Re: [PATCH v1] media: i2c: ov5640: Implement get_mbus_config

[Philipp Zabel]

Hi Jacopo,

On Sat, Jan 28, 2023 at 11:06:11AM +0100, Jacopo Mondi wrote:
> Hi Laurent
> 
> On Fri, Jan 27, 2023 at 08:34:07PM +0200, Laurent Pinchart wrote:
> > On Fri, Jan 27, 2023 at 06:50:03PM +0100, Jacopo Mondi wrote:
> > > On Fri, Jan 27, 2023 at 04:12:44PM +0100, Marcel Ziswiler wrote:
> > > > From: Aishwarya Kothari <aishwarya.kothari@xxxxxxxxxxx>
> > > >
> > > > Implement the introduced get_mbus_config operation to report the
> > > > number of used data lanes on the MIPI CSI-2 interface.
> > > >
> > >
> > > OV5640 can operate in parallel mode too.
> > >
> > > You can check how it currently configured with ov5640_is_csi2() and
> > > populate struct v4l2_mbus_config accordingly.
> >
> > I'm also wondering which CSI-2 receiver needs .get_mbus_config() for the
> > ov5640. The number of lanes is usually specified in DT, on both sides of
> > the link. It's only when selecting a number of lanes dynamically at
> > runtime that .get_mbus_config() is needed.
> >
> 
> iirc Aishwarya and Marcel reported issues on i.MX6 so I presume they
> need get_mbus_config as a drivers in staging/media/imx/ requires
> that:
> 
> drivers/staging/media/imx/imx6-mipi-csi2.c
> Fetches the remote mbus config to get the number of lanes and make
> sure the bus type is CSI-2
> 
> drivers/staging/media/imx/imx-media-csi.c
> Fetches the remote mbus config to deduce the bus type in use
> 
> In both cases I concur the callers can be fixed to parse their
> endpoints but looking at commit 7318abface486d6a6389731810f5b60650daedb5
> it seems that was not the plan (reason not clear to me)

The tc358743 driver dynamically changes the number of active lanes
depending on bandwidth requirements.

regards
Philipp