On Wed, Jan 08, 2014 at 04:43:47PM +0100, Helmut Schaa wrote:
> Due to a race condition when bringing up an internal port on Linux
> some interface flags (e.g. IFF_MULTICAST) are falsely reset. This
> happens because netlink events may be processed after the according
> netdev has been brought up (which sets interface flags).
>
> Fix this by reading the interface flags just before updating them
> if they have not been updated by from the kernel yet.
>
> Signed-off-by: Helmut Schaa <[email protected]>
Hmm. I see the problem. Thanks for finding and reporting it, and for
the patch.
I have two ideas here:
1. Add (back) VALID_FLAGS and check for it here. (It will
generally be valid, except in the case of initial
construction of internal devices.) I think that this would
probably be better than abusing VALID_DRVINFO.
2. Instead of using an ioctl and ifreq to set interface flags,
use rtnetlink with RTM_SETLINK, with ifi_flags and
ifi_change. In Linux 2.6.22 and later, this allows one to
set a subset of flags, rather than all flags overall, and
that generally reflects OVS userspace intent, which never
tries to change more than one flag at a time. I think that
we could even look at ifi_change in the Netlink response to
see whether the change actually made a difference and (in
pre-2.6.22) spot whether it had some unintended effect so
that we could reverse it.
Ideally, we'd do both.
What do you think?
Thanks,
Ben.
_______________________________________________
dev mailing list
[email protected]
http://openvswitch.org/mailman/listinfo/dev
