merge wnj changes plus lint

[unix-history] / usr / src / sys / netinet / tcp_usrreq.c

diff --git a/usr/src/sys/netinet/tcp_usrreq.c b/usr/src/sys/netinet/tcp_usrreq.c
index 17007fa..229cbfc 100644 (file)
--- a/usr/src/sys/netinet/tcp_usrreq.c
+++ b/usr/src/sys/netinet/tcp_usrreq.c
@@ -1,4 +1,4 @@
-/* tcp_usrreq.c 1.41 81/12/12 */
+/*     tcp_usrreq.c    1.78    83/05/27        */

 
 #include "../h/param.h"
 #include "../h/systm.h"
 
 #include "../h/param.h"
 #include "../h/systm.h"


@@ -6,146 +6,225 @@
 #include "../h/socket.h"
 #include "../h/socketvar.h"
 #include "../h/protosw.h"
 #include "../h/socket.h"
 #include "../h/socketvar.h"
 #include "../h/protosw.h"

-#include "../net/in.h"
-#include "../net/in_pcb.h"
-#include "../net/in_systm.h"
+#include "../h/errno.h"
+

 #include "../net/if.h"
 #include "../net/if.h"

-#include "../net/ip.h"
-#include "../net/ip_var.h"
-#include "../net/tcp.h"
-#include "../net/tcp_fsm.h"
-#include "../net/tcp_seq.h"
-#include "../net/tcp_timer.h"
-#include "../net/tcp_var.h"
-#include "../net/tcpip.h"
-#include "../errno.h"
-
-extern char *tcpstates[];
+#include "../net/route.h"
+
+#include "../netinet/in.h"
+#include "../netinet/in_pcb.h"
+#include "../netinet/in_systm.h"
+#include "../netinet/ip.h"
+#include "../netinet/ip_var.h"
+#include "../netinet/tcp.h"
+#include "../netinet/tcp_fsm.h"
+#include "../netinet/tcp_seq.h"
+#include "../netinet/tcp_timer.h"
+#include "../netinet/tcp_var.h"
+#include "../netinet/tcpip.h"
+#include "../netinet/tcp_debug.h"
+
+/*
+ * TCP protocol interface to socket abstraction.
+ */
+extern char *tcpstates[];

 struct tcpcb *tcp_newtcpcb();
 struct tcpcb *tcp_newtcpcb();

+int    tcpsenderrors;
+

 /*
 /*

- * Process a TCP user request for tcp tb.  If this is a send request
+ * Process a TCP user request for TCP tb.  If this is a send request

  * then m is the mbuf chain of send data.  If this is a timer expiration
  * (called from the software clock routine), then timertype tells which timer.
  */
  * then m is the mbuf chain of send data.  If this is a timer expiration
  * (called from the software clock routine), then timertype tells which timer.
  */

-tcp_usrreq(so, req, m, addr)
+/*ARGSUSED*/
+tcp_usrreq(so, req, m, nam, rights)

        struct socket *so;
        int req;
        struct socket *so;
        int req;

-       struct mbuf *m;
-       caddr_t addr;
+       struct mbuf *m, *nam, *rights;

 {
        register struct inpcb *inp = sotoinpcb(so);
        register struct tcpcb *tp;
        int s = splnet();
        int error = 0;
 {
        register struct inpcb *inp = sotoinpcb(so);
        register struct tcpcb *tp;
        int s = splnet();
        int error = 0;

-COUNT(TCP_USRREQ);
+       int ostate;

 
 

+       if (rights && rights->m_len) {
+               splx(s);
+               return (EINVAL);
+       }

        /*
        /*

-        * Make sure attached.  If not,
-        * only PRU_ATTACH is valid.
+        * When a TCP is attached to a socket, then there will be
+        * a (struct inpcb) pointed at by the socket, and this
+        * structure will point at a subsidary (struct tcpcb).

         */
        if (inp == 0 && req != PRU_ATTACH) {
                splx(s);
         */
        if (inp == 0 && req != PRU_ATTACH) {
                splx(s);

-               return (EINVAL);
+               return (EINVAL);                /* XXX */

        }
        if (inp) {
                tp = intotcpcb(inp);
        }
        if (inp) {
                tp = intotcpcb(inp);

+               /* WHAT IF TP IS 0? */

 #ifdef KPROF
                tcp_acounts[tp->t_state][req]++;
 #endif
 #ifdef KPROF
                tcp_acounts[tp->t_state][req]++;
 #endif

-       }
+               ostate = tp->t_state;
+       } else
+               ostate = 0;

        switch (req) {
 
        switch (req) {
 

+       /*
+        * TCP attaches to socket via PRU_ATTACH, reserving space,
+        * and an internet control block.
+        */

        case PRU_ATTACH:
                if (inp) {
                        error = EISCONN;
                        break;
                }
        case PRU_ATTACH:
                if (inp) {
                        error = EISCONN;
                        break;
                }

-               error = in_pcbattach(so, &tcb, 2048, 2048, (struct sockaddr_in *)addr);
+               error = tcp_attach(so);

                if (error)
                        break;
                if (error)
                        break;

-               inp = (struct inpcb *)so->so_pcb;
-               tp = tcp_newtcpcb(inp);
-               if (so->so_options & SO_ACCEPTCONN) {
-                       if (tp == 0) {
-                               in_pcbdetach(inp);
-                               error = ENOBUFS;
-                               break;
-                       }
-                       tp->t_state = TCPS_LISTEN;
-               } else
-                       tp->t_state = TCPS_CLOSED;
+               if ((so->so_options & SO_LINGER) && so->so_linger == 0)
+                       so->so_linger = TCP_LINGERTIME;
+               tp = sototcpcb(so);

                break;
 
                break;
 

+       /*
+        * PRU_DETACH detaches the TCP protocol from the socket.
+        * If the protocol state is non-embryonic, then can't
+        * do this directly: have to initiate a PRU_DISCONNECT,
+        * which may finish later; embryonic TCB's can just
+        * be discarded here.
+        */

        case PRU_DETACH:
        case PRU_DETACH:

-               in_pcbdetach(inp);
+               if (tp->t_state > TCPS_LISTEN)
+                       tp = tcp_disconnect(tp);
+               else
+                       tp = tcp_close(tp);
+               break;
+
+       /*
+        * Give the socket an address.
+        */
+       case PRU_BIND:
+               error = in_pcbbind(inp, nam);
+               if (error)
+                       break;
+               break;
+
+       /*
+        * Prepare to accept connections.
+        */
+       case PRU_LISTEN:
+               if (inp->inp_lport == 0)
+                       error = in_pcbbind(inp, (struct mbuf *)0);
+               if (error == 0)
+                       tp->t_state = TCPS_LISTEN;

                break;
 
                break;
 

+       /*
+        * Initiate connection to peer.
+        * Create a template for use in transmissions on this connection.
+        * Enter SYN_SENT state, and mark socket as connecting.
+        * Start keep-alive timer, and seed output sequence space.
+        * Send initial segment on connection.
+        */

        case PRU_CONNECT:
        case PRU_CONNECT:

-               error = in_pcbconnect(inp, (struct sockaddr_in *)addr);
+               if (inp->inp_lport == 0) {
+                       error = in_pcbbind(inp, (struct mbuf *)0);
+                       if (error)
+                               break;
+               }
+               error = in_pcbconnect(inp, nam);

                if (error)
                        break;
                if (error)
                        break;

-               tp = tcp_newtcpcb(inp);
-               if (tp == 0)
-                       goto badcon;

                tp->t_template = tcp_template(tp);
                tp->t_template = tcp_template(tp);

-               if (tp->t_template == 0)
-                       goto badcon2;
-               tp->t_inpcb = inp;
-               inp->inp_ppcb = (caddr_t)tp;
+               if (tp->t_template == 0) {
+                       in_pcbdisconnect(inp);
+                       error = ENOBUFS;
+                       break;
+               }

                soisconnecting(so);
                tp->t_state = TCPS_SYN_SENT;
                tp->t_timer[TCPT_KEEP] = TCPTV_KEEP;
                tp->iss = tcp_iss; tcp_iss += TCP_ISSINCR/2;
                tcp_sendseqinit(tp);
                soisconnecting(so);
                tp->t_state = TCPS_SYN_SENT;
                tp->t_timer[TCPT_KEEP] = TCPTV_KEEP;
                tp->iss = tcp_iss; tcp_iss += TCP_ISSINCR/2;
                tcp_sendseqinit(tp);

-               (void) tcp_output(tp);
+               error = tcp_output(tp);

                break;
 
                break;
 

-badcon2:
-               (void) m_free(dtom(tp));
-badcon:
-               in_pcbdisconnect(inp);
-               error = ENOBUFS;
+       /*
+        * Initiate disconnect from peer.
+        * If connection never passed embryonic stage, just drop;
+        * else if don't need to let data drain, then can just drop anyways,
+        * else have to begin TCP shutdown process: mark socket disconnecting,
+        * drain unread data, state switch to reflect user close, and
+        * send segment (e.g. FIN) to peer.  Socket will be really disconnected
+        * when peer sends FIN and acks ours.
+        *
+        * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB.
+        */
+       case PRU_DISCONNECT:
+               tp = tcp_disconnect(tp);

                break;
 
                break;
 

-       case PRU_ACCEPT:
-               soisconnected(so);
-               break;
+       /*
+        * Accept a connection.  Essentially all the work is
+        * done at higher levels; just return the address
+        * of the peer, storing through addr.
+        */
+       case PRU_ACCEPT: {
+               struct sockaddr_in *sin = mtod(nam, struct sockaddr_in *);

 
 

-       case PRU_DISCONNECT:
-               if (tp->t_state < TCPS_ESTABLISHED)
-                       tcp_close(tp);
-               else {
-                       soisdisconnecting(so);
-                       tcp_usrclosed(tp);
-                       (void) tcp_output(tp);
-               }
+               nam->m_len = sizeof (struct sockaddr_in);
+               sin->sin_family = AF_INET;
+               sin->sin_port = inp->inp_fport;
+               sin->sin_addr = inp->inp_faddr;

                break;
                break;

+               }

 
 

+       /*
+        * Mark the connection as being incapable of further output.
+        */

        case PRU_SHUTDOWN:
                socantsendmore(so);
        case PRU_SHUTDOWN:
                socantsendmore(so);

-               tcp_usrclosed(tp);
-               (void) tcp_output(tp);
+               tp = tcp_usrclosed(tp);
+               if (tp)
+                       error = tcp_output(tp);

                break;
 
                break;
 

+       /*
+        * After a receive, possibly send window update to peer.
+        */

        case PRU_RCVD:
                (void) tcp_output(tp);
                break;
 
        case PRU_RCVD:
                (void) tcp_output(tp);
                break;
 

+       /*
+        * Do a send by putting data in output queue and updating urgent
+        * marker if URG set.  Possibly send more data.
+        */

        case PRU_SEND:
                sbappend(&so->so_snd, m);
        case PRU_SEND:
                sbappend(&so->so_snd, m);

-/*
+#ifdef notdef

                if (tp->t_flags & TF_PUSH)
                        tp->snd_end = tp->snd_una + so->so_snd.sb_cc;
                if (tp->t_flags & TF_PUSH)
                        tp->snd_end = tp->snd_una + so->so_snd.sb_cc;

- */
-               if (tp->t_flags & TF_URG)
-                       tp->snd_up = tp->snd_una + so->so_snd.sb_cc + 1;
-               (void) tcp_output(tp);
+#endif
+               error = tcp_output(tp);
+               if (error) {            /* XXX fix to use other path */
+                       if (error == ENOBUFS)           /* XXX */
+                               error = 0;              /* XXX */
+                       tcpsenderrors++;
+               }

                break;
 
                break;
 

+       /*
+        * Abort the TCP.
+        */

        case PRU_ABORT:
        case PRU_ABORT:

-               tcp_drop(tp, ECONNABORTED);
+               tp = tcp_drop(tp, ECONNABORTED);

                break;
 
                break;
 

+/* SOME AS YET UNIMPLEMENTED HOOKS */

        case PRU_CONTROL:
                error = EOPNOTSUPP;
                break;
        case PRU_CONTROL:
                error = EOPNOTSUPP;
                break;


@@ -153,36 +232,141 @@ badcon:
        case PRU_SENSE:
                error = EOPNOTSUPP;
                break;
        case PRU_SENSE:
                error = EOPNOTSUPP;
                break;

+/* END UNIMPLEMENTED HOOKS */

 
        case PRU_RCVOOB:
 
        case PRU_RCVOOB:

-               error = EOPNOTSUPP;
+               if (so->so_oobmark == 0 &&
+                   (so->so_state & SS_RCVATMARK) == 0) {
+                       error = EINVAL;
+                       break;
+               }
+               if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) {
+                       error = EWOULDBLOCK;
+                       break;
+               }
+               m->m_len = 1;
+               *mtod(m, caddr_t) = tp->t_iobc;

                break;
 
        case PRU_SENDOOB:
                break;
 
        case PRU_SENDOOB:

-               error = EOPNOTSUPP;
+               if (sbspace(&so->so_snd) < -512) {
+                       m_freem(m);
+                       error = ENOBUFS;
+                       break;
+               }
+               tp->snd_up = tp->snd_una + so->so_snd.sb_cc + 1;
+               sbappend(&so->so_snd, m);
+               tp->t_force = 1;
+               error = tcp_output(tp);
+               tp->t_force = 0;

                break;
 
                break;
 

+       case PRU_SOCKADDR:
+               in_setsockaddr(inp, nam);
+               break;
+
+       /*
+        * TCP slow timer went off; going through this
+        * routine for tracing's sake.
+        */

        case PRU_SLOWTIMO:
        case PRU_SLOWTIMO:

-               tcp_timers(tp, (int)addr);
+               tp = tcp_timers(tp, (int)nam);
+               req |= (int)nam << 8;           /* for debug's sake */

                break;
 
        default:
                panic("tcp_usrreq");
        }
                break;
 
        default:
                panic("tcp_usrreq");
        }

+       if (tp && (so->so_options & SO_DEBUG))
+               tcp_trace(TA_USER, ostate, tp, (struct tcpiphdr *)0, req);

        splx(s);
        return (error);
 }
 
        splx(s);
        return (error);
 }
 

+int    tcp_sendspace = 1024*2;
+int    tcp_recvspace = 1024*2;
+/*
+ * Attach TCP protocol to socket, allocating
+ * internet protocol control block, tcp control block,
+ * bufer space, and entering LISTEN state if to accept connections.
+ */
+tcp_attach(so)
+       struct socket *so;
+{
+       register struct tcpcb *tp;
+       struct inpcb *inp;
+       int error;
+
+       error = soreserve(so, tcp_sendspace, tcp_recvspace);
+       if (error)
+               goto bad;
+       error = in_pcballoc(so, &tcb);
+       if (error)
+               goto bad;
+       inp = sotoinpcb(so);
+       tp = tcp_newtcpcb(inp);
+       if (tp == 0) {
+               error = ENOBUFS;
+               goto bad2;
+       }
+       tp->t_state = TCPS_CLOSED;
+       return (0);
+bad2:
+       in_pcbdetach(inp);
+bad:
+       return (error);
+}
+
+/*
+ * Initiate (or continue) disconnect.
+ * If embryonic state, just send reset (once).
+ * If not in ``let data drain'' option, just drop.
+ * Otherwise (hard), mark socket disconnecting and drop
+ * current input data; switch states based on user close, and
+ * send segment to peer (with FIN).
+ */
+struct tcpcb *
+tcp_disconnect(tp)
+       register struct tcpcb *tp;
+{
+       struct socket *so = tp->t_inpcb->inp_socket;
+
+       if (tp->t_state < TCPS_ESTABLISHED)
+               tp = tcp_close(tp);
+       else if (so->so_linger == 0)
+               tp = tcp_drop(tp, 0);
+       else {
+               soisdisconnecting(so);
+               sbflush(&so->so_rcv);
+               tp = tcp_usrclosed(tp);
+               if (tp)
+                       (void) tcp_output(tp);
+       }
+       return (tp);
+}
+
+/*
+ * User issued close, and wish to trail through shutdown states:
+ * if never received SYN, just forget it.  If got a SYN from peer,
+ * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN.
+ * If already got a FIN from peer, then almost done; go to LAST_ACK
+ * state.  In all other cases, have already sent FIN to peer (e.g.
+ * after PRU_SHUTDOWN), and just have to play tedious game waiting
+ * for peer to send FIN or not respond to keep-alives, etc.
+ * We can let the user exit from the close as soon as the FIN is acked.
+ */
+struct tcpcb *

 tcp_usrclosed(tp)
 tcp_usrclosed(tp)

-       struct tcpcb *tp;
+       register struct tcpcb *tp;

 {
 
        switch (tp->t_state) {
 
 {
 
        switch (tp->t_state) {
 

+       case TCPS_CLOSED:

        case TCPS_LISTEN:
        case TCPS_SYN_SENT:
                tp->t_state = TCPS_CLOSED;
        case TCPS_LISTEN:
        case TCPS_SYN_SENT:
                tp->t_state = TCPS_CLOSED;

-               tcp_close(tp);
+               tp = tcp_close(tp);

                break;
 
        case TCPS_SYN_RECEIVED:
                break;
 
        case TCPS_SYN_RECEIVED:


@@ -194,4 +378,7 @@ tcp_usrclosed(tp)
                tp->t_state = TCPS_LAST_ACK;
                break;
        }
                tp->t_state = TCPS_LAST_ACK;
                break;
        }

+       if (tp && tp->t_state >= TCPS_FIN_WAIT_2)
+               soisdisconnected(tp->t_inpcb->inp_socket);
+       return (tp);

 }
 }