On Mon, May 19, 2025 at 11:33:50AM +0100, Lexi Winter wrote:
although it's possible everyone who is affected by this is already aware
of the change, i thought i should send a heads up anyway, if only to
have a single place to discuss this (since there was quite a lot of
discussion).
in short, following this commit...
b61850c4e6f "bridge(4): default net.link.bridge.member_ifaddrs to false"
https://cgit.freebsd.org/src/commit/?id=b61850c4e6f6b0f21b36da7238db969d9090309e
...it is now impossible to use a network interface which has an IP
address assigned to it as a bridge member, or to configure an IP
address on an interface which is a member of a bridge.
Hi, for the sake of clarity, when you say "IP addresses assigned to it as
a bridge member", do you mean assigned via eg rc.conf on the host,
or assigned, for example within a VM, or assigned within a bridge statement? [1]
I have a machine with 2x NICs with static ips assigned in the
usual way in rc.conf. They are also bridge members (they have
to be otherwise the tap interfaces on the bhyve VMs wouldn't work)
Within each vm the interfaces are assigned either static or dynamic
IPs. I don't use vm-bhyve. Do I need to worry? [2]
[1] example - /etc/rc.conf snippet on the bhyve host
ifconfig_bge1="inet REDACTED.REAL.IP netmask 255.255.255.248 mtu 1500 media
1000baseT mediaopt full-duplex,master"
defaultrouter="REDACTED.REAL.GATEWAY"
ifconfig_bge1_ipv6="inet6 accept_rtadv"
#
#
ifconfig_bridge1="addm bge1 addm tap10 addm tap11 addm tap12 \
addm tap13 addm tap14 addm tap15 addm tap16 addm tap17 addm tap18 addm tap19"
#
[2] because here bge1 has an ip addigned to it and is a bridge member
--
