Re: [RFC][PATCH] gpu:drm:i915:intel_detect_pch: back to check devfn instead of check class type

["Chen, Tiejun" <tiejun.chen@intel.com>]

On 2014/6/25 14:48, Paolo Bonzini wrote:
> Il 22/06/2014 10:25, Chen, Tiejun ha scritto:
>> In qemu-upstream, as you commented we can't create this as a ISA class
>> type explicitly.
>
> Note I didn't say that QEMU doesn't like having two ISA bridges.
>
> I commented that the firmware will see two ISA bridges and will try to
> initialize both of them.  It currently doesn't just because it only
> knows of two southbridge PCI IDs, and both of them are old or relatively
> old (PIIX3/4 and ICH9).
>
> But what would happen if you did graphics passthrough on a machine with
> an ICH9 southbridge?  Your code will create the PIIX4 ISA bridge, and
> will create an ICH9 southbridge just to please the i915 driver.  The
> BIOS will recognize the ICH9's vendor/device ids and try to initialize
> it.  It will write to the configuration space to set up PCI PIRQ[A-H]
> routing, for example, which makes no sense since your ICH9 is just a
> dummy device.
>

Thanks for your detailed explanation.

> Second problem.  Your IGD passthrough code currently works with QEMU's
> PIIX4-based machine.  But what happens if you try to extend it, so that

Yes, current xen machine, xenpv, is based on pii4, and also I don't 
known if we will plan to migrate to q35 or others. So its hard to 
further say more now.

> it works with QEMU's ICH9-based machine?  That's a more modern machine
> that has a PCIe chipset and hence has its ISA bridge at 00:1f.0.  Now

But even in this case, could we set the real vendor/device ids for that 
ISA bridge at 00:1f.0? If not, what's broken?

> you have a conflict.
>
> In other words, if you want IGD passthrough in QEMU, you need a solution
> that is future-proof.
>
>> So we compromise by faking this ISA bridge without ISA
>> class type setting (as I recall you already said this way is slightly
>> better).
>
> It is only slightly better, but the right solution is to fix the driver.
>   There is absolutely zero reason why a graphics driver should know
> about the vendor/device ids of the PCH.

This means we have to fix this both on Linux and Windows but I'm not 
sure if this is feasible to us.

>
> The right way could be to make QEMU add a vendor-specific capability to
> the video device. The driver can probe for that capability before

Do you mean we can pick two unused offsets in the configuration space of 
the video device as a vendor-specific capability to hold the 
vendor/device ids of the PCH?

> looking at the PCI bus.  QEMU can add the capability to the list, it is
> easy because all accesses to the video device's configuration space trap
> to QEMU.  Then you do not need to add fake devices to the machine.
>
> In fact, it would be nice if Intel added such a capability on the next
> generation of integrated graphics, too.  On real hardware, ACPI or some

Maybe, but even this would be implemented, shouldn't we need to be 
compatible with those old generations?

Thanks
Tiejun

> other kind of firmware can probe the PCH at 00:1f.0 and initialize that
> vendor-specific capability.  QEMU would just forward the value from the
> host, so that virtualized guests never depend on the PCH at 00:1f.0.
>
> Paolo
>
>> Maybe we will figure better way in the future. But anyway, this
>> is always registered as 00:15.0, right? So I think the i915 driver can
>> go back to probe the devfn like the native environment.
>
>
>