Hello,
I am developing a device driver for the AxiomTek AX5621H data
acquisition card, and I am encountering some problems on a particular
machine. This driver works pretty fine on normal machines, but crashes
on an Industrial PC with intel 830 2-piece board (with the main board
going into a PCI and an ISA slot on the expansion board(which houses
all the PCI, ISA and AGP slots) at the same time. The error I get is
given below :
CPU 0 : Machine Check Exception : 0000000000000004
Bank 0 : a200000084010400
Kernel panic : CPU context corrupt
In interrupt handler - not syncing
The DMA channel I use is DMA1, and i check for free interrupts and then
allocate them accordingly. So there is no conflicts, I assume. Also there
are no conflicts on the address range I have allocated (It is for now
0x300). The card supports only DMA channels 1 and 3. I have tried both, to
the same result. If anyone among you have had experience with such a
problem, any help in fixing this matter would be of great help.
Thank You.
Harish Harshan.
-----------------------------------------
This email was sent using Amrita Mail.
"Amrita Vishwa Vidyapeetham [Deemed University] - Amritapuri Campus"
http://amritapuri.amrita.edu
On Wed, Apr 20, 2005 at 12:36:45PM +0530, Harish K Harshan wrote:
> CPU 0 : Machine Check Exception : 0000000000000004
> Bank 0 : a200000084010400
> Kernel panic : CPU context corrupt
> In interrupt handler - not syncing
CPU got messed up... could be a bad CPU/cache/chipset or simply it's
over heating or has a bad powersupply.
Hello Chris,
But the system works pretty fine when other applications are running.
Oncei load the driver, the system gets messed up. Could it be the
problem with the way I handle DMA and interrupts?? I mean, is it
possible to mess up everything by wrong programming??? This driver
works prefectly on the other IPCs we have, but not on the two-piece
board (Chino-Laxons) systems we have. The DMA channels are both free
before loading the drivers all the time, and one it is loaded, the
/proc/dma file shows the DMA has been hooked properly. Could it still
be the problem with the CPU/Cache/Chipset as you said? If yes, then why
is it throwing up the error only when I load the driver? Or is it
really the problem with the driver programming? Please let me know as
soon as possible.
Thank You in advance,
Harish K Harshan.
On Wed, April 20, 2005 12:27 pm, Chris Wedgwood said:
> On Wed, Apr 20, 2005 at 12:36:45PM +0530, Harish K Harshan wrote:
>
>> CPU 0 : Machine Check Exception : 0000000000000004
>> Bank 0 : a200000084010400
>> Kernel panic : CPU context corrupt
>> In interrupt handler - not syncing
>
> CPU got messed up... could be a bad CPU/cache/chipset or simply it's
> over heating or has a bad powersupply.
>
-----------------------------------------
This email was sent using Amrita Mail.
"Amrita Vishwa Vidyapeetham [Deemed University] - Amritapuri Campus"
http://amritapuri.amrita.edu
On Fri, Apr 22, 2005 at 05:39:02PM +0530, Harish K Harshan wrote:
> But the system works pretty fine when other applications are
> running.
so either the driver is poking something that is causing problems or
maybe the card when operating makes the system unstabel
> Oncei load the driver, the system gets messed up. Could it be the
> problem with the way I handle DMA and interrupts??
maybe, but i would wouldn't think so
> I mean, is it possible to mess up everything by wrong programming???
yes
> This driver works prefectly on the other IPCs we have, but not on
> the two-piece board (Chino-Laxons) systems we have.
if these are embedded system they might have some weird wiring or
similar (embedded PCs used to have lots of strangeness to them, i
assume that hasn't changed)
> The DMA channels are both free before loading the drivers all the
> time, and one it is loaded, the /proc/dma file shows the DMA has
> been hooked properly. Could it still be the problem with the
> CPU/Cache/Chipset as you said?
i would ask the mainboard maker and also make sure you have the most
recent bios
also make sure your driver isn't stomping on bad memory in the slight
chance it whacking chipset registers I guess