[svlug] Kernel panic?

[Rick Moen]

Quoting Sanatan Rai (sanatan at gmail.com):

> Would anyone know what is going on?

Either a fatal bug in kernelspace (kernel or a driver) or a serious
hardware glitch.

> How do I troubleshoot this?

Well, let's see.  I've never seriously tried to interpret kernel oops
data before, so I'm trying to learn this on the fly.

The Extended Instruction Pointer (EIP) data is said to be often
fruitful, storing as it does the address of the faulty instruction:

> Dec 15 16:00:03 fractal kernel: [333635.322295] EIP: 0060:[<c10dc258>]
> EFLAGS: 00010286 CPU: 1
> Dec 15 16:00:03 fractal kernel: [333635.322300] EIP is at
> time_out_leases+0x33/0x45

So, whatever was going on, it involved executing a function called
time_out_leases, and that the processor was byte 33 (hex) of a
45-hex-byte-long function, at the time.  I'm pretty sure that the
time_out_leases in question is the one in fs/locks.c, dealing in 'file
leases' (for filesystems' file locking), but you'll want to investigate
locally in the kernel source tree of whatever kernel you're running.

Now, that in all likelihood is simply not enough data to narrow down the
root cause, so you may wish to tweak the logging of data by the kernel
printk function to give you more and better information the _next_ time
this thing happens.  Here's what looks like a decent guide to doing so:
http://madwifi-project.org/wiki/DevDocs/KernelOops

-- 
Rick Moen               "Never diagram a sentence when under the influence of 
rick at linuxmafia.com     alcohol or strong narcotics, even if it IS hilarious."
McQ!  (4x80)                                              -- FakeAPStylebook