Re: [PATCH v9 00/21] media: i2c: add Maxim GMSL2/3 serializer and deserializer drivers

[Martin Hecht]

Hi Sakari, hi Dumitru,

I'm testing v8 already and will migrate next week to that version. Below 
some further comments.

On 3/11/26 08:58, Sakari Ailus wrote:
> Hi Dumitru,
>
> On Wed, Mar 11, 2026 at 09:17:12AM +0200, Dumitru Ceclan via B4 Relay wrote:
> > The following deserializers are supported:
> > * MAX96712 (already exists in staging)
> > * MAX96714 (already exists)
> > * MAX96714F (already exists)
> > * MAX96714R (GMSL2)
> > * MAX96716 (GMSL2)
> > * MAX96724 (already exists as part of existing MAX96712 driver)
> > * MAX96724F (GMSL2)
> > * MAX96724R (GMSL2)
> > * MAX9296A (GMSL2)
> > * MAX96792A (GMSL3)
> >
> > The following serializers are supported:
> > * MAX96717 (already exists)
> > * MAX9295A (GMSL2)
> > * MAX96793 (GMSL3)
>
> It'd be nice to have more reviews as well as comments from the users of the
> existing drivers especially on how this works for them.
>
> Are there differences in functionality or UAPI compared to the drivers
> already in upstream?

Compared with the current mainline drivers from Julian (what offere a 
good basic functionality) there are a lot or more features enabled. One 
ov them is support for routing the channels as well as i2c-atr instead 
of i2c-gate only. Both helps to setup up more complex and reconfigurable 
routing of the video channels in complex systems with many 
sensors/cameras/streams.

Additionally a lot more and never devices are supported by that patchset 
including the new GMSL3 devices what are also backward compatible to GMSL2.

Both link types are supported now. Generation of test patterns in pixel 
mode has been added.

Also log-status returns a lot of very useful information about the link 
mode (tunnel mode versus pixel mode) and potential errors / packet 
losses over cable by presenting the counter registers. That helps to 
monitor the quality and reliability of the GMSL links as well as CSI links.

> ...
>
> > The following v4l2-compliance test still fails:
> >                  fail: v4l2-test-subdevs.cpp(371): fmt.code == 0 || fmt.code == ~0U
> >                  fail: v4l2-test-subdevs.cpp(418): checkMBusFrameFmt(node, fmt.format)
> >          test Active VIDIOC_SUBDEV_G/S_FMT: FAIL
>
> Could you post the full report here, please?

BR Martin