Re: [Qemu-devel] [RFC v2 0/3] virtio-net: Add support to MTU feature

[Michael S. Tsirkin]

On Wed, Nov 23, 2016 at 09:02:53AM -0500, Aaron Conole wrote:
> "Michael S. Tsirkin" <address@hidden> writes:
> 
> > On Wed, Nov 23, 2016 at 11:42:52AM +0800, Jason Wang wrote:
> >> > > > > > > >  Seems to me like an easy way to get out of sync.
> >> > > > > >
> >> > > > > >If we send it to the backend, that has a chance to check
> >> > > > > >mtu and disconnect on error.
> >> > > >
> >> > > >For vhost-user backend, we can send it the MTU value with a
> >> > > >vhost-user protocol feature.
> >> > > >
> >> > > >For tun/macvtap, how do you do without adding a new ioctl ?
> >> > Have management configure same mtu on the backend and in qemu.
> >> > 
> >> > 
> >> 
> >> Then why not do same for vhost-user (instead of using two different
> >> methods)?
> >
> > That's what I'm saying. If backend supports that, we can also
> > check the mtu in some way to make sure it matches.
> 
> I'm not sure why we need a new ioctl (or an ioctl at all - netlink
> supports all of this)?
> 
> ex:
> 
> 08:58:34 aconole {fast-datapath-beta-rhel-7} ~/rhpkg/openvswitch$ sudo ip 
> tuntap add dev tap0 mode tap
> [sudo] password for aconole: 
> 08:58:40 aconole {fast-datapath-beta-rhel-7} ~/rhpkg/openvswitch$ ip l
> 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode 
> DEFAULT group default qlen 1000
>     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
> ...
> 7: tap0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT 
> group default qlen 1000
>     link/ether 46:e0:fc:83:54:1c brd ff:ff:ff:ff:ff:ff
> 08:58:51 aconole {fast-datapath-beta-rhel-7} ~/rhpkg/openvswitch$ sudo ip l 
> set tap0 mtu 8000
> 08:58:54 aconole {fast-datapath-beta-rhel-7} ~/rhpkg/openvswitch$ ip l
> 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode 
> DEFAULT group default qlen 1000
>     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
> ...
> 7: tap0: <BROADCAST,MULTICAST> mtu 8000 qdisc noop state DOWN mode DEFAULT 
> group default qlen 1000
>     link/ether 46:e0:fc:83:54:1c brd ff:ff:ff:ff:ff:ff
> 
> So, at least with iproute2, we can already read and write using the netlink
> interface for tuntap devices.  I haven't played with macvtap, but I
> think it's similar support - just do a netlink query, get the configured
> MTU, and advertise it.  I might be missing something though - I'm a
> simple guy with simple ideas.  Maybe there's a cross-platform issue or
> something?
> 
> -Aaron

qemu is generally not running with enough priveledges to
allow access to netlink.