Description of problem: ----------------------- When invoking the add a printer functionality using cups HTTP interface, the cups server crashes. Version-Release number of selected component (if applicable): cups-1.1.20-1 How reproducible: ----------------- Always Steps to Reproduce: ------------------- 1. Make sure the cups server has been started by running "service cups start" 2. Open up a Web browser window locally. 3. Navigate to URL http://localhost:631 4. From the CUPS main web page, click on the "Do Administration Tasks" link 5. To access the previous link, you'll need to authenticate as root. 6. From the "Admin on localhost" page, click on the "Add Printer" link beneath the "Printers" section. 7. You'll see an error page with the following text: An error occurred while loading http://localhost:631/admin/?op=add-printer: Connection to localhost is broken 8. The cups server has crashed. Actual results: --------------- I'm unable to add printers to CUPS using the HTTP web interface. Expected results: ----------------- CUPS shouldn't crash when trying to add a printer using the HTTP interface. Additional info: ---------------- Before trying to reproduce the crash, I modified "/etc/cups/cupsd.conf" and set the "LogLevel" parameter to "debug2". Attached to this bug report there are two files: - access_log - error_log Both files do contain the logs of a cleanly installed cups server when performing the steps mentioned above.
Created attachment 96798 [details] access_log from cups after trying to add a printer via HTTP interface
Created attachment 96799 [details] error_log from cups after trying to add a printer via HTTP interface
I can't seem to reproduce this here. :-/ Please try running 'strace -fp 734' (or whatever PID cupsd has) shortly before clicking 'Add printer', and attach the output here.
Created attachment 96805 [details] strace of the cupsd process before trying to add a new printer via HTTP interface rawhide up2date machine kernel 2.6.0-rc2 zebra is running IPv6 support is enabled and working fine cupsd pristine configuration with two printers added via system-config-printer tool
What's in your /etc/pam.d/system-auth file?
Created attachment 96810 [details] /etc/pam.d/system-auth Yeah! I forgot to mention I was using Kerberos V authentication.
Is it possible for you to recompile the CUPS packages from the 1.1.20-1 src.rpm file? I'd like you to make a change to the spec file: remove the line that says '%patch16 -p1 -b .pie'. That way we'll be able to look at what cupsd is up to in the debugger.
OK, Now, I've got cups-1.1.20-2 installed with no program independent code support (i.e. no -fPIE gcc flag during build). What do you want me to do next?
Curiously, after removing PIE support, cupsd doesn't crash anymore when trying to add a printer via the Web interface but instead turns itself into a CPU hog, wasting tons of CPU cycles with no apparent effect. Once cupsd goes wild, the HTTP interface stops responding and the cupsd daemon stops serving requests. Aditionally, an "strace" of the cupsd process shows nothing at all.
I believe my problem is somewhat similar; therefore I add it to this bug. I get random delays, error messages, and mangled requests when I do a POST to a password protected Location (/admin) in cups-1.1.20-2 (rebuilt on RH 7.2/Alpha). It might have something to do with cups/auth.c, which was added in CUPS 1.1.20, since I didn´t have any problems with cups-1.1.19-8. Also, the problems go away if I comment out the AuthType and AuthClass lines for the /admin Location. I´ll attach "before" and "after" log snippets for a config-printer request in cups-1.1.19-8 and cups-1.1.20-2. cupsd.conf was not changed before, during or after the upgrade. You´ll note that config-printer works flawlessly in 1.1.19-8, but produces errors, long delays (for the ´gimp-print´ printer, a short delay for the ´hpijs´ printer), and mangled requests (random ppd noise) in 1.1.20-2.
Created attachment 97479 [details] config-printer request in cups-1.1.19-8
Created attachment 97480 [details] config-printer request in cups-1.1.20-2
Robert: if you do not have krb5 enabled in your /etc/pam.d/system-auth file, please file a separate bug report.
I do not have krb5 enabled and have opened a new bug, #114999.
Thanks. *This* bug is in libkrb5 I think.
I can't reproduce this here (krb5 1.3.1-7, pam_krb5 2.0.4-1/2.0.5-1, cups 1.1.20-2, kernel-smp 2.6.1-1.65).
Nalin: you added the pam_krb5.so lines to system-auth? I was able to produce a busy-loop (corrupted linked list had a cycle) inside libkrb5 using that set-up.
Yes, I even set up a test realm so that I could have a principal named "root", and the administration tool asked for printer information at the point where cupsd should have crashed. Is there a version skew somewhere between my test system and the original bug reporter's (or yours)?
'Red Hat Raw Hide' refers to the development tree for Red Hat Linux. Red Hat Linux is no longer supported by Red Hat, Inc. If you are still running Red Hat Linux, you are strongly advised to upgrade to a current Fedora Core release or Red Hat Enterprise Linux or comparable. Some information on which option may be right for you is available at http://www.redhat.com/rhel/migrate/redhatlinux/. Red Hat apologizes that these issues were not resolved in a more timely manner. However, we do want to make sure that important don't slip through the cracks. If these issues are still present in a current release, such as Fedora Core 5, please move these bugs to that product and version. Note that any remaining Red Hat Raw Hide bugs will be closed as 'CANTFIX' on September 30, 2006. Thanks again for your help.
Red Hat Linux is no longer supported by Red Hat, Inc. If you are still running Red Hat Linux, you are strongly advised to upgrade to a current Fedora Core release or Red Hat Enterprise Linux or comparable. Some information on which option may be right for you is available at http://www.redhat.com/rhel/migrate/redhatlinux/. Closing as CANTFIX.