| Summary: | System Settings crashes in KNSCore::KPackageJob::~KPackageJob() when installing new themes | ||
|---|---|---|---|
| Product: | [Frameworks and Libraries] frameworks-knewstuff | Reporter: | jbclark325 |
| Component: | general | Assignee: | Dan Leinir Turthra Jensen <admin> |
| Status: | RESOLVED FIXED | ||
| Severity: | crash | CC: | admin, aritra.sett, hara75gr, joseph.d.harris78, karl, kde, kdelibs-bugs, lakshaygera63, mateusz.walas, nate, rolandhuu, shkumar006, techxgames, ucanfindmr.nero |
| Priority: | VHI | Keywords: | drkonqi |
| Version: | 5.20.0 | ||
| Target Milestone: | --- | ||
| Platform: | openSUSE | ||
| OS: | Linux | ||
| Latest Commit: | https://invent.kde.org/frameworks/knewstuff/commit/e1917b6a3bc876ef852ef22769fde07943ea3847 | Version Fixed In: | 5.76 |
| Sentry Crash Report: | |||
| Attachments: |
New crash information added by DrKonqi
New crash information added by DrKonqi New crash information added by DrKonqi |
||
|
Description
jbclark325
2020-10-18 09:02:05 UTC
*** Bug 427915 has been marked as a duplicate of this bug. *** Still crashing in KNSCore::KPackageJob::~KPackageJob() even with Frameworks 5.75. :( *** Bug 427821 has been marked as a duplicate of this bug. *** Still crashing in KNSCore::KPackageJob::~KPackageJob() even with Frameworks 5.75. :( Indeed... i am discovering new and exciting things, and a patch is arriving shortly which i would greatly appreciate if someone who can reproduce the crash would be so kind as to test out for me :) A possibly relevant merge request was started @ https://invent.kde.org/frameworks/knewstuff/-/merge_requests/66 Created attachment 132625 [details]
New crash information added by DrKonqi
systemsettings5 (5.20.0) using Qt 5.15.1
- What I was doing when the application crashed:
just downloaded some global themes and it crashed after I finished
-- Backtrace (Reduced):
#4 0x00007efec98e0c90 in QScopedPointerDeleter<KPackage::PackageStructure>::cleanup (pointer=0x55876d919b40) at /usr/include/qt5/QtCore/qscopedpointer.h:60
#5 QScopedPointer<KPackage::PackageStructure, QScopedPointerDeleter<KPackage::PackageStructure> >::~QScopedPointer (this=<optimized out>, this=<optimized out>) at /usr/include/qt5/QtCore/qscopedpointer.h:107
#6 KPackageTask::~KPackageTask (this=<optimized out>, this=<optimized out>) at /usr/src/debug/knewstuff-5.75.0-1.1.x86_64/src/core/jobs/kpackagejob.cpp:62
#7 KPackageTask::~KPackageTask (this=<optimized out>, this=<optimized out>) at /usr/src/debug/knewstuff-5.75.0-1.1.x86_64/src/core/jobs/kpackagejob.cpp:62
#8 0x00007eff2b59fc7e in QObjectPrivate::deleteChildren (this=0x55876f654570) at kernel/qobject.cpp:2104
Git commit e1917b6a3bc876ef852ef22769fde07943ea3847 by Dan Leinir Turthra Jensen. Committed on 21/10/2020 at 14:24. Pushed by leinir into branch 'master'. Fix double-delete crash in kpackagejob Don't hang onto a pointer that's already been deleted, as QScopedPointer will just delete that for us. The semantics of KPackage::PackageStructures is that someone else /might/ own it, while the semantics of KPackage::Package is that we definitely own it, so we need to track the one we don't really know about. M +2 -0 src/core/jobs/kpackagejob.cpp https://invent.kde.org/frameworks/knewstuff/commit/e1917b6a3bc876ef852ef22769fde07943ea3847 *** Bug 428055 has been marked as a duplicate of this bug. *** Created attachment 132687 [details]
New crash information added by DrKonqi
plasma-discover (5.20.1) using Qt 5.15.0
- What I was doing when the application crashed:
14 themes show up everytime under "updates" in Discover. I can't get rid of them. I've tried updating them and when I do Discover crashes. When reopened they are there again.
-- Backtrace (Reduced):
#5 0x00007f07bb744d67 in QScopedPointerDeleter<KPackage::PackageStructure>::cleanup (pointer=<optimized out>) at /usr/include/x86_64-linux-gnu/qt5/QtCore/qscopedpointer.h:52
#6 QScopedPointer<KPackage::PackageStructure, QScopedPointerDeleter<KPackage::PackageStructure> >::~QScopedPointer (this=0x5628a2526730, __in_chrg=<optimized out>) at /usr/include/x86_64-linux-gnu/qt5/QtCore/qscopedpointer.h:107
#7 KPackageTask::~KPackageTask (this=0x5628a25266f0, __in_chrg=<optimized out>) at ./src/core/jobs/kpackagejob.cpp:62
#8 KPackageTask::~KPackageTask (this=0x5628a25266f0, __in_chrg=<optimized out>) at ./src/core/jobs/kpackagejob.cpp:62
#9 0x00007f082f7d6bae in QObjectPrivate::deleteChildren (this=0x5628a27b5b50) at kernel/qobject.cpp:2104
Created attachment 132703 [details]
New crash information added by DrKonqi
plasma-discover (5.20.1) using Qt 5.15.0
- What I was doing when the application crashed:
I was just updating themes on Discover
Operating System: KDE neon Testing Edition
KDE Plasma Version: 5.20.1
KDE Frameworks Version: 5.76.0
Qt Version: 5.15.0
Kernel Version: 5.8.0-23-generic
OS Type: 64-bit
Processors: 4 × Intel® Core™ i5-6500 CPU @ 3.20GHz
Memory: 15.6 GiB of RAM
Graphics Processor: GeForce GTX 1060 6GB/PCIe/SSE2
-- Backtrace (Reduced):
#5 0x00007f96ae0a761e in QScopedPointerDeleter<QObjectData>::cleanup(QObjectData*) (pointer=<optimized out>) at ../../include/QtCore/../../src/corelib/tools/qscopedpointer.h:52
#6 QScopedPointer<QObjectData, QScopedPointerDeleter<QObjectData> >::~QScopedPointer() (this=0x5580805f58e8, __in_chrg=<optimized out>) at ../../include/QtCore/../../src/corelib/tools/qscopedpointer.h:107
#7 QObject::~QObject() (this=<optimized out>, __in_chrg=<optimized out>) at kernel/qobject.cpp:972
#8 0x00007f9634fe25f7 in ComicPackage::~ComicPackage() (this=0x5580805f58e0, __in_chrg=<optimized out>) at ./dataengines/comic/comic_package.h:24
#9 ComicPackage::~ComicPackage() (this=0x5580805f58e0, __in_chrg=<optimized out>) at ./dataengines/comic/comic_package.h:24
It was pointed out to me that the logic i assumed QScopedPointer was using was not incorrect. On Saturday afternoon EU time, i pushed https://invent.kde.org/frameworks/knewstuff/commit/a95e5037b79b648f663f8db7ded6800ec6e49c75 which should hopefully sort it. If it doesn't, i guess we'll just have to leak that pointer and trust KPackage to clean up after itself. Thank you also to everybody who's been testing this out, since as it happens i have nothing to hand that provokes this crash... (if that sounds like frustration, it's because it kind of is - it's no fun watching something you spent this long on crashing left and right, but it is heartening to have people helping out with the testing of fixes, and seeing the neon dev testing branch working as intended for that purpose!) (In reply to Dan Leinir Turthra Jensen from comment #12) > It was pointed out to me that the logic i assumed QScopedPointer was using > was not incorrect. (also i can grammar, honest, double negative unintended, though i imagine that's clear from the context anyway ;) ) *** Bug 428370 has been marked as a duplicate of this bug. *** *** Bug 427786 has been marked as a duplicate of this bug. *** *** Bug 428505 has been marked as a duplicate of this bug. *** *** Bug 428576 has been marked as a duplicate of this bug. *** *** Bug 428622 has been marked as a duplicate of this bug. *** *** Bug 435153 has been marked as a duplicate of this bug. *** |