ip_fw.h revision 134022
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: head/sys/netinet/ip_fw.h 134022 2004-08-19 17:38:47Z andre $ 26 */ 27 28#ifndef _IPFW2_H 29#define _IPFW2_H 30#define IPFW2 1 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 metadata) 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 * Note that the code is written so that individual instructions 42 * have a size which is a multiple of 32 bits. This means that, if 43 * such structures contain pointers or other 64-bit entities, 44 * (there is just one instance now) they may end up unaligned on 45 * 64-bit architectures, so the must be handled with care. 46 * 47 * "enum ipfw_opcodes" are the opcodes supported. We can have up 48 * to 256 different opcodes. When adding new opcodes, they should 49 * be appended to the end of the opcode list before O_LAST_OPCODE, 50 * this will prevent the ABI from being broken, otherwise users 51 * will have to recompile ipfw(8) when they update the kernel. 52 */ 53 54enum ipfw_opcodes { /* arguments (4 byte each) */ 55 O_NOP, 56 57 O_IP_SRC, /* u32 = IP */ 58 O_IP_SRC_MASK, /* ip = IP/mask */ 59 O_IP_SRC_ME, /* none */ 60 O_IP_SRC_SET, /* u32=base, arg1=len, bitmap */ 61 62 O_IP_DST, /* u32 = IP */ 63 O_IP_DST_MASK, /* ip = IP/mask */ 64 O_IP_DST_ME, /* none */ 65 O_IP_DST_SET, /* u32=base, arg1=len, bitmap */ 66 67 O_IP_SRCPORT, /* (n)port list:mask 4 byte ea */ 68 O_IP_DSTPORT, /* (n)port list:mask 4 byte ea */ 69 O_PROTO, /* arg1=protocol */ 70 71 O_MACADDR2, /* 2 mac addr:mask */ 72 O_MAC_TYPE, /* same as srcport */ 73 74 O_LAYER2, /* none */ 75 O_IN, /* none */ 76 O_FRAG, /* none */ 77 78 O_RECV, /* none */ 79 O_XMIT, /* none */ 80 O_VIA, /* none */ 81 82 O_IPOPT, /* arg1 = 2*u8 bitmap */ 83 O_IPLEN, /* arg1 = len */ 84 O_IPID, /* arg1 = id */ 85 86 O_IPTOS, /* arg1 = id */ 87 O_IPPRECEDENCE, /* arg1 = precedence << 5 */ 88 O_IPTTL, /* arg1 = TTL */ 89 90 O_IPVER, /* arg1 = version */ 91 O_UID, /* u32 = id */ 92 O_GID, /* u32 = id */ 93 O_ESTAB, /* none (tcp established) */ 94 O_TCPFLAGS, /* arg1 = 2*u8 bitmap */ 95 O_TCPWIN, /* arg1 = desired win */ 96 O_TCPSEQ, /* u32 = desired seq. */ 97 O_TCPACK, /* u32 = desired seq. */ 98 O_ICMPTYPE, /* u32 = icmp bitmap */ 99 O_TCPOPTS, /* arg1 = 2*u8 bitmap */ 100 101 O_VERREVPATH, /* none */ 102 O_VERSRCREACH, /* none */ 103 104 O_PROBE_STATE, /* none */ 105 O_KEEP_STATE, /* none */ 106 O_LIMIT, /* ipfw_insn_limit */ 107 O_LIMIT_PARENT, /* dyn_type, not an opcode. */ 108 109 /* 110 * These are really 'actions'. 111 */ 112 113 O_LOG, /* ipfw_insn_log */ 114 O_PROB, /* u32 = match probability */ 115 116 O_CHECK_STATE, /* none */ 117 O_ACCEPT, /* none */ 118 O_DENY, /* none */ 119 O_REJECT, /* arg1=icmp arg (same as deny) */ 120 O_COUNT, /* none */ 121 O_SKIPTO, /* arg1=next rule number */ 122 O_PIPE, /* arg1=pipe number */ 123 O_QUEUE, /* arg1=queue number */ 124 O_DIVERT, /* arg1=port number */ 125 O_TEE, /* arg1=port number */ 126 O_FORWARD_IP, /* fwd sockaddr */ 127 O_FORWARD_MAC, /* fwd mac */ 128 129 /* 130 * More opcodes. 131 */ 132 O_IPSEC, /* has ipsec history */ 133 O_IP_SRC_LOOKUP, /* arg1=table number, u32=value */ 134 O_IP_DST_LOOKUP, /* arg1=table number, u32=value */ 135 O_ANTISPOOF, /* none */ 136 O_JAIL, /* u32 = id */ 137 138 O_LAST_OPCODE /* not an opcode! */ 139}; 140 141/* 142 * Template for instructions. 143 * 144 * ipfw_insn is used for all instructions which require no operands, 145 * a single 16-bit value (arg1), or a couple of 8-bit values. 146 * 147 * For other instructions which require different/larger arguments 148 * we have derived structures, ipfw_insn_*. 149 * 150 * The size of the instruction (in 32-bit words) is in the low 151 * 6 bits of "len". The 2 remaining bits are used to implement 152 * NOT and OR on individual instructions. Given a type, you can 153 * compute the length to be put in "len" using F_INSN_SIZE(t) 154 * 155 * F_NOT negates the match result of the instruction. 156 * 157 * F_OR is used to build or blocks. By default, instructions 158 * are evaluated as part of a logical AND. An "or" block 159 * { X or Y or Z } contains F_OR set in all but the last 160 * instruction of the block. A match will cause the code 161 * to skip past the last instruction of the block. 162 * 163 * NOTA BENE: in a couple of places we assume that 164 * sizeof(ipfw_insn) == sizeof(u_int32_t) 165 * this needs to be fixed. 166 * 167 */ 168typedef struct _ipfw_insn { /* template for instructions */ 169 enum ipfw_opcodes opcode:8; 170 u_int8_t len; /* numer of 32-byte words */ 171#define F_NOT 0x80 172#define F_OR 0x40 173#define F_LEN_MASK 0x3f 174#define F_LEN(cmd) ((cmd)->len & F_LEN_MASK) 175 176 u_int16_t arg1; 177} ipfw_insn; 178 179/* 180 * The F_INSN_SIZE(type) computes the size, in 4-byte words, of 181 * a given type. 182 */ 183#define F_INSN_SIZE(t) ((sizeof (t))/sizeof(u_int32_t)) 184 185/* 186 * This is used to store an array of 16-bit entries (ports etc.) 187 */ 188typedef struct _ipfw_insn_u16 { 189 ipfw_insn o; 190 u_int16_t ports[2]; /* there may be more */ 191} ipfw_insn_u16; 192 193/* 194 * This is used to store an array of 32-bit entries 195 * (uid, single IPv4 addresses etc.) 196 */ 197typedef struct _ipfw_insn_u32 { 198 ipfw_insn o; 199 u_int32_t d[1]; /* one or more */ 200} ipfw_insn_u32; 201 202/* 203 * This is used to store IP addr-mask pairs. 204 */ 205typedef struct _ipfw_insn_ip { 206 ipfw_insn o; 207 struct in_addr addr; 208 struct in_addr mask; 209} ipfw_insn_ip; 210 211/* 212 * This is used to forward to a given address (ip). 213 */ 214typedef struct _ipfw_insn_sa { 215 ipfw_insn o; 216 struct sockaddr_in sa; 217} ipfw_insn_sa; 218 219/* 220 * This is used for MAC addr-mask pairs. 221 */ 222typedef struct _ipfw_insn_mac { 223 ipfw_insn o; 224 u_char addr[12]; /* dst[6] + src[6] */ 225 u_char mask[12]; /* dst[6] + src[6] */ 226} ipfw_insn_mac; 227 228/* 229 * This is used for interface match rules (recv xx, xmit xx). 230 */ 231typedef struct _ipfw_insn_if { 232 ipfw_insn o; 233 union { 234 struct in_addr ip; 235 int glob; 236 } p; 237 char name[IFNAMSIZ]; 238} ipfw_insn_if; 239 240/* 241 * This is used for pipe and queue actions, which need to store 242 * a single pointer (which can have different size on different 243 * architectures. 244 * Note that, because of previous instructions, pipe_ptr might 245 * be unaligned in the overall structure, so it needs to be 246 * manipulated with care. 247 */ 248typedef struct _ipfw_insn_pipe { 249 ipfw_insn o; 250 void *pipe_ptr; /* XXX */ 251} ipfw_insn_pipe; 252 253/* 254 * This is used for limit rules. 255 */ 256typedef struct _ipfw_insn_limit { 257 ipfw_insn o; 258 u_int8_t _pad; 259 u_int8_t limit_mask; /* combination of DYN_* below */ 260#define DYN_SRC_ADDR 0x1 261#define DYN_SRC_PORT 0x2 262#define DYN_DST_ADDR 0x4 263#define DYN_DST_PORT 0x8 264 265 u_int16_t conn_limit; 266} ipfw_insn_limit; 267 268/* 269 * This is used for log instructions. 270 */ 271typedef struct _ipfw_insn_log { 272 ipfw_insn o; 273 u_int32_t max_log; /* how many do we log -- 0 = all */ 274 u_int32_t log_left; /* how many left to log */ 275} ipfw_insn_log; 276 277/* 278 * Here we have the structure representing an ipfw rule. 279 * 280 * It starts with a general area (with link fields and counters) 281 * followed by an array of one or more instructions, which the code 282 * accesses as an array of 32-bit values. 283 * 284 * Given a rule pointer r: 285 * 286 * r->cmd is the start of the first instruction. 287 * ACTION_PTR(r) is the start of the first action (things to do 288 * once a rule matched). 289 * 290 * When assembling instruction, remember the following: 291 * 292 * + if a rule has a "keep-state" (or "limit") option, then the 293 * first instruction (at r->cmd) MUST BE an O_PROBE_STATE 294 * + if a rule has a "log" option, then the first action 295 * (at ACTION_PTR(r)) MUST be O_LOG 296 * 297 * NOTE: we use a simple linked list of rules because we never need 298 * to delete a rule without scanning the list. We do not use 299 * queue(3) macros for portability and readability. 300 */ 301 302struct ip_fw { 303 struct ip_fw *next; /* linked list of rules */ 304 struct ip_fw *next_rule; /* ptr to next [skipto] rule */ 305 /* 'next_rule' is used to pass up 'set_disable' status */ 306 307 u_int16_t act_ofs; /* offset of action in 32-bit units */ 308 u_int16_t cmd_len; /* # of 32-bit words in cmd */ 309 u_int16_t rulenum; /* rule number */ 310 u_int8_t set; /* rule set (0..31) */ 311#define RESVD_SET 31 /* set for default and persistent rules */ 312 u_int8_t _pad; /* padding */ 313 314 /* These fields are present in all rules. */ 315 u_int64_t pcnt; /* Packet counter */ 316 u_int64_t bcnt; /* Byte counter */ 317 u_int32_t timestamp; /* tv_sec of last match */ 318 319 ipfw_insn cmd[1]; /* storage for commands */ 320}; 321 322#define ACTION_PTR(rule) \ 323 (ipfw_insn *)( (u_int32_t *)((rule)->cmd) + ((rule)->act_ofs) ) 324 325#define RULESIZE(rule) (sizeof(struct ip_fw) + \ 326 ((struct ip_fw *)(rule))->cmd_len * 4 - 4) 327 328/* 329 * This structure is used as a flow mask and a flow id for various 330 * parts of the code. 331 */ 332struct ipfw_flow_id { 333 u_int32_t dst_ip; 334 u_int32_t src_ip; 335 u_int16_t dst_port; 336 u_int16_t src_port; 337 u_int8_t proto; 338 u_int8_t flags; /* protocol-specific flags */ 339}; 340 341/* 342 * Dynamic ipfw rule. 343 */ 344typedef struct _ipfw_dyn_rule ipfw_dyn_rule; 345 346struct _ipfw_dyn_rule { 347 ipfw_dyn_rule *next; /* linked list of rules. */ 348 struct ip_fw *rule; /* pointer to rule */ 349 /* 'rule' is used to pass up the rule number (from the parent) */ 350 351 ipfw_dyn_rule *parent; /* pointer to parent rule */ 352 u_int64_t pcnt; /* packet match counter */ 353 u_int64_t bcnt; /* byte match counter */ 354 struct ipfw_flow_id id; /* (masked) flow id */ 355 u_int32_t expire; /* expire time */ 356 u_int32_t bucket; /* which bucket in hash table */ 357 u_int32_t state; /* state of this rule (typically a 358 * combination of TCP flags) 359 */ 360 u_int32_t ack_fwd; /* most recent ACKs in forward */ 361 u_int32_t ack_rev; /* and reverse directions (used */ 362 /* to generate keepalives) */ 363 u_int16_t dyn_type; /* rule type */ 364 u_int16_t count; /* refcount */ 365}; 366 367/* 368 * Definitions for IP option names. 369 */ 370#define IP_FW_IPOPT_LSRR 0x01 371#define IP_FW_IPOPT_SSRR 0x02 372#define IP_FW_IPOPT_RR 0x04 373#define IP_FW_IPOPT_TS 0x08 374 375/* 376 * Definitions for TCP option names. 377 */ 378#define IP_FW_TCPOPT_MSS 0x01 379#define IP_FW_TCPOPT_WINDOW 0x02 380#define IP_FW_TCPOPT_SACK 0x04 381#define IP_FW_TCPOPT_TS 0x08 382#define IP_FW_TCPOPT_CC 0x10 383 384#define ICMP_REJECT_RST 0x100 /* fake ICMP code (send a TCP RST) */ 385 386/* 387 * These are used for lookup tables. 388 */ 389typedef struct _ipfw_table_entry { 390 in_addr_t addr; /* network address */ 391 u_int32_t value; /* value */ 392 u_int16_t tbl; /* table number */ 393 u_int8_t masklen; /* mask length */ 394} ipfw_table_entry; 395 396typedef struct _ipfw_table { 397 u_int32_t size; /* size of entries in bytes */ 398 u_int32_t cnt; /* # of entries */ 399 u_int16_t tbl; /* table number */ 400 ipfw_table_entry ent[0]; /* entries */ 401} ipfw_table; 402 403/* 404 * Main firewall chains definitions and global var's definitions. 405 */ 406#ifdef _KERNEL 407 408#define IP_FW_PORT_DYNT_FLAG 0x10000 409#define IP_FW_PORT_TEE_FLAG 0x20000 410#define IP_FW_PORT_DENY_FLAG 0x40000 411 412/* 413 * Arguments for calling ipfw_chk() and dummynet_io(). We put them 414 * all into a structure because this way it is easier and more 415 * efficient to pass variables around and extend the interface. 416 */ 417struct ip_fw_args { 418 struct mbuf *m; /* the mbuf chain */ 419 struct ifnet *oif; /* output interface */ 420 struct sockaddr_in *next_hop; /* forward address */ 421 struct ip_fw *rule; /* matching rule */ 422 struct ether_header *eh; /* for bridged packets */ 423 424 int flags; /* for dummynet */ 425 426 struct ipfw_flow_id f_id; /* grabbed from IP header */ 427 u_int32_t retval; 428}; 429 430/* 431 * Function definitions. 432 */ 433 434/* Firewall hooks */ 435struct sockopt; 436struct dn_flow_set; 437 438int ipfw_check_in(void *, struct mbuf **, struct ifnet *, int); 439int ipfw_check_out(void *, struct mbuf **, struct ifnet *, int); 440 441int ipfw_chk(struct ip_fw_args *); 442 443int ipfw_init(void); 444void ipfw_destroy(void); 445 446void flush_pipe_ptrs(struct dn_flow_set *match); /* used by dummynet */ 447 448typedef int ip_fw_ctl_t(struct sockopt *); 449extern ip_fw_ctl_t *ip_fw_ctl_ptr; 450extern int fw_one_pass; 451extern int fw_enable; 452 453/* For kernel ipfw_ether and ipfw_bridge. */ 454typedef int ip_fw_chk_t(struct ip_fw_args *args); 455extern ip_fw_chk_t *ip_fw_chk_ptr; 456#define IPFW_LOADED (ip_fw_chk_ptr != NULL) 457 458#endif /* _KERNEL */ 459#endif /* _IPFW2_H */ 460