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  */
  27
  28 #ifndef _IPFW2_H
  29 #define _IPFW2_H
  30
  31 #ifdef _KERNEL
  32 #include <net/netisr2.h>
  33 #endif
  34
  35 /*
  36  * The kernel representation of ipfw rules is made of a list of
  37  * 'instructions' (for all practical purposes equivalent to BPF
  38  * instructions), which specify which fields of the packet
  39  * (or its metatada) should be analysed.
  40  *
  41  * Each instruction is stored in a structure which begins with
  42  * "ipfw_insn", and can contain extra fields depending on the
  43  * instruction type (listed below).
  44  *
  45  * "enum ipfw_opcodes" are the opcodes supported. We can have up
  46  * to 256 different opcodes.
  47  */
  48
  49 enum ipfw_opcodes {             /* arguments (4 byte each)      */
  50         O_NOP,
  51
  52         O_IP_SRC,               /* u32 = IP                     */
  53         O_IP_SRC_MASK,          /* ip = IP/mask                 */
  54         O_IP_SRC_ME,            /* none                         */
  55         O_IP_SRC_SET,           /* u32=base, arg1=len, bitmap   */
  56
  57         O_IP_DST,               /* u32 = IP                     */
  58         O_IP_DST_MASK,          /* ip = IP/mask                 */
  59         O_IP_DST_ME,            /* none                         */
  60         O_IP_DST_SET,           /* u32=base, arg1=len, bitmap   */
  61
  62         O_IP_SRCPORT,           /* (n)port list:mask 4 byte ea  */
  63         O_IP_DSTPORT,           /* (n)port list:mask 4 byte ea  */
  64         O_PROTO,                /* arg1=protocol                */
  65
  66         O_MACADDR2,             /* 2 mac addr:mask              */
  67         O_MAC_TYPE,             /* same as srcport              */
  68
  69         O_LAYER2,               /* none                         */
  70         O_IN,                   /* none                         */
  71         O_FRAG,                 /* none                         */
  72
  73         O_RECV,                 /* none                         */
  74         O_XMIT,                 /* none                         */
  75         O_VIA,                  /* none                         */
  76
  77         O_IPOPT,                /* arg1 = 2*u8 bitmap           */
  78         O_IPLEN,                /* arg1 = len                   */
  79         O_IPID,                 /* arg1 = id                    */
  80
  81         O_IPTOS,                /* arg1 = id                    */
  82         O_IPPRECEDENCE,         /* arg1 = precedence << 5       */
  83         O_IPTTL,                /* arg1 = TTL                   */
  84
  85         O_IPVER,                /* arg1 = version               */
  86         O_UID,                  /* u32 = id                     */
  87         O_GID,                  /* u32 = id                     */
  88         O_ESTAB,                /* none (tcp established)       */
  89         O_TCPFLAGS,             /* arg1 = 2*u8 bitmap           */
  90         O_TCPWIN,               /* arg1 = desired win           */
  91         O_TCPSEQ,               /* u32 = desired seq.           */
  92         O_TCPACK,               /* u32 = desired seq.           */
  93         O_ICMPTYPE,             /* u32 = icmp bitmap            */
  94         O_TCPOPTS,              /* arg1 = 2*u8 bitmap           */
  95
  96         O_PROBE_STATE,          /* none                         */
  97         O_KEEP_STATE,           /* none                         */
  98         O_LIMIT,                /* ipfw_insn_limit              */
  99         O_LIMIT_PARENT,         /* dyn_type, not an opcode.     */
 100         /*
 101          * these are really 'actions', and must be last in the list.
 102          */
 103
 104         O_LOG,                  /* ipfw_insn_log                */
 105         O_PROB,                 /* u32 = match probability      */
 106
 107         O_CHECK_STATE,          /* none                         */
 108         O_ACCEPT,               /* none                         */
 109         O_DENY,                 /* none                         */
 110         O_REJECT,               /* arg1=icmp arg (same as deny) */
 111         O_COUNT,                /* none                         */
 112         O_SKIPTO,               /* arg1=next rule number        */
 113         O_PIPE,                 /* arg1=pipe number             */
 114         O_QUEUE,                /* arg1=queue number            */
 115         O_DIVERT,               /* arg1=port number             */
 116         O_TEE,                  /* arg1=port number             */
 117         O_FORWARD_IP,           /* fwd sockaddr                 */
 118         O_FORWARD_MAC,          /* fwd mac                      */
 119         O_LAST_OPCODE           /* not an opcode!               */
 120 };
 121
 122 /*
 123  * Template for instructions.
 124  *
 125  * ipfw_insn is used for all instructions which require no operands,
 126  * a single 16-bit value (arg1), or a couple of 8-bit values.
 127  *
 128  * For other instructions which require different/larger arguments
 129  * we have derived structures, ipfw_insn_*.
 130  *
 131  * The size of the instruction (in 32-bit words) is in the low
 132  * 6 bits of "len". The 2 remaining bits are used to implement
 133  * NOT and OR on individual instructions. Given a type, you can
 134  * compute the length to be put in "len" using F_INSN_SIZE(t)
 135  *
 136  * F_NOT        negates the match result of the instruction.
 137  *
 138  * F_OR         is used to build or blocks. By default, instructions
 139  *              are evaluated as part of a logical AND. An "or" block
 140  *              { X or Y or Z } contains F_OR set in all but the last
 141  *              instruction of the block. A match will cause the code
 142  *              to skip past the last instruction of the block.
 143  *
 144  * NOTA BENE: in a couple of places we assume that
 145  *      sizeof(ipfw_insn) == sizeof(uint32_t)
 146  * this needs to be fixed.
 147  *
 148  */
 149 typedef struct  _ipfw_insn {    /* template for instructions */
 150         enum ipfw_opcodes       opcode:8;
 151         uint8_t         len;    /* numer of 32-byte words */
 152 #define F_NOT           0x80
 153 #define F_OR            0x40
 154 #define F_LEN_MASK      0x3f
 155 #define F_LEN(cmd)      ((cmd)->len & F_LEN_MASK)
 156
 157         uint16_t        arg1;
 158 } ipfw_insn;
 159
 160 /*
 161  * The F_INSN_SIZE(type) computes the size, in 4-byte words, of
 162  * a given type.
 163  */
 164 #define F_INSN_SIZE(t)  ((sizeof (t))/sizeof(uint32_t))
 165
 166 /*
 167  * This is used to store an array of 16-bit entries (ports etc.)
 168  */
 169 typedef struct  _ipfw_insn_u16 {
 170         ipfw_insn o;
 171         uint16_t ports[2];      /* there may be more */
 172 } ipfw_insn_u16;
 173
 174 /*
 175  * This is used to store an array of 32-bit entries
 176  * (uid, single IPv4 addresses etc.)
 177  */
 178 typedef struct  _ipfw_insn_u32 {
 179         ipfw_insn o;
 180         uint32_t d[1];  /* one or more */
 181 } ipfw_insn_u32;
 182
 183 /*
 184  * This is used to store IP addr-mask pairs.
 185  */
 186 typedef struct  _ipfw_insn_ip {
 187         ipfw_insn o;
 188         struct in_addr  addr;
 189         struct in_addr  mask;
 190 } ipfw_insn_ip;
 191
 192 /*
 193  * This is used to forward to a given address (ip)
 194  */
 195 typedef struct  _ipfw_insn_sa {
 196         ipfw_insn o;
 197         struct sockaddr_in sa;
 198 } ipfw_insn_sa;
 199
 200 /*
 201  * This is used for MAC addr-mask pairs.
 202  */
 203 typedef struct  _ipfw_insn_mac {
 204         ipfw_insn o;
 205         u_char addr[12];        /* dst[6] + src[6] */
 206         u_char mask[12];        /* dst[6] + src[6] */
 207 } ipfw_insn_mac;
 208
 209 /*
 210  * This is used for interface match rules (recv xx, xmit xx)
 211  */
 212 typedef struct  _ipfw_insn_if {
 213         ipfw_insn o;
 214         union {
 215                 struct in_addr ip;
 216                 int glob;
 217         } p;
 218         char name[IFNAMSIZ];
 219 } ipfw_insn_if;
 220
 221 /*
 222  * This is used for pipe and queue actions, which need to store
 223  * a single pointer (which can have different size on different
 224  * architectures.
 225  */
 226 typedef struct  _ipfw_insn_pipe {
 227         ipfw_insn       o;
 228         void            *pipe_ptr;
 229 } ipfw_insn_pipe;
 230
 231 /*
 232  * This is used for limit rules.
 233  */
 234 typedef struct  _ipfw_insn_limit {
 235         ipfw_insn o;
 236         uint8_t _pad;
 237         uint8_t limit_mask;     /* combination of DYN_* below   */
 238 #define DYN_SRC_ADDR    0x1
 239 #define DYN_SRC_PORT    0x2
 240 #define DYN_DST_ADDR    0x4
 241 #define DYN_DST_PORT    0x8
 242
 243         uint16_t conn_limit;
 244 } ipfw_insn_limit;
 245
 246 /*
 247  * This is used for log instructions
 248  */
 249 typedef struct  _ipfw_insn_log {
 250         ipfw_insn o;
 251         uint32_t max_log;       /* how many do we log -- 0 = all */
 252         uint32_t log_left;      /* how many left to log         */
 253 } ipfw_insn_log;
 254
 255 #ifdef _KERNEL
 256
 257 struct ip_fw;
 258 struct ip_fw_stub {
 259         struct ip_fw    *rule[1];
 260 };
 261
 262 /*
 263  * Here we have the structure representing an ipfw rule.
 264  *
 265  * It starts with a general area (with link fields and counters)
 266  * followed by an array of one or more instructions, which the code
 267  * accesses as an array of 32-bit values.
 268  *
 269  * Given a rule pointer  r:
 270  *
 271  *  r->cmd              is the start of the first instruction.
 272  *  ACTION_PTR(r)       is the start of the first action (things to do
 273  *                      once a rule matched).
 274  *
 275  * When assembling instruction, remember the following:
 276  *
 277  *  + if a rule has a "keep-state" (or "limit") option, then the
 278  *      first instruction (at r->cmd) MUST BE an O_PROBE_STATE
 279  *  + if a rule has a "log" option, then the first action
 280  *      (at ACTION_PTR(r)) MUST be O_LOG
 281  *
 282  * NOTE: we use a simple linked list of rules because we never need
 283  *      to delete a rule without scanning the list. We do not use
 284  *      queue(3) macros for portability and readability.
 285  */
 286
 287 struct ip_fw {
 288         struct ip_fw    *next;          /* linked list of rules         */
 289         struct ip_fw    *next_rule;     /* ptr to next [skipto] rule    */
 290         uint16_t        act_ofs;        /* offset of action in 32-bit units */
 291         uint16_t        cmd_len;        /* # of 32-bit words in cmd     */
 292         uint16_t        rulenum;        /* rule number                  */
 293         uint8_t         set;            /* rule set (0..31)             */
 294         uint8_t         usr_flags;      /* IPFW_USR_F_                  */
 295
 296         /* These fields are present in all rules.                       */
 297         uint64_t        pcnt;           /* Packet counter               */
 298         uint64_t        bcnt;           /* Byte counter                 */
 299         uint32_t        timestamp;      /* tv_sec of last match         */
 300
 301         struct ip_fw_stub *stub;        /* back pointers to clones      */
 302         struct ip_fw    *sibling;       /* clone on next cpu            */
 303         int             cpuid;          /* owner cpu                    */
 304
 305         uint32_t        refcnt;         /* Ref count for transit pkts   */
 306         uint32_t        rule_flags;     /* IPFW_RULE_F_                 */
 307
 308         ipfw_insn       cmd[1];         /* storage for commands         */
 309 };
 310
 311 #define IPFW_RULE_F_INVALID     0x1
 312 #define IPFW_RULE_F_STATE       0x2
 313
 314 #define RULESIZE(rule)  (sizeof(struct ip_fw) + (rule)->cmd_len * 4 - 4)
 315
 316 /*
 317  * This structure is used as a flow mask and a flow id for various
 318  * parts of the code.
 319  */
 320 struct ipfw_flow_id {
 321         uint32_t        dst_ip;
 322         uint32_t        src_ip;
 323         uint16_t        dst_port;
 324         uint16_t        src_port;
 325         uint8_t         proto;
 326         uint8_t         flags;  /* protocol-specific flags */
 327 };
 328
 329 /*
 330  * dynamic ipfw rule
 331  */
 332 typedef struct _ipfw_dyn_rule ipfw_dyn_rule;
 333
 334 struct _ipfw_dyn_rule {
 335         ipfw_dyn_rule   *next;          /* linked list of rules.        */
 336         struct ipfw_flow_id id;         /* (masked) flow id             */
 337 #ifdef notyet
 338         struct ip_fw *rule;             /* pointer to rule              */
 339 #else
 340         const struct ip_fw_stub *stub;  /* pointer to rule's stub       */
 341 #endif
 342         ipfw_dyn_rule *parent;          /* pointer to parent rule       */
 343         uint32_t        expire;         /* expire time                  */
 344         uint64_t        pcnt;           /* packet match counter         */
 345         uint64_t        bcnt;           /* byte match counter           */
 346         uint32_t        bucket;         /* which bucket in hash table   */
 347         uint32_t        state;          /* state of this rule (typically a
 348                                          * combination of TCP flags)
 349                                          */
 350         uint32_t        ack_fwd;        /* most recent ACKs in forward  */
 351         uint32_t        ack_rev;        /* and reverse directions (used */
 352                                         /* to generate keepalives)      */
 353         uint16_t        dyn_type;       /* rule type                    */
 354         uint16_t        count;          /* refcount                     */
 355         time_t          keep_alive;     /* last keep-alive sending time */
 356 };
 357
 358 /*
 359  * Main firewall chains definitions and global var's definitions.
 360  */
 361
 362 /* ipfw_chk/ip_fw_chk_ptr return values */
 363 #define IP_FW_PASS      0
 364 #define IP_FW_DENY      1
 365 #define IP_FW_DIVERT    2
 366 #define IP_FW_TEE       3
 367 #define IP_FW_DUMMYNET  4
 368
 369 /*
 370  * arguments for calling ipfw_chk() and dummynet_io(). We put them
 371  * all into a structure because this way it is easier and more
 372  * efficient to pass variables around and extend the interface.
 373  */
 374 struct ip_fw_args {
 375         struct mbuf     *m;             /* the mbuf chain               */
 376         struct ifnet    *oif;           /* output interface             */
 377         struct ip_fw    *rule;          /* matching rule                */
 378         struct ether_header *eh;        /* for bridged packets          */
 379
 380         struct ipfw_flow_id f_id;       /* grabbed from IP header       */
 381
 382         /*
 383          * Depend on the return value of ipfw_chk/ip_fw_chk_ptr
 384          * 'cookie' field may save following information:
 385          *
 386          * IP_FW_TEE or IP_FW_DIVERT
 387          *   The divert port number
 388          *
 389          * IP_FW_DUMMYNET
 390          *   The pipe or queue number
 391          */
 392         uint32_t        cookie;
 393 };
 394
 395 /*
 396  * Function definitions.
 397  */
 398 int     ip_fw_sockopt(struct sockopt *);
 399
 400 /* Firewall hooks */
 401 struct sockopt;
 402 struct dn_flow_set;
 403
 404 typedef int     ip_fw_chk_t(struct ip_fw_args *);
 405 typedef int     ip_fw_ctl_t(struct sockopt *);
 406 typedef void    ip_fw_dn_io_t(struct mbuf *, int, int, struct ip_fw_args *);
 407
 408 extern ip_fw_chk_t      *ip_fw_chk_ptr;
 409 extern ip_fw_ctl_t      *ip_fw_ctl_ptr;
 410 extern ip_fw_dn_io_t    *ip_fw_dn_io_ptr;
 411
 412 extern int fw_one_pass;
 413 extern int fw_enable;
 414
 415 extern int ip_fw_loaded;
 416 #define IPFW_LOADED     (ip_fw_loaded)
 417
 418 #define IPFW_CFGCPUID   0
 419 #define IPFW_CFGPORT    netisr_cpuport(IPFW_CFGCPUID)
 420 #define IPFW_ASSERT_CFGPORT(msgport)    \
 421         KASSERT((msgport) == IPFW_CFGPORT, ("not IPFW CFGPORT"))
 422
 423 #endif /* _KERNEL */
 424
 425 #define ACTION_PTR(rule)        \
 426         (ipfw_insn *)((uint32_t *)((rule)->cmd) + ((rule)->act_ofs))
 427
 428 struct ipfw_ioc_rule {
 429         uint16_t        act_ofs;        /* offset of action in 32-bit units */
 430         uint16_t        cmd_len;        /* # of 32-bit words in cmd     */
 431         uint16_t        rulenum;        /* rule number                  */
 432         uint8_t         set;            /* rule set (0..31)             */
 433         uint8_t         usr_flags;      /* IPFW_USR_F_                  */
 434
 435         /* Rule set information */
 436         uint32_t        set_disable;    /* disabled rule sets           */
 437         uint32_t        static_count;   /* # of static rules            */
 438         uint32_t        static_len;     /* total length of static rules */
 439
 440         /* Statistics */
 441         uint64_t        pcnt;           /* Packet counter               */
 442         uint64_t        bcnt;           /* Byte counter                 */
 443         uint32_t        timestamp;      /* tv_sec of last match         */
 444
 445         uint8_t         reserved[16];
 446
 447         ipfw_insn       cmd[1];         /* storage for commands         */
 448 };
 449
 450 #define IPFW_USR_F_NORULE       0x01
 451
 452 #define IPFW_RULE_SIZE_MAX      255     /* unit: uint32_t */
 453
 454 #define IOC_RULESIZE(rule)      \
 455         (sizeof(struct ipfw_ioc_rule) + (rule)->cmd_len * 4 - 4)
 456
 457 struct ipfw_ioc_flowid {
 458         uint16_t        type;   /* ETHERTYPE_ */
 459         uint16_t        pad;
 460         union {
 461                 struct {
 462                         uint32_t dst_ip;
 463                         uint32_t src_ip;
 464                         uint16_t dst_port;
 465                         uint16_t src_port;
 466                         uint8_t proto;
 467                 } ip;
 468                 uint8_t pad[64];
 469         } u;
 470 };
 471
 472 struct ipfw_ioc_state {
 473         uint32_t        expire;         /* expire time                  */
 474         uint64_t        pcnt;           /* packet match counter         */
 475         uint64_t        bcnt;           /* byte match counter           */
 476
 477         uint16_t        dyn_type;       /* rule type                    */
 478         uint16_t        count;          /* refcount                     */
 479
 480         uint16_t        rulenum;
 481         uint16_t        pad;
 482
 483         int             cpu;            /* reserved                     */
 484
 485         struct ipfw_ioc_flowid id;      /* (masked) flow id             */
 486         uint8_t         reserved[16];
 487 };
 488
 489 /*
 490  * Definitions for IP option names.
 491  */
 492 #define IP_FW_IPOPT_LSRR        0x01
 493 #define IP_FW_IPOPT_SSRR        0x02
 494 #define IP_FW_IPOPT_RR          0x04
 495 #define IP_FW_IPOPT_TS          0x08
 496
 497 /*
 498  * Definitions for TCP option names.
 499  */
 500 #define IP_FW_TCPOPT_MSS        0x01
 501 #define IP_FW_TCPOPT_WINDOW     0x02
 502 #define IP_FW_TCPOPT_SACK       0x04
 503 #define IP_FW_TCPOPT_TS         0x08
 504 #define IP_FW_TCPOPT_CC         0x10
 505
 506 #define ICMP_REJECT_RST         0x100   /* fake ICMP code (send a TCP RST) */
 507
 508 #endif /* _IPFW2_H */