Actually, I believe this can now be closed as invalid.
A dependent contributing factor to this issue was the combination of a
network bridge (using brctl) with the VLANs. This worked slightly
differently under several other networking drivers, including Intel and
Broadcom. The way the Realtek drivers are implemented is apparently a
bit different, though I guess there isn't an exact specification around
this type of functionality, so one may not necessarily be more correct
than the other.
It seems that the bridging and VLAN stages are reversed between the
Realtek drivers and the other drivers I had been using previously. In
the other drivers, the VLAN processing happened first, such that
bridging the default physical adapter would not be aware of VLAN
traffic, but would only bridge packets without 802.1q VLAN tags,
allowing these packets to properly forward to the virtual VLAN
interfaces.
In the Realtek drivers, this functionality seems to be implemented in
reverse order. Bridging the default physical adapter includes all
traffic, including packets with 802.1q VLAN tags. As such, these
packets were never forwarded to the virtual VLAN interface. This can be
worked around by using something like ebtables: ebtables -t broute -A
BROUTING -p 8021q -j DROP . See also: Debian Bug # 506133 .
** Changed in: linux (Ubuntu)
Status: New => Invalid
--
Bridging breaks VLAN support for RTL8111/8168B PCI Express Gigabit Ethernet
controller on Karmic
https://bugs.launchpad.net/bugs/499766
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
--
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
