Problems enabling NTP on ELDK

[Johan Borkhuis]

Wolfgang Denk wrote:
> In message <46DE9322.7080901 at dutchspace.nl> you wrote:
>   
>> I tried enabling NTP on my embedded machine (using ELDK 4.1, ppc_85xx, 
>> kernel version 2.6.14 with Xenomai 2.3.2). After I did that I noticed 
>> that ksoftirqd/0 suddenly took almost all processor time (> 95%). What 
>> is causing this? Is there a way around this, or is this something that 
>> does not cause any problems?
>>     
>
> You probably have some poroblems in your kernel port - check the  RTC
> driver  and  related  modules like I2C etc. Of course this depends on
> the NTP configuration you use, but the only kernel interaction I  can
> think  of  that  could cause this is if you interact with the RTC. [I
> guess that normal network traffic is working fine on your board.]
>   

All other functions are working fine: no problem with network or other 
PCI devices.

I disabled all RTC and I2C devices, and the problem disappeared. After 
enabling all items one by one the cause of the problem seems to be the 
DS1375 RTC: after I enable this and start NTPD ksoftirqd goes wild 
again. I do not see any other special activity: no extra interrupts in 
/proc/interrupts, and also no other RTC related errors or messages. Also 
OpenPIC does not show any extra interrupts.
Is there a way to find out what interrupt is causing ksoftirqd to be 
triggered?

> BTW: ELDK 4.1 comes with a 2.6.19.2  kernel  -  is  there  a  special
> reason that you use such an obsoletel kernel tree?
>   

I am using an MVME3100 board together with Xenomai. This board has 
support for version 2.6.14-ppc and 2.6.20-ppc, while Xenomai/Adeos is 
supported on 2.6.14 and 2.6.19. I did not yet find time to backport the 
support to 2.6.19.

Kind regards,
    Johan Borkhuis