Bug#880979: libreoffice-calc: Crash after starting NLPsolver
severity 880979 grave
reassign 880979 src:linux
forcemerge 880979 865303
affects 880979 libreoffice-wiki-publisher
affects 880979 libreoffice-nlpsolver
affects 880979 libreoffice-calc
thanks
Hi,
On Mon, Nov 06, 2017 at 04:36:29PM +0100, Sulyok Attila wrote:
> I experinenced an immediate crach when starting NLPsolver extension.
The NLPsolver extension is in libreoffice-nlpsolver, not in -calc...
> I am regular user of this non-linear extension package.
> Recently, I updated my Libreoffice-set to 5.4.2 from stretch-backport.
... and your kernel, probably ... longer ago...
> I have worked on my sheets for a week, and now I started an NLPsolver step that caused a prompt crash.
Yes, known. It's Java.
> I repeated it a few times with the same crashing end.
> I checked the installation and everything was good.
> I deleted and reinstalled the NLP extension, but it did not help.
>
> A cleared the whole libreoffice installation and went back to the 5.2.7 version of the stable, beacause it worked well previously.
> I received the same crashing result.
Exactly. Because it's not LO at fault.
> Because the NLP package itself is not a new one, I assume there is some glitch in the debian package and/or there is
> some faulty error treating in the LibreOffice.
No, in Java and the kernel.
> Now I see two problems.
> 1. It is not a normal situation if the Calc of LibreOffice crashes when something is not proper in its environment.
Correct.
> 2. What do I need to do to restore my LibreOffice ability to use the NLP again ?
Fix the Linux kernel and/or Java. Or use amd64 instead of i386. This is
a know problem and doesn't happen on amd64.
See
#865303 [G|u|=☣] [src:linux] libreoffice: Libreoffice Java features crash with Linux 3.16.43-2+deb8u1
(and the bugs merged with it)
#876051 [i|+|☣] [openjdk-8-jre-headless] i386 applications using JNI may crash due to Hotspot workaround for Exec Shield
#876068 [i|+|☣] [openjdk-7-jre-headless] i386 applications using JNI may crash due to Hotspot workaround for Exec Shield
and
#876069 [i|+|☺☣] [openjdk-9-jre-headless] i386 applications using JNI may crash due to Hotspot workaround for Exec Shield
(I just noticed that this fixed, not available in stable yet. Maybe I should
force that newe(er) OpenJDK 9 for i386 for buster/sid/stretch-backports if it's
available there.)
See also the press coverage months ago:
https://lwn.net/Articles/727206/
https://lwn.net/Articles/727703/
Regards,
Rene
Reply to: