On Tue, Nov 21, 2017 at 01:32:42PM +0000, Steve McIntyre wrote:
>On Tue, Nov 21, 2017 at 02:27:36PM +0100, Nicolas Dechesne wrote:
>>On Tue, Nov 21, 2017 at 2:21 PM, Steve McIntyre <steve.mcintyre@linaro.org>
>>wrote:
>>    On Tue, Nov 21, 2017 at 12:49:42PM +0000, Milosz Wasilewski wrote:
>>    >>
>>    >> Ideally, the Android builds would have a rootfs configured to be as
>>    >> small as possible - it's automatically resized to fit the available
>>    >> space at first boot, and leaving space otherwise is
>>    >> inefficient. However, if we're going to add the LAVA test overlay into
>>    >> the rootfs, space needs to be left for it to fit. That's not
>>    >
>>    >Could you give a bit more context/use case on why the overlay is
>>    >needed? I have similar discussions with YongQin and the main issue he
>>    >raises is that binaries in such potential overlay might not be
>>    >compatible with rootfs that is used. So I'm wondering if the overlay
>>    >only contains scripts or the mentioned limitation was somehow solved?
>>
>>    The overlay we're talking about here is the LAVA test overlay with the
>>    test scripts, requested git repos etc. I'd expect it to be quite rare
>>    that you'd have compiled binaries in it, but it's possible - it
>>    depends on the particular test use cases.
>>
>>
>>and the main issue is that we (e.g. the one who creates the image) does not
>>know in advance what the size of the overlay will be, since the size is
>>dependent on who is going to write the tests. So the ask, is for LAVA to
>>increase the resize the image if needed. 
>
>Nod. We're putting stories in for this today. I'll post links shortly.

See https://projects.linaro.org/browse/LAVA-1141 . The exact delivery
date is still open to change as we prioritise work, but hopefully this
captures what we need to do.

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