[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Kernel panics



On Tue, 2004-09-14 at 07:54, Mark Dadgar wrote:
> On Sep 13, 2004, at 10:51 PM, Arjan van de Ven wrote:
> >> Yah.  While I don't have a lot of kernel panics, I'm wondering what 
> >> the
> >> best crash reporting tools are in RHEL3.  There's netdump, crash, 
> >> etc...
> >
> > get us the oops.. that's the most valuable piece of information.
> > If you don't want to write it down, there is serial console and
> > netconsole that are quite neat for this.
> 
> It's particularly painful to do in a location with only one Linux 
> machine (as both of mine were), but I'll do what I can next time.

if you're really lucky the oops makes it into /var/log/messages ... but
yeah it's a almost-half-the-time thing.

Btw if you write downt he oops by hand we generally don't care THAT much
about all those hex numbers; what we care about most are
* the EIP value and "decoded function name" that that value represents
  (which is on the line after it)
* the functions in the stack backtrace)

eg on the screen it looks like:
Call Trace:   
    [<c01ca4ea>] __handle_sysrq_nolock [kernel] 0x7a (0xf2cb5f38)
    [<c01ca449>] handle_sysrq [kernel] 0x59 (0xf2cb5f58) 
    [<c01944c3>] write_sysrq_trigger [kernel] 0x53 (0xf2cb5f7c)
    [<c01608a7>] sys_write [kernel] 0x97 (0xf2cb5f94)

and what we care about is:
  __handle_sysrq_nolock
  handle_sysrq
  write_sysrq_trigger
  sys_write


Attachment: signature.asc
Description: This is a digitally signed message part


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]