Re: Re: your mail

Re: Re: your mail

[Huo Zhigang]

  Remove the driver first befor reboot! It works. But what is relation between the reboot process and my driver? When I remove the driver module myself, nothing goes wrong. What is the difference?
 
  Thank you. 
  Joining in the lkml is so exciting. :) 
  I love this community.

>> 
>> >INIT: Switching to runlevel: 6
>> >INIT: Send processes the TERM signal
>> >Unable to handle kernel NULL pointer dereference
>>   
>>   What's wrong with my machines?  They are all running linux-2.2.18(SMP-supported) with a kernel module which is a driver of Myricom NIC M3S-PCI64C-2 written by my group.
>>   Thank you in advance 8-)
>
>If you boot the machije without your driver, then reboot does the
>same happen ? If not then it may well be your driver has an error but only
>when it closes down

Re: Re: your mail

[Andreas Dilger]

On Apr 24, 2002  17:44 +0800, Huo Zhigang wrote:
>   Remove the driver first befor reboot! It works. But what is relation
> between the reboot process and my driver? When I remove the driver module
> myself, nothing goes wrong. What is the difference?

Does your module have a timer or thread which may be active at shutdown?
It may be that if it has a kernel thread that the TERM will kill the
thread and your driver does not expect this.

Cheers, Andreas
--
Andreas Dilger
http://www-mddsp.enel.ucalgary.ca/People/adilger/
http://sourceforge.net/projects/ext2resize/

Re: Re: your mail

[Huo Zhigang]

  I boot all the nodes in my cluster without my driver and it is "insmod"ed manually. 
  Now, I will try to "reboot" my machine after the driver is removed. Great. Thank you. 

>> 
>> >INIT: Switching to runlevel: 6
>> >INIT: Send processes the TERM signal
>> >Unable to handle kernel NULL pointer dereference
>>   
>>   What's wrong with my machines?  They are all running linux-2.2.18(SMP-supported) with a kernel module which is a driver of Myricom NIC M3S-PCI64C-2 written by my group.
>>   Thank you in advance 8-)

>Alan:
>If you boot the machije without your driver, then reboot does the
>same happen ? If not then it may well be your driver has an error but only
>when it closes down

Re: Re: your mail

[Huo Zhigang]

  Thank you.
  There is no oops information available. The mechine is just freezed there deadly. In a machine, when this happens, the beep in its box will beep endlessly.

>
>>   Hi, all.
>>   My cluster go wrong these days. So many times when I "/sbin/reboot" a node, the following message will be displayed on the console.
>> 
>> >INIT: Switching to runlevel: 6
>> >INIT: Send processes the TERM signal
>> >Unable to handle kernel NULL pointer dereference
>>   
>>   What's wrong with my machines?  They are all running linux-2.2.18(SMP-supported) with a kernel module which is a driver of Myricom NIC M3S-PCI64C-2 written by my group.
>>   Thank you in advance 8-)
>>   
>>             Zhigang Huo
>>             zghuo@ncic.ac.cn
>
>Have you tried decoding the oops? Have a look at  
>linux/Documentation/oops-tracing.txt
>
>	Zwane
>
>--