Re: [netbsd-5] machine losing time by the hour

[Hauke Fath <hf%spg.tu-darmstadt.de@localhost>]

At 19:44 Uhr +0100 23.03.2010, Matthias Drochner wrote:
>hf%spg.tu-darmstadt.de@localhost said:
>> where the new one has a whizzing
>> mainbus0 (root) cpu0 at mainbus0 apid 0: Intel 686-class, 8379MHz, id
>> 0xf33 cpu1 at mainbus0 apid 1: Intel 686-class, 8379MHz, id 0xf33
>
>So is this the same kernel where you got (in an earlier mail)?:
>
>> kern.timecounter.choice = TSC(q=3000, f=2992666350 Hz) clockinterrupt(q=0,
>> f=100 Hz) ACPI-Fast(q=1000, f=3579545 Hz) lapic(q=-100, f=199580162 Hz)
>> i8254(q=100, f=1193182 Hz) dummy(q=-1000000, f=1000000 Hz)

You got me there... above is the 2009-09 kernel's output.

Output from both kernels looked identical at first glance, which is why I
removed the second one; and they are, except for the TSC frequency. The
newer kernel has

% uname -a
NetBSD bounce 5.0_STABLE NetBSD 5.0_STABLE (BOUNCE) #0: Tue Mar 23 13:45:42
CET 2010
hf@Hochstuhl:/var/obj/netbsd-builds/5/i386/sys/arch/i386/compile/BOUNCE i386
%  sysctl -a|grep timecounter
kern.timecounter.choice = TSC(q=3000, f=8379204600 Hz) clockinterrupt(q=0,
f=100 Hz) ACPI-Fast(q=1000, f=3579545 Hz) lapic(q=-100, f=199573063 Hz)
i8254(q=100, f=1193182 Hz) dummy(q=-1000000, f=1000000 Hz)
kern.timecounter.hardware = ACPI-Fast
kern.timecounter.timestepwarnings = 0
%

>In theory, the CPU frequency reported and the "f" value for "TSC"
>should be the same.

I guess they are, then.

>I can't spot significant changes within the last 6 month in the netbsd-5
>branch -- could you tell the version number differences between the
>two versions in the sys/arch/x86/{x86,isa,include} subdirs?

I'll look at that tomorrow.

        hauke

-- 
     The ASCII Ribbon Campaign                    Hauke Fath
()     No HTML/RTF in email            Institut für Nachrichtentechnik
/\     No Word docs in email                     TU Darmstadt
     Respect for open standards              Ruf +49-6151-16-3281