Bug#715448: cups-filters: pdftopdf segfaults
severity 715448 important
thanks
On Tue 09 Jul 2013 at 10:07:47 +0200, Johannes Stezenbach wrote:
Hello Johannes,
Thank you for the detail you have supplied.
> usually I'm using Kyocera FS-1020D Foomatic/hpijs-pcl5e via socket://,
> but the bug is reproducible with CUPS-PDF Printer both when
> trying to print PDFs and using "Print Test Page" from Maintenance menu.
>
> Driver: Generic CUPS-PDF Printer (color, 2-sided printing)
> Connection: cups-pdf:/
> Defaults: job-sheets=none, none media=iso_a4_210x297mm sides=one-sided
>
> In dmesg I get:
> kernel: [1678587.402484] pdftopdf[31537]: segfault at 7ffff7986aa0 ip 00007ffff7986aa0 sp 00007ffff7986ac0 error 15
> (addresses are random on each crash)
>
> In error_log with LogLevel=debug2:
>
> I [09/Jul/2013:09:43:29 +0200] [Job 24] Started filter /usr/lib/cups/filter/pdftopdf (PID 31258)
> d [09/Jul/2013:09:43:29 +0200] PDF: File "/usr/lib/cups/filter/pdftops" permissions OK (040755/uid=0/gid=0).
> d [09/Jul/2013:09:43:29 +0200] cupsdStartProcess(command="/usr/lib/cups/filter/pdftops", argv=0x7f035bdffc90, envp=0x7fff2699dba0, infd=25, outfd=24, errfd=18, backfd=19, sidefd=21, root=0, profile=(nil), job=0x7f035bdf63b0(24), pid=0x7f035bdf64ac) = 31259
> ...
> d [09/Jul/2013:09:43:29 +0200] cupsdFinishProcess(pid=31258, name=0x7fff269ae080, namelen=1024, job_id=0x7fff269ae07c(24)) = "/usr
> /lib/cups/filter/pdftopdf"
> D [09/Jul/2013:09:43:29 +0200] [Job 24] PID 31258 (/usr/lib/cups/filter/pdftopdf) crashed on signal 11.
>
> Since the cups-filter package has not been updated recently
> Maybe some other library change broke it, but I've no idea how to debug.
> I hope you can reproduce it.
On a minimal, newly installed unstable amd64 system I set up a print
queue for a file device with the hpijs-pcl5e PPD. Printing a text file
to file:/tmp/fs1020d and through cups-pdf failed with the 'crashed on
signal 11' line in the error log.
However, the Gutenprint and postscript driver both worked to produce
an expected output in /tmp. This is reason enough to question whether
the package is rendered unusable and consquently class its severity
as 'important' at the most.
The bug may also not be in cups-filters.
Regards,
Brian.
Reply to: