| Summary: | Crash on start with KDE integration installed | ||
|---|---|---|---|
| Product: | LibreOffice | Reporter: | krzysiek |
| Component: | LibreOffice | Assignee: | Not Assigned <libreoffice-bugs> |
| Status: | VERIFIED FIXED | ||
| Severity: | normal | CC: | glogow, ilmari.lauhakangas, m.weghorn |
| Priority: | medium | ||
| Version: | 6.2.0.3 release | ||
| Hardware: | x86-64 (AMD64) | ||
| OS: | Linux (All) | ||
| See Also: |
https://bugs.documentfoundation.org/show_bug.cgi?id=124124 https://bugs.documentfoundation.org/show_bug.cgi?id=128738 |
||
| Whiteboard: | target:6.4.0 target:6.3.4 | ||
| Crash report or crash signature: | Regression By: | ||
| Bug Depends on: | |||
| Bug Blocks: | 102495 | ||
| Attachments: |
backtrace with master as of 8e3f4317823aa83e0ad9e7dc008a1550ef3c9798
User profile to reprodue GDB trace of crash |
||
|
Description
krzysiek
2019-03-04 21:54:15 UTC
Thanks for your bug report. Could you please retest with a current daily build for the LibreOffice 6.2 branch? They are provided at [1] and [2] describes how you can test those without actually installing the deb packages. If it works with those, the problem has probably either already fixed in the meantime or it's a problem in the packaging done by KDE Neon. [1] https://dev-builds.libreoffice.org/daily/libreoffice-6-2/ [2] https://wiki.documentfoundation.org/Installing_in_parallel/Linux Created attachment 152842 [details] backtrace with master as of 8e3f4317823aa83e0ad9e7dc008a1550ef3c9798 I had such a crash today with current master (as of commit 8e3f4317823aa83e0ad9e7dc008a1550ef3c9798) on Debian testing (kde-style-breeze 4:5.14.5-1, libqt5core5a:amd64 5.11.3+dfsg1-1). New backtrace with all relevant debug symbols installed attached. Created attachment 152843 [details]
User profile to reprodue
I can even reliably reproduce now with the attached user profile (from instdir/user in my dev setup). This also works when copying the 'instdir/user' dir contained therein into ~/.config/libreoffice/ for a release build on another host where I have a different user name.
*** Bug 128738 has been marked as a duplicate of this bug. *** (In reply to Michael Weghorn from comment #4) > I can even reliably reproduce now with the attached user profile (from > instdir/user in my dev setup). To clarify, steps to reproduce with a current master build (all commands to run from the source dir): 1) delete current user profile $ rm -r instdir/user 2) extract the attached profile $ tar xvf .../configdir_tdf123851.tar.gz 3) start LO -> needs to be run twice after fresh LO build SAL_USE_VCLPLUGIN=kf5 ./instdir/program/soffice.bin --writer -> the document recovery dialog shows up 4) click "Cancel" -> Crash Note that using './instdir/program/soffice' instead of './instdir/program/soffice.bin' in step 3 does not work, because the document recovery dialog is not shown in this case. Still reproducible for me with master as of eeae2032476e8c8645ab438fbb379f3151c8e472 and Qt packages as of version 5.12.5+dfsg-2 on Debian testing) I can repro the crash with the profile. I have Qt debug symbols installed. Please advise how I can get a trace in these circumstances, where it crashes on startup. I tried make debugrun, but bt said "No stack". Created attachment 155771 [details] GDB trace of crash I managed to do it by launching soffice.bin with a sleep command of some seconds and then in another terminal doing gdb --pid `pgrep soffice` The trace is different, though. It's about fontconfig. Arch Linux 64-bit Version: 6.4.0.0.alpha1+ Build ID: bcc701e7c544331f0b5992df532919232b5ae5a0 CPU threads: 8; OS: Linux 5.3; UI render: default; VCL: kf5; Locale: fi-FI (fi_FI.UTF-8); UI-Language: en-US Calc: threaded Built on 12 November 2019 (In reply to Buovjaga from comment #8) > I managed to do it by launching soffice.bin with a sleep command of some > seconds and then in another terminal doing gdb --pid `pgrep soffice` I just did 'gdb --args ./instdir/program/soffice.bin --writer' > The trace is different, though. It's about fontconfig. That's actually different. Does it mention anything about the crash in addition to the plain backtrace (e.g. segfault, ...)? Oh, and just to make sure: Did you press 'c' in gdb to continue running LO after attaching to the process? (In reply to Michael Weghorn from comment #9) > (In reply to Buovjaga from comment #8) > > I managed to do it by launching soffice.bin with a sleep command of some > > seconds and then in another terminal doing gdb --pid `pgrep soffice` > > I just did 'gdb --args ./instdir/program/soffice.bin --writer' Thanks, with this, I was able to get the same trace as before. I don't know why I did not notice it in your trace :( (I had to use 'run' twice). Sadly, like you I still get rControlRegion=... instead of the value - Jan-Marek was hoping to see it. Which debug symbols are we missing?? (In reply to Buovjaga from comment #10) > Sadly, like you I still get rControlRegion=... instead of the value - > Jan-Marek was hoping to see it. Which debug symbols are we missing?? Hm, that is 'const tools::Rectangle& rControlRegion', i.e. an LO-internal type, so this doesn't look like missing debug symbols. Maybe going to the frame in gdb and explicitly printing with 'p rControlRegion' might have worked. Anyway, Jan-Marek mentioned he can reproduce himself now. Just for the docs: I used instdir/program/soffice instead of instdir/program/soffice.bin from my master build, which has a local instdir/program/ooenv, which itself contains an "export OOO_DISABLE_RECOVERY=1", which prevented the recovery to run. Actually I didn't even register the difference, until the steps were - again - explicitly written in comment 6. And then also "SAL_USE_VCLPLUGIN=kf5 ./instdir/program/soffice.bin -env:UserInstallation=file:///tmp/instdir --writer" crashed for me. There is now: https://gerrit.libreoffice.org/#/c/82600/ waiting for the Jenkins run to finish. Jan-Marek Glogowski committed a patch related to this issue. It has been pushed to "master": https://git.libreoffice.org/core/commit/9bc848cf0d301aa57eabcffa101a1cf87bad6470 tdf#123851 Qt5 handle broken ScrollBar values It will be available in 6.4.0. The patch should be included in the daily builds available at https://dev-builds.libreoffice.org/daily/ in the next 24-48 hours. More information about daily builds can be found at: https://wiki.documentfoundation.org/Testing_Daily_Builds Affected users are encouraged to test the fix and report feedback. Jan-Marek Glogowski committed a patch related to this issue. It has been pushed to "libreoffice-6-3": https://git.libreoffice.org/core/commit/c5946dfb6fe36b558967659a85d3dce8317fd7c5 tdf#123851 Qt5 handle broken ScrollBar values It will be available in 6.3.4. The patch should be included in the daily builds available at https://dev-builds.libreoffice.org/daily/ in the next 24-48 hours. More information about daily builds can be found at: https://wiki.documentfoundation.org/Testing_Daily_Builds Affected users are encouraged to test the fix and report feedback. Thanks for everyone involved |