umount.cifs cpu load issue


My CPU load increases by 1 everytime I use the umount.cifs command – anyone else seen this issue?


Same problem here cifs-mount-3.2.6-0.3.1. (opensuse 11.1)

ps ax | grep cifs give this after umount :
3653 ? S< 0:00 [cifsoplockd]
3654 ? S< 0:00 [cifsdnotifyd]
3663 ? D< 0:00 [cifsd]

I can’t get rid of cifsd whithout reboot and my load average become :

load average: 3.00, 2.34, 1.16

I don’t have this issue with opensue 11.0 and cifs-mount-3.2.6-0.3.1.

Any clue anyone?

This raises another issue – how is it the CPU load can be so out of sync with the processes in “ps” or “top”? Is that attributable to the kernel? If a kernel driver is using up CPU resources wouldn’t that show up in the calling process? How many resources does the kernel use that aren’t part of some calling process, even getty, network and init (which manage the computer) show up as individual processes with resources attributed to them.

Though, I did notice that “mount” isn’t a process, “mount” must command the kernel, then exits. Maybe that’s why a “umount.cifs” command may leave the kernel busy without having an associated process.

BTW: “hald-addon-stor” did the same thing to me, but killing that process recovered CPU resources.

Apparently it’s been recorded as a kernel bug, and apparently no one from OpenSuse pays attention to these forums, or bugs identified by the Fedora or kernel guys.

Yep, here is the bug: