[Pkg-e-devel] Bug#750828: libedje-bin: edje_cc (was: Re: [pkg-fso-maint] Bug#750828: libphone-ui-shr: FTBFS on powerpc but built there in the past (blocks testing migration))

[Albin Tonnerre]

On Sun, Jun 8, 2014 at 2:10 PM, Niels Thykier <niels at thykier.net> wrote:
> severity 750828 important
> retitle 750828 libedje-bin: edje_cc sometimes triggers SIGILL on buildds
> reassign libedje-bin 1.8.6-2
> affects 750828 + libphone-ui-shr e17
> thanks
>
> On 2014-06-08 13:02, Albin Tonnerre wrote:
>> [...]
>>
>> It appears I'm having the same issue with the latest upload of e17
>> (0.17.6) on powerpc. The funny thing though is that edje_cc is the
>> exact same version as the previous time, which built correctly. I've
>> tried reproducing the failure on the ppc porterbox, to no avail;
>> things work fine there. I don't really know what's next; do you have
>> any suggestion on what we could try?
>>
>> Thanks,
>>
>
> @Albin: I have retired to libphone-ui-shr build on powerpc and now it
> works.  I have tried to reschedule e17 - we will see how that goes.

Looks like it failed again :/

> Dear maintainer(s) of efl  / libedje-bin,
>
> We have noticed that edje_cc on powerpc /sometimes/ triggers a SIGILL,
> which causes e17 and libphone-ui-shr to FTBFS randomly[1].  I have
> therefore reassigned this bug to you.

As I can't reproduce the failure on the ppc porterbox, do you if
there's any way to find out exactly what instruction causes the SIGILL
on the buildds?

Thanks,
-- 
Albin