Re: [PATCH 8/9] vfio/pci: use x86 naming instead of igd

[Matthew Rosato]

On 2/11/21 10:47 AM, Max Gurtovoy wrote:
> On 2/2/2021 7:10 PM, Jason Gunthorpe wrote:
> > On Tue, Feb 02, 2021 at 05:06:59PM +0100, Cornelia Huck wrote:
> >
> > > On the other side, we have the zdev support, which both requires s390
> > > and applies to any pci device on s390.
> > Is there a reason why CONFIG_VFIO_PCI_ZDEV exists? Why not just always
> > return the s390 specific data in VFIO_DEVICE_GET_INFO if running on
> > s390?
> >
> > It would be like returning data from ACPI on other platforms.
>
> Agree.
>
> all agree that I remove it ?

I did some archives digging on the discussions around 
CONFIG_VFIO_PCI_ZDEV and whether we should/should not have a Kconfig 
switch around this; it was something that was carried over various 
attempts to get the zdev support upstream, but I can't really find (or 
think of) a compelling reason that a Kconfig switch must be kept for it. 
 The bottom line is if you're on s390, you really want zdev support.

So: I don't have an objection so long as the net result is that 
vfio_pci_zdev.o is always built in to vfio-pci(-core) for s390.

> we already have a check in the code:
>
> if (ret && ret != -ENODEV) {
>                                  pci_warn(vdev->vpdev.pdev, "Failed to 
> setup zPCI info capabilities\n");
>                                  return ret;
> }
>
> so in case its not zdev we should get -ENODEV and continue in the good 
> flow.
>
> > It really seems like part of vfio-pci-core
> >
> > Jason