Functions
int	nfq_fd (struct nfq_handle *h)

struct nfq_q_handle *	nfq_create_queue (struct nfq_handle h, uint16_t num, nfq_callback cb, void *data)

int	nfq_destroy_queue (struct nfq_q_handle *qh)

int	nfq_handle_packet (struct nfq_handle h, char buf, int len)

int	nfq_set_mode (struct nfq_q_handle *qh, uint8_t mode, uint32_t range)

int	nfq_set_queue_flags (struct nfq_q_handle *qh, uint32_t mask, uint32_t flags)

int	nfq_set_queue_maxlen (struct nfq_q_handle *qh, uint32_t queuelen)

int	nfq_set_verdict (struct nfq_q_handle qh, uint32_t id, uint32_t verdict, uint32_t data_len, const unsigned char buf)

int	nfq_set_verdict2 (struct nfq_q_handle qh, uint32_t id, uint32_t verdict, uint32_t mark, uint32_t data_len, const unsigned char buf)

int	nfq_set_verdict_batch (struct nfq_q_handle *qh, uint32_t id, uint32_t verdict)

int	nfq_set_verdict_batch2 (struct nfq_q_handle *qh, uint32_t id, uint32_t verdict, uint32_t mark)

int	nfq_set_verdict_mark (struct nfq_q_handle qh, uint32_t id, uint32_t verdict, uint32_t mark, uint32_t data_len, const unsigned char buf)

Detailed Description

Once libnetfilter_queue library has been initialised (See LibrarySetup), it is possible to bind the program to a specific queue. This can be done by using nfq_create_queue().

The queue can then be tuned via nfq_set_mode() or nfq_set_queue_maxlen().

Here's a little code snippet that create queue numbered 0:

        printf("binding this socket to queue '0'\n");
        qh = nfq_create_queue(h,  0, &cb, NULL);
        if (!qh) {
                fprintf(stderr, "error during nfq_create_queue()\n");
                exit(1);
        }

        printf("setting copy_packet mode\n");
        if (nfq_set_mode(qh, NFQNL_COPY_PACKET, 0xffff) < 0) {
                fprintf(stderr, "can't set packet_copy mode\n");
                exit(1);
        }

Next step is the handling of incoming packets which can be done via a loop:

        fd = nfq_fd(h);

        while ((rv = recv(fd, buf, sizeof(buf), 0)) >= 0) {
                printf("pkt received\n");
                nfq_handle_packet(h, buf, rv);
        }

When the decision on a packet has been choosed, the verdict has to be given by calling nfq_set_verdict() or nfq_set_verdict2(). The verdict determines the destiny of the packet as follows:

NF_DROP discarded the packet
NF_ACCEPT the packet passes, continue iterations
NF_QUEUE inject the packet into a different queue (the target queue number is in the high 16 bits of the verdict)
NF_REPEAT iterate the same cycle once more
NF_STOP accept, but don't continue iterations

The verdict NF_STOLEN must not be used, as it has special meaning in the kernel. When using NF_REPEAT, one way to prevent re-queueing of the same packet is to also set an nfmark using nfq_set_verdict2, and set up the nefilter rules to only queue a packet when the mark is not (yet) set.

Data and information about the packet can be fetch by using message parsing functions (See Parsing).

Function Documentation

struct nfq_q_handle* nfq_create_queue	(	struct nfq_handle *	h,
		uint16_t	num,
		nfq_callback *	cb,
		void *	data
	)

nfq_create_queue - create a new queue handle and return it.

Parameters

h	Netfilter queue connection handle obtained via call to nfq_open()
num	the number of the queue to bind to
cb	callback function to call for each queued packet
data	custom data to pass to the callback function

Returns: a nfq_q_handle pointing to the newly created queue

Creates a new queue handle, and returns it. The new queue is identified by #num, and the callback specified by #cb will be called for each enqueued packet. The #data argument will be passed unchanged to the callback. If a queue entry with id #num already exists, this function will return failure and the existing entry is unchanged.

The nfq_callback type is defined in libnetfilter_queue.h as:

typedef int nfq_callback(struct nfq_q_handle *qh,
                         struct nfgenmsg *nfmsg,
                         struct nfq_data *nfad, void *data);

Parameters:

qh The queue handle returned by nfq_create_queue
nfmsg message objetc that contains the packet
nfad Netlink packet data handle
data the value passed to the data parameter of nfq_create_queue

The callback should return < 0 to stop processing.

Definition at line 527 of file libnetfilter_queue.c.

int nfq_destroy_queue ( struct nfq_q_handle * qh )

nfq_destroy_queue - destroy a queue handle

Parameters

qh	queue handle that we want to destroy created via nfq_create_queue

Removes the binding for the specified queue handle. This call also unbind from the nfqueue handler, so you don't have to call nfq_unbind_pf.

Definition at line 576 of file libnetfilter_queue.c.

int nfq_fd ( struct nfq_handle * h )

nfq_fd - get the file descriptor associated with the nfqueue handler

Parameters

h	Netfilter queue connection handle obtained via call to nfq_open()

Returns: a file descriptor for the netlink connection associated with the given queue connection handle. The file descriptor can then be used for receiving the queued packets for processing.

This function returns a file descriptor that can be used for communication over the netlink connection associated with the given queue connection handle.

Definition at line 297 of file libnetfilter_queue.c.

int nfq_handle_packet	(	struct nfq_handle *	h,
		char *	buf,
		int	len
	)

nfq_handle_packet - handle a packet received from the nfqueue subsystem

Parameters

h	Netfilter queue connection handle obtained via call to nfq_open()
buf	data to pass to the callback
len	length of packet data in buffer

Triggers an associated callback for the given packet received from the queue. Packets can be read from the queue using nfq_fd() and recv(). See example code for nfq_fd().

Returns: 0 on success, non-zero on failure.

Definition at line 600 of file libnetfilter_queue.c.

int nfq_set_mode	(	struct nfq_q_handle *	qh,
		uint8_t	mode,
		uint32_t	range
	)

nfq_set_mode - set the amount of packet data that nfqueue copies to userspace

Parameters

qh	Netfilter queue handle obtained by call to nfq_create_queue().
mode	the part of the packet that we are interested in
range	size of the packet that we want to get

Sets the amount of data to be copied to userspace for each packet queued to the given queue.

NFQNL_COPY_NONE - noop, do not use it
NFQNL_COPY_META - copy only packet metadata
NFQNL_COPY_PACKET - copy entire packet

Returns: -1 on error; >=0 otherwise.

Definition at line 621 of file libnetfilter_queue.c.

int nfq_set_queue_flags	(	struct nfq_q_handle *	qh,
		uint32_t	mask,
		uint32_t	flags
	)

nfq_set_queue_flags - set flags (options) for the kernel queue

Parameters

qh	Netfilter queue handle obtained by call to nfq_create_queue().
mask	specifies which flag bits to modify
flag	bitmask of flags

Existing flags, that you may want to combine, are:

NFQA_CFG_F_FAIL_OPEN (requires Linux kernel >= 3.6): the kernel will accept the packets if the kernel queue gets full. If this flag is not set, the default action in this case is to drop packets.
NFQA_CFG_F_CONNTRACK (requires Linux kernel >= 3.6): the kernel will include the Connection Tracking system information.
NFQA_CFG_F_GSO (requires Linux kernel >= 3.10): the kernel will not normalize offload packets, i.e. your application will need to be able to handle packets larger than the mtu.

Normalization is expensive, so this flag should always be set. Because attributes in netlink messages are limited to 65531 bytes, you also need to check the NFQA_CAP_LEN attribute, it contains the original size of the captured packet on the kernel side. If it is set and differs from the payload length, the packet was truncated. This also happens when limiting capture size with the NFQNL_COPY_PACKET setting, or when e.g. a local user sends a very large packet.

If your application validates checksums (e.g., tcp checksum), then you must also check if the NFQA_SKB_INFO attribute is present. If it is, you need to test the NFQA_SKB_CSUMNOTREADY bit:
```
        if (attr[NFQA_SKB_INFO]) {
                uint32_t info = ntohl(mnl_attr_get_u32(attr[NFQA_SKB_INFO]));
                if (info & NFQA_SKB_CSUMNOTREADY)
                        validate_checksums = false;
        }
```
if this bit is set, the layer 3/4 checksums of the packet appear incorrect, but are not (because they will be corrected later by the kernel). Please see example/nf-queue.c in the libnetfilter_queue source for more details.
- NFQA_CFG_F_UID_GID: the kernel will dump UID and GID of the socket to which each packet belongs.

Here's a little code snippet to show how to use this API:

        uint32_t flags = NFQA_CFG_F_FAIL_OPEN;
        uint32_t mask = NFQA_CFG_F_FAIL_OPEN;

        printf("Enabling fail-open on this q\n");
        err = nfq_set_queue_flags(qh, mask, flags);

        printf("Disabling fail-open on this q\n");
        flags &= ~NFQA_CFG_F_FAIL_OPEN;
        err = nfq_set_queue_flags(qh, mask, flags);

NFQA_CFG_F_SECCTX: the kernel will dump security context of the socket to which each packet belongs.

Warning: When fragmentation occurs and NFQA_CFG_F_GSO is NOT set then the kernel dumps UID/GID and security context fields only for one fragment. To deal with this limitation always set NFQA_CFG_F_GSO.

Returns: -1 on error with errno set appropriately; =0 otherwise.

Definition at line 711 of file libnetfilter_queue.c.

int nfq_set_queue_maxlen	(	struct nfq_q_handle *	qh,
		uint32_t	queuelen
	)

nfq_set_queue_maxlen - Set kernel queue maximum length parameter

Parameters

qh	Netfilter queue handle obtained by call to nfq_create_queue().
queuelen	the length of the queue

Sets the size of the queue in kernel. This fixes the maximum number of packets the kernel will store before internally before dropping upcoming packets.

Returns: -1 on error; >=0 otherwise.

Definition at line 745 of file libnetfilter_queue.c.

int nfq_set_verdict	(	struct nfq_q_handle *	qh,
		uint32_t	id,
		uint32_t	verdict,
		uint32_t	data_len,
		const unsigned char *	buf
	)

nfq_set_verdict - issue a verdict on a packet

Parameters

qh	Netfilter queue handle obtained by call to nfq_create_queue().
id	ID assigned to packet by netfilter.
verdict	verdict to return to netfilter (NF_ACCEPT, NF_DROP)
data_len	number of bytes of data pointed to by #buf
buf	the buffer that contains the packet data

Can be obtained by:

        int id;
        struct nfqnl_msg_packet_hdr *ph = nfq_get_msg_packet_hdr(tb);
        if (ph)
                id = ntohl(ph->packet_id);

Notifies netfilter of the userspace verdict for the given packet. Every queued packet must have a verdict specified by userspace, either by calling this function, the nfq_set_verdict2() function, or the _batch versions of these functions.

Returns: -1 on error; >= 0 otherwise.

Definition at line 850 of file libnetfilter_queue.c.

int nfq_set_verdict2	(	struct nfq_q_handle *	qh,
		uint32_t	id,
		uint32_t	verdict,
		uint32_t	mark,
		uint32_t	data_len,
		const unsigned char *	buf
	)

nfq_set_verdict2 - like nfq_set_verdict, but you can set the mark.

Parameters

qh	Netfilter queue handle obtained by call to nfq_create_queue().
id	ID assigned to packet by netfilter.
verdict	verdict to return to netfilter (NF_ACCEPT, NF_DROP)
mark	mark to put on packet
data_len	number of bytes of data pointed to by #buf
buf	the buffer that contains the packet data

Definition at line 868 of file libnetfilter_queue.c.

int nfq_set_verdict_batch	(	struct nfq_q_handle *	qh,
		uint32_t	id,
		uint32_t	verdict
	)

nfq_set_verdict_batch - issue verdicts on several packets at once

Parameters

qh	Netfilter queue handle obtained by call to nfq_create_queue().
id	maximum ID of the packets that the verdict should be applied to.
verdict	verdict to return to netfilter (NF_ACCEPT, NF_DROP)

Unlike nfq_set_verdict, the verdict is applied to all queued packets whose packet id is smaller or equal to #id.

batch support was added in Linux 3.1. These functions will fail silently on older kernels.

Definition at line 889 of file libnetfilter_queue.c.

int nfq_set_verdict_batch2	(	struct nfq_q_handle *	qh,
		uint32_t	id,
		uint32_t	verdict,
		uint32_t	mark
	)

nfq_set_verdict_batch2 - like nfq_set_verdict_batch, but you can set a mark.

Parameters

qh	Netfilter queue handle obtained by call to nfq_create_queue().
id	maximum ID of the packets that the verdict should be applied to.
verdict	verdict to return to netfilter (NF_ACCEPT, NF_DROP)
mark	mark to put on packet

Definition at line 904 of file libnetfilter_queue.c.

int nfq_set_verdict_mark	(	struct nfq_q_handle *	qh,
		uint32_t	id,
		uint32_t	verdict,
		uint32_t	mark,
		uint32_t	data_len,
		const unsigned char *	buf
	)

nfq_set_verdict_mark - like nfq_set_verdict, but you can set the mark.

Parameters

qh	Netfilter queue handle obtained by call to nfq_create_queue().
id	ID assigned to packet by netfilter.
verdict	verdict to return to netfilter (NF_ACCEPT, NF_DROP)
mark	the mark to put on the packet, in network byte order.
data_len	number of bytes of data pointed to by #buf
buf	the buffer that contains the packet data

Returns: -1 on error; >= 0 otherwise.

This function is deprecated since it is broken, its use is highly discouraged. Please, use nfq_set_verdict2 instead.

Definition at line 926 of file libnetfilter_queue.c.

Functions

Detailed Description

Function Documentation