hsr_ping test failed

Hi, Sebastian, Yunshui,

I tried to run hsr_ping recently on x86_64 arch (build via vng --build --config tools/testing/selftests/net/hsr/config, latest net branch, iproute2-6.15.0) and the test always failed due to error "Expect to send and receive 10 packets and no duplicates."

I checked the normal ping test and found it also has duplicates. e.g.

PING 100.64.0.3 (100.64.0.3) 56(84) bytes of data.

--- 100.64.0.3 ping statistics ---
2 packets transmitted, 2 received, +3 duplicates, 0% packet loss, time 1012ms

I'm not sure if I missed some configurations or anything else. How do we make sure there is not duplicates?

Another issue during my testing is that the test checks

        while [ ${WAIT} -gt 0 ]
        do
                grep 00:00:00:00:00:00 /sys/kernel/debug/hsr/hsr*/node_table
                if [ $? -ne 0 ]
                then
                        break
                fi
                sleep 1
                let "WAIT = WAIT - 1"
        done

While in my testing the debug log shows 00:00:00:00:00:00 for mac B address during the whole 5 seconds.

/sys/kernel/debug/hsr/hsr1/node_table:00:11:22:00:02:01 00:00:00:00:00:00  10026a200,  100265ef5,              0,     1
/sys/kernel/debug/hsr/hsr1/node_table:00:11:22:00:02:02 00:00:00:00:00:00  100265ef5,  100269800,              0,     1
/sys/kernel/debug/hsr/hsr1/node_table:00:11:22:00:03:01 00:00:00:00:00:00  100266140,  10026a200,              0,     1
/sys/kernel/debug/hsr/hsr1/node_table:00:11:22:00:03:02 00:00:00:00:00:00  100269000,  100266140,              0,     1
/sys/kernel/debug/hsr/hsr2/node_table:00:11:22:00:01:01 00:00:00:00:00:00  100269000,  100265ef5,              0,     1
/sys/kernel/debug/hsr/hsr2/node_table:00:11:22:00:01:02 00:00:00:00:00:00  100265ef5,  100269000,              0,     1
/sys/kernel/debug/hsr/hsr2/node_table:00:11:22:00:03:02 00:00:00:00:00:00  100266140,  10026a200,              0,     1
/sys/kernel/debug/hsr/hsr2/node_table:00:11:22:00:03:01 00:00:00:00:00:00  10026a200,  100266140,              0,     1
/sys/kernel/debug/hsr/hsr3/node_table:00:11:22:00:01:02 00:00:00:00:00:00  100269000,  100265ef5,              0,     1
/sys/kernel/debug/hsr/hsr3/node_table:00:11:22:00:01:01 00:00:00:00:00:00  100265ef5,  100269000,              0,     1
/sys/kernel/debug/hsr/hsr3/node_table:00:11:22:00:02:02 00:00:00:00:00:00  100265ef5,  10026a200,              0,     1
/sys/kernel/debug/hsr/hsr3/node_table:00:11:22:00:02:01 00:00:00:00:00:00  10026a200,  100265f98,              0,     1

Thanks
Hangbin

Hi,

Something is broken. I traced it down to commit
9cfb5e7f0ded2 ("net: hsr: fix hsr_init_sk() vs network/transport headers.")

haven't put much into it but appears that skb_reset_network_header() is expected to be initialized "late" which leads to the large mac-len.

Sebastian