Re: [lwip-members] new udp.c issue / sharing of UDP port between broadcast listeners

[Jani Monoses]

Ok Leon go ahead
I think I also found something in udp_bind
it seems that a pcb can't be rebound if
it keeps the port but changes the address(not too frequent but still)
so I put a continue so pcb does not compare with itself instead
it goes to the next in thelist

because after rebind is set to 1 the following 

 if (pcb == ipcb) {
            rebind = 1;
            continue;   <<HERE
    }
    /* port matches that of PCB in list? */
    if ((ipcb->local_port == port) &&
       /* IP address matches, or one is IP_ADDR_ANY? */
      (ip_addr_isany(&(ipcb->local_ip)) ||
             ip_addr_isany(ipaddr) ||
             ip_addr_cmp(&(ipcb->local_ip), ipaddr))) {


      
> Hello,
> 
> found another issue in recent udp.c:
> 
> udp_bind(struct udp_pcb *pcb, struct ip_addr *ipaddr, u16_t port)
> 
> does no longer allow two PCB's to share a local port.
> 
> This is what we use to mimick SO_REUSEADDR behaviour where
> multiple UDP ports are allowed to listen to the same broadcast
> traffic on a certain UDP port. This is allowed under *BSD/Unix/Linux
> type sockets.
> 
> As said earlier, I will revert the changes made to udp.c by simply
> disabling the new code using #if 0 constructs. This will probably
> break some issues in the netcon_ layer once I commit.
> 
> But I rather choose the lower layer to break the upper, then vice
> versa. (The lower layers must not restrict functionality needed by
> upper layers.)
> 
> Regards,
> 
> Leon.
> 
> 
> 
> _______________________________________________
> lwip-members mailing list
> address@hidden
> http://mail.nongnu.org/mailman/listinfo/lwip-members