Summary: | KDE 4.5.77: kded 100% CPU apparently in Solid module | ||
---|---|---|---|
Product: | [Frameworks and Libraries] solid | Reporter: | Andrey Borzenkov <arvidjaar> |
Component: | libsolid-fstab | Assignee: | Mario Bensi <nef> |
Status: | RESOLVED WORKSFORME | ||
Severity: | normal | CC: | afiestas, amg1127, mail, majewsky, mbensi, mmodem00, niko.sams, perezmeyer, stephan.diestelhorst |
Priority: | NOR | ||
Version: | unspecified | ||
Target Milestone: | --- | ||
Platform: | Mandriva RPMs | ||
OS: | Linux | ||
Latest Commit: | Version Fixed In: |
Description
Andrey Borzenkov
2010-11-21 10:21:58 UTC
Seem related to the fstab backend. Sounds like something difficult to reproduce though. Reassigning on the right component. *** Bug 263508 has been marked as a duplicate of this bug. *** *** Bug 259388 has been marked as a duplicate of this bug. *** *** This bug has been confirmed by popular vote. *** Same issue here with KDE 4.6.2 on Gentoo. Exactly the same backtrace for 4 running kded4 processes as Andrey posted. Unfortunately this is a very common bug. It seems that kded never exits properly, it always hits the qt_xio_errhandler(). See also http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=621358 The problem is that qt_xio_errhandler() kills the QCoreApplication event delivering mechanism, however QFileSystemWatcher requires it in order to destruct itself. It needs to send an event to its polling thread to instruct it to quit, however that event cannot be delivered inside the polling thread because the event delivering mechanism is dead and it stucks in the event loop calling poll(), which returns immediately (because there *are* pending events) and thus keeps the cpu busy entering and exiting the poll() system call... Unfortunately, there seems to be no way to handle this once qt_xio_errhandler() has run. This is really a Qt bug imho, but the Qt docs mention it: "Note: To avoid deadlocks on shutdown, all instances of QFileSystemWatcher need to be destroyed before QCoreApplication." ...so it seems that one should not be using a QFileSystemWatcher this way. I removed the deletion of QFileSystemWatcher from KDE 4.6.2 I don't know how QFileSystemWatcher could be destroy by fstab solid backend. If you want check the change, i give you the urls of this change : https://projects.kde.org/projects/kde/kdelibs/repository/revisions/42d40d1d351588a71bef0af1d62a8f6dc586f141 https://projects.kde.org/projects/kde/kdelibs/repository/revisions/350a5d8de016b6daa36c6e29d5d5f83ad6c2b38d Actually i can't reproduce this bug on master and on my pc with 4.6.2 and 4.6.3, if you have a way to reproduce this, I will able to fix this Mario I am getting this now too, after a recent upgrade in debian testing. Googling around, I found another issue that seems related. https://bugs.archlinux.org/task/23612 In [0] it has been suggested that this bug is the same as [1] and can be worked around by: You can use ~/.kde/shutdown/kded4.sh script with this command: #!/bin/sh killall -9 kded4 [0] <http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=621358#34> [1] <https://bugs.kde.org/show_bug.cgi?id=265400> (In reply to comment #9) > In [0] it has been suggested that this bug is the same as [1] This bug is unrelated to logout, it appears during the session. This should be fixed now, both ntrack and fstab issue. Can anyone reproduce this? Closing the thread for lack of activity. Please, please! if you are still able to reproduce this with KDE 4.10 or 4.11 feel free to reopen the bug! Thanks for reporting ! |