2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
4 * Copyright (c) 2004 Andre Oppermann, Internet Business Solutions AG
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
20 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 #include <sys/cdefs.h>
30 __FBSDID("$FreeBSD$");
34 #include "opt_inet6.h"
36 #error IPFIREWALL requires INET.
39 #include <sys/param.h>
40 #include <sys/systm.h>
41 #include <sys/malloc.h>
43 #include <sys/module.h>
44 #include <sys/kernel.h>
46 #include <sys/rwlock.h>
47 #include <sys/socket.h>
48 #include <sys/sysctl.h>
51 #include <net/if_var.h>
52 #include <net/route.h>
53 #include <net/ethernet.h>
57 #include <netinet/in.h>
58 #include <netinet/in_systm.h>
59 #include <netinet/ip.h>
60 #include <netinet/ip_var.h>
61 #include <netinet/ip_fw.h>
63 #include <netinet/ip6.h>
64 #include <netinet6/ip6_var.h>
65 #include <netinet6/scope6_var.h>
68 #include <netgraph/ng_ipfw.h>
70 #include <netpfil/ipfw/ip_fw_private.h>
72 #include <machine/in_cksum.h>
74 VNET_DEFINE_STATIC(int, fw_enable) = 1;
75 #define V_fw_enable VNET(fw_enable)
78 VNET_DEFINE_STATIC(int, fw6_enable) = 1;
79 #define V_fw6_enable VNET(fw6_enable)
82 VNET_DEFINE_STATIC(int, fwlink_enable) = 0;
83 #define V_fwlink_enable VNET(fwlink_enable)
85 int ipfw_chg_hook(SYSCTL_HANDLER_ARGS);
87 /* Forward declarations. */
88 static int ipfw_divert(struct mbuf **, struct ip_fw_args *, bool);
94 SYSCTL_DECL(_net_inet_ip_fw);
95 SYSCTL_PROC(_net_inet_ip_fw, OID_AUTO, enable,
96 CTLFLAG_VNET | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE3,
97 &VNET_NAME(fw_enable), 0, ipfw_chg_hook, "I", "Enable ipfw");
99 SYSCTL_DECL(_net_inet6_ip6_fw);
100 SYSCTL_PROC(_net_inet6_ip6_fw, OID_AUTO, enable,
101 CTLFLAG_VNET | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE3,
102 &VNET_NAME(fw6_enable), 0, ipfw_chg_hook, "I", "Enable ipfw+6");
105 SYSCTL_DECL(_net_link_ether);
106 SYSCTL_PROC(_net_link_ether, OID_AUTO, ipfw,
107 CTLFLAG_VNET | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE3,
108 &VNET_NAME(fwlink_enable), 0, ipfw_chg_hook, "I",
109 "Pass ether pkts through firewall");
113 #endif /* SYSCTL_NODE */
116 * The pfilter hook to pass packets to ipfw_chk and then to
117 * dummynet, divert, netgraph or other modules.
118 * The packet may be consumed.
121 ipfw_check_packet(struct mbuf **m0, struct ifnet *ifp, int flags,
122 void *ruleset __unused, struct inpcb *inp)
124 struct ip_fw_args args;
129 args.flags = (flags & PFIL_IN) ? IPFW_ARGS_IN : IPFW_ARGS_OUT;
130 dir = (flags & PFIL_IN) ? DIR_IN : DIR_OUT;
133 * extract and remove the tag if present. If we are left
134 * with onepass, optimize the outgoing path.
136 tag = m_tag_locate(*m0, MTAG_IPFW_RULE, 0, NULL);
138 args.rule = *((struct ipfw_rule_ref *)(tag+1));
139 m_tag_delete(*m0, tag);
140 if (args.rule.info & IPFW_ONEPASS)
142 args.flags |= IPFW_ARGS_REF;
149 ipfw = ipfw_chk(&args);
152 KASSERT(*m0 != NULL || ipfw == IP_FW_DENY, ("%s: m0 is NULL",
158 /* next_hop may be set by ipfw_chk */
159 if ((args.flags & (IPFW_ARGS_NH4 | IPFW_ARGS_NH4PTR |
160 IPFW_ARGS_NH6 | IPFW_ARGS_NH6PTR)) == 0)
162 #if (!defined(INET6) && !defined(INET))
169 if (args.flags & (IPFW_ARGS_NH4 | IPFW_ARGS_NH4PTR)) {
170 MPASS((args.flags & (IPFW_ARGS_NH4 |
171 IPFW_ARGS_NH4PTR)) != (IPFW_ARGS_NH4 |
173 MPASS((args.flags & (IPFW_ARGS_NH6 |
174 IPFW_ARGS_NH6PTR)) == 0);
175 len = sizeof(struct sockaddr_in);
176 psa = (args.flags & IPFW_ARGS_NH4) ?
177 &args.hopstore : args.next_hop;
178 if (in_localip(satosin(psa)->sin_addr))
179 (*m0)->m_flags |= M_FASTFWD_OURS;
180 (*m0)->m_flags |= M_IP_NEXTHOP;
184 if (args.flags & (IPFW_ARGS_NH6 | IPFW_ARGS_NH6PTR)) {
185 MPASS((args.flags & (IPFW_ARGS_NH6 |
186 IPFW_ARGS_NH6PTR)) != (IPFW_ARGS_NH6 |
188 MPASS((args.flags & (IPFW_ARGS_NH4 |
189 IPFW_ARGS_NH4PTR)) == 0);
190 len = sizeof(struct sockaddr_in6);
191 psa = args.next_hop6;
192 (*m0)->m_flags |= M_IP6_NEXTHOP;
196 * Incoming packets should not be tagged so we do not
197 * m_tag_find. Outgoing packets may be tagged, so we
198 * reuse the tag if present.
200 tag = (flags & PFIL_IN) ? NULL :
201 m_tag_find(*m0, PACKET_TAG_IPFORWARD, NULL);
203 m_tag_unlink(*m0, tag);
205 tag = m_tag_get(PACKET_TAG_IPFORWARD, len,
212 if ((args.flags & IPFW_ARGS_NH6) == 0)
213 bcopy(psa, tag + 1, len);
214 m_tag_prepend(*m0, tag);
217 /* IPv6 next hop needs additional handling */
218 if (args.flags & (IPFW_ARGS_NH6 | IPFW_ARGS_NH6PTR)) {
219 struct sockaddr_in6 *sa6;
221 sa6 = satosin6(tag + 1);
222 if (args.flags & IPFW_ARGS_NH6) {
223 sa6->sin6_family = AF_INET6;
224 sa6->sin6_len = sizeof(*sa6);
225 sa6->sin6_addr = args.hopstore6.sin6_addr;
226 sa6->sin6_port = args.hopstore6.sin6_port;
228 args.hopstore6.sin6_scope_id;
231 * If nh6 address is link-local we should convert
232 * it to kernel internal form before doing any
235 if (sa6_embedscope(sa6, V_ip6_use_defzone) != 0) {
239 if (in6_localip(&sa6->sin6_addr))
240 (*m0)->m_flags |= M_FASTFWD_OURS;
244 #endif /* INET || INET6 */
252 if (ip_dn_io_ptr == NULL) {
256 MPASS(args.flags & IPFW_ARGS_REF);
257 if (mtod(*m0, struct ip *)->ip_v == 4)
258 (void )ip_dn_io_ptr(m0, dir, &args);
259 else if (mtod(*m0, struct ip *)->ip_v == 6)
260 (void )ip_dn_io_ptr(m0, dir | PROTO_IPV6, &args);
266 * XXX should read the return value.
267 * dummynet normally eats the packet and sets *m0=NULL
268 * unless the packet can be sent immediately. In this
269 * case args is updated and we should re-run the
270 * check without clearing args.
279 if (ip_divert_ptr == NULL) {
283 MPASS(args.flags & IPFW_ARGS_REF);
284 (void )ipfw_divert(m0, &args, ipfw == IP_FW_TEE);
285 /* continue processing for the original packet (tee). */
293 if (ng_ipfw_input_p == NULL) {
297 MPASS(args.flags & IPFW_ARGS_REF);
298 (void )ng_ipfw_input_p(m0, &args, ipfw == IP_FW_NGTEE);
299 if (ipfw == IP_FW_NGTEE) /* ignore errors for NGTEE */
300 goto again; /* continue with packet */
305 /* honor one-pass in case of successful nat */
311 goto again; /* continue with packet */
318 KASSERT(0, ("%s: unknown retval", __func__));
321 if (ret != PFIL_PASS) {
331 * ipfw processing for ethernet packets (in and out).
334 ipfw_check_frame(struct mbuf **m0, struct ifnet *ifp, int dir,
335 void *ruleset __unused, struct inpcb *inp)
337 struct ip_fw_args args;
338 struct ether_header save_eh;
339 struct ether_header *eh;
345 args.flags = IPFW_ARGS_ETHER;
346 args.flags |= (dir & PFIL_IN) ? IPFW_ARGS_IN : IPFW_ARGS_OUT;
348 /* fetch start point from rule, if any. remove the tag if present. */
349 mtag = m_tag_locate(*m0, MTAG_IPFW_RULE, 0, NULL);
351 args.rule = *((struct ipfw_rule_ref *)(mtag+1));
352 m_tag_delete(*m0, mtag);
353 if (args.rule.info & IPFW_ONEPASS)
355 args.flags |= IPFW_ARGS_REF;
358 /* I need some amt of data to be contiguous */
360 i = min(m->m_pkthdr.len, max_protohdr);
368 eh = mtod(m, struct ether_header *);
369 save_eh = *eh; /* save copy for restore below */
370 m_adj(m, ETHER_HDR_LEN); /* strip ethernet header */
372 args.m = m; /* the packet we are looking at */
374 args.eh = &save_eh; /* MAC header for bridged/MAC packets */
375 args.inp = inp; /* used by ipfw uid/gid/jail rules */
380 * Restore Ethernet header, as needed, in case the
381 * mbuf chain was replaced by ipfw.
383 M_PREPEND(m, ETHER_HDR_LEN, M_NOWAIT);
388 if (eh != mtod(m, struct ether_header *))
389 bcopy(&save_eh, mtod(m, struct ether_header *),
395 /* Check result of ipfw_chk() */
405 if (ip_dn_io_ptr == NULL) {
410 dir = (dir & PFIL_IN) ? DIR_IN : DIR_OUT;
411 MPASS(args.flags & IPFW_ARGS_REF);
412 ip_dn_io_ptr(&m, dir | PROTO_LAYER2, &args);
413 return (PFIL_CONSUMED);
417 if (ng_ipfw_input_p == NULL) {
421 MPASS(args.flags & IPFW_ARGS_REF);
422 (void )ng_ipfw_input_p(m0, &args, i == IP_FW_NGTEE);
423 if (i == IP_FW_NGTEE) /* ignore errors for NGTEE */
424 goto again; /* continue with packet */
429 KASSERT(0, ("%s: unknown retval", __func__));
432 if (ret != PFIL_PASS) {
441 /* do the divert, return 1 on error 0 on success */
443 ipfw_divert(struct mbuf **m0, struct ip_fw_args *args, bool tee)
446 * ipfw_chk() has already tagged the packet with the divert tag.
447 * If tee is set, copy packet and return original.
448 * If not tee, consume packet and send it to divert socket.
451 struct ip *ip = mtod(*m0, struct ip *);
454 /* Cloning needed for tee? */
456 clone = *m0; /* use the original mbuf */
459 clone = m_dup(*m0, M_NOWAIT);
460 /* If we cannot duplicate the mbuf, we sacrifice the divert
461 * chain and continue with the tee-ed packet.
468 * Divert listeners can normally handle non-fragmented packets,
469 * but we can only reass in the non-tee case.
470 * This means that listeners on a tee rule may get fragments,
471 * and have to live with that.
472 * Note that we now have the 'reass' ipfw option so if we care
473 * we can do it before a 'tee'.
475 if (tee == false) switch (ip->ip_v) {
477 if (ntohs(ip->ip_off) & (IP_MF | IP_OFFMASK)) {
481 reass = ip_reass(clone); /* Reassemble packet. */
483 return 0; /* not an error */
484 /* if reass = NULL then it was consumed by ip_reass */
486 * IP header checksum fixup after reassembly and leave header
487 * in network byte order.
489 ip = mtod(reass, struct ip *);
490 hlen = ip->ip_hl << 2;
492 if (hlen == sizeof(struct ip))
493 ip->ip_sum = in_cksum_hdr(ip);
495 ip->ip_sum = in_cksum(reass, hlen);
500 case IPV6_VERSION >> 4:
502 struct ip6_hdr *const ip6 = mtod(clone, struct ip6_hdr *);
504 if (ip6->ip6_nxt == IPPROTO_FRAGMENT) {
507 off = sizeof(struct ip6_hdr);
508 nxt = frag6_input(&clone, &off, 0);
509 if (nxt == IPPROTO_DONE)
517 /* attach a tag to the packet with the reinject info */
518 tag = m_tag_alloc(MTAG_IPFW_RULE, 0,
519 sizeof(struct ipfw_rule_ref), M_NOWAIT);
524 *((struct ipfw_rule_ref *)(tag+1)) = args->rule;
525 m_tag_prepend(clone, tag);
527 /* Do the dirty job... */
528 ip_divert_ptr(clone, args->flags & IPFW_ARGS_IN);
533 * attach or detach hooks for a given protocol family
535 VNET_DEFINE_STATIC(pfil_hook_t, ipfw_inet_hook);
536 #define V_ipfw_inet_hook VNET(ipfw_inet_hook)
538 VNET_DEFINE_STATIC(pfil_hook_t, ipfw_inet6_hook);
539 #define V_ipfw_inet6_hook VNET(ipfw_inet6_hook)
541 VNET_DEFINE_STATIC(pfil_hook_t, ipfw_link_hook);
542 #define V_ipfw_link_hook VNET(ipfw_link_hook)
545 ipfw_hook(int onoff, int pf)
547 struct pfil_hook_args pha;
548 struct pfil_link_args pla;
551 pha.pa_version = PFIL_VERSION;
552 pha.pa_flags = PFIL_IN | PFIL_OUT;
553 pha.pa_modname = "ipfw";
554 pha.pa_ruleset = NULL;
556 pla.pa_version = PFIL_VERSION;
557 pla.pa_flags = PFIL_IN | PFIL_OUT |
558 PFIL_HEADPTR | PFIL_HOOKPTR;
562 pha.pa_func = ipfw_check_packet;
563 pha.pa_type = PFIL_TYPE_IP4;
564 pha.pa_rulname = "default";
565 h = &V_ipfw_inet_hook;
566 pla.pa_head = V_inet_pfil_head;
570 pha.pa_func = ipfw_check_packet;
571 pha.pa_type = PFIL_TYPE_IP6;
572 pha.pa_rulname = "default6";
573 h = &V_ipfw_inet6_hook;
574 pla.pa_head = V_inet6_pfil_head;
578 pha.pa_func = ipfw_check_frame;
579 pha.pa_type = PFIL_TYPE_ETHERNET;
580 pha.pa_rulname = "default-link";
581 h = &V_ipfw_link_hook;
582 pla.pa_head = V_link_pfil_head;
587 *h = pfil_add_hook(&pha);
589 (void)pfil_link(&pla);
592 pfil_remove_hook(*h);
598 ipfw_attach_hooks(int arg)
602 if (arg == 0) /* detach */
603 ipfw_hook(0, AF_INET);
604 else if (V_fw_enable && ipfw_hook(1, AF_INET) != 0) {
605 error = ENOENT; /* see ip_fw_pfil.c::ipfw_hook() */
606 printf("ipfw_hook() error\n");
609 if (arg == 0) /* detach */
610 ipfw_hook(0, AF_INET6);
611 else if (V_fw6_enable && ipfw_hook(1, AF_INET6) != 0) {
613 printf("ipfw6_hook() error\n");
616 if (arg == 0) /* detach */
617 ipfw_hook(0, AF_LINK);
618 else if (V_fwlink_enable && ipfw_hook(1, AF_LINK) != 0) {
620 printf("ipfw_link_hook() error\n");
626 ipfw_chg_hook(SYSCTL_HANDLER_ARGS)
632 if (arg1 == &V_fw_enable)
635 else if (arg1 == &V_fw6_enable)
638 else if (arg1 == &V_fwlink_enable)
643 newval = *(int *)arg1;
644 /* Handle sysctl change */
645 error = sysctl_handle_int(oidp, &newval, 0, req);
650 /* Formalize new value */
651 newval = (newval) ? 1 : 0;
653 if (*(int *)arg1 == newval)
656 error = ipfw_hook(newval, af);
659 *(int *)arg1 = newval;