system freezes

[Nils Kassube]

Xen wrote:
> Nils Kassube schreef op 09-11-2017 11:25:
> > Xen wrote:
> >> I reset the CMOS but no difference.
> >> 
> >> [ 4276.865023] clocksource: timekeeping watchdog on CPU3: Marking
> >> clocksource 'tsc' as unstable because the skew is too large:
> >> [ 4276.865027] clocksource:                       'hpet' wd_now:
> >> e559648a wd_last: 30656759 mask: ffffffff
> >> [ 4276.865037] clocksource:                       'tsc' cs_now:
> >> a42bfbbea74 cs_last: 9f8e57ce8e2 mask: ffffffffffffffff
> >> [ 4276.870912] clocksource: Switched to clocksource hpet
> >> 
> >> It took a while I guess, more than an hour?
> >> 
> >> All I can do is replace the battery, otherwise this mobo is not fit
> >> for Linux.
> > 
> > Colin already mentioned that the battery shouldn't be a problem
> > because the RTC is read at startup only. Maybe you should rather
> > replace the CPU:
> > <https://forums.lime-technology.com/topic/45190-marking-clocksource-> > tsc-as-unstable-because-the-skew-is-too-large/>
> Thanks for the pointer. That was informative.
> 
> However.
> 
> I have ran Ubuntu 14.04 Live for days without any issue.

That is indeed an indication, that it isn't related to the CPU. OTOH it 
is also an indication that it isn't related to the battery.

> I mean, I can also just say: Yeah, spend another $40 on a DDR2
> system...
> 
> The guy also doesn't mention what he replaced it with, and the 955 was
> what I wanted to replace mine _with_ rather than get away from ;-).
> 
> I can't go randomly test $40 CPUs and the only one I have left I
> think...
> 
> Well whatever.
> 
> I could check if the hang occurs during such a message but the above
> thing you quoted happened without a hang.

I can't imagine that switching the clock source from tsc to hpet could 
cause a hang - probably those issues aren't related.

> Other than that my symptom appears to be the same as the thread you
> linked.
> 
> I'm just a bit annoyed at the thought of throwing random hardware at
> what is basically a kernel problem... :(.

Sure, thet makes sense and I don't know if the CPU really causes the 
trouble. It was just an interesting thread I found when I tried to 
understand what the kernel message means.

Of course a new battery would be less expensive than the CPU, but 
replacing the battery is also what you call "throwing random hardware at 
what is basically a kernel problem", IMHO.

> I have not had a hang thus far though after the CMOS reset.

Let's hope that it really was the cure.


Nils