[meta-chip] Yocto on the 9$ computer

[meta-chip] Yocto on the 9$ computer

[Andrei Gherzan]

[-- Attachment #1: Type: text/plain, Size: 170 bytes --]

Hi all,

Have a C.H.I.P. 9$ computer? It works with Yocto now.

http://layers.openembedded.org/layerindex/branch/master/layer/meta-chip/

Happy hacking!

Andrei

[-- Attachment #2: Type: text/html, Size: 336 bytes --]

Re: [meta-chip] Yocto on the 9$ computer

[Nicolas Aguirre]

2015-10-24 19:26 GMT+02:00 Andrei Gherzan <andrei@gherzan.ro>:
> Hi all,
>

Hi Andrei,

> Have a C.H.I.P. 9$ computer? It works with Yocto now.
>
> http://layers.openembedded.org/layerindex/branch/master/layer/meta-chip/
>
Good job.
IMO it make sense to add C.H.I.P support in meta-sunxi, don't you think ?

https://github.com/linux-sunxi/meta-sunxi

Regards,
Nicolas



> Happy hacking!
>
> Andrei
>
>
> --
> _______________________________________________
> yocto mailing list
> yocto@yoctoproject.org
> https://lists.yoctoproject.org/listinfo/yocto
>



-- 
Nicolas Aguirre
Mail: aguirre.nicolas@gmail.com
Web: http://www.calaos.fr
Blog: http://dev.enlightenment.fr/~captainigloo/

Re: [meta-chip] Yocto on the 9$ computer

[Andrei Gherzan]

On Sat, Oct 24, 2015 at 09:58:42PM +0200, Nicolas Aguirre wrote:
> 2015-10-24 19:26 GMT+02:00 Andrei Gherzan <andrei@gherzan.ro>:
> > Hi all,
> >
>
> Hi Andrei,
>
> > Have a C.H.I.P. 9$ computer? It works with Yocto now.
> >
> > http://layers.openembedded.org/layerindex/branch/master/layer/meta-chip/
> >
> Good job.
> IMO it make sense to add C.H.I.P support in meta-sunxi, don't you think ?
>
> https://github.com/linux-sunxi/meta-sunxi
>

Well. Temporary it is a separate layer. And this is mainly because of the
overhead you need for flashing the board. So I do see a benefit in keeping it
separately. We will see in time.

Thanks for feedback.

--
Andrei Gherzan

Re: [meta-chip] Yocto on the 9$ computer

[Alex J Lennon]

>> On 24 Oct 2015, at 21:13, Andrei Gherzan <andrei@gherzan.ro> wrote:
>> 
>>> On Sat, Oct 24, 2015 at 09:58:42PM +0200, Nicolas Aguirre wrote:
>>> 2015-10-24 19:26 GMT+02:00 Andrei Gherzan <andrei@gherzan.ro>:
>>> Hi all,
>> 
>> Hi Andrei,
>> 
>>> Have a C.H.I.P. 9$ computer? It works with Yocto now.
>>> 
>>> http://layers.openembedded.org/layerindex/branch/master/layer/meta-chip/
>> Good job.
>> IMO it make sense to add C.H.I.P support in meta-sunxi, don't you think ?
>> 
>> https://github.com/linux-sunxi/meta-sunxi
> 
> Well. Temporary it is a separate layer. And this is mainly because of the
> overhead you need for flashing the board. So I do see a benefit in keeping it
> separately. We will see in time.
> 
> Thanks for feedback.

Great work Andrei. I have a couple of CHiPs here and was trying to decide whether meta-sunxi would provide what was needed or whether we needed a custom layer.

I've been hugely impressed with what Docker+Resin gives me for application deployment management so I'd like to look at how easy it is to take your Linux/u-boot CHiP support and deploy via Resin. 

Are you at the point you have a base OS image for CHiP with Resin support? If not I would be interested in having a look at this with you.

Cheers,

Alex

> --
> Andrei Gherzan
> -- 
> _______________________________________________
> yocto mailing list
> yocto@yoctoproject.org
> https://lists.yoctoproject.org/listinfo/yocto

Re: [meta-chip] Yocto on the 9$ computer

[Andrei Gherzan]

On Sat, Oct 24, 2015 at 09:46:41PM +0100, Alex J Lennon wrote:
>
>
>
>
> Sent from my iPhone
> >> On 24 Oct 2015, at 21:13, Andrei Gherzan <andrei@gherzan.ro> wrote:
> >>
> >>> On Sat, Oct 24, 2015 at 09:58:42PM +0200, Nicolas Aguirre wrote:
> >>> 2015-10-24 19:26 GMT+02:00 Andrei Gherzan <andrei@gherzan.ro>:
> >>> Hi all,
> >>
> >> Hi Andrei,
> >>
> >>> Have a C.H.I.P. 9$ computer? It works with Yocto now.
> >>>
> >>> http://layers.openembedded.org/layerindex/branch/master/layer/meta-chip/
> >> Good job.
> >> IMO it make sense to add C.H.I.P support in meta-sunxi, don't you think ?
> >>
> >> https://github.com/linux-sunxi/meta-sunxi
> >
> > Well. Temporary it is a separate layer. And this is mainly because of the
> > overhead you need for flashing the board. So I do see a benefit in keeping it
> > separately. We will see in time.
> >
> > Thanks for feedback.
>
> Great work Andrei. I have a couple of CHiPs here and was trying to decide whether meta-sunxi would provide what was needed or whether we needed a custom layer.
>
> I've been hugely impressed with what Docker+Resin gives me for application deployment management so I'd like to look at how easy it is to take your Linux/u-boot CHiP support and deploy via Resin.
>
> Are you at the point you have a base OS image for CHiP with Resin support? If not I would be interested in having a look at this with you.
>

We definitely thought about that already. The problem is that we currently rely
on a BTRFS partition for the docker runtime environment. This obviously won't
work on a flash raw device. So if you want to dig into this I can give you some
hints, otherwise we will just tackle it when we will get to it.

Andrei

--
Andrei Gherzan

Re: [meta-chip] Yocto on the 9$ computer

[Alex J Lennon]

On 24/10/2015 22:06, Andrei Gherzan wrote:
> On Sat, Oct 24, 2015 at 09:46:41PM +0100, Alex J Lennon wrote:
>>
>>
>>
>> Sent from my iPhone
>>>> On 24 Oct 2015, at 21:13, Andrei Gherzan <andrei@gherzan.ro> wrote:
>>>>
>>>>> On Sat, Oct 24, 2015 at 09:58:42PM +0200, Nicolas Aguirre wrote:
>>>>> 2015-10-24 19:26 GMT+02:00 Andrei Gherzan <andrei@gherzan.ro>:
>>>>> Hi all,
>>>> Hi Andrei,
>>>>
>>>>> Have a C.H.I.P. 9$ computer? It works with Yocto now.
>>>>>
>>>>> http://layers.openembedded.org/layerindex/branch/master/layer/meta-chip/
>>>> Good job.
>>>> IMO it make sense to add C.H.I.P support in meta-sunxi, don't you think ?
>>>>
>>>> https://github.com/linux-sunxi/meta-sunxi
>>> Well. Temporary it is a separate layer. And this is mainly because of the
>>> overhead you need for flashing the board. So I do see a benefit in keeping it
>>> separately. We will see in time.
>>>
>>> Thanks for feedback.
>> Great work Andrei. I have a couple of CHiPs here and was trying to decide whether meta-sunxi would provide what was needed or whether we needed a custom layer.
>>
>> I've been hugely impressed with what Docker+Resin gives me for application deployment management so I'd like to look at how easy it is to take your Linux/u-boot CHiP support and deploy via Resin.
>>
>> Are you at the point you have a base OS image for CHiP with Resin support? If not I would be interested in having a look at this with you.
>>
> We definitely thought about that already. The problem is that we currently rely
> on a BTRFS partition for the docker runtime environment. This obviously won't
> work on a flash raw device. So if you want to dig into this I can give you some
> hints, otherwise we will just tackle it when we will get to it.
>
>

Are you making use of any BTRFS specific features or should be it fairly
straightforward to use JFFS2 or similar?

Presumably BTRFS will sit happily on other devices with eMMC?

Best, Alex

Re: [meta-chip] Yocto on the 9$ computer

[Khem Raj]

[-- Attachment #1: Type: text/plain, Size: 1420 bytes --]

Andrei

good work.

> On Oct 24, 2015, at 1:13 PM, Andrei Gherzan <andrei@gherzan.ro> wrote:
> 
> On Sat, Oct 24, 2015 at 09:58:42PM +0200, Nicolas Aguirre wrote:
>> 2015-10-24 19:26 GMT+02:00 Andrei Gherzan <andrei@gherzan.ro>:
>>> Hi all,
>>> 
>> 
>> Hi Andrei,
>> 
>>> Have a C.H.I.P. 9$ computer? It works with Yocto now.
>>> 
>>> http://layers.openembedded.org/layerindex/branch/master/layer/meta-chip/
>>> 
>> Good job.
>> IMO it make sense to add C.H.I.P support in meta-sunxi, don't you think ?
>> 
>> https://github.com/linux-sunxi/meta-sunxi
>> 
> 
> Well. Temporary it is a separate layer. And this is mainly because of the
> overhead you need for flashing the board. So I do see a benefit in keeping it
> separately. We will see in time.

what does board flashing has to do with layer infrastructure ?
when creating a new layer all we should think that it should make it easy for end user
to use the ecosystem and reduce confusion. That also might mean that you may have different layers
for chips coming from  same SOC vendor but it is contextual, e.g. if you are using same
u-boot and kernel trees as rest of sunxi then you better align it in sunxi layer e.g.

> 
> Thanks for feedback.
> 
> --
> Andrei Gherzan
> --
> _______________________________________________
> yocto mailing list
> yocto@yoctoproject.org
> https://lists.yoctoproject.org/listinfo/yocto


[-- Attachment #2: Message signed with OpenPGP using GPGMail --]
[-- Type: application/pgp-signature, Size: 211 bytes --]

Re: [meta-chip] Yocto on the 9$ computer

[Trevor Woerner]

On 10/24/15 15:58, Nicolas Aguirre wrote:
> 2015-10-24 19:26 GMT+02:00 Andrei Gherzan <andrei@gherzan.ro>:
>
>> Have a C.H.I.P. 9$ computer? It works with Yocto now.
>>
>> http://layers.openembedded.org/layerindex/branch/master/layer/meta-chip/
>>
> Good job.
> IMO it make sense to add C.H.I.P support in meta-sunxi, don't you think ?

Interesting: I had the exact same thought, had started to compose a
reply, but then thought about layers such as meta-beagleboard and
meta-raspberrypi :-)

Unfortunately there isn't much oversight/guidance in BSP-land. It's
entirely understandable (the work involved would be monstrous, and the
existing infrastructure can barely keep up with the tiny handful of BSPs
that are currently used for validation) but can be a bit frustrating
nonetheless.

Do we create layers for every board (meta-chip, meta-raspberrypi,
meta-beagleboard) or do we create layers for SoCs themselves
(meta-sunxi, meta-atmel, meta-rockchip)[1]? Currently there is no
"correct" answer. Unfortunately the distinction between SoCs, boards,
MACHINEs, DEFAULTTUNEs, and DISTROs isn't as clear and as clean as one
would like.

:-)
    Trevor



[1] and let's not forget layers such as meta-ettus and meta-gumstix:
companies who make various boards with various SoCs!

Re: [meta-chip] Yocto on the 9$ computer

[Trevor Woerner]

On 10/24/15 18:14, Trevor Woerner wrote:
> Unfortunately the distinction between SoCs, boards,
> MACHINEs, DEFAULTTUNEs, and DISTROs isn't as clear and as clean as one
> would like.

Oops! The distinction between these ideas is clear, what isn't clear is
how to divide them up into layers (imho).