xref: /freebsd/sys/netinet/ip_fw.h (revision d056fa046c6a91b90cd98165face0e42a33a5173)
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  * $FreeBSD$
26  */
27 
28 #ifndef _IPFW2_H
29 #define _IPFW2_H
30 
31 /*
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.
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 
53 enum 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_IP4,
157 
158 	O_UNREACH6,		/* arg1=icmpv6 code arg (deny)  */
159 
160 	O_TAG,   		/* arg1=tag number */
161 	O_TAGGED,		/* arg1=tag number */
162 
163 	O_LAST_OPCODE		/* not an opcode!		*/
164 };
165 
166 /*
167  * The extension header are filtered only for presence using a bit
168  * vector with a flag for each header.
169  */
170 #define EXT_FRAGMENT	0x1
171 #define EXT_HOPOPTS	0x2
172 #define EXT_ROUTING	0x4
173 #define EXT_AH		0x8
174 #define EXT_ESP		0x10
175 #define EXT_DSTOPTS	0x20
176 
177 /*
178  * Template for instructions.
179  *
180  * ipfw_insn is used for all instructions which require no operands,
181  * a single 16-bit value (arg1), or a couple of 8-bit values.
182  *
183  * For other instructions which require different/larger arguments
184  * we have derived structures, ipfw_insn_*.
185  *
186  * The size of the instruction (in 32-bit words) is in the low
187  * 6 bits of "len". The 2 remaining bits are used to implement
188  * NOT and OR on individual instructions. Given a type, you can
189  * compute the length to be put in "len" using F_INSN_SIZE(t)
190  *
191  * F_NOT	negates the match result of the instruction.
192  *
193  * F_OR		is used to build or blocks. By default, instructions
194  *		are evaluated as part of a logical AND. An "or" block
195  *		{ X or Y or Z } contains F_OR set in all but the last
196  *		instruction of the block. A match will cause the code
197  *		to skip past the last instruction of the block.
198  *
199  * NOTA BENE: in a couple of places we assume that
200  *	sizeof(ipfw_insn) == sizeof(u_int32_t)
201  * this needs to be fixed.
202  *
203  */
204 typedef struct	_ipfw_insn {	/* template for instructions */
205 	enum ipfw_opcodes	opcode:8;
206 	u_int8_t	len;	/* numer of 32-byte words */
207 #define	F_NOT		0x80
208 #define	F_OR		0x40
209 #define	F_LEN_MASK	0x3f
210 #define	F_LEN(cmd)	((cmd)->len & F_LEN_MASK)
211 
212 	u_int16_t	arg1;
213 } ipfw_insn;
214 
215 /*
216  * The F_INSN_SIZE(type) computes the size, in 4-byte words, of
217  * a given type.
218  */
219 #define	F_INSN_SIZE(t)	((sizeof (t))/sizeof(u_int32_t))
220 
221 #define MTAG_IPFW	1148380143	/* IPFW-tagged cookie */
222 
223 /*
224  * This is used to store an array of 16-bit entries (ports etc.)
225  */
226 typedef struct	_ipfw_insn_u16 {
227 	ipfw_insn o;
228 	u_int16_t ports[2];	/* there may be more */
229 } ipfw_insn_u16;
230 
231 /*
232  * This is used to store an array of 32-bit entries
233  * (uid, single IPv4 addresses etc.)
234  */
235 typedef struct	_ipfw_insn_u32 {
236 	ipfw_insn o;
237 	u_int32_t d[1];	/* one or more */
238 } ipfw_insn_u32;
239 
240 /*
241  * This is used to store IP addr-mask pairs.
242  */
243 typedef struct	_ipfw_insn_ip {
244 	ipfw_insn o;
245 	struct in_addr	addr;
246 	struct in_addr	mask;
247 } ipfw_insn_ip;
248 
249 /*
250  * This is used to forward to a given address (ip).
251  */
252 typedef struct  _ipfw_insn_sa {
253 	ipfw_insn o;
254 	struct sockaddr_in sa;
255 } ipfw_insn_sa;
256 
257 /*
258  * This is used for MAC addr-mask pairs.
259  */
260 typedef struct	_ipfw_insn_mac {
261 	ipfw_insn o;
262 	u_char addr[12];	/* dst[6] + src[6] */
263 	u_char mask[12];	/* dst[6] + src[6] */
264 } ipfw_insn_mac;
265 
266 /*
267  * This is used for interface match rules (recv xx, xmit xx).
268  */
269 typedef struct	_ipfw_insn_if {
270 	ipfw_insn o;
271 	union {
272 		struct in_addr ip;
273 		int glob;
274 	} p;
275 	char name[IFNAMSIZ];
276 } ipfw_insn_if;
277 
278 /*
279  * This is used for storing an altq queue id number.
280  */
281 typedef struct _ipfw_insn_altq {
282 	ipfw_insn	o;
283 	u_int32_t	qid;
284 } ipfw_insn_altq;
285 
286 /*
287  * This is used for limit rules.
288  */
289 typedef struct	_ipfw_insn_limit {
290 	ipfw_insn o;
291 	u_int8_t _pad;
292 	u_int8_t limit_mask;	/* combination of DYN_* below	*/
293 #define	DYN_SRC_ADDR	0x1
294 #define	DYN_SRC_PORT	0x2
295 #define	DYN_DST_ADDR	0x4
296 #define	DYN_DST_PORT	0x8
297 
298 	u_int16_t conn_limit;
299 } ipfw_insn_limit;
300 
301 /*
302  * This is used for log instructions.
303  */
304 typedef struct  _ipfw_insn_log {
305         ipfw_insn o;
306 	u_int32_t max_log;	/* how many do we log -- 0 = all */
307 	u_int32_t log_left;	/* how many left to log 	*/
308 } ipfw_insn_log;
309 
310 /* Apply ipv6 mask on ipv6 addr */
311 #define APPLY_MASK(addr,mask)                          \
312     (addr)->__u6_addr.__u6_addr32[0] &= (mask)->__u6_addr.__u6_addr32[0]; \
313     (addr)->__u6_addr.__u6_addr32[1] &= (mask)->__u6_addr.__u6_addr32[1]; \
314     (addr)->__u6_addr.__u6_addr32[2] &= (mask)->__u6_addr.__u6_addr32[2]; \
315     (addr)->__u6_addr.__u6_addr32[3] &= (mask)->__u6_addr.__u6_addr32[3];
316 
317 /* Structure for ipv6 */
318 typedef struct _ipfw_insn_ip6 {
319        ipfw_insn o;
320        struct in6_addr addr6;
321        struct in6_addr mask6;
322 } ipfw_insn_ip6;
323 
324 /* Used to support icmp6 types */
325 typedef struct _ipfw_insn_icmp6 {
326        ipfw_insn o;
327        uint32_t d[7]; /* XXX This number si related to the netinet/icmp6.h
328                        *     define ICMP6_MAXTYPE
329                        *     as follows: n = ICMP6_MAXTYPE/32 + 1
330                         *     Actually is 203
331                        */
332 } ipfw_insn_icmp6;
333 
334 /*
335  * Here we have the structure representing an ipfw rule.
336  *
337  * It starts with a general area (with link fields and counters)
338  * followed by an array of one or more instructions, which the code
339  * accesses as an array of 32-bit values.
340  *
341  * Given a rule pointer  r:
342  *
343  *  r->cmd		is the start of the first instruction.
344  *  ACTION_PTR(r)	is the start of the first action (things to do
345  *			once a rule matched).
346  *
347  * When assembling instruction, remember the following:
348  *
349  *  + if a rule has a "keep-state" (or "limit") option, then the
350  *	first instruction (at r->cmd) MUST BE an O_PROBE_STATE
351  *  + if a rule has a "log" option, then the first action
352  *	(at ACTION_PTR(r)) MUST be O_LOG
353  *  + if a rule has an "altq" option, it comes after "log"
354  *  + if a rule has an O_TAG option, it comes after "log" and "altq"
355  *
356  * NOTE: we use a simple linked list of rules because we never need
357  * 	to delete a rule without scanning the list. We do not use
358  *	queue(3) macros for portability and readability.
359  */
360 
361 struct ip_fw {
362 	struct ip_fw	*next;		/* linked list of rules		*/
363 	struct ip_fw	*next_rule;	/* ptr to next [skipto] rule	*/
364 	/* 'next_rule' is used to pass up 'set_disable' status		*/
365 
366 	u_int16_t	act_ofs;	/* offset of action in 32-bit units */
367 	u_int16_t	cmd_len;	/* # of 32-bit words in cmd	*/
368 	u_int16_t	rulenum;	/* rule number			*/
369 	u_int8_t	set;		/* rule set (0..31)		*/
370 #define	RESVD_SET	31	/* set for default and persistent rules */
371 	u_int8_t	_pad;		/* padding			*/
372 
373 	/* These fields are present in all rules.			*/
374 	u_int64_t	pcnt;		/* Packet counter		*/
375 	u_int64_t	bcnt;		/* Byte counter			*/
376 	u_int32_t	timestamp;	/* tv_sec of last match		*/
377 
378 	ipfw_insn	cmd[1];		/* storage for commands		*/
379 };
380 
381 #define ACTION_PTR(rule)				\
382 	(ipfw_insn *)( (u_int32_t *)((rule)->cmd) + ((rule)->act_ofs) )
383 
384 #define RULESIZE(rule)  (sizeof(struct ip_fw) + \
385 	((struct ip_fw *)(rule))->cmd_len * 4 - 4)
386 
387 /*
388  * This structure is used as a flow mask and a flow id for various
389  * parts of the code.
390  */
391 struct ipfw_flow_id {
392 	u_int32_t	dst_ip;
393 	u_int32_t	src_ip;
394 	u_int16_t	dst_port;
395 	u_int16_t	src_port;
396 	u_int8_t	proto;
397 	u_int8_t	flags;	/* protocol-specific flags */
398 	uint8_t		addr_type; /* 4 = ipv4, 6 = ipv6, 1=ether ? */
399 	struct in6_addr dst_ip6;	/* could also store MAC addr! */
400 	struct in6_addr src_ip6;
401 	u_int32_t	flow_id6;
402 	u_int32_t	frag_id6;
403 };
404 
405 #define IS_IP6_FLOW_ID(id)	((id)->addr_type == 6)
406 
407 /*
408  * Dynamic ipfw rule.
409  */
410 typedef struct _ipfw_dyn_rule ipfw_dyn_rule;
411 
412 struct _ipfw_dyn_rule {
413 	ipfw_dyn_rule	*next;		/* linked list of rules.	*/
414 	struct ip_fw *rule;		/* pointer to rule		*/
415 	/* 'rule' is used to pass up the rule number (from the parent)	*/
416 
417 	ipfw_dyn_rule *parent;		/* pointer to parent rule	*/
418 	u_int64_t	pcnt;		/* packet match counter		*/
419 	u_int64_t	bcnt;		/* byte match counter		*/
420 	struct ipfw_flow_id id;		/* (masked) flow id		*/
421 	u_int32_t	expire;		/* expire time			*/
422 	u_int32_t	bucket;		/* which bucket in hash table	*/
423 	u_int32_t	state;		/* state of this rule (typically a
424 					 * combination of TCP flags)
425 					 */
426 	u_int32_t	ack_fwd;	/* most recent ACKs in forward	*/
427 	u_int32_t	ack_rev;	/* and reverse directions (used	*/
428 					/* to generate keepalives)	*/
429 	u_int16_t	dyn_type;	/* rule type			*/
430 	u_int16_t	count;		/* refcount			*/
431 };
432 
433 /*
434  * Definitions for IP option names.
435  */
436 #define	IP_FW_IPOPT_LSRR	0x01
437 #define	IP_FW_IPOPT_SSRR	0x02
438 #define	IP_FW_IPOPT_RR		0x04
439 #define	IP_FW_IPOPT_TS		0x08
440 
441 /*
442  * Definitions for TCP option names.
443  */
444 #define	IP_FW_TCPOPT_MSS	0x01
445 #define	IP_FW_TCPOPT_WINDOW	0x02
446 #define	IP_FW_TCPOPT_SACK	0x04
447 #define	IP_FW_TCPOPT_TS		0x08
448 #define	IP_FW_TCPOPT_CC		0x10
449 
450 #define	ICMP_REJECT_RST		0x100	/* fake ICMP code (send a TCP RST) */
451 #define	ICMP6_UNREACH_RST	0x100	/* fake ICMPv6 code (send a TCP RST) */
452 
453 /*
454  * These are used for lookup tables.
455  */
456 typedef 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 
463 typedef 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 #define IP_FW_TABLEARG	65535
471 
472 /*
473  * Main firewall chains definitions and global var's definitions.
474  */
475 #ifdef _KERNEL
476 
477 /* Return values from ipfw_chk() */
478 enum {
479 	IP_FW_PASS = 0,
480 	IP_FW_DENY,
481 	IP_FW_DIVERT,
482 	IP_FW_TEE,
483 	IP_FW_DUMMYNET,
484 	IP_FW_NETGRAPH,
485 	IP_FW_NGTEE,
486 };
487 
488 /* flags for divert mtag */
489 #define	IP_FW_DIVERT_LOOPBACK_FLAG	0x00080000
490 #define	IP_FW_DIVERT_OUTPUT_FLAG	0x00100000
491 
492 /*
493  * Structure for collecting parameters to dummynet for ip6_output forwarding
494  */
495 struct _ip6dn_args {
496        struct ip6_pktopts *opt_or;
497        struct route_in6 ro_or;
498        int flags_or;
499        struct ip6_moptions *im6o_or;
500        struct ifnet *origifp_or;
501        struct ifnet *ifp_or;
502        struct sockaddr_in6 dst_or;
503        u_long mtu_or;
504        struct route_in6 ro_pmtu_or;
505 };
506 
507 /*
508  * Arguments for calling ipfw_chk() and dummynet_io(). We put them
509  * all into a structure because this way it is easier and more
510  * efficient to pass variables around and extend the interface.
511  */
512 struct ip_fw_args {
513 	struct mbuf	*m;		/* the mbuf chain		*/
514 	struct ifnet	*oif;		/* output interface		*/
515 	struct sockaddr_in *next_hop;	/* forward address		*/
516 	struct ip_fw	*rule;		/* matching rule		*/
517 	struct ether_header *eh;	/* for bridged packets		*/
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 */
531 struct sockopt;
532 struct dn_flow_set;
533 
534 int ipfw_check_in(void *, struct mbuf **, struct ifnet *, int, struct inpcb *inp);
535 int ipfw_check_out(void *, struct mbuf **, struct ifnet *, int, struct inpcb *inp);
536 
537 int ipfw_chk(struct ip_fw_args *);
538 
539 int ipfw_init(void);
540 void ipfw_destroy(void);
541 
542 typedef int ip_fw_ctl_t(struct sockopt *);
543 extern ip_fw_ctl_t *ip_fw_ctl_ptr;
544 extern int fw_one_pass;
545 extern int fw_enable;
546 #ifdef INET6
547 extern int fw6_enable;
548 #endif
549 
550 /* For kernel ipfw_ether and ipfw_bridge. */
551 typedef	int ip_fw_chk_t(struct ip_fw_args *args);
552 extern	ip_fw_chk_t	*ip_fw_chk_ptr;
553 #define	IPFW_LOADED	(ip_fw_chk_ptr != NULL)
554 
555 #endif /* _KERNEL */
556 #endif /* _IPFW2_H */
557