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/route.h>
52 #include <net/ethernet.h>
56 #include <netinet/in.h>
57 #include <netinet/in_systm.h>
58 #include <netinet/ip.h>
59 #include <netinet/ip_var.h>
60 #include <netinet/ip_fw.h>
62 #include <netinet/ip6.h>
63 #include <netinet6/ip6_var.h>
64 #include <netinet6/scope6_var.h>
67 #include <netgraph/ng_ipfw.h>
69 #include <netpfil/ipfw/ip_fw_private.h>
71 #include <machine/in_cksum.h>
73 VNET_DEFINE_STATIC(int, fw_enable) = 1;
74 #define V_fw_enable VNET(fw_enable)
77 VNET_DEFINE_STATIC(int, fw6_enable) = 1;
78 #define V_fw6_enable VNET(fw6_enable)
81 VNET_DEFINE_STATIC(int, fwlink_enable) = 0;
82 #define V_fwlink_enable VNET(fwlink_enable)
84 int ipfw_chg_hook(SYSCTL_HANDLER_ARGS);
86 /* Forward declarations. */
87 static int ipfw_divert(struct mbuf **, int, struct ipfw_rule_ref *, int);
88 int ipfw_check_packet(void *, struct mbuf **, struct ifnet *, int,
90 int ipfw_check_frame(void *, struct mbuf **, struct ifnet *, int,
97 SYSCTL_DECL(_net_inet_ip_fw);
98 SYSCTL_PROC(_net_inet_ip_fw, OID_AUTO, enable,
99 CTLFLAG_VNET | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE3,
100 &VNET_NAME(fw_enable), 0, ipfw_chg_hook, "I", "Enable ipfw");
102 SYSCTL_DECL(_net_inet6_ip6_fw);
103 SYSCTL_PROC(_net_inet6_ip6_fw, OID_AUTO, enable,
104 CTLFLAG_VNET | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE3,
105 &VNET_NAME(fw6_enable), 0, ipfw_chg_hook, "I", "Enable ipfw+6");
108 SYSCTL_DECL(_net_link_ether);
109 SYSCTL_PROC(_net_link_ether, OID_AUTO, ipfw,
110 CTLFLAG_VNET | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE3,
111 &VNET_NAME(fwlink_enable), 0, ipfw_chg_hook, "I",
112 "Pass ether pkts through firewall");
116 #endif /* SYSCTL_NODE */
119 * The pfilter hook to pass packets to ipfw_chk and then to
120 * dummynet, divert, netgraph or other modules.
121 * The packet may be consumed.
124 ipfw_check_packet(void *arg, struct mbuf **m0, struct ifnet *ifp, int dir,
127 struct ip_fw_args args;
131 /* convert dir to IPFW values */
132 dir = (dir == PFIL_IN) ? DIR_IN : DIR_OUT;
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)
145 args.flags |= IPFW_ARGS_REF;
149 args.oif = dir == DIR_OUT ? ifp : NULL;
152 ipfw = ipfw_chk(&args);
155 KASSERT(*m0 != NULL || ipfw == IP_FW_DENY, ("%s: m0 is NULL",
158 /* breaking out of the switch means drop */
161 /* next_hop may be set by ipfw_chk */
162 if ((args.flags & (IPFW_ARGS_NH4 | IPFW_ARGS_NH4PTR |
163 IPFW_ARGS_NH6 | IPFW_ARGS_NH6PTR)) == 0) {
167 #if (!defined(INET6) && !defined(INET))
174 if (args.flags & (IPFW_ARGS_NH4 | IPFW_ARGS_NH4PTR)) {
175 MPASS((args.flags & (IPFW_ARGS_NH4 |
176 IPFW_ARGS_NH4PTR)) != (IPFW_ARGS_NH4 |
178 MPASS((args.flags & (IPFW_ARGS_NH6 |
179 IPFW_ARGS_NH6PTR)) == 0);
180 len = sizeof(struct sockaddr_in);
181 psa = (args.flags & IPFW_ARGS_NH4) ?
182 &args.hopstore : args.next_hop;
183 if (in_localip(satosin(psa)->sin_addr))
184 (*m0)->m_flags |= M_FASTFWD_OURS;
185 (*m0)->m_flags |= M_IP_NEXTHOP;
189 if (args.flags & (IPFW_ARGS_NH6 | IPFW_ARGS_NH6PTR)) {
190 MPASS((args.flags & (IPFW_ARGS_NH6 |
191 IPFW_ARGS_NH6PTR)) != (IPFW_ARGS_NH6 |
193 MPASS((args.flags & (IPFW_ARGS_NH4 |
194 IPFW_ARGS_NH4PTR)) == 0);
195 len = sizeof(struct sockaddr_in6);
196 psa = args.next_hop6;
197 (*m0)->m_flags |= M_IP6_NEXTHOP;
201 * Incoming packets should not be tagged so we do not
202 * m_tag_find. Outgoing packets may be tagged, so we
203 * reuse the tag if present.
205 tag = (dir == DIR_IN) ? NULL :
206 m_tag_find(*m0, PACKET_TAG_IPFORWARD, NULL);
208 m_tag_unlink(*m0, tag);
210 tag = m_tag_get(PACKET_TAG_IPFORWARD, len,
214 break; /* i.e. drop */
217 if ((args.flags & IPFW_ARGS_NH6) == 0)
218 bcopy(psa, tag + 1, len);
219 m_tag_prepend(*m0, tag);
222 /* IPv6 next hop needs additional handling */
223 if (args.flags & (IPFW_ARGS_NH6 | IPFW_ARGS_NH6PTR)) {
224 struct sockaddr_in6 *sa6;
226 sa6 = satosin6(tag + 1);
227 if (args.flags & IPFW_ARGS_NH6) {
228 sa6->sin6_family = AF_INET6;
229 sa6->sin6_len = sizeof(*sa6);
230 sa6->sin6_addr = args.hopstore6.sin6_addr;
231 sa6->sin6_port = args.hopstore6.sin6_port;
233 args.hopstore6.sin6_scope_id;
236 * If nh6 address is link-local we should convert
237 * it to kernel internal form before doing any
240 if (sa6_embedscope(sa6, V_ip6_use_defzone) != 0) {
244 if (in6_localip(&sa6->sin6_addr))
245 (*m0)->m_flags |= M_FASTFWD_OURS;
249 #endif /* INET || INET6 */
254 break; /* i.e. drop */
258 if (ip_dn_io_ptr == NULL)
259 break; /* i.e. drop */
260 MPASS(args.flags & IPFW_ARGS_REF);
261 if (mtod(*m0, struct ip *)->ip_v == 4)
262 ret = ip_dn_io_ptr(m0, dir, &args);
263 else if (mtod(*m0, struct ip *)->ip_v == 6)
264 ret = ip_dn_io_ptr(m0, dir | PROTO_IPV6, &args);
268 * XXX should read the return value.
269 * dummynet normally eats the packet and sets *m0=NULL
270 * unless the packet can be sent immediately. In this
271 * case args is updated and we should re-run the
272 * check without clearing args.
280 if (ip_divert_ptr == NULL) {
282 break; /* i.e. drop */
284 MPASS(args.flags & IPFW_ARGS_REF);
285 ret = ipfw_divert(m0, dir, &args.rule,
286 (ipfw == IP_FW_TEE) ? 1 : 0);
287 /* continue processing for the original packet (tee). */
294 if (ng_ipfw_input_p == NULL) {
296 break; /* i.e. drop */
298 MPASS(args.flags & IPFW_ARGS_REF);
299 ret = ng_ipfw_input_p(m0, dir, &args,
300 (ipfw == IP_FW_NGTEE) ? 1 : 0);
301 if (ipfw == IP_FW_NGTEE) /* ignore errors for NGTEE */
302 goto again; /* continue with packet */
306 /* honor one-pass in case of successful nat */
314 goto again; /* continue with packet */
317 KASSERT(0, ("%s: unknown retval", __func__));
330 * ipfw processing for ethernet packets (in and out).
333 ipfw_check_frame(void *arg, struct mbuf **m0, struct ifnet *ifp, int dir,
336 struct ip_fw_args args;
337 struct ether_header save_eh;
338 struct ether_header *eh;
343 args.flags = IPFW_ARGS_ETHER;
345 /* fetch start point from rule, if any. remove the tag if present. */
346 mtag = m_tag_locate(*m0, MTAG_IPFW_RULE, 0, NULL);
348 args.rule = *((struct ipfw_rule_ref *)(mtag+1));
349 m_tag_delete(*m0, mtag);
350 if (args.rule.info & IPFW_ONEPASS)
352 args.flags |= IPFW_ARGS_REF;
355 /* I need some amt of data to be contiguous */
357 i = min(m->m_pkthdr.len, max_protohdr);
365 eh = mtod(m, struct ether_header *);
366 save_eh = *eh; /* save copy for restore below */
367 m_adj(m, ETHER_HDR_LEN); /* strip ethernet header */
369 args.m = m; /* the packet we are looking at */
370 args.oif = dir == PFIL_OUT ? ifp: NULL; /* destination, if any */
371 args.eh = &save_eh; /* MAC header for bridged/MAC packets */
372 args.inp = inp; /* used by ipfw uid/gid/jail rules */
377 * Restore Ethernet header, as needed, in case the
378 * mbuf chain was replaced by ipfw.
380 M_PREPEND(m, ETHER_HDR_LEN, M_NOWAIT);
385 if (eh != mtod(m, struct ether_header *))
386 bcopy(&save_eh, mtod(m, struct ether_header *),
392 /* Check result of ipfw_chk() */
399 break; /* i.e. drop */
404 if (ip_dn_io_ptr == NULL)
405 break; /* i.e. drop */
408 dir = (dir == PFIL_IN) ? DIR_IN : DIR_OUT;
409 MPASS(args.flags & IPFW_ARGS_REF);
410 ip_dn_io_ptr(&m, dir | PROTO_LAYER2, &args);
415 if (ng_ipfw_input_p == NULL) {
417 break; /* i.e. drop */
419 MPASS(args.flags & IPFW_ARGS_REF);
420 ret = ng_ipfw_input_p(m0, (dir == PFIL_IN) ? DIR_IN : DIR_OUT,
421 &args, (i == IP_FW_NGTEE) ? 1 : 0);
422 if (i == IP_FW_NGTEE) /* ignore errors for NGTEE */
423 goto again; /* continue with packet */
427 KASSERT(0, ("%s: unknown retval", __func__));
439 /* do the divert, return 1 on error 0 on success */
441 ipfw_divert(struct mbuf **m0, int incoming, struct ipfw_rule_ref *rule,
445 * ipfw_chk() has already tagged the packet with the divert tag.
446 * If tee is set, copy packet and return original.
447 * If not tee, consume packet and send it to divert socket.
450 struct ip *ip = mtod(*m0, struct ip *);
453 /* Cloning needed for tee? */
455 clone = *m0; /* use the original mbuf */
458 clone = m_dup(*m0, M_NOWAIT);
459 /* If we cannot duplicate the mbuf, we sacrifice the divert
460 * chain and continue with the tee-ed packet.
467 * Divert listeners can normally handle non-fragmented packets,
468 * but we can only reass in the non-tee case.
469 * This means that listeners on a tee rule may get fragments,
470 * and have to live with that.
471 * Note that we now have the 'reass' ipfw option so if we care
472 * we can do it before a 'tee'.
474 if (!tee) switch (ip->ip_v) {
476 if (ntohs(ip->ip_off) & (IP_MF | IP_OFFMASK)) {
480 reass = ip_reass(clone); /* Reassemble packet. */
482 return 0; /* not an error */
483 /* if reass = NULL then it was consumed by ip_reass */
485 * IP header checksum fixup after reassembly and leave header
486 * in network byte order.
488 ip = mtod(reass, struct ip *);
489 hlen = ip->ip_hl << 2;
491 if (hlen == sizeof(struct ip))
492 ip->ip_sum = in_cksum_hdr(ip);
494 ip->ip_sum = in_cksum(reass, hlen);
499 case IPV6_VERSION >> 4:
501 struct ip6_hdr *const ip6 = mtod(clone, struct ip6_hdr *);
503 if (ip6->ip6_nxt == IPPROTO_FRAGMENT) {
506 off = sizeof(struct ip6_hdr);
507 nxt = frag6_input(&clone, &off, 0);
508 if (nxt == IPPROTO_DONE)
516 /* attach a tag to the packet with the reinject info */
517 tag = m_tag_alloc(MTAG_IPFW_RULE, 0,
518 sizeof(struct ipfw_rule_ref), M_NOWAIT);
523 *((struct ipfw_rule_ref *)(tag+1)) = *rule;
524 m_tag_prepend(clone, tag);
526 /* Do the dirty job... */
527 ip_divert_ptr(clone, incoming);
532 * attach or detach hooks for a given protocol family
535 ipfw_hook(int onoff, int pf)
537 struct pfil_head *pfh;
538 pfil_func_t hook_func;
540 pfh = pfil_head_get(PFIL_TYPE_AF, pf);
544 hook_func = (pf == AF_LINK) ? ipfw_check_frame : ipfw_check_packet;
546 (void) (onoff ? pfil_add_hook : pfil_remove_hook)
547 (hook_func, NULL, PFIL_IN | PFIL_OUT | PFIL_WAITOK, pfh);
553 ipfw_attach_hooks(int arg)
557 if (arg == 0) /* detach */
558 ipfw_hook(0, AF_INET);
559 else if (V_fw_enable && ipfw_hook(1, AF_INET) != 0) {
560 error = ENOENT; /* see ip_fw_pfil.c::ipfw_hook() */
561 printf("ipfw_hook() error\n");
564 if (arg == 0) /* detach */
565 ipfw_hook(0, AF_INET6);
566 else if (V_fw6_enable && ipfw_hook(1, AF_INET6) != 0) {
568 printf("ipfw6_hook() error\n");
571 if (arg == 0) /* detach */
572 ipfw_hook(0, AF_LINK);
573 else if (V_fwlink_enable && ipfw_hook(1, AF_LINK) != 0) {
575 printf("ipfw_link_hook() error\n");
581 ipfw_chg_hook(SYSCTL_HANDLER_ARGS)
587 if (arg1 == &V_fw_enable)
590 else if (arg1 == &V_fw6_enable)
593 else if (arg1 == &V_fwlink_enable)
598 newval = *(int *)arg1;
599 /* Handle sysctl change */
600 error = sysctl_handle_int(oidp, &newval, 0, req);
605 /* Formalize new value */
606 newval = (newval) ? 1 : 0;
608 if (*(int *)arg1 == newval)
611 error = ipfw_hook(newval, af);
614 *(int *)arg1 = newval;