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>
64 #include <netgraph/ng_ipfw.h>
66 #include <netpfil/ipfw/ip_fw_private.h>
68 #include <machine/in_cksum.h>
70 static VNET_DEFINE(int, fw_enable) = 1;
71 #define V_fw_enable VNET(fw_enable)
74 static VNET_DEFINE(int, fw6_enable) = 1;
75 #define V_fw6_enable VNET(fw6_enable)
78 static VNET_DEFINE(int, fwlink_enable) = 0;
79 #define V_fwlink_enable VNET(fwlink_enable)
81 int ipfw_chg_hook(SYSCTL_HANDLER_ARGS);
83 /* Forward declarations. */
84 static int ipfw_divert(struct mbuf **, int, struct ipfw_rule_ref *, int);
85 static int ipfw_check_packet(void *, struct mbuf **, struct ifnet *, int,
87 static int ipfw_check_frame(void *, struct mbuf **, struct ifnet *, int,
94 SYSCTL_DECL(_net_inet_ip_fw);
95 SYSCTL_VNET_PROC(_net_inet_ip_fw, OID_AUTO, enable,
96 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE3, &VNET_NAME(fw_enable), 0,
97 ipfw_chg_hook, "I", "Enable ipfw");
99 SYSCTL_DECL(_net_inet6_ip6_fw);
100 SYSCTL_VNET_PROC(_net_inet6_ip6_fw, OID_AUTO, enable,
101 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE3, &VNET_NAME(fw6_enable), 0,
102 ipfw_chg_hook, "I", "Enable ipfw+6");
105 SYSCTL_DECL(_net_link_ether);
106 SYSCTL_VNET_PROC(_net_link_ether, OID_AUTO, ipfw,
107 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE3, &VNET_NAME(fwlink_enable), 0,
108 ipfw_chg_hook, "I", "Pass ether pkts through firewall");
112 #endif /* SYSCTL_NODE */
115 * The pfilter hook to pass packets to ipfw_chk and then to
116 * dummynet, divert, netgraph or other modules.
117 * The packet may be consumed.
120 ipfw_check_packet(void *arg, struct mbuf **m0, struct ifnet *ifp, int dir,
123 struct ip_fw_args args;
128 /* convert dir to IPFW values */
129 dir = (dir == PFIL_IN) ? DIR_IN : DIR_OUT;
130 bzero(&args, sizeof(args));
134 * extract and remove the tag if present. If we are left
135 * with onepass, optimize the outgoing path.
137 tag = m_tag_locate(*m0, MTAG_IPFW_RULE, 0, NULL);
139 args.rule = *((struct ipfw_rule_ref *)(tag+1));
140 m_tag_delete(*m0, tag);
141 if (args.rule.info & IPFW_ONEPASS)
146 args.oif = dir == DIR_OUT ? ifp : NULL;
149 ipfw = ipfw_chk(&args);
152 KASSERT(*m0 != NULL || ipfw == IP_FW_DENY, ("%s: m0 is NULL",
155 /* breaking out of the switch means drop */
156 ret = 0; /* default return value for pass */
159 /* next_hop may be set by ipfw_chk */
160 if (args.next_hop == NULL && args.next_hop6 == NULL)
162 #if (!defined(INET6) && !defined(INET))
166 struct m_tag *fwd_tag;
169 KASSERT(args.next_hop == NULL || args.next_hop6 == NULL,
170 ("%s: both next_hop=%p and next_hop6=%p not NULL", __func__,
171 args.next_hop, args.next_hop6));
173 if (args.next_hop6 != NULL)
174 len = sizeof(struct sockaddr_in6);
177 if (args.next_hop != NULL)
178 len = sizeof(struct sockaddr_in);
181 /* Incoming packets should not be tagged so we do not
182 * m_tag_find. Outgoing packets may be tagged, so we
183 * reuse the tag if present.
185 fwd_tag = (dir == DIR_IN) ? NULL :
186 m_tag_find(*m0, PACKET_TAG_IPFORWARD, NULL);
187 if (fwd_tag != NULL) {
188 m_tag_unlink(*m0, fwd_tag);
190 fwd_tag = m_tag_get(PACKET_TAG_IPFORWARD, len,
192 if (fwd_tag == NULL) {
194 break; /* i.e. drop */
198 if (args.next_hop6 != NULL) {
199 bcopy(args.next_hop6, (fwd_tag+1), len);
200 if (in6_localip(&args.next_hop6->sin6_addr))
201 (*m0)->m_flags |= M_FASTFWD_OURS;
202 (*m0)->m_flags |= M_IP6_NEXTHOP;
206 if (args.next_hop != NULL) {
207 bcopy(args.next_hop, (fwd_tag+1), len);
208 if (in_localip(args.next_hop->sin_addr))
209 (*m0)->m_flags |= M_FASTFWD_OURS;
210 (*m0)->m_flags |= M_IP_NEXTHOP;
213 m_tag_prepend(*m0, fwd_tag);
215 #endif /* INET || INET6 */
220 break; /* i.e. drop */
224 if (ip_dn_io_ptr == NULL)
225 break; /* i.e. drop */
226 if (mtod(*m0, struct ip *)->ip_v == 4)
227 ret = ip_dn_io_ptr(m0, dir, &args);
228 else if (mtod(*m0, struct ip *)->ip_v == 6)
229 ret = ip_dn_io_ptr(m0, dir | PROTO_IPV6, &args);
233 * XXX should read the return value.
234 * dummynet normally eats the packet and sets *m0=NULL
235 * unless the packet can be sent immediately. In this
236 * case args is updated and we should re-run the
237 * check without clearing args.
245 if (ip_divert_ptr == NULL) {
247 break; /* i.e. drop */
249 ret = ipfw_divert(m0, dir, &args.rule,
250 (ipfw == IP_FW_TEE) ? 1 : 0);
251 /* continue processing for the original packet (tee). */
258 if (ng_ipfw_input_p == NULL) {
260 break; /* i.e. drop */
262 ret = ng_ipfw_input_p(m0, dir, &args,
263 (ipfw == IP_FW_NGTEE) ? 1 : 0);
264 if (ipfw == IP_FW_NGTEE) /* ignore errors for NGTEE */
265 goto again; /* continue with packet */
269 /* honor one-pass in case of successful nat */
271 break; /* ret is already 0 */
275 goto again; /* continue with packet */
278 KASSERT(0, ("%s: unknown retval", __func__));
291 * ipfw processing for ethernet packets (in and out).
292 * Inteface is NULL from ether_demux, and ifp from
293 * ether_output_frame.
296 ipfw_check_frame(void *arg, struct mbuf **m0, struct ifnet *dst, int dir,
299 struct ether_header *eh;
300 struct ether_header save_eh;
303 struct ip_fw_args args;
306 /* fetch start point from rule, if any */
307 mtag = m_tag_locate(*m0, MTAG_IPFW_RULE, 0, NULL);
311 /* dummynet packet, already partially processed */
312 struct ipfw_rule_ref *r;
314 /* XXX can we free it after use ? */
315 mtag->m_tag_id = PACKET_TAG_NONE;
316 r = (struct ipfw_rule_ref *)(mtag + 1);
317 if (r->info & IPFW_ONEPASS)
322 /* I need some amt of data to be contiguous */
324 i = min(m->m_pkthdr.len, max_protohdr);
332 eh = mtod(m, struct ether_header *);
333 save_eh = *eh; /* save copy for restore below */
334 m_adj(m, ETHER_HDR_LEN); /* strip ethernet header */
336 args.m = m; /* the packet we are looking at */
337 args.oif = dir == PFIL_OUT ? dst: NULL; /* destination, if any */
338 args.next_hop = NULL; /* we do not support forward yet */
339 args.next_hop6 = NULL; /* we do not support forward yet */
340 args.eh = &save_eh; /* MAC header for bridged/MAC packets */
341 args.inp = NULL; /* used by ipfw uid/gid/jail rules */
346 * Restore Ethernet header, as needed, in case the
347 * mbuf chain was replaced by ipfw.
349 M_PREPEND(m, ETHER_HDR_LEN, M_NOWAIT);
354 if (eh != mtod(m, struct ether_header *))
355 bcopy(&save_eh, mtod(m, struct ether_header *),
361 /* Check result of ipfw_chk() */
368 break; /* i.e. drop */
374 if (ip_dn_io_ptr == NULL)
375 break; /* i.e. drop */
378 dir = PROTO_LAYER2 | (dst ? DIR_OUT : DIR_IN);
379 ip_dn_io_ptr(&m, dir, &args);
383 KASSERT(0, ("%s: unknown retval", __func__));
395 /* do the divert, return 1 on error 0 on success */
397 ipfw_divert(struct mbuf **m0, int incoming, struct ipfw_rule_ref *rule,
401 * ipfw_chk() has already tagged the packet with the divert tag.
402 * If tee is set, copy packet and return original.
403 * If not tee, consume packet and send it to divert socket.
406 struct ip *ip = mtod(*m0, struct ip *);
409 /* Cloning needed for tee? */
411 clone = *m0; /* use the original mbuf */
414 clone = m_dup(*m0, M_NOWAIT);
415 /* If we cannot duplicate the mbuf, we sacrifice the divert
416 * chain and continue with the tee-ed packet.
423 * Divert listeners can normally handle non-fragmented packets,
424 * but we can only reass in the non-tee case.
425 * This means that listeners on a tee rule may get fragments,
426 * and have to live with that.
427 * Note that we now have the 'reass' ipfw option so if we care
428 * we can do it before a 'tee'.
430 if (!tee) switch (ip->ip_v) {
432 if (ntohs(ip->ip_off) & (IP_MF | IP_OFFMASK)) {
436 reass = ip_reass(clone); /* Reassemble packet. */
438 return 0; /* not an error */
439 /* if reass = NULL then it was consumed by ip_reass */
441 * IP header checksum fixup after reassembly and leave header
442 * in network byte order.
444 ip = mtod(reass, struct ip *);
445 hlen = ip->ip_hl << 2;
447 if (hlen == sizeof(struct ip))
448 ip->ip_sum = in_cksum_hdr(ip);
450 ip->ip_sum = in_cksum(reass, hlen);
455 case IPV6_VERSION >> 4:
457 struct ip6_hdr *const ip6 = mtod(clone, struct ip6_hdr *);
459 if (ip6->ip6_nxt == IPPROTO_FRAGMENT) {
462 off = sizeof(struct ip6_hdr);
463 nxt = frag6_input(&clone, &off, 0);
464 if (nxt == IPPROTO_DONE)
472 /* attach a tag to the packet with the reinject info */
473 tag = m_tag_alloc(MTAG_IPFW_RULE, 0,
474 sizeof(struct ipfw_rule_ref), M_NOWAIT);
479 *((struct ipfw_rule_ref *)(tag+1)) = *rule;
480 m_tag_prepend(clone, tag);
482 /* Do the dirty job... */
483 ip_divert_ptr(clone, incoming);
488 * attach or detach hooks for a given protocol family
491 ipfw_hook(int onoff, int pf)
493 struct pfil_head *pfh;
496 pfh = pfil_head_get(PFIL_TYPE_AF, pf);
500 hook_func = (pf == AF_LINK) ? ipfw_check_frame : ipfw_check_packet;
502 (void) (onoff ? pfil_add_hook : pfil_remove_hook)
503 (hook_func, NULL, PFIL_IN | PFIL_OUT | PFIL_WAITOK, pfh);
509 ipfw_attach_hooks(int arg)
513 if (arg == 0) /* detach */
514 ipfw_hook(0, AF_INET);
515 else if (V_fw_enable && ipfw_hook(1, AF_INET) != 0) {
516 error = ENOENT; /* see ip_fw_pfil.c::ipfw_hook() */
517 printf("ipfw_hook() error\n");
520 if (arg == 0) /* detach */
521 ipfw_hook(0, AF_INET6);
522 else if (V_fw6_enable && ipfw_hook(1, AF_INET6) != 0) {
524 printf("ipfw6_hook() error\n");
527 if (arg == 0) /* detach */
528 ipfw_hook(0, AF_LINK);
529 else if (V_fwlink_enable && ipfw_hook(1, AF_LINK) != 0) {
531 printf("ipfw_link_hook() error\n");
537 ipfw_chg_hook(SYSCTL_HANDLER_ARGS)
543 if (arg1 == &V_fw_enable)
546 else if (arg1 == &V_fw6_enable)
549 else if (arg1 == &V_fwlink_enable)
554 newval = *(int *)arg1;
555 /* Handle sysctl change */
556 error = sysctl_handle_int(oidp, &newval, 0, req);
561 /* Formalize new value */
562 newval = (newval) ? 1 : 0;
564 if (*(int *)arg1 == newval)
567 error = ipfw_hook(newval, af);
570 *(int *)arg1 = newval;