Re: [PATCH] Document kfree and vfree NULL usage

[Pekka Enberg <penberg@cs.helsinki.fi>]

On Sat, Nov 27, 2004 at 09:43:17PM +0100, Marcel Sebek wrote:
> > diff -urpN linux-2.6.10/sound/core/init.c linux-2.6.10-new/sound/core/init.c
> > --- linux-2.6.10/sound/core/init.c	2004-10-23 10:55:09.000000000 +0200
> > +++ linux-2.6.10-new/sound/core/init.c	2004-11-27 21:21:50.000000000 +0100
> > @@ -665,9 +665,8 @@ int snd_card_file_remove(snd_card_t *car
> >  	spin_unlock(&card->files_lock);
> >  	if (card->files == NULL)
> >  		wake_up(&card->shutdown_sleep);
> > -	if (mfile) {
> > -		kfree(mfile);
> > -	} else {
> > +	kfree(mfile);
> > +	if (!mfile) {
> >  		snd_printk(KERN_ERR "ALSA card file remove problem (%p)\n", file);
> >  		return -ENOENT;
> >  	}

On Sat, 2004-11-27 at 13:23 -0800, Phil Oester wrote:
> The above change seems to always trigger the ENOENT return, no?

No it doesn't. kfree() does not set mfile to NULL. However, I think the
above would be more readable if we did the kfree() _after_ the NULL
check. Marcel, what do you think?

		Pekka