Bugzilla – Bug 398932
"trap divide error" appears when launching the Virtual Machine Manager
Last modified: 2009-05-29 16:16:18 UTC
I have again updated my system with "zypper dup" and rebooted the server. I am running now with the kernel-xen version uname -a Linux server 2.6.25.5-1.1-xen #1 SMP 2008-06-07 01:55:22 +0200 x86_64 x86_64 x86_64 GNU/Linux I begin the process to create a Xen guest with "xm create" xm create -c /home/test/xen/os11.cfg I see at the Console the continuation until NET: Registered protocol family 17 RPC: Registered udp transport module. RPC: Registered tcp transport module. loop: module loaded xen-vbd: registered block device major 202 blkfront: xvda: barriers enabled xvda: unknown partition table netfront: Initialising virtual ethernet driver. Then I make a connect to the server with VNC screen and launch the "Virtual Machine Manager" to see the Guest Console. But now not as before the program starts quickly and then stops then. I see for only a moment the GUI of the Virtual Machine Manager. The output in the /var/log/messages System log shows me Jun 10 07:17:55 server kernel: /usr/share/virt[4156] trap divide error ip:7f326840a5ac sp:7fff7d135c70 error:0 in libvirt.so.0.4.0[7f32683e9000+5e000] Jun 10 07:18:09 server gconfd (root-4147): GConf server is not in use, shutting down. Jun 10 07:18:09 server gconfd (root-4147): Exiting I can repeat this problem each time. In this way the Virtual Machine Manager can not be used at all.
I have now installed our System as the FINAL OpenSuse 11.0 and can say that the same probelm happens. In the System log I me see as before Jun 19 08:34:35 server gconfd (root-4377): starting (version 2.22.0), pid 4377 user 'root' Jun 19 08:34:35 server gconfd (root-4377): Resolved address "xml:readonly:/etc/gconf/gconf.xml.mandatory" to a read-only configuration source at position 0 Jun 19 08:34:35 server gconfd (root-4377): Resolved address "xml:readwrite:/root/.gconf" to a writable configuration source at position 1 Jun 19 08:34:35 server gconfd (root-4377): Resolved address "xml:readonly:/etc/gconf/gconf.xml.defaults" to a read-only configuration source at position 2 Jun 19 08:34:35 server gconfd (root-4377): Resolved address "xml:readonly:/etc/gconf/gconf.xml.vendor" to a read-only configuration source at position 3 Jun 19 08:34:35 server gconfd (root-4377): Resolved address "xml:readonly:/etc/gconf/gconf.xml.schemas" to a read-only configuration source at position 4 Jun 19 08:34:35 server kernel: /usr/share/virt[4375] trap divide error ip:7fe28d5085ac sp:7fffa2231cd0 error:0 in libvirt.so.0.4.0[7fe28d4e7000+5e000]
Are you running the Gnome desktop on your host? Also, please attach the /root/.virt-manager/virt-manager.log file.
Hello Doug! I am not using Gnome Desktop. Instead I am using XFCE4 Desktop. It is smaller and simpler and I think then better for resources in Domain 0. I can use it with success for other Distrubitions, for example Centos and Ubuntu, with Xen. Also in earlier Opensuse Factory I was successfull to use it. Only then there were some changes in Opensuse before the Final release. As you have made the request the outputt of the log file in the same event is [Thu, 19 Jun 2008 15:35:15 virt-manager 22971] INFO (virt-manager:126) Application startup [Thu, 19 Jun 2008 15:35:16 virt-manager 22972] WARNING (keyring:31) No support for gnome-keyring [Thu, 19 Jun 2008 15:35:17 virt-manager 22972] DEBUG (engine:74) About to connect to uris ['xen:///'] [Thu, 19 Jun 2008 15:35:17 virt-manager 22972] DEBUG (manager:112) minimize_minutes set to 120 [Thu, 19 Jun 2008 15:35:17 virt-manager 22972] DEBUG (connection:311) Scheduling background open thread for xen:/// [Thu, 19 Jun 2008 15:35:17 virt-manager 22972] DEBUG (connection:318) Background thread is running [Thu, 19 Jun 2008 15:35:17 virt-manager 22972] DEBUG (connection:347) Background open thread complete, scheduling notify [Thu, 19 Jun 2008 15:35:17 virt-manager 22972] DEBUG (connection:356) Notifying open result [Thu, 19 Jun 2008 15:35:17 virt-manager 22972] DEBUG (manager:427) About to append vm: Domain-0 [Thu, 19 Jun 2008 15:35:17 virt-manager 22972] DEBUG (manager:417) VM Domain-0 started
Is maybe there some additionale information that I can add to this Bug Report? There is no more news in time so I thought I shoud ask again. Its now a situation where I cannot work with the Release of Opensuse because of this and I am hopefull to help with the solution any way.
I have a problem with exactly same symptoms, here's some information that mostly repeats what the original submitter provided. I can provide any additional info that might help in diagnosing, or give access to the offending box to the dev if the issue is not reproducible elsewhere. Thanks --------------------- Hardware: ASUS P5K Deluxe, Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz --------------------- # uname -a Linux gateway 2.6.25.5-1.1-xen #1 SMP 2008-06-07 01:55:22 +0200 x86_64 x86_64 x86_64 GNU/Linux --------------------- --------------------- /var/log/messages --------------------- Jun 28 19:33:07 gateway gconfd (root-22054): starting (version 2.22.0), pid 22054 user 'root' Jun 28 19:33:07 gateway gconfd (root-22054): Resolved address "xml:readonly:/etc/gconf/gconf.xml.mandatory" to a read-only configuration source at position 0 Jun 28 19:33:07 gateway gconfd (root-22054): Resolved address "xml:readwrite:/root/.gconf" to a writable configuration source at position 1 Jun 28 19:33:07 gateway gconfd (root-22054): Resolved address "xml:readonly:/etc/gconf/gconf.xml.defaults" to a read-only configuration source at position 2 Jun 28 19:33:07 gateway gconfd (root-22054): Resolved address "xml:readonly:/etc/gconf/gconf.xml.vendor" to a read-only configuration source at position 3 Jun 28 19:33:07 gateway gconfd (root-22054): Resolved address "xml:readonly:/etc/gconf/gconf.xml.schemas" to a read-only configuration source at position 4 Jun 28 19:33:08 gateway kernel: /usr/share/virt[22050] trap divide error ip:7fc7b6bd95ac sp:7fffcb3c9720 error:0 in libvirt.so.0.4.0[7fc7b6bb8000+5e000]
No idea what information is needed by Coolo.
(In reply to comment #6 from Andreas Jaeger) > No idea what information is needed by Coolo. > You would also have the same idea if you read as I have read in the Devel list once weeks have passed, http://lists.opensuse.org/opensuse-factory/2008-06/msg00893.html, the instructions by Mr. Cristian Rodriguez from Novell say "set the bug report NEEDEDINFO to coolo@novell" I have not I think spoken with you Andreas so it is I think much more helpful if you do not only change things in my report without any question or explanation. If it is incorrect what are then the instructions that you say?
Tesuichi, Coolo is on vacation and I'm his proxy - so I'm authorized to do this ;-) What Cristian missed to say was: Add a comment on what you think is missing and then set NEEDINFO to Coolo. Doug, could you have a look at this bug, please? Does it reproduce for you?
(In reply to comment #8 from Andreas Jaeger) > Tesuichi, Coolo is on vacation and I'm his proxy - so I'm authorized to do this Its polite to know this earlier than such changes. Thank you for your explaination. > What Cristian missed to say was: Add a comment on what you think is missing and > then set NEEDINFO to Coolo. I am sorry I have also not explained more thinking that its clear. The message from C Rodriguez made a suggestion if we have no reponse or solution after couple of weeks, direct communications with the Assigned, and so fourth. The missing thing only is a farther discussion for 10 or more days, and any answer to my question for how to help with additional information. I understand that Novell will do or not do things because of its own schedules and opinions. For me I have made an arrangement with my Manager to use Opensuse because the Official Releases are strongly trusted. This is now difficult only for me and, it seems for Mr Dmitry Nikitin too, because I can not RollOut the Opensuse Release for using Xen. He has shown me that Centos Version 5.2 is available and that there are no problems so I think the problem indeed is in Opensuse. So I am trying here again to get answers and help if its possible I can.
We have had trouble reproducing this, but we can see the problem now. It looks like an error in the libvirt module's virNodeGetInfo() function, so I'll get Jim to look at this with me on Friday, when he returns. In the mean time, can you verify your version of the "libvirt-python" rpm? I think it should be 0.4.0-59.1 but I want to make sure we know, in case it matters.
*** Bug 401600 has been marked as a duplicate of this bug. ***
Hello Doug (In reply to comment #10 from Doug Merrill) > In the mean time, can you verify your version of the "libvirt-python" rpm? I > think it should be 0.4.0-59.1 but I want to make sure we know, in case it > matters. Yes I certainly have the same version. rpm -qa libvirt-python libvirt-python-0.4.0-59.1
*** Bug 405767 has been marked as a duplicate of this bug. ***
For other reasons, I've updated libvirt to latest release in my OBS home project. RPMs for 11.0 are available here http://download.opensuse.org/repositories/home:/jfehlig/openSUSE_11.0/ Could you try 0.4.4? Would be interesting to know if solves your problem. However, I found nothing regarding "trap divide error" in libvirt-devel archives so I would be surprised if the new version helped. NB, I have not tested the 0.4.4 RPMs. Will do so upon returning to the office on Monday. I'll also have access to a machine exhibiting the issue then.
Yes it works with your packaged Libvirt 0.4.4, I have no more floating point exception when I start Virt-manager. I think the problem is specific to Libvirt 0.4.0 (more info on https://bugzilla.novell.com/show_bug.cgi?id=405767).
Hello Jim (In reply to comment #15 from Jim Fehlig) > Could you try 0.4.4? Would be interesting to know if solves your problem. > However, I found nothing regarding "trap divide error" in libvirt-devel > archives so I would be surprised if the new version helped. I have installed the RPM files you've made availabel. libvirt-0.4.4-3.1.x86_64.rpm libvirt-doc-0.4.4-3.1.x86_64.rpm libvirt-python-0.4.4-3.1.x86_64.rpm And also I have reinstalled zypper install virt-manager Now I can launch the Virtual Machine Manager application in the XFCE Desktop and it launches permanenntly. It doesn't disappear at once any more. But I do also see in the logs I have cheked before /var/log/messages Jul 6 14:48:02 server gconfd (root-23674): starting (version 2.22.0), pid 23674 user 'root' Jul 6 14:48:02 server gconfd (root-23674): Resolved address "xml:readonly:/etc/gconf/gconf.xml.mandatory" to a read-only configuration source at position 0 Jul 6 14:48:02 server gconfd (root-23674): Resolved address "xml:readwrite:/root/.gconf" to a writable configuration source at position 1 Jul 6 14:48:02 server gconfd (root-23674): Resolved address "xml:readonly:/etc/gconf/gconf.xml.defaults" to a read-only configuration source at position 2 Jul 6 14:48:02 server gconfd (root-23674): Resolved address "xml:readonly:/etc/gconf/gconf.xml.vendor" to a read-only configuration source at position 3 Jul 6 14:48:02 server gconfd (root-23674): Resolved address "xml:readonly:/etc/gconf/gconf.xml.schemas" to a read-only configuration source at position 4 /root/.virt-manager/virt-manager.log [Sun, 06 Jul 2008 14:48:00 virt-manager 23671] INFO (virt-manager:126) Application startup [Sun, 06 Jul 2008 14:48:01 virt-manager 23672] WARNING (keyring:31) No support for gnome-keyring [Sun, 06 Jul 2008 14:48:02 virt-manager 23672] DEBUG (engine:74) About to connect to uris ['xen:///'] [Sun, 06 Jul 2008 14:48:03 virt-manager 23672] DEBUG (manager:112) minimize_minutes set to 120 [Sun, 06 Jul 2008 14:48:03 virt-manager 23672] DEBUG (connection:311) Scheduling background open thread for xen:/// [Sun, 06 Jul 2008 14:48:03 virt-manager 23672] DEBUG (connection:318) Background thread is running [Sun, 06 Jul 2008 14:48:03 virt-manager 23672] DEBUG (connection:347) Background open thread complete, scheduling notify [Sun, 06 Jul 2008 14:48:03 virt-manager 23672] DEBUG (connection:356) Notifying open result [Sun, 06 Jul 2008 14:48:03 virt-manager 23672] WARNING (connection:487) Unable to list active networks [Sun, 06 Jul 2008 14:48:03 virt-manager 23672] WARNING (connection:491) Unable to list inactive networks [Sun, 06 Jul 2008 14:48:03 virt-manager 23672] DEBUG (manager:427) About to append vm: Domain-0 [Sun, 06 Jul 2008 14:48:03 virt-manager 23672] DEBUG (manager:417) VM Domain-0 started [Sun, 06 Jul 2008 14:48:03 virt-manager 23672] WARNING (connection:487) Unable to list active networks [Sun, 06 Jul 2008 14:48:03 virt-manager 23672] WARNING (connection:491) Unable to list inactive networks [Sun, 06 Jul 2008 14:48:04 virt-manager 23672] WARNING (connection:487) Unable to list active networks [Sun, 06 Jul 2008 14:48:04 virt-manager 23672] WARNING (connection:491) Unable to list inactive networks [Sun, 06 Jul 2008 14:48:05 virt-manager 23672] WARNING (connection:487) Unable to list active networks [Sun, 06 Jul 2008 14:48:05 virt-manager 23672] WARNING (connection:491) Unable to list inactive networks [Sun, 06 Jul 2008 14:48:06 virt-manager 23672] WARNING (connection:487) Unable to list active networks [Sun, 06 Jul 2008 14:48:06 virt-manager 23672] WARNING (connection:491) Unable to list inactive networks And so this output continues untill I stop the VirtualManager application. So I dont see the "trap divide" Error for the moment but these other messages.
*** Bug 405779 has been marked as a duplicate of this bug. ***
Created attachment 226355 [details] Patch to fix divide by zero error in libvirt 0.4.0 OK, I've gotten to the bottom of this issue in libvirt 0.4.0. With this patch I'm able to start virt-manager fine on several hosts that previously exhibited the divide by zero error.
(In reply to comment #18 from Tetsuichi Asaka) > [Sun, 06 Jul 2008 14:48:03 virt-manager 23672] WARNING (connection:487) Unable > to list active networks > [Sun, 06 Jul 2008 14:48:03 virt-manager 23672] WARNING (connection:491) Unable > to list inactive networks > [Sun, 06 Jul 2008 14:48:04 virt-manager 23672] WARNING (connection:487) Unable > to list active networks > [Sun, 06 Jul 2008 14:48:04 virt-manager 23672] WARNING (connection:491) Unable > to list inactive networks ... > > And so this output continues untill I stop the VirtualManager application. > > So I dont see the "trap divide" Error for the moment but these other messages. That output is indicative of libvirtd not running. Do you have libvirtd daemon running? I might have some problems in the 0.4.4 spec file since I took a stab at making the same spec file work for all products (SLES*, openSUSE*) and on arch's not supported by Xen. Anyhow, I'll attach the patched 0.4.0 rpms for your testing pleasure.
Created attachment 226358 [details] Patched libvirt 0.4.0 for x86_64
Created attachment 226359 [details] libvirt-python for x86_64
Hello Jim (In reply to comment #21 from Jim Fehlig) > That output is indicative of libvirtd not running. Do you have libvirtd daemon > running? I might have some problems in the 0.4.4 spec file since I took a stab > at making the same spec file work for all products (SLES*, openSUSE*) and on > arch's not supported by Xen. > > Anyhow, I'll attach the patched 0.4.0 rpms for your testing pleasure. I have installed the new RPMs. rpm -qa libvirt\* virt-manager libvirt-0.4.0-60 libvirt-python-0.4.0-60 virt-manager-0.5.3-21.1 On repetition in the logs I see, /var/log/messages Jul 7 15:01:56 server gconfd (root-12141): starting (version 2.22.0), pid 12141 user 'root' Jul 7 15:01:56 server gconfd (root-12141): Resolved address "xml:readonly:/etc/gconf/gconf.xml.mandatory" to a read-only configuration source at position 0 Jul 7 15:01:56 server gconfd (root-12141): Resolved address "xml:readwrite:/root/.gconf" to a writable configuration source at position 1 Jul 7 15:01:56 server gconfd (root-12141): Resolved address "xml:readonly:/etc/gconf/gconf.xml.defaults" to a read-only configuration source at position 2 Jul 7 15:01:56 server gconfd (root-12141): Resolved address "xml:readonly:/etc/gconf/gconf.xml.vendor" to a read-only configuration source at position 3 Jul 7 15:01:56 server gconfd (root-12141): Resolved address "xml:readonly:/etc/gconf/gconf.xml.schemas" to a read-only configuration source at position 4 /root/.virt-manager/virt-manager.log [Mon, 07 Jul 2008 15:01:54 virt-manager 12138] INFO (virt-manager:126) Application startup [Mon, 07 Jul 2008 15:01:55 virt-manager 12139] WARNING (keyring:31) No support for gnome-keyring [Mon, 07 Jul 2008 15:01:56 virt-manager 12139] DEBUG (engine:74) About to connect to uris ['xen:///'] [Mon, 07 Jul 2008 15:01:56 virt-manager 12139] DEBUG (manager:112) minimize_minutes set to 120 [Mon, 07 Jul 2008 15:01:56 virt-manager 12139] DEBUG (connection:311) Scheduling background open thread for xen:/// [Mon, 07 Jul 2008 15:01:56 virt-manager 12139] DEBUG (connection:318) Background thread is running [Mon, 07 Jul 2008 15:01:56 virt-manager 12139] DEBUG (connection:347) Background open thread complete, scheduling notify [Mon, 07 Jul 2008 15:01:56 virt-manager 12139] DEBUG (connection:356) Notifying open result [Mon, 07 Jul 2008 15:01:56 virt-manager 12139] WARNING (connection:487) Unable to list active networks [Mon, 07 Jul 2008 15:01:56 virt-manager 12139] WARNING (connection:491) Unable to list inactive networks [Mon, 07 Jul 2008 15:01:56 virt-manager 12139] DEBUG (manager:427) About to append vm: Domain-0 [Mon, 07 Jul 2008 15:01:56 virt-manager 12139] DEBUG (manager:417) VM Domain-0 started [Mon, 07 Jul 2008 15:01:57 virt-manager 12139] WARNING (connection:487) Unable to list active networks [Mon, 07 Jul 2008 15:01:57 virt-manager 12139] WARNING (connection:491) Unable to list inactive networks And so on repeatedley. For the libvirtd status I can check, I think correctly, service libvirtd status Checking status of libvirtd unused chkconfig --list | grep virt libvirtd 0:off 1:off 2:off 3:on 4:off 5:on 6:off So its not running. Is it required? I know from my reading that "libvirtd provides virbr0, an isolated netdev connected to external using forwarding+NAT" and since I will not use virbr0, but only several other bridge interfaces I will name, then I did think it is not required. It has never earlier been something required for me.
(In reply to comment #25 from Tetsuichi Asaka) > On repetition in the logs I see, > > /var/log/messages > Jul 7 15:01:56 server gconfd (root-12141): starting (version 2.22.0), pid > 12141 user 'root' > Jul 7 15:01:56 server gconfd (root-12141): Resolved address > "xml:readonly:/etc/gconf/gconf.xml.mandatory" to a read-only configuration > source at position 0 > Jul 7 15:01:56 server gconfd (root-12141): Resolved address > "xml:readwrite:/root/.gconf" to a writable configuration source at position 1 > Jul 7 15:01:56 server gconfd (root-12141): Resolved address > "xml:readonly:/etc/gconf/gconf.xml.defaults" to a read-only configuration > source at position 2 > Jul 7 15:01:56 server gconfd (root-12141): Resolved address > "xml:readonly:/etc/gconf/gconf.xml.vendor" to a read-only configuration source > at position 3 > Jul 7 15:01:56 server gconfd (root-12141): Resolved address > "xml:readonly:/etc/gconf/gconf.xml.schemas" to a read-only configuration source > at position 4 I have no clue what this stuff is. Doug, is this something virt-manager is causing? > [Mon, 07 Jul 2008 15:01:57 virt-manager 12139] WARNING (connection:487) Unable > to list active networks > [Mon, 07 Jul 2008 15:01:57 virt-manager 12139] WARNING (connection:491) Unable > to list inactive networks > > And so on repeatedley. > > For the libvirtd status I can check, I think correctly, > > service libvirtd status > Checking status of libvirtd unused > > chkconfig --list | grep virt > libvirtd 0:off 1:off 2:off 3:on 4:off 5:on 6:off So it should be running in run levels 3 and 5. You certainly should be in one of those run levels :-). > So its not running. Is it required? No, not required unless you want to manage the node remotely. You can connect virt-manager to a remote node via libvirtd running on that remote node. > I know from my reading that "libvirtd provides virbr0, an isolated netdev > connected to external using forwarding+NAT" and since I will not use virbr0, > but only several other bridge interfaces I will name, then I did think it is > not required. It has never earlier been something required for me. Well, you do get a noisy log file without it running.
(In reply to comment #26 from James Fehlig) > I have no clue what this stuff is. Doug, is this something virt-manager is > causing? I have read "Most Gnome applications hook into gconfd which is a daemon that monitors a registry-like database of user settings" So is it maybe a result from my earlier reply to Doug Merrill? "I am not using Gnome Desktop. Instead I am using XFCE4 Desktop." > > service libvirtd status > > Checking status of libvirtd unused > > > > chkconfig --list | grep virt > > libvirtd 0:off 1:off 2:off 3:on 4:off 5:on 6:off > > So it should be running in run levels 3 and 5. You certainly should be in one > of those run levels :-). As you say it should be running, because the '3:on' and '5:on' settings were made, I was suspicious so I restarted the server. After the restart, I now see, service libvirtd status Checking status of libvirtd running So it is now running. And now as above, checking the logs, I see the difference, /var/log/messages Jul 7 15:37:50 server gconfd (root-4133): starting (version 2.22.0), pid 4133 user 'root' Jul 7 15:37:51 server gconfd (root-4133): Resolved address "xml:readonly:/etc/gconf/gconf.xml.mandatory" to a read-only configuration source at position 0 Jul 7 15:37:51 server gconfd (root-4133): Resolved address "xml:readwrite:/root/.gconf" to a writable configuration source at position 1 Jul 7 15:37:51 server gconfd (root-4133): Resolved address "xml:readonly:/etc/gconf/gconf.xml.defaults" to a read-only configuration source at position 2 Jul 7 15:37:51 server gconfd (root-4133): Resolved address "xml:readonly:/etc/gconf/gconf.xml.vendor" to a read-only configuration source at position 3 Jul 7 15:37:51 server gconfd (root-4133): Resolved address "xml:readonly:/etc/gconf/gconf.xml.schemas" to a read-only configuration source at position 4 /root/.virt-manager/virt-manager.log [Mon, 07 Jul 2008 15:37:49 virt-manager 4130] INFO (virt-manager:126) Application startup [Mon, 07 Jul 2008 15:37:50 virt-manager 4131] WARNING (keyring:31) No support for gnome-keyring [Mon, 07 Jul 2008 15:37:51 virt-manager 4131] DEBUG (engine:74) About to connect to uris ['xen:///'] [Mon, 07 Jul 2008 15:37:51 virt-manager 4131] DEBUG (manager:112) minimize_minutes set to 120 [Mon, 07 Jul 2008 15:37:51 virt-manager 4131] DEBUG (connection:311) Scheduling background open thread for xen:/// [Mon, 07 Jul 2008 15:37:51 virt-manager 4131] DEBUG (connection:318) Background thread is running [Mon, 07 Jul 2008 15:37:51 virt-manager 4131] DEBUG (connection:347) Background open thread complete, scheduling notify [Mon, 07 Jul 2008 15:37:51 virt-manager 4131] DEBUG (connection:356) Notifying open result [Mon, 07 Jul 2008 15:37:51 virt-manager 4131] DEBUG (manager:427) About to append vm: Domain-0 [Mon, 07 Jul 2008 15:37:51 virt-manager 4131] DEBUG (manager:417) VM Domain-0 started And nothing repetitive any more. So I think maybe the last problem was because the libvirtd was not running, as you said.
Jim, your patch seems to correct the problem on my fresh OpenSUSE 11.0 AMD64 install. When it will be integrated on the OpenSUSE update repository ?
Hello all, I had the very same problem as Tetsuichi did and found this bug on an HP AMD Dual-Core AMD Opteron(tm) Processor 1220 based system (HP xw4550). I am confirming that installing the two attached patched RPMs (libvirt and libvirt-python) fixed the problem for me. The Virtual Machine Manager in Yast now runs. Thank you all for your help (and for everything you do, Mr. Jaeger). If I can provide anything helpful, please let me know.
released
*** Bug 391157 has been marked as a duplicate of this bug. ***