2 * Copyright (c) 2011 Jakub Zawadzki
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
9 * 1. Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
14 * 3. The name of the author may not be used to endorse or promote
15 * products derived from this software without specific prior written
18 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
19 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
20 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
21 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
22 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
23 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
24 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
25 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
26 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
27 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
28 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
37 #ifdef NEED_STRERROR_H
45 #include <sys/socket.h>
46 #include <arpa/inet.h>
50 #include <netinet/in.h>
51 #include <linux/types.h>
53 #include <linux/netlink.h>
54 #include <linux/netfilter.h>
55 #include <linux/netfilter/nfnetlink.h>
56 #include <linux/netfilter/nfnetlink_log.h>
57 #include <linux/netfilter/nfnetlink_queue.h>
59 /* NOTE: if your program drops privilages after pcap_activate() it WON'T work with nfqueue.
60 * It took me quite some time to debug ;/
62 * Sending any data to nfnetlink socket requires CAP_NET_ADMIN privilages,
63 * and in nfqueue we need to send verdict reply after recving packet.
65 * In tcpdump you can disable dropping privilages with -Z root
68 #include "pcap-netfilter-linux.h"
70 #define HDR_LENGTH (NLMSG_LENGTH(NLMSG_ALIGN(sizeof(struct nfgenmsg))))
72 #define NFLOG_IFACE "nflog"
73 #define NFQUEUE_IFACE "nfqueue"
75 typedef enum { OTHER
= -1, NFLOG
, NFQUEUE
} nftype_t
;
78 * Private data for capturing on Linux netfilter sockets.
80 struct pcap_netfilter
{
81 u_int packets_read
; /* count of packets read with recvfrom() */
84 static int nfqueue_send_verdict(const pcap_t
*handle
, u_int16_t group_id
, u_int32_t id
, u_int32_t verdict
);
87 netfilter_read_linux(pcap_t
*handle
, int max_packets
, pcap_handler callback
, u_char
*user
)
89 struct pcap_netfilter
*handlep
= handle
->priv
;
90 const unsigned char *buf
;
94 /* ignore interrupt system call error */
96 len
= recv(handle
->fd
, handle
->buffer
, handle
->bufsize
, 0);
97 if (handle
->break_loop
) {
98 handle
->break_loop
= 0;
101 } while ((len
== -1) && (errno
== EINTR
));
104 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "Can't receive packet %d:%s", errno
, pcap_strerror(errno
));
108 buf
= handle
->buffer
;
109 while (len
>= NLMSG_SPACE(0)) {
110 const struct nlmsghdr
*nlh
= (const struct nlmsghdr
*) buf
;
112 nftype_t type
= OTHER
;
114 if (nlh
->nlmsg_len
< sizeof(struct nlmsghdr
) || len
< nlh
->nlmsg_len
) {
115 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "Message truncated: (got: %d) (nlmsg_len: %u)", len
, nlh
->nlmsg_len
);
119 if (NFNL_SUBSYS_ID(nlh
->nlmsg_type
) == NFNL_SUBSYS_ULOG
&&
120 NFNL_MSG_TYPE(nlh
->nlmsg_type
) == NFULNL_MSG_PACKET
)
122 else if (NFNL_SUBSYS_ID(nlh
->nlmsg_type
) == NFNL_SUBSYS_QUEUE
&&
123 NFNL_MSG_TYPE(nlh
->nlmsg_type
) == NFQNL_MSG_PACKET
)
127 const unsigned char *payload
= NULL
;
128 struct pcap_pkthdr pkth
;
130 const struct nfgenmsg
*nfg
= NULL
;
133 if (handle
->linktype
!= DLT_NFLOG
) {
134 const struct nfattr
*payload_attr
= NULL
;
136 if (nlh
->nlmsg_len
< HDR_LENGTH
) {
137 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "Malformed message: (nlmsg_len: %u)", nlh
->nlmsg_len
);
141 nfg
= NLMSG_DATA(nlh
);
142 if (nlh
->nlmsg_len
> HDR_LENGTH
) {
143 struct nfattr
*attr
= NFM_NFA(nfg
);
144 int attr_len
= nlh
->nlmsg_len
- NLMSG_ALIGN(HDR_LENGTH
);
146 while (NFA_OK(attr
, attr_len
)) {
147 if (type
== NFQUEUE
) {
148 switch (NFA_TYPE(attr
)) {
149 case NFQA_PACKET_HDR
:
151 const struct nfqnl_msg_packet_hdr
*pkt_hdr
= (const struct nfqnl_msg_packet_hdr
*) NFA_DATA(attr
);
153 id
= ntohl(pkt_hdr
->packet_id
);
161 } else if (type
== NFLOG
) {
162 switch (NFA_TYPE(attr
)) {
168 attr
= NFA_NEXT(attr
, attr_len
);
173 payload
= NFA_DATA(payload_attr
);
174 pkth
.len
= pkth
.caplen
= NFA_PAYLOAD(payload_attr
);
178 payload
= NLMSG_DATA(nlh
);
179 pkth
.caplen
= pkth
.len
= nlh
->nlmsg_len
-NLMSG_ALIGN(sizeof(struct nlmsghdr
));
183 /* pkth.caplen = min (payload_len, handle->snapshot); */
185 gettimeofday(&pkth
.ts
, NULL
);
186 if (handle
->fcode
.bf_insns
== NULL
||
187 bpf_filter(handle
->fcode
.bf_insns
, payload
, pkth
.len
, pkth
.caplen
))
189 handlep
->packets_read
++;
190 callback(user
, &pkth
, payload
);
195 if (type
== NFQUEUE
) {
196 /* XXX, possible responses: NF_DROP, NF_ACCEPT, NF_STOLEN, NF_QUEUE, NF_REPEAT, NF_STOP */
197 /* if type == NFQUEUE, handle->linktype is always != DLT_NFLOG,
198 so nfg is always initialized to NLMSG_DATA(nlh). */
200 nfqueue_send_verdict(handle
, ntohs(nfg
->res_id
), id
, NF_ACCEPT
);
204 msg_len
= NLMSG_ALIGN(nlh
->nlmsg_len
);
215 netfilter_set_datalink(pcap_t
*handle
, int dlt
)
217 handle
->linktype
= dlt
;
222 netfilter_stats_linux(pcap_t
*handle
, struct pcap_stat
*stats
)
224 struct pcap_netfilter
*handlep
= handle
->priv
;
226 stats
->ps_recv
= handlep
->packets_read
;
228 stats
->ps_ifdrop
= 0;
233 netfilter_inject_linux(pcap_t
*handle
, const void *buf
, size_t size
)
235 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "inject not supported on netfilter devices");
246 netfilter_send_config_msg(const pcap_t
*handle
, u_int16_t msg_type
, int ack
, u_int8_t family
, u_int16_t res_id
, const struct my_nfattr
*mynfa
)
248 char buf
[1024] __attribute__ ((aligned
));
250 struct nlmsghdr
*nlh
= (struct nlmsghdr
*) buf
;
251 struct nfgenmsg
*nfg
= (struct nfgenmsg
*) (buf
+ sizeof(struct nlmsghdr
));
253 struct sockaddr_nl snl
;
254 static unsigned int seq_id
;
260 nlh
->nlmsg_len
= NLMSG_LENGTH(sizeof(struct nfgenmsg
));
261 nlh
->nlmsg_type
= msg_type
;
262 nlh
->nlmsg_flags
= NLM_F_REQUEST
| (ack
? NLM_F_ACK
: 0);
263 nlh
->nlmsg_pid
= 0; /* to kernel */
264 nlh
->nlmsg_seq
= seq_id
;
266 nfg
->nfgen_family
= family
;
267 nfg
->version
= NFNETLINK_V0
;
268 nfg
->res_id
= htons(res_id
);
271 struct nfattr
*nfa
= (struct nfattr
*) (buf
+ NLMSG_ALIGN(nlh
->nlmsg_len
));
273 nfa
->nfa_type
= mynfa
->nfa_type
;
274 nfa
->nfa_len
= NFA_LENGTH(mynfa
->nfa_len
);
275 memcpy(NFA_DATA(nfa
), mynfa
->data
, mynfa
->nfa_len
);
276 nlh
->nlmsg_len
= NLMSG_ALIGN(nlh
->nlmsg_len
) + NFA_ALIGN(nfa
->nfa_len
);
279 memset(&snl
, 0, sizeof(snl
));
280 snl
.nl_family
= AF_NETLINK
;
282 if (sendto(handle
->fd
, nlh
, nlh
->nlmsg_len
, 0, (struct sockaddr
*) &snl
, sizeof(snl
)) == -1)
288 /* waiting for reply loop */
290 socklen_t addrlen
= sizeof(snl
);
293 /* ignore interrupt system call error */
295 len
= recvfrom(handle
->fd
, buf
, sizeof(buf
), 0, (struct sockaddr
*) &snl
, &addrlen
);
296 } while ((len
== -1) && (errno
== EINTR
));
301 if (addrlen
!= sizeof(snl
) || snl
.nl_family
!= AF_NETLINK
) {
306 nlh
= (struct nlmsghdr
*) buf
;
307 if (snl
.nl_pid
!= 0 || seq_id
!= nlh
->nlmsg_seq
) /* if not from kernel or wrong sequence skip */
310 while (len
>= NLMSG_SPACE(0) && NLMSG_OK(nlh
, len
)) {
311 if (nlh
->nlmsg_type
== NLMSG_ERROR
|| (nlh
->nlmsg_type
== NLMSG_DONE
&& nlh
->nlmsg_flags
& NLM_F_MULTI
)) {
312 if (nlh
->nlmsg_len
< NLMSG_ALIGN(sizeof(struct nlmsgerr
))) {
316 errno
= -(*((int *)NLMSG_DATA(nlh
)));
317 return (errno
== 0) ? 0 : -1;
319 nlh
= NLMSG_NEXT(nlh
, len
);
323 return -1; /* never here */
327 nflog_send_config_msg(const pcap_t
*handle
, u_int8_t family
, u_int16_t group_id
, const struct my_nfattr
*mynfa
)
329 return netfilter_send_config_msg(handle
, (NFNL_SUBSYS_ULOG
<< 8) | NFULNL_MSG_CONFIG
, 1, family
, group_id
, mynfa
);
333 nflog_send_config_cmd(const pcap_t
*handle
, u_int16_t group_id
, u_int8_t cmd
, u_int8_t family
)
335 struct nfulnl_msg_config_cmd msg
;
336 struct my_nfattr nfa
;
341 nfa
.nfa_type
= NFULA_CFG_CMD
;
342 nfa
.nfa_len
= sizeof(msg
);
344 return nflog_send_config_msg(handle
, family
, group_id
, &nfa
);
348 nflog_send_config_mode(const pcap_t
*handle
, u_int16_t group_id
, u_int8_t copy_mode
, u_int32_t copy_range
)
350 struct nfulnl_msg_config_mode msg
;
351 struct my_nfattr nfa
;
353 msg
.copy_range
= htonl(copy_range
);
354 msg
.copy_mode
= copy_mode
;
357 nfa
.nfa_type
= NFULA_CFG_MODE
;
358 nfa
.nfa_len
= sizeof(msg
);
360 return nflog_send_config_msg(handle
, AF_UNSPEC
, group_id
, &nfa
);
364 nfqueue_send_verdict(const pcap_t
*handle
, u_int16_t group_id
, u_int32_t id
, u_int32_t verdict
)
366 struct nfqnl_msg_verdict_hdr msg
;
367 struct my_nfattr nfa
;
370 msg
.verdict
= htonl(verdict
);
373 nfa
.nfa_type
= NFQA_VERDICT_HDR
;
374 nfa
.nfa_len
= sizeof(msg
);
376 return netfilter_send_config_msg(handle
, (NFNL_SUBSYS_QUEUE
<< 8) | NFQNL_MSG_VERDICT
, 0, AF_UNSPEC
, group_id
, &nfa
);
380 nfqueue_send_config_msg(const pcap_t
*handle
, u_int8_t family
, u_int16_t group_id
, const struct my_nfattr
*mynfa
)
382 return netfilter_send_config_msg(handle
, (NFNL_SUBSYS_QUEUE
<< 8) | NFQNL_MSG_CONFIG
, 1, family
, group_id
, mynfa
);
386 nfqueue_send_config_cmd(const pcap_t
*handle
, u_int16_t group_id
, u_int8_t cmd
, u_int16_t pf
)
388 struct nfqnl_msg_config_cmd msg
;
389 struct my_nfattr nfa
;
395 nfa
.nfa_type
= NFQA_CFG_CMD
;
396 nfa
.nfa_len
= sizeof(msg
);
398 return nfqueue_send_config_msg(handle
, AF_UNSPEC
, group_id
, &nfa
);
402 nfqueue_send_config_mode(const pcap_t
*handle
, u_int16_t group_id
, u_int8_t copy_mode
, u_int32_t copy_range
)
404 struct nfqnl_msg_config_params msg
;
405 struct my_nfattr nfa
;
407 msg
.copy_range
= htonl(copy_range
);
408 msg
.copy_mode
= copy_mode
;
411 nfa
.nfa_type
= NFQA_CFG_PARAMS
;
412 nfa
.nfa_len
= sizeof(msg
);
414 return nfqueue_send_config_msg(handle
, AF_UNSPEC
, group_id
, &nfa
);
418 netfilter_activate(pcap_t
* handle
)
420 const char *dev
= handle
->opt
.source
;
421 unsigned short groups
[32];
423 nftype_t type
= OTHER
;
426 if (strncmp(dev
, NFLOG_IFACE
, strlen(NFLOG_IFACE
)) == 0) {
427 dev
+= strlen(NFLOG_IFACE
);
430 } else if (strncmp(dev
, NFQUEUE_IFACE
, strlen(NFQUEUE_IFACE
)) == 0) {
431 dev
+= strlen(NFQUEUE_IFACE
);
435 if (type
!= OTHER
&& *dev
== ':') {
441 if (group_count
== 32) {
442 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
,
443 "Maximum 32 netfilter groups! dev: %s",
448 group_id
= strtol(dev
, &end_dev
, 0);
449 if (end_dev
!= dev
) {
450 if (group_id
< 0 || group_id
> 65535) {
451 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
,
452 "Netfilter group range from 0 to 65535 (got %ld)",
457 groups
[group_count
++] = (unsigned short) group_id
;
466 if (type
== OTHER
|| *dev
) {
467 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
,
468 "Can't get netfilter group(s) index from %s",
473 /* if no groups, add default: 0 */
479 /* Initialize some components of the pcap structure. */
480 handle
->bufsize
= 128 + handle
->snapshot
;
482 handle
->read_op
= netfilter_read_linux
;
483 handle
->inject_op
= netfilter_inject_linux
;
484 handle
->setfilter_op
= install_bpf_program
; /* no kernel filtering */
485 handle
->setdirection_op
= NULL
;
486 handle
->set_datalink_op
= netfilter_set_datalink
;
487 handle
->getnonblock_op
= pcap_getnonblock_fd
;
488 handle
->setnonblock_op
= pcap_setnonblock_fd
;
489 handle
->stats_op
= netfilter_stats_linux
;
491 /* Create netlink socket */
492 handle
->fd
= socket(AF_NETLINK
, SOCK_RAW
, NETLINK_NETFILTER
);
493 if (handle
->fd
< 0) {
494 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "Can't create raw socket %d:%s", errno
, pcap_strerror(errno
));
499 handle
->linktype
= DLT_NFLOG
;
500 handle
->dlt_list
= (u_int
*) malloc(sizeof(u_int
) * 2);
501 if (handle
->dlt_list
!= NULL
) {
502 handle
->dlt_list
[0] = DLT_NFLOG
;
503 handle
->dlt_list
[1] = DLT_IPV4
;
504 handle
->dlt_count
= 2;
508 handle
->linktype
= DLT_IPV4
;
510 handle
->buffer
= malloc(handle
->bufsize
);
511 if (!handle
->buffer
) {
512 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "Can't allocate dump buffer: %s", pcap_strerror(errno
));
517 if (nflog_send_config_cmd(handle
, 0, NFULNL_CFG_CMD_PF_UNBIND
, AF_INET
) < 0) {
518 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "NFULNL_CFG_CMD_PF_UNBIND: %s", pcap_strerror(errno
));
522 if (nflog_send_config_cmd(handle
, 0, NFULNL_CFG_CMD_PF_BIND
, AF_INET
) < 0) {
523 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "NFULNL_CFG_CMD_PF_BIND: %s", pcap_strerror(errno
));
527 /* Bind socket to the nflog groups */
528 for (i
= 0; i
< group_count
; i
++) {
529 if (nflog_send_config_cmd(handle
, groups
[i
], NFULNL_CFG_CMD_BIND
, AF_UNSPEC
) < 0) {
530 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "Can't listen on group group index: %s", pcap_strerror(errno
));
534 if (nflog_send_config_mode(handle
, groups
[i
], NFULNL_COPY_PACKET
, handle
->snapshot
) < 0) {
535 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "NFULNL_COPY_PACKET: %s", pcap_strerror(errno
));
541 if (nfqueue_send_config_cmd(handle
, 0, NFQNL_CFG_CMD_PF_UNBIND
, AF_INET
) < 0) {
542 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "NFQNL_CFG_CMD_PF_UNBIND: %s", pcap_strerror(errno
));
546 if (nfqueue_send_config_cmd(handle
, 0, NFQNL_CFG_CMD_PF_BIND
, AF_INET
) < 0) {
547 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "NFQNL_CFG_CMD_PF_BIND: %s", pcap_strerror(errno
));
551 /* Bind socket to the nfqueue groups */
552 for (i
= 0; i
< group_count
; i
++) {
553 if (nfqueue_send_config_cmd(handle
, groups
[i
], NFQNL_CFG_CMD_BIND
, AF_UNSPEC
) < 0) {
554 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "Can't listen on group group index: %s", pcap_strerror(errno
));
558 if (nfqueue_send_config_mode(handle
, groups
[i
], NFQNL_COPY_PACKET
, handle
->snapshot
) < 0) {
559 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "NFQNL_COPY_PACKET: %s", pcap_strerror(errno
));
565 if (handle
->opt
.rfmon
) {
567 * Monitor mode doesn't apply to netfilter devices.
569 pcap_cleanup_live_common(handle
);
570 return PCAP_ERROR_RFMON_NOTSUP
;
573 if (handle
->opt
.buffer_size
!= 0) {
575 * Set the socket buffer size to the specified value.
577 if (setsockopt(handle
->fd
, SOL_SOCKET
, SO_RCVBUF
, &handle
->opt
.buffer_size
, sizeof(handle
->opt
.buffer_size
)) == -1) {
578 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "SO_RCVBUF: %s", pcap_strerror(errno
));
583 handle
->selectable_fd
= handle
->fd
;
587 pcap_cleanup_live_common(handle
);
592 netfilter_create(const char *device
, char *ebuf
, int *is_ours
)
597 /* Does this look like an netfilter device? */
598 cp
= strrchr(device
, '/');
602 /* Does it begin with NFLOG_IFACE or NFQUEUE_IFACE? */
603 if (strncmp(cp
, NFLOG_IFACE
, sizeof NFLOG_IFACE
- 1) == 0)
604 cp
+= sizeof NFLOG_IFACE
- 1;
605 else if (strncmp(cp
, NFQUEUE_IFACE
, sizeof NFQUEUE_IFACE
- 1) == 0)
606 cp
+= sizeof NFQUEUE_IFACE
- 1;
608 /* Nope, doesn't begin with NFLOG_IFACE nor NFQUEUE_IFACE */
614 * Yes - is that either the end of the name, or is it followed
617 if (*cp
!= ':' && *cp
!= '\0') {
623 /* OK, it's probably ours. */
626 p
= pcap_create_common(device
, ebuf
, sizeof (struct pcap_netfilter
));
630 p
->activate_op
= netfilter_activate
;
635 netfilter_findalldevs(pcap_if_t
**alldevsp
, char *err_str
)
639 sock
= socket(AF_NETLINK
, SOCK_RAW
, NETLINK_NETFILTER
);
641 /* if netlink is not supported this is not fatal */
642 if (errno
== EAFNOSUPPORT
|| errno
== EPROTONOSUPPORT
)
644 snprintf(err_str
, PCAP_ERRBUF_SIZE
, "Can't open netlink socket %d:%s",
645 errno
, pcap_strerror(errno
));
650 if (pcap_add_if(alldevsp
, NFLOG_IFACE
, 0, "Linux netfilter log (NFLOG) interface", err_str
) < 0)
652 if (pcap_add_if(alldevsp
, NFQUEUE_IFACE
, 0, "Linux netfilter queue (NFQUEUE) interface", err_str
) < 0)