xref: /dragonfly/sys/net/ipfw3/ip_fw3.c (revision f0e61bb7)
1 /*
2  * Copyright (c) 1993 Daniel Boulet
3  * Copyright (c) 1994 Ugen J.S.Antsilevich
4  * Copyright (c) 2002 Luigi Rizzo, Universita` di Pisa
5  * Copyright (c) 2014 - 2018 The DragonFly Project.  All rights reserved.
6  *
7  * This code is derived from software contributed to The DragonFly Project
8  * by Bill Yuan <bycn82@dragonflybsd.org>
9  *
10  * Redistribution and use in source and binary forms, with or without
11  * modification, are permitted provided that the following conditions
12  * are met:
13  *
14  * 1. Redistributions of source code must retain the above copyright
15  *    notice, this list of conditions and the following disclaimer.
16  * 2. Redistributions in binary form must reproduce the above copyright
17  *    notice, this list of conditions and the following disclaimer in
18  *    the documentation and/or other materials provided with the
19  *    distribution.
20  * 3. Neither the name of The DragonFly Project nor the names of its
21  *    contributors may be used to endorse or promote products derived
22  *    from this software without specific, prior written permission.
23  *
24  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
25  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
26  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
27  * FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE
28  * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
29  * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
30  * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
31  * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
32  * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
33  * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
34  * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35  * SUCH DAMAGE.
36  *
37  */
38 
39 #include "opt_ipfw.h"
40 #include "opt_inet.h"
41 #ifndef INET
42 #error IPFIREWALL3 requires INET.
43 #endif /* INET */
44 
45 #include <sys/param.h>
46 #include <sys/kernel.h>
47 #include <sys/malloc.h>
48 #include <sys/mbuf.h>
49 #include <sys/socketvar.h>
50 #include <sys/sysctl.h>
51 #include <sys/systimer.h>
52 #include <sys/in_cksum.h>
53 #include <sys/systm.h>
54 #include <sys/proc.h>
55 #include <sys/socket.h>
56 #include <sys/syslog.h>
57 #include <sys/ucred.h>
58 #include <sys/lock.h>
59 
60 #include <net/if.h>
61 #include <net/radix.h>
62 #include <net/route.h>
63 #include <net/pfil.h>
64 #include <net/netmsg2.h>
65 
66 #include <netinet/in.h>
67 #include <netinet/in_systm.h>
68 #include <netinet/in_var.h>
69 #include <netinet/in_pcb.h>
70 #include <netinet/ip.h>
71 #include <netinet/ip_var.h>
72 #include <netinet/ip_icmp.h>
73 #include <netinet/tcp.h>
74 #include <netinet/tcp_timer.h>
75 #include <netinet/tcp_var.h>
76 #include <netinet/tcpip.h>
77 #include <netinet/udp.h>
78 #include <netinet/udp_var.h>
79 #include <netinet/ip_divert.h>
80 #include <netinet/if_ether.h>
81 
82 #include <net/ipfw3/ip_fw.h>
83 #include <net/ipfw3/ip_fw3_set.h>
84 #include <net/ipfw3_basic/ip_fw3_log.h>
85 #include <net/ipfw3_basic/ip_fw3_table.h>
86 #include <net/ipfw3_basic/ip_fw3_sync.h>
87 #include <net/ipfw3_basic/ip_fw3_basic.h>
88 #include <net/ipfw3_basic/ip_fw3_state.h>
89 #include <net/ipfw3_nat/ip_fw3_nat.h>
90 #include <net/dummynet3/ip_dummynet3.h>
91 
92 MALLOC_DEFINE(M_IPFW3, "IPFW3", "ipfw3 module");
93 
94 #define MAX_MODULE		10
95 #define MAX_OPCODE_PER_MODULE	100
96 
97 #define IPFW_AUTOINC_STEP_MIN	1
98 #define IPFW_AUTOINC_STEP_MAX	1000
99 #define IPFW_AUTOINC_STEP_DEF	100
100 
101 
102 struct netmsg_ipfw {
103 	struct netmsg_base base;
104 	const struct ipfw_ioc_rule *ioc_rule;
105 	struct ip_fw	*rule;
106 	struct ip_fw	*next_rule;
107 	struct ip_fw	*prev_rule;
108 	struct ip_fw	*sibling;	/* sibling in prevous CPU */
109 };
110 
111 struct netmsg_del {
112 	struct netmsg_base base;
113 	struct ip_fw	*rule;
114 	struct ip_fw	*start_rule;
115 	struct ip_fw	*prev_rule;
116 	struct ipfw_ioc_state *ioc_state;
117 	uint16_t	rulenum;
118 	uint8_t		set_from;
119 	uint8_t		set_to;
120 	int		kill_default;
121 };
122 
123 struct netmsg_zent {
124 	struct netmsg_base base;
125 	struct ip_fw	*start_rule;
126 	uint16_t	rulenum;
127 	uint16_t	log_only;
128 };
129 
130 ip_fw_ctl_t	*ip_fw3_ctl_nat_ptr = NULL;
131 ip_fw_ctl_t	*ip_fw3_ctl_state_ptr = NULL;
132 ip_fw_ctl_t	*ip_fw3_ctl_table_ptr = NULL;
133 ip_fw_ctl_t	*ip_fw3_ctl_sync_ptr = NULL;
134 ip_fw_log_t	*ip_fw3_log_ptr = NULL;
135 
136 extern int ip_fw_loaded;
137 extern struct ipfw3_state_context 	*fw3_state_ctx[MAXCPU];
138 int 			sysctl_var_fw3_enable = 1;
139 int 			sysctl_var_fw3_one_pass = 1;
140 int 			sysctl_var_fw3_verbose = 0;
141 static int 		sysctl_var_fw3_flushing;
142 static int 		sysctl_var_fw3_debug;
143 static int 		sysctl_var_autoinc_step = IPFW_AUTOINC_STEP_DEF;
144 
145 int	ip_fw3_sysctl_enable(SYSCTL_HANDLER_ARGS);
146 int	ip_fw3_sysctl_autoinc_step(SYSCTL_HANDLER_ARGS);
147 
148 SYSCTL_NODE(_net_inet_ip, OID_AUTO, fw3, CTLFLAG_RW, 0, "Firewall");
149 SYSCTL_PROC(_net_inet_ip_fw3, OID_AUTO, enable, CTLTYPE_INT | CTLFLAG_RW,
150 	&sysctl_var_fw3_enable, 0, ip_fw3_sysctl_enable, "I", "Enable ipfw");
151 SYSCTL_PROC(_net_inet_ip_fw3, OID_AUTO, sysctl_var_autoinc_step,
152 	CTLTYPE_INT | CTLFLAG_RW, &sysctl_var_autoinc_step, 0,
153 	ip_fw3_sysctl_autoinc_step, "I", "Rule number autincrement step");
154 SYSCTL_INT(_net_inet_ip_fw3, OID_AUTO,one_pass,CTLFLAG_RW,
155 	&sysctl_var_fw3_one_pass, 0,
156 	"Only do a single pass through ipfw3 when using dummynet(4)");
157 SYSCTL_INT(_net_inet_ip_fw3, OID_AUTO, debug, CTLFLAG_RW,
158 	&sysctl_var_fw3_debug, 0, "Enable printing of debug ip_fw statements");
159 SYSCTL_INT(_net_inet_ip_fw3, OID_AUTO, verbose, CTLFLAG_RW,
160 	&sysctl_var_fw3_verbose, 0, "Log matches to ipfw3 rules");
161 
162 
163 filter_func 			filter_funcs[MAX_MODULE][MAX_OPCODE_PER_MODULE];
164 struct ipfw3_module 		fw3_modules[MAX_MODULE];
165 struct ipfw3_context 		*fw3_ctx[MAXCPU];
166 struct ipfw3_sync_context 	fw3_sync_ctx;
167 
168 
169 void
170 ip_fw3_register_module(int module_id,char *module_name)
171 {
172 	struct ipfw3_module *tmp;
173 	int i;
174 
175 	tmp = fw3_modules;
176 	for (i=0; i < MAX_MODULE; i++) {
177 		if (tmp->type == 0) {
178 			tmp->type = 1;
179 			tmp->id = module_id;
180 			strncpy(tmp->name, module_name, strlen(module_name));
181 			break;
182 		}
183 		tmp++;
184 	}
185 	kprintf("ipfw3 module %s loaded\n", module_name);
186 }
187 
188 int
189 ip_fw3_unregister_module(int module_id)
190 {
191 	struct ipfw3_module *tmp;
192 	struct ip_fw *fw;
193 	ipfw_insn *cmd;
194 	int i, len, cmdlen, found;
195 
196 	found = 0;
197 	tmp = fw3_modules;
198 	struct ipfw3_context *ctx = fw3_ctx[mycpuid];
199 	fw = ctx->rules;
200 	for (; fw; fw = fw->next) {
201 		for (len = fw->cmd_len, cmd = fw->cmd; len > 0;
202 			len -= cmdlen,
203 			cmd = (ipfw_insn *)((uint32_t *)cmd + cmdlen)) {
204 			cmdlen = F_LEN(cmd);
205 			if (cmd->module == 0 &&
206 				(cmd->opcode == 0 || cmd->opcode == 1)) {
207 				//action accept or deny
208 			} else if (cmd->module == module_id) {
209 				found = 1;
210 				goto decide;
211 			}
212 		}
213 	}
214 decide:
215 	if (found) {
216 		return 1;
217 	} else {
218 		for (i = 0; i < MAX_MODULE; i++) {
219 			if (tmp->type == 1 && tmp->id == module_id) {
220 				tmp->type = 0;
221 				kprintf("ipfw3 module %s unloaded\n",
222 						tmp->name);
223 				break;
224 			}
225 			tmp++;
226 		}
227 
228 		for (i = 0; i < MAX_OPCODE_PER_MODULE; i++) {
229 			if (module_id == 0) {
230 				if (i ==0 || i == 1) {
231 					continue;
232 				}
233 			}
234 			filter_funcs[module_id][i] = NULL;
235 		}
236 		return 0;
237 	}
238 }
239 
240 void
241 ip_fw3_register_filter_funcs(int module, int opcode, filter_func func)
242 {
243 	filter_funcs[module][opcode] = func;
244 }
245 
246 void
247 check_accept(int *cmd_ctl, int *cmd_val, struct ip_fw_args **args,
248 		struct ip_fw **f, ipfw_insn *cmd, uint16_t ip_len)
249 {
250 	*cmd_val = IP_FW_PASS;
251 	*cmd_ctl = IP_FW_CTL_DONE;
252 	if (cmd->arg3 && ip_fw3_log_ptr != NULL) {
253 		ip_fw3_log_ptr((*args)->m, (*args)->eh, cmd->arg1);
254 	}
255 }
256 
257 void
258 check_deny(int *cmd_ctl, int *cmd_val, struct ip_fw_args **args,
259 		struct ip_fw **f, ipfw_insn *cmd, uint16_t ip_len)
260 {
261 	*cmd_val = IP_FW_DENY;
262 	*cmd_ctl = IP_FW_CTL_DONE;
263 	if (cmd->arg3 && ip_fw3_log_ptr != NULL) {
264 		ip_fw3_log_ptr((*args)->m, (*args)->eh, cmd->arg1);
265 	}
266 }
267 
268 void
269 init_module(void)
270 {
271 	memset(fw3_modules, 0, sizeof(struct ipfw3_module) * MAX_MODULE);
272 	memset(filter_funcs, 0, sizeof(filter_func) *
273 			MAX_OPCODE_PER_MODULE * MAX_MODULE);
274 	ip_fw3_register_filter_funcs(0, O_BASIC_ACCEPT,
275 			(filter_func)check_accept);
276 	ip_fw3_register_filter_funcs(0, O_BASIC_DENY, (filter_func)check_deny);
277 }
278 
279 int
280 ip_fw3_free_rule(struct ip_fw *rule)
281 {
282 	kfree(rule, M_IPFW3);
283 	rule = NULL;
284 	return 1;
285 }
286 
287 static struct ip_fw *
288 lookup_next_rule(struct ip_fw *me)
289 {
290 	struct ip_fw *rule = NULL;
291 	ipfw_insn *cmd;
292 
293 	/* look for action, in case it is a skipto */
294 	cmd = ACTION_PTR(me);
295 	if ((int)cmd->module == MODULE_BASIC_ID &&
296 		(int)cmd->opcode == O_BASIC_SKIPTO) {
297 		for (rule = me->next; rule; rule = rule->next) {
298 			if (rule->rulenum >= cmd->arg1)
299 				break;
300 		}
301 	}
302 	if (rule == NULL) {	/* failure or not a skipto */
303 		rule = me->next;
304 	}
305 	me->next_rule = rule;
306 	return rule;
307 }
308 
309 /*
310  * rules are stored in ctx->ipfw_rule_chain.
311  * and each rule is combination of multiple cmds.(ipfw_insn)
312  * in each rule, it begin with filter cmds. and end with action cmds.
313  * 'outer/inner loop' are looping the rules/cmds.
314  * it will invoke the cmds relatived function according to the cmd's
315  * module id and opcode id. and process according to return value.
316  */
317 int
318 ip_fw3_chk(struct ip_fw_args *args)
319 {
320 	struct tcphdr *tcp;
321 	struct udphdr *udp;
322 	struct icmp *icmp;
323 
324 	struct mbuf *m = args->m;
325 	struct ip *ip = mtod(m, struct ip *);
326 	struct ip_fw *f = NULL;		/* matching rule */
327 	int cmd_val = IP_FW_PASS;
328 	struct m_tag *mtag;
329 	struct divert_info *divinfo;
330 
331 	/*
332 	 * hlen	The length of the IPv4 header.
333 	 *	hlen >0 means we have an IPv4 packet.
334 	 */
335 	u_int hlen = 0;		/* hlen >0 means we have an IP pkt */
336 
337 	/*
338 	 * offset	The offset of a fragment. offset != 0 means that
339 	 *	we have a fragment at this offset of an IPv4 packet.
340 	 *	offset == 0 means that (if this is an IPv4 packet)
341 	 *	this is the first or only fragment.
342 	 */
343 	u_short offset = 0;
344 
345 	uint8_t proto;
346 	uint16_t src_port = 0, dst_port = 0;	/* NOTE: host format	*/
347 	struct in_addr src_ip, dst_ip;		/* NOTE: network format	*/
348 	uint16_t ip_len = 0;
349 	uint8_t prev_module = -1, prev_opcode = -1; /* previous module & opcode */
350 	struct ipfw3_context *ctx = fw3_ctx[mycpuid];
351 
352 	if (m->m_pkthdr.fw_flags & IPFW_MBUF_GENERATED)
353 		return IP_FW_PASS;	/* accept */
354 
355 	if (args->eh == NULL ||		/* layer 3 packet */
356 		(m->m_pkthdr.len >= sizeof(struct ip) &&
357 		 ntohs(args->eh->ether_type) == ETHERTYPE_IP))
358 		hlen = ip->ip_hl << 2;
359 
360 	/*
361 	 * Collect parameters into local variables for faster matching.
362 	 */
363 	if (hlen == 0) {	/* do not grab addresses for non-ip pkts */
364 		proto = args->f_id.proto = 0;	/* mark f_id invalid */
365 		goto after_ip_checks;
366 	}
367 
368 	proto = args->f_id.proto = ip->ip_p;
369 	src_ip = ip->ip_src;
370 	dst_ip = ip->ip_dst;
371 	offset = ntohs(ip->ip_off) & IP_OFFMASK;
372 	ip_len = ntohs(ip->ip_len);
373 
374 #define PULLUP_TO(len)					\
375 do {							\
376 	if (m->m_len < (len)) {				\
377 		args->m = m = m_pullup(m, (len));	\
378 			if (m == NULL)			\
379 				goto pullup_failed;	\
380 		ip = mtod(m, struct ip *);		\
381 	}						\
382 } while (0)
383 
384 	if (offset == 0) {
385 		switch (proto) {
386 			case IPPROTO_TCP:
387 				PULLUP_TO(hlen + sizeof(struct tcphdr));
388 				tcp = L3HDR(struct tcphdr, ip);
389 				dst_port = tcp->th_dport;
390 				src_port = tcp->th_sport;
391 				args->f_id.flags = tcp->th_flags;
392 				break;
393 			case IPPROTO_UDP:
394 				PULLUP_TO(hlen + sizeof(struct udphdr));
395 				udp = L3HDR(struct udphdr, ip);
396 				dst_port = udp->uh_dport;
397 				src_port = udp->uh_sport;
398 				break;
399 			case IPPROTO_ICMP:
400 				PULLUP_TO(hlen + 4);
401 				icmp = L3HDR(struct icmp, ip);
402   				args->f_id.flags = icmp->icmp_type;
403 				dst_port = icmp->icmp_id;
404 				src_port = dst_port;
405 				break;
406 			default:
407 				break;
408 		}
409 	}
410 
411 #undef PULLUP_TO
412 
413 	args->f_id.src_ip = ntohl(src_ip.s_addr);
414 	args->f_id.dst_ip = ntohl(dst_ip.s_addr);
415 	args->f_id.src_port = src_port = ntohs(src_port);
416 	args->f_id.dst_port = dst_port = ntohs(dst_port);
417 
418 after_ip_checks:
419 	if (args->rule) {
420 		/*
421 		 * Packet has already been tagged. Look for the next rule
422 		 * to restart processing.
423 		 *
424 		 * If sysctl_var_fw3_one_pass != 0 then just accept it.
425 		 * XXX should not happen here, but optimized out in
426 		 * the caller.
427 		 */
428 		if (sysctl_var_fw3_one_pass)
429 			return IP_FW_PASS;
430 
431 		/* This rule is being/has been flushed */
432 		if (sysctl_var_fw3_flushing)
433 			return IP_FW_DENY;
434 
435 		f = args->rule->next_rule;
436 		if (f == NULL)
437 			f = lookup_next_rule(args->rule);
438 	} else {
439 		/*
440 		 * Find the starting rule. It can be either the first
441 		 * one, or the one after divert_rule if asked so.
442 		 */
443 		int skipto;
444 
445 		mtag = m_tag_find(m, PACKET_TAG_IPFW_DIVERT, NULL);
446 		if (mtag != NULL) {
447 			divinfo = m_tag_data(mtag);
448 			skipto = divinfo->skipto;
449 		} else {
450 			skipto = 0;
451 		}
452 
453 		f = ctx->rules;
454 		if (args->eh == NULL && skipto != 0) {
455 			/* No skipto during rule flushing */
456 			if (sysctl_var_fw3_flushing) {
457 				return IP_FW_DENY;
458 			}
459 			if (skipto >= IPFW_DEFAULT_RULE) {
460 				return IP_FW_DENY; /* invalid */
461 			}
462 			while (f && f->rulenum <= skipto) {
463 				f = f->next;
464 			}
465 			if (f == NULL) {	/* drop packet */
466 				return IP_FW_DENY;
467 			}
468 		} else if (sysctl_var_fw3_flushing) {
469 			/* Rules are being flushed; skip to default rule */
470 			f = ctx->default_rule;
471 		}
472 	}
473 	if ((mtag = m_tag_find(m, PACKET_TAG_IPFW_DIVERT, NULL)) != NULL) {
474 		m_tag_delete(m, mtag);
475 	}
476 
477 	/*
478 	 * Now scan the rules, and parse microinstructions for each rule.
479 	 */
480 	int prev_val;	/*  previous result of 'or' filter */
481 	int l, cmdlen;
482 	ipfw_insn *cmd;
483 	int cmd_ctl;
484 	/* foreach rule in chain */
485 	for (; f; f = f->next) {
486 again:  /* check the rule again*/
487 		if (ctx->sets & (1 << f->set)) {
488 			continue;
489 		}
490 
491 		prev_val = -1;
492 		 /* foreach cmd in rule */
493 		for (l = f->cmd_len, cmd = f->cmd; l > 0; l -= cmdlen,
494 			cmd = (ipfw_insn *)((uint32_t *)cmd+ cmdlen)) {
495 			cmdlen = F_LEN(cmd);
496 
497 			/* skip 'or' filter when already match */
498 			if (cmd->len & F_OR &&
499 				cmd->module == prev_module &&
500 				cmd->opcode == prev_opcode &&
501 				prev_val == 1) {
502 				goto next_cmd;
503 			}
504 
505 check_body: /* check the body of the rule again.*/
506 			(filter_funcs[cmd->module][cmd->opcode])
507 				(&cmd_ctl, &cmd_val, &args, &f, cmd, ip_len);
508 			switch(cmd_ctl) {
509 				case IP_FW_CTL_DONE:
510 					if (prev_val == 0) /* but 'or' failed */
511 						goto next_rule;
512 					goto done;
513 				case IP_FW_CTL_AGAIN:
514 					goto again;
515 				case IP_FW_CTL_NEXT:
516 					goto next_rule;
517 				case IP_FW_CTL_NAT:
518 					args->rule=f;
519 					goto done;
520 				case IP_FW_CTL_CHK_STATE:
521 					/* update the cmd and l */
522 					cmd = ACTION_PTR(f);
523 					l = f->cmd_len - f->act_ofs;
524 					goto check_body;
525 			}
526 			if (cmd->len & F_NOT)
527 				cmd_val= !cmd_val;
528 
529 			if (cmd->len & F_OR) {	/* has 'or' */
530 				if (!cmd_val) {	/* not matched */
531 					if(prev_val == -1){	/* first 'or' */
532 						prev_val = 0;
533 						prev_module = cmd->module;
534 						prev_opcode = cmd->opcode;
535 					} else if (prev_module == cmd->module &&
536 						prev_opcode == cmd->opcode) {
537 						/* continuous 'or' filter */
538 					} else if (prev_module != cmd->module ||
539 						prev_opcode != cmd->opcode) {
540 						/* 'or' filter changed */
541 						if(prev_val == 0){
542 							goto next_rule;
543 						} else {
544 							prev_val = 0;
545 							prev_module = cmd->module;
546 							prev_opcode = cmd->opcode;
547 						}
548 					}
549 				} else { /* has 'or' and matched */
550 					prev_val = 1;
551 					prev_module = cmd->module;
552 					prev_opcode = cmd->opcode;
553 				}
554 			} else { /* no or */
555 				if (!cmd_val) {	/* not matched */
556 					goto next_rule;
557 				} else {
558 					if (prev_val == 0) {
559 						/* previous 'or' not matched */
560 						goto next_rule;
561 					} else {
562 						prev_val = -1;
563 					}
564 				}
565 			}
566 next_cmd:;
567 		}	/* end of inner for, scan opcodes */
568 next_rule:;		/* try next rule		*/
569 	}		/* end of outer for, scan rules */
570 	kprintf("+++ ipfw: ouch!, skip past end of rules, denying packet\n");
571 	return IP_FW_DENY;
572 
573 done:
574 	/* Update statistics */
575 	f->pcnt++;
576 	f->bcnt += ip_len;
577 	f->timestamp = time_second;
578 	return cmd_val;
579 
580 pullup_failed:
581 	if (sysctl_var_fw3_verbose)
582 		kprintf("pullup failed\n");
583 	return IP_FW_DENY;
584 }
585 
586 struct mbuf *
587 ip_fw3_dummynet_io(struct mbuf *m, int pipe_nr, int dir, struct ip_fw_args *fwa)
588 {
589 	struct m_tag *mtag;
590 	struct dn_pkt *pkt;
591 	ipfw_insn *cmd;
592 	const struct ipfw_flow_id *id;
593 	struct dn_flow_id *fid;
594 
595 	M_ASSERTPKTHDR(m);
596 
597 	mtag = m_tag_get(PACKET_TAG_DUMMYNET, sizeof(*pkt),
598 	    M_INTWAIT | M_NULLOK);
599 	if (mtag == NULL) {
600 		m_freem(m);
601 		return (NULL);
602 	}
603 	m_tag_prepend(m, mtag);
604 
605 	pkt = m_tag_data(mtag);
606 	bzero(pkt, sizeof(*pkt));
607 
608 	cmd = (ipfw_insn *)((uint32_t *)fwa->rule->cmd + fwa->rule->act_ofs);
609 	KASSERT(cmd->opcode == O_DUMMYNET_PIPE ||
610 			cmd->opcode == O_DUMMYNET_QUEUE,
611 			("Rule is not PIPE or QUEUE, opcode %d", cmd->opcode));
612 
613 	pkt->dn_m = m;
614 	pkt->dn_flags = (dir & DN_FLAGS_DIR_MASK);
615 	pkt->ifp = fwa->oif;
616 	pkt->pipe_nr = pipe_nr;
617 
618 	pkt->cpuid = mycpuid;
619 	pkt->msgport = netisr_curport();
620 
621 	id = &fwa->f_id;
622 	fid = &pkt->id;
623 	fid->fid_dst_ip = id->dst_ip;
624 	fid->fid_src_ip = id->src_ip;
625 	fid->fid_dst_port = id->dst_port;
626 	fid->fid_src_port = id->src_port;
627 	fid->fid_proto = id->proto;
628 	fid->fid_flags = id->flags;
629 
630 	pkt->dn_priv = fwa->rule;
631 
632 	if ((int)cmd->opcode == O_DUMMYNET_PIPE)
633 		pkt->dn_flags |= DN_FLAGS_IS_PIPE;
634 
635 	m->m_pkthdr.fw_flags |= DUMMYNET_MBUF_TAGGED;
636 	return (m);
637 }
638 
639 
640 void
641 add_rule_dispatch(netmsg_t nmsg)
642 {
643 	struct netmsg_ipfw *fwmsg = (struct netmsg_ipfw *)nmsg;
644 	struct ipfw3_context *ctx = fw3_ctx[mycpuid];
645 	struct ip_fw *rule, *prev,*next;
646 	const struct ipfw_ioc_rule *ioc_rule;
647 
648 	ioc_rule = fwmsg->ioc_rule;
649 	 // create rule by ioc_rule
650 	rule = kmalloc(RULESIZE(ioc_rule), M_IPFW3, M_WAITOK | M_ZERO);
651 	rule->act_ofs = ioc_rule->act_ofs;
652 	rule->cmd_len = ioc_rule->cmd_len;
653 	rule->rulenum = ioc_rule->rulenum;
654 	rule->set = ioc_rule->set;
655 	bcopy(ioc_rule->cmd, rule->cmd, rule->cmd_len * 4);
656 
657 	for (prev = NULL, next = ctx->rules;
658 		next; prev = next, next = next->next) {
659 		if (ioc_rule->insert) {
660 			if (next->rulenum >= ioc_rule->rulenum) {
661 				break;
662 			}
663 		} else {
664 			if (next->rulenum > ioc_rule->rulenum) {
665 				break;
666 			}
667 		}
668 	}
669 	KASSERT(next != NULL, ("no default rule?!"));
670 
671 	/*
672 	 * Insert rule into the pre-determined position
673 	 */
674 	if (prev != NULL) {
675 		rule->next = next;
676 		prev->next = rule;
677 	} else {
678 		rule->next = ctx->rules;
679 		ctx->rules = rule;
680 	}
681 
682 	/*
683 	 * if sibiling in last CPU is exists,
684 	 * then it's sibling should be current rule
685 	 */
686 	if (fwmsg->sibling != NULL) {
687 		fwmsg->sibling->sibling = rule;
688 	}
689 	/* prepare for next CPU */
690 	fwmsg->sibling = rule;
691 
692 	netisr_forwardmsg_all(&nmsg->base, mycpuid + 1);
693 }
694 
695 /*
696  * confirm the rulenumber
697  * call dispatch function to add rule into the list
698  * Update the statistic
699  */
700 void
701 ip_fw3_add_rule(struct ipfw_ioc_rule *ioc_rule)
702 {
703 	struct ipfw3_context *ctx = fw3_ctx[mycpuid];
704 	struct netmsg_ipfw fwmsg;
705 	struct netmsg_base *nmsg;
706 	struct ip_fw *f;
707 
708 	IPFW_ASSERT_CFGPORT(&curthread->td_msgport);
709 
710 	/*
711 	 * If rulenum is 0, find highest numbered rule before the
712 	 * default rule, and add rule number incremental step.
713 	 */
714 	if (ioc_rule->rulenum == 0) {
715 		int step = sysctl_var_autoinc_step;
716 
717 		KKASSERT(step >= IPFW_AUTOINC_STEP_MIN &&
718 				step <= IPFW_AUTOINC_STEP_MAX);
719 
720 		/*
721 		 * Locate the highest numbered rule before default
722 		 */
723 		for (f = ctx->rules; f; f = f->next) {
724 			if (f->rulenum == IPFW_DEFAULT_RULE)
725 				break;
726 			ioc_rule->rulenum = f->rulenum;
727 		}
728 		if (ioc_rule->rulenum < IPFW_DEFAULT_RULE - step)
729 			ioc_rule->rulenum += step;
730 	}
731 	KASSERT(ioc_rule->rulenum != IPFW_DEFAULT_RULE &&
732 			ioc_rule->rulenum != 0,
733 			("invalid rule num %d", ioc_rule->rulenum));
734 
735 	bzero(&fwmsg, sizeof(fwmsg));
736 	nmsg = &fwmsg.base;
737 	netmsg_init(nmsg, NULL, &curthread->td_msgport,
738 			0, add_rule_dispatch);
739 	fwmsg.ioc_rule = ioc_rule;
740 
741 	netisr_domsg(nmsg, 0);
742 
743 	IPFW3_DEBUG("++ installed rule %d, static count now %d\n",
744 			ioc_rule->rulenum, static_count);
745 }
746 
747 /**
748  * Free storage associated with a static rule (including derived
749  * dynamic rules).
750  * The caller is in charge of clearing rule pointers to avoid
751  * dangling pointers.
752  * @return a pointer to the next entry.
753  * Arguments are not checked, so they better be correct.
754  * Must be called at splimp().
755  */
756 struct ip_fw *
757 ip_fw3_delete_rule(struct ipfw3_context *ctx,
758 		 struct ip_fw *prev, struct ip_fw *rule)
759 {
760 	if (prev == NULL)
761 		ctx->rules = rule->next;
762 	else
763 		prev->next = rule->next;
764 
765 	kfree(rule, M_IPFW3);
766 	rule = NULL;
767 	return NULL;
768 }
769 
770 void
771 flush_rule_dispatch(netmsg_t nmsg)
772 {
773 	struct netmsg_del *dmsg = (struct netmsg_del *)nmsg;
774 	struct ipfw3_context *ctx = fw3_ctx[mycpuid];
775 	struct ip_fw *rule, *the_rule;
776 	int kill_default = dmsg->kill_default;
777 
778 	rule = ctx->rules;
779 	while (rule != NULL) {
780 		if (rule->rulenum == IPFW_DEFAULT_RULE && kill_default == 0) {
781 			ctx->rules = rule;
782 			break;
783 		}
784 		the_rule = rule;
785 		rule = rule->next;
786 
787 		kfree(the_rule, M_IPFW3);
788 	}
789 
790 	netisr_forwardmsg_all(&nmsg->base, mycpuid + 1);
791 }
792 
793 
794 /*
795  * Deletes all rules from a chain (including the default rule
796  * if the second argument is set).
797  * Must be called at splimp().
798  */
799 void
800 ip_fw3_ctl_flush_rule(int kill_default)
801 {
802 	struct netmsg_del dmsg;
803 
804 	IPFW_ASSERT_CFGPORT(&curthread->td_msgport);
805 
806 	if (!kill_default) {
807 		sysctl_var_fw3_flushing = 1;
808 		netmsg_service_sync();
809 	}
810 	/*
811 	 * Press the 'flush' button
812 	 */
813 	bzero(&dmsg, sizeof(dmsg));
814 	netmsg_init(&dmsg.base, NULL, &curthread->td_msgport,
815 			0, flush_rule_dispatch);
816 	dmsg.kill_default = kill_default;
817 	netisr_domsg(&dmsg.base, 0);
818 
819 	/* Flush is done */
820 	sysctl_var_fw3_flushing = 0;
821 }
822 
823 void
824 delete_rule_dispatch(netmsg_t nmsg)
825 {
826 	struct netmsg_del *dmsg = (struct netmsg_del *)nmsg;
827 	struct ipfw3_context *ctx = fw3_ctx[mycpuid];
828 	struct ip_fw *rule, *prev = NULL;
829 
830 	rule = ctx->rules;
831 	while (rule!=NULL) {
832 		if (rule->rulenum == dmsg->rulenum) {
833 			ip_fw3_delete_rule(ctx, prev, rule);
834 			break;
835 		}
836 		prev = rule;
837 		rule = rule->next;
838 	}
839 
840 	netisr_forwardmsg_all(&nmsg->base, mycpuid + 1);
841 }
842 
843 int
844 ip_fw3_ctl_delete_rule(struct sockopt *sopt)
845 {
846 	struct netmsg_del dmsg;
847 	struct netmsg_base *nmsg;
848 	int *rulenum;
849 
850 	rulenum = (int *) sopt->sopt_val;
851 
852 
853 	/*
854 	 * Get rid of the rule duplications on all CPUs
855 	 */
856 	bzero(&dmsg, sizeof(dmsg));
857 	nmsg = &dmsg.base;
858 	netmsg_init(nmsg, NULL, &curthread->td_msgport,
859 			0, delete_rule_dispatch);
860 	dmsg.rulenum = *rulenum;
861 	netisr_domsg(nmsg, 0);
862 	return 0;
863 }
864 
865 /*
866  * Clear counters for a specific rule.
867  */
868 void
869 ip_fw3_clear_counters(struct ip_fw *rule)
870 {
871 	rule->bcnt = rule->pcnt = 0;
872 	rule->timestamp = 0;
873 }
874 
875 void
876 ip_fw3_zero_entry_dispatch(netmsg_t nmsg)
877 {
878 	struct netmsg_zent *zmsg = (struct netmsg_zent *)nmsg;
879 	struct ipfw3_context *ctx = fw3_ctx[mycpuid];
880 	struct ip_fw *rule;
881 
882 	if (zmsg->rulenum == 0) {
883 		for (rule = ctx->rules; rule; rule = rule->next) {
884 			ip_fw3_clear_counters(rule);
885 		}
886 	} else {
887 		for (rule = ctx->rules; rule; rule = rule->next) {
888 			if (rule->rulenum == zmsg->rulenum) {
889 				ip_fw3_clear_counters(rule);
890 			}
891 		}
892 	}
893 	ip_fw3_clear_counters(ctx->default_rule);
894 	netisr_forwardmsg_all(&nmsg->base, mycpuid + 1);
895 }
896 
897 /**
898  * Reset some or all counters on firewall rules.
899  * @arg frwl is null to clear all entries, or contains a specific
900  * rule number.
901  * @arg log_only is 1 if we only want to reset logs, zero otherwise.
902  */
903 int
904 ip_fw3_ctl_zero_entry(int rulenum, int log_only)
905 {
906 	struct netmsg_zent zmsg;
907 	struct netmsg_base *nmsg;
908 	const char *msg;
909 	struct ipfw3_context *ctx = fw3_ctx[mycpuid];
910 
911 	bzero(&zmsg, sizeof(zmsg));
912 	nmsg = &zmsg.base;
913 	netmsg_init(nmsg, NULL, &curthread->td_msgport,
914 			0, ip_fw3_zero_entry_dispatch);
915 	zmsg.log_only = log_only;
916 
917 	if (rulenum == 0) {
918 		msg = log_only ? "ipfw: All logging counts reset.\n"
919 				   : "ipfw: Accounting cleared.\n";
920 	} else {
921 		struct ip_fw *rule;
922 
923 		/*
924 		 * Locate the first rule with 'rulenum'
925 		 */
926 		for (rule = ctx->rules; rule; rule = rule->next) {
927 			if (rule->rulenum == rulenum)
928 				break;
929 		}
930 		if (rule == NULL) /* we did not find any matching rules */
931 			return (EINVAL);
932 		zmsg.start_rule = rule;
933 		zmsg.rulenum = rulenum;
934 
935 		msg = log_only ? "ipfw: Entry %d logging count reset.\n"
936 				   : "ipfw: Entry %d cleared.\n";
937 	}
938 	netisr_domsg(nmsg, 0);
939 	KKASSERT(zmsg.start_rule == NULL);
940 
941 	if (sysctl_var_fw3_verbose)
942 		log(LOG_SECURITY | LOG_NOTICE, msg, rulenum);
943 	return (0);
944 }
945 
946 /*
947  * Get the ioc_rule from the sopt
948  * call ip_fw3_add_rule to add the rule
949  */
950 int
951 ip_fw3_ctl_add_rule(struct sockopt *sopt)
952 {
953 	struct ipfw_ioc_rule *ioc_rule;
954 	size_t size;
955 
956 	size = sopt->sopt_valsize;
957 	if (size > (sizeof(uint32_t) * IPFW_RULE_SIZE_MAX) ||
958 			size < sizeof(*ioc_rule) - sizeof(ipfw_insn)) {
959 		return EINVAL;
960 	}
961 	if (size != (sizeof(uint32_t) * IPFW_RULE_SIZE_MAX)) {
962 		sopt->sopt_val = krealloc(sopt->sopt_val, sizeof(uint32_t) *
963 				IPFW_RULE_SIZE_MAX, M_TEMP, M_WAITOK);
964 	}
965 	ioc_rule = sopt->sopt_val;
966 
967 	ip_fw3_add_rule(ioc_rule);
968 	return 0;
969 }
970 
971 int
972 ip_fw3_ctl_get_modules(struct sockopt *sopt)
973 {
974 	int i;
975 	struct ipfw3_module *mod;
976 	char module_str[1024];
977 	memset(module_str,0,1024);
978 	for (i = 0, mod = fw3_modules; i < MAX_MODULE; i++, mod++) {
979 		if (mod->type != 0) {
980 			if (i > 0)
981 				strcat(module_str,",");
982 			strcat(module_str,mod->name);
983 		}
984 	}
985 	bzero(sopt->sopt_val, sopt->sopt_valsize);
986 	bcopy(module_str, sopt->sopt_val, strlen(module_str));
987 	sopt->sopt_valsize = strlen(module_str);
988 	return 0;
989 }
990 
991 /*
992  * Copy all static rules and states on all CPU
993  */
994 int
995 ip_fw3_ctl_get_rules(struct sockopt *sopt)
996 {
997 	struct ipfw3_context *ctx = fw3_ctx[mycpuid];
998 	struct ip_fw *rule;
999 	struct ipfw_ioc_rule *ioc;
1000 	const struct ip_fw *sibling;
1001 	int total_len = 0;
1002 
1003 	ioc = (struct ipfw_ioc_rule *)sopt->sopt_val;
1004 
1005 	for (rule = ctx->rules; rule; rule = rule->next) {
1006 		total_len += IOC_RULESIZE(rule);
1007 		if (total_len > sopt->sopt_valsize) {
1008 			bzero(sopt->sopt_val, sopt->sopt_valsize);
1009 			return 0;
1010 		}
1011 		ioc->act_ofs = rule->act_ofs;
1012 		ioc->cmd_len = rule->cmd_len;
1013 		ioc->rulenum = rule->rulenum;
1014 		ioc->set = rule->set;
1015 
1016 		ioc->sets = fw3_ctx[mycpuid]->sets;
1017 		ioc->pcnt = 0;
1018 		ioc->bcnt = 0;
1019 		ioc->timestamp = 0;
1020 		for (sibling = rule; sibling != NULL; sibling = sibling->sibling) {
1021 			ioc->pcnt += sibling->pcnt;
1022 			ioc->bcnt += sibling->bcnt;
1023 			if (sibling->timestamp > ioc->timestamp)
1024 				ioc->timestamp = sibling->timestamp;
1025 		}
1026 		bcopy(rule->cmd, ioc->cmd, ioc->cmd_len * 4);
1027 		ioc = (struct ipfw_ioc_rule *)((uint8_t *)ioc + IOC_RULESIZE(ioc));
1028 	}
1029 	sopt->sopt_valsize = total_len;
1030 	return 0;
1031 }
1032 
1033 
1034 /*
1035  * ip_fw3_ctl_x - extended version of ip_fw3_ctl
1036  * remove the x_header, and adjust the sopt_name, sopt_val and sopt_valsize.
1037  */
1038 int
1039 ip_fw3_ctl_x(struct sockopt *sopt)
1040 {
1041 	ip_fw_x_header *x_header;
1042 	x_header = (ip_fw_x_header *)(sopt->sopt_val);
1043 	sopt->sopt_name = x_header->opcode;
1044 	sopt->sopt_valsize -= sizeof(ip_fw_x_header);
1045 	bcopy(++x_header, sopt->sopt_val, sopt->sopt_valsize);
1046 	return ip_fw3_ctl(sopt);
1047 }
1048 
1049 
1050 /**
1051  * {set|get}sockopt parser.
1052  */
1053 int
1054 ip_fw3_ctl(struct sockopt *sopt)
1055 {
1056 	int error = 0;
1057 	switch (sopt->sopt_name) {
1058 		case IP_FW_X:
1059 			ip_fw3_ctl_x(sopt);
1060 			break;
1061 		case IP_FW_GET:
1062 		case IP_FW_MODULE:
1063 		case IP_FW_FLUSH:
1064 		case IP_FW_ADD:
1065 		case IP_FW_DEL:
1066 		case IP_FW_ZERO:
1067 		case IP_FW_RESETLOG:
1068 			error = ip_fw3_ctl_sockopt(sopt);
1069 			break;
1070 		case IP_FW_SET_GET:
1071 		case IP_FW_SET_MOVE_RULE:
1072 		case IP_FW_SET_MOVE_SET:
1073 		case IP_FW_SET_SWAP:
1074 		case IP_FW_SET_TOGGLE:
1075 			error = ip_fw3_ctl_set_sockopt(sopt);
1076 			break;
1077 		case IP_FW_NAT_ADD:
1078 		case IP_FW_NAT_DEL:
1079 		case IP_FW_NAT_FLUSH:
1080 		case IP_FW_NAT_GET:
1081 		case IP_FW_NAT_GET_RECORD:
1082 			if (ip_fw3_ctl_nat_ptr != NULL) {
1083 				error = ip_fw3_ctl_nat_ptr(sopt);
1084 			}
1085 			break;
1086 		case IP_DUMMYNET_GET:
1087 		case IP_DUMMYNET_CONFIGURE:
1088 		case IP_DUMMYNET_DEL:
1089 		case IP_DUMMYNET_FLUSH:
1090 			error = ip_dn_sockopt(sopt);
1091 			break;
1092 		case IP_FW_STATE_ADD:
1093 		case IP_FW_STATE_DEL:
1094 		case IP_FW_STATE_FLUSH:
1095 		case IP_FW_STATE_GET:
1096 			if (ip_fw3_ctl_state_ptr != NULL) {
1097 				error = ip_fw3_ctl_state_ptr(sopt);
1098 			}
1099 			break;
1100 		case IP_FW_TABLE_CREATE:
1101 		case IP_FW_TABLE_DELETE:
1102 		case IP_FW_TABLE_APPEND:
1103 		case IP_FW_TABLE_REMOVE:
1104 		case IP_FW_TABLE_LIST:
1105 		case IP_FW_TABLE_FLUSH:
1106 		case IP_FW_TABLE_SHOW:
1107 		case IP_FW_TABLE_TEST:
1108 		case IP_FW_TABLE_RENAME:
1109 			if (ip_fw3_ctl_table_ptr != NULL) {
1110 				error = ip_fw3_ctl_table_ptr(sopt);
1111 			}
1112 			break;
1113 		case IP_FW_SYNC_SHOW_CONF:
1114 		case IP_FW_SYNC_SHOW_STATUS:
1115 		case IP_FW_SYNC_EDGE_CONF:
1116 		case IP_FW_SYNC_EDGE_START:
1117 		case IP_FW_SYNC_EDGE_STOP:
1118 		case IP_FW_SYNC_EDGE_TEST:
1119 		case IP_FW_SYNC_EDGE_CLEAR:
1120 		case IP_FW_SYNC_CENTRE_CONF:
1121 		case IP_FW_SYNC_CENTRE_START:
1122 		case IP_FW_SYNC_CENTRE_STOP:
1123 		case IP_FW_SYNC_CENTRE_TEST:
1124 		case IP_FW_SYNC_CENTRE_CLEAR:
1125 			if (ip_fw3_ctl_sync_ptr != NULL) {
1126 				error = ip_fw3_ctl_sync_ptr(sopt);
1127 			}
1128 			break;
1129 		default:
1130 			kprintf("ip_fw3_ctl invalid option %d\n",
1131 				sopt->sopt_name);
1132 			error = EINVAL;
1133 	}
1134 	return error;
1135 }
1136 
1137 int
1138 ip_fw3_ctl_sockopt(struct sockopt *sopt)
1139 {
1140 	int error = 0, rulenum;
1141 
1142 	switch (sopt->sopt_name) {
1143 		case IP_FW_GET:
1144 			error = ip_fw3_ctl_get_rules(sopt);
1145 			break;
1146 		case IP_FW_MODULE:
1147 			error = ip_fw3_ctl_get_modules(sopt);
1148 			break;
1149 		case IP_FW_FLUSH:
1150 			ip_fw3_ctl_flush_rule(0);
1151 			break;
1152 		case IP_FW_ADD:
1153 			error = ip_fw3_ctl_add_rule(sopt);
1154 			break;
1155 		case IP_FW_DEL:
1156 			error = ip_fw3_ctl_delete_rule(sopt);
1157 			break;
1158 		case IP_FW_ZERO:
1159 		case IP_FW_RESETLOG: /* argument is an int, the rule number */
1160 			rulenum = 0;
1161 			if (sopt->sopt_valsize != 0) {
1162 				error = soopt_to_kbuf(sopt, &rulenum,
1163 						sizeof(int), sizeof(int));
1164 				if (error) {
1165 					break;
1166 				}
1167 			}
1168 			error = ip_fw3_ctl_zero_entry(rulenum,
1169 					sopt->sopt_name == IP_FW_RESETLOG);
1170 			break;
1171 		default:
1172 			kprintf("ip_fw3_ctl invalid option %d\n",
1173 				sopt->sopt_name);
1174 			error = EINVAL;
1175 	}
1176 	return error;
1177 }
1178 
1179 int
1180 ip_fw3_check_in(void *arg, struct mbuf **m0, struct ifnet *ifp, int dir)
1181 {
1182 	struct ip_fw_args args;
1183 	struct mbuf *m = *m0;
1184 	struct m_tag *mtag;
1185 	int tee = 0, error = 0, ret;
1186 	// again:
1187 	if (m->m_pkthdr.fw_flags & DUMMYNET_MBUF_TAGGED) {
1188 		/* Extract info from dummynet tag */
1189 		mtag = m_tag_find(m, PACKET_TAG_DUMMYNET, NULL);
1190 		KKASSERT(mtag != NULL);
1191 		args.rule = ((struct dn_pkt *)m_tag_data(mtag))->dn_priv;
1192 		KKASSERT(args.rule != NULL);
1193 
1194 		m_tag_delete(m, mtag);
1195 		m->m_pkthdr.fw_flags &= ~DUMMYNET_MBUF_TAGGED;
1196 	} else {
1197 		args.rule = NULL;
1198 	}
1199 
1200 	args.eh = NULL;
1201 	args.oif = NULL;
1202 	args.m = m;
1203 	ret = ip_fw3_chk(&args);
1204 	m = args.m;
1205 
1206 	if (m == NULL) {
1207 		error = EACCES;
1208 		goto back;
1209 	}
1210 	switch (ret) {
1211 		case IP_FW_PASS:
1212 			break;
1213 
1214 		case IP_FW_DENY:
1215 			m_freem(m);
1216 			m = NULL;
1217 			error = EACCES;
1218 			break;
1219 
1220 		case IP_FW_DUMMYNET:
1221 			/* Send packet to the appropriate pipe */
1222 			m = ip_fw3_dummynet_io(m, args.cookie, DN_TO_IP_IN,
1223 			    &args);
1224 			break;
1225 		case IP_FW_TEE:
1226 			tee = 1;
1227 			/* FALL THROUGH */
1228 		case IP_FW_DIVERT:
1229 			/*
1230 			 * Must clear bridge tag when changing
1231 			 */
1232 			m->m_pkthdr.fw_flags &= ~BRIDGE_MBUF_TAGGED;
1233 			if (ip_divert_p != NULL) {
1234 				m = ip_divert_p(m, tee, 1);
1235 			} else {
1236 				m_freem(m);
1237 				m = NULL;
1238 				/* not sure this is the right error msg */
1239 				error = EACCES;
1240 			}
1241 			break;
1242 		case IP_FW_NAT:
1243 			break;
1244 		case IP_FW_ROUTE:
1245 			break;
1246 		default:
1247 			panic("unknown ipfw3 return value: %d", ret);
1248 	}
1249 back:
1250 	*m0 = m;
1251 	return error;
1252 }
1253 
1254 int
1255 ip_fw3_check_out(void *arg, struct mbuf **m0, struct ifnet *ifp, int dir)
1256 {
1257 	struct ip_fw_args args;
1258 	struct mbuf *m = *m0;
1259 	struct m_tag *mtag;
1260 	int tee = 0, error = 0, ret;
1261 	// again:
1262 	if (m->m_pkthdr.fw_flags & DUMMYNET_MBUF_TAGGED) {
1263 		/* Extract info from dummynet tag */
1264 		mtag = m_tag_find(m, PACKET_TAG_DUMMYNET, NULL);
1265 		KKASSERT(mtag != NULL);
1266 		args.rule = ((struct dn_pkt *)m_tag_data(mtag))->dn_priv;
1267 		KKASSERT(args.rule != NULL);
1268 
1269 		m_tag_delete(m, mtag);
1270 		m->m_pkthdr.fw_flags &= ~DUMMYNET_MBUF_TAGGED;
1271 	} else {
1272 		args.rule = NULL;
1273 	}
1274 
1275 	args.eh = NULL;
1276 	args.m = m;
1277 	args.oif = ifp;
1278 	ret = ip_fw3_chk(&args);
1279 	m = args.m;
1280 
1281 	if (m == NULL) {
1282 		error = EACCES;
1283 		goto back;
1284 	}
1285 
1286 	switch (ret) {
1287 		case IP_FW_PASS:
1288 			break;
1289 
1290 		case IP_FW_DENY:
1291 			m_freem(m);
1292 			m = NULL;
1293 			error = EACCES;
1294 			break;
1295 
1296 		case IP_FW_DUMMYNET:
1297 			m = ip_fw3_dummynet_io(m, args.cookie, DN_TO_IP_OUT,
1298 			    &args);
1299 			break;
1300 
1301 		case IP_FW_TEE:
1302 			tee = 1;
1303 			/* FALL THROUGH */
1304 
1305 		case IP_FW_DIVERT:
1306 			if (ip_divert_p != NULL) {
1307 				m = ip_divert_p(m, tee, 0);
1308 			} else {
1309 				m_freem(m);
1310 				m = NULL;
1311 				/* not sure this is the right error msg */
1312 				error = EACCES;
1313 			}
1314 			break;
1315 
1316 		case IP_FW_NAT:
1317 			break;
1318 		case IP_FW_ROUTE:
1319 			break;
1320 		default:
1321 			panic("unknown ipfw3 return value: %d", ret);
1322 	}
1323 back:
1324 	*m0 = m;
1325 	return error;
1326 }
1327 
1328 void
1329 ip_fw3_hook(void)
1330 {
1331 	struct pfil_head *pfh;
1332 	IPFW_ASSERT_CFGPORT(&curthread->td_msgport);
1333 
1334 	pfh = pfil_head_get(PFIL_TYPE_AF, AF_INET);
1335 	if (pfh == NULL)
1336 		return;
1337 
1338 	pfil_add_hook(ip_fw3_check_in, NULL, PFIL_IN, pfh);
1339 	pfil_add_hook(ip_fw3_check_out, NULL, PFIL_OUT, pfh);
1340 }
1341 
1342 void
1343 ip_fw3_dehook(void)
1344 {
1345 	struct pfil_head *pfh;
1346 
1347 	IPFW_ASSERT_CFGPORT(&curthread->td_msgport);
1348 
1349 	pfh = pfil_head_get(PFIL_TYPE_AF, AF_INET);
1350 	if (pfh == NULL)
1351 		return;
1352 
1353 	pfil_remove_hook(ip_fw3_check_in, NULL, PFIL_IN, pfh);
1354 	pfil_remove_hook(ip_fw3_check_out, NULL, PFIL_OUT, pfh);
1355 }
1356 
1357 void
1358 ip_fw3_sysctl_enable_dispatch(netmsg_t nmsg)
1359 {
1360 	struct lwkt_msg *lmsg = &nmsg->lmsg;
1361 	int enable = lmsg->u.ms_result;
1362 
1363 	if (sysctl_var_fw3_enable == enable)
1364 		goto reply;
1365 
1366 	sysctl_var_fw3_enable = enable;
1367 	if (sysctl_var_fw3_enable)
1368 		ip_fw3_hook();
1369 	else
1370 		ip_fw3_dehook();
1371 
1372 reply:
1373 	lwkt_replymsg(lmsg, 0);
1374 }
1375 
1376 int
1377 ip_fw3_sysctl_enable(SYSCTL_HANDLER_ARGS)
1378 {
1379 	struct netmsg_base nmsg;
1380 	struct lwkt_msg *lmsg;
1381 	int enable, error;
1382 
1383 	enable = sysctl_var_fw3_enable;
1384 	error = sysctl_handle_int(oidp, &enable, 0, req);
1385 	if (error || req->newptr == NULL)
1386 		return error;
1387 
1388 	netmsg_init(&nmsg, NULL, &curthread->td_msgport,
1389 			0, ip_fw3_sysctl_enable_dispatch);
1390 	lmsg = &nmsg.lmsg;
1391 	lmsg->u.ms_result = enable;
1392 
1393 	return lwkt_domsg(IPFW_CFGPORT, lmsg, 0);
1394 }
1395 
1396 int
1397 ip_fw3_sysctl_autoinc_step(SYSCTL_HANDLER_ARGS)
1398 {
1399 	return sysctl_int_range(oidp, arg1, arg2, req,
1400 			IPFW_AUTOINC_STEP_MIN, IPFW_AUTOINC_STEP_MAX);
1401 }
1402 
1403 void
1404 ctx_init_dispatch(netmsg_t nmsg)
1405 {
1406 	struct netmsg_ipfw *fwmsg = (struct netmsg_ipfw *)nmsg;
1407 	struct ipfw3_context *ctx;
1408 	struct ip_fw *def_rule;
1409 
1410 	ctx = kmalloc(LEN_FW3_CTX, M_IPFW3, M_WAITOK | M_ZERO);
1411 	fw3_ctx[mycpuid] = ctx;
1412 	ctx->sets = IPFW_ALL_SETS;
1413 
1414 	def_rule = kmalloc(LEN_FW3, M_IPFW3, M_WAITOK | M_ZERO);
1415 	def_rule->act_ofs = 0;
1416 	def_rule->rulenum = IPFW_DEFAULT_RULE;
1417 	def_rule->cmd_len = 2;
1418 	def_rule->set = IPFW_DEFAULT_SET;
1419 
1420 	def_rule->cmd[0].len = LEN_OF_IPFWINSN;
1421 	def_rule->cmd[0].module = MODULE_BASIC_ID;
1422 #ifdef IPFIREWALL_DEFAULT_TO_ACCEPT
1423 	def_rule->cmd[0].opcode = O_BASIC_ACCEPT;
1424 #else
1425 	if (filters_default_to_accept)
1426 		def_rule->cmd[0].opcode = O_BASIC_ACCEPT;
1427 	else
1428 		def_rule->cmd[0].opcode = O_BASIC_DENY;
1429 #endif
1430 
1431 	/* Install the default rule */
1432 	ctx->default_rule = def_rule;
1433 	ctx->rules = def_rule;
1434 
1435 	/*
1436 	 * if sibiling in last CPU is exists,
1437 	 * then it's sibling should be current rule
1438 	 */
1439 	if (fwmsg->sibling != NULL) {
1440 		fwmsg->sibling->sibling = def_rule;
1441 	}
1442 	/* prepare for next CPU */
1443 	fwmsg->sibling = def_rule;
1444 
1445 	netisr_forwardmsg_all(&nmsg->base, mycpuid + 1);
1446 }
1447 
1448 void
1449 init_dispatch(netmsg_t nmsg)
1450 {
1451 	struct netmsg_ipfw fwmsg;
1452 	int error = 0;
1453 	if (IPFW3_LOADED) {
1454 		kprintf("ipfw3 already loaded\n");
1455 		error = EEXIST;
1456 		goto reply;
1457 	}
1458 
1459 	bzero(&fwmsg, sizeof(fwmsg));
1460 	netmsg_init(&fwmsg.base, NULL, &curthread->td_msgport,
1461 			0, ctx_init_dispatch);
1462 	netisr_domsg(&fwmsg.base, 0);
1463 
1464 	ip_fw_chk_ptr = ip_fw3_chk;
1465 	ip_fw_ctl_x_ptr = ip_fw3_ctl_x;
1466 	ip_fw_dn_io_ptr = ip_fw3_dummynet_io;
1467 
1468 	kprintf("ipfw3 initialized, default to %s\n",
1469 			filters_default_to_accept ? "accept" : "deny");
1470 
1471 	ip_fw3_loaded = 1;
1472 	if (sysctl_var_fw3_enable)
1473 		ip_fw3_hook();
1474 reply:
1475 	lwkt_replymsg(&nmsg->lmsg, error);
1476 }
1477 
1478 int
1479 ip_fw3_init(void)
1480 {
1481 	struct netmsg_base smsg;
1482 	int error;
1483 
1484 	init_module();
1485 	netmsg_init(&smsg, NULL, &curthread->td_msgport,
1486 			0, init_dispatch);
1487 	error = lwkt_domsg(IPFW_CFGPORT, &smsg.lmsg, 0);
1488 	return error;
1489 }
1490 
1491 #ifdef KLD_MODULE
1492 
1493 void
1494 fini_dispatch(netmsg_t nmsg)
1495 {
1496 	int error = 0, cpu;
1497 
1498 	ip_fw3_loaded = 0;
1499 
1500 	ip_fw3_dehook();
1501 	netmsg_service_sync();
1502 	ip_fw_chk_ptr = NULL;
1503 	ip_fw_ctl_x_ptr = NULL;
1504 	ip_fw_dn_io_ptr = NULL;
1505 	ip_fw3_ctl_flush_rule(1);
1506 	/* Free pre-cpu context */
1507 	for (cpu = 0; cpu < ncpus; ++cpu) {
1508 		if (fw3_ctx[cpu] != NULL) {
1509 			kfree(fw3_ctx[cpu], M_IPFW3);
1510 			fw3_ctx[cpu] = NULL;
1511 		}
1512 	}
1513 	kprintf("ipfw3 unloaded\n");
1514 
1515 	lwkt_replymsg(&nmsg->lmsg, error);
1516 }
1517 
1518 int
1519 ip_fw3_fini(void)
1520 {
1521 	struct netmsg_base smsg;
1522 
1523 	netmsg_init(&smsg, NULL, &curthread->td_msgport,
1524 			0, fini_dispatch);
1525 	return lwkt_domsg(IPFW_CFGPORT, &smsg.lmsg, 0);
1526 }
1527 
1528 #endif	/* KLD_MODULE */
1529 
1530 static int
1531 ip_fw3_modevent(module_t mod, int type, void *unused)
1532 {
1533 	int err = 0;
1534 
1535 	switch (type) {
1536 		case MOD_LOAD:
1537 			err = ip_fw3_init();
1538 			break;
1539 
1540 		case MOD_UNLOAD:
1541 
1542 #ifndef KLD_MODULE
1543 			kprintf("ipfw3 statically compiled, cannot unload\n");
1544 			err = EBUSY;
1545 #else
1546 			err = ip_fw3_fini();
1547 #endif
1548 			break;
1549 		default:
1550 			break;
1551 	}
1552 	return err;
1553 }
1554 
1555 static moduledata_t ipfw3mod = {
1556 	"ipfw3",
1557 	ip_fw3_modevent,
1558 	0
1559 };
1560 /* ipfw3 must init before ipfw3_basic */
1561 DECLARE_MODULE(ipfw3, ipfw3mod, SI_SUB_PROTO_END, SI_ORDER_FIRST);
1562 MODULE_VERSION(ipfw3, 1);
1563