Re: [PATCH 1/4] lockdep: lock_set_subclass() fix

From: Yong Zhang
Date: Tue Nov 08 2011 - 04:37:52 EST


On Tue, Nov 08, 2011 at 05:07:35PM +0800, Yong Zhang wrote:
> On Tue, Nov 08, 2011 at 09:46:20AM +0100, Peter Zijlstra wrote:
> > On Tue, 2011-11-08 at 16:14 +0800, Yong Zhang wrote:
> > >
> > > But how do we deal with ->class_cache? Always set it in
> > > loop_up_lock_class()?
> >
> > Hrm.. good point, aside from that there's another problem as well, I
> > think we can deal with the cache being NULL, but is memset() an atomic
> > write? If not a read could observe an intermediate state and go funny.
>
> Yup.
>
> >
> > I'm tempted to go with the pure kmemcheck_mark_initialized() thing for
> > now.
>
> me too :)
>
> So something like below?
>
> Thanks,
> Yong
> ---
> From: Yong Zhang <yong.zhang0@xxxxxxxxx>
> Subject: [PATCH] lockdep: kmemcheck: annotate ->lock in lockdep_init_map()
>
> Since commit f59de89 [lockdep: Clear whole lockdep_map on initialization],
> lockdep_init_map() will clear all the struct. But it will break
> lock_set_class()/lock_set_subclass(). A typical race condition
> is like below:
>
> CPU A CPU B
> lock_set_subclass(lockA);
> lock_set_class(lockA);
> lockdep_init_map(lockA);
> /* lockA->name is cleared */
> memset(lockA);
> __lock_acquire(lockA);
> /* lockA->class_cache[] is cleared */
> register_lock_class(lockA);
> look_up_lock_class(lockA);
> WARN_ON_ONCE(class->name !=
> lock->name);
>
> lock->name = name;
>
> So annotate ->lock with kmemcheck_mark_initialized() to cure this problem
> and the one reported in commit f59de89.
>
> Reported-by: Sergey Senozhatsky <sergey.senozhatsky@xxxxxxxxx>
> Reported-by: Borislav Petkov <bp@xxxxxxxxx>
> Suggested-by: Vegard Nossum <vegard.nossum@xxxxxxxxx>
> Signed-off-by: Yong Zhang <yong.zhang0@xxxxxxxxx>
> Cc: Peter Zijlstra <a.p.zijlstra@xxxxxxxxx>
> Cc: Ingo Molnar <mingo@xxxxxxx>
> Cc: Tejun Heo <tj@xxxxxxxxxx>
> Cc: David Rientjes <rientjes@xxxxxxxxxx>
> ---
> kernel/lockdep.c | 2 +-
> 1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/kernel/lockdep.c b/kernel/lockdep.c
> index e69434b..08a2b1b 100644
> --- a/kernel/lockdep.c
> +++ b/kernel/lockdep.c
> @@ -2948,7 +2948,7 @@ static int mark_lock(struct task_struct *curr, struct held_lock *this,
> void lockdep_init_map(struct lockdep_map *lock, const char *name,
> struct lock_class_key *key, int subclass)
> {
> - memset(lock, 0, sizeof(*lock));
> + kmemcheck_mark_initialized(lock, sizeof(*lock));

But for a lock in initializing stage (like spin_lock_init()), we still
need to guarantee that lock->class_cache[] doesn't contain garbage.

Need more thought here.

Thanks,
Yong

>
> #ifdef CONFIG_LOCK_STAT
> lock->cpu = raw_smp_processor_id();
> --
> 1.7.5.4
>

--
Only stand for myself
--
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/