X-Git-Url: https://git.tcpdump.org/libpcap/blobdiff_plain/cc8520ff5294900d93509eaf843684c51af102a9..HEAD:/pcap-netfilter-linux.c

diff --git a/pcap-netfilter-linux.c b/pcap-netfilter-linux.c
index d55b5c35..90296be7 100644
--- a/pcap-netfilter-linux.c
+++ b/pcap-netfilter-linux.c
@@ -11,8 +11,8 @@
  * 2. Redistributions in binary form must reproduce the above copyright
  * notice, this list of conditions and the following disclaimer in the
  * documentation and/or other materials provided with the distribution.
- * 3. The name of the author may not be used to endorse or promote 
- * products derived from this software without specific prior written 
+ * 3. The name of the author may not be used to endorse or promote
+ * products derived from this software without specific prior written
  * permission.
  *
  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
@@ -28,15 +28,10 @@
  * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  */
 
-#ifdef HAVE_CONFIG_H
-#include "config.h"
-#endif
+#include <config.h>
 
 #include "pcap-int.h"
-
-#ifdef NEED_STRERROR_H
-#include "strerror.h"
-#endif
+#include "diag-control.h"
 
 #include <errno.h>
 #include <stdlib.h>
@@ -51,138 +46,320 @@
 #include <linux/types.h>
 
 #include <linux/netlink.h>
+#include <linux/netfilter.h>
 #include <linux/netfilter/nfnetlink.h>
 #include <linux/netfilter/nfnetlink_log.h>
+#include <linux/netfilter/nfnetlink_queue.h>
+
+/* NOTE: if your program drops privileges after pcap_activate() it WON'T work with nfqueue.
+ *       It took me quite some time to debug ;/
+ *
+ *       Sending any data to nfnetlink socket requires CAP_NET_ADMIN privileges,
+ *       and in nfqueue we need to send verdict reply after receiving packet.
+ *
+ *       In tcpdump you can disable dropping privileges with -Z root
+ */
+
+#include "pcap-netfilter-linux.h"
 
 #define HDR_LENGTH (NLMSG_LENGTH(NLMSG_ALIGN(sizeof(struct nfgenmsg))))
 
 #define NFLOG_IFACE "nflog"
+#define NFQUEUE_IFACE "nfqueue"
+
+typedef enum { OTHER = -1, NFLOG, NFQUEUE } nftype_t;
+
+/*
+ * Private data for capturing on Linux netfilter sockets.
+ */
+struct pcap_netfilter {
+	u_int	packets_read;	/* count of packets read with recvfrom() */
+	u_int   packets_nobufs; /* ENOBUFS counter */
+};
+
+static int nfqueue_send_verdict(const pcap_t *handle, uint16_t group_id, u_int32_t id, u_int32_t verdict);
+
 
 static int
-nflog_read_linux(pcap_t *handle, int max_packets, pcap_handler callback, u_char *user)
+netfilter_read_linux(pcap_t *handle, int max_packets, pcap_handler callback, u_char *user)
 {
-	const unsigned char *buf;
+	struct pcap_netfilter *handlep = handle->priv;
+	register u_char *bp, *ep;
 	int count = 0;
-	int len;
-
-	/* ignore interrupt system call error */
-	do {
-		len = recv(handle->fd, handle->buffer, handle->bufsize, 0);
-		if (handle->break_loop) {
-			handle->break_loop = 0;
-			return -2;
-		}
-	} while ((len == -1) && (errno == EINTR));
+	u_int cc;
 
-	if (len < 0) {
-		snprintf(handle->errbuf, PCAP_ERRBUF_SIZE, "Can't receive packet %d:%s", errno, pcap_strerror(errno));
-		return -1;
+	/*
+	 * Has "pcap_breakloop()" been called?
+	 */
+	if (handle->break_loop) {
+		/*
+		 * Yes - clear the flag that indicates that it
+		 * has, and return PCAP_ERROR_BREAK to indicate
+		 * that we were told to break out of the loop.
+		 */
+		handle->break_loop = 0;
+		return PCAP_ERROR_BREAK;
 	}
+	cc = handle->cc;
+	if (cc == 0) {
+		/*
+		 * The buffer is empty; refill it.
+		 *
+		 * We ignore EINTR, as that might just be due to a signal
+		 * being delivered - if the signal should interrupt the
+		 * loop, the signal handler should call pcap_breakloop()
+		 * to set handle->break_loop (we ignore it on other
+		 * platforms as well).
+		 */
+		ssize_t read_ret;
 
-	buf = handle->buffer;
-	while (len >= NLMSG_SPACE(0)) {
-		const struct nlmsghdr *nlh = (const struct nlmsghdr *) buf;
-		u_int32_t msg_len;
+		do {
+			read_ret = recv(handle->fd, handle->buffer, handle->bufsize, 0);
+			if (handle->break_loop) {
+				handle->break_loop = 0;
+				return PCAP_ERROR_BREAK;
+			}
+			if (read_ret == -1 && errno == ENOBUFS)
+				handlep->packets_nobufs++;
+		} while ((read_ret == -1) && (errno == EINTR || errno == ENOBUFS));
+
+		if (read_ret < 0) {
+			pcapint_fmt_errmsg_for_errno(handle->errbuf,
+			    PCAP_ERRBUF_SIZE, errno, "Can't receive packet");
+			return PCAP_ERROR;
+		}
 
-		if (nlh->nlmsg_len < sizeof(struct nlmsghdr) || len < nlh->nlmsg_len) {
-			snprintf(handle->errbuf, PCAP_ERRBUF_SIZE, "Message truncated: (got: %d) (nlmsg_len: %u)", len, nlh->nlmsg_len);
+		/*
+		 * At this point, read_ret is guaranteed to be
+		 * >= 0 and < p->bufsize; p->bufsize is a u_int,
+		 * so its value is guaranteed to fit in cc, which
+		 * is also a u_int.
+		 */
+		cc = (u_int)read_ret;
+		bp = (unsigned char *)handle->buffer;
+	} else
+		bp = handle->bp;
+
+	/*
+	 * Loop through each message.
+	 *
+	 * This assumes that a single buffer of message will have
+	 * <= INT_MAX packets, so the message count doesn't overflow.
+	 */
+	ep = bp + cc;
+	while (bp < ep) {
+		const struct nlmsghdr *nlh = (const struct nlmsghdr *) bp;
+		uint32_t msg_len;
+		nftype_t type = OTHER;
+		/*
+		 * Has "pcap_breakloop()" been called?
+		 * If so, return immediately - if we haven't read any
+		 * packets, clear the flag and return PCAP_ERROR_BREAK
+		 * to indicate that we were told to break out of the loop,
+		 * otherwise leave the flag set, so that the *next* call
+		 * will break out of the loop without having read any
+		 * packets, and return the number of packets we've
+		 * processed so far.
+		 */
+		if (handle->break_loop) {
+			handle->bp = bp;
+			handle->cc = (u_int)(ep - bp);
+			if (count == 0) {
+				handle->break_loop = 0;
+				return PCAP_ERROR_BREAK;
+			} else
+				return count;
+		}
+		/*
+		 * NLMSG_SPACE(0) might be signed or might be unsigned,
+		 * depending on whether the kernel defines NLMSG_ALIGNTO
+		 * as 4, which older kernels do, or as 4U, which newer
+		 * kernels do.
+		 *
+		 * ep - bp is of type ptrdiff_t, which is signed.
+		 *
+		 * To squelch warnings, we cast both to size_t, which
+		 * is unsigned; ep >= bp, so the cast is safe.
+		 */
+		if ((size_t)(ep - bp) < (size_t)NLMSG_SPACE(0)) {
+			/*
+			 * There's less than one netlink message left
+			 * in the buffer.  Give up.
+			 */
+			break;
+		}
+
+		if (nlh->nlmsg_len < sizeof(struct nlmsghdr) || cc < nlh->nlmsg_len) {
+			snprintf(handle->errbuf, PCAP_ERRBUF_SIZE, "Message truncated: (got: %u) (nlmsg_len: %u)", cc, nlh->nlmsg_len);
 			return -1;
 		}
 
-		if (NFNL_SUBSYS_ID(nlh->nlmsg_type) == NFNL_SUBSYS_ULOG && 
-			NFNL_MSG_TYPE(nlh->nlmsg_type) == NFULNL_MSG_PACKET) 
-		{
-			const struct nfattr *payload_attr = NULL;
+		if (NFNL_SUBSYS_ID(nlh->nlmsg_type) == NFNL_SUBSYS_ULOG &&
+		    NFNL_MSG_TYPE(nlh->nlmsg_type) == NFULNL_MSG_PACKET)
+			type = NFLOG;
+		else if (NFNL_SUBSYS_ID(nlh->nlmsg_type) == NFNL_SUBSYS_QUEUE &&
+		         NFNL_MSG_TYPE(nlh->nlmsg_type) == NFQNL_MSG_PACKET)
+			type = NFQUEUE;
 
-			if (nlh->nlmsg_len < HDR_LENGTH) {
-				snprintf(handle->errbuf, PCAP_ERRBUF_SIZE, "Malformed message: (nlmsg_len: %u)", nlh->nlmsg_len);
-				return -1;
-			}
+		if (type != OTHER) {
+			const unsigned char *payload = NULL;
+			struct pcap_pkthdr pkth;
+
+			const struct nfgenmsg *nfg = NULL;
+			int id = 0;
 
-			if (nlh->nlmsg_len > HDR_LENGTH) {
-				struct nfattr *attr = NFM_NFA(NLMSG_DATA(nlh));
-				int attr_len = nlh->nlmsg_len - NLMSG_ALIGN(HDR_LENGTH);
+			if (handle->linktype != DLT_NFLOG) {
+				const struct nfattr *payload_attr = NULL;
 
-				while (NFA_OK(attr, attr_len)) {
-					switch (NFA_TYPE(attr)) {
-						case NFULA_PAYLOAD:
-							payload_attr = attr;
-							break;
+				if (nlh->nlmsg_len < HDR_LENGTH) {
+					snprintf(handle->errbuf, PCAP_ERRBUF_SIZE, "Malformed message: (nlmsg_len: %u)", nlh->nlmsg_len);
+					return -1;
+				}
+
+				nfg = NLMSG_DATA(nlh);
+				if (nlh->nlmsg_len > HDR_LENGTH) {
+					struct nfattr *attr = NFM_NFA(nfg);
+					int attr_len = nlh->nlmsg_len - NLMSG_ALIGN(HDR_LENGTH);
+
+					while (NFA_OK(attr, attr_len)) {
+						if (type == NFQUEUE) {
+							switch (NFA_TYPE(attr)) {
+								case NFQA_PACKET_HDR:
+									{
+										const struct nfqnl_msg_packet_hdr *pkt_hdr = (const struct nfqnl_msg_packet_hdr *) NFA_DATA(attr);
+
+										id = ntohl(pkt_hdr->packet_id);
+										break;
+									}
+								case NFQA_PAYLOAD:
+									payload_attr = attr;
+									break;
+							}
+
+						} else if (type == NFLOG) {
+							switch (NFA_TYPE(attr)) {
+								case NFULA_PAYLOAD:
+									payload_attr = attr;
+									break;
+							}
+						}
+						attr = NFA_NEXT(attr, attr_len);
 					}
-					attr = NFA_NEXT(attr, attr_len);
 				}
-			}
 
-			if (payload_attr) {
-				struct pcap_pkthdr pkth;
+				if (payload_attr) {
+					payload = NFA_DATA(payload_attr);
+					pkth.len = pkth.caplen = NFA_PAYLOAD(payload_attr);
+				}
 
-				const unsigned char *payload = NFA_DATA(payload_attr);
-				int payload_len = NFA_PAYLOAD(payload_attr);
+			} else {
+				payload = NLMSG_DATA(nlh);
+				pkth.caplen = pkth.len = nlh->nlmsg_len-NLMSG_ALIGN(sizeof(struct nlmsghdr));
+			}
 
-				pkth.len = pkth.caplen = payload_len;
+			if (payload) {
 				/* pkth.caplen = min (payload_len, handle->snapshot); */
 
 				gettimeofday(&pkth.ts, NULL);
 				if (handle->fcode.bf_insns == NULL ||
-						bpf_filter(handle->fcode.bf_insns, payload, pkth.len, pkth.caplen)) 
+						pcapint_filter(handle->fcode.bf_insns, payload, pkth.len, pkth.caplen))
 				{
-					handle->md.packets_read++;
+					handlep->packets_read++;
 					callback(user, &pkth, payload);
 					count++;
 				}
 			}
+
+			if (type == NFQUEUE) {
+				/* XXX, possible responses: NF_DROP, NF_ACCEPT, NF_STOLEN, NF_QUEUE, NF_REPEAT, NF_STOP */
+				/* if type == NFQUEUE, handle->linktype is always != DLT_NFLOG,
+				   so nfg is always initialized to NLMSG_DATA(nlh). */
+				if (nfg != NULL)
+					nfqueue_send_verdict(handle, ntohs(nfg->res_id), id, NF_ACCEPT);
+			}
 		}
 
 		msg_len = NLMSG_ALIGN(nlh->nlmsg_len);
-		if (msg_len > len)
-			msg_len = len;
-
-		len -= msg_len;
-		buf += msg_len;
+		/*
+		 * If the message length would run past the end of the
+		 * buffer, truncate it to the remaining space in the
+		 * buffer.
+		 *
+		 * To squelch warnings, we cast ep - bp to uint32_t, which
+		 * is unsigned and is the type of msg_len; ep >= bp, and
+		 * len should fit in 32 bits (either it's set from an int
+		 * or it's set from a recv() call with a buffer size that's
+		 * an int, and we're assuming either ILP32 or LP64), so
+		 * the cast is safe.
+		 */
+		if (msg_len > (uint32_t)(ep - bp))
+			msg_len = (uint32_t)(ep - bp);
+
+		bp += msg_len;
+		if (count >= max_packets && !PACKET_COUNT_IS_UNLIMITED(max_packets)) {
+			handle->bp = bp;
+			handle->cc = (u_int)(ep - bp);
+			return count;
+		}
 	}
+
+	handle->cc = 0;
 	return count;
 }
 
+static int
+netfilter_set_datalink(pcap_t *handle, int dlt)
+{
+	handle->linktype = dlt;
+	return 0;
+}
+
 static int
 netfilter_stats_linux(pcap_t *handle, struct pcap_stat *stats)
 {
-	stats->ps_recv = handle->md.packets_read;
-	stats->ps_drop = 0;
+	struct pcap_netfilter *handlep = handle->priv;
+
+	stats->ps_recv = handlep->packets_read;
+	stats->ps_drop = handlep->packets_nobufs;
 	stats->ps_ifdrop = 0;
 	return 0;
 }
 
 static int
-netfilter_inject_linux(pcap_t *handle, const void *buf, size_t size)
+netfilter_inject_linux(pcap_t *handle, const void *buf _U_, int size _U_)
 {
-	snprintf(handle->errbuf, PCAP_ERRBUF_SIZE, "inject not supported on netfilter devices");
+	snprintf(handle->errbuf, PCAP_ERRBUF_SIZE,
+	    "Packet injection is not supported on netfilter devices");
 	return (-1);
-}                           
+}
 
 struct my_nfattr {
-	u_int16_t nfa_len;
-	u_int16_t nfa_type;
+	uint16_t nfa_len;
+	uint16_t nfa_type;
 	void *data;
 };
 
 static int
-nflog_send_config_msg(const pcap_t *handle, u_int8_t family, u_int16_t res_id, const struct my_nfattr *mynfa)
+netfilter_send_config_msg(const pcap_t *handle, uint16_t msg_type, int ack, u_int8_t family, u_int16_t res_id, const struct my_nfattr *mynfa)
 {
 	char buf[1024] __attribute__ ((aligned));
+	memset(buf, 0, sizeof(buf));
 
 	struct nlmsghdr *nlh = (struct nlmsghdr *) buf;
 	struct nfgenmsg *nfg = (struct nfgenmsg *) (buf + sizeof(struct nlmsghdr));
 
 	struct sockaddr_nl snl;
 	static unsigned int seq_id;
-	
+
 	if (!seq_id)
+DIAG_OFF_NARROWING
 		seq_id = time(NULL);
+DIAG_ON_NARROWING
 	++seq_id;
 
 	nlh->nlmsg_len = NLMSG_LENGTH(sizeof(struct nfgenmsg));
-	nlh->nlmsg_type = (NFNL_SUBSYS_ULOG << 8) | NFULNL_MSG_CONFIG;
-	nlh->nlmsg_flags = NLM_F_REQUEST | NLM_F_ACK;
+	nlh->nlmsg_type = msg_type;
+	nlh->nlmsg_flags = NLM_F_REQUEST | (ack ? NLM_F_ACK : 0);
 	nlh->nlmsg_pid = 0;	/* to kernel */
 	nlh->nlmsg_seq = seq_id;
 
@@ -205,6 +382,9 @@ nflog_send_config_msg(const pcap_t *handle, u_int8_t family, u_int16_t res_id, c
 	if (sendto(handle->fd, nlh, nlh->nlmsg_len, 0, (struct sockaddr *) &snl, sizeof(snl)) == -1)
 		return -1;
 
+	if (!ack)
+		return 0;
+
 	/* waiting for reply loop */
 	do {
 		socklen_t addrlen = sizeof(snl);
@@ -212,7 +392,11 @@ nflog_send_config_msg(const pcap_t *handle, u_int8_t family, u_int16_t res_id, c
 
 		/* ignore interrupt system call error */
 		do {
-			len = recvfrom(handle->fd, buf, sizeof(buf), 0, (struct sockaddr *) &snl, &addrlen);
+			/*
+			 * The buffer is not so big that its size won't
+			 * fit into an int.
+			 */
+			len = (int)recvfrom(handle->fd, buf, sizeof(buf), 0, (struct sockaddr *) &snl, &addrlen);
 		} while ((len == -1) && (errno == EINTR));
 
 		if (len <= 0)
@@ -227,7 +411,7 @@ nflog_send_config_msg(const pcap_t *handle, u_int8_t family, u_int16_t res_id, c
 		if (snl.nl_pid != 0 || seq_id != nlh->nlmsg_seq)	/* if not from kernel or wrong sequence skip */
 			continue;
 
-		while (len >= NLMSG_SPACE(0) && NLMSG_OK(nlh, len)) {
+		while ((u_int)len >= NLMSG_SPACE(0) && NLMSG_OK(nlh, (u_int)len)) {
 			if (nlh->nlmsg_type == NLMSG_ERROR || (nlh->nlmsg_type == NLMSG_DONE && nlh->nlmsg_flags & NLM_F_MULTI)) {
 				if (nlh->nlmsg_len < NLMSG_ALIGN(sizeof(struct nlmsgerr))) {
 					errno = EBADMSG;
@@ -244,7 +428,13 @@ nflog_send_config_msg(const pcap_t *handle, u_int8_t family, u_int16_t res_id, c
 }
 
 static int
-nflog_send_config_cmd(const pcap_t *handle, u_int16_t group_id, u_int8_t cmd, u_int8_t family)
+nflog_send_config_msg(const pcap_t *handle, uint8_t family, u_int16_t group_id, const struct my_nfattr *mynfa)
+{
+	return netfilter_send_config_msg(handle, (NFNL_SUBSYS_ULOG << 8) | NFULNL_MSG_CONFIG, 1, family, group_id, mynfa);
+}
+
+static int
+nflog_send_config_cmd(const pcap_t *handle, uint16_t group_id, u_int8_t cmd, u_int8_t family)
 {
 	struct nfulnl_msg_config_cmd msg;
 	struct my_nfattr nfa;
@@ -258,8 +448,8 @@ nflog_send_config_cmd(const pcap_t *handle, u_int16_t group_id, u_int8_t cmd, u_
 	return nflog_send_config_msg(handle, family, group_id, &nfa);
 }
 
-static int 
-nflog_send_config_mode(const pcap_t *handle, u_int16_t group_id, u_int8_t copy_mode, u_int32_t copy_range)
+static int
+nflog_send_config_mode(const pcap_t *handle, uint16_t group_id, u_int8_t copy_mode, u_int32_t copy_range)
 {
 	struct nfulnl_msg_config_mode msg;
 	struct my_nfattr nfa;
@@ -275,28 +465,87 @@ nflog_send_config_mode(const pcap_t *handle, u_int16_t group_id, u_int8_t copy_m
 }
 
 static int
-nflog_activate(pcap_t* handle)
+nfqueue_send_verdict(const pcap_t *handle, uint16_t group_id, u_int32_t id, u_int32_t verdict)
 {
-	const char *dev = handle->opt.source;
+	struct nfqnl_msg_verdict_hdr msg;
+	struct my_nfattr nfa;
+
+	msg.id = htonl(id);
+	msg.verdict = htonl(verdict);
+
+	nfa.data = &msg;
+	nfa.nfa_type = NFQA_VERDICT_HDR;
+	nfa.nfa_len = sizeof(msg);
+
+	return netfilter_send_config_msg(handle, (NFNL_SUBSYS_QUEUE << 8) | NFQNL_MSG_VERDICT, 0, AF_UNSPEC, group_id, &nfa);
+}
+
+static int
+nfqueue_send_config_msg(const pcap_t *handle, uint8_t family, u_int16_t group_id, const struct my_nfattr *mynfa)
+{
+	return netfilter_send_config_msg(handle, (NFNL_SUBSYS_QUEUE << 8) | NFQNL_MSG_CONFIG, 1, family, group_id, mynfa);
+}
+
+static int
+nfqueue_send_config_cmd(const pcap_t *handle, uint16_t group_id, u_int8_t cmd, u_int16_t pf)
+{
+	struct nfqnl_msg_config_cmd msg;
+	struct my_nfattr nfa;
+
+	msg.command = cmd;
+	msg.pf = htons(pf);
+
+	nfa.data = &msg;
+	nfa.nfa_type = NFQA_CFG_CMD;
+	nfa.nfa_len = sizeof(msg);
+
+	return nfqueue_send_config_msg(handle, AF_UNSPEC, group_id, &nfa);
+}
+
+static int
+nfqueue_send_config_mode(const pcap_t *handle, uint16_t group_id, u_int8_t copy_mode, u_int32_t copy_range)
+{
+	struct nfqnl_msg_config_params msg;
+	struct my_nfattr nfa;
+
+	msg.copy_range = htonl(copy_range);
+	msg.copy_mode = copy_mode;
+
+	nfa.data = &msg;
+	nfa.nfa_type = NFQA_CFG_PARAMS;
+	nfa.nfa_len = sizeof(msg);
+
+	return nfqueue_send_config_msg(handle, AF_UNSPEC, group_id, &nfa);
+}
+
+static int
+netfilter_activate(pcap_t* handle)
+{
+	const char *dev = handle->opt.device;
 	unsigned short groups[32];
 	int group_count = 0;
+	nftype_t type = OTHER;
 	int i;
 
 	if (strncmp(dev, NFLOG_IFACE, strlen(NFLOG_IFACE)) == 0) {
 		dev += strlen(NFLOG_IFACE);
+		type = NFLOG;
 
-		/* nflog:30,33,42 looks nice, allow it */
-		if (*dev == ':')
-			dev++;
+	} else if (strncmp(dev, NFQUEUE_IFACE, strlen(NFQUEUE_IFACE)) == 0) {
+		dev += strlen(NFQUEUE_IFACE);
+		type = NFQUEUE;
+	}
 
+	if (type != OTHER && *dev == ':') {
+		dev++;
 		while (*dev) {
 			long int group_id;
 			char *end_dev;
 
 			if (group_count == 32) {
 				snprintf(handle->errbuf, PCAP_ERRBUF_SIZE,
-						"Maximum 32 netfilter groups! dev: %s", 
-						handle->opt.source);
+						"Maximum 32 netfilter groups! dev: %s",
+						handle->opt.device);
 				return PCAP_ERROR;
 			}
 
@@ -318,10 +567,10 @@ nflog_activate(pcap_t* handle)
 		}
 	}
 
-	if (*dev) {
+	if (type == OTHER || *dev) {
 		snprintf(handle->errbuf, PCAP_ERRBUF_SIZE,
-				"Can't get netfilter group(s) index from %s", 
-				handle->opt.source);
+				"Can't get netfilter group(s) index from %s",
+				handle->opt.device);
 		return PCAP_ERROR;
 	}
 
@@ -331,60 +580,126 @@ nflog_activate(pcap_t* handle)
 		group_count = 1;
 	}
 
+	/*
+	 * Turn a negative snapshot value (invalid), a snapshot value of
+	 * 0 (unspecified), or a value bigger than the normal maximum
+	 * value, into the maximum allowed value.
+	 *
+	 * If some application really *needs* a bigger snapshot
+	 * length, we should just increase MAXIMUM_SNAPLEN.
+	 */
+	if (handle->snapshot <= 0 || handle->snapshot > MAXIMUM_SNAPLEN)
+		handle->snapshot = MAXIMUM_SNAPLEN;
+
 	/* Initialize some components of the pcap structure. */
 	handle->bufsize = 128 + handle->snapshot;
 	handle->offset = 0;
-	handle->linktype = DLT_IPV4;
-	handle->read_op = nflog_read_linux;
+	handle->read_op = netfilter_read_linux;
 	handle->inject_op = netfilter_inject_linux;
-	handle->setfilter_op = install_bpf_program; /* no kernel filtering */
+	handle->setfilter_op = pcapint_install_bpf_program; /* no kernel filtering */
 	handle->setdirection_op = NULL;
-	handle->set_datalink_op = NULL;
-	handle->getnonblock_op = pcap_getnonblock_fd;
-	handle->setnonblock_op = pcap_setnonblock_fd;
+	handle->set_datalink_op = netfilter_set_datalink;
+	handle->getnonblock_op = pcapint_getnonblock_fd;
+	handle->setnonblock_op = pcapint_setnonblock_fd;
 	handle->stats_op = netfilter_stats_linux;
 
 	/* Create netlink socket */
 	handle->fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_NETFILTER);
 	if (handle->fd < 0) {
-		snprintf(handle->errbuf, PCAP_ERRBUF_SIZE, "Can't create raw socket %d:%s", errno, pcap_strerror(errno));
+		pcapint_fmt_errmsg_for_errno(handle->errbuf, PCAP_ERRBUF_SIZE,
+		    errno, "Can't create raw socket");
 		return PCAP_ERROR;
 	}
 
+	if (type == NFLOG) {
+		handle->linktype = DLT_NFLOG;
+		handle->dlt_list = (u_int *) malloc(sizeof(u_int) * 2);
+		if (handle->dlt_list == NULL) {
+			pcapint_fmt_errmsg_for_errno(handle->errbuf,
+			    PCAP_ERRBUF_SIZE, errno,
+			    "Can't allocate DLT list");
+			goto close_fail;
+		}
+		handle->dlt_list[0] = DLT_NFLOG;
+		handle->dlt_list[1] = DLT_IPV4;
+		handle->dlt_count = 2;
+	} else
+		handle->linktype = DLT_IPV4;
+
 	handle->buffer = malloc(handle->bufsize);
 	if (!handle->buffer) {
-		snprintf(handle->errbuf, PCAP_ERRBUF_SIZE, "Can't allocate dump buffer: %s", pcap_strerror(errno));
+		pcapint_fmt_errmsg_for_errno(handle->errbuf, PCAP_ERRBUF_SIZE,
+		    errno, "Can't allocate dump buffer");
 		goto close_fail;
 	}
 
-	if (nflog_send_config_cmd(handle, 0, NFULNL_CFG_CMD_PF_UNBIND, AF_INET) < 0) {
-		snprintf(handle->errbuf, PCAP_ERRBUF_SIZE, "NFULNL_CFG_CMD_PF_UNBIND: %s", pcap_strerror(errno));
-		goto close_fail;
-	}
+	if (type == NFLOG) {
+		if (nflog_send_config_cmd(handle, 0, NFULNL_CFG_CMD_PF_UNBIND, AF_INET) < 0) {
+			pcapint_fmt_errmsg_for_errno(handle->errbuf,
+			    PCAP_ERRBUF_SIZE, errno,
+			    "NFULNL_CFG_CMD_PF_UNBIND");
+			goto close_fail;
+		}
 
-	if (nflog_send_config_cmd(handle, 0, NFULNL_CFG_CMD_PF_BIND, AF_INET) < 0) {
-		snprintf(handle->errbuf, PCAP_ERRBUF_SIZE, "NFULNL_CFG_CMD_PF_BIND: %s", pcap_strerror(errno));
-		goto close_fail;
-	}
+		if (nflog_send_config_cmd(handle, 0, NFULNL_CFG_CMD_PF_BIND, AF_INET) < 0) {
+			pcapint_fmt_errmsg_for_errno(handle->errbuf,
+			    PCAP_ERRBUF_SIZE, errno, "NFULNL_CFG_CMD_PF_BIND");
+			goto close_fail;
+		}
+
+		/* Bind socket to the nflog groups */
+		for (i = 0; i < group_count; i++) {
+			if (nflog_send_config_cmd(handle, groups[i], NFULNL_CFG_CMD_BIND, AF_UNSPEC) < 0) {
+				pcapint_fmt_errmsg_for_errno(handle->errbuf,
+				    PCAP_ERRBUF_SIZE, errno,
+				    "Can't listen on group index");
+				goto close_fail;
+			}
 
-	/* Bind socket to the nflog groups */
-	for (i = 0; i < group_count; i++) {
-		if (nflog_send_config_cmd(handle, groups[i], NFULNL_CFG_CMD_BIND, AF_UNSPEC) < 0) {
-			snprintf(handle->errbuf, PCAP_ERRBUF_SIZE, "Can't listen on group group index: %s", pcap_strerror(errno));
+			if (nflog_send_config_mode(handle, groups[i], NFULNL_COPY_PACKET, handle->snapshot) < 0) {
+				pcapint_fmt_errmsg_for_errno(handle->errbuf,
+				    PCAP_ERRBUF_SIZE, errno,
+				    "NFULNL_COPY_PACKET");
+				goto close_fail;
+			}
+		}
+
+	} else {
+		if (nfqueue_send_config_cmd(handle, 0, NFQNL_CFG_CMD_PF_UNBIND, AF_INET) < 0) {
+			pcapint_fmt_errmsg_for_errno(handle->errbuf,
+			    PCAP_ERRBUF_SIZE, errno, "NFQNL_CFG_CMD_PF_UNBIND");
 			goto close_fail;
 		}
 
-		if (nflog_send_config_mode(handle, groups[i], NFULNL_COPY_PACKET, handle->snapshot) < 0) {
-			snprintf(handle->errbuf, PCAP_ERRBUF_SIZE, "NFULNL_COPY_PACKET: %s", pcap_strerror(errno));
+		if (nfqueue_send_config_cmd(handle, 0, NFQNL_CFG_CMD_PF_BIND, AF_INET) < 0) {
+			pcapint_fmt_errmsg_for_errno(handle->errbuf,
+			    PCAP_ERRBUF_SIZE, errno, "NFQNL_CFG_CMD_PF_BIND");
 			goto close_fail;
 		}
+
+		/* Bind socket to the nfqueue groups */
+		for (i = 0; i < group_count; i++) {
+			if (nfqueue_send_config_cmd(handle, groups[i], NFQNL_CFG_CMD_BIND, AF_UNSPEC) < 0) {
+				pcapint_fmt_errmsg_for_errno(handle->errbuf,
+				    PCAP_ERRBUF_SIZE, errno,
+				    "Can't listen on group index");
+				goto close_fail;
+			}
+
+			if (nfqueue_send_config_mode(handle, groups[i], NFQNL_COPY_PACKET, handle->snapshot) < 0) {
+				pcapint_fmt_errmsg_for_errno(handle->errbuf,
+				    PCAP_ERRBUF_SIZE, errno,
+				    "NFQNL_COPY_PACKET");
+				goto close_fail;
+			}
+		}
 	}
 
 	if (handle->opt.rfmon) {
 		/*
 		 * Monitor mode doesn't apply to netfilter devices.
 		 */
-		pcap_cleanup_live_common(handle);
+		pcapint_cleanup_live_common(handle);
 		return PCAP_ERROR_RFMON_NOTSUP;
 	}
 
@@ -393,7 +708,8 @@ nflog_activate(pcap_t* handle)
 		 * Set the socket buffer size to the specified value.
 		 */
 		if (setsockopt(handle->fd, SOL_SOCKET, SO_RCVBUF, &handle->opt.buffer_size, sizeof(handle->opt.buffer_size)) == -1) {
-			snprintf(handle->errbuf, PCAP_ERRBUF_SIZE, "SO_RCVBUF: %s", pcap_strerror(errno));
+			pcapint_fmt_errmsg_for_errno(handle->errbuf,
+			    PCAP_ERRBUF_SIZE, errno, "SO_RCVBUF");
 			goto close_fail;
 		}
 	}
@@ -402,42 +718,78 @@ nflog_activate(pcap_t* handle)
 	return 0;
 
 close_fail:
-	pcap_cleanup_live_common(handle);
+	pcapint_cleanup_live_common(handle);
 	return PCAP_ERROR;
 }
 
 pcap_t *
-nflog_create(const char *device, char *ebuf)
+netfilter_create(const char *device, char *ebuf, int *is_ours)
 {
+	const char *cp;
 	pcap_t *p;
 
-	p = pcap_create_common(device, ebuf);
+	/* Does this look like an netfilter device? */
+	cp = device;
+
+	/* Does it begin with NFLOG_IFACE or NFQUEUE_IFACE? */
+	if (strncmp(cp, NFLOG_IFACE, sizeof NFLOG_IFACE - 1) == 0)
+		cp += sizeof NFLOG_IFACE - 1;
+	else if (strncmp(cp, NFQUEUE_IFACE, sizeof NFQUEUE_IFACE - 1) == 0)
+		cp += sizeof NFQUEUE_IFACE - 1;
+	else {
+		/* Nope, doesn't begin with NFLOG_IFACE nor NFQUEUE_IFACE */
+		*is_ours = 0;
+		return NULL;
+	}
+
+	/*
+	 * Yes - is that either the end of the name, or is it followed
+	 * by a colon?
+	 */
+	if (*cp != ':' && *cp != '\0') {
+		/* Nope */
+		*is_ours = 0;
+		return NULL;
+	}
+
+	/* OK, it's probably ours. */
+	*is_ours = 1;
+
+	p = PCAP_CREATE_COMMON(ebuf, struct pcap_netfilter);
 	if (p == NULL)
 		return (NULL);
 
-	p->activate_op = nflog_activate;
+	p->activate_op = netfilter_activate;
 	return (p);
 }
 
-int 
-netfilter_platform_finddevs(pcap_if_t **alldevsp, char *err_str)
+int
+netfilter_findalldevs(pcap_if_list_t *devlistp, char *err_str)
 {
-	pcap_if_t *found_dev = *alldevsp;
 	int sock;
-	
+
 	sock = socket(AF_NETLINK, SOCK_RAW, NETLINK_NETFILTER);
 	if (sock < 0) {
-		/* if netlink is not supported this this is not fatal */
-		if (errno == EAFNOSUPPORT)
+		/* if netlink is not supported this is not fatal */
+		if (errno == EAFNOSUPPORT || errno == EPROTONOSUPPORT)
 			return 0;
-		snprintf(err_str, PCAP_ERRBUF_SIZE, "Can't open netlink socket %d:%s",
-			errno, pcap_strerror(errno));
+		pcapint_fmt_errmsg_for_errno(err_str, PCAP_ERRBUF_SIZE,
+		    errno, "Can't open netlink socket");
 		return -1;
 	}
 	close(sock);
 
-	if (pcap_add_if(&found_dev, NFLOG_IFACE, 0, "Linux netfilter log (NFLOG) interface", err_str) < 0)
+	/*
+	 * The notion of "connected" vs. "disconnected" doesn't apply.
+	 * XXX - what about "up" and "running"?
+	 */
+	if (pcapint_add_dev(devlistp, NFLOG_IFACE,
+	    PCAP_IF_CONNECTION_STATUS_NOT_APPLICABLE,
+	    "Linux netfilter log (NFLOG) interface", err_str) == NULL)
+		return -1;
+	if (pcapint_add_dev(devlistp, NFQUEUE_IFACE,
+	    PCAP_IF_CONNECTION_STATUS_NOT_APPLICABLE,
+	    "Linux netfilter queue (NFQUEUE) interface", err_str) == NULL)
 		return -1;
 	return 0;
 }
-