ip_fw.h revision 215179
1/*- 2 * Copyright (c) 2002-2009 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: head/sys/netinet/ip_fw.h 215179 2010-11-12 13:05:17Z luigi $ 26 */ 27 28#ifndef _IPFW2_H 29#define _IPFW2_H 30 31/* 32 * The default rule number. By the design of ip_fw, the default rule 33 * is the last one, so its number can also serve as the highest number 34 * allowed for a rule. The ip_fw code relies on both meanings of this 35 * constant. 36 */ 37#define IPFW_DEFAULT_RULE 65535 38 39/* 40 * The number of ipfw tables. The maximum allowed table number is the 41 * (IPFW_TABLES_MAX - 1). 42 */ 43#define IPFW_TABLES_MAX 128 44 45/* 46 * Most commands (queue, pipe, tag, untag, limit...) can have a 16-bit 47 * argument between 1 and 65534. The value 0 is unused, the value 48 * 65535 (IP_FW_TABLEARG) is used to represent 'tablearg', i.e. the 49 * can be 1..65534, or 65535 to indicate the use of a 'tablearg' 50 * result of the most recent table() lookup. 51 * Note that 16bit is only a historical limit, resulting from 52 * the use of a 16-bit fields for that value. In reality, we can have 53 * 2^32 pipes, queues, tag values and so on, and use 0 as a tablearg. 54 */ 55#define IPFW_ARG_MIN 1 56#define IPFW_ARG_MAX 65534 57#define IP_FW_TABLEARG 65535 /* XXX should use 0 */ 58 59/* 60 * The kernel representation of ipfw rules is made of a list of 61 * 'instructions' (for all practical purposes equivalent to BPF 62 * instructions), which specify which fields of the packet 63 * (or its metadata) should be analysed. 64 * 65 * Each instruction is stored in a structure which begins with 66 * "ipfw_insn", and can contain extra fields depending on the 67 * instruction type (listed below). 68 * Note that the code is written so that individual instructions 69 * have a size which is a multiple of 32 bits. This means that, if 70 * such structures contain pointers or other 64-bit entities, 71 * (there is just one instance now) they may end up unaligned on 72 * 64-bit architectures, so the must be handled with care. 73 * 74 * "enum ipfw_opcodes" are the opcodes supported. We can have up 75 * to 256 different opcodes. When adding new opcodes, they should 76 * be appended to the end of the opcode list before O_LAST_OPCODE, 77 * this will prevent the ABI from being broken, otherwise users 78 * will have to recompile ipfw(8) when they update the kernel. 79 */ 80 81enum ipfw_opcodes { /* arguments (4 byte each) */ 82 O_NOP, 83 84 O_IP_SRC, /* u32 = IP */ 85 O_IP_SRC_MASK, /* ip = IP/mask */ 86 O_IP_SRC_ME, /* none */ 87 O_IP_SRC_SET, /* u32=base, arg1=len, bitmap */ 88 89 O_IP_DST, /* u32 = IP */ 90 O_IP_DST_MASK, /* ip = IP/mask */ 91 O_IP_DST_ME, /* none */ 92 O_IP_DST_SET, /* u32=base, arg1=len, bitmap */ 93 94 O_IP_SRCPORT, /* (n)port list:mask 4 byte ea */ 95 O_IP_DSTPORT, /* (n)port list:mask 4 byte ea */ 96 O_PROTO, /* arg1=protocol */ 97 98 O_MACADDR2, /* 2 mac addr:mask */ 99 O_MAC_TYPE, /* same as srcport */ 100 101 O_LAYER2, /* none */ 102 O_IN, /* none */ 103 O_FRAG, /* none */ 104 105 O_RECV, /* none */ 106 O_XMIT, /* none */ 107 O_VIA, /* none */ 108 109 O_IPOPT, /* arg1 = 2*u8 bitmap */ 110 O_IPLEN, /* arg1 = len */ 111 O_IPID, /* arg1 = id */ 112 113 O_IPTOS, /* arg1 = id */ 114 O_IPPRECEDENCE, /* arg1 = precedence << 5 */ 115 O_IPTTL, /* arg1 = TTL */ 116 117 O_IPVER, /* arg1 = version */ 118 O_UID, /* u32 = id */ 119 O_GID, /* u32 = id */ 120 O_ESTAB, /* none (tcp established) */ 121 O_TCPFLAGS, /* arg1 = 2*u8 bitmap */ 122 O_TCPWIN, /* arg1 = desired win */ 123 O_TCPSEQ, /* u32 = desired seq. */ 124 O_TCPACK, /* u32 = desired seq. */ 125 O_ICMPTYPE, /* u32 = icmp bitmap */ 126 O_TCPOPTS, /* arg1 = 2*u8 bitmap */ 127 128 O_VERREVPATH, /* none */ 129 O_VERSRCREACH, /* none */ 130 131 O_PROBE_STATE, /* none */ 132 O_KEEP_STATE, /* none */ 133 O_LIMIT, /* ipfw_insn_limit */ 134 O_LIMIT_PARENT, /* dyn_type, not an opcode. */ 135 136 /* 137 * These are really 'actions'. 138 */ 139 140 O_LOG, /* ipfw_insn_log */ 141 O_PROB, /* u32 = match probability */ 142 143 O_CHECK_STATE, /* none */ 144 O_ACCEPT, /* none */ 145 O_DENY, /* none */ 146 O_REJECT, /* arg1=icmp arg (same as deny) */ 147 O_COUNT, /* none */ 148 O_SKIPTO, /* arg1=next rule number */ 149 O_PIPE, /* arg1=pipe number */ 150 O_QUEUE, /* arg1=queue number */ 151 O_DIVERT, /* arg1=port number */ 152 O_TEE, /* arg1=port number */ 153 O_FORWARD_IP, /* fwd sockaddr */ 154 O_FORWARD_MAC, /* fwd mac */ 155 O_NAT, /* nope */ 156 O_REASS, /* none */ 157 158 /* 159 * More opcodes. 160 */ 161 O_IPSEC, /* has ipsec history */ 162 O_IP_SRC_LOOKUP, /* arg1=table number, u32=value */ 163 O_IP_DST_LOOKUP, /* arg1=table number, u32=value */ 164 O_ANTISPOOF, /* none */ 165 O_JAIL, /* u32 = id */ 166 O_ALTQ, /* u32 = altq classif. qid */ 167 O_DIVERTED, /* arg1=bitmap (1:loop, 2:out) */ 168 O_TCPDATALEN, /* arg1 = tcp data len */ 169 O_IP6_SRC, /* address without mask */ 170 O_IP6_SRC_ME, /* my addresses */ 171 O_IP6_SRC_MASK, /* address with the mask */ 172 O_IP6_DST, 173 O_IP6_DST_ME, 174 O_IP6_DST_MASK, 175 O_FLOW6ID, /* for flow id tag in the ipv6 pkt */ 176 O_ICMP6TYPE, /* icmp6 packet type filtering */ 177 O_EXT_HDR, /* filtering for ipv6 extension header */ 178 O_IP6, 179 180 /* 181 * actions for ng_ipfw 182 */ 183 O_NETGRAPH, /* send to ng_ipfw */ 184 O_NGTEE, /* copy to ng_ipfw */ 185 186 O_IP4, 187 188 O_UNREACH6, /* arg1=icmpv6 code arg (deny) */ 189 190 O_TAG, /* arg1=tag number */ 191 O_TAGGED, /* arg1=tag number */ 192 193 O_SETFIB, /* arg1=FIB number */ 194 O_FIB, /* arg1=FIB desired fib number */ 195 196 O_SOCKARG, /* socket argument */ 197 198 O_LAST_OPCODE /* not an opcode! */ 199}; 200 201 202/* 203 * The extension header are filtered only for presence using a bit 204 * vector with a flag for each header. 205 */ 206#define EXT_FRAGMENT 0x1 207#define EXT_HOPOPTS 0x2 208#define EXT_ROUTING 0x4 209#define EXT_AH 0x8 210#define EXT_ESP 0x10 211#define EXT_DSTOPTS 0x20 212#define EXT_RTHDR0 0x40 213#define EXT_RTHDR2 0x80 214 215/* 216 * Template for instructions. 217 * 218 * ipfw_insn is used for all instructions which require no operands, 219 * a single 16-bit value (arg1), or a couple of 8-bit values. 220 * 221 * For other instructions which require different/larger arguments 222 * we have derived structures, ipfw_insn_*. 223 * 224 * The size of the instruction (in 32-bit words) is in the low 225 * 6 bits of "len". The 2 remaining bits are used to implement 226 * NOT and OR on individual instructions. Given a type, you can 227 * compute the length to be put in "len" using F_INSN_SIZE(t) 228 * 229 * F_NOT negates the match result of the instruction. 230 * 231 * F_OR is used to build or blocks. By default, instructions 232 * are evaluated as part of a logical AND. An "or" block 233 * { X or Y or Z } contains F_OR set in all but the last 234 * instruction of the block. A match will cause the code 235 * to skip past the last instruction of the block. 236 * 237 * NOTA BENE: in a couple of places we assume that 238 * sizeof(ipfw_insn) == sizeof(u_int32_t) 239 * this needs to be fixed. 240 * 241 */ 242typedef struct _ipfw_insn { /* template for instructions */ 243 u_int8_t opcode; 244 u_int8_t len; /* number of 32-bit words */ 245#define F_NOT 0x80 246#define F_OR 0x40 247#define F_LEN_MASK 0x3f 248#define F_LEN(cmd) ((cmd)->len & F_LEN_MASK) 249 250 u_int16_t arg1; 251} ipfw_insn; 252 253/* 254 * The F_INSN_SIZE(type) computes the size, in 4-byte words, of 255 * a given type. 256 */ 257#define F_INSN_SIZE(t) ((sizeof (t))/sizeof(u_int32_t)) 258 259/* 260 * This is used to store an array of 16-bit entries (ports etc.) 261 */ 262typedef struct _ipfw_insn_u16 { 263 ipfw_insn o; 264 u_int16_t ports[2]; /* there may be more */ 265} ipfw_insn_u16; 266 267/* 268 * This is used to store an array of 32-bit entries 269 * (uid, single IPv4 addresses etc.) 270 */ 271typedef struct _ipfw_insn_u32 { 272 ipfw_insn o; 273 u_int32_t d[1]; /* one or more */ 274} ipfw_insn_u32; 275 276/* 277 * This is used to store IP addr-mask pairs. 278 */ 279typedef struct _ipfw_insn_ip { 280 ipfw_insn o; 281 struct in_addr addr; 282 struct in_addr mask; 283} ipfw_insn_ip; 284 285/* 286 * This is used to forward to a given address (ip). 287 */ 288typedef struct _ipfw_insn_sa { 289 ipfw_insn o; 290 struct sockaddr_in sa; 291} ipfw_insn_sa; 292 293/* 294 * This is used for MAC addr-mask pairs. 295 */ 296typedef struct _ipfw_insn_mac { 297 ipfw_insn o; 298 u_char addr[12]; /* dst[6] + src[6] */ 299 u_char mask[12]; /* dst[6] + src[6] */ 300} ipfw_insn_mac; 301 302/* 303 * This is used for interface match rules (recv xx, xmit xx). 304 */ 305typedef struct _ipfw_insn_if { 306 ipfw_insn o; 307 union { 308 struct in_addr ip; 309 int glob; 310 } p; 311 char name[IFNAMSIZ]; 312} ipfw_insn_if; 313 314/* 315 * This is used for storing an altq queue id number. 316 */ 317typedef struct _ipfw_insn_altq { 318 ipfw_insn o; 319 u_int32_t qid; 320} ipfw_insn_altq; 321 322/* 323 * This is used for limit rules. 324 */ 325typedef struct _ipfw_insn_limit { 326 ipfw_insn o; 327 u_int8_t _pad; 328 u_int8_t limit_mask; /* combination of DYN_* below */ 329#define DYN_SRC_ADDR 0x1 330#define DYN_SRC_PORT 0x2 331#define DYN_DST_ADDR 0x4 332#define DYN_DST_PORT 0x8 333 334 u_int16_t conn_limit; 335} ipfw_insn_limit; 336 337/* 338 * This is used for log instructions. 339 */ 340typedef struct _ipfw_insn_log { 341 ipfw_insn o; 342 u_int32_t max_log; /* how many do we log -- 0 = all */ 343 u_int32_t log_left; /* how many left to log */ 344} ipfw_insn_log; 345 346/* 347 * Data structures required by both ipfw(8) and ipfw(4) but not part of the 348 * management API are protected by IPFW_INTERNAL. 349 */ 350#ifdef IPFW_INTERNAL 351/* Server pool support (LSNAT). */ 352struct cfg_spool { 353 LIST_ENTRY(cfg_spool) _next; /* chain of spool instances */ 354 struct in_addr addr; 355 u_short port; 356}; 357#endif 358 359/* Redirect modes id. */ 360#define REDIR_ADDR 0x01 361#define REDIR_PORT 0x02 362#define REDIR_PROTO 0x04 363 364#ifdef IPFW_INTERNAL 365/* Nat redirect configuration. */ 366struct cfg_redir { 367 LIST_ENTRY(cfg_redir) _next; /* chain of redir instances */ 368 u_int16_t mode; /* type of redirect mode */ 369 struct in_addr laddr; /* local ip address */ 370 struct in_addr paddr; /* public ip address */ 371 struct in_addr raddr; /* remote ip address */ 372 u_short lport; /* local port */ 373 u_short pport; /* public port */ 374 u_short rport; /* remote port */ 375 u_short pport_cnt; /* number of public ports */ 376 u_short rport_cnt; /* number of remote ports */ 377 int proto; /* protocol: tcp/udp */ 378 struct alias_link **alink; 379 /* num of entry in spool chain */ 380 u_int16_t spool_cnt; 381 /* chain of spool instances */ 382 LIST_HEAD(spool_chain, cfg_spool) spool_chain; 383}; 384#endif 385 386#define NAT_BUF_LEN 1024 387 388#ifdef IPFW_INTERNAL 389/* Nat configuration data struct. */ 390struct cfg_nat { 391 /* chain of nat instances */ 392 LIST_ENTRY(cfg_nat) _next; 393 int id; /* nat id */ 394 struct in_addr ip; /* nat ip address */ 395 char if_name[IF_NAMESIZE]; /* interface name */ 396 int mode; /* aliasing mode */ 397 struct libalias *lib; /* libalias instance */ 398 /* number of entry in spool chain */ 399 int redir_cnt; 400 /* chain of redir instances */ 401 LIST_HEAD(redir_chain, cfg_redir) redir_chain; 402}; 403#endif 404 405#define SOF_NAT sizeof(struct cfg_nat) 406#define SOF_REDIR sizeof(struct cfg_redir) 407#define SOF_SPOOL sizeof(struct cfg_spool) 408 409/* Nat command. */ 410typedef struct _ipfw_insn_nat { 411 ipfw_insn o; 412 struct cfg_nat *nat; 413} ipfw_insn_nat; 414 415/* Apply ipv6 mask on ipv6 addr */ 416#define APPLY_MASK(addr,mask) \ 417 (addr)->__u6_addr.__u6_addr32[0] &= (mask)->__u6_addr.__u6_addr32[0]; \ 418 (addr)->__u6_addr.__u6_addr32[1] &= (mask)->__u6_addr.__u6_addr32[1]; \ 419 (addr)->__u6_addr.__u6_addr32[2] &= (mask)->__u6_addr.__u6_addr32[2]; \ 420 (addr)->__u6_addr.__u6_addr32[3] &= (mask)->__u6_addr.__u6_addr32[3]; 421 422/* Structure for ipv6 */ 423typedef struct _ipfw_insn_ip6 { 424 ipfw_insn o; 425 struct in6_addr addr6; 426 struct in6_addr mask6; 427} ipfw_insn_ip6; 428 429/* Used to support icmp6 types */ 430typedef struct _ipfw_insn_icmp6 { 431 ipfw_insn o; 432 uint32_t d[7]; /* XXX This number si related to the netinet/icmp6.h 433 * define ICMP6_MAXTYPE 434 * as follows: n = ICMP6_MAXTYPE/32 + 1 435 * Actually is 203 436 */ 437} ipfw_insn_icmp6; 438 439/* 440 * Here we have the structure representing an ipfw rule. 441 * 442 * It starts with a general area (with link fields and counters) 443 * followed by an array of one or more instructions, which the code 444 * accesses as an array of 32-bit values. 445 * 446 * Given a rule pointer r: 447 * 448 * r->cmd is the start of the first instruction. 449 * ACTION_PTR(r) is the start of the first action (things to do 450 * once a rule matched). 451 * 452 * When assembling instruction, remember the following: 453 * 454 * + if a rule has a "keep-state" (or "limit") option, then the 455 * first instruction (at r->cmd) MUST BE an O_PROBE_STATE 456 * + if a rule has a "log" option, then the first action 457 * (at ACTION_PTR(r)) MUST be O_LOG 458 * + if a rule has an "altq" option, it comes after "log" 459 * + if a rule has an O_TAG option, it comes after "log" and "altq" 460 * 461 * NOTE: we use a simple linked list of rules because we never need 462 * to delete a rule without scanning the list. We do not use 463 * queue(3) macros for portability and readability. 464 */ 465 466struct ip_fw { 467 struct ip_fw *x_next; /* linked list of rules */ 468 struct ip_fw *next_rule; /* ptr to next [skipto] rule */ 469 /* 'next_rule' is used to pass up 'set_disable' status */ 470 471 uint16_t act_ofs; /* offset of action in 32-bit units */ 472 uint16_t cmd_len; /* # of 32-bit words in cmd */ 473 uint16_t rulenum; /* rule number */ 474 uint8_t set; /* rule set (0..31) */ 475#define RESVD_SET 31 /* set for default and persistent rules */ 476 uint8_t _pad; /* padding */ 477 uint32_t id; /* rule id */ 478 479 /* These fields are present in all rules. */ 480 uint64_t pcnt; /* Packet counter */ 481 uint64_t bcnt; /* Byte counter */ 482 uint32_t timestamp; /* tv_sec of last match */ 483 484 ipfw_insn cmd[1]; /* storage for commands */ 485}; 486 487#define ACTION_PTR(rule) \ 488 (ipfw_insn *)( (u_int32_t *)((rule)->cmd) + ((rule)->act_ofs) ) 489 490#define RULESIZE(rule) (sizeof(struct ip_fw) + \ 491 ((struct ip_fw *)(rule))->cmd_len * 4 - 4) 492 493#if 1 // should be moved to in.h 494/* 495 * This structure is used as a flow mask and a flow id for various 496 * parts of the code. 497 * addr_type is used in userland and kernel to mark the address type. 498 * fib is used in the kernel to record the fib in use. 499 * _flags is used in the kernel to store tcp flags for dynamic rules. 500 */ 501struct ipfw_flow_id { 502 uint32_t dst_ip; 503 uint32_t src_ip; 504 uint16_t dst_port; 505 uint16_t src_port; 506 uint8_t fib; 507 uint8_t proto; 508 uint8_t _flags; /* protocol-specific flags */ 509 uint8_t addr_type; /* 4=ip4, 6=ip6, 1=ether ? */ 510 struct in6_addr dst_ip6; 511 struct in6_addr src_ip6; 512 uint32_t flow_id6; 513 uint32_t extra; /* queue/pipe or frag_id */ 514}; 515#endif 516 517#define IS_IP6_FLOW_ID(id) ((id)->addr_type == 6) 518 519/* 520 * Dynamic ipfw rule. 521 */ 522typedef struct _ipfw_dyn_rule ipfw_dyn_rule; 523 524struct _ipfw_dyn_rule { 525 ipfw_dyn_rule *next; /* linked list of rules. */ 526 struct ip_fw *rule; /* pointer to rule */ 527 /* 'rule' is used to pass up the rule number (from the parent) */ 528 529 ipfw_dyn_rule *parent; /* pointer to parent rule */ 530 u_int64_t pcnt; /* packet match counter */ 531 u_int64_t bcnt; /* byte match counter */ 532 struct ipfw_flow_id id; /* (masked) flow id */ 533 u_int32_t expire; /* expire time */ 534 u_int32_t bucket; /* which bucket in hash table */ 535 u_int32_t state; /* state of this rule (typically a 536 * combination of TCP flags) 537 */ 538 u_int32_t ack_fwd; /* most recent ACKs in forward */ 539 u_int32_t ack_rev; /* and reverse directions (used */ 540 /* to generate keepalives) */ 541 u_int16_t dyn_type; /* rule type */ 542 u_int16_t count; /* refcount */ 543}; 544 545/* 546 * Definitions for IP option names. 547 */ 548#define IP_FW_IPOPT_LSRR 0x01 549#define IP_FW_IPOPT_SSRR 0x02 550#define IP_FW_IPOPT_RR 0x04 551#define IP_FW_IPOPT_TS 0x08 552 553/* 554 * Definitions for TCP option names. 555 */ 556#define IP_FW_TCPOPT_MSS 0x01 557#define IP_FW_TCPOPT_WINDOW 0x02 558#define IP_FW_TCPOPT_SACK 0x04 559#define IP_FW_TCPOPT_TS 0x08 560#define IP_FW_TCPOPT_CC 0x10 561 562#define ICMP_REJECT_RST 0x100 /* fake ICMP code (send a TCP RST) */ 563#define ICMP6_UNREACH_RST 0x100 /* fake ICMPv6 code (send a TCP RST) */ 564 565/* 566 * These are used for lookup tables. 567 */ 568typedef struct _ipfw_table_entry { 569 in_addr_t addr; /* network address */ 570 u_int32_t value; /* value */ 571 u_int16_t tbl; /* table number */ 572 u_int8_t masklen; /* mask length */ 573} ipfw_table_entry; 574 575typedef struct _ipfw_table { 576 u_int32_t size; /* size of entries in bytes */ 577 u_int32_t cnt; /* # of entries */ 578 u_int16_t tbl; /* table number */ 579 ipfw_table_entry ent[0]; /* entries */ 580} ipfw_table; 581 582#endif /* _IPFW2_H */ 583