Re: pre-2.1.66 Oops

Panu Matilainen (pjm@popcorn.fi)
Wed, 26 Nov 1997 13:40:29 +0200 (EET)

On Tue, 25 Nov 1997, Richard B. Johnson wrote:
>
> > Code: c012dd2f <select_dcache+10b/19c> 89 7c 24 10 movl %edi,0x10(%esp,1)
> > Code: c012dd33 <select_dcache+10f/19c> 8b 7c 24 14 movl 0x14(%esp,1),%edi
> > Code: c012dd3d <select_dcache+119/19c> 39 7c 24 00 cmpl %edi,0x0(%esp,1)
> > Code: c012dd41 <select_dcache+11d/19c> 00 00 addb %al,(%eax)
> > Code: c012dd43 <select_dcache+11f/19c> 00 00 addb %al,(%eax)
> > Code: c012dd45 <select_dcache+121/19c> 00 00 addb %al,(%eax)
> ^^^^^^^^^
> This does not seem to be good code. It looks like
> some zeros got written here.

Thanks for pointing this out. But... it seems that my ksymoops is
somewhat broken, because the original oops has more sensible looking
code-section:

Unable to handle kernel NULL pointer dereference
current->tss.cr3 = 0082a000, ^Hr3 = 0082a000
*pde = 00000000
Oops: 0000
CPU: 0
EIP: 0010:[<c012dd2c>]
EFLAGS: 00010286
eax: 0000644f ebx: c0b8af20 ecx: c0af55e0
esi: ffffffff edi: 00000064 ebp: 00000000
ds: 0018 es: 0018 ss: 0018
Process find (pid: 2828, process nr: 43, stackpa
Stack: 00000001 00000008 c01fde28 00000000 00000
00000008 00000000 00000003 c0b8aca0 c012d
0000679a c012d8bf 00000008 0000679a c0773
Call Trace: [<c012d4a9>] [<c012d8bf>] [<c0139a20
[<c01273a2>] [<c010920a>]
Code: 8b 7e 40 89 7c 24 10 8b 7c 24 14 39 7c 24

Seems a bit more reasonable than the zeros (still doesn't make any sense
to me, though :) Anyway, I checked the other oopses too and it's just
ksymoops that's barfing halfway through the code for some reason or other.
I'll investigate and be more careful next time...

Panu