Re: [PATCH 1/3] xen/pci: do not register devices outside of PCI segment scope

[Roger Pau Monné]

On Mon, Jan 13, 2025 at 08:17:23AM +0100, Jan Beulich wrote:
> On 10.01.2025 15:01, Roger Pau Monne wrote:
> > The PCI segment value is limited to 16 bits, however there are buses like VMD
> > that fake being part of the PCI topology by adding segment with a number
> > outside the scope of the PCI firmware specification range (>= 0x10000). The
> > MCFG ACPI Table "PCI Segment Group Number" field is defined as having a 16 bit
> > width.
> > 
> > Attempting to register or manage those devices with Xen would result in errors
> > at best, or overlaps with existing devices living on the truncated equivalent
> > segment values.
> > 
> > Skip notifying Xen about those devices.
> 
> Hmm, is simply omitting the notification really all it takes? How would Xen
> manage MSI / MSI-X, for example, without knowing of the device? As per the
> BoF on the summit in Prague(?), I continue to think we need partial driver
> logic in Xen for VMD ...

The basic mode of operation of devices behind a VMD bridge is to
reference the interrupts of the bridge device in its MSI(-X) entries,
so the VMD bridge acts as a de-multiplexer and forwards the interrupts
to the device behind the VMD bridge.  See vmd_alloc_irqs() (and
calling context) in the VMD driver for a reference about how this is
setup and operates.  Note also that devices behind a VMD bridge
operate using a different MSI domain, that uses a custom
irq_compose_msi_msg hook.

Thanks, Roger.