Unkillable OpenOffice.org process in 5.3p2
jeff at sailorfej.net
Sat Jan 15 13:23:54 PST 2005
When you rebooted, was the computer actually able to halt on its own?
Every time mine has locked up, I have had to do a hard power down, the
OS is unable to kill the process even when halting. I have noticed that
the OO app freezes typically after having Mozilla to use java and/or
adobe plugins, I am not sure which is the trigger yet. I am wondering
if it is a problem with the interaction between the OO and java, or OO
the pdf libraries, or a some shared library that facilitates the
interaction between these components. I am afraid I am not programmer
my self so I can't be of much help. But I have attached my previous
email, and the response I got below your quoted mail.
> Peter Jeremy wrote:
>> On Sat, 2005-Jan-15 10:28:10 -0800, Jeffrey Williams wrote:
>>>Have you had any good responses to this, I have been having the exact
>>>same problem. The last time I posted it, I was told it was a threading
>>>problem, and that if I cvsuped with RELENG_5 instead of RELENG_5_3 that
>>>would fix it. Unfortunately I haven't been able to test it, as from
>>>what I have seen RELENG_5 just isn't stable enough for my primary
>> Actually, I've had no response at all to this. My system hung with
>> a filesystem deadlock yesterday so I've had to reboot anyway. (I'm
>> not sure if the deadlock is related to OOo, but it seems possible).
>> I might look at the differences between RELENG_5 and RELENG_5_3 and
>> see if there's anything obvious.
My first message posted to the freebsd-openoffice mail list:
> Normally I would take this to the openoffice support forum, the problem is unusual enough that I thought you guys might want to know. I could not find any mentions of a similar problem anywhere else. The problem is severe, but has only occurred twice and is not consistent repeatable.
> I am running FreeBSD 5.3, Xorg-6.7.0, fluxbox-0.1.14_2, and jdk-1.4.2p6_7. I installed Openoffice-1.1.3_1 from the port.
> When I started "openoffice-1.1.3-swriter", the splash screen appears fine, when the window title bar and frame appears with out contents, and hangs. At this point, the second time it happened I waited for about half an hour with no change. In both cases it was when I tried to kill the process that the fun really began.
> The following are the processes that appear in the "ps -waux" listing:
> user xxxx1 0.0 0.4 1672 972 ?? Is 9:42PM 0:00.01 /bin/sh -c openoffice-1.1.3-swriter
> user xxxx2 0.0 0.4 1672 1000 ?? I 9:42PM 0:00.01 /bin/sh /usr/local/bin/openoffice-1.1.3-swriter
> user xxxx6 0.0 22.6 98072 58268 ?? S 9:42PM 0:08.81 /usr/local/OpenOffice.org1.1.3/program/soffice.bin -writer
> Now normally when killing a hung program, will start with the last process created and work my way backwards until the whole thing is gone. However when I tried that this time, the third process could not be killed, even with signal 9. So tried kill the other two, they died just fine, the third would still not die. So I tried shutting down X, which shutdown fine, but still left "soffice.bin -writer" still running, I tried the killing with signal 9 again, again nothing happened. At this the first time, I tried "shutdown -r now", the only unusual message (and this if from memory so forgive me if it is not exact) was at the point where the system drops into single user mode, it was:
> "some processes would not stop, ps axl notified"
> at which point it goes on clearing buffers and syncing the disks, the disk sync completes, there is a message about no buffers in use, at which point it hangs, and will not complete the shutdown.
> The second time openoffice writer hung, instead of trying to shutdown and reboot all at once, I used "shutdown now" to drop into single user mode. In single user mode using "ps -waux" the "soffice.bin -writer" process was still showing up, I tried one more time to kill it, and still it would not die. At this point I tried to complete the shutdown with "halt", as the first the shutdown hung immediately upon completing the disk sync.
> At this stage in both instances when the shutdown was hung, I tried turning the computer off with the power button on the front (soft power button) in both cases it would not shut off the computer. I got a acpi error message saying that the button was ignored because the computer wasn't ready. In both cases I then resorted to the hard power switch located on the power supply in the back of the case.
> As I said this problem has only occurred twice but was significant enough I thought you would want to know. It has also only occurred with writer, I have also used calc fairly often with out issue. I have not used any other applications significantly yet.
> As near as I can remember, in both instances at the time the problem occurred I was also running thunderbird-1.0rc and Mozilla 1.7.3 at the same time. Also I don't know if it would be related but OO has been core dumping whenever I try to print a MS Word doc, however it usually works fine if I save the doc in an openoffice format, and re-open it, and then print. That last is repeatable, and if you want I get you a copy of the core.
> Feel free to e-mail me, if you have any other questions.
> Jeffrey Williams
> jeff at sailorfej.net
> freebsd-openoffice at freebsd.org mailing list
> To unsubscribe, send any mail to "freebsd-openoffice-unsubscribe at freebsd.org"
And the response from Martin Blapp:
> This sounds like a threading problem. Please upgrade your box to
> FreeBSD 5.3 Stable (there have been removed a lot of edge cases
> in the threading libraries since 5.3 release. If the problem persists,
> we will work on that problem.
More information about the freebsd-stable