Description of problem: Combinations with various Bengali consonant characters (0995-09B9)+09CB+special characters (0981-0983) cause Openoffice.org writer to crash unexpectedly, after ibus, ibus-libs, ibus-gtk2, ibus-gtk3 packages are updated. Version-Release number of selected component (if applicable): openoffice.org-writer-3.3.0-9.3.fc14 ibus-libs-1.3.9-3.fc14 ibus-gtk2-1.3.9-3.fc14 ibus-1.3.9-3.fc14 ibus-gtk3-1.3.9-3.fc14 How reproducible: Steps to Reproduce: 1. Add Bengali=>Probhat and Inscript keyboards from ibus-preferences to enable them 2. Start openoffice.org writer 3. Enable Probhat keyboard by choosing it from the ibus panel applet 4. Write the following combination: * 099C + 09CB + 0981 actual key strokes: j]> to display: জোঁ * 0995 + 09CB + 0982 actual key strokes: k]L to display: কোং * 09B9 + 09CB + 0983 actual key strokes: h]H to display: হোঃ Actual results: Openoffice.org writer crashes Expected results: Combinations to be displayed and application should not crash. Additional info: 1. The bug can be replicated with direct unicode input as well 2. The consonant value can be replaced with any in the range, and same results are seen. The special character may or may not be different in each case. 3. With no change in the Openoffice.org (and its dependency) packages, these combinations worked as expected on the ibus GA versions (1.3.9-3.fc14)
Also tested with openoffice.org-writer-3.3.0-20.2.fc14. Crash replicated.
Created attachment 492372 [details] An ODT file containing the bug combinations. This ODT carries the three combinations reported in the bug.
Created attachment 492373 [details] PDF created from the attached ODT file. This PDF contains all the three reported combinations.
Packages installed in my machine are: openoffice.org-writer-1:3.3.0-20.5.fc14.i686 ibus-1.3.9-3.fc14.i686 ibus-gtk2-1.3.9-3.fc14.i686 ibus-libs-1.3.9-3.fc14.i686 And no package ibus-gtk3 installed. As you can see, I could not reproduce the bug in the first try. But, this was in Xfce, my default desktop. Then I went to Gnome, that is Bhattacharya's (the creator of the bug) default. And now, openoffice writer crashed when I tried to write. Even when I opened the already created ODT, it crashed, and then openoffice writer started taking infinite time to open. And this proved to be contagious, the next time when I returned to Xfce, the opened ODT file crashed once.
I feel pleasure to make Runa convinced of the said bug after more than 20 mails in last 3 weeks. The same problem is recurring from f13. there are also 2 more problems in connection with bengali typing - (1) .ttf fonts are found in font list of open office writer at installation but at typing these fonts are found as lohit bengali like. (2) text typed in solaimanlipi are jumbled at print,but found ok in monitor, I doubt this may be the problem of hplip. Some few more problems in fedora 14 - (1) qt & python can't be installed completely. –skip-broken leaves some resedue. (2) picasa can not run without wine and without cp /usr/bin/wine-preloader /opt/google/picasa/3.0/wine/bin/wine-preloader to fix Picasa Segmentation Fault. (3) The weak most area of free software is at Sound & Video. Most of the entertainment dvds can not be played with Movie Player. Even the VLC or sm player fails to play all dvds. I find little bit solution of this problem at installation of gstreamer* ffmpeg* lame* libXp mjpegtools libdvdcss libdvdcss2 libdvd* lsdvd streamdvd. But it crashes at rapid forward-backward playing of the videos.
(In reply to comment #5) The issues (1), (2) & (3) in Comment #5 are not relevant to the problem concerning this bug and also mentions third party/non-Fedora components. Kindly please file separate bugs for Fedora related problems and contact the upstream bug tracking system relevant for the third-party components. Thank you.
I guess your problem is not reproduced with ibus-1.3.99.20110408 and libreoffice in f15 ? http://koji.fedoraproject.org/koji/buildinfo?buildID=239582
I'd have to check and get back to you on this.
Created attachment 493336 [details] Patch for ibus-m17n engine.c I could not reproduce your problem strictly. But I can see ibus-daemon freezes when ibus-engine-m17n exits accidentally in f14. Could you try the following rpm if your problem can be fixed? http://fujiwara.fedorapeople.org/ibus/20110420/3012071/ To reproduce: 1. Invoke soffice's writer 2. Choose Bengali=>Probhat with left click on ibus panel icon 3. Exit soffice 4. Invoke soffice's writer again 5. Choose Bengali=>Probhat again Then ibus-m17n-engine exits: #0 0x00007f00080938e0 in exit () from /lib64/libc.so.6 #1 0x00007f00091e061c in g_type_module_use_plugin (plugin=0x100d080) from /lib64/libgobject-2.0.so.0 #2 0x00007f00091e105b in g_type_plugin_use () from /lib64/libgobject-2.0.so.0 #3 0x00007f00091d91b2 in type_data_ref_Wm () from /lib64/libgobject-2.0.so.0 #4 0x00007f00091dd359 in g_type_class_ref () from /lib64/libgobject-2.0.so.0 #5 0x00007f00091be816 in g_object_new_valist () from /lib64/libgobject-2.0.so.0 #6 0x00007f00091bdf03 in g_object_new () from /lib64/libgobject-2.0.so.0 #7 0x00007f00094158e4 in ibus_factory_ibus_message () from /usr/lib64/libibus.so.2 The g_type_module_use() checks type_info->loaded is FALSE. I think when soffice exits, bus_engine_proxy_real_destroy() in ibus-daemon calls g_type_module_unuse() in ibus-engine-m17n.
(In reply to comment #9) > The g_type_module_use() checks type_info->loaded is FALSE. > > I think when soffice exits, bus_engine_proxy_real_destroy() in ibus-daemon > calls g_type_module_unuse() in ibus-engine-m17n. Thank you for the investigation and patch. I could reproduce the crash with your steps even with gedit. For the patch looks fine, I'll file a review request at the upsream.
ibus-m17n-1.3.2-2.fc14 has been submitted as an update for Fedora 14. https://admin.fedoraproject.org/updates/ibus-m17n-1.3.2-2.fc14
Package ibus-m17n-1.3.2-2.fc14: * should fix your issue, * was pushed to the Fedora 14 testing repository, * should be available at your local mirror within two days. Update it with: # su -c 'yum update --enablerepo=updates-testing ibus-m17n-1.3.2-2.fc14' as soon as you are able to. Please go to the following url: https://admin.fedoraproject.org/updates/ibus-m17n-1.3.2-2.fc14 then log in and leave karma (feedback).
*** Bug 697771 has been marked as a duplicate of this bug. ***
Created attachment 494813 [details] abrt logger This bug has been reproduced on F15 beta (32 bit installation)
Above testing has done with following components ibus-m17n-1.3.2-1.fc15.i686 libreoffice-writer-3.3.2.2-6.fc15.i686 ibus-1.3.99.20110408-2.fc15.i686
> This bug has been reproduced on F15 beta (32 bit installation) From the backtrace, it seems to be a different issue from the one addressed by ibus-m17n-1.3.2-2.fc14 (see comment 9 to 12 above), though I still can't reproduce the original problem. Reassigning the component to icu to get more feedack.
I could reproduce this problem. I agreed, this could be happened in 32bit only but not x86_64. Program received signal SIGABRT, Aborted. 0x00c6c424 in __kernel_vsyscall () #0 0x00c6c424 in __kernel_vsyscall () #1 0x017fd67f in raise () from /lib/libc.so.6 #2 0x017fefb5 in abort () from /lib/libc.so.6 #3 0x01839fd7 in __libc_message () from /lib/libc.so.6 #4 0x01840f95 in malloc_printerr () from /lib/libc.so.6 #5 0x0323eb9c in uprv_free_44 (buffer=0x97bf5b8) at cmemory.c:79 #6 0x034b0aa4 in icu_44::LEGlyphStorage::reset (this=0x9851020) at LEGlyphStorage.cpp:58 #7 0x034a4286 in icu_44::LayoutEngine::reset (this=0x9850fc8) at LayoutEngine.cpp:508 #8 0x0349fc1d in icu_44::OpenTypeLayoutEngine::reset (this=0x9850fc8) at OpenTypeLayoutEngine.cpp:118 It seems the fGlyphs is not the right buffer. The from.fGlyphs was assigned by OpenTypeLayoutEngine::computeGlyphs(). #0 icu_44::LEGlyphStorage::adoptGlyphArray (this=0x9851020, from=...) at LEGlyphStorage.cpp:463 #1 0x034a04f1 in icu_44::OpenTypeLayoutEngine::glyphPostProcessing (this= 0x9850fc8, tempGlyphStorage=..., glyphStorage=..., success=@0xbfada47c) at OpenTypeLayoutEngine.cpp:291 #2 0x034a0718 in icu_44::OpenTypeLayoutEngine::computeGlyphs (this=0x9850fc8, chars=0xb088f710, offset=2, count=1, max=3, rightToLeft=0 '\000', glyphStorage=..., success=@0xbfada47c) at OpenTypeLayoutEngine.cpp:333 #3 0x034a41e7 in icu_44::LayoutEngine::layoutChars (this=0x9850fc8, chars= 0xb088f710, offset=2, count=1, max=3, rightToLeft=0 '\000', x=0, y=0, success=@0xbfada47c) at LayoutEngine.cpp:499 #4 0x045b803d in IcuLayoutEngine::operator()(ServerFontLayout&, ImplLayoutArgs&) () from /usr/lib/libreoffice/program/../basis-link/program/libvclli.so #5 0x045b873c in ServerFontLayout::LayoutText(ImplLayoutArgs&) () from /usr/lib/libreoffice/program/../basis-link/program/libvclli.so #6 0x0440c46c in OutputDevice::ImplLayout(String const&, unsigned short, unsigned short, Point const&, long, long const*, bool) const () from /usr/lib/libreoffice/program/../basis-link/program/libvclli.so #7 0x0440da8d in OutputDevice::GetTextArray(String const&, long*, unsigned short, unsigned short) const () Maybe the fakeGlyphStorage is not correct in OpenTypeLayoutEngine::computeGlyphs().
ibus-m17n-1.3.2-2.fc14 has been pushed to the Fedora 14 stable repository. If problems still persist, please make note of it in this bug report.
This package has changed ownership in the Fedora Package Database. Reassigning to the new owner of this component.
Does this issue still persist on f16 say?
This problem still persist on f16. But it was not active in Fedora 15.
Is this still an issue in f16? Next: could someone please try if it persists in f17 and f18/rawhide? We have ICU 4.6 in f16, 4.8 in f17 and switched to ICU 49 in rawhide.
I am not able to reproduce this bug in F17. Can reporter or anyone please confirm by testing on Fedora 17?