[GPL GS WORKS] MASTER [apport] gs-esp crashed with SIGSEGV in pdf_find_same_resource()

[manu]

Binary package hint: gs-esp

When converting a .ps file to pdf using ps2pdf through kile a segmentation fault was reported and the translation failed. The ps file was created from a .dvi file.

ProblemType: Crash
Architecture: amd64
CrashCounter: 1
Date: Fri Apr 6 00:53:53 2007
Disassembly: 0x0:
DistroRelease: Ubuntu 7.04
ExecutablePath: /usr/bin/gs-esp
Package: gs-esp 8.15.4.dfsg.1-0ubuntu1
PackageArchitecture: amd64
ProcCmdline: gs -dSAFER -dCompatibilityLevel=1.4 -q -dNOPAUSE -dBATCH -sDEVICE=pdfwrite -sOutputFile=final.pdf -dSAFER -dCompatibilityLevel=1.4 -c .setpdfwrite -f final.ps
ProcCwd: /home/manu/Documents/Tex/buckling
ProcEnviron:
 PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/bin/X11:/usr/games
 LANG=en_US.UTF-8
 SHELL=/bin/bash
Signal: 11
SourcePackage: gs-esp
StacktraceTop:
 ?? ()
 ?? () from /usr/lib/libgs-esp.so.8
 ?? () from /usr/lib/libgs-esp.so.8
 ?? () from /usr/lib/libgs-esp.so.8
 pdf_find_same_resource ()
Uname: Linux manu-desktop 2.6.20-13-generic #2 SMP Sat Mar 24 23:12:02 UTC 2007 x86_64 GNU/Linux
UserGroups: adm admin audio cdrom dialout dip floppy lpadmin plugdev scanner video

[Apport retracing service (apport)]

StacktraceTop:?? ()
cos_value_equal (pcv0=<value optimized out>, pcv1=<value optimized out>, pdev=0x67f058) at ./src/gdevpdfo.c:397
cos_dict_equal (pco0=0xc6a618, pco1=0xbc1260, pdev=0x67f058) at ./src/gdevpdfo.c:1050
cos_stream_equal (pco0=0xbb6ab8, pco1=0xbce2f8, pdev=0x67f058) at ./src/gdevpdfo.c:1205
pdf_find_same_resource (pdev=0x67f058, rtype=<value optimized out>, ppres=0xc2c9e8, eq=0x2b89a2e1b490 <nocheck>) at ./src/gdevpdfu.c:545

[Brian Murray (brian-murray)]

I have been unable to recreate this on amd64 using gs-esp version 8.15.4.dfsg.1-0ubuntu1.

[manu (mrcluee)]

I noticed that when switching to gs-gpl ps2pdf works fine. Is there any reason to keep gs-esp? It is based on an older version of gs-gpl and it is not developed anymore since its fork has merged into gs-gpl. In my opinion it is also buggy. Here is the quote from www.cups.org/espgs:
In 2006, Artifex and the ESP Ghostscript developers agreed to merge the changes from ESP Ghostscript into GPL Ghostscript, and to collaborate on a single version of Ghostscript for all platforms. All development on ESP Ghostscript has ended with the 8.15.4 release.

[Till Kamppeter (till-kamppeter)]

The merger will happen soon. Feisty is the last Ubuntu with ESP GhostScript. Note the [GPL GS WORKS] in the title. Please whenever you find a GhostScript bug try also gs-gpl and add the [GPL GS WORKS] to the bug title if gs-gpl works. This helps us on doing the merger without passing bugs from ESP to GPL GhostScript. Please also add the offending PostScript or PDF files in such cases, so that they can be added to the regression testing at ghostscript.com.

[Thomas Nemeth (tnemeth)]

I have the same problem with gs-esp crashing when printing to PDF.

I tested gs-gpl instead of gs-esp but if it didn't crash, it couldn't produce any readable PDF,
even with other applications (kwrite, kate, kontact...). gs-esp crashes only with konqueror
on my kubuntu/feisty: other apps can produce nice PDF files.

[Till Kamppeter (till-kamppeter)]

For everyone who has still this problem in Feisty, here is a way to get to the newest Ghostscript version:

Download the files from

http://www.linux-foundation.org/~till/tmp/ubuntu/gutsy/ghostscript/

into the same directory. Then do

sudo apt-get install build-essential fakeroot
sudo apt-get build-dep gs-esp gs-gpl
dpkg-source -x ghostscript_8.60.dfsg.2-0ubuntu1.dsc
cd ghostscript-8.60.dfsg.1
dpkg-buildpackage -rfakeroot
cd ..
sudo dpkg -i *gs-esp*.deb *gs-afpl*.deb *gs-gpl*.deb *libgs*.deb *ghostscript*.deb

and you will have the latest Ghostscript.

Answer every question which appears with "y", so that all needed packages get installed.

Please tell us your results with the test packages.

We will look into making the new Ghostscript available in the backports repository.

[josesuarez1983 (j-suarez-agapito)]

On Saturday 26 May 2007 15:20:09 Till Kamppeter wrote:
> cd ghostscript-8.60.dfsg.1

I'm currently compiling ghostscript with all the files Till said.
Just a little thing: in the instructions where you say

cd ghostscript-8.60.dfsg.1

it should say:

cd ghostscript-8.60.dfsg.2

Just telling this if anybody gets confused.

All the Best, José

[josesuarez1983 (j-suarez-agapito)]

The issue seems to be fixed with gs-gpl, since I'm now able to print to PDF
the .ps file I had attached the first time when I reported the problem.
So with me it's ok if you guys commit the change into the repossitories.
Good job!, and thanks for dealing with the bug.

All the best, José

[josesuarez1983 (j-suarez-agapito)]

On Saturday 26 May 2007 15:20:09 Till Kamppeter wrote:
> For everyone who has still this problem in Feisty, here is a way to get
> to the newest Ghostscript version:
>
> Download the files from
>
> http://www.linux-foundation.org/~till/tmp/ubuntu/gutsy/ghostscript/
>
> into the same directory. Then do
>
> sudo apt-get install build-essential fakeroot
> sudo apt-get build-dep gs-esp gs-gpl
> dpkg-source -x ghostscript_8.60.dfsg.2-0ubuntu1.dsc
> cd ghostscript-8.60.dfsg.1
> dpkg-buildpackage -rfakeroot
> cd ..
> sudo dpkg -i *gs-esp*.deb *gs-afpl*.deb *gs-gpl*.deb *libgs*.deb
> *ghostscript*.deb
>
> and you will have the latest Ghostscript.
>
> Answer every question which appears with "y", so that all needed
> packages get installed.
>
> Please tell us your results with the test packages.
>
> We will look into making the new Ghostscript available in the backports
> repository.

I have just notived that after the installation of the new gs packages cups
gives an error while trying to print with a 'true' printer related
to /usr/lib/cups/filter/foomatic-rip. However, after uninstalling the new gs
packages, and reinstalling the original feisty packages, the printer prints
correctly.

[Till Kamppeter (till-kamppeter)]

josesuarez1983, please open a new bug report for this problem and tell which printer model and driver you are using. This looks like that a certain printer driver in the new Ghostscript is broken and we must know which one to fix it upstream.