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
)
123 if (NFNL_SUBSYS_ID(nlh
->nlmsg_type
) == NFNL_SUBSYS_QUEUE
&&
124 NFNL_MSG_TYPE(nlh
->nlmsg_type
) == NFQNL_MSG_PACKET
)
128 const unsigned char *payload
= NULL
;
129 struct pcap_pkthdr pkth
;
131 const struct nfgenmsg
*nfg
;
134 if (handle
->linktype
!= DLT_NFLOG
) {
135 const struct nfattr
*payload_attr
= NULL
;
137 if (nlh
->nlmsg_len
< HDR_LENGTH
) {
138 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "Malformed message: (nlmsg_len: %u)", nlh
->nlmsg_len
);
142 nfg
= NLMSG_DATA(nlh
);
143 if (nlh
->nlmsg_len
> HDR_LENGTH
) {
144 struct nfattr
*attr
= NFM_NFA(nfg
);
145 int attr_len
= nlh
->nlmsg_len
- NLMSG_ALIGN(HDR_LENGTH
);
147 while (NFA_OK(attr
, attr_len
)) {
148 if (type
== NFQUEUE
) {
149 switch (NFA_TYPE(attr
)) {
150 case NFQA_PACKET_HDR
:
152 const struct nfqnl_msg_packet_hdr
*pkt_hdr
= (const struct nfqnl_msg_packet_hdr
*) NFA_DATA(attr
);
154 id
= ntohl(pkt_hdr
->packet_id
);
162 } else if (type
== NFLOG
) {
163 switch (NFA_TYPE(attr
)) {
169 attr
= NFA_NEXT(attr
, attr_len
);
174 payload
= NFA_DATA(payload_attr
);
175 pkth
.len
= pkth
.caplen
= NFA_PAYLOAD(payload_attr
);
179 payload
= NLMSG_DATA(nlh
);
180 pkth
.caplen
= pkth
.len
= nlh
->nlmsg_len
-NLMSG_ALIGN(sizeof(struct nlmsghdr
));
184 /* pkth.caplen = min (payload_len, handle->snapshot); */
186 gettimeofday(&pkth
.ts
, NULL
);
187 if (handle
->fcode
.bf_insns
== NULL
||
188 bpf_filter(handle
->fcode
.bf_insns
, payload
, pkth
.len
, pkth
.caplen
))
190 handlep
->packets_read
++;
191 callback(user
, &pkth
, payload
);
196 if (type
== NFQUEUE
) {
197 /* XXX, possible responses: NF_DROP, NF_ACCEPT, NF_STOLEN, NF_QUEUE, NF_REPEAT, NF_STOP */
198 nfqueue_send_verdict(handle
, ntohs(nfg
->res_id
), id
, NF_ACCEPT
);
202 msg_len
= NLMSG_ALIGN(nlh
->nlmsg_len
);
213 netfilter_set_datalink(pcap_t
*handle
, int dlt
)
215 handle
->linktype
= dlt
;
220 netfilter_stats_linux(pcap_t
*handle
, struct pcap_stat
*stats
)
222 struct pcap_netfilter
*handlep
= handle
->priv
;
224 stats
->ps_recv
= handlep
->packets_read
;
226 stats
->ps_ifdrop
= 0;
231 netfilter_inject_linux(pcap_t
*handle
, const void *buf
, size_t size
)
233 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "inject not supported on netfilter devices");
244 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
)
246 char buf
[1024] __attribute__ ((aligned
));
248 struct nlmsghdr
*nlh
= (struct nlmsghdr
*) buf
;
249 struct nfgenmsg
*nfg
= (struct nfgenmsg
*) (buf
+ sizeof(struct nlmsghdr
));
251 struct sockaddr_nl snl
;
252 static unsigned int seq_id
;
258 nlh
->nlmsg_len
= NLMSG_LENGTH(sizeof(struct nfgenmsg
));
259 nlh
->nlmsg_type
= msg_type
;
260 nlh
->nlmsg_flags
= NLM_F_REQUEST
| (ack
? NLM_F_ACK
: 0);
261 nlh
->nlmsg_pid
= 0; /* to kernel */
262 nlh
->nlmsg_seq
= seq_id
;
264 nfg
->nfgen_family
= family
;
265 nfg
->version
= NFNETLINK_V0
;
266 nfg
->res_id
= htons(res_id
);
269 struct nfattr
*nfa
= (struct nfattr
*) (buf
+ NLMSG_ALIGN(nlh
->nlmsg_len
));
271 nfa
->nfa_type
= mynfa
->nfa_type
;
272 nfa
->nfa_len
= NFA_LENGTH(mynfa
->nfa_len
);
273 memcpy(NFA_DATA(nfa
), mynfa
->data
, mynfa
->nfa_len
);
274 nlh
->nlmsg_len
= NLMSG_ALIGN(nlh
->nlmsg_len
) + NFA_ALIGN(nfa
->nfa_len
);
277 memset(&snl
, 0, sizeof(snl
));
278 snl
.nl_family
= AF_NETLINK
;
280 if (sendto(handle
->fd
, nlh
, nlh
->nlmsg_len
, 0, (struct sockaddr
*) &snl
, sizeof(snl
)) == -1)
286 /* waiting for reply loop */
288 socklen_t addrlen
= sizeof(snl
);
291 /* ignore interrupt system call error */
293 len
= recvfrom(handle
->fd
, buf
, sizeof(buf
), 0, (struct sockaddr
*) &snl
, &addrlen
);
294 } while ((len
== -1) && (errno
== EINTR
));
299 if (addrlen
!= sizeof(snl
) || snl
.nl_family
!= AF_NETLINK
) {
304 nlh
= (struct nlmsghdr
*) buf
;
305 if (snl
.nl_pid
!= 0 || seq_id
!= nlh
->nlmsg_seq
) /* if not from kernel or wrong sequence skip */
308 while (len
>= NLMSG_SPACE(0) && NLMSG_OK(nlh
, len
)) {
309 if (nlh
->nlmsg_type
== NLMSG_ERROR
|| (nlh
->nlmsg_type
== NLMSG_DONE
&& nlh
->nlmsg_flags
& NLM_F_MULTI
)) {
310 if (nlh
->nlmsg_len
< NLMSG_ALIGN(sizeof(struct nlmsgerr
))) {
314 errno
= -(*((int *)NLMSG_DATA(nlh
)));
315 return (errno
== 0) ? 0 : -1;
317 nlh
= NLMSG_NEXT(nlh
, len
);
321 return -1; /* never here */
325 nflog_send_config_msg(const pcap_t
*handle
, u_int8_t family
, u_int16_t group_id
, const struct my_nfattr
*mynfa
)
327 return netfilter_send_config_msg(handle
, (NFNL_SUBSYS_ULOG
<< 8) | NFULNL_MSG_CONFIG
, 1, family
, group_id
, mynfa
);
331 nflog_send_config_cmd(const pcap_t
*handle
, u_int16_t group_id
, u_int8_t cmd
, u_int8_t family
)
333 struct nfulnl_msg_config_cmd msg
;
334 struct my_nfattr nfa
;
339 nfa
.nfa_type
= NFULA_CFG_CMD
;
340 nfa
.nfa_len
= sizeof(msg
);
342 return nflog_send_config_msg(handle
, family
, group_id
, &nfa
);
346 nflog_send_config_mode(const pcap_t
*handle
, u_int16_t group_id
, u_int8_t copy_mode
, u_int32_t copy_range
)
348 struct nfulnl_msg_config_mode msg
;
349 struct my_nfattr nfa
;
351 msg
.copy_range
= htonl(copy_range
);
352 msg
.copy_mode
= copy_mode
;
355 nfa
.nfa_type
= NFULA_CFG_MODE
;
356 nfa
.nfa_len
= sizeof(msg
);
358 return nflog_send_config_msg(handle
, AF_UNSPEC
, group_id
, &nfa
);
362 nfqueue_send_verdict(const pcap_t
*handle
, u_int16_t group_id
, u_int32_t id
, u_int32_t verdict
)
364 struct nfqnl_msg_verdict_hdr msg
;
365 struct my_nfattr nfa
;
368 msg
.verdict
= htonl(verdict
);
371 nfa
.nfa_type
= NFQA_VERDICT_HDR
;
372 nfa
.nfa_len
= sizeof(msg
);
374 return netfilter_send_config_msg(handle
, (NFNL_SUBSYS_QUEUE
<< 8) | NFQNL_MSG_VERDICT
, 0, AF_UNSPEC
, group_id
, &nfa
);
378 nfqueue_send_config_msg(const pcap_t
*handle
, u_int8_t family
, u_int16_t group_id
, const struct my_nfattr
*mynfa
)
380 return netfilter_send_config_msg(handle
, (NFNL_SUBSYS_QUEUE
<< 8) | NFQNL_MSG_CONFIG
, 1, family
, group_id
, mynfa
);
384 nfqueue_send_config_cmd(const pcap_t
*handle
, u_int16_t group_id
, u_int8_t cmd
, u_int16_t pf
)
386 struct nfqnl_msg_config_cmd msg
;
387 struct my_nfattr nfa
;
393 nfa
.nfa_type
= NFQA_CFG_CMD
;
394 nfa
.nfa_len
= sizeof(msg
);
396 return nfqueue_send_config_msg(handle
, AF_UNSPEC
, group_id
, &nfa
);
400 nfqueue_send_config_mode(const pcap_t
*handle
, u_int16_t group_id
, u_int8_t copy_mode
, u_int32_t copy_range
)
402 struct nfqnl_msg_config_params msg
;
403 struct my_nfattr nfa
;
405 msg
.copy_range
= htonl(copy_range
);
406 msg
.copy_mode
= copy_mode
;
409 nfa
.nfa_type
= NFQA_CFG_PARAMS
;
410 nfa
.nfa_len
= sizeof(msg
);
412 return nfqueue_send_config_msg(handle
, AF_UNSPEC
, group_id
, &nfa
);
416 netfilter_activate(pcap_t
* handle
)
418 const char *dev
= handle
->opt
.source
;
419 unsigned short groups
[32];
421 nftype_t type
= OTHER
;
424 if (strncmp(dev
, NFLOG_IFACE
, strlen(NFLOG_IFACE
)) == 0) {
425 dev
+= strlen(NFLOG_IFACE
);
428 } else if (strncmp(dev
, NFQUEUE_IFACE
, strlen(NFQUEUE_IFACE
)) == 0) {
429 dev
+= strlen(NFQUEUE_IFACE
);
433 if (type
!= OTHER
&& *dev
== ':') {
439 if (group_count
== 32) {
440 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
,
441 "Maximum 32 netfilter groups! dev: %s",
446 group_id
= strtol(dev
, &end_dev
, 0);
447 if (end_dev
!= dev
) {
448 if (group_id
< 0 || group_id
> 65535) {
449 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
,
450 "Netfilter group range from 0 to 65535 (got %ld)",
455 groups
[group_count
++] = (unsigned short) group_id
;
464 if (type
== OTHER
|| *dev
) {
465 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
,
466 "Can't get netfilter group(s) index from %s",
471 /* if no groups, add default: 0 */
477 /* Initialize some components of the pcap structure. */
478 handle
->bufsize
= 128 + handle
->snapshot
;
480 handle
->read_op
= netfilter_read_linux
;
481 handle
->inject_op
= netfilter_inject_linux
;
482 handle
->setfilter_op
= install_bpf_program
; /* no kernel filtering */
483 handle
->setdirection_op
= NULL
;
484 handle
->set_datalink_op
= netfilter_set_datalink
;
485 handle
->getnonblock_op
= pcap_getnonblock_fd
;
486 handle
->setnonblock_op
= pcap_setnonblock_fd
;
487 handle
->stats_op
= netfilter_stats_linux
;
489 /* Create netlink socket */
490 handle
->fd
= socket(AF_NETLINK
, SOCK_RAW
, NETLINK_NETFILTER
);
491 if (handle
->fd
< 0) {
492 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "Can't create raw socket %d:%s", errno
, pcap_strerror(errno
));
497 handle
->linktype
= DLT_NFLOG
;
498 handle
->dlt_list
= (u_int
*) malloc(sizeof(u_int
) * 2);
499 if (handle
->dlt_list
!= NULL
) {
500 handle
->dlt_list
[0] = DLT_NFLOG
;
501 handle
->dlt_list
[1] = DLT_IPV4
;
502 handle
->dlt_count
= 2;
506 handle
->linktype
= DLT_IPV4
;
508 handle
->buffer
= malloc(handle
->bufsize
);
509 if (!handle
->buffer
) {
510 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "Can't allocate dump buffer: %s", pcap_strerror(errno
));
515 if (nflog_send_config_cmd(handle
, 0, NFULNL_CFG_CMD_PF_UNBIND
, AF_INET
) < 0) {
516 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "NFULNL_CFG_CMD_PF_UNBIND: %s", pcap_strerror(errno
));
520 if (nflog_send_config_cmd(handle
, 0, NFULNL_CFG_CMD_PF_BIND
, AF_INET
) < 0) {
521 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "NFULNL_CFG_CMD_PF_BIND: %s", pcap_strerror(errno
));
525 /* Bind socket to the nflog groups */
526 for (i
= 0; i
< group_count
; i
++) {
527 if (nflog_send_config_cmd(handle
, groups
[i
], NFULNL_CFG_CMD_BIND
, AF_UNSPEC
) < 0) {
528 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "Can't listen on group group index: %s", pcap_strerror(errno
));
532 if (nflog_send_config_mode(handle
, groups
[i
], NFULNL_COPY_PACKET
, handle
->snapshot
) < 0) {
533 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "NFULNL_COPY_PACKET: %s", pcap_strerror(errno
));
539 if (nfqueue_send_config_cmd(handle
, 0, NFQNL_CFG_CMD_PF_UNBIND
, AF_INET
) < 0) {
540 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "NFQNL_CFG_CMD_PF_UNBIND: %s", pcap_strerror(errno
));
544 if (nfqueue_send_config_cmd(handle
, 0, NFQNL_CFG_CMD_PF_BIND
, AF_INET
) < 0) {
545 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "NFQNL_CFG_CMD_PF_BIND: %s", pcap_strerror(errno
));
549 /* Bind socket to the nfqueue groups */
550 for (i
= 0; i
< group_count
; i
++) {
551 if (nfqueue_send_config_cmd(handle
, groups
[i
], NFQNL_CFG_CMD_BIND
, AF_UNSPEC
) < 0) {
552 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "Can't listen on group group index: %s", pcap_strerror(errno
));
556 if (nfqueue_send_config_mode(handle
, groups
[i
], NFQNL_COPY_PACKET
, handle
->snapshot
) < 0) {
557 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "NFQNL_COPY_PACKET: %s", pcap_strerror(errno
));
563 if (handle
->opt
.rfmon
) {
565 * Monitor mode doesn't apply to netfilter devices.
567 pcap_cleanup_live_common(handle
);
568 return PCAP_ERROR_RFMON_NOTSUP
;
571 if (handle
->opt
.buffer_size
!= 0) {
573 * Set the socket buffer size to the specified value.
575 if (setsockopt(handle
->fd
, SOL_SOCKET
, SO_RCVBUF
, &handle
->opt
.buffer_size
, sizeof(handle
->opt
.buffer_size
)) == -1) {
576 snprintf(handle
->errbuf
, PCAP_ERRBUF_SIZE
, "SO_RCVBUF: %s", pcap_strerror(errno
));
581 handle
->selectable_fd
= handle
->fd
;
585 pcap_cleanup_live_common(handle
);
590 netfilter_create(const char *device
, char *ebuf
, int *is_ours
)
595 /* Does this look like an netfilter device? */
596 cp
= strrchr(device
, '/');
600 /* Does it begin with NFLOG_IFACE or NFQUEUE_IFACE? */
601 if (strncmp(cp
, NFLOG_IFACE
, sizeof NFLOG_IFACE
- 1) == 0)
602 cp
+= sizeof NFLOG_IFACE
- 1;
603 else if (strncmp(cp
, NFQUEUE_IFACE
, sizeof NFQUEUE_IFACE
- 1) == 0)
604 cp
+= sizeof NFQUEUE_IFACE
- 1;
606 /* Nope, doesn't begin with NFLOG_IFACE nor NFQUEUE_IFACE */
612 * Yes - is that either the end of the name, or is it followed
615 if (*cp
!= ':' && *cp
!= '\0') {
621 /* OK, it's probably ours. */
624 p
= pcap_create_common(device
, ebuf
, sizeof (struct pcap_netfilter
));
628 p
->activate_op
= netfilter_activate
;
633 netfilter_findalldevs(pcap_if_t
**alldevsp
, char *err_str
)
637 sock
= socket(AF_NETLINK
, SOCK_RAW
, NETLINK_NETFILTER
);
639 /* if netlink is not supported this is not fatal */
640 if (errno
== EAFNOSUPPORT
|| errno
== EPROTONOSUPPORT
)
642 snprintf(err_str
, PCAP_ERRBUF_SIZE
, "Can't open netlink socket %d:%s",
643 errno
, pcap_strerror(errno
));
648 if (pcap_add_if(alldevsp
, NFLOG_IFACE
, 0, "Linux netfilter log (NFLOG) interface", err_str
) < 0)
650 if (pcap_add_if(alldevsp
, NFQUEUE_IFACE
, 0, "Linux netfilter queue (NFQUEUE) interface", err_str
) < 0)