Re: [PATCH v5 0/9] Output raw touch data via V4L2

[Hans Verkuil]

On 06/27/2016 01:57 PM, Nick Dyer wrote:
> Hi Hans-
> 
> Thanks for reviewing this again in such detail.
> 
> On 27/06/2016 12:26, Hans Verkuil wrote:
>> On 06/23/2016 12:08 AM, Nick Dyer wrote:
>>> This is a series of patches to add output of raw touch diagnostic data via V4L2
>>> to the Atmel maXTouch and Synaptics RMI4 drivers.
>>>
>>> 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.
>>>
>>> We have a utility which can read the data and display it in a useful format:
>>>     https://github.com/ndyer/heatmap/commits/heatmap-v4l
>>>
>>> These patches are also available from
>>>     https://github.com/ndyer/linux/commits/v4l-touch-2016-06-22
>>>
>>> Changes in v5 (Hans Verkuil review):
>>> - Update v4l2-core:
>>>   - Add VFL_TYPE_TOUCH, V4L2_BUF_TYPE_TOUCH_CAPTURE and V4L2_CAP_TOUCH
>>
>> The use of V4L2_CAP_TOUCH and V4L2_BUF_TYPE_TOUCH_CAPTURE is very inconsistent.
>> What is the rationale of adding V4L2_BUF_TYPE_TOUCH_CAPTURE? I can't remember
>> asking for it.
> 
> I am afraid that I missed updating atmel_mxt_ts from
> V4L2_BUF_TYPE_VIDEO_CAPTURE to V4L2_BUF_TYPE_TOUCH_CAPTURE, which has
> confused the situation.
> 
> Perhaps I read too much into your request that I look at the way that SDR
> is treated. When I started going through the code paths in v4l2-core and
> v4l2-compliance, it seemed cleaner to treat touch as completely separate,
> hence introducing the new BUF_TYPE. I'm happy to try it without this.

Yeah, I didn't mean that you had to add a new BUF_TYPE. My remark was related to
ensuring that all occurrences in the spec where they talk about the various
/dev/video/radio/etc. devices are extended with v4l-touch as well.

> 
>> And wouldn't the use of V4L2_BUF_TYPE_TOUCH_CAPTURE break userspace for sur40?
> 
> I think it is likely, yes. And it looks like that would make Florian unhappy.
> 
>> I'm ambiguous towards having a V4L2_BUF_TYPE_TOUCH_CAPTURE, to be honest.
>>
>> I would also recommend renaming V4L2_CAP_TOUCH to V4L2_CAP_TOUCH_CAPTURE.
> 
> Do you agree with the following changes:
> 
> - Rename V4L2_CAP_TOUCH to V4L2_CAP_TOUCH_CAPTURE.
> 
> - Touch devices should register both V4L2_CAP_VIDEO_CAPTURE and
> V4L2_CAP_TOUCH_CAPTURE.
> 
> - Get rid of V4L2_BUF_TYPE_TOUCH_CAPTURE and use
> V4L2_BUF_TYPE_VIDEO_CAPTURE. In v4l2-ioctl.c if we need to force particular
> pix formats for touch, it will need to look at V4L2_CAP_TOUCH_CAPTURE.

Actually, I think we have two choices, depending on whether we use a
BUF_TYPE_TOUCH_CAPTURE or not.

1) If we go with a BUF_TYPE_TOUCH_CAPTURE, then:

- we need a V4L2_CAP_TOUCH_CAPTURE
- no V4L2_CAP_VIDEO_CAPTURE will be set (since that would indicate support for
  BUF_TYPE_VIDEO_CAPTURE, which we don't have anymore).
- new callbacks for g/s/try/enum_fmt_tch_cap should be added to v4l2-ioctl.h.

2) Alternatively, if we want to keep using BUF_TYPE_VIDEO_CAPTURE, then:

- we keep V4L2_CAP_TOUCH which is combined with CAP_VIDEO_CAPTURE (and perhaps
  VIDEO_OUTPUT in the future). The CAP_TOUCH just says that this is a touch
  device, not a video device, but otherwise it acts the same.

I'd go with 2, since I see no reason to add a new BUF_TYPE for this.
It acts exactly like video after all, with only a few restrictions (i.e. no
colorspace info or interlaced). And adding a new BUF_TYPE will likely break
the existing sur40 app.

> 
> Your other review comments look straightforward to address - thanks.
> 
> I should say, you can see my current changes to v4l2-compliance here:
> https://github.com/ndyer/v4l-utils/commit/07e00c33
> 
> Should I post them along with the kernel patches next time?

Yes, please.

Regards,

	Hans

> 
>>
>> I can imagine an embedded usb gadget device that outputs touch data to a PC.
>>
>> Regards,
>>
>> 	Hans
> --
> To unsubscribe from this list: send the line "unsubscribe linux-media" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>