On Sat, May 03, 2008 at 11:41:16AM -0700, ron minnich wrote:
The geode is a very odd beast
Yes.
It would be better if we do not restrict ourselves unnecessarily.
I don't think we'll go all the way on this anytime soon. :\
Which really means the PIR table must die! die! die! die!
I would like to specify all about interrupts in the dts. I still don't know what we need in the syntax..
dev_pci_find_device?
I don't see the problem.
The point is that the device tree model should allow code to be associated with the right device, so that no find_device() calls are needed.
deep-rooted IRQ problems in the LX port, even down into the VSA,
VSA is very much involved at least in setting up interrupts so I suspect anything about interrupts on Geode has to at least be seen by VSA so it can update the neccessary MSRs.
Until the Geode architecture is supported in Linux we have to deal with the PCI emulation. Maybe time to look into OpenVSA a bit more?
//Peter