Re: [PATCH] content: Replace guest OS with driver

["Michael S. Tsirkin" <mst@redhat.com>]

On Tue, May 16, 2023 at 09:31:43PM +0000, Parav Pandit wrote:
> 
> > From: Michael S. Tsirkin <mst@redhat.com>
> > Sent: Tuesday, May 16, 2023 5:25 PM
> 
> > > Flavor of a RHEL has inbuilt.
> > 
> > What does this mean exactly? That virtio core is compiled into kernel?
> > That does not matter at all.
> > 
> Why does it does not matter? What is the definition of driver, is the device_structure structure or module binary?

Neither. It's the part that handles transport or specific device.




> > > > > Does driver only matter with device_driver structure or module binary?...
> > > >
> > > > Can't parse your question.
> > > >
> > > > > Driver is largely the software entity that drives the device.
> > > > > I think we can keep the spec simple enough to not mix these
> > > > > details and just
> > > > call it a "driver".
> > > >
> > > > Not just linux there are lots of drivers like this.  the two bits
> > > > pass useful information the way you changed it this distinction is lots.
> > > > I agree it is worth thinking what exactly does it mean.
> > > > Since you researched it - what exactly do drivers such as uefi and
> > > > the unnamed "some OS variant" do exactly?
> > > There is just one "driver" virtio_net_pci that has sets the required bits.
> > >
> > > > when do they set ACKNOWLEDGE and when DRIVER?
> > > >
> > > Not any different flow.
> > > Entity is one.
> > 
> > question is, what happens
> > - before ACKNOWLEDGE
> > - after ACKNOWLEDGE before DRIVER
> > ?
> It follows the same sequence described in the spec as "driver" not as "guest OS".


Looks like my description is spot on then - check vendor ID,
set ACKNOWLEDGE, check device ID, see that it's a known good value
matching device type we expect, set DRIVER.

Maybe replace with "transport part of the driver" and
"device type specific part of the driver"? A bit verbose but if
it rocks your boat ...

-- 
MST