I'm using KDE 4.14.3 on Fedora 20. Whenever I open the Open or Save As dialogs in an app, I see the message "The process for the file protocol died unexpectedly", instead of a list of files. This makes them unusable. I was previously also seeing this in Dolphin, where my saved windows would appear on login but any attempt to browse to another directory would also give the message. The latter seems to have been fixed by deleting /tmp/kde-nick & /var/tmp/kdecache-nick. It's difficult to be sure because KDE's behaviour is unpredictable, particularly after a crash, which happens quite often, mostly because of this qxl bug.

This is what is written to ~/.xsession-errors when I do File -> Open in KWrite:

Code: Select all

NETWM: Warning readIcon() needs buffer adjustment!
kwrite(22918)/kio (KDirListerCache) KDirListerCache::listDir: Entry in cache: KUrl("trash:/")
kwrite(22918)/kio (Scheduler) KIO::SchedulerPrivate::doJob: KIO::SimpleJob(0x2e4a8b0)
kwrite(22918)/kio (KDirListerCache) KDirListerCache::listDir: Entry already in use: KUrl("trash:/")
kwrite(22918)/kio (KDirListerCache) KDirListerCache::stopListingUrl: KDirLister(0x2eaa320)  url= KUrl("trash:/")
kwrite(22918)/kio (Slave) KIO::Slave::createSlave: createSlave "file" for KUrl("file:///home/nick/Dolphin_crash.txt")
kwrite(22918)/kio (KIOConnection) KIO::ConnectionServer::listenForRemote: Listening on  "local:/run/user/1000/ksocket-nick/kwritem22918.slave-socket"
kwrite(22918)/kio (Slave) KIO::Slave::timeout: slave failed to connect to application pid= 3566  protocol= "file"
kwrite(22918)/kio (Slave) KIO::Slave::timeout: Houston, we lost our slave, pid= 3566
kwrite(22918)/kio (Slave) KIO::Slave::timeout: slave died pid =  3566
kwrite(22918)/kio (Scheduler) KIO::SchedulerPrivate::jobFinished: KIO::StatJob(0x2e4a8b0) KIO::Slave(0x2e47c70)
kwrite(22918)/kio (Scheduler) KIO::SchedulerPrivate::slotSlaveDied: KIO::Slave(0x2e47c70)
kwrite(22918)/kio (KDirListerCache) KDirListerCache::listDir: Listing directory: KUrl("file:///home/nick")
kwrite(22918)/kio (Scheduler) KIO::SchedulerPrivate::doJob: KIO::SimpleJob(0x2e30550)
kwin(22662) KWin::Client::readUserTimeMapTimestamp: User timestamp, ASN: 79352700
kwin(22662) KWin::Client::readUserTimeMapTimestamp: User timestamp, final: 'ID: 62915023 ;WMCLASS: "kwrite" : "kwrite" ;Caption: "Open File – KWrite" ' : 79352700
kwin(22662) KWin::Workspace::allowClientActivation: Activation: Belongs to active application
kwrite(22918)/kio (Slave) KIO::Slave::createSlave: createSlave "file" for KUrl("file:///home/nick")
kwrite(22918)/kio (KIOConnection) KIO::ConnectionServer::listenForRemote: Listening on  "local:/run/user/1000/ksocket-nick/kwriteZ22918.slave-socket"
kwin(22662) KWin::Workspace::updateClientArea: screens:  1 desktops:  1
kwin(22662) KWin::Workspace::updateClientArea: Done.
plasma-desktop(22682)/plasma TaskManager::TaskGroup::winIds: "Kwrite" 2
plasma-desktop(22682)/plasma TaskManager::TaskGroup::winIds: 2
plasma-desktop(22682)/plasma TaskManager::TaskGroup::winIds: "Kwrite" 2
plasma-desktop(22682)/plasma TaskManager::TaskGroup::winIds: 2
NETWM: Warning readIcon() needs buffer adjustment!
kwrite(22918)/kio (Slave) KIO::Slave::timeout: slave failed to connect to application pid= 3570  protocol= "file"
kwrite(22918)/kio (Slave) KIO::Slave::timeout: Houston, we lost our slave, pid= 3570
kwrite(22918)/kio (Slave) KIO::Slave::timeout: slave died pid =  3570
kwrite(22918)/kio (Scheduler) KIO::SchedulerPrivate::jobFinished: KIO::ListJob(0x2e30550) KIO::Slave(0x2e60990)
kwrite(22918)/kio (KDirListerCache) KDirListerCache::slotResult: finished listing KUrl("file:///home/nick")
kwrite(22918)/kio (Scheduler) KIO::SchedulerPrivate::slotSlaveDied: KIO::Slave(0x2e60990)
kwin(22662) KWin::Client::readUserTimeMapTimestamp: User timestamp, ASN: 79352700
kwin(22662) KWin::Client::readUserTimeMapTimestamp: User timestamp, final: 'ID: 62915060 ;WMCLASS: "kwrite" : "kwrite" ;Caption: "Error – KWrite" ' : 79352700
kwin(22662) KWin::Workspace::allowClientActivation: Activation: Belongs to active application
kwin(22662) KWin::Workspace::updateClientArea: screens:  1 desktops:  1
kwin(22662) KWin::Workspace::updateClientArea: Done.
kmix(22748) source_cb: Ignoring Monitor Source:  Monitor of Built-in Audio Analog Stereo
kded(22625)/kmix source_cb: Ignoring Monitor Source:  Monitor of Built-in Audio Analog Stereo
plasma-desktop(22682)/plasma TaskManager::TaskGroup::winIds: "Kwrite" 2
plasma-desktop(22682)/plasma TaskManager::TaskGroup::winIds: 2
plasma-desktop(22682)/plasma TaskManager::TaskGroup::winIds: "Kwrite" 2
plasma-desktop(22682)/plasma TaskManager::TaskGroup::winIds: 2
NETWM: Warning readIcon() needs buffer adjustment!
kwrite(22918)/kdeui (KNotification) KNotificationManager::notificationClosed: 13
kded(22625)/kmix Mixer_PULSE::removeWidget: Removing  2  widget notified for index  10  but I cannot find it in my list :s
kmix(22748) Mixer_PULSE::removeWidget: Removing  2  widget notified for index  10  but I cannot find it in my list :s


How can I stop this from happening?
How can I restore KDE to a known, stable state for a user or all users?

Thanks

It seems I must treat KDE as a black box. I started again by deleting ~/.kde as well as the /tmp/kde-user & /var/tmp/kdecache-user directories and so far the problem hasn't reappeared.