STABLE kernel panic: privileged instruction fault

[Alexey Tarasov]

On Aug 16, 2010, at 11:39 PM, Kostik Belousov wrote:

> On Mon, Aug 16, 2010 at 11:35:36PM +0400, Alexey Tarasov wrote:
>> 
>> On Aug 16, 2010, at 11:31 PM, Kostik Belousov wrote:
>> 
>>> On Mon, Aug 16, 2010 at 11:21:15PM +0400, Alexey Tarasov wrote:
>>>> Hello Kostik!
>>>> 
>>>> On Aug 16, 2010, at 10:48 PM, Kostik Belousov wrote:
>>>> 
>>>>>> 
>>>>> The backtrace make absolutely no sense. I would not trust kgdb anyway.
>>>>> 
>>>>> Compile ddb in and do backtrace in console on the panic. Also, disassemble
>>>>> the kernel at the fault address. I am very curious which instruction causes
>>>>> this. This is stock GENERIC on the bare metal booted, right ?
>>>> 
>>>> Yes, stock GENERIC.
>>>> 
>>>> Please, check this out:
>>>> 
>>>> Dump of assembler code from 0xffffff0060c0b700 to 0xffffff0060c0b780:
>>> 
>>> Would be nice if you keep all requested data in one place, so that
>>> we do not need to search for the old mails to see the context.
>>> 
>>> According to your previous mail, the fault happen at the
>>> address
>>> instruction pointer     = 0x20:0xffffff8040d2cc83
>>> Your disassembled the stack instead. Please just do
>>> disass 0xffffff8040d2cc83,0xffffff8040d2cca0
>>> in kgdb.
>>> 
>>> But also, I want to see the backtrace and disassembly output from ddb.
>> 
>> (kgdb) disass 0xffffff8040d2cc83,0xffffff8040d2cca0
>> No function contains specified address.
> Err, it seems that old gdb accepts only spaces. Please try
> disass 0xffffff8040d2cc83 0xffffff8040d2cca0 instead.

(kgdb) disass 0xffffff8040d2cc83 0xffffff8040d2cca0
Dump of assembler code from 0xffffff8040d2cc83 to 0xffffff8040d2cca0:
0xffffff8040d2cc83:	(bad)  
0xffffff8040d2cc84:	(bad)  
0xffffff8040d2cc85:	jg     0xffffff8040d2cc87
0xffffff8040d2cc87:	add    %al,(%rax)
0xffffff8040d2cc89:	add    %al,(%rax)
0xffffff8040d2cc8b:	add    %al,(%rax)
0xffffff8040d2cc8d:	add    %al,(%rax)
0xffffff8040d2cc8f:	add    %al,(%rax)
0xffffff8040d2cc91:	add    %al,(%rax)
0xffffff8040d2cc93:	add    %al,(%rax)
0xffffff8040d2cc95:	add    %al,(%rax)
0xffffff8040d2cc97:	add    %al,(%rcx)
0xffffff8040d2cc99:	add    %al,(%rax)
0xffffff8040d2cc9b:	add    %al,(%rax)
0xffffff8040d2cc9d:	add    %al,(%rax)
0xffffff8040d2cc9f:	add    %al,(%rax)
End of assembler dump.



> 
>> 
>> I will build kernel with DDB tomorrow, install it on some servers and wait for the panic occurs.

> Ok. Did you checked for such things as rootkits ?


I am noticing such panics only on this model of supermicro servers for a long! time under FreeBSD.
This servers were tested on huge workload under Linux and there were no problems.

I've installed and run chkrootkit now, there are no rootkits.

--
Alexey Tarasov

(\__/) 
(='.'=) 
E[: | | | | :]З 
(")_(")