Re: PATCH convert libata-core to the new debugging scheme

[Jens Axboe <axboe@suse.de>]

On Wed, Jan 25 2006, Jeff Garzik wrote:
> On Wed, Jan 25, 2006 at 04:04:04PM +0100, Bartlomiej Zolnierkiewicz wrote:
> > On 1/24/06, Jeff Garzik <jgarzik@pobox.com> wrote:
> > > On Tue, Jan 24, 2006 at 08:43:43PM +0100, Borislav Petkov wrote:
> > > > for the debugging macros which to me _is_ kinda too much. In addition, the
> > > > ata_msg_xxx() macros are going to be pretty useless if we wrap them one more
> > > > time into other macros but this is only my opinion.
> > >
> > > Yes, I like the current stuff, and am not really interested in further
> > > macros.
> > 
> > What about using Randy's suggestion the other way around:
> > embedding printk-s into ata_msg_* macros?
> > 
> > This way we would get rid of all these if-s in the code
> > without adding more macros.
> > 
> > i.e. we would have
> > 
> > +       ata_msg_ctl(ap, "ata%u: cmd 0x%X\n", ap->id, tf->command);
> > 
> > instead of
> > 
> > +       if (ata_msg_ctl(ap))
> > +                       printk(KERN_DEBUG "%s: ata%u: cmd 0x%X\n",
> > +                               __FUNCTION__, ap->id, tf->command);
> 
> Those are the further macros that I was objecting to...
> 
> The current ata_msg_ctl() style macros are in the kernel, and thus
> accepted.

It looks cleaner, though. printk() lines are usually the worst to keep
from wrapping, saving a tab indentation there seems like it would be
worth it imho.

-- 
Jens Axboe