Re: [PATCH] staging: greybus: Clear up precedence for gcam logging macros

[Alex Elder]

On 4/6/24 4:09 AM, Dan Carpenter wrote:
> On Fri, Apr 05, 2024 at 02:22:05PM -0700, Jackson Chui wrote:
> > On Thu, Apr 04, 2024 at 05:05:09PM -0500, Alex Elder wrote:
> > > On 4/3/24 7:16 PM, Jackson Chui wrote:
> > > > Reported by checkpatch:
> > > >
> > > > CHECK: Macro argument 'gcam' may be better as '(gcam)' to avoid
> > > > precedence issues
> > >
> > > I agree with your argument about the way the macro should be
> > > defined.  But perhaps these gcam_*() functions could just
> > > be eliminated?
> > >
> > > I see 15 calls to gcam_err(), 1 call to gcam_dbg(), and none
> > > to gcam_info().  It would be a different patch, but maybe
> > > you could do that instead?
> > >
> > > 					-Alex
> > >
> > >
> > > > Disambiguates '&' (address-of) operator and '->' operator precedence,
> > > > accounting for how '(gcam)->bundle->dev' is a 'struct device' and not a
> > > > 'struct device*', which is required by the dev_{dbg,info,err} driver
> > > > model diagnostic macros. Issue found by checkpatch.
> > > >
> > > > Signed-off-by: Jackson Chui <jacksonchui.qwerty@xxxxxxxxx>
> > > > ---
> > > >    drivers/staging/greybus/camera.c | 6 +++---
> > > >    1 file changed, 3 insertions(+), 3 deletions(-)
> > > >
> > > > diff --git a/drivers/staging/greybus/camera.c b/drivers/staging/greybus/camera.c
> > > > index a8173aa3a995..d82a2d2abdca 100644
> > > > --- a/drivers/staging/greybus/camera.c
> > > > +++ b/drivers/staging/greybus/camera.c
> > > > @@ -180,9 +180,9 @@ static const struct gb_camera_fmt_info *gb_camera_get_format_info(u16 gb_fmt)
> > > >    #define GB_CAMERA_MAX_SETTINGS_SIZE	8192
> > > > -#define gcam_dbg(gcam, format...)	dev_dbg(&gcam->bundle->dev, format)
> > > > -#define gcam_info(gcam, format...)	dev_info(&gcam->bundle->dev, format)
> > > > -#define gcam_err(gcam, format...)	dev_err(&gcam->bundle->dev, format)
> > > > +#define gcam_dbg(gcam, format...)	dev_dbg(&((gcam)->bundle->dev), format)
> > > > +#define gcam_info(gcam, format...)	dev_info(&((gcam)->bundle->dev), format)
> > > > +#define gcam_err(gcam, format...)	dev_err(&((gcam)->bundle->dev), format)
> > > >    static int gb_camera_operation_sync_flags(struct gb_connection *connection,
> > > >    					  int type, unsigned int flags,
> >
> > Thanks for the feedback, Alex!
> >
> > I thought about refactoring it, but I feel it is worth keeping
> > the macro around. It acts as an apdater between callers, who
> > have 'gcam' and want to log and what the dynamic debug macros
> > expect. Without it, the code gets pretty ugly.
>
> Another idea would be to create a function:
>
> struct device *gcam_dev(struct gb_camera *gcam)
> {
> 	return &gcam->bundle->dev;
> }
>
> 	dev_dbg(gcam_dev(gcam), "received metadata ...
>
> (I don't know how to actually compile this code so I haven't tried it).

Yes, I prefer this over the original suggestion.  But
even here the gcam_dev() function doesn't add all that
much value; it saves four characters I guess.

Jackson, the basic principle that makes me say I don't
like the wrapper macros is that the wrapper obscures
the simple call(s) to dev_dbg(), etc.  If there was
something you wanted to do every time--along with
calling dev_dbg()--then maybe the wrapper would be
helpful, but instead it simply hides the standard call.
Better to have the code just use the functions kernel
programmers recognize.

					-Alex
> regards,
> dan carpenter