Re: Problem with a SunFire V120 -- Kernel panics when trying to install
On Mon, Feb 8, 2010 at 5:40 PM, ewe2 <ewetoo@gmail.com> wrote:
> On Tue, Feb 9, 2010 at 8:53 AM, Jurij Smakov <jurij@wooyd.org> wrote:
>> On Fri, Feb 05, 2010 at 09:10:04AM -0500, Yan Morissette wrote:
>>> Every single time I try to install Debian with the sparc netboot image
>>> (Of course, I'm using latest), after one to a few dialogs (usually
>>> right after selecting country, sometimes after locale and sometimes
>>> after DHCP config) the V120 crashes with the following being sent out
>>> the LOM port:
>>>
>>> [ 254.915075] Unable to handle kernel NULL pointer dereference
>>> [ 254.915075] Unable to handle kernel NULL pointer dereference
>>> [ 255.109349] tsk->{mm,active_mm}->context = 0000000000000ad8
>>> [ 255.109349] tsk->{mm,active_mm}->context = 0000000000000ad8
>>> [ 255.302290] tsk->{mm,active_mm}->pgd = fffff8001d838000
>>> [ 255.302290] tsk->{mm,active_mm}->pgd = fffff8001d838000
>>> [ 255.486760] Kernel panic - not syncing: Aiee, killing interrupt handler!
>>> [ 255.486760] Kernel panic - not syncing: Aiee, killing interrupt handler!
>>> [ 255.707837] Press Stop-A (L1-A) to return to the boot prom
>>> [ 255.707837] Press Stop-A (L1-A) to return to the boot prom
>>>
>>> I tried searching the mailing lists and google, the only thing I've
>>> found points to this being about usb-storage, but I have no idea how
>>> to disable this or fix this problem.
>>
>> It's a bit strange that the messages do not include any information on
>> the address where NULL pointer dereference takes place, without it
>> it's pretty hard to tell what happens. If you are able to return to
>> the PROM 'ok' prompt after these messages, you can try to get a stack
>> trace using 'ctrace' and register dump using '.registers' commands. If
>> they are reproducible from one crash to another, then we should be
>> able to track down the location of the crash using it.
>
> Any ideas how to get to that prompt? How would you do that on a pc at keyboard?
>
> --
> Emacs vs. Vi flamewars are a pointless waste of time. Nano is the best
>
This is a completely headless machine, so I don't actually have any
Sun keyboard to hit STOP+A or L1+A. I used minicom's break here, and
here's output.
ok ctrace
PC: f0056fe4
Last leaf: jmpl f00618b8 from 54caf4
0 w %o0-%o7: (420000 96 f0000000 fffff8001fe93cb8 29e1
800000026228f0b8 fffc5681 54c)
Fast Data Access MMU Miss
ok .registers
Normal Alternate MMU Vector
0: 0 0 0 0
1: 0 fffdc6d0 410010 ff3c5c76f7ffc838
2: 0 f0000000 3ffffe00001 794000
3: 7de52e 0 e000000000788036 79beb0
4: fffff8001f038bc0 0 fffff80000790000 400000
5: 18 f e000000000788036 0
6: fffff8001d828000 fffff8001d828000 3ffffe00001 7afbe0
7: 0 3840 4000 b5bd1dd63fd72dbe
%PC f0056fe4 %nPC f0056fe8
%TBA 420000 %CCR 88 XCC:Nzvc ICC:Nzvc
ok
I'm not quite sure how to interpret that "Fast Data Access MMU Miss"
right there. :(
It mostly usually happens when someone messes up and uses set instead
of setenv in OpenBoot, but I have no idea what it's doing here.
Reply to: