Date:	Fri, 1 Jul 2016 12:19:27 -0700
From:	Darren Hart <dvhart@...radead.org>
To:	Jonathan Woithe <jwoithe@...t42.net>
Cc:	Matej Groma <matejgroma@...il.com>,
	platform-driver-x86@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] fujitsu-laptop: Unify max brightness of exported leds

On Thu, Jun 30, 2016 at 10:30:57PM +0930, Jonathan Woithe wrote:
> On Wed, Jun 29, 2016 at 09:28:03AM +0200, Matej Groma wrote:
> > Exported leds had maximum brightness previously unset, thus having
> > value 255. Set maximum brightness of leds that can only be turned
> > off or on to 1, making the behavior more consistent with other leds
> > having binary state.
> 
> I understand the motivation here.  Is this change consistent with the LED
> API though?  Currently for the binary state Fujitsu LEDs, LED_OFF (0) means
> off and LED_FULL (255) means on.  It seems to me that this is the most
> obvious way to use the LED API with binary state LEDs.  This patch changes

That was my reading as well. There is no "enum led_toggle", so using LED_OFF (0)
and LED_FULL (255) seems like the natural set of values. That said, the
"max_brightness" field seems somewhat orthogonal to to the enum led_brightness
mechanism.... hrm...


> the "on" brightness to 1, which is clearly not LED_FULL (it does however
> match the max_brightness configured for the LED).  Thus there is an
> inconsistency: the brightness isn't LED_FULL but in reality the LED is as
> bright as it goes.  Could this create problems within the LED subsystem, or
> is there nothing special attached to the LED_FULL value?  A quick look
> through the leds driver didn't turn up anything obvious, but I might have
> missed some subtlties.
> 
> Darren: any thoughts?

According to Documentation/leds/leds-class.txt:

"
In its simplest form, the LED class just allows control of LEDs from userspace.
LEDs appear in /sys/class/leds/. The maximum brightness of the LED is defined in
max_brightness file. The brightness file will set the brightness of the LED
(taking a value 0-max_brightness). Most LEDs don't have hardware brightness
support so will just be turned on for non-zero brightness settings.
"

So, using 1 is consistent with the above, but it isn't clear that it's "better"
or "more correct". It does change the user interface to the LEDs on fujitsu
laptops, which we have to consider very carefully. If we do this, we need to
introduce a version attribute. See Documentation/sysfs-rules.txt:

"
Userspace applications can, however, expect the format and contents of the
attribute files to remain consistent in the absence of a version attribute
change in the context of a given attribute.
"

Next steps:

1) Matej, why do you want to make this change? You mention consistency with
other LEDs drivers - which ones? Also, why? Is there a tool or something that
expects this behavior which you want to use? Why is this better than sticking
with LED_OFF and LED_FULL?

2) If we change it, we need to first introduce a version attribute (separate
patch), and this patch following it should increment that version indicating the
value policy change.

--
Darren


> 
> Regards
>   jonathan
> 
> > Signed-off-by: Matej Groma <matejgroma@...il.com>
> > ---
> > Changes from v1:
> > 	made commit message more clear
> > 	rebased on testing
> > 
> >  drivers/platform/x86/fujitsu-laptop.c | 10 ++++++----
> >  1 file changed, 6 insertions(+), 4 deletions(-)
> > 
> > diff --git a/drivers/platform/x86/fujitsu-laptop.c b/drivers/platform/x86/fujitsu-laptop.c
> > index 6ce8e78..abb7c62 100644
> > --- a/drivers/platform/x86/fujitsu-laptop.c
> > +++ b/drivers/platform/x86/fujitsu-laptop.c
> > @@ -192,6 +192,7 @@ static void kblamps_set(struct led_classdev *cdev,
> > 
> >  static struct led_classdev kblamps_led = {
> >   .name = "fujitsu::kblamps",
> > + .max_brightness = 1,
> >   .brightness_get = kblamps_get,
> >   .brightness_set = kblamps_set
> >  };
> > @@ -202,6 +203,7 @@ static void radio_led_set(struct led_classdev *cdev,
> > 
> >  static struct led_classdev radio_led = {
> >   .name = "fujitsu::radio_led",
> > + .max_brightness = 1,
> >   .brightness_get = radio_led_get,
> >   .brightness_set = radio_led_set
> >  };
> > @@ -285,7 +287,7 @@ static void logolamp_set(struct led_classdev *cdev,
> >  static void kblamps_set(struct led_classdev *cdev,
> >  			       enum led_brightness brightness)
> >  {
> > -	if (brightness >= LED_FULL)
> > +	if (brightness)
> >  		call_fext_func(FUNC_LEDS, 0x1, KEYBOARD_LAMPS, FUNC_LED_ON);
> >  	else
> >  		call_fext_func(FUNC_LEDS, 0x1, KEYBOARD_LAMPS, FUNC_LED_OFF);
> > @@ -294,7 +296,7 @@ static void kblamps_set(struct led_classdev *cdev,
> >  static void radio_led_set(struct led_classdev *cdev,
> >  				enum led_brightness brightness)
> >  {
> > -	if (brightness >= LED_FULL)
> > +	if (brightness)
> >  		call_fext_func(FUNC_RFKILL, 0x5, RADIO_LED_ON, RADIO_LED_ON);
> >  	else
> >  		call_fext_func(FUNC_RFKILL, 0x5, RADIO_LED_ON, 0x0);
> > @@ -332,7 +334,7 @@ static enum led_brightness kblamps_get(struct led_classdev *cdev)
> >  	enum led_brightness brightness = LED_OFF;
> > 
> >  	if (call_fext_func(FUNC_LEDS, 0x2, KEYBOARD_LAMPS, 0x0) == FUNC_LED_ON)
> > -		brightness = LED_FULL;
> > +		brightness = cdev->max_brightness;
> > 
> >  	return brightness;
> >  }
> > @@ -342,7 +344,7 @@ static enum led_brightness radio_led_get(struct led_classdev *cdev)
> >  	enum led_brightness brightness = LED_OFF;
> > 
> >  	if (call_fext_func(FUNC_RFKILL, 0x4, 0x0, 0x0) & RADIO_LED_ON)
> > -		brightness = LED_FULL;
> > +		brightness = cdev->max_brightness;
> > 
> >  	return brightness;
> >  }
> 
> -- 
> 

-- 
Darren Hart
Intel Open Source Technology Center

Powered by blists - more mailing lists