Re: [PATCH v3] vfio-pci: Provide reviewers and acceptance criteria for vendor drivers

[Alex Williamson <alex.williamson@redhat.com>]

On Tue, 15 Mar 2022 12:53:04 -0300
Jason Gunthorpe <jgg@nvidia.com> wrote:

> On Tue, Mar 15, 2022 at 10:26:17AM +0100, Cornelia Huck wrote:
> > On Mon, Mar 14 2022, Alex Williamson <alex.williamson@redhat.com> wrote:
> >   
> > > Vendor or device specific extensions for devices exposed to userspace
> > > through the vfio-pci-core library open both new functionality and new
> > > risks.  Here we attempt to provided formalized requirements and
> > > expectations to ensure that future drivers both collaborate in their
> > > interaction with existing host drivers, as well as receive additional
> > > reviews from community members with experience in this area.
> > >
> > > Cc: Jason Gunthorpe <jgg@nvidia.com>
> > > Cc: Yishai Hadas <yishaih@nvidia.com>
> > > Cc: Kevin Tian <kevin.tian@intel.com>
> > > Acked-by: Shameer Kolothum <shameerali.kolothum.thodi@huawei.com>
> > > Signed-off-by: Alex Williamson <alex.williamson@redhat.com>  
> > 
> > (...)
> >   
> > > diff --git a/Documentation/driver-api/vfio-pci-vendor-driver-acceptance.rst b/Documentation/driver-api/vfio-pci-vendor-driver-acceptance.rst
> > > new file mode 100644
> > > index 000000000000..3a108d748681
> > > +++ b/Documentation/driver-api/vfio-pci-vendor-driver-acceptance.rst  
> > 
> > What about Christoph's request to drop the "vendor" name?
> > vfio-pci-device-specific-driver-acceptance.rst would match the actual
> > title of the document, and the only drawback I see is that it is a bit
> > longer.  
> 
> I agree we should not use the vendor name
> 
> In general I wonder if this is a bit too specific to PCI, really this
> is just review criteria for any driver making a struct vfio_device_ops
> implementation, and we have some specific guidance for migration here
> as well.
> 
> Like if IBM makes s390 migration drivers all of this applies just as
> well even though they are not PCI.

Are you volunteering to be a reviewer under drivers/vfio/?  Careful,
I'll add you ;)

What you're saying is true of course and it could be argued that this
sort of criteria is true for any new driver, I think the unique thing
here that raises it to a point where we want to formalize the breadth
of reviews is how significantly lower the bar is to create a device
specific driver now that we have a vfio-pci-core library.  Shameer's
stub driver is 100 LoC.  I also expect that the pool of people willing
to volunteer to be reviewers for PCI related device specific drivers is
large than we might see for arbitrary drivers.

> > > +New driver submissions are therefore requested to have approval via
> > > +Sign-off/Acked-by/etc for any interactions with parent drivers.  
> > 
> > s/Sign-off/Reviewed-by/ ?
> > 
> > I would not generally expect the reviewers listed to sign off on other
> > people's patches.  
> 
> It happens quite a lot when those people help write the patches too :)

This is what "etc" is for, the owners are involved and have endorsed it
in some way, that's all we care about.  Thanks,

Alex