Re: [PATCH] fix race in mark_mounts_for_expiry()

[Miklos Szeredi]

> > How about this patch?  It tries to solve this race without additional
> > locking.  If refcount is already zero, it will increment and decrement
> > it.  So be careful to only call grab_namespace() with vfsmount_lock
> > held, otherwise it could race with itself.  (vfsmount_lock is also
> > needed in this case so that mnt->mnt_namespace, doesn't change, while
> > grabbing the namespace)
> 
> How about using cmpxchg?

How?  If the count is nonzero, an incremented count must be stored.
You can't do that atomically with cmpxchg.

Miklos

-
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/