BRIDGE(4) BSD Kernel Interfaces Manual BRIDGE(4)NAMEbridge — network bridge device
SYNOPSIS
pseudo-device bridgeDESCRIPTION
The bridge driver creates a logical link between two or more IEEE 802
networks that use the same (or “similar enough”) framing format. For
example, it is possible to bridge Ethernet and 802.11 networks together,
but it is not possible to bridge Ethernet and Token Ring together.
To use bridge, the administrator must first create the interface and con‐
figure the bridge parameters. The bridge is created using the
ifconfig(8) create subcommand. See the ifconfig(8) manual page for fur‐
ther information on configuring bridges.
A bridge can be used to provide several services, such as a simple
802.11-to-Ethernet bridge for wireless hosts, and traffic isolation.
A bridge works like a hub, forwarding traffic from one interface to
another. Multicast and broadcast packets are always forwarded to all
interfaces that are part of the bridge. For unicast traffic, the bridge
learns which MAC addresses are associated with which interfaces and will
forward the traffic selectively.
The bridge operates in a safe mode by default, setting the MAC source in
the link header on outgoing packets to the outgoing interface MAC. This
reduces the chance that the layer-2 switching in your switches will
become confused.
The bridge supports various special features via link options.
The link0 option enables transparent bridging mode. The bridge will make
every effort to retain the ethernet header when forwarding packets
between interfaces, making the bridging function work more like a hard‐
ware bridge device.
The link1 option enables keepalive transmission and automatically places
a member into a special blocked mode if no keepalive reception occurs.
If either sides of the link uses this option then both sides must use
this option. This option is impemented by sending CFG updates on the
hello interval to the remote. The link is considered lost after 10
intervals (typically 20 seconds).
The link2 option enables channel bonding (see also ifbondweight). All
member interfaces with the same mac address are considered to be in a
bonding group. When something like tap(4) is used, you can manually con‐
trol or copy the mac to create bonding groups. When interface bonding is
enabled normally blocked interfaces belonging to the same bonding group
as an active forwarding interface will be changed to the bonding state.
Both sides of link the member represents must operate in bonding mode for
this to work, otherwise the remote end may decide to throw away half your
packets.
If your network becomes glitchy, with long pauses in tcp sessions, then
transparent bridging mode is likely the cause. This mode should only be
used when you are bridging networks with devices that do MAC-based secu‐
rity or firewalling (for example, the supremely braindead at&t uverse
router), or which impose severe limitations on MAC:IP assignments.
If member interfaces constantly enter a 'blocked (link1)' state then the
other end of those interfaces is not implementing the link1 keepalive.
Both sides must implement the keepalive.
If you get an enormous amount of packet loss and are using link2-based
bonding, then the other side of those member interfaces are probably not
implementing link2-based bonding.
The bridge driver implements the IEEE 802.1D Spanning Tree protocol
(STP). Spanning Tree is used to detect and remove loops in a network
topology.
Packet filtering can be used with any firewall package that hooks in via
the pfil(9) framework. When filtering is enabled, bridged packets will
pass through the filter inbound on the originating interface, on the
bridge interface and outbound on the appropriate interfaces. Either
stage can be disabled, this behaviour can be controlled using sysctl(8):
Set net.link.bridge.pfil_member to 1 to enable filtering on the incoming
and outgoing member interfaces and set net.link.bridge.pfil_bridge to 1
to enable filtering on the bridge interface.
ARP and REVARP packets are forwarded without being filtered and others
that are not IP nor IPv6 packets are not forwarded when filtering is
enabled.
Note that packets to and from the bridging host will be seen by the fil‐
ter on the interface with the appropriate address configured as well as
on the interface on which the packet arrives or departs.
The MTU of the first member interface to be added is used as the bridge
MTU, all additional members are required to have exactly the same value.
EXTRA FEATURES
DragonFly implements two additional features to make spanning tree opera‐
tion more resilient.
Specifying link0 on the bridge interface places the bridge in transparent
bridging mode. The bridge will make every attempt to retain the original
source MAC in the ethernet link header.
Specifying link1 on the bridge interface forces the bridge to generate a
802.11d CFG message on every hello interval for all interfaces partici‐
pating in the STP protocol. Normally CFG messages are only generated by
the root bridge interface or during topology changes. In addition the
bridge code expects to receive 802.11d frames from all interface partici‐
pating in the STP protocol.
An interface which fails to receive a 802.11d frame within 10 times the
hello interval (usually 20 seconds) automatically goes into l1blocking
mode, which can be observed in the ifconfig output for the bridge. This
removes the interface from consideration and the bridge code automati‐
cally routes around it.
Using link0 and link1 together between two DragonFly boxes allows you to
maintain multiple parallel vpns between those boxes via different net‐
works (if you happen to be on more than one with internet access). Use
separate openvpn instances and tap devices for each vpn link to accom‐
plish this, placing them in the same bridge interface on the two end‐
points. The tap devices do not need any IP configuration when bridged
and can be assigned the same ether MAC (in fact they have to be if you
want the failover to work nicely).
SEE ALSOpf(4), ifconfig(8)HISTORY
The bridge driver first appeared in OpenBSD 2.5 and found its way into
DragonFly 1.3. Transparent bridging (link0) was added in DragonFly 2.9
in 2011.
AUTHORS
The bridge driver was originally written by Jason L. Wright
⟨jason@thought.net⟩ as part of an undergraduate independent study at the
University of North Carolina at Greensboro.
This version of the bridge driver has been heavily modified from the
original version by Jason R. Thorpe ⟨thorpej@wasabisystems.com⟩.
BUGS
The bridge driver currently supports only Ethernet and Ethernet-like
(e.g. 802.11) network devices, with exactly the same interface MTU size
as the bridge device.
BSD February 22, 2011 BSD
