ip_fw.h revision 145864
1237433Skib/*- 2237433Skib * Copyright (c) 2002 Luigi Rizzo, Universita` di Pisa 3237433Skib * 4237433Skib * Redistribution and use in source and binary forms, with or without 5237433Skib * modification, are permitted provided that the following conditions 6237433Skib * are met: 7237433Skib * 1. Redistributions of source code must retain the above copyright 8237433Skib * notice, this list of conditions and the following disclaimer. 9237433Skib * 2. Redistributions in binary form must reproduce the above copyright 10237433Skib * notice, this list of conditions and the following disclaimer in the 11237433Skib * documentation and/or other materials provided with the distribution. 12237433Skib * 13237433Skib * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 14237433Skib * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 15237433Skib * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 16237433Skib * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 17237433Skib * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 18237433Skib * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 19237433Skib * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 20237433Skib * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 21237433Skib * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 22237433Skib * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 23237433Skib * SUCH DAMAGE. 24237433Skib * 25237433Skib * $FreeBSD: head/sys/netinet/ip_fw.h 145864 2005-05-04 13:12:52Z glebius $ 26237433Skib */ 27237433Skib 28237433Skib#ifndef _IPFW2_H 29237433Skib#define _IPFW2_H 30237433Skib 31237433Skib/* 32237433Skib * The kernel representation of ipfw rules is made of a list of 33237433Skib * 'instructions' (for all practical purposes equivalent to BPF 34237433Skib * 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. When adding new opcodes, they should 48 * be appended to the end of the opcode list before O_LAST_OPCODE, 49 * this will prevent the ABI from being broken, otherwise users 50 * will have to recompile ipfw(8) when they update the kernel. 51 */ 52 53enum ipfw_opcodes { /* arguments (4 byte each) */ 54 O_NOP, 55 56 O_IP_SRC, /* u32 = IP */ 57 O_IP_SRC_MASK, /* ip = IP/mask */ 58 O_IP_SRC_ME, /* none */ 59 O_IP_SRC_SET, /* u32=base, arg1=len, bitmap */ 60 61 O_IP_DST, /* u32 = IP */ 62 O_IP_DST_MASK, /* ip = IP/mask */ 63 O_IP_DST_ME, /* none */ 64 O_IP_DST_SET, /* u32=base, arg1=len, bitmap */ 65 66 O_IP_SRCPORT, /* (n)port list:mask 4 byte ea */ 67 O_IP_DSTPORT, /* (n)port list:mask 4 byte ea */ 68 O_PROTO, /* arg1=protocol */ 69 70 O_MACADDR2, /* 2 mac addr:mask */ 71 O_MAC_TYPE, /* same as srcport */ 72 73 O_LAYER2, /* none */ 74 O_IN, /* none */ 75 O_FRAG, /* none */ 76 77 O_RECV, /* none */ 78 O_XMIT, /* none */ 79 O_VIA, /* none */ 80 81 O_IPOPT, /* arg1 = 2*u8 bitmap */ 82 O_IPLEN, /* arg1 = len */ 83 O_IPID, /* arg1 = id */ 84 85 O_IPTOS, /* arg1 = id */ 86 O_IPPRECEDENCE, /* arg1 = precedence << 5 */ 87 O_IPTTL, /* arg1 = TTL */ 88 89 O_IPVER, /* arg1 = version */ 90 O_UID, /* u32 = id */ 91 O_GID, /* u32 = id */ 92 O_ESTAB, /* none (tcp established) */ 93 O_TCPFLAGS, /* arg1 = 2*u8 bitmap */ 94 O_TCPWIN, /* arg1 = desired win */ 95 O_TCPSEQ, /* u32 = desired seq. */ 96 O_TCPACK, /* u32 = desired seq. */ 97 O_ICMPTYPE, /* u32 = icmp bitmap */ 98 O_TCPOPTS, /* arg1 = 2*u8 bitmap */ 99 100 O_VERREVPATH, /* none */ 101 O_VERSRCREACH, /* none */ 102 103 O_PROBE_STATE, /* none */ 104 O_KEEP_STATE, /* none */ 105 O_LIMIT, /* ipfw_insn_limit */ 106 O_LIMIT_PARENT, /* dyn_type, not an opcode. */ 107 108 /* 109 * These are really 'actions'. 110 */ 111 112 O_LOG, /* ipfw_insn_log */ 113 O_PROB, /* u32 = match probability */ 114 115 O_CHECK_STATE, /* none */ 116 O_ACCEPT, /* none */ 117 O_DENY, /* none */ 118 O_REJECT, /* arg1=icmp arg (same as deny) */ 119 O_COUNT, /* none */ 120 O_SKIPTO, /* arg1=next rule number */ 121 O_PIPE, /* arg1=pipe number */ 122 O_QUEUE, /* arg1=queue number */ 123 O_DIVERT, /* arg1=port number */ 124 O_TEE, /* arg1=port number */ 125 O_FORWARD_IP, /* fwd sockaddr */ 126 O_FORWARD_MAC, /* fwd mac */ 127 128 /* 129 * More opcodes. 130 */ 131 O_IPSEC, /* has ipsec history */ 132 O_IP_SRC_LOOKUP, /* arg1=table number, u32=value */ 133 O_IP_DST_LOOKUP, /* arg1=table number, u32=value */ 134 O_ANTISPOOF, /* none */ 135 O_JAIL, /* u32 = id */ 136 O_ALTQ, /* u32 = altq classif. qid */ 137 O_DIVERTED, /* arg1=bitmap (1:loop, 2:out) */ 138 O_TCPDATALEN, /* arg1 = tcp data len */ 139 O_IP6_SRC, /* address without mask */ 140 O_IP6_SRC_ME, /* my addresses */ 141 O_IP6_SRC_MASK, /* address with the mask */ 142 O_IP6_DST, 143 O_IP6_DST_ME, 144 O_IP6_DST_MASK, 145 O_FLOW6ID, /* for flow id tag in the ipv6 pkt */ 146 O_ICMP6TYPE, /* icmp6 packet type filtering */ 147 O_EXT_HDR, /* filtering for ipv6 extension header */ 148 O_IP6, 149 150 /* 151 * actions for ng_ipfw 152 */ 153 O_NETGRAPH, /* send to ng_ipfw */ 154 O_NGTEE, /* copy to ng_ipfw */ 155 156 O_LAST_OPCODE /* not an opcode! */ 157}; 158 159/* 160 * The extension header are filtered only for presence using a bit 161 * vector with a flag for each header. 162 */ 163#define EXT_FRAGMENT 0x1 164#define EXT_HOPOPTS 0x2 165#define EXT_ROUTING 0x4 166#define EXT_AH 0x8 167#define EXT_ESP 0x10 168 169/* 170 * Template for instructions. 171 * 172 * ipfw_insn is used for all instructions which require no operands, 173 * a single 16-bit value (arg1), or a couple of 8-bit values. 174 * 175 * For other instructions which require different/larger arguments 176 * we have derived structures, ipfw_insn_*. 177 * 178 * The size of the instruction (in 32-bit words) is in the low 179 * 6 bits of "len". The 2 remaining bits are used to implement 180 * NOT and OR on individual instructions. Given a type, you can 181 * compute the length to be put in "len" using F_INSN_SIZE(t) 182 * 183 * F_NOT negates the match result of the instruction. 184 * 185 * F_OR is used to build or blocks. By default, instructions 186 * are evaluated as part of a logical AND. An "or" block 187 * { X or Y or Z } contains F_OR set in all but the last 188 * instruction of the block. A match will cause the code 189 * to skip past the last instruction of the block. 190 * 191 * NOTA BENE: in a couple of places we assume that 192 * sizeof(ipfw_insn) == sizeof(u_int32_t) 193 * this needs to be fixed. 194 * 195 */ 196typedef struct _ipfw_insn { /* template for instructions */ 197 enum ipfw_opcodes opcode:8; 198 u_int8_t len; /* numer of 32-byte words */ 199#define F_NOT 0x80 200#define F_OR 0x40 201#define F_LEN_MASK 0x3f 202#define F_LEN(cmd) ((cmd)->len & F_LEN_MASK) 203 204 u_int16_t arg1; 205} ipfw_insn; 206 207/* 208 * The F_INSN_SIZE(type) computes the size, in 4-byte words, of 209 * a given type. 210 */ 211#define F_INSN_SIZE(t) ((sizeof (t))/sizeof(u_int32_t)) 212 213/* 214 * This is used to store an array of 16-bit entries (ports etc.) 215 */ 216typedef struct _ipfw_insn_u16 { 217 ipfw_insn o; 218 u_int16_t ports[2]; /* there may be more */ 219} ipfw_insn_u16; 220 221/* 222 * This is used to store an array of 32-bit entries 223 * (uid, single IPv4 addresses etc.) 224 */ 225typedef struct _ipfw_insn_u32 { 226 ipfw_insn o; 227 u_int32_t d[1]; /* one or more */ 228} ipfw_insn_u32; 229 230/* 231 * This is used to store IP addr-mask pairs. 232 */ 233typedef struct _ipfw_insn_ip { 234 ipfw_insn o; 235 struct in_addr addr; 236 struct in_addr mask; 237} ipfw_insn_ip; 238 239/* 240 * This is used to forward to a given address (ip). 241 */ 242typedef struct _ipfw_insn_sa { 243 ipfw_insn o; 244 struct sockaddr_in sa; 245} ipfw_insn_sa; 246 247/* 248 * This is used for MAC addr-mask pairs. 249 */ 250typedef struct _ipfw_insn_mac { 251 ipfw_insn o; 252 u_char addr[12]; /* dst[6] + src[6] */ 253 u_char mask[12]; /* dst[6] + src[6] */ 254} ipfw_insn_mac; 255 256/* 257 * This is used for interface match rules (recv xx, xmit xx). 258 */ 259typedef struct _ipfw_insn_if { 260 ipfw_insn o; 261 union { 262 struct in_addr ip; 263 int glob; 264 } p; 265 char name[IFNAMSIZ]; 266} ipfw_insn_if; 267 268/* 269 * This is used for pipe and queue actions, which need to store 270 * a single pointer (which can have different size on different 271 * architectures. 272 * Note that, because of previous instructions, pipe_ptr might 273 * be unaligned in the overall structure, so it needs to be 274 * manipulated with care. 275 */ 276typedef struct _ipfw_insn_pipe { 277 ipfw_insn o; 278 void *pipe_ptr; /* XXX */ 279} ipfw_insn_pipe; 280 281/* 282 * This is used for storing an altq queue id number. 283 */ 284typedef struct _ipfw_insn_altq { 285 ipfw_insn o; 286 u_int32_t qid; 287} ipfw_insn_altq; 288 289/* 290 * This is used for limit rules. 291 */ 292typedef struct _ipfw_insn_limit { 293 ipfw_insn o; 294 u_int8_t _pad; 295 u_int8_t limit_mask; /* combination of DYN_* below */ 296#define DYN_SRC_ADDR 0x1 297#define DYN_SRC_PORT 0x2 298#define DYN_DST_ADDR 0x4 299#define DYN_DST_PORT 0x8 300 301 u_int16_t conn_limit; 302} ipfw_insn_limit; 303 304/* 305 * This is used for log instructions. 306 */ 307typedef struct _ipfw_insn_log { 308 ipfw_insn o; 309 u_int32_t max_log; /* how many do we log -- 0 = all */ 310 u_int32_t log_left; /* how many left to log */ 311} ipfw_insn_log; 312 313/* Apply ipv6 mask on ipv6 addr */ 314#define APPLY_MASK(addr,mask) \ 315 (addr)->__u6_addr.__u6_addr32[0] &= (mask)->__u6_addr.__u6_addr32[0]; \ 316 (addr)->__u6_addr.__u6_addr32[1] &= (mask)->__u6_addr.__u6_addr32[1]; \ 317 (addr)->__u6_addr.__u6_addr32[2] &= (mask)->__u6_addr.__u6_addr32[2]; \ 318 (addr)->__u6_addr.__u6_addr32[3] &= (mask)->__u6_addr.__u6_addr32[3]; 319 320/* Structure for ipv6 */ 321typedef struct _ipfw_insn_ip6 { 322 ipfw_insn o; 323 struct in6_addr addr6; 324 struct in6_addr mask6; 325} ipfw_insn_ip6; 326 327/* Used to support icmp6 types */ 328typedef struct _ipfw_insn_icmp6 { 329 ipfw_insn o; 330 uint32_t d[7]; /* XXX This number si related to the netinet/icmp6.h 331 * define ICMP6_MAXTYPE 332 * as follows: n = ICMP6_MAXTYPE/32 + 1 333 * Actually is 203 334 */ 335} ipfw_insn_icmp6; 336 337/* 338 * Here we have the structure representing an ipfw rule. 339 * 340 * It starts with a general area (with link fields and counters) 341 * followed by an array of one or more instructions, which the code 342 * accesses as an array of 32-bit values. 343 * 344 * Given a rule pointer r: 345 * 346 * r->cmd is the start of the first instruction. 347 * ACTION_PTR(r) is the start of the first action (things to do 348 * once a rule matched). 349 * 350 * When assembling instruction, remember the following: 351 * 352 * + if a rule has a "keep-state" (or "limit") option, then the 353 * first instruction (at r->cmd) MUST BE an O_PROBE_STATE 354 * + if a rule has a "log" option, then the first action 355 * (at ACTION_PTR(r)) MUST be O_LOG 356 * + if a rule has an "altq" option, it comes after "log" 357 * 358 * NOTE: we use a simple linked list of rules because we never need 359 * to delete a rule without scanning the list. We do not use 360 * queue(3) macros for portability and readability. 361 */ 362 363struct ip_fw { 364 struct ip_fw *next; /* linked list of rules */ 365 struct ip_fw *next_rule; /* ptr to next [skipto] rule */ 366 /* 'next_rule' is used to pass up 'set_disable' status */ 367 368 u_int16_t act_ofs; /* offset of action in 32-bit units */ 369 u_int16_t cmd_len; /* # of 32-bit words in cmd */ 370 u_int16_t rulenum; /* rule number */ 371 u_int8_t set; /* rule set (0..31) */ 372#define RESVD_SET 31 /* set for default and persistent rules */ 373 u_int8_t _pad; /* padding */ 374 375 /* These fields are present in all rules. */ 376 u_int64_t pcnt; /* Packet counter */ 377 u_int64_t bcnt; /* Byte counter */ 378 u_int32_t timestamp; /* tv_sec of last match */ 379 380 ipfw_insn cmd[1]; /* storage for commands */ 381}; 382 383#define ACTION_PTR(rule) \ 384 (ipfw_insn *)( (u_int32_t *)((rule)->cmd) + ((rule)->act_ofs) ) 385 386#define RULESIZE(rule) (sizeof(struct ip_fw) + \ 387 ((struct ip_fw *)(rule))->cmd_len * 4 - 4) 388 389/* 390 * This structure is used as a flow mask and a flow id for various 391 * parts of the code. 392 */ 393struct ipfw_flow_id { 394 u_int32_t dst_ip; 395 u_int32_t src_ip; 396 u_int16_t dst_port; 397 u_int16_t src_port; 398 u_int8_t proto; 399 u_int8_t flags; /* protocol-specific flags */ 400 uint8_t addr_type; /* 4 = ipv4, 6 = ipv6, 1=ether ? */ 401 struct in6_addr dst_ip6; /* could also store MAC addr! */ 402 struct in6_addr src_ip6; 403 u_int32_t flow_id6; 404}; 405 406#define IS_IP6_FLOW_ID(id) ((id)->addr_type == 6) 407 408/* 409 * Dynamic ipfw rule. 410 */ 411typedef struct _ipfw_dyn_rule ipfw_dyn_rule; 412 413struct _ipfw_dyn_rule { 414 ipfw_dyn_rule *next; /* linked list of rules. */ 415 struct ip_fw *rule; /* pointer to rule */ 416 /* 'rule' is used to pass up the rule number (from the parent) */ 417 418 ipfw_dyn_rule *parent; /* pointer to parent rule */ 419 u_int64_t pcnt; /* packet match counter */ 420 u_int64_t bcnt; /* byte match counter */ 421 struct ipfw_flow_id id; /* (masked) flow id */ 422 u_int32_t expire; /* expire time */ 423 u_int32_t bucket; /* which bucket in hash table */ 424 u_int32_t state; /* state of this rule (typically a 425 * combination of TCP flags) 426 */ 427 u_int32_t ack_fwd; /* most recent ACKs in forward */ 428 u_int32_t ack_rev; /* and reverse directions (used */ 429 /* to generate keepalives) */ 430 u_int16_t dyn_type; /* rule type */ 431 u_int16_t count; /* refcount */ 432}; 433 434/* 435 * Definitions for IP option names. 436 */ 437#define IP_FW_IPOPT_LSRR 0x01 438#define IP_FW_IPOPT_SSRR 0x02 439#define IP_FW_IPOPT_RR 0x04 440#define IP_FW_IPOPT_TS 0x08 441 442/* 443 * Definitions for TCP option names. 444 */ 445#define IP_FW_TCPOPT_MSS 0x01 446#define IP_FW_TCPOPT_WINDOW 0x02 447#define IP_FW_TCPOPT_SACK 0x04 448#define IP_FW_TCPOPT_TS 0x08 449#define IP_FW_TCPOPT_CC 0x10 450 451#define ICMP_REJECT_RST 0x100 /* fake ICMP code (send a TCP RST) */ 452 453/* 454 * These are used for lookup tables. 455 */ 456typedef struct _ipfw_table_entry { 457 in_addr_t addr; /* network address */ 458 u_int32_t value; /* value */ 459 u_int16_t tbl; /* table number */ 460 u_int8_t masklen; /* mask length */ 461} ipfw_table_entry; 462 463typedef struct _ipfw_table { 464 u_int32_t size; /* size of entries in bytes */ 465 u_int32_t cnt; /* # of entries */ 466 u_int16_t tbl; /* table number */ 467 ipfw_table_entry ent[0]; /* entries */ 468} ipfw_table; 469 470/* 471 * Main firewall chains definitions and global var's definitions. 472 */ 473#ifdef _KERNEL 474 475/* Return values from ipfw_chk() */ 476enum { 477 IP_FW_PASS = 0, 478 IP_FW_DENY, 479 IP_FW_DIVERT, 480 IP_FW_TEE, 481 IP_FW_DUMMYNET, 482 IP_FW_NETGRAPH, 483 IP_FW_NGTEE, 484}; 485 486/* flags for divert mtag */ 487#define IP_FW_DIVERT_LOOPBACK_FLAG 0x00080000 488#define IP_FW_DIVERT_OUTPUT_FLAG 0x00100000 489 490/* 491 * Structure for collecting parameters to dummynet for ip6_output forwarding 492 */ 493struct _ip6dn_args { 494 struct ip6_pktopts *opt_or; 495 struct route_in6 ro_or; 496 int flags_or; 497 struct ip6_moptions *im6o_or; 498 struct ifnet *origifp_or; 499 struct ifnet *ifp_or; 500 struct sockaddr_in6 dst_or; 501 u_long mtu_or; 502 struct route_in6 ro_pmtu_or; 503}; 504 505/* 506 * Arguments for calling ipfw_chk() and dummynet_io(). We put them 507 * all into a structure because this way it is easier and more 508 * efficient to pass variables around and extend the interface. 509 */ 510struct ip_fw_args { 511 struct mbuf *m; /* the mbuf chain */ 512 struct ifnet *oif; /* output interface */ 513 struct sockaddr_in *next_hop; /* forward address */ 514 struct ip_fw *rule; /* matching rule */ 515 struct ether_header *eh; /* for bridged packets */ 516 517 int flags; /* for dummynet */ 518 519 struct ipfw_flow_id f_id; /* grabbed from IP header */ 520 u_int32_t cookie; /* a cookie depending on rule action */ 521 struct inpcb *inp; 522 523 struct _ip6dn_args dummypar; /* dummynet->ip6_output */ 524}; 525 526/* 527 * Function definitions. 528 */ 529 530/* Firewall hooks */ 531struct sockopt; 532struct dn_flow_set; 533 534int ipfw_check_in(void *, struct mbuf **, struct ifnet *, int, struct inpcb *inp); 535int ipfw_check_out(void *, struct mbuf **, struct ifnet *, int, struct inpcb *inp); 536 537int ipfw_chk(struct ip_fw_args *); 538 539int ipfw_init(void); 540void ipfw_destroy(void); 541 542void flush_pipe_ptrs(struct dn_flow_set *match); /* used by dummynet */ 543 544typedef int ip_fw_ctl_t(struct sockopt *); 545extern ip_fw_ctl_t *ip_fw_ctl_ptr; 546extern int fw_one_pass; 547extern int fw_enable; 548 549/* For kernel ipfw_ether and ipfw_bridge. */ 550typedef int ip_fw_chk_t(struct ip_fw_args *args); 551extern ip_fw_chk_t *ip_fw_chk_ptr; 552#define IPFW_LOADED (ip_fw_chk_ptr != NULL) 553 554#endif /* _KERNEL */ 555#endif /* _IPFW2_H */ 556