bsd/netinet/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 1.1.2.4 2003/07/17 06:03:39 luigi Exp $
  26  */
  27
  28 #ifndef _IPFW2_H
  29 #define _IPFW2_H
  30
  31 /*
  32  * The kernel representation of ipfw rules is made of a list of
  33  * 'instructions' (for all practical purposes equivalent to BPF
  34  * instructions), which specify which fields of the packet
  35  * (or its metadata) should be analysed.
  36  *
  37  * Each instruction is stored in a structure which begins with
  38  * "ipfw_insn", and can contain extra fields depending on the
  39  * instruction type (listed below).
  40  * Note that the code is written so that individual instructions
  41  * have a size which is a multiple of 32 bits. This means that, if
  42  * such structures contain pointers or other 64-bit entities,
  43  * (there is just one instance now) they may end up unaligned on
  44  * 64-bit architectures, so the must be handled with care.
  45  *
  46  * "enum ipfw_opcodes" are the opcodes supported. We can have up
  47  * to 256 different opcodes.
  48  */
  49
  50 enum ipfw_opcodes {             /* arguments (4 byte each)      */
  51         O_NOP,
  52
  53         O_IP_SRC,               /* u32 = IP                     */
  54         O_IP_SRC_MASK,          /* ip = IP/mask                 */
  55         O_IP_SRC_ME,            /* none                         */
  56         O_IP_SRC_SET,           /* u32=base, arg1=len, bitmap   */
  57
  58         O_IP_DST,               /* u32 = IP                     */
  59         O_IP_DST_MASK,          /* ip = IP/mask                 */
  60         O_IP_DST_ME,            /* none                         */
  61         O_IP_DST_SET,           /* u32=base, arg1=len, bitmap   */
  62
  63         O_IP_SRCPORT,           /* (n)port list:mask 4 byte ea  */
  64         O_IP_DSTPORT,           /* (n)port list:mask 4 byte ea  */
  65         O_PROTO,                /* arg1=protocol                */
  66
  67         O_MACADDR2,             /* 2 mac addr:mask              */
  68         O_MAC_TYPE,             /* same as srcport              */
  69
  70         O_LAYER2,               /* none                         */
  71         O_IN,                   /* none                         */
  72         O_FRAG,                 /* none                         */
  73
  74         O_RECV,                 /* none                         */
  75         O_XMIT,                 /* none                         */
  76         O_VIA,                  /* none                         */
  77
  78         O_IPOPT,                /* arg1 = 2*u8 bitmap           */
  79         O_IPLEN,                /* arg1 = len                   */
  80         O_IPID,                 /* arg1 = id                    */
  81
  82         O_IPTOS,                /* arg1 = id                    */
  83         O_IPPRECEDENCE,         /* arg1 = precedence << 5       */
  84         O_IPTTL,                /* arg1 = TTL                   */
  85
  86         O_IPVER,                /* arg1 = version               */
  87         O_UID,                  /* u32 = id                     */
  88         O_GID,                  /* u32 = id                     */
  89         O_ESTAB,                /* none (tcp established)       */
  90         O_TCPFLAGS,             /* arg1 = 2*u8 bitmap           */
  91         O_TCPWIN,               /* arg1 = desired win           */
  92         O_TCPSEQ,               /* u32 = desired seq.           */
  93         O_TCPACK,               /* u32 = desired seq.           */
  94         O_ICMPTYPE,             /* u32 = icmp bitmap            */
  95         O_TCPOPTS,              /* arg1 = 2*u8 bitmap           */
  96
  97         O_VERREVPATH,           /* none                         */
  98
  99         O_PROBE_STATE,          /* none                         */
 100         O_KEEP_STATE,           /* none                         */
 101         O_LIMIT,                /* ipfw_insn_limit              */
 102         O_LIMIT_PARENT,         /* dyn_type, not an opcode.     */
 103
 104         /*
 105          * These are really 'actions'.
 106          */
 107
 108         O_LOG,                  /* ipfw_insn_log                */
 109         O_PROB,                 /* u32 = match probability      */
 110
 111         O_CHECK_STATE,          /* none                         */
 112         O_ACCEPT,               /* none                         */
 113         O_DENY,                 /* none                         */
 114         O_REJECT,               /* arg1=icmp arg (same as deny) */
 115         O_COUNT,                /* none                         */
 116         O_SKIPTO,               /* arg1=next rule number        */
 117         O_PIPE,                 /* arg1=pipe number             */
 118         O_QUEUE,                /* arg1=queue number            */
 119         O_DIVERT,               /* arg1=port number             */
 120         O_TEE,                  /* arg1=port number             */
 121         O_FORWARD_IP,           /* fwd sockaddr                 */
 122         O_FORWARD_MAC,          /* fwd mac                      */
 123
 124         /*
 125          * More opcodes.
 126          */
 127         O_IPSEC,                /* has ipsec history            */
 128
 129         O_LAST_OPCODE           /* not an opcode!               */
 130 };
 131
 132 /*
 133  * Template for instructions.
 134  *
 135  * ipfw_insn is used for all instructions which require no operands,
 136  * a single 16-bit value (arg1), or a couple of 8-bit values.
 137  *
 138  * For other instructions which require different/larger arguments
 139  * we have derived structures, ipfw_insn_*.
 140  *
 141  * The size of the instruction (in 32-bit words) is in the low
 142  * 6 bits of "len". The 2 remaining bits are used to implement
 143  * NOT and OR on individual instructions. Given a type, you can
 144  * compute the length to be put in "len" using F_INSN_SIZE(t)
 145  *
 146  * F_NOT        negates the match result of the instruction.
 147  *
 148  * F_OR         is used to build or blocks. By default, instructions
 149  *              are evaluated as part of a logical AND. An "or" block
 150  *              { X or Y or Z } contains F_OR set in all but the last
 151  *              instruction of the block. A match will cause the code
 152  *              to skip past the last instruction of the block.
 153  *
 154  * NOTA BENE: in a couple of places we assume that
 155  *      sizeof(ipfw_insn) == sizeof(u_int32_t)
 156  * this needs to be fixed.
 157  *
 158  */
 159 typedef struct  _ipfw_insn {    /* template for instructions */
 160         enum ipfw_opcodes       opcode:8;
 161         u_int8_t        len;    /* numer of 32-byte words */
 162 #define F_NOT           0x80
 163 #define F_OR            0x40
 164 #define F_LEN_MASK      0x3f
 165 #define F_LEN(cmd)      ((cmd)->len & F_LEN_MASK)
 166
 167         u_int16_t       arg1;
 168 } ipfw_insn;
 169
 170 /*
 171  * The F_INSN_SIZE(type) computes the size, in 4-byte words, of
 172  * a given type.
 173  */
 174 #define F_INSN_SIZE(t)  ((sizeof (t))/sizeof(u_int32_t))
 175
 176 /*
 177  * This is used to store an array of 16-bit entries (ports etc.)
 178  */
 179 typedef struct  _ipfw_insn_u16 {
 180         ipfw_insn o;
 181         u_int16_t ports[2];     /* there may be more */
 182 } ipfw_insn_u16;
 183
 184 /*
 185  * This is used to store an array of 32-bit entries
 186  * (uid, single IPv4 addresses etc.)
 187  */
 188 typedef struct  _ipfw_insn_u32 {
 189         ipfw_insn o;
 190         u_int32_t d[1]; /* one or more */
 191 } ipfw_insn_u32;
 192
 193 /*
 194  * This is used to store IP addr-mask pairs.
 195  */
 196 typedef struct  _ipfw_insn_ip {
 197         ipfw_insn o;
 198         struct in_addr  addr;
 199         struct in_addr  mask;
 200 } ipfw_insn_ip;
 201
 202 /*
 203  * This is used to forward to a given address (ip).
 204  */
 205 typedef struct  _ipfw_insn_sa {
 206         ipfw_insn o;
 207         struct sockaddr_in sa;
 208 } ipfw_insn_sa;
 209
 210 /*
 211  * This is used for MAC addr-mask pairs.
 212  */
 213 typedef struct  _ipfw_insn_mac {
 214         ipfw_insn o;
 215         u_char addr[12];        /* dst[6] + src[6] */
 216         u_char mask[12];        /* dst[6] + src[6] */
 217 } ipfw_insn_mac;
 218
 219 /*
 220  * This is used for interface match rules (recv xx, xmit xx).
 221  */
 222 typedef struct  _ipfw_insn_if {
 223         ipfw_insn o;
 224         union {
 225                 struct in_addr ip;
 226                 int32_t unit;
 227         } p;
 228         char name[IFNAMSIZ];
 229 } ipfw_insn_if;
 230
 231 /*
 232  * This is used for pipe and queue actions, which need to store
 233  * a single pointer (which can have different size on different
 234  * architectures.
 235  * Note that, because of previous instructions, pipe_ptr might
 236  * be unaligned in the overall structure, so it needs to be
 237  * manipulated with care.
 238  */
 239 typedef struct  _ipfw_insn_pipe {
 240         ipfw_insn       o;
 241         void            *pipe_ptr;      /* XXX */
 242 } ipfw_insn_pipe;
 243
 244 /*
 245  * This is used for limit rules.
 246  */
 247 typedef struct  _ipfw_insn_limit {
 248         ipfw_insn o;
 249         u_int8_t _pad;
 250         u_int8_t limit_mask;    /* combination of DYN_* below   */
 251 #define DYN_SRC_ADDR    0x1
 252 #define DYN_SRC_PORT    0x2
 253 #define DYN_DST_ADDR    0x4
 254 #define DYN_DST_PORT    0x8
 255
 256         u_int16_t conn_limit;
 257 } ipfw_insn_limit;
 258
 259 /*
 260  * This is used for log instructions.
 261  */
 262 typedef struct  _ipfw_insn_log {
 263         ipfw_insn o;
 264         u_int32_t max_log;      /* how many do we log -- 0 = all */
 265         u_int32_t log_left;     /* how many left to log         */
 266 } ipfw_insn_log;
 267
 268 /* Version of this API */
 269 #define IP_FW_VERSION_NONE 0
 270 #define IP_FW_VERSION_0 10      /* old ipfw */
 271 #define IP_FW_VERSION_1 20      /* ipfw in Jaguar/Panther */
 272 #define IP_FW_VERSION_2 30      /* ipfw2 */
 273 #define IP_FW_CURRENT_API_VERSION IP_FW_VERSION_2
 274
 275 /*
 276  * Here we have the structure representing an ipfw rule.
 277  *
 278  * It starts with a general area (with link fields and counters)
 279  * followed by an array of one or more instructions, which the code
 280  * accesses as an array of 32-bit values.
 281  *
 282  * Given a rule pointer  r:
 283  *
 284  *  r->cmd              is the start of the first instruction.
 285  *  ACTION_PTR(r)       is the start of the first action (things to do
 286  *                      once a rule matched).
 287  *
 288  * When assembling instruction, remember the following:
 289  *
 290  *  + if a rule has a "keep-state" (or "limit") option, then the
 291  *      first instruction (at r->cmd) MUST BE an O_PROBE_STATE
 292  *  + if a rule has a "log" option, then the first action
 293  *      (at ACTION_PTR(r)) MUST be O_LOG
 294  *
 295  * NOTE: we use a simple linked list of rules because we never need
 296  *      to delete a rule without scanning the list. We do not use
 297  *      queue(3) macros for portability and readability.
 298  */
 299
 300 struct ip_fw {
 301         u_int32_t version;              /* Version of this structure. MUST be set */
 302                                                         /* by clients. Should always be */
 303                                                         /* set to IP_FW_CURRENT_API_VERSION. */
 304         void *context;                  /* Context that is usable by user processes to */
 305                                                         /* identify this rule. */
 306         struct ip_fw    *next;          /* linked list of rules         */
 307         struct ip_fw    *next_rule;     /* ptr to next [skipto] rule    */
 308         /* 'next_rule' is used to pass up 'set_disable' status          */
 309
 310         u_int16_t       act_ofs;        /* offset of action in 32-bit units */
 311         u_int16_t       cmd_len;        /* # of 32-bit words in cmd     */
 312         u_int16_t       rulenum;        /* rule number                  */
 313         u_int8_t        set;            /* rule set (0..31)             */
 314         u_int32_t       set_masks[2];   /* masks for manipulating sets atomically */
 315 #define RESVD_SET       31      /* set for default and persistent rules */
 316         u_int8_t        _pad;           /* padding                      */
 317
 318         /* These fields are present in all rules.                       */
 319         u_int64_t       pcnt;           /* Packet counter               */
 320         u_int64_t       bcnt;           /* Byte counter                 */
 321         u_int32_t       timestamp;      /* tv_sec of last match         */
 322
 323         u_int32_t       reserved_1;     /* reserved - set to 0 */
 324         u_int32_t       reserved_2;     /* reserved - set to 0 */
 325
 326         ipfw_insn       cmd[1];         /* storage for commands         */
 327 };
 328
 329 #define ACTION_PTR(rule)                                \
 330         (ipfw_insn *)( (u_int32_t *)((rule)->cmd) + ((rule)->act_ofs) )
 331
 332 #define RULESIZE(rule)  (sizeof(struct ip_fw) + \
 333         ((struct ip_fw *)(rule))->cmd_len * 4 - 4)
 334
 335 /*
 336  * This structure is used as a flow mask and a flow id for various
 337  * parts of the code.
 338  */
 339 struct ipfw_flow_id {
 340         u_int32_t       dst_ip;
 341         u_int32_t       src_ip;
 342         u_int16_t       dst_port;
 343         u_int16_t       src_port;
 344         u_int8_t        proto;
 345         u_int8_t        flags;  /* protocol-specific flags */
 346 };
 347
 348 /*
 349  * Dynamic ipfw rule.
 350  */
 351 typedef struct _ipfw_dyn_rule ipfw_dyn_rule;
 352
 353 struct _ipfw_dyn_rule {
 354         ipfw_dyn_rule   *next;          /* linked list of rules.        */
 355         struct ip_fw *rule;             /* pointer to rule              */
 356         /* 'rule' is used to pass up the rule number (from the parent)  */
 357
 358         ipfw_dyn_rule *parent;          /* pointer to parent rule       */
 359         u_int64_t       pcnt;           /* packet match counter         */
 360         u_int64_t       bcnt;           /* byte match counter           */
 361         struct ipfw_flow_id id;         /* (masked) flow id             */
 362         u_int32_t       expire;         /* expire time                  */
 363         u_int32_t       bucket;         /* which bucket in hash table   */
 364         u_int32_t       state;          /* state of this rule (typically a
 365                                          * combination of TCP flags)
 366                                          */
 367         u_int32_t       ack_fwd;        /* most recent ACKs in forward  */
 368         u_int32_t       ack_rev;        /* and reverse directions (used */
 369                                         /* to generate keepalives)      */
 370         u_int16_t       dyn_type;       /* rule type                    */
 371         u_int16_t       count;          /* refcount                     */
 372 };
 373
 374 /*
 375  * Definitions for IP option names.
 376  */
 377 #define IP_FW_IPOPT_LSRR        0x01
 378 #define IP_FW_IPOPT_SSRR        0x02
 379 #define IP_FW_IPOPT_RR          0x04
 380 #define IP_FW_IPOPT_TS          0x08
 381
 382 /*
 383  * Definitions for TCP option names.
 384  */
 385 #define IP_FW_TCPOPT_MSS        0x01
 386 #define IP_FW_TCPOPT_WINDOW     0x02
 387 #define IP_FW_TCPOPT_SACK       0x04
 388 #define IP_FW_TCPOPT_TS         0x08
 389 #define IP_FW_TCPOPT_CC         0x10
 390
 391 #define ICMP_REJECT_RST         0x100   /* fake ICMP code (send a TCP RST) */
 392
 393 /*
 394  * Main firewall chains definitions and global var's definitions.
 395  */
 396 #ifdef KERNEL
 397
 398 #define IP_FW_PORT_DYNT_FLAG    0x10000
 399 #define IP_FW_PORT_TEE_FLAG     0x20000
 400 #define IP_FW_PORT_DENY_FLAG    0x40000
 401
 402 /*
 403  * Arguments for calling ipfw_chk() and dummynet_io(). We put them
 404  * all into a structure because this way it is easier and more
 405  * efficient to pass variables around and extend the interface.
 406  */
 407 struct ip_fw_args {
 408         struct mbuf     *m;             /* the mbuf chain               */
 409         struct ifnet    *oif;           /* output interface             */
 410         struct sockaddr_in *next_hop;   /* forward address              */
 411         struct ip_fw    *rule;          /* matching rule                */
 412         struct ether_header *eh;        /* for bridged packets          */
 413
 414         struct route    *ro;            /* for dummynet                 */
 415         struct sockaddr_in *dst;        /* for dummynet                 */
 416         int flags;                      /* for dummynet                 */
 417
 418         struct ipfw_flow_id f_id;       /* grabbed from IP header       */
 419         u_int16_t       divert_rule;    /* divert cookie                */
 420         u_int32_t       retval;
 421 };
 422
 423 /*
 424  * Function definitions.
 425  */
 426
 427 /* Firewall hooks */
 428 struct sockopt;
 429 struct dn_flow_set;
 430
 431 void flush_pipe_ptrs(struct dn_flow_set *match); /* used by dummynet */
 432 void ipfw_init(void);   /* called from raw_ip.c: load_ipfw() */
 433
 434 typedef int ip_fw_chk_t (struct ip_fw_args *args);
 435 typedef int ip_fw_ctl_t (struct sockopt *);
 436 extern ip_fw_chk_t *ip_fw_chk_ptr;
 437 extern ip_fw_ctl_t *ip_fw_ctl_ptr;
 438 extern int fw_one_pass;
 439 extern int fw_enable;
 440 #define IPFW_LOADED     (ip_fw_chk_ptr != NULL)
 441 #endif /* KERNEL */
 442
 443 #endif /* _IPFW2_H */