Bug 113823 - gs aborts with "buffer overflow detected" for -sDEVICE=lips4v
Summary: gs aborts with "buffer overflow detected" for -sDEVICE=lips4v
Status: RESOLVED FIXED
Alias: None
Product: SUSE LINUX 10.0
Classification: openSUSE
Component: Printing (show other bugs)
Version: Beta 3
Hardware: All Linux
: P5 - None : Normal
Target Milestone: ---
Assignee: Dr. Werner Fink
QA Contact: Johannes Meixner
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2005-08-29 12:09 UTC by Johannes Meixner
Modified: 2005-08-29 15:16 UTC (History)
1 user (show)

See Also:
Found By: Development
Services Priority:
Business Priority:
Blocker: ---
Marketing QA Status: ---
IT Deployment: ---

Attachments
strace -f of the gs command on nelson.suse.de (i386) (93.13 KB, text/plain)
2005-08-29 12:12 UTC, Johannes Meixner 		Details
strace -f of the gs command on caps.suse.de (x86_64) (94.63 KB, text/plain)
2005-08-29 12:12 UTC, Johannes Meixner 		Details
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Johannes Meixner 2005-08-29 12:09:06 UTC 
On nelson.suse.de (i386) where a plain beta3 is installed:

jsmeix@nelson:~> rpm -q ghostscript-library
ghostscript-library-8.15rc1-25


jsmeix@nelson:~> rpm -q --changelog ghostscript-library | head -n 4
* Mon Aug 15 2005 - mfabian@suse.de

- Bugzilla #104656: Fix for Ghostscript to render vertical
  Japanese correctly again.


jsmeix@nelson:~> gs -q -dBATCH -dPARANOIDSAFER -dQUIET -dNOPAUSE \
                    -sDEVICE=lips4v -sOutputFile=/dev/null \
                    /usr/share/doc/packages/ghostscript/examples/colorcir.ps 
*** buffer overflow detected ***: gs terminated
Aborted



On caps.suse.de (x86_64) where a newer ghostscript-library is installed:

caps:~ # rpm -q ghostscript-library
ghostscript-library-8.15rc1-25


caps:~ # rpm -q --changelog ghostscript-library | head -n 3
* Thu Aug 25 2005 - werner@suse.de

- Fix color model handling of pcl3 driver for gs 8.15 (bug #112659)


caps:~ # gs -q -dBATCH -dPARANOIDSAFER -dQUIET -dNOPAUSE \
            -sDEVICE=lips4v -sOutputFile=/dev/null \
            /usr/share/doc/packages/ghostscript/examples/colorcir.ps
*** buffer overflow detected ***: gs terminated
Aborted
Comment 1 Johannes Meixner 2005-08-29 12:12:12 UTC 
Created attachment 47982 [details]
strace -f of the gs command on nelson.suse.de (i386)
Comment 2 Johannes Meixner 2005-08-29 12:12:51 UTC 
Created attachment 47983 [details]
strace -f of the gs command on caps.suse.de (x86_64)
Comment 3 Dr. Werner Fink 2005-08-29 12:43:43 UTC 
The strace does not help in this case.
Comment 4 Johannes Meixner 2005-08-29 12:54:46 UTC 
What would help you?
Comment 5 Dr. Werner Fink 2005-08-29 13:07:56 UTC 
In this case of an overflow I've to debug the lips4v driver to catch
the wrong usage of the new color mode interface of gs 8.15.

Straces are usefull to catch missing files or files which were missed
by gs but installed.
Comment 6 Dr. Werner Fink 2005-08-29 13:19:36 UTC 
Hmmm ... within my change root environment I see:

g31:espgs-8.15rc1 # bin/gs -q -dBATCH -dPARANOIDSAFER -dQUIET -dNOPAUSE \
-sDEVICE=lips4v -sOutputFile=/dev/null \
/usr/share/doc/packages/ghostscript/examples/colorcir.ps
g31:espgs-8.15rc1 # 

simply nothing.
Comment 7 Dr. Werner Fink 2005-08-29 13:23:27 UTC 
OK on caps I see:

/suse/werner> gs -q -dBATCH -dPARANOIDSAFER -dQUIET -dNOPAUSE -sDEVICE=lips4v
-sOutputFile=/dev/null /usr/share/doc/packages/ghostscript/examples/colorcir.ps
*** buffer overflow detected ***: gs terminated
Abort

but

/suse/werner> tmp/gs -q -dBATCH -dPARANOIDSAFER -dQUIET -dNOPAUSE
-sDEVICE=lips4v -sOutputFile=/dev/null
/usr/share/doc/packages/ghostscript/examples/colorcir.ps
/suse/werner> 

... strange.
Comment 8 Dr. Werner Fink 2005-08-29 13:46:18 UTC 
It seems that libgmodule for OMNI was not linked with the new generated gs.
Comment 9 Dr. Werner Fink 2005-08-29 14:25:22 UTC 
Hmmm .. even if libgmodule is linked in and the same CFLAGS are used
the gs binary get from command line in the change root environment
is different as the binary generated from the build comand in
the same change root environment ....
Comment 10 Dr. Werner Fink 2005-08-29 15:16:29 UTC 
FIXED