Created attachment 53840 [details] typescript of gdb session My system is ubuntu-natty (32-bit), and I am running commit id 7a37392 built with autogen options --disable-mozilla --enable-symbols --enable-dbgutil --enable-crashdump --disable-build-mozilla --without-junit I was trying to open a .doc file. After the program closed the file open dialog but before opening a new Writer window, the program received SIGABRT. Attaching typescript of the gdb session; backtrace starts around line 461. I will send the .doc file in question to anybody who wants to look at the problem.
I see the same assertion raised, this time using only files from LibreOffice. I am putting this in the same bug report because of the similarity of backtraces, especially Thread 9: frames #2 through #7 from 2011-11-24 are the like frames #21 through #26 of today's backtrace. Now, LO is built from commit id 7a37392, roughly 2011-11-23. To reproduce the assertion ... (1) Download the attached file hyphen_rendering.odt. (2) Start LibreOffice. Program displays Writer document "Untitled 1". (3) Open hyphen_rendering.odt. (4) Navigate to the bottom of the document and drag the mouse through the last paragraph, from the word "This" to the period. Program highlights the selected text. (5) Type ctrl-C. (6) Take menu options File > New > Spreadsheet. Program displays Calc document "Untitled 2", "Sheet 1", with the cursor in cell A1. (7) Type ctrl-V. Program raises the assertion. Backtrace starts after line 212 of the (soon-to-be) attached typescript.
Created attachment 54071 [details] Writer document from which to copy
Created attachment 54072 [details] typescript of gdb session SIGABRT near line 212. Backtrace follows.
Created attachment 54493 [details] typescript of gdb session New behaviour, observed with commit id 4097499 (pulled 2011-12-08). The steps outlined in comment 1 no longer provoke the assertion, but following them by copying text to the clipboard in another application does provoke the assertion. But here is an easier way. (1) Start LibreOffice from gnome-terminal. The program displays empty Writer document "Untitled 1". (2) In "Untitled 1" type "asdf". (3) Select and copy that text by typing "alt-E A alt-E C". The program hightlights the copied text. (4) Open a spreadsheet by typing "alt-F N S". The program displays Calc document "Untitled 2" with the cursor in cell A1. (5) Paste the copied text by typing "alt-E P". The program displays "asdf" in cell A1. (6) In another application, copy text to the clipboard. For example, in the controlling terminal session, drag the mouse cursor through a few characters, right-click, and select Copy from the drop-down menu. Program raises the assertion in SwIndexReg::~SwIndexReg(). In the attached typescript of a gdb session, the SIGABRT is around line 249; backtrace follows shortly thereafter. I am running ubuntu-natty and the configuration options are --disable-mozilla --enable-symbols --enable-dbgutil --enable-crashdump --disable-build-mozilla --enable-ext-postgresql-sdbc
Drag and drop from Writer to Calc provokes the assertion immediately, i.e. without involving another application. gdb shows one more thread and the backtrace of all threads is 20 lines longer.
Not reproduceple in linux 32bits, libreoffice 3.5.4 or current master 3.7 with debug enabled, so ill mark it as fixed.
I still have the problem. What is different about my system, I wonder? Now running master, commit 741c56a, pulled 2012-09-13 and configured with --enable-symbols --enable-dbgutil --enable-crashdump --disable-build-mozilla --without-system-postgresql --enable-debug --enable-werror on ubunty-natty (11.04) with ubuntu classic (no effects) desktop.
The steps outlined in comment 4 still cause a crash. and so do the drag-and-drop mentioned in comment 5. LibreOffice is commit id 940db02, pulled on 2012-10-10, and configured with options ... --enable-symbols --enable-dbgutil --enable-crashdump --disable-build-mozilla --without-system-postgresql --enable-debug --enable-werror The build and execution environment is ubuntu-natty (11.04) ... $ uname -a Linux cougar-natty 2.6.38-16-generic #67-Ubuntu SMP Thu Sep 6 18:00:43 UTC 2012 i686 athlon i386 GNU/Linux $ gcc --version gcc (Ubuntu/Linaro 4.5.2-8ubuntu4) 4.5.2 Copyright (C) 2010 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. After filing this bug report, I made some effort to look into the mechanism of the failure, but eventually gave up without any progress to mention. If someone will provide guidance, I will resume the investigation.
Created attachment 72599 [details] bt + console logs with 3.6 sources On pc Debian x86-64 with 3.6 sources updated 2 days ago (commit 37eea402ed3b8b401c8eca841ca65c997994be90), I reproduced the crash. Terrence: FYI, with 4.0 branch or master sources, I don't reproduce this.
Cedric/Michael: one for you?
With recent master, 2e367c pulled 2013-02-16, built and running on ubuntu-natty (11.04), I do not see the problem. With this observation and Julien's remark in comment 9, I am closing the bug WORKSFORME. However version 3.6 was afflicted as of 2013-01-04, and presumably still is. I think that there is no plan to have release 3.7, so there is no need to go looking for the change which fixed the problem. Right?
Terrence: according to https://wiki.documentfoundation.org/ReleasePlan#3.6_release, there'll be a 3.6.6 Anyway, I let the status as it is since I'm not sure what should be done.