Going by the traceback, I believe we are stuck in the seqcount read loop in ip6_tables.c:get_counters(). This could happen if a writer has taken the lock and failed to release it. However, I don't know if we can prove that theory or find out who the writer is from just the traceback. If possible, could you collect a crashdump on a system after it has hung? sosreport output from the system would also be useful.
Also, as mentioned in comment #4, I think something is running the /sbin/ip6tables command when the hang occurs. Can you take a look at your setup to see if you can determine what that might be?
Going by the traceback, I believe we are stuck in the seqcount read loop in ip6_tables. c:get_counters( ). This could happen if a writer has taken the lock and failed to release it. However, I don't know if we can prove that theory or find out who the writer is from just the traceback. If possible, could you collect a crashdump on a system after it has hung? sosreport output from the system would also be useful.
Also, as mentioned in comment #4, I think something is running the /sbin/ip6tables command when the hang occurs. Can you take a look at your setup to see if you can determine what that might be?