Bug 473506 - cups doesn't print PDF or PS
cups doesn't print PDF or PS
Product: Fedora
Classification: Fedora
Component: cups (Show other bugs)
x86_64 Linux
medium Severity high
: ---
: ---
Assigned To: Tim Waugh
Fedora Extras Quality Assurance
: 473505 (view as bug list)
Depends On:
  Show dependency treegraph
Reported: 2008-11-28 18:09 EST by Horst H. von Brand
Modified: 2009-06-04 07:47 EDT (History)
7 users (show)

See Also:
Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
Last Closed: 2009-06-04 07:47:57 EDT
Type: ---
Regression: ---
Mount Type: ---
Documentation: ---
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---

Attachments (Terms of Use)
CUPS error_log for failed print (2.14 KB, application/octet-stream)
2008-11-28 18:10 EST, Horst H. von Brand
no flags Details
Rquested troubleshoot output (4.77 KB, application/octet-stream)
2008-12-19 13:28 EST, Horst H. von Brand
no flags Details
Output for printing an PDF (29.75 KB, text/plain)
2008-12-19 13:32 EST, Horst H. von Brand
no flags Details
The requested printing troubleshoot output (19.92 KB, text/plain)
2008-12-19 13:34 EST, Horst H. von Brand
no flags Details
An error showing text/plain being rejected as an invalid format (5.20 KB, text/plain)
2008-12-19 19:07 EST, John Himpel
no flags Details

  None (edit)
Description Horst H. von Brand 2008-11-28 18:09:09 EST
Description of problem:
When trying to print to a LaserJet 3020 via IPP lpr(1) says:

 lpr: successful-ok

and nothing gets printed. Had to downgrade to cups-1.3.9-2.fc10.x86_64 and install cupsddk-drivers-1.2.3-5.fc10.x86_64.rpm to get printing to work.

Version-Release number of selected component (if applicable):

How reproducible:
With PDF and PS here

Steps to Reproduce:
1. lpr something.pdf
Actual results:
"lpr: successful-ok", no printout. /var/log/cups/error_log excerpt after restarting cups and trying to print attached.

Expected results:

Additional info:
Comment 1 Horst H. von Brand 2008-11-28 18:10:18 EST
Created attachment 325048 [details]
CUPS error_log for failed print
Comment 2 Horst H. von Brand 2008-11-28 22:54:07 EST
*** Bug 473505 has been marked as a duplicate of this bug. ***
Comment 3 Horst H. von Brand 2008-12-09 16:43:18 EST
Tried again with cups-1.4-0.b1.4.fc11.x86_64, same results
Comment 4 Jonathan Corbet 2008-12-15 13:31:38 EST

Don't know if this is related, but I also see behavior like this:

% enscript 
lpr: Unsupported format 'application/postscript'!
[ 1 pages * 1 copy ] sent to printer
Comment 5 Horst H. von Brand 2008-12-18 15:33:56 EST
Just tried cups-1.4-0.b2.2.fc11.x86_64.rpm, same results.
Comment 6 Tim Waugh 2008-12-19 08:00:26 EST
E [28/Nov/2008:19:38:07 -0300] Returning IPP client-error-document-format-not-supported for Send-Document (ipp://localhost:631/printers/LaserJet-3020) from localhost

That's quite odd.  Could you try running the printing troubleshooter and attach the resulting troubleshoot.txt file here please?
Comment 7 Horst H. von Brand 2008-12-19 13:28:06 EST
Created attachment 327478 [details]
Rquested troubleshoot output

First I had to find it...

System -> Administration -> Printing -> Help -> Troubleshoot

I tried to print the test page, it failed with a message saying the format isn't supported. The result is attached.
Comment 8 Horst H. von Brand 2008-12-19 13:32:34 EST
Created attachment 327479 [details]
Output for printing an PDF

Same as before, but now explicitly printing a PDF file.
lpr(1) again says:

  lpr: successful-ok
The "Test Page" window shows the job (77) as cancelled.
Comment 9 Horst H. von Brand 2008-12-19 13:34:10 EST
Created attachment 327480 [details]
The requested printing troubleshoot output

Output from printing the test page
Comment 10 John Himpel 2008-12-19 19:07:35 EST
Created attachment 327506 [details]
An error showing text/plain being rejected as an invalid format

I have attached an error log showing text/plain being an unsupported format.  I am printing from firefox 3.0.4 (from rawhide) to an HP 882C.  Cups=1.4-0.b2.2.fc11.i386
Comment 11 Mario Filizzola 2009-01-07 22:58:36 EST
I'm using Fedora 10 and I'm having the same problem. Here the problem seems related to Selinux. Below is the information from setroubleshoot. I already have "use_nfs_home_dirs=1"


O SELinux prevented cups-pdf reading files in a NFS file system.

Permitindo Acesso:

Changing Boolean "use_nfs_home_dirs" will allow this access:
"setsebool -P use_nfs_home_dirs=1"

Reparar Comando:

setsebool -P use_nfs_home_dirs=1

Informação adicional:

Contexto Fonte                system_u:system_r:cups_pdf_t:s0-s0:c0.c1023
Contexto Alvo                 system_u:object_r:nfs_t:s0
Objetos target                /home [ dir ]
Fonte                         cups-pdf
Caminho source                /usr/lib/cups/backend/cups-pdf
Porta                         <Desconhecido>
Máquina                       fedora
Pacotes do RPM source          cups-pdf-2.4.8-1.fc10
Pacotes do RPM target           filesystem-2.4.19-1.fc10
RPM da Política              selinux-policy-3.5.13-34.fc10
Selinux Ativado               True
Tipo de Política             targeted
MLS Ativado                   True
Modo Enforcing                Enforcing
Nome do Plugin                use_nfs_home_dirs
Nome do Host                  fedora
Plataforma                    Linux fedora #1 SMP Tue
                              Dec 16 14:47:52 EST 2008 x86_64 x86_64
Conta de Alerta               9
Visto Primeiro em             Qua 07 Jan 2009 08:20:25 BRST
Visto pela última vez em     Qui 08 Jan 2009 00:07:57 BRST
ID Local                      e913b5fc-e4c0-4aa6-98c1-c3e4e69d1c98
Números de Linha             

Mensagens de Auditoria de Mat 

node=fedora type=AVC msg=audit(1231380477.707:57): avc:  denied  { getattr } for  pid=6113 comm="cups-pdf" path="/home" dev=sda2 ino=2 scontext=system_u:system_r:cups_pdf_t:s0-s0:c0.c1023 tcontext=system_u:object_r:nfs_t:s0 tclass=dir

node=fedora type=SYSCALL msg=audit(1231380477.707:57): arch=c000003e syscall=4 success=no exit=-13 a0=7fffa53a19b0 a1=7fffa53a0850 a2=7fffa53a0850 a3=fe items=0 ppid=2635 pid=6113 auid=4294967295 uid=0 gid=7 euid=0 suid=0 fsuid=0 egid=7 sgid=7 fsgid=7 tty=(none) ses=4294967295 comm="cups-pdf" exe="/usr/lib/cups/backend/cups-pdf" subj=system_u:system_r:cups_pdf_t:s0-s0:c0.c1023 key=(null)
Comment 12 Tim Waugh 2009-01-08 07:28:06 EST
(In reply to comment #11)
> I'm using Fedora 10 and I'm having the same problem. Here the problem seems
> related to Selinux. Below is the information from setroubleshoot. I already
> have "use_nfs_home_dirs=1"

This is a different issue from the looks of it.  Please file a separate bug report against selinux-policy.
Comment 13 Horst H. von Brand 2009-02-25 14:50:14 EST
Latest cups (cups-1.4-0.b2.7.fc11.x86_64.rpm) still shows the same problem, got  


from Koji, those do work.
Comment 14 Horst H. von Brand 2009-03-04 12:09:29 EST
Again, cups-1.4-0.b2.7.fc11.1.x86_64.rpm doesn't work. Now with cupsddk-drivers-1.2.3-6.fc11.x86_64, cups{,-libs} as above.
Comment 15 Tim Waugh 2009-03-11 13:10:08 EDT
Please try cups-1.4-0.b2.11.fc11, which is based on newer code than the 1.4b2 release.  Is it any different? (There have been lots of changes that may have a bearing on this problem.)
Comment 16 Horst H. von Brand 2009-03-11 16:06:51 EDT
Got cups-{,libs-}1.4-0.b2.11.fc11 from Koji, installed that via rpm. This gave  /etc/cups/{classes,printers,subscriptions}.conf.rpmnew, overwrote the originals with the rpmnew ones and restarted cups. Had to configure the printer again from scratch (not nice!). After that, it printed a testpage and a PDF just fine. No error messages.

Seems fixed, but I'd see if it cures the problem for all affected.

Many thanks!
Comment 17 Valdis Kletnieks 2009-03-13 11:19:45 EDT
Still broken for me, in different ways...

'enscript /etc/profile' reports

% enscript /etc/profile
lpr: Unsupported format 'application/postscript'!
[ 1 pages * 1 copy ] sent to printer

and logs:

localhost - - [13/Mar/2009:10:59:01 -0400] "POST /printers/hp4300 HTTP/1.1" 200 382 Create-Job successful-ok
localhost - - [13/Mar/2009:10:59:01 -0400] "POST /printers/hp4300 HTTP/1.1" 200 15635 Send-Document client-error-document-format-not-supported

Then when I try to release the job to the printer, cupsd goes into a 100% CPU bound state, about 45% user and 55% kernel.  gdb reports:

(gdb) where
#0  0x00007f07a9be04b3 in __epoll_wait_nocancel () from /lib64/libc.so.6
#1  0x00007f07acca18fc in ?? () from /usr/sbin/cupsd
#2  0x00007f07acc7c04c in main () from /usr/sbin/cupsd

Installing debuginfo would be 14 RPMs totalling 162M, so I'm not in a mood to do that unless needed.
Comment 18 Tim Waugh 2009-03-13 11:54:30 EDT
Hmm, I don't see this at all.

I would very much like to track down the busy-loop problem first.  You don't need to download any debuginfo RPMs except cups-debuginfo -- install it with 'rpm -ivh --nodeps' if need be.  We only need to know about symbols in cupsd/libcups, as that's where the problem is.
Comment 19 Alex Dudko 2009-05-25 11:12:10 EDT
Had the same problem in FC11. The error message was "Send-Document client-error-document-format-not-supported"
Solved by changing /usr/lib/cups/filter and enclosed files owner to root.
Comment 20 Tim Waugh 2009-06-03 07:32:29 EDT
What owner did they have before?

The package manifest has them as owned by root:root so I'm not sure how they were changed in the first place.
Comment 21 Alex Dudko 2009-06-03 18:56:41 EDT
After driver installation they were owned by a not root user, the user which particularly installed the driver.
Comment 22 Tim Waugh 2009-06-04 05:36:57 EDT
Which driver?
Comment 23 Alex Dudko 2009-06-04 07:21:46 EDT
A UnifiedLinuxDriver, downloaded from http://www.samsung.com/us/support/download/supportDownDetail.do?group=&type=&subtype=&model_nm=SCX-4100&language=&cate_type=all&mType=DR&dType=D&vType=&cttID=2048258&prd_ia_cd=&disp_nm=SCX-4100&model_cd=&menu=download&menu2=detail
Was installed running the installation script.
The same driver was installed on FC9 without any problem.
Comment 24 Tim Waugh 2009-06-04 07:47:57 EDT
OK, problem is in third party driver.

I was unable to reproduce the problem noted in comment #17, so closing.

Note You need to log in before you can comment on or make changes to this bug.