Re[2]: [PATCH v6 2/2] mfd: syscon: Add non-DT support

[Alexander Shiyan]

> On Monday 11 March 2013, Dong Aisheng wrote:
> > > > > + dev = driver_find_device(&syscon_driver.driver, NULL, (void *)s,
> > > > > +                          syscon_match_pdevname);
> > > > > + if (!dev)
> > > > > +         return ERR_PTR(-ENODEV);
> > > > 
> > > > Should it be ERR_PTR(-EPROBE_DEFER)?
> > > 
> > > I have no idea what better using here. Think that is not so important
> > > since we may have only one possible error code here.
> > 
> > I'm not quite understand your meaning.
> > Since the syscon device may be still not registered,
> > so it may be better to return a EPROBE_DERFER which is the same
> > as dt version.
> 
> I'm guessing that Alexander has not encountered deferred probing yet.
> Alexander, the point here is that returning -EPROBE_DEFER has the
> advantage that a probe() callback calling this function can
> simply return that error code to the driver core. If the driver
> core sees -EPROBE_DEFER, it will retry the same probe function
> later, after all other device probe functions have been called
> and at least one of them was successful. This way you can load the
> syscon driver after loading a driver depending on it and everything
> will still work.

You are right Arnd. I usually try to control the correct order of loading
drivers. Using EPROBE_DEFER in this procedure is more correct.
Probably time for v7... I hope it will finally be the last.

---
èº{.nÇ+‰·Ÿ®‰­†+%ŠËlzwm…ébëæìr¸›zX§»®w¥Š{ayºÊÚë,j­¢f£¢·hš‹àz¹®w¥¢¸¢·¦j:+v‰¨ŠwèjØm¶Ÿÿ¾«‘êçzZ+ƒùšŽŠÝj"ú!¶iO•æ¬z·švØ^¶m§ÿðÃnÆàþY&—