Re: [patch 2.6.29-rc3] NAND: fix broken debug messages

[David Brownell <david-b@pacbell.net>]

On Tuesday 03 February 2009, Geert Uytterhoeven wrote:
> > >-            DEBUG(MTD_DEBUG_LEVEL0, "nand_read_oob: "
> > >+            DEBUG(MTD_DEBUG_LEVEL0, "do_nand_write_oob: "
> > 
> > This is wrong too.  The function is called 'nand_do_write_oob'.

That's what happens sometimes when you edit patches by hand.
Originally I just changed "read" to "write", then I noticed
there was "do" in there ... obviously, it got inserted in
the wrong place.  ;)


> > Which leads me to wonder why we aren't doing something like:
> > 
> >       DEBUG(MTD_DEBUG_LEVEL0, "%s: <message>\n", __func__);

That would be sensible.

 
> Or use
> 
>     #define pr_fmt(fmt)  "%s: " fmt, __func__
> 
> at the top of the file, and start using the official pr_*() routines?

Yeah, DEBUG should be treated globally as a reserved
symbol.  Ever notice how dev_dbg() misbhaves in MTD
code?  Ditto pr_dbg(), but that's rarely the right thing
to use when there's a device node in hand.  Yeech.

"Now" (i.e. after rc1 and well before rc-last) would
be a good time to send a big patch changing all the
debug messaging to something less aggressively counter
to the core debug messaging convention of Linux...

- Dave