xref: /dragonfly/sys/net/ipfw/ip_fw2.h (revision 65cc0652)
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: src/sys/netinet/ip_fw2.h,v 1.1.2.2 2002/08/16 11:03:11 luigi Exp $
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 metatada) 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  *
41  * "enum ipfw_opcodes" are the opcodes supported. We can have up
42  * to 256 different opcodes.
43  */
44 
45 enum ipfw_opcodes {		/* arguments (4 byte each)	*/
46 	O_NOP,
47 
48 	O_IP_SRC,		/* u32 = IP			*/
49 	O_IP_SRC_MASK,		/* ip = IP/mask			*/
50 	O_IP_SRC_ME,		/* none				*/
51 	O_IP_SRC_SET,		/* u32=base, arg1=len, bitmap	*/
52 
53 	O_IP_DST,		/* u32 = IP			*/
54 	O_IP_DST_MASK,		/* ip = IP/mask			*/
55 	O_IP_DST_ME,		/* none				*/
56 	O_IP_DST_SET,		/* u32=base, arg1=len, bitmap	*/
57 
58 	O_IP_SRCPORT,		/* (n)port list:mask 4 byte ea	*/
59 	O_IP_DSTPORT,		/* (n)port list:mask 4 byte ea	*/
60 	O_PROTO,		/* arg1=protocol		*/
61 
62 	O_MACADDR2,		/* 2 mac addr:mask		*/
63 	O_MAC_TYPE,		/* same as srcport		*/
64 
65 	O_LAYER2,		/* none				*/
66 	O_IN,			/* none				*/
67 	O_FRAG,			/* none				*/
68 
69 	O_RECV,			/* none				*/
70 	O_XMIT,			/* none				*/
71 	O_VIA,			/* none				*/
72 
73 	O_IPOPT,		/* arg1 = 2*u8 bitmap		*/
74 	O_IPLEN,		/* arg1 = len			*/
75 	O_IPID,			/* arg1 = id			*/
76 
77 	O_IPTOS,		/* arg1 = id			*/
78 	O_IPPRECEDENCE,		/* arg1 = precedence << 5	*/
79 	O_IPTTL,		/* arg1 = TTL			*/
80 
81 	O_IPVER,		/* arg1 = version		*/
82 	O_UID,			/* u32 = id			*/
83 	O_GID,			/* u32 = id			*/
84 	O_ESTAB,		/* none (tcp established)	*/
85 	O_TCPFLAGS,		/* arg1 = 2*u8 bitmap		*/
86 	O_TCPWIN,		/* arg1 = desired win		*/
87 	O_TCPSEQ,		/* u32 = desired seq.		*/
88 	O_TCPACK,		/* u32 = desired seq.		*/
89 	O_ICMPTYPE,		/* u32 = icmp bitmap		*/
90 	O_TCPOPTS,		/* arg1 = 2*u8 bitmap		*/
91 
92 	/* States. */
93 	O_PROBE_STATE,		/* none				*/
94 	O_KEEP_STATE,		/* none				*/
95 	O_LIMIT,		/* ipfw_insn_limit		*/
96 	O_LIMIT_PARENT,		/* dyn_type, not an opcode.	*/
97 
98 	/* Actions. */
99 	O_LOG,			/* ipfw_insn_log		*/
100 	O_PROB,			/* u32 = match probability	*/
101 	O_CHECK_STATE,		/* none				*/
102 	O_ACCEPT,		/* none				*/
103 	O_DENY,			/* none 			*/
104 	O_REJECT,		/* arg1=icmp arg (same as deny)	*/
105 	O_COUNT,		/* none				*/
106 	O_SKIPTO,		/* arg1=next rule number	*/
107 	O_PIPE,			/* arg1=pipe number		*/
108 	O_QUEUE,		/* arg1=queue number		*/
109 	O_DIVERT,		/* arg1=port number		*/
110 	O_TEE,			/* arg1=port number		*/
111 	O_FORWARD_IP,		/* fwd sockaddr			*/
112 	O_FORWARD_MAC,		/* fwd mac			*/
113 
114 	/* Table based filters. */
115 	O_IP_SRC_TABLE,		/* arg1 = tableid		*/
116 	O_IP_DST_TABLE,		/* arg1 = tableid		*/
117 
118 	/* Action. */
119 	O_DEFRAG,		/* none				*/
120 
121 	/* Filters. */
122 	O_IPFRAG,		/* none				*/
123 	O_IP_SRC_IFIP,		/* ipfw_insn_ifip		*/
124 	O_IP_DST_IFIP,		/* ipfw_insn_ifip		*/
125 
126 	/* Translates. */
127 	O_REDIRECT,		/* ipfw_insn_rdr		*/
128 
129 	O_LAST_OPCODE		/* not an opcode!		*/
130 };
131 #ifdef _KERNEL
132 CTASSERT(O_LAST_OPCODE <= 256);
133 #endif
134 
135 /*
136  * Template for instructions.
137  *
138  * ipfw_insn is used for all instructions which require no operands,
139  * a single 16-bit value (arg1), or a couple of 8-bit values.
140  *
141  * For other instructions which require different/larger arguments
142  * we have derived structures, ipfw_insn_*.
143  *
144  * The size of the instruction (in 32-bit words) is in the low
145  * 6 bits of "len". The 2 remaining bits are used to implement
146  * NOT and OR on individual instructions. Given a type, you can
147  * compute the length to be put in "len" using F_INSN_SIZE(t)
148  *
149  * F_NOT	negates the match result of the instruction.
150  *
151  * F_OR		is used to build or blocks. By default, instructions
152  *		are evaluated as part of a logical AND. An "or" block
153  *		{ X or Y or Z } contains F_OR set in all but the last
154  *		instruction of the block. A match will cause the code
155  *		to skip past the last instruction of the block.
156  *
157  * NOTA BENE: in a couple of places we assume that
158  *	sizeof(ipfw_insn) == sizeof(uint32_t)
159  * this needs to be fixed.
160  *
161  */
162 typedef struct	_ipfw_insn {	/* template for instructions */
163 	enum ipfw_opcodes	opcode:8;
164 	uint8_t		len;	/* numer of 32-byte words */
165 #define	F_NOT		0x80
166 #define	F_OR		0x40
167 #define	F_LEN_MASK	0x3f
168 #define	F_LEN(cmd)	((cmd)->len & F_LEN_MASK)
169 
170 	uint16_t	arg1;
171 } ipfw_insn;
172 
173 #define IPFW_INSN_SIZE_MAX	63	/* unit: uint32_t */
174 
175 /*
176  * The F_INSN_SIZE(type) computes the size, in 4-byte words, of
177  * a given type.
178  */
179 #define	F_INSN_SIZE(t)	((sizeof (t))/sizeof(uint32_t))
180 
181 /*
182  * This is used to store an array of 16-bit entries (ports etc.)
183  */
184 typedef struct	_ipfw_insn_u16 {
185 	ipfw_insn o;
186 	uint16_t ports[2];	/* there may be more */
187 } ipfw_insn_u16;
188 
189 /*
190  * This is used to store an array of 32-bit entries
191  * (uid, single IPv4 addresses etc.)
192  */
193 typedef struct	_ipfw_insn_u32 {
194 	ipfw_insn o;
195 	uint32_t d[1];	/* one or more */
196 } ipfw_insn_u32;
197 
198 /*
199  * This is used to store IP addr-mask pairs.
200  */
201 typedef struct	_ipfw_insn_ip {
202 	ipfw_insn o;
203 	struct in_addr	addr;
204 	struct in_addr	mask;
205 } ipfw_insn_ip;
206 
207 /*
208  * This is used to forward to a given address (ip)
209  */
210 typedef struct  _ipfw_insn_sa {
211 	ipfw_insn o;
212 	struct sockaddr_in sa;
213 } ipfw_insn_sa;
214 
215 /*
216  * This is used for MAC addr-mask pairs.
217  */
218 typedef struct	_ipfw_insn_mac {
219 	ipfw_insn o;
220 	u_char addr[12];	/* dst[6] + src[6] */
221 	u_char mask[12];	/* dst[6] + src[6] */
222 } ipfw_insn_mac;
223 
224 /*
225  * This is used for interface match rules (recv xx, xmit xx)
226  */
227 typedef struct	_ipfw_insn_if {
228 	ipfw_insn o;
229 	union {
230 		struct in_addr ip;
231 		int glob;
232 	} p;
233 	char name[IFNAMSIZ];
234 } ipfw_insn_if;
235 
236 /*
237  * This is used for pipe and queue actions, which need to store
238  * a single pointer (which can have different size on different
239  * architectures.
240  */
241 typedef struct	_ipfw_insn_pipe {
242 	ipfw_insn	o;
243 	void		*pipe_ptr;
244 } ipfw_insn_pipe;
245 
246 /*
247  * This is used for limit rules.
248  */
249 typedef struct	_ipfw_insn_limit {
250 	ipfw_insn o;
251 	uint8_t _pad;
252 	uint8_t limit_mask;	/* combination of DYN_* below	*/
253 #define	DYN_SRC_ADDR	0x1
254 #define	DYN_SRC_PORT	0x2
255 #define	DYN_DST_ADDR	0x4
256 #define	DYN_DST_PORT	0x8
257 
258 	uint16_t conn_limit;
259 } ipfw_insn_limit;
260 
261 /*
262  * This is used for log instructions
263  */
264 typedef struct  _ipfw_insn_log {
265         ipfw_insn o;
266 	uint32_t max_log;	/* how many do we log -- 0 = all */
267 	uint32_t log_left;	/* how many left to log 	*/
268 } ipfw_insn_log;
269 
270 /*
271  * This is used by O_IP_{SRC,DST}_IFIP.
272  */
273 typedef struct _ipfw_insn_ifip {
274 	ipfw_insn o;		/* arg1 & 0x1, addr is valid */
275 #define IPFW_IFIP_VALID		0x0001
276 #define IPFW_IFIP_NET		0x0002
277 #define IPFW_IFIP_SETTINGS	IPFW_IFIP_NET
278 	char ifname[IFNAMSIZ];
279 	struct in_addr addr;
280 	struct in_addr mask;
281 } ipfw_insn_ifip;
282 
283 /*
284  * This is used by O_REDIRECT.
285  */
286 typedef struct _ipfw_insn_rdr {
287 	ipfw_insn o;
288 	struct in_addr addr;
289 	uint16_t port;		/* network byte order, 0 = same port */
290 	uint16_t set;		/* reserved for set, 0xffff */
291 } ipfw_insn_rdr;
292 
293 #ifdef _KERNEL
294 
295 /*
296  * Here we have the structure representing an ipfw rule.
297  *
298  * It starts with a general area (with link fields and counters)
299  * followed by an array of one or more instructions, which the code
300  * accesses as an array of 32-bit values.
301  *
302  * Given a rule pointer  r:
303  *
304  *  r->cmd		is the start of the first instruction.
305  *  ACTION_PTR(r)	is the start of the first action (things to do
306  *			once a rule matched).
307  *
308  * When assembling instruction, remember the following:
309  *
310  *  + if a rule has a "keep-state" (or "limit") option, then the
311  *	first instruction (at r->cmd) MUST BE an O_PROBE_STATE
312  *  + if a rule has a "log" option, then the first action
313  *	(at ACTION_PTR(r)) MUST be O_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 	uint16_t	act_ofs;	/* offset of action in 32-bit units */
324 	uint16_t	cmd_len;	/* # of 32-bit words in cmd	*/
325 	uint16_t	rulenum;	/* rule number			*/
326 	uint8_t		set;		/* rule set (0..31)		*/
327 	uint8_t		usr_flags;	/* IPFW_USR_F_			*/
328 
329 	/* These fields are present in all rules.			*/
330 	uint64_t	pcnt;		/* Packet counter		*/
331 	uint64_t	bcnt;		/* Byte counter			*/
332 	uint32_t	timestamp;	/* tv_sec of last match		*/
333 
334 	int		cpuid;		/* owner cpu			*/
335 	struct ip_fw	*sibling;	/* clone on next cpu		*/
336 
337 	struct ip_fw	**cross_rules;	/* cross referenced rules	*/
338 	volatile uint64_t cross_refs;	/* cross references		*/
339 
340 	uint32_t	refcnt;		/* Ref count for transit pkts	*/
341 	uint32_t	rule_flags;	/* IPFW_RULE_F_			*/
342 	uintptr_t	track_ruleid;	/* ruleid for src/dst tracks	*/
343 
344 	ipfw_insn	cmd[1];		/* storage for commands		*/
345 };
346 
347 #define IPFW_RULE_F_INVALID	0x1
348 /* unused			0x2 */
349 #define IPFW_RULE_F_GENSTATE	0x4
350 #define IPFW_RULE_F_GENTRACK	0x8
351 #define IPFW_RULE_F_CROSSREF	0x10
352 #define IPFW_RULE_F_DYNIFADDR	0x20
353 
354 #define RULESIZE(rule)	(sizeof(struct ip_fw) + (rule)->cmd_len * 4 - 4)
355 
356 /*
357  * This structure is used as a flow mask and a flow id for various
358  * parts of the code.
359  */
360 struct ipfw_flow_id {
361 	uint32_t	dst_ip;		/* host byte order */
362 	uint32_t	src_ip;		/* host byte order */
363 	uint16_t	dst_port;	/* host byte order */
364 	uint16_t	src_port;	/* host byte order */
365 	uint8_t		proto;
366 	uint8_t		flags;		/* protocol-specific flags */
367 };
368 
369 /*
370  * Main firewall chains definitions and global var's definitions.
371  */
372 
373 /* ipfw_chk/ip_fw_chk_ptr return values */
374 #define IP_FW_PASS		0
375 #define IP_FW_DENY		1
376 #define IP_FW_DIVERT		2
377 #define IP_FW_TEE		3
378 #define IP_FW_DUMMYNET		4
379 #define IP_FW_REDISPATCH	6
380 
381 /*
382  * arguments for calling ipfw_chk() and dummynet_io(). We put them
383  * all into a structure because this way it is easier and more
384  * efficient to pass variables around and extend the interface.
385  */
386 struct ip_fw_args {
387 	struct mbuf	*m;		/* the mbuf chain		*/
388 	struct ifnet	*oif;		/* output interface		*/
389 	struct ip_fw	*rule;		/* matching rule		*/
390 	struct ipfw_xlat *xlat;		/* matching xlate		*/
391 	struct ether_header *eh;	/* for bridged packets		*/
392 
393 	struct ipfw_flow_id f_id;	/* grabbed from IP header	*/
394 	uint8_t		flags;
395 #define IP_FWARG_F_CONT		0x01
396 #define IP_FWARG_F_XLATINS	0x02
397 #define IP_FWARG_F_XLATFWD	0x04
398 
399 	/*
400 	 * Depend on the return value of ipfw_chk/ip_fw_chk_ptr
401 	 * 'cookie' field may save following information:
402 	 *
403 	 * IP_FW_TEE or IP_FW_DIVERT
404 	 *   The divert port number
405 	 *
406 	 * IP_FW_DUMMYNET
407 	 *   The pipe or queue number
408 	 */
409 	uint32_t	cookie;
410 };
411 
412 /*
413  * Function definitions.
414  */
415 int	ip_fw_sockopt(struct sockopt *);
416 
417 /* Firewall hooks */
418 struct sockopt;
419 struct dn_flow_set;
420 
421 typedef int	ip_fw_chk_t(struct ip_fw_args *);
422 typedef int	ip_fw_ctl_t(struct sockopt *);
423 typedef struct mbuf
424 		*ip_fw_dn_io_t(struct mbuf *, int, int, struct ip_fw_args *);
425 
426 extern ip_fw_chk_t	*ip_fw_chk_ptr;
427 extern ip_fw_ctl_t	*ip_fw_ctl_ptr;
428 extern ip_fw_dn_io_t	*ip_fw_dn_io_ptr;
429 
430 extern int fw_one_pass;
431 extern int fw_enable;
432 
433 extern int ip_fw_loaded;
434 #define	IPFW_LOADED	(ip_fw_loaded)
435 
436 #endif /* _KERNEL */
437 
438 #define ACTION_PTR(rule)	\
439 	(ipfw_insn *)((uint32_t *)((rule)->cmd) + ((rule)->act_ofs))
440 
441 struct ipfw_ioc_rule {
442 	uint16_t	act_ofs;	/* offset of action in 32-bit units */
443 	uint16_t	cmd_len;	/* # of 32-bit words in cmd	*/
444 	uint16_t	rulenum;	/* rule number			*/
445 	uint8_t		set;		/* rule set (0..31)		*/
446 	uint8_t		usr_flags;	/* IPFW_USR_F_ 			*/
447 
448 	/* Rule set information */
449 	uint32_t	set_disable;	/* disabled rule sets		*/
450 	uint32_t	static_count;	/* # of static rules		*/
451 	uint32_t	static_len;	/* total length of static rules	*/
452 
453 	/* Statistics */
454 	uint64_t	pcnt;		/* Packet counter		*/
455 	uint64_t	bcnt;		/* Byte counter			*/
456 	uint32_t	timestamp;	/* tv_sec of last match		*/
457 
458 	uint8_t		reserved[16];
459 
460 	ipfw_insn	cmd[1];		/* storage for commands		*/
461 };
462 
463 #define IPFW_USR_F_NORULE	0x01
464 
465 #define IPFW_RULE_SIZE_MAX	255	/* unit: uint32_t */
466 
467 #define IOC_RULESIZE(rule)	\
468 	(sizeof(struct ipfw_ioc_rule) + (rule)->cmd_len * 4 - 4)
469 
470 struct ipfw_ioc_flowid {
471 	uint16_t	type;	/* ETHERTYPE_ */
472 	uint16_t	pad;
473 	union {
474 		struct {
475 			uint32_t dst_ip;	/* host byte order */
476 			uint32_t src_ip;	/* host byte order */
477 			uint16_t dst_port;	/* host byte order */
478 			uint16_t src_port;	/* host byte order */
479 			uint8_t proto;
480 		} ip;
481 		uint8_t pad[64];
482 	} u;
483 };
484 
485 struct ipfw_ioc_state {
486 	uint32_t	expire;		/* expire time			*/
487 	uint64_t	pcnt;		/* packet match counter		*/
488 	uint64_t	bcnt;		/* byte match counter		*/
489 
490 	uint16_t	dyn_type;	/* rule type			*/
491 	uint16_t	count;		/* refcount			*/
492 
493 	uint16_t	rulenum;	/* rule number			*/
494 
495 	uint16_t	xlat_port;	/* xlate port, host byte order	*/
496 	uint32_t	xlat_addr;	/* xlate addr, host byte order	*/
497 
498 	struct ipfw_ioc_flowid id;	/* (masked) flow id		*/
499 	uint8_t		reserved[16];
500 };
501 
502 /*
503  * Definitions for IP option names.
504  */
505 #define	IP_FW_IPOPT_LSRR	0x01
506 #define	IP_FW_IPOPT_SSRR	0x02
507 #define	IP_FW_IPOPT_RR		0x04
508 #define	IP_FW_IPOPT_TS		0x08
509 
510 /*
511  * Definitions for TCP option names.
512  */
513 #define	IP_FW_TCPOPT_MSS	0x01
514 #define	IP_FW_TCPOPT_WINDOW	0x02
515 #define	IP_FW_TCPOPT_SACK	0x04
516 #define	IP_FW_TCPOPT_TS		0x08
517 #define	IP_FW_TCPOPT_CC		0x10
518 
519 #define	ICMP_REJECT_RST		0x100	/* fake ICMP code (send a TCP RST) */
520 
521 /*
522  * IP_FW_TBL_CREATE, tableid >= 0.
523  * IP_FW_TBL_FLUSH, tableid >= 0.
524  * IP_FW_TBL_FLUSH, tableid < 0, flush all tables.
525  * IP_FW_TBL_DESTROY, tableid >= 0.
526  * IP_FW_TBL_ZERO, tableid >= 0.
527  * IP_FW_TBL_ZERO, tableid < 0, zero all tables' counters.
528  */
529 struct ipfw_ioc_table {
530 	int		tableid;
531 };
532 
533 struct ipfw_ioc_tblent {
534 	struct sockaddr_in key;
535 	struct sockaddr_in netmask;
536 	u_long		use;
537 	time_t		last_used;
538 	long		unused[2];
539 };
540 
541 /*
542  * IP_FW_TBL_GET, tableid < 0, list of all tables.
543  */
544 struct ipfw_ioc_tbllist {
545 	int		tableid;	/* MUST be the first field */
546 	int		tablecnt;
547 	uint16_t	tables[];
548 };
549 
550 /*
551  * IP_FW_TBL_GET, tableid >= 0, entries in the table.
552  * IP_FW_TBL_ADD, tableid >= 0, entcnt == 1.
553  * IP_FW_TBL_DEL, tableid >= 0, entcnt == 1.
554  */
555 struct ipfw_ioc_tblcont {
556 	int		tableid;	/* MUST be the first field */
557 	int		entcnt;
558 	struct ipfw_ioc_tblent ent[1];
559 };
560 
561 /*
562  * IP_FW_TBL_EXPIRE, tableid < 0, expire all tables.
563  * IP_FW_TBL_EXPIRE, tableid >= 0.
564  */
565 struct ipfw_ioc_tblexp {
566 	int		tableid;
567 	int		expcnt;
568 	time_t		expire;
569 	u_long		unused1[2];
570 };
571 
572 #endif /* _IPFW2_H */
573