ip_fw_pfil.c revision 201740
1/*-
2 * Copyright (c) 2004 Andre Oppermann, Internet Business Solutions AG
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
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.
13 *
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
24 * SUCH DAMAGE.
25 */
26
27#include <sys/cdefs.h>
28__FBSDID("$FreeBSD: head/sys/netinet/ipfw/ip_fw_pfil.c 201740 2010-01-07 12:00:54Z luigi $");
29
30#if !defined(KLD_MODULE)
31#include "opt_ipfw.h"
32#include "opt_ipdn.h"
33#include "opt_inet.h"
34#ifndef INET
35#error IPFIREWALL requires INET.
36#endif /* INET */
37#endif /* KLD_MODULE */
38#include "opt_inet6.h"
39
40#include <sys/param.h>
41#include <sys/systm.h>
42#include <sys/malloc.h>
43#include <sys/mbuf.h>
44#include <sys/module.h>
45#include <sys/kernel.h>
46#include <sys/lock.h>
47#include <sys/rwlock.h>
48#include <sys/socket.h>
49#include <sys/sysctl.h>
50
51#include <net/if.h>
52#include <net/route.h>
53#include <net/pfil.h>
54#include <net/vnet.h>
55
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>
61#include <netinet/ipfw/ip_fw_private.h>
62#include <netgraph/ng_ipfw.h>
63
64#include <machine/in_cksum.h>
65
66static VNET_DEFINE(int, fw_enable) = 1;
67#define V_fw_enable	VNET(fw_enable)
68
69#ifdef INET6
70static VNET_DEFINE(int, fw6_enable) = 1;
71#define V_fw6_enable	VNET(fw6_enable)
72#endif
73
74int ipfw_chg_hook(SYSCTL_HANDLER_ARGS);
75
76/* Forward declarations. */
77static int ipfw_divert(struct mbuf **, int, struct ipfw_rule_ref *, int);
78
79#ifdef SYSCTL_NODE
80SYSCTL_DECL(_net_inet_ip_fw);
81SYSCTL_VNET_PROC(_net_inet_ip_fw, OID_AUTO, enable,
82    CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE3, &VNET_NAME(fw_enable), 0,
83    ipfw_chg_hook, "I", "Enable ipfw");
84#ifdef INET6
85SYSCTL_DECL(_net_inet6_ip6_fw);
86SYSCTL_VNET_PROC(_net_inet6_ip6_fw, OID_AUTO, enable,
87    CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE3, &VNET_NAME(fw6_enable), 0,
88    ipfw_chg_hook, "I", "Enable ipfw+6");
89#endif /* INET6 */
90#endif /* SYSCTL_NODE */
91
92/*
93 * The pfilter hook to pass packets to ipfw_chk and then to
94 * dummynet, divert, netgraph or other modules.
95 * The packet may be consumed.
96 */
97static int
98ipfw_check_hook(void *arg, struct mbuf **m0, struct ifnet *ifp, int dir,
99    struct inpcb *inp)
100{
101	struct ip_fw_args args;
102	struct m_tag *tag;
103	int ipfw;
104	int ret;
105
106	/* all the processing now uses ip_len in net format */
107	if (mtod(*m0, struct ip *)->ip_v == 4)
108		SET_NET_IPLEN(mtod(*m0, struct ip *));
109
110	/* convert dir to IPFW values */
111	dir = (dir == PFIL_IN) ? DIR_IN : DIR_OUT;
112	bzero(&args, sizeof(args));
113
114again:
115	/*
116	 * extract and remove the tag if present. If we are left
117	 * with onepass, optimize the outgoing path.
118	 */
119	tag = m_tag_locate(*m0, MTAG_IPFW_RULE, 0, NULL);
120	if (tag != NULL) {
121		args.rule = *((struct ipfw_rule_ref *)(tag+1));
122		m_tag_delete(*m0, tag);
123		if (args.rule.info & IPFW_ONEPASS) {
124			SET_HOST_IPLEN(mtod(*m0, struct ip *));
125			return 0;
126		}
127	}
128
129	args.m = *m0;
130	args.oif = dir == DIR_OUT ? ifp : NULL;
131	args.inp = inp;
132
133	ipfw = ipfw_chk(&args);
134	*m0 = args.m;
135
136	KASSERT(*m0 != NULL || ipfw == IP_FW_DENY, ("%s: m0 is NULL",
137	    __func__));
138
139	/* breaking out of the switch means drop */
140	ret = 0;	/* default return value for pass */
141	switch (ipfw) {
142	case IP_FW_PASS:
143		/* next_hop may be set by ipfw_chk */
144                if (args.next_hop == NULL)
145                        break; /* pass */
146#ifndef IPFIREWALL_FORWARD
147		ret = EACCES;
148#else
149	    {
150		struct m_tag *fwd_tag;
151
152		/* Incoming packets should not be tagged so we do not
153		 * m_tag_find. Outgoing packets may be tagged, so we
154		 * reuse the tag if present.
155		 */
156		fwd_tag = (dir == DIR_IN) ? NULL :
157			m_tag_find(*m0, PACKET_TAG_IPFORWARD, NULL);
158		if (fwd_tag != NULL) {
159			m_tag_unlink(*m0, fwd_tag);
160		} else {
161			fwd_tag = m_tag_get(PACKET_TAG_IPFORWARD,
162				sizeof(struct sockaddr_in), M_NOWAIT);
163			if (fwd_tag == NULL) {
164				ret = EACCES;
165				break; /* i.e. drop */
166			}
167		}
168		bcopy(args.next_hop, (fwd_tag+1), sizeof(struct sockaddr_in));
169		m_tag_prepend(*m0, fwd_tag);
170
171		if (in_localip(args.next_hop->sin_addr))
172			(*m0)->m_flags |= M_FASTFWD_OURS;
173	    }
174#endif
175		break;
176
177	case IP_FW_DENY:
178		ret = EACCES;
179		break; /* i.e. drop */
180
181	case IP_FW_DUMMYNET:
182		ret = EACCES;
183		if (ip_dn_io_ptr == NULL)
184			break; /* i.e. drop */
185		if (mtod(*m0, struct ip *)->ip_v == 4)
186			ret = ip_dn_io_ptr(m0, dir, &args);
187		else if (mtod(*m0, struct ip *)->ip_v == 6)
188			ret = ip_dn_io_ptr(m0, dir | PROTO_IPV6, &args);
189		else
190			break; /* drop it */
191		/*
192		 * XXX should read the return value.
193		 * dummynet normally eats the packet and sets *m0=NULL
194		 * unless the packet can be sent immediately. In this
195		 * case args is updated and we should re-run the
196		 * check without clearing args.
197		 */
198		if (*m0 != NULL)
199			goto again;
200		break;
201
202	case IP_FW_TEE:
203	case IP_FW_DIVERT:
204		if (ip_divert_ptr == NULL) {
205			ret = EACCES;
206			break; /* i.e. drop */
207		}
208		ret = ipfw_divert(m0, dir, &args.rule,
209			(ipfw == IP_FW_TEE) ? 1 : 0);
210		/* continue processing for the original packet (tee). */
211		if (*m0)
212			goto again;
213		break;
214
215	case IP_FW_NGTEE:
216	case IP_FW_NETGRAPH:
217		if (ng_ipfw_input_p == NULL) {
218			ret = EACCES;
219			break; /* i.e. drop */
220		}
221		ret = ng_ipfw_input_p(m0, dir, &args,
222			(ipfw == IP_FW_NGTEE) ? 1 : 0);
223		if (ipfw == IP_FW_NGTEE) /* ignore errors for NGTEE */
224			goto again;	/* continue with packet */
225		break;
226
227	case IP_FW_NAT:
228	case IP_FW_REASS:
229		goto again;		/* continue with packet */
230
231	default:
232		KASSERT(0, ("%s: unknown retval", __func__));
233	}
234
235	if (ret != 0) {
236		if (*m0)
237			FREE_PKT(*m0);
238		*m0 = NULL;
239	}
240	if (*m0 && mtod(*m0, struct ip *)->ip_v == 4)
241		SET_HOST_IPLEN(mtod(*m0, struct ip *));
242	return ret;
243}
244
245/* do the divert, return 1 on error 0 on success */
246static int
247ipfw_divert(struct mbuf **m0, int incoming, struct ipfw_rule_ref *rule,
248	int tee)
249{
250	/*
251	 * ipfw_chk() has already tagged the packet with the divert tag.
252	 * If tee is set, copy packet and return original.
253	 * If not tee, consume packet and send it to divert socket.
254	 */
255	struct mbuf *clone;
256	struct ip *ip;
257	struct m_tag *tag;
258
259	/* Cloning needed for tee? */
260	if (tee == 0) {
261		clone = *m0;	/* use the original mbuf */
262		*m0 = NULL;
263	} else {
264		clone = m_dup(*m0, M_DONTWAIT);
265		/* If we cannot duplicate the mbuf, we sacrifice the divert
266		 * chain and continue with the tee-ed packet.
267		 */
268		if (clone == NULL)
269			return 1;
270	}
271
272	/*
273	 * Divert listeners can normally handle non-fragmented packets,
274	 * but we can only reass in the non-tee case.
275	 * This means that listeners on a tee rule may get fragments,
276	 * and have to live with that.
277	 * Note that we now have the 'reass' ipfw option so if we care
278	 * we can do it before a 'tee'.
279	 */
280	ip = mtod(clone, struct ip *);
281	if (!tee && ntohs(ip->ip_off) & (IP_MF | IP_OFFMASK)) {
282		int hlen;
283		struct mbuf *reass;
284
285		SET_HOST_IPLEN(ip); /* ip_reass wants host order */
286		reass = ip_reass(clone); /* Reassemble packet. */
287		if (reass == NULL)
288			return 0; /* not an error */
289		/* if reass = NULL then it was consumed by ip_reass */
290		/*
291		 * IP header checksum fixup after reassembly and leave header
292		 * in network byte order.
293		 */
294		ip = mtod(reass, struct ip *);
295		hlen = ip->ip_hl << 2;
296		SET_NET_IPLEN(ip);
297		ip->ip_sum = 0;
298		if (hlen == sizeof(struct ip))
299			ip->ip_sum = in_cksum_hdr(ip);
300		else
301			ip->ip_sum = in_cksum(reass, hlen);
302		clone = reass;
303	}
304	/* attach a tag to the packet with the reinject info */
305	tag = m_tag_alloc(MTAG_IPFW_RULE, 0,
306		    sizeof(struct ipfw_rule_ref), M_NOWAIT);
307	if (tag == NULL) {
308		FREE_PKT(clone);
309		return 1;
310	}
311	*((struct ipfw_rule_ref *)(tag+1)) = *rule;
312	m_tag_prepend(clone, tag);
313
314	/* Do the dirty job... */
315	ip_divert_ptr(clone, incoming);
316	return 0;
317}
318
319/*
320 * attach or detach hooks for a given protocol family
321 */
322static int
323ipfw_hook(int onoff, int pf)
324{
325	struct pfil_head *pfh;
326
327	pfh = pfil_head_get(PFIL_TYPE_AF, pf);
328	if (pfh == NULL)
329		return ENOENT;
330
331	(void) (onoff ? pfil_add_hook : pfil_remove_hook)
332	    (ipfw_check_hook, NULL, PFIL_IN | PFIL_OUT | PFIL_WAITOK, pfh);
333
334	return 0;
335}
336
337int
338ipfw_attach_hooks(int arg)
339{
340	int error = 0;
341
342	if (arg == 0) /* detach */
343		ipfw_hook(0, AF_INET);
344        else if (V_fw_enable && ipfw_hook(1, AF_INET) != 0) {
345                error = ENOENT; /* see ip_fw_pfil.c::ipfw_hook() */
346                printf("ipfw_hook() error\n");
347        }
348#ifdef INET6
349	if (arg == 0) /* detach */
350		ipfw_hook(0, AF_INET6);
351        else if (V_fw6_enable && ipfw_hook(1, AF_INET6) != 0) {
352                error = ENOENT;
353                printf("ipfw6_hook() error\n");
354        }
355#endif
356	return error;
357}
358
359int
360ipfw_chg_hook(SYSCTL_HANDLER_ARGS)
361{
362	int enable;
363	int oldenable;
364	int error;
365	int af;
366
367	if (arg1 == &VNET_NAME(fw_enable)) {
368		enable = V_fw_enable;
369		af = AF_INET;
370	}
371#ifdef INET6
372	else if (arg1 == &VNET_NAME(fw6_enable)) {
373		enable = V_fw6_enable;
374		af = AF_INET6;
375	}
376#endif
377	else
378		return (EINVAL);
379
380	oldenable = enable;
381
382	error = sysctl_handle_int(oidp, &enable, 0, req);
383
384	if (error)
385		return (error);
386
387	enable = (enable) ? 1 : 0;
388
389	if (enable == oldenable)
390		return (0);
391
392	error = ipfw_hook(enable, af);
393	if (error)
394		return (error);
395	if (af == AF_INET)
396		V_fw_enable = enable;
397#ifdef INET6
398	else if (af == AF_INET6)
399		V_fw6_enable = enable;
400#endif
401
402	return (0);
403}
404/* end of file */
405