Openoffice deadlock if starting Impress whilst writer/calc open dialog visible

[shilliard]

Binary package hint: openoffice.org

Using openoffice version 2.0.2 in Ubuntu 6.06 I can hang openoffice using the following
sequence:

1. Start Openoffice Writer (or Calc)
2. Select File->Open, dialog box appears.
3. Start Openoffice Impress, the Presentation wizard will appear.
4. Openoffice is now hung. Cancel does not work for either writer or impress,
only way to recover is to kill the openoffice processes.

This is 100% repeatable on several Ubuntu installations

Openoffice guys say it is a problem with the Ubuntu version as it is not repeatable with the official version.

[Joey Stanford (joey)]

Confirmed. I have reproduced this 2.0.4

 5201 ? S 0:00 /bin/sh /usr/lib/openoffice/program/soffice -writer -splash-pipe=5
 5215 ? Sl 0:03 /usr/lib/openoffice/program/soffice.bin -writer -splash-pipe=5

[Joey Stanford (joey)]

vs

 /bin/sh /usr/lib/openoffice/program/soffice -impress -splash-pipe=5

I'm trying to get a some traces out of these with currently not much luck.

[Joey Stanford (joey)]

I'm not an expert but looks like the package was linked incorrectly (perhaps due to the build environment) when built for the Ubuntu repository. Suggest Matthias updates and rebuilds.

[Simon Oosthoek (simon-margo)]

I'm having similar problems on edgy's openoffice (calc in my case)

when started from the commandline, I see this:
(process:5085): GLib-GObject-CRITICAL **: gtype.c:2240: initialization assertion failed, use IA__g_type_init() prior to this function

(process:5085): Gdk-CRITICAL **: gdk_screen_get_font_options: assertion `GDK_IS_SCREEN (screen)' failed

(process:5085): GLib-GObject-CRITICAL **: gtype.c:2240: initialization assertion failed, use IA__g_type_init() prior to this function

(process:5085): Gdk-CRITICAL **: gdk_screen_get_font_options: assertion `GDK_IS_SCREEN (screen)' failed

(process:5085): GLib-GObject-CRITICAL **: gtype.c:2240: initialization assertion failed, use IA__g_type_init() prior to this function

(process:5085): Gdk-CRITICAL **: gdk_screen_get_font_options: assertion `GDK_IS_SCREEN (screen)' failed

(process:5085): GLib-GObject-CRITICAL **: gtype.c:2240: initialization assertion failed, use IA__g_type_init() prior to this function

(process:5085): Gdk-CRITICAL **: gdk_screen_get_font_options: assertion `GDK_IS_SCREEN (screen)' failed

However, it's not actually totally locked then. It is deadlocked when started from the kmenu. When trying it now to reproduce, I don't even get a window, and strace shows some sort of loop:

[pid 5217] poll([{fd=27, events=POLLIN}], 1, 1000) = 0
[pid 5217] gettimeofday({1170844252, 98260}, NULL) = 0
[pid 5217] poll([{fd=27, events=POLLIN}], 1, 1000) = 0
[pid 5217] gettimeofday({1170844253, 98252}, NULL) = 0
[pid 5217] poll([{fd=27, events=POLLIN}], 1, 1000) = 0
[pid 5217] gettimeofday({1170844254, 98243}, NULL) = 0
[pid 5217] poll([{fd=27, events=POLLIN}], 1, 1000) = 0
[pid 5217] gettimeofday({1170844255, 98229}, NULL) = 0
[pid 5217] poll([{fd=27, events=POLLIN}], 1, 1000) = 0
[pid 5217] gettimeofday({1170844256, 98219}, NULL) = 0
[pid 5217] poll([{fd=27, events=POLLIN}], 1, 1000) = 0
[pid 5217] gettimeofday({1170844257, 98219}, NULL) = 0
[pid 5217] poll( <unfinished ...>
[pid 5218] <... futex resumed> ) = -1 ETIMEDOUT (Connection timed out)
[pid 5218] gettimeofday({1170844257, 662226}, NULL) = 0
[pid 5218] futex(0xb742c580, FUTEX_WAKE, 1) = 0
[pid 5218] clock_gettime(CLOCK_REALTIME, {1170844257, 662325000}) = 0
[pid 5218] futex(0xb742c6e4, FUTEX_WAIT, 21, {9, 999901000} <unfinished ...>
[pid 5217] <... poll resumed> [{fd=27, events=POLLIN}], 1, 1000) = 0
[pid 5217] gettimeofday({1170844258, 98222}, NULL) = 0
[pid 5217] poll([{fd=27, events=POLLIN}], 1, 1000) = 0
[pid 5217] gettimeofday({1170844259, 98204}, NULL) = 0
[pid 5217] poll([{fd=27, events=POLLIN}], 1, 1000) = 0
[pid 5217] gettimeofday({1170844260, 98192}, NULL) = 0
[pid 5217] poll([{fd=27, events=POLLIN}], 1, 1000) = 0
[pid 5217] gettimeofday({1170844261, 102181}, NULL) = 0
[pid 5217] poll([{fd=27, events=POLLIN}], 1, 1000) = 0
[pid 5217] gettimeofday({1170844262, 106214}, NULL) = 0
[pid 5217] poll([{fd=27, events=POLLIN}], 1, 1000) = 0
[pid 5217] gettimeofday({1170844263, 110190}, NULL) = 0
[pid 5217] poll([{fd=27, events=POLLIN}], 1, 1000) = 0
[pid 5217] gettimeofday({1170844264, 110157}, NULL) = 0
[pid 5217] poll([{fd=27, events=POLLIN}], 1, 1000) = 0
[pid 5217] gettimeofday({1170844265...

[Simon Oosthoek (simon-margo)]

When started from the commandline it seemed to work ok, but then it hung upon saving the file (at choosing save as from the file menu). Not just openoffice hung, but the entire desktop didn't respond to mouse or keyboard events. Windows from other programs (xosview, konsole) were updating during this.

I'm going to try the packages from openoffice.org now, because it seems the unbuntu version has problems.

[Simon Oosthoek (simon-margo)]

I installed the openoffice.org 2.1 packages from the rpms after conversion by alien. I get the same type of problems with this version. I'm starting to suspect that NFS plays a role in this, since my homedir is mounted from a debian server over NFS.

current mount options are:
server:/local/home /home nfs rw,nosuid,tcp,async 0 0
I'll try changing these to see if that could make a difference.

[Simon Oosthoek (simon-margo)]

ok, found something that fixes my issues

http://www.ubuntuforums.org/showthread.php?t=332043

(says: install nfs-common on the client side)

[Matthias Klose (doko)]

2.2-0ubuntu2 now disables locking if the file is found on a NFS share. -0ubuntu3 will add an additional recommendation on nfs-common.

[Matthias Klose (doko)]

fixed in 2.2.0-1ubuntu1