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;
132 * extract and remove the tag if present. If we are left
133 * with onepass, optimize the outgoing path.
135 tag = m_tag_locate(*m0, MTAG_IPFW_RULE, 0, NULL);
137 args.rule = *((struct ipfw_rule_ref *)(tag+1));
138 m_tag_delete(*m0, tag);
139 if (args.rule.info & IPFW_ONEPASS)
141 args.flags |= IPFW_ARGS_REF;
148 ipfw = ipfw_chk(&args);
151 KASSERT(*m0 != NULL || ipfw == IP_FW_DENY ||
152 ipfw == IP_FW_NAT64, ("%s: m0 is NULL", __func__));
157 /* next_hop may be set by ipfw_chk */
158 if ((args.flags & (IPFW_ARGS_NH4 | IPFW_ARGS_NH4PTR |
159 IPFW_ARGS_NH6 | IPFW_ARGS_NH6PTR)) == 0)
161 #if (!defined(INET6) && !defined(INET))
168 if (args.flags & (IPFW_ARGS_NH4 | IPFW_ARGS_NH4PTR)) {
169 MPASS((args.flags & (IPFW_ARGS_NH4 |
170 IPFW_ARGS_NH4PTR)) != (IPFW_ARGS_NH4 |
172 MPASS((args.flags & (IPFW_ARGS_NH6 |
173 IPFW_ARGS_NH6PTR)) == 0);
174 len = sizeof(struct sockaddr_in);
175 psa = (args.flags & IPFW_ARGS_NH4) ?
176 &args.hopstore : args.next_hop;
177 if (in_localip(satosin(psa)->sin_addr))
178 (*m0)->m_flags |= M_FASTFWD_OURS;
179 (*m0)->m_flags |= M_IP_NEXTHOP;
183 if (args.flags & (IPFW_ARGS_NH6 | IPFW_ARGS_NH6PTR)) {
184 MPASS((args.flags & (IPFW_ARGS_NH6 |
185 IPFW_ARGS_NH6PTR)) != (IPFW_ARGS_NH6 |
187 MPASS((args.flags & (IPFW_ARGS_NH4 |
188 IPFW_ARGS_NH4PTR)) == 0);
189 len = sizeof(struct sockaddr_in6);
190 psa = args.next_hop6;
191 (*m0)->m_flags |= M_IP6_NEXTHOP;
195 * Incoming packets should not be tagged so we do not
196 * m_tag_find. Outgoing packets may be tagged, so we
197 * reuse the tag if present.
199 tag = (flags & PFIL_IN) ? NULL :
200 m_tag_find(*m0, PACKET_TAG_IPFORWARD, NULL);
202 m_tag_unlink(*m0, tag);
204 tag = m_tag_get(PACKET_TAG_IPFORWARD, len,
211 if ((args.flags & IPFW_ARGS_NH6) == 0)
212 bcopy(psa, tag + 1, len);
213 m_tag_prepend(*m0, tag);
216 /* IPv6 next hop needs additional handling */
217 if (args.flags & (IPFW_ARGS_NH6 | IPFW_ARGS_NH6PTR)) {
218 struct sockaddr_in6 *sa6;
220 sa6 = satosin6(tag + 1);
221 if (args.flags & IPFW_ARGS_NH6) {
222 sa6->sin6_family = AF_INET6;
223 sa6->sin6_len = sizeof(*sa6);
224 sa6->sin6_addr = args.hopstore6.sin6_addr;
225 sa6->sin6_port = args.hopstore6.sin6_port;
227 args.hopstore6.sin6_scope_id;
230 * If nh6 address is link-local we should convert
231 * it to kernel internal form before doing any
234 if (sa6_embedscope(sa6, V_ip6_use_defzone) != 0) {
238 if (in6_localip(&sa6->sin6_addr))
239 (*m0)->m_flags |= M_FASTFWD_OURS;
243 #endif /* INET || INET6 */
251 if (ip_dn_io_ptr == NULL) {
255 MPASS(args.flags & IPFW_ARGS_REF);
256 if (args.flags & (IPFW_ARGS_IP4 | IPFW_ARGS_IP6))
257 (void )ip_dn_io_ptr(m0, &args);
263 * XXX should read the return value.
264 * dummynet normally eats the packet and sets *m0=NULL
265 * unless the packet can be sent immediately. In this
266 * case args is updated and we should re-run the
267 * check without clearing args.
276 if (ip_divert_ptr == NULL) {
280 MPASS(args.flags & IPFW_ARGS_REF);
281 (void )ipfw_divert(m0, &args, ipfw == IP_FW_TEE);
282 /* continue processing for the original packet (tee). */
290 if (ng_ipfw_input_p == NULL) {
294 MPASS(args.flags & IPFW_ARGS_REF);
295 (void )ng_ipfw_input_p(m0, &args, ipfw == IP_FW_NGTEE);
296 if (ipfw == IP_FW_NGTEE) /* ignore errors for NGTEE */
297 goto again; /* continue with packet */
302 /* honor one-pass in case of successful nat */
308 goto again; /* continue with packet */
315 KASSERT(0, ("%s: unknown retval", __func__));
318 if (ret != PFIL_PASS) {
328 * ipfw processing for ethernet packets (in and out).
331 ipfw_check_frame(pfil_packet_t p, struct ifnet *ifp, int flags,
332 void *ruleset __unused, struct inpcb *inp)
334 struct ip_fw_args args;
339 if (flags & PFIL_MEMPTR) {
342 args.flags = PFIL_LENGTH(flags) | IPFW_ARGS_ETHER;
345 mem = realloc = false;
346 args.flags = IPFW_ARGS_ETHER;
348 args.flags |= (flags & PFIL_IN) ? IPFW_ARGS_IN : IPFW_ARGS_OUT;
355 * Fetch start point from rule, if any.
356 * Remove the tag if present.
360 mtag = m_tag_locate(*p.m, MTAG_IPFW_RULE, 0, NULL);
362 args.rule = *((struct ipfw_rule_ref *)(mtag+1));
363 m_tag_delete(*p.m, mtag);
364 if (args.rule.info & IPFW_ONEPASS)
366 args.flags |= IPFW_ARGS_REF;
371 ipfw = ipfw_chk(&args);
383 if (ip_dn_io_ptr == NULL) {
388 if (pfil_realloc(&p, flags, ifp) != 0) {
395 MPASS(args.flags & IPFW_ARGS_REF);
396 ip_dn_io_ptr(p.m, &args);
397 return (PFIL_CONSUMED);
401 if (ng_ipfw_input_p == NULL) {
406 if (pfil_realloc(&p, flags, ifp) != 0) {
413 MPASS(args.flags & IPFW_ARGS_REF);
414 (void )ng_ipfw_input_p(p.m, &args, ipfw == IP_FW_NGTEE);
415 if (ipfw == IP_FW_NGTEE) /* ignore errors for NGTEE */
416 goto again; /* continue with packet */
421 KASSERT(0, ("%s: unknown retval", __func__));
424 if (!mem && ret != PFIL_PASS) {
430 if (realloc && ret == PFIL_PASS)
431 ret = PFIL_REALLOCED;
436 /* do the divert, return 1 on error 0 on success */
438 ipfw_divert(struct mbuf **m0, struct ip_fw_args *args, bool tee)
441 * ipfw_chk() has already tagged the packet with the divert tag.
442 * If tee is set, copy packet and return original.
443 * If not tee, consume packet and send it to divert socket.
446 struct ip *ip = mtod(*m0, struct ip *);
449 /* Cloning needed for tee? */
451 clone = *m0; /* use the original mbuf */
454 clone = m_dup(*m0, M_NOWAIT);
455 /* If we cannot duplicate the mbuf, we sacrifice the divert
456 * chain and continue with the tee-ed packet.
463 * Divert listeners can normally handle non-fragmented packets,
464 * but we can only reass in the non-tee case.
465 * This means that listeners on a tee rule may get fragments,
466 * and have to live with that.
467 * Note that we now have the 'reass' ipfw option so if we care
468 * we can do it before a 'tee'.
470 if (tee == false) switch (ip->ip_v) {
472 if (ntohs(ip->ip_off) & (IP_MF | IP_OFFMASK)) {
476 reass = ip_reass(clone); /* Reassemble packet. */
478 return 0; /* not an error */
479 /* if reass = NULL then it was consumed by ip_reass */
481 * IP header checksum fixup after reassembly and leave header
482 * in network byte order.
484 ip = mtod(reass, struct ip *);
485 hlen = ip->ip_hl << 2;
487 if (hlen == sizeof(struct ip))
488 ip->ip_sum = in_cksum_hdr(ip);
490 ip->ip_sum = in_cksum(reass, hlen);
495 case IPV6_VERSION >> 4:
497 struct ip6_hdr *const ip6 = mtod(clone, struct ip6_hdr *);
499 if (ip6->ip6_nxt == IPPROTO_FRAGMENT) {
502 off = sizeof(struct ip6_hdr);
503 nxt = frag6_input(&clone, &off, 0);
504 if (nxt == IPPROTO_DONE)
512 /* attach a tag to the packet with the reinject info */
513 tag = m_tag_alloc(MTAG_IPFW_RULE, 0,
514 sizeof(struct ipfw_rule_ref), M_NOWAIT);
519 *((struct ipfw_rule_ref *)(tag+1)) = args->rule;
520 m_tag_prepend(clone, tag);
522 /* Do the dirty job... */
523 ip_divert_ptr(clone, args->flags & IPFW_ARGS_IN);
528 * attach or detach hooks for a given protocol family
530 VNET_DEFINE_STATIC(pfil_hook_t, ipfw_inet_hook);
531 #define V_ipfw_inet_hook VNET(ipfw_inet_hook)
533 VNET_DEFINE_STATIC(pfil_hook_t, ipfw_inet6_hook);
534 #define V_ipfw_inet6_hook VNET(ipfw_inet6_hook)
536 VNET_DEFINE_STATIC(pfil_hook_t, ipfw_link_hook);
537 #define V_ipfw_link_hook VNET(ipfw_link_hook)
542 struct pfil_hook_args pha;
545 pha.pa_version = PFIL_VERSION;
546 pha.pa_flags = PFIL_IN | PFIL_OUT;
547 pha.pa_modname = "ipfw";
548 pha.pa_ruleset = NULL;
552 pha.pa_func = ipfw_check_packet;
553 pha.pa_type = PFIL_TYPE_IP4;
554 pha.pa_rulname = "default";
555 h = &V_ipfw_inet_hook;
559 pha.pa_func = ipfw_check_packet;
560 pha.pa_type = PFIL_TYPE_IP6;
561 pha.pa_rulname = "default6";
562 h = &V_ipfw_inet6_hook;
566 pha.pa_func = ipfw_check_frame;
567 pha.pa_type = PFIL_TYPE_ETHERNET;
568 pha.pa_rulname = "default-link";
569 pha.pa_flags |= PFIL_MEMPTR;
570 h = &V_ipfw_link_hook;
574 *h = pfil_add_hook(&pha);
583 pfil_remove_hook(V_ipfw_inet_hook);
587 pfil_remove_hook(V_ipfw_inet6_hook);
591 pfil_remove_hook(V_ipfw_link_hook);
597 ipfw_link(int pf, bool unlink)
599 struct pfil_link_args pla;
601 pla.pa_version = PFIL_VERSION;
602 pla.pa_flags = PFIL_IN | PFIL_OUT | PFIL_HEADPTR | PFIL_HOOKPTR;
604 pla.pa_flags |= PFIL_UNLINK;
608 pla.pa_head = V_inet_pfil_head;
609 pla.pa_hook = V_ipfw_inet_hook;
613 pla.pa_head = V_inet6_pfil_head;
614 pla.pa_hook = V_ipfw_inet6_hook;
618 pla.pa_head = V_link_pfil_head;
619 pla.pa_hook = V_ipfw_link_hook;
623 return (pfil_link(&pla));
627 ipfw_attach_hooks(void)
632 TUNABLE_INT_FETCH("net.inet.ip.fw.enable", &V_fw_enable);
633 if (V_fw_enable && (error = ipfw_link(AF_INET, false)) != 0)
634 printf("ipfw_hook() error\n");
637 TUNABLE_INT_FETCH("net.inet6.ip6.fw.enable", &V_fw6_enable);
638 if (V_fw6_enable && (error = ipfw_link(AF_INET6, false)) != 0)
639 printf("ipfw6_hook() error\n");
642 TUNABLE_INT_FETCH("net.link.ether.ipfw", &V_fwlink_enable);
643 if (V_fwlink_enable && (error = ipfw_link(AF_LINK, false)) != 0)
644 printf("ipfw_link_hook() error\n");
650 ipfw_detach_hooks(void)
653 ipfw_unhook(AF_INET);
655 ipfw_unhook(AF_INET6);
657 ipfw_unhook(AF_LINK);
661 ipfw_chg_hook(SYSCTL_HANDLER_ARGS)
667 if (arg1 == &V_fw_enable)
670 else if (arg1 == &V_fw6_enable)
673 else if (arg1 == &V_fwlink_enable)
678 newval = *(int *)arg1;
679 /* Handle sysctl change */
680 error = sysctl_handle_int(oidp, &newval, 0, req);
685 /* Formalize new value */
686 newval = (newval) ? 1 : 0;
688 if (*(int *)arg1 == newval)
691 error = ipfw_link(af, newval == 0 ? true : false);
694 *(int *)arg1 = newval;