LXR linux/include/linux/netfilter/ipset/ip

   1/* Copyright (C) 2000-2002 Joakim Axelsson <gozem@linux.nu>
   2 *                         Patrick Schaaf <bof@bof.de>
   3 *                         Martin Josefsson <gandalf@wlug.westbo.se>
   4 * Copyright (C) 2003-2013 Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>
   5 *
   6 * This program is free software; you can redistribute it and/or modify
   7 * it under the terms of the GNU General Public License version 2 as
   8 * published by the Free Software Foundation.
   9 */
  10#ifndef _IP_SET_H
  11#define _IP_SET_H
  12
  13#include <linux/ip.h>
  14#include <linux/ipv6.h>
  15#include <linux/netlink.h>
  16#include <linux/netfilter.h>
  17#include <linux/netfilter/x_tables.h>
  18#include <linux/stringify.h>
  19#include <linux/vmalloc.h>
  20#include <net/netlink.h>
  21#include <uapi/linux/netfilter/ipset/ip_set.h>
  22
  23#define _IP_SET_MODULE_DESC(a, b, c)            \
  24        MODULE_DESCRIPTION(a " type of IP sets, revisions " b "-" c)
  25#define IP_SET_MODULE_DESC(a, b, c)             \
  26        _IP_SET_MODULE_DESC(a, __stringify(b), __stringify(c))
  27
  28/* Set features */
  29enum ip_set_feature {
  30        IPSET_TYPE_IP_FLAG = 0,
  31        IPSET_TYPE_IP = (1 << IPSET_TYPE_IP_FLAG),
  32        IPSET_TYPE_PORT_FLAG = 1,
  33        IPSET_TYPE_PORT = (1 << IPSET_TYPE_PORT_FLAG),
  34        IPSET_TYPE_MAC_FLAG = 2,
  35        IPSET_TYPE_MAC = (1 << IPSET_TYPE_MAC_FLAG),
  36        IPSET_TYPE_IP2_FLAG = 3,
  37        IPSET_TYPE_IP2 = (1 << IPSET_TYPE_IP2_FLAG),
  38        IPSET_TYPE_NAME_FLAG = 4,
  39        IPSET_TYPE_NAME = (1 << IPSET_TYPE_NAME_FLAG),
  40        IPSET_TYPE_IFACE_FLAG = 5,
  41        IPSET_TYPE_IFACE = (1 << IPSET_TYPE_IFACE_FLAG),
  42        IPSET_TYPE_MARK_FLAG = 6,
  43        IPSET_TYPE_MARK = (1 << IPSET_TYPE_MARK_FLAG),
  44        IPSET_TYPE_NOMATCH_FLAG = 7,
  45        IPSET_TYPE_NOMATCH = (1 << IPSET_TYPE_NOMATCH_FLAG),
  46        /* Strictly speaking not a feature, but a flag for dumping:
  47         * this settype must be dumped last */
  48        IPSET_DUMP_LAST_FLAG = 8,
  49        IPSET_DUMP_LAST = (1 << IPSET_DUMP_LAST_FLAG),
  50};
  51
  52/* Set extensions */
  53enum ip_set_extension {
  54        IPSET_EXT_BIT_TIMEOUT = 0,
  55        IPSET_EXT_TIMEOUT = (1 << IPSET_EXT_BIT_TIMEOUT),
  56        IPSET_EXT_BIT_COUNTER = 1,
  57        IPSET_EXT_COUNTER = (1 << IPSET_EXT_BIT_COUNTER),
  58        IPSET_EXT_BIT_COMMENT = 2,
  59        IPSET_EXT_COMMENT = (1 << IPSET_EXT_BIT_COMMENT),
  60        IPSET_EXT_BIT_SKBINFO = 3,
  61        IPSET_EXT_SKBINFO = (1 << IPSET_EXT_BIT_SKBINFO),
  62        /* Mark set with an extension which needs to call destroy */
  63        IPSET_EXT_BIT_DESTROY = 7,
  64        IPSET_EXT_DESTROY = (1 << IPSET_EXT_BIT_DESTROY),
  65};
  66
  67#define SET_WITH_TIMEOUT(s)     ((s)->extensions & IPSET_EXT_TIMEOUT)
  68#define SET_WITH_COUNTER(s)     ((s)->extensions & IPSET_EXT_COUNTER)
  69#define SET_WITH_COMMENT(s)     ((s)->extensions & IPSET_EXT_COMMENT)
  70#define SET_WITH_SKBINFO(s)     ((s)->extensions & IPSET_EXT_SKBINFO)
  71#define SET_WITH_FORCEADD(s)    ((s)->flags & IPSET_CREATE_FLAG_FORCEADD)
  72
  73/* Extension id, in size order */
  74enum ip_set_ext_id {
  75        IPSET_EXT_ID_COUNTER = 0,
  76        IPSET_EXT_ID_TIMEOUT,
  77        IPSET_EXT_ID_SKBINFO,
  78        IPSET_EXT_ID_COMMENT,
  79        IPSET_EXT_ID_MAX,
  80};
  81
  82/* Extension type */
  83struct ip_set_ext_type {
  84        /* Destroy extension private data (can be NULL) */
  85        void (*destroy)(void *ext);
  86        enum ip_set_extension type;
  87        enum ipset_cadt_flags flag;
  88        /* Size and minimal alignment */
  89        u8 len;
  90        u8 align;
  91};
  92
  93extern const struct ip_set_ext_type ip_set_extensions[];
  94
  95struct ip_set_ext {
  96        u64 packets;
  97        u64 bytes;
  98        u32 timeout;
  99        u32 skbmark;
 100        u32 skbmarkmask;
 101        u32 skbprio;
 102        u16 skbqueue;
 103        char *comment;
 104};
 105
 106struct ip_set_counter {
 107        atomic64_t bytes;
 108        atomic64_t packets;
 109};
 110
 111struct ip_set_comment_rcu {
 112        struct rcu_head rcu;
 113        char str[0];
 114};
 115
 116struct ip_set_comment {
 117        struct ip_set_comment_rcu __rcu *c;
 118};
 119
 120struct ip_set_skbinfo {
 121        u32 skbmark;
 122        u32 skbmarkmask;
 123        u32 skbprio;
 124        u16 skbqueue;
 125};
 126
 127struct ip_set;
 128
 129#define ext_timeout(e, s)       \
 130((unsigned long *)(((void *)(e)) + (s)->offset[IPSET_EXT_ID_TIMEOUT]))
 131#define ext_counter(e, s)       \
 132((struct ip_set_counter *)(((void *)(e)) + (s)->offset[IPSET_EXT_ID_COUNTER]))
 133#define ext_comment(e, s)       \
 134((struct ip_set_comment *)(((void *)(e)) + (s)->offset[IPSET_EXT_ID_COMMENT]))
 135#define ext_skbinfo(e, s)       \
 136((struct ip_set_skbinfo *)(((void *)(e)) + (s)->offset[IPSET_EXT_ID_SKBINFO]))
 137
 138typedef int (*ipset_adtfn)(struct ip_set *set, void *value,
 139                           const struct ip_set_ext *ext,
 140                           struct ip_set_ext *mext, u32 cmdflags);
 141
 142/* Kernel API function options */
 143struct ip_set_adt_opt {
 144        u8 family;              /* Actual protocol family */
 145        u8 dim;                 /* Dimension of match/target */
 146        u8 flags;               /* Direction and negation flags */
 147        u32 cmdflags;           /* Command-like flags */
 148        struct ip_set_ext ext;  /* Extensions */
 149};
 150
 151/* Set type, variant-specific part */
 152struct ip_set_type_variant {
 153        /* Kernelspace: test/add/del entries
 154         *              returns negative error code,
 155         *                      zero for no match/success to add/delete
 156         *                      positive for matching element */
 157        int (*kadt)(struct ip_set *set, const struct sk_buff *skb,
 158                    const struct xt_action_param *par,
 159                    enum ipset_adt adt, struct ip_set_adt_opt *opt);
 160
 161        /* Userspace: test/add/del entries
 162         *              returns negative error code,
 163         *                      zero for no match/success to add/delete
 164         *                      positive for matching element */
 165        int (*uadt)(struct ip_set *set, struct nlattr *tb[],
 166                    enum ipset_adt adt, u32 *lineno, u32 flags, bool retried);
 167
 168        /* Low level add/del/test functions */
 169        ipset_adtfn adt[IPSET_ADT_MAX];
 170
 171        /* When adding entries and set is full, try to resize the set */
 172        int (*resize)(struct ip_set *set, bool retried);
 173        /* Destroy the set */
 174        void (*destroy)(struct ip_set *set);
 175        /* Flush the elements */
 176        void (*flush)(struct ip_set *set);
 177        /* Expire entries before listing */
 178        void (*expire)(struct ip_set *set);
 179        /* List set header data */
 180        int (*head)(struct ip_set *set, struct sk_buff *skb);
 181        /* List elements */
 182        int (*list)(const struct ip_set *set, struct sk_buff *skb,
 183                    struct netlink_callback *cb);
 184        /* Keep listing private when resizing runs parallel */
 185        void (*uref)(struct ip_set *set, struct netlink_callback *cb,
 186                     bool start);
 187
 188        /* Return true if "b" set is the same as "a"
 189         * according to the create set parameters */
 190        bool (*same_set)(const struct ip_set *a, const struct ip_set *b);
 191};
 192
 193/* The core set type structure */
 194struct ip_set_type {
 195        struct list_head list;
 196
 197        /* Typename */
 198        char name[IPSET_MAXNAMELEN];
 199        /* Protocol version */
 200        u8 protocol;
 201        /* Set type dimension */
 202        u8 dimension;
 203        /*
 204         * Supported family: may be NFPROTO_UNSPEC for both
 205         * NFPROTO_IPV4/NFPROTO_IPV6.
 206         */
 207        u8 family;
 208        /* Type revisions */
 209        u8 revision_min, revision_max;
 210        /* Set features to control swapping */
 211        u16 features;
 212
 213        /* Create set */
 214        int (*create)(struct net *net, struct ip_set *set,
 215                      struct nlattr *tb[], u32 flags);
 216
 217        /* Attribute policies */
 218        const struct nla_policy create_policy[IPSET_ATTR_CREATE_MAX + 1];
 219        const struct nla_policy adt_policy[IPSET_ATTR_ADT_MAX + 1];
 220
 221        /* Set this to THIS_MODULE if you are a module, otherwise NULL */
 222        struct module *me;
 223};
 224
 225/* register and unregister set type */
 226extern int ip_set_type_register(struct ip_set_type *set_type);
 227extern void ip_set_type_unregister(struct ip_set_type *set_type);
 228
 229/* A generic IP set */
 230struct ip_set {
 231        /* The name of the set */
 232        char name[IPSET_MAXNAMELEN];
 233        /* Lock protecting the set data */
 234        spinlock_t lock;
 235        /* References to the set */
 236        u32 ref;
 237        /* References to the set for netlink events like dump,
 238         * ref can be swapped out by ip_set_swap
 239         */
 240        u32 ref_netlink;
 241        /* The core set type */
 242        struct ip_set_type *type;
 243        /* The type variant doing the real job */
 244        const struct ip_set_type_variant *variant;
 245        /* The actual INET family of the set */
 246        u8 family;
 247        /* The type revision */
 248        u8 revision;
 249        /* Extensions */
 250        u8 extensions;
 251        /* Create flags */
 252        u8 flags;
 253        /* Default timeout value, if enabled */
 254        u32 timeout;
 255        /* Element data size */
 256        size_t dsize;
 257        /* Offsets to extensions in elements */
 258        size_t offset[IPSET_EXT_ID_MAX];
 259        /* The type specific data */
 260        void *data;
 261};
 262
 263static inline void
 264ip_set_ext_destroy(struct ip_set *set, void *data)
 265{
 266        /* Check that the extension is enabled for the set and
 267         * call it's destroy function for its extension part in data.
 268         */
 269        if (SET_WITH_COMMENT(set))
 270                ip_set_extensions[IPSET_EXT_ID_COMMENT].destroy(
 271                        ext_comment(data, set));
 272}
 273
 274static inline int
 275ip_set_put_flags(struct sk_buff *skb, struct ip_set *set)
 276{
 277        u32 cadt_flags = 0;
 278
 279        if (SET_WITH_TIMEOUT(set))
 280                if (unlikely(nla_put_net32(skb, IPSET_ATTR_TIMEOUT,
 281                                           htonl(set->timeout))))
 282                        return -EMSGSIZE;
 283        if (SET_WITH_COUNTER(set))
 284                cadt_flags |= IPSET_FLAG_WITH_COUNTERS;
 285        if (SET_WITH_COMMENT(set))
 286                cadt_flags |= IPSET_FLAG_WITH_COMMENT;
 287        if (SET_WITH_SKBINFO(set))
 288                cadt_flags |= IPSET_FLAG_WITH_SKBINFO;
 289        if (SET_WITH_FORCEADD(set))
 290                cadt_flags |= IPSET_FLAG_WITH_FORCEADD;
 291
 292        if (!cadt_flags)
 293                return 0;
 294        return nla_put_net32(skb, IPSET_ATTR_CADT_FLAGS, htonl(cadt_flags));
 295}
 296
 297static inline void
 298ip_set_add_bytes(u64 bytes, struct ip_set_counter *counter)
 299{
 300        atomic64_add((long long)bytes, &(counter)->bytes);
 301}
 302
 303static inline void
 304ip_set_add_packets(u64 packets, struct ip_set_counter *counter)
 305{
 306        atomic64_add((long long)packets, &(counter)->packets);
 307}
 308
 309static inline u64
 310ip_set_get_bytes(const struct ip_set_counter *counter)
 311{
 312        return (u64)atomic64_read(&(counter)->bytes);
 313}
 314
 315static inline u64
 316ip_set_get_packets(const struct ip_set_counter *counter)
 317{
 318        return (u64)atomic64_read(&(counter)->packets);
 319}
 320
 321static inline void
 322ip_set_update_counter(struct ip_set_counter *counter,
 323                      const struct ip_set_ext *ext,
 324                      struct ip_set_ext *mext, u32 flags)
 325{
 326        if (ext->packets != ULLONG_MAX &&
 327            !(flags & IPSET_FLAG_SKIP_COUNTER_UPDATE)) {
 328                ip_set_add_bytes(ext->bytes, counter);
 329                ip_set_add_packets(ext->packets, counter);
 330        }
 331        if (flags & IPSET_FLAG_MATCH_COUNTERS) {
 332                mext->packets = ip_set_get_packets(counter);
 333                mext->bytes = ip_set_get_bytes(counter);
 334        }
 335}
 336
 337static inline void
 338ip_set_get_skbinfo(struct ip_set_skbinfo *skbinfo,
 339                      const struct ip_set_ext *ext,
 340                      struct ip_set_ext *mext, u32 flags)
 341{
 342                mext->skbmark = skbinfo->skbmark;
 343                mext->skbmarkmask = skbinfo->skbmarkmask;
 344                mext->skbprio = skbinfo->skbprio;
 345                mext->skbqueue = skbinfo->skbqueue;
 346}
 347static inline bool
 348ip_set_put_skbinfo(struct sk_buff *skb, struct ip_set_skbinfo *skbinfo)
 349{
 350        /* Send nonzero parameters only */
 351        return ((skbinfo->skbmark || skbinfo->skbmarkmask) &&
 352                nla_put_net64(skb, IPSET_ATTR_SKBMARK,
 353                              cpu_to_be64((u64)skbinfo->skbmark << 32 |
 354                                          skbinfo->skbmarkmask))) ||
 355               (skbinfo->skbprio &&
 356                nla_put_net32(skb, IPSET_ATTR_SKBPRIO,
 357                              cpu_to_be32(skbinfo->skbprio))) ||
 358               (skbinfo->skbqueue &&
 359                nla_put_net16(skb, IPSET_ATTR_SKBQUEUE,
 360                             cpu_to_be16(skbinfo->skbqueue)));
 361}
 362
 363static inline void
 364ip_set_init_skbinfo(struct ip_set_skbinfo *skbinfo,
 365                    const struct ip_set_ext *ext)
 366{
 367        skbinfo->skbmark = ext->skbmark;
 368        skbinfo->skbmarkmask = ext->skbmarkmask;
 369        skbinfo->skbprio = ext->skbprio;
 370        skbinfo->skbqueue = ext->skbqueue;
 371}
 372
 373static inline bool
 374ip_set_put_counter(struct sk_buff *skb, struct ip_set_counter *counter)
 375{
 376        return nla_put_net64(skb, IPSET_ATTR_BYTES,
 377                             cpu_to_be64(ip_set_get_bytes(counter))) ||
 378               nla_put_net64(skb, IPSET_ATTR_PACKETS,
 379                             cpu_to_be64(ip_set_get_packets(counter)));
 380}
 381
 382static inline void
 383ip_set_init_counter(struct ip_set_counter *counter,
 384                    const struct ip_set_ext *ext)
 385{
 386        if (ext->bytes != ULLONG_MAX)
 387                atomic64_set(&(counter)->bytes, (long long)(ext->bytes));
 388        if (ext->packets != ULLONG_MAX)
 389                atomic64_set(&(counter)->packets, (long long)(ext->packets));
 390}
 391
 392/* Netlink CB args */
 393enum {
 394        IPSET_CB_NET = 0,       /* net namespace */
 395        IPSET_CB_DUMP,          /* dump single set/all sets */
 396        IPSET_CB_INDEX,         /* set index */
 397        IPSET_CB_PRIVATE,       /* set private data */
 398        IPSET_CB_ARG0,          /* type specific */
 399        IPSET_CB_ARG1,
 400};
 401
 402/* register and unregister set references */
 403extern ip_set_id_t ip_set_get_byname(struct net *net,
 404                                     const char *name, struct ip_set **set);
 405extern void ip_set_put_byindex(struct net *net, ip_set_id_t index);
 406extern const char *ip_set_name_byindex(struct net *net, ip_set_id_t index);
 407extern ip_set_id_t ip_set_nfnl_get_byindex(struct net *net, ip_set_id_t index);
 408extern void ip_set_nfnl_put(struct net *net, ip_set_id_t index);
 409
 410/* API for iptables set match, and SET target */
 411
 412extern int ip_set_add(ip_set_id_t id, const struct sk_buff *skb,
 413                      const struct xt_action_param *par,
 414                      struct ip_set_adt_opt *opt);
 415extern int ip_set_del(ip_set_id_t id, const struct sk_buff *skb,
 416                      const struct xt_action_param *par,
 417                      struct ip_set_adt_opt *opt);
 418extern int ip_set_test(ip_set_id_t id, const struct sk_buff *skb,
 419                       const struct xt_action_param *par,
 420                       struct ip_set_adt_opt *opt);
 421
 422/* Utility functions */
 423extern void *ip_set_alloc(size_t size);
 424extern void ip_set_free(void *members);
 425extern int ip_set_get_ipaddr4(struct nlattr *nla,  __be32 *ipaddr);
 426extern int ip_set_get_ipaddr6(struct nlattr *nla, union nf_inet_addr *ipaddr);
 427extern size_t ip_set_elem_len(struct ip_set *set, struct nlattr *tb[],
 428                              size_t len, size_t align);
 429extern int ip_set_get_extensions(struct ip_set *set, struct nlattr *tb[],
 430                                 struct ip_set_ext *ext);
 431
 432static inline int
 433ip_set_get_hostipaddr4(struct nlattr *nla, u32 *ipaddr)
 434{
 435        __be32 ip;
 436        int ret = ip_set_get_ipaddr4(nla, &ip);
 437
 438        if (ret)
 439                return ret;
 440        *ipaddr = ntohl(ip);
 441        return 0;
 442}
 443
 444/* Ignore IPSET_ERR_EXIST errors if asked to do so? */
 445static inline bool
 446ip_set_eexist(int ret, u32 flags)
 447{
 448        return ret == -IPSET_ERR_EXIST && (flags & IPSET_FLAG_EXIST);
 449}
 450
 451/* Match elements marked with nomatch */
 452static inline bool
 453ip_set_enomatch(int ret, u32 flags, enum ipset_adt adt, struct ip_set *set)
 454{
 455        return adt == IPSET_TEST &&
 456               (set->type->features & IPSET_TYPE_NOMATCH) &&
 457               ((flags >> 16) & IPSET_FLAG_NOMATCH) &&
 458               (ret > 0 || ret == -ENOTEMPTY);
 459}
 460
 461/* Check the NLA_F_NET_BYTEORDER flag */
 462static inline bool
 463ip_set_attr_netorder(struct nlattr *tb[], int type)
 464{
 465        return tb[type] && (tb[type]->nla_type & NLA_F_NET_BYTEORDER);
 466}
 467
 468static inline bool
 469ip_set_optattr_netorder(struct nlattr *tb[], int type)
 470{
 471        return !tb[type] || (tb[type]->nla_type & NLA_F_NET_BYTEORDER);
 472}
 473
 474/* Useful converters */
 475static inline u32
 476ip_set_get_h32(const struct nlattr *attr)
 477{
 478        return ntohl(nla_get_be32(attr));
 479}
 480
 481static inline u16
 482ip_set_get_h16(const struct nlattr *attr)
 483{
 484        return ntohs(nla_get_be16(attr));
 485}
 486
 487#define ipset_nest_start(skb, attr) nla_nest_start(skb, attr | NLA_F_NESTED)
 488#define ipset_nest_end(skb, start)  nla_nest_end(skb, start)
 489
 490static inline int nla_put_ipaddr4(struct sk_buff *skb, int type, __be32 ipaddr)
 491{
 492        struct nlattr *__nested = ipset_nest_start(skb, type);
 493        int ret;
 494
 495        if (!__nested)
 496                return -EMSGSIZE;
 497        ret = nla_put_in_addr(skb, IPSET_ATTR_IPADDR_IPV4, ipaddr);
 498        if (!ret)
 499                ipset_nest_end(skb, __nested);
 500        return ret;
 501}
 502
 503static inline int nla_put_ipaddr6(struct sk_buff *skb, int type,
 504                                  const struct in6_addr *ipaddrptr)
 505{
 506        struct nlattr *__nested = ipset_nest_start(skb, type);
 507        int ret;
 508
 509        if (!__nested)
 510                return -EMSGSIZE;
 511        ret = nla_put_in6_addr(skb, IPSET_ATTR_IPADDR_IPV6, ipaddrptr);
 512        if (!ret)
 513                ipset_nest_end(skb, __nested);
 514        return ret;
 515}
 516
 517/* Get address from skbuff */
 518static inline __be32
 519ip4addr(const struct sk_buff *skb, bool src)
 520{
 521        return src ? ip_hdr(skb)->saddr : ip_hdr(skb)->daddr;
 522}
 523
 524static inline void
 525ip4addrptr(const struct sk_buff *skb, bool src, __be32 *addr)
 526{
 527        *addr = src ? ip_hdr(skb)->saddr : ip_hdr(skb)->daddr;
 528}
 529
 530static inline void
 531ip6addrptr(const struct sk_buff *skb, bool src, struct in6_addr *addr)
 532{
 533        memcpy(addr, src ? &ipv6_hdr(skb)->saddr : &ipv6_hdr(skb)->daddr,
 534               sizeof(*addr));
 535}
 536
 537/* Calculate the bytes required to store the inclusive range of a-b */
 538static inline int
 539bitmap_bytes(u32 a, u32 b)
 540{
 541        return 4 * ((((b - a + 8) / 8) + 3) / 4);
 542}
 543
 544#include <linux/netfilter/ipset/ip_set_timeout.h>
 545#include <linux/netfilter/ipset/ip_set_comment.h>
 546
 547int
 548ip_set_put_extensions(struct sk_buff *skb, const struct ip_set *set,
 549                      const void *e, bool active);
 550
 551#define IP_SET_INIT_KEXT(skb, opt, set)                 \
 552        { .bytes = (skb)->len, .packets = 1,            \
 553          .timeout = ip_set_adt_opt_timeout(opt, set) }
 554
 555#define IP_SET_INIT_UEXT(set)                           \
 556        { .bytes = ULLONG_MAX, .packets = ULLONG_MAX,   \
 557          .timeout = (set)->timeout }
 558
 559#define IPSET_CONCAT(a, b)              a##b
 560#define IPSET_TOKEN(a, b)               IPSET_CONCAT(a, b)
 561
 562#endif /*_IP_SET_H */
 563