Hi! I think I "suffer" from the same bug and want to add some remarks. First the system information: - Architecture: i386 (i686) - Intel Pentium III 450 MHz - 256 MHz RAM - Debian Release: testing/unstable (etch) - APT prefers testing - APT policy: (500, 'testing') - KDE (kde-core 5:45) - xserver-xorg 6.8.2.dfsg6.8.2.dfsg - openoffice.org 1.1.4-7 - Graphic card: nvidia riva TNT (Driver "nv" in /etc/X11/xorg.conf) - Shell: /bin/sh linked to /bin/bash - Kernel: Linux 2.6.12-1-686 - Locale: LANG=de_AT@euro, LC_CTYPE=de_AT@euro (charmap=ISO-8859-15) After updating quite a lot thinks via aptitude (including the kernel (2.6.8 to 2.6.12), KDE, and Openoffice (1.1.3 to 1.1.4)), the problem appeared. After launching openoffice it behaves as usual (expect that loading a document lasts may last longer as before) and after about 3 minutes working it disappears, writing the following text in the console: [begin] sh: crash_report: command not found Xlib: unexpected async reply (sequence 0x2caa4)! Xlib: sequence lost (0x3d0b9 > 0x2d0c3) in reply type 0x3! Fatal exception: Signal 11 Stack: /usr/lib/openoffice/program/libsal.so.3[0x411b73c6] /usr/lib/openoffice/program/libsal.so.3[0x411b7553] /usr/lib/openoffice/program/libsal.so.3[0x411b761e] [0xffffe420] /usr/lib/openoffice/program/libsal.so.3[0x411ac5d4] /lib/tls/libpthread.so.0[0x41180ccd] /lib/tls/libc.so.6(__clone+0x5e)[0x410eab0e] [end] Each time, the message slightly differs: [begin] Xlib: unexpected async reply (sequence 0x6751)! Fatal exception: Signal 11 Stack: /usr/lib/openoffice/program/libsal.so.3[0x411b73c6] /usr/lib/openoffice/program/libsal.so.3[0x411b7553] /usr/lib/openoffice/program/libsal.so.3[0x411b761e] [0xffffe420] /usr/lib/openoffice/program/libsal.so.3[0x411ac5d4] /lib/tls/libpthread.so.0[0x41180ccd] /lib/tls/libc.so.6(__clone+0x5e)[0x410eab0e] [end] I have made some "tests" to gain more information and found the following: - It happens (at least) in oowriter and oocalc (I have not used the other programs for my tests) - It happens independently of being started as user or root. - It doesn't seem to be related to locale-settings. Starting it with the default C locale instead of the de_AT locale results in the same async reply. - The problem does not depend on whether openoffice finds a .lock file remaining from a previos crash or not (as claimed in http://clug.ca/pipermail/clug-talk_clug.ca/2005-July/003499.html). - There are no new entries in /var/log/syslog or /var/log/messages between openoffice start and crash. - If I start openoffice writer but do not actually work with it (keeping it in the background), it doesn't crash (at least the time I waited). When I open a document and keep it in the background, it crashes, too. - Maybe it's just a "feeling", but I think when it crashes its about to do something screen-drawing-related. Typically, it crashes when I (but not always): - switch between openoffice windows. - save a document (all the icons get grayed out for a short time) - mark text (strg-a) - Some times I noticed that a - maybe - dialogbox is about to appear quite before openoffice crashes: The frame is drawn by the windowmanager but not the content. But here the maybe most interesting observation: - Booting the "old" kernel 2.6.8 with the same system and configuration, openoffice did _not_ crash, even working a whole hour very intensively with it. The differences between the kernels are: - kernel, openoffice works with: 2.6.8 (compiled by me, omitting the modules, where I have no hardware) - kernel, openoffice crashes with: 2.6.12-1-686 (standard debian kernel) The /var/log/Xorg.0.log files of the x server starts with the different kernels are identical (exect the creation time line and the kernel line). I would have compiled the 2.6.12 kernel by myself with the same .config options as 2.6.8 but 2.6.12 does not compile with gcc 4.0.2. Thank you and greetings, Philipp Spitzer
Attachment:
signature.asc
Description: Digital signature