Re: [patch] pci: revert "PCI: remove transparent bridge sizing"

[Benjamin Herrenschmidt <benh@kernel.crashing.org>]

On Wed, 2008-03-26 at 22:07 +0000, Alan Cox wrote:
> > PCI bridges at zero is perfectly valid indeed and I'm sure we have that
> > around at least for IO space. In fact, I'm surprised you don't have that
> > on x86. Typically, things like an HT segment with a P2P bridge and
> > behind that bridge an ISA bridge could well have the P2P bridge with a
> > resource forwarding 0...0x1000 IO downstream for example even on x86
> > no ? (I'm not -that- familiar with the crazyness of legacy ISA on x86
> > but I've definitely seen such setup on other archs).
> 
> On a PC system 0x00-0xFF are motherboard resources (sometimes chipset,
> some even swallowed by the CPU in certain cases) so 0 as disabled is sort
> of safe but as shown by the pci_enable_device_bars replacement code - not
> a good idea neccessarily.
> 
> A lot of driver code does assume 0 == unavailable/off/disabled including
> large chunks of serial, ata, ide and probably other subsystems.

True, and we do try to avoid allocating resources at 0 most of the time,
but it happens, and in the case of bridges, it can make somewhat sense
(more in fact than for devices I'd say).

Cheers,
Ben.