Re: [Qemu-devel] [PATCH 01/15] pc: merge pc_piix.c into pc.c

[Anthony Liguori]

On 01/27/2012 08:03 AM, Andreas Färber wrote:
> Am 27.01.2012 14:07, schrieb Anthony Liguori:
> > [...] My guess is that a SuperIO
> > chip could be an ISADevice and that we could simply make the PIIX3 has-a
> > SuperIO.  Then the ISAPC would have a trivial ISA chipset that has-a
> > SuperIO.
>
> That sounds pretty much like our construction site for PReP...

Yup.  It basically boils down to:

Root Complex[1] (PCI Host)
 - Northbridge (Memory Controller)
 - Southbridge (Super I/O chip)

Most platforms will follow this type of composition model with peripheral 
devices hanging off a bus in the Southbridge or directly from the Root Complex.

> Would you say that the SuperIO is-a ISADevice

The SuperIO device is-a Root Complex device.  With the I440FX, the Root Complex 
was PCI bus so the SuperIO chip (the PIIX3) is-a PCIDevice.  In older PCIs, the 
Root Complex was ISA (sort of).

[1] A better name for this is System Bus but I'm using Root Complex to avoid 
confusion with sysbus.

But for the PC, we can make the SuperIO chip be is-a DeviceState and just have 
it expose a MemoryRegion and a bunch of IRQs.  That would trivialize the 
implementation of an PIIX3 such that it has-a PCSuperIO and then just routes 
IRQs appropriately.

> and has-a ISADevice or
> would you want to remodel all ISADevices associated with a Super I/O
> chipset as private devices to mess with their internals without the
> whole enable/disable, etc. ugliness we ran into?
>
> I somewhat doubt that we can find a generic "SuperIO" base class btw.

Right, it's unlikely that a PC SuperIO chip would be useful outside of a PC. 
But you may find certain classes of platforms all have a common SuperI/O chip 
and can model similar things.

Regards,

Anthony Liguori

> Andreas