_IDX]); |
| 465 | |
| 466 | if (tb[MPTCP_PM_ADDR_ATTR_ID]) |
| 467 | entry->addr.id = nla_get_u8(tb[MPTCP_PM_ADDR_ATTR_ID]); |
| 468 | |
| 469 | if (tb[MPTCP_PM_ADDR_ATTR_FLAGS]) |
| 470 | entry->flags = nla_get_u32(tb[MPTCP_PM_ADDR_ATTR_FLAGS]); |
| 471 | |
| 472 | return 0; |
| 473 | } |
| 474 | |
| 475 | static struct pm_nl_pernet *genl_info_pm_nl(struct genl_info *info) |
| 476 | { |
| 477 | return net_generic(genl_info_net(info), pm_nl_pernet_id); |
| 478 | } |
| 479 | |
| 480 | static int mptcp_nl_cmd_add_addr(struct sk_buff *skb, struct genl_info *info) |
| 481 | { |
| 482 | struct nlattr *attr = info->attrs[MPTCP_PM_ATTR_ADDR]; |
| 483 | struct pm_nl_pernet *pernet = genl_info_pm_nl(info); |
| 484 | struct mptcp_pm_addr_entry addr, *entry; |
| 485 | int ret; |
| 486 | |
| 487 | ret = mptcp_pm_parse_addr(attr, info, true, &addr); |
| 488 | if (ret < 0) |
| 489 | return ret; |
| 490 | |
| 491 | entry = kmalloc(sizeof(*entry), GFP_KERNEL); |
| 492 | if (!entry) { |
| 493 | GENL_SET_ERR_MSG(info, "can't allocate addr"); |
| 494 | return -ENOMEM; |
| 495 | } |
| 496 | |
| 497 | *entry = addr; |
| 498 | ret = mptcp_pm_nl_append_new_local_addr(pernet, entry); |
| 499 | if (ret < 0) { |
| 500 | GENL_SET_ERR_MSG(info, "too many addresses or duplicate one"); |
| 501 | kfree(entry); |
| 502 | return ret; |
| 503 | } |
| 504 | |
| 505 | return 0; |
| 506 | } |
| 507 | |
| 508 | static struct mptcp_pm_addr_entry * |
| 509 | __lookup_addr_by_id(struct pm_nl_pernet *pernet, unsigned int id) |
| 510 | { |
| 511 | struct mptcp_pm_addr_entry *entry; |
| 512 | |
| 513 | list_for_each_entry(entry, &pernet->local_addr_list, list) { |
| 514 | if (entry->addr.id == id) |
| 515 | return entry; |
| 516 | } |
| 517 | return NULL; |
| 518 | } |
| 519 | |
| 520 | static int mptcp_nl_cmd_del_addr(struct sk_buff *skb, struct genl_info *info) |
| 521 | { |
| 522 | struct nlattr *attr = info->attrs[MPTCP_PM_ATTR_ADDR]; |
| 523 | struct pm_nl_pernet *pernet = genl_info_pm_nl(info); |
| 524 | struct mptcp_pm_addr_entry addr, *entry; |
| 525 | int ret; |
| 526 | |
| 527 | ret = mptcp_pm_parse_addr(attr, info, false, &addr); |
| 528 | if (ret < 0) |
| 529 | return ret; |
| 530 | |
| 531 | spin_lock_bh(&pernet->lock); |
| 532 | entry = __lookup_addr_by_id(pernet, addr.addr.id); |
| 533 | if (!entry) { |
| 534 | GENL_SET_ERR_MSG(info, "address not found"); |
| 535 | ret = -EINVAL; |
| 536 | goto out; |
| 537 | } |
| 538 | if (entry->flags & MPTCP_PM_ADDR_FLAG_SIGNAL) |
| 539 | pernet->add_addr_signal_max--; |
| 540 | if (entry->flags & MPTCP_PM_ADDR_FLAG_SUBFLOW) |
| 541 | pernet->local_addr_max--; |
| 542 | |
| 543 | pernet->addrs--; |
| 544 | list_del_rcu(&entry->list); |
| 545 | kfree_rcu(entry, rcu); |
| 546 | out: |
| 547 | spin_unlock_bh(&pernet->lock); |
| 548 | return ret; |
| 549 | } |
| 550 | |
| 551 | static void __flush_addrs(struct pm_nl_pernet *pernet) |
| 552 | { |
| 553 | while (!list_empty(&pernet->local_addr_list)) { |
| 554 | struct mptcp_pm_addr_entry *cur; |
| 555 | |
| 556 | cur = list_entry(pernet->local_addr_list.next, |
| 557 | struct mptcp_pm_addr_entry, list); |
| 558 | list_del_rcu(&cur->list); |
| 559 | kfree_rcu(cur, rcu); |
| 560 | } |
| 561 | } |
| 562 | |
| 563 | static void __reset_counters(struct pm_nl_pernet *pernet) |
| 564 | { |
| 565 | pernet->add_addr_signal_max = 0; |
| 566 | pernet->add_addr_accept_max = 0; |
| 567 | pernet->local_addr_max = 0; |
| 568 | pernet->addrs = 0; |
| 569 | } |
| 570 | |
| 571 | static int mptcp_nl_cmd_flush_addrs(struct sk_buff *skb, struct genl_info *info) |
| 572 | { |
| 573 | struct pm_nl_pernet *pernet = genl_info_pm_nl(info); |
| 574 | |
| 575 | spin_lock_bh(&pernet->lock); |
| 576 | __flush_addrs(pernet); |
| 577 | __reset_counters(pernet); |
| 578 | spin_unlock_bh(&pernet->lock); |
| 579 | return 0; |
| 580 | } |
| 581 | |
| 582 | static int mptcp_nl_fill_addr(struct sk_buff *skb, |
| 583 | struct mptcp_pm_addr_entry *entry) |
| 584 | { |
| 585 | struct mptcp_addr_info *addr = &entry->addr; |
| 586 | struct nlattr *attr; |
| 587 | |
| 588 | attr = nla_nest_start(skb, MPTCP_PM_ATTR_ADDR); |
| 589 | if (!attr) |
| 590 | return -EMSGSIZE; |
| 591 | |
| 592 | if (nla_put_u16(skb, MPTCP_PM_ADDR_ATTR_FAMILY, addr->family)) |
| 593 | goto nla_put_failure; |
| 594 | if (nla_put_u8(skb, MPTCP_PM_ADDR_ATTR_ID, addr->id)) |
| 595 | goto nla_put_failure; |
| 596 | if (nla_put_u32(skb, MPTCP_PM_ADDR_ATTR_FLAGS, entry->flags)) |
| 597 | goto nla_put_failure; |
| 598 | if (entry->ifindex && |
| 599 | nla_put_s32(skb, MPTCP_PM_ADDR_ATTR_IF_IDX, entry->ifindex)) |
| 600 | goto nla_put_failure; |
| 601 | |
Bo YU | b4e0f9a | 2020-04-23 10:10:03 +0800 | [diff] [blame] | 602 | if (addr->family == AF_INET && |
| 603 | nla_put_in_addr(skb, MPTCP_PM_ADDR_ATTR_ADDR4, |
| 604 | addr->addr.s_addr)) |
| 605 | goto nla_put_failure; |
Paolo Abeni | 01cacb0 | 2020-03-27 14:48:51 -0700 | [diff] [blame] | 606 | #if IS_ENABLED(CONFIG_MPTCP_IPV6) |
Bo YU | b4e0f9a | 2020-04-23 10:10:03 +0800 | [diff] [blame] | 607 | else if (addr->family == AF_INET6 && |
| 608 | nla_put_in6_addr(skb, MPTCP_PM_ADDR_ATTR_ADDR6, &addr->addr6)) |
| 609 | goto nla_put_failure; |
Paolo Abeni | 01cacb0 | 2020-03-27 14:48:51 -0700 | [diff] [blame] | 610 | #endif |
| 611 | nla_nest_end(skb, attr); |
| 612 | return 0; |
| 613 | |
| 614 | nla_put_failure: |
| 615 | nla_nest_cancel(skb, attr); |
| 616 | return -EMSGSIZE; |
| 617 | } |
| 618 | |
| 619 | static int mptcp_nl_cmd_get_addr(struct sk_buff *skb, struct genl_info *info) |
| 620 | { |
| 621 | struct nlattr *attr = info->attrs[MPTCP_PM_ATTR_ADDR]; |
| 622 | struct pm_nl_pernet *pernet = genl_info_pm_nl(info); |
| 623 | struct mptcp_pm_addr_entry addr, *entry; |
| 624 | struct sk_buff *msg; |
| 625 | void *reply; |
| 626 | int ret; |
| 627 | |
| 628 | ret = mptcp_pm_parse_addr(attr, info, false, &addr); |
| 629 | if (ret < 0) |
| 630 | return ret; |
| 631 | |
| 632 | msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL); |
| 633 | if (!msg) |
| 634 | return -ENOMEM; |
| 635 | |
| 636 | reply = genlmsg_put_reply(msg, info, &mptcp_genl_family, 0, |
| 637 | info->genlhdr->cmd); |
| 638 | if (!reply) { |
| 639 | GENL_SET_ERR_MSG(info, "not enough space in Netlink message"); |
| 640 | ret = -EMSGSIZE; |
| 641 | goto fail; |
| 642 | } |
| 643 | |
| 644 | spin_lock_bh(&pernet->lock); |
| 645 | entry = __lookup_addr_by_id(pernet, addr.addr.id); |
| 646 | if (!entry) { |
| 647 | GENL_SET_ERR_MSG(info, "address not found"); |
| 648 | ret = -EINVAL; |
| 649 | goto unlock_fail; |
| 650 | } |
| 651 | |
| 652 | ret = mptcp_nl_fill_addr(msg, entry); |
| 653 | if (ret) |
| 654 | goto unlock_fail; |
| 655 | |
| 656 | genlmsg_end(msg, reply); |
| 657 | ret = genlmsg_reply(msg, info); |
| 658 | spin_unlock_bh(&pernet->lock); |
| 659 | return ret; |
| 660 | |
| 661 | unlock_fail: |
| 662 | spin_unlock_bh(&pernet->lock); |
| 663 | |
| 664 | fail: |
| 665 | nlmsg_free(msg); |
| 666 | return ret; |
| 667 | } |
| 668 | |
| 669 | static int mptcp_nl_cmd_dump_addrs(struct sk_buff *msg, |
| 670 | struct netlink_callback *cb) |
| 671 | { |
| 672 | struct net *net = sock_net(msg->sk); |
| 673 | struct mptcp_pm_addr_entry *entry; |
| 674 | struct pm_nl_pernet *pernet; |
| 675 | int id = cb->args[0]; |
| 676 | void *hdr; |
| 677 | |
| 678 | pernet = net_generic(net, pm_nl_pernet_id); |
| 679 | |
| 680 | spin_lock_bh(&pernet->lock); |
| 681 | list_for_each_entry(entry, &pernet->local_addr_list, list) { |
| 682 | if (entry->addr.id <= id) |
| 683 | continue; |
| 684 | |
| 685 | hdr = genlmsg_put(msg, NETLINK_CB(cb->skb).portid, |
| 686 | cb->nlh->nlmsg_seq, &mptcp_genl_family, |
| 687 | NLM_F_MULTI, MPTCP_PM_CMD_GET_ADDR); |
| 688 | if (!hdr) |
| 689 | break; |
| 690 | |
| 691 | if (mptcp_nl_fill_addr(msg, entry) < 0) { |
| 692 | genlmsg_cancel(msg, hdr); |
| 693 | break; |
| 694 | } |
| 695 | |
| 696 | id = entry->addr.id; |
| 697 | genlmsg_end(msg, hdr); |
| 698 | } |
| 699 | spin_unlock_bh(&pernet->lock); |
| 700 | |
| 701 | cb->args[0] = id; |
| 702 | return msg->len; |
| 703 | } |
| 704 | |
| 705 | static int parse_limit(struct genl_info *info, int id, unsigned int *limit) |
| 706 | { |
| 707 | struct nlattr *attr = info->attrs[id]; |
| 708 | |
| 709 | if (!attr) |
| 710 | return 0; |
| 711 | |
| 712 | *limit = nla_get_u32(attr); |
| 713 | if (*limit > MPTCP_PM_ADDR_MAX) { |
| 714 | GENL_SET_ERR_MSG(info, "limit greater than maximum"); |
| 715 | return -EINVAL; |
| 716 | } |
| 717 | return 0; |
| 718 | } |
| 719 | |
| 720 | static int |
| 721 | mptcp_nl_cmd_set_limits(struct sk_buff *skb, struct genl_info *info) |
| 722 | { |
| 723 | struct pm_nl_pernet *pernet = genl_info_pm_nl(info); |
| 724 | unsigned int rcv_addrs, subflows; |
| 725 | int ret; |
| 726 | |
| 727 | spin_lock_bh(&pernet->lock); |
| 728 | rcv_addrs = pernet->add_addr_accept_max; |
| 729 | ret = parse_limit(info, MPTCP_PM_ATTR_RCV_ADD_ADDRS, &rcv_addrs); |
| 730 | if (ret) |
| 731 | goto unlock; |
| 732 | |
| 733 | subflows = pernet->subflows_max; |
| 734 | ret = parse_limit(info, MPTCP_PM_ATTR_SUBFLOWS, &subflows); |
| 735 | if (ret) |
| 736 | goto unlock; |
| 737 | |
| 738 | WRITE_ONCE(pernet->add_addr_accept_max, rcv_addrs); |
| 739 | WRITE_ONCE(pernet->subflows_max, subflows); |
| 740 | |
| 741 | unlock: |
| 742 | spin_unlock_bh(&pernet->lock); |
| 743 | return ret; |
| 744 | } |
| 745 | |
| 746 | static int |
| 747 | mptcp_nl_cmd_get_limits(struct sk_buff *skb, struct genl_info *info) |
| 748 | { |
| 749 | struct pm_nl_pernet *pernet = genl_info_pm_nl(info); |
| 750 | struct sk_buff *msg; |
| 751 | void *reply; |
| 752 | |
| 753 | msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL); |
| 754 | if (!msg) |
| 755 | return -ENOMEM; |
| 756 | |
| 757 | reply = genlmsg_put_reply(msg, info, &mptcp_genl_family, 0, |
| 758 | MPTCP_PM_CMD_GET_LIMITS); |
| 759 | if (!reply) |
| 760 | goto fail; |
| 761 | |
| 762 | if (nla_put_u32(msg, MPTCP_PM_ATTR_RCV_ADD_ADDRS, |
| 763 | READ_ONCE(pernet->add_addr_accept_max))) |
| 764 | goto fail; |
| 765 | |
| 766 | if (nla_put_u32(msg, MPTCP_PM_ATTR_SUBFLOWS, |
| 767 | READ_ONCE(pernet->subflows_max))) |
| 768 | goto fail; |
| 769 | |
| 770 | genlmsg_end(msg, reply); |
| 771 | return genlmsg_reply(msg, info); |
| 772 | |
| 773 | fail: |
| 774 | GENL_SET_ERR_MSG(info, "not enough space in Netlink message"); |
| 775 | nlmsg_free(msg); |
| 776 | return -EMSGSIZE; |
| 777 | } |
| 778 | |
| 779 | static struct genl_ops mptcp_pm_ops[] = { |
| 780 | { |
| 781 | .cmd = MPTCP_PM_CMD_ADD_ADDR, |
| 782 | .doit = mptcp_nl_cmd_add_addr, |
| 783 | .flags = GENL_ADMIN_PERM, |
| 784 | }, |
| 785 | { |
| 786 | .cmd = MPTCP_PM_CMD_DEL_ADDR, |
| 787 | .doit = mptcp_nl_cmd_del_addr, |
| 788 | .flags = GENL_ADMIN_PERM, |
| 789 | }, |
| 790 | { |
| 791 | .cmd = MPTCP_PM_CMD_FLUSH_ADDRS, |
| 792 | .doit = mptcp_nl_cmd_flush_addrs, |
| 793 | .flags = GENL_ADMIN_PERM, |
| 794 | }, |
| 795 | { |
| 796 | .cmd = MPTCP_PM_CMD_GET_ADDR, |
| 797 | .doit = mptcp_nl_cmd_get_addr, |
| 798 | .dumpit = mptcp_nl_cmd_dump_addrs, |
| 799 | }, |
| 800 | { |
| 801 | .cmd = MPTCP_PM_CMD_SET_LIMITS, |
| 802 | .doit = mptcp_nl_cmd_set_limits, |
| 803 | .flags = GENL_ADMIN_PERM, |
| 804 | }, |
| 805 | { |
| 806 | .cmd = MPTCP_PM_CMD_GET_LIMITS, |
| 807 | .doit = mptcp_nl_cmd_get_limits, |
| 808 | }, |
| 809 | }; |
| 810 | |
| 811 | static struct genl_family mptcp_genl_family __ro_after_init = { |
| 812 | .name = MPTCP_PM_NAME, |
| 813 | .version = MPTCP_PM_VER, |
| 814 | .maxattr = MPTCP_PM_ATTR_MAX, |
| 815 | .policy = mptcp_pm_policy, |
| 816 | .netnsok = true, |
| 817 | .module = THIS_MODULE, |
| 818 | .ops = mptcp_pm_ops, |
| 819 | .n_ops = ARRAY_SIZE(mptcp_pm_ops), |
| 820 | .mcgrps = mptcp_pm_mcgrps, |
| 821 | .n_mcgrps = ARRAY_SIZE(mptcp_pm_mcgrps), |
| 822 | }; |
| 823 | |
| 824 | static int __net_init pm_nl_init_net(struct net *net) |
| 825 | { |
| 826 | struct pm_nl_pernet *pernet = net_generic(net, pm_nl_pernet_id); |
| 827 | |
| 828 | INIT_LIST_HEAD_RCU(&pernet->local_addr_list); |
| 829 | __reset_counters(pernet); |
| 830 | pernet->next_id = 1; |
| 831 | spin_lock_init(&pernet->lock); |
| 832 | return 0; |
| 833 | } |
| 834 | |
| 835 | static void __net_exit pm_nl_exit_net(struct list_head *net_list) |
| 836 | { |
| 837 | struct net *net; |
| 838 | |
| 839 | list_for_each_entry(net, net_list, exit_list) { |
| 840 | /* net is removed from namespace list, can't race with |
| 841 | * other modifiers |
| 842 | */ |
| 843 | __flush_addrs(net_generic(net, pm_nl_pernet_id)); |
| 844 | } |
| 845 | } |
| 846 | |
| 847 | static struct pernet_operations mptcp_pm_pernet_ops = { |
| 848 | .init = pm_nl_init_net, |
| 849 | .exit_batch = pm_nl_exit_net, |
| 850 | .id = &pm_nl_pernet_id, |
| 851 | .size = sizeof(struct pm_nl_pernet), |
| 852 | }; |
| 853 | |
Paolo Abeni | d39dcec | 2020-06-26 19:29:59 +0200 | [diff] [blame^] | 854 | void __init mptcp_pm_nl_init(void) |
Paolo Abeni | 01cacb0 | 2020-03-27 14:48:51 -0700 | [diff] [blame] | 855 | { |
| 856 | if (register_pernet_subsys(&mptcp_pm_pernet_ops) < 0) |
| 857 | panic("Failed to register MPTCP PM pernet subsystem.\n"); |
| 858 | |
| 859 | if (genl_register_family(&mptcp_genl_family)) |
| 860 | panic("Failed to register MPTCP PM netlink family\n"); |
| 861 | } |