[Xenomai] Is there a specification?

[Xenomai] Is there a specification?

[Eichenberg, Marius]

Hey folks, 
we're a team of two students at a university in Germany.

Currently we're trying to evaluate how easy it is to map tasks from Autosar (OSEK) to Xenomai. 
Therefore we're looking for a specification of Xenomai. Right now we're dealing with the API, but this is acutally a little bulky.

We havn't been able to find a specification on the internet, so now we need your help.
Is there a specification at all? Is it publicly available or at least for educational use?
Can somebody forward it to us?

Thanks in advance.
Best regards from Germany
Marius and George

Re: [Xenomai] Is there a specification?

[Gilles Chanteperdrix]

On Thu, Jan 08, 2015 at 03:13:10PM +0000, Eichenberg, Marius wrote:
> Hey folks, 
> we're a team of two students at a university in Germany.
> 
> Currently we're trying to evaluate how easy it is to map tasks
> from Autosar (OSEK) to Xenomai. Therefore we're looking for a
> specification of Xenomai. Right now we're dealing with the API,
> but this is acutally a little bulky.

The API documentation is big because it documents several
interfaces, the underlying common interface as well as several skins
implemented on top of this common interface. But implementing a new
interface such as OSEK means implementing a new skin, on top of the
common interface only.

For Xenomai 2.x, this common interface resides in kernel-space and
is documented here:
https://xenomai.org/documentation/xenomai-2.6/html/api/group__nucleus.html

For Xenomai 3.x (which is recommended for starting a new
development), this common interface moved to user-space, is based on
the POSIX interface and is documented here:
https://xenomai.org/documentation/xenomai-3/html/xeno3prm/group__alchemy.html

-- 
					    Gilles.

Re: [Xenomai] Is there a specification?

[Gilles Chanteperdrix]

On Thu, Jan 08, 2015 at 04:35:19PM +0100, Gilles Chanteperdrix wrote:
> On Thu, Jan 08, 2015 at 03:13:10PM +0000, Eichenberg, Marius wrote:
> > Hey folks, 
> > we're a team of two students at a university in Germany.
> > 
> > Currently we're trying to evaluate how easy it is to map tasks
> > from Autosar (OSEK) to Xenomai. Therefore we're looking for a
> > specification of Xenomai. Right now we're dealing with the API,
> > but this is acutally a little bulky.
> 
> The API documentation is big because it documents several
> interfaces, the underlying common interface as well as several skins
> implemented on top of this common interface. But implementing a new
> interface such as OSEK means implementing a new skin, on top of the
> common interface only.
> 
> For Xenomai 2.x, this common interface resides in kernel-space and
> is documented here:
> https://xenomai.org/documentation/xenomai-2.6/html/api/group__nucleus.html
> 
> For Xenomai 3.x (which is recommended for starting a new
> development), this common interface moved to user-space, is based on
> the POSIX interface and is documented here:
> https://xenomai.org/documentation/xenomai-3/html/xeno3prm/group__alchemy.html

Sorry, wrong link. The Alchemy interface is a skin, like the OSEK
skin, so you could use it as an example for implementing a new skin.

The common interface for user-space skins is now posix, with the
copperplate library masking differences between Xenomai posix
implementation and Linux posix implementation. The alchemy or
vxworks skins are example of using posix and copperplate to
implement a new skin.

-- 
					    Gilles.

Re: [Xenomai] Is there a specification?

[Eichenberg, Marius]

Hey Gilles, 
thanks for your fast response.

Actually we don't want to implement an own skin. 
We just want to compare the tasks from OSEK and the one from Xenomai (version 3 is preferred).

For now we were using the Task Management Service of the Alchemy API as a reference.
https://xenomai.org/documentation/xenomai-3/html/xeno3prm/group__alchemy__task.html

The API is good for a specific mapping of the methods, but we're missing an abstract overview.
Especially a conceptual description how Xenomai sees tasks,
e.g. which states are available, when are transitions firing?

Thank you a lot,
Cheers
________________________________________
Von: Gilles Chanteperdrix [gilles.chanteperdrix@xenomai.org]
Gesendet: Donnerstag, 8. Januar 2015 16:39
An: Eichenberg, Marius
Cc: xenomai@xenomai.org
Betreff: Re: [Xenomai] Is there a specification?

On Thu, Jan 08, 2015 at 04:35:19PM +0100, Gilles Chanteperdrix wrote:
> On Thu, Jan 08, 2015 at 03:13:10PM +0000, Eichenberg, Marius wrote:
> > Hey folks,
> > we're a team of two students at a university in Germany.
> >
> > Currently we're trying to evaluate how easy it is to map tasks
> > from Autosar (OSEK) to Xenomai. Therefore we're looking for a
> > specification of Xenomai. Right now we're dealing with the API,
> > but this is acutally a little bulky.
>
> The API documentation is big because it documents several
> interfaces, the underlying common interface as well as several skins
> implemented on top of this common interface. But implementing a new
> interface such as OSEK means implementing a new skin, on top of the
> common interface only.
>
> For Xenomai 2.x, this common interface resides in kernel-space and
> is documented here:
> https://xenomai.org/documentation/xenomai-2.6/html/api/group__nucleus.html
>
> For Xenomai 3.x (which is recommended for starting a new
> development), this common interface moved to user-space, is based on
> the POSIX interface and is documented here:
> https://xenomai.org/documentation/xenomai-3/html/xeno3prm/group__alchemy.html

Sorry, wrong link. The Alchemy interface is a skin, like the OSEK
skin, so you could use it as an example for implementing a new skin.

The common interface for user-space skins is now posix, with the
copperplate library masking differences between Xenomai posix
implementation and Linux posix implementation. The alchemy or
vxworks skins are example of using posix and copperplate to
implement a new skin.

--
                                            Gilles.

Re: [Xenomai] Is there a specification?

[Gilles Chanteperdrix]

On Thu, Jan 08, 2015 at 04:02:44PM +0000, Eichenberg, Marius wrote:
> Hey Gilles, 
> thanks for your fast response.
> 
> Actually we don't want to implement an own skin.

It is the way to go, if you want to run OSEK over Xenomai. Note that
the OSEK API is really lightweight, so going directly to an initial
implementation of this API would take somewhere around two weeks to
someone who knows the job. There is no need to turn around the
problem and ask useless questions, if something is missing, you can
add it to Xenomai core.

> We just want to compare the tasks from OSEK and the one from
> Xenomai (version 3 is preferred).

I am not sure how this is going to achieve anything.

> 
> For now we were using the Task Management Service of the Alchemy API as a reference.
> https://xenomai.org/documentation/xenomai-3/html/xeno3prm/group__alchemy__task.html
> 
> The API is good for a specific mapping of the methods, but we're
> missing an abstract overview.

The Alchemy API is an API equivalent to the OSEK API. This is not
what you should be looking at, if you want to implement an OSEK
skin. You should look at the implementation itself, not the Alchemy
API.

> Especially a conceptual description how Xenomai sees tasks,
> e.g. which states are available, when are transitions firing?

There is an explanation of all this in Xenomai white paper. Have you
not skimmed all the available documentation at:
https://www.xenomai.org 

?

Please do not top-post.
Regards.

-- 
					    Gilles.