reboot after panic

[Stephen Clark]

Jeremy Chadwick wrote:
> On Tue, May 06, 2008 at 09:47:59AM -0400, Stephen Clark wrote:
>> Jeremy Chadwick wrote:
>>> On Fri, May 02, 2008 at 09:40:20AM -0400, Stephen Clark wrote:
>>>> Mine is a nvidia 6300 mb with a dual core amd processor. I am causing the panic
>>>> while trying to develope a DD for a EVDO usb modem - so it is not a great 
>>>> problem - I was just surprised it wasn't rebooting. This is a 6.1 system.
>>>>
>>>> Yes it is sort of discouraging that it is hard to get answers when you 
>>>> aren't running the latest and greatest kernel. In our case we have over 
>>>> 500 units in
>>>> the field running a mix of 4.9 and 6.1 and it is not feasible to 
>>>> continually upgrade them, especially since there is no documented way to 
>>>> reliably upgrade
>>>> a remote installation.
>>> Does the system reboot OK if you issue the "reboot" command?
>>>
>>> If not, then the problem is likely with the reboot method being used
>>> (ACPI vs. non-ACPI) or ACPI tweakage prior to reboot, and not anything
>>> to do with panics.  See the following two sysctls:
>>>
>>> hw.acpi.disable_on_reboot
>>> hw.acpi.handle_reboot
>> It reboots fine when I "shutdown -r now". It is only after a panic
>> that it hangs. I have it set to save the crash dump:
>> dumpdev="AUTO"          # Device to crashdump to (device name, AUTO, or NO).
>> dumpdir="/var/crash"    # Directory where crash dumps are to be stored
>>
>> but there is never one. It is like it hangs trying to dump the memory image.
>>
>> This mother board has both sata and pata controllers but I am using only pata
>> drives.
> 
> A kernel panic causes the kernel to dump all memory contents (from start
> to end) to whatever swap device is available.  It's written to the disk
> in a fairly "raw" format, with some header data of some sort I think.
> After it's done, the system should reboot.
> 
> My guess is that you either don't have any swap defined, swap is defined
> incorrectly (disklabel -r output would be useful), or your swap space is
> smaller than your total amount of memory.  (Swap should usually be 2x
> RAM).
> 
> dumpdir and dumpdev are used during the startup process, where
> savecore(8) is called.  The memory dump on the swap device is extracted
> and stored in a file in $dumpdir, which you can examine later.  Keep in
> mind that savecore(8) will use /dev/dumpdev, which is a symlink to
> whatever device your swap lives on -- and that's determined by reading
> /etc/fstab.
> 
> Does this help?  :-)
> 
Hi Jeremy,

Thanks for the response but I think I have everything set up OK.

from top:
Mem: 33M Active, 19M Inact, 56M Wired, 54M Buf, 762M Free
Swap: 2048M Total, 2048M Free

$ sudo disklabel /dev/ad0s1
# /dev/ad0s1:
8 partitions:
#        size   offset    fstype   [fsize bsize bps/cpg]
   a:   204800        0    4.2BSD     1024  8192    23
   b:  4194304   204800      swap
   c: 78156162        0    unused        0     0         # "raw" part, don't edit
   d: 45879682  4399104    4.2BSD     2048 16384    89
   e:   409600 50278786    4.2BSD     2048 16384    97
   f:  2097152 50688386    4.2BSD     2048 16384    89
   g: 12685312 52785538    4.2BSD     2048 16384    89
   h: 12685312 65470850    4.2BSD     2048 16384    89
J301002:~

1 gig of memory
$ sysctl  -a |grep physmem
hw.physmem: 929439744

$ ls -al /dev/dumpdev
lrwxr-xr-x  1 root  wheel  11 May  6 05:39 /dev/dumpdev -> /dev/ad0s1b

$ less /etc/fstab
# Device                Mountpoint      FStype  Options         Dump    Pass#
/dev/ad0s1b             none            swap    sw              0       0

Any other ideas?

Regards,
Steve
-- 

"They that give up essential liberty to obtain temporary safety,
deserve neither liberty nor safety."  (Ben Franklin)

"The course of history shows that as a government grows, liberty
decreases."  (Thomas Jefferson)