Re: [BProc] another 3.2.0 oops

[<er...@he...>]

On Mon, Oct 21, 2002 at 10:06:44AM -0400, Nicholas Henke wrote:
> Hrm -- I seems to be breaking things again. The following oops occurred
> running my 'noop' script again -- the ps script is that script, just
> remove the bpsh $node ps line. I have seen this oops twice so far -- the
> Code: was the same in both. If the trace is bogus, I would _really_
> appreciate any pointers you could give me to get better traces for you. 
> 
> Cheers!
> Nic
> 
> Unable to handle kernel paging request at virtual address 0804a59b
> *pde = 1bde1067
> Oops: 0003
> CPU: 1
> EIP: [<c0116c6c>]   Not tainted
> Using defaults from ksymoops -t elf32-i386 -a i386
> EFLAGS: 00010046
> Process bpslave (pid: 31698, stackpage=d8a45000)
> Stack: 00000001 00000282 00000003 dca85ef0 dc488cc0 00000000 dbd9800
> c010608c
> dca85ef0 dc40bfa0 00000000 e097af39 00000000 00000000 00000000 00000000
> 00000000
> 00000000 00000000 00000000 00000000 00000000 00000000 00000011
> Call Trace: [<c010608c>] [<e097af1d>] [<e096f329>] [<e097a698>]
> [<e096f313>]
> [<e097a698>]
> Code: c7 01 00 00 00 00 8b 41 3c 85 c0 75 2d a1 c0 d1 26 c0 8d 51
> 
> >>EIP; c0116c6c <__wake_up+4c/c0>   <=====
> Trace; c010608c <__up_wakeup+8/c>
> Trace; e097af1d <__module_using_checksums+5893/????>
> Trace; e096f329 <[bproc]bproc_iod_release+3d/89>
> Trace; e097a698 <__module_using_checksums+500e/????>
> Trace; e096f313 <[bproc]bproc_iod_release+27/89>
> Trace; e097a698 <__module_using_checksums+500e/????>

Hrm.  It's hard to give generic pointers.  I usually try to look at
the whole mess and try to figure it out.

This back trace looks reasonable except for the
__module_using_checksums part.  That's weird.  A lot of the time I get
the module binary and look for the code in it to see what function
it's in.  In this case you're in __wake_up so that's not that useful.

As always reproducing it is the most reliable way to go.  I just saw
some more weirdness so I'm looking into it.

- Erik