Re: [PATCH] PCI MMConfig: Detect and support the E7520 and the 945G/GZ/P/PL

[Matthew Wilcox <matthew@wil.cx>]

On Thu, Nov 23, 2006 at 08:57:41PM +0100, Arjan van de Ven wrote:
> On Thu, 2006-11-23 at 20:51 +0100, Olivier Galibert wrote:
> > It seems that the only way to reliably support mmconfig in the
> > presence of funky biosen is to detect the hostbridge and read where
> > the window is mapped from its registers.  Do that for the E7520 and
> > the 945G/GZ/P/PL on x86-64 for a start.
> 
> while I like this approach a lot, I am wondering if this shouldn't be
> done as a PCI quirk instead.... it would make a lot of sense to use that
> shared infrastructure for this...

Except the shared infrastructure kind of relies on having the pci
accesses already working ... unless we want the kernel to print messages
like:

PCI: Using type 1 config access method
...
PCI: Whoops, looks like we're going to use MMCONFIG after all.  Ignore that