drivers/net/vxlan/vxlan_vnifilter.c

   1 // SPDX-License-Identifier: GPL-2.0-only
   2 /*
   3  *      Vxlan vni filter for collect metadata mode
   4  *
   5  *      Authors: Roopa Prabhu <roopa@nvidia.com>
   6  *
   7  */
   8
   9 #include <linux/kernel.h>
  10 #include <linux/slab.h>
  11 #include <linux/etherdevice.h>
  12 #include <linux/rhashtable.h>
  13 #include <net/rtnetlink.h>
  14 #include <net/net_namespace.h>
  15 #include <net/sock.h>
  16 #include <net/vxlan.h>
  17
  18 #include "vxlan_private.h"
  19
  20 static inline int vxlan_vni_cmp(struct rhashtable_compare_arg *arg,
  21                                 const void *ptr)
  22 {
  23         const struct vxlan_vni_node *vnode = ptr;
  24         __be32 vni = *(__be32 *)arg->key;
  25
  26         return vnode->vni != vni;
  27 }
  28
  29 const struct rhashtable_params vxlan_vni_rht_params = {
  30         .head_offset = offsetof(struct vxlan_vni_node, vnode),
  31         .key_offset = offsetof(struct vxlan_vni_node, vni),
  32         .key_len = sizeof(__be32),
  33         .nelem_hint = 3,
  34         .max_size = VXLAN_N_VID,
  35         .obj_cmpfn = vxlan_vni_cmp,
  36         .automatic_shrinking = true,
  37 };
  38
  39 static void vxlan_vs_add_del_vninode(struct vxlan_dev *vxlan,
  40                                      struct vxlan_vni_node *v,
  41                                      bool del)
  42 {
  43         struct vxlan_net *vn = net_generic(vxlan->net, vxlan_net_id);
  44         struct vxlan_dev_node *node;
  45         struct vxlan_sock *vs;
  46
  47         spin_lock(&vn->sock_lock);
  48         if (del) {
  49                 if (!hlist_unhashed(&v->hlist4.hlist))
  50                         hlist_del_init_rcu(&v->hlist4.hlist);
  51 #if IS_ENABLED(CONFIG_IPV6)
  52                 if (!hlist_unhashed(&v->hlist6.hlist))
  53                         hlist_del_init_rcu(&v->hlist6.hlist);
  54 #endif
  55                 goto out;
  56         }
  57
  58 #if IS_ENABLED(CONFIG_IPV6)
  59         vs = rtnl_dereference(vxlan->vn6_sock);
  60         if (vs && v) {
  61                 node = &v->hlist6;
  62                 hlist_add_head_rcu(&node->hlist, vni_head(vs, v->vni));
  63         }
  64 #endif
  65         vs = rtnl_dereference(vxlan->vn4_sock);
  66         if (vs && v) {
  67                 node = &v->hlist4;
  68                 hlist_add_head_rcu(&node->hlist, vni_head(vs, v->vni));
  69         }
  70 out:
  71         spin_unlock(&vn->sock_lock);
  72 }
  73
  74 void vxlan_vs_add_vnigrp(struct vxlan_dev *vxlan,
  75                          struct vxlan_sock *vs,
  76                          bool ipv6)
  77 {
  78         struct vxlan_net *vn = net_generic(vxlan->net, vxlan_net_id);
  79         struct vxlan_vni_group *vg = rtnl_dereference(vxlan->vnigrp);
  80         struct vxlan_vni_node *v, *tmp;
  81         struct vxlan_dev_node *node;
  82
  83         if (!vg)
  84                 return;
  85
  86         spin_lock(&vn->sock_lock);
  87         list_for_each_entry_safe(v, tmp, &vg->vni_list, vlist) {
  88 #if IS_ENABLED(CONFIG_IPV6)
  89                 if (ipv6)
  90                         node = &v->hlist6;
  91                 else
  92 #endif
  93                         node = &v->hlist4;
  94                 node->vxlan = vxlan;
  95                 hlist_add_head_rcu(&node->hlist, vni_head(vs, v->vni));
  96         }
  97         spin_unlock(&vn->sock_lock);
  98 }
  99
 100 void vxlan_vs_del_vnigrp(struct vxlan_dev *vxlan)
 101 {
 102         struct vxlan_vni_group *vg = rtnl_dereference(vxlan->vnigrp);
 103         struct vxlan_net *vn = net_generic(vxlan->net, vxlan_net_id);
 104         struct vxlan_vni_node *v, *tmp;
 105
 106         if (!vg)
 107                 return;
 108
 109         spin_lock(&vn->sock_lock);
 110         list_for_each_entry_safe(v, tmp, &vg->vni_list, vlist) {
 111                 hlist_del_init_rcu(&v->hlist4.hlist);
 112 #if IS_ENABLED(CONFIG_IPV6)
 113                 hlist_del_init_rcu(&v->hlist6.hlist);
 114 #endif
 115         }
 116         spin_unlock(&vn->sock_lock);
 117 }
 118
 119 static u32 vnirange(struct vxlan_vni_node *vbegin,
 120                     struct vxlan_vni_node *vend)
 121 {
 122         return (be32_to_cpu(vend->vni) - be32_to_cpu(vbegin->vni));
 123 }
 124
 125 static size_t vxlan_vnifilter_entry_nlmsg_size(void)
 126 {
 127         return NLMSG_ALIGN(sizeof(struct tunnel_msg))
 128                 + nla_total_size(0) /* VXLAN_VNIFILTER_ENTRY */
 129                 + nla_total_size(sizeof(u32)) /* VXLAN_VNIFILTER_ENTRY_START */
 130                 + nla_total_size(sizeof(u32)) /* VXLAN_VNIFILTER_ENTRY_END */
 131                 + nla_total_size(sizeof(struct in6_addr));/* VXLAN_VNIFILTER_ENTRY_GROUP{6} */
 132 }
 133
 134 static bool vxlan_fill_vni_filter_entry(struct sk_buff *skb,
 135                                         struct vxlan_vni_node *vbegin,
 136                                         struct vxlan_vni_node *vend)
 137 {
 138         struct nlattr *ventry;
 139         u32 vs = be32_to_cpu(vbegin->vni);
 140         u32 ve = 0;
 141
 142         if (vbegin != vend)
 143                 ve = be32_to_cpu(vend->vni);
 144
 145         ventry = nla_nest_start(skb, VXLAN_VNIFILTER_ENTRY);
 146         if (!ventry)
 147                 return false;
 148
 149         if (nla_put_u32(skb, VXLAN_VNIFILTER_ENTRY_START, vs))
 150                 goto out_err;
 151
 152         if (ve && nla_put_u32(skb, VXLAN_VNIFILTER_ENTRY_END, ve))
 153                 goto out_err;
 154
 155         if (!vxlan_addr_any(&vbegin->remote_ip)) {
 156                 if (vbegin->remote_ip.sa.sa_family == AF_INET) {
 157                         if (nla_put_in_addr(skb, VXLAN_VNIFILTER_ENTRY_GROUP,
 158                                             vbegin->remote_ip.sin.sin_addr.s_addr))
 159                                 goto out_err;
 160 #if IS_ENABLED(CONFIG_IPV6)
 161                 } else {
 162                         if (nla_put_in6_addr(skb, VXLAN_VNIFILTER_ENTRY_GROUP6,
 163                                              &vbegin->remote_ip.sin6.sin6_addr))
 164                                 goto out_err;
 165 #endif
 166                 }
 167         }
 168
 169         nla_nest_end(skb, ventry);
 170
 171         return true;
 172
 173 out_err:
 174         nla_nest_cancel(skb, ventry);
 175
 176         return false;
 177 }
 178
 179 static void vxlan_vnifilter_notify(const struct vxlan_dev *vxlan,
 180                                    struct vxlan_vni_node *vninode, int cmd)
 181 {
 182         struct tunnel_msg *tmsg;
 183         struct sk_buff *skb;
 184         struct nlmsghdr *nlh;
 185         struct net *net = dev_net(vxlan->dev);
 186         int err = -ENOBUFS;
 187
 188         skb = nlmsg_new(vxlan_vnifilter_entry_nlmsg_size(), GFP_KERNEL);
 189         if (!skb)
 190                 goto out_err;
 191
 192         err = -EMSGSIZE;
 193         nlh = nlmsg_put(skb, 0, 0, cmd, sizeof(*tmsg), 0);
 194         if (!nlh)
 195                 goto out_err;
 196         tmsg = nlmsg_data(nlh);
 197         memset(tmsg, 0, sizeof(*tmsg));
 198         tmsg->family = AF_BRIDGE;
 199         tmsg->ifindex = vxlan->dev->ifindex;
 200
 201         if (!vxlan_fill_vni_filter_entry(skb, vninode, vninode))
 202                 goto out_err;
 203
 204         nlmsg_end(skb, nlh);
 205         rtnl_notify(skb, net, 0, RTNLGRP_TUNNEL, NULL, GFP_KERNEL);
 206
 207         return;
 208
 209 out_err:
 210         rtnl_set_sk_err(net, RTNLGRP_TUNNEL, err);
 211
 212         kfree_skb(skb);
 213 }
 214
 215 static int vxlan_vnifilter_dump_dev(const struct net_device *dev,
 216                                     struct sk_buff *skb,
 217                                     struct netlink_callback *cb)
 218 {
 219         struct vxlan_vni_node *tmp, *v, *vbegin = NULL, *vend = NULL;
 220         struct vxlan_dev *vxlan = netdev_priv(dev);
 221         struct tunnel_msg *new_tmsg;
 222         int idx = 0, s_idx = cb->args[1];
 223         struct vxlan_vni_group *vg;
 224         struct nlmsghdr *nlh;
 225         int err = 0;
 226
 227         if (!(vxlan->cfg.flags & VXLAN_F_VNIFILTER))
 228                 return -EINVAL;
 229
 230         /* RCU needed because of the vni locking rules (rcu || rtnl) */
 231         vg = rcu_dereference(vxlan->vnigrp);
 232         if (!vg || !vg->num_vnis)
 233                 return 0;
 234
 235         nlh = nlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
 236                         RTM_NEWTUNNEL, sizeof(*new_tmsg), NLM_F_MULTI);
 237         if (!nlh)
 238                 return -EMSGSIZE;
 239         new_tmsg = nlmsg_data(nlh);
 240         memset(new_tmsg, 0, sizeof(*new_tmsg));
 241         new_tmsg->family = PF_BRIDGE;
 242         new_tmsg->ifindex = dev->ifindex;
 243
 244         list_for_each_entry_safe(v, tmp, &vg->vni_list, vlist) {
 245                 if (idx < s_idx) {
 246                         idx++;
 247                         continue;
 248                 }
 249                 if (!vbegin) {
 250                         vbegin = v;
 251                         vend = v;
 252                         continue;
 253                 }
 254                 if (vnirange(vend, v) == 1 &&
 255                     vxlan_addr_equal(&v->remote_ip, &vend->remote_ip)) {
 256                         goto update_end;
 257                 } else {
 258                         if (!vxlan_fill_vni_filter_entry(skb, vbegin, vend)) {
 259                                 err = -EMSGSIZE;
 260                                 break;
 261                         }
 262                         idx += vnirange(vbegin, vend) + 1;
 263                         vbegin = v;
 264                 }
 265 update_end:
 266                 vend = v;
 267         }
 268
 269         if (!err && vbegin) {
 270                 if (!vxlan_fill_vni_filter_entry(skb, vbegin, vend))
 271                         err = -EMSGSIZE;
 272         }
 273
 274         cb->args[1] = err ? idx : 0;
 275
 276         nlmsg_end(skb, nlh);
 277
 278         return err;
 279 }
 280
 281 static int vxlan_vnifilter_dump(struct sk_buff *skb, struct netlink_callback *cb)
 282 {
 283         int idx = 0, err = 0, s_idx = cb->args[0];
 284         struct net *net = sock_net(skb->sk);
 285         struct tunnel_msg *tmsg;
 286         struct net_device *dev;
 287
 288         tmsg = nlmsg_data(cb->nlh);
 289
 290         rcu_read_lock();
 291         if (tmsg->ifindex) {
 292                 dev = dev_get_by_index_rcu(net, tmsg->ifindex);
 293                 if (!dev) {
 294                         err = -ENODEV;
 295                         goto out_err;
 296                 }
 297                 err = vxlan_vnifilter_dump_dev(dev, skb, cb);
 298                 /* if the dump completed without an error we return 0 here */
 299                 if (err != -EMSGSIZE)
 300                         goto out_err;
 301         } else {
 302                 for_each_netdev_rcu(net, dev) {
 303                         if (!netif_is_vxlan(dev))
 304                                 continue;
 305                         if (idx < s_idx)
 306                                 goto skip;
 307                         err = vxlan_vnifilter_dump_dev(dev, skb, cb);
 308                         if (err == -EMSGSIZE)
 309                                 break;
 310 skip:
 311                         idx++;
 312                 }
 313         }
 314         cb->args[0] = idx;
 315         rcu_read_unlock();
 316
 317         return skb->len;
 318
 319 out_err:
 320         rcu_read_unlock();
 321
 322         return err;
 323 }
 324
 325 static const struct nla_policy vni_filter_entry_policy[VXLAN_VNIFILTER_ENTRY_MAX + 1] = {
 326         [VXLAN_VNIFILTER_ENTRY_START] = { .type = NLA_U32 },
 327         [VXLAN_VNIFILTER_ENTRY_END] = { .type = NLA_U32 },
 328         [VXLAN_VNIFILTER_ENTRY_GROUP]   = { .type = NLA_BINARY,
 329                                             .len = sizeof_field(struct iphdr, daddr) },
 330         [VXLAN_VNIFILTER_ENTRY_GROUP6]  = { .type = NLA_BINARY,
 331                                             .len = sizeof(struct in6_addr) },
 332 };
 333
 334 static const struct nla_policy vni_filter_policy[VXLAN_VNIFILTER_MAX + 1] = {
 335         [VXLAN_VNIFILTER_ENTRY] = { .type = NLA_NESTED },
 336 };
 337
 338 static int vxlan_update_default_fdb_entry(struct vxlan_dev *vxlan, __be32 vni,
 339                                           union vxlan_addr *old_remote_ip,
 340                                           union vxlan_addr *remote_ip,
 341                                           struct netlink_ext_ack *extack)
 342 {
 343         struct vxlan_rdst *dst = &vxlan->default_dst;
 344         u32 hash_index;
 345         int err = 0;
 346
 347         hash_index = fdb_head_index(vxlan, all_zeros_mac, vni);
 348         spin_lock_bh(&vxlan->hash_lock[hash_index]);
 349         if (remote_ip && !vxlan_addr_any(remote_ip)) {
 350                 err = vxlan_fdb_update(vxlan, all_zeros_mac,
 351                                        remote_ip,
 352                                        NUD_REACHABLE | NUD_PERMANENT,
 353                                        NLM_F_APPEND | NLM_F_CREATE,
 354                                        vxlan->cfg.dst_port,
 355                                        vni,
 356                                        vni,
 357                                        dst->remote_ifindex,
 358                                        NTF_SELF, 0, true, extack);
 359                 if (err) {
 360                         spin_unlock_bh(&vxlan->hash_lock[hash_index]);
 361                         return err;
 362                 }
 363         }
 364
 365         if (old_remote_ip && !vxlan_addr_any(old_remote_ip)) {
 366                 __vxlan_fdb_delete(vxlan, all_zeros_mac,
 367                                    *old_remote_ip,
 368                                    vxlan->cfg.dst_port,
 369                                    vni, vni,
 370                                    dst->remote_ifindex,
 371                                    true);
 372         }
 373         spin_unlock_bh(&vxlan->hash_lock[hash_index]);
 374
 375         return err;
 376 }
 377
 378 static int vxlan_vni_update_group(struct vxlan_dev *vxlan,
 379                                   struct vxlan_vni_node *vninode,
 380                                   union vxlan_addr *group,
 381                                   bool create, bool *changed,
 382                                   struct netlink_ext_ack *extack)
 383 {
 384         struct vxlan_net *vn = net_generic(vxlan->net, vxlan_net_id);
 385         struct vxlan_rdst *dst = &vxlan->default_dst;
 386         union vxlan_addr *newrip = NULL, *oldrip = NULL;
 387         union vxlan_addr old_remote_ip;
 388         int ret = 0;
 389
 390         memcpy(&old_remote_ip, &vninode->remote_ip, sizeof(old_remote_ip));
 391
 392         /* if per vni remote ip is not present use vxlan dev
 393          * default dst remote ip for fdb entry
 394          */
 395         if (group && !vxlan_addr_any(group)) {
 396                 newrip = group;
 397         } else {
 398                 if (!vxlan_addr_any(&dst->remote_ip))
 399                         newrip = &dst->remote_ip;
 400         }
 401
 402         /* if old rip exists, and no newrip,
 403          * explicitly delete old rip
 404          */
 405         if (!newrip && !vxlan_addr_any(&old_remote_ip))
 406                 oldrip = &old_remote_ip;
 407
 408         if (!newrip && !oldrip)
 409                 return 0;
 410
 411         if (!create && oldrip && newrip && vxlan_addr_equal(oldrip, newrip))
 412                 return 0;
 413
 414         ret = vxlan_update_default_fdb_entry(vxlan, vninode->vni,
 415                                              oldrip, newrip,
 416                                              extack);
 417         if (ret)
 418                 goto out;
 419
 420         if (group)
 421                 memcpy(&vninode->remote_ip, group, sizeof(vninode->remote_ip));
 422
 423         if (vxlan->dev->flags & IFF_UP) {
 424                 if (vxlan_addr_multicast(&old_remote_ip) &&
 425                     !vxlan_group_used(vn, vxlan, vninode->vni,
 426                                       &old_remote_ip,
 427                                       vxlan->default_dst.remote_ifindex)) {
 428                         ret = vxlan_igmp_leave(vxlan, &old_remote_ip,
 429                                                0);
 430                         if (ret)
 431                                 goto out;
 432                 }
 433
 434                 if (vxlan_addr_multicast(&vninode->remote_ip)) {
 435                         ret = vxlan_igmp_join(vxlan, &vninode->remote_ip, 0);
 436                         if (ret == -EADDRINUSE)
 437                                 ret = 0;
 438                         if (ret)
 439                                 goto out;
 440                 }
 441         }
 442
 443         *changed = true;
 444
 445         return 0;
 446 out:
 447         return ret;
 448 }
 449
 450 int vxlan_vnilist_update_group(struct vxlan_dev *vxlan,
 451                                union vxlan_addr *old_remote_ip,
 452                                union vxlan_addr *new_remote_ip,
 453                                struct netlink_ext_ack *extack)
 454 {
 455         struct list_head *headp, *hpos;
 456         struct vxlan_vni_group *vg;
 457         struct vxlan_vni_node *vent;
 458         int ret;
 459
 460         vg = rtnl_dereference(vxlan->vnigrp);
 461
 462         headp = &vg->vni_list;
 463         list_for_each_prev(hpos, headp) {
 464                 vent = list_entry(hpos, struct vxlan_vni_node, vlist);
 465                 if (vxlan_addr_any(&vent->remote_ip)) {
 466                         ret = vxlan_update_default_fdb_entry(vxlan, vent->vni,
 467                                                              old_remote_ip,
 468                                                              new_remote_ip,
 469                                                              extack);
 470                         if (ret)
 471                                 return ret;
 472                 }
 473         }
 474
 475         return 0;
 476 }
 477
 478 static void vxlan_vni_delete_group(struct vxlan_dev *vxlan,
 479                                    struct vxlan_vni_node *vninode)
 480 {
 481         struct vxlan_net *vn = net_generic(vxlan->net, vxlan_net_id);
 482         struct vxlan_rdst *dst = &vxlan->default_dst;
 483
 484         /* if per vni remote_ip not present, delete the
 485          * default dst remote_ip previously added for this vni
 486          */
 487         if (!vxlan_addr_any(&vninode->remote_ip) ||
 488             !vxlan_addr_any(&dst->remote_ip))
 489                 __vxlan_fdb_delete(vxlan, all_zeros_mac,
 490                                    (vxlan_addr_any(&vninode->remote_ip) ?
 491                                    dst->remote_ip : vninode->remote_ip),
 492                                    vxlan->cfg.dst_port,
 493                                    vninode->vni, vninode->vni,
 494                                    dst->remote_ifindex,
 495                                    true);
 496
 497         if (vxlan->dev->flags & IFF_UP) {
 498                 if (vxlan_addr_multicast(&vninode->remote_ip) &&
 499                     !vxlan_group_used(vn, vxlan, vninode->vni,
 500                                       &vninode->remote_ip,
 501                                       dst->remote_ifindex)) {
 502                         vxlan_igmp_leave(vxlan, &vninode->remote_ip, 0);
 503                 }
 504         }
 505 }
 506
 507 static int vxlan_vni_update(struct vxlan_dev *vxlan,
 508                             struct vxlan_vni_group *vg,
 509                             __be32 vni, union vxlan_addr *group,
 510                             bool *changed,
 511                             struct netlink_ext_ack *extack)
 512 {
 513         struct vxlan_vni_node *vninode;
 514         int ret;
 515
 516         vninode = rhashtable_lookup_fast(&vg->vni_hash, &vni,
 517                                          vxlan_vni_rht_params);
 518         if (!vninode)
 519                 return 0;
 520
 521         ret = vxlan_vni_update_group(vxlan, vninode, group, false, changed,
 522                                      extack);
 523         if (ret)
 524                 return ret;
 525
 526         if (changed)
 527                 vxlan_vnifilter_notify(vxlan, vninode, RTM_NEWTUNNEL);
 528
 529         return 0;
 530 }
 531
 532 static void __vxlan_vni_add_list(struct vxlan_vni_group *vg,
 533                                  struct vxlan_vni_node *v)
 534 {
 535         struct list_head *headp, *hpos;
 536         struct vxlan_vni_node *vent;
 537
 538         headp = &vg->vni_list;
 539         list_for_each_prev(hpos, headp) {
 540                 vent = list_entry(hpos, struct vxlan_vni_node, vlist);
 541                 if (be32_to_cpu(v->vni) < be32_to_cpu(vent->vni))
 542                         continue;
 543                 else
 544                         break;
 545         }
 546         list_add_rcu(&v->vlist, hpos);
 547         vg->num_vnis++;
 548 }
 549
 550 static void __vxlan_vni_del_list(struct vxlan_vni_group *vg,
 551                                  struct vxlan_vni_node *v)
 552 {
 553         list_del_rcu(&v->vlist);
 554         vg->num_vnis--;
 555 }
 556
 557 static struct vxlan_vni_node *vxlan_vni_alloc(struct vxlan_dev *vxlan,
 558                                               __be32 vni)
 559 {
 560         struct vxlan_vni_node *vninode;
 561
 562         vninode = kzalloc(sizeof(*vninode), GFP_ATOMIC);
 563         if (!vninode)
 564                 return NULL;
 565         vninode->vni = vni;
 566         vninode->hlist4.vxlan = vxlan;
 567 #if IS_ENABLED(CONFIG_IPV6)
 568         vninode->hlist6.vxlan = vxlan;
 569 #endif
 570
 571         return vninode;
 572 }
 573
 574 static int vxlan_vni_add(struct vxlan_dev *vxlan,
 575                          struct vxlan_vni_group *vg,
 576                          u32 vni, union vxlan_addr *group,
 577                          struct netlink_ext_ack *extack)
 578 {
 579         struct vxlan_vni_node *vninode;
 580         __be32 v = cpu_to_be32(vni);
 581         bool changed = false;
 582         int err = 0;
 583
 584         if (vxlan_vnifilter_lookup(vxlan, v))
 585                 return vxlan_vni_update(vxlan, vg, v, group, &changed, extack);
 586
 587         err = vxlan_vni_in_use(vxlan->net, vxlan, &vxlan->cfg, v);
 588         if (err) {
 589                 NL_SET_ERR_MSG(extack, "VNI in use");
 590                 return err;
 591         }
 592
 593         vninode = vxlan_vni_alloc(vxlan, v);
 594         if (!vninode)
 595                 return -ENOMEM;
 596
 597         err = rhashtable_lookup_insert_fast(&vg->vni_hash,
 598                                             &vninode->vnode,
 599                                             vxlan_vni_rht_params);
 600         if (err) {
 601                 kfree(vninode);
 602                 return err;
 603         }
 604
 605         __vxlan_vni_add_list(vg, vninode);
 606
 607         if (vxlan->dev->flags & IFF_UP)
 608                 vxlan_vs_add_del_vninode(vxlan, vninode, false);
 609
 610         err = vxlan_vni_update_group(vxlan, vninode, group, true, &changed,
 611                                      extack);
 612
 613         if (changed)
 614                 vxlan_vnifilter_notify(vxlan, vninode, RTM_NEWTUNNEL);
 615
 616         return err;
 617 }
 618
 619 static void vxlan_vni_node_rcu_free(struct rcu_head *rcu)
 620 {
 621         struct vxlan_vni_node *v;
 622
 623         v = container_of(rcu, struct vxlan_vni_node, rcu);
 624         kfree(v);
 625 }
 626
 627 static int vxlan_vni_del(struct vxlan_dev *vxlan,
 628                          struct vxlan_vni_group *vg,
 629                          u32 vni, struct netlink_ext_ack *extack)
 630 {
 631         struct vxlan_vni_node *vninode;
 632         __be32 v = cpu_to_be32(vni);
 633         int err = 0;
 634
 635         vg = rtnl_dereference(vxlan->vnigrp);
 636
 637         vninode = rhashtable_lookup_fast(&vg->vni_hash, &v,
 638                                          vxlan_vni_rht_params);
 639         if (!vninode) {
 640                 err = -ENOENT;
 641                 goto out;
 642         }
 643
 644         vxlan_vni_delete_group(vxlan, vninode);
 645
 646         err = rhashtable_remove_fast(&vg->vni_hash,
 647                                      &vninode->vnode,
 648                                      vxlan_vni_rht_params);
 649         if (err)
 650                 goto out;
 651
 652         __vxlan_vni_del_list(vg, vninode);
 653
 654         vxlan_vnifilter_notify(vxlan, vninode, RTM_DELTUNNEL);
 655
 656         if (vxlan->dev->flags & IFF_UP)
 657                 vxlan_vs_add_del_vninode(vxlan, vninode, true);
 658
 659         call_rcu(&vninode->rcu, vxlan_vni_node_rcu_free);
 660
 661         return 0;
 662 out:
 663         return err;
 664 }
 665
 666 static int vxlan_vni_add_del(struct vxlan_dev *vxlan, __u32 start_vni,
 667                              __u32 end_vni, union vxlan_addr *group,
 668                              int cmd, struct netlink_ext_ack *extack)
 669 {
 670         struct vxlan_vni_group *vg;
 671         int v, err = 0;
 672
 673         vg = rtnl_dereference(vxlan->vnigrp);
 674
 675         for (v = start_vni; v <= end_vni; v++) {
 676                 switch (cmd) {
 677                 case RTM_NEWTUNNEL:
 678                         err = vxlan_vni_add(vxlan, vg, v, group, extack);
 679                         break;
 680                 case RTM_DELTUNNEL:
 681                         err = vxlan_vni_del(vxlan, vg, v, extack);
 682                         break;
 683                 default:
 684                         err = -EOPNOTSUPP;
 685                         break;
 686                 }
 687                 if (err)
 688                         goto out;
 689         }
 690
 691         return 0;
 692 out:
 693         return err;
 694 }
 695
 696 static int vxlan_process_vni_filter(struct vxlan_dev *vxlan,
 697                                     struct nlattr *nlvnifilter,
 698                                     int cmd, struct netlink_ext_ack *extack)
 699 {
 700         struct nlattr *vattrs[VXLAN_VNIFILTER_ENTRY_MAX + 1];
 701         u32 vni_start = 0, vni_end = 0;
 702         union vxlan_addr group;
 703         int err;
 704
 705         err = nla_parse_nested(vattrs,
 706                                VXLAN_VNIFILTER_ENTRY_MAX,
 707                                nlvnifilter, vni_filter_entry_policy,
 708                                extack);
 709         if (err)
 710                 return err;
 711
 712         if (vattrs[VXLAN_VNIFILTER_ENTRY_START]) {
 713                 vni_start = nla_get_u32(vattrs[VXLAN_VNIFILTER_ENTRY_START]);
 714                 vni_end = vni_start;
 715         }
 716
 717         if (vattrs[VXLAN_VNIFILTER_ENTRY_END])
 718                 vni_end = nla_get_u32(vattrs[VXLAN_VNIFILTER_ENTRY_END]);
 719
 720         if (!vni_start && !vni_end) {
 721                 NL_SET_ERR_MSG_ATTR(extack, nlvnifilter,
 722                                     "vni start nor end found in vni entry");
 723                 return -EINVAL;
 724         }
 725
 726         if (vattrs[VXLAN_VNIFILTER_ENTRY_GROUP]) {
 727                 group.sin.sin_addr.s_addr =
 728                         nla_get_in_addr(vattrs[VXLAN_VNIFILTER_ENTRY_GROUP]);
 729                 group.sa.sa_family = AF_INET;
 730         } else if (vattrs[VXLAN_VNIFILTER_ENTRY_GROUP6]) {
 731                 group.sin6.sin6_addr =
 732                         nla_get_in6_addr(vattrs[VXLAN_VNIFILTER_ENTRY_GROUP6]);
 733                 group.sa.sa_family = AF_INET6;
 734         } else {
 735                 memset(&group, 0, sizeof(group));
 736         }
 737
 738         if (vxlan_addr_multicast(&group) && !vxlan->default_dst.remote_ifindex) {
 739                 NL_SET_ERR_MSG(extack,
 740                                "Local interface required for multicast remote group");
 741
 742                 return -EINVAL;
 743         }
 744
 745         err = vxlan_vni_add_del(vxlan, vni_start, vni_end, &group, cmd,
 746                                 extack);
 747         if (err)
 748                 return err;
 749
 750         return 0;
 751 }
 752
 753 void vxlan_vnigroup_uninit(struct vxlan_dev *vxlan)
 754 {
 755         struct vxlan_vni_node *v, *tmp;
 756         struct vxlan_vni_group *vg;
 757
 758         vg = rtnl_dereference(vxlan->vnigrp);
 759         list_for_each_entry_safe(v, tmp, &vg->vni_list, vlist) {
 760                 rhashtable_remove_fast(&vg->vni_hash, &v->vnode,
 761                                        vxlan_vni_rht_params);
 762                 hlist_del_init_rcu(&v->hlist4.hlist);
 763 #if IS_ENABLED(CONFIG_IPV6)
 764                 hlist_del_init_rcu(&v->hlist6.hlist);
 765 #endif
 766                 __vxlan_vni_del_list(vg, v);
 767                 vxlan_vnifilter_notify(vxlan, v, RTM_DELTUNNEL);
 768                 call_rcu(&v->rcu, vxlan_vni_node_rcu_free);
 769         }
 770         rhashtable_destroy(&vg->vni_hash);
 771         kfree(vg);
 772 }
 773
 774 int vxlan_vnigroup_init(struct vxlan_dev *vxlan)
 775 {
 776         struct vxlan_vni_group *vg;
 777         int ret;
 778
 779         vg = kzalloc(sizeof(*vg), GFP_KERNEL);
 780         if (!vg)
 781                 return -ENOMEM;
 782         ret = rhashtable_init(&vg->vni_hash, &vxlan_vni_rht_params);
 783         if (ret) {
 784                 kfree(vg);
 785                 return ret;
 786         }
 787         INIT_LIST_HEAD(&vg->vni_list);
 788         rcu_assign_pointer(vxlan->vnigrp, vg);
 789
 790         return 0;
 791 }
 792
 793 static int vxlan_vnifilter_process(struct sk_buff *skb, struct nlmsghdr *nlh,
 794                                    struct netlink_ext_ack *extack)
 795 {
 796         struct net *net = sock_net(skb->sk);
 797         struct tunnel_msg *tmsg;
 798         struct vxlan_dev *vxlan;
 799         struct net_device *dev;
 800         struct nlattr *attr;
 801         int err, vnis = 0;
 802         int rem;
 803
 804         /* this should validate the header and check for remaining bytes */
 805         err = nlmsg_parse(nlh, sizeof(*tmsg), NULL, VXLAN_VNIFILTER_MAX,
 806                           vni_filter_policy, extack);
 807         if (err < 0)
 808                 return err;
 809
 810         tmsg = nlmsg_data(nlh);
 811         dev = __dev_get_by_index(net, tmsg->ifindex);
 812         if (!dev)
 813                 return -ENODEV;
 814
 815         if (!netif_is_vxlan(dev)) {
 816                 NL_SET_ERR_MSG_MOD(extack, "The device is not a vxlan device");
 817                 return -EINVAL;
 818         }
 819
 820         vxlan = netdev_priv(dev);
 821
 822         if (!(vxlan->cfg.flags & VXLAN_F_VNIFILTER))
 823                 return -EOPNOTSUPP;
 824
 825         nlmsg_for_each_attr(attr, nlh, sizeof(*tmsg), rem) {
 826                 switch (nla_type(attr)) {
 827                 case VXLAN_VNIFILTER_ENTRY:
 828                         err = vxlan_process_vni_filter(vxlan, attr,
 829                                                        nlh->nlmsg_type, extack);
 830                         break;
 831                 default:
 832                         continue;
 833                 }
 834                 vnis++;
 835                 if (err)
 836                         break;
 837         }
 838
 839         if (!vnis) {
 840                 NL_SET_ERR_MSG_MOD(extack, "No vnis found to process");
 841                 err = -EINVAL;
 842         }
 843
 844         return err;
 845 }
 846
 847 void vxlan_vnifilter_init(void)
 848 {
 849         rtnl_register_module(THIS_MODULE, PF_BRIDGE, RTM_GETTUNNEL, NULL,
 850                              vxlan_vnifilter_dump, 0);
 851         rtnl_register_module(THIS_MODULE, PF_BRIDGE, RTM_NEWTUNNEL,
 852                              vxlan_vnifilter_process, NULL, 0);
 853         rtnl_register_module(THIS_MODULE, PF_BRIDGE, RTM_DELTUNNEL,
 854                              vxlan_vnifilter_process, NULL, 0);
 855 }
 856
 857 void vxlan_vnifilter_uninit(void)
 858 {
 859         rtnl_unregister(PF_BRIDGE, RTM_GETTUNNEL);
 860         rtnl_unregister(PF_BRIDGE, RTM_NEWTUNNEL);
 861         rtnl_unregister(PF_BRIDGE, RTM_DELTUNNEL);
 862 }