lockups

[Jason King]

John Baldwin wrote:

>On Tuesday 25 May 2004 12:40 am, Jason King wrote:
>  
>
>>John Baldwin wrote:
>>    
>>
>>>On Sunday 23 May 2004 12:30 am, Jason King wrote:
>>>      
>>>
>>>>John Baldwin wrote:
>>>>        
>>>>
>>>>>On Friday 14 May 2004 08:54 pm, Jason King wrote:
>>>>>          
>>>>>
>>>>>>ok i'm a dumbass... in my haste this morning, i commented out the wrong
>>>>>>line in my config file, so when i tried disabling mixed mode, apic
>>>>>>support wasn't even compiled in the kernel.  So to make extra sure, I
>>>>>>went back and explicitly tried it with the 5.2.1-RELEASE cd (disabling
>>>>>>apic worked, disabling mixed mode did not), so its really something
>>>>>>related to apic support.  So next question, what can I do to track down
>>>>>>the problem?  Any suggestions for resources to understand how the apic
>>>>>>stuff works?  Might there be a way to grab the kernel messages even
>>>>>>after a reboot since the lockups don't even my to drop to the debugger?
>>>>>>Doing a boot -v generated a bunch of text that I couldn't read as it
>>>>>>scrolled by too fast, and I'm wondering if it might possibly help shed
>>>>>>some light as to what's going on...
>>>>>>            
>>>>>>
>>>>>Note that you can't disable mixed mode via a tunable on 5.2.1, so you
>>>>>might want to retest disabling mixed mode.  However, the first steps are
>>>>>to work on the non-ACPI case as that is easiest.  I'll need 'mptable'
>>>>>output and dmesg output from a verbose boot with APIC enabled and ACPI
>>>>>disabled if you can get that.  Thanks.
>>>>>          
>>>>>
>>>>Unfortunately, booting with ACPI disabled and apic enabled still caused
>>>>the system to hang.  So anytime apic is enabled, the system locks up.
>>>>(i've tried just the kernel with no modules loaded as well as with the
>>>>normal modules I use loaded, locks up, but in a different place).  I was
>>>>able to pull off the output from a boot -v using a serial console.  I've
>>>>attached that.  Hopefully that helps...
>>>>        
>>>>
>>>Can you provide mptable output?
>>>      
>>>
>>Here is the mptable output.
>>    
>>
>
>Hmm, your box has lots of issues. :-/  The $PIR table doesn't include entries 
>for several devices on PCI bus 0, and the mptable doesn't include entries for 
>almost all of your PCI devices.  I would first look for a BIOS upgrade.
>
>  
>
Unfortunately, it has the latest BIOS on there (I tried updating it 
before posting my original question to try to eliminate that as a 
possibility).  Any suggestions on how to proceed?? The system does work 
with no issues under XP (possibly a way to gather some info from that?)