I have a Kontron ThinkIO with Kontron CE 4.2 BSP (modified CEPC) Industrial
Controller, which has been working successfully for some time. The device
has 2 Ethernet ports, and I've added the gateway components (NAT, ICS, UPNP
+ IGD, firewall, DHCP allocator, etc). the private interface is set up as
192.168.0.1, and on that side is a Rockwell L34 PLC, statically assigned at
192.168.0.2. I've configured port forwarding, so that from the public
interface port 8080 TCP forwards to internal 192.168.0.2:80, also the
EthernetIP port 44818 (TCP and UCP) are forwarded to the internal PLC.
This works, such that when I point my external web browser at port 8080 on
the public interface (DHCP allocated at 192.168.221.58), it brings up the
startup page of the PLC's http interface. However thats all I see, the CE OS
immediately locks up. I don't get ant diagnostic debug messages on the
serial port, and don't know what's happening. The lockup also happens when I
connect a laptop to the hub on the private side. It get DHCP allocation, and
I can ping devices on the public side (NAT translation works), but if I
tracert I get 3 hits on the CE interface (192.168.0.1), then one from the
device on the public side (eg 192.168.221.4) and then CE locks up again.
I suspect something in NAT if forwarding to itself in a tight loop, but I
can't diagnose with the debugger, as the BSP doesn't support KITL over
serial, or VMINI with the two Ethernet interfaces in use, as far as I can
ascertain.
Does anyone have any suggestions here, have I missed something?
One other clue, early on in bootup the CE box actually sends two DHCP
packets (the second an offer) with a source address of 192.168.0.1 out over
the 192.168.221.58 public interface, which should never happen...

Thanks for any assistance,

David

Re: Gateway NAT use locking up OS by John

John
Tue Mar 01 14:59:00 CST 2005

I've asked around and none of my coworkers are fimiliar with this problem.
(The giving out the DHCP packet address on the public interface is a known
issue but isn't causing this).

Some thoughts for further investigation:
* The IP addresses 192.168.0.1 and 192.168.221.58 might be confusing the
routing part if the subnet masks on the two interfaces allow for overlap.
* Run the CETK platform validation tests to rule out BSP/driver issues.


--
John Spaith
Software Design Engineer, Windows CE
Microsoft Corporation

Check out the new CE Networking Team Blog at http://blogs.msdn.com/cenet/.

This posting is provided "AS IS" with no warranties, and confers no rights.
You assume all risk for your use. © 2003 Microsoft Corporation. All rights
reserved.

"David Varley" <David.Varley@cborn.com> wrote in message
news:OY08HfeHFHA.400@TK2MSFTNGP14.phx.gbl...
>I have a Kontron ThinkIO with Kontron CE 4.2 BSP (modified CEPC) Industrial
> Controller, which has been working successfully for some time. The device
> has 2 Ethernet ports, and I've added the gateway components (NAT, ICS,
> UPNP
> + IGD, firewall, DHCP allocator, etc). the private interface is set up as
> 192.168.0.1, and on that side is a Rockwell L34 PLC, statically assigned
> at
> 192.168.0.2. I've configured port forwarding, so that from the public
> interface port 8080 TCP forwards to internal 192.168.0.2:80, also the
> EthernetIP port 44818 (TCP and UCP) are forwarded to the internal PLC.
> This works, such that when I point my external web browser at port 8080 on
> the public interface (DHCP allocated at 192.168.221.58), it brings up the
> startup page of the PLC's http interface. However thats all I see, the CE
> OS
> immediately locks up. I don't get ant diagnostic debug messages on the
> serial port, and don't know what's happening. The lockup also happens when
> I
> connect a laptop to the hub on the private side. It get DHCP allocation,
> and
> I can ping devices on the public side (NAT translation works), but if I
> tracert I get 3 hits on the CE interface (192.168.0.1), then one from the
> device on the public side (eg 192.168.221.4) and then CE locks up again.
> I suspect something in NAT if forwarding to itself in a tight loop, but I
> can't diagnose with the debugger, as the BSP doesn't support KITL over
> serial, or VMINI with the two Ethernet interfaces in use, as far as I can
> ascertain.
> Does anyone have any suggestions here, have I missed something?
> One other clue, early on in bootup the CE box actually sends two DHCP
> packets (the second an offer) with a source address of 192.168.0.1 out
> over
> the 192.168.221.58 public interface, which should never happen...
>
> Thanks for any assistance,
>
> David
>
>



Re: Gateway NAT use locking up OS by David

David
Wed Mar 02 05:57:29 CST 2005

Hi John,

Thanks for your response.

The subnet masks are both 255.255.255.0, so that part should be ok.

I tried testing with CETK as you suggested, not a lot of luck so far.

The One-Card Network Card Mini-Port Driver test fails, but with a 0-sized
log file, so I can't get any info on what is wrong. I can't use the 2-Card
test because I only have the 2 ports.

The driver is the Intel E100CE for 82551, and in normal operation it seems
fine, I am driving a PLC directly from it, and have had continuous operation
over one port for weeks without fault, although that doesn't rule out some
conflict when using both at once, I guess. But they both seem to work
together ok until I forward something between them.

I don't know much about IP6 and UPNP, so I haven't done anything to
configure them, just in as they came across from the catalog.

This is rather frustrating, something that seems so simple has held the
project up for over a week. I've asked for support from Kontron (Germany)
who provide the hardware and BSP, but nothing has come back so far. We did
wast a month trying to track down software problems with the first 3 units
they supplied, which turned out to be a hardware fault in that batch, but in
this case it feels more like software...

Any more advice much appreciated.

David


"John Spaith [MS]" <jspaith@ONLINE.microsoft.com> wrote in message
news:ODdjCGqHFHA.2276@TK2MSFTNGP15.phx.gbl...
> I've asked around and none of my coworkers are fimiliar with this problem.
> (The giving out the DHCP packet address on the public interface is a known
> issue but isn't causing this).
>
> Some thoughts for further investigation:
> * The IP addresses 192.168.0.1 and 192.168.221.58 might be confusing the
> routing part if the subnet masks on the two interfaces allow for overlap.
> * Run the CETK platform validation tests to rule out BSP/driver issues.
>
>
> --
> John Spaith
> Software Design Engineer, Windows CE
> Microsoft Corporation
>
> Check out the new CE Networking Team Blog at http://blogs.msdn.com/cenet/.
>
> This posting is provided "AS IS" with no warranties, and confers no
rights.
> You assume all risk for your use. © 2003 Microsoft Corporation. All rights
> reserved.
>
> "David Varley" <David.Varley@cborn.com> wrote in message
> news:OY08HfeHFHA.400@TK2MSFTNGP14.phx.gbl...
> >I have a Kontron ThinkIO with Kontron CE 4.2 BSP (modified CEPC)
Industrial
> > Controller, which has been working successfully for some time. The
device
> > has 2 Ethernet ports, and I've added the gateway components (NAT, ICS,
> > UPNP
> > + IGD, firewall, DHCP allocator, etc). the private interface is set up
as
> > 192.168.0.1, and on that side is a Rockwell L34 PLC, statically assigned
> > at
> > 192.168.0.2. I've configured port forwarding, so that from the public
> > interface port 8080 TCP forwards to internal 192.168.0.2:80, also the
> > EthernetIP port 44818 (TCP and UCP) are forwarded to the internal PLC.
> > This works, such that when I point my external web browser at port 8080
on
> > the public interface (DHCP allocated at 192.168.221.58), it brings up
the
> > startup page of the PLC's http interface. However thats all I see, the
CE
> > OS
> > immediately locks up. I don't get ant diagnostic debug messages on the
> > serial port, and don't know what's happening. The lockup also happens
when
> > I
> > connect a laptop to the hub on the private side. It get DHCP allocation,
> > and
> > I can ping devices on the public side (NAT translation works), but if I
> > tracert I get 3 hits on the CE interface (192.168.0.1), then one from
the
> > device on the public side (eg 192.168.221.4) and then CE locks up again.
> > I suspect something in NAT if forwarding to itself in a tight loop, but
I
> > can't diagnose with the debugger, as the BSP doesn't support KITL over
> > serial, or VMINI with the two Ethernet interfaces in use, as far as I
can
> > ascertain.
> > Does anyone have any suggestions here, have I missed something?
> > One other clue, early on in bootup the CE box actually sends two DHCP
> > packets (the second an offer) with a source address of 192.168.0.1 out
> > over
> > the 192.168.221.58 public interface, which should never happen...
> >
> > Thanks for any assistance,
> >
> > David
> >
> >
>
>



Re: Gateway NAT use locking up OS by John

John
Fri Mar 04 11:13:49 CST 2005

Sorry to hear about all the problems you're having on this. We're still not
sure what's going on at our end. Some more debugging thoughts

* Verify there's no way you can get KITL on this connection. Maybe not
possible, but if there's even a chance of it it'll make debugging a lot
easier.
* Can you try setting up a CEPC or other MS provided BSP? This could help
verify whether it was an MS or hardware/BSP problem.

--
John Spaith
Software Design Engineer, Windows CE
Microsoft Corporation

Check out the new CE Networking Team Blog at http://blogs.msdn.com/cenet/.

This posting is provided "AS IS" with no warranties, and confers no rights.
You assume all risk for your use. © 2003 Microsoft Corporation. All rights
reserved.

"David Varley" <David.Varley@cborn.com> wrote in message
news:u75HH8xHFHA.156@TK2MSFTNGP10.phx.gbl...
> Hi John,
>
> Thanks for your response.
>
> The subnet masks are both 255.255.255.0, so that part should be ok.
>
> I tried testing with CETK as you suggested, not a lot of luck so far.
>
> The One-Card Network Card Mini-Port Driver test fails, but with a 0-sized
> log file, so I can't get any info on what is wrong. I can't use the 2-Card
> test because I only have the 2 ports.
>
> The driver is the Intel E100CE for 82551, and in normal operation it seems
> fine, I am driving a PLC directly from it, and have had continuous
> operation
> over one port for weeks without fault, although that doesn't rule out some
> conflict when using both at once, I guess. But they both seem to work
> together ok until I forward something between them.
>
> I don't know much about IP6 and UPNP, so I haven't done anything to
> configure them, just in as they came across from the catalog.
>
> This is rather frustrating, something that seems so simple has held the
> project up for over a week. I've asked for support from Kontron (Germany)
> who provide the hardware and BSP, but nothing has come back so far. We did
> wast a month trying to track down software problems with the first 3 units
> they supplied, which turned out to be a hardware fault in that batch, but
> in
> this case it feels more like software...
>
> Any more advice much appreciated.
>
> David
>
>
> "John Spaith [MS]" <jspaith@ONLINE.microsoft.com> wrote in message
> news:ODdjCGqHFHA.2276@TK2MSFTNGP15.phx.gbl...
>> I've asked around and none of my coworkers are fimiliar with this
>> problem.
>> (The giving out the DHCP packet address on the public interface is a
>> known
>> issue but isn't causing this).
>>
>> Some thoughts for further investigation:
>> * The IP addresses 192.168.0.1 and 192.168.221.58 might be confusing the
>> routing part if the subnet masks on the two interfaces allow for overlap.
>> * Run the CETK platform validation tests to rule out BSP/driver issues.
>>
>>
>> --
>> John Spaith
>> Software Design Engineer, Windows CE
>> Microsoft Corporation
>>
>> Check out the new CE Networking Team Blog at
>> http://blogs.msdn.com/cenet/.
>>
>> This posting is provided "AS IS" with no warranties, and confers no
> rights.
>> You assume all risk for your use. © 2003 Microsoft Corporation. All
>> rights
>> reserved.
>>
>> "David Varley" <David.Varley@cborn.com> wrote in message
>> news:OY08HfeHFHA.400@TK2MSFTNGP14.phx.gbl...
>> >I have a Kontron ThinkIO with Kontron CE 4.2 BSP (modified CEPC)
> Industrial
>> > Controller, which has been working successfully for some time. The
> device
>> > has 2 Ethernet ports, and I've added the gateway components (NAT, ICS,
>> > UPNP
>> > + IGD, firewall, DHCP allocator, etc). the private interface is set up
> as
>> > 192.168.0.1, and on that side is a Rockwell L34 PLC, statically
>> > assigned
>> > at
>> > 192.168.0.2. I've configured port forwarding, so that from the public
>> > interface port 8080 TCP forwards to internal 192.168.0.2:80, also the
>> > EthernetIP port 44818 (TCP and UCP) are forwarded to the internal PLC.
>> > This works, such that when I point my external web browser at port 8080
> on
>> >