Re: [PATCH 0/8] Input: atmel_mxt_ts - output raw touch diagnostic data via V4L

From: Hans Verkuil
Date: Fri Apr 22 2016 - 11:18:35 EST

On 04/22/2016 05:07 PM, Nick Dyer wrote:
> On 22/04/2016 15:45, Mauro Carvalho Chehab wrote:
>> Em Fri, 22 Apr 2016 10:26:37 +0200
>> Hans Verkuil <hverkuil@xxxxxxxxx> escreveu:
>>> On 04/21/2016 11:31 AM, Nick Dyer wrote:
>>>> This is a series of patches to add diagnostic data support to the Atmel
>>>> maXTouch driver. It's a rewrite of the previous implementation which output via
>>>> debugfs: it now uses a V4L2 device in a similar way to the sur40 driver.
>>>> There are significant performance advantages to putting this code into the
>>>> driver. The algorithm for retrieving the data has been fairly consistent across
>>>> a range of chips, with the exception of the mXT1386 series (see patch).
>>>> We have a utility which can read the data and display it in a useful format:
>>>> These patches are also available from
>>>> Any feedback appreciated.
>>> FYI: we're working on a new buffer type for meta data:
>> One of the things I missed on your patchset is the content of the
>> new format you added (V4L2_PIX_FMT_YS16). You should be patching
>> the V4L2 docbook too, in order to add it there.
> OK, will do. I also see that I forgot Kconfig changes for CONFIG_VIDEO_V4L2
> etc.
>> That's said, if the output is really an image, I don't think it
>> should be mapped via the new V4L2_BUF_TYPE_META_CAPTURE. This type of
>> buffer is meant to be used on non-image metadata, like image statistics
>> to feed auto whitebalance and other similar AAA algorithms.
> The output is raw touch data - i.e. a rectangular grid of nodes each having
> an integer value. I think it is an image in some senses, although perhaps
> it's a matter of opinion!
> You can see an example of a Atmel MXT capacitive touch device here (using
> this patchset):
> There are touch devices which can deliver much higher resolution/framerate.
> For example here's the data coming from a SUR40 which is an optical touch
> sensor but uses V4L in a similar way:
>> It could still make sense to use the new device type (VFL_TYPE_META) for
>> such drivers, as we don't want applications to identify those devices as
>> if they are a webcam.
> I agree it may be a little confusing if things like Skype start picking up
> these devices. Could we #define V4L2_INPUT_TYPE_TOUCH_SENSOR to solve that
> problem?

That might be an idea. I have to admit that I didn't look at the patches in
detail. It mentioned diagnostics, so I didn't realize that it is a image
with a width and height, even though it is not a regular video input.

Adding a new input type won't prevent anyone from picking it up, since
nobody tests that field :-)

On the other hand, it would be a good place to tell the user that it
is from a touch sensor.

Using the upcoming metadata feature wouldn't work since there is no width
and height in the metadata format.

I wonder what others think about adding a new type value.