On 01/02/17 14:51, Jean-Michel Hautbois wrote:
Hi,
2016-12-30 21:26 GMT+01:00 Steve Longerbeam <steve_longerb...@mentor.com>:
On 12/30/2016 11:06 AM, Marek Vasut wrote:
On 12/29/2016 09:51 PM, Robert Schwebel wrote:
Hi Jean-Michel,
Hi,
On Thu, Dec 29, 2016 at 04:08:33PM +0100, Jean-Michel Hautbois wrote:
What is the status of this work?
Philipp's patches have been reworked with the review feedback from the
last round and a new version will be posted when he is back from
holidays.
IMO Philipp's patches are better integrated and well structured, so I'd
rather like to see his work in at some point.
Granted I am biased, but I will state my case. "Better integrated" - my
patches
are also well integrated with the media core infrastructure. Philipp's
patches
in fact require modification to media core, whereas mine require none.
Some changes are needed of course (more subdev type definitions for
one).
As for "well structured", I don't really understand what is meant by that,
but my driver is also well structured.
Philipp's driver only supports unconverted image capture from the SMFC. In
addition
to that, mine allows for all the hardware links supported by the IPU,
including routing
frames from the CSI directly to the Image converter for scaling up to
4096x4096,
colorspace conversion, rotation, and motion compensated de-interlace. Yes
all these
conversion can be carried out post-capture via a mem2mem device, but
conversion
directly from CSI capture has advantages, including minimized CPU
utilization and
lower AXI bus traffic. In any case, Freescale added these hardware paths,
and my
driver supports them.
I had a deeper look to both drivers, and I must say the features of
Steve's one are really interesting.
I don't think any of those has been tested with digital inputs (I have
ADV76xx chips on my board, which I hope will be available one day for
those interested) and so, I can test and help adding some of the
missing parts.
And for at least a week or two, I have all of my time for it, so it
would be of great help to know which one has the bigger chance to be
integrated... :)
Steve's series is definitely preferred from my point of view. The feature
set is clearly superior to Philipp's driver.
I plan on reviewing Steve's series soonish but a quick scan didn't see anything
suspicious. The code looks clean, and I am leaning towards getting this in
sooner
rather than later, so if you have the time to work on this, then go for it!
Steve, I have a SabreLite and a ov5642 sensor, so I should be able to test
that driver.
There is also an ov5642 sensor driver in drivers/media/i2/soc_camera/ov5642.c.
But nobody AFAIK is using it, so I would be inclined to take your code and
remove the soc_camera driver.
Regards,
Hans
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
