https://bugs.kde.org/show_bug.cgi?id=528373
Bug ID: 528373
Summary: Unreachable NFS server freezes the Plasma session:
Solid FstabStorageAccess calls access() on the mount
point in the GUI thread
Classification: Frameworks and Libraries
Product: frameworks-solid
Version First 6.31.0
Reported In:
Platform: Fedora RPMs
OS: Linux
Status: REPORTED
Severity: normal
Priority: NOR
Component: general
Assignee: [email protected]
Reporter: [email protected]
CC: [email protected], [email protected]
Target Milestone: ---
SYMPTOMS
- Volume keys react only after a long delay.
- Right-click on the desktop shows no context menu.
- Actions taken in the meantime (new windows, menu clicks) are executed later,
all at once.
- Dolphin no longer responds.
- Restarting plasmashell does not help: the new instance blocks during startup
(see stack trace below).
- Restarting kded6 helps: after "systemctl --user restart plasma-kded6.service"
the session responds again.
STEPS TO REPRODUCE
1. Mount an NFS share without options (Fedora defaults):
sudo mount t410.home:/storage/data1 /mnt/t410/hdd-3tb
sudo mount t410.home:/storage/data2 /mnt/t410/raid-3tb
2. Make the NFS server unreachable.
3. Use the Plasma session (desktop, volume keys, Dolphin).
OBSERVED RESULT
Kernel, from the moment the server became unreachable, about 6 messages per
minute, still ongoing the next day:
2026-11-04T09:23:51+01:00 sudo: COMMAND=/usr/sbin/mount
t410.home:/storage/data1 /mnt/t410/hdd-3tb
2026-11-04T09:23:52+01:00 sudo: COMMAND=/usr/sbin/mount
t410.home:/storage/data2 /mnt/t410/raid-3tb
2026-11-04T10:13:59+01:00 kernel: nfs: server t410.home not responding, timed
out
2026-11-04T10:23:37+01:00 kernel: nfs: server t410.home not responding, still
trying
...
2026-11-05T10:53:59+01:00 kernel: nfs: server t410.home not responding, timed
out
Unmounting does not release the processes:
2026-11-04T12:41:42+01:00 krusader[7344]: QProcess: Destroyed while process
("/usr/bin/umount") is still running.
2026-11-04T14:05:28+01:00 sudo: COMMAND=/usr/sbin/umount /mnt/t410/hdd-3tb
2026-11-04T14:06:06+01:00 sudo: COMMAND=/usr/sbin/umount -l
/mnt/t410/raid-3tb
2026-11-04T14:08:12+01:00 sudo: COMMAND=/usr/sbin/umount -fl
/mnt/t410/raid-3tb
After "umount -fl" the mount is gone from /proc/self/mountinfo, but the NFS
superblock stays (/sys/fs/nfs/0:87) and the kernel keeps sending RPCs.
/proc/<pid>/wchan showed rpc_wait_bit_killable (state D) for:
- kded6 (PID 4548)
- dolphin (PID 54577, main thread)
plasmashell cannot be restarted; each new instance hangs and is killed by
systemd after the stop timeout:
2026-11-04T14:04:11+01:00 systemd[4221]: Started plasma-plasmashell.service -
KDE Plasma Workspace.
2026-11-04T14:06:14+01:00 systemd[4221]: Stopping plasma-plasmashell.service
- KDE Plasma Workspace...
2026-11-04T14:06:54+01:00 systemd[4221]: plasma-plasmashell.service: State
'stop-sigterm' timed out. Aborting.
2026-11-04T14:06:54+01:00 systemd[4221]: plasma-plasmashell.service: Killing
process 102434 (plasmashell) with signal SIGABRT.
2026-11-04T14:07:51+01:00 systemd[4221]: plasma-plasmashell.service: State
'stop-sigterm' timed out. Aborting.
2026-11-04T14:07:51+01:00 systemd[4221]: plasma-plasmashell.service: Killing
process 195026 (plasmashell) with signal SIGABRT.
2026-11-04T14:07:48+01:00 ksystemstats[211734]: Could not determine Plasma
version, got: "Did not receive a reply. [...]"
2026-11-04T14:08:19+01:00 abrt-applet[5170]: unable to send notifications
through org.freedesktop.Notifications: Timeout was reached
kded6 does not react to SIGTERM or SIGABRT either; restarting the service ends
the hang:
2026-11-04T14:12:43+01:00 systemd[4221]: Stopping plasma-kded6.service - KDE
Daemon 6...
2026-11-04T14:12:48+01:00 systemd[4221]: plasma-kded6.service: State
'stop-sigterm' timed out. Aborting.
2026-11-04T14:12:48+01:00 systemd[4221]: plasma-kded6.service: Killing
process 4548 (kded6) with signal SIGABRT.
2026-11-04T14:12:53+01:00 systemd[4221]: plasma-kded6.service: Killing
process 4548 (kded6) with signal SIGKILL.
2026-11-04T14:12:53+01:00 systemd[4221]: Started plasma-kded6.service - KDE
Daemon 6.
STACK TRACE
Main thread of plasmashell PID 102434 (started 14:04:11, before the umount
attempts), from the core dump written after SIGABRT at 14:06:55. The thread is
in access() on the mount point, called from the Solid Fstab backend while the
Device Notifier applet is created during shell startup:
#0 access (libc.so.6 + 0xe442b)
#1 QFileSystemEngine::fillMetaData(QFileSystemEntry const&,
QFileSystemMetaData&, QFlags<QFileSystemMetaData::MetaDataFlag>)
(libQt6Core.so.6 + 0x2f8113)
#2 QDir::isReadable() const (libQt6Core.so.6 + 0x916ff)
#3
Solid::Backends::Fstab::FstabStorageAccess::FstabStorageAccess(Solid::Backends::Fstab::FstabDevice*)
(libKF6Solid.so.6 + 0x61ec6)
#4
Solid::Backends::Fstab::FstabDevice::createDeviceInterface(Solid::DeviceInterface::Type
const&) (libKF6Solid.so.6 + 0x61fa1)
#5 Solid::Device::asDeviceInterface(Solid::DeviceInterface::Type const&)
const (libKF6Solid.so.6 + 0x170cf)
#6 DeviceControl::onDeviceAdded(QString const&)
(org.kde.plasma.devicenotifier.so + 0x220ad)
#7 DeviceFilterControl::DeviceFilterControl(QObject*)
(org.kde.plasma.devicenotifier.so + 0x442e6)
#8 void QQmlPrivate::createInto<DeviceFilterControl>(void*, void*)
(org.kde.plasma.devicenotifier.so + 0x5835)
...
#19 PlasmaQuick::AppletQuickItem::itemForApplet(Plasma::Applet*)
(libPlasmaQuick.so.8 + 0x10619)
#20 ContainmentItem::init() (libPlasmaQuick.so.8 + 0x3fef4)
...
#30 PlasmaQuick::ContainmentViewPrivate::setContainment(Plasma::Containment*)
(libPlasmaQuick.so.8 + 0x2540b)
#31 ShellCorona::createWaitingPanels() (plasmashell + 0x46e4f)
#32 ShellCorona::checkAllDesktopsUiReady() (plasmashell + 0x65ba2)
...
#48 QCoreApplication::exec() (libQt6Core.so.6 + 0x115ad2)
EXPECTED RESULT
An unreachable network mount must not block the GUI thread of plasmashell or
Dolphin, nor kded6. FstabStorageAccess should not call
access()/QDir::isReadable() on network file systems synchronously in its
constructor; the check should be deferred, run in a worker thread, or skipped
for NFS/CIFS.
WORKAROUND
- systemctl --user restart plasma-kded6.service (the session responds again)
- echo 1 | sudo tee /sys/fs/nfs/<dev>/shutdown (makes outstanding RPCs fail
with EIO, kernel 6.8 and later)
SOFTWARE
- Fedora 44, kernel 7.2.15-200.fc44.x86_64
- Plasma 6.8.2 (plasma-workspace, plasma-desktop), Wayland session
- KDE Frameworks 6.31.0 (kf6-solid, kf6-kio-core, kf6-kded)
- Dolphin 26.08.2
- Qt 6.11.2
- nfs-utils 2.8.7
--
You are receiving this mail because:
You are watching all bug changes.
