Re: savecore writes files after every reboot

[Julio Merino <jmmv%NetBSD.org@localhost>]

On 6/13/11 12:18 PM, Fredrik Pettai wrote:
> Hi,
> 
> During reboots (shutdown -r now) on a particular machine, I've started to see 
> this during boot:
> 
> Jun 13 10:06:39 netbsd-vm savecore: warning: (null) version mismatch:        
> NetBSD 5.1 (GENERIC) #0: Sun Nov  7 14:39:56 UTC 2010   
> builds%b6.netbsd.org@localhost:/home/builds/ab/netbsd-5-1-RELEASE/i386/201011061943Z-obj/home/builds/ab/netbsd-5-1-RELEASE
> /src/sys/arch/i386/compile/GENERIC  and 5DsP4^F^Yg6@'>7^BM    
> 2M[^L3^O1;1I^Ob0^KeU;W"h:^UH_8Sv^F9^Q^\1<^ 4=^\`^C?Z^Z>^X4m
> Jun 13 10:06:39 netbsd-vm savecore: reboot after panic: panic: trap
> Jun 13 10:06:39 netbsd-vm savecore: writing compressed core to 
> /var/crash/netbsd.1.core.gz
> Jun 13 10:06:42 netbsd-vm savecore: writing compressed kernel to 
> /var/crash/netbsd.1.gz
> Jun 13 10:06:42 netbsd-vm savecore: (null): Bad address
> 
> During reboots, I don't notice any panic on the console. 
> What could be triggering savecore to write a crash file? What (files?) does 
> it compare?

I think it looks for some magic number on the swap partition.  It seems
like savecore is crashing right after writing the crash dump to
/var/crash/ and thus it might be failing to do any post-dump cleanup.
(This is just my understanding, but it could be completely wrong.)

Could you file a PR with this information?  Also, if savecore has
actually crashed, maybe you can find a .core file in the root directory.

Another thing to try would be to boot into single user mode and run
savecore by hand so that you can get a stack trace out of it, if it is
actually crashing.  That'd probably be the most useful for debugging
purposes.

-- 
Julio Merino / @jmmv