On Thu, 2005-01-27 at 02:17 +0000, Kentaro A. Kurahone wrote:
> On Wed, Jan 26, 2005 at 10:06:12PM +0000, Kentaro A. Kurahone wrote:
> > On Wed, Jan 26, 2005 at 05:56:38PM -0400, Jared D. McNeill wrote:
> > > On Wed, 2005-01-26 at 20:34 +0000, Kentaro A. Kurahone wrote:
> > > > Looks good.  I have cbb, auich, agp_i810, and pms fixed up.  I'm going to
> > > > rework those to use pci_state_saveoff/restore.
> > > 
> > > I've commited my changes (modified as per Matt Thomas's suggestions), so
> > > you might want to update and use pci_conf_capture/restore along with
> > > struct pci_conf_state.
> > 
> > Ok, will do.
> 
> Done.  Haven't sysctlized the VBIOS stuff yet, but I think it might be wiser
> to hold off on that till the "new and improved" powermanagement API stuff
> stabilises.

I may have a solution that will fix it so we don't need to initialize
the VBIOS in the wakecode at all.. going to take me a bit of time to
figure out how I want it done, but basically I'm running into some
problems. With the laptop in a docking station, S3 sleep/resume works
like a charm. Without a docking station, the video fails to recover on a
resume. Resetting the video BIOS in the wakecode causes my laptop to
reboot on resume.

It seems that I need to defer my video BIOS initialization until _after_
1. the PCI chipset is reinitialized, and 2. pci_conf_state is restored
to vga_pci. I plan on adding an MD powerhook to drop back to real mode
after the standard powerhooks run so I can reinitialize the display
adapter then. Stay tuned for a patch for this.

In the meantime, I'll have a look at your patch. Thanks for the great
work!

Cheers,
Jared