Re: enigmail FTBFS on mipsel (RFH)

[Willi Mann <foss-ml@wm1.at>]

Am 2010-06-03 13:23, schrieb Shane McDonald:
> On Tue, Jun 1, 2010 at 11:38 PM, Willi Mann <foss-ml@wm1.at> wrote:
>>
>> Could anyone with a testing/unstable mipsel box please download the current
>> enigmail sources and try to compile it? I'd like to know whether the build
>> error message "Bus error" is a temporary problem on this buildd or a
>> permanent problem.
>>
>> https://buildd.debian.org/fetch.cgi?pkg=enigmail;ver=2%3A1.0.1-2;arch=mipsel;stamp=1275423629
> 
> I just tried building it in a sid pbuilder on a mipsel box running the
> 2.6.34 kernel,
> and I see it crash in the exact same place.  With my system, though,
> it reports a "Segmentation fault" rather than a "Bus error".
> 
>   inflating: debian/enigmail/usr/lib/xul-ext/enigmail/platform/Linux_/components/libenigmime-.so
>   inflating: debian/enigmail/usr/lib/xul-ext/enigmail/chrome.manifest
>   inflating: debian/enigmail/usr/lib/xul-ext/enigmail/install.rdf
> Segmentation fault
> make[1]: *** [override_dh_install] Error 139
> make[1]: Leaving directory `/tmp/buildd/enigmail-1.0.1'
> make: *** [binary] Error 2
> dpkg-buildpackage: error: fakeroot debian/rules binary gave error exit status 2
> 
> Looks like a permanent problem, rather than a temporary buildd problem.

Thank you, this is interesting. In the mean time, I've set up a qemu
mipsel installation running 2.6.26, where it also ends with "Bus error".
So I found out that the "Bus error" problem happens in a python library,
python-rdflib. (Bug#584327)

I'm now trying to find out which package's fault it is. It could even be
a now fixed compiler problem, because a test case I created succeeds
with a rebuilt rdflib (albeit with noopt).

WM