pcap_compile(3PCAP) man page

pcap_compile(3PCAP) man page

Updated: 11 September 2026 • View in plain text • Return to Main Contents

NAME

pcap_compile - compile a filter expression

SYNOPSIS

#include <pcap/pcap.h>

int pcap_compile(pcap_t *p, struct bpf_program *fp,
    const char *str, int optimize, bpf_u_int32 netmask);

DESCRIPTION

pcap_compile() is used to compile the string str into a filter program. See pcap-filter(7) for the syntax of that string. fp is a pointer to a bpf_program structure and is filled in by pcap_compile(). optimize controls whether optimization on the resulting code is performed. netmask specifies the IPv4 netmask (in host byte order) of the network on which packets are being captured; it is used only when checking for IPv4 broadcast addresses in the filter program. If the netmask of the network on which packets are being captured isn't known to the program, or if the network interface does not have exactly one IPv4 netmask, or if packets are being captured on the Linux “any” pseudo-interface, which can capture on more than one network interface, a value of PCAP_NETMASK_UNKNOWN can be supplied; in this case an attempt to compile a filter expression that contains the ip broadcast primitive will fail.

On Linux, if the pcap_t handle corresponds to a live packet capture, the resulting filter program may use Linux BPF extensions. This works transparently if the filter program is used to filter packets on the same pcap_t handle, which should be done when possible. In other use cases trying to use a filter program with BPF extensions in pcap_offline_filter(3PCAP) or for filtering an input savefile would reject more packets than expected because the extensions depend on auxiliary packet data, which would not be available. The workaround is to compile the filter without the extensions by using a pcap_t handle from pcap_open_dead(3PCAP) or pcap_open_offline(3PCAP) rather than a handle from pcap_create(3PCAP) or pcap_open_live(3PCAP).

If BPF extensions are disabled as described above or the OS is not Linux, pcap_compile() may start rejecting some filter expressions for some link-layer header types, this is the expected behaviour. For example, the ifindex keyword is valid for any live capture on Linux, but when reading packets from a savefile, regardless of the OS it is valid for DLT_LINUX_SLL2 only.

By convention, in this function both a null pointer and an empty string are valid arguments for the str input parameter and mean the same: to accept all packets. This should not be confused with an empty resulting filter program: even a program that accepts or rejects all packets comprises at least one BPF instruction, so on success fp.bf_insns is never a null pointer and fp.bf_len is always greater than zero.

RETURN VALUE

pcap_compile() returns 0 on success and PCAP_ERROR on failure. If PCAP_ERROR is returned, pcap_geterr(3PCAP) or pcap_perror(3PCAP) may be called with p as an argument to fetch or display the error text.

BACKWARD COMPATIBILITY

The PCAP_NETMASK_UNKNOWN constant became available in libpcap release 1.1.0.

In libpcap 1.8.0 and later, pcap_compile() can be used in multiple threads within a single process. However, in earlier versions of libpcap, it is not safe to use pcap_compile() in multiple threads in a single process without some form of mutual exclusion allowing only one thread to call it at any given time.

SEE ALSO

pcap(3PCAP), pcap_setfilter(3PCAP), pcap_freecode(3PCAP)

COLOPHON

This HTML man page was generated at 15:15:19 GMT, September 13, 2026 from a source man page in “The Tcpdump Group” git repositories using man2html and other tools.