Current-Users archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: firefox slowness -- X server problem?

On Mon, Aug 23, 2010 at 07:17:12PM +0100, Patrick Welche wrote:
> > This should be fixed now.
> Any clues on what was up?

This was just a bug, nothing that interesting.

More specifically, the cpu_need_resched(9) function needs to send an inter-
processor interrupt (IPI) for a CPU that is idling using the STI-HLT sequence
("enable interrupts, halt"). A hard-coded value was used to check if "halt"
is being used.  This was unnoticed because acpicpu(4) was initially written
and primarily tested on Intel CPUs, which typically use the alternative MWAIT
instruction; with the accompanying MONITOR instruction, the CPU will wake
automatically when the monitored "need rescheduling" flag changes.

- Jukka.

Home | Main Index | Thread Index | Old Index