[okl4-developer] Cell isolation

[Xavier.LANGELLIER at fr.thalesgroup.com]

Hi,

Thanks for your answer.

I understand that the Linux freeze can be caused by a higher priority of the
other cell but I have not found where to change priority setting in this
example.
Secondly, when I generate a segmentation fault in the decrypt cell, I don't
understand why the Linux freezes too, I supposed that the decrypt cell is no
longer scheduled as soon as it crashes, so it shouldn't block the Linux
cell. How is it possible?

Regards,
Xavier. 

-----Message d'origine-----
De : Josh Matthews [mailto:jmatthews at ok-labs.com] 
Envoyé : vendredi 10 juillet 2009 00:11
À : Xavier.LANGELLIER at fr.thalesgroup.com
Cc : developer at okl4.org
Objet : Re: [okl4-developer] Cell isolation

Hi Xavier,

On Fri, July 3, 2009 10:40 am, Xavier.LANGELLIER at fr.thalesgroup.com wrote:
> Hello,
>
> I would like to know what is the isolation policy between cells in OKL4
> examples (and how to change it).
> Indeed I modified the decrypt example distributed with OKL4 to crash the
> decrypt cell, and as soon as the OKL4 cell crashes, the Linux cell freezes
> too.
> Furthermore, when I try to block the OKL4 celle with a while(1)}; the OKL4
> cell still freezes.
> Since I try to crash/block the cell when there is no communication between
> the cells, I don't understand why the Linux cells is affected.
> Does anyone has an answer?

The isolation policy is very strong - cells can't impact each other unless
they have received a capability to communicate, or are sharing some
memory.

This points me to an issue with your cell priorities - is the blocked cell
(with the while(1); which is always doing work) at a higher priority than
Linux?

Cheers,
Josh