Hi, all

I got the following kernel panic on an i386 box running a slightly modified
1.4.2 kernel.  The modification is in the atm_input() of net/if_atmsubr.c.  I
basically check beyond the ATM/LLC header into the IP header.  If it is the
packet I want, make some change to the ATM header (VC number).  Sometimes I 
prepend another IP-IP header in front of the original one.  For some packets,
I also duplicate the packet (with m_copypacket()) and send it to another VC
(IF_ENQUEUE() etc).

This works fine at our own site, but it causes this panic (see DDB trace
below) at another site.  The changes made are just IP address, routing table,
and stop of some services.  

Can someone give me some insight what could cause this messages?  Where should
I look to find the problem?

Thanks

Ruibiao

===<<One panic>>===
There was the followin messages on the screen:

... Some CRC errors (5 of them with different # of cells) ...

kernel: page fault trap, code=0
stopped at _m_freemX0x27:  movswl 0x10(%ebx),%eax

and then gives you the kernel debuger prompt.

A trace gives you the following:

db> tr
  _m_freem(c0be1950) at _m_mfree+0x27
  _ipintr(c0410010,10,......) at _ipintr+0X5ae

  Bad frame pointer: 0XC0413ed8

db> show map

  kernel: page fault trape, code=0
  faulted in DDB, continuing


===<<Another panic>>==
 
_m_freem(c0be1f18) at _m_freem+0x27
_ipintr(c7910010,10,c77f3618,c77f3618,c7919dec) at _ipintr+0x5ae

>kernel: page fault trap, code=0
>Stopped at _m_freem+0x27: movswl ...
>
>A trace in the kerner debuger says:
>
>_m_freem ...
>_ipintr ...
>Bad frame pointer: 0xc7919d78