RE: [PATCH] leds: leds-dual-gpio: Add dual GPIO LEDs driver

From: Hermes Zhang <Hermes.Zhang@axis.com>
To: Alexander Dahl <ada@thorsis.com>,
	"linux-leds@vger.kernel.org" <linux-leds@vger.kernel.org>
Cc: Pavel Machek <pavel@ucw.cz>, Dan Murphy <dmurphy@ti.com>,
	kernel <kernel@axis.com>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	Marek Behun <marek.behun@nic.cz>
Subject: RE: [PATCH] leds: leds-dual-gpio: Add dual GPIO LEDs driver
Date: Fri, 12 Mar 2021 08:48:55 +0000	[thread overview]
Message-ID: <e56946d1d85948c7b6ce94982a221981@XBOX03.axis.com> (raw)
In-Reply-To: <2315048.uTtSMl1LR1@ada>

Hi Alexander,

> Am Donnerstag, 11. März 2021, 14:04:08 CET schrieb Hermes Zhang:
> > From: Hermes Zhang <chenhuiz@axis.com>
> >
> > Introduce a new Dual GPIO LED driver. These two GPIOs LED will act as
> > one LED as normal GPIO LED but give the possibility to change the
> > intensity in four levels: OFF, LOW, MIDDLE and HIGH.
> 
> Interesting use case. Is there any real world hardware wired like that you
> could point to?
> 

Yes, we have the HW, it's not a chip but just some circuit to made of.

> > +config LEDS_DUAL_GPIO
> > +	tristate "LED Support for Dual GPIO connected LEDs"
> > +	depends on LEDS_CLASS
> > +	depends on GPIOLIB || COMPILE_TEST
> > +	help
> > +	  This option enables support for the two LEDs connected to GPIO
> > +	  outputs. These two GPIO LEDs act as one LED in the sysfs and
> > +	  perform different intensity by enable either one of them or both.
> 
> Well, although I never had time to implement that, I suspect that could
> conflict if someone will eventually write a driver for two pin dual color LEDs
> connected to GPIO pins.  We actually do that on our hardware and I know
> others do, too.
> 
> I asked about that back in 2019, see this thread:
> 
> https://www.spinics.net/lists/linux-leds/msg11665.html
> 
> At the time the multicolor framework was not yet merged, so today I would
> probably make something which either uses the multicolor framework or at
> least has a similar interface to userspace. However, it probably won't surprise
> you all, this is not highest priority on my ToDo list. ;-)
> 
> (What we actually do is pretend those are separate LEDs and ignore the
> conflicting case where both GPIOs are on and the LED is dark then.)
> 

Yes, that case seems conflict with mine, the pattern for me is like:

P1 | P2 | LED
-- + -- + -----
 0 |  0 | off
 0 |  1 | Any color
 1 |  0 | Any color
 1 |  1 | both on

Now I'm investigate another way from Marek's suggestion by using REGULATOR_GPIO, to see if could meet my requirement. If yes, then I do think no new  driver is needed.

Best Regards,
Hermes