On Sat, 18 Jul 2009 10:59:17 +0100, Mark Brown wrote:
> On Sat, Jul 18, 2009 at 11:17:12AM +0200, Jean Delvare wrote:
>
> > > +WM8350_NAMED_VOLTAGE(0, USB);
> > > +WM8350_NAMED_VOLTAGE(1, BATT);
> > > +WM8350_NAMED_VOLTAGE(2, LINE);
>
> > I would suggest passing the whole WM8350_AUXADC_USB, etc. string as the
> > 2nd parameter. This makes grepping the source code for users (as LXR
> > does for example) possible.
>
> The reason it's formatted like that is that the AUXADC input names match
> the names of the supplies that are being monitored so the second parameter
> is also used to provide a label to userspace.
This isn't how I read the code. I see a look-up table for labels, so the
macro implementation is just a detail which shouldn't matter.
> > > +static struct platform_driver wm8350_hwmon_driver = {
> > > + .probe = wm8350_hwmon_probe,
> > > + .remove = __devexit_p(wm8350_hwmon_remove),
> > > + .driver = {
> > > + .name = "wm8350-hwmon",
>
> > I think you are supposed to set .owner to THIS_MODULE?
>
> Hrm, yeah. I need to figure out where I cut'n'pasted this from since I
> suspect the error came along with that.
Maybe from an i2c driver, where the owner is set at run-time. Back when
this was implemented, I did object that it would cause confusion
because it would make different subsystems have different
requirements... Looks like I was right.
> > When you are done, what is your merge plan? I would be happy to take
> > the patch in my unofficial hwmon tree and push it in 2.6.32, but due to
> > wm8350-core dependencies, you may prefer it to be merged differently?
>
> I'm happy to go with whatever is easiest for the people actually doing
> the merging. wm8350 is fairly static at the minute - it's much more
> likely that there would be merge issues from the build infrastructure.
> It might be a little easier to merge via mfd.
Fine with me.
> I'll fix the rest of your comments, thanks.
OK.
--
Jean Delvare
