[lwip-devel] [bug #51687] tcp_close(): is it OK to return ERR_MEM if TF_CLOSEPEND is set?

[Simon Goldschmidt]

Follow-up Comment #1, bug #51687 (project lwip):

> is it OK to return ERR_MEM if TF_CLOSEPEND is set?

Actually, it's not. Up to and including 2.0.1, tcp_close() returning != ERR_OK
means the pcb is *not* gone. Applications like the httpd retry closing in
their 'poll' function and are not prepared for the pcb to vanish by tcp_tmr()
closing it.

This is a change I have introduced after 2.0.1, so it's OK to change it
again.

We can either make tcp_close() return ERR_OK in this case (should be OK since
the pcb is not freed anyway from tcp_close(), normally) or let the user app
call another function to achieve dalayed closing.
Something like:

if(tcp_close(pcb) != ERR_OK) {
  tcp_close_delayed(pcb);
}


or

tcp_close_force(pcb);


    _______________________________________________________

Reply to this item at:

  <http://savannah.nongnu.org/bugs/?51687>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.nongnu.org/