[Software] Debian12 as print server: segfault

[jameslast29]

Hello All,

For some time I'm having trouble printing to Kyocera printers via a Debian12 print server.
Actualy it's not just me but 220 users who use this server.
When printing certain types of docs I see this EM :

feb 13 14:51:07 printserver2hofkw kernel: gs[528]: segfault at 55c9e239add0 ip 00007f752de0000a sp 00007fff14c81c10 error 4 in libgs.so.10.00[7f752dc95000+3ff000] likely on CPU 1 (core 2, socket 0)
feb 13 14:51:07 printserver2hofkw kernel: Code: 03 48 39 c2 0f 82 f0 01 00 00 85 c9 0f 84 d4 01 00 00 31 c0 eb 11 66 2e 0f 1f 84 00 00 00 00 00 41 8b b7 d8 00 00 00 8d 14 06 <48> 8b 14 d7 48 81 fa ff 00 00 00 0>

Now today I took another print server who was on Debian10 ----> no issues.
I updated it to Debian11 ---> no issues.
After update to Debian12 ----> same issue, same EM.

My first idea was to downgrade cups-common but there seem to be no earlier packages available:
root@printserver2hofkw:~# apt list --all-versions cups-common
Bezig met oplijsten... Klaar
cups-common/stable,stable,stable-security,stable-security,now 2.4.2-3+deb12u8 all [geïnstalleerd,automatisch]

Any help would be greatly appreciated.
Greetings, j.

[RedGreen925]

Hello All,

Now today I took another print server who was on Debian10 ----> no issues.
I updated it to Debian11 ---> no issues.
After update to Debian12 ----> same issue, same EM.

Any help would be greatly appreciated.

The solution seems obvious to me use the Debian 11 or 10 they just work the 12 does not. It is not like you need the latest software to be running on a print server its job is to allow people to print which the Debian 12 does not do all the time. Either that or struggle along with the Debian 12 and hope your bug report gets solved.

[Aki]

Hello,

For some time I'm having trouble printing to Kyocera printers via a Debian12 print server.
[..]
When printing certain types of docs I see this EM :

Code: Select all

feb 13 14:51:07 printserver2hofkw kernel: gs[528]: segfault at 55c9e239add0 ip 00007f752de0000a sp 00007fff14c81c10 error 4 in libgs.so.10.00[7f752dc95000+3ff000] likely on CPU 1 (core 2, socket 0)
feb 13 14:51:07 printserver2hofkw kernel: Code: 03 48 39 c2 0f 82 f0 01 00 00 85 c9 0f 84 d4 01 00 00 31 c0 eb 11 66 2e 0f 1f 84 00 00 00 00 00 41 8b b7 d8 00 00 00 8d 14 06 <48> 8b 14 d7 48 81 fa ff 00 00 00 0>

Now today I took another print server who was on Debian10 ----> no issues.
I updated it to Debian11 ---> no issues.
After update to Debian12 ----> same issue, same EM.

No such a report is currently reported in the Debian Bug Tracking System (DBTS):

• Debian Bug report logs: Bugs in package ghostscript

According to the previous log, the gs program (source package: https://packages.debian.org/source/bookworm/ghostscript) generates a segmentation error in the libgs.so.10.00 library when it renders a postscript file.

It may be caused by a malformed postscript file: the offending file should be isolated and shared to replicate the problem.

You can also install the libgs10 package debug symbols to get more information from the error message when it happens.

What is the exact model of the Kyocera printer?

Are there any proprietary printer drivers installed on the CUPS server for the Kyocera printer?

How do clients print to the server ? Are they windows or linux clients ?

How often does the segfault happen ?

Hope this helps.

[jameslast29]

Hello,

thanks for the replies, I realy appreciate it.

I rather not do the setup again. It's complicated. The server is on Google Cloud and the printers are obviously on site. Also Papercut is attached to it.
I was kinda hoping there was some fix to this.

What is the exact model of the Kyocera printer? -----> Taskalfa 6053ci and Taskalfa307ci

Are there any proprietary printer drivers installed on the CUPS server for the Kyocera printer? -----> yes, the ppd comes from a .deb package I downloaded from Kyocera site

How do clients print to the server ? Are they windows or linux clients ? This happens only when printing from Chromebook, document opened in Google Docs, document with images in it. Printing the same doc from Google Docs from Debian laptop works.

How often does the segfault happen ? -----> every time the above mentioned are conditions are fullfilled. I can provide a example doc if needed.

Greetings, j.

[Aki]

What is the exact model of the Kyocera printer? -----> Taskalfa 6053ci and Taskalfa307ci

Are there any proprietary printer drivers installed on the CUPS server for the Kyocera printer? -----> yes, the ppd comes from a .deb package I downloaded from Kyocera site

Can you please provide the deb file ?

How do clients print to the server ? Are they windows or linux clients ? This happens only when printing from Chromebook, document opened in Google Docs, document with images in it. Printing the same doc from Google Docs from Debian laptop works.

How often does the segfault happen ? -----> every time the above mentioned are conditions are fullfilled. I can provide a example doc if needed.

It might be useful to share a postscript file that causes the gs program (running on the CUPS server) to fail with a segmentation error.

You could print a sample document with images from your Chromebook (document opened in Google Docs) and collect the resulting postscript file generated by the Kyocera printer drivers in the affected CUPS server after the error has showen up in the logs for that print job.

[jameslast29]

Hello,

the .deb file is too large to add here but I added the resulting ppd
( I had to rename it to ppd.pdf to be able to add it )
I got it from the Kyocera download page, the name of the download is
Linux driver (v.8.7114)
Getting the ghostscript file ---> where would I find that after I print & see the error message?

Greetings, j.

[Aki]

Hello,

Getting the ghostscript file ---> where would I find that after I print & see the error message?

According to [1] scheduled jobs should be stored in /var/spool/cups :

Job Files

The scheduler stores job files in a spool directory, typically /var/spool/cups. Two types of files will be found in the spool directory: control files starting with the letter "c" ("c00001", "c99999", "c100000", etc.) and data files starting with the letter "d" ("d00001-001", "d99999-001", "d100000-001", etc.) Control files are IPP messages based on the original IPP Print-Job or Create-Job messages, while data files are the original print files that were submitted for printing. There is one control file for every job known to the system and 0 or more data files for each job.

Control files are normally cleaned out after the 500th job is submitted, while data files are removed immediately after a job has successfully printed. Both behaviors can be configured.

Hope this helps.

--
[1] CUPS Design Description

[jameslast29]

Hello,

here is the c file:
^B^@^@^E^@^@^@^D^AG^@^Rattributes-charset^@^Eutf-8H^@^[attributes-natural-language^@^Ben^BE^@^Kprinter-uri^@ ipp://localhost/printers/hofkw-3B^@^Yjob-originating-user-name^@^Ojohan.vermeulenB^@^Hjob-name^@/VERBLIJFSOVEREENKOMST.docx. - Google Documenten!^@^Fcopies^@^D^@^@^@^A#^@
finishings^@^D^@^@^@^C!^@^Pjob-cancel-after^@^D^@^@*0D^@^Njob-hold-until^@^Gno-hold!^@^Ljob-priority^@^D^@^@^@2B^@
job-sheets^@^DnoneB^@^@^@^DnoneD^@^Emedia^@^BA4!^@ number-up^@^D^@^@^@^AB^@
PCClientIP^@^K81.83.1.160D^@^Pprint-color-mode^@^EcolorB^@^KPrintSource^@^MMobilityPrintD^@^Esides^@^Stwo-sided-long-edgeE^@^Hjob-uuid^@-urn:uuid:aabdeb75-78f6-36aa-44db-8bc88547be2aB^@^Yjob-originating-host-name^@ localhost1^@^Vdate-time-at-completed^@^K^Gé^B^T^K^V^P^@+^@^@1^@^Udate-time-at-creation^@^K^Gé^B^T^K^S!^@+^@^@1^@^Wdate-time-at-processing^@^K^Gé^B^T^K^S!^@+^@^@!^@^Qtime-at-completed^@^Dg·^Ph!^@^Ptime-at-creation^@^Dg·^OÅ!^@^Rtime-at-processing^@^Dg·^OÅ!^@^Fjob-id^@^D^@^@E #^@ job-state^@^D^@^@^@^GD^@^Qjob-state-reasons^@^Tjob-canceled-by-user!^@^Yjob-impressions-completed^@^D^@^@^@^@!^@^Zjob-media-sheets-completed^@^D^@^@^@^@E^@^Ojob-printer-uri^@#ipp://papercut:631/printers/hofkw-3!^@^Ljob-k-octets^@^D^@^@^@ÎI^@^Xdocument-format-supplied^@^Xapplication/octet-streamI^@^Xdocument-format-detected^@^Oapplication/pdfI^@^Odocument-format^@^Oapplication/pdfB^@^Vdocument-name-supplied^@
017665.pdfA^@^Yjob-printer-state-message^@^MFilter failedD^@^Yjob-printer-state-reasons^@^Dnone^C

The d file is very long & I don't think there is anything in it.

Greetings, j.

[Aki]

Hello,

The d file is very long & I don't think there is anything in it.

The d file should contain the actual data that is causing your Ghostscript interpreter to fail. So both c and d files should be useful for analysis. You can send them as a compressed file as an attachment to a follow-up post.