LX2160ARDB ip forwarding not working

[LeNormich]

Hello all,

I'm trying to use the LX2160ARDB as a router but I can't manage to have ip forwarding working.

I'm using the setup in attached files.

The traffic generator sends frames with the following properties:

- ip src=192.85.1.3

- ip dst = 192.85.2.3

- mac src=00:10:94:00:00:01

- mac dst = 00:10:94:00:00:02

- protocol udp

- size 1400 bytes

I use the following commands on LX2160:

ls-addni -nq=16 -t=8 dpmac.3    //creates eth2
ls-addni -nq=16 -t=8 dpmac.4    //creates eth3
ifconfig eth2 hw ether 00:E0:0C:00:77:01
ifconfig eth2 192.85.1.1/24 up
ifconfig eth3 hw ether 00:E0:0C:00:77:02
ifconfig eth3 192.85.2.1/24 up
arp -s 192.85.1.3 00:10:94:00:00:01
arp -s 192.85.2.3 00:10:94:00:00:02
echo 1 > /proc/sys/net/ipv4/ip_forward
echo 1 > /proc/sys/net/ipv4/conf/all/forwarding​

I see the frames arriving on eth2 but they are never transmitted to eth3 and back to the generator.

using restool dpni, I see that the frames arrive on interface as "ingress_filtered_frames" increases.

Does someone know why is traffic not sent to the other interface as ip forward should do?

Thanks,

[LeNormich]

the following command solved it:

sudo iptables -P FORWARD ACCEPT

[bpe]

If you want your frames to be processed by a host in any way, the frames must
contain the MAC address of that host in L2 destination address field. You set your
eth2 MAC address to 00:E0:0C:00:77:01 (ifconfig eth2 hw ether 00:E0:0C:00:77:01)
and send test frames with mac dst = 00:10:94:00:00:02. It is therefore no surprise that
they are dropped at eth2 by ingress filtering (DPAA2RM, Section 7.3.25).

Best Regards,
Platon

[LeNormich]

Thanks for the answer!

I changed the dst mac address to 00:E0:0C:00:77:01

but now frames are discarded at ingress and not transmitted to the other interface.

Do you know what causes this behavior? Shouldn't the board transmit frames in this configuration?

[bpe]

It is not fully clear how the behaviour of the system changed, if at all.
Frames drop at ingress were the problem described in the initial post in this
tread. Do you observe any changes in the system reaction to the test stream 
after you adjusted the destination MAC, or it is just the same?

Best Regrds
Platon

[LeNormich]

Sorry, i wasn't precise enough: now the "ingress_discarded_frames" counter increases instead of "filtered_frames".

Regards,