Re: [RFC PATCH v5 1/2] Use lock_device_hotplug() in cpu_probe_store() and cpu_release_store()

[Li Zhong <zhong@linux.vnet.ibm.com>]

On Wed, 2014-04-23 at 10:39 -0400, Tejun Heo wrote:
> On Wed, Apr 23, 2014 at 10:00:26AM +0800, Li Zhong wrote:
> > If you remove cpu0, then the cpu0 directory will be removed, together
> > with the "online" file in the directory, while some other process might
> > be writing 0 or 1 to it.
> > 
> > But here, for the probe/release, take "release" for example, if user
> > writes something that stands for cpu0 to it, the cpu0 will be removed, 
> > and the cpu0 directory and the files under it will be removed. But
> > "release" itself is not removed.
> > 
> > They are attributes of cpu_subsys, not of some specific cpus.
> 
> OIC, so the file itself which triggers removal doesn't get removed.
> Hmmm... but can't you still fall into deadlock?  If on/offline takes
> the same lock under active protection which is also taken while
> removing the cpu files, it doesn't really matter whether the release
> file itself is removed in the process or not.  You can still have ABBA
> deadlock.  What am I missing here?

After thinking it harder, I still couldn't see ABBA here ... 

the active protection taken here is for "probe/release" which will not
be waited for removing something like "online" under cpu#? Or my
assumption that s_active for different files are different locks are
completely wrong? Or I missed something else? 

Thanks, Zhong
> 
> Thanks.
>