Re: pci bridge suspend/resume (Re: PCI not restored correctly after suspend to ram)

[Greg KH <greg-U8xfFu+wG4EAvxtiuMwx3w@public.gmane.org>]

On Mon, Mar 22, 2004 at 10:34:25PM +0000, Adam Belay wrote:
> > Yes, this will happen in 2.7 when we add the ability to have "generic"
> > and "specific" type drivers for devices.
> 
> What do you think would be the best way to go about this?
> 
> I was able to come up with the following options:
> 
> 1.) flags in struct device_driver

I think this is the best place for it.  We will have to do something
like weight the different ways to specify a driver can handle a device
differently.  Something like this kind of priority:
	- device id
	- device class

So if a driver matches a device by the id, it should be bound to it,
instead of a driver that can bind to id by the "class".

> 2.) return codes from (*match)
> 3.) a seperate bus specific function to evaluate driver priority

I think this will be necessary, as every bus will have different ways to
specify priority.

> Also how can we prevent the order in which drivers are loaded from
> determining which driver is bound?  It seems that we would need to
> delay device matching.

Or we can disconnect a driver from a device when a driver with a higher
"weight" comes into the system.

But I don't really know, I haven't given it all that much thought yet :)

thanks,

greg k-h


-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click