Re: [Xen-devel] [PATCH for-2.10 1/2] hw/acpi: Call acpi_set_pci_info when no ACPI tables needed

[Anthony PERARD <anthony.perard@xxxxxxxxxx>]

On Mon, Aug 14, 2017 at 06:53:14PM +0300, Michael S. Tsirkin wrote:
> On Mon, Aug 14, 2017 at 03:55:50PM +0100, Anthony PERARD wrote:
> > On Fri, Aug 11, 2017 at 08:18:28PM +0300, Michael S. Tsirkin wrote:
> > > On Fri, Aug 11, 2017 at 04:11:37PM +0100, Anthony PERARD wrote:
> > > > To do PCI passthrough with Xen, the property acpi-pcihp-bsel needs to be
> > > > set, but this was done only when ACPI tables are built which is not
> > > > needed for a Xen guest. The need for the property starts with commit
> > > > "pc: pcihp: avoid adding ACPI_PCIHP_PROP_BSEL twice"
> > > > (f0c9d64a68b776374ec4732424a3e27753ce37b6).
> > > > 
> > > > Set pci info before checking for the needs to build ACPI tables.
> > > > 
> > > > Reported-by: Sander Eikelenboom <linux@xxxxxxxxxxxxxx>
> > > > Tested-by: Sander Eikelenboom <linux@xxxxxxxxxxxxxx>
> > > > Signed-off-by: Anthony PERARD <anthony.perard@xxxxxxxxxx>
> > > 
> > > I am worried that Xen will come to depend on specific
> > > assignment of bsel which isn't guaranteed. Thoughts on
> > > how to avoid that?
> > 
> > Is it possible to have a different BSEL than 0 with PIIX ?
> 
> With PCI to PCI bridges, yes.
> 
> > Also, I don't known if having more than on PCI bus is going to work on
> > Xen, there is nothing in our ACPI tables beyond _SB.PCI0, and nothing to
> > use a different BSEL.
> 
> My worry is someone might decide to implement hotplug for pci to pci
> bridges on Xen. If doing that, it's important to use the qemu supplied
> acpi tables.

I can always add assert((xen_enable() && !bsel) || !xen_enable()) in
acpi_set_bsel(), so if someone was going to make any change, he would
find out about bsel quicker. But I don't see Xen using QEMU supplied
ACPI tables anytime soon.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel