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

Re: sudden power loss on tibook

> Michel Dänzer wrote:
> > My assumption is that the sudden shutdowns are triggered by some kind of
> > race condition when switching CPU speed. If powernowd switches less
> > often than cpudyn, it may indeed be significantly less likely to
> > trigger. Of course, cpudyn could be configured to switch less often to
> > verify that theory.
>  From watching Gnome's cpu-frequency-monitor, it would appear that
> cpufreqd, in the configuration I had it, did not adjust my cpu frequency
> at all except when changing between AC and battery. The system would

Well, in that case, powernowd seems to be the better choice anyway. Except
it also locks up on TiBooks :-(

> appear to be fixed at full CPU speed ('performance' profile), and
> occasionally shut down spontaneously anyway.

Odd. Unless cpufreqd would drop the CPU frequency only for a very short
period - anything in the logs about frequency switching?

> Now I'm using powernowd which switches the cpu frequently according to
> CPU load (these switches are visible in the Gnome applet).
> Also, is it significant that the hardware clock gets reset (to
> 1904-01-01) after one of these shutdowns? Does this imply anything about
> where the fault is occurring?

Not really. This is typical for loss of communication between host CPU and
PMU (the PMU has a timeout on that, and just cuts power after timeout,
resetting the system clock). I'm not sure whether this only applies to
host initiated PMU comms - that would mean we have a lockup happen while
talking to the PMU. Otherwise, it might be just any lockup with interrupts
disabled (does it happen on the text console as well? Does the cursor keep


Reply to: