Message-ID: <d054a18a-10d1-7aa5-31d5-44af1e5a6c76@gmail.com>
Date:   Wed, 14 Dec 2016 12:05:50 -0800
From:   David Daney <ddaney.cavm@...il.com>
To:     Arnd Bergmann <arnd@...db.de>,
        Neil Armstrong <narmstrong@...libre.com>
Cc:     Thomas Petazzoni <thomas.petazzoni@...e-electrons.com>,
        mark.rutland@....com,
        Jaghathiswari Rankappagounder Natarajan <jaghu@...gle.com>,
        devicetree@...r.kernel.org, Greg KH <gregkh@...uxfoundation.org>,
        openbmc@...ts.ozlabs.org, linux@...linux.org.uk,
        linux-kernel@...r.kernel.org, robh+dt@...nel.org, joel@....id.au,
        linux-arm-kernel@...ts.infradead.org
Subject: Re: [PATCH linux v1 0/4] Seven segment display support

On 12/14/2016 06:15 AM, Arnd Bergmann wrote:
> On Wednesday, December 14, 2016 2:12:41 PM CET Neil Armstrong wrote:
>> On 12/14/2016 01:56 PM, Greg KH wrote:
>>> On Wed, Dec 14, 2016 at 01:45:30PM +0100, Thomas Petazzoni wrote:
>>>> Hello,
>>>>
>>>> On Tue, 13 Dec 2016 23:55:00 -0800, Jaghathiswari Rankappagounder
>>>> Natarajan wrote:
>>>>
>>>>> Documentation for the binding which provides an interface for adding clock,
>>>>> data and clear signal GPIO lines to control seven segment display.
>>>>>
>>>>> The platform device driver provides an API for displaying on two 7-segment
>>>>> displays, and implements the required bit-banging. The hardware assumed is
>>>>> 74HC164 wired to two 7-segment displays.
>>>>>
>>>>> The character device driver implements the user-space API for letting a user
>>>>> write to two 7-segment displays including any conversion methods necessary
>>>>> to map the user input to two 7-segment displays.
>>>>>
>>>>> Adding clock, data and clear signal GPIO lines in the devicetree to control
>>>>> seven segment display on zaius platform.
>>>>>
>>>>> The platform driver matches on the device tree node; the platform driver also
>>>>> initializes the character device.
>>>>>
>>>>> Tested that the seven segment display works properly by writing to the
>>>>> character device file on a EVB AST2500 board which also has 74HC164 wired
>>>>> to two 7-segment displays.
>>>>
>>>> FWIW, I proposed a driver for seven segment displays back in 2013:
>>>>
>>>>    http://lists.infradead.org/pipermail/linux-arm-kernel/2013-January/139986.html
>>>>
>>>> And the feedback from Greg KH was: we don't need a driver for that, do
>>>> it from userspace. See:
>>>>
>>>>    http://lists.infradead.org/pipermail/linux-arm-kernel/2013-January/139992.html
>>>>
>>>> So: good luck
>>>
>>> Did anyone ever write a library for this type of thing?
>>>
>>> Again, I don't want to see one-off drivers for random devices like this
>>> that should be able to all be controlled from userspace in a common
>>> manner.  Much like we did for fingerprint readers a long long time
>>> ago...
>>>
>
>> Actually, it's more than a random interface, a lot of SoCs and boards actually have such displays
>> and it's a pity to use UIO, sysfs gpio bitbanging and all sort of ugly stuff to only print a few
>> characters a simple and clean driver could achieve.
>> Some very well known SoCs even have integrated registers to lower the BOM and bypass the need for
>> a 74HC164 like component and avoid gpio bit banging.
>>
>> My personal concern is that you could also need to drive more segments, thus 7-segments
>> is too restrictive.
>>
>> But this driver is well structured, the gpio-bitbanging sub-driver is welcome.
>
> Maybe we can find a way to fit this into the existing drivers/leds/ subsystem?
>
> That already supports blinking, brightness and colour attributes of LEDs,
> so could this be extended to support (one of) digit, number, character
> or string with a common sysfs attribute and a way to hook up a led driver
> to that?

We have a lot of boards with an 8-cell dot matrix LED. Each cell is 
programmed with an 8-bit value.  The mapping of these values to the dots 
defaults to ASCII character rendering, but there is the facility to 
install other bitmaps as well.

Really I view these things not as part of the LED subsystem, but more as 
a very small frame buffer.

We like to display entire words, and the most useful interface from a 
user point of view is something that consumes entire strings rather than 
having to manage each cell independently.

You could imagine that if the text to be displayed were longer than the 
display, that the driver would make it continuously scroll.  I would 
like to see a framework where a simple character device were exposed, 
and from userspace you could do: "echo message > /dev/small-display" and 
get something sensible.

>
> 	Arnd
>
> _______________________________________________
> linux-arm-kernel mailing list
> linux-arm-kernel@...ts.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
>

Powered by blists - more mailing lists