netfilter: passive OS fingerprint xtables match
authorEvgeniy Polyakov <zbr@ioremap.net>
Mon, 8 Jun 2009 15:01:51 +0000 (17:01 +0200)
committerPatrick McHardy <kaber@trash.net>
Mon, 8 Jun 2009 15:01:51 +0000 (17:01 +0200)
Passive OS fingerprinting netfilter module allows to passively detect
remote OS and perform various netfilter actions based on that knowledge.
This module compares some data (WS, MSS, options and it's order, ttl, df
and others) from packets with SYN bit set with dynamically loaded OS
fingerprints.

Fingerprint matching rules can be downloaded from OpenBSD source tree
or found in archive and loaded via netfilter netlink subsystem into
the kernel via special util found in archive.

Archive contains library file (also attached), which was shipped
with iptables extensions some time ago (at least when ipt_osf existed
in patch-o-matic).

Following changes were made in this release:
 * added NLM_F_CREATE/NLM_F_EXCL checks
 * dropped _rcu list traversing helpers in the protected add/remove calls
 * dropped unneded structures, debug prints, obscure comment and check

Fingerprints can be downloaded from
http://www.openbsd.org/cgi-bin/cvsweb/src/etc/pf.os
or can be found in archive

Example usage:
-d switch removes fingerprints

Please consider for inclusion.
Thank you.

Passive OS fingerprint homepage (archives, examples):
http://www.ioremap.net/projects/osf

Signed-off-by: Evgeniy Polyakov <zbr@ioremap.net>
Signed-off-by: Patrick McHardy <kaber@trash.net>
include/linux/netfilter/Kbuild
include/linux/netfilter/nfnetlink.h
include/linux/netfilter/xt_osf.h [new file with mode: 0644]
net/netfilter/Kconfig
net/netfilter/Makefile
net/netfilter/xt_osf.c [new file with mode: 0644]

index af9d2fb97212966ae2661894f3fe0d7e872ae376..2aea50399c0b0f761eaed0d11098a0c2a4c97ddc 100644 (file)
@@ -33,6 +33,7 @@ header-y += xt_limit.h
 header-y += xt_mac.h
 header-y += xt_mark.h
 header-y += xt_multiport.h
+header-y += xt_osf.h
 header-y += xt_owner.h
 header-y += xt_pkttype.h
 header-y += xt_quota.h
index 2214e516146105104668e4b6d72bb6d03013cc23..bff4d5741d98c19ad32a5e87a401713514a67dbc 100644 (file)
@@ -46,7 +46,8 @@ struct nfgenmsg {
 #define NFNL_SUBSYS_CTNETLINK_EXP      2
 #define NFNL_SUBSYS_QUEUE              3
 #define NFNL_SUBSYS_ULOG               4
-#define NFNL_SUBSYS_COUNT              5
+#define NFNL_SUBSYS_OSF                        5
+#define NFNL_SUBSYS_COUNT              6
 
 #ifdef __KERNEL__
 
diff --git a/include/linux/netfilter/xt_osf.h b/include/linux/netfilter/xt_osf.h
new file mode 100644 (file)
index 0000000..fd2272e
--- /dev/null
@@ -0,0 +1,133 @@
+/*
+ * Copyright (c) 2003+ Evgeniy Polyakov <johnpol@2ka.mxt.ru>
+ *
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#ifndef _XT_OSF_H
+#define _XT_OSF_H
+
+#define MAXGENRELEN            32
+
+#define XT_OSF_GENRE           (1<<0)
+#define        XT_OSF_TTL              (1<<1)
+#define XT_OSF_LOG             (1<<2)
+#define XT_OSF_INVERT          (1<<3)
+
+#define XT_OSF_LOGLEVEL_ALL    0       /* log all matched fingerprints */
+#define XT_OSF_LOGLEVEL_FIRST  1       /* log only the first matced fingerprint */
+#define XT_OSF_LOGLEVEL_ALL_KNOWN      2 /* do not log unknown packets */
+
+#define XT_OSF_TTL_TRUE                0       /* True ip and fingerprint TTL comparison */
+#define XT_OSF_TTL_LESS                1       /* Check if ip TTL is less than fingerprint one */
+#define XT_OSF_TTL_NOCHECK     2       /* Do not compare ip and fingerprint TTL at all */
+
+struct xt_osf_info {
+       char                    genre[MAXGENRELEN];
+       __u32                   len;
+       __u32                   flags;
+       __u32                   loglevel;
+       __u32                   ttl;
+};
+
+/*
+ * Wildcard MSS (kind of).
+ * It is used to implement a state machine for the different wildcard values
+ * of the MSS and window sizes.
+ */
+struct xt_osf_wc {
+       __u32                   wc;
+       __u32                   val;
+};
+
+/*
+ * This struct represents IANA options
+ * http://www.iana.org/assignments/tcp-parameters
+ */
+struct xt_osf_opt {
+       __u16                   kind, length;
+       struct xt_osf_wc        wc;
+};
+
+struct xt_osf_user_finger {
+       struct xt_osf_wc        wss;
+
+       __u8                    ttl, df;
+       __u16                   ss, mss;
+       __u16                   opt_num;
+
+       char                    genre[MAXGENRELEN];
+       char                    version[MAXGENRELEN];
+       char                    subtype[MAXGENRELEN];
+
+       /* MAX_IPOPTLEN is maximum if all options are NOPs or EOLs */
+       struct xt_osf_opt       opt[MAX_IPOPTLEN];
+};
+
+struct xt_osf_nlmsg {
+       struct xt_osf_user_finger       f;
+       struct iphdr            ip;
+       struct tcphdr           tcp;
+};
+
+/* Defines for IANA option kinds */
+
+enum iana_options {
+       OSFOPT_EOL = 0,         /* End of options */
+       OSFOPT_NOP,             /* NOP */
+       OSFOPT_MSS,             /* Maximum segment size */
+       OSFOPT_WSO,             /* Window scale option */
+       OSFOPT_SACKP,           /* SACK permitted */
+       OSFOPT_SACK,            /* SACK */
+       OSFOPT_ECHO,
+       OSFOPT_ECHOREPLY,
+       OSFOPT_TS,              /* Timestamp option */
+       OSFOPT_POCP,            /* Partial Order Connection Permitted */
+       OSFOPT_POSP,            /* Partial Order Service Profile */
+
+       /* Others are not used in the current OSF */
+       OSFOPT_EMPTY = 255,
+};
+
+/*
+ * Initial window size option state machine: multiple of mss, mtu or
+ * plain numeric value. Can also be made as plain numeric value which
+ * is not a multiple of specified value.
+ */
+enum xt_osf_window_size_options {
+       OSF_WSS_PLAIN   = 0,
+       OSF_WSS_MSS,
+       OSF_WSS_MTU,
+       OSF_WSS_MODULO,
+       OSF_WSS_MAX,
+};
+
+/*
+ * Add/remove fingerprint from the kernel.
+ */
+enum xt_osf_msg_types {
+       OSF_MSG_ADD,
+       OSF_MSG_REMOVE,
+       OSF_MSG_MAX,
+};
+
+enum xt_osf_attr_type {
+       OSF_ATTR_UNSPEC,
+       OSF_ATTR_FINGER,
+       OSF_ATTR_MAX,
+};
+
+#endif                         /* _XT_OSF_H */
index cb3ad741ebf878f2131e557b84d055a0bfbbe4ac..79ba47f042c0f40614c5f7302fd997c7de70ff7b 100644 (file)
@@ -917,6 +917,19 @@ config NETFILTER_XT_MATCH_U32
 
          Details and examples are in the kernel module source.
 
+config NETFILTER_XT_MATCH_OSF
+       tristate '"osf" Passive OS fingerprint match'
+       depends on NETFILTER_ADVANCED && NETFILTER_NETLINK
+       help
+         This option selects the Passive OS Fingerprinting match module
+         that allows to passively match the remote operating system by
+         analyzing incoming TCP SYN packets.
+
+         Rules and loading software can be downloaded from
+         http://www.ioremap.net/projects/osf
+
+         To compile it as a module, choose M here.  If unsure, say N.
+
 endif # NETFILTER_XTABLES
 
 endmenu
index 6282060fbda91245a331cf8d283f22da2d1c5790..49f62ee4e9ffe2d96d9362bdd91c45fa9c193134 100644 (file)
@@ -77,6 +77,7 @@ obj-$(CONFIG_NETFILTER_XT_MATCH_LIMIT) += xt_limit.o
 obj-$(CONFIG_NETFILTER_XT_MATCH_MAC) += xt_mac.o
 obj-$(CONFIG_NETFILTER_XT_MATCH_MARK) += xt_mark.o
 obj-$(CONFIG_NETFILTER_XT_MATCH_MULTIPORT) += xt_multiport.o
+obj-$(CONFIG_NETFILTER_XT_MATCH_OSF) += xt_osf.o
 obj-$(CONFIG_NETFILTER_XT_MATCH_OWNER) += xt_owner.o
 obj-$(CONFIG_NETFILTER_XT_MATCH_PHYSDEV) += xt_physdev.o
 obj-$(CONFIG_NETFILTER_XT_MATCH_PKTTYPE) += xt_pkttype.o
diff --git a/net/netfilter/xt_osf.c b/net/netfilter/xt_osf.c
new file mode 100644 (file)
index 0000000..863e409
--- /dev/null
@@ -0,0 +1,428 @@
+/*
+ * Copyright (c) 2003+ Evgeniy Polyakov <zbr@ioremap.net>
+ *
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+
+#include <linux/if.h>
+#include <linux/inetdevice.h>
+#include <linux/ip.h>
+#include <linux/list.h>
+#include <linux/rculist.h>
+#include <linux/skbuff.h>
+#include <linux/slab.h>
+#include <linux/tcp.h>
+
+#include <net/ip.h>
+#include <net/tcp.h>
+
+#include <linux/netfilter/nfnetlink.h>
+#include <linux/netfilter/x_tables.h>
+#include <net/netfilter/nf_log.h>
+#include <linux/netfilter/xt_osf.h>
+
+struct xt_osf_finger {
+       struct rcu_head                 rcu_head;
+       struct list_head                finger_entry;
+       struct xt_osf_user_finger       finger;
+};
+
+enum osf_fmatch_states {
+       /* Packet does not match the fingerprint */
+       FMATCH_WRONG = 0,
+       /* Packet matches the fingerprint */
+       FMATCH_OK,
+       /* Options do not match the fingerprint, but header does */
+       FMATCH_OPT_WRONG,
+};
+
+/*
+ * Indexed by dont-fragment bit.
+ * It is the only constant value in the fingerprint.
+ */
+static struct list_head xt_osf_fingers[2];
+
+static const struct nla_policy xt_osf_policy[OSF_ATTR_MAX + 1] = {
+       [OSF_ATTR_FINGER]       = { .len = sizeof(struct xt_osf_user_finger) },
+};
+
+static void xt_osf_finger_free_rcu(struct rcu_head *rcu_head)
+{
+       struct xt_osf_finger *f = container_of(rcu_head, struct xt_osf_finger, rcu_head);
+
+       kfree(f);
+}
+
+static int xt_osf_add_callback(struct sock *ctnl, struct sk_buff *skb,
+                       struct nlmsghdr *nlh, struct nlattr *osf_attrs[])
+{
+       struct xt_osf_user_finger *f;
+       struct xt_osf_finger *kf = NULL, *sf;
+       int err = 0;
+
+       if (!osf_attrs[OSF_ATTR_FINGER])
+               return -EINVAL;
+
+       if (!(nlh->nlmsg_flags & NLM_F_CREATE))
+               return -EINVAL;
+
+       f = nla_data(osf_attrs[OSF_ATTR_FINGER]);
+
+       kf = kmalloc(sizeof(struct xt_osf_finger), GFP_KERNEL);
+       if (!kf)
+               return -ENOMEM;
+
+       memcpy(&kf->finger, f, sizeof(struct xt_osf_user_finger));
+
+       list_for_each_entry(sf, &xt_osf_fingers[!!f->df], finger_entry) {
+               if (memcmp(&sf->finger, f, sizeof(struct xt_osf_user_finger)))
+                       continue;
+
+               kfree(kf);
+               kf = NULL;
+
+               if (nlh->nlmsg_flags & NLM_F_EXCL)
+                       err = -EEXIST;
+               break;
+       }
+
+       /*
+        * We are protected by nfnl mutex.
+        */
+       if (kf)
+               list_add_tail_rcu(&kf->finger_entry, &xt_osf_fingers[!!f->df]);
+
+       return err;
+}
+
+static int xt_osf_remove_callback(struct sock *ctnl, struct sk_buff *skb,
+                       struct nlmsghdr *nlh, struct nlattr *osf_attrs[])
+{
+       struct xt_osf_user_finger *f;
+       struct xt_osf_finger *sf;
+       int err = ENOENT;
+
+       if (!osf_attrs[OSF_ATTR_FINGER])
+               return -EINVAL;
+
+       f = nla_data(osf_attrs[OSF_ATTR_FINGER]);
+
+       list_for_each_entry(sf, &xt_osf_fingers[!!f->df], finger_entry) {
+               if (memcmp(&sf->finger, f, sizeof(struct xt_osf_user_finger)))
+                       continue;
+
+               /*
+                * We are protected by nfnl mutex.
+                */
+               list_del_rcu(&sf->finger_entry);
+               call_rcu(&sf->rcu_head, xt_osf_finger_free_rcu);
+
+               err = 0;
+               break;
+       }
+
+       return err;
+}
+
+static const struct nfnl_callback xt_osf_nfnetlink_callbacks[OSF_MSG_MAX] = {
+       [OSF_MSG_ADD]   = {
+               .call           = xt_osf_add_callback,
+               .attr_count     = OSF_ATTR_MAX,
+               .policy         = xt_osf_policy,
+       },
+       [OSF_MSG_REMOVE]        = {
+               .call           = xt_osf_remove_callback,
+               .attr_count     = OSF_ATTR_MAX,
+               .policy         = xt_osf_policy,
+       },
+};
+
+static const struct nfnetlink_subsystem xt_osf_nfnetlink = {
+       .name                   = "osf",
+       .subsys_id              = NFNL_SUBSYS_OSF,
+       .cb_count               = OSF_MSG_MAX,
+       .cb                     = xt_osf_nfnetlink_callbacks,
+};
+
+static inline int xt_osf_ttl(const struct sk_buff *skb, const struct xt_osf_info *info,
+                           unsigned char f_ttl)
+{
+       const struct iphdr *ip = ip_hdr(skb);
+
+       if (info->flags & XT_OSF_TTL) {
+               if (info->ttl == XT_OSF_TTL_TRUE)
+                       return ip->ttl == f_ttl;
+               if (info->ttl == XT_OSF_TTL_NOCHECK)
+                       return 1;
+               else if (ip->ttl <= f_ttl)
+                       return 1;
+               else {
+                       struct in_device *in_dev = __in_dev_get_rcu(skb->dev);
+                       int ret = 0;
+
+                       for_ifa(in_dev) {
+                               if (inet_ifa_match(ip->saddr, ifa)) {
+                                       ret = (ip->ttl == f_ttl);
+                                       break;
+                               }
+                       }
+                       endfor_ifa(in_dev);
+
+                       return ret;
+               }
+       }
+
+       return ip->ttl == f_ttl;
+}
+
+static bool xt_osf_match_packet(const struct sk_buff *skb,
+               const struct xt_match_param *p)
+{
+       const struct xt_osf_info *info = p->matchinfo;
+       const struct iphdr *ip = ip_hdr(skb);
+       const struct tcphdr *tcp;
+       struct tcphdr _tcph;
+       int fmatch = FMATCH_WRONG, fcount = 0;
+       unsigned int optsize = 0, check_WSS = 0;
+       u16 window, totlen, mss = 0;
+       bool df;
+       const unsigned char *optp = NULL, *_optp = NULL;
+       unsigned char opts[MAX_IPOPTLEN];
+       const struct xt_osf_finger *kf;
+       const struct xt_osf_user_finger *f;
+
+       if (!info)
+               return false;
+
+       tcp = skb_header_pointer(skb, ip_hdrlen(skb), sizeof(struct tcphdr), &_tcph);
+       if (!tcp)
+               return false;
+
+       if (!tcp->syn)
+               return false;
+
+       totlen = ntohs(ip->tot_len);
+       df = ntohs(ip->frag_off) & IP_DF;
+       window = ntohs(tcp->window);
+
+       if (tcp->doff * 4 > sizeof(struct tcphdr)) {
+               optsize = tcp->doff * 4 - sizeof(struct tcphdr);
+
+               _optp = optp = skb_header_pointer(skb, ip_hdrlen(skb) +
+                               sizeof(struct tcphdr), optsize, opts);
+       }
+
+       rcu_read_lock();
+       list_for_each_entry_rcu(kf, &xt_osf_fingers[df], finger_entry) {
+               f = &kf->finger;
+
+               if (!(info->flags & XT_OSF_LOG) && strcmp(info->genre, f->genre))
+                       continue;
+
+               optp = _optp;
+               fmatch = FMATCH_WRONG;
+
+               if (totlen == f->ss && xt_osf_ttl(skb, info, f->ttl)) {
+                       int foptsize, optnum;
+
+                       /*
+                        * Should not happen if userspace parser was written correctly.
+                        */
+                       if (f->wss.wc >= OSF_WSS_MAX)
+                               continue;
+
+                       /* Check options */
+
+                       foptsize = 0;
+                       for (optnum = 0; optnum < f->opt_num; ++optnum)
+                               foptsize += f->opt[optnum].length;
+
+                       if (foptsize > MAX_IPOPTLEN ||
+                               optsize > MAX_IPOPTLEN ||
+                               optsize != foptsize)
+                               continue;
+
+                       check_WSS = f->wss.wc;
+
+                       for (optnum = 0; optnum < f->opt_num; ++optnum) {
+                               if (f->opt[optnum].kind == (*optp)) {
+                                       __u32 len = f->opt[optnum].length;
+                                       const __u8 *optend = optp + len;
+                                       int loop_cont = 0;
+
+                                       fmatch = FMATCH_OK;
+
+                                       switch (*optp) {
+                                       case OSFOPT_MSS:
+                                               mss = optp[3];
+                                               mss <<= 8;
+                                               mss |= optp[2];
+
+                                               mss = ntohs(mss);
+                                               break;
+                                       case OSFOPT_TS:
+                                               loop_cont = 1;
+                                               break;
+                                       }
+
+                                       optp = optend;
+                               } else
+                                       fmatch = FMATCH_OPT_WRONG;
+
+                               if (fmatch != FMATCH_OK)
+                                       break;
+                       }
+
+                       if (fmatch != FMATCH_OPT_WRONG) {
+                               fmatch = FMATCH_WRONG;
+
+                               switch (check_WSS) {
+                               case OSF_WSS_PLAIN:
+                                       if (f->wss.val == 0 || window == f->wss.val)
+                                               fmatch = FMATCH_OK;
+                                       break;
+                               case OSF_WSS_MSS:
+                                       /*
+                                        * Some smart modems decrease mangle MSS to 
+                                        * SMART_MSS_2, so we check standard, decreased
+                                        * and the one provided in the fingerprint MSS
+                                        * values.
+                                        */
+#define SMART_MSS_1    1460
+#define SMART_MSS_2    1448
+                                       if (window == f->wss.val * mss ||
+                                           window == f->wss.val * SMART_MSS_1 ||
+                                           window == f->wss.val * SMART_MSS_2)
+                                               fmatch = FMATCH_OK;
+                                       break;
+                               case OSF_WSS_MTU:
+                                       if (window == f->wss.val * (mss + 40) ||
+                                           window == f->wss.val * (SMART_MSS_1 + 40) ||
+                                           window == f->wss.val * (SMART_MSS_2 + 40))
+                                               fmatch = FMATCH_OK;
+                                       break;
+                               case OSF_WSS_MODULO:
+                                       if ((window % f->wss.val) == 0)
+                                               fmatch = FMATCH_OK;
+                                       break;
+                               }
+                       }
+
+                       if (fmatch != FMATCH_OK)
+                               continue;
+
+                       fcount++;
+
+                       if (info->flags & XT_OSF_LOG)
+                               nf_log_packet(p->hooknum, 0, skb, p->in, p->out, NULL,
+                                       "%s [%s:%s] : %pi4:%d -> %pi4:%d hops=%d\n",
+                                       f->genre, f->version, f->subtype,
+                                       &ip->saddr, ntohs(tcp->source),
+                                       &ip->daddr, ntohs(tcp->dest),
+                                       f->ttl - ip->ttl);
+
+                       if ((info->flags & XT_OSF_LOG) &&
+                           info->loglevel == XT_OSF_LOGLEVEL_FIRST)
+                               break;
+               }
+       }
+       rcu_read_unlock();
+
+       if (!fcount && (info->flags & XT_OSF_LOG))
+               nf_log_packet(p->hooknum, 0, skb, p->in, p->out, NULL,
+                       "Remote OS is not known: %pi4:%u -> %pi4:%u\n",
+                               &ip->saddr, ntohs(tcp->source),
+                               &ip->daddr, ntohs(tcp->dest));
+
+       if (fcount)
+               fmatch = FMATCH_OK;
+
+       return fmatch == FMATCH_OK;
+}
+
+static struct xt_match xt_osf_match = {
+       .name           = "osf",
+       .revision       = 0,
+       .family         = NFPROTO_IPV4,
+       .proto          = IPPROTO_TCP,
+       .hooks          = (1 << NF_INET_LOCAL_IN) |
+                               (1 << NF_INET_PRE_ROUTING) |
+                               (1 << NF_INET_FORWARD),
+       .match          = xt_osf_match_packet,
+       .matchsize      = sizeof(struct xt_osf_info),
+       .me             = THIS_MODULE,
+};
+
+static int __init xt_osf_init(void)
+{
+       int err = -EINVAL;
+       int i;
+
+       for (i=0; i<ARRAY_SIZE(xt_osf_fingers); ++i)
+               INIT_LIST_HEAD(&xt_osf_fingers[i]);
+
+       err = nfnetlink_subsys_register(&xt_osf_nfnetlink);
+       if (err < 0) {
+               printk(KERN_ERR "Failed (%d) to register OSF nsfnetlink helper.\n", err);
+               goto err_out_exit;
+       }
+
+       err = xt_register_match(&xt_osf_match);
+       if (err) {
+               printk(KERN_ERR "Failed (%d) to register OS fingerprint "
+                               "matching module.\n", err);
+               goto err_out_remove;
+       }
+
+       return 0;
+
+err_out_remove:
+       nfnetlink_subsys_unregister(&xt_osf_nfnetlink);
+err_out_exit:
+       return err;
+}
+
+static void __exit xt_osf_fini(void)
+{
+       struct xt_osf_finger *f;
+       int i;
+
+       nfnetlink_subsys_unregister(&xt_osf_nfnetlink);
+       xt_unregister_match(&xt_osf_match);
+
+       rcu_read_lock();
+       for (i=0; i<ARRAY_SIZE(xt_osf_fingers); ++i) {
+
+               list_for_each_entry_rcu(f, &xt_osf_fingers[i], finger_entry) {
+                       list_del_rcu(&f->finger_entry);
+                       call_rcu(&f->rcu_head, xt_osf_finger_free_rcu);
+               }
+       }
+       rcu_read_unlock();
+
+       rcu_barrier();
+}
+
+module_init(xt_osf_init);
+module_exit(xt_osf_fini);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Evgeniy Polyakov <zbr@ioremap.net>");
+MODULE_DESCRIPTION("Passive OS fingerprint matching.");
+MODULE_ALIAS_NFNL_SUBSYS(NFNL_SUBSYS_OSF);