Question

I have a trouble with tcp_fin() function. It should process incoming TCP-segments with FIN flag, but when I add tracepoint or just printk at the begin of this function, this tracepoint handler is never calls (or no any messages from printk).

My actions:

1. Add tracepoint or printk to tcp_fin()
2. Build & boot new kernel

3. Run something like this:

   #!/bin/bash
   
   nflows=50
   
   on_int()
   {
       echo "$nflows skeeped"
       exit 0
   }
   
   trap 'on_int' INT
   
   while [ $nflows -ne 0 ]
   do
       iperf -n 5M -c X.X.X.X
       nflows=$(( $nflows - 1))
       echo "======================"
       echo $nflows
       echo "======================"
   done
   

And as the result I should to observe calls of tcp_fin(), but nothing happens.

Answer 1

As I wrote a few days ago tcp_fin() processed incoming FIN-segment, but in contrast to RFC-793 not each connection ended by two FIN-segments (A->B, B->A). So in some cases we have only outgoing FIN and no more.