Improving SMP Support for RTEMS (GSOC 2014)

[Sebastian Huber]

Hello,

the SMP support is work in progress so at the moment it is hard to find an area 
suitable for a GSoC project from my point of view.

On 2014-02-25 18:46, Pulkit Mittal wrote:
> Hi all,
>
> I am a 4th-year CSE BTech student and I wish to participating  the GSOC-2014. I
> liked the following ideas in the design header of 'Improvements to SMP support'
>
> Processor Affinity
> Global Scheduler

The scheduler API will change with the support for partitioned/clustered 
scheduling.

http://www.rtems.org/wiki/index.php?title=SMP#Clustered_Scheduling_2

I think RTEMS will be a good platform to experiment with hard read time SMP 
schedulers once the basic work is done.  All projects in this area are 
challenging since it requires a fair amount of knowledge of RTEMS internals.

With respect to processor affinity for threads I am skeptical if this is useful 
for schedulers with priorities:

http://www.rtems.org/wiki/index.php?title=SMP#Processor_Affinity

> SMP / ISR Locks

In terms of locks everything is done for now.  I have an SMP rwlock 
implementation from the last GSoC project that needs to be 
reviewed/tested/merged.  We have currently no use case for this lock, but this 
will likely change if I start with the fine grained locking support:

http://www.rtems.org/wiki/index.php?title=SMP#Fine_Grained_Locking

-- 
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.