Re: [PATCH] leds: Add Chrome OS keyboard backlight LEDs driver

[Dmitry Torokhov <dmitry.torokhov@gmail.com>]

On Fri, Mar 04, 2016 at 03:41:36PM -0500, Evan McClain wrote:
> On Fri, 2016-03-04 at 11:13 -0800, Dmitry Torokhov wrote:
> > On Fri, Mar 04, 2016 at 10:38:40AM +0100, Jacek Anaszewski wrote:
> > > 
> > > Hi Evan,
> > > 
> > > On 03/04/2016 09:38 AM, Evan McClain wrote:
> > > > 
> > > > On Thu, 2016-03-03 at 15:46 -0800, Dmitry Torokhov wrote:
> > > > > 
> > > > > From: Simon Que <sque@chromium.org>
> > > > > 
> > > > > This is a driver for ACPI-based keyboard backlight LEDs found
> > > > > on
> > > > > Chromebooks. The driver locates \\_SB.KBLT ACPI device and
> > > > > exports
> > > > > backlight as "chromeos::kbd_backlight" LED class device in
> > > > > sysfs.
> > > > Was it ever decided where this driver should live? I was planning
> > > > on
> > > > submitting to platform/chrome since most keyboard backlights seem
> > > > to
> > > > live over there but I don't think I got a response.
> > > > 
> > > It hasn't been decided yet. I can take it, but could you submit one
> > > more
> > > version, without
> > > 
> > > 'owner	= THIS_MODULE' in struct platform_driver
> > > keyboard_led_driver ?
> > > 
> > > It is redundant, because the core will do it.
> > > 
> > > Also the line with devm_kzalloc has over 80 characters.
> > Also:
> > 
> > - preferably use sizeof(*cdev) instead of sizeof(struct ...)
> > - do not check cdev->flags & LED_SUSPENDED in
> >   keyboard_led_set_brightness() as it is not going to be called when
> > led
> >   device is suspended anyway
> 
> Your patch is definitely better, I was only taking Simon's original
> submission and doing the minimal cleanup to help get it submitted (as
> someone using mainline linux on a pixel 2/samus).
> 
> > - change the MODULE_LICENSE from "GPL v2" to "GPL" as to match the
> >   actual license notice
> 
> I think this got changed in one of the revisions in error.
> 
> > - report ACPI errors in error messages (since we clobber them)
> > - preferably use ENXIO instead of ENODEV
> 
> Most other drivers seem to use ENODEV on probe, but I'm in the 'learn
> through grep' level of understanding for parts of linux.
> 
> > - maybe add "depends on CHROME_PLATFORMS || COMPILE_TEST" so that we
> > do
> >   not prompt for it on non-Chrome platforms
> 
> Adding depends on CHROME_PLATFORMS definitely makes sense, but also
> might support putting this driver in platform/chrome. Either way I just

Olof, do you want it in platform/chrome?

> selfishly want better mainline support for this laptop.
> 
> The only change I made (other than the changes suggested by Jacek) was
> to remove the line setting brightness to max_brightness on probe. I can

I wonder if we should actually read the current brightness before
registering the led device.

Thanks.

-- 
Dmitry