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 */
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 */
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 */
151 * actions for ng_ipfw
153 O_NETGRAPH, /* send to ng_ipfw */
154 O_NGTEE, /* copy to ng_ipfw */
158 O_UNREACH6, /* arg1=icmpv6 code arg (deny) */
160 O_LAST_OPCODE /* not an opcode! */
164 * The extension header are filtered only for presence using a bit
165 * vector with a flag for each header.
167 #define EXT_FRAGMENT 0x1
168 #define EXT_HOPOPTS 0x2
169 #define EXT_ROUTING 0x4
172 #define EXT_DSTOPTS 0x20
175 * Template for instructions.
177 * ipfw_insn is used for all instructions which require no operands,
178 * a single 16-bit value (arg1), or a couple of 8-bit values.
180 * For other instructions which require different/larger arguments
181 * we have derived structures, ipfw_insn_*.
183 * The size of the instruction (in 32-bit words) is in the low
184 * 6 bits of "len". The 2 remaining bits are used to implement
185 * NOT and OR on individual instructions. Given a type, you can
186 * compute the length to be put in "len" using F_INSN_SIZE(t)
188 * F_NOT negates the match result of the instruction.
190 * F_OR is used to build or blocks. By default, instructions
191 * are evaluated as part of a logical AND. An "or" block
192 * { X or Y or Z } contains F_OR set in all but the last
193 * instruction of the block. A match will cause the code
194 * to skip past the last instruction of the block.
196 * NOTA BENE: in a couple of places we assume that
197 * sizeof(ipfw_insn) == sizeof(u_int32_t)
198 * this needs to be fixed.
201 typedef struct _ipfw_insn { /* template for instructions */
202 enum ipfw_opcodes opcode:8;
203 u_int8_t len; /* numer of 32-byte words */
206 #define F_LEN_MASK 0x3f
207 #define F_LEN(cmd) ((cmd)->len & F_LEN_MASK)
213 * The F_INSN_SIZE(type) computes the size, in 4-byte words, of
216 #define F_INSN_SIZE(t) ((sizeof (t))/sizeof(u_int32_t))
219 * This is used to store an array of 16-bit entries (ports etc.)
221 typedef struct _ipfw_insn_u16 {
223 u_int16_t ports[2]; /* there may be more */
227 * This is used to store an array of 32-bit entries
228 * (uid, single IPv4 addresses etc.)
230 typedef struct _ipfw_insn_u32 {
232 u_int32_t d[1]; /* one or more */
236 * This is used to store IP addr-mask pairs.
238 typedef struct _ipfw_insn_ip {
245 * This is used to forward to a given address (ip).
247 typedef struct _ipfw_insn_sa {
249 struct sockaddr_in sa;
253 * This is used for MAC addr-mask pairs.
255 typedef struct _ipfw_insn_mac {
257 u_char addr[12]; /* dst[6] + src[6] */
258 u_char mask[12]; /* dst[6] + src[6] */
262 * This is used for interface match rules (recv xx, xmit xx).
264 typedef struct _ipfw_insn_if {
274 * This is used for storing an altq queue id number.
276 typedef struct _ipfw_insn_altq {
282 * This is used for limit rules.
284 typedef struct _ipfw_insn_limit {
287 u_int8_t limit_mask; /* combination of DYN_* below */
288 #define DYN_SRC_ADDR 0x1
289 #define DYN_SRC_PORT 0x2
290 #define DYN_DST_ADDR 0x4
291 #define DYN_DST_PORT 0x8
293 u_int16_t conn_limit;
297 * This is used for log instructions.
299 typedef struct _ipfw_insn_log {
301 u_int32_t max_log; /* how many do we log -- 0 = all */
302 u_int32_t log_left; /* how many left to log */
305 /* Apply ipv6 mask on ipv6 addr */
306 #define APPLY_MASK(addr,mask) \
307 (addr)->__u6_addr.__u6_addr32[0] &= (mask)->__u6_addr.__u6_addr32[0]; \
308 (addr)->__u6_addr.__u6_addr32[1] &= (mask)->__u6_addr.__u6_addr32[1]; \
309 (addr)->__u6_addr.__u6_addr32[2] &= (mask)->__u6_addr.__u6_addr32[2]; \
310 (addr)->__u6_addr.__u6_addr32[3] &= (mask)->__u6_addr.__u6_addr32[3];
312 /* Structure for ipv6 */
313 typedef struct _ipfw_insn_ip6 {
315 struct in6_addr addr6;
316 struct in6_addr mask6;
319 /* Used to support icmp6 types */
320 typedef struct _ipfw_insn_icmp6 {
322 uint32_t d[7]; /* XXX This number si related to the netinet/icmp6.h
323 * define ICMP6_MAXTYPE
324 * as follows: n = ICMP6_MAXTYPE/32 + 1
330 * Here we have the structure representing an ipfw rule.
332 * It starts with a general area (with link fields and counters)
333 * followed by an array of one or more instructions, which the code
334 * accesses as an array of 32-bit values.
336 * Given a rule pointer r:
338 * r->cmd is the start of the first instruction.
339 * ACTION_PTR(r) is the start of the first action (things to do
340 * once a rule matched).
342 * When assembling instruction, remember the following:
344 * + if a rule has a "keep-state" (or "limit") option, then the
345 * first instruction (at r->cmd) MUST BE an O_PROBE_STATE
346 * + if a rule has a "log" option, then the first action
347 * (at ACTION_PTR(r)) MUST be O_LOG
348 * + if a rule has an "altq" option, it comes after "log"
350 * NOTE: we use a simple linked list of rules because we never need
351 * to delete a rule without scanning the list. We do not use
352 * queue(3) macros for portability and readability.
356 struct ip_fw *next; /* linked list of rules */
357 struct ip_fw *next_rule; /* ptr to next [skipto] rule */
358 /* 'next_rule' is used to pass up 'set_disable' status */
360 u_int16_t act_ofs; /* offset of action in 32-bit units */
361 u_int16_t cmd_len; /* # of 32-bit words in cmd */
362 u_int16_t rulenum; /* rule number */
363 u_int8_t set; /* rule set (0..31) */
364 #define RESVD_SET 31 /* set for default and persistent rules */
365 u_int8_t _pad; /* padding */
367 /* These fields are present in all rules. */
368 u_int64_t pcnt; /* Packet counter */
369 u_int64_t bcnt; /* Byte counter */
370 u_int32_t timestamp; /* tv_sec of last match */
372 ipfw_insn cmd[1]; /* storage for commands */
375 #define ACTION_PTR(rule) \
376 (ipfw_insn *)( (u_int32_t *)((rule)->cmd) + ((rule)->act_ofs) )
378 #define RULESIZE(rule) (sizeof(struct ip_fw) + \
379 ((struct ip_fw *)(rule))->cmd_len * 4 - 4)
382 * This structure is used as a flow mask and a flow id for various
385 struct ipfw_flow_id {
391 u_int8_t flags; /* protocol-specific flags */
392 uint8_t addr_type; /* 4 = ipv4, 6 = ipv6, 1=ether ? */
393 struct in6_addr dst_ip6; /* could also store MAC addr! */
394 struct in6_addr src_ip6;
399 #define IS_IP6_FLOW_ID(id) ((id)->addr_type == 6)
404 typedef struct _ipfw_dyn_rule ipfw_dyn_rule;
406 struct _ipfw_dyn_rule {
407 ipfw_dyn_rule *next; /* linked list of rules. */
408 struct ip_fw *rule; /* pointer to rule */
409 /* 'rule' is used to pass up the rule number (from the parent) */
411 ipfw_dyn_rule *parent; /* pointer to parent rule */
412 u_int64_t pcnt; /* packet match counter */
413 u_int64_t bcnt; /* byte match counter */
414 struct ipfw_flow_id id; /* (masked) flow id */
415 u_int32_t expire; /* expire time */
416 u_int32_t bucket; /* which bucket in hash table */
417 u_int32_t state; /* state of this rule (typically a
418 * combination of TCP flags)
420 u_int32_t ack_fwd; /* most recent ACKs in forward */
421 u_int32_t ack_rev; /* and reverse directions (used */
422 /* to generate keepalives) */
423 u_int16_t dyn_type; /* rule type */
424 u_int16_t count; /* refcount */
428 * Definitions for IP option names.
430 #define IP_FW_IPOPT_LSRR 0x01
431 #define IP_FW_IPOPT_SSRR 0x02
432 #define IP_FW_IPOPT_RR 0x04
433 #define IP_FW_IPOPT_TS 0x08
436 * Definitions for TCP option names.
438 #define IP_FW_TCPOPT_MSS 0x01
439 #define IP_FW_TCPOPT_WINDOW 0x02
440 #define IP_FW_TCPOPT_SACK 0x04
441 #define IP_FW_TCPOPT_TS 0x08
442 #define IP_FW_TCPOPT_CC 0x10
444 #define ICMP_REJECT_RST 0x100 /* fake ICMP code (send a TCP RST) */
445 #define ICMP6_UNREACH_RST 0x100 /* fake ICMPv6 code (send a TCP RST) */
448 * These are used for lookup tables.
450 typedef struct _ipfw_table_entry {
451 in_addr_t addr; /* network address */
452 u_int32_t value; /* value */
453 u_int16_t tbl; /* table number */
454 u_int8_t masklen; /* mask length */
457 typedef struct _ipfw_table {
458 u_int32_t size; /* size of entries in bytes */
459 u_int32_t cnt; /* # of entries */
460 u_int16_t tbl; /* table number */
461 ipfw_table_entry ent[0]; /* entries */
464 #define IP_FW_TABLEARG 65535
467 * Main firewall chains definitions and global var's definitions.
471 /* Return values from ipfw_chk() */
482 /* flags for divert mtag */
483 #define IP_FW_DIVERT_LOOPBACK_FLAG 0x00080000
484 #define IP_FW_DIVERT_OUTPUT_FLAG 0x00100000
487 * Structure for collecting parameters to dummynet for ip6_output forwarding
490 struct ip6_pktopts *opt_or;
491 struct route_in6 ro_or;
493 struct ip6_moptions *im6o_or;
494 struct ifnet *origifp_or;
495 struct ifnet *ifp_or;
496 struct sockaddr_in6 dst_or;
498 struct route_in6 ro_pmtu_or;
502 * Arguments for calling ipfw_chk() and dummynet_io(). We put them
503 * all into a structure because this way it is easier and more
504 * efficient to pass variables around and extend the interface.
507 struct mbuf *m; /* the mbuf chain */
508 struct ifnet *oif; /* output interface */
509 struct sockaddr_in *next_hop; /* forward address */
510 struct ip_fw *rule; /* matching rule */
511 struct ether_header *eh; /* for bridged packets */
513 struct ipfw_flow_id f_id; /* grabbed from IP header */
514 u_int32_t cookie; /* a cookie depending on rule action */
517 struct _ip6dn_args dummypar; /* dummynet->ip6_output */
521 * Function definitions.
528 int ipfw_check_in(void *, struct mbuf **, struct ifnet *, int, struct inpcb *inp);
529 int ipfw_check_out(void *, struct mbuf **, struct ifnet *, int, struct inpcb *inp);
531 int ipfw_chk(struct ip_fw_args *);
534 void ipfw_destroy(void);
536 typedef int ip_fw_ctl_t(struct sockopt *);
537 extern ip_fw_ctl_t *ip_fw_ctl_ptr;
538 extern int fw_one_pass;
539 extern int fw_enable;
541 /* For kernel ipfw_ether and ipfw_bridge. */
542 typedef int ip_fw_chk_t(struct ip_fw_args *args);
543 extern ip_fw_chk_t *ip_fw_chk_ptr;
544 #define IPFW_LOADED (ip_fw_chk_ptr != NULL)
547 #endif /* _IPFW2_H */