[OE-core] [PATCH v4 0/6] systemd patches

[Jonas Bonn]

Hi Richard,

On 30/01/2019 22:54, Richard Purdie wrote:
> On Mon, 2019-01-28 at 21:58 +0100, Jonas Bonn wrote:
>> Changed in v4:
>> - add patch to make systemd-firstboot a non-default option to systemd
>> to
>>    prevent unexpected prompts at runtime
> 
> There were still some failures:
> 
> https://autobuilder.yoctoproject.org/typhoon/#/builders/72/builds/237
> 
> (steps 5c, 6c and 7c)

OK, thanks.

I looked into these failures and have a couple of comments:

i)  There are seemingly two failures here:  unable to sync time and 
unable to connect to network (by looks of things).  These are related 
because the network failure leads to the timesync failure, AFAICT.

ii)  I have seen these failures locally; however, I even get these 
failures on origin/master, i.e. without any of my systemd patches.

iii)  The failure behaviour on this end is strange:  sometimes I get 
warnings from the ethernet driver about "incomplete frames" being 
detected.  Same kernel on the same hardware with Thud doesn't produce 
these warnings (with or without the systemd patches in this series).

iv)  connman-1.36 crashes (something about Wispr despite 'wispr' being 
disabled in my connman build... it seems bits of wispr are built despite 
this) immediately after getting a DHCP address and is unable to get an 
address when it's restarted.

v)  Disabling connman altogether and things work much better. 
systemd-network can bring up the network without ethernet driver frame 
errors, strangely enough.

So with that:

a)  Does buildbot use connman, systemd-networkd, both, or something 
else?  How do I find this out?

b)  I'll poke at the patch series again once I get a working 
origin/master build so that I have sane state to work from.  The systemd 
patches work fine on Thud... I suspect the problem lies elsewhere.

c)  Are others seeing similar errors with connman?

Thanks,
/Jonas

> 
> Cheers,
> 
> Richard
>