Date: Sat, 27 Jun 2009 02:14:50 +0300
From: Sergey Senozhatsky <sergey.senozhatsky@mail.by>
To: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Sergey Senozhatsky <sergey.senozhatsky@mail.by>, Greg KH <gregkh@suse.de>,
       Kay Sievers <kay.sievers@vrfy.org>, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] kobject_set_name_vargs memory leak
Message-ID: <20090626231450.GC3858@localdomain.by>
References: <20090626143652.GB6281@localdomain.by>
 <20090626144949.GA24173@suse.de>
 <20090626222937.GA3858@localdomain.by>
 <m17hyyfvnv.fsf@fess.ebiederm.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <m17hyyfvnv.fsf@fess.ebiederm.org>
User-Agent: Mutt/1.5.20 (2009-06-14)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 2249
Lines: 71

On (06/26/09 16:00), Eric W. Biederman wrote:
> >> > Fix memory leak when kobject_set_name_vargs returns -ENOMEM.
> >> > 
> >> > Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@mail.by>
> >> > ---
> >> > diff --git a/lib/kobject.c b/lib/kobject.c
> >> > index b512b74..922cd8c 100644
> >> > --- a/lib/kobject.c
> >> > +++ b/lib/kobject.c
> >> > @@ -222,8 +222,10 @@ int kobject_set_name_vargs(struct kobject *kobj, const char *fmt,
> >> >  		return 0;
> >> >  
> >> >  	kobj->name = kvasprintf(GFP_KERNEL, fmt, vargs);
> >> > -	if (!kobj->name)
> >> > +	if (!kobj->name) {
> >> > +		kfree(old_name);
> >> >  		return -ENOMEM;
> >> > +	}
> >> 
> >> We've been through this before (search lkml archives).  If kvasprintf
> >> fails, then we don't want to free old_name, as the caller might want to
> >> do something with it.
> >> 
> > Hello Greg,
> >
> > int kobject_set_name_vargs.... {
> > 	const char *old_name = kobj->name;
> >
> > old_name is local variable.
> >
> > In the following lines we overwrite kobject->name.
> >
> > 	kobj->name = kvasprintf(GFP_KERNEL, fmt, vargs);
> > 	if (!kobj->name)
> > 		return -ENOMEM;
> >
> > It's not clear to me how we can do anything (including kfree) with old_name after 'return -ENOMEM'.
> 
> My feel is that if we fail we should restore kobject->name to old_name.
> 
> That should also prevent the leak without getting us into trouble elsewhere.
> 
> Eric
> 
Or work with 'new_name' and overwrite kobject->name only 'if(new_name)'.
I thought about restoring. ( Blue or Red Pill? :) )

diff --git a/lib/kobject.c b/lib/kobject.c
index b512b74..d6b1502 100644
--- a/lib/kobject.c
+++ b/lib/kobject.c
@@ -222,8 +222,10 @@ int kobject_set_name_vargs(struct kobject *kobj, const char *fmt,
 		return 0;
 
 	kobj->name = kvasprintf(GFP_KERNEL, fmt, vargs);
-	if (!kobj->name)
+	if (!kobj->name) {
+		kobj->name = old_name;
 		return -ENOMEM;
+	}
 
 	/* ewww... some of these buggers have '/' in the name ... */
 	while ((s = strchr(kobj->name, '/')))


	Sergey
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/