[plasmashell] [Bug 483163] blank screen on lock screen activation when using breeze plasma style

[Jakob Petsovits]

https://bugs.kde.org/show_bug.cgi?id=483163

--- Comment #21 from Jakob Petsovits <jpetso at petsovits.com> ---
If it's not a caching issue, and it's not necessarily a race condition, it
could also be an issue with invalid memory access somewhere. I decided to run
`kscreenlocker_greet --testing` in valgrind to see what it says.

Here's the excerpt that should roughly line up with window creation (I forgot
to specify more verbose QT_LOGGING_RULES):

kf.windowsystem: Loaded plugin
"/home/kpetso/build/prefix/lib/plugins/kf6/kwindowsystem/KF6WindowSystemX11Plugin.so"
for platform "xcb"
kf.coreaddons.kdirwatch: Available methods:  QList("Stat", "INotify",
"QFileSystemWatcher") preferred= INotify
kf.kirigami.platform: Loading style plugin from
"/home/kpetso/build/prefix/lib/plugins/kf6/kirigami/platform/KirigamiPlasmaStyle.so"
==10885== Conditional jump or move depends on uninitialised value(s)
==10885==    at 0x1D0CD789: ???
==10885==    by 0x218D1E4F: ???
==10885== 
Locked at 1711932039
==10885== Conditional jump or move depends on uninitialised value(s)
==10885==    at 0x1D0CD789: ???
==10885==    by 0x1D2D08DF: ???
==10885== 
==10885== Conditional jump or move depends on uninitialised value(s)
==10885==    at 0x1D0CD789: ???
==10885==    by 0x2140FF5F: ???
==10885== 
kscreenlocker_greet: PamAuthenticators: starting authenticators
kscreenlocker_greet: PamAuthenticators: state changing from
PamAuthenticators::Idle to PamAuthenticators::Authenticating

And after the PAM authentication workers have started, we've got two invalid
reads with an actually useful backtrace for which I just created
https://bugreports.qt.io/browse/QTBUG-123878. Either I'm stupid and can't read
code / valgrind output, or this is actually a non-trivial bug in central Qt
code.

It's somewhat plausible that our bug is coming from one of these invalid memory
accesses. I wish I had an actually useful backtrace for the "Conditional jump"
ones too. Not sure why valgrind denies me that one. Missing frame pointers on
Arch perhaps?

If you have a debug build of kscreenlocker at hand, maybe you can see if your
invocation of `valgrind ${KDE_PREFIX}/lib/libexec/kscreenlocker_greet
--testing` gives you something more useful than the pasted question marks
above.

-- 
You are receiving this mail because:
You are on the CC list for the bug.