Re: workqueue deadlock

[Srivatsa Vaddagiri]

On Sun, Dec 10, 2006 at 09:26:16AM +0100, Ingo Molnar wrote:
> something like the pseudocode further below - when applied to a data
> structure it has semantics and scalability close to that of
> preempt_disable(), but it is still preemptible and the lock is specific.

Ingo,
	The psuedo-code you have provided can still fail to avoid
the deadlock reported by Bjorn Helgaas earlier in this thread:

	http://lkml.org/lkml/2006/12/6/352

Thread1->flush_workqueue->mutex_lock(cpu4's hotplug_lock)

Thread2(keventd)->run_workqueue->som_work_fn-> ..
		flush_workqueue->mutex_lock(cpu4's hotplug_lock)

Both deadlock with each other.

All this mess could easily be avoided if we implement a reference-count
based cpu_hotplug_lock(), as suggested by Arjan and Linus before and
implemented by Gautham here:

	http://lkml.org/lkml/2006/10/26/65

-- 
Regards,
vatsa
-
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/