Re: [RFC][PATCH 2/6] PCI PM: capability probing and setup

[Greg KH <gregkh@suse.de>]

On Wed, Nov 16, 2005 at 02:21:01AM -0500, Adam Belay wrote:
> On Tue, 2005-11-15 at 22:21 -0800, Greg KH wrote:
> > On Tue, Nov 15, 2005 at 10:31:17PM -0500, Adam Belay wrote:
> > > +int pci_setup_device_pm(struct pci_dev *dev)
> > 
> > Care to give kernel doc for this new function?
> 
> Absolutely.  I was planing to do this but must have forgotten.
> 
> > > +	unsigned char	state_mask;	/* a mask of supported power states */
> > > +	unsigned char	pme_mask;	/* a mask of power states that allow #PME */ 
> > 
> > Trailing space, use quilt it strips this :)
> 
> Sorry about that :)
> 
> > 
> > > +	struct pci_dev_pm *pm;		/* power management information */
> > 
> > Why make this a pointer and not just part of this structure?  Don't all
> > pci devices need this?
> 
> Actually, not every PCI device supports the PCI PM spec.  There are many
> devices, even in modern systems, that can only be in D0.  I was thinking
> we could save some memory and allocate this structure when PCI PM is
> detected.  Would that be ok?

That would be ok, but you better remember to check for the pointer when
doing things like suspend and resume of the config space.

thanks,

greg k-h