6.0 RC1 - problems with HighPoint RocketRAID 1640?

[Neil Hoggarth]

On Mon, 24 Oct 2005, Scot Hetzel wrote:

> On 10/24/05, Neil Hoggarth <neil.hoggarth at physiol.ox.ac.uk> wrote:
>> I tried a source-upgrade from RELENG_5 to RELENG_6 over the weekend
>> (cvsuped RELENG_6 at around 10am Sunday morning, UK time). I was using a
>> GENERIC kernel. The procedure from UPDATING was working fine, up to the
>> point when I ran "make installworld". The system froze solid very early
>> in the installworld process (the comment printed on the screen prior to
>> the freeze suggests that it was copying, or preparing to copy, include
>> files).
>
> Did you reboot after installing the 6.0 Kernel and before the installworld?
> This ensures that the 6.0 Kernel works on your system before the 6.0
> binaries are installed.

Yes. I'd successfully rebooted single-user on the 6.0 kernel, and run 
"mergemaster -p" without incident.

>> Any pointers on how I might go about diagnosing this, and hopefully
>> producing a more useful problem report concerning the difficultes that
>> my hardware seems to have with 6.0 kernels?
>>
> You'll need to ensure that you have the debugging options enabled in
> the 6.0 kernel, this way your kernel will hopefully break to the
> debugger when the problem reoccurs.
>
> See the handbooks chapter on debugging for more details.

I had another go this weekend. I created a new kernel config by adding 
the following to GENERIC:

   options 	KDB
   options 	DDB
   options 	GDB
   options 	INVARIANTS
   options 	INVARIANT_SUPPORT
   options 	WITNESS

On this kernel the "boot single-user / mergemaster -p / make 
installworld / mergemaster -i" sequence worked okay. Unfortunately the 
system wedged on me once again when I rebooted multi-user following the 
install (system got as far as presenting a login prompt, but was 
completely unresponsive to the keyboard or the network there after).

After reset the system would repeatably lock up just after the point
where the SATA disks are probed and geom_mirror picks up my mirroring
setup.

The debugging kernel doesn't panic, break to the debugger, or print any 
additional diagnostics that I can see. CTRL+ALT+ESC does not break to 
the debugger.

I was going to try online debugging using remote GDB (as described in 
section 10.6 of the Developers' Handbook), using my laptop (which I have 
successfully upgraded to RELENG_6) as the debugging machine; however it 
appears that "gdb -k" doesn't work any more? I get:

   gdb: unrecognized option `-k'
   Use `gdb --help' for a complete list of options.

Regards,
-- 
Neil Hoggarth                                Departmental Computing Manager
<neil.hoggarth at physiol.ox.ac.uk>                   Laboratory of Physiology
http://www.physiol.ox.ac.uk/~njh/                  University of Oxford, UK