Re: [virtio] [PATCH v2 06/17] charter: drop attempts to list all devices

[Halil Pasic <pasic@linux.ibm.com>]

On Tue, 9 Feb 2021 05:02:10 -0500
"Michael S. Tsirkin" <mst@redhat.com> wrote:

> On Mon, Feb 08, 2021 at 05:15:55PM +0100, Halil Pasic wrote:
> > On Mon, 8 Feb 2021 07:15:57 -0500
> > "Michael S. Tsirkin" <mst@redhat.com> wrote:
> >   
> > > we are adding new device types all the time.
> > > no point in trying to list them all in the charter.
> > > 
> > > Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> > > Reviewed-by: Cornelia Huck <cohuck@redhat.com>
> > > ---
> > >  charter.html | 17 +----------------
> > >  1 file changed, 1 insertion(+), 16 deletions(-)
> > > 
> > > diff --git a/charter.html b/charter.html
> > > index 38cad98..82fb981 100644
> > > --- a/charter.html
> > > +++ b/charter.html
> > > @@ -80,22 +80,7 @@
> > >            Specification of device-specific configuration.
> > >            <ol class="vanilla-lower-alpha">
> > >              <li>
> > > -              For network devices.
> > > -            </li>
> > > -            <li>
> > > -              For block storage devices.
> > > -            </li>
> > > -            <li>
> > > -              For entropy devices.
> > > -            </li>
> > > -            <li>
> > > -              For console devices.
> > > -            </li>
> > > -            <li>
> > > -              For memory ballooning devices.
> > > -            </li>
> > > -            <li>
> > > -              For SCSI endpoint devices.
> > > +              Normative requirements for specified device types.
> > >              </li>  
> > 
> > Does it make sense to keep the list with only a single element?
> >   
> > >              <li>
> > >                Non-normative advice for designing new device types.  
> > 
> > I believe the the rest of the items starting with the one immediately
> > above should be moved up one level higher. These don't seem to be sub
> > deliverables of 'device-specific configuration' but top level
> > deliverables...  
> 
> Well if you look at the spec they are all under section
> 5 Device Types. The only exception is
> Appendix B. Creating New Device Types.
> I will move that one out.
>

What confuses my is the configuration word in the phrase
'device-specific configuration'. I assumed that is about the config
space, but it seems your intention for deliverable 2 is to cover
everything that is device type specific (config space layout, virtqueues
and their purpose, the communication protocol between driver and device
(message formats, etc)).
 
I had something like this in mind (please compare to the attachment):

4 Deliverables

    1 Specification of feature negotiation, configuration and queues, from both driver and device points of view.
        1 Specifically for virtio over PCI.
        2 Specifically for virtio over mmio.
        3 Specifically for virtio over channel I/O.
        4 Specifically for other transports as decided by TC.
    2 Normative requirements for specified device types.
    3 Non-normative advice for designing new device types.
    4 List of reserved device numbers for non-specified device types. (This section is likely to see ongoing maintenance).
    5 List of reserved feature numbers for non-specified features for each device type. (This section is likely to see ongoing maintenance).
    6 Non-normative code examples for operation of driver/device side of buffers.
    7 Non-normative guide for creating devices and drivers which also support a legacy mode.


>

Attachment: Screen Shot 2021-02-09 at 12.26.32.png
Description: PNG image