On Wed, Jan 26, 2005 at 12:26:10AM +0900, Takayoshi Kochi wrote:
[snip]
> IMO, powerhook interface should also be reconsiderd for supporting
> suspend or software suspend (hibernation).  Just using the current
> interface will never improve maintenanceability in the long run.
> So I think the patch isn't mergeable to the tree.

Understood.  I would have thought that the powerhook interface would
be "sufficient", either by adding another pair of states, or shoehorning
the save/restore code into the PWR_SUSPEND/PWR_RESUME locations for each
driver, but a more elegant solution would be ideal.

Unfortunately a lot of the device drivers were not coded with code to
transition from D3 to D0 state, so it's a tedious cycle of "fix one
device, reboot, find the next device that's busted".

I'm going to port over pci_cfg_save/pci_cfg_restore from FreeBSD,
since it's becoming painfully aparent as to why they have that.

> With that said, if you can work on this area, it's very welcome!

I'll be more than happy to, especially since my laptop won't suspend until
I get this working. ;)

-- 
Kentaro A. Kurahone
SIGUSR1 Research and Development