Processed: Re: Bug#940127: ghostscript makes c2esp autopkgtest timeout
Processing control commands: > severity -1 important Bug #940127 [src:ghostscript, src:c2esp] gs segfaults in c2esp filter chain Severity set to 'important' from 'serious' -- 940127: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=940127 Debian Bug Tracking System Contact ow...@bugs.debian.org with problems
Re: Bug#940127: ghostscript makes c2esp autopkgtest timeout
Control: severity -1 important Quoting Jonas Smedegaard (2019-11-14 14:28:35) > @Didier: Since you reassigned this to (only) ghostscript, would you > please consider re-reassigning to (only) c2esp instead? > > Reason I ask is that ghostscript is now security-buggy in testing > since a month, seemingly blocked only by this issue. If reassigning > does not seem sensible, then how about lowering severity (maybe only > temporarily)? Lowering severity to prioritize general security over use with c2esp. - Jonas -- * Jonas Smedegaard - idealist & Internet-arkitekt * Tlf.: +45 40843136 Website: http://dr.jones.dk/ [x] quote me freely [ ] ask before reusing [ ] keep private signature.asc Description: signature
Re: Bug#940127: ghostscript makes c2esp autopkgtest timeout
Quoting Didier 'OdyX' Raboud (2019-09-22 15:30:53) > Le dimanche, 22 septembre 2019, 13.25:19 h CEST Brian Potkin a écrit : > > On Sat 21 Sep 2019 at 17:39:20 +0200, Didier 'OdyX' Raboud wrote: > > > Le samedi, 21 septembre 2019, 16.24:30 h CEST Brian Potkin a écrit : > > > > > There's clearly a regression in ghostscript 9.28 that started > > > > > segfaulting > > > > > in the c2esp filter chain. But I can't manage to reproduce it outside > > > > > of > > > > > the "cups + c2esp + cups-filters (gstoraster) + ghostscript" > > > > > environment. > > > > > > > > > > Brian; Till: any idea? > > > > > > > > No ideas from me really. I too get gstoraster stopping when attempting > > > > to print /usr/share/cups/data/form_russian.pdf; but the same is true for > > > > form_english.pdf. > > > > > > Ah, sorry; I formulated my inquiry weakly. Let me try again: > > > > > > Do you have a hint on how to reproduce the failing ghostscript call (or > > > the > > > gstoraster call) directly, without using CUPS in the middle? > > > > Would this do? > > > > cat /usr/share/cups/data/form_russian.pdf | gs -dQUIET -dPARANOIDSAFER > > -dNOPAUSE -dBATCH -dNOINTERPOLATE -dNOMEDIAATTRS -dShowAcroForm > > -sstdout=%stderr -sOutputFile=%stdout -sDEVICE=cups -r600x600 > > -dDEVICEWIDTHPOINTS=595 -dDEVICEHEIGHTPOINTS=842 -dcupsBitsPerColor=8 > > -dcupsColorOrder=0 -dcupsColorSpace=4 -scupsPageSizeName=A4 > > -I/usr/share/cups/fonts -c '< > 3.00] /Margins[00]>>setpagedevice' -f -_ > out.ras > > The problem is… This doesn't segfault. :-( This bugreport seems to only really describe broken behaviour of c2esp. Yes, it smells quite strongly of being _caused_ by some bug in ghostscript, but when only proven breakable in an environment created inside of c2esp, it seems more sensible to me that this bugreport is tied to c2esp rather than ghostscript. @Didier: Since you reassigned this to (only) ghostscript, would you please consider re-reassigning to (only) c2esp instead? Reason I ask is that ghostscript is now security-buggy in testing since a month, seemingly blocked only by this issue. If reassigning does not seem sensible, then how about lowering severity (maybe only temporarily)? - Jonas -- * Jonas Smedegaard - idealist & Internet-arkitekt * Tlf.: +45 40843136 Website: http://dr.jones.dk/ [x] quote me freely [ ] ask before reusing [ ] keep private signature.asc Description: signature
Re: Bug#940127: ghostscript makes c2esp autopkgtest timeout
On 22/09/2019 13:25, Brian Potkin wrote: Would this do? cat /usr/share/cups/data/form_russian.pdf | gs -dQUIET -dPARANOIDSAFER -dNOPAUSE -dBATCH -dNOINTERPOLATE -dNOMEDIAATTRS -dShowAcroForm -sstdout=%stderr -sOutputFile=%stdout -sDEVICE=cups -r600x600 -dDEVICEWIDTHPOINTS=595 -dDEVICEHEIGHTPOINTS=842 -dcupsBitsPerColor=8 -dcupsColorOrder=0 -dcupsColorSpace=4 -scupsPageSizeName=A4 -I/usr/share/cups/fonts -c '<>setpagedevice' -f -_ > out.ras I have tried this command line (with gs 9.27 as of Ubuntu Eoan) and it did neither crash nor error. I got a valid out.ras file without any problem. For the given color space (CMY, -dcupsColorSpace=4) I got broken output (I have checked with rasterview). All other important color spaces (0, 1, 17, 18, 19, 20) give correct output for me (could be another bug in Ghostscript). Till
Re: Bug#940127: ghostscript makes c2esp autopkgtest timeout
Le dimanche, 22 septembre 2019, 13.25:19 h CEST Brian Potkin a écrit : > On Sat 21 Sep 2019 at 17:39:20 +0200, Didier 'OdyX' Raboud wrote: > > Le samedi, 21 septembre 2019, 16.24:30 h CEST Brian Potkin a écrit : > > > > There's clearly a regression in ghostscript 9.28 that started > > > > segfaulting > > > > in the c2esp filter chain. But I can't manage to reproduce it outside > > > > of > > > > the "cups + c2esp + cups-filters (gstoraster) + ghostscript" > > > > environment. > > > > > > > > Brian; Till: any idea? > > > > > > No ideas from me really. I too get gstoraster stopping when attempting > > > to print /usr/share/cups/data/form_russian.pdf; but the same is true for > > > form_english.pdf. > > > > Ah, sorry; I formulated my inquiry weakly. Let me try again: > > > > Do you have a hint on how to reproduce the failing ghostscript call (or > > the > > gstoraster call) directly, without using CUPS in the middle? > > Would this do? > > cat /usr/share/cups/data/form_russian.pdf | gs -dQUIET -dPARANOIDSAFER > -dNOPAUSE -dBATCH -dNOINTERPOLATE -dNOMEDIAATTRS -dShowAcroForm > -sstdout=%stderr -sOutputFile=%stdout -sDEVICE=cups -r600x600 > -dDEVICEWIDTHPOINTS=595 -dDEVICEHEIGHTPOINTS=842 -dcupsBitsPerColor=8 > -dcupsColorOrder=0 -dcupsColorSpace=4 -scupsPageSizeName=A4 > -I/usr/share/cups/fonts -c '< 3.00] /Margins[00]>>setpagedevice' -f -_ > out.ras The problem is… This doesn't segfault. :-( -- OdyX signature.asc Description: This is a digitally signed message part.
Re: Bug#940127: ghostscript makes c2esp autopkgtest timeout
On Sat 21 Sep 2019 at 17:39:20 +0200, Didier 'OdyX' Raboud wrote: > Le samedi, 21 septembre 2019, 16.24:30 h CEST Brian Potkin a écrit : > > > There's clearly a regression in ghostscript 9.28 that started segfaulting > > > in the c2esp filter chain. But I can't manage to reproduce it outside of > > > the "cups + c2esp + cups-filters (gstoraster) + ghostscript" environment. > > > > > > Brian; Till: any idea? > > > > No ideas from me really. I too get gstoraster stopping when attempting > > to print /usr/share/cups/data/form_russian.pdf; but the same is true for > > form_english.pdf. > > Ah, sorry; I formulated my inquiry weakly. Let me try again: > > Do you have a hint on how to reproduce the failing ghostscript call (or the > gstoraster call) directly, without using CUPS in the middle? Would this do? cat /usr/share/cups/data/form_russian.pdf | gs -dQUIET -dPARANOIDSAFER -dNOPAUSE -dBATCH -dNOINTERPOLATE -dNOMEDIAATTRS -dShowAcroForm -sstdout=%stderr -sOutputFile=%stdout -sDEVICE=cups -r600x600 -dDEVICEWIDTHPOINTS=595 -dDEVICEHEIGHTPOINTS=842 -dcupsBitsPerColor=8 -dcupsColorOrder=0 -dcupsColorSpace=4 -scupsPageSizeName=A4 -I/usr/share/cups/fonts -c '<>setpagedevice' -f -_ > out.ras Regards, Brian.
Re: Bug#940127: ghostscript makes c2esp autopkgtest timeout
On Sat 21 Sep 2019 at 17:39:20 +0200, Didier 'OdyX' Raboud wrote: > Le samedi, 21 septembre 2019, 16.24:30 h CEST Brian Potkin a écrit : > > > There's clearly a regression in ghostscript 9.28 that started segfaulting > > > in the c2esp filter chain. But I can't manage to reproduce it outside of > > > the "cups + c2esp + cups-filters (gstoraster) + ghostscript" environment. > > > > > > Brian; Till: any idea? > > > > No ideas from me really. I too get gstoraster stopping when attempting > > to print /usr/share/cups/data/form_russian.pdf; but the same is true for > > form_english.pdf. > > Ah, sorry; I formulated my inquiry weakly. Let me try again: > > Do you have a hint on how to reproduce the failing ghostscript call (or the > gstoraster call) directly, without using CUPS in the middle?
Re: Bug#940127: ghostscript makes c2esp autopkgtest timeout
All, I was the author of c2esp. It's hard for me to test it now because I no longer have kodak printers. I'm wondering if other backends also use the gstoraster filter (I'd expect there would be others?) and do they also cause this error? Or did I do something unusual in c2espC that's contributing to the problem? Paul Newall On 21/09/2019 15:24, Brian Potkin wrote: On Sat 21 Sep 2019 at 10:22:22 +0200, Didier 'OdyX' Raboud wrote: [...] Using snapshot.debian.org: * ghostscript 9.27~dfsg-3.1 works * ghostscript 9.28~~rc1~dfsg-1 and all the later versions segfault So… There's clearly a regression in ghostscript 9.28 that started segfaulting in the c2esp filter chain. But I can't manage to reproduce it outside of the "cups + c2esp + cups-filters (gstoraster) + ghostscript" environment. Brian; Till: any idea? No ideas from me really. I too get gstoraster stopping when attempting to print /usr/share/cups/data/form_russian.pdf; but the same is true for form_english.pdf. OTOH, /usr/share/doc/shared-mime-info/shared-mime-info-spec.pdf and text files complete successfully. Regards, Brian.
Re: Bug#940127: ghostscript makes c2esp autopkgtest timeout
Le samedi, 21 septembre 2019, 16.24:30 h CEST Brian Potkin a écrit : > > There's clearly a regression in ghostscript 9.28 that started segfaulting > > in the c2esp filter chain. But I can't manage to reproduce it outside of > > the "cups + c2esp + cups-filters (gstoraster) + ghostscript" environment. > > > > Brian; Till: any idea? > > No ideas from me really. I too get gstoraster stopping when attempting > to print /usr/share/cups/data/form_russian.pdf; but the same is true for > form_english.pdf. Ah, sorry; I formulated my inquiry weakly. Let me try again: Do you have a hint on how to reproduce the failing ghostscript call (or the gstoraster call) directly, without using CUPS in the middle? signature.asc Description: This is a digitally signed message part.
Re: Bug#940127: ghostscript makes c2esp autopkgtest timeout
On Sat 21 Sep 2019 at 10:22:22 +0200, Didier 'OdyX' Raboud wrote: [...] > Using snapshot.debian.org: > > * ghostscript 9.27~dfsg-3.1 works > * ghostscript 9.28~~rc1~dfsg-1 and all the later versions segfault > > So… > > There's clearly a regression in ghostscript 9.28 that started segfaulting in > the c2esp filter chain. But I can't manage to reproduce it outside of the > "cups + c2esp + cups-filters (gstoraster) + ghostscript" environment. > > Brian; Till: any idea? No ideas from me really. I too get gstoraster stopping when attempting to print /usr/share/cups/data/form_russian.pdf; but the same is true for form_english.pdf. OTOH, /usr/share/doc/shared-mime-info/shared-mime-info-spec.pdf and text files complete successfully. Regards, Brian.
Processed: Re: Bug#940127: ghostscript makes c2esp autopkgtest timeout
Processing control commands: > notfound -1 c2esp/27-4 Bug #940127 [src:ghostscript, src:c2esp] ghostscript makes c2esp autopkgtest timeout No longer marked as found in versions c2esp/27-4. > affects -1 c2esp Bug #940127 [src:ghostscript, src:c2esp] ghostscript makes c2esp autopkgtest timeout Added indication that 940127 affects c2esp > found -1 ghostscript/9.28~~rc1~dfsg-1 Bug #940127 [src:ghostscript, src:c2esp] ghostscript makes c2esp autopkgtest timeout Marked as found in versions ghostscript/9.28~~rc1~dfsg-1. > retitle -1 gs segfaults in c2esp filter chain Bug #940127 [src:ghostscript, src:c2esp] ghostscript makes c2esp autopkgtest timeout Changed Bug title to 'gs segfaults in c2esp filter chain' from 'ghostscript makes c2esp autopkgtest timeout'. -- 940127: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=940127 Debian Bug Tracking System Contact ow...@bugs.debian.org with problems
Re: Bug#940127: ghostscript makes c2esp autopkgtest timeout
Control: notfound -1 c2esp/27-4 Control: affects -1 c2esp Control: found -1 ghostscript/9.28~~rc1~dfsg-1 Control: retitle -1 gs segfaults in c2esp filter chain Hi Paul, thanks for the detailed report. Le jeudi, 12 septembre 2019, 19.27:46 h CEST Paul Gevers a écrit : > Dear maintainers, > > With a recent upload of ghostscript the autopkgtest of c2esp timeouts > [1] in testing when that autopkgtest is run with the binary packages of > ghostscript from unstable. It passes when run with only packages from > testing. In tabular form: >passfail > ghostscriptfrom testing9.28~~rc2~dfsg-1 > c2esp from testing27-4 > all others from testingfrom testing > > The autopkgtest of c2esp also times out in unstable since the upload of > ghostscript. > > Due to the nature of this issue, I filed this bug report against both > packages. Can you please investigate the situation and reassign the bug > to the right package? Sure. So… It fails with the c2espC filter only, the c2esp filter seems fine. It asks for the following filter chain to process a PDF file such as /usr/share/cups/data/form_russian.pdf: /usr/lib/cups/filter/pdftopdf /usr/lib/cups/filter/gstoraster /usr/lib/cups/filter/c2espC In this chain, from the CUPS logs: /usr/lib/cups/filter/gstoraster stopped with status 1 And from my kernel log: Sep 21 09:10:37 gyllingar kernel: [ 9755.656484] gs[10134]: segfault at 7f93b77ea000 ip 7f93b91b0e2a sp 7ffe368515e0 error 4 in libgs.so.9.28[7f93b9135000+367000] Sep 21 09:10:37 gyllingar kernel: [ 9755.656492] Code: 01 48 8d 5c 44 42 31 c0 8b 94 24 10 01 00 00 45 31 f6 85 d2 0f 8e ce 01 00 00 66 0f 1f 44 00 00 48 8b 44 24 18 44 89 74 24 38 <42> 0f b7 34 70 89 f0 8d 50 01 81 e2 fe 00 00 00 0f 85 b0 00 00 00 … but I can't manage to reproduce this failure by manually running the filters. :( But when I go through cups, the ghostscript segfault is repeatable. So it really seems like a ghostscript bug. The corresponding lines from cups' error log are the following: D [21/Sep/2019:09:20:30 +0200] [Job 2201] Ghostscript using Any-Part-of-Pixel method to fill paths. D [21/Sep/2019:09:20:30 +0200] [Job 2201] Ghostscript command line: gs -dQUIET -dPARANOIDSAFER -dNOPAUSE -dBATCH -dNOINTERPOLATE -dNOMEDIAATTRS -dShowAcroForm -sstdout=%stderr -sOutputFile=%stdout -sDEVICE=cups -r600x600 -dDEVICEWIDTHPOINTS=595 -dDEVICEHEIGHTPOINTS=842 -dcupsBitsPerColor=8 -dcupsColorOrder=0 -dcupsColorSpace=4 -scupsPageSizeName=A4 -I/usr/share/cups/fonts -c \'<>setpagedevice\' -f -_ D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[0]=\"CUPS_CACHEDIR=/var/cache/cups\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[1]=\"CUPS_DATADIR=/usr/share/cups\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[2]=\"CUPS_DOCROOT=/usr/share/cups/doc-root\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[3]=\"CUPS_FONTPATH=/usr/share/cups/fonts\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[4]=\"CUPS_REQUESTROOT=/var/spool/cups\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[5]=\"CUPS_SERVERBIN=/usr/lib/cups\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[6]=\"CUPS_SERVERROOT=/etc/cups\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[7]=\"CUPS_STATEDIR=/run/cups\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[8]=\"HOME=/var/spool/cups/tmp\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[9]=\"PATH=/usr/lib/cups/filter:/usr/bin:/usr/sbin:/bin:/usr/bin\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[10]=\"SERVER_ADMIN=root@gyllingar\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[11]=\"SOFTWARE=CUPS/2.3.0\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[12]=\"USER=root\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[13]=\"CUPS_MAX_MESSAGE=2047\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[14]=\"CUPS_SERVER=/var/run/cups/cups.sock\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[15]=\"CUPS_ENCRYPTION=IfRequested\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[16]=\"IPP_PORT=631\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[17]=\"CHARSET=utf-8\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[18]=\"LANG=fr_CH.UTF-8\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[19]=\"PPD=/etc/cups/ppd/test-printer0.ppd\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[20]=\"RIP_MAX_CACHE=128m\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[21]=\"CONTENT_TYPE=application/pdf\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[22]=\"DEVICE_URI=file:///dev/null\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[23]=\"PRINTER_INFO=test-printer0\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[24]=\"PRINTER_LOCATION=\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[25]=\"PRINTER=test-printer0\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[26]=\"PRINTER_STATE_REASONS=none\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[27]=\"CUPS_FILETYPE=document\" D [21/Sep/2019:09:20:30 +0200] [Job 2201] envp[28]=\"FINAL_CONTENT_TYPE=appl
Bug#940127: ghostscript makes c2esp autopkgtest timeout
Source: ghostscript, c2esp Control: found -1 ghostscript/9.28~~rc2~dfsg-1 Control: found -1 c2esp/27-4 Severity: serious Tags: sid bullseye X-Debbugs-CC: debian...@lists.debian.org User: debian...@lists.debian.org Usertags: breaks needs-update timeout Dear maintainers, With a recent upload of ghostscript the autopkgtest of c2esp timeouts [1] in testing when that autopkgtest is run with the binary packages of ghostscript from unstable. It passes when run with only packages from testing. In tabular form: passfail ghostscriptfrom testing9.28~~rc2~dfsg-1 c2esp from testing27-4 all others from testingfrom testing The autopkgtest of c2esp also times out in unstable since the upload of ghostscript. Due to the nature of this issue, I filed this bug report against both packages. Can you please investigate the situation and reassign the bug to the right package? Paul [1] https://ci.debian.net/data/autopkgtest/testing/amd64/c/c2esp/2944626/log.gz signature.asc Description: OpenPGP digital signature
