2 * Copyright (c) 2002-2009 Luigi Rizzo, Universita` di Pisa
4 * Redistribution and use in source and binary forms, with or without
5 * modification, are permitted provided that the following conditions
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.
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
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
37 #define IPFW_DEFAULT_RULE 65535
40 * Default number of ipfw tables.
42 #define IPFW_TABLES_MAX 128
45 * Most commands (queue, pipe, tag, untag, limit...) can have a 16-bit
46 * argument between 1 and 65534. The value 0 is unused, the value
47 * 65535 (IP_FW_TABLEARG) is used to represent 'tablearg', i.e. the
48 * can be 1..65534, or 65535 to indicate the use of a 'tablearg'
49 * result of the most recent table() lookup.
50 * Note that 16bit is only a historical limit, resulting from
51 * the use of a 16-bit fields for that value. In reality, we can have
52 * 2^32 pipes, queues, tag values and so on, and use 0 as a tablearg.
54 #define IPFW_ARG_MIN 1
55 #define IPFW_ARG_MAX 65534
56 #define IP_FW_TABLEARG 65535 /* XXX should use 0 */
59 * Number of entries in the call stack of the call/return commands.
60 * Call stack currently is an uint16_t array with rule numbers.
62 #define IPFW_CALLSTACK_SIZE 16
64 /* IP_FW3 header/opcodes */
65 typedef struct _ip_fw3_opheader {
66 uint16_t opcode; /* Operation opcode */
67 uint16_t reserved[3]; /* Align to 64-bit boundary */
71 /* IPFW extented tables support */
72 #define IP_FW_TABLE_XADD 86 /* add entry */
73 #define IP_FW_TABLE_XDEL 87 /* delete entry */
74 #define IP_FW_TABLE_XGETSIZE 88 /* get table size */
75 #define IP_FW_TABLE_XLIST 89 /* list table contents */
78 * The kernel representation of ipfw rules is made of a list of
79 * 'instructions' (for all practical purposes equivalent to BPF
80 * instructions), which specify which fields of the packet
81 * (or its metadata) should be analysed.
83 * Each instruction is stored in a structure which begins with
84 * "ipfw_insn", and can contain extra fields depending on the
85 * instruction type (listed below).
86 * Note that the code is written so that individual instructions
87 * have a size which is a multiple of 32 bits. This means that, if
88 * such structures contain pointers or other 64-bit entities,
89 * (there is just one instance now) they may end up unaligned on
90 * 64-bit architectures, so the must be handled with care.
92 * "enum ipfw_opcodes" are the opcodes supported. We can have up
93 * to 256 different opcodes. When adding new opcodes, they should
94 * be appended to the end of the opcode list before O_LAST_OPCODE,
95 * this will prevent the ABI from being broken, otherwise users
96 * will have to recompile ipfw(8) when they update the kernel.
99 enum ipfw_opcodes { /* arguments (4 byte each) */
102 O_IP_SRC, /* u32 = IP */
103 O_IP_SRC_MASK, /* ip = IP/mask */
104 O_IP_SRC_ME, /* none */
105 O_IP_SRC_SET, /* u32=base, arg1=len, bitmap */
107 O_IP_DST, /* u32 = IP */
108 O_IP_DST_MASK, /* ip = IP/mask */
109 O_IP_DST_ME, /* none */
110 O_IP_DST_SET, /* u32=base, arg1=len, bitmap */
112 O_IP_SRCPORT, /* (n)port list:mask 4 byte ea */
113 O_IP_DSTPORT, /* (n)port list:mask 4 byte ea */
114 O_PROTO, /* arg1=protocol */
116 O_MACADDR2, /* 2 mac addr:mask */
117 O_MAC_TYPE, /* same as srcport */
127 O_IPOPT, /* arg1 = 2*u8 bitmap */
128 O_IPLEN, /* arg1 = len */
129 O_IPID, /* arg1 = id */
131 O_IPTOS, /* arg1 = id */
132 O_IPPRECEDENCE, /* arg1 = precedence << 5 */
133 O_IPTTL, /* arg1 = TTL */
135 O_IPVER, /* arg1 = version */
136 O_UID, /* u32 = id */
137 O_GID, /* u32 = id */
138 O_ESTAB, /* none (tcp established) */
139 O_TCPFLAGS, /* arg1 = 2*u8 bitmap */
140 O_TCPWIN, /* arg1 = desired win */
141 O_TCPSEQ, /* u32 = desired seq. */
142 O_TCPACK, /* u32 = desired seq. */
143 O_ICMPTYPE, /* u32 = icmp bitmap */
144 O_TCPOPTS, /* arg1 = 2*u8 bitmap */
146 O_VERREVPATH, /* none */
147 O_VERSRCREACH, /* none */
149 O_PROBE_STATE, /* none */
150 O_KEEP_STATE, /* none */
151 O_LIMIT, /* ipfw_insn_limit */
152 O_LIMIT_PARENT, /* dyn_type, not an opcode. */
155 * These are really 'actions'.
158 O_LOG, /* ipfw_insn_log */
159 O_PROB, /* u32 = match probability */
161 O_CHECK_STATE, /* none */
164 O_REJECT, /* arg1=icmp arg (same as deny) */
166 O_SKIPTO, /* arg1=next rule number */
167 O_PIPE, /* arg1=pipe number */
168 O_QUEUE, /* arg1=queue number */
169 O_DIVERT, /* arg1=port number */
170 O_TEE, /* arg1=port number */
171 O_FORWARD_IP, /* fwd sockaddr */
172 O_FORWARD_MAC, /* fwd mac */
179 O_IPSEC, /* has ipsec history */
180 O_IP_SRC_LOOKUP, /* arg1=table number, u32=value */
181 O_IP_DST_LOOKUP, /* arg1=table number, u32=value */
182 O_ANTISPOOF, /* none */
183 O_JAIL, /* u32 = id */
184 O_ALTQ, /* u32 = altq classif. qid */
185 O_DIVERTED, /* arg1=bitmap (1:loop, 2:out) */
186 O_TCPDATALEN, /* arg1 = tcp data len */
187 O_IP6_SRC, /* address without mask */
188 O_IP6_SRC_ME, /* my addresses */
189 O_IP6_SRC_MASK, /* address with the mask */
193 O_FLOW6ID, /* for flow id tag in the ipv6 pkt */
194 O_ICMP6TYPE, /* icmp6 packet type filtering */
195 O_EXT_HDR, /* filtering for ipv6 extension header */
199 * actions for ng_ipfw
201 O_NETGRAPH, /* send to ng_ipfw */
202 O_NGTEE, /* copy to ng_ipfw */
206 O_UNREACH6, /* arg1=icmpv6 code arg (deny) */
208 O_TAG, /* arg1=tag number */
209 O_TAGGED, /* arg1=tag number */
211 O_SETFIB, /* arg1=FIB number */
212 O_FIB, /* arg1=FIB desired fib number */
214 O_SOCKARG, /* socket argument */
216 O_CALLRETURN, /* arg1=called rule number */
218 O_FORWARD_IP6, /* fwd sockaddr_in6 */
220 O_LAST_OPCODE /* not an opcode! */
225 * The extension header are filtered only for presence using a bit
226 * vector with a flag for each header.
228 #define EXT_FRAGMENT 0x1
229 #define EXT_HOPOPTS 0x2
230 #define EXT_ROUTING 0x4
233 #define EXT_DSTOPTS 0x20
234 #define EXT_RTHDR0 0x40
235 #define EXT_RTHDR2 0x80
238 * Template for instructions.
240 * ipfw_insn is used for all instructions which require no operands,
241 * a single 16-bit value (arg1), or a couple of 8-bit values.
243 * For other instructions which require different/larger arguments
244 * we have derived structures, ipfw_insn_*.
246 * The size of the instruction (in 32-bit words) is in the low
247 * 6 bits of "len". The 2 remaining bits are used to implement
248 * NOT and OR on individual instructions. Given a type, you can
249 * compute the length to be put in "len" using F_INSN_SIZE(t)
251 * F_NOT negates the match result of the instruction.
253 * F_OR is used to build or blocks. By default, instructions
254 * are evaluated as part of a logical AND. An "or" block
255 * { X or Y or Z } contains F_OR set in all but the last
256 * instruction of the block. A match will cause the code
257 * to skip past the last instruction of the block.
259 * NOTA BENE: in a couple of places we assume that
260 * sizeof(ipfw_insn) == sizeof(u_int32_t)
261 * this needs to be fixed.
264 typedef struct _ipfw_insn { /* template for instructions */
266 u_int8_t len; /* number of 32-bit words */
269 #define F_LEN_MASK 0x3f
270 #define F_LEN(cmd) ((cmd)->len & F_LEN_MASK)
276 * The F_INSN_SIZE(type) computes the size, in 4-byte words, of
279 #define F_INSN_SIZE(t) ((sizeof (t))/sizeof(u_int32_t))
282 * This is used to store an array of 16-bit entries (ports etc.)
284 typedef struct _ipfw_insn_u16 {
286 u_int16_t ports[2]; /* there may be more */
290 * This is used to store an array of 32-bit entries
291 * (uid, single IPv4 addresses etc.)
293 typedef struct _ipfw_insn_u32 {
295 u_int32_t d[1]; /* one or more */
299 * This is used to store IP addr-mask pairs.
301 typedef struct _ipfw_insn_ip {
308 * This is used to forward to a given address (ip).
310 typedef struct _ipfw_insn_sa {
312 struct sockaddr_in sa;
316 * This is used to forward to a given address (ipv6).
318 typedef struct _ipfw_insn_sa6 {
320 struct sockaddr_in6 sa;
324 * This is used for MAC addr-mask pairs.
326 typedef struct _ipfw_insn_mac {
328 u_char addr[12]; /* dst[6] + src[6] */
329 u_char mask[12]; /* dst[6] + src[6] */
333 * This is used for interface match rules (recv xx, xmit xx).
335 typedef struct _ipfw_insn_if {
345 * This is used for storing an altq queue id number.
347 typedef struct _ipfw_insn_altq {
353 * This is used for limit rules.
355 typedef struct _ipfw_insn_limit {
358 u_int8_t limit_mask; /* combination of DYN_* below */
359 #define DYN_SRC_ADDR 0x1
360 #define DYN_SRC_PORT 0x2
361 #define DYN_DST_ADDR 0x4
362 #define DYN_DST_PORT 0x8
364 u_int16_t conn_limit;
368 * This is used for log instructions.
370 typedef struct _ipfw_insn_log {
372 u_int32_t max_log; /* how many do we log -- 0 = all */
373 u_int32_t log_left; /* how many left to log */
377 * Data structures required by both ipfw(8) and ipfw(4) but not part of the
378 * management API are protected by IPFW_INTERNAL.
381 /* Server pool support (LSNAT). */
383 LIST_ENTRY(cfg_spool) _next; /* chain of spool instances */
389 /* Redirect modes id. */
390 #define REDIR_ADDR 0x01
391 #define REDIR_PORT 0x02
392 #define REDIR_PROTO 0x04
395 /* Nat redirect configuration. */
397 LIST_ENTRY(cfg_redir) _next; /* chain of redir instances */
398 u_int16_t mode; /* type of redirect mode */
399 struct in_addr laddr; /* local ip address */
400 struct in_addr paddr; /* public ip address */
401 struct in_addr raddr; /* remote ip address */
402 u_short lport; /* local port */
403 u_short pport; /* public port */
404 u_short rport; /* remote port */
405 u_short pport_cnt; /* number of public ports */
406 u_short rport_cnt; /* number of remote ports */
407 int proto; /* protocol: tcp/udp */
408 struct alias_link **alink;
409 /* num of entry in spool chain */
411 /* chain of spool instances */
412 LIST_HEAD(spool_chain, cfg_spool) spool_chain;
417 /* Nat configuration data struct. */
419 /* chain of nat instances */
420 LIST_ENTRY(cfg_nat) _next;
422 struct in_addr ip; /* nat ip address */
423 char if_name[IF_NAMESIZE]; /* interface name */
424 int mode; /* aliasing mode */
425 struct libalias *lib; /* libalias instance */
426 /* number of entry in spool chain */
428 /* chain of redir instances */
429 LIST_HEAD(redir_chain, cfg_redir) redir_chain;
433 #define SOF_NAT sizeof(struct cfg_nat)
434 #define SOF_REDIR sizeof(struct cfg_redir)
435 #define SOF_SPOOL sizeof(struct cfg_spool)
438 typedef struct _ipfw_insn_nat {
443 /* Apply ipv6 mask on ipv6 addr */
444 #define APPLY_MASK(addr,mask) \
445 (addr)->__u6_addr.__u6_addr32[0] &= (mask)->__u6_addr.__u6_addr32[0]; \
446 (addr)->__u6_addr.__u6_addr32[1] &= (mask)->__u6_addr.__u6_addr32[1]; \
447 (addr)->__u6_addr.__u6_addr32[2] &= (mask)->__u6_addr.__u6_addr32[2]; \
448 (addr)->__u6_addr.__u6_addr32[3] &= (mask)->__u6_addr.__u6_addr32[3];
450 /* Structure for ipv6 */
451 typedef struct _ipfw_insn_ip6 {
453 struct in6_addr addr6;
454 struct in6_addr mask6;
457 /* Used to support icmp6 types */
458 typedef struct _ipfw_insn_icmp6 {
460 uint32_t d[7]; /* XXX This number si related to the netinet/icmp6.h
461 * define ICMP6_MAXTYPE
462 * as follows: n = ICMP6_MAXTYPE/32 + 1
468 * Here we have the structure representing an ipfw rule.
470 * It starts with a general area (with link fields and counters)
471 * followed by an array of one or more instructions, which the code
472 * accesses as an array of 32-bit values.
474 * Given a rule pointer r:
476 * r->cmd is the start of the first instruction.
477 * ACTION_PTR(r) is the start of the first action (things to do
478 * once a rule matched).
480 * When assembling instruction, remember the following:
482 * + if a rule has a "keep-state" (or "limit") option, then the
483 * first instruction (at r->cmd) MUST BE an O_PROBE_STATE
484 * + if a rule has a "log" option, then the first action
485 * (at ACTION_PTR(r)) MUST be O_LOG
486 * + if a rule has an "altq" option, it comes after "log"
487 * + if a rule has an O_TAG option, it comes after "log" and "altq"
489 * NOTE: we use a simple linked list of rules because we never need
490 * to delete a rule without scanning the list. We do not use
491 * queue(3) macros for portability and readability.
495 struct ip_fw *x_next; /* linked list of rules */
496 struct ip_fw *next_rule; /* ptr to next [skipto] rule */
497 /* 'next_rule' is used to pass up 'set_disable' status */
499 uint16_t act_ofs; /* offset of action in 32-bit units */
500 uint16_t cmd_len; /* # of 32-bit words in cmd */
501 uint16_t rulenum; /* rule number */
502 uint8_t set; /* rule set (0..31) */
503 #define RESVD_SET 31 /* set for default and persistent rules */
504 uint8_t _pad; /* padding */
505 uint32_t id; /* rule id */
507 /* These fields are present in all rules. */
508 uint64_t pcnt; /* Packet counter */
509 uint64_t bcnt; /* Byte counter */
510 uint32_t timestamp; /* tv_sec of last match */
512 ipfw_insn cmd[1]; /* storage for commands */
515 #define ACTION_PTR(rule) \
516 (ipfw_insn *)( (u_int32_t *)((rule)->cmd) + ((rule)->act_ofs) )
518 #define RULESIZE(rule) (sizeof(struct ip_fw) + \
519 ((struct ip_fw *)(rule))->cmd_len * 4 - 4)
521 #if 1 // should be moved to in.h
523 * This structure is used as a flow mask and a flow id for various
525 * addr_type is used in userland and kernel to mark the address type.
526 * fib is used in the kernel to record the fib in use.
527 * _flags is used in the kernel to store tcp flags for dynamic rules.
529 struct ipfw_flow_id {
536 uint8_t _flags; /* protocol-specific flags */
537 uint8_t addr_type; /* 4=ip4, 6=ip6, 1=ether ? */
538 struct in6_addr dst_ip6;
539 struct in6_addr src_ip6;
541 uint32_t extra; /* queue/pipe or frag_id */
545 #define IS_IP6_FLOW_ID(id) ((id)->addr_type == 6)
550 typedef struct _ipfw_dyn_rule ipfw_dyn_rule;
552 struct _ipfw_dyn_rule {
553 ipfw_dyn_rule *next; /* linked list of rules. */
554 struct ip_fw *rule; /* pointer to rule */
555 /* 'rule' is used to pass up the rule number (from the parent) */
557 ipfw_dyn_rule *parent; /* pointer to parent rule */
558 u_int64_t pcnt; /* packet match counter */
559 u_int64_t bcnt; /* byte match counter */
560 struct ipfw_flow_id id; /* (masked) flow id */
561 u_int32_t expire; /* expire time */
562 u_int32_t bucket; /* which bucket in hash table */
563 u_int32_t state; /* state of this rule (typically a
564 * combination of TCP flags)
566 u_int32_t ack_fwd; /* most recent ACKs in forward */
567 u_int32_t ack_rev; /* and reverse directions (used */
568 /* to generate keepalives) */
569 u_int16_t dyn_type; /* rule type */
570 u_int16_t count; /* refcount */
574 * Definitions for IP option names.
576 #define IP_FW_IPOPT_LSRR 0x01
577 #define IP_FW_IPOPT_SSRR 0x02
578 #define IP_FW_IPOPT_RR 0x04
579 #define IP_FW_IPOPT_TS 0x08
582 * Definitions for TCP option names.
584 #define IP_FW_TCPOPT_MSS 0x01
585 #define IP_FW_TCPOPT_WINDOW 0x02
586 #define IP_FW_TCPOPT_SACK 0x04
587 #define IP_FW_TCPOPT_TS 0x08
588 #define IP_FW_TCPOPT_CC 0x10
590 #define ICMP_REJECT_RST 0x100 /* fake ICMP code (send a TCP RST) */
591 #define ICMP6_UNREACH_RST 0x100 /* fake ICMPv6 code (send a TCP RST) */
594 * These are used for lookup tables.
597 #define IPFW_TABLE_CIDR 1 /* Table for holding IPv4/IPv6 prefixes */
598 #define IPFW_TABLE_INTERFACE 2 /* Table for holding interface names */
599 #define IPFW_TABLE_MAXTYPE 2 /* Maximum valid number */
601 typedef struct _ipfw_table_entry {
602 in_addr_t addr; /* network address */
603 u_int32_t value; /* value */
604 u_int16_t tbl; /* table number */
605 u_int8_t masklen; /* mask length */
608 typedef struct _ipfw_table_xentry {
609 uint16_t len; /* Total entry length */
610 uint8_t type; /* entry type */
611 uint8_t masklen; /* mask length */
612 uint16_t tbl; /* table number */
613 uint32_t value; /* value */
615 /* Longest field needs to be aligned by 4-byte boundary */
616 struct in6_addr addr6; /* IPv6 address */
617 char iface[IF_NAMESIZE]; /* interface name */
621 typedef struct _ipfw_table {
622 u_int32_t size; /* size of entries in bytes */
623 u_int32_t cnt; /* # of entries */
624 u_int16_t tbl; /* table number */
625 ipfw_table_entry ent[0]; /* entries */
628 typedef struct _ipfw_xtable {
629 ip_fw3_opheader opheader; /* eXtended tables are controlled via IP_FW3 */
630 uint32_t size; /* size of entries in bytes */
631 uint32_t cnt; /* # of entries */
632 uint16_t tbl; /* table number */
633 uint8_t type; /* table type */
634 ipfw_table_xentry xent[0]; /* entries */
637 #endif /* _IPFW2_H */