Bug 129689

Summary:	Crash in: libmergedlo.so launch libreoffice from a script FILEOPEN
Product:	LibreOffice	Reporter:	Joona Tolonen <joona.tolonen>
Component:	Impress	Assignee:	Not Assigned <libreoffice-bugs>
Status:	RESOLVED WORKSFORME
Severity:	normal	CC:	ilmari.lauhakangas
Priority:	medium
Version:	6.3.3.2 release
Hardware:	x86-64 (AMD64)
OS:	Linux (All)
Whiteboard:
Crash report or crash signature:	["libmergedlo.so"]	Regression By:
Attachments:	Script used

Description Joona Tolonen 2019-12-29 16:26:35 UTC

This bug was filed from the crash reporting server and is br-a406f680-5f5e-45c8-adb0-ebeee19d51c5.
=========================================

Failed to start libreoffice from a script. Mainly Python (ver 3.5) but also from bash.

- Used commands: os.system() and subprocess.run() where libreoffice were tried to launch in headless, nologo, accept (URP) mode on a localhost (port 8100) e.g.
os.system('/opt/libreoffice6.3/program/soffice --impress --headless --nologo --accept="socket,host=localhost,port=8100;urp" &')
- libreoffice were called /usr/local/bin/libreoffice6.3. Also via /opt/libreoffice6.3/program/soffice --impress
- Libreoffice were tried to open with a specific presentation (with --show flag) and without a presentation as a parameter, both failed

All failed with segfault (ID#139)

Comment 1 Julien Nabet 2019-12-29 17:24:52 UTC

You can try 2 things:
- https://wiki.documentfoundation.org/QA/FirstSteps
- upgrade to 6.3.4

Comment 2 Buovjaga 2020-04-29 18:13:27 UTC

Can you share the script?

Comment 3 Joona Tolonen 2020-04-29 18:37:06 UTC

Created attachment 160087 [details]
Script used

File contains all 'try-outs' around the problem. They are in comments.

Comment 4 Joona Tolonen 2020-04-29 18:37:42 UTC

Yeah sure, why not - I guess it was this one... Nothing special though, just calling os.system from Python 3 script with parameters to make it 'command-able' through UNO.

I've since moved to using Gentoo and latest version of LibreOffice and it's not having this issue - but I thought to share the bug since I thought it was interesting since LibreOffice has been quite stable otherwise.

Comment 5 Buovjaga 2020-04-29 18:48:27 UTC

(In reply to Joona Tolonen from comment #4)
> Yeah sure, why not - I guess it was this one... Nothing special though, just
> calling os.system from Python 3 script with parameters to make it
> 'command-able' through UNO.
> 
> I've since moved to using Gentoo and latest version of LibreOffice and it's
> not having this issue - but I thought to share the bug since I thought it
> was interesting since LibreOffice has been quite stable otherwise.

Ok, thanks. I could not repro the crash either (on Arch Linux). Let's close and celebrate vappu.