Bug 693882 - latest glibc update breaks Evolution on RHEL 5.6
Summary: latest glibc update breaks Evolution on RHEL 5.6
Status: CLOSED DUPLICATE of bug 694655
Alias: None
Product: Red Hat Enterprise Linux 5
Classification: Red Hat
Component: glibc
Version: 5.6
Hardware: x86_64
OS: Linux
Target Milestone: rc
: ---
Assignee: Andreas Schwab
QA Contact: qe-baseos-tools-bugs
: 696431 697095 (view as bug list)
Depends On:
Blocks: 719046 668957
TreeView+ depends on / blocked
Reported: 2011-04-05 19:49 UTC by Andy Feldt
Modified: 2018-12-06 20:45 UTC (History)
38 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Last Closed: 2011-08-16 14:33:18 UTC
Target Upstream Version:

Attachments (Terms of Use)
Backtrace of crash, after clicking on new mail (8.42 KB, application/octet-stream)
2011-04-06 12:48 UTC, Stefan Fuchs
no flags Details
output of valgrind - evolution did not crash (37.02 KB, text/plain)
2011-04-08 09:04 UTC, Jan Kees
no flags Details
user .xsession-errors showing gnome-panel crash (19.39 KB, text/plain)
2011-04-08 20:23 UTC, Andy Feldt
no flags Details
bug buddy report of gnome-panel crash (3.40 KB, text/plain)
2011-04-11 16:47 UTC, Lee Whatley
no flags Details
Valgrind log (5.94 MB, text/plain)
2011-04-19 12:55 UTC, John Bradshaw
no flags Details

Description Andy Feldt 2011-04-05 19:49:41 UTC
Description of problem:
The latest glibc update (to glibc-2.5-58.el5_6.2), or
possibly the glibc-common update of the same version causes
Evolution to not lauch from the desktop icon.  Attempts
to launch from a command line vary in degrees of failure.

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

How reproducible:
Fails on all systems for all users

Steps to Reproduce:
1. yum update glibc
2. log in as ordinary user and click on Evolution icon

Actual results:
no Evolution window appears

Expected results:
expect Evolution window appear

Additional info:
This is with the 64-bit version of Evolution on 64-bit systems.
The version of Evolution is 2.12.3-19

Comment 1 Andy Feldt 2011-04-05 19:51:49 UTC
updated to indicate RHEL version, platform and severity

Comment 2 Andreas Schwab 2011-04-06 06:55:59 UTC
What does it do instead?

Comment 3 Stefan Fuchs 2011-04-06 12:48:37 UTC
Created attachment 490262 [details]
Backtrace of crash, after clicking on new mail

Comment 4 Stefan Fuchs 2011-04-06 12:53:16 UTC
After installing the latetest glibc patch evolution crashes, when I click on new mail. Evolution starts fine. I'm on 32-Bit RHEL 5.6.

Comment 5 Andy Feldt 2011-04-06 14:01:10 UTC

These are all 64-bit systems.  With the 64-bit version of Evolution installed, when a user clicks on the icon, although no window appears (and so from the user's viewpoint, nothing happens) the output of 'ps -ef | grep evolution' shows:

feldt    17050     1  0 08:57 ?        00:00:00 evolution --component=mail
feldt    17068     1  0 08:57 ?        00:00:00 /usr/libexec/evolution-data-server-1.12 --oaf-activate-iid=OAFIID:GNOME_Evolution_DataServer_InterfaceCheck --oaf-ior-fd=41

Furthermore, if I uninstall the 64-bit version of Evolution and then reinstall the 32-bit version, Evolution starts and shows its normal window, reads mail and allow one to view mail.  However, any attempt to send mail then causes Evolution to crash. (Exactly as described above by Steven.)

So, in either case I have a lot of unhappy faculty members! ;-(

Comment 6 Andy Feldt 2011-04-06 19:58:15 UTC
More information - the impacts are somewhat random.  One user who was completely unable to get Evolution to run yesterday tried today and found that it would open and work EXCEPT for clicking on Reply (which crashed Evolution) while creating a new message worked fine.   Another found that it still would not open at all.  Another found that it would open but would crash on a Reply or a New Message.  My own system is running Evolution without any errors at all today (including being able to Reply, create new messages.)

This smacks strongly of memory issues (stack?, heap?)...

Comment 7 Jan Kees 2011-04-07 09:33:38 UTC
After starting evolution from the terminal I get this:

*** glibc detected *** evolution: munmap_chunk(): invalid pointer: 0x000000001014dda0 ***
======= Backtrace: =========
======= Memory map: ========
00400000-0041e000 r-xp 00000000 08:05 6810619                            /usr/bin/evolution
0061d000-00621000 rw-p 0001d000 08:05 6810619                            /usr/bin/evolution
100a9000-102c8000 rw-p 100a9000 00:00 0                                  [heap]
35fa200000-35fa202000 r-xp 00000000 08:05 5434390                        /lib64/libkeyutils-1.2.so
35fa202000-35fa401000 ---p 00002000 08:05 5434390                        /lib64/libkeyutils-1.2.so
35fa401000-35fa402000 rw-p 00001000 08:05 5434390                        /lib64/libkeyutils-1.2.so
35fba00000-35fba68000 r-xp 00000000 08:05 8250957                        /usr/lib64/libbonoboui-2.so.0.0.0
35fba68000-35fbc68000 ---p 00068000 08:05 8250957                        /usr/lib64/libbonoboui-2.so.0.0.0
35fbc68000-35fbc6d000 rw-p 00068000 08:05 8250957                        /usr/lib64/libbonoboui-2.so.0.0.0
35fbe00000-35fbe93000 r-xp 00000000 08:05 8249962                        /usr/lib64/libgnomeui-2.so.0.1600.0
35fbe93000-35fc093000 ---p 00093000 08:05 8249962                        /usr/lib64/libgnomeui-2.so.0.1600.0
35fc093000-35fc099000 rw-p 00093000 08:05 8249962                        /usr/lib64/libgnomeui-2.so.0.1600.0
35fc200000-35fc216000 r-xp 00000000 08:05 8250700                        /usr/lib64/libgnome-2.so.0.1600.0
35fc216000-35fc415000 ---p 00016000 08:05 8250700                        /usr/lib64/libgnome-2.so.0.1600.0
35fc415000-35fc417000 rw-p 00015000 08:05 8250700                        /usr/lib64/libgnome-2.so.0.1600.0
3a50600000-3a5060a000 r-xp 00000000 08:05 8251132                        /usr/lib64/libhal.so.1.0.0
3a5060a000-3a50809000 ---p 0000a000 08:05 8251132                        /usr/lib64/libhal.so.1.0.0
3a50809000-3a5080a000 rw-p 00009000 08:05 8251132                        /usr/lib64/libhal.so.1.0.0
3a52600000-3a52607000 r-xp 00000000 08:05 8252563                        /usr/lib64/libnotify.so.1.1.0
3a52607000-3a52807000 ---p 00007000 08:05 8252563                        /usr/lib64/libnotify.so.1.1.0
3a52807000-3a52808000 rw-p 00007000 08:05 8252563                        /usr/lib64/libnotify.so.1.1.0
3c72c00000-3c72c29000 r-xp 00000000 08:05 8250139                        /usr/lib64/libfontconfig.so.1.1.0
3c72c29000-3c72e29000 ---p 00029000 08:05 8250139                        /usr/lib64/libfontconfig.so.1.1.0
3c72e29000-3c72e33000 rw-p 00029000 08:05 8250139                        /usr/lib64/libfontconfig.so.1.1.0
3c72e33000-3c72e34000 rw-p 3c72e33000 00:00 0 
3c76200000-3c7622d000 r-xp 00000000 08:05 8252541                        /usr/lib64/libgnomecanvas-2.so.0.1400.0
3c7622d000-3c7642c000 ---p 0002d000 08:05 8252541                        /usr/lib64/libgnomecanvas-2.so.0.1400.0
3c7642c000-3c7642e000 rw-p 0002c000 08:05 8252541                        /usr/lib64/libgnomecanvas-2.so.0.1400.0
3c77200000-3c77218000 r-xp 00000000 08:05 8249961                        /usr/lib64/libglade-2.0.so.0.0.7
3c77218000-3c77417000 ---p 00018000 08:05 8249961                        /usr/lib64/libglade-2.0.so.0.0.7
3c77417000-3c77419000 rw-p 00

Comment 8 Andreas Schwab 2011-04-07 11:40:17 UTC
Please try valgrind.

Comment 9 Troy Dawson 2011-04-07 14:11:59 UTC
Just so you know this is widespread, it's affecting my users as well.  (I don't use evolution so it's not affecting me.)  I've been able to reproduce this on every RHEL5 machine I've tried it on, both 32 bit and 64 bit.
I tried using valgrind as you said, ... and it didn't fail.  Tried it again without valgrind, failed as soon as I clicked on "Send Mail".

Comment 10 Akemi Yagi 2011-04-08 01:37:28 UTC
I can reproduce the problem on my RHEL 5.6 (32-bit) system.  I tested evolution before and after updating to the latest glibc. After the update, clicking on "Send Mail" crashed evolution as has been reported here.

Comment 11 Jan Kees 2011-04-08 09:04:03 UTC
Created attachment 490738 [details]
output of valgrind - evolution did not crash

I also tried valgrind and then evolution also didn't fail

Comment 12 Andy Feldt 2011-04-08 20:23:26 UTC
Created attachment 490868 [details]
user .xsession-errors showing gnome-panel crash

We have now had users (so far only of our 32-bit systems) report that their Gnome panels are frozen.  The .xsession-error files for these users show that gnome-panel has crashed with the initial line stating:

*** glibc detected *** gnome-panel: malloc(): memory corruption: 0x09e20e40 ***

which is then followed by a backtrace.  This does not happen to all users or even all the time to a given user.

So, the impact of the glibc update appears to be broader than just the impact on Evolution.

Comment 13 Orrie Gartner 2011-04-08 21:08:07 UTC
We are seeing identical behaviour on our 32-bit RHEL 5.6 systems with:

*** glibc detected *** gnome-panel: malloc(): memory corruption: 0x09e20e40 ***

the gnome-panel crashes - not on every login, but most.  The evolution issue is occuring on 64-bit systems as well.  We have a temporary workaround as follows:

 sudo yum downgrade glibc\*

followed by excluding glibc in /etc/yum.conf until this is fixed.  We can quickly duplicate the problem if after running the "yum downgrade", we "yum install glibc\*" so this is easily reproducible.

Comment 14 Lee Whatley 2011-04-11 16:45:39 UTC
We are seeing the gnome-panel freeze/crash issue here as well.  So far we have only seein it on 32-bit systems.  Attaching the output that "bug buddy" produces when it crashes...

Comment 15 Lee Whatley 2011-04-11 16:47:30 UTC
Created attachment 491280 [details]
bug buddy report of gnome-panel crash

Comment 16 Troy Dawson 2011-04-12 16:09:06 UTC
Here is the results of my research into this bug.

I have recompiled the glibc 6 different ways.
1 - Patch 249 not applied
2 - Patch 250 not applied
3 - Patch 251 not applied
4 - Patch 252 not applied
5 - Patches 249,250,251,252 not applied
6 - Changed patch 245 back to what it was in glibc-2.5-58, and Patch 249 not applied.  (I tried to just change patch 245 back, but then patch 249 wouldn't apply)

In all of my tests, evolution failed as before (when pushing the send button), except for test 6.
Test 6 worked correctly.

So, in summary, in my opinion, there is a bug in the new patch 245.

(For those that don't know, patch 245 was changed between glibc-2.5-58 and glibc-2.5-58.el5_6.2

I haven't tracked down the real bug.  I'm just sharing my results.

Comment 17 Akemi Yagi 2011-04-12 17:10:11 UTC
I see that patch 245 is glibc-rh643822.patch. Unfortunately bugzilla #643822 is not accessible. Could someone from Red Hat please tell us if there is any risk factor (vulnerability) by temporarily removing this patch?

Comment 18 Jack Neely 2011-04-14 20:41:35 UTC
I've opened case # 00451533 with Red Hat support on this issue as well.

Comment 19 Lee Whatley 2011-04-15 19:21:56 UTC
I've also opened case # 00452071 with Red Hat in regards specifically to the gnome-panel issue on 32-bit systems.

Comment 20 Jenny Aquilino 2011-04-15 20:08:56 UTC
I too have multiple systems that are experiencing this issue and have opened a case # 00452086 with Red Hat.

Comment 21 Leonard den Ottolander 2011-04-17 14:39:16 UTC
On a Sempron system (i686) I experienced the issues with evolution crashing and gnome panels not showing some of the times only after I recovered from a broken boot loader (hanging at "GRUB") after the April 15th update on CentOS 5.6.

Could it be that the presence of this broken glibc caused this boot loader issue in combination with a kernel update? Apart from the glibc update last Friday's update also contained kernel-2.6.18-238.9.1.el5 and mkinitrd- a.o.

Has anyone else seen grub breakage in relation to this issue or is this just a coincidence?

Comment 22 John Bradshaw 2011-04-19 12:26:14 UTC
Same issue here on 32-bit evolution and gnome panels in 5.6
Evolution crashes on new mail (either the button or the menu item).
Output below.

Was all rock solid with 5.5

Seems to be non-printing characters in this error line... not sure if that's in the filename when searching for library or just in the error message...

/usr/lib/bonobo/monikers/libmoniker_std_2.so�: No such file or directory.

[user@hostname~]$ xxd
/usr/lib/bonobo/monikers/libmoniker_std_2.so�: No s
0000000: 2f75 7372 2f6c 6962 2f62 6f6e 6f62 6f2f  /usr/lib/bonobo/
0000010: 6d6f 6e69 6b65 7273 2f6c 6962 6d6f 6e69  monikers/libmoni
0000020: 6b65 725f 7374 645f 322e 736f efbf bd3a  ker_std_2.so...:
0000030: 204e 6f20 730a                            No s.

Comment 23 John Bradshaw 2011-04-19 12:55:43 UTC
Created attachment 493183 [details]
Valgrind log

Valgrind log of the following steps:
1. Open evolution
2. New mail (from button)
3. Compose and send mail
4. Close evolution.

This works with valgrind running, without valgrind Evolution crashes when clicking the new mail button or selecting File/New/Mail Message menu item.

Comment 24 John Bradshaw 2011-04-19 14:49:35 UTC
yum downgrade glibc fixed Evolution and panel problems.  Latest version was working fine on my box until today. 32-bit.

Comment 25 Boris Folgmann 2011-04-19 15:44:01 UTC
gnome-panel was also crashing on my 34-bit Athlon system.

yum downgrade glibc* nscd nss_ldap

And adding

to /etc/yum.conf

I hope a fix will be released real soon.

Comment 26 John Bradshaw 2011-04-20 08:44:51 UTC
Evolution run per comment 23 with glibc-2.5-58 and glibc-common-2.5-58.
Works fine without valgrind and not getting the "no such file or directory" error.
System back to normally stablity (panel works, evolution mail notification popup back to bottom right of screen instead of top left, etc.).

[jbradsha@host ~]$ evolution 
** (evolution:4168): DEBUG: mailto URL command: evolution --component=mail %s
** (evolution:4168): DEBUG: mailto URL program: evolution
libnm_glib_nm_state_cb: dbus returned an error.
  (org.freedesktop.DBus.Error.ServiceUnknown) The name org.freedesktop.NetworkManager was not provided by any .service files
Increasing itip formatter search count to 1
Increasing itip formatter search count to 2
Decreasing itip formatter search count to 1
Decreasing itip formatter search count to 0
(evolution:4168): e-data-server-DEBUG: Loading categories from "/home/jbradsha/.evolution/categories.xml"
(evolution:4168): e-data-server-DEBUG: Loaded 29 categories
[jbradsha@host ~]$ 

Hope this helps narrow down the problem.

Comment 28 bugreports2005 2011-04-26 05:21:30 UTC
Is there any news on this? The issue with gnome-panel breaks the desktop pretty bad.

Comment 29 Need Real Name 2011-04-26 14:15:01 UTC
Cannot access the gnome pannel (can see it but not click on it) from any remote X session. A downgrade to the previous glibc and everything works ok.

Comment 30 John Bradshaw 2011-04-26 22:50:56 UTC
This errata https://rhn.redhat.com/errata/RHBA-2011-0466.html seems to be the fix... can RH confirm this is for the same issue?  (The corrupted library filenames in comment 22 could be due to this). Thanks.

Comment 31 Troy Dawson 2011-04-27 14:28:15 UTC
I can verify that glibc.i686 0:2.5-58.el5_6.3 (from https://rhn.redhat.com/errata/RHBA-2011-0466.html) fixes the evolution problem.

Comment 32 Andy Feldt 2011-05-03 20:20:46 UTC
Does anyone know if the equivalent glibc update for RHEL 6 has the same problem that was exposed under RHEL 5?  We only have one RHEL 6 system and it is our primary server, so we can't mess around to find out...  (The test system is still on my "todo" list! ;-)  The package would be:


which has the same CVE fixes that the broken glibc had.

Comment 33 Greg Matthews 2011-05-31 14:53:43 UTC
still seeing occasional gnome-panel lock-ups. Also, our data acquisition software GDA still crashes with this version, while fully stable with glibc-2.5-49.el5_5.6.i686

Comment 34 Andreas Schwab 2011-08-16 14:33:18 UTC

*** This bug has been marked as a duplicate of bug 694655 ***

Comment 35 Michael Firth 2011-08-16 14:43:16 UTC

Bug 694655 doesn't seem to be visible - can some details of it be made available?



Comment 36 Alan Bartlett 2012-09-19 18:45:00 UTC
(In reply to comment #34)
> *** This bug has been marked as a duplicate of bug 694655 ***

Please do not mark this bug as a duplicate of bug 694655 without ensuring that bug 694655 is accessible to persons on the CC-list of this bug.

Either open up bug 694655 or copy the contents thereof to this report and mark bug 694655 as a duplicate of this.

<Big sigh!>

Comment 37 Ray Strode [halfline] 2014-03-10 15:40:00 UTC
*** Bug 697095 has been marked as a duplicate of this bug. ***

Comment 38 Ray Strode [halfline] 2014-03-10 15:40:05 UTC
*** Bug 696431 has been marked as a duplicate of this bug. ***

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