2 * Copyright (c) 2004 Andre Oppermann, Internet Business Solutions AG
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
14 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 #include <sys/cdefs.h>
28 __FBSDID("$FreeBSD$");
32 #include "opt_inet6.h"
34 #error IPFIREWALL requires INET.
37 #include <sys/param.h>
38 #include <sys/systm.h>
39 #include <sys/malloc.h>
41 #include <sys/module.h>
42 #include <sys/kernel.h>
44 #include <sys/rwlock.h>
45 #include <sys/socket.h>
46 #include <sys/sysctl.h>
49 #include <net/route.h>
50 #include <net/ethernet.h>
54 #include <netinet/in.h>
55 #include <netinet/in_systm.h>
56 #include <netinet/ip.h>
57 #include <netinet/ip_var.h>
58 #include <netinet/ip_fw.h>
60 #include <netinet/ip6.h>
61 #include <netinet6/ip6_var.h>
62 #include <netinet6/scope6_var.h>
65 #include <netgraph/ng_ipfw.h>
67 #include <netpfil/ipfw/ip_fw_private.h>
69 #include <machine/in_cksum.h>
71 static VNET_DEFINE(int, fw_enable) = 1;
72 #define V_fw_enable VNET(fw_enable)
75 static VNET_DEFINE(int, fw6_enable) = 1;
76 #define V_fw6_enable VNET(fw6_enable)
79 static VNET_DEFINE(int, fwlink_enable) = 0;
80 #define V_fwlink_enable VNET(fwlink_enable)
82 int ipfw_chg_hook(SYSCTL_HANDLER_ARGS);
84 /* Forward declarations. */
85 static int ipfw_divert(struct mbuf **, int, struct ipfw_rule_ref *, int);
86 int ipfw_check_packet(void *, struct mbuf **, struct ifnet *, int,
88 int ipfw_check_frame(void *, struct mbuf **, struct ifnet *, int,
95 SYSCTL_DECL(_net_inet_ip_fw);
96 SYSCTL_PROC(_net_inet_ip_fw, OID_AUTO, enable,
97 CTLFLAG_VNET | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE3,
98 &VNET_NAME(fw_enable), 0, ipfw_chg_hook, "I", "Enable ipfw");
100 SYSCTL_DECL(_net_inet6_ip6_fw);
101 SYSCTL_PROC(_net_inet6_ip6_fw, OID_AUTO, enable,
102 CTLFLAG_VNET | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE3,
103 &VNET_NAME(fw6_enable), 0, ipfw_chg_hook, "I", "Enable ipfw+6");
106 SYSCTL_DECL(_net_link_ether);
107 SYSCTL_PROC(_net_link_ether, OID_AUTO, ipfw,
108 CTLFLAG_VNET | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE3,
109 &VNET_NAME(fwlink_enable), 0, ipfw_chg_hook, "I",
110 "Pass ether pkts through firewall");
114 #endif /* SYSCTL_NODE */
117 * The pfilter hook to pass packets to ipfw_chk and then to
118 * dummynet, divert, netgraph or other modules.
119 * The packet may be consumed.
122 ipfw_check_packet(void *arg, struct mbuf **m0, struct ifnet *ifp, int dir,
125 struct ip_fw_args args;
130 /* convert dir to IPFW values */
131 dir = (dir == PFIL_IN) ? DIR_IN : DIR_OUT;
132 bzero(&args, sizeof(args));
136 * extract and remove the tag if present. If we are left
137 * with onepass, optimize the outgoing path.
139 tag = m_tag_locate(*m0, MTAG_IPFW_RULE, 0, NULL);
141 args.rule = *((struct ipfw_rule_ref *)(tag+1));
142 m_tag_delete(*m0, tag);
143 if (args.rule.info & IPFW_ONEPASS)
148 args.oif = dir == DIR_OUT ? ifp : NULL;
151 ipfw = ipfw_chk(&args);
154 KASSERT(*m0 != NULL || ipfw == IP_FW_DENY, ("%s: m0 is NULL",
157 /* breaking out of the switch means drop */
158 ret = 0; /* default return value for pass */
161 /* next_hop may be set by ipfw_chk */
162 if (args.next_hop == NULL && args.next_hop6 == NULL)
164 #if (!defined(INET6) && !defined(INET))
168 struct m_tag *fwd_tag;
171 KASSERT(args.next_hop == NULL || args.next_hop6 == NULL,
172 ("%s: both next_hop=%p and next_hop6=%p not NULL", __func__,
173 args.next_hop, args.next_hop6));
175 if (args.next_hop6 != NULL)
176 len = sizeof(struct sockaddr_in6);
179 if (args.next_hop != NULL)
180 len = sizeof(struct sockaddr_in);
183 /* Incoming packets should not be tagged so we do not
184 * m_tag_find. Outgoing packets may be tagged, so we
185 * reuse the tag if present.
187 fwd_tag = (dir == DIR_IN) ? NULL :
188 m_tag_find(*m0, PACKET_TAG_IPFORWARD, NULL);
189 if (fwd_tag != NULL) {
190 m_tag_unlink(*m0, fwd_tag);
192 fwd_tag = m_tag_get(PACKET_TAG_IPFORWARD, len,
194 if (fwd_tag == NULL) {
196 break; /* i.e. drop */
200 if (args.next_hop6 != NULL) {
201 struct sockaddr_in6 *sa6;
203 sa6 = (struct sockaddr_in6 *)(fwd_tag + 1);
204 bcopy(args.next_hop6, sa6, len);
206 * If nh6 address is link-local we should convert
207 * it to kernel internal form before doing any
210 if (sa6_embedscope(sa6, V_ip6_use_defzone) != 0) {
214 if (in6_localip(&sa6->sin6_addr))
215 (*m0)->m_flags |= M_FASTFWD_OURS;
216 (*m0)->m_flags |= M_IP6_NEXTHOP;
220 if (args.next_hop != NULL) {
221 bcopy(args.next_hop, (fwd_tag+1), len);
222 if (in_localip(args.next_hop->sin_addr))
223 (*m0)->m_flags |= M_FASTFWD_OURS;
224 (*m0)->m_flags |= M_IP_NEXTHOP;
227 m_tag_prepend(*m0, fwd_tag);
229 #endif /* INET || INET6 */
234 break; /* i.e. drop */
238 if (ip_dn_io_ptr == NULL)
239 break; /* i.e. drop */
240 if (mtod(*m0, struct ip *)->ip_v == 4)
241 ret = ip_dn_io_ptr(m0, dir, &args);
242 else if (mtod(*m0, struct ip *)->ip_v == 6)
243 ret = ip_dn_io_ptr(m0, dir | PROTO_IPV6, &args);
247 * XXX should read the return value.
248 * dummynet normally eats the packet and sets *m0=NULL
249 * unless the packet can be sent immediately. In this
250 * case args is updated and we should re-run the
251 * check without clearing args.
259 if (ip_divert_ptr == NULL) {
261 break; /* i.e. drop */
263 ret = ipfw_divert(m0, dir, &args.rule,
264 (ipfw == IP_FW_TEE) ? 1 : 0);
265 /* continue processing for the original packet (tee). */
272 if (ng_ipfw_input_p == NULL) {
274 break; /* i.e. drop */
276 ret = ng_ipfw_input_p(m0, dir, &args,
277 (ipfw == IP_FW_NGTEE) ? 1 : 0);
278 if (ipfw == IP_FW_NGTEE) /* ignore errors for NGTEE */
279 goto again; /* continue with packet */
283 /* honor one-pass in case of successful nat */
285 break; /* ret is already 0 */
289 goto again; /* continue with packet */
292 KASSERT(0, ("%s: unknown retval", __func__));
305 * ipfw processing for ethernet packets (in and out).
306 * Inteface is NULL from ether_demux, and ifp from
307 * ether_output_frame.
310 ipfw_check_frame(void *arg, struct mbuf **m0, struct ifnet *dst, int dir,
313 struct ether_header *eh;
314 struct ether_header save_eh;
317 struct ip_fw_args args;
320 /* fetch start point from rule, if any */
321 mtag = m_tag_locate(*m0, MTAG_IPFW_RULE, 0, NULL);
325 /* dummynet packet, already partially processed */
326 struct ipfw_rule_ref *r;
328 /* XXX can we free it after use ? */
329 mtag->m_tag_id = PACKET_TAG_NONE;
330 r = (struct ipfw_rule_ref *)(mtag + 1);
331 if (r->info & IPFW_ONEPASS)
336 /* I need some amt of data to be contiguous */
338 i = min(m->m_pkthdr.len, max_protohdr);
346 eh = mtod(m, struct ether_header *);
347 save_eh = *eh; /* save copy for restore below */
348 m_adj(m, ETHER_HDR_LEN); /* strip ethernet header */
350 args.m = m; /* the packet we are looking at */
351 args.oif = dir == PFIL_OUT ? dst: NULL; /* destination, if any */
352 args.next_hop = NULL; /* we do not support forward yet */
353 args.next_hop6 = NULL; /* we do not support forward yet */
354 args.eh = &save_eh; /* MAC header for bridged/MAC packets */
355 args.inp = NULL; /* used by ipfw uid/gid/jail rules */
360 * Restore Ethernet header, as needed, in case the
361 * mbuf chain was replaced by ipfw.
363 M_PREPEND(m, ETHER_HDR_LEN, M_NOWAIT);
368 if (eh != mtod(m, struct ether_header *))
369 bcopy(&save_eh, mtod(m, struct ether_header *),
375 /* Check result of ipfw_chk() */
382 break; /* i.e. drop */
388 if (ip_dn_io_ptr == NULL)
389 break; /* i.e. drop */
392 dir = PROTO_LAYER2 | (dst ? DIR_OUT : DIR_IN);
393 ip_dn_io_ptr(&m, dir, &args);
397 KASSERT(0, ("%s: unknown retval", __func__));
409 /* do the divert, return 1 on error 0 on success */
411 ipfw_divert(struct mbuf **m0, int incoming, struct ipfw_rule_ref *rule,
415 * ipfw_chk() has already tagged the packet with the divert tag.
416 * If tee is set, copy packet and return original.
417 * If not tee, consume packet and send it to divert socket.
420 struct ip *ip = mtod(*m0, struct ip *);
423 /* Cloning needed for tee? */
425 clone = *m0; /* use the original mbuf */
428 clone = m_dup(*m0, M_NOWAIT);
429 /* If we cannot duplicate the mbuf, we sacrifice the divert
430 * chain and continue with the tee-ed packet.
437 * Divert listeners can normally handle non-fragmented packets,
438 * but we can only reass in the non-tee case.
439 * This means that listeners on a tee rule may get fragments,
440 * and have to live with that.
441 * Note that we now have the 'reass' ipfw option so if we care
442 * we can do it before a 'tee'.
444 if (!tee) switch (ip->ip_v) {
446 if (ntohs(ip->ip_off) & (IP_MF | IP_OFFMASK)) {
450 reass = ip_reass(clone); /* Reassemble packet. */
452 return 0; /* not an error */
453 /* if reass = NULL then it was consumed by ip_reass */
455 * IP header checksum fixup after reassembly and leave header
456 * in network byte order.
458 ip = mtod(reass, struct ip *);
459 hlen = ip->ip_hl << 2;
461 if (hlen == sizeof(struct ip))
462 ip->ip_sum = in_cksum_hdr(ip);
464 ip->ip_sum = in_cksum(reass, hlen);
469 case IPV6_VERSION >> 4:
471 struct ip6_hdr *const ip6 = mtod(clone, struct ip6_hdr *);
473 if (ip6->ip6_nxt == IPPROTO_FRAGMENT) {
476 off = sizeof(struct ip6_hdr);
477 nxt = frag6_input(&clone, &off, 0);
478 if (nxt == IPPROTO_DONE)
486 /* attach a tag to the packet with the reinject info */
487 tag = m_tag_alloc(MTAG_IPFW_RULE, 0,
488 sizeof(struct ipfw_rule_ref), M_NOWAIT);
493 *((struct ipfw_rule_ref *)(tag+1)) = *rule;
494 m_tag_prepend(clone, tag);
496 /* Do the dirty job... */
497 ip_divert_ptr(clone, incoming);
502 * attach or detach hooks for a given protocol family
505 ipfw_hook(int onoff, int pf)
507 struct pfil_head *pfh;
508 pfil_func_t hook_func;
510 pfh = pfil_head_get(PFIL_TYPE_AF, pf);
514 hook_func = (pf == AF_LINK) ? ipfw_check_frame : ipfw_check_packet;
516 (void) (onoff ? pfil_add_hook : pfil_remove_hook)
517 (hook_func, NULL, PFIL_IN | PFIL_OUT | PFIL_WAITOK, pfh);
523 ipfw_attach_hooks(int arg)
527 if (arg == 0) /* detach */
528 ipfw_hook(0, AF_INET);
529 else if (V_fw_enable && ipfw_hook(1, AF_INET) != 0) {
530 error = ENOENT; /* see ip_fw_pfil.c::ipfw_hook() */
531 printf("ipfw_hook() error\n");
534 if (arg == 0) /* detach */
535 ipfw_hook(0, AF_INET6);
536 else if (V_fw6_enable && ipfw_hook(1, AF_INET6) != 0) {
538 printf("ipfw6_hook() error\n");
541 if (arg == 0) /* detach */
542 ipfw_hook(0, AF_LINK);
543 else if (V_fwlink_enable && ipfw_hook(1, AF_LINK) != 0) {
545 printf("ipfw_link_hook() error\n");
551 ipfw_chg_hook(SYSCTL_HANDLER_ARGS)
557 if (arg1 == &V_fw_enable)
560 else if (arg1 == &V_fw6_enable)
563 else if (arg1 == &V_fwlink_enable)
568 newval = *(int *)arg1;
569 /* Handle sysctl change */
570 error = sysctl_handle_int(oidp, &newval, 0, req);
575 /* Formalize new value */
576 newval = (newval) ? 1 : 0;
578 if (*(int *)arg1 == newval)
581 error = ipfw_hook(newval, af);
584 *(int *)arg1 = newval;