Created attachment 72676 [details] system log when khubd took 100% cpu after lid close Does not happen every time - but seems to be more likely after a sleep them being used for a day and using bluetooth mouse. Scenario A) Close laptop lid - does not sleep - re-open lid and find khubd with 100% CPU according to top. System log is attached (file log-khubd-100pctcpu) Scenario B) Close laptop lid - does not sleep - open lid see oops on screen. Attached cell phone pic of screen: (file name: oops-2012-03-21_18.38.37.jpg) (virtualbox module is loaded but not used - so kernel is tainted from that). Problem did not happen under 3.2.x This might be similar to:https://bugzilla.kernel.org/show_bug.cgi?id=42788 laptop is lenovo w520 - ver_linux output attached and lspci -vvv. gene
Created attachment 72677 [details] ver_linux output
Created attachment 72678 [details] lspci -vvv
Created attachment 72679 [details] cell phone image of oops
I have the same problem, khubd goes on rampage when I want to shutdown or suspend, preventing both actions from successfully completing. This happens when I use my bluetooth headset, in which case I am unable to perform a clean shutdown because the system hangs waiting for khubd to stop. If I do not use the headset, the shutdown process is a breeze. Alessandro
Same problem with bluetooth mouse. Arch Linux x86_64, linux 3.3.1, laptop Thinkpad T510
Created attachment 72917 [details] khubd.txt More or less the same here. No 100% CPU eating process, I just cannot put my _desktop_ PC to sleep after I had some troubles with a (probably faulty) USB flash stick.
(In reply to comment #0) khubd is a USB arbitrator process ( http://www.linuxjournal.com/article/8093 ) in the kernel, so this bug is in the USB subsystem, not somewhere else. If you can reproduce this bug every time you boot, probably you should try finding the bad commit using git-bisect (of course, you'll have to learn how to compile and use vanillla kernel).