[<prev] [next>] [day] [month] [year] [list]
Message-ID: <658307f3-7abe-5738-7a8a-2eff6e0902a1@st.com>
Date: Tue, 27 Jun 2017 07:57:02 +0000
From: Hugues FRUCHET <hugues.fruchet@...com>
To: "H. Nikolaus Schaller" <hns@...delico.com>,
Suman Anna <s-anna@...com>,
Laurent Pinchart <laurent.pinchart@...asonboard.com>
CC: Pavel Machek <pavel@....cz>, Mark Rutland <mark.rutland@....com>,
devicetree <devicetree@...r.kernel.org>,
Benjamin Gaignard <benjamin.gaignard@...aro.org>,
Alexandre TORGUE <alexandre.torgue@...com>,
LKML <linux-kernel@...r.kernel.org>,
Hans Verkuil <hverkuil@...all.nl>,
"Rob Herring" <robh+dt@...nel.org>,
Maxime Coquelin <mcoquelin.stm32@...il.com>,
Yannick FERTRE <yannick.fertre@...com>,
Sylwester Nawrocki <sylvester.nawrocki@...il.com>,
Mauro Carvalho Chehab <mchehab@...nel.org>,
Guennadi Liakhovetski <g.liakhovetski@....de>,
linux-arm-kernel <linux-arm-kernel@...ts.infradead.org>,
"linux-media@...r.kernel.org" <linux-media@...r.kernel.org>,
Discussions about the Letux Kernel
<letux-kernel@...nphoenux.org>
Subject: Re: omap3isp camera was Re: [PATCH v1 0/6] Add support of OV9655
camera
Hi Nikolaus,
I would propose to work first on YUV support, so you can test a YUV VGA
grabbing using your OMPA3 setup, I will add this support then in patch
serie.
For the co-work, let's continue on IRC (irc.freenode.net), chat #v4l, my
pseudo is "hfr".
BR,
Hugues.
On 06/26/2017 06:28 PM, H. Nikolaus Schaller wrote:
> Hi Hugues,
>
>> Am 26.06.2017 um 15:19 schrieb Hugues FRUCHET <hugues.fruchet@...com>:
>>
>> Nikolaus,
>> some comments about pixel format/resolution below:
>>
>> On 06/26/2017 10:39 AM, Pavel Machek wrote:
>>> On Mon 2017-06-26 08:05:04, H. Nikolaus Schaller wrote:
>>>> Hi Pavel,
>>>>
>>>>> Am 25.06.2017 um 11:18 schrieb Pavel Machek <pavel@....cz>:
>>>>>
>>>>> Hi!
>>>>>
>>>>>> * unfortunately we still get no image :(
>>>>>>
>>>>>> The latter is likely a setup issue of our camera interface (OMAP3 ISP = Image Signal Processor) which
>>>>>> we were not yet able to solve. Oscilloscoping signals on the interface indicated that signals and
>>>>>> sync are correct. But we do not know since mplayer only shows a green screen.
>>>>>
>>>>> What mplayer command line do you use? How did you set up the pipeline
>>>>> with media-ctl?
>>>>>
>>>>> On kernel.org, I have tree called camera-fw5-6 , where camera works
>>>>> for me on n900. On gitlab, there's modifed fcam-dev, which can be used
>>>>> for testing.
>>>>
>>>> We did have yet another (non-DT) camera driver and media-ctl working in with 3.12.37,
>>>> but had no success yet to update it to work with modern kernels or drivers. It
>>>> is either that the (newer) drivers missing something or the media-ctl has changed.
>>>>
>>>> Here is the log of our scripts with Hugues' driver and our latest setup:
>>>>
>>>> root@...ux:~# ./camera-demo sxga
>>>> DISPLAY=:0
>>>> XAUTHORITY=tcp
>>>> Camera: /dev/v4l-subdev8
>>>> Setting mode sxga
>>>> media-ctl -r
>>>> media-ctl -l '"ov965x":0 -> "OMAP3 ISP CCDC":0[1]'
>>>> media-ctl -l '"OMAP3 ISP CCDC":1 -> "OMAP3 ISP CCDC output":0[1]'
>>>> media-ctl -V '"ov965x":0 [UYVY2X8 1280x1024]'
>>>> media-ctl -V '"OMAP3 ISP CCDC":0 [UYVY2X8 1280x1024]'
>>>> media-ctl -V '"OMAP3 ISP CCDC":1 [UYVY 1280x1024]'
>>>
>>> Ok, so you are using capture, not preview.
>>>
>>> You may want to try this one:
>>>
>>> commit 0eae9d2a8f096f703cbc8f9a0ab155cd3cc14cef
>>> Author: Pavel <pavel@....cz>
>>> Date: Mon Feb 13 21:26:51 2017 +0100
>>>
>>> omap3isp: fix VP2SDR bit so capture (not preview) works
>>>
>>> This is neccessary for capture (not preview) to work properly on
>>> N900. Why is unknown.
>>>
>>> Pavel
>>>
>>>> ### starting mplayer in sxga mode ###
>>>> mplayer tv:// -vf rotate=2 -tv driver=v4l2:device=/dev/video2:outfmt=uyvy:width=1280:height=1024:fps=15 -vo x11
>>
>> => "outfmt=uyvy:width=1280:height=1024"
>>
>> Nikolaus,
>> Be careful that only VGA/RGB565 is coded in this basic version of OV9655,
>> perhaps this explain partly your troubles ?
>
> Ah, I see. The driver should support SXGA and UYVY2X8 (because our 3.12 compatible driver did).
>
> This very old (but working) non-DT driver for 3.12 kernels
> was not based on the ov9650.c code but mt9p031.c:
>
> http://git.goldelico.com/?p=gta04-kernel.git;a=blob;f=drivers/media/i2c/ov9655.c;hb=refs/heads/3.12.37
>
> We abandoned this independent driver because we felt (like you) that extending the existing
> ov9650 driver is a better solution for mainline.
>
>
> At least in theory. Therefore I assumed your submission supports SXGA and UYVY as well,
> since your work is based on ours.
>
> Nevertheless, VGA resolution doesn't work either.
>
> root@...ux:~# ./camera-demo vga
> DISPLAY=:0
> XAUTHORITY=tcp
> Camera: /dev/v4l-subdev8
> Setting mode vga
> media-ctl -r
> media-ctl -l '"ov965x":0 -> "OMAP3 ISP CCDC":0[1]'
> media-ctl -l '"OMAP3 ISP CCDC":1 -> "OMAP3 ISP CCDC output":0[1]'
> media-ctl -V '"ov965x":0 [UYVY2X8 640x480]'
> media-ctl -V '"OMAP3 ISP CCDC":0 [UYVY2X8 640x480]'
> media-ctl -V '"OMAP3 ISP CCDC":1 [UYVY 640x480]'
> ### starting mplayer in vga mode ###
> mplayer tv:// -vf rotate=2 -tv driver=v4l2:device=/dev/video2:outfmt=uyvy:width=640:height=480:fps=30 -vo x11
>
>
> A little later it reports a NULL pointer dereference in ccdc_link_validate() / ccdc_is_shiftable().
>
> It appears as if the input format translates into a NULL pointer by
> omap3isp_video_format_info(0x00001008).
>
> This NULL pointer is not catched by ccdc_is_shiftable().
>
> And it indicates that the camera driver is doing a format that is not
> supported by omap3isp...
>
>
> So how should we proceed?
>
> It looks as if your driver supports your scenario (STM32F746G-DISCO) in VGA/RGB565
> and our drivers (basically) support ours (omap3isp) in VGA/SXGA but UYVY.
>
> We certainly need a generic driver that supports all platforms and formats. So
> we somehow need to get all this stuff into a single driver.
>
> Working on two different patch sets doesn't make sense and would be rejected
> my maintainers...
>
> Basically I could work on your basis to add what we need, but this requires your
> basis to be stabilized first. Or I would spend more time rebasing my code than
> fixing things.
>
> Or you implement the missing formats / features and I test and try to pinpoint
> and report issues to you for integration.
>
> So we have to agree on some way to coordinate the work, especially who submits
> patch sets for review here. Since you were faster than us to submit anything,
> so you should continue with this series.
>
> BR and thanks,
> Nikolaus
>
Powered by blists - more mailing lists