KDE 4.4.4 in its default configuration (or the powerdevil daemon?) did not react to the ACPI event of the sleep-button. But instead of configuring something in my desktop-environment I used acpid for it. It allows defining actions under /etc/acpid/default.sh Now with KDE 4.6.2, those events are handled and the system correctly hibernates. But with acpid configured to also run hibernate or hibernate-ram, it is configured two times! Usually it would just create an odd double-hibernate... once woken up, my laptop would hibernate again right away. But hibernate on Linux is not always the most robust thing. Rather often (~20%) the immediat hibernate-ram after the first one would cause the kernel to panic and crash! (2.6.36, 37, 38 tested) It took me alot of testing before I figured that out. Maybe to save others some time, there could be some hint. For example if powerdevil is handling these acpi events, it could be some elog message like: if grep 'hibernate' /etc/acpi/default.sh > /dev/null; then echo Powerdevil handles certain ACPI events like hibernate on sleep-button. You might want to check your /etc/acpid/default.sh for conflicing scripts.; fi Or if there is some KDE upgrade guide, mention it there. I could imagine that some people think that it is the users own fault and he should be aware of such things himself. Feel free to close this bug then. Thought I could save someone some trouble with this =) Reproducible: Always
Thanks, added to the guide (work in progress) http://www.gentoo.org/proj/en/desktop/kde/kde44-46-upgrade.xml
Can you please add some information to the upgrade guide on how one can prevent KDE from handling these events? I prefer acpid scripts over automagic handling of these keys.