Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 21 Mar 2010 00:39:01 -0400
From:      jhell <jhell@DataIX.net>
To:        "O. Hartmann" <ohartman@mail.zedat.fu-berlin.de>
Cc:        freebsd-stable@freebsd.org, Matthew Fleming <matthew.fleming@isilon.com>
Subject:   Re: Fatal trap 12: page fault while in kernel mode/current process: 12 (swi2: cambio)
Message-ID:  <alpine.BSF.2.00.1003210033050.3725@pragry.qngnvk.ybpny>
In-Reply-To: <4BA294F7.4030209@mail.zedat.fu-berlin.de>
References:  <4B9E2DFD.2000701@zedat.fu-berlin.de> <06D5F9F6F655AD4C92E28B662F7F853E037DDA15@seaxch09.desktop.isilon.com> <4B9EBCE5.7080303@mail.zedat.fu-berlin.de> <4BA294F7.4030209@mail.zedat.fu-berlin.de>

next in thread | previous in thread | raw e-mail | index | archive | help

On Thu, 18 Mar 2010 17:02, O. Hartmann wrote:
In Message-Id: <4BA294F7.4030209@mail.zedat.fu-berlin.de>

> On 03/16/10 00:04, O. Hartmann wrote:
>> On 03/15/10 18:30, Matthew Fleming wrote:
>>>> Since the last update and make world on Friday, 12th March I get a
>>> crash
>>>> on one of my FreeBSD SMP boxes (it is always the same core message),
>>>> saying something about
>>>> 
>>>> Fatal trap 12: page fault while in kernel mode [...] current process:
>>> 12
>>>> (swi2: cambio)
>>> 
>>> Can you show the stack traceback from the kernel core?
>>> 
>>> We had a problem a while ago at Isilon that I can't tell if it's
>>> related. In our case, the camisr() routine was called after panic(9)
>>> started and before the halt of other processors. This did Bad
>>> Things(TM) since the mtx_lock is a no-op after panicstr is set.
>>> 
>>> We solved it locally by wrapping camisr() in a local cambio_swi()
>>> routine that only called camisr(NULL) when panicstr == NULL.
>>> 
>>> Thanks,
>>> matthew
>> 
>> Hello.
>> 
>> I will do as soon as possible. The box is in production at the moment
>> and I've less time to put everything into debugging to provide more
>> details.
>> 
>> Just in case: does the kernel automatically save the screen with the
>> dump information? If not, I have no other terminal facility to get a
>> dump via the classical way.
>> 
>> Regards,
>> Oliver
>
> Since yesterday, this problem went away! This is mystical. After deactivating 
> radeon.ko and the virtual box stuff I tried again with a new build of world 
> and - voila! - everything worked again. This is strange ...
>
> Oliver
>

If possible set a dump device and use the following in your rc.conf:

crashinfo_enable="YES"
dumpdev="ADD-DEVICE-HERE"

After the crash happens look for core.txt.N files in /var/crash.

You will probably want to look over the crash info and scrub it of 
vital information to comply with your companies policies. It is very 
verbose.

DDB as I have heard can be configured AFAIR to textdump but I have no 
knowledge of that.

Good Luck,

-- 

  jhell




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?alpine.BSF.2.00.1003210033050.3725>