Re: [lwip-users] Nonblocking sockets problem

[Åke Forslund]

Hi again, this is not the original code but a implementation of the problem in 
our simple telnet-server for debug. Code below. In our original code we use 
ioctlsocket() to set non-blocking but using lwip_ fcntl() makes no difference.

void task_telnet(void * arg)
{
  int lSocketRecv;
  u32_t lAddrSize;
  struct addrinfo *sListenAddr, hints;
  char buffer[1024];
  unsigned short u16L_buffPtr = 0;
  int nbytes;
  unsigned char one = 1;

  LWIP_DEBUGF(SOCKETS_DEBUG, ("Recv Task started!\n"));
  memset(&hints, 0, sizeof(struct addrinfo));
  hints.ai_family   = AF_UNSPEC;
  hints.ai_socktype = SOCK_STREAM;

  lwip_getaddrinfo("0.0.0.0", "23", &hints, &sListenAddr);
    
  lSocketRecv = lwip_socket(sListenAddr->ai_family, sListenAddr->ai_socktype, 
sListenAddr->ai_protocol);
  
  
  lwip_bind(lSocketRecv, sListenAddr->ai_addr, sListenAddr->ai_addrlen);
  //ioctlsocket(lSocketRecv, FIONBIO, &one); 
  lwip_fcntl(lSocketRecv, F_SETFL, lwip_fcntl(lSocketRecv, F_GETFL, 0) | 
O_NONBLOCK);
  lwip_listen(lSocketRecv, 128);
    
  while (1)
  {
    os_dly_wait(100);
    lAddrSize = sizeof(sListenAddr);
    new_fd = lwip_accept(lSocketRecv, (struct sockaddr *)&sListenAddr, 
&lAddrSize);
    lwip_write(new_fd, "Welcome to the jungle\n\n\r>", 25);
    str_tcpTty.active = 1;
    while (1)
    {
      nbytes = lwip_read(new_fd, &buffer[u16L_buffPtr], 1);
      if (nbytes > 0)
      {
        u16L_buffPtr += nbytes;
        if (buffer[u16L_buffPtr-1] == 'q')    // q ends the session
          break;  // Terminate session
        
        if (!testParseBuff((char *)buffer, u16L_buffPtr, &str_tcpTty)) // 
complete command found and parsed
        {
          u16L_buffPtr = 0;
          lwip_write(new_fd, "\r>", 2);
        }
      }

      os_dly_wait(6);
    }
    str_tcpTty.active = 0;
    u16L_buffPtr = 0;
    lwip_close(new_fd);
  }
}

At the moment we can manage without nonblocking sockets using lwip_setsockopt() 
to set a timeout for the socket, but we are still curious as to why 
nonblocking-sockets don't work for us.

Best regards
/Åke

-----Ursprungligt meddelande-----
Från: address@hidden [mailto:address@hidden För Simon Goldschmidt
Skickat: den 13 mars 2012 09:06
Till: Mailing list for lwIP users
Ämne: Re: [lwip-users] Nonblocking sockets problem

"Åke Forslund" <address@hidden> wrote:
> We added some debug-prints in the case NETCONN_EVT_RCVPLUS in 
> event_callback() showing that  rcvevent was only incremented when the 
> socket was set as blocking.
> 
> I'm having trouble seeing where to probe for the error so if anyone 
> has an idea of where to look, I would be very thankful. Is there any 
> options in lwipopt that could trigger this behavior?
> From where is event_callback() triggered normally, it's added as a 
> callback to the netconnection-object but where is this callback called?

The callback is called via the macro API_EVENT(), which is mainly called with 
RECVPLUS from the raw API recv- and accept-callbacks in api_msg.c (those are 
called from tcpip_thread). However, it is also called with RECVMINUS from 
netconn_recv_data() in api_lib.c (called from your application thread).

Which type of socket are you using (tcp, udp, raw)?

Could you provide a simple test code that reproduces this, that would make it 
much easier for me to check.

Thanks,
Simon
-- 
NEU: FreePhone 3-fach-Flat mit kostenlosem Smartphone!                          
        
Jetzt informieren: http://mobile.1und1.de/?ac=OM.PW.PW003K20328T7073a

_______________________________________________
lwip-users mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/lwip-users