[Lava-users] Customizing the deployment and booting actions

Steve McIntyre steve.mcintyre at linaro.org
Wed Nov 8 16:38:57 UTC 2017

Hi Alfonso,

On Wed, Nov 08, 2017 at 04:24:41PM +0000, Ros Dos Santos, Alfonso (CT RDA DS EVO OPS DIA SE 1) wrote:
>in our current project we have some devices that are not "directly" 
>supported by lava. I would like to ask for you opinion on which would be 
>the correct way to proceed.
>The main problem is that our devices have a EFI implementation in the 
>firmware that is making the task of installing uboot very hard. To avoid 
>this, we thought about serving the image through an "emulated" usb 
>stick. On that regard, we made some progress by setting up a secondary 
>device that would use the linux g_mass_storage module to serve the image 
>through the USB otg port.
>Our setup is then:
>1) The actual testing device
>2) The secondary device which only serves the image with g_mass_storage
>3) The host machine running the lava-slave application.
>We thought that we could add a new device-type template to the lava 
>server that would somehow override the deployment and boot actions to 
>address our setup. We tried to look into the base.jinja2 file for some 
>sort of entry point that would allow us to, for example, run a script 
>that would first send the image to the secondary device and secondly run 
>the g_mass_storage module with the image file.

It sounds like you're trying to do something like what's supported via
LAVA's "secondary media" support?


although you might be adding more complexity to your tests that
way. If your devices already boot via UEFI, could you configure them
to net-boot from UEFI instead, and run your tests that way?

Steve McIntyre                                steve.mcintyre at linaro.org
<http://www.linaro.org/> Linaro.org | Open source software for ARM SoCs

More information about the Lava-users mailing list