[PATCH v3 1/3] mfd: mc13xxx: add device tree probe support
shawn.guo at freescale.com
Tue Dec 20 08:52:52 EST 2011
On Tue, Dec 20, 2011 at 11:25:11AM +0000, Mark Brown wrote:
> On Tue, Dec 20, 2011 at 11:03:48AM +0800, Shawn Guo wrote:
> > On Tue, Dec 20, 2011 at 01:59:32AM +0000, Mark Brown wrote:
> > > You're missing the point - someone reading the documentation needs to be
> > > able to figure out what strings they need to use for all the different
> > > regulators on the chip are without groveling through the driver code.
> > > That's *not* something that we should be exposing in the device tree
> > > bindings. This is an implementation detail of the Linux driver which
> > > might well change in future.
> > Any suggestion on a better binding for mc13892 regulator device?
> Well, removing the random extra _s would be a big start (though I'd just
> drop the chip name entirely from the name of the regulators since by the
> time we're looking at the regulator we've already identified the chip)
> and as I keep saying you need to document what the names mean - what are
> the possible names and how do they map onto the hardware?
I just came up with an idea which can totally avoid matching name. It
seems that we can identify a regulator using register plus enable bit,
which is basically 'reg' and 'enable_bit' in 'mc13xxx_regulator'. As
these data must be coming from hardware manual, they should be stable
enough for binding a regulator. What do you think?
More information about the linux-arm-kernel