Re: unexpected IO-APIC, 2.4.36 SMP, AMD dual core Opteron

Bill Davidsen <[email protected]> Thu, 07 Feb 2008 15:52:25 -0500
Newsgroups org.kernel.vger.linux-smp
Message-ID <[email protected]>
Tuynman, William (JSC-DT/USA)[USA] wrote:
> Thanks for the response. 
>
> For now, we are stuck with running a 2.4 kernel. We have applications
> that will not run under a 2.6 kernel with a current Red Hat software
> release. Yes we did try but figured it was easier and safer to run 2.4
> on the new hardware. 
>
> We do plan to upgrade but it's going to take a year or 2 because of the
> amount of code and the need to purchase new versions of COTS products,
> testing and all that. In the meantime, these new machines need to run
> the same applications as the other machines that we have.
>
>   
If you have the luxury of a serious test, I believe that CPU has the 
hardware support needed to run a 2.4 virtual machine under xen or kvm, 
and you might be able to postpone the upgrade of applications somewhat 
longer.

> I tried 2.4.36 because it was just released and improvements have been
> incorporated from time to time.
>
>   
Clearly worth trying.

> NOTE: The kernel appears to run correctly in spite of the IO-APIC
> message. So, it appears that the message is more of a warning than an
> error? Possibly there is some degradation that I'm not aware of but the
> system doesn't need to do a lot of IO.
>
>   
I believe you're right, but there are some boot options to control 
lapic, apic, io-apic, apci, etc. Some selective boot options might get 
rid of the warnings. I can't remember just what were 2.4, I think a lot 
of the ones I'm remembering ar 2.6, but there still may be some joy in 
trying what you can.

> The IO-APIC message does not occur with a 2.6 kernel. I have one of the
> machines running with 2.6.22.13 So, it  appears to be 2.4 related. I
> assume that's why you suggested the "live CD"
>
>   
Yes, to allow testing without any chance of messing up your existing 
setup. It's a safe way to find out how it's going to work.

> For this build I specified the Athlon processor, which is the same as I
> specified for the 2.6 kernel. I'll try  the Opteron and possibly a more
> generic type to see what happens.
>
>   
Let me know how you make out, bet you guys are busy right now. ;-)

> Bill
>  
>
>
>    
> -----Original Message-----
> From: Bill Davidsen [mailto:[email protected]] 
> Sent: Wednesday, February 06, 2008 6:07 PM
> To: Tuynman, William (JSC-DT/USA)[USA]
> Cc: [email protected]
> Subject: Re: unexpected IO-APIC, 2.4.36 SMP, AMD dual core Opteron
>
> Tuynman, William (JSC-DT/USA)[USA] wrote:
>   
>> Hello,
>> 	I have been getting this "An unexpected IO-APIC was found"
>>     
> message. In checking, it occurs on 2.4.34 also.
>   
>> 	The kernel was compiled for 32 bit execution and is running on
>>     
> an AMD dual core Opteron. The machine is an HP Proliant, DL 585.
>   
>> I'm attaching, as text, a portion of the messages file which contains
>>     
> all of the boot information.
>   
>> I would be happy to assist with providing additional information that
>>     
> would be helpful to you.
>   
>>   
>>     
> The first question I have is, why are you running a 2.4 kernel on a
> modern CPU? I suspect the SMP logic in 2.4.26 is a decade older than the
> hardware. SMP was pretty unusual then, multi-core was just not an end
> user feature. I don't doubt that the IO-APIC may have unsupported
> features.
>
> Might I suggest that you boot off a recent "live CD" like Fedora or
> Puppy Linux, just to see if you still see a problem. If not, you will
> have more information to evaluate your next step.
>
>
> --
> Bill Davidsen <[email protected]>
>   "Woe unto the statesman who makes war without a reason that will still
>   be valid when the war is over..." Otto von Bismark 
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-smp" in
> the body of a message to [email protected]
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
>   


-- 
Bill Davidsen <[email protected]>
  "Woe unto the statesman who makes war without a reason that will still
  be valid when the war is over..." Otto von Bismark