Re: debug of a kernel panic leading nowhere...

[Keith Owens]

On Wed, 25 Jul 2001 21:57:18 -0400, 
"Gerald Walden" <[e-mail]> wrote:
>During the rmmod of a fairly simple driver, I get a kernel
>panic which crashes the system.  I believe I have taken all
>the proper steps to attempt to debug the problem to no
>avail.
>perhaps there is a more appropriate list that I should be
>sending this to) would greatly be appreciated.
>before <1>Unable to handle kernel paging request at virtual
>address 5a5a5a5e

That bit pattern indicates an attempt to access storage that has been
freed and poisoned by the slab cache.

>>>EIP; c482424a <__module_parm_watchdog+4fd2/????>   <=====
>Trace; c482777b <END_OF_CODE+8503/????>
>Trace; c482779d <END_OF_CODE+8525/????>
>Trace; c482190f <__module_parm_watchdog+2697/????>
>Trace; c4854c20 <END_OF_CODE+359a8/????>

At the time the oops occurred, the module symbols had already been
removed.  You need to run ksymoops against ksyms and lsmod data from
*before* the rmmod.  man insmod, see ksymoops assistance, in particular
directory /var/log/ksymoops.  Run ksymoops against the symbols saved
after the module was loaded.