2 * Copyright (c) 2002 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 kernel representation of ipfw rules is made of a list of
33 * 'instructions' (for all practical purposes equivalent to BPF
34 * instructions), which specify which fields of the packet
35 * (or its metadata) should be analysed.
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.
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.
53 enum ipfw_opcodes { /* arguments (4 byte each) */
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 */
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 */
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 */
70 O_MACADDR2, /* 2 mac addr:mask */
71 O_MAC_TYPE, /* same as srcport */
81 O_IPOPT, /* arg1 = 2*u8 bitmap */
82 O_IPLEN, /* arg1 = len */
83 O_IPID, /* arg1 = id */
85 O_IPTOS, /* arg1 = id */
86 O_IPPRECEDENCE, /* arg1 = precedence << 5 */
87 O_IPTTL, /* arg1 = TTL */
89 O_IPVER, /* arg1 = version */
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 */
100 O_VERREVPATH, /* none */
101 O_VERSRCREACH, /* none */
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. */
109 * These are really 'actions'.
112 O_LOG, /* ipfw_insn_log */
113 O_PROB, /* u32 = match probability */
115 O_CHECK_STATE, /* none */
118 O_REJECT, /* arg1=icmp arg (same as deny) */
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 */
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 O_ALTQ, /* u32 = altq classif. qid */
138 O_DIVERTED, /* arg1=bitmap (1:loop, 2:out) */
139 O_TCPDATALEN, /* arg1 = tcp data len */
140 O_IP6_SRC, /* address without mask */
141 O_IP6_SRC_ME, /* my addresses */
142 O_IP6_SRC_MASK, /* address with the mask */
146 O_FLOW6ID, /* for flow id tag in the ipv6 pkt */
147 O_ICMP6TYPE, /* icmp6 packet type filtering */
148 O_EXT_HDR, /* filtering for ipv6 extension header */
152 * actions for ng_ipfw
154 O_NETGRAPH, /* send to ng_ipfw */
155 O_NGTEE, /* copy to ng_ipfw */
159 O_UNREACH6, /* arg1=icmpv6 code arg (deny) */
161 O_TAG, /* arg1=tag number */
162 O_TAGGED, /* arg1=tag number */
164 O_LAST_OPCODE /* not an opcode! */
168 * The extension header are filtered only for presence using a bit
169 * vector with a flag for each header.
171 #define EXT_FRAGMENT 0x1
172 #define EXT_HOPOPTS 0x2
173 #define EXT_ROUTING 0x4
176 #define EXT_DSTOPTS 0x20
177 #define EXT_RTHDR0 0x40
178 #define EXT_RTHDR2 0x80
181 * Template for instructions.
183 * ipfw_insn is used for all instructions which require no operands,
184 * a single 16-bit value (arg1), or a couple of 8-bit values.
186 * For other instructions which require different/larger arguments
187 * we have derived structures, ipfw_insn_*.
189 * The size of the instruction (in 32-bit words) is in the low
190 * 6 bits of "len". The 2 remaining bits are used to implement
191 * NOT and OR on individual instructions. Given a type, you can
192 * compute the length to be put in "len" using F_INSN_SIZE(t)
194 * F_NOT negates the match result of the instruction.
196 * F_OR is used to build or blocks. By default, instructions
197 * are evaluated as part of a logical AND. An "or" block
198 * { X or Y or Z } contains F_OR set in all but the last
199 * instruction of the block. A match will cause the code
200 * to skip past the last instruction of the block.
202 * NOTA BENE: in a couple of places we assume that
203 * sizeof(ipfw_insn) == sizeof(u_int32_t)
204 * this needs to be fixed.
207 typedef struct _ipfw_insn { /* template for instructions */
208 enum ipfw_opcodes opcode:8;
209 u_int8_t len; /* numer of 32-byte words */
212 #define F_LEN_MASK 0x3f
213 #define F_LEN(cmd) ((cmd)->len & F_LEN_MASK)
219 * The F_INSN_SIZE(type) computes the size, in 4-byte words, of
222 #define F_INSN_SIZE(t) ((sizeof (t))/sizeof(u_int32_t))
224 #define MTAG_IPFW 1148380143 /* IPFW-tagged cookie */
227 * This is used to store an array of 16-bit entries (ports etc.)
229 typedef struct _ipfw_insn_u16 {
231 u_int16_t ports[2]; /* there may be more */
235 * This is used to store an array of 32-bit entries
236 * (uid, single IPv4 addresses etc.)
238 typedef struct _ipfw_insn_u32 {
240 u_int32_t d[1]; /* one or more */
244 * This is used to store IP addr-mask pairs.
246 typedef struct _ipfw_insn_ip {
253 * This is used to forward to a given address (ip).
255 typedef struct _ipfw_insn_sa {
257 struct sockaddr_in sa;
261 * This is used for MAC addr-mask pairs.
263 typedef struct _ipfw_insn_mac {
265 u_char addr[12]; /* dst[6] + src[6] */
266 u_char mask[12]; /* dst[6] + src[6] */
270 * This is used for interface match rules (recv xx, xmit xx).
272 typedef struct _ipfw_insn_if {
282 * This is used for storing an altq queue id number.
284 typedef struct _ipfw_insn_altq {
290 * This is used for limit rules.
292 typedef struct _ipfw_insn_limit {
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
301 u_int16_t conn_limit;
305 * This is used for log instructions.
307 typedef struct _ipfw_insn_log {
309 u_int32_t max_log; /* how many do we log -- 0 = all */
310 u_int32_t log_left; /* how many left to log */
313 /* Server pool support (LSNAT). */
315 LIST_ENTRY(cfg_spool) _next; /* chain of spool instances */
320 /* Redirect modes id. */
321 #define REDIR_ADDR 0x01
322 #define REDIR_PORT 0x02
323 #define REDIR_PROTO 0x04
325 /* Nat redirect configuration. */
327 LIST_ENTRY(cfg_redir) _next; /* chain of redir instances */
328 u_int16_t mode; /* type of redirect mode */
329 struct in_addr laddr; /* local ip address */
330 struct in_addr paddr; /* public ip address */
331 struct in_addr raddr; /* remote ip address */
332 u_short lport; /* local port */
333 u_short pport; /* public port */
334 u_short rport; /* remote port */
335 u_short pport_cnt; /* number of public ports */
336 u_short rport_cnt; /* number of remote ports */
337 int proto; /* protocol: tcp/udp */
338 struct alias_link **alink;
339 /* num of entry in spool chain */
341 /* chain of spool instances */
342 LIST_HEAD(spool_chain, cfg_spool) spool_chain;
345 #define NAT_BUF_LEN 1024
346 /* Nat configuration data struct. */
348 /* chain of nat instances */
349 LIST_ENTRY(cfg_nat) _next;
351 struct in_addr ip; /* nat ip address */
352 char if_name[IF_NAMESIZE]; /* interface name */
353 int mode; /* aliasing mode */
354 struct libalias *lib; /* libalias instance */
355 /* number of entry in spool chain */
357 /* chain of redir instances */
358 LIST_HEAD(redir_chain, cfg_redir) redir_chain;
361 #define SOF_NAT sizeof(struct cfg_nat)
362 #define SOF_REDIR sizeof(struct cfg_redir)
363 #define SOF_SPOOL sizeof(struct cfg_spool)
366 typedef struct _ipfw_insn_nat {
371 /* Apply ipv6 mask on ipv6 addr */
372 #define APPLY_MASK(addr,mask) \
373 (addr)->__u6_addr.__u6_addr32[0] &= (mask)->__u6_addr.__u6_addr32[0]; \
374 (addr)->__u6_addr.__u6_addr32[1] &= (mask)->__u6_addr.__u6_addr32[1]; \
375 (addr)->__u6_addr.__u6_addr32[2] &= (mask)->__u6_addr.__u6_addr32[2]; \
376 (addr)->__u6_addr.__u6_addr32[3] &= (mask)->__u6_addr.__u6_addr32[3];
378 /* Structure for ipv6 */
379 typedef struct _ipfw_insn_ip6 {
381 struct in6_addr addr6;
382 struct in6_addr mask6;
385 /* Used to support icmp6 types */
386 typedef struct _ipfw_insn_icmp6 {
388 uint32_t d[7]; /* XXX This number si related to the netinet/icmp6.h
389 * define ICMP6_MAXTYPE
390 * as follows: n = ICMP6_MAXTYPE/32 + 1
396 * Here we have the structure representing an ipfw rule.
398 * It starts with a general area (with link fields and counters)
399 * followed by an array of one or more instructions, which the code
400 * accesses as an array of 32-bit values.
402 * Given a rule pointer r:
404 * r->cmd is the start of the first instruction.
405 * ACTION_PTR(r) is the start of the first action (things to do
406 * once a rule matched).
408 * When assembling instruction, remember the following:
410 * + if a rule has a "keep-state" (or "limit") option, then the
411 * first instruction (at r->cmd) MUST BE an O_PROBE_STATE
412 * + if a rule has a "log" option, then the first action
413 * (at ACTION_PTR(r)) MUST be O_LOG
414 * + if a rule has an "altq" option, it comes after "log"
415 * + if a rule has an O_TAG option, it comes after "log" and "altq"
417 * NOTE: we use a simple linked list of rules because we never need
418 * to delete a rule without scanning the list. We do not use
419 * queue(3) macros for portability and readability.
423 struct ip_fw *next; /* linked list of rules */
424 struct ip_fw *next_rule; /* ptr to next [skipto] rule */
425 /* 'next_rule' is used to pass up 'set_disable' status */
427 u_int16_t act_ofs; /* offset of action in 32-bit units */
428 u_int16_t cmd_len; /* # of 32-bit words in cmd */
429 u_int16_t rulenum; /* rule number */
430 u_int8_t set; /* rule set (0..31) */
431 #define RESVD_SET 31 /* set for default and persistent rules */
432 u_int8_t _pad; /* padding */
434 /* These fields are present in all rules. */
435 u_int64_t pcnt; /* Packet counter */
436 u_int64_t bcnt; /* Byte counter */
437 u_int32_t timestamp; /* tv_sec of last match */
439 ipfw_insn cmd[1]; /* storage for commands */
442 #define ACTION_PTR(rule) \
443 (ipfw_insn *)( (u_int32_t *)((rule)->cmd) + ((rule)->act_ofs) )
445 #define RULESIZE(rule) (sizeof(struct ip_fw) + \
446 ((struct ip_fw *)(rule))->cmd_len * 4 - 4)
449 * This structure is used as a flow mask and a flow id for various
452 struct ipfw_flow_id {
458 u_int8_t flags; /* protocol-specific flags */
459 uint8_t addr_type; /* 4 = ipv4, 6 = ipv6, 1=ether ? */
460 struct in6_addr dst_ip6; /* could also store MAC addr! */
461 struct in6_addr src_ip6;
466 #define IS_IP6_FLOW_ID(id) ((id)->addr_type == 6)
471 typedef struct _ipfw_dyn_rule ipfw_dyn_rule;
473 struct _ipfw_dyn_rule {
474 ipfw_dyn_rule *next; /* linked list of rules. */
475 struct ip_fw *rule; /* pointer to rule */
476 /* 'rule' is used to pass up the rule number (from the parent) */
478 ipfw_dyn_rule *parent; /* pointer to parent rule */
479 u_int64_t pcnt; /* packet match counter */
480 u_int64_t bcnt; /* byte match counter */
481 struct ipfw_flow_id id; /* (masked) flow id */
482 u_int32_t expire; /* expire time */
483 u_int32_t bucket; /* which bucket in hash table */
484 u_int32_t state; /* state of this rule (typically a
485 * combination of TCP flags)
487 u_int32_t ack_fwd; /* most recent ACKs in forward */
488 u_int32_t ack_rev; /* and reverse directions (used */
489 /* to generate keepalives) */
490 u_int16_t dyn_type; /* rule type */
491 u_int16_t count; /* refcount */
495 * Definitions for IP option names.
497 #define IP_FW_IPOPT_LSRR 0x01
498 #define IP_FW_IPOPT_SSRR 0x02
499 #define IP_FW_IPOPT_RR 0x04
500 #define IP_FW_IPOPT_TS 0x08
503 * Definitions for TCP option names.
505 #define IP_FW_TCPOPT_MSS 0x01
506 #define IP_FW_TCPOPT_WINDOW 0x02
507 #define IP_FW_TCPOPT_SACK 0x04
508 #define IP_FW_TCPOPT_TS 0x08
509 #define IP_FW_TCPOPT_CC 0x10
511 #define ICMP_REJECT_RST 0x100 /* fake ICMP code (send a TCP RST) */
512 #define ICMP6_UNREACH_RST 0x100 /* fake ICMPv6 code (send a TCP RST) */
515 * These are used for lookup tables.
517 typedef struct _ipfw_table_entry {
518 in_addr_t addr; /* network address */
519 u_int32_t value; /* value */
520 u_int16_t tbl; /* table number */
521 u_int8_t masklen; /* mask length */
524 typedef struct _ipfw_table {
525 u_int32_t size; /* size of entries in bytes */
526 u_int32_t cnt; /* # of entries */
527 u_int16_t tbl; /* table number */
528 ipfw_table_entry ent[0]; /* entries */
531 #define IP_FW_TABLEARG 65535
534 * Main firewall chains definitions and global var's definitions.
538 /* Return values from ipfw_chk() */
550 /* flags for divert mtag */
551 #define IP_FW_DIVERT_LOOPBACK_FLAG 0x00080000
552 #define IP_FW_DIVERT_OUTPUT_FLAG 0x00100000
555 * Structure for collecting parameters to dummynet for ip6_output forwarding
558 struct ip6_pktopts *opt_or;
559 struct route_in6 ro_or;
561 struct ip6_moptions *im6o_or;
562 struct ifnet *origifp_or;
563 struct ifnet *ifp_or;
564 struct sockaddr_in6 dst_or;
566 struct route_in6 ro_pmtu_or;
570 * Arguments for calling ipfw_chk() and dummynet_io(). We put them
571 * all into a structure because this way it is easier and more
572 * efficient to pass variables around and extend the interface.
575 struct mbuf *m; /* the mbuf chain */
576 struct ifnet *oif; /* output interface */
577 struct sockaddr_in *next_hop; /* forward address */
578 struct ip_fw *rule; /* matching rule */
579 struct ether_header *eh; /* for bridged packets */
581 struct ipfw_flow_id f_id; /* grabbed from IP header */
582 u_int32_t cookie; /* a cookie depending on rule action */
585 struct _ip6dn_args dummypar; /* dummynet->ip6_output */
586 struct sockaddr_in hopstore; /* store here if cannot use a pointer */
590 * Function definitions.
597 int ipfw_check_in(void *, struct mbuf **, struct ifnet *, int, struct inpcb *inp);
598 int ipfw_check_out(void *, struct mbuf **, struct ifnet *, int, struct inpcb *inp);
600 int ipfw_chk(struct ip_fw_args *);
603 void ipfw_destroy(void);
605 typedef int ip_fw_ctl_t(struct sockopt *);
606 extern ip_fw_ctl_t *ip_fw_ctl_ptr;
607 extern int fw_one_pass;
608 extern int fw_enable;
610 extern int fw6_enable;
613 /* For kernel ipfw_ether and ipfw_bridge. */
614 typedef int ip_fw_chk_t(struct ip_fw_args *args);
615 extern ip_fw_chk_t *ip_fw_chk_ptr;
616 #define IPFW_LOADED (ip_fw_chk_ptr != NULL)
619 #endif /* _IPFW2_H */