ip_fw.h revision 115793
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: head/sys/netinet/ip_fw.h 115793 2003-06-04 01:17:37Z ticso $
26 */
27
28#ifndef _IPFW2_H
29#define _IPFW2_H
30#define IPFW2  1
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.
48 */
49
50enum ipfw_opcodes {		/* arguments (4 byte each)	*/
51	O_NOP,
52
53	O_IP_SRC,		/* u32 = IP			*/
54	O_IP_SRC_MASK,		/* ip = IP/mask			*/
55	O_IP_SRC_ME,		/* none				*/
56	O_IP_SRC_SET,		/* u32=base, arg1=len, bitmap	*/
57
58	O_IP_DST,		/* u32 = IP			*/
59	O_IP_DST_MASK,		/* ip = IP/mask			*/
60	O_IP_DST_ME,		/* none				*/
61	O_IP_DST_SET,		/* u32=base, arg1=len, bitmap	*/
62
63	O_IP_SRCPORT,		/* (n)port list:mask 4 byte ea	*/
64	O_IP_DSTPORT,		/* (n)port list:mask 4 byte ea	*/
65	O_PROTO,		/* arg1=protocol		*/
66
67	O_MACADDR2,		/* 2 mac addr:mask		*/
68	O_MAC_TYPE,		/* same as srcport		*/
69
70	O_LAYER2,		/* none				*/
71	O_IN,			/* none				*/
72	O_FRAG,			/* none				*/
73
74	O_RECV,			/* none				*/
75	O_XMIT,			/* none				*/
76	O_VIA,			/* none				*/
77
78	O_IPOPT,		/* arg1 = 2*u8 bitmap		*/
79	O_IPLEN,		/* arg1 = len			*/
80	O_IPID,			/* arg1 = id			*/
81
82	O_IPTOS,		/* arg1 = id			*/
83	O_IPPRECEDENCE,		/* arg1 = precedence << 5	*/
84	O_IPTTL,		/* arg1 = TTL			*/
85
86	O_IPVER,		/* arg1 = version		*/
87	O_UID,			/* u32 = id			*/
88	O_GID,			/* u32 = id			*/
89	O_ESTAB,		/* none (tcp established)	*/
90	O_TCPFLAGS,		/* arg1 = 2*u8 bitmap		*/
91	O_TCPWIN,		/* arg1 = desired win		*/
92	O_TCPSEQ,		/* u32 = desired seq.		*/
93	O_TCPACK,		/* u32 = desired seq.		*/
94	O_ICMPTYPE,		/* u32 = icmp bitmap		*/
95	O_TCPOPTS,		/* arg1 = 2*u8 bitmap		*/
96
97	O_VERREVPATH,		/* none				*/
98
99	O_PROBE_STATE,		/* none				*/
100	O_KEEP_STATE,		/* none				*/
101	O_LIMIT,		/* ipfw_insn_limit		*/
102	O_LIMIT_PARENT,		/* dyn_type, not an opcode.	*/
103	/*
104	 * these are really 'actions', and must be last in the list.
105	 */
106
107	O_LOG,			/* ipfw_insn_log		*/
108	O_PROB,			/* u32 = match probability	*/
109
110	O_CHECK_STATE,		/* none				*/
111	O_ACCEPT,		/* none				*/
112	O_DENY,			/* none 			*/
113	O_REJECT,		/* arg1=icmp arg (same as deny)	*/
114	O_COUNT,		/* none				*/
115	O_SKIPTO,		/* arg1=next rule number	*/
116	O_PIPE,			/* arg1=pipe number		*/
117	O_QUEUE,		/* arg1=queue number		*/
118	O_DIVERT,		/* arg1=port number		*/
119	O_TEE,			/* arg1=port number		*/
120	O_FORWARD_IP,		/* fwd sockaddr			*/
121	O_FORWARD_MAC,		/* fwd mac			*/
122	O_LAST_OPCODE		/* not an opcode!		*/
123};
124
125/*
126 * Template for instructions.
127 *
128 * ipfw_insn is used for all instructions which require no operands,
129 * a single 16-bit value (arg1), or a couple of 8-bit values.
130 *
131 * For other instructions which require different/larger arguments
132 * we have derived structures, ipfw_insn_*.
133 *
134 * The size of the instruction (in 32-bit words) is in the low
135 * 6 bits of "len". The 2 remaining bits are used to implement
136 * NOT and OR on individual instructions. Given a type, you can
137 * compute the length to be put in "len" using F_INSN_SIZE(t)
138 *
139 * F_NOT	negates the match result of the instruction.
140 *
141 * F_OR		is used to build or blocks. By default, instructions
142 *		are evaluated as part of a logical AND. An "or" block
143 *		{ X or Y or Z } contains F_OR set in all but the last
144 *		instruction of the block. A match will cause the code
145 *		to skip past the last instruction of the block.
146 *
147 * NOTA BENE: in a couple of places we assume that
148 *	sizeof(ipfw_insn) == sizeof(u_int32_t)
149 * this needs to be fixed.
150 *
151 */
152typedef struct	_ipfw_insn {	/* template for instructions */
153	enum ipfw_opcodes	opcode:8;
154	u_int8_t	len;	/* numer of 32-byte words */
155#define	F_NOT		0x80
156#define	F_OR		0x40
157#define	F_LEN_MASK	0x3f
158#define	F_LEN(cmd)	((cmd)->len & F_LEN_MASK)
159
160	u_int16_t	arg1;
161} ipfw_insn;
162
163/*
164 * The F_INSN_SIZE(type) computes the size, in 4-byte words, of
165 * a given type.
166 */
167#define	F_INSN_SIZE(t)	((sizeof (t))/sizeof(u_int32_t))
168
169/*
170 * This is used to store an array of 16-bit entries (ports etc.)
171 */
172typedef struct	_ipfw_insn_u16 {
173	ipfw_insn o;
174	u_int16_t ports[2];	/* there may be more */
175} ipfw_insn_u16;
176
177/*
178 * This is used to store an array of 32-bit entries
179 * (uid, single IPv4 addresses etc.)
180 */
181typedef struct	_ipfw_insn_u32 {
182	ipfw_insn o;
183	u_int32_t d[1];	/* one or more */
184} ipfw_insn_u32;
185
186/*
187 * This is used to store IP addr-mask pairs.
188 */
189typedef struct	_ipfw_insn_ip {
190	ipfw_insn o;
191	struct in_addr	addr;
192	struct in_addr	mask;
193} ipfw_insn_ip;
194
195/*
196 * This is used to forward to a given address (ip)
197 */
198typedef struct  _ipfw_insn_sa {
199	ipfw_insn o;
200	struct sockaddr_in sa;
201} ipfw_insn_sa;
202
203/*
204 * This is used for MAC addr-mask pairs.
205 */
206typedef struct	_ipfw_insn_mac {
207	ipfw_insn o;
208	u_char addr[12];	/* dst[6] + src[6] */
209	u_char mask[12];	/* dst[6] + src[6] */
210} ipfw_insn_mac;
211
212/*
213 * This is used for interface match rules (recv xx, xmit xx)
214 */
215typedef struct	_ipfw_insn_if {
216	ipfw_insn o;
217	union {
218		struct in_addr ip;
219		int32_t unit;
220	} p;
221	char name[IFNAMSIZ];
222} ipfw_insn_if;
223
224/*
225 * This is used for pipe and queue actions, which need to store
226 * a single pointer (which can have different size on different
227 * architectures.
228 * Note that, because of previous instructions, pipe_ptr might
229 * be unaligned in the overall structure, so it needs to be
230 * manipulated with care.
231 */
232typedef struct	_ipfw_insn_pipe {
233	ipfw_insn	o;
234	void		*pipe_ptr;	/* XXX */
235} ipfw_insn_pipe;
236
237/*
238 * This is used for limit rules.
239 */
240typedef struct	_ipfw_insn_limit {
241	ipfw_insn o;
242	u_int8_t _pad;
243	u_int8_t limit_mask;	/* combination of DYN_* below	*/
244#define	DYN_SRC_ADDR	0x1
245#define	DYN_SRC_PORT	0x2
246#define	DYN_DST_ADDR	0x4
247#define	DYN_DST_PORT	0x8
248
249	u_int16_t conn_limit;
250} ipfw_insn_limit;
251
252/*
253 * This is used for log instructions
254 */
255typedef struct  _ipfw_insn_log {
256        ipfw_insn o;
257	u_int32_t max_log;	/* how many do we log -- 0 = all */
258	u_int32_t log_left;	/* how many left to log 	*/
259} ipfw_insn_log;
260
261/*
262 * Here we have the structure representing an ipfw rule.
263 *
264 * It starts with a general area (with link fields and counters)
265 * followed by an array of one or more instructions, which the code
266 * accesses as an array of 32-bit values.
267 *
268 * Given a rule pointer  r:
269 *
270 *  r->cmd		is the start of the first instruction.
271 *  ACTION_PTR(r)	is the start of the first action (things to do
272 *			once a rule matched).
273 *
274 * When assembling instruction, remember the following:
275 *
276 *  + if a rule has a "keep-state" (or "limit") option, then the
277 *	first instruction (at r->cmd) MUST BE an O_PROBE_STATE
278 *  + if a rule has a "log" option, then the first action
279 *	(at ACTION_PTR(r)) MUST be O_LOG
280 *
281 * NOTE: we use a simple linked list of rules because we never need
282 * 	to delete a rule without scanning the list. We do not use
283 *	queue(3) macros for portability and readability.
284 */
285
286struct ip_fw {
287	struct ip_fw	*next;		/* linked list of rules		*/
288	struct ip_fw	*next_rule;	/* ptr to next [skipto] rule	*/
289#if 0	/* passed up using 'next_rule' */
290	u_int32_t	set_disable;	/* disabled sets (for userland)	*/
291#endif
292	u_int16_t	act_ofs;	/* offset of action in 32-bit units */
293	u_int16_t	cmd_len;	/* # of 32-bit words in cmd	*/
294	u_int16_t	rulenum;	/* rule number			*/
295	u_int8_t	set;		/* rule set (0..31)		*/
296	u_int8_t	_pad;		/* padding			*/
297
298	/* These fields are present in all rules.			*/
299	u_int64_t	pcnt;		/* Packet counter		*/
300	u_int64_t	bcnt;		/* Byte counter			*/
301	u_int32_t	timestamp;	/* tv_sec of last match		*/
302
303	ipfw_insn	cmd[1];		/* storage for commands		*/
304};
305
306#define ACTION_PTR(rule)				\
307	(ipfw_insn *)( (u_int32_t *)((rule)->cmd) + ((rule)->act_ofs) )
308
309#define RULESIZE(rule)  (sizeof(struct ip_fw) + \
310	((struct ip_fw *)(rule))->cmd_len * 4 - 4)
311
312/*
313 * This structure is used as a flow mask and a flow id for various
314 * parts of the code.
315 */
316struct ipfw_flow_id {
317	u_int32_t	dst_ip;
318	u_int32_t	src_ip;
319	u_int16_t	dst_port;
320	u_int16_t	src_port;
321	u_int8_t	proto;
322	u_int8_t	flags;	/* protocol-specific flags */
323};
324
325/*
326 * dynamic ipfw rule
327 */
328typedef struct _ipfw_dyn_rule ipfw_dyn_rule;
329
330struct _ipfw_dyn_rule {
331	ipfw_dyn_rule	*next;		/* linked list of rules.	*/
332	struct ip_fw *rule;		/* pointer to rule		*/
333	ipfw_dyn_rule *parent;		/* pointer to parent rule	*/
334	u_int64_t	pcnt;		/* packet match counter		*/
335	u_int64_t	bcnt;		/* byte match counter		*/
336	struct ipfw_flow_id id;		/* (masked) flow id		*/
337	u_int32_t	expire;		/* expire time			*/
338	u_int32_t	bucket;		/* which bucket in hash table	*/
339	u_int32_t	state;		/* state of this rule (typically a
340					 * combination of TCP flags)
341					 */
342	u_int32_t	ack_fwd;	/* most recent ACKs in forward	*/
343	u_int32_t	ack_rev;	/* and reverse directions (used	*/
344					/* to generate keepalives)	*/
345	u_int16_t	dyn_type;	/* rule type			*/
346	u_int16_t	count;		/* refcount			*/
347#if 0	/* passed up with 'rule' */
348	u_int16_t	rulenum;	/* rule number (for userland)	*/
349#endif
350};
351
352/*
353 * Definitions for IP option names.
354 */
355#define	IP_FW_IPOPT_LSRR	0x01
356#define	IP_FW_IPOPT_SSRR	0x02
357#define	IP_FW_IPOPT_RR		0x04
358#define	IP_FW_IPOPT_TS		0x08
359
360/*
361 * Definitions for TCP option names.
362 */
363#define	IP_FW_TCPOPT_MSS	0x01
364#define	IP_FW_TCPOPT_WINDOW	0x02
365#define	IP_FW_TCPOPT_SACK	0x04
366#define	IP_FW_TCPOPT_TS		0x08
367#define	IP_FW_TCPOPT_CC		0x10
368
369#define	ICMP_REJECT_RST		0x100	/* fake ICMP code (send a TCP RST) */
370
371/*
372 * Main firewall chains definitions and global var's definitions.
373 */
374#ifdef _KERNEL
375
376#define	IP_FW_PORT_DYNT_FLAG	0x10000
377#define	IP_FW_PORT_TEE_FLAG	0x20000
378#define	IP_FW_PORT_DENY_FLAG	0x40000
379
380/*
381 * arguments for calling ipfw_chk() and dummynet_io(). We put them
382 * all into a structure because this way it is easier and more
383 * efficient to pass variables around and extend the interface.
384 */
385struct ip_fw_args {
386	struct mbuf	*m;		/* the mbuf chain		*/
387	struct ifnet	*oif;		/* output interface		*/
388	struct sockaddr_in *next_hop;	/* forward address		*/
389	struct ip_fw	*rule;		/* matching rule		*/
390	struct ether_header *eh;	/* for bridged packets		*/
391
392	struct route	*ro;		/* for dummynet			*/
393	struct sockaddr_in *dst;	/* for dummynet			*/
394	int flags;			/* for dummynet			*/
395
396	struct ipfw_flow_id f_id;	/* grabbed from IP header	*/
397	u_int16_t	divert_rule;	/* divert cookie		*/
398	u_int32_t	retval;
399};
400
401/*
402 * Function definitions.
403 */
404
405/* Firewall hooks */
406struct sockopt;
407struct dn_flow_set;
408
409void flush_pipe_ptrs(struct dn_flow_set *match); /* used by dummynet */
410
411typedef int ip_fw_chk_t (struct ip_fw_args *args);
412typedef int ip_fw_ctl_t (struct sockopt *);
413extern ip_fw_chk_t *ip_fw_chk_ptr;
414extern ip_fw_ctl_t *ip_fw_ctl_ptr;
415extern int fw_one_pass;
416extern int fw_enable;
417#define	IPFW_LOADED	(ip_fw_chk_ptr != NULL)
418#endif /* _KERNEL */
419
420#endif /* _IPFW2_H */
421