[Gta04-owner] On camera driver

[Andrey Utkin]

On Tue, Aug 09, 2016 at 01:44:10PM +0200, H. Nikolaus Schaller wrote:
> Hi Andrey,
> 
> > Am 09.08.2016 um 12:45 schrieb Andrey Utkin <andrey_utkin at fastmail.com>:
> > 
> > On Fri, Jul 29, 2016 at 08:55:53AM +0200, H. Nikolaus Schaller wrote:
> >> This allows me to continue to work on the camera driver where I wasn't able
> >> to install the v4l-utils package to control the ISP (Image Signal Processor)
> >> of the DM3730 SoC.
> >> 
> >> Now the tools are sharp again. Let's start looking into the camera driver...
> > 
> > Please keep me posted about any issues you have with it. (Or give links
> > to existing tickets.)
> 
> you mean the camera driver?
> 
> Well, here are all links:
> 
> http://projects.goldelico.com/p/gta04-kernel/issues/467/
> http://git.goldelico.com/?p=gta04-kernel.git;a=shortlog;h=refs/heads/work/hns/video/ov9655
> http://git.goldelico.com/?p=gta04-kernel.git;a=blob;f=Letux/root/camera-demo;h=2858acac4fad5b684a0ec3a899a0c7b706f64292;hb=07d63901af40477eaa958e18ff0ae9da4c8e03fb
> http://git.goldelico.com/?p=gta04-kernel.git;a=blob;f=Letux/root/capture-demo;h=e6c86844f04ac668fda6ed34006c533fd1050f11;hb=07d63901af40477eaa958e18ff0ae9da4c8e03fb
> http://git.goldelico.com/?p=gta04-kernel.git;a=blob;f=Letux/root/capture.c;h=c2a3bc9d176a78850be52afb8aca1eacfab76392;hb=07d63901af40477eaa958e18ff0ae9da4c8e03fb
> http://electricstuff.co.uk/OV9655-datasheet-annotated.pdf
> 
> The chip is now recognised and probed successfully. This means the DT setup for the I2C command interface is ok.
> 
> What has to be done:
> a) check if the data (video stream) interface setup is correct (does DT choose the correct interface and send out the right clock frequency)
> but it is likely so since the camera chip needs the system clock to respond on I2C. But there could be pinmux issues.
> 
> b) which registers have to be programmed in which way through the v4l2 ioctls...
> We have a very old (working) driver and the ov9650 driver to mix&merge ideas.
> Probably a very basic and static setup could already work and leave gamma correction, pan, crop, auto-gain, resolution switching for further improvements.
> 
> c) I do not know if the mediactl setup is still matching the kernel driver's expectations
> 
> But for all this a developer should have a device with built-in camera for being able
> to make experiments...

Thanks for sharing.
The only way I can help now is to mainline your patches to ov9650
driver. BTW are they complete, i.e. does the driver from
work/hns/video/ov9655 do all expected things?

To test robustness of driver, I'd suggest running

v4l2-compliance -s -d /dev/videoX

(after all the media-ctl invocations) and sharing the output with me.
v4l2-compliance is a lot like Letux/root/capture.c only doing much more
and carrying authoritative opinion of V4L subsystem maintainers.

I don't possess gta04 and currnetly not going to obtain one, but if
there's incentive for me to tackle with it, a lot could be done via ssh
I believe.