On Fri, Oct 30, 2015 at 06:48:09PM +0100, Thomas Monjalon wrote:
> 2015-10-18 10:04, Michael S. Tsirkin:
> > On Fri, Oct 16, 2015 at 02:52:30PM +0100, Bruce Richardson wrote:
> > > On Thu, Oct 15, 2015 at 04:18:59PM +0300, Michael S. Tsirkin wrote:
> > > > On Thu, Oct 15, 2015 at 02:08:39PM +0300, Marcel Apfelbaum wrote:
> > > > > Make vhost-user virtio 1.0 compatible by adding it to the
> > > > > supported features and keeping the header length
> > > > > the same as for mergeable RX buffers.
> > > > >
> > > > > Signed-off-by: Marcel Apfelbaum <marcel at redhat.com>
> > > >
> > > > Looks good to me
> > > >
> > > > Acked-by: Michael S. Tsirkin <mst at redhat.com>
> > > >
> > > > Just one question: dpdk is only supported on little-endian
> > > > platforms at the moment, right?
> > >
> > > A recent release added in support for PPC (patches supplied by IBM). For
> > > example, see:
> > > http://dpdk.org/browse/dpdk/commit/?id=704ba3770032c5a901719d3837845581d5a56b58
> > >
> > > /Bruce
> >
> > This will require more work then as 1.0 is a different
> > endian-ness from 0.9. It's up to you guys to decide
> > whether correct BE support is now a requirement for all
> > new dpdk code. Let us know.
>
> I'm not sure to understand.
> Yes DPDK must work on big endian platforms.
> Does this patch prevent from using virtio 0.9 with big endian?
No it doesn't. virtio 0.9 with big endian most likely does not work
reliably with vhost-user at the moment since qemu does not tell vhost-user
about guest endian-ness (though it's common to have it match host,
then it will work by luck).
> Does it work with old QEMU not supporting virtio 1.0?
Yes it does.
virtio 1 requires a bunch of fields to be little endian.
If someone uses this patch with new QEMU on BE machine,
things won't work: you need
if (virtio_1)
cpu_to_le
On LE, cpu_to_le is a NOP so things work by luck.
In other words this patch is fine, more is needed to make virtio 1
work on BE hosts.
--
MST
