Re: [PATCH v5 7/8] media: i2c: add DS90UB913 driver

From: Laurent Pinchart
Date: Wed Jan 04 2023 - 08:33:03 EST


On Tue, Dec 27, 2022 at 10:09:43PM +0200, Andy Shevchenko wrote:
> On Mon, Dec 26, 2022 at 07:01:11PM +0200, Laurent Pinchart wrote:
> > On Wed, Dec 14, 2022 at 08:29:48AM +0200, Tomi Valkeinen wrote:
> > > On 11/12/2022 20:33, Laurent Pinchart wrote:
> > > > On Thu, Dec 08, 2022 at 12:40:05PM +0200, Tomi Valkeinen wrote:
>
> ...
>
> > > >> + dev_dbg(dev, "Bind %s\n", source_subdev->name);
> > > >
> > > > I'd drop this message.
>
> +1 here.
>
> > > Why is that? Do we get this easily from the v4l2 core? These debug
> > > prints in the bind/unbind process have been valuable for me.
> >
> > Because debug messages are not meant to be a tracing infrastructure, and
> > because, if we want to keep this message, it would be best handled in
> > the v4l2-async core instead of being duplicated across drivers. Same for
> > the messages at the end of the function.
>
> I don't think v4l2 needs debug prints. If we consider the above case, the
> ftrace already provides that. If we consider something specific to v4l2 to
> trace only critical parts, then trace events should be implemented.

V4L2 has no support for trace events, and I agree that they would be
useful. It doesn't mean that dev_dbg() is entirely useless though, and
sometimes the line between the two isn't very clear. In any case, for
this very specific debug message, I don't think it should be printed in
individual drivers, but it should be handled in the V4L2 core, either as
a debug message or a trace event.

--
Regards,

Laurent Pinchart