Crash Diagnosis

Ruler2112 · 01-18-2005, 03:55 PM

I have a single-processor AMD Athlon PC running Slackware 9. I have this machine installed as a video security system using the ZoneMinder security software (great stuff). It has 2 Promise IDE controllers, 2 Spectra-8 (bt878) video capture cards, an 80 gig boot drive, 4 250-gig drives for data in a RAID-0 configuration, DVD+-RW for backup if an incident occurs, and a gig of RAM. This machine is running 24/7, recording video from 16 cameras to the terrabyte RAID array and recycling disk space as needed. I had to re-compile the kernel to use bit-banging for the capture cards and removed a bunch of unused features (mostly esoteric features, some of which I've never even heard of) while doing so. I had one crash (a kernel panic) shortly after installing the system before it was even fully functional, so I assumed that I hadn't set something up correctly. Rebooting the system apparently fixed the problem, so I attributed the problem to a quirk. However, I recently had another kernel panic; I wrote down details about this one in the hope that somebody can assist me in troubleshooting whatever problem is causing the crashes. The system was used for several weeks between the two panics. I'm not certain if I've mis-configured something, if there's a software bug somewhere that builds until the system dies, or if there is a problem with a hardware component. (I'm leaning towards one of the first two, especially the first one; I wouldn't think that a hardware problem wouldn't be so intermittent and a software bug would most likely have documentation somewhere because everything I use is used by many people.)

I've included everything I wrote down below. If anybody has any ideas, or could even point me somewhere that I could learn about diagnosing crashes of this sort, I would be appreciative. I don't even have a clue where to start - there's nothing in /var/log that gave me anything but the approx crash time. I've tried googling for different strings that appeared in the crash information as shown below, but haven't found anything substantiative even remotely pertinent to my own situation.

(Anything in this block enclosed in {} are comments from me, everything else is verbatim from the screen displayed when I found it dead. I'm guessing that the high ascii characters in the process name of the first item are bad and indicative of a problem somewhere, which most likely caused the successive messages...)

Code:

Code: ff 0e 94 c0 84 c0 0f 85 8c 00 00 00 89 3c 24 e8 46 5f 00
<1> Unable to handle kernel NULL pointer dereference at virtual address 0000000d
printing eip:
c01216ca
*Pde = 00000000
Oops: 0000
CPU: 0
EIP: 0010:[<c01216ca>] Not tainted
{snip several lines of hex codes}
Process {there were a few high ASCII chars here} (pid 12535, stackpage=f107d000)
{snip about two pages of hex codes}
Code: 8b 1a 89 54 24 04 89 04 24 e8 78 ec 00 00 ff 0d a0 0b 38 c0
<1>Unable to handle kernel NULL pointer dereference at virtual address 00000004
printing eip:
c0130a6c
*pde=00000000
Oops: 0002
CPU: 0
EIP: 0010:[<c0130a6c>] Not tainted
{snip several lines of hex codes}
Process (pid: 0, stackpage=f1067000)
{snip about two pages of hex codes}
Code:89 48 04 89 01 89 51 04 89 4b 08 eb ce 8b 01 8b 51 04 89 02
<1>Unable to handle kernel paging request at virtual address a24cc58f
printing eip:
c011487b
*pde=00000000
Oops: 0002
CPU: 0
EIP: 0010[<c011487b>] Not tainted
{snip several lines of hex codes}
Process  (pid: 0, stackpage=f1067000)
{snip about two pages of hex codes}
Code: ff 00 0f 88 6c 04 00 00 89 1c 24 89 6c 24 04 e8 f1 40 01 00
<0>Kernel panic: Attempted to kill the idle task!
In idle task - not syncing
<1>Unable to handle kernel paging request at virtual address 07630762
printing eip:
c0120e20
Oops: 0000
EIP: 0010:[<c0120e20>] Not tainted
{snip several lines of hex codes}
Process (pid: 0, stackframe=f1067000)
{snip about two pages of hex codes}
Code: 8b 02 85 c0 74 06 8b 02 74 06 89 c1 00 08 00 00 8b 52
<0>Kernel panic: Aiee, killing the interrupt handler!
In Interrupt handler - not syncing

Ruler2112 · 01-20-2005, 06:04 PM

Correction - I'm running Slackware 10.0. (I got too many computers...

)

Ruler2112 · 01-25-2005, 12:19 PM

Anyone... anyone... anyone have any ideas?

reddyl · 01-25-2005, 12:35 PM

check if there is a core dump file called "core" in the filesystem and user the command #file /path/to/core to check which process has crashed the system.

Ruler2112 · 01-25-2005, 01:17 PM

When I execute that command, I get the type of file the core is - ELF 32-bit LSB core file Intel blah blah blah...

The core file is just over 952 meg and is binary - how do I go about parsing the useful information out of it? (I've tried strings on the file, but haven't been able to get anything substantiative out of the display.)

Thanks for the reply reddyl.