| Summary: | Crash in Garbage collection after Plasma::DataEngine::connectSource | ||
|---|---|---|---|
| Product: | [Plasma] plasmashell | Reporter: | Cristian Castro <lahostiafulluser> |
| Component: | general | Assignee: | David Edmundson <kde> |
| Status: | RESOLVED FIXED | ||
| Severity: | crash | CC: | bhush94, cpatrick08, eng.mahs, gustavo.antunez2000, heikki, netphilos, plasma-bugs, renda.krell, squan |
| Priority: | NOR | Keywords: | drkonqi |
| Version: | 5.4.0 | ||
| Target Milestone: | 1.0 | ||
| Platform: | openSUSE | ||
| OS: | Linux | ||
| Latest Commit: | Version Fixed In: | ||
| Sentry Crash Report: | |||
| Attachments: |
my backtrace
latest backtrace with debug-symbols installed |
||
|
Description
Cristian Castro
2015-09-21 23:30:11 UTC
Created attachment 94690 [details]
my backtrace
Plasma crashes for me too at startup. Using openSUSE Tumbleweed, Plasma 5.3.95 Qt 5.4.2 Kernel 4.1.6-3 x64.
*** Bug 353145 has been marked as a duplicate of this bug. *** *** Bug 353250 has been marked as a duplicate of this bug. *** *** Bug 353239 has been marked as a duplicate of this bug. *** last time we had a crash in GC it meant that there was an exec() being called in the main thread when called from QML I want to find a trace that includes symbols for plasma-workspace so we can see what's happening here. #25 0x00007f80511bf455 in ?? () #26 0x0000000004d439a0 in ?? () #27 0x0000000001fbb9e0 in ?? () #28 0x0003000000000001 in ?? () #29 0x00007f8118baf985 in Plasma::DataEnginePrivate::requestSource Created attachment 94772 [details]
latest backtrace with debug-symbols installed
Deleting plasma-configuration files in ~/.config and ~/.local/share first helped and Plasma didn't crash at startup but now it does it again. In attachemnt is backtrace after I installed more debug-symbols. I'm using openSUSE TW x64, Plasma 5.4.1 and Qt 5.4.2, Kernel 4.1.6-3.2 on a Dell Latitude e5420 laptop with external monitor.
I get that crash after login when I'm using laptop with external monitor (with HDMI) on and internal monitor off. If external monitor is unplugged I don't seem t crash. System info: Machine: System: Dell product: Latitude E5420 version: 01 Mobo: Dell model: 0H5TG2 version: A01 Bios: Dell version: A14 date: 12/26/2013 CPU: Dual core Intel Core i5-2520M CPU (-MCP-) cache: 3072 KB flags: (lm nx sse sse2 sse3 sse4_1 sse4_2 ssse3 vmx) Graphics: Card: Intel 2nd Generation Core Processor Family Integrated Graphics Controller X.org: 1.17.2 driver: intel Resolution: 114x37 @Henri Thanks, but I'm afraid that's a different crash (https://bugs.kde.org/show_bug.cgi?id=344550) fixed with Qt5.5 *** Bug 353392 has been marked as a duplicate of this bug. *** (In reply to David Edmundson from comment #5) > last time we had a crash in GC it meant that there was an exec() being > called in the main thread when called from QML > > I want to find a trace that includes symbols for plasma-workspace so we can > see what's happening here. > > #25 0x00007f80511bf455 in ?? () #26 0x0000000004d439a0 in ?? () #27 > 0x0000000001fbb9e0 in ?? () #28 0x0003000000000001 in ?? () #29 > 0x00007f8118baf985 in Plasma::DataEnginePrivate::requestSource See my backtrace from the duplicate bug 353392: #69 0x00007f6e7d20eb55 in Plasma::DataEnginePrivate::requestSource (this=0x7f6e503c40c0, sourceName=..., newSource=0x7ffcfde986d0) at /usr/src/debug/plasma-framework-5.14.0/src/plasma/dataengine.cpp:603 argh, I need to see the sourceName, which is the bit supressed here. which either means symbols from plasma-workspace where it will have a call from that dataengine or mad gdb skills from someone who can reproduce this. if you get this again if you could: open in gdb (there's a button in drkonqui) http://silmor.de/qtstuff.printqstring.php then type frame 69 printqs5dynamic sourceName (In reply to David Edmundson from comment #11) > argh, I need to see the sourceName, which is the bit supressed here. > > which either means symbols from plasma-workspace where it will have a call > from that dataengine > > or > > mad gdb skills from someone who can reproduce this. > > if you get this again if you could: > open in gdb (there's a button in drkonqui) > http://silmor.de/qtstuff.printqstring.php > > then type > frame 69 > printqs5dynamic sourceName I can reproduce this, but need some advice regarding the gdb: (gdb) frame 69 #69 0x00007fc11855ab55 in Plasma::DataEnginePrivate::requestSource (this=0x7fc0e6fff0c0, sourceName=..., newSource=0x7ffcabbaa860) at /usr/src/debug/plasma-framework-5.14.0/src/plasma/dataengine.cpp:603 603 /usr/src/debug/plasma-framework-5.14.0/src/plasma/dataengine.cpp: No such file or directory. (gdb) printqs5dynamic sourceName Undefined command: "printqs5dynamic". Try "help". I missed a line, copy and paste everything from teh final box here: http://silmor.de/qtstuff.printqstring.php (In reply to David Edmundson from comment #13) > I missed a line, > > copy and paste everything from teh final box here: > http://silmor.de/qtstuff.printqstring.php After setting up $HOME/.gdbinit with the last box from the link: (gdb) frame 69 #69 0x00007fc11855ab55 in Plasma::DataEnginePrivate::requestSource (this=0x7fc0e6fff0c0, sourceName=..., newSource=0x7ffcabbaa860) at /usr/src/debug/plasma-framework-5.14.0/src/plasma/dataengine.cpp:603 603 /usr/src/debug/plasma-framework-5.14.0/src/plasma/dataengine.cpp: No such file or directory. (gdb) printqs5dynamic sourceName (Qt5 QString)0x2261f60 length=19: "Cannot evaluate function -- may be inlined Thanks ever so much only dataengine with length == 19 devicenotifications I'll take a look there. In the meantime if you can install plasma-workspace debug symbols that'd really help too ..unless that's null terminated. In which case org.kde.activities also fits Can someone please install plasma-workspace-dbg (In reply to David Edmundson from comment #16) > Can someone please install plasma-workspace-dbg I installed all according packages from the openSUSE distribution: plasma5-workspace-debuginfo, plasma5-workspace-debugsource,plasma5-workspace-libs-debuginfo, plasma5-workspace-branding-openSUSE-debuginfo and all libKF5* and libQt5* debuginfo packages I could find, there are still unresolved lines: #64 QV4::Runtime::callProperty (context=0x7ffcabbaa8c0, name=0x7fc0ec64a220, callData=0x7fc0e6fff080) at /usr/src/debug/qtdeclarative-opensource-src-5.4.2/src/qml/jsruntime/qv4runtime.cpp:930 #65 0x00007fc040d742d5 in ?? () #66 0x0000000002264210 in ?? () #67 0x00000000023efb80 in ?? () #68 0x0003000000000001 in ?? () #69 0x00007fc11855ab55 in Plasma::DataEnginePrivate::requestSource (this=0x7fc0e6fff0c0, sourceName=..., newSource=0x7ffcabbaa860) at /usr/src/debug/plasma-framework-5.14.0/src/plasma/dataengine.cpp:603 #70 0x00007fc11855abb2 in Plasma::DataEngine::connectSource (this=0x7fc050160880, source=..., visualization=0x7fc1171b5736 <QV4::SimpleScriptFunction::call(QV4::Managed*, QV4::CallData*)+438>, pollingInterval=3966574592, intervalAlignment=(unknown: 20494912)) at /usr/src/debug/plasma-framework-5.14.0/src/plasma/dataengine.cpp:124 #71 0x00007fc0ec6d2000 in ?? () #72 0x00007fc050013910 in ?? () #73 0x0000000000000000 in ?? () Any idea what might be missing in particular? A packaging problem? Oh, maybe I'm wrong that line prior to the missing bits is: emit q->sourceAdded(sourceName); which does go roughly direct to QML New update (sorry, you all get spammed with my train of thoughts.. but if I don't write things down I lose them) It can't be the Plasma core declarative import as that connects via a queued connection, so we'd see the eventloop beneath it in the stack From a grep that leaves: applets/systemtray/plugin/protocols/dbussystemtray/dbussystemtrayprotocol.cpp: dataengines/weather/weatherengine.cpp: runners/solid/solidrunner.cpp: Top one makes the most sense, as it's then creating an object which matches the trace. At a guess, removing the system tray is going to stop this crash. (In reply to David Edmundson from comment #19) > At a guess, removing the system tray is going to stop this crash. Yes sir :-) It is absolutely sure for me - Plasma is crashing with the system tray. After removing it and restart it comes up smoothly. So what now? *** Bug 353420 has been marked as a duplicate of this bug. *** *** Bug 353328 has been marked as a duplicate of this bug. *** (In reply to René Krell from comment #20) > It is absolutely sure for me - Plasma is crashing with the system tray. > After removing it and restart it comes up smoothly. So what now? Interesting here is also that after re-adding the system tray no crashes appear any longer for me, just like a configuration migration problem or whatever. *** Bug 353478 has been marked as a duplicate of this bug. *** Given I narrowed this down to the applets/systemtray/plugin/protocols/dbussystemtray/dbussystemtrayprotocol.cpp: and this has been rewritten since Plasma 5.5, I'm going to close this. We've not got any duplicates newer than Plasma 5.4 |