On Wed, Sep 14, 2022 at 5:14 PM Mike Pattrick <[email protected]> wrote: > > On Tue, Sep 13, 2022 at 5:21 AM David Marchand > <[email protected]> wrote: > > > > At some point in OVS history, some virtio features were announced as > > supported (ECN and UFO virtio features). > > > > The userspace TSO code, which has been added later, does not support > > those features and tries to disable them. > > > > This breaks OVS upgrades: if an existing VM already negotiated such > > features, their lack on reconnection to an upgraded OVS triggers a > > vhost socket disconnection by Qemu. > > This results in an endless loop because Qemu then retries with the same > > set of virtio features. > > > > This patch proposes to try and detect those vhost socket disconnection > > and fallback restoring the old virtio features (and disabling TSO for this > > vhost port). > > > > Signed-off-by: David Marchand <[email protected]> > > --- > > Changelog since v1: > > - added a note in the documentation, > > - fixed vhost unregister trigger (so that both disabling and re-enabling > > TSO is handled), > > - cleared netdev features when disabling TSO, > > - changed level and ratelimited log message on vhost socket disconnect, > > > > This does solve the issue. My only real comment is it may be opaque to > an end user. Do you think it makes sense to add this information to > netdev_dpdk_vhost_user_get_status? Or do you think the log message > should suffice?
Well, log messages may be lost after some time. I agree that we need a place where to get the latest up-to-date info for a particular vhost port. And netdev_dpdk_vhost_user_get_status() seems like a good place, indeed. I'll update the documentation accordingly. -- David Marchand _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
