Re: [PATCH 1/8] soc_camera: Use soc_camera_device::sizeimage to compute buffer sizes

[Laurent Pinchart <laurent.pinchart@ideasonboard.com>]

Hi Guennadi,

On Wednesday 25 April 2012 00:06:03 Guennadi Liakhovetski wrote:
> Hi Laurent
> 
> Sorry for a slightly delayed reply;-)

So slightly :-)

> On Thu, 26 Jan 2012, Laurent Pinchart wrote:
> 
> [snip]
> 
> > > > diff --git a/drivers/media/video/sh_mobile_ceu_camera.c
> > > > b/drivers/media/video/sh_mobile_ceu_camera.c index c51decf..f4eb9e1
> > > > 100644
> > > > --- a/drivers/media/video/sh_mobile_ceu_camera.c
> > > > +++ b/drivers/media/video/sh_mobile_ceu_camera.c
> 
> [snip]
> 
> > > Looks like sh_mobile_ceu_set_rect() can also be simplified, since there
> > > bytes_per_line is calculated for data-fetch mode, for which the
> > > ->bytesperline can also be used?
> > 
> > Is sh_mobile_ceu_set_rect() guaranteed to be called after try_fmt(), with
> > the ->bytesperline value set to the correct value for the current format
> > ?
>
> I think it is, yes. soc_camera.c always configures the pipeline upon the
> first .open() call by calling soc_camera_set_fmt(), at which point
> ->bytesperline is set too. Also, just to avoid confusion - above you meant
> set_fmt(), not try_fmt(), right? *try* are not supposed to set anything.

Yes, I meant set_fmt(), sorry.

I've just checked the code paths in which sh_mobile_ceu_set_rect() is called, 
and I don't see any issue there. We can thus simplify the data-fetch mode code 
in sh_mobile_ceu_set_rect().

> So, if you agree, either you can do a patch 11/9 or I can do it myself. Or
> you could do a v3 of just one patch 3/9.

I'd rather avoid a v3 if possible :-) If you could add a 11/9 patch that would 
be great.

-- 
Regards,

Laurent Pinchart