socket.c

一个Windows下的Linux专用虚拟机

	sb->sb_rptr += nn;
	if (sb->sb_rptr >= (sb->sb_data + sb->sb_datalen))
		sb->sb_rptr -= sb->sb_datalen;
	
	/*
	 * If in DRAIN mode, and there's no more data, set
	 * it CANTSENDMORE
	 */
	if ((so->so_state & SS_FWDRAIN) && sb->sb_cc == 0)
		sofcantsendmore(so);
	
	return nn;
}

/*
 * recvfrom() a UDP socket
 */
void
sorecvfrom(so)
	struct socket *so;
{
	struct sockaddr_in addr;
	socklen_t addrlen = sizeof(struct sockaddr_in);
	
	DEBUG_CALL("sorecvfrom");
	DEBUG_ARG("so = %lx", (long)so);
	
	if (so->so_type == IPPROTO_ICMP) {   /* This is a "ping" reply */
	  char buff[256];
	  int len;
		
	  len = recvfrom(so->s, buff, 256, 0, 
			 (struct sockaddr *)&addr, &addrlen);
	  /* XXX Check if reply is "correct"? */

#ifdef WSAECONNRESET
	  /*
	   * WSAECONNRESET: Connection reset by peer. An existing
	   * connection was forcibly closed by the remote host
	   * WIN32: Connection is closed after "ping" replay.
	   * Hack: It's not an error, it's good response here.
	   */
          if (len == -1 && errno == WSAECONNRESET) {
	    icmp_reflect(so->so_m);
	    so->so_m = 0; /* Don't m_free() it again! */
	  } else
#endif
	  if(len == -1 || len == 0) {
	    u_char code=ICMP_UNREACH_PORT;

	    if(errno == EHOSTUNREACH) code=ICMP_UNREACH_HOST;
	    else if(errno == ENETUNREACH) code=ICMP_UNREACH_NET;
	    
	    DEBUG_MISC((dfd," udp icmp rx errno = %d-%s\n",
			errno,strerror(errno)));
	    icmp_error(so->so_m, ICMP_UNREACH,code, 0,strerror(errno));
	  } else {
	    icmp_reflect(so->so_m);
	    so->so_m = 0; /* Don't m_free() it again! */
	  }
	  /* No need for this socket anymore, udp_detach it */
	  udp_detach(so);
	} else {                            	/* A "normal" UDP packet */
	  struct mbuf *m;
	  int len;
	  unsigned long n;

	  if (!(m = m_get())) return;
	  m->m_data += if_maxlinkhdr;
		
	  /* 
	   * XXX Shouldn't FIONREAD packets destined for port 53,
	   * but I don't know the max packet size for DNS lookups
	   */
	  len = M_FREEROOM(m);
	  /* if (so->so_fport != htons(53)) { */
	  ioctlsocket(so->s, FIONREAD, &n);
	  
	  if (n > len) {
	    n = (m->m_data - m->m_dat) + m->m_len + n + 1;
	    m_inc(m, n);
	    len = M_FREEROOM(m);
	  }
	  /* } */
		
	  m->m_len = recvfrom(so->s, m->m_data, len, 0,
			      (struct sockaddr *)&addr, &addrlen);
	  DEBUG_MISC((dfd, " did recvfrom %d, errno = %d-%s\n", 
		      m->m_len, errno,strerror(errno)));
	  if(m->m_len<0) {
	    u_char code=ICMP_UNREACH_PORT;

	    if(errno == EHOSTUNREACH) code=ICMP_UNREACH_HOST;
	    else if(errno == ENETUNREACH) code=ICMP_UNREACH_NET;
	    
	    DEBUG_MISC((dfd," rx error, tx icmp ICMP_UNREACH:%i\n", code));
	    icmp_error(so->so_m, ICMP_UNREACH,code, 0,strerror(errno));
	    m_free(m);
	  } else {
	  /*
	   * Hack: domain name lookup will be used the most for UDP,
	   * and since they'll only be used once there's no need
	   * for the 4 minute (or whatever) timeout... So we time them
	   * out much quicker (10 seconds  for now...)
	   */
	    if (so->so_expire) {
	      if (so->so_fport == htons(53))
		so->so_expire = curtime + SO_EXPIREFAST;
	      else
		so->so_expire = curtime + SO_EXPIRE;
	    }

	    /*		if (m->m_len == len) {
	     *			m_inc(m, MINCSIZE);
	     *			m->m_len = 0;
	     *		}
	     */
	    
	    /* 
	     * If this packet was destined for CTL_ADDR,
	     * make it look like that's where it came from, done by udp_output
	     */
	    udp_output(so, m, &addr);
	  } /* rx error */
	} /* if ping packet */
}

/*
 * sendto() a socket
 */
int
sosendto(so, m)
	struct socket *so;
	struct mbuf *m;
{
	int ret;
	struct sockaddr_in addr;

	DEBUG_CALL("sosendto");
	DEBUG_ARG("so = %lx", (long)so);
	DEBUG_ARG("m = %lx", (long)m);
	
        addr.sin_family = AF_INET;
	if ((so->so_faddr.s_addr & htonl(0xffffff00)) == special_addr.s_addr) {
	  /* It's an alias */
	  switch(ntohl(so->so_faddr.s_addr) & 0xff) {
	  case CTL_DNS:
	    addr.sin_addr = cached_dns_addr();
	    break;
	  case CTL_ALIAS:
	  default:
	    addr.sin_addr = loopback_addr;
	    break;
	  }
	} else
	  addr.sin_addr = so->so_faddr;
	addr.sin_port = so->so_fport;

	DEBUG_MISC((dfd, " sendto()ing, addr.sin_port=%d, addr.sin_addr.s_addr=%.16s\n", ntohs(addr.sin_port), inet_ntoa(addr.sin_addr)));
	
	/* Don't care what port we get */
	ret = sendto(so->s, m->m_data, m->m_len, 0,
		     (struct sockaddr *)&addr, sizeof (struct sockaddr));
	if (ret < 0)
		return -1;
	
	/*
	 * Kill the socket if there's no reply in 4 minutes,
	 * but only if it's an expirable socket
	 */
	if (so->so_expire)
		so->so_expire = curtime + SO_EXPIRE;
	so->so_state = SS_ISFCONNECTED; /* So that it gets select()ed */
	return 0;
}

/*
 * XXX This should really be tcp_listen
 */
struct socket *
solisten(port, laddr, lport, flags)
	u_int port;
	u_int32_t laddr;
	u_int lport;
	int flags;
{
	struct sockaddr_in addr;
	struct socket *so;
	socklen_t addrlen = sizeof(addr);
	int s, opt = 1;

	DEBUG_CALL("solisten");
	DEBUG_ARG("port = %d", port);
	DEBUG_ARG("laddr = %x", laddr);
	DEBUG_ARG("lport = %d", lport);
	DEBUG_ARG("flags = %x", flags);
	
	if ((so = socreate()) == NULL) {
	  /* free(so);      Not sofree() ??? free(NULL) == NOP */
	  return NULL;
	}
	
	/* Don't tcp_attach... we don't need so_snd nor so_rcv */
	if ((so->so_tcpcb = tcp_newtcpcb(so)) == NULL) {
		free(so);
		return NULL;
	}
	insque(so,&tcb);
	
	/* 
	 * SS_FACCEPTONCE sockets must time out.
	 */
	if (flags & SS_FACCEPTONCE)
	   so->so_tcpcb->t_timer[TCPT_KEEP] = TCPTV_KEEP_INIT*2;
	
	so->so_state = (SS_FACCEPTCONN|flags);
	so->so_lport = lport; /* Kept in network format */
	so->so_laddr.s_addr = laddr; /* Ditto */

	addr.sin_family = AF_INET;
	addr.sin_addr.s_addr = INADDR_ANY;
	addr.sin_port = port;
	
	if (((s = socket(AF_INET,SOCK_STREAM,0)) < 0) ||
	    (setsockopt(s,SOL_SOCKET,SO_REUSEADDR,(char *)&opt,sizeof(int)) < 0) ||
	    (bind(s,(struct sockaddr *)&addr, sizeof(addr)) < 0) ||
	    (listen(s,1) < 0)) {
		int tmperrno = errno; /* Don't clobber the real reason we failed */
		
		close(s);
		sofree(so);
		/* Restore the real errno */
#ifdef _WIN32
		WSASetLastError(tmperrno);
#else
		errno = tmperrno;
#endif
		return NULL;
	}
	setsockopt(s,SOL_SOCKET,SO_OOBINLINE,(char *)&opt,sizeof(int));
	
	getsockname(s,(struct sockaddr *)&addr,&addrlen);
	so->so_fport = addr.sin_port;

	/* Translate connections from localhost to the alias hostname */
	if (is_localhost(addr.sin_addr))
	   so->so_faddr = alias_addr;
	else
	   so->so_faddr = addr.sin_addr;

	so->s = s;
	return so;
}

/* 
 * Data is available in so_rcv
 * Just write() the data to the socket
 * XXX not yet...
 */
void
sorwakeup(so)
	struct socket *so;
{
/*	sowrite(so); */
/*	FD_CLR(so->s,&writefds); */
}
	
/*
 * Data has been freed in so_snd
 * We have room for a read() if we want to
 * For now, don't read, it'll be done in the main loop
 */
void
sowwakeup(so)
	struct socket *so;
{
	/* Nothing, yet */
}

/*
 * Various session state calls
 * XXX Should be #define's
 * The socket state stuff needs work, these often get call 2 or 3
 * times each when only 1 was needed
 */
void
soisfconnecting(so)
	register struct socket *so;
{
	so->so_state &= ~(SS_NOFDREF|SS_ISFCONNECTED|SS_FCANTRCVMORE|
			  SS_FCANTSENDMORE|SS_FWDRAIN);
	so->so_state |= SS_ISFCONNECTING; /* Clobber other states */
}

void
soisfconnected(so)
        register struct socket *so;
{
	so->so_state &= ~(SS_ISFCONNECTING|SS_FWDRAIN|SS_NOFDREF);
	so->so_state |= SS_ISFCONNECTED; /* Clobber other states */
}

void
sofcantrcvmore(so)
	struct  socket *so;
{
	if ((so->so_state & SS_NOFDREF) == 0) {
		shutdown(so->s,0);
		if(global_writefds) {
		  FD_CLR(so->s,global_writefds);
		}
	}
	so->so_state &= ~(SS_ISFCONNECTING);
	if (so->so_state & SS_FCANTSENDMORE)
	   so->so_state = SS_NOFDREF; /* Don't select it */ /* XXX close() here as well? */
	else
	   so->so_state |= SS_FCANTRCVMORE;
}

void
sofcantsendmore(so)
	struct socket *so;
{
	if ((so->so_state & SS_NOFDREF) == 0) {
            shutdown(so->s,1);           /* send FIN to fhost */
            if (global_readfds) {
                FD_CLR(so->s,global_readfds);
            }
            if (global_xfds) {
                FD_CLR(so->s,global_xfds);
            }
	}
	so->so_state &= ~(SS_ISFCONNECTING);
	if (so->so_state & SS_FCANTRCVMORE)
	   so->so_state = SS_NOFDREF; /* as above */
	else
	   so->so_state |= SS_FCANTSENDMORE;
}

void
soisfdisconnected(so)
	struct socket *so;
{
/*	so->so_state &= ~(SS_ISFCONNECTING|SS_ISFCONNECTED); */
/*	close(so->s); */
/*	so->so_state = SS_ISFDISCONNECTED; */
	/*
	 * XXX Do nothing ... ?
	 */
}

/*
 * Set write drain mode
 * Set CANTSENDMORE once all data has been write()n
 */
void
sofwdrain(so)
	struct socket *so;
{
	if (so->so_rcv.sb_cc)
		so->so_state |= SS_FWDRAIN;
	else
		sofcantsendmore(so);
}