Re: [Qemu-devel] [PATCH v5] vhost-user: add multi queue support

[Martin Kletzander <mkletzan@redhat.com>]

[-- Attachment #1: Type: text/plain, Size: 2979 bytes --]

On Thu, Jul 09, 2015 at 10:06:35AM +0300, Michael S. Tsirkin wrote:
>On Thu, Jul 09, 2015 at 12:00:59AM +0200, Maxime Leroy wrote:
>> Hi Michael,
>>
>> On Wed, Jul 8, 2015 at 4:29 PM, Michael S. Tsirkin <mst@redhat.com> wrote:
>> > On Thu, May 28, 2015 at 09:23:06AM +0800, Ouyang Changchun wrote:
>> >> Based on patch by Nikolay Nikolaev:
>> >> Vhost-user will implement the multi queue support in a similar way
>> >> to what vhost already has - a separate thread for each queue.
>> >> To enable the multi queue functionality - a new command line parameter
>> >> "queues" is introduced for the vhost-user netdev.
>> >>
>> >> Signed-off-by: Nikolay Nikolaev <n.nikolaev@virtualopensystems.com>
>> >> Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
>> >
>> > So testing turned up a significant issue with the protocol extension in this
>> > one.  Specifically, remote has no idea how many queues guest actually
>> > wants to use (it's dynamic, guest changes this at any time).
>> > We need support for enabling and disabling queues dynamically.
>> >
>> > Given we are past hard freeze, and given no one uses this yet
>> > (dpdk upstream did not merge supporting protocol),
>> > I think the best thing to do is to disable this functionality for 2.4.
>> > I will send a patch to do this shortly.
>>
>> You are making a wrong statement, we already use multiqueue for
>> vhost-user and we expected to have this support officially integrated
>> in qemu 2.4.
>>
>> Libvirt 1.2.17 has been released with multiqueue support for
>> vhost-user. (http://libvirt.org/git/?p=libvirt.git;a=commit;h=366c22f2bcf1ddb8253c123f93fd18d1ba9eacd6)
>> It checks against the version of qemu (i.e. 2.4)  to know if
>> multiqueue is supported or not by qemu.
>> (http://libvirt.org/git/?p=libvirt.git;a=commitdiff;h=7971723b985b9adc27122a3503e7ab38ced2b57f;hp=e7f5510ef2d28ca0ae0ed5751b1fd3018130d6c1)
>
>Ouch. Just another example showing how version checks are evil.
>We don't want to break libvirt, I agree.
>

Yes, exactly.  Unfortunately if QEMU doesn't expose it in any way we
don't have many more options.

>I think what we can do is accept the command line created
>by libvirt, just ignore it and use a single queue only.
>

Anyway, I think it would be pretty OK to disable it *if and only if*
you error out with a sensible error message (e.g. "multiple queues are
not supported yet").

And from 2.5 and next libvirt release we can fix this properly
(QEMU - exposing the capability and libvirt - checking for it).

>> Dynamically enabling/disabling queue between host/guest is a nice
>> feature to have.
>> But it's not mandatory.
>
>It's mandated by the virtio spec.
>
>> You can still configure manually guest and
>> host to use the same number of queues.
>>
>> I think this feature can be implemented later in qemu 2.5.
>>
>> Regards,
>>
>> Maxime
>
>Sorry, that's not how virtio is supposed to work. It must downgrade
>gracefully, not just stop working or crash guest.
>
>--
>MST

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]