xref: /titanic_44/usr/src/uts/common/inet/ipf/fil.c (revision bda1f129971950880940a17bab0bf096d5744b0c)
1 /*
2  * Copyright (C) 1993-2003 by Darren Reed.
3  *
4  * See the IPFILTER.LICENCE file for details on licencing.
5  *
6  * Copyright 2009 Sun Microsystems, Inc.  All rights reserved.
7  * Use is subject to license terms.
8  */
9 
10 #if defined(KERNEL) || defined(_KERNEL)
11 # undef KERNEL
12 # undef _KERNEL
13 # define        KERNEL	1
14 # define        _KERNEL	1
15 #endif
16 #include <sys/errno.h>
17 #include <sys/types.h>
18 #include <sys/param.h>
19 #include <sys/time.h>
20 #if defined(__NetBSD__)
21 # if (NetBSD >= 199905) && !defined(IPFILTER_LKM) && defined(_KERNEL)
22 #  include "opt_ipfilter_log.h"
23 # endif
24 #endif
25 #if defined(_KERNEL) && defined(__FreeBSD_version) && \
26     (__FreeBSD_version >= 220000)
27 # if (__FreeBSD_version >= 400000)
28 #  if !defined(IPFILTER_LKM)
29 #   include "opt_inet6.h"
30 #  endif
31 #  if (__FreeBSD_version == 400019)
32 #   define CSUM_DELAY_DATA
33 #  endif
34 # endif
35 # include <sys/filio.h>
36 #else
37 # include <sys/ioctl.h>
38 #endif
39 #if !defined(_AIX51)
40 # include <sys/fcntl.h>
41 #endif
42 #if defined(_KERNEL)
43 # include <sys/systm.h>
44 # include <sys/file.h>
45 #else
46 # include <stdio.h>
47 # include <string.h>
48 # include <stdlib.h>
49 # include <stddef.h>
50 # include <sys/file.h>
51 # define _KERNEL
52 # ifdef __OpenBSD__
53 struct file;
54 # endif
55 # include <sys/uio.h>
56 # undef _KERNEL
57 #endif
58 #if !defined(__SVR4) && !defined(__svr4__) && !defined(__hpux) && \
59     !defined(linux)
60 # include <sys/mbuf.h>
61 #else
62 # if !defined(linux)
63 #  include <sys/byteorder.h>
64 # endif
65 # if (SOLARIS2 < 5) && defined(sun)
66 #  include <sys/dditypes.h>
67 # endif
68 #endif
69 #ifdef __hpux
70 # define _NET_ROUTE_INCLUDED
71 #endif
72 #if !defined(linux)
73 # include <sys/protosw.h>
74 #endif
75 #include <sys/socket.h>
76 #include <net/if.h>
77 #ifdef sun
78 # include <net/af.h>
79 #endif
80 #if !defined(_KERNEL) && defined(__FreeBSD__)
81 # include "radix_ipf.h"
82 #endif
83 #include <net/route.h>
84 #include <netinet/in.h>
85 #include <netinet/in_systm.h>
86 #include <netinet/ip.h>
87 #if !defined(linux)
88 # include <netinet/ip_var.h>
89 #endif
90 #if defined(__sgi) && defined(IFF_DRVRLOCK) /* IRIX 6 */
91 # include <sys/hashing.h>
92 # include <netinet/in_var.h>
93 #endif
94 #include <netinet/tcp.h>
95 #if (!defined(__sgi) && !defined(AIX)) || defined(_KERNEL)
96 # include <netinet/udp.h>
97 # include <netinet/ip_icmp.h>
98 #endif
99 #ifdef __hpux
100 # undef _NET_ROUTE_INCLUDED
101 #endif
102 #include "netinet/ip_compat.h"
103 #ifdef	USE_INET6
104 # include <netinet/icmp6.h>
105 # if !SOLARIS && defined(_KERNEL) && !defined(__osf__) && !defined(__hpux)
106 #  include <netinet6/in6_var.h>
107 # endif
108 #endif
109 #include <netinet/tcpip.h>
110 #include "netinet/ip_fil.h"
111 #include "netinet/ip_nat.h"
112 #include "netinet/ip_frag.h"
113 #include "netinet/ip_state.h"
114 #include "netinet/ip_proxy.h"
115 #include "netinet/ip_auth.h"
116 #include "netinet/ipf_stack.h"
117 #ifdef IPFILTER_SCAN
118 # include "netinet/ip_scan.h"
119 #endif
120 #ifdef IPFILTER_SYNC
121 # include "netinet/ip_sync.h"
122 #endif
123 #include "netinet/ip_pool.h"
124 #include "netinet/ip_htable.h"
125 #ifdef IPFILTER_COMPILED
126 # include "netinet/ip_rules.h"
127 #endif
128 #if defined(IPFILTER_BPF) && defined(_KERNEL)
129 # include <net/bpf.h>
130 #endif
131 #if defined(__FreeBSD_version) && (__FreeBSD_version >= 300000)
132 # include <sys/malloc.h>
133 # if defined(_KERNEL) && !defined(IPFILTER_LKM)
134 #  include "opt_ipfilter.h"
135 # endif
136 #endif
137 #include "netinet/ipl.h"
138 /* END OF INCLUDES */
139 
140 #if !defined(lint)
141 static const char sccsid[] = "@(#)fil.c	1.36 6/5/96 (C) 1993-2000 Darren Reed";
142 static const char rcsid[] = "@(#)$Id: fil.c,v 2.243.2.64 2005/08/13 05:19:59 darrenr Exp $";
143 #endif
144 
145 #ifndef	_KERNEL
146 # include "ipf.h"
147 # include "ipt.h"
148 # include "bpf-ipf.h"
149 extern	int	opts;
150 
151 # define	FR_VERBOSE(verb_pr)			verbose verb_pr
152 # define	FR_DEBUG(verb_pr)			debug verb_pr
153 #else /* #ifndef _KERNEL */
154 # define	FR_VERBOSE(verb_pr)
155 # define	FR_DEBUG(verb_pr)
156 #endif /* _KERNEL */
157 
158 
159 char	ipfilter_version[] = IPL_VERSION;
160 int	fr_features = 0
161 #ifdef	IPFILTER_LKM
162 		| IPF_FEAT_LKM
163 #endif
164 #ifdef	IPFILTER_LOG
165 		| IPF_FEAT_LOG
166 #endif
167 #ifdef	IPFILTER_LOOKUP
168 		| IPF_FEAT_LOOKUP
169 #endif
170 #ifdef	IPFILTER_BPF
171 		| IPF_FEAT_BPF
172 #endif
173 #ifdef	IPFILTER_COMPILED
174 		| IPF_FEAT_COMPILED
175 #endif
176 #ifdef	IPFILTER_CKSUM
177 		| IPF_FEAT_CKSUM
178 #endif
179 #ifdef	IPFILTER_SYNC
180 		| IPF_FEAT_SYNC
181 #endif
182 #ifdef	IPFILTER_SCAN
183 		| IPF_FEAT_SCAN
184 #endif
185 #ifdef	USE_INET6
186 		| IPF_FEAT_IPV6
187 #endif
188 	;
189 
190 #define	IPF_BUMP(x)	(x)++
191 
192 static	INLINE int	fr_ipfcheck __P((fr_info_t *, frentry_t *, int));
193 static	INLINE int	fr_ipfcheck __P((fr_info_t *, frentry_t *, int));
194 static	int		fr_portcheck __P((frpcmp_t *, u_short *));
195 static	int		frflushlist __P((int, minor_t, int *, frentry_t **,
196 					 ipf_stack_t *));
197 static	ipfunc_t	fr_findfunc __P((ipfunc_t));
198 static	frentry_t	*fr_firewall __P((fr_info_t *, u_32_t *));
199 static	int		fr_funcinit __P((frentry_t *fr, ipf_stack_t *));
200 static	INLINE void	frpr_ah __P((fr_info_t *));
201 static	INLINE void	frpr_esp __P((fr_info_t *));
202 static	INLINE void	frpr_gre __P((fr_info_t *));
203 static	INLINE void	frpr_udp __P((fr_info_t *));
204 static	INLINE void	frpr_tcp __P((fr_info_t *));
205 static	INLINE void	frpr_icmp __P((fr_info_t *));
206 static	INLINE void	frpr_ipv4hdr __P((fr_info_t *));
207 static	INLINE int	frpr_pullup __P((fr_info_t *, int));
208 static	INLINE void	frpr_short __P((fr_info_t *, int));
209 static	INLINE void	frpr_tcpcommon __P((fr_info_t *));
210 static	INLINE void	frpr_udpcommon __P((fr_info_t *));
211 static	INLINE int	fr_updateipid __P((fr_info_t *));
212 #ifdef	IPFILTER_LOOKUP
213 static	int		fr_grpmapinit __P((frentry_t *fr, ipf_stack_t *));
214 static	INLINE void	*fr_resolvelookup __P((u_int, u_int, lookupfunc_t *,
215 					       ipf_stack_t *));
216 #endif
217 static	void		frsynclist __P((int, int, void *, char *, frentry_t *,
218     ipf_stack_t *));
219 static	void		*fr_ifsync __P((int, int, char *, char *,
220 					void *, void *, ipf_stack_t *));
221 static	ipftuneable_t	*fr_findtunebyname __P((const char *, ipf_stack_t *));
222 static	ipftuneable_t	*fr_findtunebycookie __P((void *, void **, ipf_stack_t *));
223 
224 /*
225  * bit values for identifying presence of individual IP options
226  * All of these tables should be ordered by increasing key value on the left
227  * hand side to allow for binary searching of the array and include a trailer
228  * with a 0 for the bitmask for linear searches to easily find the end with.
229  */
230 const	struct	optlist	ipopts[20] = {
231 	{ IPOPT_NOP,	0x000001 },
232 	{ IPOPT_RR,	0x000002 },
233 	{ IPOPT_ZSU,	0x000004 },
234 	{ IPOPT_MTUP,	0x000008 },
235 	{ IPOPT_MTUR,	0x000010 },
236 	{ IPOPT_ENCODE,	0x000020 },
237 	{ IPOPT_TS,	0x000040 },
238 	{ IPOPT_TR,	0x000080 },
239 	{ IPOPT_SECURITY, 0x000100 },
240 	{ IPOPT_LSRR,	0x000200 },
241 	{ IPOPT_E_SEC,	0x000400 },
242 	{ IPOPT_CIPSO,	0x000800 },
243 	{ IPOPT_SATID,	0x001000 },
244 	{ IPOPT_SSRR,	0x002000 },
245 	{ IPOPT_ADDEXT,	0x004000 },
246 	{ IPOPT_VISA,	0x008000 },
247 	{ IPOPT_IMITD,	0x010000 },
248 	{ IPOPT_EIP,	0x020000 },
249 	{ IPOPT_FINN,	0x040000 },
250 	{ 0,		0x000000 }
251 };
252 
253 #ifdef USE_INET6
254 struct optlist ip6exthdr[] = {
255 	{ IPPROTO_HOPOPTS,		0x000001 },
256 	{ IPPROTO_IPV6,			0x000002 },
257 	{ IPPROTO_ROUTING,		0x000004 },
258 	{ IPPROTO_FRAGMENT,		0x000008 },
259 	{ IPPROTO_ESP,			0x000010 },
260 	{ IPPROTO_AH,			0x000020 },
261 	{ IPPROTO_NONE,			0x000040 },
262 	{ IPPROTO_DSTOPTS,		0x000080 },
263 	{ 0,				0 }
264 };
265 #endif
266 
267 struct optlist tcpopts[] = {
268 	{ TCPOPT_NOP,			0x000001 },
269 	{ TCPOPT_MAXSEG,		0x000002 },
270 	{ TCPOPT_WINDOW,		0x000004 },
271 	{ TCPOPT_SACK_PERMITTED,	0x000008 },
272 	{ TCPOPT_SACK,			0x000010 },
273 	{ TCPOPT_TIMESTAMP,		0x000020 },
274 	{ 0,				0x000000 }
275 };
276 
277 /*
278  * bit values for identifying presence of individual IP security options
279  */
280 const	struct	optlist	secopt[8] = {
281 	{ IPSO_CLASS_RES4,	0x01 },
282 	{ IPSO_CLASS_TOPS,	0x02 },
283 	{ IPSO_CLASS_SECR,	0x04 },
284 	{ IPSO_CLASS_RES3,	0x08 },
285 	{ IPSO_CLASS_CONF,	0x10 },
286 	{ IPSO_CLASS_UNCL,	0x20 },
287 	{ IPSO_CLASS_RES2,	0x40 },
288 	{ IPSO_CLASS_RES1,	0x80 }
289 };
290 
291 
292 /*
293  * Table of functions available for use with call rules.
294  */
295 static ipfunc_resolve_t fr_availfuncs[] = {
296 #ifdef	IPFILTER_LOOKUP
297 	{ "fr_srcgrpmap", fr_srcgrpmap, fr_grpmapinit },
298 	{ "fr_dstgrpmap", fr_dstgrpmap, fr_grpmapinit },
299 #endif
300 	{ "", NULL }
301 };
302 
303 
304 /*
305  * Below we declare a list of constants used only by the ipf_extraflush()
306  * routine.  We are placing it here, instead of in ipf_extraflush() itself,
307  * because we want to make it visible to tools such as mdb, nm etc., so the
308  * values can easily be altered during debugging.
309  */
310 static	const	int	idletime_tab[] = {
311 	IPF_TTLVAL(30),		/* 30 seconds */
312 	IPF_TTLVAL(1800),	/* 30 minutes */
313 	IPF_TTLVAL(43200),	/* 12 hours */
314 	IPF_TTLVAL(345600),	/* 4 days */
315 };
316 
317 
318 /*
319  * The next section of code is a a collection of small routines that set
320  * fields in the fr_info_t structure passed based on properties of the
321  * current packet.  There are different routines for the same protocol
322  * for each of IPv4 and IPv6.  Adding a new protocol, for which there
323  * will "special" inspection for setup, is now more easily done by adding
324  * a new routine and expanding the frpr_ipinit*() function rather than by
325  * adding more code to a growing switch statement.
326  */
327 #ifdef USE_INET6
328 static	INLINE int	frpr_ah6 __P((fr_info_t *));
329 static	INLINE void	frpr_esp6 __P((fr_info_t *));
330 static	INLINE void	frpr_gre6 __P((fr_info_t *));
331 static	INLINE void	frpr_udp6 __P((fr_info_t *));
332 static	INLINE void	frpr_tcp6 __P((fr_info_t *));
333 static	INLINE void	frpr_icmp6 __P((fr_info_t *));
334 static	INLINE void	frpr_ipv6hdr __P((fr_info_t *));
335 static	INLINE void	frpr_short6 __P((fr_info_t *, int));
336 static	INLINE int	frpr_hopopts6 __P((fr_info_t *));
337 static	INLINE int	frpr_routing6 __P((fr_info_t *));
338 static	INLINE int	frpr_dstopts6 __P((fr_info_t *));
339 static	INLINE int	frpr_fragment6 __P((fr_info_t *));
340 static	INLINE int	frpr_ipv6exthdr __P((fr_info_t *, int, int));
341 
342 
343 /* ------------------------------------------------------------------------ */
344 /* Function:    frpr_short6                                                 */
345 /* Returns:     void                                                        */
346 /* Parameters:  fin(I) - pointer to packet information                      */
347 /*                                                                          */
348 /* IPv6 Only                                                                */
349 /* This is function enforces the 'is a packet too short to be legit' rule   */
350 /* for IPv6 and marks the packet with FI_SHORT if so.  See function comment */
351 /* for frpr_short() for more details.                                       */
352 /* ------------------------------------------------------------------------ */
353 static INLINE void frpr_short6(fin, xmin)
354 fr_info_t *fin;
355 int xmin;
356 {
357 
358 	if (fin->fin_dlen < xmin)
359 		fin->fin_flx |= FI_SHORT;
360 }
361 
362 
363 /* ------------------------------------------------------------------------ */
364 /* Function:    frpr_ipv6hdr                                                */
365 /* Returns:     Nil                                                         */
366 /* Parameters:  fin(I) - pointer to packet information                      */
367 /*                                                                          */
368 /* IPv6 Only                                                                */
369 /* Copy values from the IPv6 header into the fr_info_t struct and call the  */
370 /* per-protocol analyzer if it exists.                                      */
371 /* ------------------------------------------------------------------------ */
372 static INLINE void frpr_ipv6hdr(fin)
373 fr_info_t *fin;
374 {
375 	ip6_t *ip6 = (ip6_t *)fin->fin_ip;
376 	int p, go = 1, i, hdrcount;
377 	fr_ip_t *fi = &fin->fin_fi;
378 
379 	fin->fin_off = 0;
380 
381 	fi->fi_tos = 0;
382 	fi->fi_optmsk = 0;
383 	fi->fi_secmsk = 0;
384 	fi->fi_auth = 0;
385 
386 	p = ip6->ip6_nxt;
387 	fi->fi_ttl = ip6->ip6_hlim;
388 	fi->fi_src.in6 = ip6->ip6_src;
389 	fi->fi_dst.in6 = ip6->ip6_dst;
390 	fin->fin_id = 0;
391 
392 	hdrcount = 0;
393 	while (go && !(fin->fin_flx & (FI_BAD|FI_SHORT))) {
394 		switch (p)
395 		{
396 		case IPPROTO_UDP :
397 			frpr_udp6(fin);
398 			go = 0;
399 			break;
400 
401 		case IPPROTO_TCP :
402 			frpr_tcp6(fin);
403 			go = 0;
404 			break;
405 
406 		case IPPROTO_ICMPV6 :
407 			frpr_icmp6(fin);
408 			go = 0;
409 			break;
410 
411 		case IPPROTO_GRE :
412 			frpr_gre6(fin);
413 			go = 0;
414 			break;
415 
416 		case IPPROTO_HOPOPTS :
417 			/*
418 			 * hop by hop ext header is only allowed
419 			 * right after IPv6 header.
420 			 */
421 			if (hdrcount != 0) {
422 				fin->fin_flx |= FI_BAD;
423 				p = IPPROTO_NONE;
424 			} else {
425 				p = frpr_hopopts6(fin);
426 			}
427 			break;
428 
429 		case IPPROTO_DSTOPTS :
430 			p = frpr_dstopts6(fin);
431 			break;
432 
433 		case IPPROTO_ROUTING :
434 			p = frpr_routing6(fin);
435 			break;
436 
437 		case IPPROTO_AH :
438 			p = frpr_ah6(fin);
439 			break;
440 
441 		case IPPROTO_ESP :
442 			frpr_esp6(fin);
443 			go = 0;
444 			break;
445 
446 		case IPPROTO_IPV6 :
447 			for (i = 0; ip6exthdr[i].ol_bit != 0; i++)
448 				if (ip6exthdr[i].ol_val == p) {
449 					fin->fin_flx |= ip6exthdr[i].ol_bit;
450 					break;
451 				}
452 			go = 0;
453 			break;
454 
455 		case IPPROTO_NONE :
456 			go = 0;
457 			break;
458 
459 		case IPPROTO_FRAGMENT :
460 			p = frpr_fragment6(fin);
461 			if (fin->fin_off != 0)  /* Not the first frag */
462 				go = 0;
463 			break;
464 
465 		default :
466 			go = 0;
467 			break;
468 		}
469 		hdrcount++;
470 
471 		/*
472 		 * It is important to note that at this point, for the
473 		 * extension headers (go != 0), the entire header may not have
474 		 * been pulled up when the code gets to this point.  This is
475 		 * only done for "go != 0" because the other header handlers
476 		 * will all pullup their complete header.  The other indicator
477 		 * of an incomplete packet is that this was just an extension
478 		 * header.
479 		 */
480 		if ((go != 0) && (p != IPPROTO_NONE) &&
481 		    (frpr_pullup(fin, 0) == -1)) {
482 			p = IPPROTO_NONE;
483 			go = 0;
484 		}
485 	}
486 	fi->fi_p = p;
487 }
488 
489 
490 /* ------------------------------------------------------------------------ */
491 /* Function:    frpr_ipv6exthdr                                             */
492 /* Returns:     int    - value of the next header or IPPROTO_NONE if error  */
493 /* Parameters:  fin(I)      - pointer to packet information                 */
494 /*              multiple(I) - flag indicating yes/no if multiple occurances */
495 /*                            of this extension header are allowed.         */
496 /*              proto(I)    - protocol number for this extension header     */
497 /*                                                                          */
498 /* IPv6 Only                                                                */
499 /* ------------------------------------------------------------------------ */
500 static INLINE int frpr_ipv6exthdr(fin, multiple, proto)
501 fr_info_t *fin;
502 int multiple, proto;
503 {
504 	struct ip6_ext *hdr;
505 	u_short shift;
506 	int i;
507 
508 	fin->fin_flx |= FI_V6EXTHDR;
509 
510 				/* 8 is default length of extension hdr */
511 	if ((fin->fin_dlen - 8) < 0) {
512 		fin->fin_flx |= FI_SHORT;
513 		return IPPROTO_NONE;
514 	}
515 
516 	if (frpr_pullup(fin, 8) == -1)
517 		return IPPROTO_NONE;
518 
519 	hdr = fin->fin_dp;
520 	shift = 8 + (hdr->ip6e_len << 3);
521 	if (shift > fin->fin_dlen) {	/* Nasty extension header length? */
522 		fin->fin_flx |= FI_BAD;
523 		return IPPROTO_NONE;
524 	}
525 
526 	for (i = 0; ip6exthdr[i].ol_bit != 0; i++)
527 		if (ip6exthdr[i].ol_val == proto) {
528 			/*
529 			 * Most IPv6 extension headers are only allowed once.
530 			 */
531 			if ((multiple == 0) &&
532 			    ((fin->fin_optmsk & ip6exthdr[i].ol_bit) != 0))
533 				fin->fin_flx |= FI_BAD;
534 			else
535 				fin->fin_optmsk |= ip6exthdr[i].ol_bit;
536 			break;
537 		}
538 
539 	fin->fin_dp = (char *)fin->fin_dp + shift;
540 	fin->fin_dlen -= shift;
541 
542 	return hdr->ip6e_nxt;
543 }
544 
545 
546 /* ------------------------------------------------------------------------ */
547 /* Function:    frpr_hopopts6                                               */
548 /* Returns:     int    - value of the next header or IPPROTO_NONE if error  */
549 /* Parameters:  fin(I) - pointer to packet information                      */
550 /*                                                                          */
551 /* IPv6 Only                                                                */
552 /* This is function checks pending hop by hop options extension header      */
553 /* ------------------------------------------------------------------------ */
554 static INLINE int frpr_hopopts6(fin)
555 fr_info_t *fin;
556 {
557 	return frpr_ipv6exthdr(fin, 0, IPPROTO_HOPOPTS);
558 }
559 
560 
561 /* ------------------------------------------------------------------------ */
562 /* Function:    frpr_routing6                                               */
563 /* Returns:     int    - value of the next header or IPPROTO_NONE if error  */
564 /* Parameters:  fin(I) - pointer to packet information                      */
565 /*                                                                          */
566 /* IPv6 Only                                                                */
567 /* This is function checks pending routing extension header                 */
568 /* ------------------------------------------------------------------------ */
569 static INLINE int frpr_routing6(fin)
570 fr_info_t *fin;
571 {
572 	struct ip6_ext *hdr;
573 	int shift;
574 
575 	hdr = fin->fin_dp;
576 	if (frpr_ipv6exthdr(fin, 0, IPPROTO_ROUTING) == IPPROTO_NONE)
577 		return IPPROTO_NONE;
578 
579 	shift = 8 + (hdr->ip6e_len << 3);
580 	/*
581 	 * Nasty extension header length?
582 	 */
583 	if ((hdr->ip6e_len << 3) & 15) {
584 		fin->fin_flx |= FI_BAD;
585 		/*
586 		 * Compensate for the changes made in frpr_ipv6exthdr()
587 		 */
588 		fin->fin_dlen += shift;
589 		fin->fin_dp = (char *)fin->fin_dp - shift;
590 		return IPPROTO_NONE;
591 	}
592 
593 	return hdr->ip6e_nxt;
594 }
595 
596 
597 /* ------------------------------------------------------------------------ */
598 /* Function:    frpr_fragment6                                              */
599 /* Returns:     int    - value of the next header or IPPROTO_NONE if error  */
600 /* Parameters:  fin(I) - pointer to packet information                      */
601 /*                                                                          */
602 /* IPv6 Only                                                                */
603 /* Examine the IPv6 fragment header and extract fragment offset information.*/
604 /*                                                                          */
605 /* We don't know where the transport layer header (or whatever is next is), */
606 /* as it could be behind destination options (amongst others).  Because     */
607 /* there is no fragment cache, there is no knowledge about whether or not an*/
608 /* upper layer header has been seen (or where it ends) and thus we are not  */
609 /* able to continue processing beyond this header with any confidence.      */
610 /* ------------------------------------------------------------------------ */
611 static INLINE int frpr_fragment6(fin)
612 fr_info_t *fin;
613 {
614 	struct ip6_frag *frag;
615 	int dlen;
616 
617 	fin->fin_flx |= FI_FRAG;
618 
619 	dlen = fin->fin_dlen;
620 	if (frpr_ipv6exthdr(fin, 0, IPPROTO_FRAGMENT) == IPPROTO_NONE)
621 		return IPPROTO_NONE;
622 
623 	if (frpr_pullup(fin, sizeof(*frag)) == -1)
624 		return IPPROTO_NONE;
625 
626 	frpr_short6(fin, sizeof(*frag));
627 
628 	if ((fin->fin_flx & FI_SHORT) != 0)
629 		return IPPROTO_NONE;
630 
631 	frag = (struct ip6_frag *)((char *)fin->fin_dp - sizeof(*frag));
632 	/*
633 	 * Fragment but no fragmentation info set?  Bad packet...
634 	 */
635 	if (frag->ip6f_offlg == 0) {
636 		fin->fin_flx |= FI_BAD;
637 		return IPPROTO_NONE;
638 	}
639 
640 	fin->fin_id = frag->ip6f_ident;
641 	fin->fin_off = frag->ip6f_offlg & IP6F_OFF_MASK;
642 	fin->fin_off = ntohs(fin->fin_off);
643 	if (fin->fin_off != 0)
644 		fin->fin_flx |= FI_FRAGBODY;
645 
646 	fin->fin_dp = (char *)frag + sizeof(*frag);
647 	fin->fin_dlen = dlen - sizeof(*frag);
648 
649 	/* length of hdrs(after frag hdr) + data */
650 	fin->fin_flen = fin->fin_dlen;
651 
652 	/*
653 	 * If the frag is not the last one and the payload length
654 	 * is not multiple of 8, it must be dropped.
655 	 */
656 	if ((frag->ip6f_offlg & IP6F_MORE_FRAG) && (dlen % 8)) {
657 		fin->fin_flx |= FI_BAD;
658 		return IPPROTO_NONE;
659 	}
660 
661 	return frag->ip6f_nxt;
662 }
663 
664 
665 /* ------------------------------------------------------------------------ */
666 /* Function:    frpr_dstopts6                                               */
667 /* Returns:     int    - value of the next header or IPPROTO_NONE if error  */
668 /* Parameters:  fin(I) - pointer to packet information                      */
669 /*              nextheader(I) - stores next header value                    */
670 /*                                                                          */
671 /* IPv6 Only                                                                */
672 /* This is function checks pending destination options extension header     */
673 /* ------------------------------------------------------------------------ */
674 static INLINE int frpr_dstopts6(fin)
675 fr_info_t *fin;
676 {
677 	return frpr_ipv6exthdr(fin, 1, IPPROTO_DSTOPTS);
678 }
679 
680 
681 /* ------------------------------------------------------------------------ */
682 /* Function:    frpr_icmp6                                                  */
683 /* Returns:     void                                                        */
684 /* Parameters:  fin(I) - pointer to packet information                      */
685 /*                                                                          */
686 /* IPv6 Only                                                                */
687 /* This routine is mainly concerned with determining the minimum valid size */
688 /* for an ICMPv6 packet.                                                    */
689 /* ------------------------------------------------------------------------ */
690 static INLINE void frpr_icmp6(fin)
691 fr_info_t *fin;
692 {
693 	int minicmpsz = sizeof(struct icmp6_hdr);
694 	struct icmp6_hdr *icmp6;
695 
696 	if (frpr_pullup(fin, ICMP6ERR_MINPKTLEN - sizeof(ip6_t)) == -1)
697 		return;
698 
699 	if (fin->fin_dlen > 1) {
700 		icmp6 = fin->fin_dp;
701 
702 		fin->fin_data[0] = *(u_short *)icmp6;
703 
704 		if ((icmp6->icmp6_type & ICMP6_INFOMSG_MASK) != 0)
705 			fin->fin_flx |= FI_ICMPQUERY;
706 
707 		switch (icmp6->icmp6_type)
708 		{
709 		case ICMP6_ECHO_REPLY :
710 		case ICMP6_ECHO_REQUEST :
711 			if (fin->fin_dlen >= 6)
712 				fin->fin_data[1] = icmp6->icmp6_id;
713 			minicmpsz = ICMP6ERR_MINPKTLEN - sizeof(ip6_t);
714 			break;
715 		case ICMP6_DST_UNREACH :
716 		case ICMP6_PACKET_TOO_BIG :
717 		case ICMP6_TIME_EXCEEDED :
718 		case ICMP6_PARAM_PROB :
719 			if ((fin->fin_m != NULL) &&
720 			    (M_LEN(fin->fin_m) < fin->fin_plen)) {
721 				if (fr_coalesce(fin) != 1)
722 					return;
723 			}
724 			fin->fin_flx |= FI_ICMPERR;
725 			minicmpsz = ICMP6ERR_IPICMPHLEN - sizeof(ip6_t);
726 			break;
727 		default :
728 			break;
729 		}
730 	}
731 
732 	frpr_short6(fin, minicmpsz);
733 	fin->fin_flen -= fin->fin_dlen - minicmpsz;
734 }
735 
736 
737 /* ------------------------------------------------------------------------ */
738 /* Function:    frpr_udp6                                                   */
739 /* Returns:     void                                                        */
740 /* Parameters:  fin(I) - pointer to packet information                      */
741 /*                                                                          */
742 /* IPv6 Only                                                                */
743 /* Analyse the packet for IPv6/UDP properties.                              */
744 /* Is not expected to be called for fragmented packets.                     */
745 /* ------------------------------------------------------------------------ */
746 static INLINE void frpr_udp6(fin)
747 fr_info_t *fin;
748 {
749 
750 	fr_checkv6sum(fin);
751 
752 	frpr_short6(fin, sizeof(struct udphdr));
753 	if (frpr_pullup(fin, sizeof(struct udphdr)) == -1)
754 		return;
755 
756 	fin->fin_flen -= fin->fin_dlen - sizeof(struct udphdr);
757 
758 	frpr_udpcommon(fin);
759 }
760 
761 
762 /* ------------------------------------------------------------------------ */
763 /* Function:    frpr_tcp6                                                   */
764 /* Returns:     void                                                        */
765 /* Parameters:  fin(I) - pointer to packet information                      */
766 /*                                                                          */
767 /* IPv6 Only                                                                */
768 /* Analyse the packet for IPv6/TCP properties.                              */
769 /* Is not expected to be called for fragmented packets.                     */
770 /* ------------------------------------------------------------------------ */
771 static INLINE void frpr_tcp6(fin)
772 fr_info_t *fin;
773 {
774 
775 	fr_checkv6sum(fin);
776 
777 	frpr_short6(fin, sizeof(struct tcphdr));
778 	if (frpr_pullup(fin, sizeof(struct tcphdr)) == -1)
779 		return;
780 
781 	fin->fin_flen -= fin->fin_dlen - sizeof(struct tcphdr);
782 
783 	frpr_tcpcommon(fin);
784 }
785 
786 
787 /* ------------------------------------------------------------------------ */
788 /* Function:    frpr_esp6                                                   */
789 /* Returns:     void                                                        */
790 /* Parameters:  fin(I) - pointer to packet information                      */
791 /*                                                                          */
792 /* IPv6 Only                                                                */
793 /* Analyse the packet for ESP properties.                                   */
794 /* The minimum length is taken to be the SPI (32bits) plus a tail (32bits)  */
795 /* even though the newer ESP packets must also have a sequence number that  */
796 /* is 32bits as well, it is not possible(?) to determine the version from a */
797 /* simple packet header.                                                    */
798 /* ------------------------------------------------------------------------ */
799 static INLINE void frpr_esp6(fin)
800 fr_info_t *fin;
801 {
802 	int i;
803 	frpr_short6(fin, sizeof(grehdr_t));
804 
805 	(void) frpr_pullup(fin, 8);
806 
807 	for (i = 0; ip6exthdr[i].ol_bit != 0; i++)
808 		if (ip6exthdr[i].ol_val == IPPROTO_ESP) {
809 			fin->fin_optmsk |= ip6exthdr[i].ol_bit;
810 			break;
811 		}
812 }
813 
814 
815 /* ------------------------------------------------------------------------ */
816 /* Function:    frpr_ah6                                                    */
817 /* Returns:     void                                                        */
818 /* Parameters:  fin(I) - pointer to packet information                      */
819 /*                                                                          */
820 /* IPv6 Only                                                                */
821 /* Analyse the packet for AH properties.                                    */
822 /* The minimum length is taken to be the combination of all fields in the   */
823 /* header being present and no authentication data (null algorithm used.)   */
824 /* ------------------------------------------------------------------------ */
825 static INLINE int frpr_ah6(fin)
826 fr_info_t *fin;
827 {
828 	authhdr_t *ah;
829 	int i, shift;
830 
831 	frpr_short6(fin, 12);
832 
833 	if (frpr_pullup(fin, sizeof(*ah)) == -1)
834 		return IPPROTO_NONE;
835 
836 	for (i = 0; ip6exthdr[i].ol_bit != 0; i++)
837 		if (ip6exthdr[i].ol_val == IPPROTO_AH) {
838 			fin->fin_optmsk |= ip6exthdr[i].ol_bit;
839 			break;
840 		}
841 
842 	ah = (authhdr_t *)fin->fin_dp;
843 
844 	shift = (ah->ah_plen + 2) * 4;
845 	fin->fin_dlen -= shift;
846 	fin->fin_dp = (char*)fin->fin_dp + shift;
847 
848 	return ah->ah_next;
849 }
850 
851 
852 /* ------------------------------------------------------------------------ */
853 /* Function:    frpr_gre6                                                   */
854 /* Returns:     void                                                        */
855 /* Parameters:  fin(I) - pointer to packet information                      */
856 /*                                                                          */
857 /* Analyse the packet for GRE properties.                                   */
858 /* ------------------------------------------------------------------------ */
859 static INLINE void frpr_gre6(fin)
860 fr_info_t *fin;
861 {
862 	grehdr_t *gre;
863 
864 	frpr_short6(fin, sizeof(grehdr_t));
865 
866 	if (frpr_pullup(fin, sizeof(grehdr_t)) == -1)
867 		return;
868 
869 	gre = fin->fin_dp;
870 	if (GRE_REV(gre->gr_flags) == 1)
871 		fin->fin_data[0] = gre->gr_call;
872 }
873 #endif	/* USE_INET6 */
874 
875 
876 /* ------------------------------------------------------------------------ */
877 /* Function:    frpr_pullup                                                 */
878 /* Returns:     int     - 0 == pullup succeeded, -1 == failure              */
879 /* Parameters:  fin(I)  - pointer to packet information                     */
880 /*              plen(I) - length (excluding L3 header) to pullup            */
881 /*                                                                          */
882 /* Short inline function to cut down on code duplication to perform a call  */
883 /* to fr_pullup to ensure there is the required amount of data,             */
884 /* consecutively in the packet buffer.                                      */
885 /* ------------------------------------------------------------------------ */
886 static INLINE int frpr_pullup(fin, plen)
887 fr_info_t *fin;
888 int plen;
889 {
890 #if defined(_KERNEL)
891 	if (fin->fin_m != NULL) {
892 		int ipoff;
893 
894 		ipoff = (char *)fin->fin_ip - MTOD(fin->fin_m, char *);
895 
896 		if (fin->fin_dp != NULL)
897 			plen += (char *)fin->fin_dp -
898 				((char *)fin->fin_ip + fin->fin_hlen);
899 		plen += fin->fin_hlen;
900 		/*
901 		 * We don't do 'plen += ipoff;' here. The fr_pullup() will
902 		 * do it for us.
903 		 */
904 		if (M_LEN(fin->fin_m) < plen + ipoff) {
905 			if (fr_pullup(fin->fin_m, fin, plen) == NULL)
906 				return -1;
907 		}
908 	}
909 #endif
910 	return 0;
911 }
912 
913 
914 /* ------------------------------------------------------------------------ */
915 /* Function:    frpr_short                                                  */
916 /* Returns:     void                                                        */
917 /* Parameters:  fin(I)  - pointer to packet information                     */
918 /*              xmin(I) - minimum header size                               */
919 /*                                                                          */
920 /* Check if a packet is "short" as defined by xmin.  The rule we are        */
921 /* applying here is that the packet must not be fragmented within the layer */
922 /* 4 header.  That is, it must not be a fragment that has its offset set to */
923 /* start within the layer 4 header (hdrmin) or if it is at offset 0, the    */
924 /* entire layer 4 header must be present (min).                             */
925 /* ------------------------------------------------------------------------ */
926 static INLINE void frpr_short(fin, xmin)
927 fr_info_t *fin;
928 int xmin;
929 {
930 
931 	if (fin->fin_off == 0) {
932 		if (fin->fin_dlen < xmin)
933 			fin->fin_flx |= FI_SHORT;
934 	} else if (fin->fin_off < xmin) {
935 		fin->fin_flx |= FI_SHORT;
936 	}
937 }
938 
939 
940 /* ------------------------------------------------------------------------ */
941 /* Function:    frpr_icmp                                                   */
942 /* Returns:     void                                                        */
943 /* Parameters:  fin(I) - pointer to packet information                      */
944 /*                                                                          */
945 /* IPv4 Only                                                                */
946 /* Do a sanity check on the packet for ICMP (v4).  In nearly all cases,     */
947 /* except extrememly bad packets, both type and code will be present.       */
948 /* The expected minimum size of an ICMP packet is very much dependent on    */
949 /* the type of it.                                                          */
950 /*                                                                          */
951 /* XXX - other ICMP sanity checks?                                          */
952 /* ------------------------------------------------------------------------ */
953 static INLINE void frpr_icmp(fin)
954 fr_info_t *fin;
955 {
956 	int minicmpsz = sizeof(struct icmp);
957 	icmphdr_t *icmp;
958 	ip_t *oip;
959 	ipf_stack_t *ifs = fin->fin_ifs;
960 
961 	if (fin->fin_off != 0) {
962 		frpr_short(fin, ICMPERR_ICMPHLEN);
963 		return;
964 	}
965 
966 	if (frpr_pullup(fin, ICMPERR_ICMPHLEN) == -1)
967 		return;
968 
969 	fr_checkv4sum(fin);
970 
971 	/*
972 	 * This is a right place to set icmp pointer, since the memory
973 	 * referenced by fin_dp could get reallocated. The code down below can
974 	 * rely on fact icmp variable always points to ICMP header.
975 	 */
976 	icmp = fin->fin_dp;
977 	fin->fin_data[0] = *(u_short *)icmp;
978 	fin->fin_data[1] = icmp->icmp_id;
979 
980 	switch (icmp->icmp_type)
981 	{
982 	case ICMP_ECHOREPLY :
983 	case ICMP_ECHO :
984 	/* Router discovery messaes - RFC 1256 */
985 	case ICMP_ROUTERADVERT :
986 	case ICMP_ROUTERSOLICIT :
987 		minicmpsz = ICMP_MINLEN;
988 		break;
989 	/*
990 	 * type(1) + code(1) + cksum(2) + id(2) seq(2) +
991 	 * 3 * timestamp(3 * 4)
992 	 */
993 	case ICMP_TSTAMP :
994 	case ICMP_TSTAMPREPLY :
995 		minicmpsz = 20;
996 		break;
997 	/*
998 	 * type(1) + code(1) + cksum(2) + id(2) seq(2) +
999 	 * mask(4)
1000 	 */
1001 	case ICMP_MASKREQ :
1002 	case ICMP_MASKREPLY :
1003 			minicmpsz = 12;
1004 			break;
1005 	/*
1006 	 * type(1) + code(1) + cksum(2) + id(2) seq(2) + ip(20+)
1007 	 */
1008 	case ICMP_UNREACH :
1009 		if (icmp->icmp_code == ICMP_UNREACH_NEEDFRAG) {
1010 			if (icmp->icmp_nextmtu < ifs->ifs_fr_icmpminfragmtu)
1011 				fin->fin_flx |= FI_BAD;
1012 		}
1013 		/* FALLTHRU */
1014 	case ICMP_SOURCEQUENCH :
1015 	case ICMP_REDIRECT :
1016 	case ICMP_TIMXCEED :
1017 	case ICMP_PARAMPROB :
1018 		fin->fin_flx |= FI_ICMPERR;
1019 		if (fr_coalesce(fin) != 1)
1020 			return;
1021 		/*
1022 		 * ICMP error packets should not be generated for IP
1023 		 * packets that are a fragment that isn't the first
1024 		 * fragment.
1025 		 */
1026 		oip = (ip_t *)((char *)fin->fin_dp + ICMPERR_ICMPHLEN);
1027 		if ((ntohs(oip->ip_off) & IP_OFFMASK) != 0)
1028 			fin->fin_flx |= FI_BAD;
1029 		break;
1030 	default :
1031 		break;
1032 	}
1033 
1034 	frpr_short(fin, minicmpsz);
1035 }
1036 
1037 
1038 /* ------------------------------------------------------------------------ */
1039 /* Function:    frpr_tcpcommon                                              */
1040 /* Returns:     void                                                        */
1041 /* Parameters:  fin(I) - pointer to packet information                      */
1042 /*                                                                          */
1043 /* TCP header sanity checking.  Look for bad combinations of TCP flags,     */
1044 /* and make some checks with how they interact with other fields.           */
1045 /* If compiled with IPFILTER_CKSUM, check to see if the TCP checksum is     */
1046 /* valid and mark the packet as bad if not.                                 */
1047 /* ------------------------------------------------------------------------ */
1048 static INLINE void frpr_tcpcommon(fin)
1049 fr_info_t *fin;
1050 {
1051 	int flags, tlen;
1052 	tcphdr_t *tcp;
1053 
1054 	fin->fin_flx |= FI_TCPUDP;
1055 	if (fin->fin_off != 0)
1056 		return;
1057 
1058 	if (frpr_pullup(fin, sizeof(*tcp)) == -1)
1059 		return;
1060 	tcp = fin->fin_dp;
1061 
1062 	if (fin->fin_dlen > 3) {
1063 		fin->fin_sport = ntohs(tcp->th_sport);
1064 		fin->fin_dport = ntohs(tcp->th_dport);
1065 	}
1066 
1067 	if ((fin->fin_flx & FI_SHORT) != 0)
1068 		return;
1069 
1070 	/*
1071 	 * Use of the TCP data offset *must* result in a value that is at
1072 	 * least the same size as the TCP header.
1073 	 */
1074 	tlen = TCP_OFF(tcp) << 2;
1075 	if (tlen < sizeof(tcphdr_t)) {
1076 		fin->fin_flx |= FI_BAD;
1077 		return;
1078 	}
1079 
1080 	flags = tcp->th_flags;
1081 	fin->fin_tcpf = tcp->th_flags;
1082 
1083 	/*
1084 	 * If the urgent flag is set, then the urgent pointer must
1085 	 * also be set and vice versa.  Good TCP packets do not have
1086 	 * just one of these set.
1087 	 */
1088 	if ((flags & TH_URG) != 0 && (tcp->th_urp == 0)) {
1089 		fin->fin_flx |= FI_BAD;
1090 	} else if ((flags & TH_URG) == 0 && (tcp->th_urp != 0)) {
1091 		/* Ignore this case, it shows up in "real" traffic with */
1092 		/* bogus values in the urgent pointer field. */
1093 		flags = flags; /* LINT */
1094 	} else if (((flags & (TH_SYN|TH_FIN)) != 0) &&
1095 		   ((flags & (TH_RST|TH_ACK)) == TH_RST)) {
1096 		/* TH_FIN|TH_RST|TH_ACK seems to appear "naturally" */
1097 		fin->fin_flx |= FI_BAD;
1098 	} else if (!(flags & TH_ACK)) {
1099 		/*
1100 		 * If the ack bit isn't set, then either the SYN or
1101 		 * RST bit must be set.  If the SYN bit is set, then
1102 		 * we expect the ACK field to be 0.  If the ACK is
1103 		 * not set and if URG, PSH or FIN are set, consdier
1104 		 * that to indicate a bad TCP packet.
1105 		 */
1106 		if ((flags == TH_SYN) && (tcp->th_ack != 0)) {
1107 			/*
1108 			 * Cisco PIX sets the ACK field to a random value.
1109 			 * In light of this, do not set FI_BAD until a patch
1110 			 * is available from Cisco to ensure that
1111 			 * interoperability between existing systems is
1112 			 * achieved.
1113 			 */
1114 			/*fin->fin_flx |= FI_BAD*/;
1115 			flags = flags; /* LINT */
1116 		} else if (!(flags & (TH_RST|TH_SYN))) {
1117 			fin->fin_flx |= FI_BAD;
1118 		} else if ((flags & (TH_URG|TH_PUSH|TH_FIN)) != 0) {
1119 			fin->fin_flx |= FI_BAD;
1120 		}
1121 	}
1122 
1123 	/*
1124 	 * At this point, it's not exactly clear what is to be gained by
1125 	 * marking up which TCP options are and are not present.  The one we
1126 	 * are most interested in is the TCP window scale.  This is only in
1127 	 * a SYN packet [RFC1323] so we don't need this here...?
1128 	 * Now if we were to analyse the header for passive fingerprinting,
1129 	 * then that might add some weight to adding this...
1130 	 */
1131 	if (tlen == sizeof(tcphdr_t))
1132 		return;
1133 
1134 	if (frpr_pullup(fin, tlen) == -1)
1135 		return;
1136 
1137 #if 0
1138 	ip = fin->fin_ip;
1139 	s = (u_char *)(tcp + 1);
1140 	off = IP_HL(ip) << 2;
1141 # ifdef _KERNEL
1142 	if (fin->fin_mp != NULL) {
1143 		mb_t *m = *fin->fin_mp;
1144 
1145 		if (off + tlen > M_LEN(m))
1146 			return;
1147 	}
1148 # endif
1149 	for (tlen -= (int)sizeof(*tcp); tlen > 0; ) {
1150 		opt = *s;
1151 		if (opt == '\0')
1152 			break;
1153 		else if (opt == TCPOPT_NOP)
1154 			ol = 1;
1155 		else {
1156 			if (tlen < 2)
1157 				break;
1158 			ol = (int)*(s + 1);
1159 			if (ol < 2 || ol > tlen)
1160 				break;
1161 		}
1162 
1163 		for (i = 9, mv = 4; mv >= 0; ) {
1164 			op = ipopts + i;
1165 			if (opt == (u_char)op->ol_val) {
1166 				optmsk |= op->ol_bit;
1167 				break;
1168 			}
1169 		}
1170 		tlen -= ol;
1171 		s += ol;
1172 	}
1173 #endif /* 0 */
1174 }
1175 
1176 
1177 
1178 /* ------------------------------------------------------------------------ */
1179 /* Function:    frpr_udpcommon                                              */
1180 /* Returns:     void                                                        */
1181 /* Parameters:  fin(I) - pointer to packet information                      */
1182 /*                                                                          */
1183 /* Extract the UDP source and destination ports, if present.  If compiled   */
1184 /* with IPFILTER_CKSUM, check to see if the UDP checksum is valid.          */
1185 /* ------------------------------------------------------------------------ */
1186 static INLINE void frpr_udpcommon(fin)
1187 fr_info_t *fin;
1188 {
1189 	udphdr_t *udp;
1190 
1191 	fin->fin_flx |= FI_TCPUDP;
1192 
1193 	if (!fin->fin_off && (fin->fin_dlen > 3)) {
1194 		if (frpr_pullup(fin, sizeof(*udp)) == -1) {
1195 			fin->fin_flx |= FI_SHORT;
1196 			return;
1197 		}
1198 
1199 		udp = fin->fin_dp;
1200 
1201 		fin->fin_sport = ntohs(udp->uh_sport);
1202 		fin->fin_dport = ntohs(udp->uh_dport);
1203 	}
1204 }
1205 
1206 
1207 /* ------------------------------------------------------------------------ */
1208 /* Function:    frpr_tcp                                                    */
1209 /* Returns:     void                                                        */
1210 /* Parameters:  fin(I) - pointer to packet information                      */
1211 /*                                                                          */
1212 /* IPv4 Only                                                                */
1213 /* Analyse the packet for IPv4/TCP properties.                              */
1214 /* ------------------------------------------------------------------------ */
1215 static INLINE void frpr_tcp(fin)
1216 fr_info_t *fin;
1217 {
1218 
1219 	fr_checkv4sum(fin);
1220 
1221 	frpr_short(fin, sizeof(tcphdr_t));
1222 
1223 	frpr_tcpcommon(fin);
1224 }
1225 
1226 
1227 /* ------------------------------------------------------------------------ */
1228 /* Function:    frpr_udp                                                    */
1229 /* Returns:     void                                                        */
1230 /* Parameters:  fin(I) - pointer to packet information                      */
1231 /*                                                                          */
1232 /* IPv4 Only                                                                */
1233 /* Analyse the packet for IPv4/UDP properties.                              */
1234 /* ------------------------------------------------------------------------ */
1235 static INLINE void frpr_udp(fin)
1236 fr_info_t *fin;
1237 {
1238 
1239 	fr_checkv4sum(fin);
1240 
1241 	frpr_short(fin, sizeof(udphdr_t));
1242 
1243 	frpr_udpcommon(fin);
1244 }
1245 
1246 
1247 /* ------------------------------------------------------------------------ */
1248 /* Function:    frpr_esp                                                    */
1249 /* Returns:     void                                                        */
1250 /* Parameters:  fin(I) - pointer to packet information                      */
1251 /*                                                                          */
1252 /* Analyse the packet for ESP properties.                                   */
1253 /* The minimum length is taken to be the SPI (32bits) plus a tail (32bits)  */
1254 /* even though the newer ESP packets must also have a sequence number that  */
1255 /* is 32bits as well, it is not possible(?) to determine the version from a */
1256 /* simple packet header.                                                    */
1257 /* ------------------------------------------------------------------------ */
1258 static INLINE void frpr_esp(fin)
1259 fr_info_t *fin;
1260 {
1261 	if ((fin->fin_off == 0) && (frpr_pullup(fin, 8) == -1))
1262 		return;
1263 
1264 	frpr_short(fin, 8);
1265 }
1266 
1267 
1268 /* ------------------------------------------------------------------------ */
1269 /* Function:    frpr_ah                                                     */
1270 /* Returns:     void                                                        */
1271 /* Parameters:  fin(I) - pointer to packet information                      */
1272 /*                                                                          */
1273 /* Analyse the packet for AH properties.                                    */
1274 /* The minimum length is taken to be the combination of all fields in the   */
1275 /* header being present and no authentication data (null algorithm used.)   */
1276 /* ------------------------------------------------------------------------ */
1277 static INLINE void frpr_ah(fin)
1278 fr_info_t *fin;
1279 {
1280 	authhdr_t *ah;
1281 	int len;
1282 
1283 	if ((fin->fin_off == 0) && (frpr_pullup(fin, sizeof(*ah)) == -1))
1284 		return;
1285 
1286 	ah = (authhdr_t *)fin->fin_dp;
1287 
1288 	len = (ah->ah_plen + 2) << 2;
1289 	frpr_short(fin, len);
1290 }
1291 
1292 
1293 /* ------------------------------------------------------------------------ */
1294 /* Function:    frpr_gre                                                    */
1295 /* Returns:     void                                                        */
1296 /* Parameters:  fin(I) - pointer to packet information                      */
1297 /*                                                                          */
1298 /* Analyse the packet for GRE properties.                                   */
1299 /* ------------------------------------------------------------------------ */
1300 static INLINE void frpr_gre(fin)
1301 fr_info_t *fin;
1302 {
1303 	grehdr_t *gre;
1304 
1305 	if ((fin->fin_off == 0) && (frpr_pullup(fin, sizeof(grehdr_t)) == -1))
1306 		return;
1307 
1308 	frpr_short(fin, sizeof(grehdr_t));
1309 
1310 	if (fin->fin_off == 0) {
1311 		gre = fin->fin_dp;
1312 		if (GRE_REV(gre->gr_flags) == 1)
1313 			fin->fin_data[0] = gre->gr_call;
1314 	}
1315 }
1316 
1317 
1318 /* ------------------------------------------------------------------------ */
1319 /* Function:    frpr_ipv4hdr                                                */
1320 /* Returns:     void                                                        */
1321 /* Parameters:  fin(I) - pointer to packet information                      */
1322 /*                                                                          */
1323 /* IPv4 Only                                                                */
1324 /* Analyze the IPv4 header and set fields in the fr_info_t structure.       */
1325 /* Check all options present and flag their presence if any exist.          */
1326 /* ------------------------------------------------------------------------ */
1327 static INLINE void frpr_ipv4hdr(fin)
1328 fr_info_t *fin;
1329 {
1330 	u_short optmsk = 0, secmsk = 0, auth = 0;
1331 	int hlen, ol, mv, p, i;
1332 	const struct optlist *op;
1333 	u_char *s, opt;
1334 	u_short off;
1335 	fr_ip_t *fi;
1336 	ip_t *ip;
1337 
1338 	fi = &fin->fin_fi;
1339 	hlen = fin->fin_hlen;
1340 
1341 	ip = fin->fin_ip;
1342 	p = ip->ip_p;
1343 	fi->fi_p = p;
1344 	fi->fi_tos = ip->ip_tos;
1345 	fin->fin_id = ip->ip_id;
1346 	off = ip->ip_off;
1347 
1348 	/* Get both TTL and protocol */
1349 	fi->fi_p = ip->ip_p;
1350 	fi->fi_ttl = ip->ip_ttl;
1351 #if 0
1352 	(*(((u_short *)fi) + 1)) = (*(((u_short *)ip) + 4));
1353 #endif
1354 
1355 	/* Zero out bits not used in IPv6 address */
1356 	fi->fi_src.i6[1] = 0;
1357 	fi->fi_src.i6[2] = 0;
1358 	fi->fi_src.i6[3] = 0;
1359 	fi->fi_dst.i6[1] = 0;
1360 	fi->fi_dst.i6[2] = 0;
1361 	fi->fi_dst.i6[3] = 0;
1362 
1363 	fi->fi_saddr = ip->ip_src.s_addr;
1364 	fi->fi_daddr = ip->ip_dst.s_addr;
1365 
1366 	/*
1367 	 * set packet attribute flags based on the offset and
1368 	 * calculate the byte offset that it represents.
1369 	 */
1370 	off &= IP_MF|IP_OFFMASK;
1371 	if (off != 0) {
1372 		fi->fi_flx |= FI_FRAG;
1373 		off &= IP_OFFMASK;
1374 		if (off != 0) {
1375 			fin->fin_flx |= FI_FRAGBODY;
1376 			off <<= 3;
1377 			if ((off + fin->fin_dlen > 65535) ||
1378 			    (fin->fin_dlen == 0) ||
1379 			    ((ip->ip_off & IP_MF) && (fin->fin_dlen & 7))) {
1380 				/*
1381 				 * The length of the packet, starting at its
1382 				 * offset cannot exceed 65535 (0xffff) as the
1383 				 * length of an IP packet is only 16 bits.
1384 				 *
1385 				 * Any fragment that isn't the last fragment
1386 				 * must have a length greater than 0 and it
1387 				 * must be an even multiple of 8.
1388 				 */
1389 				fi->fi_flx |= FI_BAD;
1390 			}
1391 		}
1392 	}
1393 	fin->fin_off = off;
1394 
1395 	/*
1396 	 * Call per-protocol setup and checking
1397 	 */
1398 	switch (p)
1399 	{
1400 	case IPPROTO_UDP :
1401 		frpr_udp(fin);
1402 		break;
1403 	case IPPROTO_TCP :
1404 		frpr_tcp(fin);
1405 		break;
1406 	case IPPROTO_ICMP :
1407 		frpr_icmp(fin);
1408 		break;
1409 	case IPPROTO_AH :
1410 		frpr_ah(fin);
1411 		break;
1412 	case IPPROTO_ESP :
1413 		frpr_esp(fin);
1414 		break;
1415 	case IPPROTO_GRE :
1416 		frpr_gre(fin);
1417 		break;
1418 	}
1419 
1420 	ip = fin->fin_ip;
1421 	if (ip == NULL)
1422 		return;
1423 
1424 	/*
1425 	 * If it is a standard IP header (no options), set the flag fields
1426 	 * which relate to options to 0.
1427 	 */
1428 	if (hlen == sizeof(*ip)) {
1429 		fi->fi_optmsk = 0;
1430 		fi->fi_secmsk = 0;
1431 		fi->fi_auth = 0;
1432 		return;
1433 	}
1434 
1435 	/*
1436 	 * So the IP header has some IP options attached.  Walk the entire
1437 	 * list of options present with this packet and set flags to indicate
1438 	 * which ones are here and which ones are not.  For the somewhat out
1439 	 * of date and obscure security classification options, set a flag to
1440 	 * represent which classification is present.
1441 	 */
1442 	fi->fi_flx |= FI_OPTIONS;
1443 
1444 	for (s = (u_char *)(ip + 1), hlen -= (int)sizeof(*ip); hlen > 0; ) {
1445 		opt = *s;
1446 		if (opt == '\0')
1447 			break;
1448 		else if (opt == IPOPT_NOP)
1449 			ol = 1;
1450 		else {
1451 			if (hlen < 2)
1452 				break;
1453 			ol = (int)*(s + 1);
1454 			if (ol < 2 || ol > hlen)
1455 				break;
1456 		}
1457 		for (i = 9, mv = 4; mv >= 0; ) {
1458 			op = ipopts + i;
1459 			if ((opt == (u_char)op->ol_val) && (ol > 4)) {
1460 				optmsk |= op->ol_bit;
1461 				if (opt == IPOPT_SECURITY) {
1462 					const struct optlist *sp;
1463 					u_char	sec;
1464 					int j, m;
1465 
1466 					sec = *(s + 2);	/* classification */
1467 					for (j = 3, m = 2; m >= 0; ) {
1468 						sp = secopt + j;
1469 						if (sec == sp->ol_val) {
1470 							secmsk |= sp->ol_bit;
1471 							auth = *(s + 3);
1472 							auth *= 256;
1473 							auth += *(s + 4);
1474 							break;
1475 						}
1476 						if (sec < sp->ol_val)
1477 							j -= m;
1478 						else
1479 							j += m;
1480 						m--;
1481 					}
1482 				}
1483 				break;
1484 			}
1485 			if (opt < op->ol_val)
1486 				i -= mv;
1487 			else
1488 				i += mv;
1489 			mv--;
1490 		}
1491 		hlen -= ol;
1492 		s += ol;
1493 	}
1494 
1495 	/*
1496 	 *
1497 	 */
1498 	if (auth && !(auth & 0x0100))
1499 		auth &= 0xff00;
1500 	fi->fi_optmsk = optmsk;
1501 	fi->fi_secmsk = secmsk;
1502 	fi->fi_auth = auth;
1503 }
1504 
1505 
1506 /* ------------------------------------------------------------------------ */
1507 /* Function:    fr_makefrip                                                 */
1508 /* Returns:     int - 1 == hdr checking error, 0 == OK                      */
1509 /* Parameters:  hlen(I) - length of IP packet header                        */
1510 /*              ip(I)   - pointer to the IP header                          */
1511 /*              fin(IO) - pointer to packet information                     */
1512 /*                                                                          */
1513 /* Compact the IP header into a structure which contains just the info.     */
1514 /* which is useful for comparing IP headers with and store this information */
1515 /* in the fr_info_t structure pointer to by fin.  At present, it is assumed */
1516 /* this function will be called with either an IPv4 or IPv6 packet.         */
1517 /* ------------------------------------------------------------------------ */
1518 int	fr_makefrip(hlen, ip, fin)
1519 int hlen;
1520 ip_t *ip;
1521 fr_info_t *fin;
1522 {
1523 	int v;
1524 
1525 	fin->fin_depth = 0;
1526 	fin->fin_hlen = (u_short)hlen;
1527 	fin->fin_ip = ip;
1528 	fin->fin_rule = 0xffffffff;
1529 	fin->fin_group[0] = -1;
1530 	fin->fin_group[1] = '\0';
1531 	fin->fin_dlen = fin->fin_plen - hlen;
1532 	fin->fin_dp = (char *)ip + hlen;
1533 
1534 	v = fin->fin_v;
1535 	if (v == 4)
1536 		frpr_ipv4hdr(fin);
1537 #ifdef	USE_INET6
1538 	else if (v == 6)
1539 		frpr_ipv6hdr(fin);
1540 #endif
1541 	if (fin->fin_ip == NULL)
1542 		return -1;
1543 	return 0;
1544 }
1545 
1546 
1547 /* ------------------------------------------------------------------------ */
1548 /* Function:    fr_portcheck                                                */
1549 /* Returns:     int - 1 == port matched, 0 == port match failed             */
1550 /* Parameters:  frp(I) - pointer to port check `expression'                 */
1551 /*              pop(I) - pointer to port number to evaluate                 */
1552 /*                                                                          */
1553 /* Perform a comparison of a port number against some other(s), using a     */
1554 /* structure with compare information stored in it.                         */
1555 /* ------------------------------------------------------------------------ */
1556 static INLINE int fr_portcheck(frp, pop)
1557 frpcmp_t *frp;
1558 u_short *pop;
1559 {
1560 	u_short tup, po;
1561 	int err = 1;
1562 
1563 	tup = *pop;
1564 	po = frp->frp_port;
1565 
1566 	/*
1567 	 * Do opposite test to that required and continue if that succeeds.
1568 	 */
1569 	switch (frp->frp_cmp)
1570 	{
1571 	case FR_EQUAL :
1572 		if (tup != po) /* EQUAL */
1573 			err = 0;
1574 		break;
1575 	case FR_NEQUAL :
1576 		if (tup == po) /* NOTEQUAL */
1577 			err = 0;
1578 		break;
1579 	case FR_LESST :
1580 		if (tup >= po) /* LESSTHAN */
1581 			err = 0;
1582 		break;
1583 	case FR_GREATERT :
1584 		if (tup <= po) /* GREATERTHAN */
1585 			err = 0;
1586 		break;
1587 	case FR_LESSTE :
1588 		if (tup > po) /* LT or EQ */
1589 			err = 0;
1590 		break;
1591 	case FR_GREATERTE :
1592 		if (tup < po) /* GT or EQ */
1593 			err = 0;
1594 		break;
1595 	case FR_OUTRANGE :
1596 		if (tup >= po && tup <= frp->frp_top) /* Out of range */
1597 			err = 0;
1598 		break;
1599 	case FR_INRANGE :
1600 		if (tup <= po || tup >= frp->frp_top) /* In range */
1601 			err = 0;
1602 		break;
1603 	case FR_INCRANGE :
1604 		if (tup < po || tup > frp->frp_top) /* Inclusive range */
1605 			err = 0;
1606 		break;
1607 	default :
1608 		break;
1609 	}
1610 	return err;
1611 }
1612 
1613 
1614 /* ------------------------------------------------------------------------ */
1615 /* Function:    fr_tcpudpchk                                                */
1616 /* Returns:     int - 1 == protocol matched, 0 == check failed              */
1617 /* Parameters:  fin(I) - pointer to packet information                      */
1618 /*              ft(I)  - pointer to structure with comparison data          */
1619 /*                                                                          */
1620 /* Compares the current pcket (assuming it is TCP/UDP) information with a   */
1621 /* structure containing information that we want to match against.          */
1622 /* ------------------------------------------------------------------------ */
1623 int fr_tcpudpchk(fin, ft)
1624 fr_info_t *fin;
1625 frtuc_t *ft;
1626 {
1627 	int err = 1;
1628 
1629 	/*
1630 	 * Both ports should *always* be in the first fragment.
1631 	 * So far, I cannot find any cases where they can not be.
1632 	 *
1633 	 * compare destination ports
1634 	 */
1635 	if (ft->ftu_dcmp)
1636 		err = fr_portcheck(&ft->ftu_dst, &fin->fin_dport);
1637 
1638 	/*
1639 	 * compare source ports
1640 	 */
1641 	if (err && ft->ftu_scmp)
1642 		err = fr_portcheck(&ft->ftu_src, &fin->fin_sport);
1643 
1644 	/*
1645 	 * If we don't have all the TCP/UDP header, then how can we
1646 	 * expect to do any sort of match on it ?  If we were looking for
1647 	 * TCP flags, then NO match.  If not, then match (which should
1648 	 * satisfy the "short" class too).
1649 	 */
1650 	if (err && (fin->fin_p == IPPROTO_TCP)) {
1651 		if (fin->fin_flx & FI_SHORT)
1652 			return !(ft->ftu_tcpf | ft->ftu_tcpfm);
1653 		/*
1654 		 * Match the flags ?  If not, abort this match.
1655 		 */
1656 		if (ft->ftu_tcpfm &&
1657 		    ft->ftu_tcpf != (fin->fin_tcpf & ft->ftu_tcpfm)) {
1658 			FR_DEBUG(("f. %#x & %#x != %#x\n", fin->fin_tcpf,
1659 				 ft->ftu_tcpfm, ft->ftu_tcpf));
1660 			err = 0;
1661 		}
1662 	}
1663 	return err;
1664 }
1665 
1666 
1667 /* ------------------------------------------------------------------------ */
1668 /* Function:    fr_ipfcheck                                                 */
1669 /* Returns:     int - 0 == match, 1 == no match                             */
1670 /* Parameters:  fin(I)     - pointer to packet information                  */
1671 /*              fr(I)      - pointer to filter rule                         */
1672 /*              portcmp(I) - flag indicating whether to attempt matching on */
1673 /*                           TCP/UDP port data.                             */
1674 /*                                                                          */
1675 /* Check to see if a packet matches an IPFilter rule.  Checks of addresses, */
1676 /* port numbers, etc, for "standard" IPFilter rules are all orchestrated in */
1677 /* this function.                                                           */
1678 /* ------------------------------------------------------------------------ */
1679 static INLINE int fr_ipfcheck(fin, fr, portcmp)
1680 fr_info_t *fin;
1681 frentry_t *fr;
1682 int portcmp;
1683 {
1684 	u_32_t	*ld, *lm, *lip;
1685 	fripf_t *fri;
1686 	fr_ip_t *fi;
1687 	int i;
1688 	ipf_stack_t *ifs = fin->fin_ifs;
1689 
1690 	fi = &fin->fin_fi;
1691 	fri = fr->fr_ipf;
1692 	lip = (u_32_t *)fi;
1693 	lm = (u_32_t *)&fri->fri_mip;
1694 	ld = (u_32_t *)&fri->fri_ip;
1695 
1696 	/*
1697 	 * first 32 bits to check coversion:
1698 	 * IP version, TOS, TTL, protocol
1699 	 */
1700 	i = ((*lip & *lm) != *ld);
1701 	FR_DEBUG(("0. %#08x & %#08x != %#08x\n",
1702 		   *lip, *lm, *ld));
1703 	if (i)
1704 		return 1;
1705 
1706 	/*
1707 	 * Next 32 bits is a constructed bitmask indicating which IP options
1708 	 * are present (if any) in this packet.
1709 	 */
1710 	lip++, lm++, ld++;
1711 	i |= ((*lip & *lm) != *ld);
1712 	FR_DEBUG(("1. %#08x & %#08x != %#08x\n",
1713 		   *lip, *lm, *ld));
1714 	if (i)
1715 		return 1;
1716 
1717 	lip++, lm++, ld++;
1718 	/*
1719 	 * Unrolled loops (4 each, for 32 bits) for address checks.
1720 	 */
1721 	/*
1722 	 * Check the source address.
1723 	 */
1724 #ifdef	IPFILTER_LOOKUP
1725 	if (fr->fr_satype == FRI_LOOKUP) {
1726 		i = (*fr->fr_srcfunc)(fr->fr_srcptr, fi->fi_v, lip, ifs);
1727 		if (i == -1)
1728 			return 1;
1729 		lip += 3;
1730 		lm += 3;
1731 		ld += 3;
1732 	} else {
1733 #endif
1734 		i = ((*lip & *lm) != *ld);
1735 		FR_DEBUG(("2a. %#08x & %#08x != %#08x\n",
1736 			   *lip, *lm, *ld));
1737 		if (fi->fi_v == 6) {
1738 			lip++, lm++, ld++;
1739 			i |= ((*lip & *lm) != *ld);
1740 			FR_DEBUG(("2b. %#08x & %#08x != %#08x\n",
1741 				   *lip, *lm, *ld));
1742 			lip++, lm++, ld++;
1743 			i |= ((*lip & *lm) != *ld);
1744 			FR_DEBUG(("2c. %#08x & %#08x != %#08x\n",
1745 				   *lip, *lm, *ld));
1746 			lip++, lm++, ld++;
1747 			i |= ((*lip & *lm) != *ld);
1748 			FR_DEBUG(("2d. %#08x & %#08x != %#08x\n",
1749 				   *lip, *lm, *ld));
1750 		} else {
1751 			lip += 3;
1752 			lm += 3;
1753 			ld += 3;
1754 		}
1755 #ifdef	IPFILTER_LOOKUP
1756 	}
1757 #endif
1758 	i ^= (fr->fr_flags & FR_NOTSRCIP) >> 6;
1759 	if (i)
1760 		return 1;
1761 
1762 	/*
1763 	 * Check the destination address.
1764 	 */
1765 	lip++, lm++, ld++;
1766 #ifdef	IPFILTER_LOOKUP
1767 	if (fr->fr_datype == FRI_LOOKUP) {
1768 		i = (*fr->fr_dstfunc)(fr->fr_dstptr, fi->fi_v, lip, ifs);
1769 		if (i == -1)
1770 			return 1;
1771 		lip += 3;
1772 		lm += 3;
1773 		ld += 3;
1774 	} else {
1775 #endif
1776 		i = ((*lip & *lm) != *ld);
1777 		FR_DEBUG(("3a. %#08x & %#08x != %#08x\n",
1778 			   *lip, *lm, *ld));
1779 		if (fi->fi_v == 6) {
1780 			lip++, lm++, ld++;
1781 			i |= ((*lip & *lm) != *ld);
1782 			FR_DEBUG(("3b. %#08x & %#08x != %#08x\n",
1783 				   *lip, *lm, *ld));
1784 			lip++, lm++, ld++;
1785 			i |= ((*lip & *lm) != *ld);
1786 			FR_DEBUG(("3c. %#08x & %#08x != %#08x\n",
1787 				   *lip, *lm, *ld));
1788 			lip++, lm++, ld++;
1789 			i |= ((*lip & *lm) != *ld);
1790 			FR_DEBUG(("3d. %#08x & %#08x != %#08x\n",
1791 				   *lip, *lm, *ld));
1792 		} else {
1793 			lip += 3;
1794 			lm += 3;
1795 			ld += 3;
1796 		}
1797 #ifdef	IPFILTER_LOOKUP
1798 	}
1799 #endif
1800 	i ^= (fr->fr_flags & FR_NOTDSTIP) >> 7;
1801 	if (i)
1802 		return 1;
1803 	/*
1804 	 * IP addresses matched.  The next 32bits contains:
1805 	 * mast of old IP header security & authentication bits.
1806 	 */
1807 	lip++, lm++, ld++;
1808 	i |= ((*lip & *lm) != *ld);
1809 	FR_DEBUG(("4. %#08x & %#08x != %#08x\n",
1810 		   *lip, *lm, *ld));
1811 
1812 	/*
1813 	 * Next we have 32 bits of packet flags.
1814 	 */
1815 	lip++, lm++, ld++;
1816 	i |= ((*lip & *lm) != *ld);
1817 	FR_DEBUG(("5. %#08x & %#08x != %#08x\n",
1818 		   *lip, *lm, *ld));
1819 
1820 	if (i == 0) {
1821 		/*
1822 		 * If a fragment, then only the first has what we're
1823 		 * looking for here...
1824 		 */
1825 		if (portcmp) {
1826 			if (!fr_tcpudpchk(fin, &fr->fr_tuc))
1827 				i = 1;
1828 		} else {
1829 			if (fr->fr_dcmp || fr->fr_scmp ||
1830 			    fr->fr_tcpf || fr->fr_tcpfm)
1831 				i = 1;
1832 			if (fr->fr_icmpm || fr->fr_icmp) {
1833 				if (((fi->fi_p != IPPROTO_ICMP) &&
1834 				     (fi->fi_p != IPPROTO_ICMPV6)) ||
1835 				    fin->fin_off || (fin->fin_dlen < 2))
1836 					i = 1;
1837 				else if ((fin->fin_data[0] & fr->fr_icmpm) !=
1838 					 fr->fr_icmp) {
1839 					FR_DEBUG(("i. %#x & %#x != %#x\n",
1840 						 fin->fin_data[0],
1841 						 fr->fr_icmpm, fr->fr_icmp));
1842 					i = 1;
1843 				}
1844 			}
1845 		}
1846 	}
1847 	return i;
1848 }
1849 
1850 
1851 /* ------------------------------------------------------------------------ */
1852 /* Function:    fr_scanlist                                                 */
1853 /* Returns:     int - result flags of scanning filter list                  */
1854 /* Parameters:  fin(I) - pointer to packet information                      */
1855 /*              pass(I) - default result to return for filtering            */
1856 /*                                                                          */
1857 /* Check the input/output list of rules for a match to the current packet.  */
1858 /* If a match is found, the value of fr_flags from the rule becomes the     */
1859 /* return value and fin->fin_fr points to the matched rule.                 */
1860 /*                                                                          */
1861 /* This function may be called recusively upto 16 times (limit inbuilt.)    */
1862 /* When unwinding, it should finish up with fin_depth as 0.                 */
1863 /*                                                                          */
1864 /* Could be per interface, but this gets real nasty when you don't have,    */
1865 /* or can't easily change, the kernel source code to .                      */
1866 /* ------------------------------------------------------------------------ */
1867 int fr_scanlist(fin, pass)
1868 fr_info_t *fin;
1869 u_32_t pass;
1870 {
1871 	int rulen, portcmp, off, logged, skip;
1872 	struct frentry *fr, *fnext;
1873 	u_32_t passt, passo;
1874 	ipf_stack_t *ifs = fin->fin_ifs;
1875 
1876 	/*
1877 	 * Do not allow nesting deeper than 16 levels.
1878 	 */
1879 	if (fin->fin_depth >= 16)
1880 		return pass;
1881 
1882 	fr = fin->fin_fr;
1883 
1884 	/*
1885 	 * If there are no rules in this list, return now.
1886 	 */
1887 	if (fr == NULL)
1888 		return pass;
1889 
1890 	skip = 0;
1891 	logged = 0;
1892 	portcmp = 0;
1893 	fin->fin_depth++;
1894 	fin->fin_fr = NULL;
1895 	off = fin->fin_off;
1896 
1897 	if ((fin->fin_flx & FI_TCPUDP) && (fin->fin_dlen > 3) && !off)
1898 		portcmp = 1;
1899 
1900 	for (rulen = 0; fr; fr = fnext, rulen++) {
1901 		fnext = fr->fr_next;
1902 		if (skip != 0) {
1903 			FR_VERBOSE(("%d (%#x)\n", skip, fr->fr_flags));
1904 			skip--;
1905 			continue;
1906 		}
1907 
1908 		/*
1909 		 * In all checks below, a null (zero) value in the
1910 		 * filter struture is taken to mean a wildcard.
1911 		 *
1912 		 * check that we are working for the right interface
1913 		 */
1914 #ifdef	_KERNEL
1915 		if (fr->fr_ifa && fr->fr_ifa != fin->fin_ifp)
1916 			continue;
1917 #else
1918 		if (opts & (OPT_VERBOSE|OPT_DEBUG))
1919 			printf("\n");
1920 		FR_VERBOSE(("%c", FR_ISSKIP(pass) ? 's' :
1921 				  FR_ISPASS(pass) ? 'p' :
1922 				  FR_ISACCOUNT(pass) ? 'A' :
1923 				  FR_ISAUTH(pass) ? 'a' :
1924 				  (pass & FR_NOMATCH) ? 'n' :'b'));
1925 		if (fr->fr_ifa && fr->fr_ifa != fin->fin_ifp)
1926 			continue;
1927 		FR_VERBOSE((":i"));
1928 #endif
1929 
1930 		switch (fr->fr_type)
1931 		{
1932 		case FR_T_IPF :
1933 		case FR_T_IPF|FR_T_BUILTIN :
1934 			if (fr_ipfcheck(fin, fr, portcmp))
1935 				continue;
1936 			break;
1937 #if defined(IPFILTER_BPF)
1938 		case FR_T_BPFOPC :
1939 		case FR_T_BPFOPC|FR_T_BUILTIN :
1940 		    {
1941 			u_char *mc;
1942 
1943 			if (*fin->fin_mp == NULL)
1944 				continue;
1945 			if (fin->fin_v != fr->fr_v)
1946 				continue;
1947 			mc = (u_char *)fin->fin_m;
1948 			if (!bpf_filter(fr->fr_data, mc, fin->fin_plen, 0))
1949 				continue;
1950 			break;
1951 		    }
1952 #endif
1953 		case FR_T_CALLFUNC|FR_T_BUILTIN :
1954 		    {
1955 			frentry_t *f;
1956 
1957 			f = (*fr->fr_func)(fin, &pass);
1958 			if (f != NULL)
1959 				fr = f;
1960 			else
1961 				continue;
1962 			break;
1963 		    }
1964 		default :
1965 			break;
1966 		}
1967 
1968 		if ((fin->fin_out == 0) && (fr->fr_nattag.ipt_num[0] != 0)) {
1969 			if (fin->fin_nattag == NULL)
1970 				continue;
1971 			if (fr_matchtag(&fr->fr_nattag, fin->fin_nattag) == 0)
1972 				continue;
1973 		}
1974 		FR_VERBOSE(("=%s.%d *", fr->fr_group, rulen));
1975 
1976 		passt = fr->fr_flags;
1977 
1978 		/*
1979 		 * Allowing a rule with the "keep state" flag set to match
1980 		 * packets that have been tagged "out of window" by the TCP
1981 		 * state tracking is foolish as the attempt to add a new
1982 		 * state entry to the table will fail.
1983 		 */
1984 		if ((passt & FR_KEEPSTATE) && (fin->fin_flx & FI_OOW))
1985 			continue;
1986 
1987 		/*
1988 		 * If the rule is a "call now" rule, then call the function
1989 		 * in the rule, if it exists and use the results from that.
1990 		 * If the function pointer is bad, just make like we ignore
1991 		 * it, except for increasing the hit counter.
1992 		 */
1993 		if ((passt & FR_CALLNOW) != 0) {
1994 			IPF_BUMP(fr->fr_hits);
1995 			if ((fr->fr_func != NULL) &&
1996 			    (fr->fr_func != (ipfunc_t)-1)) {
1997 				frentry_t *frs;
1998 
1999 				frs = fin->fin_fr;
2000 				fin->fin_fr = fr;
2001 				fr = (*fr->fr_func)(fin, &passt);
2002 				if (fr == NULL) {
2003 					fin->fin_fr = frs;
2004 					continue;
2005 				}
2006 				passt = fr->fr_flags;
2007 				fin->fin_fr = fr;
2008 			}
2009 		} else {
2010 			fin->fin_fr = fr;
2011 		}
2012 
2013 #ifdef  IPFILTER_LOG
2014 		/*
2015 		 * Just log this packet...
2016 		 */
2017 		if ((passt & FR_LOGMASK) == FR_LOG) {
2018 			if (ipflog(fin, passt) == -1) {
2019 				if (passt & FR_LOGORBLOCK) {
2020 					passt &= ~FR_CMDMASK;
2021 					passt |= FR_BLOCK|FR_QUICK;
2022 				}
2023 				IPF_BUMP(ifs->ifs_frstats[fin->fin_out].fr_skip);
2024 			}
2025 			IPF_BUMP(ifs->ifs_frstats[fin->fin_out].fr_pkl);
2026 			logged = 1;
2027 		}
2028 #endif /* IPFILTER_LOG */
2029 		fr->fr_bytes += (U_QUAD_T)fin->fin_plen;
2030 		passo = pass;
2031 		if (FR_ISSKIP(passt))
2032 			skip = fr->fr_arg;
2033 		else if ((passt & FR_LOGMASK) != FR_LOG)
2034 			pass = passt;
2035 		if (passt & (FR_RETICMP|FR_FAKEICMP))
2036 			fin->fin_icode = fr->fr_icode;
2037 		FR_DEBUG(("pass %#x\n", pass));
2038 		IPF_BUMP(fr->fr_hits);
2039 		fin->fin_rule = rulen;
2040 		(void) strncpy(fin->fin_group, fr->fr_group, FR_GROUPLEN);
2041 		if (fr->fr_grp != NULL) {
2042 			fin->fin_fr = *fr->fr_grp;
2043 			pass = fr_scanlist(fin, pass);
2044 			if (fin->fin_fr == NULL) {
2045 				fin->fin_rule = rulen;
2046 				(void) strncpy(fin->fin_group, fr->fr_group,
2047 					       FR_GROUPLEN);
2048 				fin->fin_fr = fr;
2049 			}
2050 			if (fin->fin_flx & FI_DONTCACHE)
2051 				logged = 1;
2052 		}
2053 
2054 		if (pass & FR_QUICK) {
2055 			/*
2056 			 * Finally, if we've asked to track state for this
2057 			 * packet, set it up.  Add state for "quick" rules
2058 			 * here so that if the action fails we can consider
2059 			 * the rule to "not match" and keep on processing
2060 			 * filter rules.
2061 			 */
2062 			if ((pass & FR_KEEPSTATE) &&
2063 			    !(fin->fin_flx & FI_STATE)) {
2064 				int out = fin->fin_out;
2065 
2066 				if (fr_addstate(fin, NULL, 0) != NULL) {
2067 					IPF_BUMP(ifs->ifs_frstats[out].fr_ads);
2068 				} else {
2069 					IPF_BUMP(ifs->ifs_frstats[out].fr_bads);
2070 					pass = passo;
2071 					continue;
2072 				}
2073 			}
2074 			break;
2075 		}
2076 	}
2077 	if (logged)
2078 		fin->fin_flx |= FI_DONTCACHE;
2079 	fin->fin_depth--;
2080 	return pass;
2081 }
2082 
2083 
2084 /* ------------------------------------------------------------------------ */
2085 /* Function:    fr_acctpkt                                                  */
2086 /* Returns:     frentry_t* - always returns NULL                            */
2087 /* Parameters:  fin(I) - pointer to packet information                      */
2088 /*              passp(IO) - pointer to current/new filter decision (unused) */
2089 /*                                                                          */
2090 /* Checks a packet against accounting rules, if there are any for the given */
2091 /* IP protocol version.                                                     */
2092 /*                                                                          */
2093 /* N.B.: this function returns NULL to match the prototype used by other    */
2094 /* functions called from the IPFilter "mainline" in fr_check().             */
2095 /* ------------------------------------------------------------------------ */
2096 frentry_t *fr_acctpkt(fin, passp)
2097 fr_info_t *fin;
2098 u_32_t *passp;
2099 {
2100 	char group[FR_GROUPLEN];
2101 	frentry_t *fr, *frsave;
2102 	u_32_t pass, rulen;
2103 	ipf_stack_t *ifs = fin->fin_ifs;
2104 
2105 	passp = passp;
2106 #ifdef	USE_INET6
2107 	if (fin->fin_v == 6)
2108 		fr = ifs->ifs_ipacct6[fin->fin_out][ifs->ifs_fr_active];
2109 	else
2110 #endif
2111 		fr = ifs->ifs_ipacct[fin->fin_out][ifs->ifs_fr_active];
2112 
2113 	if (fr != NULL) {
2114 		frsave = fin->fin_fr;
2115 		bcopy(fin->fin_group, group, FR_GROUPLEN);
2116 		rulen = fin->fin_rule;
2117 		fin->fin_fr = fr;
2118 		pass = fr_scanlist(fin, FR_NOMATCH);
2119 		if (FR_ISACCOUNT(pass)) {
2120 			IPF_BUMP(ifs->ifs_frstats[0].fr_acct);
2121 		}
2122 		fin->fin_fr = frsave;
2123 		bcopy(group, fin->fin_group, FR_GROUPLEN);
2124 		fin->fin_rule = rulen;
2125 	}
2126 	return NULL;
2127 }
2128 
2129 
2130 /* ------------------------------------------------------------------------ */
2131 /* Function:    fr_firewall                                                 */
2132 /* Returns:     frentry_t* - returns pointer to matched rule, if no matches */
2133 /*                           were found, returns NULL.                      */
2134 /* Parameters:  fin(I) - pointer to packet information                      */
2135 /*              passp(IO) - pointer to current/new filter decision (unused) */
2136 /*                                                                          */
2137 /* Applies an appropriate set of firewall rules to the packet, to see if    */
2138 /* there are any matches.  The first check is to see if a match can be seen */
2139 /* in the cache.  If not, then search an appropriate list of rules.  Once a */
2140 /* matching rule is found, take any appropriate actions as defined by the   */
2141 /* rule - except logging.                                                   */
2142 /* ------------------------------------------------------------------------ */
2143 static frentry_t *fr_firewall(fin, passp)
2144 fr_info_t *fin;
2145 u_32_t *passp;
2146 {
2147 	frentry_t *fr;
2148 	u_32_t pass;
2149 	int out;
2150 	ipf_stack_t *ifs = fin->fin_ifs;
2151 
2152 	out = fin->fin_out;
2153 	pass = *passp;
2154 
2155 #ifdef	USE_INET6
2156 	if (fin->fin_v == 6)
2157 		fin->fin_fr = ifs->ifs_ipfilter6[out][ifs->ifs_fr_active];
2158 	else
2159 #endif
2160 		fin->fin_fr = ifs->ifs_ipfilter[out][ifs->ifs_fr_active];
2161 	if (fin->fin_fr != NULL)
2162 		pass = fr_scanlist(fin, ifs->ifs_fr_pass);
2163 
2164 	if ((pass & FR_NOMATCH)) {
2165 		IPF_BUMP(ifs->ifs_frstats[out].fr_nom);
2166 	}
2167 	fr = fin->fin_fr;
2168 
2169 	/*
2170 	 * Apply packets per second rate-limiting to a rule as required.
2171 	 */
2172 	if ((fr != NULL) && (fr->fr_pps != 0) &&
2173 	    !ppsratecheck(&fr->fr_lastpkt, &fr->fr_curpps, fr->fr_pps)) {
2174 		pass &= ~(FR_CMDMASK|FR_DUP|FR_RETICMP|FR_RETRST);
2175 		pass |= FR_BLOCK;
2176 		IPF_BUMP(ifs->ifs_frstats[out].fr_ppshit);
2177 	}
2178 
2179 	/*
2180 	 * If we fail to add a packet to the authorization queue, then we
2181 	 * drop the packet later.  However, if it was added then pretend
2182 	 * we've dropped it already.
2183 	 */
2184 	if (FR_ISAUTH(pass)) {
2185 		if (fr_newauth(fin->fin_m, fin) != 0) {
2186 #ifdef	_KERNEL
2187 			fin->fin_m = *fin->fin_mp = NULL;
2188 #else
2189 			;
2190 #endif
2191 			fin->fin_error = 0;
2192 		} else
2193 			fin->fin_error = ENOSPC;
2194 	}
2195 
2196 	if ((fr != NULL) && (fr->fr_func != NULL) &&
2197 	    (fr->fr_func != (ipfunc_t)-1) && !(pass & FR_CALLNOW))
2198 		(void) (*fr->fr_func)(fin, &pass);
2199 
2200 	/*
2201 	 * If a rule is a pre-auth rule, check again in the list of rules
2202 	 * loaded for authenticated use.  It does not particulary matter
2203 	 * if this search fails because a "preauth" result, from a rule,
2204 	 * is treated as "not a pass", hence the packet is blocked.
2205 	 */
2206 	if (FR_ISPREAUTH(pass)) {
2207 		if ((fin->fin_fr = ifs->ifs_ipauth) != NULL)
2208 			pass = fr_scanlist(fin, ifs->ifs_fr_pass);
2209 	}
2210 
2211 	/*
2212 	 * If the rule has "keep frag" and the packet is actually a fragment,
2213 	 * then create a fragment state entry.
2214 	 */
2215 	if ((pass & (FR_KEEPFRAG|FR_KEEPSTATE)) == FR_KEEPFRAG) {
2216 		if (fin->fin_flx & FI_FRAG) {
2217 			if (fr_newfrag(fin, pass) == -1) {
2218 				IPF_BUMP(ifs->ifs_frstats[out].fr_bnfr);
2219 			} else {
2220 				IPF_BUMP(ifs->ifs_frstats[out].fr_nfr);
2221 			}
2222 		} else {
2223 			IPF_BUMP(ifs->ifs_frstats[out].fr_cfr);
2224 		}
2225 	}
2226 
2227 	/*
2228 	 * Finally, if we've asked to track state for this packet, set it up.
2229 	 */
2230 	if ((pass & FR_KEEPSTATE) && !(fin->fin_flx & FI_STATE)) {
2231 		if (fr_addstate(fin, NULL, 0) != NULL) {
2232 			IPF_BUMP(ifs->ifs_frstats[out].fr_ads);
2233 		} else {
2234 			IPF_BUMP(ifs->ifs_frstats[out].fr_bads);
2235 			if (FR_ISPASS(pass)) {
2236 				pass &= ~FR_CMDMASK;
2237 				pass |= FR_BLOCK;
2238 			}
2239 		}
2240 	}
2241 
2242 	fr = fin->fin_fr;
2243 
2244 	if (passp != NULL)
2245 		*passp = pass;
2246 
2247 	return fr;
2248 }
2249 
2250 /* ------------------------------------------------------------------------ */
2251 /* Function:    fr_check                                                    */
2252 /* Returns:     int -  0 == packet allowed through,                         */
2253 /*              User space:                                                 */
2254 /*                    -1 == packet blocked                                  */
2255 /*                     1 == packet not matched                              */
2256 /*                    -2 == requires authentication                         */
2257 /*              Kernel:                                                     */
2258 /*                   > 0 == filter error # for packet                       */
2259 /* Parameters: ip(I)   - pointer to start of IPv4/6 packet                  */
2260 /*             hlen(I) - length of header                                   */
2261 /*             ifp(I)  - pointer to interface this packet is on             */
2262 /*             out(I)  - 0 == packet going in, 1 == packet going out        */
2263 /*             mp(IO)  - pointer to caller's buffer pointer that holds this */
2264 /*                       IP packet.                                         */
2265 /* Solaris & HP-UX ONLY :                                                   */
2266 /*             qpi(I)  - pointer to STREAMS queue information for this      */
2267 /*                       interface & direction.                             */
2268 /*                                                                          */
2269 /* fr_check() is the master function for all IPFilter packet processing.    */
2270 /* It orchestrates: Network Address Translation (NAT), checking for packet  */
2271 /* authorisation (or pre-authorisation), presence of related state info.,   */
2272 /* generating log entries, IP packet accounting, routing of packets as      */
2273 /* directed by firewall rules and of course whether or not to allow the     */
2274 /* packet to be further processed by the kernel.                            */
2275 /*                                                                          */
2276 /* For packets blocked, the contents of "mp" will be NULL'd and the buffer  */
2277 /* freed.  Packets passed may be returned with the pointer pointed to by    */
2278 /* by "mp" changed to a new buffer.                                         */
2279 /* ------------------------------------------------------------------------ */
2280 int fr_check(ip, hlen, ifp, out
2281 #if defined(_KERNEL) && defined(MENTAT)
2282 , qif, mp, ifs)
2283 void *qif;
2284 #else
2285 , mp, ifs)
2286 #endif
2287 mb_t **mp;
2288 ip_t *ip;
2289 int hlen;
2290 void *ifp;
2291 int out;
2292 ipf_stack_t *ifs;
2293 {
2294 	/*
2295 	 * The above really sucks, but short of writing a diff
2296 	 */
2297 	fr_info_t frinfo;
2298 	fr_info_t *fin = &frinfo;
2299 	u_32_t pass;
2300 	frentry_t *fr = NULL;
2301 	int v = IP_V(ip);
2302 	mb_t *mc = NULL;
2303 	mb_t *m;
2304 #ifdef USE_INET6
2305 	ip6_t *ip6;
2306 #endif
2307 #ifdef	_KERNEL
2308 # ifdef MENTAT
2309 	qpktinfo_t *qpi = qif;
2310 #endif
2311 #endif
2312 
2313 	SPL_INT(s);
2314 	pass = ifs->ifs_fr_pass;
2315 
2316 	/*
2317 	 * The first part of fr_check() deals with making sure that what goes
2318 	 * into the filtering engine makes some sense.  Information about the
2319 	 * the packet is distilled, collected into a fr_info_t structure and
2320 	 * the an attempt to ensure the buffer the packet is in is big enough
2321 	 * to hold all the required packet headers.
2322 	 */
2323 #ifdef	_KERNEL
2324 # ifdef MENTAT
2325 	if (!OK_32PTR(ip))
2326 		return 2;
2327 # endif
2328 
2329 
2330 	if (ifs->ifs_fr_running <= 0) {
2331 		return 0;
2332 	}
2333 
2334 	bzero((char *)fin, sizeof(*fin));
2335 
2336 # ifdef MENTAT
2337 	fin->fin_flx = qpi->qpi_flags & (FI_NOCKSUM|FI_MBCAST|FI_MULTICAST|
2338 					 FI_BROADCAST);
2339 	m = qpi->qpi_m;
2340 	fin->fin_qfm = m;
2341 	fin->fin_qpi = qpi;
2342 # else /* MENTAT */
2343 
2344 	m = *mp;
2345 
2346 #  if defined(M_MCAST)
2347 	if ((m->m_flags & M_MCAST) != 0)
2348 		fin->fin_flx |= FI_MBCAST|FI_MULTICAST;
2349 #  endif
2350 #  if defined(M_MLOOP)
2351 	if ((m->m_flags & M_MLOOP) != 0)
2352 		fin->fin_flx |= FI_MBCAST|FI_MULTICAST;
2353 #  endif
2354 #  if defined(M_BCAST)
2355 	if ((m->m_flags & M_BCAST) != 0)
2356 		fin->fin_flx |= FI_MBCAST|FI_BROADCAST;
2357 #  endif
2358 #  ifdef M_CANFASTFWD
2359 	/*
2360 	 * XXX For now, IP Filter and fast-forwarding of cached flows
2361 	 * XXX are mutually exclusive.  Eventually, IP Filter should
2362 	 * XXX get a "can-fast-forward" filter rule.
2363 	 */
2364 	m->m_flags &= ~M_CANFASTFWD;
2365 #  endif /* M_CANFASTFWD */
2366 #  ifdef CSUM_DELAY_DATA
2367 	/*
2368 	 * disable delayed checksums.
2369 	 */
2370 	if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
2371 		in_delayed_cksum(m);
2372 		m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA;
2373 	}
2374 #  endif /* CSUM_DELAY_DATA */
2375 # endif /* MENTAT */
2376 #else
2377 
2378 	bzero((char *)fin, sizeof(*fin));
2379 	m = *mp;
2380 #endif /* _KERNEL */
2381 
2382 	fin->fin_v = v;
2383 	fin->fin_m = m;
2384 	fin->fin_ip = ip;
2385 	fin->fin_mp = mp;
2386 	fin->fin_out = out;
2387 	fin->fin_ifp = ifp;
2388 	fin->fin_error = ENETUNREACH;
2389 	fin->fin_hlen = (u_short)hlen;
2390 	fin->fin_dp = (char *)ip + hlen;
2391 	fin->fin_ipoff = (char *)ip - MTOD(m, char *);
2392 	fin->fin_ifs = ifs;
2393 
2394 	SPL_NET(s);
2395 
2396 #ifdef	USE_INET6
2397 	if (v == 6) {
2398 		IPF_BUMP(ifs->ifs_frstats[out].fr_ipv6);
2399 		/*
2400 		 * Jumbo grams are quite likely too big for internal buffer
2401 		 * structures to handle comfortably, for now, so just drop
2402 		 * them.
2403 		 */
2404 		ip6 = (ip6_t *)ip;
2405 		fin->fin_plen = ntohs(ip6->ip6_plen);
2406 		if (fin->fin_plen == 0) {
2407 			READ_ENTER(&ifs->ifs_ipf_mutex);
2408 			pass = FR_BLOCK|FR_NOMATCH;
2409 			goto filtered;
2410 		}
2411 		fin->fin_plen += sizeof(ip6_t);
2412 	} else
2413 #endif
2414 	{
2415 #if (OpenBSD >= 200311) && defined(_KERNEL)
2416 		ip->ip_len = ntohs(ip->ip_len);
2417 		ip->ip_off = ntohs(ip->ip_off);
2418 #endif
2419 		fin->fin_plen = ip->ip_len;
2420 	}
2421 
2422 	if (fr_makefrip(hlen, ip, fin) == -1) {
2423 		READ_ENTER(&ifs->ifs_ipf_mutex);
2424 		pass = FR_BLOCK;
2425 		goto filtered;
2426 	}
2427 
2428 	/*
2429 	 * For at least IPv6 packets, if a m_pullup() fails then this pointer
2430 	 * becomes NULL and so we have no packet to free.
2431 	 */
2432 	if (*fin->fin_mp == NULL)
2433 		goto finished;
2434 
2435 	if (!out) {
2436 		if (v == 4) {
2437 #ifdef _KERNEL
2438 			if (ifs->ifs_fr_chksrc && !fr_verifysrc(fin)) {
2439 				IPF_BUMP(ifs->ifs_frstats[0].fr_badsrc);
2440 				fin->fin_flx |= FI_BADSRC;
2441 			}
2442 #endif
2443 			if (fin->fin_ip->ip_ttl < ifs->ifs_fr_minttl) {
2444 				IPF_BUMP(ifs->ifs_frstats[0].fr_badttl);
2445 				fin->fin_flx |= FI_LOWTTL;
2446 			}
2447 		}
2448 #ifdef USE_INET6
2449 		else  if (v == 6) {
2450 			ip6 = (ip6_t *)ip;
2451 #ifdef _KERNEL
2452 			if (ifs->ifs_fr_chksrc && !fr_verifysrc(fin)) {
2453 				IPF_BUMP(ifs->ifs_frstats[0].fr_badsrc);
2454 				fin->fin_flx |= FI_BADSRC;
2455 			}
2456 #endif
2457 			if (ip6->ip6_hlim < ifs->ifs_fr_minttl) {
2458 				IPF_BUMP(ifs->ifs_frstats[0].fr_badttl);
2459 				fin->fin_flx |= FI_LOWTTL;
2460 			}
2461 		}
2462 #endif
2463 	}
2464 
2465 	if (fin->fin_flx & FI_SHORT) {
2466 		IPF_BUMP(ifs->ifs_frstats[out].fr_short);
2467 	}
2468 
2469 	READ_ENTER(&ifs->ifs_ipf_mutex);
2470 
2471 	/*
2472 	 * Check auth now.  This, combined with the check below to see if apass
2473 	 * is 0 is to ensure that we don't count the packet twice, which can
2474 	 * otherwise occur when we reprocess it.  As it is, we only count it
2475 	 * after it has no auth. table matchup.  This also stops NAT from
2476 	 * occuring until after the packet has been auth'd.
2477 	 */
2478 	fr = fr_checkauth(fin, &pass);
2479 	if (!out) {
2480 		switch (fin->fin_v)
2481 		{
2482 		case 4 :
2483 			if (fr_checknatin(fin, &pass) == -1) {
2484 				RWLOCK_EXIT(&ifs->ifs_ipf_mutex);
2485 				goto finished;
2486 			}
2487 			break;
2488 #ifdef	USE_INET6
2489 		case 6 :
2490 			if (fr_checknat6in(fin, &pass) == -1) {
2491 				RWLOCK_EXIT(&ifs->ifs_ipf_mutex);
2492 				goto finished;
2493 			}
2494 			break;
2495 #endif
2496 		default :
2497 			break;
2498 		}
2499 	}
2500 	if (!out)
2501 		(void) fr_acctpkt(fin, NULL);
2502 
2503 	if (fr == NULL)
2504 		if ((fin->fin_flx & (FI_FRAG|FI_BAD)) == FI_FRAG)
2505 			fr = fr_knownfrag(fin, &pass);
2506 	if (fr == NULL)
2507 		fr = fr_checkstate(fin, &pass);
2508 
2509 	if ((pass & FR_NOMATCH) || (fr == NULL))
2510 		fr = fr_firewall(fin, &pass);
2511 
2512 	fin->fin_fr = fr;
2513 
2514 	/*
2515 	 * Only count/translate packets which will be passed on, out the
2516 	 * interface.
2517 	 */
2518 	if (out && FR_ISPASS(pass)) {
2519 		(void) fr_acctpkt(fin, NULL);
2520 
2521 		switch (fin->fin_v)
2522 		{
2523 		case 4 :
2524 			if (fr_checknatout(fin, &pass) == -1) {
2525 				RWLOCK_EXIT(&ifs->ifs_ipf_mutex);
2526 				goto finished;
2527 			}
2528 			break;
2529 #ifdef	USE_INET6
2530 		case 6 :
2531 			if (fr_checknat6out(fin, &pass) == -1) {
2532 				RWLOCK_EXIT(&ifs->ifs_ipf_mutex);
2533 				goto finished;
2534 			}
2535 			break;
2536 #endif
2537 		default :
2538 			break;
2539 		}
2540 
2541 		if ((ifs->ifs_fr_update_ipid != 0) && (v == 4)) {
2542 			if (fr_updateipid(fin) == -1) {
2543 				IPF_BUMP(ifs->ifs_frstats[1].fr_ipud);
2544 				pass &= ~FR_CMDMASK;
2545 				pass |= FR_BLOCK;
2546 			} else {
2547 				IPF_BUMP(ifs->ifs_frstats[0].fr_ipud);
2548 			}
2549 		}
2550 	}
2551 
2552 #ifdef	IPFILTER_LOG
2553 	if ((ifs->ifs_fr_flags & FF_LOGGING) || (pass & FR_LOGMASK)) {
2554 		(void) fr_dolog(fin, &pass);
2555 	}
2556 #endif
2557 
2558 	/*
2559 	 * The FI_STATE flag is cleared here so that calling fr_checkstate
2560 	 * will work when called from inside of fr_fastroute.  Although
2561 	 * there is a similar flag, FI_NATED, for NAT, it does have the same
2562 	 * impact on code execution.
2563 	 */
2564 	fin->fin_flx &= ~FI_STATE;
2565 
2566 	/*
2567 	 * Only allow FR_DUP to work if a rule matched - it makes no sense to
2568 	 * set FR_DUP as a "default" as there are no instructions about where
2569 	 * to send the packet.  Use fin_m here because it may have changed
2570 	 * (without an update of 'm') in prior processing.
2571 	 */
2572 	if ((fr != NULL) && (pass & FR_DUP)) {
2573 		mc = M_DUPLICATE(fin->fin_m);
2574 	}
2575 
2576 	if (pass & (FR_RETRST|FR_RETICMP)) {
2577 		/*
2578 		 * Should we return an ICMP packet to indicate error
2579 		 * status passing through the packet filter ?
2580 		 * WARNING: ICMP error packets AND TCP RST packets should
2581 		 * ONLY be sent in repsonse to incoming packets.  Sending them
2582 		 * in response to outbound packets can result in a panic on
2583 		 * some operating systems.
2584 		 */
2585 		if (!out) {
2586 			if (pass & FR_RETICMP) {
2587 				int dst;
2588 
2589 				if ((pass & FR_RETMASK) == FR_FAKEICMP)
2590 					dst = 1;
2591 				else
2592 					dst = 0;
2593 #if defined(_KERNEL) && (SOLARIS2 >= 10)
2594 				/*
2595 				 * Assume it's possible to enter insane rule:
2596 				 * 	pass return-icmp in proto udp ...
2597 				 * then we have no other option than to forward
2598 				 * packet on loopback and give up any attempt
2599 				 * to create a fake response.
2600 				 */
2601 				if (IPF_IS_LOOPBACK(qpi->qpi_flags) &&
2602 				    FR_ISBLOCK(pass)) {
2603 
2604 					if (fr_make_icmp(fin) == 0) {
2605 						IPF_BUMP(
2606 						ifs->ifs_frstats[out].fr_ret);
2607 					}
2608 					/*
2609 					 * we drop packet silently in case we
2610 					 * failed assemble fake response for it
2611 					 */
2612 					else if (*mp != NULL) {
2613 						FREE_MB_T(*mp);
2614 						m = *mp = NULL;
2615 					}
2616 
2617 					IPF_BUMP(
2618 					    ifs->ifs_frstats[out].fr_block);
2619 					RWLOCK_EXIT(&ifs->ifs_ipf_mutex);
2620 
2621 					return (0);
2622 				}
2623 #endif	/* _KERNEL && SOLARIS2 >= 10 */
2624 
2625 				(void) fr_send_icmp_err(ICMP_UNREACH, fin, dst);
2626 				IPF_BUMP(ifs->ifs_frstats[out].fr_ret);
2627 
2628 			} else if (((pass & FR_RETMASK) == FR_RETRST) &&
2629 				   !(fin->fin_flx & FI_SHORT)) {
2630 
2631 #if defined(_KERNEL) && (SOLARIS2 >= 10)
2632 				/*
2633 				 * Assume it's possible to enter insane rule:
2634 				 * 	pass return-rst in proto tcp ...
2635 				 * then we have no other option than to forward
2636 				 * packet on loopback and give up any attempt
2637 				 * to create a fake response.
2638 				 */
2639 				if (IPF_IS_LOOPBACK(qpi->qpi_flags) &&
2640 				    FR_ISBLOCK(pass)) {
2641 					if (fr_make_rst(fin) == 0) {
2642 						IPF_BUMP(
2643 						ifs->ifs_frstats[out].fr_ret);
2644 					}
2645 					else if (mp != NULL) {
2646 					/*
2647 					 * we drop packet silently in case we
2648 					 * failed assemble fake response for it
2649 					 */
2650 						FREE_MB_T(*mp);
2651 						m = *mp = NULL;
2652 					}
2653 
2654 					IPF_BUMP(
2655 					    ifs->ifs_frstats[out].fr_block);
2656 					RWLOCK_EXIT(&ifs->ifs_ipf_mutex);
2657 
2658 					return (0);
2659 				 }
2660 #endif /* _KERNEL && _SOLARIS2 >= 10 */
2661 				if (fr_send_reset(fin) == 0) {
2662 					IPF_BUMP(ifs->ifs_frstats[1].fr_ret);
2663 				}
2664 			}
2665 		} else {
2666 			if (pass & FR_RETRST)
2667 				fin->fin_error = ECONNRESET;
2668 		}
2669 	}
2670 
2671 	/*
2672 	 * If we didn't drop off the bottom of the list of rules (and thus
2673 	 * the 'current' rule fr is not NULL), then we may have some extra
2674 	 * instructions about what to do with a packet.
2675 	 * Once we're finished return to our caller, freeing the packet if
2676 	 * we are dropping it (* BSD ONLY *).
2677 	 * Reassign m from fin_m as we may have a new buffer, now.
2678 	 */
2679 filtered:
2680 	m = fin->fin_m;
2681 
2682 	if (fr != NULL) {
2683 		frdest_t *fdp;
2684 
2685 		fdp = &fr->fr_tifs[fin->fin_rev];
2686 
2687 		if (!out && (pass & FR_FASTROUTE)) {
2688 			/*
2689 			 * For fastroute rule, no destioation interface defined
2690 			 * so pass NULL as the frdest_t parameter
2691 			 */
2692 			(void) fr_fastroute(m, mp, fin, NULL);
2693 			m = *mp = NULL;
2694 		} else if ((fdp->fd_ifp != NULL) &&
2695 			   (fdp->fd_ifp != (struct ifnet *)-1)) {
2696 			/* this is for to rules: */
2697 			(void) fr_fastroute(m, mp, fin, fdp);
2698 			m = *mp = NULL;
2699 		}
2700 
2701 		/*
2702 		 * Generate a duplicated packet.
2703 		 */
2704 		if (mc != NULL)
2705 			(void) fr_fastroute(mc, &mc, fin, &fr->fr_dif);
2706 	}
2707 
2708 	if (FR_ISBLOCK(pass) && (fin->fin_flx & FI_NEWNAT))
2709 		nat_uncreate(fin);
2710 
2711 	/*
2712 	 * This late because the likes of fr_fastroute() use fin_fr.
2713 	 */
2714 	RWLOCK_EXIT(&ifs->ifs_ipf_mutex);
2715 
2716 finished:
2717 	if (!FR_ISPASS(pass)) {
2718 		IPF_BUMP(ifs->ifs_frstats[out].fr_block);
2719 		if (*mp != NULL) {
2720 			FREE_MB_T(*mp);
2721 			m = *mp = NULL;
2722 		}
2723 	} else {
2724 		IPF_BUMP(ifs->ifs_frstats[out].fr_pass);
2725 #if defined(_KERNEL) && defined(__sgi)
2726 		if ((fin->fin_hbuf != NULL) &&
2727 		    (mtod(fin->fin_m, struct ip *) != fin->fin_ip)) {
2728 			COPYBACK(m, 0, fin->fin_plen, fin->fin_hbuf);
2729 		}
2730 #endif
2731 	}
2732 
2733 	SPL_X(s);
2734 
2735 #ifdef _KERNEL
2736 # if OpenBSD >= 200311
2737 	if (FR_ISPASS(pass) && (v == 4)) {
2738 		ip = fin->fin_ip;
2739 		ip->ip_len = ntohs(ip->ip_len);
2740 		ip->ip_off = ntohs(ip->ip_off);
2741 	}
2742 # endif
2743 	return (FR_ISPASS(pass)) ? 0 : fin->fin_error;
2744 #else /* _KERNEL */
2745 	FR_VERBOSE(("fin_flx %#x pass %#x ", fin->fin_flx, pass));
2746 	if ((pass & FR_NOMATCH) != 0)
2747 		return 1;
2748 
2749 	if ((pass & FR_RETMASK) != 0)
2750 		switch (pass & FR_RETMASK)
2751 		{
2752 		case FR_RETRST :
2753 			return 3;
2754 		case FR_RETICMP :
2755 			return 4;
2756 		case FR_FAKEICMP :
2757 			return 5;
2758 		}
2759 
2760 	switch (pass & FR_CMDMASK)
2761 	{
2762 	case FR_PASS :
2763 		return 0;
2764 	case FR_BLOCK :
2765 		return -1;
2766 	case FR_AUTH :
2767 		return -2;
2768 	case FR_ACCOUNT :
2769 		return -3;
2770 	case FR_PREAUTH :
2771 		return -4;
2772 	}
2773 	return 2;
2774 #endif /* _KERNEL */
2775 }
2776 
2777 
2778 #ifdef	IPFILTER_LOG
2779 /* ------------------------------------------------------------------------ */
2780 /* Function:    fr_dolog                                                    */
2781 /* Returns:     frentry_t* - returns contents of fin_fr (no change made)    */
2782 /* Parameters:  fin(I) - pointer to packet information                      */
2783 /*              passp(IO) - pointer to current/new filter decision (unused) */
2784 /*                                                                          */
2785 /* Checks flags set to see how a packet should be logged, if it is to be    */
2786 /* logged.  Adjust statistics based on its success or not.                  */
2787 /* ------------------------------------------------------------------------ */
2788 frentry_t *fr_dolog(fin, passp)
2789 fr_info_t *fin;
2790 u_32_t *passp;
2791 {
2792 	u_32_t pass;
2793 	int out;
2794 	ipf_stack_t *ifs = fin->fin_ifs;
2795 
2796 	out = fin->fin_out;
2797 	pass = *passp;
2798 
2799 	if ((ifs->ifs_fr_flags & FF_LOGNOMATCH) && (pass & FR_NOMATCH)) {
2800 		pass |= FF_LOGNOMATCH;
2801 		IPF_BUMP(ifs->ifs_frstats[out].fr_npkl);
2802 		goto logit;
2803 	} else if (((pass & FR_LOGMASK) == FR_LOGP) ||
2804 	    (FR_ISPASS(pass) && (ifs->ifs_fr_flags & FF_LOGPASS))) {
2805 		if ((pass & FR_LOGMASK) != FR_LOGP)
2806 			pass |= FF_LOGPASS;
2807 		IPF_BUMP(ifs->ifs_frstats[out].fr_ppkl);
2808 		goto logit;
2809 	} else if (((pass & FR_LOGMASK) == FR_LOGB) ||
2810 		   (FR_ISBLOCK(pass) && (ifs->ifs_fr_flags & FF_LOGBLOCK))) {
2811 		if ((pass & FR_LOGMASK) != FR_LOGB)
2812 			pass |= FF_LOGBLOCK;
2813 		IPF_BUMP(ifs->ifs_frstats[out].fr_bpkl);
2814 logit:
2815 		if (ipflog(fin, pass) == -1) {
2816 			IPF_BUMP(ifs->ifs_frstats[out].fr_skip);
2817 
2818 			/*
2819 			 * If the "or-block" option has been used then
2820 			 * block the packet if we failed to log it.
2821 			 */
2822 			if ((pass & FR_LOGORBLOCK) &&
2823 			    FR_ISPASS(pass)) {
2824 				pass &= ~FR_CMDMASK;
2825 				pass |= FR_BLOCK;
2826 			}
2827 		}
2828 		*passp = pass;
2829 	}
2830 
2831 	return fin->fin_fr;
2832 }
2833 #endif /* IPFILTER_LOG */
2834 
2835 
2836 /* ------------------------------------------------------------------------ */
2837 /* Function:    ipf_cksum                                                   */
2838 /* Returns:     u_short - IP header checksum                                */
2839 /* Parameters:  addr(I) - pointer to start of buffer to checksum            */
2840 /*              len(I)  - length of buffer in bytes                         */
2841 /*                                                                          */
2842 /* Calculate the two's complement 16 bit checksum of the buffer passed.     */
2843 /*                                                                          */
2844 /* N.B.: addr should be 16bit aligned.                                      */
2845 /* ------------------------------------------------------------------------ */
2846 u_short ipf_cksum(addr, len)
2847 u_short *addr;
2848 int len;
2849 {
2850 	u_32_t sum = 0;
2851 
2852 	for (sum = 0; len > 1; len -= 2)
2853 		sum += *addr++;
2854 
2855 	/* mop up an odd byte, if necessary */
2856 	if (len == 1)
2857 		sum += *(u_char *)addr;
2858 
2859 	/*
2860 	 * add back carry outs from top 16 bits to low 16 bits
2861 	 */
2862 	sum = (sum >> 16) + (sum & 0xffff);	/* add hi 16 to low 16 */
2863 	sum += (sum >> 16);			/* add carry */
2864 	return (u_short)(~sum);
2865 }
2866 
2867 
2868 /* ------------------------------------------------------------------------ */
2869 /* Function:    fr_cksum                                                    */
2870 /* Returns:     u_short - layer 4 checksum                                  */
2871 /* Parameters:  m(I  )     - pointer to buffer holding packet               */
2872 /*              ip(I)      - pointer to IP header                           */
2873 /*              l4proto(I) - protocol to caclulate checksum for             */
2874 /*              l4hdr(I)   - pointer to layer 4 header                      */
2875 /*                                                                          */
2876 /* Calculates the TCP checksum for the packet held in "m", using the data   */
2877 /* in the IP header "ip" to seed it.                                        */
2878 /*                                                                          */
2879 /* NB: This function assumes we've pullup'd enough for all of the IP header */
2880 /* and the TCP header.  We also assume that data blocks aren't allocated in */
2881 /* odd sizes.                                                               */
2882 /*                                                                          */
2883 /* Expects ip_len to be in host byte order when called.                     */
2884 /* ------------------------------------------------------------------------ */
2885 u_short fr_cksum(m, ip, l4proto, l4hdr)
2886 mb_t *m;
2887 ip_t *ip;
2888 int l4proto;
2889 void *l4hdr;
2890 {
2891 	u_short *sp, slen, sumsave, l4hlen, *csump;
2892 	u_int sum, sum2;
2893 	int hlen;
2894 #ifdef	USE_INET6
2895 	ip6_t *ip6;
2896 #endif
2897 
2898 	csump = NULL;
2899 	sumsave = 0;
2900 	l4hlen = 0;
2901 	sp = NULL;
2902 	slen = 0;
2903 	hlen = 0;
2904 	sum = 0;
2905 
2906 	/*
2907 	 * Add up IP Header portion
2908 	 */
2909 #ifdef	USE_INET6
2910 	if (IP_V(ip) == 4) {
2911 #endif
2912 		hlen = IP_HL(ip) << 2;
2913 		slen = ip->ip_len - hlen;
2914 		sum = htons((u_short)l4proto);
2915 		sum += htons(slen);
2916 		sp = (u_short *)&ip->ip_src;
2917 		sum += *sp++;	/* ip_src */
2918 		sum += *sp++;
2919 		sum += *sp++;	/* ip_dst */
2920 		sum += *sp++;
2921 #ifdef	USE_INET6
2922 	} else if (IP_V(ip) == 6) {
2923 		ip6 = (ip6_t *)ip;
2924 		hlen = sizeof(*ip6);
2925 		slen = ntohs(ip6->ip6_plen);
2926 		sum = htons((u_short)l4proto);
2927 		sum += htons(slen);
2928 		sp = (u_short *)&ip6->ip6_src;
2929 		sum += *sp++;	/* ip6_src */
2930 		sum += *sp++;
2931 		sum += *sp++;
2932 		sum += *sp++;
2933 		sum += *sp++;
2934 		sum += *sp++;
2935 		sum += *sp++;
2936 		sum += *sp++;
2937 		sum += *sp++;	/* ip6_dst */
2938 		sum += *sp++;
2939 		sum += *sp++;
2940 		sum += *sp++;
2941 		sum += *sp++;
2942 		sum += *sp++;
2943 		sum += *sp++;
2944 		sum += *sp++;
2945 	}
2946 #endif
2947 
2948 	switch (l4proto)
2949 	{
2950 	case IPPROTO_UDP :
2951 		csump = &((udphdr_t *)l4hdr)->uh_sum;
2952 		l4hlen = sizeof(udphdr_t);
2953 		break;
2954 
2955 	case IPPROTO_TCP :
2956 		csump = &((tcphdr_t *)l4hdr)->th_sum;
2957 		l4hlen = sizeof(tcphdr_t);
2958 		break;
2959 	case IPPROTO_ICMP :
2960 		csump = &((icmphdr_t *)l4hdr)->icmp_cksum;
2961 		l4hlen = 4;
2962 		sum = 0;
2963 		break;
2964 	default :
2965 		break;
2966 	}
2967 
2968 	if (csump != NULL) {
2969 		sumsave = *csump;
2970 		*csump = 0;
2971 	}
2972 
2973 	l4hlen = l4hlen;	/* LINT */
2974 
2975 #ifdef	_KERNEL
2976 # ifdef MENTAT
2977 	{
2978 	void *rp = m->b_rptr;
2979 
2980 	if ((unsigned char *)ip > m->b_rptr && (unsigned char *)ip < m->b_wptr)
2981 		m->b_rptr = (u_char *)ip;
2982 	sum2 = ip_cksum(m, hlen, sum);	/* hlen == offset */
2983 	m->b_rptr = rp;
2984 	sum2 = (sum2 & 0xffff) + (sum2 >> 16);
2985 	sum2 = ~sum2 & 0xffff;
2986 	}
2987 # else /* MENTAT */
2988 #  if defined(BSD) || defined(sun)
2989 #   if BSD >= 199103
2990 	m->m_data += hlen;
2991 #   else
2992 	m->m_off += hlen;
2993 #   endif
2994 	m->m_len -= hlen;
2995 	sum2 = in_cksum(m, slen);
2996 	m->m_len += hlen;
2997 #   if BSD >= 199103
2998 	m->m_data -= hlen;
2999 #   else
3000 	m->m_off -= hlen;
3001 #   endif
3002 	/*
3003 	 * Both sum and sum2 are partial sums, so combine them together.
3004 	 */
3005 	sum += ~sum2 & 0xffff;
3006 	while (sum > 0xffff)
3007 		sum = (sum & 0xffff) + (sum >> 16);
3008 	sum2 = ~sum & 0xffff;
3009 #  else /* defined(BSD) || defined(sun) */
3010 {
3011 	union {
3012 		u_char	c[2];
3013 		u_short	s;
3014 	} bytes;
3015 	u_short len = ip->ip_len;
3016 #   if defined(__sgi)
3017 	int add;
3018 #   endif
3019 
3020 	/*
3021 	 * Add up IP Header portion
3022 	 */
3023 	if (sp != (u_short *)l4hdr)
3024 		sp = (u_short *)l4hdr;
3025 
3026 	switch (l4proto)
3027 	{
3028 	case IPPROTO_UDP :
3029 		sum += *sp++;	/* sport */
3030 		sum += *sp++;	/* dport */
3031 		sum += *sp++;	/* udp length */
3032 		sum += *sp++;	/* checksum */
3033 		break;
3034 
3035 	case IPPROTO_TCP :
3036 		sum += *sp++;	/* sport */
3037 		sum += *sp++;	/* dport */
3038 		sum += *sp++;	/* seq */
3039 		sum += *sp++;
3040 		sum += *sp++;	/* ack */
3041 		sum += *sp++;
3042 		sum += *sp++;	/* off */
3043 		sum += *sp++;	/* win */
3044 		sum += *sp++;	/* checksum */
3045 		sum += *sp++;	/* urp */
3046 		break;
3047 	case IPPROTO_ICMP :
3048 		sum = *sp++;	/* type/code */
3049 		sum += *sp++;	/* checksum */
3050 		break;
3051 	}
3052 
3053 #   ifdef	__sgi
3054 	/*
3055 	 * In case we had to copy the IP & TCP header out of mbufs,
3056 	 * skip over the mbuf bits which are the header
3057 	 */
3058 	if ((caddr_t)ip != mtod(m, caddr_t)) {
3059 		hlen = (caddr_t)sp - (caddr_t)ip;
3060 		while (hlen) {
3061 			add = MIN(hlen, m->m_len);
3062 			sp = (u_short *)(mtod(m, caddr_t) + add);
3063 			hlen -= add;
3064 			if (add == m->m_len) {
3065 				m = m->m_next;
3066 				if (!hlen) {
3067 					if (!m)
3068 						break;
3069 					sp = mtod(m, u_short *);
3070 				}
3071 				PANIC((!m),("fr_cksum(1): not enough data"));
3072 			}
3073 		}
3074 	}
3075 #   endif
3076 
3077 	len -= (l4hlen + hlen);
3078 	if (len <= 0)
3079 		goto nodata;
3080 
3081 	while (len > 1) {
3082 		if (((caddr_t)sp - mtod(m, caddr_t)) >= m->m_len) {
3083 			m = m->m_next;
3084 			PANIC((!m),("fr_cksum(2): not enough data"));
3085 			sp = mtod(m, u_short *);
3086 		}
3087 		if (((caddr_t)(sp + 1) - mtod(m, caddr_t)) > m->m_len) {
3088 			bytes.c[0] = *(u_char *)sp;
3089 			m = m->m_next;
3090 			PANIC((!m),("fr_cksum(3): not enough data"));
3091 			sp = mtod(m, u_short *);
3092 			bytes.c[1] = *(u_char *)sp;
3093 			sum += bytes.s;
3094 			sp = (u_short *)((u_char *)sp + 1);
3095 		}
3096 		if ((u_long)sp & 1) {
3097 			bcopy((char *)sp++, (char *)&bytes.s, sizeof(bytes.s));
3098 			sum += bytes.s;
3099 		} else
3100 			sum += *sp++;
3101 		len -= 2;
3102 	}
3103 
3104 	if (len != 0)
3105 		sum += ntohs(*(u_char *)sp << 8);
3106 nodata:
3107 	while (sum > 0xffff)
3108 		sum = (sum & 0xffff) + (sum >> 16);
3109 	sum2 = (u_short)(~sum & 0xffff);
3110 }
3111 #  endif /*  defined(BSD) || defined(sun) */
3112 # endif /* MENTAT */
3113 #else /* _KERNEL */
3114 	for (; slen > 1; slen -= 2)
3115 	        sum += *sp++;
3116 	if (slen)
3117 		sum += ntohs(*(u_char *)sp << 8);
3118 	while (sum > 0xffff)
3119 		sum = (sum & 0xffff) + (sum >> 16);
3120 	sum2 = (u_short)(~sum & 0xffff);
3121 #endif /* _KERNEL */
3122 	if (csump != NULL)
3123 		*csump = sumsave;
3124 	return sum2;
3125 }
3126 
3127 
3128 #if defined(_KERNEL) && ( ((BSD < 199103) && !defined(MENTAT)) || \
3129     defined(__sgi) ) && !defined(linux) && !defined(_AIX51)
3130 /*
3131  * Copyright (c) 1982, 1986, 1988, 1991, 1993
3132  *	The Regents of the University of California.  All rights reserved.
3133  *
3134  * Redistribution and use in source and binary forms, with or without
3135  * modification, are permitted provided that the following conditions
3136  * are met:
3137  * 1. Redistributions of source code must retain the above copyright
3138  *    notice, this list of conditions and the following disclaimer.
3139  * 2. Redistributions in binary form must reproduce the above copyright
3140  *    notice, this list of conditions and the following disclaimer in the
3141  *    documentation and/or other materials provided with the distribution.
3142  * 3. Neither the name of the University nor the names of its contributors
3143  *    may be used to endorse or promote products derived from this software
3144  *    without specific prior written permission.
3145  *
3146  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
3147  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
3148  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
3149  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
3150  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
3151  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
3152  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
3153  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
3154  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
3155  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
3156  * SUCH DAMAGE.
3157  *
3158  *	@(#)uipc_mbuf.c	8.2 (Berkeley) 1/4/94
3159  * $Id: fil.c,v 2.243.2.64 2005/08/13 05:19:59 darrenr Exp $
3160  */
3161 /*
3162  * Copy data from an mbuf chain starting "off" bytes from the beginning,
3163  * continuing for "len" bytes, into the indicated buffer.
3164  */
3165 void
3166 m_copydata(m, off, len, cp)
3167 	mb_t *m;
3168 	int off;
3169 	int len;
3170 	caddr_t cp;
3171 {
3172 	unsigned count;
3173 
3174 	if (off < 0 || len < 0)
3175 		panic("m_copydata");
3176 	while (off > 0) {
3177 		if (m == 0)
3178 			panic("m_copydata");
3179 		if (off < m->m_len)
3180 			break;
3181 		off -= m->m_len;
3182 		m = m->m_next;
3183 	}
3184 	while (len > 0) {
3185 		if (m == 0)
3186 			panic("m_copydata");
3187 		count = MIN(m->m_len - off, len);
3188 		bcopy(mtod(m, caddr_t) + off, cp, count);
3189 		len -= count;
3190 		cp += count;
3191 		off = 0;
3192 		m = m->m_next;
3193 	}
3194 }
3195 
3196 
3197 /*
3198  * Copy data from a buffer back into the indicated mbuf chain,
3199  * starting "off" bytes from the beginning, extending the mbuf
3200  * chain if necessary.
3201  */
3202 void
3203 m_copyback(m0, off, len, cp)
3204 	struct	mbuf *m0;
3205 	int off;
3206 	int len;
3207 	caddr_t cp;
3208 {
3209 	int mlen;
3210 	struct mbuf *m = m0, *n;
3211 	int totlen = 0;
3212 
3213 	if (m0 == 0)
3214 		return;
3215 	while (off > (mlen = m->m_len)) {
3216 		off -= mlen;
3217 		totlen += mlen;
3218 		if (m->m_next == 0) {
3219 			n = m_getclr(M_DONTWAIT, m->m_type);
3220 			if (n == 0)
3221 				goto out;
3222 			n->m_len = min(MLEN, len + off);
3223 			m->m_next = n;
3224 		}
3225 		m = m->m_next;
3226 	}
3227 	while (len > 0) {
3228 		mlen = min(m->m_len - off, len);
3229 		bcopy(cp, off + mtod(m, caddr_t), (unsigned)mlen);
3230 		cp += mlen;
3231 		len -= mlen;
3232 		mlen += off;
3233 		off = 0;
3234 		totlen += mlen;
3235 		if (len == 0)
3236 			break;
3237 		if (m->m_next == 0) {
3238 			n = m_get(M_DONTWAIT, m->m_type);
3239 			if (n == 0)
3240 				break;
3241 			n->m_len = min(MLEN, len);
3242 			m->m_next = n;
3243 		}
3244 		m = m->m_next;
3245 	}
3246 out:
3247 #if 0
3248 	if (((m = m0)->m_flags & M_PKTHDR) && (m->m_pkthdr.len < totlen))
3249 		m->m_pkthdr.len = totlen;
3250 #endif
3251 	return;
3252 }
3253 #endif /* (_KERNEL) && ( ((BSD < 199103) && !MENTAT) || __sgi) */
3254 
3255 
3256 /* ------------------------------------------------------------------------ */
3257 /* Function:    fr_findgroup                                                */
3258 /* Returns:     frgroup_t * - NULL = group not found, else pointer to group */
3259 /* Parameters:  group(I) - group name to search for                         */
3260 /*              unit(I)  - device to which this group belongs               */
3261 /*              set(I)   - which set of rules (inactive/inactive) this is   */
3262 /*              fgpp(O)  - pointer to place to store pointer to the pointer */
3263 /*                         to where to add the next (last) group or where   */
3264 /*                         to delete group from.                            */
3265 /*                                                                          */
3266 /* Search amongst the defined groups for a particular group number.         */
3267 /* ------------------------------------------------------------------------ */
3268 frgroup_t *fr_findgroup(group, unit, set, fgpp, ifs)
3269 char *group;
3270 minor_t unit;
3271 int set;
3272 frgroup_t ***fgpp;
3273 ipf_stack_t *ifs;
3274 {
3275 	frgroup_t *fg, **fgp;
3276 
3277 	/*
3278 	 * Which list of groups to search in is dependent on which list of
3279 	 * rules are being operated on.
3280 	 */
3281 	fgp = &ifs->ifs_ipfgroups[unit][set];
3282 
3283 	while ((fg = *fgp) != NULL) {
3284 		if (strncmp(group, fg->fg_name, FR_GROUPLEN) == 0)
3285 			break;
3286 		else
3287 			fgp = &fg->fg_next;
3288 	}
3289 	if (fgpp != NULL)
3290 		*fgpp = fgp;
3291 	return fg;
3292 }
3293 
3294 
3295 /* ------------------------------------------------------------------------ */
3296 /* Function:    fr_addgroup                                                 */
3297 /* Returns:     frgroup_t * - NULL == did not create group,                 */
3298 /*                            != NULL == pointer to the group               */
3299 /* Parameters:  num(I)   - group number to add                              */
3300 /*              head(I)  - rule pointer that is using this as the head      */
3301 /*              flags(I) - rule flags which describe the type of rule it is */
3302 /*              unit(I)  - device to which this group will belong to        */
3303 /*              set(I)   - which set of rules (inactive/inactive) this is   */
3304 /* Write Locks: ipf_mutex                                                   */
3305 /*                                                                          */
3306 /* Add a new group head, or if it already exists, increase the reference    */
3307 /* count to it.                                                             */
3308 /* ------------------------------------------------------------------------ */
3309 frgroup_t *fr_addgroup(group, head, flags, unit, set, ifs)
3310 char *group;
3311 void *head;
3312 u_32_t flags;
3313 minor_t unit;
3314 int set;
3315 ipf_stack_t *ifs;
3316 {
3317 	frgroup_t *fg, **fgp;
3318 	u_32_t gflags;
3319 
3320 	if (group == NULL)
3321 		return NULL;
3322 
3323 	if (unit == IPL_LOGIPF && *group == '\0')
3324 		return NULL;
3325 
3326 	fgp = NULL;
3327 	gflags = flags & FR_INOUT;
3328 
3329 	fg = fr_findgroup(group, unit, set, &fgp, ifs);
3330 	if (fg != NULL) {
3331 		if (fg->fg_flags == 0)
3332 			fg->fg_flags = gflags;
3333 		else if (gflags != fg->fg_flags)
3334 			return NULL;
3335 		fg->fg_ref++;
3336 		return fg;
3337 	}
3338 	KMALLOC(fg, frgroup_t *);
3339 	if (fg != NULL) {
3340 		fg->fg_head = head;
3341 		fg->fg_start = NULL;
3342 		fg->fg_next = *fgp;
3343 		bcopy(group, fg->fg_name, FR_GROUPLEN);
3344 		fg->fg_flags = gflags;
3345 		fg->fg_ref = 1;
3346 		*fgp = fg;
3347 	}
3348 	return fg;
3349 }
3350 
3351 
3352 /* ------------------------------------------------------------------------ */
3353 /* Function:    fr_delgroup                                                 */
3354 /* Returns:     Nil                                                         */
3355 /* Parameters:  group(I) - group name to delete                             */
3356 /*              unit(I)  - device to which this group belongs               */
3357 /*              set(I)   - which set of rules (inactive/inactive) this is   */
3358 /* Write Locks: ipf_mutex                                                   */
3359 /*                                                                          */
3360 /* Attempt to delete a group head.                                          */
3361 /* Only do this when its reference count reaches 0.                         */
3362 /* ------------------------------------------------------------------------ */
3363 void fr_delgroup(group, unit, set, ifs)
3364 char *group;
3365 minor_t unit;
3366 int set;
3367 ipf_stack_t *ifs;
3368 {
3369 	frgroup_t *fg, **fgp;
3370 
3371 	fg = fr_findgroup(group, unit, set, &fgp, ifs);
3372 	if (fg == NULL)
3373 		return;
3374 
3375 	fg->fg_ref--;
3376 	if (fg->fg_ref == 0) {
3377 		*fgp = fg->fg_next;
3378 		KFREE(fg);
3379 	}
3380 }
3381 
3382 
3383 /* ------------------------------------------------------------------------ */
3384 /* Function:    fr_getrulen                                                 */
3385 /* Returns:     frentry_t * - NULL == not found, else pointer to rule n     */
3386 /* Parameters:  unit(I)  - device for which to count the rule's number      */
3387 /*              flags(I) - which set of rules to find the rule in           */
3388 /*              group(I) - group name                                       */
3389 /*              n(I)     - rule number to find                              */
3390 /*                                                                          */
3391 /* Find rule # n in group # g and return a pointer to it.  Return NULl if   */
3392 /* group # g doesn't exist or there are less than n rules in the group.     */
3393 /* ------------------------------------------------------------------------ */
3394 frentry_t *fr_getrulen(unit, group, n, ifs)
3395 int unit;
3396 char *group;
3397 u_32_t n;
3398 ipf_stack_t *ifs;
3399 {
3400 	frentry_t *fr;
3401 	frgroup_t *fg;
3402 
3403 	fg = fr_findgroup(group, unit, ifs->ifs_fr_active, NULL, ifs);
3404 	if (fg == NULL)
3405 		return NULL;
3406 	for (fr = fg->fg_head; fr && n; fr = fr->fr_next, n--)
3407 		;
3408 	if (n != 0)
3409 		return NULL;
3410 	return fr;
3411 }
3412 
3413 
3414 /* ------------------------------------------------------------------------ */
3415 /* Function:    fr_rulen                                                    */
3416 /* Returns:     int - >= 0 - rule number, -1 == search failed               */
3417 /* Parameters:  unit(I) - device for which to count the rule's number       */
3418 /*              fr(I)   - pointer to rule to match                          */
3419 /*                                                                          */
3420 /* Return the number for a rule on a specific filtering device.             */
3421 /* ------------------------------------------------------------------------ */
3422 int fr_rulen(unit, fr, ifs)
3423 int unit;
3424 frentry_t *fr;
3425 ipf_stack_t *ifs;
3426 {
3427 	frentry_t *fh;
3428 	frgroup_t *fg;
3429 	u_32_t n = 0;
3430 
3431 	if (fr == NULL)
3432 		return -1;
3433 	fg = fr_findgroup(fr->fr_group, unit, ifs->ifs_fr_active, NULL, ifs);
3434 	if (fg == NULL)
3435 		return -1;
3436 	for (fh = fg->fg_head; fh; n++, fh = fh->fr_next)
3437 		if (fh == fr)
3438 			break;
3439 	if (fh == NULL)
3440 		return -1;
3441 	return n;
3442 }
3443 
3444 
3445 /* ------------------------------------------------------------------------ */
3446 /* Function:    frflushlist                                                 */
3447 /* Returns:     int - >= 0 - number of flushed rules                        */
3448 /* Parameters:  set(I)   - which set of rules (inactive/inactive) this is   */
3449 /*              unit(I)  - device for which to flush rules                  */
3450 /*              flags(I) - which set of rules to flush                      */
3451 /*              nfreedp(O) - pointer to int where flush count is stored     */
3452 /*              listp(I)   - pointer to list to flush pointer               */
3453 /* Write Locks: ipf_mutex                                                   */
3454 /*                                                                          */
3455 /* Recursively flush rules from the list, descending groups as they are     */
3456 /* encountered.  if a rule is the head of a group and it has lost all its   */
3457 /* group members, then also delete the group reference.  nfreedp is needed  */
3458 /* to store the accumulating count of rules removed, whereas the returned   */
3459 /* value is just the number removed from the current list.  The latter is   */
3460 /* needed to correctly adjust reference counts on rules that define groups. */
3461 /*                                                                          */
3462 /* NOTE: Rules not loaded from user space cannot be flushed.                */
3463 /* ------------------------------------------------------------------------ */
3464 static int frflushlist(set, unit, nfreedp, listp, ifs)
3465 int set;
3466 minor_t unit;
3467 int *nfreedp;
3468 frentry_t **listp;
3469 ipf_stack_t *ifs;
3470 {
3471 	int freed = 0;
3472 	frentry_t *fp;
3473 
3474 	while ((fp = *listp) != NULL) {
3475 		if ((fp->fr_type & FR_T_BUILTIN) ||
3476 		    !(fp->fr_flags & FR_COPIED)) {
3477 			listp = &fp->fr_next;
3478 			continue;
3479 		}
3480 		*listp = fp->fr_next;
3481 		if (fp->fr_grp != NULL) {
3482 			(void) frflushlist(set, unit, nfreedp, fp->fr_grp, ifs);
3483 		}
3484 
3485 		if (fp->fr_grhead != NULL) {
3486 			fr_delgroup(fp->fr_grhead, unit, set, ifs);
3487 			*fp->fr_grhead = '\0';
3488 		}
3489 
3490 		ASSERT(fp->fr_ref > 0);
3491 		fp->fr_next = NULL;
3492 		if (fr_derefrule(&fp, ifs) == 0)
3493 			freed++;
3494 	}
3495 	*nfreedp += freed;
3496 	return freed;
3497 }
3498 
3499 
3500 /* ------------------------------------------------------------------------ */
3501 /* Function:    frflush                                                     */
3502 /* Returns:     int - >= 0 - number of flushed rules                        */
3503 /* Parameters:  unit(I)  - device for which to flush rules                  */
3504 /*              flags(I) - which set of rules to flush                      */
3505 /*                                                                          */
3506 /* Calls flushlist() for all filter rules (accounting, firewall - both IPv4 */
3507 /* and IPv6) as defined by the value of flags.                              */
3508 /* ------------------------------------------------------------------------ */
3509 int frflush(unit, proto, flags, ifs)
3510 minor_t unit;
3511 int proto, flags;
3512 ipf_stack_t *ifs;
3513 {
3514 	int flushed = 0, set;
3515 
3516 	WRITE_ENTER(&ifs->ifs_ipf_mutex);
3517 
3518 	set = ifs->ifs_fr_active;
3519 	if ((flags & FR_INACTIVE) == FR_INACTIVE)
3520 		set = 1 - set;
3521 
3522 	if (flags & FR_OUTQUE) {
3523 		if (proto == 0 || proto == 6) {
3524 			(void) frflushlist(set, unit,
3525 			    &flushed, &ifs->ifs_ipfilter6[1][set], ifs);
3526 			(void) frflushlist(set, unit,
3527 			    &flushed, &ifs->ifs_ipacct6[1][set], ifs);
3528 		}
3529 		if (proto == 0 || proto == 4) {
3530 			(void) frflushlist(set, unit,
3531 			    &flushed, &ifs->ifs_ipfilter[1][set], ifs);
3532 			(void) frflushlist(set, unit,
3533 			    &flushed, &ifs->ifs_ipacct[1][set], ifs);
3534 		}
3535 	}
3536 	if (flags & FR_INQUE) {
3537 		if (proto == 0 || proto == 6) {
3538 			(void) frflushlist(set, unit,
3539 			    &flushed, &ifs->ifs_ipfilter6[0][set], ifs);
3540 			(void) frflushlist(set, unit,
3541 			    &flushed, &ifs->ifs_ipacct6[0][set], ifs);
3542 		}
3543 		if (proto == 0 || proto == 4) {
3544 			(void) frflushlist(set, unit,
3545 			    &flushed, &ifs->ifs_ipfilter[0][set], ifs);
3546 			(void) frflushlist(set, unit,
3547 			    &flushed, &ifs->ifs_ipacct[0][set], ifs);
3548 		}
3549 	}
3550 	RWLOCK_EXIT(&ifs->ifs_ipf_mutex);
3551 
3552 	if (unit == IPL_LOGIPF) {
3553 		int tmp;
3554 
3555 		tmp = frflush(IPL_LOGCOUNT, proto, flags, ifs);
3556 		if (tmp >= 0)
3557 			flushed += tmp;
3558 	}
3559 	return flushed;
3560 }
3561 
3562 
3563 /* ------------------------------------------------------------------------ */
3564 /* Function:    memstr                                                      */
3565 /* Returns:     char *  - NULL if failed, != NULL pointer to matching bytes */
3566 /* Parameters:  src(I)  - pointer to byte sequence to match                 */
3567 /*              dst(I)  - pointer to byte sequence to search                */
3568 /*              slen(I) - match length                                      */
3569 /*              dlen(I) - length available to search in                     */
3570 /*                                                                          */
3571 /* Search dst for a sequence of bytes matching those at src and extend for  */
3572 /* slen bytes.                                                              */
3573 /* ------------------------------------------------------------------------ */
3574 char *memstr(src, dst, slen, dlen)
3575 char *src, *dst;
3576 int slen, dlen;
3577 {
3578 	char *s = NULL;
3579 
3580 	while (dlen >= slen) {
3581 		if (bcmp(src, dst, slen) == 0) {
3582 			s = dst;
3583 			break;
3584 		}
3585 		dst++;
3586 		dlen--;
3587 	}
3588 	return s;
3589 }
3590 /* ------------------------------------------------------------------------ */
3591 /* Function:    fr_fixskip                                                  */
3592 /* Returns:     Nil                                                         */
3593 /* Parameters:  listp(IO)    - pointer to start of list with skip rule      */
3594 /*              rp(I)        - rule added/removed with skip in it.          */
3595 /*              addremove(I) - adjustment (-1/+1) to make to skip count,    */
3596 /*                             depending on whether a rule was just added   */
3597 /*                             or removed.                                  */
3598 /*                                                                          */
3599 /* Adjust all the rules in a list which would have skip'd past the position */
3600 /* where we are inserting to skip to the right place given the change.      */
3601 /* ------------------------------------------------------------------------ */
3602 void fr_fixskip(listp, rp, addremove)
3603 frentry_t **listp, *rp;
3604 int addremove;
3605 {
3606 	int rules, rn;
3607 	frentry_t *fp;
3608 
3609 	rules = 0;
3610 	for (fp = *listp; (fp != NULL) && (fp != rp); fp = fp->fr_next)
3611 		rules++;
3612 
3613 	if (!fp)
3614 		return;
3615 
3616 	for (rn = 0, fp = *listp; fp && (fp != rp); fp = fp->fr_next, rn++)
3617 		if (FR_ISSKIP(fp->fr_flags) && (rn + fp->fr_arg >= rules))
3618 			fp->fr_arg += addremove;
3619 }
3620 
3621 
3622 #ifdef	_KERNEL
3623 /* ------------------------------------------------------------------------ */
3624 /* Function:    count4bits                                                  */
3625 /* Returns:     int - >= 0 - number of consecutive bits in input            */
3626 /* Parameters:  ip(I) - 32bit IP address                                    */
3627 /*                                                                          */
3628 /* IPv4 ONLY                                                                */
3629 /* count consecutive 1's in bit mask.  If the mask generated by counting    */
3630 /* consecutive 1's is different to that passed, return -1, else return #    */
3631 /* of bits.                                                                 */
3632 /* ------------------------------------------------------------------------ */
3633 int	count4bits(ip)
3634 u_32_t	ip;
3635 {
3636 	u_32_t	ipn;
3637 	int	cnt = 0, i, j;
3638 
3639 	ip = ipn = ntohl(ip);
3640 	for (i = 32; i; i--, ipn *= 2)
3641 		if (ipn & 0x80000000)
3642 			cnt++;
3643 		else
3644 			break;
3645 	ipn = 0;
3646 	for (i = 32, j = cnt; i; i--, j--) {
3647 		ipn *= 2;
3648 		if (j > 0)
3649 			ipn++;
3650 	}
3651 	if (ipn == ip)
3652 		return cnt;
3653 	return -1;
3654 }
3655 
3656 
3657 #ifdef USE_INET6
3658 /* ------------------------------------------------------------------------ */
3659 /* Function:    count6bits                                                  */
3660 /* Returns:     int - >= 0 - number of consecutive bits in input            */
3661 /* Parameters:  msk(I) - pointer to start of IPv6 bitmask                   */
3662 /*                                                                          */
3663 /* IPv6 ONLY                                                                */
3664 /* count consecutive 1's in bit mask.                                       */
3665 /* ------------------------------------------------------------------------ */
3666 int count6bits(msk)
3667 u_32_t *msk;
3668 {
3669 	int i = 0, k;
3670 	u_32_t j;
3671 
3672 	for (k = 3; k >= 0; k--)
3673 		if (msk[k] == 0xffffffff)
3674 			i += 32;
3675 		else {
3676 			for (j = msk[k]; j; j <<= 1)
3677 				if (j & 0x80000000)
3678 					i++;
3679 		}
3680 	return i;
3681 }
3682 # endif
3683 #endif /* _KERNEL */
3684 
3685 
3686 /* ------------------------------------------------------------------------ */
3687 /* Function:    fr_ifsync                                                   */
3688 /* Returns:     void *    - new interface identifier                        */
3689 /* Parameters:  action(I)  - type of synchronisation to do                  */
3690 /*              v(I)       - IP version being sync'd (v4 or v6)             */
3691 /*              newifp(I)  - interface identifier being introduced/removed  */
3692 /*              oldifp(I)  - interface identifier in a filter rule          */
3693 /*              newname(I) - name associated with newifp interface          */
3694 /*              oldname(I) - name associated with oldifp interface          */
3695 /*		ifs       - pointer to IPF stack instance		    */
3696 /*                                                                          */
3697 /* This function returns what the new value for "oldifp" should be for its  */
3698 /* caller.  In some cases it will not change, in some it will.              */
3699 /* action == IPFSYNC_RESYNC                                                 */
3700 /*   a new value for oldifp will always be looked up, according to oldname, */
3701 /*   the values of newname and newifp are ignored.                          */
3702 /* action == IPFSYNC_NEWIFP                                                 */
3703 /*   if oldname matches newname then we are doing a sync for the matching   */
3704 /*   interface, so we return newifp to be used in place of oldifp.  If the  */
3705 /*   the names don't match, just return oldifp.                             */
3706 /* action == IPFSYNC_OLDIFP                                                 */
3707 /*   if oldifp matches newifp then we are are doing a sync to remove any    */
3708 /*   references to oldifp, so we return "-1".                               */
3709 /* -----								    */
3710 /* NOTE:								    */
3711 /* This function processes NIC event from PF_HOOKS. The action parameter    */
3712 /* is set in ipf_nic_event_v4()/ipf_nic_event_v6() function. There is	    */
3713 /* one single switch statement() in ipf_nic_event_vx() function, which	    */
3714 /* translates the HOOK event type to action parameter passed to fr_ifsync.  */
3715 /* The translation table looks as follows:				    */
3716 /*	event		| action					    */
3717 /*	----------------+-------------					    */
3718 /*	NE_PLUMB	| IPFSYNC_NEWIFP				    */
3719 /*	NE_UNPLUMB	| IPFSYNC_OLDIFP				    */
3720 /*    NE_ADDRESS_CHANGE	| IPFSYNC_RESYNC				    */
3721 /*									    */
3722 /* The oldname and oldifp parameters are taken from IPF entry (rule, state  */
3723 /* table entry, NAT table entry, fragment ...). The newname and newifp	    */
3724 /* parameters come from hook event data, parameters are taken from event    */
3725 /* in ipf_nic_event_vx() functions. Any time NIC changes, the IPF is	    */
3726 /* notified by hook function.						    */
3727 /*									    */
3728 /* We get NE_UNPLUMB event from PF_HOOKS even if someone coincidently tries */
3729 /* to plumb the interface, which is already plumbed. In such case we always */
3730 /* get the event from PF_HOOKS as follows:				    */
3731 /*	event:	NE_PLUMB						    */
3732 /*	NIC:	0x0							    */
3733 /* ------------------------------------------------------------------------ */
3734 static void *fr_ifsync(action, v, newname, oldname, newifp, oldifp, ifs)
3735 int action, v;
3736 char *newname, *oldname;
3737 void *newifp, *oldifp;
3738 ipf_stack_t *ifs;
3739 {
3740 	void *rval = oldifp;
3741 
3742 	switch (action)
3743 	{
3744 	case IPFSYNC_RESYNC :
3745 		if (oldname[0] != '\0') {
3746 			rval = fr_resolvenic(oldname, v, ifs);
3747 		}
3748 		break;
3749 	case IPFSYNC_NEWIFP :
3750 		if (!strncmp(newname, oldname, LIFNAMSIZ))
3751 			rval = newifp;
3752 		break;
3753 	case IPFSYNC_OLDIFP :
3754 		/*
3755 		 * If interface gets unplumbed it must be invalidated, which
3756 		 * means set all existing references to the interface to -1.
3757 		 * We don't want to invalidate references for wildcard
3758 		 * (unbound) rules (entries).
3759 		 */
3760 		if (newifp == oldifp)
3761 			rval = (oldifp) ? (void *)-1 : NULL;
3762 		break;
3763 	}
3764 
3765 	return rval;
3766 }
3767 
3768 
3769 /* ------------------------------------------------------------------------ */
3770 /* Function:    frsynclist                                                  */
3771 /* Returns:     void                                                        */
3772 /* Parameters:  action(I) - type of synchronisation to do                   */
3773 /*              v(I)      - IP version being sync'd (v4 or v6)              */
3774 /*              ifp(I)    - interface identifier associated with action     */
3775 /*              ifname(I) - name associated with ifp parameter              */
3776 /*              fr(I)     - pointer to filter rule                          */
3777 /*		ifs       - pointer to IPF stack instance		    */
3778 /* Write Locks: ipf_mutex                                                   */
3779 /*                                                                          */
3780 /* Walk through a list of filter rules and resolve any interface names into */
3781 /* pointers.  Where dynamic addresses are used, also update the IP address  */
3782 /* used in the rule.  The interface pointer is used to limit the lookups to */
3783 /* a specific set of matching names if it is non-NULL.                      */
3784 /* ------------------------------------------------------------------------ */
3785 static void frsynclist(action, v, ifp, ifname, fr, ifs)
3786 int action, v;
3787 void *ifp;
3788 char *ifname;
3789 frentry_t *fr;
3790 ipf_stack_t *ifs;
3791 {
3792 	frdest_t *fdp;
3793 	int rv, i;
3794 
3795 	for (; fr; fr = fr->fr_next) {
3796 		rv = fr->fr_v;
3797 		if (v != 0 && v != rv)
3798 			continue;
3799 
3800 		/*
3801 		 * Lookup all the interface names that are part of the rule.
3802 		 */
3803 		for (i = 0; i < 4; i++) {
3804 			fr->fr_ifas[i] = fr_ifsync(action, rv, ifname,
3805 						   fr->fr_ifnames[i],
3806 						   ifp, fr->fr_ifas[i],
3807 						   ifs);
3808 		}
3809 
3810 		fdp = &fr->fr_tifs[0];
3811 		fdp->fd_ifp = fr_ifsync(action, rv, ifname, fdp->fd_ifname,
3812 					   ifp, fdp->fd_ifp, ifs);
3813 
3814 		fdp = &fr->fr_tifs[1];
3815 		fdp->fd_ifp = fr_ifsync(action, rv, ifname, fdp->fd_ifname,
3816 					   ifp, fdp->fd_ifp, ifs);
3817 
3818 		fdp = &fr->fr_dif;
3819 		fdp->fd_ifp = fr_ifsync(action, rv, ifname, fdp->fd_ifname,
3820 					   ifp, fdp->fd_ifp, ifs);
3821 
3822 		if (action != IPFSYNC_RESYNC)
3823 			continue;
3824 
3825 		if (fr->fr_type == FR_T_IPF) {
3826 			if (fr->fr_satype != FRI_NORMAL &&
3827 			    fr->fr_satype != FRI_LOOKUP) {
3828 				(void)fr_ifpaddr(rv, fr->fr_satype,
3829 						 fr->fr_ifas[fr->fr_sifpidx],
3830 						 &fr->fr_src, &fr->fr_smsk,
3831 						 ifs);
3832 			}
3833 			if (fr->fr_datype != FRI_NORMAL &&
3834 			    fr->fr_datype != FRI_LOOKUP) {
3835 				(void)fr_ifpaddr(rv, fr->fr_datype,
3836 						 fr->fr_ifas[fr->fr_difpidx],
3837 						 &fr->fr_dst, &fr->fr_dmsk,
3838 						 ifs);
3839 			}
3840 		}
3841 
3842 #ifdef	IPFILTER_LOOKUP
3843 		if (fr->fr_type == FR_T_IPF && fr->fr_satype == FRI_LOOKUP &&
3844 		    fr->fr_srcptr == NULL) {
3845 			fr->fr_srcptr = fr_resolvelookup(fr->fr_srctype,
3846 							 fr->fr_srcnum,
3847 							 &fr->fr_srcfunc, ifs);
3848 		}
3849 		if (fr->fr_type == FR_T_IPF && fr->fr_datype == FRI_LOOKUP &&
3850 		    fr->fr_dstptr == NULL) {
3851 			fr->fr_dstptr = fr_resolvelookup(fr->fr_dsttype,
3852 							 fr->fr_dstnum,
3853 							 &fr->fr_dstfunc, ifs);
3854 		}
3855 #endif
3856 	}
3857 }
3858 
3859 
3860 #ifdef	_KERNEL
3861 /* ------------------------------------------------------------------------ */
3862 /* Function:    frsync                                                      */
3863 /* Returns:     void                                                        */
3864 /* Parameters:  action(I) - type of synchronisation to do                   */
3865 /*              v(I)      - IP version being sync'd (v4 or v6)              */
3866 /*              ifp(I)    - interface identifier associated with action     */
3867 /*              name(I)   - name associated with ifp parameter              */
3868 /*                                                                          */
3869 /* frsync() is called when we suspect that the interface list or            */
3870 /* information about interfaces (like IP#) has changed.  Go through all     */
3871 /* filter rules, NAT entries and the state table and check if anything      */
3872 /* needs to be changed/updated.                                             */
3873 /* With the filtering hooks added to Solaris, we needed to change the manner*/
3874 /* in which this was done to support three different types of sync:         */
3875 /* - complete resync of all interface name/identifiers                      */
3876 /* - new interface being announced with its name and identifier             */
3877 /* - interface removal being announced by only its identifier               */
3878 /* ------------------------------------------------------------------------ */
3879 void frsync(action, v, ifp, name, ifs)
3880 int action, v;
3881 void *ifp;
3882 char *name;
3883 ipf_stack_t *ifs;
3884 {
3885 	int i;
3886 
3887 	WRITE_ENTER(&ifs->ifs_ipf_mutex);
3888 	frsynclist(action, v, ifp, name, ifs->ifs_ipacct[0][ifs->ifs_fr_active], ifs);
3889 	frsynclist(action, v, ifp, name, ifs->ifs_ipacct[1][ifs->ifs_fr_active], ifs);
3890 	frsynclist(action, v, ifp, name, ifs->ifs_ipfilter[0][ifs->ifs_fr_active], ifs);
3891 	frsynclist(action, v, ifp, name, ifs->ifs_ipfilter[1][ifs->ifs_fr_active], ifs);
3892 	frsynclist(action, v, ifp, name, ifs->ifs_ipacct6[0][ifs->ifs_fr_active], ifs);
3893 	frsynclist(action, v, ifp, name, ifs->ifs_ipacct6[1][ifs->ifs_fr_active], ifs);
3894 	frsynclist(action, v, ifp, name, ifs->ifs_ipfilter6[0][ifs->ifs_fr_active], ifs);
3895 	frsynclist(action, v, ifp, name, ifs->ifs_ipfilter6[1][ifs->ifs_fr_active], ifs);
3896 
3897 	for (i = 0; i < IPL_LOGSIZE; i++) {
3898 		frgroup_t *g;
3899 
3900 		for (g = ifs->ifs_ipfgroups[i][0]; g != NULL; g = g->fg_next)
3901 			frsynclist(action, v, ifp, name, g->fg_start, ifs);
3902 		for (g = ifs->ifs_ipfgroups[i][1]; g != NULL; g = g->fg_next)
3903 			frsynclist(action, v, ifp, name, g->fg_start, ifs);
3904 	}
3905 	RWLOCK_EXIT(&ifs->ifs_ipf_mutex);
3906 }
3907 
3908 
3909 /*
3910  * In the functions below, bcopy() is called because the pointer being
3911  * copied _from_ in this instance is a pointer to a char buf (which could
3912  * end up being unaligned) and on the kernel's local stack.
3913  */
3914 /* ------------------------------------------------------------------------ */
3915 /* Function:    copyinptr                                                   */
3916 /* Returns:     int - 0 = success, else failure                             */
3917 /* Parameters:  src(I)  - pointer to the source address                     */
3918 /*              dst(I)  - destination address                               */
3919 /*              size(I) - number of bytes to copy                           */
3920 /*                                                                          */
3921 /* Copy a block of data in from user space, given a pointer to the pointer  */
3922 /* to start copying from (src) and a pointer to where to store it (dst).    */
3923 /* NB: src - pointer to user space pointer, dst - kernel space pointer      */
3924 /* ------------------------------------------------------------------------ */
3925 int copyinptr(src, dst, size)
3926 void *src, *dst;
3927 size_t size;
3928 {
3929 	caddr_t ca;
3930 	int err;
3931 
3932 # if SOLARIS
3933 	err = COPYIN(src, (caddr_t)&ca, sizeof(ca));
3934 	if (err != 0)
3935 		return err;
3936 # else
3937 	bcopy(src, (caddr_t)&ca, sizeof(ca));
3938 # endif
3939 	err = COPYIN(ca, dst, size);
3940 	return err;
3941 }
3942 
3943 
3944 /* ------------------------------------------------------------------------ */
3945 /* Function:    copyoutptr                                                  */
3946 /* Returns:     int - 0 = success, else failure                             */
3947 /* Parameters:  src(I)  - pointer to the source address                     */
3948 /*              dst(I)  - destination address                               */
3949 /*              size(I) - number of bytes to copy                           */
3950 /*                                                                          */
3951 /* Copy a block of data out to user space, given a pointer to the pointer   */
3952 /* to start copying from (src) and a pointer to where to store it (dst).    */
3953 /* NB: src - kernel space pointer, dst - pointer to user space pointer.     */
3954 /* ------------------------------------------------------------------------ */
3955 int copyoutptr(src, dst, size)
3956 void *src, *dst;
3957 size_t size;
3958 {
3959 	caddr_t ca;
3960 	int err;
3961 
3962 # if SOLARIS
3963 	err = COPYIN(dst, (caddr_t)&ca, sizeof(ca));
3964 	if (err != 0)
3965 		return err;
3966 # else
3967 	bcopy(dst, (caddr_t)&ca, sizeof(ca));
3968 # endif
3969 	err = COPYOUT(src, ca, size);
3970 	return err;
3971 }
3972 #endif
3973 
3974 
3975 /* ------------------------------------------------------------------------ */
3976 /* Function:    fr_lock                                                     */
3977 /* Returns:	int - 0 = success, else error				    */
3978 /* Parameters:  data(I)  - pointer to lock value to set                     */
3979 /*              lockp(O) - pointer to location to store old lock value      */
3980 /*                                                                          */
3981 /* Get the new value for the lock integer, set it and return the old value  */
3982 /* in *lockp.                                                               */
3983 /* ------------------------------------------------------------------------ */
3984 int fr_lock(data, lockp)
3985 caddr_t data;
3986 int *lockp;
3987 {
3988 	int arg, err;
3989 
3990 	err = BCOPYIN(data, (caddr_t)&arg, sizeof(arg));
3991 	if (err != 0)
3992 		return (EFAULT);
3993 	err = BCOPYOUT((caddr_t)lockp, data, sizeof(*lockp));
3994 	if (err != 0)
3995 		return (EFAULT);
3996 	*lockp = arg;
3997 	return (0);
3998 }
3999 
4000 
4001 /* ------------------------------------------------------------------------ */
4002 /* Function:    fr_getstat                                                  */
4003 /* Returns:     Nil                                                         */
4004 /* Parameters:  fiop(I)  - pointer to ipfilter stats structure              */
4005 /*                                                                          */
4006 /* Stores a copy of current pointers, counters, etc, in the friostat        */
4007 /* structure.                                                               */
4008 /* ------------------------------------------------------------------------ */
4009 void fr_getstat(fiop, ifs)
4010 friostat_t *fiop;
4011 ipf_stack_t *ifs;
4012 {
4013 	int i, j;
4014 
4015 	bcopy((char *)&ifs->ifs_frstats, (char *)fiop->f_st,
4016 	    sizeof(filterstats_t) * 2);
4017 	fiop->f_locks[IPL_LOGSTATE] = ifs->ifs_fr_state_lock;
4018 	fiop->f_locks[IPL_LOGNAT] = ifs->ifs_fr_nat_lock;
4019 	fiop->f_locks[IPL_LOGIPF] = ifs->ifs_fr_frag_lock;
4020 	fiop->f_locks[IPL_LOGAUTH] = ifs->ifs_fr_auth_lock;
4021 
4022 	for (i = 0; i < 2; i++)
4023 		for (j = 0; j < 2; j++) {
4024 			fiop->f_ipf[i][j] = ifs->ifs_ipfilter[i][j];
4025 			fiop->f_acct[i][j] = ifs->ifs_ipacct[i][j];
4026 			fiop->f_ipf6[i][j] = ifs->ifs_ipfilter6[i][j];
4027 			fiop->f_acct6[i][j] = ifs->ifs_ipacct6[i][j];
4028 		}
4029 
4030 	fiop->f_ticks = ifs->ifs_fr_ticks;
4031 	fiop->f_active = ifs->ifs_fr_active;
4032 	fiop->f_froute[0] = ifs->ifs_fr_frouteok[0];
4033 	fiop->f_froute[1] = ifs->ifs_fr_frouteok[1];
4034 
4035 	fiop->f_running = ifs->ifs_fr_running;
4036 	for (i = 0; i < IPL_LOGSIZE; i++) {
4037 		fiop->f_groups[i][0] = ifs->ifs_ipfgroups[i][0];
4038 		fiop->f_groups[i][1] = ifs->ifs_ipfgroups[i][1];
4039 	}
4040 #ifdef  IPFILTER_LOG
4041 	fiop->f_logging = 1;
4042 #else
4043 	fiop->f_logging = 0;
4044 #endif
4045 	fiop->f_defpass = ifs->ifs_fr_pass;
4046 	fiop->f_features = fr_features;
4047 	(void) strncpy(fiop->f_version, ipfilter_version,
4048 		       sizeof(fiop->f_version));
4049 }
4050 
4051 
4052 #ifdef	USE_INET6
4053 int icmptoicmp6types[ICMP_MAXTYPE+1] = {
4054 	ICMP6_ECHO_REPLY,	/* 0: ICMP_ECHOREPLY */
4055 	-1,			/* 1: UNUSED */
4056 	-1,			/* 2: UNUSED */
4057 	ICMP6_DST_UNREACH,	/* 3: ICMP_UNREACH */
4058 	-1,			/* 4: ICMP_SOURCEQUENCH */
4059 	ND_REDIRECT,		/* 5: ICMP_REDIRECT */
4060 	-1,			/* 6: UNUSED */
4061 	-1,			/* 7: UNUSED */
4062 	ICMP6_ECHO_REQUEST,	/* 8: ICMP_ECHO */
4063 	-1,			/* 9: UNUSED */
4064 	-1,			/* 10: UNUSED */
4065 	ICMP6_TIME_EXCEEDED,	/* 11: ICMP_TIMXCEED */
4066 	ICMP6_PARAM_PROB,	/* 12: ICMP_PARAMPROB */
4067 	-1,			/* 13: ICMP_TSTAMP */
4068 	-1,			/* 14: ICMP_TSTAMPREPLY */
4069 	-1,			/* 15: ICMP_IREQ */
4070 	-1,			/* 16: ICMP_IREQREPLY */
4071 	-1,			/* 17: ICMP_MASKREQ */
4072 	-1,			/* 18: ICMP_MASKREPLY */
4073 };
4074 
4075 
4076 int	icmptoicmp6unreach[ICMP_MAX_UNREACH] = {
4077 	ICMP6_DST_UNREACH_ADDR,		/* 0: ICMP_UNREACH_NET */
4078 	ICMP6_DST_UNREACH_ADDR,		/* 1: ICMP_UNREACH_HOST */
4079 	-1,				/* 2: ICMP_UNREACH_PROTOCOL */
4080 	ICMP6_DST_UNREACH_NOPORT,	/* 3: ICMP_UNREACH_PORT */
4081 	-1,				/* 4: ICMP_UNREACH_NEEDFRAG */
4082 	ICMP6_DST_UNREACH_NOTNEIGHBOR,	/* 5: ICMP_UNREACH_SRCFAIL */
4083 	ICMP6_DST_UNREACH_ADDR,		/* 6: ICMP_UNREACH_NET_UNKNOWN */
4084 	ICMP6_DST_UNREACH_ADDR,		/* 7: ICMP_UNREACH_HOST_UNKNOWN */
4085 	-1,				/* 8: ICMP_UNREACH_ISOLATED */
4086 	ICMP6_DST_UNREACH_ADMIN,	/* 9: ICMP_UNREACH_NET_PROHIB */
4087 	ICMP6_DST_UNREACH_ADMIN,	/* 10: ICMP_UNREACH_HOST_PROHIB */
4088 	-1,				/* 11: ICMP_UNREACH_TOSNET */
4089 	-1,				/* 12: ICMP_UNREACH_TOSHOST */
4090 	ICMP6_DST_UNREACH_ADMIN,	/* 13: ICMP_UNREACH_ADMIN_PROHIBIT */
4091 };
4092 int	icmpreplytype6[ICMP6_MAXTYPE + 1];
4093 #endif
4094 
4095 int	icmpreplytype4[ICMP_MAXTYPE + 1];
4096 
4097 
4098 /* ------------------------------------------------------------------------ */
4099 /* Function:    fr_matchicmpqueryreply                                      */
4100 /* Returns:     int - 1 if "icmp" is a valid reply to "ic" else 0.          */
4101 /* Parameters:  v(I)    - IP protocol version (4 or 6)                      */
4102 /*              ic(I)   - ICMP information                                  */
4103 /*              icmp(I) - ICMP packet header                                */
4104 /*              rev(I)  - direction (0 = forward/1 = reverse) of packet     */
4105 /*                                                                          */
4106 /* Check if the ICMP packet defined by the header pointed to by icmp is a   */
4107 /* reply to one as described by what's in ic.  If it is a match, return 1,  */
4108 /* else return 0 for no match.                                              */
4109 /* ------------------------------------------------------------------------ */
4110 int fr_matchicmpqueryreply(v, ic, icmp, rev)
4111 int v;
4112 icmpinfo_t *ic;
4113 icmphdr_t *icmp;
4114 int rev;
4115 {
4116 	int ictype;
4117 
4118 	ictype = ic->ici_type;
4119 
4120 	if (v == 4) {
4121 		/*
4122 		 * If we matched its type on the way in, then when going out
4123 		 * it will still be the same type.
4124 		 */
4125 		if ((!rev && (icmp->icmp_type == ictype)) ||
4126 		    (rev && (icmpreplytype4[ictype] == icmp->icmp_type))) {
4127 			if (icmp->icmp_type != ICMP_ECHOREPLY)
4128 				return 1;
4129 			if (icmp->icmp_id == ic->ici_id)
4130 				return 1;
4131 		}
4132 	}
4133 #ifdef	USE_INET6
4134 	else if (v == 6) {
4135 		if ((!rev && (icmp->icmp_type == ictype)) ||
4136 		    (rev && (icmpreplytype6[ictype] == icmp->icmp_type))) {
4137 			if (icmp->icmp_type != ICMP6_ECHO_REPLY)
4138 				return 1;
4139 			if (icmp->icmp_id == ic->ici_id)
4140 				return 1;
4141 		}
4142 	}
4143 #endif
4144 	return 0;
4145 }
4146 
4147 
4148 #ifdef	IPFILTER_LOOKUP
4149 /* ------------------------------------------------------------------------ */
4150 /* Function:    fr_resolvelookup                                            */
4151 /* Returns:     void * - NULL = failure, else success.                      */
4152 /* Parameters:  type(I)     - type of lookup these parameters are for.      */
4153 /*              number(I)   - table number to use when searching            */
4154 /*              funcptr(IO) - pointer to pointer for storing IP address     */
4155 /*                           searching function.                            */
4156 /*                                                                          */
4157 /* Search for the "table" number passed in amongst those configured for     */
4158 /* that particular type.  If the type is recognised then the function to    */
4159 /* call to do the IP address search will be change, regardless of whether   */
4160 /* or not the "table" number exists.                                        */
4161 /* ------------------------------------------------------------------------ */
4162 static void *fr_resolvelookup(type, number, funcptr, ifs)
4163 u_int type, number;
4164 lookupfunc_t *funcptr;
4165 ipf_stack_t *ifs;
4166 {
4167 	char name[FR_GROUPLEN];
4168 	iphtable_t *iph;
4169 	ip_pool_t *ipo;
4170 	void *ptr;
4171 
4172 #if defined(SNPRINTF) && defined(_KERNEL)
4173 	(void) SNPRINTF(name, sizeof(name), "%u", number);
4174 #else
4175 	(void) sprintf(name, "%u", number);
4176 #endif
4177 
4178 	READ_ENTER(&ifs->ifs_ip_poolrw);
4179 
4180 	switch (type)
4181 	{
4182 	case IPLT_POOL :
4183 # if (defined(__osf__) && defined(_KERNEL))
4184 		ptr = NULL;
4185 		*funcptr = NULL;
4186 # else
4187 		ipo = ip_pool_find(IPL_LOGIPF, name, ifs);
4188 		ptr = ipo;
4189 		if (ipo != NULL) {
4190 			ATOMIC_INC32(ipo->ipo_ref);
4191 		}
4192 		*funcptr = ip_pool_search;
4193 # endif
4194 		break;
4195 	case IPLT_HASH :
4196 		iph = fr_findhtable(IPL_LOGIPF, name, ifs);
4197 		ptr = iph;
4198 		if (iph != NULL) {
4199 			ATOMIC_INC32(iph->iph_ref);
4200 		}
4201 		*funcptr = fr_iphmfindip;
4202 		break;
4203 	default:
4204 		ptr = NULL;
4205 		*funcptr = NULL;
4206 		break;
4207 	}
4208 	RWLOCK_EXIT(&ifs->ifs_ip_poolrw);
4209 
4210 	return ptr;
4211 }
4212 #endif
4213 
4214 
4215 /* ------------------------------------------------------------------------ */
4216 /* Function:    frrequest                                                   */
4217 /* Returns:     int - 0 == success, > 0 == errno value                      */
4218 /* Parameters:  unit(I)     - device for which this is for                  */
4219 /*              req(I)      - ioctl command (SIOC*)                         */
4220 /*              data(I)     - pointr to ioctl data                          */
4221 /*              set(I)      - 1 or 0 (filter set)                           */
4222 /*              makecopy(I) - flag indicating whether data points to a rule */
4223 /*                            in kernel space & hence doesn't need copying. */
4224 /*                                                                          */
4225 /* This function handles all the requests which operate on the list of      */
4226 /* filter rules.  This includes adding, deleting, insertion.  It is also    */
4227 /* responsible for creating groups when a "head" rule is loaded.  Interface */
4228 /* names are resolved here and other sanity checks are made on the content  */
4229 /* of the rule structure being loaded.  If a rule has user defined timeouts */
4230 /* then make sure they are created and initialised before exiting.          */
4231 /* ------------------------------------------------------------------------ */
4232 int frrequest(unit, req, data, set, makecopy, ifs)
4233 int unit;
4234 ioctlcmd_t req;
4235 int set, makecopy;
4236 caddr_t data;
4237 ipf_stack_t *ifs;
4238 {
4239 	frentry_t frd, *fp, *f, **fprev, **ftail;
4240 	int error = 0, in, v;
4241 	void *ptr, *uptr;
4242 	u_int *p, *pp;
4243 	frgroup_t *fg;
4244 	char *group;
4245 
4246 	fg = NULL;
4247 	fp = &frd;
4248 	if (makecopy != 0) {
4249 		error = fr_inobj(data, fp, IPFOBJ_FRENTRY);
4250 		if (error)
4251 			return EFAULT;
4252 		if ((fp->fr_flags & FR_T_BUILTIN) != 0)
4253 			return EINVAL;
4254 		fp->fr_ref = 0;
4255 		fp->fr_flags |= FR_COPIED;
4256 	} else {
4257 		fp = (frentry_t *)data;
4258 		if ((fp->fr_type & FR_T_BUILTIN) == 0)
4259 			return EINVAL;
4260 		fp->fr_flags &= ~FR_COPIED;
4261 	}
4262 
4263 	if (((fp->fr_dsize == 0) && (fp->fr_data != NULL)) ||
4264 	    ((fp->fr_dsize != 0) && (fp->fr_data == NULL)))
4265 		return EINVAL;
4266 
4267 	v = fp->fr_v;
4268 	uptr = fp->fr_data;
4269 
4270 	/*
4271 	 * Only filter rules for IPv4 or IPv6 are accepted.
4272 	 */
4273 	if (v == 4)
4274 		/*EMPTY*/;
4275 #ifdef	USE_INET6
4276 	else if (v == 6)
4277 		/*EMPTY*/;
4278 #endif
4279 	else {
4280 		return EINVAL;
4281 	}
4282 
4283 	/*
4284 	 * If the rule is being loaded from user space, i.e. we had to copy it
4285 	 * into kernel space, then do not trust the function pointer in the
4286 	 * rule.
4287 	 */
4288 	if ((makecopy == 1) && (fp->fr_func != NULL)) {
4289 		if (fr_findfunc(fp->fr_func) == NULL)
4290 			return ESRCH;
4291 		error = fr_funcinit(fp, ifs);
4292 		if (error != 0)
4293 			return error;
4294 	}
4295 
4296 	ptr = NULL;
4297 	/*
4298 	 * Check that the group number does exist and that its use (in/out)
4299 	 * matches what the rule is.
4300 	 */
4301 	if (!strncmp(fp->fr_grhead, "0", FR_GROUPLEN))
4302 		*fp->fr_grhead = '\0';
4303 	group = fp->fr_group;
4304 	if (!strncmp(group, "0", FR_GROUPLEN))
4305 		*group = '\0';
4306 
4307 	if (FR_ISACCOUNT(fp->fr_flags))
4308 		unit = IPL_LOGCOUNT;
4309 
4310 	if ((req != (int)SIOCZRLST) && (*group != '\0')) {
4311 		fg = fr_findgroup(group, unit, set, NULL, ifs);
4312 		if (fg == NULL)
4313 			return ESRCH;
4314 		if (fg->fg_flags == 0)
4315 			fg->fg_flags = fp->fr_flags & FR_INOUT;
4316 		else if (fg->fg_flags != (fp->fr_flags & FR_INOUT))
4317 			return ESRCH;
4318 	}
4319 
4320 	in = (fp->fr_flags & FR_INQUE) ? 0 : 1;
4321 
4322 	/*
4323 	 * Work out which rule list this change is being applied to.
4324 	 */
4325 	ftail = NULL;
4326 	fprev = NULL;
4327 	if (unit == IPL_LOGAUTH)
4328 		fprev = &ifs->ifs_ipauth;
4329 	else if (v == 4) {
4330 		if (FR_ISACCOUNT(fp->fr_flags))
4331 			fprev = &ifs->ifs_ipacct[in][set];
4332 		else if ((fp->fr_flags & (FR_OUTQUE|FR_INQUE)) != 0)
4333 			fprev = &ifs->ifs_ipfilter[in][set];
4334 	} else if (v == 6) {
4335 		if (FR_ISACCOUNT(fp->fr_flags))
4336 			fprev = &ifs->ifs_ipacct6[in][set];
4337 		else if ((fp->fr_flags & (FR_OUTQUE|FR_INQUE)) != 0)
4338 			fprev = &ifs->ifs_ipfilter6[in][set];
4339 	}
4340 	if (fprev == NULL)
4341 		return ESRCH;
4342 
4343 	if (*group != '\0') {
4344 	    if (!fg && !(fg = fr_findgroup(group, unit, set, NULL, ifs)))
4345 			return ESRCH;
4346 		fprev = &fg->fg_start;
4347 	}
4348 
4349 	ftail = fprev;
4350 	for (f = *ftail; (f = *ftail) != NULL; ftail = &f->fr_next) {
4351 		if (fp->fr_collect <= f->fr_collect) {
4352 			ftail = fprev;
4353 			f = NULL;
4354 			break;
4355 		}
4356 		fprev = ftail;
4357 	}
4358 
4359 	/*
4360 	 * Copy in extra data for the rule.
4361 	 */
4362 	if (fp->fr_dsize != 0) {
4363 		if (makecopy != 0) {
4364 			KMALLOCS(ptr, void *, fp->fr_dsize);
4365 			if (!ptr)
4366 				return ENOMEM;
4367 			error = COPYIN(uptr, ptr, fp->fr_dsize);
4368 		} else {
4369 			ptr = uptr;
4370 			error = 0;
4371 		}
4372 		if (error != 0) {
4373 			KFREES(ptr, fp->fr_dsize);
4374 			return ENOMEM;
4375 		}
4376 		fp->fr_data = ptr;
4377 	} else
4378 		fp->fr_data = NULL;
4379 
4380 	/*
4381 	 * Perform per-rule type sanity checks of their members.
4382 	 */
4383 	switch (fp->fr_type & ~FR_T_BUILTIN)
4384 	{
4385 #if defined(IPFILTER_BPF)
4386 	case FR_T_BPFOPC :
4387 		if (fp->fr_dsize == 0)
4388 			return EINVAL;
4389 		if (!bpf_validate(ptr, fp->fr_dsize/sizeof(struct bpf_insn))) {
4390 			if (makecopy && fp->fr_data != NULL) {
4391 				KFREES(fp->fr_data, fp->fr_dsize);
4392 			}
4393 			return EINVAL;
4394 		}
4395 		break;
4396 #endif
4397 	case FR_T_IPF :
4398 		if (fp->fr_dsize != sizeof(fripf_t))
4399 			return EINVAL;
4400 
4401 		/*
4402 		 * Allowing a rule with both "keep state" and "with oow" is
4403 		 * pointless because adding a state entry to the table will
4404 		 * fail with the out of window (oow) flag set.
4405 		 */
4406 		if ((fp->fr_flags & FR_KEEPSTATE) && (fp->fr_flx & FI_OOW))
4407 			return EINVAL;
4408 
4409 		switch (fp->fr_satype)
4410 		{
4411 		case FRI_BROADCAST :
4412 		case FRI_DYNAMIC :
4413 		case FRI_NETWORK :
4414 		case FRI_NETMASKED :
4415 		case FRI_PEERADDR :
4416 			if (fp->fr_sifpidx < 0 || fp->fr_sifpidx > 3) {
4417 				if (makecopy && fp->fr_data != NULL) {
4418 					KFREES(fp->fr_data, fp->fr_dsize);
4419 				}
4420 				return EINVAL;
4421 			}
4422 			break;
4423 #ifdef	IPFILTER_LOOKUP
4424 		case FRI_LOOKUP :
4425 			fp->fr_srcptr = fr_resolvelookup(fp->fr_srctype,
4426 							 fp->fr_srcnum,
4427 							 &fp->fr_srcfunc, ifs);
4428 			break;
4429 #endif
4430 		default :
4431 			break;
4432 		}
4433 
4434 		switch (fp->fr_datype)
4435 		{
4436 		case FRI_BROADCAST :
4437 		case FRI_DYNAMIC :
4438 		case FRI_NETWORK :
4439 		case FRI_NETMASKED :
4440 		case FRI_PEERADDR :
4441 			if (fp->fr_difpidx < 0 || fp->fr_difpidx > 3) {
4442 				if (makecopy && fp->fr_data != NULL) {
4443 					KFREES(fp->fr_data, fp->fr_dsize);
4444 				}
4445 				return EINVAL;
4446 			}
4447 			break;
4448 #ifdef	IPFILTER_LOOKUP
4449 		case FRI_LOOKUP :
4450 			fp->fr_dstptr = fr_resolvelookup(fp->fr_dsttype,
4451 							 fp->fr_dstnum,
4452 							 &fp->fr_dstfunc, ifs);
4453 			break;
4454 #endif
4455 		default :
4456 			break;
4457 		}
4458 		break;
4459 	case FR_T_NONE :
4460 		break;
4461 	case FR_T_CALLFUNC :
4462 		break;
4463 	case FR_T_COMPIPF :
4464 		break;
4465 	default :
4466 		if (makecopy && fp->fr_data != NULL) {
4467 			KFREES(fp->fr_data, fp->fr_dsize);
4468 		}
4469 		return EINVAL;
4470 	}
4471 
4472 	/*
4473 	 * Lookup all the interface names that are part of the rule.
4474 	 */
4475 	frsynclist(0, 0, NULL, NULL, fp, ifs);
4476 	fp->fr_statecnt = 0;
4477 
4478 	/*
4479 	 * Look for an existing matching filter rule, but don't include the
4480 	 * next or interface pointer in the comparison (fr_next, fr_ifa).
4481 	 * This elminates rules which are indentical being loaded.  Checksum
4482 	 * the constant part of the filter rule to make comparisons quicker
4483 	 * (this meaning no pointers are included).
4484 	 */
4485 	for (fp->fr_cksum = 0, p = (u_int *)&fp->fr_func, pp = &fp->fr_cksum;
4486 	     p < pp; p++)
4487 		fp->fr_cksum += *p;
4488 	pp = (u_int *)(fp->fr_caddr + fp->fr_dsize);
4489 	for (p = (u_int *)fp->fr_data; p < pp; p++)
4490 		fp->fr_cksum += *p;
4491 
4492 	WRITE_ENTER(&ifs->ifs_ipf_mutex);
4493 
4494 	for (; (f = *ftail) != NULL; ftail = &f->fr_next) {
4495 		if ((fp->fr_cksum != f->fr_cksum) ||
4496 		    (f->fr_dsize != fp->fr_dsize))
4497 			continue;
4498 		if (bcmp((char *)&f->fr_func, (char *)&fp->fr_func, FR_CMPSIZ))
4499 			continue;
4500 		if ((!ptr && !f->fr_data) ||
4501 		    (ptr && f->fr_data &&
4502 		     !bcmp((char *)ptr, (char *)f->fr_data, f->fr_dsize)))
4503 			break;
4504 	}
4505 
4506 	/*
4507 	 * If zero'ing statistics, copy current to caller and zero.
4508 	 */
4509 	if (req == (ioctlcmd_t)SIOCZRLST) {
4510 		if (f == NULL)
4511 			error = ESRCH;
4512 		else {
4513 			/*
4514 			 * Copy and reduce lock because of impending copyout.
4515 			 * Well we should, but if we do then the atomicity of
4516 			 * this call and the correctness of fr_hits and
4517 			 * fr_bytes cannot be guaranteed.  As it is, this code
4518 			 * only resets them to 0 if they are successfully
4519 			 * copied out into user space.
4520 			 */
4521 			bcopy((char *)f, (char *)fp, sizeof(*f));
4522 
4523 			/*
4524 			 * When we copy this rule back out, set the data
4525 			 * pointer to be what it was in user space.
4526 			 */
4527 			fp->fr_data = uptr;
4528 			error = fr_outobj(data, fp, IPFOBJ_FRENTRY);
4529 
4530 			if (error == 0) {
4531 				if ((f->fr_dsize != 0) && (uptr != NULL))
4532 					error = COPYOUT(f->fr_data, uptr,
4533 							f->fr_dsize);
4534 				if (error == 0) {
4535 					f->fr_hits = 0;
4536 					f->fr_bytes = 0;
4537 				}
4538 			}
4539 		}
4540 
4541 		if ((ptr != NULL) && (makecopy != 0)) {
4542 			KFREES(ptr, fp->fr_dsize);
4543 		}
4544 		RWLOCK_EXIT(&ifs->ifs_ipf_mutex);
4545 		return error;
4546 	}
4547 
4548 	if (!f) {
4549 		/*
4550 		 * At the end of this, ftail must point to the place where the
4551 		 * new rule is to be saved/inserted/added.
4552 		 * For SIOCAD*FR, this should be the last rule in the group of
4553 		 * rules that have equal fr_collect fields.
4554 		 * For SIOCIN*FR, ...
4555 		 */
4556 		if (req == (ioctlcmd_t)SIOCADAFR ||
4557 		    req == (ioctlcmd_t)SIOCADIFR) {
4558 
4559 			for (ftail = fprev; (f = *ftail) != NULL; ) {
4560 				if (f->fr_collect > fp->fr_collect)
4561 					break;
4562 				ftail = &f->fr_next;
4563 			}
4564 			f = NULL;
4565 			ptr = NULL;
4566 			error = 0;
4567 		} else if (req == (ioctlcmd_t)SIOCINAFR ||
4568 			   req == (ioctlcmd_t)SIOCINIFR) {
4569 			while ((f = *fprev) != NULL) {
4570 				if (f->fr_collect >= fp->fr_collect)
4571 					break;
4572 				fprev = &f->fr_next;
4573 			}
4574 			ftail = fprev;
4575 			if (fp->fr_hits != 0) {
4576 				while (fp->fr_hits && (f = *ftail)) {
4577 					if (f->fr_collect != fp->fr_collect)
4578 						break;
4579 					fprev = ftail;
4580 					ftail = &f->fr_next;
4581 					fp->fr_hits--;
4582 				}
4583 			}
4584 			f = NULL;
4585 			ptr = NULL;
4586 			error = 0;
4587 		}
4588 	}
4589 
4590 	/*
4591 	 * Request to remove a rule.
4592 	 */
4593 	if (req == (ioctlcmd_t)SIOCRMAFR || req == (ioctlcmd_t)SIOCRMIFR) {
4594 		if (!f)
4595 			error = ESRCH;
4596 		else {
4597 			/*
4598 			 * Do not allow activity from user space to interfere
4599 			 * with rules not loaded that way.
4600 			 */
4601 			if ((makecopy == 1) && !(f->fr_flags & FR_COPIED)) {
4602 				error = EPERM;
4603 				goto done;
4604 			}
4605 
4606 			/*
4607 			 * Return EBUSY if the rule is being reference by
4608 			 * something else (eg state information.
4609 			 */
4610 			if (f->fr_ref > 1) {
4611 				error = EBUSY;
4612 				goto done;
4613 			}
4614 #ifdef	IPFILTER_SCAN
4615 			if (f->fr_isctag[0] != '\0' &&
4616 			    (f->fr_isc != (struct ipscan *)-1))
4617 				ipsc_detachfr(f);
4618 #endif
4619 			if (unit == IPL_LOGAUTH) {
4620 				error = fr_preauthcmd(req, f, ftail, ifs);
4621 				goto done;
4622 			}
4623 			if (*f->fr_grhead != '\0')
4624 				fr_delgroup(f->fr_grhead, unit, set, ifs);
4625 			fr_fixskip(ftail, f, -1);
4626 			*ftail = f->fr_next;
4627 			f->fr_next = NULL;
4628 			(void)fr_derefrule(&f, ifs);
4629 		}
4630 	} else {
4631 		/*
4632 		 * Not removing, so we must be adding/inserting a rule.
4633 		 */
4634 		if (f)
4635 			error = EEXIST;
4636 		else {
4637 			if (unit == IPL_LOGAUTH) {
4638 				error = fr_preauthcmd(req, fp, ftail, ifs);
4639 				goto done;
4640 			}
4641 			if (makecopy) {
4642 				KMALLOC(f, frentry_t *);
4643 			} else
4644 				f = fp;
4645 			if (f != NULL) {
4646 				if (fp != f)
4647 					bcopy((char *)fp, (char *)f,
4648 					      sizeof(*f));
4649 				MUTEX_NUKE(&f->fr_lock);
4650 				MUTEX_INIT(&f->fr_lock, "filter rule lock");
4651 #ifdef	IPFILTER_SCAN
4652 				if (f->fr_isctag[0] != '\0' &&
4653 				    ipsc_attachfr(f))
4654 					f->fr_isc = (struct ipscan *)-1;
4655 #endif
4656 				f->fr_hits = 0;
4657 				if (makecopy != 0)
4658 					f->fr_ref = 1;
4659 				f->fr_next = *ftail;
4660 				*ftail = f;
4661 				if (req == (ioctlcmd_t)SIOCINIFR ||
4662 				    req == (ioctlcmd_t)SIOCINAFR)
4663 					fr_fixskip(ftail, f, 1);
4664 				f->fr_grp = NULL;
4665 				group = f->fr_grhead;
4666 				if (*group != '\0') {
4667 					fg = fr_addgroup(group, f, f->fr_flags,
4668 							 unit, set, ifs);
4669 					if (fg != NULL)
4670 						f->fr_grp = &fg->fg_start;
4671 				}
4672 			} else
4673 				error = ENOMEM;
4674 		}
4675 	}
4676 done:
4677 	RWLOCK_EXIT(&ifs->ifs_ipf_mutex);
4678 	if ((ptr != NULL) && (error != 0) && (makecopy != 0)) {
4679 		KFREES(ptr, fp->fr_dsize);
4680 	}
4681 	return (error);
4682 }
4683 
4684 
4685 /* ------------------------------------------------------------------------ */
4686 /* Function:    fr_funcinit                                                 */
4687 /* Returns:     int - 0 == success, else ESRCH: cannot resolve rule details */
4688 /* Parameters:  fr(I) - pointer to filter rule                              */
4689 /*                                                                          */
4690 /* If a rule is a call rule, then check if the function it points to needs  */
4691 /* an init function to be called now the rule has been loaded.              */
4692 /* ------------------------------------------------------------------------ */
4693 static int fr_funcinit(fr, ifs)
4694 frentry_t *fr;
4695 ipf_stack_t *ifs;
4696 {
4697 	ipfunc_resolve_t *ft;
4698 	int err;
4699 
4700 	err = ESRCH;
4701 
4702 	for (ft = fr_availfuncs; ft->ipfu_addr != NULL; ft++)
4703 		if (ft->ipfu_addr == fr->fr_func) {
4704 			err = 0;
4705 			if (ft->ipfu_init != NULL)
4706 				err = (*ft->ipfu_init)(fr, ifs);
4707 			break;
4708 		}
4709 	return err;
4710 }
4711 
4712 
4713 /* ------------------------------------------------------------------------ */
4714 /* Function:    fr_findfunc                                                 */
4715 /* Returns:     ipfunc_t - pointer to function if found, else NULL          */
4716 /* Parameters:  funcptr(I) - function pointer to lookup                     */
4717 /*                                                                          */
4718 /* Look for a function in the table of known functions.                     */
4719 /* ------------------------------------------------------------------------ */
4720 static ipfunc_t fr_findfunc(funcptr)
4721 ipfunc_t funcptr;
4722 {
4723 	ipfunc_resolve_t *ft;
4724 
4725 	for (ft = fr_availfuncs; ft->ipfu_addr != NULL; ft++)
4726 		if (ft->ipfu_addr == funcptr)
4727 			return funcptr;
4728 	return NULL;
4729 }
4730 
4731 
4732 /* ------------------------------------------------------------------------ */
4733 /* Function:    fr_resolvefunc                                              */
4734 /* Returns:     int - 0 == success, else error                              */
4735 /* Parameters:  data(IO) - ioctl data pointer to ipfunc_resolve_t struct    */
4736 /*                                                                          */
4737 /* Copy in a ipfunc_resolve_t structure and then fill in the missing field. */
4738 /* This will either be the function name (if the pointer is set) or the     */
4739 /* function pointer if the name is set.  When found, fill in the other one  */
4740 /* so that the entire, complete, structure can be copied back to user space.*/
4741 /* ------------------------------------------------------------------------ */
4742 int fr_resolvefunc(data)
4743 void *data;
4744 {
4745 	ipfunc_resolve_t res, *ft;
4746 	int err;
4747 
4748 	err = BCOPYIN(data, &res, sizeof(res));
4749 	if (err != 0)
4750 		return EFAULT;
4751 
4752 	if (res.ipfu_addr == NULL && res.ipfu_name[0] != '\0') {
4753 		for (ft = fr_availfuncs; ft->ipfu_addr != NULL; ft++)
4754 			if (strncmp(res.ipfu_name, ft->ipfu_name,
4755 				    sizeof(res.ipfu_name)) == 0) {
4756 				res.ipfu_addr = ft->ipfu_addr;
4757 				res.ipfu_init = ft->ipfu_init;
4758 				if (COPYOUT(&res, data, sizeof(res)) != 0)
4759 					return EFAULT;
4760 				return 0;
4761 			}
4762 	}
4763 	if (res.ipfu_addr != NULL && res.ipfu_name[0] == '\0') {
4764 		for (ft = fr_availfuncs; ft->ipfu_addr != NULL; ft++)
4765 			if (ft->ipfu_addr == res.ipfu_addr) {
4766 				(void) strncpy(res.ipfu_name, ft->ipfu_name,
4767 					       sizeof(res.ipfu_name));
4768 				res.ipfu_init = ft->ipfu_init;
4769 				if (COPYOUT(&res, data, sizeof(res)) != 0)
4770 					return EFAULT;
4771 				return 0;
4772 			}
4773 	}
4774 	return ESRCH;
4775 }
4776 
4777 
4778 #if !defined(_KERNEL) || (!defined(__NetBSD__) && !defined(__OpenBSD__) && !defined(__FreeBSD__)) || \
4779     (defined(__FreeBSD__) && (__FreeBSD_version < 490000)) || \
4780     (defined(__NetBSD__) && (__NetBSD_Version__ < 105000000)) || \
4781     (defined(__OpenBSD__) && (OpenBSD < 200006))
4782 /*
4783  * From: NetBSD
4784  * ppsratecheck(): packets (or events) per second limitation.
4785  */
4786 int
4787 ppsratecheck(lasttime, curpps, maxpps)
4788 	struct timeval *lasttime;
4789 	int *curpps;
4790 	int maxpps;	/* maximum pps allowed */
4791 {
4792 	struct timeval tv, delta;
4793 	int rv;
4794 
4795 	GETKTIME(&tv);
4796 
4797 	delta.tv_sec = tv.tv_sec - lasttime->tv_sec;
4798 	delta.tv_usec = tv.tv_usec - lasttime->tv_usec;
4799 	if (delta.tv_usec < 0) {
4800 		delta.tv_sec--;
4801 		delta.tv_usec += 1000000;
4802 	}
4803 
4804 	/*
4805 	 * check for 0,0 is so that the message will be seen at least once.
4806 	 * if more than one second have passed since the last update of
4807 	 * lasttime, reset the counter.
4808 	 *
4809 	 * we do increment *curpps even in *curpps < maxpps case, as some may
4810 	 * try to use *curpps for stat purposes as well.
4811 	 */
4812 	if ((lasttime->tv_sec == 0 && lasttime->tv_usec == 0) ||
4813 	    delta.tv_sec >= 1) {
4814 		*lasttime = tv;
4815 		*curpps = 0;
4816 		rv = 1;
4817 	} else if (maxpps < 0)
4818 		rv = 1;
4819 	else if (*curpps < maxpps)
4820 		rv = 1;
4821 	else
4822 		rv = 0;
4823 	*curpps = *curpps + 1;
4824 
4825 	return (rv);
4826 }
4827 #endif
4828 
4829 
4830 /* ------------------------------------------------------------------------ */
4831 /* Function:    fr_derefrule                                                */
4832 /* Returns:     int   - 0 == rule freed up, else rule not freed             */
4833 /* Parameters:  fr(I) - pointer to filter rule                              */
4834 /*                                                                          */
4835 /* Decrement the reference counter to a rule by one.  If it reaches zero,   */
4836 /* free it and any associated storage space being used by it.               */
4837 /* ------------------------------------------------------------------------ */
4838 int fr_derefrule(frp, ifs)
4839 frentry_t **frp;
4840 ipf_stack_t *ifs;
4841 {
4842 	frentry_t *fr;
4843 
4844 	fr = *frp;
4845 
4846 	MUTEX_ENTER(&fr->fr_lock);
4847 	fr->fr_ref--;
4848 	if (fr->fr_ref == 0) {
4849 		MUTEX_EXIT(&fr->fr_lock);
4850 		MUTEX_DESTROY(&fr->fr_lock);
4851 
4852 #ifdef IPFILTER_LOOKUP
4853 		if (fr->fr_type == FR_T_IPF && fr->fr_satype == FRI_LOOKUP)
4854 		    ip_lookup_deref(fr->fr_srctype, fr->fr_srcptr, ifs);
4855 		if (fr->fr_type == FR_T_IPF && fr->fr_datype == FRI_LOOKUP)
4856 		    ip_lookup_deref(fr->fr_dsttype, fr->fr_dstptr, ifs);
4857 #endif
4858 
4859 		if (fr->fr_dsize) {
4860 			KFREES(fr->fr_data, fr->fr_dsize);
4861 		}
4862 		if ((fr->fr_flags & FR_COPIED) != 0) {
4863 			KFREE(fr);
4864 			return 0;
4865 		}
4866 		return 1;
4867 	} else {
4868 		MUTEX_EXIT(&fr->fr_lock);
4869 	}
4870 	*frp = NULL;
4871 	return -1;
4872 }
4873 
4874 
4875 #ifdef	IPFILTER_LOOKUP
4876 /* ------------------------------------------------------------------------ */
4877 /* Function:    fr_grpmapinit                                               */
4878 /* Returns:     int - 0 == success, else ESRCH because table entry not found*/
4879 /* Parameters:  fr(I) - pointer to rule to find hash table for              */
4880 /*                                                                          */
4881 /* Looks for group hash table fr_arg and stores a pointer to it in fr_ptr.  */
4882 /* fr_ptr is later used by fr_srcgrpmap and fr_dstgrpmap.                   */
4883 /* ------------------------------------------------------------------------ */
4884 static int fr_grpmapinit(fr, ifs)
4885 frentry_t *fr;
4886 ipf_stack_t *ifs;
4887 {
4888 	char name[FR_GROUPLEN];
4889 	iphtable_t *iph;
4890 
4891 #if defined(SNPRINTF) && defined(_KERNEL)
4892 	(void) SNPRINTF(name, sizeof(name), "%d", fr->fr_arg);
4893 #else
4894 	(void) sprintf(name, "%d", fr->fr_arg);
4895 #endif
4896 	iph = fr_findhtable(IPL_LOGIPF, name, ifs);
4897 	if (iph == NULL)
4898 		return ESRCH;
4899 	if ((iph->iph_flags & FR_INOUT) != (fr->fr_flags & FR_INOUT))
4900 		return ESRCH;
4901 	fr->fr_ptr = iph;
4902 	return 0;
4903 }
4904 
4905 
4906 /* ------------------------------------------------------------------------ */
4907 /* Function:    fr_srcgrpmap                                                */
4908 /* Returns:     frentry_t * - pointer to "new last matching" rule or NULL   */
4909 /* Parameters:  fin(I)    - pointer to packet information                   */
4910 /*              passp(IO) - pointer to current/new filter decision (unused) */
4911 /*                                                                          */
4912 /* Look for a rule group head in a hash table, using the source address as  */
4913 /* the key, and descend into that group and continue matching rules against */
4914 /* the packet.                                                              */
4915 /* ------------------------------------------------------------------------ */
4916 frentry_t *fr_srcgrpmap(fin, passp)
4917 fr_info_t *fin;
4918 u_32_t *passp;
4919 {
4920 	frgroup_t *fg;
4921 	void *rval;
4922 	ipf_stack_t *ifs = fin->fin_ifs;
4923 
4924 	rval = fr_iphmfindgroup(fin->fin_fr->fr_ptr, fin->fin_v, &fin->fin_src, ifs);
4925 	if (rval == NULL)
4926 		return NULL;
4927 
4928 	fg = rval;
4929 	fin->fin_fr = fg->fg_start;
4930 	(void) fr_scanlist(fin, *passp);
4931 	return fin->fin_fr;
4932 }
4933 
4934 
4935 /* ------------------------------------------------------------------------ */
4936 /* Function:    fr_dstgrpmap                                                */
4937 /* Returns:     frentry_t * - pointer to "new last matching" rule or NULL   */
4938 /* Parameters:  fin(I)    - pointer to packet information                   */
4939 /*              passp(IO) - pointer to current/new filter decision (unused) */
4940 /*                                                                          */
4941 /* Look for a rule group head in a hash table, using the destination        */
4942 /* address as the key, and descend into that group and continue matching    */
4943 /* rules against  the packet.                                               */
4944 /* ------------------------------------------------------------------------ */
4945 frentry_t *fr_dstgrpmap(fin, passp)
4946 fr_info_t *fin;
4947 u_32_t *passp;
4948 {
4949 	frgroup_t *fg;
4950 	void *rval;
4951 	ipf_stack_t *ifs = fin->fin_ifs;
4952 
4953 	rval = fr_iphmfindgroup(fin->fin_fr->fr_ptr, fin->fin_v, &fin->fin_dst, ifs);
4954 	if (rval == NULL)
4955 		return NULL;
4956 
4957 	fg = rval;
4958 	fin->fin_fr = fg->fg_start;
4959 	(void) fr_scanlist(fin, *passp);
4960 	return fin->fin_fr;
4961 }
4962 #endif /* IPFILTER_LOOKUP */
4963 
4964 /*
4965  * Queue functions
4966  * ===============
4967  * These functions manage objects on queues for efficient timeouts.  There are
4968  * a number of system defined queues as well as user defined timeouts.  It is
4969  * expected that a lock is held in the domain in which the queue belongs
4970  * (i.e. either state or NAT) when calling any of these functions that prevents
4971  * fr_freetimeoutqueue() from being called at the same time as any other.
4972  */
4973 
4974 
4975 /* ------------------------------------------------------------------------ */
4976 /* Function:    fr_addtimeoutqueue                                          */
4977 /* Returns:     struct ifqtq * - NULL if malloc fails, else pointer to      */
4978 /*                               timeout queue with given interval.         */
4979 /* Parameters:  parent(I)  - pointer to pointer to parent node of this list */
4980 /*                           of interface queues.                           */
4981 /*              seconds(I) - timeout value in seconds for this queue.       */
4982 /*                                                                          */
4983 /* This routine first looks for a timeout queue that matches the interval   */
4984 /* being requested.  If it finds one, increments the reference counter and  */
4985 /* returns a pointer to it.  If none are found, it allocates a new one and  */
4986 /* inserts it at the top of the list.                                       */
4987 /*                                                                          */
4988 /* Locking.                                                                 */
4989 /* It is assumed that the caller of this function has an appropriate lock   */
4990 /* held (exclusively) in the domain that encompases 'parent'.               */
4991 /* ------------------------------------------------------------------------ */
4992 ipftq_t *fr_addtimeoutqueue(parent, seconds, ifs)
4993 ipftq_t **parent;
4994 u_int seconds;
4995 ipf_stack_t *ifs;
4996 {
4997 	ipftq_t *ifq;
4998 	u_int period;
4999 
5000 	period = seconds * IPF_HZ_DIVIDE;
5001 
5002 	MUTEX_ENTER(&ifs->ifs_ipf_timeoutlock);
5003 	for (ifq = *parent; ifq != NULL; ifq = ifq->ifq_next) {
5004 		if (ifq->ifq_ttl == period) {
5005 			/*
5006 			 * Reset the delete flag, if set, so the structure
5007 			 * gets reused rather than freed and reallocated.
5008 			 */
5009 			MUTEX_ENTER(&ifq->ifq_lock);
5010 			ifq->ifq_flags &= ~IFQF_DELETE;
5011 			ifq->ifq_ref++;
5012 			MUTEX_EXIT(&ifq->ifq_lock);
5013 			MUTEX_EXIT(&ifs->ifs_ipf_timeoutlock);
5014 
5015 			return ifq;
5016 		}
5017 	}
5018 
5019 	KMALLOC(ifq, ipftq_t *);
5020 	if (ifq != NULL) {
5021 		ifq->ifq_ttl = period;
5022 		ifq->ifq_head = NULL;
5023 		ifq->ifq_tail = &ifq->ifq_head;
5024 		ifq->ifq_next = *parent;
5025 		ifq->ifq_pnext = parent;
5026 		ifq->ifq_ref = 1;
5027 		ifq->ifq_flags = IFQF_USER;
5028 		*parent = ifq;
5029 		ifs->ifs_fr_userifqs++;
5030 		MUTEX_NUKE(&ifq->ifq_lock);
5031 		MUTEX_INIT(&ifq->ifq_lock, "ipftq mutex");
5032 	}
5033 	MUTEX_EXIT(&ifs->ifs_ipf_timeoutlock);
5034 	return ifq;
5035 }
5036 
5037 
5038 /* ------------------------------------------------------------------------ */
5039 /* Function:    fr_deletetimeoutqueue                                       */
5040 /* Returns:     int    - new reference count value of the timeout queue     */
5041 /* Parameters:  ifq(I) - timeout queue which is losing a reference.         */
5042 /* Locks:       ifq->ifq_lock                                               */
5043 /*                                                                          */
5044 /* This routine must be called when we're discarding a pointer to a timeout */
5045 /* queue object, taking care of the reference counter.                      */
5046 /*                                                                          */
5047 /* Now that this just sets a DELETE flag, it requires the expire code to    */
5048 /* check the list of user defined timeout queues and call the free function */
5049 /* below (currently commented out) to stop memory leaking.  It is done this */
5050 /* way because the locking may not be sufficient to safely do a free when   */
5051 /* this function is called.                                                 */
5052 /* ------------------------------------------------------------------------ */
5053 int fr_deletetimeoutqueue(ifq)
5054 ipftq_t *ifq;
5055 {
5056 
5057 	ifq->ifq_ref--;
5058 	if ((ifq->ifq_ref == 0) && ((ifq->ifq_flags & IFQF_USER) != 0)) {
5059 		ifq->ifq_flags |= IFQF_DELETE;
5060 	}
5061 
5062 	return ifq->ifq_ref;
5063 }
5064 
5065 
5066 /* ------------------------------------------------------------------------ */
5067 /* Function:    fr_freetimeoutqueue                                         */
5068 /* Parameters:  ifq(I) - timeout queue which is losing a reference.         */
5069 /* Returns:     Nil                                                         */
5070 /*                                                                          */
5071 /* Locking:                                                                 */
5072 /* It is assumed that the caller of this function has an appropriate lock   */
5073 /* held (exclusively) in the domain that encompases the callers "domain".   */
5074 /* The ifq_lock for this structure should not be held.                      */
5075 /*                                                                          */
5076 /* Remove a user definde timeout queue from the list of queues it is in and */
5077 /* tidy up after this is done.                                              */
5078 /* ------------------------------------------------------------------------ */
5079 void fr_freetimeoutqueue(ifq, ifs)
5080 ipftq_t *ifq;
5081 ipf_stack_t *ifs;
5082 {
5083 
5084 
5085 	if (((ifq->ifq_flags & IFQF_DELETE) == 0) || (ifq->ifq_ref != 0) ||
5086 	    ((ifq->ifq_flags & IFQF_USER) == 0)) {
5087 		printf("fr_freetimeoutqueue(%lx) flags 0x%x ttl %d ref %d\n",
5088 		       (u_long)ifq, ifq->ifq_flags, ifq->ifq_ttl,
5089 		       ifq->ifq_ref);
5090 		return;
5091 	}
5092 
5093 	/*
5094 	 * Remove from its position in the list.
5095 	 */
5096 	*ifq->ifq_pnext = ifq->ifq_next;
5097 	if (ifq->ifq_next != NULL)
5098 		ifq->ifq_next->ifq_pnext = ifq->ifq_pnext;
5099 
5100 	MUTEX_DESTROY(&ifq->ifq_lock);
5101 	ifs->ifs_fr_userifqs--;
5102 	KFREE(ifq);
5103 }
5104 
5105 
5106 /* ------------------------------------------------------------------------ */
5107 /* Function:    fr_deletequeueentry                                         */
5108 /* Returns:     Nil                                                         */
5109 /* Parameters:  tqe(I) - timeout queue entry to delete                      */
5110 /*              ifq(I) - timeout queue to remove entry from                 */
5111 /*                                                                          */
5112 /* Remove a tail queue entry from its queue and make it an orphan.          */
5113 /* fr_deletetimeoutqueue is called to make sure the reference count on the  */
5114 /* queue is correct.  We can't, however, call fr_freetimeoutqueue because   */
5115 /* the correct lock(s) may not be held that would make it safe to do so.    */
5116 /* ------------------------------------------------------------------------ */
5117 void fr_deletequeueentry(tqe)
5118 ipftqent_t *tqe;
5119 {
5120 	ipftq_t *ifq;
5121 
5122 	ifq = tqe->tqe_ifq;
5123 	if (ifq == NULL)
5124 		return;
5125 
5126 	MUTEX_ENTER(&ifq->ifq_lock);
5127 
5128 	if (tqe->tqe_pnext != NULL) {
5129 		*tqe->tqe_pnext = tqe->tqe_next;
5130 		if (tqe->tqe_next != NULL)
5131 			tqe->tqe_next->tqe_pnext = tqe->tqe_pnext;
5132 		else    /* we must be the tail anyway */
5133 			ifq->ifq_tail = tqe->tqe_pnext;
5134 
5135 		tqe->tqe_pnext = NULL;
5136 		tqe->tqe_ifq = NULL;
5137 	}
5138 
5139 	(void) fr_deletetimeoutqueue(ifq);
5140 
5141 	MUTEX_EXIT(&ifq->ifq_lock);
5142 }
5143 
5144 
5145 /* ------------------------------------------------------------------------ */
5146 /* Function:    fr_queuefront                                               */
5147 /* Returns:     Nil                                                         */
5148 /* Parameters:  tqe(I) - pointer to timeout queue entry                     */
5149 /*                                                                          */
5150 /* Move a queue entry to the front of the queue, if it isn't already there. */
5151 /* ------------------------------------------------------------------------ */
5152 void fr_queuefront(tqe)
5153 ipftqent_t *tqe;
5154 {
5155 	ipftq_t *ifq;
5156 
5157 	ifq = tqe->tqe_ifq;
5158 	if (ifq == NULL)
5159 		return;
5160 
5161 	MUTEX_ENTER(&ifq->ifq_lock);
5162 	if (ifq->ifq_head != tqe) {
5163 		*tqe->tqe_pnext = tqe->tqe_next;
5164 		if (tqe->tqe_next)
5165 			tqe->tqe_next->tqe_pnext = tqe->tqe_pnext;
5166 		else
5167 			ifq->ifq_tail = tqe->tqe_pnext;
5168 
5169 		tqe->tqe_next = ifq->ifq_head;
5170 		ifq->ifq_head->tqe_pnext = &tqe->tqe_next;
5171 		ifq->ifq_head = tqe;
5172 		tqe->tqe_pnext = &ifq->ifq_head;
5173 	}
5174 	MUTEX_EXIT(&ifq->ifq_lock);
5175 }
5176 
5177 
5178 /* ------------------------------------------------------------------------ */
5179 /* Function:    fr_queueback                                                */
5180 /* Returns:     Nil                                                         */
5181 /* Parameters:  tqe(I) - pointer to timeout queue entry                     */
5182 /*                                                                          */
5183 /* Move a queue entry to the back of the queue, if it isn't already there.  */
5184 /* ------------------------------------------------------------------------ */
5185 void fr_queueback(tqe, ifs)
5186 ipftqent_t *tqe;
5187 ipf_stack_t *ifs;
5188 {
5189 	ipftq_t *ifq;
5190 
5191 	ifq = tqe->tqe_ifq;
5192 	if (ifq == NULL)
5193 		return;
5194 	tqe->tqe_die = ifs->ifs_fr_ticks + ifq->ifq_ttl;
5195 
5196 	MUTEX_ENTER(&ifq->ifq_lock);
5197 	if (tqe->tqe_next == NULL) {		/* at the end already ? */
5198 		MUTEX_EXIT(&ifq->ifq_lock);
5199 		return;
5200 	}
5201 
5202 	/*
5203 	 * Remove from list
5204 	 */
5205 	*tqe->tqe_pnext = tqe->tqe_next;
5206 	tqe->tqe_next->tqe_pnext = tqe->tqe_pnext;
5207 
5208 	/*
5209 	 * Make it the last entry.
5210 	 */
5211 	tqe->tqe_next = NULL;
5212 	tqe->tqe_pnext = ifq->ifq_tail;
5213 	*ifq->ifq_tail = tqe;
5214 	ifq->ifq_tail = &tqe->tqe_next;
5215 	MUTEX_EXIT(&ifq->ifq_lock);
5216 }
5217 
5218 
5219 /* ------------------------------------------------------------------------ */
5220 /* Function:    fr_queueappend                                              */
5221 /* Returns:     Nil                                                         */
5222 /* Parameters:  tqe(I)    - pointer to timeout queue entry                  */
5223 /*              ifq(I)    - pointer to timeout queue                        */
5224 /*              parent(I) - owing object pointer                            */
5225 /*                                                                          */
5226 /* Add a new item to this queue and put it on the very end.                 */
5227 /* ------------------------------------------------------------------------ */
5228 void fr_queueappend(tqe, ifq, parent, ifs)
5229 ipftqent_t *tqe;
5230 ipftq_t *ifq;
5231 void *parent;
5232 ipf_stack_t *ifs;
5233 {
5234 
5235 	MUTEX_ENTER(&ifq->ifq_lock);
5236 	tqe->tqe_parent = parent;
5237 	tqe->tqe_pnext = ifq->ifq_tail;
5238 	*ifq->ifq_tail = tqe;
5239 	ifq->ifq_tail = &tqe->tqe_next;
5240 	tqe->tqe_next = NULL;
5241 	tqe->tqe_ifq = ifq;
5242 	tqe->tqe_die = ifs->ifs_fr_ticks + ifq->ifq_ttl;
5243 	ifq->ifq_ref++;
5244 	MUTEX_EXIT(&ifq->ifq_lock);
5245 }
5246 
5247 
5248 /* ------------------------------------------------------------------------ */
5249 /* Function:    fr_movequeue                                                */
5250 /* Returns:     Nil                                                         */
5251 /* Parameters:  tq(I)   - pointer to timeout queue information              */
5252 /*              oifp(I) - old timeout queue entry was on                    */
5253 /*              nifp(I) - new timeout queue to put entry on                 */
5254 /*		ifs	- ipf stack instance				    */
5255 /*                                                                          */
5256 /* Move a queue entry from one timeout queue to another timeout queue.      */
5257 /* If it notices that the current entry is already last and does not need   */
5258 /* to move queue, the return.                                               */
5259 /* ------------------------------------------------------------------------ */
5260 void fr_movequeue(tqe, oifq, nifq, ifs)
5261 ipftqent_t *tqe;
5262 ipftq_t *oifq, *nifq;
5263 ipf_stack_t *ifs;
5264 {
5265 	/*
5266 	 * If the queue isn't changing, and the clock hasn't ticked
5267 	 * since the last update, the operation will be a no-op.
5268 	 */
5269 	if (oifq == nifq && tqe->tqe_touched == ifs->ifs_fr_ticks)
5270 		return;
5271 
5272 	/*
5273 	 * Grab the lock and update the timers.
5274 	 */
5275 	MUTEX_ENTER(&oifq->ifq_lock);
5276 	tqe->tqe_touched = ifs->ifs_fr_ticks;
5277 	tqe->tqe_die = ifs->ifs_fr_ticks + nifq->ifq_ttl;
5278 
5279 	/*
5280 	 * The remainder of the operation can still be a no-op.
5281 	 *
5282 	 * If the queue isn't changing, check to see if
5283 	 * an update would be meaningless.
5284 	 */
5285 	if (oifq == nifq) {
5286 		if ((tqe->tqe_next == NULL) ||
5287 		    (tqe->tqe_next->tqe_die == tqe->tqe_die)) {
5288 			MUTEX_EXIT(&oifq->ifq_lock);
5289 			return;
5290 		}
5291 	}
5292 
5293 	/*
5294 	 * Remove from the old queue
5295 	 */
5296 	*tqe->tqe_pnext = tqe->tqe_next;
5297 	if (tqe->tqe_next)
5298 		tqe->tqe_next->tqe_pnext = tqe->tqe_pnext;
5299 	else
5300 		oifq->ifq_tail = tqe->tqe_pnext;
5301 	tqe->tqe_next = NULL;
5302 
5303 	/*
5304 	 * If we're moving from one queue to another, release the lock on the
5305 	 * old queue and get a lock on the new queue.  For user defined queues,
5306 	 * if we're moving off it, call delete in case it can now be freed.
5307 	 */
5308 	if (oifq != nifq) {
5309 		tqe->tqe_ifq = NULL;
5310 
5311 		(void) fr_deletetimeoutqueue(oifq);
5312 
5313 		MUTEX_EXIT(&oifq->ifq_lock);
5314 
5315 		MUTEX_ENTER(&nifq->ifq_lock);
5316 
5317 		tqe->tqe_ifq = nifq;
5318 		nifq->ifq_ref++;
5319 	}
5320 
5321 	/*
5322 	 * Add to the bottom of the new queue
5323 	 */
5324 	tqe->tqe_pnext = nifq->ifq_tail;
5325 	*nifq->ifq_tail = tqe;
5326 	nifq->ifq_tail = &tqe->tqe_next;
5327 	MUTEX_EXIT(&nifq->ifq_lock);
5328 }
5329 
5330 
5331 /* ------------------------------------------------------------------------ */
5332 /* Function:    fr_updateipid                                               */
5333 /* Returns:     int - 0 == success, -1 == error (packet should be droppped) */
5334 /* Parameters:  fin(I) - pointer to packet information                      */
5335 /*                                                                          */
5336 /* When we are doing NAT, change the IP of every packet to represent a      */
5337 /* single sequence of packets coming from the host, hiding any host         */
5338 /* specific sequencing that might otherwise be revealed.  If the packet is  */
5339 /* a fragment, then store the 'new' IPid in the fragment cache and look up  */
5340 /* the fragment cache for non-leading fragments.  If a non-leading fragment */
5341 /* has no match in the cache, return an error.                              */
5342 /* ------------------------------------------------------------------------ */
5343 static INLINE int fr_updateipid(fin)
5344 fr_info_t *fin;
5345 {
5346 	u_short id, ido, sums;
5347 	u_32_t sumd, sum;
5348 	ip_t *ip;
5349 
5350 	if (fin->fin_off != 0) {
5351 		sum = fr_ipid_knownfrag(fin);
5352 		if (sum == 0xffffffff)
5353 			return -1;
5354 		sum &= 0xffff;
5355 		id = (u_short)sum;
5356 	} else {
5357 		id = fr_nextipid(fin);
5358 		if (fin->fin_off == 0 && (fin->fin_flx & FI_FRAG) != 0)
5359 			(void) fr_ipid_newfrag(fin, (u_32_t)id);
5360 	}
5361 
5362 	ip = fin->fin_ip;
5363 	ido = ntohs(ip->ip_id);
5364 	if (id == ido)
5365 		return 0;
5366 	ip->ip_id = htons(id);
5367 	CALC_SUMD(ido, id, sumd);	/* DESTRUCTIVE MACRO! id,ido change */
5368 	sum = (~ntohs(ip->ip_sum)) & 0xffff;
5369 	sum += sumd;
5370 	sum = (sum >> 16) + (sum & 0xffff);
5371 	sum = (sum >> 16) + (sum & 0xffff);
5372 	sums = ~(u_short)sum;
5373 	ip->ip_sum = htons(sums);
5374 	return 0;
5375 }
5376 
5377 
5378 #ifdef	NEED_FRGETIFNAME
5379 /* ------------------------------------------------------------------------ */
5380 /* Function:    fr_getifname                                                */
5381 /* Returns:     char *    - pointer to interface name                       */
5382 /* Parameters:  ifp(I)    - pointer to network interface                    */
5383 /*              buffer(O) - pointer to where to store interface name        */
5384 /*                                                                          */
5385 /* Constructs an interface name in the buffer passed.  The buffer passed is */
5386 /* expected to be at least LIFNAMSIZ in bytes big.  If buffer is passed in  */
5387 /* as a NULL pointer then return a pointer to a static array.               */
5388 /* ------------------------------------------------------------------------ */
5389 char *fr_getifname(ifp, buffer)
5390 struct ifnet *ifp;
5391 char *buffer;
5392 {
5393 	static char namebuf[LIFNAMSIZ];
5394 # if defined(MENTAT) || defined(__FreeBSD__) || defined(__osf__) || \
5395      defined(__sgi) || defined(linux) || defined(_AIX51) || \
5396      (defined(sun) && !defined(__SVR4) && !defined(__svr4__))
5397 	int unit, space;
5398 	char temp[20];
5399 	char *s;
5400 # endif
5401 
5402 	ASSERT(buffer != NULL);
5403 #ifdef notdef
5404 	if (buffer == NULL)
5405 		buffer = namebuf;
5406 #endif
5407 	(void) strncpy(buffer, ifp->if_name, LIFNAMSIZ);
5408 	buffer[LIFNAMSIZ - 1] = '\0';
5409 # if defined(MENTAT) || defined(__FreeBSD__) || defined(__osf__) || \
5410      defined(__sgi) || defined(_AIX51) || \
5411      (defined(sun) && !defined(__SVR4) && !defined(__svr4__))
5412 	for (s = buffer; *s; s++)
5413 		;
5414 	unit = ifp->if_unit;
5415 	space = LIFNAMSIZ - (s - buffer);
5416 	if (space > 0) {
5417 #  if defined(SNPRINTF) && defined(_KERNEL)
5418 		(void) SNPRINTF(temp, sizeof(temp), "%d", unit);
5419 #  else
5420 		(void) sprintf(temp, "%d", unit);
5421 #  endif
5422 		(void) strncpy(s, temp, space);
5423 	}
5424 # endif
5425 	return buffer;
5426 }
5427 #endif
5428 
5429 
5430 /* ------------------------------------------------------------------------ */
5431 /* Function:    fr_ioctlswitch                                              */
5432 /* Returns:     int     - -1 continue processing, else ioctl return value   */
5433 /* Parameters:  unit(I) - device unit opened                                */
5434 /*              data(I) - pointer to ioctl data                             */
5435 /*              cmd(I)  - ioctl command                                     */
5436 /*              mode(I) - mode value                                        */
5437 /*                                                                          */
5438 /* Based on the value of unit, call the appropriate ioctl handler or return */
5439 /* EIO if ipfilter is not running.   Also checks if write perms are req'd   */
5440 /* for the device in order to execute the ioctl.                            */
5441 /* ------------------------------------------------------------------------ */
5442 INLINE int fr_ioctlswitch(unit, data, cmd, mode, uid, ctx, ifs)
5443 int unit, mode, uid;
5444 ioctlcmd_t cmd;
5445 void *data, *ctx;
5446 ipf_stack_t *ifs;
5447 {
5448 	int error = 0;
5449 
5450 	switch (unit)
5451 	{
5452 	case IPL_LOGIPF :
5453 		error = -1;
5454 		break;
5455 	case IPL_LOGNAT :
5456 		if (ifs->ifs_fr_running > 0)
5457 			error = fr_nat_ioctl(data, cmd, mode, uid, ctx, ifs);
5458 		else
5459 			error = EIO;
5460 		break;
5461 	case IPL_LOGSTATE :
5462 		if (ifs->ifs_fr_running > 0)
5463 			error = fr_state_ioctl(data, cmd, mode, uid, ctx, ifs);
5464 		else
5465 			error = EIO;
5466 		break;
5467 	case IPL_LOGAUTH :
5468 		if (ifs->ifs_fr_running > 0) {
5469 			if ((cmd == (ioctlcmd_t)SIOCADAFR) ||
5470 			    (cmd == (ioctlcmd_t)SIOCRMAFR)) {
5471 				if (!(mode & FWRITE)) {
5472 					error = EPERM;
5473 				} else {
5474 					error = frrequest(unit, cmd, data,
5475 						  ifs->ifs_fr_active, 1, ifs);
5476 				}
5477 			} else {
5478 				error = fr_auth_ioctl(data, cmd, mode, uid, ctx, ifs);
5479 			}
5480 		} else
5481 			error = EIO;
5482 		break;
5483 	case IPL_LOGSYNC :
5484 #ifdef IPFILTER_SYNC
5485 		if (ifs->ifs_fr_running > 0)
5486 			error = fr_sync_ioctl(data, cmd, mode, ifs);
5487 		else
5488 #endif
5489 			error = EIO;
5490 		break;
5491 	case IPL_LOGSCAN :
5492 #ifdef IPFILTER_SCAN
5493 		if (ifs->ifs_fr_running > 0)
5494 			error = fr_scan_ioctl(data, cmd, mode, ifs);
5495 		else
5496 #endif
5497 			error = EIO;
5498 		break;
5499 	case IPL_LOGLOOKUP :
5500 #ifdef IPFILTER_LOOKUP
5501 		if (ifs->ifs_fr_running > 0)
5502 			error = ip_lookup_ioctl(data, cmd, mode, uid, ctx, ifs);
5503 		else
5504 #endif
5505 			error = EIO;
5506 		break;
5507 	default :
5508 		error = EIO;
5509 		break;
5510 	}
5511 
5512 	return error;
5513 }
5514 
5515 
5516 /*
5517  * This array defines the expected size of objects coming into the kernel
5518  * for the various recognised object types.
5519  */
5520 #define	NUM_OBJ_TYPES	19
5521 
5522 static	int	fr_objbytes[NUM_OBJ_TYPES][2] = {
5523 	{ 1,	sizeof(struct frentry) },		/* frentry */
5524 	{ 0,	sizeof(struct friostat) },
5525 	{ 0,	sizeof(struct fr_info) },
5526 	{ 0,	sizeof(struct fr_authstat) },
5527 	{ 0,	sizeof(struct ipfrstat) },
5528 	{ 0,	sizeof(struct ipnat) },
5529 	{ 0,	sizeof(struct natstat) },
5530 	{ 0,	sizeof(struct ipstate_save) },
5531 	{ 1,	sizeof(struct nat_save) },		/* nat_save */
5532 	{ 0,	sizeof(struct natlookup) },
5533 	{ 1,	sizeof(struct ipstate) },		/* ipstate */
5534 	{ 0,	sizeof(struct ips_stat) },
5535 	{ 0,	sizeof(struct frauth) },
5536 	{ 0,	sizeof(struct ipftune) },
5537 	{ 0,	sizeof(struct nat) },                   /* nat_t */
5538 	{ 0,	sizeof(struct ipfruleiter) },
5539 	{ 0,	sizeof(struct ipfgeniter) },
5540 	{ 0,	sizeof(struct ipftable) },
5541 	{ 0,	sizeof(struct ipflookupiter) }
5542 };
5543 
5544 
5545 /* ------------------------------------------------------------------------ */
5546 /* Function:    fr_inobj                                                    */
5547 /* Returns:     int     - 0 = success, else failure                         */
5548 /* Parameters:  data(I) - pointer to ioctl data                             */
5549 /*              ptr(I)  - pointer to store real data in                     */
5550 /*              type(I) - type of structure being moved                     */
5551 /*                                                                          */
5552 /* Copy in the contents of what the ipfobj_t points to.  In future, we      */
5553 /* add things to check for version numbers, sizes, etc, to make it backward */
5554 /* compatible at the ABI for user land.                                     */
5555 /* ------------------------------------------------------------------------ */
5556 int fr_inobj(data, ptr, type)
5557 void *data;
5558 void *ptr;
5559 int type;
5560 {
5561 	ipfobj_t obj;
5562 	int error = 0;
5563 
5564 	if ((type < 0) || (type > NUM_OBJ_TYPES-1))
5565 		return EINVAL;
5566 
5567 	error = BCOPYIN((caddr_t)data, (caddr_t)&obj, sizeof(obj));
5568 	if (error != 0)
5569 		return EFAULT;
5570 
5571 	if (obj.ipfo_type != type)
5572 		return EINVAL;
5573 
5574 #ifndef	IPFILTER_COMPAT
5575 	if ((fr_objbytes[type][0] & 1) != 0) {
5576 		if (obj.ipfo_size < fr_objbytes[type][1])
5577 			return EINVAL;
5578 	} else if (obj.ipfo_size != fr_objbytes[type][1])
5579 		return EINVAL;
5580 #else
5581 	if (obj.ipfo_rev != IPFILTER_VERSION) {
5582 		error = fr_incomptrans(&obj, ptr);
5583 		return error;
5584 	}
5585 
5586 	if ((fr_objbytes[type][0] & 1) != 0 &&
5587 	    obj.ipfo_size < fr_objbytes[type][1] ||
5588 	    obj.ipfo_size != fr_objbytes[type][1])
5589 		return EINVAL;
5590 #endif
5591 
5592 	if ((fr_objbytes[type][0] & 1) != 0) {
5593 		error = COPYIN((caddr_t)obj.ipfo_ptr, (caddr_t)ptr,
5594 				fr_objbytes[type][1]);
5595 	} else {
5596 		error = COPYIN((caddr_t)obj.ipfo_ptr, (caddr_t)ptr,
5597 				obj.ipfo_size);
5598 	}
5599 	return error;
5600 }
5601 
5602 
5603 /* ------------------------------------------------------------------------ */
5604 /* Function:    fr_inobjsz                                                  */
5605 /* Returns:     int     - 0 = success, else failure                         */
5606 /* Parameters:  data(I) - pointer to ioctl data                             */
5607 /*              ptr(I)  - pointer to store real data in                     */
5608 /*              type(I) - type of structure being moved                     */
5609 /*              sz(I)   - size of data to copy                              */
5610 /*                                                                          */
5611 /* As per fr_inobj, except the size of the object to copy in is passed in   */
5612 /* but it must not be smaller than the size defined for the type and the    */
5613 /* type must allow for varied sized objects.  The extra requirement here is */
5614 /* that sz must match the size of the object being passed in - this is not  */
5615 /* not possible nor required in fr_inobj().                                 */
5616 /* ------------------------------------------------------------------------ */
5617 int fr_inobjsz(data, ptr, type, sz)
5618 void *data;
5619 void *ptr;
5620 int type, sz;
5621 {
5622 	ipfobj_t obj;
5623 	int error;
5624 
5625 	if ((type < 0) || (type > NUM_OBJ_TYPES-1))
5626 		return EINVAL;
5627 	if (((fr_objbytes[type][0] & 1) == 0) || (sz < fr_objbytes[type][1]))
5628 		return EINVAL;
5629 
5630 	error = BCOPYIN((caddr_t)data, (caddr_t)&obj, sizeof(obj));
5631 	if (error != 0)
5632 		return EFAULT;
5633 
5634 	if (obj.ipfo_type != type)
5635 		return EINVAL;
5636 
5637 #ifndef	IPFILTER_COMPAT
5638 	if (obj.ipfo_size != sz)
5639 		return EINVAL;
5640 #else
5641 	if (obj.ipfo_rev != IPFILTER_VERSION)
5642 		/*XXX compatibility hook here */
5643 		/*EMPTY*/;
5644 	if (obj.ipfo_size != sz)
5645 		/* XXX compatibility hook here */
5646 		return EINVAL;
5647 #endif
5648 
5649 	error = COPYIN((caddr_t)obj.ipfo_ptr, (caddr_t)ptr, sz);
5650 	return error;
5651 }
5652 
5653 
5654 /* ------------------------------------------------------------------------ */
5655 /* Function:    fr_outobjsz                                                 */
5656 /* Returns:     int     - 0 = success, else failure                         */
5657 /* Parameters:  data(I) - pointer to ioctl data                             */
5658 /*              ptr(I)  - pointer to store real data in                     */
5659 /*              type(I) - type of structure being moved                     */
5660 /*              sz(I)   - size of data to copy                              */
5661 /*                                                                          */
5662 /* As per fr_outobj, except the size of the object to copy out is passed in */
5663 /* but it must not be smaller than the size defined for the type and the    */
5664 /* type must allow for varied sized objects.  The extra requirement here is */
5665 /* that sz must match the size of the object being passed in - this is not  */
5666 /* not possible nor required in fr_outobj().                                */
5667 /* ------------------------------------------------------------------------ */
5668 int fr_outobjsz(data, ptr, type, sz)
5669 void *data;
5670 void *ptr;
5671 int type, sz;
5672 {
5673 	ipfobj_t obj;
5674 	int error;
5675 
5676 	if ((type < 0) || (type > NUM_OBJ_TYPES-1) ||
5677 	    ((fr_objbytes[type][0] & 1) == 0) ||
5678 	    (sz < fr_objbytes[type][1]))
5679 		return EINVAL;
5680 
5681 	error = BCOPYIN((caddr_t)data, (caddr_t)&obj, sizeof(obj));
5682 	if (error != 0)
5683 		return EFAULT;
5684 
5685 	if (obj.ipfo_type != type)
5686 		return EINVAL;
5687 
5688 #ifndef	IPFILTER_COMPAT
5689 	if (obj.ipfo_size != sz)
5690 		return EINVAL;
5691 #else
5692 	if (obj.ipfo_rev != IPFILTER_VERSION)
5693 		/* XXX compatibility hook here */
5694 		/*EMPTY*/;
5695 	if (obj.ipfo_size != sz)
5696 		/* XXX compatibility hook here */
5697 		return EINVAL;
5698 #endif
5699 
5700 	error = COPYOUT((caddr_t)ptr, (caddr_t)obj.ipfo_ptr, sz);
5701 	return error;
5702 }
5703 
5704 
5705 /* ------------------------------------------------------------------------ */
5706 /* Function:    fr_outobj                                                   */
5707 /* Returns:     int     - 0 = success, else failure                         */
5708 /* Parameters:  data(I) - pointer to ioctl data                             */
5709 /*              ptr(I)  - pointer to store real data in                     */
5710 /*              type(I) - type of structure being moved                     */
5711 /*                                                                          */
5712 /* Copy out the contents of what ptr is to where ipfobj points to.  In      */
5713 /* future, we add things to check for version numbers, sizes, etc, to make  */
5714 /* it backward  compatible at the ABI for user land.                        */
5715 /* ------------------------------------------------------------------------ */
5716 int fr_outobj(data, ptr, type)
5717 void *data;
5718 void *ptr;
5719 int type;
5720 {
5721 	ipfobj_t obj;
5722 	int error;
5723 
5724 	if ((type < 0) || (type > NUM_OBJ_TYPES-1))
5725 		return EINVAL;
5726 
5727 	error = BCOPYIN((caddr_t)data, (caddr_t)&obj, sizeof(obj));
5728 	if (error != 0)
5729 		return EFAULT;
5730 
5731 	if (obj.ipfo_type != type)
5732 		return EINVAL;
5733 
5734 #ifndef	IPFILTER_COMPAT
5735 	if ((fr_objbytes[type][0] & 1) != 0) {
5736 		if (obj.ipfo_size < fr_objbytes[type][1])
5737 			return EINVAL;
5738 	} else if (obj.ipfo_size != fr_objbytes[type][1])
5739 		return EINVAL;
5740 #else
5741 	if (obj.ipfo_rev != IPFILTER_VERSION) {
5742 		error = fr_outcomptrans(&obj, ptr);
5743 		return error;
5744 	}
5745 
5746 	if ((fr_objbytes[type][0] & 1) != 0 &&
5747 	    obj.ipfo_size < fr_objbytes[type][1] ||
5748 	    obj.ipfo_size != fr_objbytes[type][1])
5749 		return EINVAL;
5750 #endif
5751 
5752 	error = COPYOUT((caddr_t)ptr, (caddr_t)obj.ipfo_ptr, obj.ipfo_size);
5753 	return error;
5754 }
5755 
5756 
5757 /* ------------------------------------------------------------------------ */
5758 /* Function:    fr_checkl4sum                                               */
5759 /* Returns:     int     - 0 = good, -1 = bad, 1 = cannot check              */
5760 /* Parameters:  fin(I) - pointer to packet information                      */
5761 /*                                                                          */
5762 /* If possible, calculate the layer 4 checksum for the packet.  If this is  */
5763 /* not possible, return without indicating a failure or success but in a    */
5764 /* way that is ditinguishable.                                              */
5765 /* ------------------------------------------------------------------------ */
5766 int fr_checkl4sum(fin)
5767 fr_info_t *fin;
5768 {
5769 	u_short sum, hdrsum, *csump;
5770 	udphdr_t *udp;
5771 	int dosum;
5772 	ipf_stack_t *ifs = fin->fin_ifs;
5773 
5774 #if SOLARIS && defined(_KERNEL) && (SOLARIS2 >= 6)
5775 	net_handle_t net_data_p;
5776 	if (fin->fin_v == 4)
5777 		net_data_p = ifs->ifs_ipf_ipv4;
5778 	else
5779 		net_data_p = ifs->ifs_ipf_ipv6;
5780 #endif
5781 
5782 	if ((fin->fin_flx & FI_NOCKSUM) != 0)
5783 		return 0;
5784 
5785 	/*
5786 	 * If the TCP packet isn't a fragment, isn't too short and otherwise
5787 	 * isn't already considered "bad", then validate the checksum.  If
5788 	 * this check fails then considered the packet to be "bad".
5789 	 */
5790 	if ((fin->fin_flx & (FI_FRAG|FI_SHORT|FI_BAD)) != 0)
5791 		return 1;
5792 
5793 	csump = NULL;
5794 	hdrsum = 0;
5795 	dosum = 0;
5796 	sum = 0;
5797 
5798 #if SOLARIS && defined(_KERNEL) && (SOLARIS2 >= 6)
5799 	ASSERT(fin->fin_m != NULL);
5800 	if (NET_IS_HCK_L4_FULL(net_data_p, fin->fin_m) ||
5801 	    NET_IS_HCK_L4_PART(net_data_p, fin->fin_m)) {
5802 			hdrsum = 0;
5803 			sum = 0;
5804 	} else {
5805 #endif
5806 		switch (fin->fin_p)
5807 		{
5808 		case IPPROTO_TCP :
5809 			csump = &((tcphdr_t *)fin->fin_dp)->th_sum;
5810 			dosum = 1;
5811 			break;
5812 
5813 		case IPPROTO_UDP :
5814 			udp = fin->fin_dp;
5815 			if (udp->uh_sum != 0) {
5816 				csump = &udp->uh_sum;
5817 				dosum = 1;
5818 			}
5819 			break;
5820 
5821 		case IPPROTO_ICMP :
5822 			csump = &((struct icmp *)fin->fin_dp)->icmp_cksum;
5823 			dosum = 1;
5824 			break;
5825 
5826 		default :
5827 			return 1;
5828 			/*NOTREACHED*/
5829 		}
5830 
5831 		if (csump != NULL)
5832 			hdrsum = *csump;
5833 
5834 		if (dosum)
5835 			sum = fr_cksum(fin->fin_m, fin->fin_ip,
5836 				       fin->fin_p, fin->fin_dp);
5837 #if SOLARIS && defined(_KERNEL) && (SOLARIS2 >= 6)
5838 	}
5839 #endif
5840 #if !defined(_KERNEL)
5841 	if (sum == hdrsum) {
5842 		FR_DEBUG(("checkl4sum: %hx == %hx\n", sum, hdrsum));
5843 	} else {
5844 		FR_DEBUG(("checkl4sum: %hx != %hx\n", sum, hdrsum));
5845 	}
5846 #endif
5847 	if (hdrsum == sum)
5848 		return 0;
5849 	return -1;
5850 }
5851 
5852 
5853 /* ------------------------------------------------------------------------ */
5854 /* Function:    fr_ifpfillv4addr                                            */
5855 /* Returns:     int     - 0 = address update, -1 = address not updated      */
5856 /* Parameters:  atype(I)   - type of network address update to perform      */
5857 /*              sin(I)     - pointer to source of address information       */
5858 /*              mask(I)    - pointer to source of netmask information       */
5859 /*              inp(I)     - pointer to destination address store           */
5860 /*              inpmask(I) - pointer to destination netmask store           */
5861 /*                                                                          */
5862 /* Given a type of network address update (atype) to perform, copy          */
5863 /* information from sin/mask into inp/inpmask.  If ipnmask is NULL then no  */
5864 /* netmask update is performed unless FRI_NETMASKED is passed as atype, in  */
5865 /* which case the operation fails.  For all values of atype other than      */
5866 /* FRI_NETMASKED, if inpmask is non-NULL then the mask is set to an all 1s  */
5867 /* value.                                                                   */
5868 /* ------------------------------------------------------------------------ */
5869 int fr_ifpfillv4addr(atype, sin, mask, inp, inpmask)
5870 int atype;
5871 struct sockaddr_in *sin, *mask;
5872 struct in_addr *inp, *inpmask;
5873 {
5874 	if (inpmask != NULL && atype != FRI_NETMASKED)
5875 		inpmask->s_addr = 0xffffffff;
5876 
5877 	if (atype == FRI_NETWORK || atype == FRI_NETMASKED) {
5878 		if (atype == FRI_NETMASKED) {
5879 			if (inpmask == NULL)
5880 				return -1;
5881 			inpmask->s_addr = mask->sin_addr.s_addr;
5882 		}
5883 		inp->s_addr = sin->sin_addr.s_addr & mask->sin_addr.s_addr;
5884 	} else {
5885 		inp->s_addr = sin->sin_addr.s_addr;
5886 	}
5887 	return 0;
5888 }
5889 
5890 
5891 #ifdef	USE_INET6
5892 /* ------------------------------------------------------------------------ */
5893 /* Function:    fr_ifpfillv6addr                                            */
5894 /* Returns:     int     - 0 = address update, -1 = address not updated      */
5895 /* Parameters:  atype(I)   - type of network address update to perform      */
5896 /*              sin(I)     - pointer to source of address information       */
5897 /*              mask(I)    - pointer to source of netmask information       */
5898 /*              inp(I)     - pointer to destination address store           */
5899 /*              inpmask(I) - pointer to destination netmask store           */
5900 /*                                                                          */
5901 /* Given a type of network address update (atype) to perform, copy          */
5902 /* information from sin/mask into inp/inpmask.  If ipnmask is NULL then no  */
5903 /* netmask update is performed unless FRI_NETMASKED is passed as atype, in  */
5904 /* which case the operation fails.  For all values of atype other than      */
5905 /* FRI_NETMASKED, if inpmask is non-NULL then the mask is set to an all 1s  */
5906 /* value.                                                                   */
5907 /* ------------------------------------------------------------------------ */
5908 int fr_ifpfillv6addr(atype, sin, mask, inp, inpmask)
5909 int atype;
5910 struct sockaddr_in6 *sin, *mask;
5911 struct in_addr *inp, *inpmask;
5912 {
5913 	i6addr_t *src, *dst, *and, *dmask;
5914 
5915 	src = (i6addr_t *)&sin->sin6_addr;
5916 	and = (i6addr_t *)&mask->sin6_addr;
5917 	dst = (i6addr_t *)inp;
5918 	dmask = (i6addr_t *)inpmask;
5919 
5920 	if (inpmask != NULL && atype != FRI_NETMASKED) {
5921 		dmask->i6[0] = 0xffffffff;
5922 		dmask->i6[1] = 0xffffffff;
5923 		dmask->i6[2] = 0xffffffff;
5924 		dmask->i6[3] = 0xffffffff;
5925 	}
5926 
5927 	if (atype == FRI_NETWORK || atype == FRI_NETMASKED) {
5928 		if (atype == FRI_NETMASKED) {
5929 			if (inpmask == NULL)
5930 				return -1;
5931 			dmask->i6[0] = and->i6[0];
5932 			dmask->i6[1] = and->i6[1];
5933 			dmask->i6[2] = and->i6[2];
5934 			dmask->i6[3] = and->i6[3];
5935 		}
5936 
5937 		dst->i6[0] = src->i6[0] & and->i6[0];
5938 		dst->i6[1] = src->i6[1] & and->i6[1];
5939 		dst->i6[2] = src->i6[2] & and->i6[2];
5940 		dst->i6[3] = src->i6[3] & and->i6[3];
5941 	} else {
5942 		dst->i6[0] = src->i6[0];
5943 		dst->i6[1] = src->i6[1];
5944 		dst->i6[2] = src->i6[2];
5945 		dst->i6[3] = src->i6[3];
5946 	}
5947 	return 0;
5948 }
5949 #endif
5950 
5951 
5952 /* ------------------------------------------------------------------------ */
5953 /* Function:    fr_matchtag                                                 */
5954 /* Returns:     0 == mismatch, 1 == match.                                  */
5955 /* Parameters:  tag1(I) - pointer to first tag to compare                   */
5956 /*              tag2(I) - pointer to second tag to compare                  */
5957 /*                                                                          */
5958 /* Returns true (non-zero) or false(0) if the two tag structures can be     */
5959 /* considered to be a match or not match, respectively.  The tag is 16      */
5960 /* bytes long (16 characters) but that is overlayed with 4 32bit ints so    */
5961 /* compare the ints instead, for speed. tag1 is the master of the           */
5962 /* comparison.  This function should only be called with both tag1 and tag2 */
5963 /* as non-NULL pointers.                                                    */
5964 /* ------------------------------------------------------------------------ */
5965 int fr_matchtag(tag1, tag2)
5966 ipftag_t *tag1, *tag2;
5967 {
5968 	if (tag1 == tag2)
5969 		return 1;
5970 
5971 	if ((tag1->ipt_num[0] == 0) && (tag2->ipt_num[0] == 0))
5972 		return 1;
5973 
5974 	if ((tag1->ipt_num[0] == tag2->ipt_num[0]) &&
5975 	    (tag1->ipt_num[1] == tag2->ipt_num[1]) &&
5976 	    (tag1->ipt_num[2] == tag2->ipt_num[2]) &&
5977 	    (tag1->ipt_num[3] == tag2->ipt_num[3]))
5978 		return 1;
5979 	return 0;
5980 }
5981 
5982 
5983 /* ------------------------------------------------------------------------ */
5984 /* Function:    fr_coalesce                                                 */
5985 /* Returns:     1 == success, -1 == failure, 0 == no change                 */
5986 /* Parameters:  fin(I) - pointer to packet information                      */
5987 /*                                                                          */
5988 /* Attempt to get all of the packet data into a single, contiguous buffer.  */
5989 /* If this call returns a failure then the buffers have also been freed.    */
5990 /* ------------------------------------------------------------------------ */
5991 int fr_coalesce(fin)
5992 fr_info_t *fin;
5993 {
5994 	ipf_stack_t *ifs = fin->fin_ifs;
5995 	if ((fin->fin_flx & FI_COALESCE) != 0)
5996 		return 1;
5997 
5998 	/*
5999 	 * If the mbuf pointers indicate that there is no mbuf to work with,
6000 	 * return but do not indicate success or failure.
6001 	 */
6002 	if (fin->fin_m == NULL || fin->fin_mp == NULL)
6003 		return 0;
6004 
6005 #if defined(_KERNEL)
6006 	if (fr_pullup(fin->fin_m, fin, fin->fin_plen) == NULL) {
6007 		IPF_BUMP(ifs->ifs_fr_badcoalesces[fin->fin_out]);
6008 # ifdef MENTAT
6009 		FREE_MB_T(*fin->fin_mp);
6010 # endif
6011 		*fin->fin_mp = NULL;
6012 		fin->fin_m = NULL;
6013 		return -1;
6014 	}
6015 #else
6016 	fin = fin;	/* LINT */
6017 #endif
6018 	return 1;
6019 }
6020 
6021 
6022 /*
6023  * The following table lists all of the tunable variables that can be
6024  * accessed via SIOCIPFGET/SIOCIPFSET/SIOCIPFGETNEXT.  The format of each row
6025  * in the table below is as follows:
6026  *
6027  * pointer to value, name of value, minimum, maximum, size of the value's
6028  *     container, value attribute flags
6029  *
6030  * For convienience, IPFT_RDONLY means the value is read-only, IPFT_WRDISABLED
6031  * means the value can only be written to when IPFilter is loaded but disabled.
6032  * The obvious implication is if neither of these are set then the value can be
6033  * changed at any time without harm.
6034  */
6035 ipftuneable_t lcl_ipf_tuneables[] = {
6036 	/* filtering */
6037 	{ { NULL },	"fr_flags",		0,	0xffffffff,
6038 			0,		0 },
6039 	{ { NULL },	"fr_active",		0,	0,
6040 			0,		IPFT_RDONLY },
6041 	{ { NULL },	"fr_control_forwarding",	0, 1,
6042 			0,		0 },
6043 	{ { NULL },	"fr_update_ipid",	0,	1,
6044 			0,		0 },
6045 	{ { NULL },	"fr_chksrc",		0,	1,
6046 			0,		0 },
6047 	{ { NULL },	"fr_minttl",		0,	1,
6048 			0,		0 },
6049 	{ { NULL }, 	"fr_icmpminfragmtu",	0,	1,
6050 			0,		0 },
6051 	{ { NULL },		"fr_pass",		0,	0xffffffff,
6052 			0,		0 },
6053 #if SOLARIS2 >= 10
6054 	{ { NULL },	"ipf_loopback",		0,	1,
6055 			0,		IPFT_WRDISABLED },
6056 #endif
6057 	/* state */
6058 	{ { NULL }, "fr_tcpidletimeout",	1,	0x7fffffff,
6059 			0,	IPFT_WRDISABLED },
6060 	{ { NULL },	"fr_tcpclosewait",	1,	0x7fffffff,
6061 			0,	IPFT_WRDISABLED },
6062 	{ { NULL },	"fr_tcplastack",	1,	0x7fffffff,
6063 			0,		IPFT_WRDISABLED },
6064 	{ { NULL },	"fr_tcptimeout",	1,	0x7fffffff,
6065 			0,		IPFT_WRDISABLED },
6066 	{ { NULL },	"fr_tcpclosed",		1,	0x7fffffff,
6067 			0,		IPFT_WRDISABLED },
6068 	{ { NULL },	"fr_tcphalfclosed",	1,	0x7fffffff,
6069 			0,	IPFT_WRDISABLED },
6070 	{ { NULL },	"fr_udptimeout",	1,	0x7fffffff,
6071 			0,		IPFT_WRDISABLED },
6072 	{ { NULL }, "fr_udpacktimeout",	1,	0x7fffffff,
6073 			0,	IPFT_WRDISABLED },
6074 	{ { NULL },	"fr_icmptimeout",	1,	0x7fffffff,
6075 			0,		IPFT_WRDISABLED },
6076 	{ { NULL }, "fr_icmpacktimeout",	1,	0x7fffffff,
6077 			0,	IPFT_WRDISABLED },
6078 	{ { NULL }, "fr_iptimeout",		1,	0x7fffffff,
6079 			0,		IPFT_WRDISABLED },
6080 	{ { NULL },	"fr_statemax",		1,	0x7fffffff,
6081 			0,		0 },
6082 	{ { NULL },	"fr_statesize",		1,	0x7fffffff,
6083 			0,		IPFT_WRDISABLED },
6084 	{ { NULL },	"fr_state_lock",	0,	1,
6085 			0,		IPFT_RDONLY },
6086 	{ { NULL }, "fr_state_maxbucket", 1,	0x7fffffff,
6087 			0,	IPFT_WRDISABLED },
6088 	{ { NULL }, "fr_state_maxbucket_reset",	0, 1,
6089 			0, IPFT_WRDISABLED },
6090 	{ { NULL },	"ipstate_logging",	0,	1,
6091 			0,	0 },
6092 	{ { NULL },	"state_flush_level_hi",	1,	100,
6093 			0,		0 },
6094 	{ { NULL },	"state_flush_level_lo",	1,	100,
6095 			0,		0 },
6096 	/* nat */
6097 	{ { NULL },		"fr_nat_lock",		0,	1,
6098 			0,		IPFT_RDONLY },
6099 	{ { NULL },	"ipf_nattable_sz",	1,	0x7fffffff,
6100 			0,	IPFT_WRDISABLED },
6101 	{ { NULL }, "ipf_nattable_max",	1,	0x7fffffff,
6102 			0,	0 },
6103 	{ { NULL },	"ipf_natrules_sz",	1,	0x7fffffff,
6104 			0,	IPFT_WRDISABLED },
6105 	{ { NULL },	"ipf_rdrrules_sz",	1,	0x7fffffff,
6106 			0,	IPFT_WRDISABLED },
6107 	{ { NULL },	"ipf_hostmap_sz",	1,	0x7fffffff,
6108 			0,		IPFT_WRDISABLED },
6109 	{ { NULL }, "fr_nat_maxbucket",	1,	0x7fffffff,
6110 			0,	IPFT_WRDISABLED },
6111 	{ { NULL },	"fr_nat_maxbucket_reset",	0, 1,
6112 			0,	IPFT_WRDISABLED },
6113 	{ { NULL },		"nat_logging",		0,	1,
6114 			0,		0 },
6115 	{ { NULL },	"fr_defnatage",		1,	0x7fffffff,
6116 			0,		IPFT_WRDISABLED },
6117 	{ { NULL },	"fr_defnatipage",	1,	0x7fffffff,
6118 			0,		IPFT_WRDISABLED },
6119 	{ { NULL }, "fr_defnaticmpage",	1,	0x7fffffff,
6120 			0,	IPFT_WRDISABLED },
6121 	{ { NULL },	"nat_flush_level_hi",	1,	100,
6122 			0,		0 },
6123 	{ { NULL },	"nat_flush_level_lo",	1,	100,
6124 			0,		0 },
6125 	/* frag */
6126 	{ { NULL },	"ipfr_size",		1,	0x7fffffff,
6127 			0,		IPFT_WRDISABLED },
6128 	{ { NULL },	"fr_ipfrttl",		1,	0x7fffffff,
6129 			0,		IPFT_WRDISABLED },
6130 #ifdef IPFILTER_LOG
6131 	/* log */
6132 	{ { NULL },	"ipl_suppress",		0,	1,
6133 			0,		0 },
6134 	{ { NULL },	"ipl_buffer_sz",	0,	0,
6135 			0,		IPFT_RDONLY },
6136 	{ { NULL },	"ipl_logmax",		0,	0x7fffffff,
6137 			0,		IPFT_WRDISABLED },
6138 	{ { NULL },	"ipl_logall",		0,	1,
6139 			0,		0 },
6140 	{ { NULL },	"ipl_logsize",		0,	0x80000,
6141 			0,		0 },
6142 #endif
6143 	{ { NULL },		NULL,			0,	0 }
6144 };
6145 
6146 static ipftuneable_t *
6147 tune_lookup(ipf_stack_t *ifs, char *name)
6148 {
6149     int i;
6150 
6151     for (i = 0; ifs->ifs_ipf_tuneables[i].ipft_name != NULL; i++) {
6152 	if (strcmp(ifs->ifs_ipf_tuneables[i].ipft_name, name) == 0)
6153 	    return (&ifs->ifs_ipf_tuneables[i]);
6154     }
6155     return (NULL);
6156 }
6157 
6158 #ifdef _KERNEL
6159 extern dev_info_t *ipf_dev_info;
6160 extern int ipf_property_update __P((dev_info_t *, ipf_stack_t *));
6161 #endif
6162 
6163 /* -------------------------------------------------------------------- */
6164 /* Function:	ipftuneable_setdefs()					*/
6165 /* Returns:		void						*/
6166 /* Parameters:	ifs - pointer to newly allocated IPF instance		*/
6167 /*				assigned to	IP instance		*/
6168 /*									*/
6169 /* Function initializes IPF instance variables. Function is invoked	*/
6170 /* from	ipftuneable_alloc(). ipftuneable_alloc() is called only one	*/
6171 /* time during IP instance lifetime - at the time of IP instance	*/
6172 /* creation. Anytime IP	instance is being created new private IPF	*/
6173 /* instance is allocated and assigned to it. The moment of IP 		*/
6174 /* instance creation is the right time to initialize those IPF 		*/
6175 /* variables.								*/
6176 /*									*/
6177 /* -------------------------------------------------------------------- */
6178 static void ipftuneable_setdefs(ipf_stack_t *ifs)
6179 {
6180 	ifs->ifs_ipfr_size = IPFT_SIZE;
6181 	ifs->ifs_fr_ipfrttl = 120;	/* 60 seconds */
6182 
6183 	/* it comes from fr_authinit() in IPF auth */
6184 	ifs->ifs_fr_authsize = FR_NUMAUTH;
6185 	ifs->ifs_fr_defaultauthage = 600;
6186 
6187 	/* it comes from fr_stateinit() in IPF state */
6188 	ifs->ifs_fr_tcpidletimeout = IPF_TTLVAL(3600 * 24 * 5);	/* five days */
6189 	ifs->ifs_fr_tcpclosewait = IPF_TTLVAL(TCP_MSL);
6190 	ifs->ifs_fr_tcplastack = IPF_TTLVAL(TCP_MSL);
6191 	ifs->ifs_fr_tcptimeout = IPF_TTLVAL(TCP_MSL);
6192 	ifs->ifs_fr_tcpclosed = IPF_TTLVAL(60);
6193 	ifs->ifs_fr_tcphalfclosed = IPF_TTLVAL(2 * 3600);	/* 2 hours */
6194 	ifs->ifs_fr_udptimeout = IPF_TTLVAL(120);
6195 	ifs->ifs_fr_udpacktimeout = IPF_TTLVAL(12);
6196 	ifs->ifs_fr_icmptimeout = IPF_TTLVAL(60);
6197 	ifs->ifs_fr_icmpacktimeout = IPF_TTLVAL(6);
6198 	ifs->ifs_fr_iptimeout = IPF_TTLVAL(60);
6199 	ifs->ifs_fr_statemax = IPSTATE_MAX;
6200 	ifs->ifs_fr_statesize = IPSTATE_SIZE;
6201 	ifs->ifs_fr_state_maxbucket_reset = 1;
6202 	ifs->ifs_state_flush_level_hi = ST_FLUSH_HI;
6203 	ifs->ifs_state_flush_level_lo = ST_FLUSH_LO;
6204 
6205 	/* it comes from fr_natinit() in ipnat */
6206 	ifs->ifs_ipf_nattable_sz = NAT_TABLE_SZ;
6207 	ifs->ifs_ipf_nattable_max = NAT_TABLE_MAX;
6208 	ifs->ifs_ipf_natrules_sz = NAT_SIZE;
6209 	ifs->ifs_ipf_rdrrules_sz = RDR_SIZE;
6210 	ifs->ifs_ipf_hostmap_sz = HOSTMAP_SIZE;
6211 	ifs->ifs_fr_nat_maxbucket_reset = 1;
6212 	ifs->ifs_fr_defnatage = DEF_NAT_AGE;
6213 	ifs->ifs_fr_defnatipage = 120;		/* 60 seconds */
6214 	ifs->ifs_fr_defnaticmpage = 6;		/* 3 seconds */
6215 	ifs->ifs_nat_flush_level_hi = NAT_FLUSH_HI;
6216 	ifs->ifs_nat_flush_level_lo = NAT_FLUSH_LO;
6217 
6218 #ifdef IPFILTER_LOG
6219 	/* it comes from fr_loginit() in IPF log */
6220 	ifs->ifs_ipl_suppress = 1;
6221 	ifs->ifs_ipl_logmax = IPL_LOGMAX;
6222 	ifs->ifs_ipl_logsize = IPFILTER_LOGSIZE;
6223 
6224 	/* from fr_natinit() */
6225 	ifs->ifs_nat_logging = 1;
6226 
6227 	/* from fr_stateinit() */
6228 	ifs->ifs_ipstate_logging = 1;
6229 #else
6230 	/* from fr_natinit() */
6231 	ifs->ifs_nat_logging = 0;
6232 
6233 	/* from fr_stateinit() */
6234 	ifs->ifs_ipstate_logging = 0;
6235 #endif
6236 	ifs->ifs_ipf_loopback = 0;
6237 
6238 }
6239 /*
6240  * Allocate a per-stack tuneable and copy in the names. Then
6241  * set it to point to each of the per-stack tunables.
6242  */
6243 void
6244 ipftuneable_alloc(ipf_stack_t *ifs)
6245 {
6246     ipftuneable_t *item;
6247 
6248     KMALLOCS(ifs->ifs_ipf_tuneables, ipftuneable_t *,
6249 	sizeof (lcl_ipf_tuneables));
6250     bcopy(lcl_ipf_tuneables, ifs->ifs_ipf_tuneables,
6251 	sizeof (lcl_ipf_tuneables));
6252 
6253 #define TUNE_SET(_ifs, _name, _field)			\
6254     item = tune_lookup((_ifs), (_name));		\
6255     if (item != NULL) {					\
6256 	item->ipft_una.ipftp_int = (unsigned int *)&((_ifs)->_field);	\
6257 	item->ipft_sz = sizeof ((_ifs)->_field);	\
6258     }
6259 
6260     TUNE_SET(ifs, "fr_flags", ifs_fr_flags);
6261     TUNE_SET(ifs, "fr_active", ifs_fr_active);
6262     TUNE_SET(ifs, "fr_control_forwarding", ifs_fr_control_forwarding);
6263     TUNE_SET(ifs, "fr_update_ipid", ifs_fr_update_ipid);
6264     TUNE_SET(ifs, "fr_chksrc", ifs_fr_chksrc);
6265     TUNE_SET(ifs, "fr_minttl", ifs_fr_minttl);
6266     TUNE_SET(ifs, "fr_icmpminfragmtu", ifs_fr_icmpminfragmtu);
6267     TUNE_SET(ifs, "fr_pass", ifs_fr_pass);
6268     TUNE_SET(ifs, "fr_tcpidletimeout", ifs_fr_tcpidletimeout);
6269     TUNE_SET(ifs, "fr_tcpclosewait", ifs_fr_tcpclosewait);
6270     TUNE_SET(ifs, "fr_tcplastack", ifs_fr_tcplastack);
6271     TUNE_SET(ifs, "fr_tcptimeout", ifs_fr_tcptimeout);
6272     TUNE_SET(ifs, "fr_tcpclosed", ifs_fr_tcpclosed);
6273     TUNE_SET(ifs, "fr_tcphalfclosed", ifs_fr_tcphalfclosed);
6274     TUNE_SET(ifs, "fr_udptimeout", ifs_fr_udptimeout);
6275     TUNE_SET(ifs, "fr_udpacktimeout", ifs_fr_udpacktimeout);
6276     TUNE_SET(ifs, "fr_icmptimeout", ifs_fr_icmptimeout);
6277     TUNE_SET(ifs, "fr_icmpacktimeout", ifs_fr_icmpacktimeout);
6278     TUNE_SET(ifs, "fr_iptimeout", ifs_fr_iptimeout);
6279     TUNE_SET(ifs, "fr_statemax", ifs_fr_statemax);
6280     TUNE_SET(ifs, "fr_statesize", ifs_fr_statesize);
6281     TUNE_SET(ifs, "fr_state_lock", ifs_fr_state_lock);
6282     TUNE_SET(ifs, "fr_state_maxbucket", ifs_fr_state_maxbucket);
6283     TUNE_SET(ifs, "fr_state_maxbucket_reset", ifs_fr_state_maxbucket_reset);
6284     TUNE_SET(ifs, "ipstate_logging", ifs_ipstate_logging);
6285     TUNE_SET(ifs, "fr_nat_lock", ifs_fr_nat_lock);
6286     TUNE_SET(ifs, "ipf_nattable_sz", ifs_ipf_nattable_sz);
6287     TUNE_SET(ifs, "ipf_nattable_max", ifs_ipf_nattable_max);
6288     TUNE_SET(ifs, "ipf_natrules_sz", ifs_ipf_natrules_sz);
6289     TUNE_SET(ifs, "ipf_rdrrules_sz", ifs_ipf_rdrrules_sz);
6290     TUNE_SET(ifs, "ipf_hostmap_sz", ifs_ipf_hostmap_sz);
6291     TUNE_SET(ifs, "fr_nat_maxbucket", ifs_fr_nat_maxbucket);
6292     TUNE_SET(ifs, "fr_nat_maxbucket_reset", ifs_fr_nat_maxbucket_reset);
6293     TUNE_SET(ifs, "nat_logging", ifs_nat_logging);
6294     TUNE_SET(ifs, "fr_defnatage", ifs_fr_defnatage);
6295     TUNE_SET(ifs, "fr_defnatipage", ifs_fr_defnatipage);
6296     TUNE_SET(ifs, "fr_defnaticmpage", ifs_fr_defnaticmpage);
6297     TUNE_SET(ifs, "nat_flush_level_hi", ifs_nat_flush_level_hi);
6298     TUNE_SET(ifs, "nat_flush_level_lo", ifs_nat_flush_level_lo);
6299     TUNE_SET(ifs, "state_flush_level_hi", ifs_state_flush_level_hi);
6300     TUNE_SET(ifs, "state_flush_level_lo", ifs_state_flush_level_lo);
6301     TUNE_SET(ifs, "ipfr_size", ifs_ipfr_size);
6302     TUNE_SET(ifs, "fr_ipfrttl", ifs_fr_ipfrttl);
6303     TUNE_SET(ifs, "ipf_loopback", ifs_ipf_loopback);
6304 #ifdef IPFILTER_LOG
6305     TUNE_SET(ifs, "ipl_suppress", ifs_ipl_suppress);
6306     TUNE_SET(ifs, "ipl_buffer_sz", ifs_ipl_buffer_sz);
6307     TUNE_SET(ifs, "ipl_logmax", ifs_ipl_logmax);
6308     TUNE_SET(ifs, "ipl_logall", ifs_ipl_logall);
6309     TUNE_SET(ifs, "ipl_logsize", ifs_ipl_logsize);
6310 #endif
6311 #undef TUNE_SET
6312 
6313 	ipftuneable_setdefs(ifs);
6314 
6315 #ifdef _KERNEL
6316     (void) ipf_property_update(ipf_dev_info, ifs);
6317 #endif
6318 }
6319 
6320 void
6321 ipftuneable_free(ipf_stack_t *ifs)
6322 {
6323 	KFREES(ifs->ifs_ipf_tuneables, sizeof (lcl_ipf_tuneables));
6324 	ifs->ifs_ipf_tuneables = NULL;
6325 }
6326 
6327 /* ------------------------------------------------------------------------ */
6328 /* Function:    fr_findtunebycookie                                         */
6329 /* Returns:     NULL = search failed, else pointer to tune struct           */
6330 /* Parameters:  cookie(I) - cookie value to search for amongst tuneables    */
6331 /*              next(O)   - pointer to place to store the cookie for the    */
6332 /*                          "next" tuneable, if it is desired.              */
6333 /*                                                                          */
6334 /* This function is used to walk through all of the existing tunables with  */
6335 /* successive calls.  It searches the known tunables for the one which has  */
6336 /* a matching value for "cookie" - ie its address.  When returning a match, */
6337 /* the next one to be found may be returned inside next.                    */
6338 /* ------------------------------------------------------------------------ */
6339 static ipftuneable_t *fr_findtunebycookie(cookie, next, ifs)
6340 void *cookie, **next;
6341 ipf_stack_t * ifs;
6342 {
6343 	ipftuneable_t *ta, **tap;
6344 
6345 	for (ta = ifs->ifs_ipf_tuneables; ta->ipft_name != NULL; ta++)
6346 		if (ta == cookie) {
6347 			if (next != NULL) {
6348 				/*
6349 				 * If the next entry in the array has a name
6350 				 * present, then return a pointer to it for
6351 				 * where to go next, else return a pointer to
6352 				 * the dynaminc list as a key to search there
6353 				 * next.  This facilitates a weak linking of
6354 				 * the two "lists" together.
6355 				 */
6356 				if ((ta + 1)->ipft_name != NULL)
6357 					*next = ta + 1;
6358 				else
6359 					*next = &ifs->ifs_ipf_tunelist;
6360 			}
6361 			return ta;
6362 		}
6363 
6364 	for (tap = &ifs->ifs_ipf_tunelist; (ta = *tap) != NULL; tap = &ta->ipft_next)
6365 		if (tap == cookie) {
6366 			if (next != NULL)
6367 				*next = &ta->ipft_next;
6368 			return ta;
6369 		}
6370 
6371 	if (next != NULL)
6372 		*next = NULL;
6373 	return NULL;
6374 }
6375 
6376 
6377 /* ------------------------------------------------------------------------ */
6378 /* Function:    fr_findtunebyname                                           */
6379 /* Returns:     NULL = search failed, else pointer to tune struct           */
6380 /* Parameters:  name(I) - name of the tuneable entry to find.               */
6381 /*                                                                          */
6382 /* Search the static array of tuneables and the list of dynamic tuneables   */
6383 /* for an entry with a matching name.  If we can find one, return a pointer */
6384 /* to the matching structure.                                               */
6385 /* ------------------------------------------------------------------------ */
6386 static ipftuneable_t *fr_findtunebyname(name, ifs)
6387 const char *name;
6388 ipf_stack_t *ifs;
6389 {
6390 	ipftuneable_t *ta;
6391 
6392 	for (ta = ifs->ifs_ipf_tuneables; ta->ipft_name != NULL; ta++)
6393 		if (!strcmp(ta->ipft_name, name)) {
6394 			return ta;
6395 		}
6396 
6397 	for (ta = ifs->ifs_ipf_tunelist; ta != NULL; ta = ta->ipft_next)
6398 		if (!strcmp(ta->ipft_name, name)) {
6399 			return ta;
6400 		}
6401 
6402 	return NULL;
6403 }
6404 
6405 
6406 /* ------------------------------------------------------------------------ */
6407 /* Function:    fr_addipftune                                               */
6408 /* Returns:     int - 0 == success, else failure                            */
6409 /* Parameters:  newtune - pointer to new tune struct to add to tuneables    */
6410 /*                                                                          */
6411 /* Appends the tune structure pointer to by "newtune" to the end of the     */
6412 /* current list of "dynamic" tuneable parameters.  Once added, the owner    */
6413 /* of the object is not expected to ever change "ipft_next".                */
6414 /* ------------------------------------------------------------------------ */
6415 int fr_addipftune(newtune, ifs)
6416 ipftuneable_t *newtune;
6417 ipf_stack_t *ifs;
6418 {
6419 	ipftuneable_t *ta, **tap;
6420 
6421 	ta = fr_findtunebyname(newtune->ipft_name, ifs);
6422 	if (ta != NULL)
6423 		return EEXIST;
6424 
6425 	for (tap = &ifs->ifs_ipf_tunelist; *tap != NULL; tap = &(*tap)->ipft_next)
6426 		;
6427 
6428 	newtune->ipft_next = NULL;
6429 	*tap = newtune;
6430 	return 0;
6431 }
6432 
6433 
6434 /* ------------------------------------------------------------------------ */
6435 /* Function:    fr_delipftune                                               */
6436 /* Returns:     int - 0 == success, else failure                            */
6437 /* Parameters:  oldtune - pointer to tune struct to remove from the list of */
6438 /*                        current dynamic tuneables                         */
6439 /*                                                                          */
6440 /* Search for the tune structure, by pointer, in the list of those that are */
6441 /* dynamically added at run time.  If found, adjust the list so that this   */
6442 /* structure is no longer part of it.                                       */
6443 /* ------------------------------------------------------------------------ */
6444 int fr_delipftune(oldtune, ifs)
6445 ipftuneable_t *oldtune;
6446 ipf_stack_t *ifs;
6447 {
6448 	ipftuneable_t *ta, **tap;
6449 
6450 	for (tap = &ifs->ifs_ipf_tunelist; (ta = *tap) != NULL; tap = &ta->ipft_next)
6451 		if (ta == oldtune) {
6452 			*tap = oldtune->ipft_next;
6453 			oldtune->ipft_next = NULL;
6454 			return 0;
6455 		}
6456 
6457 	return ESRCH;
6458 }
6459 
6460 
6461 /* ------------------------------------------------------------------------ */
6462 /* Function:    fr_ipftune                                                  */
6463 /* Returns:     int - 0 == success, else failure                            */
6464 /* Parameters:  cmd(I)  - ioctl command number                              */
6465 /*              data(I) - pointer to ioctl data structure                   */
6466 /*                                                                          */
6467 /* Implement handling of SIOCIPFGETNEXT, SIOCIPFGET and SIOCIPFSET.  These  */
6468 /* three ioctls provide the means to access and control global variables    */
6469 /* within IPFilter, allowing (for example) timeouts and table sizes to be   */
6470 /* changed without rebooting, reloading or recompiling.  The initialisation */
6471 /* and 'destruction' routines of the various components of ipfilter are all */
6472 /* each responsible for handling their own values being too big.            */
6473 /* ------------------------------------------------------------------------ */
6474 int fr_ipftune(cmd, data, ifs)
6475 ioctlcmd_t cmd;
6476 void *data;
6477 ipf_stack_t *ifs;
6478 {
6479 	ipftuneable_t *ta;
6480 	ipftune_t tu;
6481 	void *cookie;
6482 	int error;
6483 
6484 	error = fr_inobj(data, &tu, IPFOBJ_TUNEABLE);
6485 	if (error != 0)
6486 		return error;
6487 
6488 	tu.ipft_name[sizeof(tu.ipft_name) - 1] = '\0';
6489 	cookie = tu.ipft_cookie;
6490 	ta = NULL;
6491 
6492 	switch (cmd)
6493 	{
6494 	case SIOCIPFGETNEXT :
6495 		/*
6496 		 * If cookie is non-NULL, assume it to be a pointer to the last
6497 		 * entry we looked at, so find it (if possible) and return a
6498 		 * pointer to the next one after it.  The last entry in the
6499 		 * the table is a NULL entry, so when we get to it, set cookie
6500 		 * to NULL and return that, indicating end of list, erstwhile
6501 		 * if we come in with cookie set to NULL, we are starting anew
6502 		 * at the front of the list.
6503 		 */
6504 		if (cookie != NULL) {
6505 			ta = fr_findtunebycookie(cookie, &tu.ipft_cookie, ifs);
6506 		} else {
6507 			ta = ifs->ifs_ipf_tuneables;
6508 			tu.ipft_cookie = ta + 1;
6509 		}
6510 		if (ta != NULL) {
6511 			/*
6512 			 * Entry found, but does the data pointed to by that
6513 			 * row fit in what we can return?
6514 			 */
6515 			if (ta->ipft_sz > sizeof(tu.ipft_un))
6516 				return EINVAL;
6517 
6518 			tu.ipft_vlong = 0;
6519 			if (ta->ipft_sz == sizeof(u_long))
6520 				tu.ipft_vlong = *ta->ipft_plong;
6521 			else if (ta->ipft_sz == sizeof(u_int))
6522 				tu.ipft_vint = *ta->ipft_pint;
6523 			else if (ta->ipft_sz == sizeof(u_short))
6524 				tu.ipft_vshort = *ta->ipft_pshort;
6525 			else if (ta->ipft_sz == sizeof(u_char))
6526 				tu.ipft_vchar = *ta->ipft_pchar;
6527 
6528 			tu.ipft_sz = ta->ipft_sz;
6529 			tu.ipft_min = ta->ipft_min;
6530 			tu.ipft_max = ta->ipft_max;
6531 			tu.ipft_flags = ta->ipft_flags;
6532 			bcopy(ta->ipft_name, tu.ipft_name,
6533 			      MIN(sizeof(tu.ipft_name),
6534 				  strlen(ta->ipft_name) + 1));
6535 		}
6536 		error = fr_outobj(data, &tu, IPFOBJ_TUNEABLE);
6537 		break;
6538 
6539 	case SIOCIPFGET :
6540 	case SIOCIPFSET :
6541 		/*
6542 		 * Search by name or by cookie value for a particular entry
6543 		 * in the tuning paramter table.
6544 		 */
6545 		error = ESRCH;
6546 		if (cookie != NULL) {
6547 			ta = fr_findtunebycookie(cookie, NULL, ifs);
6548 			if (ta != NULL)
6549 				error = 0;
6550 		} else if (tu.ipft_name[0] != '\0') {
6551 			ta = fr_findtunebyname(tu.ipft_name, ifs);
6552 			if (ta != NULL)
6553 				error = 0;
6554 		}
6555 		if (error != 0)
6556 			break;
6557 
6558 		if (cmd == (ioctlcmd_t)SIOCIPFGET) {
6559 			/*
6560 			 * Fetch the tuning parameters for a particular value
6561 			 */
6562 			tu.ipft_vlong = 0;
6563 			if (ta->ipft_sz == sizeof(u_long))
6564 				tu.ipft_vlong = *ta->ipft_plong;
6565 			else if (ta->ipft_sz == sizeof(u_int))
6566 				tu.ipft_vint = *ta->ipft_pint;
6567 			else if (ta->ipft_sz == sizeof(u_short))
6568 				tu.ipft_vshort = *ta->ipft_pshort;
6569 			else if (ta->ipft_sz == sizeof(u_char))
6570 				tu.ipft_vchar = *ta->ipft_pchar;
6571 			tu.ipft_cookie = ta;
6572 			tu.ipft_sz = ta->ipft_sz;
6573 			tu.ipft_min = ta->ipft_min;
6574 			tu.ipft_max = ta->ipft_max;
6575 			tu.ipft_flags = ta->ipft_flags;
6576 			error = fr_outobj(data, &tu, IPFOBJ_TUNEABLE);
6577 
6578 		} else if (cmd == (ioctlcmd_t)SIOCIPFSET) {
6579 			/*
6580 			 * Set an internal parameter.  The hard part here is
6581 			 * getting the new value safely and correctly out of
6582 			 * the kernel (given we only know its size, not type.)
6583 			 */
6584 			u_long in;
6585 
6586 			if (((ta->ipft_flags & IPFT_WRDISABLED) != 0) &&
6587 			    (ifs->ifs_fr_running > 0)) {
6588 				error = EBUSY;
6589 				break;
6590 			}
6591 
6592 			in = tu.ipft_vlong;
6593 			if (in < ta->ipft_min || in > ta->ipft_max) {
6594 				error = EINVAL;
6595 				break;
6596 			}
6597 
6598 			if (ta->ipft_sz == sizeof(u_long)) {
6599 				tu.ipft_vlong = *ta->ipft_plong;
6600 				*ta->ipft_plong = in;
6601 			} else if (ta->ipft_sz == sizeof(u_int)) {
6602 				tu.ipft_vint = *ta->ipft_pint;
6603 				*ta->ipft_pint = (u_int)(in & 0xffffffff);
6604 			} else if (ta->ipft_sz == sizeof(u_short)) {
6605 				tu.ipft_vshort = *ta->ipft_pshort;
6606 				*ta->ipft_pshort = (u_short)(in & 0xffff);
6607 			} else if (ta->ipft_sz == sizeof(u_char)) {
6608 				tu.ipft_vchar = *ta->ipft_pchar;
6609 				*ta->ipft_pchar = (u_char)(in & 0xff);
6610 			}
6611 			error = fr_outobj(data, &tu, IPFOBJ_TUNEABLE);
6612 		}
6613 		break;
6614 
6615 	default :
6616 		error = EINVAL;
6617 		break;
6618 	}
6619 
6620 	return error;
6621 }
6622 
6623 
6624 /* ------------------------------------------------------------------------ */
6625 /* Function:    fr_initialise                                               */
6626 /* Returns:     int - 0 == success,  < 0 == failure                         */
6627 /* Parameters:  None.                                                       */
6628 /*                                                                          */
6629 /* Call of the initialise functions for all the various subsystems inside   */
6630 /* of IPFilter.  If any of them should fail, return immeadiately a failure  */
6631 /* BUT do not try to recover from the error here.                           */
6632 /* ------------------------------------------------------------------------ */
6633 int fr_initialise(ifs)
6634 ipf_stack_t *ifs;
6635 {
6636 	int i;
6637 
6638 #ifdef IPFILTER_LOG
6639 	i = fr_loginit(ifs);
6640 	if (i < 0)
6641 		return -10 + i;
6642 #endif
6643 	i = fr_natinit(ifs);
6644 	if (i < 0)
6645 		return -20 + i;
6646 
6647 	i = fr_stateinit(ifs);
6648 	if (i < 0)
6649 		return -30 + i;
6650 
6651 	i = fr_authinit(ifs);
6652 	if (i < 0)
6653 		return -40 + i;
6654 
6655 	i = fr_fraginit(ifs);
6656 	if (i < 0)
6657 		return -50 + i;
6658 
6659 	i = appr_init(ifs);
6660 	if (i < 0)
6661 		return -60 + i;
6662 
6663 #ifdef IPFILTER_SYNC
6664 	i = ipfsync_init(ifs);
6665 	if (i < 0)
6666 		return -70 + i;
6667 #endif
6668 #ifdef IPFILTER_SCAN
6669 	i = ipsc_init(ifs);
6670 	if (i < 0)
6671 		return -80 + i;
6672 #endif
6673 #ifdef IPFILTER_LOOKUP
6674 	i = ip_lookup_init(ifs);
6675 	if (i < 0)
6676 		return -90 + i;
6677 #endif
6678 #ifdef IPFILTER_COMPILED
6679 	ipfrule_add(ifs);
6680 #endif
6681 	return 0;
6682 }
6683 
6684 
6685 /* ------------------------------------------------------------------------ */
6686 /* Function:    fr_deinitialise                                             */
6687 /* Returns:     None.                                                       */
6688 /* Parameters:  None.                                                       */
6689 /*                                                                          */
6690 /* Call all the various subsystem cleanup routines to deallocate memory or  */
6691 /* destroy locks or whatever they've done that they need to now undo.       */
6692 /* The order here IS important as there are some cross references of        */
6693 /* internal data structures.                                                */
6694 /* ------------------------------------------------------------------------ */
6695 void fr_deinitialise(ifs)
6696 ipf_stack_t *ifs;
6697 {
6698 	fr_fragunload(ifs);
6699 	fr_authunload(ifs);
6700 	fr_natunload(ifs);
6701 	fr_stateunload(ifs);
6702 #ifdef IPFILTER_SCAN
6703 	fr_scanunload(ifs);
6704 #endif
6705 	appr_unload(ifs);
6706 
6707 #ifdef IPFILTER_COMPILED
6708 	ipfrule_remove(ifs);
6709 #endif
6710 
6711 	(void) frflush(IPL_LOGIPF, 0, FR_INQUE|FR_OUTQUE|FR_INACTIVE, ifs);
6712 	(void) frflush(IPL_LOGIPF, 0, FR_INQUE|FR_OUTQUE, ifs);
6713 	(void) frflush(IPL_LOGCOUNT, 0, FR_INQUE|FR_OUTQUE|FR_INACTIVE, ifs);
6714 	(void) frflush(IPL_LOGCOUNT, 0, FR_INQUE|FR_OUTQUE, ifs);
6715 
6716 #ifdef IPFILTER_LOOKUP
6717 	ip_lookup_unload(ifs);
6718 #endif
6719 
6720 #ifdef IPFILTER_LOG
6721 	fr_logunload(ifs);
6722 #endif
6723 }
6724 
6725 
6726 /* ------------------------------------------------------------------------ */
6727 /* Function:    fr_zerostats                                                */
6728 /* Returns:     int - 0 = success, else failure                             */
6729 /* Parameters:  data(O) - pointer to pointer for copying data back to       */
6730 /*                                                                          */
6731 /* Copies the current statistics out to userspace and then zero's the       */
6732 /* current ones in the kernel. The lock is only held across the bzero() as  */
6733 /* the copyout may result in paging (ie network activity.)                  */
6734 /* ------------------------------------------------------------------------ */
6735 int	fr_zerostats(data, ifs)
6736 caddr_t	data;
6737 ipf_stack_t *ifs;
6738 {
6739 	friostat_t fio;
6740 	int error;
6741 
6742 	fr_getstat(&fio, ifs);
6743 	error = copyoutptr(&fio, data, sizeof(fio));
6744 	if (error)
6745 		return EFAULT;
6746 
6747 	WRITE_ENTER(&ifs->ifs_ipf_mutex);
6748 	bzero((char *)ifs->ifs_frstats, sizeof(*ifs->ifs_frstats) * 2);
6749 	RWLOCK_EXIT(&ifs->ifs_ipf_mutex);
6750 
6751 	return 0;
6752 }
6753 
6754 
6755 #ifdef _KERNEL
6756 /* ------------------------------------------------------------------------ */
6757 /* Function:    fr_resolvedest                                              */
6758 /* Returns:     Nil                                                         */
6759 /* Parameters:  fdp(IO) - pointer to destination information to resolve     */
6760 /*              v(I)    - IP protocol version to match                      */
6761 /*                                                                          */
6762 /* Looks up an interface name in the frdest structure pointed to by fdp and */
6763 /* if a matching name can be found for the particular IP protocol version   */
6764 /* then store the interface pointer in the frdest struct.  If no match is   */
6765 /* found, then set the interface pointer to be -1 as NULL is considered to  */
6766 /* indicate there is no information at all in the structure.                */
6767 /* ------------------------------------------------------------------------ */
6768 void fr_resolvedest(fdp, v, ifs)
6769 frdest_t *fdp;
6770 int v;
6771 ipf_stack_t *ifs;
6772 {
6773 	fdp->fd_ifp = NULL;
6774 
6775   	if (*fdp->fd_ifname != '\0') {
6776  		fdp->fd_ifp = GETIFP(fdp->fd_ifname, v, ifs);
6777 		if (fdp->fd_ifp == NULL)
6778 			fdp->fd_ifp = (void *)-1;
6779 	}
6780 }
6781 #endif /* _KERNEL */
6782 
6783 
6784 /* ------------------------------------------------------------------------ */
6785 /* Function:    fr_resolvenic                                               */
6786 /* Returns:     void* - NULL = wildcard name, -1 = failed to find NIC, else */
6787 /*                      pointer to interface structure for NIC              */
6788 /* Parameters:  name(I) - complete interface name                           */
6789 /*              v(I)    - IP protocol version                               */
6790 /*                                                                          */
6791 /* Look for a network interface structure that firstly has a matching name  */
6792 /* to that passed in and that is also being used for that IP protocol       */
6793 /* version (necessary on some platforms where there are separate listings   */
6794 /* for both IPv4 and IPv6 on the same physical NIC.                         */
6795 /*                                                                          */
6796 /* One might wonder why name gets terminated with a \0 byte in here.  The   */
6797 /* reason is an interface name could get into the kernel structures of ipf  */
6798 /* in any number of ways and so long as they all use the same sized array   */
6799 /* to put the name in, it makes sense to ensure it gets null terminated     */
6800 /* before it is used for its intended purpose - finding its match in the    */
6801 /* kernel's list of configured interfaces.                                  */
6802 /*                                                                          */
6803 /* NOTE: This SHOULD ONLY be used with IPFilter structures that have an     */
6804 /*       array for the name that is LIFNAMSIZ bytes (at least) in length.   */
6805 /* ------------------------------------------------------------------------ */
6806 void *fr_resolvenic(name, v, ifs)
6807 char *name;
6808 int v;
6809 ipf_stack_t *ifs;
6810 {
6811 	void *nic;
6812 
6813 	if (name[0] == '\0')
6814 		return NULL;
6815 
6816 	if ((name[1] == '\0') && ((name[0] == '-') || (name[0] == '*'))) {
6817 		return NULL;
6818 	}
6819 
6820 	name[LIFNAMSIZ - 1] = '\0';
6821 
6822 	nic = GETIFP(name, v, ifs);
6823 	if (nic == NULL)
6824 		nic = (void *)-1;
6825 	return nic;
6826 }
6827 
6828 
6829 /* ------------------------------------------------------------------------ */
6830 /* Function:    ipf_expiretokens                                            */
6831 /* Returns:     None.                                                       */
6832 /* Parameters:  ifs - ipf stack instance                                    */
6833 /*                                                                          */
6834 /* This function is run every ipf tick to see if there are any tokens that  */
6835 /* have been held for too long and need to be freed up.                     */
6836 /* ------------------------------------------------------------------------ */
6837 void ipf_expiretokens(ifs)
6838 ipf_stack_t *ifs;
6839 {
6840 	ipftoken_t *it;
6841 
6842 	WRITE_ENTER(&ifs->ifs_ipf_tokens);
6843 	while ((it = ifs->ifs_ipftokenhead) != NULL) {
6844 		if (it->ipt_die > ifs->ifs_fr_ticks)
6845 			break;
6846 
6847 		ipf_freetoken(it, ifs);
6848 	}
6849 	RWLOCK_EXIT(&ifs->ifs_ipf_tokens);
6850 }
6851 
6852 
6853 /* ------------------------------------------------------------------------ */
6854 /* Function:    ipf_deltoken                                                */
6855 /* Returns:     int - 0 = success, else error                               */
6856 /* Parameters:  type(I) - the token type to match                           */
6857 /*              uid(I)  - uid owning the token                              */
6858 /*              ptr(I)  - context pointer for the token                     */
6859 /*              ifs - ipf stack instance                                    */
6860 /*                                                                          */
6861 /* This function looks for a a token in the current list that matches up    */
6862 /* the fields (type, uid, ptr).  If none is found, ESRCH is returned, else  */
6863 /* call ipf_freetoken() to remove it from the list.                         */
6864 /* ------------------------------------------------------------------------ */
6865 int ipf_deltoken(type, uid, ptr, ifs)
6866 int type, uid;
6867 void *ptr;
6868 ipf_stack_t *ifs;
6869 {
6870 	ipftoken_t *it;
6871 	int error = ESRCH;
6872 
6873 	WRITE_ENTER(&ifs->ifs_ipf_tokens);
6874 	for (it = ifs->ifs_ipftokenhead; it != NULL; it = it->ipt_next)
6875 		if (ptr == it->ipt_ctx && type == it->ipt_type &&
6876 		    uid == it->ipt_uid) {
6877 			ipf_freetoken(it, ifs);
6878 			error = 0;
6879 			break;
6880 	}
6881 	RWLOCK_EXIT(&ifs->ifs_ipf_tokens);
6882 
6883 	return error;
6884 }
6885 
6886 
6887 /* ------------------------------------------------------------------------ */
6888 /* Function:    ipf_unlinktoken                                             */
6889 /* Returns:     None.                                                       */
6890 /* Parameters:  token(I) - pointer to token structure                       */
6891 /*              ifs - ipf stack instance                                    */
6892 /*                                                                          */
6893 /* This function unlinks a token structure from the linked list of tokens   */
6894 /* that it belongs to.  The head pointer never needs to be explicitly       */
6895 /* adjusted, but the tail does due to the linked list implementation.       */
6896 /* ------------------------------------------------------------------------ */
6897 static void ipf_unlinktoken(token, ifs)
6898 ipftoken_t *token;
6899 ipf_stack_t *ifs;
6900 {
6901 
6902 	if (ifs->ifs_ipftokentail == &token->ipt_next)
6903 		ifs->ifs_ipftokentail = token->ipt_pnext;
6904 
6905 	*token->ipt_pnext = token->ipt_next;
6906 	if (token->ipt_next != NULL)
6907 		token->ipt_next->ipt_pnext = token->ipt_pnext;
6908 }
6909 
6910 
6911 /* ------------------------------------------------------------------------ */
6912 /* Function:    ipf_findtoken                                               */
6913 /* Returns:     ipftoken_t * - NULL if no memory, else pointer to token     */
6914 /* Parameters:  type(I) - the token type to match                           */
6915 /*              uid(I) - uid owning the token                               */
6916 /*              ptr(I) - context pointer for the token                      */
6917 /*              ifs - ipf stack instance                                    */
6918 /*                                                                          */
6919 /* This function looks for a live token in the list of current tokens that  */
6920 /* matches the tuple (type, uid, ptr).  If one cannot be found then one is  */
6921 /* allocated.  If one is found then it is moved to the top of the list of   */
6922 /* currently active tokens.                                                 */
6923 /*                                                                          */
6924 /* NOTE: It is by design that this function returns holding a read lock on  */
6925 /*       ipf_tokens.  Callers must make sure they release it!               */
6926 /* ------------------------------------------------------------------------ */
6927 ipftoken_t *ipf_findtoken(type, uid, ptr, ifs)
6928 int type, uid;
6929 void *ptr;
6930 ipf_stack_t *ifs;
6931 {
6932 	ipftoken_t *it, *new;
6933 
6934 	KMALLOC(new, ipftoken_t *);
6935 
6936 	WRITE_ENTER(&ifs->ifs_ipf_tokens);
6937 	for (it = ifs->ifs_ipftokenhead; it != NULL; it = it->ipt_next) {
6938 		if (it->ipt_alive == 0)
6939 			continue;
6940 		if (ptr == it->ipt_ctx && type == it->ipt_type &&
6941 		    uid == it->ipt_uid)
6942 			break;
6943 	}
6944 
6945 	if (it == NULL) {
6946 		it = new;
6947 		new = NULL;
6948 		if (it == NULL)
6949 			return NULL;
6950 		it->ipt_data = NULL;
6951 		it->ipt_ctx = ptr;
6952 		it->ipt_uid = uid;
6953 		it->ipt_type = type;
6954 		it->ipt_next = NULL;
6955 		it->ipt_alive = 1;
6956 	} else {
6957 		if (new != NULL) {
6958 			KFREE(new);
6959 			new = NULL;
6960 		}
6961 
6962 		ipf_unlinktoken(it, ifs);
6963 	}
6964 	it->ipt_pnext = ifs->ifs_ipftokentail;
6965 	*ifs->ifs_ipftokentail = it;
6966 	ifs->ifs_ipftokentail = &it->ipt_next;
6967 	it->ipt_next = NULL;
6968 
6969 	it->ipt_die = ifs->ifs_fr_ticks + 2;
6970 
6971 	MUTEX_DOWNGRADE(&ifs->ifs_ipf_tokens);
6972 
6973 	return it;
6974 }
6975 
6976 
6977 /* ------------------------------------------------------------------------ */
6978 /* Function:    ipf_freetoken                                               */
6979 /* Returns:     None.                                                       */
6980 /* Parameters:  token(I) - pointer to token structure                       */
6981 /*              ifs - ipf stack instance                                    */
6982 /*                                                                          */
6983 /* This function unlinks a token from the linked list and on the path to    */
6984 /* free'ing the data, it calls the dereference function that is associated  */
6985 /* with the type of data pointed to by the token as it is considered to     */
6986 /* hold a reference to it.                                                  */
6987 /* ------------------------------------------------------------------------ */
6988 void ipf_freetoken(token, ifs)
6989 ipftoken_t *token;
6990 ipf_stack_t *ifs;
6991 {
6992 	void *data, **datap;
6993 
6994 	ipf_unlinktoken(token, ifs);
6995 
6996 	data = token->ipt_data;
6997 	datap = &data;
6998 
6999 	if ((data != NULL) && (data != (void *)-1)) {
7000 		switch (token->ipt_type)
7001 		{
7002 		case IPFGENITER_IPF :
7003 			(void)fr_derefrule((frentry_t **)datap, ifs);
7004 			break;
7005 		case IPFGENITER_IPNAT :
7006 			WRITE_ENTER(&ifs->ifs_ipf_nat);
7007 			fr_ipnatderef((ipnat_t **)datap, ifs);
7008 			RWLOCK_EXIT(&ifs->ifs_ipf_nat);
7009 			break;
7010 		case IPFGENITER_NAT :
7011 			fr_natderef((nat_t **)datap, ifs);
7012 			break;
7013 		case IPFGENITER_STATE :
7014 			fr_statederef((ipstate_t **)datap, ifs);
7015 			break;
7016 		case IPFGENITER_FRAG :
7017 			fr_fragderef((ipfr_t **)datap, &ifs->ifs_ipf_frag, ifs);
7018 			break;
7019 		case IPFGENITER_NATFRAG :
7020  			fr_fragderef((ipfr_t **)datap,
7021 				     &ifs->ifs_ipf_natfrag, ifs);
7022 			break;
7023 		case IPFGENITER_HOSTMAP :
7024 			WRITE_ENTER(&ifs->ifs_ipf_nat);
7025 			fr_hostmapdel((hostmap_t **)datap);
7026 			RWLOCK_EXIT(&ifs->ifs_ipf_nat);
7027 			break;
7028 		default :
7029 			(void) ip_lookup_iterderef(token->ipt_type, data, ifs);
7030 			break;
7031 		}
7032 	}
7033 
7034 	KFREE(token);
7035 }
7036 
7037 
7038 /* ------------------------------------------------------------------------ */
7039 /* Function:    ipf_getnextrule                                             */
7040 /* Returns:     int - 0 = success, else error                               */
7041 /* Parameters:  t(I)   - pointer to destination information to resolve      */
7042 /*              ptr(I) - pointer to ipfobj_t to copyin from user space      */
7043 /*              ifs - ipf stack instance                                    */
7044 /*                                                                          */
7045 /* This function's first job is to bring in the ipfruleiter_t structure via */
7046 /* the ipfobj_t structure to determine what should be the next rule to      */
7047 /* return. Once the ipfruleiter_t has been brought in, it then tries to     */
7048 /* find the 'next rule'.  This may include searching rule group lists or    */
7049 /* just be as simple as looking at the 'next' field in the rule structure.  */
7050 /* When we have found the rule to return, increase its reference count and  */
7051 /* if we used an existing rule to get here, decrease its reference count.   */
7052 /* ------------------------------------------------------------------------ */
7053 int ipf_getnextrule(t, ptr, ifs)
7054 ipftoken_t *t;
7055 void *ptr;
7056 ipf_stack_t *ifs;
7057 {
7058 	frentry_t *fr, *next, zero;
7059 	int error, out, count;
7060 	ipfruleiter_t it;
7061 	frgroup_t *fg;
7062 	char *dst;
7063 
7064 	if (t == NULL || ptr == NULL)
7065 		return EFAULT;
7066 	error = fr_inobj(ptr, &it, IPFOBJ_IPFITER);
7067 	if (error != 0)
7068 		return error;
7069 	if ((it.iri_ver != AF_INET) && (it.iri_ver != AF_INET6))
7070 		return EINVAL;
7071 	if ((it.iri_inout < 0) || (it.iri_inout > 3))
7072 		return EINVAL;
7073 	if (it.iri_nrules == 0)
7074 		return EINVAL;
7075 	if ((it.iri_active != 0) && (it.iri_active != 1))
7076 		return EINVAL;
7077 	if (it.iri_rule == NULL)
7078 		return EFAULT;
7079 
7080 	/*
7081 	 * Use bitmask on it.iri_inout to determine direction.
7082 	 * F_OUT (1) and F_ACOUT (3) mask to out = 1, while
7083 	 * F_IN (0) and F_ACIN (2) mask to out = 0.
7084 	 */
7085 	out = it.iri_inout & F_OUT;
7086 	READ_ENTER(&ifs->ifs_ipf_mutex);
7087 
7088 	/*
7089 	 * Retrieve "previous" entry from token and find the next entry.
7090 	 */
7091 	fr = t->ipt_data;
7092 	if (fr == NULL) {
7093 		if (*it.iri_group == '\0') {
7094 			/*
7095 			 * Use bitmask again to determine accounting or not.
7096 			 * F_ACIN will mask to accounting cases F_ACIN (2)
7097 			 * or F_ACOUT (3), but not F_IN or F_OUT.
7098 			 */
7099 			if ((it.iri_inout & F_ACIN) != 0) {
7100 				if (it.iri_ver == AF_INET)
7101 					next = ifs->ifs_ipacct
7102 					    [out][it.iri_active];
7103 				else
7104 					next = ifs->ifs_ipacct6
7105 					    [out][it.iri_active];
7106 			} else {
7107 				if (it.iri_ver == AF_INET)
7108 					next = ifs->ifs_ipfilter
7109 					    [out][it.iri_active];
7110 				else
7111 					next = ifs->ifs_ipfilter6
7112 					    [out][it.iri_active];
7113 			}
7114 		} else {
7115 			fg = fr_findgroup(it.iri_group, IPL_LOGIPF,
7116 					  it.iri_active, NULL, ifs);
7117 			if (fg != NULL)
7118 				next = fg->fg_start;
7119 			else
7120 				next = NULL;
7121 		}
7122 	} else {
7123 		next = fr->fr_next;
7124 	}
7125 
7126 	dst = (char *)it.iri_rule;
7127 	/*
7128 	 * The ipfruleiter may ask for more than 1 rule at a time to be
7129 	 * copied out, so long as that many exist in the list to start with!
7130 	 */
7131 	for (count = it.iri_nrules; count > 0; count--) {
7132 		/*
7133 		 * If we found an entry, add reference to it and update token.
7134 		 * Otherwise, zero out data to be returned and NULL out token.
7135 		 */
7136 		if (next != NULL) {
7137 			MUTEX_ENTER(&next->fr_lock);
7138 			next->fr_ref++;
7139 			MUTEX_EXIT(&next->fr_lock);
7140 			t->ipt_data = next;
7141 		} else {
7142 			bzero(&zero, sizeof(zero));
7143 			next = &zero;
7144 			t->ipt_data = NULL;
7145 		}
7146 
7147 		/*
7148 		 * Now that we have ref, it's save to give up lock.
7149 		 */
7150 		RWLOCK_EXIT(&ifs->ifs_ipf_mutex);
7151 
7152 		/*
7153 		 * Copy out data and clean up references and token as needed.
7154 		 */
7155 		error = COPYOUT(next, dst, sizeof(*next));
7156 		if (error != 0)
7157 			error = EFAULT;
7158 		if (t->ipt_data == NULL) {
7159 			ipf_freetoken(t, ifs);
7160 			break;
7161 		} else {
7162 			if (fr != NULL)
7163 				(void) fr_derefrule(&fr, ifs);
7164 			if (next->fr_data != NULL) {
7165 				dst += sizeof(*next);
7166 				error = COPYOUT(next->fr_data, dst,
7167 						next->fr_dsize);
7168 				if (error != 0)
7169 					error = EFAULT;
7170 				else
7171 					dst += next->fr_dsize;
7172 			}
7173 			if (next->fr_next == NULL) {
7174 				ipf_freetoken(t, ifs);
7175 				break;
7176 			}
7177 		}
7178 
7179 		if ((count == 1) || (error != 0))
7180 			break;
7181 
7182 		READ_ENTER(&ifs->ifs_ipf_mutex);
7183 		fr = next;
7184 		next = fr->fr_next;
7185 	}
7186 
7187 	return error;
7188 }
7189 
7190 
7191 /* ------------------------------------------------------------------------ */
7192 /* Function:    fr_frruleiter                                               */
7193 /* Returns:     int - 0 = success, else error                               */
7194 /* Parameters:  data(I) - the token type to match                           */
7195 /*              uid(I) - uid owning the token                               */
7196 /*              ptr(I) - context pointer for the token                      */
7197 /*              ifs - ipf stack instance                                    */
7198 /*                                                                          */
7199 /* This function serves as a stepping stone between fr_ipf_ioctl and        */
7200 /* ipf_getnextrule.  It's role is to find the right token in the kernel for */
7201 /* the process doing the ioctl and use that to ask for the next rule.       */
7202 /* ------------------------------------------------------------------------ */
7203 int ipf_frruleiter(data, uid, ctx, ifs)
7204 void *data, *ctx;
7205 int uid;
7206 ipf_stack_t *ifs;
7207 {
7208 	ipftoken_t *token;
7209 	int error;
7210 
7211 	token = ipf_findtoken(IPFGENITER_IPF, uid, ctx, ifs);
7212 	if (token != NULL)
7213 		error = ipf_getnextrule(token, data, ifs);
7214 	else
7215 		error = EFAULT;
7216 	RWLOCK_EXIT(&ifs->ifs_ipf_tokens);
7217 
7218 	return error;
7219 }
7220 
7221 
7222 /* ------------------------------------------------------------------------ */
7223 /* Function:    ipf_geniter                                                 */
7224 /* Returns:     int - 0 = success, else error                               */
7225 /* Parameters:  token(I) - pointer to ipftoken structure                    */
7226 /*              itp(I) - pointer to ipfgeniter structure                    */
7227 /*              ifs - ipf stack instance                                    */
7228 /*                                                                          */
7229 /* Generic iterator called from ipf_genericiter.  Currently only used for   */
7230 /* walking through list of fragments.                                       */
7231 /* ------------------------------------------------------------------------ */
7232 int ipf_geniter(token, itp, ifs)
7233 ipftoken_t *token;
7234 ipfgeniter_t *itp;
7235 ipf_stack_t *ifs;
7236 {
7237 	int error;
7238 
7239 	switch (itp->igi_type)
7240 	{
7241 	case IPFGENITER_FRAG :
7242 		error = fr_nextfrag(token, itp, &ifs->ifs_ipfr_list,
7243 				    &ifs->ifs_ipfr_tail, &ifs->ifs_ipf_frag,
7244 				    ifs);
7245 		break;
7246 	default :
7247 		error = EINVAL;
7248 		break;
7249 	}
7250 
7251 	return error;
7252 }
7253 
7254 
7255 /* ------------------------------------------------------------------------ */
7256 /* Function:    ipf_genericiter                                             */
7257 /* Returns:     int - 0 = success, else error                               */
7258 /* Parameters:  data(I) - the token type to match                           */
7259 /*              uid(I) - uid owning the token                               */
7260 /*              ptr(I) - context pointer for the token                      */
7261 /*              ifs - ipf stack instance                                    */
7262 /*                                                                          */
7263 /* This function serves as a stepping stone between fr_ipf_ioctl and        */
7264 /* ipf_geniter when handling SIOCGENITER.  It's role is to find the right   */
7265 /* token in the kernel for the process using the ioctl, and to use that     */
7266 /* token when calling ipf_geniter.                                          */
7267 /* ------------------------------------------------------------------------ */
7268 int ipf_genericiter(data, uid, ctx, ifs)
7269 void *data, *ctx;
7270 int uid;
7271 ipf_stack_t *ifs;
7272 {
7273 	ipftoken_t *token;
7274 	ipfgeniter_t iter;
7275 	int error;
7276 
7277 	error = fr_inobj(data, &iter, IPFOBJ_GENITER);
7278 	if (error != 0)
7279 		return error;
7280 
7281 	token = ipf_findtoken(iter.igi_type, uid, ctx, ifs);
7282 	if (token != NULL) {
7283 		token->ipt_subtype = iter.igi_type;
7284 		error = ipf_geniter(token, &iter, ifs);
7285 	} else
7286 		error = EFAULT;
7287 	RWLOCK_EXIT(&ifs->ifs_ipf_tokens);
7288 
7289 	return error;
7290 }
7291 
7292 
7293 /* --------------------------------------------------------------------- */
7294 /* Function:    ipf_earlydrop                                            */
7295 /* Returns:     number of dropped/removed entries from the queue         */
7296 /* Parameters:	flushtype - which table we're cleaning (NAT or State)	 */
7297 /*              ifq	- pointer to queue with entries to be deleted    */
7298 /*              idletime - entry must be idle this long to be deleted    */
7299 /*              ifs     - ipf stack instance                             */
7300 /*                                                                       */
7301 /* Function is invoked from state/NAT flush routines to remove entries   */
7302 /* from specified timeout queue, based on how long they've sat idle,     */
7303 /* without waiting for it to happen on its own.                          */
7304 /* --------------------------------------------------------------------- */
7305 int ipf_earlydrop(flushtype, ifq, idletime, ifs)
7306 int flushtype;
7307 ipftq_t *ifq;
7308 int idletime;
7309 ipf_stack_t *ifs;
7310 {
7311         ipftqent_t *tqe, *tqn;
7312         unsigned int dropped;
7313         int droptick;
7314 	void *ent;
7315 
7316         if (ifq == NULL)
7317                 return (0);
7318 
7319         dropped = 0;
7320 
7321         /*
7322          * Determine the tick representing the idle time we're interested
7323          * in.  If an entry exists in the queue, and it was touched before
7324          * that tick, then it's been idle longer than idletime, so it should
7325 	 * be deleted.
7326          */
7327         droptick = ifs->ifs_fr_ticks - idletime;
7328         tqn = ifq->ifq_head;
7329         while ((tqe = tqn) != NULL && tqe->tqe_touched < droptick) {
7330                 tqn = tqe->tqe_next;
7331 		ent = tqe->tqe_parent;
7332 		switch (flushtype)
7333 		{
7334 		case NAT_FLUSH:
7335 			if (nat_delete((nat_t *)ent, NL_FLUSH, ifs) == 0)
7336 				dropped++;
7337 			break;
7338 		case STATE_FLUSH:
7339 			if (fr_delstate((ipstate_t *)ent, ISL_FLUSH, ifs) == 0)
7340 				dropped++;
7341 			break;
7342 		default:
7343 			return (0);
7344 		}
7345         }
7346         return (dropped);
7347 }
7348 
7349 
7350 /* --------------------------------------------------------------------- */
7351 /* Function:    ipf_flushclosing                                         */
7352 /* Returns:     int - number of entries deleted                          */
7353 /* Parameters:	flushtype - which table we're cleaning (NAT or State)	 */
7354 /*              stateval - TCP state at which to start removing entries  */
7355 /*              ipfqs - pointer to timeout queues                        */
7356 /*              userqs - pointer to user defined queues                  */
7357 /*              ifs  - ipf stack instance                                */
7358 /*                                                                       */
7359 /* Remove state/NAT table entries for TCP connections which are in the   */
7360 /* process of closing, and have at least reached the state specified by  */
7361 /* the 'stateval' parameter.                                             */
7362 /* --------------------------------------------------------------------- */
7363 int ipf_flushclosing(flushtype, stateval, ipfqs, userqs, ifs)
7364 int flushtype, stateval;
7365 ipftq_t *ipfqs, *userqs;
7366 ipf_stack_t *ifs;
7367 {
7368 	ipftq_t *ifq, *ifqn;
7369         ipftqent_t *tqe, *tqn;
7370         int dropped;
7371 	void *ent;
7372 	nat_t *nat;
7373 	ipstate_t *is;
7374 
7375         dropped = 0;
7376 
7377         /*
7378          * Start by deleting any entries in specific timeout queues.
7379          */
7380 	ifqn = &ipfqs[stateval];
7381         while ((ifq = ifqn) != NULL) {
7382                 ifqn = ifq->ifq_next;
7383                 dropped += ipf_earlydrop(flushtype, ifq, (int)0, ifs);
7384         }
7385 
7386         /*
7387          * Next, look through user defined queues for closing entries.
7388          */
7389 	ifqn = userqs;
7390         while ((ifq = ifqn) != NULL) {
7391                 ifqn = ifq->ifq_next;
7392                 tqn = ifq->ifq_head;
7393                 while ((tqe = tqn) != NULL) {
7394                         tqn = tqe->tqe_next;
7395 			ent = tqe->tqe_parent;
7396 			switch (flushtype)
7397 			{
7398 			case NAT_FLUSH:
7399 				nat = (nat_t *)ent;
7400 				if ((nat->nat_p == IPPROTO_TCP) &&
7401 				    (nat->nat_tcpstate[0] >= stateval) &&
7402 				    (nat->nat_tcpstate[1] >= stateval) &&
7403 				    (nat_delete(nat, NL_EXPIRE, ifs) == 0))
7404 					dropped++;
7405 				break;
7406 			case STATE_FLUSH:
7407 				is = (ipstate_t *)ent;
7408 				if ((is->is_p == IPPROTO_TCP) &&
7409 				    (is->is_state[0] >= stateval) &&
7410 				    (is->is_state[1] >= stateval) &&
7411 				    (fr_delstate(is, ISL_EXPIRE, ifs) == 0))
7412 					dropped++;
7413 				break;
7414 			default:
7415 				return (0);
7416 			}
7417                 }
7418         }
7419         return (dropped);
7420 }
7421 
7422 
7423 /* --------------------------------------------------------------------- */
7424 /* Function:    ipf_extraflush                                           */
7425 /* Returns:     int - number of entries flushed (0 = none)               */
7426 /* Parameters:	flushtype - which table we're cleaning (NAT or State)	 */
7427 /*              ipfqs - pointer to 'established' timeout queue           */
7428 /*              userqs - pointer to user defined queues                  */
7429 /*              ifs  - ipf stack instance                                */
7430 /*                                                                       */
7431 /* This function gets called when either NAT or state tables fill up.    */
7432 /* We need to try a bit harder to free up some space.  The function will */
7433 /* flush entries for TCP connections which have been idle a long time.   */
7434 /*                                                                       */
7435 /* Currently, the idle time is checked using values from ideltime_tab[]	 */
7436 /* --------------------------------------------------------------------- */
7437 int ipf_extraflush(flushtype, ipfqs, userqs, ifs)
7438 int flushtype;
7439 ipftq_t *ipfqs, *userqs;
7440 ipf_stack_t *ifs;
7441 {
7442 	ipftq_t *ifq, *ifqn;
7443 	int idletime, removed, idle_idx;
7444 
7445 	removed = 0;
7446 
7447 	/*
7448   	 * Determine initial threshold for minimum idle time based on
7449 	 * how long ipfilter has been running.  Ipfilter needs to have
7450 	 * been up as long as the smallest interval to continue on.
7451 	 *
7452 	 * Minimum idle times stored in idletime_tab and indexed by
7453 	 * idle_idx.  Start at upper end of array and work backwards.
7454 	 *
7455 	 * Once the index is found, set the initial idle time to the
7456 	 * first interval before the current ipfilter run time.
7457 	 */
7458 	if (ifs->ifs_fr_ticks < idletime_tab[0])
7459 		return (0);
7460 	idle_idx = (sizeof (idletime_tab) / sizeof (int)) - 1;
7461 	if (ifs->ifs_fr_ticks > idletime_tab[idle_idx]) {
7462 		idletime = idletime_tab[idle_idx];
7463 	} else {
7464 		while ((idle_idx > 0) &&
7465 		    (ifs->ifs_fr_ticks < idletime_tab[idle_idx]))
7466 			idle_idx--;
7467 
7468 		idletime = (ifs->ifs_fr_ticks /
7469 			    idletime_tab[idle_idx]) *
7470 			    idletime_tab[idle_idx];
7471 	}
7472 
7473 	while (idle_idx >= 0) {
7474 		/*
7475 		 * Check to see if we need to delete more entries.
7476 		 * If we do, start with appropriate timeout queue.
7477 		 */
7478 		if (flushtype == NAT_FLUSH) {
7479 			if (NAT_TAB_WATER_LEVEL(ifs) <=
7480 			    ifs->ifs_nat_flush_level_lo)
7481 				break;
7482 		} else if (flushtype == STATE_FLUSH) {
7483 			if (ST_TAB_WATER_LEVEL(ifs) <=
7484 			    ifs->ifs_state_flush_level_lo)
7485 				break;
7486 		} else {
7487 			break;
7488 		}
7489 
7490 		removed += ipf_earlydrop(flushtype, ipfqs, idletime, ifs);
7491 
7492 		/*
7493 		 * Next, check the user defined queues.  But first, make
7494 		 * certain that timeout queue deletions didn't do enough.
7495 		 */
7496 		if (flushtype == NAT_FLUSH) {
7497 			if (NAT_TAB_WATER_LEVEL(ifs) <=
7498 			    ifs->ifs_nat_flush_level_lo)
7499 				break;
7500 		} else {
7501 			if (ST_TAB_WATER_LEVEL(ifs) <=
7502 			    ifs->ifs_state_flush_level_lo)
7503 				break;
7504 		}
7505 		ifqn = userqs;
7506 		while ((ifq = ifqn) != NULL) {
7507 			ifqn = ifq->ifq_next;
7508 			removed += ipf_earlydrop(flushtype, ifq, idletime, ifs);
7509 		}
7510 
7511 		/*
7512 		 * Adjust the granularity of idle time.
7513 		 *
7514 		 * If we reach an interval boundary, we need to
7515 		 * either adjust the idle time accordingly or exit
7516 		 * the loop altogether (if this is very last check).
7517 		 */
7518 		idletime -= idletime_tab[idle_idx];
7519 		if (idletime < idletime_tab[idle_idx]) {
7520 			if (idle_idx != 0) {
7521 				idletime = idletime_tab[idle_idx] -
7522 				idletime_tab[idle_idx - 1];
7523 				idle_idx--;
7524 			} else {
7525 				break;
7526 			}
7527 		}
7528 	}
7529 
7530 	return (removed);
7531 }
7532