https://bugs.kde.org/show_bug.cgi?id=427191
Bug ID: 427191
Summary: bad config for kscreen causes lots of things to
coredump
Product: KScreen
Version: 5.18.5
Platform: Fedora RPMs
OS: Linux
Status: REPORTED
Severity: normal
Priority: NOR
Component: common
Assignee: kscreen-bugs-n...@kde.org
Reporter: dklot...@gmail.com
Target Milestone: ---
SUMMARY
This was hard to track down because there's nothing in the logs to indicate
that the config is bad, or that the monitors aren't happy
STEPS TO REPRODUCE
1. Make a bogus config for 2 monitors that doesnt work in
~/.local/share/kscreen/<GUID>
2. Try and log in via SDDM
OBSERVED RESULT
SDDM either crashes and leaves you at a flashing cursor, drops you back to the
login, or leaves you at a black screen with a cursor. Log messages from
journalctl are like:
Process 6590 (kwin_wayland) of user 1000 dumped core.
Stack trace of
thread 6590:
#0
0x00007f36a2e0f9e5 n/a (n/a + 0x0)
systemd-coredump[7949]: Process 7462 (org_kde_powerde) of user 1000 dumped
core.
Stack trace of
thread 7462:
#0
0x00007feb9ff799e5 raise (libc.so.6 + 0x3c9e5)
#1
0x00007feb9ff79a70 __restore_rt (libc.so.6 + 0x3ca70)
#2
0x00007feb9ff799e5 raise (libc.so.6 + 0x3c9e5)
#3
0x00007feb9ff62895 abort (libc.so.6 + 0x25895)
#4
0x00007feba038aa7f _ZNK14QMessageLogger5fatalEPKcz (libQt5Core.so.5 + 0x91a7f)
#5
0x00007feb8f0928c9 _ZNK15QtWaylandClient15QWaylandDisplay10checkErrorEv.cold
(libQt5WaylandClient.so.5 + 0x5b8c9)
#6
0x00007feb8f0a0ede _ZN15QtWaylandClient15QWaylandDisplay13flushRequestsEv
(libQt5WaylandClient.so.5 + 0x69ede)
#7
0x00007feba0582fb0 _Z10doActivateILb0EEvP7QObjectiPPv (libQt5Core.so.5 +
0x289fb0)
#8
0x00007feba0585f4c _ZN15QSocketNotifier9activatedEiNS_14QPrivateSignalE
(libQt5Core.so.5 + 0x28cf4c)
polkit-kde-authentication-agent-1[7946]: Failed to create wl_display (No such
file or directory)
polkit-kde-authentication-agent-1[7946]: qt.qpa.plugin: Could not load the Qt
platform plugin "wayland" in "" even though it was found.
systemd-coredump[7864]: Process 7609 (kwalletd5) of user 1000 dumped core.
Stack trace of
thread 7609:
#0
0x00007f3eec6b99e5 raise (libc.so.6 + 0x3c9e5)
#1
0x00007f3eec6a2895 abort (libc.so.6 + 0x25895)
systemd-coredump[26514]: Process 8389 (plasmashell) of user 1000 dumped core.
Stack trace of
thread 25673:
#0
0x00007f34da7369e5 raise (libc.so.6 + 0x3c9e5)
#1
0x00007f34da736a70 __restore_rt (libc.so.6 + 0x3ca70)
#2
0x00007f34da63e7f9 wl_proxy_set_queue (libwayland-client.so.0 + 0x77f9)
#3
0x00007f34c9b2cf15 _ZN15QtWaylandClient14QWaylandWindow16waitForFrameSyncEi
(libQt5WaylandClient.so.5 + 0x6ef15)
EXPECTED RESULT
Usable Debug logging would be helpful.
kwin_wayland, plasmashell, org_kde_powerde, kactivitymanage, kglobalaccel5,
kwalletd5 shouldn't all coredump in this case, and should exit gracefully.
possibly wait until the screen config is applied to start all these services
and fail faster.
if the kscreen config is invalid it should have some input validation and
either back out the config (leaving it in someplace safe to review) and revert
to a default
log that there is a problem with the display adapters.
SOFTWARE/OS VERSIONS
Linux/KDE Plasma: Fedora 32, plasma-workspace-wayland 5.18.5-2.fc32
(available in About System)
KDE Plasma Version: 5.18.5-2.fc32
KDE Frameworks Version: 5.70.1-1.fc32
Qt Version: 5.14.2-4.fc32
ADDITIONAL INFORMATION
Fedora 32, amdgpu, qt5, wayland
--
You are receiving this mail because:
You are watching all bug changes.
