Re: [virtio-comment] Re: [PATCH 0/3] Rename queue index to queue number

[Cornelia Huck <cohuck@redhat.com>]

On Wed, Mar 01 2023, "Michael S. Tsirkin" <mst@redhat.com> wrote:

> On Wed, Mar 01, 2023 at 06:22:07PM +0100, Halil Pasic wrote:
>> On Mon, 27 Feb 2023 09:45:31 +0100
>> Cornelia Huck <cohuck@redhat.com> wrote:
>> 
>> > For the vq index/number, I'm not that sure that "virtqueue number" is
>> > better that "virtqueue index" -- actually, I'd prefer the latter. We'd
>> > need some renaming either way.
>> 
>> I prefer index as well. Especially that we start indexing with 0. Also
>> seems to be the more common term for such stuff in both Mathematics and
>> CS.
>> 
>> Regards,
>> Halil 
>
>
> Basically I am saying that this:
>
> /* Queue size for the currently selected queue - Write Only */
> #define VIRTIO_MMIO_QUEUE_NUM           0x038
>
> is a bad name because queue number seems to be ambiguous.
>
> Maybe start with just getting rid of uses of QUEUE_NUM meaning size?

Nod, that seems to be uncontroversial.

>
>
> If we want to use queue index we need to fix RSS spec in networking
> since that seems to want to use queue index in a very weird way:
> networking has this idea of calling queues like this:
> receiveq1 receiveq2 ....
> why 1-based? I guess we wanted this to be clear even to a 5 year olds ;)
>
> And then for extra fun, in the RSS section we say "0 based index" where we
> seem to mean "this number in the queue name, but subtract 1 in your head".
> Why subtract 1? I guess we wanted these 5 year olds to practice math ...

Heh :)

Yeah, that looks like a mess... I don't think we should use a weird
substraction scheme. I haven't looked at the RSS stuff much, would it be
hard to fix it up?