Re: safety of *mutex_unlock() (Was: [BUG] signal: sighand unprotected when accessed by /proc)

[Oleg Nesterov]

On 06/09, Steven Rostedt wrote:
>
> On Mon, 9 Jun 2014 20:15:53 +0200
> Oleg Nesterov <oleg@xxxxxxxxxx> wrote:
>
> > > That would indeed be a bad thing, as it could potentially lead to
> > > use-after-free bugs.  Though one could argue that any code that resulted
> > > in use-after-free would be quite aggressive.  But still...
> >
> > And once again, note that the normal mutex is already unsafe (unless I missed
> > something).
>
> Is it unsafe?

Only in a sense that UNLOCK is not atomic.

IOW, you can't, say, declare a mutex or semaphore on stack, and use lock/unlock
to serialize with another thread.

But rt_mutex seems fine in this case, and for example rcu_boost() does this.
I do not know if this is by design or not, and can we rely on this or not.

> This thread was started because of a bug we triggered in -rt, which
> ended up being a change specific to -rt that modified the way slub
> handled SLAB_DESTROY_BY_RCU. What else was wrong with it?

And I specially changed the subject to avoid the confusion with
SLAB_DESTROY_BY_RCU bug we discussed before, but apparently I need to
apologize for confusion again ;)

But. Note that if rt_mutex is changed so that UNLOCK becomes non-atomic
in a sense above, then lock_task_sighand()/unlock_task_sighand() will be
buggy in -rt.

Oleg.

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