Re: [PATCH 1/2] pci id: clarify pci subsystem id usage

["Daniel P. Berrangé" <berrange@redhat.com>]

On Thu, Oct 15, 2026 at 12:20:59PM +0200, Gerd Hoffmann wrote:
> On Thu, Oct 15, 2026 at 09:45:45AM +0100, Daniel P. Berrangé wrote:
> > >  1af4:1100
> > > -  Used as PCI Subsystem ID for existing hardware devices emulated
> > > -  by QEMU.
> > > +  Used as PCI Subsystem ID for existing hardware devices emulated by
> > > +  QEMU, such as the intel chipset devices.  Also used for virtio
> > > +  devices and generic pci class devices (1b36:00xx, see below).
> > 
> > When talking virtio devices, does that include virtio devices
> > implemented outside QEMU via vhostuser, or only in-QEMU virtio
> > devices ?
> 
> For vhost devices the pci config space emulation is still handled by
> qemu.  On a quick check I don't see anything special done for vhost,
> so I expect the same subsystem ID being used.

So IIUC, the QEMU subsystem ID applies to every PCI device exposed
from QEMU, except for VFIO passthrough of host devices.

> 
> Maybe we should extend the vhost protocol to allow overriding the
> subsystem id, so linux kernel / rust-vmm  / ... can have their own
> subsystem ids?

Would be a good idea as a way to distinguish the QEMU impls for
virtio devices, from 3rd party ones accessed via vhostuser.

With regards,
Daniel
-- 
|: https://berrange.com       ~~        https://hachyderm.io/@berrange :|
|: https://libvirt.org          ~~          https://entangle-photo.org :|
|: https://pixelfed.art/berrange   ~~    https://fstop138.berrange.com :|