Re: [RFC][PATCH 0/4] Redesign cpu_hotplug locking.

[Dipankar Sarma <dipankar@in.ibm.com>]

On Sun, Aug 27, 2006 at 10:21:16AM -0700, Andrew Morton wrote:
> On Sun, 27 Aug 2006 16:36:58 +0530
> Dipankar Sarma <dipankar@in.ibm.com> wrote:
> 
> > > Did you look?  workqueue_mutex is used to protect per-cpu workqueue
> > > resources.  The lock is taken prior to modification of per-cpu resources
> > > and is released after their modification.  Very very simple.
> > 
> > I did and there is no lock named workqueue_mutex. workqueue_cpu_callback()
> > is farily simple and doesn't have the issues in cpufreq that
> > we are talking about (lock_cpu_hotplug() in cpu callback path).
> 
> http://www.kernel.org/git/gitweb.cgi?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=9b41ea7289a589993d3daabc61f999b4147872c4

Ah, I didn't realize that it was already in git. It does take
care of create_workqueue callers, however I don't see why this
is needed - 

+ break;
+
+ case CPU_DOWN_PREPARE:
+ 	mutex_lock(&workqueue_mutex);
+ 	break;
+
+ case CPU_DOWN_FAILED:
+ 	mutex_unlock(&workqueue_mutex);
	break;

This seems like some implicit code locking to me. Why is it not
sufficient to hold the lock in the CPU_DEAD code while walking
the workqueues ?

Thanks
Dipankar