Re: system clock speed too high? hyperthread problem?

[george anzinger]

Andreas Haumer wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Hi!
> 
> george anzinger wrote:
> 
>>Andreas Haumer wrote:
>>
> 
> [...]
> 
>>>I have a quite strange phenomenon here: I see a ~2.5 times
>>>speed up of system time on a Asus AP1700-S5 server with
>>>Linux-2.4.21-rc6-ac1.
>>>Simple proof: a "sleep 300" command terminates after exactly
>>>120 seconds of wall clock time.
>>
>>
>>Just as a wild shot in the dark, what speed does the kernel think the
>>cpu is running at and does this match what the BIOS thinks?
>>
>>It sounds like the CLOCK_TICK_RATE is wrong.  This would show up as the
>>kernel thinking the cpu was fast also.
>>
> 
> Hm, I don't think this is the case.
> BIOS reports (correctly) a single, hyperthreaded Xeon CPU
> with 2.4GHz

Ok, that means that the PIT is doing the right thing.  Since the clock 
is only advanced on PIT interrupts and it is doing the right thing, it 
follows that you are servicing the interrupt more than once.  This 
leads one to suspect the hyperthread stuff.  We need someone who 
understands APIC/ hyperthread issues.

-g
> 
> Kernel reports the same:
> root@setup:~ {503} $ uname -a
> Linux setup 2.4.21-rc6-ac1 #2 SMP Tue Jun 3 09:45:13 CEST 2003 i686 unknown
> 
> root@setup:~ {504} $ cat /proc/cpuinfo
> processor       : 0
> vendor_id       : GenuineIntel
> cpu family      : 15
> model           : 2
> model name      : Intel(R) Xeon(TM) CPU 2.40GHz
> stepping        : 7
> cpu MHz         : 2392.065
> cache size      : 512 KB
> physical id     : 0
> siblings        : 2
> fdiv_bug        : no
> hlt_bug         : no
> f00f_bug        : no
> coma_bug        : no
> fpu             : yes
> fpu_exception   : yes
> cpuid level     : 2
> wp              : yes
> flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm
> bogomips        : 4771.02
> 
> processor       : 1
> vendor_id       : GenuineIntel
> cpu family      : 15
> model           : 2
> model name      : Intel(R) Xeon(TM) CPU 2.40GHz
> stepping        : 7
> cpu MHz         : 2392.065
> cache size      : 512 KB
> physical id     : 0
> siblings        : 2
> fdiv_bug        : no
> hlt_bug         : no
> f00f_bug        : no
> coma_bug        : no
> fpu             : yes
> fpu_exception   : yes
> cpuid level     : 2
> wp              : yes
> flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm
> bogomips        : 4771.02
> 
> root@setup:~ {505} $ ntpdate ntp.xss.co.at; sleep 500; ntpdate ntp.xss.co.at
>  3 Jun 11:58:16 ntpdate[1118]: step time server 194.152.162.17 offset -177.268071 sec
>  3 Jun 12:01:36 ntpdate[1120]: step time server 194.152.162.17 offset -300.106769 sec
> 
> (Sleeping 500 "system seconds" takes 200 "wall clock seconds")
> 
> 
>>You pin this to a particular kernel version.  Do other kernel versions
>>do a better job?
>>
> 
> So far I tried with:
> 
> 2.4.21-rc2-ac2 (ACPI compiled as module)
> 2.4.21-rc4 (ACPI compiled as module)
> 2.4.21-rc6-ac1 (ACPI compiled as module)
> 2.4.21-rc6-ac1 (no ACPI comiled at all)
> 
> Time acceleration is the same for all kernels.
> 
> I'll try it with some older kernels, too.
> 
> - - andreas
> 
> - --
> Andreas Haumer                     | mailto:[e-mail]
> *x Software + Systeme              | http://www.xss.co.at/
> Karmarschgasse 51/2/20             | Tel: [phone]
> A-1100 Vienna, Austria             | Fax: [phone]
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.2.1 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
> 
> iD8DBQE+3HKxxJmyeGcXPhERAkZ/AJ9tjgI6K2HtM/tL15nKDFIKzdcvQgCfYOGY
> i96/o6Nyxlem5yOp1A2Q//s=
> =plFh
> -----END PGP SIGNATURE-----
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [e-mail]
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

-- 
George Anzinger   [e-mail]
High-res-timers:  http://sourceforge.net/projects/high-res-timers/
Preemption patch: http://www.kernel.org/pub/linux/kernel/people/rml