panic: sorele

[Evren Yurtesen]

John Baldwin wrote:
> On Tuesday 28 September 2004 06:57 am, Evren Yurtesen wrote:
> 
>>John Baldwin wrote:
>>
>>>On Monday 27 September 2004 09:34 pm, Evren Yurtesen wrote:
>>>
>>>>Robert Watson wrote:
>>>>
>>>>>On Thu, 23 Sep 2004, Evren Yurtesen wrote:
>>>>>
>>>>>>Does anybody know what is panic: sorele? :)
>>>>>
>>>>>It's the result of an inconsistency detected by the kernel regarding
>>>>>kernel invariants for sockets.  The somewhat uninformative nature of the
>>>>>message is historical, and should probably be fixed.  If this is
>>>>>reproduceable, could you rebuild your kernel with INVARIANTS support, as
>>>>>well as DDB so you can generate a stack trace and other debugging
>>>>>information (or, if you have a core, could you use gdb to extract
>>>>>debugging information per the Handbook?)
>>>>
>>>>I am not able to compile the kernel with DDB and INVARIANTS either...
>>>>When I put these options to kernel conf, I get this error
>>>>
>>>>Options Added:
>>>>
>>>>options         INVARIANTS
>>>>options         KDB
>>>>options         DDB
>>>
>>>INVARIANTS requires the option 'INVARIANT_SUPPORT'.
>>
>>Good point, one sometimes wonders why options INVARIANTS doesnt
>>automatically include options INVARIANT_SUPPORT :)
> 
> 
> So you can build a kernel with INVARIANT_SUPPORT but no INVARIANTS but still 
> build modules with INVARIANTS.  Useful when working on a device driver for 
> example.  I think the dependency there is noted in sys/conf/NOTES btw.

Yes, probably, but it was just a suggestion, because DDB said that it 
needs KDB to compile :) but INVARIANTS just stopped compiling with an 
error so I couldnt guess what was the problem. The error could be more 
informative maybe?

Just a cosmetic thing though. Nothing vital. But I guess there is a real 
problem about this panic: sorele when the kernel is compiled with SMP 
support. If this happens again and I cant get a kernel dump or 
something, I will compile my kernel without SMP... since I dont even 
have multiple processors, I was just testing hyperthreading...

Evren