Patchwork Ceph: Fix a use-after-free bug in ceph_destroy_client().

login
register
mail settings
Submitter Jesper Juhl
Date Dec. 25, 2010, 6:17 p.m.
Message ID <alpine.LNX.2.00.1012251914370.10759@swampdragon.chaosbits.net>
Download mbox | patch
Permalink /patch/76696/
State Awaiting Upstream
Delegated to: David Miller
Headers show

Comments

Jesper Juhl - Dec. 25, 2010, 6:17 p.m.
Hello,

In net/ceph/ceph_common.c::ceph_destroy_client() the pointer 'client' is 
freed by kfree() and subsequently used in a call to dout() - use after 
free bug.
Easily fixed by simply moving the kfree() call after the dout() call.

Signed-off-by: Jesper Juhl <jj@chaosbits.net>
---
 ceph_common.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
richard -rw- weinberger - Dec. 25, 2010, 9:12 p.m.
On Sat, Dec 25, 2010 at 7:17 PM, Jesper Juhl <jj@chaosbits.net> wrote:
> Hello,
>
> In net/ceph/ceph_common.c::ceph_destroy_client() the pointer 'client' is
> freed by kfree() and subsequently used in a call to dout() - use after
> free bug.

Not really. %p reads only the address of "client".
kfree() does not alter this address.

> Easily fixed by simply moving the kfree() call after the dout() call.
>
> Signed-off-by: Jesper Juhl <jj@chaosbits.net>
> ---
>  ceph_common.c |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> diff --git a/net/ceph/ceph_common.c b/net/ceph/ceph_common.c
> index f3e4a13..890bbbf 100644
> --- a/net/ceph/ceph_common.c
> +++ b/net/ceph/ceph_common.c
> @@ -408,8 +408,8 @@ void ceph_destroy_client(struct ceph_client *client)
>
>        ceph_destroy_options(client->options);
>
> -       kfree(client);
>        dout("destroy_client %p done\n", client);
> +       kfree(client);
>  }
>  EXPORT_SYMBOL(ceph_destroy_client);
>
>
> --
> Jesper Juhl <jj@chaosbits.net>            http://www.chaosbits.net/
> Don't top-post http://www.catb.org/~esr/jargon/html/T/top-post.html
> Plain text mails only, please.
>
> --
> 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/
>
Jesper Juhl - Dec. 25, 2010, 9:24 p.m.
On Sat, 25 Dec 2010, richard -rw- weinberger wrote:

> On Sat, Dec 25, 2010 at 7:17 PM, Jesper Juhl <jj@chaosbits.net> wrote:
> > Hello,
> >
> > In net/ceph/ceph_common.c::ceph_destroy_client() the pointer 'client' is
> > freed by kfree() and subsequently used in a call to dout() - use after
> > free bug.
> 
> Not really. %p reads only the address of "client".
> kfree() does not alter this address.
> 

Ok, I see your point and you are correct. But still, the patch does not 
change behaviour and it makes it absolutely clear that there's no 
use-after-free bug, so it might still have merit... or?
Jesper Juhl - Dec. 25, 2010, 9:35 p.m.
On Sat, 25 Dec 2010, richard -rw- weinberger wrote:

> On Sat, Dec 25, 2010 at 10:24 PM, Jesper Juhl <jj@chaosbits.net> wrote:
> > On Sat, 25 Dec 2010, richard -rw- weinberger wrote:
> >
> >> On Sat, Dec 25, 2010 at 7:17 PM, Jesper Juhl <jj@chaosbits.net> wrote:
> >> > Hello,
> >> >
> >> > In net/ceph/ceph_common.c::ceph_destroy_client() the pointer 'client' is
> >> > freed by kfree() and subsequently used in a call to dout() - use after
> >> > free bug.
> >>
> >> Not really. %p reads only the address of "client".
> >> kfree() does not alter this address.
> >>
> >
> > Ok, I see your point and you are correct. But still, the patch does not
> > change behaviour and it makes it absolutely clear that there's no
> > use-after-free bug, so it might still have merit... or?
> 
> Your patch does not fix a bug.
> I would say it's a style fix.
> 
At this point in time I'd agree. :-)
richard -rw- weinberger - Dec. 25, 2010, 9:40 p.m.
On Sat, Dec 25, 2010 at 10:24 PM, Jesper Juhl <jj@chaosbits.net> wrote:
> On Sat, 25 Dec 2010, richard -rw- weinberger wrote:
>
>> On Sat, Dec 25, 2010 at 7:17 PM, Jesper Juhl <jj@chaosbits.net> wrote:
>> > Hello,
>> >
>> > In net/ceph/ceph_common.c::ceph_destroy_client() the pointer 'client' is
>> > freed by kfree() and subsequently used in a call to dout() - use after
>> > free bug.
>>
>> Not really. %p reads only the address of "client".
>> kfree() does not alter this address.
>>
>
> Ok, I see your point and you are correct. But still, the patch does not
> change behaviour and it makes it absolutely clear that there's no
> use-after-free bug, so it might still have merit... or?

Your patch does not fix a bug.
I would say it's a style fix.

> --
> Jesper Juhl <jj@chaosbits.net>            http://www.chaosbits.net/
> Don't top-post http://www.catb.org/~esr/jargon/html/T/top-post.html
> Plain text mails only, please.
>
>
Dan Carpenter - Dec. 25, 2010, 10:46 p.m.
On Sat, Dec 25, 2010 at 10:24:57PM +0100, Jesper Juhl wrote:
> On Sat, 25 Dec 2010, richard -rw- weinberger wrote:
> 
> > On Sat, Dec 25, 2010 at 7:17 PM, Jesper Juhl <jj@chaosbits.net> wrote:
> > > Hello,
> > >
> > > In net/ceph/ceph_common.c::ceph_destroy_client() the pointer 'client' is
> > > freed by kfree() and subsequently used in a call to dout() - use after
> > > free bug.
> > 
> > Not really. %p reads only the address of "client".
> > kfree() does not alter this address.
> > 
> 
> Ok, I see your point and you are correct. But still, the patch does not 
> change behaviour and it makes it absolutely clear that there's no 
> use-after-free bug, so it might still have merit... or?
> 

I see these with Smatch as well.  This type of usage is quite common.
People do it deliberately and I guess they feel it's readable.  Don't
change them.

If it were something that a static checker couldn't figure out, then
I'd say change it, but really the static checkers should just be made
smarter.  Some day I'm going to make Smatch complain if it's a %s in
the string instead of a %p, but for now I just ignore the false
positives.

regards,
dan carpenter
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/net/ceph/ceph_common.c b/net/ceph/ceph_common.c
index f3e4a13..890bbbf 100644
--- a/net/ceph/ceph_common.c
+++ b/net/ceph/ceph_common.c
@@ -408,8 +408,8 @@  void ceph_destroy_client(struct ceph_client *client)
 
 	ceph_destroy_options(client->options);
 
-	kfree(client);
 	dout("destroy_client %p done\n", client);
+	kfree(client);
 }
 EXPORT_SYMBOL(ceph_destroy_client);