In article <am1y3sv49c.fsf@nutcracker.stacken.kth.se>
lha@stacken.kth.se wrote:

> Should not the pci_set_powerstate() be use instead of tinkering with the
> powerstate register themself.

I agree that PMCSR code in all PCI drivers should be merged
in some MI place, but it will be done in the next stage.

> When the power state changes, do you really
> need to save parts of the configration state register like vr does ? If you
> need, shouldn't pci_set_powerstate() do that too then ?

Sorry, I'm not familiar with the PCI spec, but I think if you change
power state from D3 to D0, some registers should be restored
because it causes "D0 uninitialized state transition."
---
Izumi Tsutsui
tsutsui@ceres.dti.ne.jp