Prev: [git pull] fuse bug fix
Next: b44.c box lockup fix (netconsole): ratelimit NAPI poll error message
From: Eduardo Valentin on 30 Nov 2009 05:50
Hello Mark and Liam,

I'm writing to ask about VBAT use case. What is the expected
way to use regulator framework in case of rail coming from battery?
Should it be added to the regulator framework at all?

In that case, the rail should not be controllable. So I don't see
any reason to add it to the regulator framework board definitions,
as we should not be controlling it.

However, drivers for devices on that rail would require their regulator anyway.
And I guess the point would be that drivers should not be aware that they are on VBAT
or any other rail.

So, what's the correct way to solve this?

- Should drivers fail nicely if a regulator_get fail? And continue even if one fails.
- Should drivers disable regfw usage completely in the driver if regulator_get doesn't
give valid regulator ?
- or Should a fake fixed regulator be added for vbat so drivers can still get a valid
regulator with regulator_get.

The last options seams to be the one that does not require much changes on drivers.
But it will be adding a regulator that does basically nothing in the system.

BR,

--
Eduardo Valentin
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Eduardo Valentin on 30 Nov 2009 06:00
Forgot to add Liam into Cc. Doing it so.


On Mon, Nov 30, 2009 at 11:41:02AM +0100, Valentin Eduardo (Nokia-D/Helsinki) wrote:
> Hello Mark and Liam,
>
> I'm writing to ask about VBAT use case. What is the expected
> way to use regulator framework in case of rail coming from battery?
> Should it be added to the regulator framework at all?
>
> In that case, the rail should not be controllable. So I don't see
> any reason to add it to the regulator framework board definitions,
> as we should not be controlling it.
>
> However, drivers for devices on that rail would require their regulator anyway.
> And I guess the point would be that drivers should not be aware that they are on VBAT
> or any other rail.
>
> So, what's the correct way to solve this?
>
> - Should drivers fail nicely if a regulator_get fail? And continue even if one fails.
> - Should drivers disable regfw usage completely in the driver if regulator_get doesn't
> give valid regulator ?
> - or Should a fake fixed regulator be added for vbat so drivers can still get a valid
> regulator with regulator_get.
>
> The last options seams to be the one that does not require much changes on drivers.
> But it will be adding a regulator that does basically nothing in the system.
>
> BR,
>
> --
> Eduardo Valentin
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo(a)vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/

--
Eduardo Valentin
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Mark Brown on 30 Nov 2009 06:50
On Mon, Nov 30, 2009 at 12:41:02PM +0200, Eduardo Valentin wrote:

> I'm writing to ask about VBAT use case. What is the expected
> way to use regulator framework in case of rail coming from battery?
> Should it be added to the regulator framework at all?

....

> However, drivers for devices on that rail would require their regulator anyway.
> And I guess the point would be that drivers should not be aware that they are on VBAT
> or any other rail.

I'd add it as a fixed voltage regulator and either not specify the
voltage or specify the nominal voltage.

> - Should drivers fail nicely if a regulator_get fail? And continue even if one fails.
> - Should drivers disable regfw usage completely in the driver if regulator_get doesn't
> give valid regulator ?

These are reasonable approaches if the supply is an optional one that
the device does not need - the driver shouldn't be failing if it doesn't
need to.

> - or Should a fake fixed regulator be added for vbat so drivers can still get a valid
> regulator with regulator_get.

That's the way to do it, yes.

> The last options seams to be the one that does not require much changes on drivers.
> But it will be adding a regulator that does basically nothing in the system.

It's not quite doing nothing, it's mapping out the supply on the board.
Otherwise there's no good way to tell the difference between a supply
not being available because the regulator failed to initialise and the
supply not being available because it's provided by some other invisible
means.

If the regulator API starts getting a lot of usage on boards that have
primarily fixed regulators we may want to have some support in the core
for automatically faking up supplies if requested by the board but
there's not been much demand for that yet and there's risks.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: =?utf-8?Q?H=C3=B6gander?= Jouni on 2 Dec 2009 06:40
ext Mark Brown <broonie(a)opensource.wolfsonmicro.com> writes:

> On Mon, Nov 30, 2009 at 12:41:02PM +0200, Eduardo Valentin wrote:
>
>> I'm writing to ask about VBAT use case. What is the expected
>> way to use regulator framework in case of rail coming from battery?
>> Should it be added to the regulator framework at all?
>
> ...
>
>> However, drivers for devices on that rail would require their regulator anyway.
>> And I guess the point would be that drivers should not be aware that they are on VBAT
>> or any other rail.
>
> I'd add it as a fixed voltage regulator and either not specify the
> voltage or specify the nominal voltage.

I was preparing patch for this when I noticed that fixed voltage
regulator can control only one regulator. Are you aware of any ongoing
work to enable control for multiple regulators in
fixed-voltage-regulator?

--
Jouni Högander
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Mark Brown on 2 Dec 2009 07:10
On Wed, Dec 02, 2009 at 01:32:06PM +0200, H??gander Jouni wrote:
> ext Mark Brown <broonie(a)opensource.wolfsonmicro.com> writes:

> > I'd add it as a fixed voltage regulator and either not specify the
> > voltage or specify the nominal voltage.

> I was preparing patch for this when I noticed that fixed voltage
> regulator can control only one regulator. Are you aware of any ongoing
> work to enable control for multiple regulators in
> fixed-voltage-regulator?

It's just a platform device configured with platform data, you can
instantiate as many fixed voltage regulators in your system as you care
to using the current driver.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
 |  Next  |  Last
Pages: 1 2
Prev: [git pull] fuse bug fix
Next: b44.c box lockup fix (netconsole): ratelimit NAPI poll error message