RE: [RFC] Regulator state after regulator_get

[kalle.jokiniemi]

Hi,

 > -----Original Message-----
 > From: ext Mark Brown [mailto:broonie@xxxxxxxxxxxxxxxxxxxxxxxxxxx]
 > Sent: 28. huhtikuuta 2011 13:06
 > To: Jokiniemi Kalle (Nokia-SD/Tampere)
 > Cc: lrg@xxxxxxxxxxxxxxx; mchehab@xxxxxxxxxxxxx; svarbatov@xxxxxxxxxx;
 > saaguirre@xxxxxx; grosikopulos@xxxxxxxxxx; Zutshi Vimarsh (Nokia-
 > SD/Helsinki); Ailus Sakari (Nokia-SD/Helsinki); linux-kernel@xxxxxxxxxxxxxxx;
 > linux-media@xxxxxxxxxxxxxxx
 > Subject: Re: [RFC] Regulator state after regulator_get
 > 
 > On Thu, Apr 28, 2011 at 09:01:03AM +0000, kalle.jokiniemi@xxxxxxxxx wrote:
 > 
 > > If the device driver using the regulator does not enable and disable the
 > > regulator after regulator_get, the regulator is left in the state that it was
 > > after bootloader. In case of N900 this is a problem as the regulator is left
 > > on to leak current. Of course there is the option to let regulator FW disable
 > > all unused regulators, but this will break the N900 functionality, as the
 > > regulator handling is not in place for many drivers.
 > 
 > You should use regulator_full_constraints() if your board has a fully
 > described set of regulators.  This will cause the framework to power
 > down any regulators which aren't in use after init has completed.  If
 > you have some regulators with no consumers or missing consumers you need
 > to mark them as always_on in their constraints.

I don't have a full set of regulators described, that's why things broke when
I tried the regulator_full_constraints call earlier. But I don't think it would be too
big issue to check the current after boot configuration and define all the
regulators as you suggest. I will try this approach.

 > 
 > > So reset the regulator on first regulator_get call to make
 > > sure that any regulator that has users is not left active
 > > needlessly.
 > 
 > This would cause lots of breakage, it would mean that all regulators
 > that aren't always_on would get bounced off at least once during startup
 > - that's not going to be great for things like the backlight.

OK, this is not a viable solution.

Thanks for the comments,
Kalle

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/