[cups] Sending mail from a CentOS7 virtual machine to Fuji Xerox ApeosPort-IV C4475 problem revisted

Zdenek Dohnal zdohnal at redhat.com
Mon Oct 22 01:57:22 PDT 2018


On 10/22/18 4:11 AM, マスターズ・イアン wrote:
> Hi Zdenek
>
>> Ian, do you find such logs in journal?
>>
>> |Aug 07 17:00:59 pserver.example.com kernel: pdftopjlfx[60882]:
>> segfault at 0 ip 0000000000401bf3 sp 00007ffc30231020 error 4 in
>> pdftopjlfx[400000+3000] Aug 07 17:11:51 pserver.example.com kernel:
>> pdftopjlfx[65152]: segfault at 0 ip 0000000000401bf3 sp
>> 00007ffd3d52e570 error 4 in pdftopjlfx[400000+3000] Aug 07 17:16:49
>> pserver.example.com kernel: pdftopjlfx[2326]: segfault at 0 ip
>> 0000000000401bf3 sp 00007fff5e18f0b0 error 4 in
>> pdftopjlfx[400000+3000] |
>
> Yes, I have such messages in my /var/log/messages file ...
>
>> |||If you do, then you encountered known problem - you have these
>> filters from fxlinuxprint package from vendor, correct? These rpms
>> seems to be built with different versions of required packages than
>> we have in CentOS, so when you try to run the filter, it segfaults
>> because of binary incompatibility. The solution would be to rebuild
>> with packages from CentOS, but I cannot do the rebuild, because it is
>> not in CentOS distribution (because nobody puts it in distribution...).
>
> What isn't in the CentOS distribution?
I meant fxlinuxprint project - source code, which contains
implementation of filters, which you are trying to use. Nobody took the
project and packaged it into our OS.
>
>>
>> |But the project is open source, it seems -
>> https://github.com/rogers0/fxlinuxprint - you can compile it for
>> yourself and segfault would go away. I bookmarked the package for
>> adding into Fedora, so I hope I will fix it permanently in the future
>> (but the path is long - first it needs to be added in Fedora, then
>> get it into RHEL/CentOS, which needs agreement from other people than
>> me..., so I'm sorry, I wouldn't expect to solve it quickly...)
>
> I tried to compile it from the source I got from Xerox
> (http://download.fujixerox.co.jp/pub/exe/apeosport/c4300series/fxlinuxprint-src-1.0.1.tar.gz),
> but that segfaulted too.

Hmm, that is against my suspicion of binary incompatibility, if you
already use binaries compiled by yourself on CentOS... ok, then you can
look into core dump by gdb (binary - the filter - needs to be compiled
with '-g' option, then you can see correct debug info in gdb) and share
full backtrace here (as an attachment) or if you have some experience
with C, you can find out something by yourself.

You should get core dump of segfaulted binary when you do:

1) $ ulimit -c unlimited

2) check if you only have 'core' in /proc/sys/kernel/core_pattern

Then run the binary and when it segfaults, you should have file
core.<PID> in your current directory.  You can open the core dump by gdb by:

$ gdb /path/to/your/filter core.<PID>

and check what is there.

>
> If there's anything I can send you to help, please let me know.
>
> ian
>
>>
>> ||
>>
>> On 10/19/18 8:58 AM, マスターズ・イアン wrote:
>>> Gernot and anyone else still reading ...
>>>
>>> I decided to try and run each filter one by one to see what happens.
>>>
>>> I started with:
>>>
>>> /usr/lib/cups/filter/pdftopdffx 1 3496 '' 1 ''
>>> </home/user1/Desktop/test1.pdf >pdftopdffx.pdf
>>>
>>> That was OK, it produced a readable PDF from a readable PDF
>>>
>>> Then I tried:
>>> /usr/lib/cups/filter/pdftopjlfx 1 3496 '' 1 ''
>>> </home/user1/pdftopdffx.pdf >pdftopdffx.pjlfx
>>>
>>> and I got "Segmentation fault (コアダンプ)"
>>>
>>> I thought that might mean that /usr/lib/cups/filter/pdftopjlfx was
>>> corrupted but I extracted it from the RPM and diffed the 2 and they
>>> were identical.
>>>
>>> I'm not sure why it's core dumping but I guess that's why it's not
>>> producing a useable print job.
>>>
>>> Any idea what I can do about this?
>>>
>>> ian
>>> _______________________________________________
>>> cups mailing list
>>> cups at cups.org
>>> https://lists.cups.org/mailman/listinfo/cups
>>
>> -- 
>> Zdenek Dohnal
>> Associate Software Engineer
>> Red Hat Czech - Brno TPB-C
>>
-- 
Zdenek Dohnal
Associate Software Engineer
Red Hat Czech - Brno TPB-C




More information about the cups mailing list