On 30 Jul 2008, at 12:39, John Martin wrote: > Chris Ridd wrote: >> >> >> >> Here's the output of "intrstat 5". Steady state is around the >> 20000 mark, and dragging a big Thunderbird window around (causing >> a lot of gnome-terminal windows to white out and not redraw >> themselves immediately) increases it to around 24000. > nvidia#0, uhci#0 and uhci#3 are sharing a legacy interrupt so each > driver's > interrupt handler is being called each time the interrupt goes off. > A 20K number > is usually an indication the interrupt is stuck on and is only being > limited by > how fast the CPU assigned to it (cpu3) can process them.
Nod. > When you originally reported this it was happening on two systems: a > desktop > box sith NVIDIA Quadro FX 540 and a laptop with ATI FireGL 9000. Do > you see the same interrupt issue on the notebook? No, the notebook's reporting numbers of around 100-200. > If the SBIOS allows it can you selectively shut off each of your USB > controllers > to isolate which one is the source of the interrupts? It (a Dell Precision 470) appears to just have a single switch to disable "the" USB controller. My kbd and mouse are both USB (looks like they're on uhci#2), so disabling USB entirely would make the machine hard to use :-) I've just rebooted into snv94, and the intrstat figures are back down in the 100 range. I guess I'll keep watching out to see if they go bezerk again. Cheers, Chris
