xref: /freebsd/sys/netinet/ip_fw.h (revision 6af83ee0d2941d18880b6aaa2b4facd1d30c6106)
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 #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 
54 enum 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 	O_ALTQ,			/* u32 = altq classif. qid	*/
138 	O_DIVERTED,		/* arg1=bitmap (1:loop, 2:out)	*/
139 	O_TCPDATALEN,		/* arg1 = tcp data len		*/
140 
141 	/*
142 	 * actions for ng_ipfw
143 	 */
144 	O_NETGRAPH,		/* send to ng_ipfw		*/
145 	O_NGTEE,		/* copy to ng_ipfw		*/
146 
147 	O_LAST_OPCODE		/* not an opcode!		*/
148 };
149 
150 /*
151  * Template for instructions.
152  *
153  * ipfw_insn is used for all instructions which require no operands,
154  * a single 16-bit value (arg1), or a couple of 8-bit values.
155  *
156  * For other instructions which require different/larger arguments
157  * we have derived structures, ipfw_insn_*.
158  *
159  * The size of the instruction (in 32-bit words) is in the low
160  * 6 bits of "len". The 2 remaining bits are used to implement
161  * NOT and OR on individual instructions. Given a type, you can
162  * compute the length to be put in "len" using F_INSN_SIZE(t)
163  *
164  * F_NOT	negates the match result of the instruction.
165  *
166  * F_OR		is used to build or blocks. By default, instructions
167  *		are evaluated as part of a logical AND. An "or" block
168  *		{ X or Y or Z } contains F_OR set in all but the last
169  *		instruction of the block. A match will cause the code
170  *		to skip past the last instruction of the block.
171  *
172  * NOTA BENE: in a couple of places we assume that
173  *	sizeof(ipfw_insn) == sizeof(u_int32_t)
174  * this needs to be fixed.
175  *
176  */
177 typedef struct	_ipfw_insn {	/* template for instructions */
178 	enum ipfw_opcodes	opcode:8;
179 	u_int8_t	len;	/* numer of 32-byte words */
180 #define	F_NOT		0x80
181 #define	F_OR		0x40
182 #define	F_LEN_MASK	0x3f
183 #define	F_LEN(cmd)	((cmd)->len & F_LEN_MASK)
184 
185 	u_int16_t	arg1;
186 } ipfw_insn;
187 
188 /*
189  * The F_INSN_SIZE(type) computes the size, in 4-byte words, of
190  * a given type.
191  */
192 #define	F_INSN_SIZE(t)	((sizeof (t))/sizeof(u_int32_t))
193 
194 /*
195  * This is used to store an array of 16-bit entries (ports etc.)
196  */
197 typedef struct	_ipfw_insn_u16 {
198 	ipfw_insn o;
199 	u_int16_t ports[2];	/* there may be more */
200 } ipfw_insn_u16;
201 
202 /*
203  * This is used to store an array of 32-bit entries
204  * (uid, single IPv4 addresses etc.)
205  */
206 typedef struct	_ipfw_insn_u32 {
207 	ipfw_insn o;
208 	u_int32_t d[1];	/* one or more */
209 } ipfw_insn_u32;
210 
211 /*
212  * This is used to store IP addr-mask pairs.
213  */
214 typedef struct	_ipfw_insn_ip {
215 	ipfw_insn o;
216 	struct in_addr	addr;
217 	struct in_addr	mask;
218 } ipfw_insn_ip;
219 
220 /*
221  * This is used to forward to a given address (ip).
222  */
223 typedef struct  _ipfw_insn_sa {
224 	ipfw_insn o;
225 	struct sockaddr_in sa;
226 } ipfw_insn_sa;
227 
228 /*
229  * This is used for MAC addr-mask pairs.
230  */
231 typedef struct	_ipfw_insn_mac {
232 	ipfw_insn o;
233 	u_char addr[12];	/* dst[6] + src[6] */
234 	u_char mask[12];	/* dst[6] + src[6] */
235 } ipfw_insn_mac;
236 
237 /*
238  * This is used for interface match rules (recv xx, xmit xx).
239  */
240 typedef struct	_ipfw_insn_if {
241 	ipfw_insn o;
242 	union {
243 		struct in_addr ip;
244 		int glob;
245 	} p;
246 	char name[IFNAMSIZ];
247 } ipfw_insn_if;
248 
249 /*
250  * This is used for pipe and queue actions, which need to store
251  * a single pointer (which can have different size on different
252  * architectures.
253  * Note that, because of previous instructions, pipe_ptr might
254  * be unaligned in the overall structure, so it needs to be
255  * manipulated with care.
256  */
257 typedef struct	_ipfw_insn_pipe {
258 	ipfw_insn	o;
259 	void		*pipe_ptr;	/* XXX */
260 } ipfw_insn_pipe;
261 
262 /*
263  * This is used for storing an altq queue id number.
264  */
265 typedef struct _ipfw_insn_altq {
266 	ipfw_insn	o;
267 	u_int32_t	qid;
268 } ipfw_insn_altq;
269 
270 /*
271  * This is used for limit rules.
272  */
273 typedef struct	_ipfw_insn_limit {
274 	ipfw_insn o;
275 	u_int8_t _pad;
276 	u_int8_t limit_mask;	/* combination of DYN_* below	*/
277 #define	DYN_SRC_ADDR	0x1
278 #define	DYN_SRC_PORT	0x2
279 #define	DYN_DST_ADDR	0x4
280 #define	DYN_DST_PORT	0x8
281 
282 	u_int16_t conn_limit;
283 } ipfw_insn_limit;
284 
285 /*
286  * This is used for log instructions.
287  */
288 typedef struct  _ipfw_insn_log {
289         ipfw_insn o;
290 	u_int32_t max_log;	/* how many do we log -- 0 = all */
291 	u_int32_t log_left;	/* how many left to log 	*/
292 } ipfw_insn_log;
293 
294 /*
295  * Here we have the structure representing an ipfw rule.
296  *
297  * It starts with a general area (with link fields and counters)
298  * followed by an array of one or more instructions, which the code
299  * accesses as an array of 32-bit values.
300  *
301  * Given a rule pointer  r:
302  *
303  *  r->cmd		is the start of the first instruction.
304  *  ACTION_PTR(r)	is the start of the first action (things to do
305  *			once a rule matched).
306  *
307  * When assembling instruction, remember the following:
308  *
309  *  + if a rule has a "keep-state" (or "limit") option, then the
310  *	first instruction (at r->cmd) MUST BE an O_PROBE_STATE
311  *  + if a rule has a "log" option, then the first action
312  *	(at ACTION_PTR(r)) MUST be O_LOG
313  *  + if a rule has an "altq" option, it comes after "log"
314  *
315  * NOTE: we use a simple linked list of rules because we never need
316  * 	to delete a rule without scanning the list. We do not use
317  *	queue(3) macros for portability and readability.
318  */
319 
320 struct ip_fw {
321 	struct ip_fw	*next;		/* linked list of rules		*/
322 	struct ip_fw	*next_rule;	/* ptr to next [skipto] rule	*/
323 	/* 'next_rule' is used to pass up 'set_disable' status		*/
324 
325 	u_int16_t	act_ofs;	/* offset of action in 32-bit units */
326 	u_int16_t	cmd_len;	/* # of 32-bit words in cmd	*/
327 	u_int16_t	rulenum;	/* rule number			*/
328 	u_int8_t	set;		/* rule set (0..31)		*/
329 #define	RESVD_SET	31	/* set for default and persistent rules */
330 	u_int8_t	_pad;		/* padding			*/
331 
332 	/* These fields are present in all rules.			*/
333 	u_int64_t	pcnt;		/* Packet counter		*/
334 	u_int64_t	bcnt;		/* Byte counter			*/
335 	u_int32_t	timestamp;	/* tv_sec of last match		*/
336 
337 	ipfw_insn	cmd[1];		/* storage for commands		*/
338 };
339 
340 #define ACTION_PTR(rule)				\
341 	(ipfw_insn *)( (u_int32_t *)((rule)->cmd) + ((rule)->act_ofs) )
342 
343 #define RULESIZE(rule)  (sizeof(struct ip_fw) + \
344 	((struct ip_fw *)(rule))->cmd_len * 4 - 4)
345 
346 /*
347  * This structure is used as a flow mask and a flow id for various
348  * parts of the code.
349  */
350 struct ipfw_flow_id {
351 	u_int32_t	dst_ip;
352 	u_int32_t	src_ip;
353 	u_int16_t	dst_port;
354 	u_int16_t	src_port;
355 	u_int8_t	proto;
356 	u_int8_t	flags;	/* protocol-specific flags */
357 };
358 
359 /*
360  * Dynamic ipfw rule.
361  */
362 typedef struct _ipfw_dyn_rule ipfw_dyn_rule;
363 
364 struct _ipfw_dyn_rule {
365 	ipfw_dyn_rule	*next;		/* linked list of rules.	*/
366 	struct ip_fw *rule;		/* pointer to rule		*/
367 	/* 'rule' is used to pass up the rule number (from the parent)	*/
368 
369 	ipfw_dyn_rule *parent;		/* pointer to parent rule	*/
370 	u_int64_t	pcnt;		/* packet match counter		*/
371 	u_int64_t	bcnt;		/* byte match counter		*/
372 	struct ipfw_flow_id id;		/* (masked) flow id		*/
373 	u_int32_t	expire;		/* expire time			*/
374 	u_int32_t	bucket;		/* which bucket in hash table	*/
375 	u_int32_t	state;		/* state of this rule (typically a
376 					 * combination of TCP flags)
377 					 */
378 	u_int32_t	ack_fwd;	/* most recent ACKs in forward	*/
379 	u_int32_t	ack_rev;	/* and reverse directions (used	*/
380 					/* to generate keepalives)	*/
381 	u_int16_t	dyn_type;	/* rule type			*/
382 	u_int16_t	count;		/* refcount			*/
383 };
384 
385 /*
386  * Definitions for IP option names.
387  */
388 #define	IP_FW_IPOPT_LSRR	0x01
389 #define	IP_FW_IPOPT_SSRR	0x02
390 #define	IP_FW_IPOPT_RR		0x04
391 #define	IP_FW_IPOPT_TS		0x08
392 
393 /*
394  * Definitions for TCP option names.
395  */
396 #define	IP_FW_TCPOPT_MSS	0x01
397 #define	IP_FW_TCPOPT_WINDOW	0x02
398 #define	IP_FW_TCPOPT_SACK	0x04
399 #define	IP_FW_TCPOPT_TS		0x08
400 #define	IP_FW_TCPOPT_CC		0x10
401 
402 #define	ICMP_REJECT_RST		0x100	/* fake ICMP code (send a TCP RST) */
403 
404 /*
405  * These are used for lookup tables.
406  */
407 typedef struct	_ipfw_table_entry {
408 	in_addr_t	addr;		/* network address		*/
409 	u_int32_t	value;		/* value			*/
410 	u_int16_t	tbl;		/* table number			*/
411 	u_int8_t	masklen;	/* mask length			*/
412 } ipfw_table_entry;
413 
414 typedef struct	_ipfw_table {
415 	u_int32_t	size;		/* size of entries in bytes	*/
416 	u_int32_t	cnt;		/* # of entries			*/
417 	u_int16_t	tbl;		/* table number			*/
418 	ipfw_table_entry ent[0];	/* entries			*/
419 } ipfw_table;
420 
421 /*
422  * Main firewall chains definitions and global var's definitions.
423  */
424 #ifdef _KERNEL
425 
426 /* Return values from ipfw_chk() */
427 enum {
428 	IP_FW_PASS = 0,
429 	IP_FW_DENY,
430 	IP_FW_DIVERT,
431 	IP_FW_TEE,
432 	IP_FW_DUMMYNET,
433 	IP_FW_NETGRAPH,
434 	IP_FW_NGTEE,
435 };
436 
437 /* flags for divert mtag */
438 #define	IP_FW_DIVERT_LOOPBACK_FLAG	0x00080000
439 #define	IP_FW_DIVERT_OUTPUT_FLAG	0x00100000
440 
441 /*
442  * Arguments for calling ipfw_chk() and dummynet_io(). We put them
443  * all into a structure because this way it is easier and more
444  * efficient to pass variables around and extend the interface.
445  */
446 struct ip_fw_args {
447 	struct mbuf	*m;		/* the mbuf chain		*/
448 	struct ifnet	*oif;		/* output interface		*/
449 	struct sockaddr_in *next_hop;	/* forward address		*/
450 	struct ip_fw	*rule;		/* matching rule		*/
451 	struct ether_header *eh;	/* for bridged packets		*/
452 
453 	int flags;			/* for dummynet			*/
454 
455 	struct ipfw_flow_id f_id;	/* grabbed from IP header	*/
456 	u_int32_t	cookie;		/* a cookie depending on rule action */
457 	struct inpcb	*inp;
458 };
459 
460 /*
461  * Function definitions.
462  */
463 
464 /* Firewall hooks */
465 struct sockopt;
466 struct dn_flow_set;
467 
468 int ipfw_check_in(void *, struct mbuf **, struct ifnet *, int, struct inpcb *inp);
469 int ipfw_check_out(void *, struct mbuf **, struct ifnet *, int, struct inpcb *inp);
470 
471 int ipfw_chk(struct ip_fw_args *);
472 
473 int ipfw_init(void);
474 void ipfw_destroy(void);
475 
476 void flush_pipe_ptrs(struct dn_flow_set *match); /* used by dummynet */
477 
478 typedef int ip_fw_ctl_t(struct sockopt *);
479 extern ip_fw_ctl_t *ip_fw_ctl_ptr;
480 extern int fw_one_pass;
481 extern int fw_enable;
482 
483 /* For kernel ipfw_ether and ipfw_bridge. */
484 typedef	int ip_fw_chk_t(struct ip_fw_args *args);
485 extern	ip_fw_chk_t	*ip_fw_chk_ptr;
486 #define	IPFW_LOADED	(ip_fw_chk_ptr != NULL)
487 
488 #endif /* _KERNEL */
489 #endif /* _IPFW2_H */
490