Re: [PATCH 1/3] driver core: add probe_err log helper

[Joe Perches <joe@perches.com>]

On Wed, 2018-10-17 at 12:29 +0100, Russell King - ARM Linux wrote:
> On Tue, Oct 16, 2018 at 04:55:00PM +0300, Andy Shevchenko wrote:
> > On Tue, Oct 16, 2018 at 3:55 PM Andrzej Hajda <a.hajda@samsung.com> wrote:
> > > On 16.10.2018 13:29, Andrzej Hajda wrote:
> > > > On 16.10.2018 13:01, Andy Shevchenko wrote:
> > > > > On Tue, Oct 16, 2018 at 10:22 AM Andrzej Hajda <a.hajda@samsung.com> wrote:
> > > > > > During probe every time driver gets resource it should usually check for error
> > > > > > printk some message if it is not -EPROBE_DEFER and return the error. This
> > > > > > pattern is simple but requires adding few lines after any resource acquisition
> > > > > > code, as a result it is often omited or implemented only partially.
> > > > > > probe_err helps to replace such code seqences with simple call, so code:
> > > > > >         if (err != -EPROBE_DEFER)
> > > > > >                 dev_err(dev, ...);
> > > > > >         return err;
> > > > > > becomes:
> > > > > >         return probe_err(dev, err, ...);
> > > > > > +               va_start(args, fmt);
> > > > > > +
> > > > > > +               vaf.fmt = fmt;
> > > > > > +               vaf.va = &args;
> > > > > > +
> > > > > > +               __dev_printk(KERN_ERR, dev, &vaf);
> > > > > It would be nice to print an error code as well, wouldn't it?
> > > > Hmm, on probe fail error is printed anyway (with exception of
> > > > EPROBE_DEFER, ENODEV and ENXIO):
> > > >     "probe of %s failed with error %d\n"
> > > > On the other side currently some drivers prints the error code anyway
> > > > via dev_err or similar, so I guess during conversion to probe_err it
> > > > should be removed then.
> > > > 
> > > > If we add error code to probe_err is it OK to report it this way?
> > > >     dev_err(dev, "%V, %d\n", &vaf, err);
> > > 
> > > Ups, I forgot that message passed to probe_err will contain already
> > > newline character.
> > 
> > You may consider not to pass it.
> 
> It's normal to pass the '\n', so by doing this, we create the situation
> where this function becomes the exception to the norm.  That's not a
> good idea - we will see people forget that appending '\n' should not
> be done for this particular function.
> 
> While we could add a checkpatch rule, that's hassle (extra rework).

It would not be a simple checkpatch rule with high confidence
because of pr_cont uses that may not be in the patch context.

> In I think the message would be much better formatted if we did:
> 
> 	dev_err(dev, "error %d: %V", err, &vaf);

s/%V/%pV/