Red Hat Bugzilla – Bug 248879
tpb eating CPU (> 25% Xorg)
Last modified: 2007-11-30 17:12:10 EST
My Z60t ThinkPad 'idling': PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 2128 root 15 0 36516 2444 1356 S 12.0 0.2 138:15.30 NetworkManager 2597 sd 15 0 598m 311m 73m R 0.7 31.1 256:16.31 firefox-bin 2224 root 15 0 579m 181m 9216 R 0.3 18.1 48:07.32 Xorg 2332 sd 15 0 15576 1264 1000 S 0.3 0.1 13:52.50 tpb tpb is using more than 25% as much CPU as Xorg (and I'm not mashing on the volume buttons!). uptime above: 2 days. I don't recall seeing tpb that high before, but I haven't been looking since I upgraded from FC6 so I can't be sure when this started happening. kernel-2.6.21-1.3228.fc7 tpb-0.6.4-8.fc7
(Sorry for the delay here, I was out of town...) quite odd. What arch is your install? i686? or x86_64? Can you see if the latest update kernel fixes the issue?
Arch is i686. I've updated to kernel-2.6.22.1-41.fc7 (and all other updates) and the problem is still there. I also tried lowering POLLTIME in /etc/tpbrc from the default 200000 to 400000 but it didn't seem to make much difference to the CPU consumption, although the volume buttons often failed to work... :-) Is there a way to get rid of tpb altogether? I thought I read on planet Gnome (Richard Hughes?) that the thinkpad extra buttons were supported as ACPI events. What listens for these events? Is there a package I can install, so that the polling tpb can be dropped?
Perhaps you mean this post: http://hughsient.livejournal.com/29730.html If you can bypass tpb that would be a great solution. tpb seems pretty much dead upstream. ;(
Yeah, that's the one. So, looks like tpb is dead but HAL will handle this in F8. Groovy. I guess this bug can be closed? Thanks for looking!
Sure, I'd be happy to close this... Do you have any way to test and see if you can get things working with the new setup? I hate to close this without being sure that it will get solved for you. Possibly booting the f8test1 live image?
I'll go ahead and close this now. Feel free to reopen or file a new bug if you see it again...