Hello, I tried to process pdf file via Ghostscript 8.62. It crashed. I tried 8.63svn - it crashed too. I have follows: $ ../bin/gs -dSAFER -dBATCH -dNOPAUSE -sDEVICE=tiffsep -sOutputFile=l01.tif L01-001-K-12345600-00DSG60k.pdf GPL Ghostscript SVN PRE-RELEASE 8.63 (2008-03-01) Copyright (C) 2008 Artifex Software, Inc. All rights reserved. This software comes with NO WARRANTY: see the file PUBLIC for details. Processing pages 1 through 1. Page 1 Segmentation fault Best regards Leonid Lukomskij
Created attachment 4144 [details] L01-001-K-12345600-00DSG60k.pdf This is problem file
This is possibly a duplicate of #689581 which is also a crash using tiffsep and PDF files. The crash in that case is due to transparency and compressed colour representation, I'm not certain if this file contains transparency, but its moderately complicated, so I expect it may. Note that 689581 appears to be a memory corruption problem, if this is the same issue it may be difficult to reproduce. I could only reproduce with a release build, not a debug build, under Windows.
I think this is not tiffsep problem. I have the same result with jpeg device: $ gs -dSAFER -dBATCH -dNOPAUSE -sDEVICE=jpeg -sOutputFile=l01.tif L01-001-K-12345600-00DSG60k.pdf GPL Ghostscript 8.62 (2008-02-29) Copyright (C) 2008 Artifex Software, Inc. All rights reserved. This software comes with NO WARRANTY: see the file PUBLIC for details. Processing pages 1 through 1. Page 1 Segmentation fault I tried this file with 8.61 and it was processed without problems
I confirm that: v. 8.61 renders the file correctly. v. 8.62 crashes current revision renders the file incorrectly but doesn't crash. It would be great to find at what revision occurred the changes.
The crash was introduced by the rev: r8510 | leonardo | 2008-01-28 18:02:59 -0500 (Mon, 28 Jan 2008) | 28 lines Fix (graphics) : Optimize filling a path with a shading color. And fixed by the rev: r8597 | leonardo | 2008-03-12 17:32:03 -0500 (Wed, 12 Mar 2008) | 13 lines Fix (graphics) : Optimize filling a path with a shading color (continuewd 3). Bug 689748 "gs segfaults". I've applied the diff between rev. 8596 and 8597 to the rev. 8510. The resulting version doesn't crash but renders the file incorrectly. This proves that rev. 8510 also introduced incorrect rendering. Reassigning to the author of the patch.
Created attachment 4161 [details] vg,txt Valgring reports quite a few warnings in the current build when it runs the sample file. -dNOGC avoids GC-related warnings. $ valgrind ./gs/debugobj/gs -dNOGC -dNOPAUSE -dBATCH -sDEVICE=bmp16m \ -sOutputFile=/dev/null ../foo.pdf >vg.txt 2>&1
Sedtting P2 for a crashing bug.
Patch http://ghostscript.com/pipermail/gs-cvs/2008-July/008461.html should fix nthe indeterminizm in cmd_write_trapezoid_cmd .
The incorrect rendering is moved to a new bug 689982 "A wrong shading raster". I'll close this one because there is no crash with current revision. Likely the crash was fixwd with r8597 (see Comment #5). See also Comment #8.