ipfilter - Introduction to IP packet filtering
IP Filter is a TCP/IP packet filter, suitable for use in a firewall environment.
To use, it can either be used as a loadable kernel module or incorporated into
your UNIX kernel; use as a loadable kernel module where possible is highly
recommended. Scripts are provided to install and patch system files, as
required.
The IP packet filter can:
- explicitly deny/permit any packet from passing through
- distinguish between various interfaces
- filter by IP networks or hosts
- selectively filter any IP protocol
- selectively filter fragmented IP packets
- selectively filter packets with IP options
- send back an ICMP error/TCP reset for blocked packets
- keep packet state information for TCP, UDP and ICMP packet
flows
- keep fragment state information for any IP packet, applying
the same rule to all fragments.
- act as a Network Address Translator (NAT)
- use redirection to setup true transparent proxy
connections
- provide packet header details to a user program for
authentication
- in addition, supports temporary storage of
pre-authenticated rules for passing packets through
Special provision is made for the three most common Internet protocols, TCP, UDP
and ICMP. The IP Packet filter allows filtering of:
- Inverted host/net matchingTCP/UDP packets by port number or
a port number range
- ICMP packets by type/code
- "established" TCP packets
- On any arbitrary combination of TCP flags
- "short" (fragmented) IP packets with incomplete
headers can be filtered
- any of the 19 IP options or 8 registered IP security
classes TOS (Type of Service) field in packets
To keep track of the performance of the IP packet filter, a logging device is
used which supports logging of:
- the TCP/UDP/ICMP and IP packet headers
- the first 128 bytes of the packet (including headers)
A packet can be logged when:
- it is successfully passed through
- it is blocked from passing through
- it matches a rule setup to look for suspicious packets
IP Filter keeps its own set of statistics on:
- packets blocked
- packets (and bytes!) used for accounting
- packets passed
- packets logged
- attempts to log which failed (buffer full)
- and much more, for packets going both in and out.
The current implementation provides a small set of tools, which can easily be
used and integrated with regular unix shells and tools. A brief description of
the tools provided:
ipf(8) reads in a set of rules, from either stdin or a file, and adds them to
the kernels current list (appending them). It can also be used to flush the
current filter set or delete individual filter rules. The file format is
described in
ipf(5).
ipfs(8) is a utility to temporarily lock the IP Filter kernel tables (state
tables and NAT mappings) and write them to disk. After that the system can be
rebooted, and ipfs can be used to read these tables from disk and restore them
into the kernel. This way the system can be rebooted without the connections
being terminated.
ipfstat(8) interrogates the kernel for statistics on packet filtering, so far,
and retrieves the list of filters in operation for inbound and outbound
packets.
ipftest(1) reads in a filter rule file and then applies sample IP packets to the
rule file. This allows for testing of filter list and examination of how a
packet is passed along through it.
ipmon(8) reads buffered data from the logging device (default is /dev/ipl) for
output to either:
- screen (standard output)
- file
- syslog
ipsend(1) generates arbitary IP packets for ethernet connected machines.
ipresend(1) reads in a data file of saved IP packets (ie snoop/tcpdump/etherfind
output) and sends it back across the network.
iptest(1) contains a set of test "programs" which send out a series of
IP packets, aimed at testing the strength of the TCP/IP stack at which it is
aimed at. WARNING: this may crash machine(s) targeted!
ipnat(8) reads in a set of rules, from either stdin or a file and adds them to
the kernels current list of active NAT rules. NAT rules can also be deleted
using ipnat. The format of the configuration file to be used with ipnat is
described in
ipnat(5).
For use in your own programs (e.g. for writing of transparent application
proxies), the programming interface and the associated ioctl's are documented
in
ipf(4).
Documentation on ioctl's and the format of data saved to the logging character
device is provided in
ipl(4) so that you may develop your own applications to
work with or in place of any of the above.
Similar, the interface to the NAT code is documented in
ipnat(4).
The following diagram illustrates the flow of TCP/IP packets through the various
stages introduced by IP Filter.
IN
|
V
+-------------------------+--------------------------+
| | |
| V |
| Network Address Translation |
| | |
| authenticated | |
| +-------<---------+ |
| | | |
| | V |
| V IP Accounting |
| | | |
| | V |
| | Fragment Cache Check--+ |
| | | | |
| V V V |
| | Packet State Check-->+ |
| | | | |
| | +->--+ | | |
| | | | V | |
| V groups IP Filtering V |
| | | | | | |
| | +--<-+ | | |
| | | | |
| +---------------->|<-----------+ |
| | |
| V |
| +---<----+ |
| | | |
| function | |
| | V |
| +--->----+ |
| | |
| V |
+--|---<--- fast-route ---<--+ |
| | | |
| | V |
| +-------------------------+--------------------------+
| |
| pass only
| |
| V
V [KERNEL TCP/IP Processing]
| |
| +-------------------------+--------------------------+
| | | |
| | V |
| | Fragment Cache Check--+ |
| | | | |
| | V V |
| | Packet State Check-->+ |
| | | | |
| | V | |
V | IP Filtering | |
| | | V |
| | |<-----------+ |
| | V |
| | IP Accounting |
| | | |
| | V |
| | Network Address Translation |
| | | |
| | V |
| +-------------------------+--------------------------+
| |
| pass only
V |
+--------------------------->|
V
OUT
More information (including pointers to the FAQ and the mailing list) can be
obtained from the sofware's official homepage:
www.ipfilter.org
ipf(4),
ipf(5),
ipf(8),
ipfilter(5),
ipfs(8),
ipfstat(8),
ipftest(1),
ipl(4),
ipmon(8),
ipnat(8),
ipnat(4),