Re: Version 2.1.39 SMP

Trevor Johnson (trevor@blues.jpj.net)
Fri, 23 May 1997 18:33:17 +0000 (Local time zone must be set--see zic manual pagtrevor@blues.jpj.net


On Tue, 20 May 1997, Richard B. Johnson wrote:

> I have been trying to break linux version 2.1.39 SMP. The damn thing seems
> to work!
[...]
> Of course there is at least one BUG! Attempt to execute vmlinux, er...
> dismount your disks first. The result will be:
>
> general protection: 0000
> CPU: 0
> EIP: 0010:[<c0109916>]
> EFLAGS: 00010282
> etc..
> Data segment was at 002b and extra segment the same (user-space).
>
>
> I think the appropriate action is supposed to be a core-dump. This is
> a user task that did a bad thing. Instead, the machine halts.

That bug "worked" for me with 3.1.39, non-SMP. With 2.1.40, there's still
the general protection fault^H^H^H^H^Herror but the kernel keeps running
afterward.

general protection: 0000
CPU: 0
EIP: 0010:[<c0109466>]
EFLAGS: 00010282
eax: 00000000 ebx: 080b3008 ecx: 0809faf8 edx: 00000000
esi: 4012b144 edi: 080b3008 ebp: bfff733c esp: c3409fec
ds: 002b es: 002b ss: 0018
Process vmlinux (pid: 140, process nr: 4, stackpage=c3409000)
Stack: c0100000 00000023 00000206 bffffac0 0000002b
Call Trace: [<c0100000>]
Code: cf 89 f6 8d bc 27 00 00 00 00 f7 44 24 30 00 00 02 00 54 75
Segmentation fault

from System.map:

c0100000 A _text
c0100000 T _stext
c0100000 T stext
c0100000 t startup_32
c01000df t isnew

c0109430 T ret_from_sys_call
c0109441 t ret_with_reschedule
c0109470 t signal_return
c01094a0 t v86_signal_return
c01094c0 t tracesys
c01094e6 t badsys
___
Trevor Johnson <trevor@jpj.net>