Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Wed, 8 Jan 2020 13:40:00 +0100
From:   Guido =?iso-8859-1?Q?G=FCnther?= <agx@sigxcpu.org>
To:     Pavel Machek <pavel@ucw.cz>
Cc:     Jacek Anaszewski <jacek.anaszewski@gmail.com>,
        Dan Murphy <dmurphy@ti.com>, Rob Herring <robh+dt@kernel.org>,
        Mark Rutland <mark.rutland@arm.com>,
        linux-leds@vger.kernel.org, devicetree@vger.kernel.org,
        linux-kernel@vger.kernel.org
Subject: Re: Exponential LED brightness Re: [PATCH v4 0/6] leds: lm3692x:
 Allow to set ovp and brigthness mode
Message-ID: <20200108124000.GA21245@bogon.m.sigxcpu.org>
References: <cover.1578324703.git.agx@sigxcpu.org>
 <20200107133119.GA3825@duo.ucw.cz>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: 8BIT
In-Reply-To: <20200107133119.GA3825@duo.ucw.cz>
User-Agent: Mutt/1.10.1 (2018-07-13)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Hi Pavel,
On Tue, Jan 07, 2020 at 02:31:20PM +0100, Pavel Machek wrote:
> Hi!
> 
> > Overvoltage protection and brightness mode are currently hardcoded
> > as 29V and disabled in the driver. Make these configurable via DT.
> > 
> > This v4 moves the exponential brightness mode to the back of the series
> > as per Pavel's request:
> > 
> >   https://lore.kernel.org/linux-next/20200106103233.GA32426@amd/T/#m93270a9bf10b88e060f4e4cf5701c527476de985
> > 
> > The end result is identical and i've tested everything still works when
> > dropping the last to patches and checked compiltion via
> 
> Thank you. Applied 1-4 (with some reformatting of changelog, and
> led->LED).
> 
> Exponential mode:
> 
> We should decide if LEDs should be linear or not. Most LEDs are linear
> now, and we may want to make it part of the API. Additional advantage
> is that linear is "well defined". It is actually quite important for
> RGB LEDs, because you get wrong colors otherwise.
> 
> (Non-linear can have advantages, too... like needing less bits.)
> 
> So, my suggestion is to document LEDs as linear, and leave
> exponential->linear conversion to someone else.

That would mean doing a conversion in the kernel that can be done by the
chip. Would exposing non-linearity like in
/sys/class/backlight/<backlight>/scale be an option?
Cheers,
 -- Guido


> 
> Best regards,
> 
> 									Pavel
> -- 
> (english) http://www.livejournal.com/~pavelmachek
> (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html