Re: [RFC][PATCH] Power domains for platform bus type

[Grant Likely <grant.likely@secretlab.ca>]

On Mon, Jan 31, 2011 at 03:16:15PM -0800, Kevin Hilman wrote:
> Hi Rafael,
> 
> "Rafael J. Wysocki" <rjw@sisk.pl> writes:
> 
> > Hi,
> >
> > This is something we discussed during the last Linux Plumbers Conference.
> >
> > The problem appears to be that the same device may be used in different
> > systems in different configurations such that actions necessary for the
> > device's power management can vary from one system to another.  In those
> > cases the drivers' power management callbacks are generally not sufficient,
> > because they can't take the configuration of the whole system into account.
> >
> > I think this issue may be addressed by adding objects that will represent
> > power domains and will provide power management callbacks to be executed
> > in addition to the device driver's PM callbacks, which is done by the patch
> > below.
> >
> > Please have a look at it and tell me what you think.
> 
> Very nice.  I like the approach and the fact that it allows grouping of
> only devices we care about customizing, instead of every device on the
> platform_bus (I know Grant will like that part too.  :)

:-D

> I experimented with something similar before I took the easy way out
> with platform_bus_set_pm_ops() :/
> 
> This approach might also solve the problem(s) we've been discussing
> around the conflicts between runtime PM callbacks and system PM
> callbacks (that RTC vs. i2c issue.)  With this approach, I shouldn't
> have to to add the callbacks like I did for the i2c driver, but rather
> handle them in common code.  I'll experiment with this...
> 
> The primary question for me is whether this should rather be at the
> 'struct device' level instead of at the platform_device level.  While
> we're currently only using this customization on platform_devices, I
> don't think we should limit it to that.  Part of these discussions at
> LPC was also about whether or not to move to using custom SoC-specific
> devices and busses.  If/when we go that route, we'd want these power
> domains part of struct device, not platform_device.

Definitely should be at the struct device level IMNSHO.