sys/net/ipfw/ip_fw2.h

   1 /*
   2  * Copyright (c) 2002 Luigi Rizzo, Universita` di Pisa
   3  *
   4  * Redistribution and use in source and binary forms, with or without
   5  * modification, are permitted provided that the following conditions
   6  * are met:
   7  * 1. Redistributions of source code must retain the above copyright
   8  *    notice, this list of conditions and the following disclaimer.
   9  * 2. Redistributions in binary form must reproduce the above copyright
  10  *    notice, this list of conditions and the following disclaimer in the
  11  *    documentation and/or other materials provided with the distribution.
  12  *
  13  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
  14  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  15  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  16  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
  17  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  18  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  19  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  20  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  21  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  22  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  23  * SUCH DAMAGE.
  24  *
  25  * $FreeBSD: src/sys/netinet/ip_fw2.h,v 1.1.2.2 2002/08/16 11:03:11 luigi Exp $
  26  * $DragonFly: src/sys/net/ipfw/ip_fw2.h,v 1.17 2008/09/26 12:12:36 sephe Exp $
  27  */
  28
  29 #ifndef _IPFW2_H
  30 #define _IPFW2_H
  31
  32 /*
  33  * The kernel representation of ipfw rules is made of a list of
  34  * 'instructions' (for all practical purposes equivalent to BPF
  35  * instructions), which specify which fields of the packet
  36  * (or its metatada) should be analysed.
  37  *
  38  * Each instruction is stored in a structure which begins with
  39  * "ipfw_insn", and can contain extra fields depending on the
  40  * instruction type (listed below).
  41  *
  42  * "enum ipfw_opcodes" are the opcodes supported. We can have up
  43  * to 256 different opcodes.
  44  */
  45
  46 enum ipfw_opcodes {             /* arguments (4 byte each)      */
  47         O_NOP,
  48
  49         O_IP_SRC,               /* u32 = IP                     */
  50         O_IP_SRC_MASK,          /* ip = IP/mask                 */
  51         O_IP_SRC_ME,            /* none                         */
  52         O_IP_SRC_SET,           /* u32=base, arg1=len, bitmap   */
  53
  54         O_IP_DST,               /* u32 = IP                     */
  55         O_IP_DST_MASK,          /* ip = IP/mask                 */
  56         O_IP_DST_ME,            /* none                         */
  57         O_IP_DST_SET,           /* u32=base, arg1=len, bitmap   */
  58
  59         O_IP_SRCPORT,           /* (n)port list:mask 4 byte ea  */
  60         O_IP_DSTPORT,           /* (n)port list:mask 4 byte ea  */
  61         O_PROTO,                /* arg1=protocol                */
  62
  63         O_MACADDR2,             /* 2 mac addr:mask              */
  64         O_MAC_TYPE,             /* same as srcport              */
  65
  66         O_LAYER2,               /* none                         */
  67         O_IN,                   /* none                         */
  68         O_FRAG,                 /* none                         */
  69
  70         O_RECV,                 /* none                         */
  71         O_XMIT,                 /* none                         */
  72         O_VIA,                  /* none                         */
  73
  74         O_IPOPT,                /* arg1 = 2*u8 bitmap           */
  75         O_IPLEN,                /* arg1 = len                   */
  76         O_IPID,                 /* arg1 = id                    */
  77
  78         O_IPTOS,                /* arg1 = id                    */
  79         O_IPPRECEDENCE,         /* arg1 = precedence << 5       */
  80         O_IPTTL,                /* arg1 = TTL                   */
  81
  82         O_IPVER,                /* arg1 = version               */
  83         O_UID,                  /* u32 = id                     */
  84         O_GID,                  /* u32 = id                     */
  85         O_ESTAB,                /* none (tcp established)       */
  86         O_TCPFLAGS,             /* arg1 = 2*u8 bitmap           */
  87         O_TCPWIN,               /* arg1 = desired win           */
  88         O_TCPSEQ,               /* u32 = desired seq.           */
  89         O_TCPACK,               /* u32 = desired seq.           */
  90         O_ICMPTYPE,             /* u32 = icmp bitmap            */
  91         O_TCPOPTS,              /* arg1 = 2*u8 bitmap           */
  92
  93         O_PROBE_STATE,          /* none                         */
  94         O_KEEP_STATE,           /* none                         */
  95         O_LIMIT,                /* ipfw_insn_limit              */
  96         O_LIMIT_PARENT,         /* dyn_type, not an opcode.     */
  97         /*
  98          * these are really 'actions', and must be last in the list.
  99          */
 100
 101         O_LOG,                  /* ipfw_insn_log                */
 102         O_PROB,                 /* u32 = match probability      */
 103
 104         O_CHECK_STATE,          /* none                         */
 105         O_ACCEPT,               /* none                         */
 106         O_DENY,                 /* none                         */
 107         O_REJECT,               /* arg1=icmp arg (same as deny) */
 108         O_COUNT,                /* none                         */
 109         O_SKIPTO,               /* arg1=next rule number        */
 110         O_PIPE,                 /* arg1=pipe number             */
 111         O_QUEUE,                /* arg1=queue number            */
 112         O_DIVERT,               /* arg1=port number             */
 113         O_TEE,                  /* arg1=port number             */
 114         O_FORWARD_IP,           /* fwd sockaddr                 */
 115         O_FORWARD_MAC,          /* fwd mac                      */
 116         O_LAST_OPCODE           /* not an opcode!               */
 117 };
 118
 119 /*
 120  * Template for instructions.
 121  *
 122  * ipfw_insn is used for all instructions which require no operands,
 123  * a single 16-bit value (arg1), or a couple of 8-bit values.
 124  *
 125  * For other instructions which require different/larger arguments
 126  * we have derived structures, ipfw_insn_*.
 127  *
 128  * The size of the instruction (in 32-bit words) is in the low
 129  * 6 bits of "len". The 2 remaining bits are used to implement
 130  * NOT and OR on individual instructions. Given a type, you can
 131  * compute the length to be put in "len" using F_INSN_SIZE(t)
 132  *
 133  * F_NOT        negates the match result of the instruction.
 134  *
 135  * F_OR         is used to build or blocks. By default, instructions
 136  *              are evaluated as part of a logical AND. An "or" block
 137  *              { X or Y or Z } contains F_OR set in all but the last
 138  *              instruction of the block. A match will cause the code
 139  *              to skip past the last instruction of the block.
 140  *
 141  * NOTA BENE: in a couple of places we assume that
 142  *      sizeof(ipfw_insn) == sizeof(uint32_t)
 143  * this needs to be fixed.
 144  *
 145  */
 146 typedef struct  _ipfw_insn {    /* template for instructions */
 147         enum ipfw_opcodes       opcode:8;
 148         uint8_t         len;    /* numer of 32-byte words */
 149 #define F_NOT           0x80
 150 #define F_OR            0x40
 151 #define F_LEN_MASK      0x3f
 152 #define F_LEN(cmd)      ((cmd)->len & F_LEN_MASK)
 153
 154         uint16_t        arg1;
 155 } ipfw_insn;
 156
 157 /*
 158  * The F_INSN_SIZE(type) computes the size, in 4-byte words, of
 159  * a given type.
 160  */
 161 #define F_INSN_SIZE(t)  ((sizeof (t))/sizeof(uint32_t))
 162
 163 /*
 164  * This is used to store an array of 16-bit entries (ports etc.)
 165  */
 166 typedef struct  _ipfw_insn_u16 {
 167         ipfw_insn o;
 168         uint16_t ports[2];      /* there may be more */
 169 } ipfw_insn_u16;
 170
 171 /*
 172  * This is used to store an array of 32-bit entries
 173  * (uid, single IPv4 addresses etc.)
 174  */
 175 typedef struct  _ipfw_insn_u32 {
 176         ipfw_insn o;
 177         uint32_t d[1];  /* one or more */
 178 } ipfw_insn_u32;
 179
 180 /*
 181  * This is used to store IP addr-mask pairs.
 182  */
 183 typedef struct  _ipfw_insn_ip {
 184         ipfw_insn o;
 185         struct in_addr  addr;
 186         struct in_addr  mask;
 187 } ipfw_insn_ip;
 188
 189 /*
 190  * This is used to forward to a given address (ip)
 191  */
 192 typedef struct  _ipfw_insn_sa {
 193         ipfw_insn o;
 194         struct sockaddr_in sa;
 195 } ipfw_insn_sa;
 196
 197 /*
 198  * This is used for MAC addr-mask pairs.
 199  */
 200 typedef struct  _ipfw_insn_mac {
 201         ipfw_insn o;
 202         u_char addr[12];        /* dst[6] + src[6] */
 203         u_char mask[12];        /* dst[6] + src[6] */
 204 } ipfw_insn_mac;
 205
 206 /*
 207  * This is used for interface match rules (recv xx, xmit xx)
 208  */
 209 typedef struct  _ipfw_insn_if {
 210         ipfw_insn o;
 211         union {
 212                 struct in_addr ip;
 213                 int glob;
 214         } p;
 215         char name[IFNAMSIZ];
 216 } ipfw_insn_if;
 217
 218 /*
 219  * This is used for pipe and queue actions, which need to store
 220  * a single pointer (which can have different size on different
 221  * architectures.
 222  */
 223 typedef struct  _ipfw_insn_pipe {
 224         ipfw_insn       o;
 225         void            *pipe_ptr;
 226 } ipfw_insn_pipe;
 227
 228 /*
 229  * This is used for limit rules.
 230  */
 231 typedef struct  _ipfw_insn_limit {
 232         ipfw_insn o;
 233         uint8_t _pad;
 234         uint8_t limit_mask;     /* combination of DYN_* below   */
 235 #define DYN_SRC_ADDR    0x1
 236 #define DYN_SRC_PORT    0x2
 237 #define DYN_DST_ADDR    0x4
 238 #define DYN_DST_PORT    0x8
 239
 240         uint16_t conn_limit;
 241 } ipfw_insn_limit;
 242
 243 /*
 244  * This is used for log instructions
 245  */
 246 typedef struct  _ipfw_insn_log {
 247         ipfw_insn o;
 248         uint32_t max_log;       /* how many do we log -- 0 = all */
 249         uint32_t log_left;      /* how many left to log         */
 250 } ipfw_insn_log;
 251
 252 #ifdef _KERNEL
 253
 254 struct ip_fw;
 255 struct ip_fw_stub {
 256         struct ip_fw    *rule[1];
 257 };
 258
 259 /*
 260  * Here we have the structure representing an ipfw rule.
 261  *
 262  * It starts with a general area (with link fields and counters)
 263  * followed by an array of one or more instructions, which the code
 264  * accesses as an array of 32-bit values.
 265  *
 266  * Given a rule pointer  r:
 267  *
 268  *  r->cmd              is the start of the first instruction.
 269  *  ACTION_PTR(r)       is the start of the first action (things to do
 270  *                      once a rule matched).
 271  *
 272  * When assembling instruction, remember the following:
 273  *
 274  *  + if a rule has a "keep-state" (or "limit") option, then the
 275  *      first instruction (at r->cmd) MUST BE an O_PROBE_STATE
 276  *  + if a rule has a "log" option, then the first action
 277  *      (at ACTION_PTR(r)) MUST be O_LOG
 278  *
 279  * NOTE: we use a simple linked list of rules because we never need
 280  *      to delete a rule without scanning the list. We do not use
 281  *      queue(3) macros for portability and readability.
 282  */
 283
 284 struct ip_fw {
 285         struct ip_fw    *next;          /* linked list of rules         */
 286         struct ip_fw    *next_rule;     /* ptr to next [skipto] rule    */
 287         uint16_t        act_ofs;        /* offset of action in 32-bit units */
 288         uint16_t        cmd_len;        /* # of 32-bit words in cmd     */
 289         uint16_t        rulenum;        /* rule number                  */
 290         uint8_t         set;            /* rule set (0..31)             */
 291         uint8_t         usr_flags;      /* IPFW_USR_F_                  */
 292
 293         /* These fields are present in all rules.                       */
 294         uint64_t        pcnt;           /* Packet counter               */
 295         uint64_t        bcnt;           /* Byte counter                 */
 296         uint32_t        timestamp;      /* tv_sec of last match         */
 297
 298         struct ip_fw_stub *stub;        /* back pointers to clones      */
 299         struct ip_fw    *sibling;       /* clone on next cpu            */
 300         int             cpuid;          /* owner cpu                    */
 301
 302         uint32_t        refcnt;         /* Ref count for transit pkts   */
 303         uint32_t        rule_flags;     /* IPFW_RULE_F_                 */
 304
 305         ipfw_insn       cmd[1];         /* storage for commands         */
 306 };
 307
 308 #define IPFW_RULE_F_INVALID     0x1
 309 #define IPFW_RULE_F_STATE       0x2
 310
 311 #define RULESIZE(rule)  (sizeof(struct ip_fw) + (rule)->cmd_len * 4 - 4)
 312
 313 /*
 314  * This structure is used as a flow mask and a flow id for various
 315  * parts of the code.
 316  */
 317 struct ipfw_flow_id {
 318         uint32_t        dst_ip;
 319         uint32_t        src_ip;
 320         uint16_t        dst_port;
 321         uint16_t        src_port;
 322         uint8_t         proto;
 323         uint8_t         flags;  /* protocol-specific flags */
 324 };
 325
 326 /*
 327  * dynamic ipfw rule
 328  */
 329 typedef struct _ipfw_dyn_rule ipfw_dyn_rule;
 330
 331 struct _ipfw_dyn_rule {
 332         ipfw_dyn_rule   *next;          /* linked list of rules.        */
 333         struct ipfw_flow_id id;         /* (masked) flow id             */
 334 #ifdef notyet
 335         struct ip_fw *rule;             /* pointer to rule              */
 336 #else
 337         const struct ip_fw_stub *stub;  /* pointer to rule's stub       */
 338 #endif
 339         ipfw_dyn_rule *parent;          /* pointer to parent rule       */
 340         uint32_t        expire;         /* expire time                  */
 341         uint64_t        pcnt;           /* packet match counter         */
 342         uint64_t        bcnt;           /* byte match counter           */
 343         uint32_t        bucket;         /* which bucket in hash table   */
 344         uint32_t        state;          /* state of this rule (typically a
 345                                          * combination of TCP flags)
 346                                          */
 347         uint32_t        ack_fwd;        /* most recent ACKs in forward  */
 348         uint32_t        ack_rev;        /* and reverse directions (used */
 349                                         /* to generate keepalives)      */
 350         uint16_t        dyn_type;       /* rule type                    */
 351         uint16_t        count;          /* refcount                     */
 352         time_t          keep_alive;     /* last keep-alive sending time */
 353 };
 354
 355 /*
 356  * Main firewall chains definitions and global var's definitions.
 357  */
 358
 359 /* ipfw_chk/ip_fw_chk_ptr return values */
 360 #define IP_FW_PASS      0
 361 #define IP_FW_DENY      1
 362 #define IP_FW_DIVERT    2
 363 #define IP_FW_TEE       3
 364 #define IP_FW_DUMMYNET  4
 365
 366 /*
 367  * arguments for calling ipfw_chk() and dummynet_io(). We put them
 368  * all into a structure because this way it is easier and more
 369  * efficient to pass variables around and extend the interface.
 370  */
 371 struct ip_fw_args {
 372         struct mbuf     *m;             /* the mbuf chain               */
 373         struct ifnet    *oif;           /* output interface             */
 374         struct ip_fw    *rule;          /* matching rule                */
 375         struct ether_header *eh;        /* for bridged packets          */
 376
 377         struct ipfw_flow_id f_id;       /* grabbed from IP header       */
 378
 379         /*
 380          * Depend on the return value of ipfw_chk/ip_fw_chk_ptr
 381          * 'cookie' field may save following information:
 382          *
 383          * IP_FW_TEE or IP_FW_DIVERT
 384          *   The divert port number
 385          *
 386          * IP_FW_DUMMYNET
 387          *   The pipe or queue number
 388          */
 389         uint32_t        cookie;
 390 };
 391
 392 /*
 393  * Function definitions.
 394  */
 395 int     ip_fw_sockopt(struct sockopt *);
 396
 397 /* Firewall hooks */
 398 struct sockopt;
 399 struct dn_flow_set;
 400
 401 typedef int     ip_fw_chk_t(struct ip_fw_args *);
 402 typedef int     ip_fw_ctl_t(struct sockopt *);
 403 typedef void    ip_fw_dn_io_t(struct mbuf *, int, int, struct ip_fw_args *);
 404
 405 extern ip_fw_chk_t      *ip_fw_chk_ptr;
 406 extern ip_fw_ctl_t      *ip_fw_ctl_ptr;
 407 extern ip_fw_dn_io_t    *ip_fw_dn_io_ptr;
 408
 409 extern int fw_one_pass;
 410 extern int fw_enable;
 411
 412 extern int ip_fw_loaded;
 413 #define IPFW_LOADED     (ip_fw_loaded)
 414
 415 #define IPFW_CFGCPUID   0
 416 #define IPFW_CFGPORT    cpu_portfn(IPFW_CFGCPUID)
 417 #define IPFW_ASSERT_CFGPORT(msgport)    \
 418         KASSERT((msgport) == IPFW_CFGPORT, ("not IPFW CFGPORT"))
 419
 420 #endif /* _KERNEL */
 421
 422 #define ACTION_PTR(rule)        \
 423         (ipfw_insn *)((uint32_t *)((rule)->cmd) + ((rule)->act_ofs))
 424
 425 struct ipfw_ioc_rule {
 426         uint16_t        act_ofs;        /* offset of action in 32-bit units */
 427         uint16_t        cmd_len;        /* # of 32-bit words in cmd     */
 428         uint16_t        rulenum;        /* rule number                  */
 429         uint8_t         set;            /* rule set (0..31)             */
 430         uint8_t         usr_flags;      /* IPFW_USR_F_                  */
 431
 432         /* Rule set information */
 433         uint32_t        set_disable;    /* disabled rule sets           */
 434         uint32_t        static_count;   /* # of static rules            */
 435         uint32_t        static_len;     /* total length of static rules */
 436
 437         /* Statistics */
 438         uint64_t        pcnt;           /* Packet counter               */
 439         uint64_t        bcnt;           /* Byte counter                 */
 440         uint32_t        timestamp;      /* tv_sec of last match         */
 441
 442         uint8_t         reserved[16];
 443
 444         ipfw_insn       cmd[1];         /* storage for commands         */
 445 };
 446
 447 #define IPFW_USR_F_NORULE       0x01
 448
 449 #define IPFW_RULE_SIZE_MAX      255     /* unit: uint32_t */
 450
 451 #define IOC_RULESIZE(rule)      \
 452         (sizeof(struct ipfw_ioc_rule) + (rule)->cmd_len * 4 - 4)
 453
 454 struct ipfw_ioc_flowid {
 455         uint16_t        type;   /* ETHERTYPE_ */
 456         uint16_t        pad;
 457         union {
 458                 struct {
 459                         uint32_t dst_ip;
 460                         uint32_t src_ip;
 461                         uint16_t dst_port;
 462                         uint16_t src_port;
 463                         uint8_t proto;
 464                 } ip;
 465                 uint8_t pad[64];
 466         } u;
 467 };
 468
 469 struct ipfw_ioc_state {
 470         uint32_t        expire;         /* expire time                  */
 471         uint64_t        pcnt;           /* packet match counter         */
 472         uint64_t        bcnt;           /* byte match counter           */
 473
 474         uint16_t        dyn_type;       /* rule type                    */
 475         uint16_t        count;          /* refcount                     */
 476
 477         uint16_t        rulenum;
 478         uint16_t        pad;
 479
 480         int             cpu;            /* reserved                     */
 481
 482         struct ipfw_ioc_flowid id;      /* (masked) flow id             */
 483         uint8_t         reserved[16];
 484 };
 485
 486 /*
 487  * Definitions for IP option names.
 488  */
 489 #define IP_FW_IPOPT_LSRR        0x01
 490 #define IP_FW_IPOPT_SSRR        0x02
 491 #define IP_FW_IPOPT_RR          0x04
 492 #define IP_FW_IPOPT_TS          0x08
 493
 494 /*
 495  * Definitions for TCP option names.
 496  */
 497 #define IP_FW_TCPOPT_MSS        0x01
 498 #define IP_FW_TCPOPT_WINDOW     0x02
 499 #define IP_FW_TCPOPT_SACK       0x04
 500 #define IP_FW_TCPOPT_TS         0x08
 501 #define IP_FW_TCPOPT_CC         0x10
 502
 503 #define ICMP_REJECT_RST         0x100   /* fake ICMP code (send a TCP RST) */
 504
 505 #endif /* _IPFW2_H */