[yocto-ab] YP HW test strategy

[Philip Balister]

On 12/11/2013 08:45 AM, Richard Purdie wrote:
> On Wed, 2013-12-11 at 08:15 -0500, William Mills wrote:
>> Last meeting I was discussing YP testing on HW and suggesting we adopt LAVA.
>> Since that time:
>> * I have found out that Minnow board was already being worked on for 
>> inclusion in LAVA
>> * I formally requested for Linaro to complete the Minnow board LAVA 
>> integration happen
>> * Linaro has announced that the LMP board IP have been donated to the 
>> open compute project
>>      OCP plans to manufacture boards and give them away
>>      The IP can be obtained from OCP under an open HW license and OCP 
>> will be accepting/expecting patches
>>      [LMP are the little boards that help test a device under test, SD 
>> card mux, USB mux, HDMI mux etc]
>> [Both Beagleboard and Beaglebone black are already supported in LAVA]
>> * Khem Raj met with the LAVA team at ARM tech con and was impressed
>>      * Khem said he was interested in adopting LAVA for his own work
>>      * Khem said he was going to suggest adopting LAVA at the OE level

I also was at that meeting. Tyler Baker ran us quickly through the
capabilities of LAVA. The LAVA team has definitely worked through a lot
of the problems for a board level test system. I'd hate to have to go
through the process they have, because I think we would end up
duplicating a lot of their work

>>
>> I will be on the call today if we wish to discuss or we can continue here.
>>
>> Richard,
>> I am happy to make introductions between the LAVA team and YP if they 
>> have not already introduced themselves.
>> Who should be on the list for that?
> 
> I'm assuming you saw the email from Paul Eggleton where he evaluated the
> current position with LAVA and some of the other technologies out there?
> LAVA has some attractive features but also some downsides. I believe
> some of the LAVA people did see and respond to that email.
> 

The key point from Paul's email is that there are some features of LAVA
that need changing to make LAVA usable as a general purpose test system.
I do not see any of the issues as impossible technical problems.

> What isn't clear right now is who would be prepared to step up and
> address some of the downsides. The missing piece in getting strong
> hardware testing for YP has always been the engineering and I think
> whoever steps up and makes that happen is going to have the most
> influence in which technologies get used in what capacities.
> 

Richard is correct. Nothing is going to happen without people assigned
to work on the problem.

But step one is getting buy in from Linaro management agreeing that they
are interested in outside contributions and willing to work together
with the Yocto Project to generalize LAVA. (By generalize, I mean
improve installation methods, address things like the root issue Paul
mentioned, add support for more build types, improve documentation, etc)

I have no doubts that we could develop a good relationship at the level
of the people that would be doing the work, but they need management
support for these efforts.

> Being perfectly honest, right now we're struggling with resources even
> to tackle issues in the day to day running/bug fixing/updating of the
> core of the project. I'm planning to talk more about this on today's
> call.
> 

This is exactly why I would like to see the Yocto Project develop a
collaborative relationship with Linaro to develop LAVA into a gneral
purpose test environment, rather than develop something completely new.

Philip


> Cheers,
> 
> Richard
> 
> 
> 
> _______________________________________________
> yocto-ab mailing list
> yocto-ab at yoctoproject.org
> https://lists.yoctoproject.org/listinfo/yocto-ab
> 
>