Dual kernel made easy

Dual kernel made easy

[Philippe Gerum]

Motivated by the pain we continue to inflict upon ourselves everytime
we need to rebase the I-pipe over a recent kernel release, I started
working to define its future replacement. Now that we have an
implementation of its successor running reliably, my next step is to
rebase the Cobalt ARM port on this new interface called Dovetail [1].

This work is part of a larger effort to rethink the way independent
software cores like Xenomai's integrate into the main kernel [2].

This led to not only implement Dovetail, but also to showcase it with
a new real-time co-kernel originally derived from Cobalt [3], which is
simple and compact enough to be used as a learning tool for everyone
interested in the dual kernel technology based on Linux.

The purpose of this work is about making dual kernel easy, at least as
simple as possible for people to develop and maintain their own
software core of choice over the current mainline kernel.

[1] https://evenless.org/dovetail/
[2] https://evenless.org/
[3] https://evenless.org/core/

-- 
Philippe.

Re: Dual kernel made easy

[Jorge Ramirez]

On 4/1/19 22:11, Philippe Gerum wrote:
> Motivated by the pain we continue to inflict upon ourselves everytime
> we need to rebase the I-pipe over a recent kernel release, I started
> working to define its future replacement. Now that we have an
> implementation of its successor running reliably, my next step is to
> rebase the Cobalt ARM port on this new interface called Dovetail [1].
>
> This work is part of a larger effort to rethink the way independent
> software cores like Xenomai's integrate into the main kernel [2].
>
> This led to not only implement Dovetail, but also to showcase it with
> a new real-time co-kernel originally derived from Cobalt [3], which is
> simple and compact enough to be used as a learning tool for everyone
> interested in the dual kernel technology based on Linux.
>
> The purpose of this work is about making dual kernel easy, at least as
> simple as possible for people to develop and maintain their own
> software core of choice over the current mainline kernel.
>
> [1] https://evenless.org/dovetail/
> [2] https://evenless.org/
> [3] https://evenless.org/core/
>
Hi Philippe,

For a number of reasons - some personal and some professional - I am
interested in porting an IP stack to EVL.

I have asked around and it seems one of the most popular IP stacks
these days is LwIP [1]. In fact this stack seems to align well with
one of the goals you set for EVL, targeting simplicity.

Is this activity something you (or others) would be interested in
supporting?

I am also extending the question to anyone on the ML willing to take
part on this of course.

[0] https://savannah.nongnu.org/projects/lwip/
[1] http://www.nongnu.org/lwip/2_1_x/index.html

Re: Dual kernel made easy

[Philippe Gerum]

On 4/29/19 10:24 PM, Jorge Ramirez wrote:
> On 4/1/19 22:11, Philippe Gerum wrote:
>> Motivated by the pain we continue to inflict upon ourselves everytime
>> we need to rebase the I-pipe over a recent kernel release, I started
>> working to define its future replacement. Now that we have an
>> implementation of its successor running reliably, my next step is to
>> rebase the Cobalt ARM port on this new interface called Dovetail [1].
>>
>> This work is part of a larger effort to rethink the way independent
>> software cores like Xenomai's integrate into the main kernel [2].
>>
>> This led to not only implement Dovetail, but also to showcase it with
>> a new real-time co-kernel originally derived from Cobalt [3], which is
>> simple and compact enough to be used as a learning tool for everyone
>> interested in the dual kernel technology based on Linux.
>>
>> The purpose of this work is about making dual kernel easy, at least as
>> simple as possible for people to develop and maintain their own
>> software core of choice over the current mainline kernel.
>>
>> [1] https://evenless.org/dovetail/
>> [2] https://evenless.org/
>> [3] https://evenless.org/core/
>>
> Hi Philippe,
> 
> For a number of reasons - some personal and some professional - I am
> interested in porting an IP stack to EVL.
> 
> I have asked around and it seems one of the most popular IP stacks
> these days is LwIP [1]. In fact this stack seems to align well with
> one of the goals you set for EVL, targeting simplicity.
> 
> Is this activity something you (or others) would be interested in
> supporting?
> 

Yes, definitely, count me in. A lightweight IP stack running from
userland is something I had in mind for EVL too.

-- 
Philippe.

Re: Dual kernel made easy

[Greg Gallagher]

On Tue, Apr 30, 2019 at 5:19 AM Philippe Gerum via Xenomai
<xenomai@xenomai.org> wrote:
>
> On 4/29/19 10:24 PM, Jorge Ramirez wrote:
> > On 4/1/19 22:11, Philippe Gerum wrote:
> >> Motivated by the pain we continue to inflict upon ourselves everytime
> >> we need to rebase the I-pipe over a recent kernel release, I started
> >> working to define its future replacement. Now that we have an
> >> implementation of its successor running reliably, my next step is to
> >> rebase the Cobalt ARM port on this new interface called Dovetail [1].
> >>
> >> This work is part of a larger effort to rethink the way independent
> >> software cores like Xenomai's integrate into the main kernel [2].
> >>
> >> This led to not only implement Dovetail, but also to showcase it with
> >> a new real-time co-kernel originally derived from Cobalt [3], which is
> >> simple and compact enough to be used as a learning tool for everyone
> >> interested in the dual kernel technology based on Linux.
> >>
> >> The purpose of this work is about making dual kernel easy, at least as
> >> simple as possible for people to develop and maintain their own
> >> software core of choice over the current mainline kernel.
> >>
> >> [1] https://evenless.org/dovetail/
> >> [2] https://evenless.org/
> >> [3] https://evenless.org/core/
> >>
> > Hi Philippe,
> >
> > For a number of reasons - some personal and some professional - I am
> > interested in porting an IP stack to EVL.
> >
> > I have asked around and it seems one of the most popular IP stacks
> > these days is LwIP [1]. In fact this stack seems to align well with
> > one of the goals you set for EVL, targeting simplicity.
> >
> > Is this activity something you (or others) would be interested in
> > supporting?
> >
>
> Yes, definitely, count me in. A lightweight IP stack running from
> userland is something I had in mind for EVL too.
>
> --
> Philippe.
>

I'll make some time to help.

-Greg