336435 - When printing a web page, the page prints but the print progress window hangs, and Firefox with it.

Status: RESOLVED DUPLICATE of bug 294879

(Core Graveyard :: GFX: Gtk, defect)

Description

[Paul Abrahams]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.0.3) Gecko/20060425 SUSE/1.5.0.3-4.1 Firefox/1.5.0.3
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.0.3) Gecko/20060425 SUSE/1.5.0.3-4.1 Firefox/1.5.0.3

When I attempt to print a web page, the page prints but the progress window hangs (usually on the Preparing message) and Firefox itself hangs also.  I've tried printing both with CUPS (my default) and with Postscript, and get the same result in either case.  This happens both with the tarball from mozilla.com and with the SuSE rpm from "projects/mozilla".  It happens both with glibc 2.3.5 and glibc 2.4.  glibc seems to be involved because if I start Firefox from a command shell I get the message

  *** glibc detected *** /usr/lib/firefox/firefox-bin: corrupted double-linked list: 0x094110a8 ***


Reproducible: Always

Steps to Reproduce:
1.Print a web page (!!!)
2.
3.

Actual Results:  
The page prints and the print progress window hangs.

Expected Results:  
Not hang!

See the attachments

Comment 1

[Paul Abrahams]

Attachment: Traceback, .xsession-errors

Comment 2

[Svetoslav Milenov]

Same here:

Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.8.0.3) Gecko/20060425 SUSE/1.5.0.3-4.1 Firefox/1.5.0.3

Wondering if this is only a SuSE bug.

Comment 3

[jsmeix]

By accident the original reporter found a strange way how to "solve" it, see
http://lists.suse.com/archive/suse-linux-e/2006-May/1444.html
-----------------------------------------------------------------------------
In my case, I solved the problem but the cure (unintentional) was worse
than the disease: I deleted my .kderc file.
-----------------------------------------------------------------------------
It seems to be caused by some strange interaction with KDE-related stuff.
Isn't there special GTK/Qt stuff used in Mozilla/firefox which does some
strange magic to translate KDE or Qt themes into GTK stuff?
E.g. the GTK-Qt Theme Engine (Suse RPM "gtk-qt-engine") form
http://gtk-qt.freedesktop.org/wiki/Software_2fgtk_2dqt

Comment 4

[Svetoslav Milenov]

Yes, we discussed it on that list, but it did not solve the problem for me, so I guess there should be something else. Anyway, even if a change in the environment (KDE) settings do cause the problem, I still will consider it as a bug in FF. It may be only SUSE specific, so later I'll post it in opensuse's bug tracking system as well.

Comment 5

[Paul Abrahams]

Enough different people have encountered this problem so that it's clear it's not just an oddity of my system, but the problem does seem to be quite environment-dependent and is manifested by updating Firefox.  Since I no longer have the bug -- killing my .kderc made it go away -- I can't do this, but a good way to track it down would be to find an environment in which the bug does not appear and compare it to the environment in which the bug does appear, on the same machine.  In my case, while I still had the bug, if I logged in (under KDE) as root, I did not have the bug.

Comment 6

[Wolfgang Rosenauer [:wolfiR]]

Can someone affected please try if it also is a problem under Gnome or any other window manager besides KDE?

Comment 7

[Svetoslav Milenov]

I tried with WindowMaker, TVM and FVWM and the problem is there. I have no gnome installed.

I also tried in KDE to use not the CUPS/laserjet printer, but Default/postscript - i.e. using KPrint. Still the same.

I just saw that there is MozillaFirefox-debuginfo package provided. If someone has a suggestion how to use it, I can install it and produce more meaningful output.

Comment 8

[Paul Abrahams]

I even had the problem with Print Preview (when I still had it), so I now doubt if  CUPS is involved.  

Comment 9

[Robert Basch]

We have been victimized by what is likely the same problem, on our RHEL-4
machines running 1.5.0.5.  I debugged the problem (using valgrind), and found
a couple of double-free bugs in gfx/src/ps/nsFontMetricsPS.cpp.  One bug is
a failure to increment the reference count on a CharSet following a look-up
in fontconfig.  The other is a failure to zero out an FT_Face pointer after
calling into freetype to free it.  I will provide a patch soon.

Comment 10

[Robert Basch]

Attachment: Fix some double-free bugs

Comment 11

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

So, this bug languished in an incorrect Firefox component without being moved to Core.

Half the patch ended up being bug 294879 and half was bug 355059.

*** This bug has been marked as a duplicate of 294879 ***