[PATCH 11/11] smp: Move secondary CPU initialization request

[Sebastian Huber]

On 05/27/2013 06:12 PM, Joel Sherrill wrote:
> Are you sure it was a scheduling bug and not a dispatch disable nesting problem? If the count does not go back to zero, then a blocked thread will continue to execute.

Yes, I am sure that it is a scheduling bug.  I reach a state in which the 
executing task is not ready after the dispatch loop in _Thread_Dispatch().

>
> If it is a bug I would like to know the detailed scenario so I can reproduce it in the simulator.

The bug can be reproduced with smptests/smp08 on Qemu.

After the current patch set review and integration I will propose a new 
spinlock API.  After that I am ready to commit the ARM SMP support with the 
existing scheduler.  We can then look this problem more closely.

I have a working alternative simple scheduler.

-- 
Sebastian Huber, embedded brains GmbH

Address : Dornierstr. 4, D-82178 Puchheim, Germany
Phone   : +49 89 189 47 41-16
Fax     : +49 89 189 47 41-09
E-Mail  : sebastian.huber at embedded-brains.de
PGP     : Public key available on request.

Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG.