Evolution-data-server crashes with floating point exception

[Tristan Tarrant]

I'm using Hoary on amd64 and evolution-data-server 1.1.3-0ubuntu10 crashes with
a floating point exception, making it impossible to access calendars and contacts.
Here's what I get on syslog

evolution-data-[24384] trap divide error rip:2a9555e6cd rsp:7fbffff360 error:0
evolution-excha[24386] trap divide error rip:2a9555e6cd rsp:7fbffff350 error:0
evolution-data-[24388] trap divide error rip:2a9555e6cd rsp:7fbffff360 error:0
evolution-excha[24390] trap divide error rip:2a9555e6cd rsp:7fbffff350 error:0
evolution-data-[24392] trap divide error rip:2a9555e6cd rsp:7fbffff360 error:0
evolution-excha[24394] trap divide error rip:2a9555e6cd rsp:7fbffff350 error:0
evolution-data-[24396] trap divide error rip:2a9555e6cd rsp:7fbffff360 error:0
evolution-excha[24398] trap divide error rip:2a9555e6cd rsp:7fbffff350 error:0

[Tristan Tarrant (tristan-tarrant)]

I've upgraded to evolution 2.1.4-0ubuntu1 and this problem now happens with them
main evolution executable.

Here is the syslog:

evolution-2.2[10308] trap divide error rip:2a9555e6cd rsp:7fbffff630 error:0
evolution[10330] trap divide error rip:2a9555e6cd rsp:7fbffff3b0 error:0
evolution[10371] trap divide error rip:2a9555e6cd rsp:7fbffff390 error:0

[Tristan Tarrant (tristan-tarrant)]

Raised to blocker and P1 and changed package to evolution as Evolution itself
doesn't even startup now

[Blake Drager (artifact-alone)]

FWIW, I can corroborate on this bug.

Hoary on AMD64.

[Oliver Grawert (ogra)]

same here even the upgrade to evolution 2.1.4-0ubuntu2 didnt help

[Tristan Tarrant (tristan-tarrant)]

As I've mentioned on the forum, compiling evolution and evolution-data-server by
hand produced working executables. Could it be a problem with the build daemons ?

[Rob Adams (readams)]

If I try to build the source package, I get an whole lot of this:
gcc.real: ": No such file or directory
gcc.real: ": No such file or directory
cc1: error: bad value (Medium) for -march= switch
cc1: error: bad value (Medium) for -mcpu= switch
mv -f lock.o .libs/lock.lo
cc -c -I. -I./.. -I./../dbinc -D_GNU_SOURCE -D_REENTRANT -g -Wall -O2
../lock/lock.c >/dev/null 2>&1
mv -f .libs/lock.lo lock.lo
/bin/sh ./libtool --mode=compile cc -c -I. -I./.. -I./../dbinc -D_GNU_SOURCE
-D_REENTRANT -g -Wall -O2 ../lock/lock_deadlock.c
rm -f .libs/lock_deadlock.lo
cc -c -I. -I./.. -I./../dbinc -D_GNU_SOURCE -D_REENTRANT -g -Wall -O2
../lock/lock_deadlock.c -fPIC -DPIC
gcc.real: ": No such file or directory
gcc.real: ": No such file or directory
cc1: error: bad value (Medium) for -march= switch
cc1: error: bad value (Medium) for -mcpu= switch

[Jason Toffaletti (jason)]

I've been fighting with this issue for a while. See my posts on
https://bugzilla.ubuntu.com/show_bug.cgi?id=3619

[Gabriel de Perthuis (g2p)]

I've finally been able to build an evolution that launches, when my binaries got
SIGFPE before reaching main one or two weeks ago.
I used gcc-3.4 and dpkg-buildpackage in both cases; this time I have a fully
up-to-date hoary too.
I'll check if that works with gcc-3.3 too.

[Gabriel de Perthuis (g2p)]

There is no problem with gcc-3.3

[Jim McLaughlin (jimmclaughlin)]

recompiling evolution and evolution-data-server with gcc-3.3 and gcc-3.4 does
not help for me. I still get a floating point exception.

[Gabriel de Perthuis (g2p)]

You're right; I tried to do it again after upgrading to 2.1.5 and e-d-s 1.2.5,
and it didn't work anymore. I tried with gcc 3.3, 3.4 and 4.0 (!). For now I am
stuck with a chrooted verion - see
http://www.ubuntulinux.org/wiki/DebootstrapChroot on the wiki.
There is a report at ximian's bugzilla too, they should now have enough info to
reproduce the bug: http://bugzilla.ximian.com/show_bug.cgi?id=71776 .

[Matt Zimmerman (mdz)]

I believe Tollef was looking at this; any update?

[Sebastien Bacher (seb128)]

should be fixed now (phtread/krb4 issue)