[LITMUS^RT] [LITMUS-RT] Modified pfp_schedule for migration

[Sebastiano Catellani]

I have already looked into that, but as you suggested every function has
its context and I suppose that there are a lot of differences among the
schedul_plugin and the litmus_lock_ops callbacks.

I probably need to release the lock on the local cpu before calling the
migration function.

Another thing that isn't unclear to me is why the error generated by the
task set is raised by the cpu 0, which should not be affected by any
migration, instead of cpu 1 or 2.


2014/1/13 Björn Brandenburg <bbb at mpi-sws.org>

>
> On 13 Jan 2014, at 20:10, Sebastiano Catellani <zebganzo at gmail.com> wrote:
>
> > First of all thanks for the answer, I know that It isn't correct to make
> a task migrate from a cpu to another in a partitioned system, but It's a
> prototype for a more complex scheduler that allows a task to migrate only
> when particular conditions are met.
>
> Ah, ok, I see.
>
> >
> > I understand my mistake, I will look under the hood hoping to find out
> how to avoid deadlock by calling the migrate_to function.
>
> Have a look at how pfp_migrate_to() is used by the DPCP implementation
> (for which it was introduced).
>
> Regards,
> Björn
>
>
>
> _______________________________________________
> litmus-dev mailing list
> litmus-dev at lists.litmus-rt.org
> https://lists.litmus-rt.org/listinfo/litmus-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.litmus-rt.org/pipermail/litmus-dev/attachments/20140113/3e31b7a5/attachment.html>