4.4BSD snapshot (revision 8.1); add 1993 to copyright

[unix-history] / usr / src / sys / kern / uipc_socket2.c

diff --git a/usr/src/sys/kern/uipc_socket2.c b/usr/src/sys/kern/uipc_socket2.c
index 24a3ed2..1b22cbc 100644 (file)
--- a/usr/src/sys/kern/uipc_socket2.c
+++ b/usr/src/sys/kern/uipc_socket2.c
@@ -1,28 +1,39 @@
-/*     uipc_socket2.c  4.22    82/03/19        */
-
-#include "../h/param.h"
-#include "../h/systm.h"
-#include "../h/dir.h"
-#include "../h/user.h"
-#include "../h/proc.h"
-#include "../h/file.h"
-#include "../h/inode.h"
-#include "../h/buf.h"
-#include "../h/mbuf.h"
-#include "../h/protosw.h"
-#include "../h/socket.h"
-#include "../h/socketvar.h"
-#include "../net/in.h"
-#include "../net/in_systm.h"
+/*
+ * Copyright (c) 1982, 1986, 1988, 1990, 1993
+ *     The Regents of the University of California.  All rights reserved.
+ *
+ * %sccs.include.redist.c%
+ *
+ *     @(#)uipc_socket2.c      8.1 (Berkeley) %G%
+ */
+
+#include <sys/param.h>
+#include <sys/systm.h>
+#include <sys/proc.h>
+#include <sys/file.h>
+#include <sys/buf.h>
+#include <sys/malloc.h>
+#include <sys/mbuf.h>
+#include <sys/protosw.h>
+#include <sys/socket.h>
+#include <sys/socketvar.h>

 
 /*
  * Primitive routines for operating on sockets and socket buffers
  */
 
 
 /*
  * Primitive routines for operating on sockets and socket buffers
  */
 

+/* strings for sleep message: */
+char   netio[] = "netio";
+char   netcon[] = "netcon";
+char   netcls[] = "netcls";
+
+u_long sb_max = SB_MAX;                /* patchable */
+

 /*
  * Procedures to manipulate state flags of socket
 /*
  * Procedures to manipulate state flags of socket

- * and do appropriate wakeups.  Normal sequence is that
- * soisconnecting() is called during processing of connect() call,
+ * and do appropriate wakeups.  Normal sequence from the
+ * active (originating) side is that soisconnecting() is
+ * called during processing of connect() call,

  * resulting in an eventual call to soisconnected() if/when the
  * connection is established.  When the connection is torn down
  * soisdisconnecting() is called during processing of disconnect() call,
  * resulting in an eventual call to soisconnected() if/when the
  * connection is established.  When the connection is torn down
  * soisdisconnecting() is called during processing of disconnect() call,


@@ -32,33 +43,50 @@
  * only, bypassing the in-progress calls when setting up a ``connection''
  * takes no time.
  *
  * only, bypassing the in-progress calls when setting up a ``connection''
  * takes no time.
  *

- * When higher level protocols are implemented in
+ * From the passive side, a socket is created with
+ * two queues of sockets: so_q0 for connections in progress
+ * and so_q for connections already made and awaiting user acceptance.
+ * As a protocol is preparing incoming connections, it creates a socket
+ * structure queued on so_q0 by calling sonewconn().  When the connection
+ * is established, soisconnected() is called, and transfers the
+ * socket structure to so_q, making it available to accept().
+ * 
+ * If a socket is closed with sockets on either
+ * so_q0 or so_q, these sockets are dropped.
+ *
+ * If higher level protocols are implemented in

  * the kernel, the wakeups done here will sometimes
  * the kernel, the wakeups done here will sometimes

- * be implemented as software-interrupt process scheduling.
+ * cause software-interrupt process scheduling.

  */
 
 soisconnecting(so)
  */
 
 soisconnecting(so)

-       struct socket *so;
+       register struct socket *so;

 {
 
        so->so_state &= ~(SS_ISCONNECTED|SS_ISDISCONNECTING);
        so->so_state |= SS_ISCONNECTING;
 {
 
        so->so_state &= ~(SS_ISCONNECTED|SS_ISDISCONNECTING);
        so->so_state |= SS_ISCONNECTING;

-       wakeup((caddr_t)&so->so_timeo);

 }
 
 soisconnected(so)
 }
 
 soisconnected(so)

-       struct socket *so;
+       register struct socket *so;

 {
 {

+       register struct socket *head = so->so_head;

 
 

-       so->so_state &= ~(SS_ISCONNECTING|SS_ISDISCONNECTING);
+       so->so_state &= ~(SS_ISCONNECTING|SS_ISDISCONNECTING|SS_ISCONFIRMING);

        so->so_state |= SS_ISCONNECTED;
        so->so_state |= SS_ISCONNECTED;

-       wakeup((caddr_t)&so->so_timeo);
-       sorwakeup(so);
-       sowwakeup(so);
+       if (head && soqremque(so, 0)) {
+               soqinsque(head, so, 1);
+               sorwakeup(head);
+               wakeup((caddr_t)&head->so_timeo);
+       } else {
+               wakeup((caddr_t)&so->so_timeo);
+               sorwakeup(so);
+               sowwakeup(so);
+       }

 }
 
 soisdisconnecting(so)
 }
 
 soisdisconnecting(so)

-       struct socket *so;
+       register struct socket *so;

 {
 
        so->so_state &= ~SS_ISCONNECTING;
 {
 
        so->so_state &= ~SS_ISCONNECTING;


@@ -69,7 +97,7 @@ soisdisconnecting(so)
 }
 
 soisdisconnected(so)
 }
 
 soisdisconnected(so)

-       struct socket *so;
+       register struct socket *so;

 {
 
        so->so_state &= ~(SS_ISCONNECTING|SS_ISCONNECTED|SS_ISDISCONNECTING);
 {
 
        so->so_state &= ~(SS_ISCONNECTING|SS_ISCONNECTED|SS_ISDISCONNECTING);


@@ -79,6 +107,103 @@ soisdisconnected(so)
        sorwakeup(so);
 }
 
        sorwakeup(so);
 }
 

+/*
+ * When an attempt at a new connection is noted on a socket
+ * which accepts connections, sonewconn is called.  If the
+ * connection is possible (subject to space constraints, etc.)
+ * then we allocate a new structure, propoerly linked into the
+ * data structure of the original socket, and return this.
+ * Connstatus may be 0, or SO_ISCONFIRMING, or SO_ISCONNECTED.
+ *
+ * Currently, sonewconn() is defined as sonewconn1() in socketvar.h
+ * to catch calls that are missing the (new) second parameter.
+ */
+struct socket *
+sonewconn1(head, connstatus)
+       register struct socket *head;
+       int connstatus;
+{
+       register struct socket *so;
+       int soqueue = connstatus ? 1 : 0;
+
+       if (head->so_qlen + head->so_q0len > 3 * head->so_qlimit / 2)
+               return ((struct socket *)0);
+       MALLOC(so, struct socket *, sizeof(*so), M_SOCKET, M_DONTWAIT);
+       if (so == NULL) 
+               return ((struct socket *)0);
+       bzero((caddr_t)so, sizeof(*so));
+       so->so_type = head->so_type;
+       so->so_options = head->so_options &~ SO_ACCEPTCONN;
+       so->so_linger = head->so_linger;
+       so->so_state = head->so_state | SS_NOFDREF;
+       so->so_proto = head->so_proto;
+       so->so_timeo = head->so_timeo;
+       so->so_pgid = head->so_pgid;
+       (void) soreserve(so, head->so_snd.sb_hiwat, head->so_rcv.sb_hiwat);
+       soqinsque(head, so, soqueue);
+       if ((*so->so_proto->pr_usrreq)(so, PRU_ATTACH,
+           (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0)) {
+               (void) soqremque(so, soqueue);
+               (void) free((caddr_t)so, M_SOCKET);
+               return ((struct socket *)0);
+       }
+       if (connstatus) {
+               sorwakeup(head);
+               wakeup((caddr_t)&head->so_timeo);
+               so->so_state |= connstatus;
+       }
+       return (so);
+}
+
+soqinsque(head, so, q)
+       register struct socket *head, *so;
+       int q;
+{
+
+       register struct socket **prev;
+       so->so_head = head;
+       if (q == 0) {
+               head->so_q0len++;
+               so->so_q0 = 0;
+               for (prev = &(head->so_q0); *prev; )
+                       prev = &((*prev)->so_q0);
+       } else {
+               head->so_qlen++;
+               so->so_q = 0;
+               for (prev = &(head->so_q); *prev; )
+                       prev = &((*prev)->so_q);
+       }
+       *prev = so;
+}
+
+soqremque(so, q)
+       register struct socket *so;
+       int q;
+{
+       register struct socket *head, *prev, *next;
+
+       head = so->so_head;
+       prev = head;
+       for (;;) {
+               next = q ? prev->so_q : prev->so_q0;
+               if (next == so)
+                       break;
+               if (next == 0)
+                       return (0);
+               prev = next;
+       }
+       if (q == 0) {
+               prev->so_q0 = next->so_q0;
+               head->so_q0len--;
+       } else {
+               prev->so_q = next->so_q;
+               head->so_qlen--;
+       }
+       next->so_q0 = next->so_q = 0;
+       next->so_head = 0;
+       return (1);
+}
+

 /*
  * Socantsendmore indicates that no more data will be sent on the
  * socket; it would normally be applied to a socket when the user
 /*
  * Socantsendmore indicates that no more data will be sent on the
  * socket; it would normally be applied to a socket when the user


@@ -106,82 +231,61 @@ socantrcvmore(so)
 }
 
 /*
 }
 
 /*

- * Socket select/wakeup routines.
- */
-
-/*
- * Interface routine to select() system
- * call for sockets.
- */
-soselect(so, rw)
-       register struct socket *so;
-       int rw;
-{
-       int s = splnet();
-
-       switch (rw) {
-
-       case FREAD:
-               if (soreadable(so)) {
-                       splx(s);
-                       return (1);
-               }
-               sbselqueue(&so->so_rcv);
-               break;
-
-       case FWRITE:
-               if (sowriteable(so)) {
-                       splx(s);
-                       return (1);
-               }
-               sbselqueue(&so->so_snd);
-               break;
-       }
-       splx(s);
-       return (0);
-}
-
-/*
- * Queue a process for a select on a socket buffer.
+ * Wait for data to arrive at/drain from a socket buffer.

  */
  */

-sbselqueue(sb)
+sbwait(sb)

        struct sockbuf *sb;
 {
        struct sockbuf *sb;
 {

-       register struct proc *p;

 
 

-       if ((p = sb->sb_sel) && p->p_wchan == (caddr_t)&selwait)
-               sb->sb_flags |= SB_COLL;
-       else
-               sb->sb_sel = u.u_procp;
+       sb->sb_flags |= SB_WAIT;
+       return (tsleep((caddr_t)&sb->sb_cc,
+           (sb->sb_flags & SB_NOINTR) ? PSOCK : PSOCK | PCATCH, netio,
+           sb->sb_timeo));

 }
 
 }
 

-/*
- * Wait for data to arrive at/drain from a socket buffer.
+/* 
+ * Lock a sockbuf already known to be locked;
+ * return any error returned from sleep (EINTR).

  */
  */

-sbwait(sb)
-       struct sockbuf *sb;
+sb_lock(sb)
+       register struct sockbuf *sb;

 {
 {

-
-       sb->sb_flags |= SB_WAIT;
-       sleep((caddr_t)&sb->sb_cc, PZERO+1);
+       int error;
+
+       while (sb->sb_flags & SB_LOCK) {
+               sb->sb_flags |= SB_WANT;
+               if (error = tsleep((caddr_t)&sb->sb_flags, 
+                   (sb->sb_flags & SB_NOINTR) ? PSOCK : PSOCK|PCATCH,
+                   netio, 0))
+                       return (error);
+       }
+       sb->sb_flags |= SB_LOCK;
+       return (0);

 }
 
 /*
  * Wakeup processes waiting on a socket buffer.
 }
 
 /*
  * Wakeup processes waiting on a socket buffer.

+ * Do asynchronous notification via SIGIO
+ * if the socket has the SS_ASYNC flag set.

  */
  */

-sbwakeup(sb)
-       struct sockbuf *sb;
+sowakeup(so, sb)
+       register struct socket *so;
+       register struct sockbuf *sb;

 {
 {

+       struct proc *p;

 
 

-       if (sb->sb_sel) {
-               selwakeup(sb->sb_sel, sb->sb_flags & SB_COLL);
-               sb->sb_sel = 0;
-               sb->sb_flags &= ~SB_COLL;
-       }
+       selwakeup(&sb->sb_sel);
+       sb->sb_flags &= ~SB_SEL;

        if (sb->sb_flags & SB_WAIT) {
                sb->sb_flags &= ~SB_WAIT;
                wakeup((caddr_t)&sb->sb_cc);
        }
        if (sb->sb_flags & SB_WAIT) {
                sb->sb_flags &= ~SB_WAIT;
                wakeup((caddr_t)&sb->sb_cc);
        }

+       if (so->so_state & SS_ASYNC) {
+               if (so->so_pgid < 0)
+                       gsignal(-so->so_pgid, SIGIO);
+               else if (so->so_pgid > 0 && (p = pfind(so->so_pgid)) != 0)
+                       psignal(p, SIGIO);
+       }

 }
 
 /*
 }
 
 /*


@@ -193,46 +297,67 @@ sbwakeup(sb)
  * queue, and other fields allowing select() statements and notification
  * on data availability to be implemented.
  *
  * queue, and other fields allowing select() statements and notification
  * on data availability to be implemented.
  *

- * Before using a new socket structure it is first necessary to reserve
- * buffer space to the socket, by calling sbreserve.  This commits
- * some of the available buffer space in the system buffer pool for the
- * socket.  The space should be released by calling sbrelease when the
- * socket is destroyed.
+ * Data stored in a socket buffer is maintained as a list of records.
+ * Each record is a list of mbufs chained together with the m_next
+ * field.  Records are chained together with the m_nextpkt field. The upper
+ * level routine soreceive() expects the following conventions to be
+ * observed when placing information in the receive buffer:

  *
  *

- * The routine sbappend() is normally called to append new mbufs
- * to a socket buffer, after checking that adequate space is available
- * comparing the function spspace() with the amount of data to be added.
- * Data is normally removed from a socket buffer in a protocol by
- * first calling m_copy on the socket buffer mbuf chain and sending this
- * to a peer, and then removing the data from the socket buffer with
- * sbdrop when the data is acknowledged by the peer (or immediately
- * in the case of unreliable protocols.)
+ * 1. If the protocol requires each message be preceded by the sender's
+ *    name, then a record containing that name must be present before
+ *    any associated data (mbuf's must be of type MT_SONAME).
+ * 2. If the protocol supports the exchange of ``access rights'' (really
+ *    just additional data associated with the message), and there are
+ *    ``rights'' to be received, then a record containing this data
+ *    should be present (mbuf's must be of type MT_RIGHTS).
+ * 3. If a name or rights record exists, then it must be followed by
+ *    a data record, perhaps of zero length.

  *
  *

- * Protocols which do not require connections place both source address
- * and data information in socket buffer queues.  The source addresses
- * are stored in single mbufs after each data item, and are easily found
- * as the data items are all marked with end of record markers.  The
- * sbappendaddr() routine stores a datum and associated address in
- * a socket buffer.  Note that, unlike sbappend(), this routine checks
- * for the caller that there will be enough space to store the data.
- * It fails if there is not enough space, or if it cannot find
- * a mbuf to store the address in.
- *
- * The higher-level routines sosend and soreceive (in socket.c)
- * also add data to, and remove data from socket buffers repectively.
+ * Before using a new socket structure it is first necessary to reserve
+ * buffer space to the socket, by calling sbreserve().  This should commit
+ * some of the available buffer space in the system buffer pool for the
+ * socket (currently, it does nothing but enforce limits).  The space
+ * should be released by calling sbrelease() when the socket is destroyed.

  */
 
  */
 

+soreserve(so, sndcc, rcvcc)
+       register struct socket *so;
+       u_long sndcc, rcvcc;
+{
+
+       if (sbreserve(&so->so_snd, sndcc) == 0)
+               goto bad;
+       if (sbreserve(&so->so_rcv, rcvcc) == 0)
+               goto bad2;
+       if (so->so_rcv.sb_lowat == 0)
+               so->so_rcv.sb_lowat = 1;
+       if (so->so_snd.sb_lowat == 0)
+               so->so_snd.sb_lowat = MCLBYTES;
+       if (so->so_snd.sb_lowat > so->so_snd.sb_hiwat)
+               so->so_snd.sb_lowat = so->so_snd.sb_hiwat;
+       return (0);
+bad2:
+       sbrelease(&so->so_snd);
+bad:
+       return (ENOBUFS);
+}
+

 /*
  * Allot mbufs to a sockbuf.
 /*
  * Allot mbufs to a sockbuf.

+ * Attempt to scale mbmax so that mbcnt doesn't become limiting
+ * if buffering efficiency is near the normal case.

  */
 sbreserve(sb, cc)
        struct sockbuf *sb;
  */
 sbreserve(sb, cc)
        struct sockbuf *sb;

+       u_long cc;

 {
 
 {
 

-       if (m_reserve((cc*2)/MSIZE) == 0)
+       if (cc > sb_max * MCLBYTES / (MSIZE + MCLBYTES))

                return (0);
        sb->sb_hiwat = cc;
                return (0);
        sb->sb_hiwat = cc;

-       sb->sb_mbmax = cc*2;
+       sb->sb_mbmax = min(cc * 2, sb_max);
+       if (sb->sb_lowat > sb->sb_hiwat)
+               sb->sb_lowat = sb->sb_hiwat;

        return (1);
 }
 
        return (1);
 }
 


@@ -244,36 +369,259 @@ sbrelease(sb)
 {
 
        sbflush(sb);
 {
 
        sbflush(sb);

-       m_release(sb->sb_mbmax/MSIZE);

        sb->sb_hiwat = sb->sb_mbmax = 0;
 }
 
 /*
        sb->sb_hiwat = sb->sb_mbmax = 0;
 }
 
 /*

- * Routines to add (at the end) and remove (from the beginning)
- * data from a mbuf queue.
+ * Routines to add and remove
+ * data from an mbuf queue.
+ *
+ * The routines sbappend() or sbappendrecord() are normally called to
+ * append new mbufs to a socket buffer, after checking that adequate
+ * space is available, comparing the function sbspace() with the amount
+ * of data to be added.  sbappendrecord() differs from sbappend() in
+ * that data supplied is treated as the beginning of a new record.
+ * To place a sender's address, optional access rights, and data in a
+ * socket receive buffer, sbappendaddr() should be used.  To place
+ * access rights and data in a socket receive buffer, sbappendrights()
+ * should be used.  In either case, the new data begins a new record.
+ * Note that unlike sbappend() and sbappendrecord(), these routines check
+ * for the caller that there will be enough space to store the data.
+ * Each fails if there is not enough space, or if it cannot find mbufs
+ * to store additional information in.
+ *
+ * Reliable protocols may use the socket send buffer to hold data
+ * awaiting acknowledgement.  Data is normally copied from a socket
+ * send buffer in a protocol with m_copy for output to a peer,
+ * and then removing the data from the socket buffer with sbdrop()
+ * or sbdroprecord() when the data is acknowledged by the peer.

  */
 
 /*
  */
 
 /*

- * Append mbuf queue m to sockbuf sb.
+ * Append mbuf chain m to the last record in the
+ * socket buffer sb.  The additional space associated
+ * the mbuf chain is recorded in sb.  Empty mbufs are
+ * discarded and mbufs are compacted where possible.

  */
 sbappend(sb, m)
  */
 sbappend(sb, m)

+       struct sockbuf *sb;
+       struct mbuf *m;
+{
+       register struct mbuf *n, *n0;
+
+       if (m == 0)
+               return;
+       if (n = sb->sb_mb) {
+               while (n->m_nextpkt)
+                       n = n->m_nextpkt;
+       }
+       sbcompress(sb, m, n);
+}
+
+#ifdef SOCKBUF_DEBUG
+sbcheck(sb)
+       register struct sockbuf *sb;
+{
+       register struct mbuf *m;
+       register int len = 0, mbcnt = 0;
+
+       for (m = sb->sb_mb; m; m = m->m_next) {
+               len += m->m_len;
+               mbcnt += MSIZE;
+               if (m->m_flags & M_EXT)
+                       mbcnt += m->m_ext.ext_size;
+               if (m->m_nextpkt)
+                       panic("sbcheck nextpkt");
+       }
+       if (len != sb->sb_cc || mbcnt != sb->sb_mbcnt) {
+               printf("cc %d != %d || mbcnt %d != %d\n", len, sb->sb_cc,
+                   mbcnt, sb->sb_mbcnt);
+               panic("sbcheck");
+       }
+}
+#endif
+
+/*
+ * As above, except the mbuf chain
+ * begins a new record.
+ */
+sbappendrecord(sb, m0)
+       register struct sockbuf *sb;
+       register struct mbuf *m0;
+{

        register struct mbuf *m;
        register struct mbuf *m;

+
+       if (m0 == 0)
+               return;
+       if (m = sb->sb_mb)
+               while (m->m_nextpkt)
+                       m = m->m_nextpkt;
+       /*
+        * Put the first mbuf on the queue.
+        * Note this permits zero length records.
+        */
+       sballoc(sb, m0);
+       if (m)
+               m->m_nextpkt = m0;
+       else
+               sb->sb_mb = m0;
+       m = m0->m_next;
+       m0->m_next = 0;
+       sbcompress(sb, m, m0);
+}
+
+/*
+ * As above except that OOB data
+ * is inserted at the beginning of the sockbuf,
+ * but after any other OOB data.
+ */
+sbinsertoob(sb, m0)

        register struct sockbuf *sb;
        register struct sockbuf *sb;

+       register struct mbuf *m0;

 {
 {

-       register struct mbuf *n;
+       register struct mbuf *m;
+       register struct mbuf **mp;
+
+       if (m0 == 0)
+               return;
+       for (mp = &sb->sb_mb; m = *mp; mp = &((*mp)->m_nextpkt)) {
+           again:
+               switch (m->m_type) {
+
+               case MT_OOBDATA:
+                       continue;               /* WANT next train */

 
 

-       n = sb->sb_mb;
+               case MT_CONTROL:
+                       if (m = m->m_next)
+                               goto again;     /* inspect THIS train further */
+               }
+               break;
+       }
+       /*
+        * Put the first mbuf on the queue.
+        * Note this permits zero length records.
+        */
+       m0->m_nextpkt = *mp;
+       *mp = m0;
+       for (m = m0; m; m = m->m_next)
+               sballoc(sb, m);
+}
+
+/*
+ * Append address and data, and optionally, control (ancillary) data
+ * to the receive queue of a socket.  If present,
+ * m0 must include a packet header with total length.
+ * Returns 0 if no space in sockbuf or insufficient mbufs.
+ */
+sbappendaddr(sb, asa, m0, control)
+       register struct sockbuf *sb;
+       struct sockaddr *asa;
+       struct mbuf *m0, *control;
+{
+       register struct mbuf *m, *n;
+       int space = asa->sa_len, eor = 0;
+
+if (m0 && (m0->m_flags & M_PKTHDR) == 0)
+panic("sbappendaddr");
+       if (m0)
+               space += m0->m_pkthdr.len;
+       for (n = control; n; n = n->m_next) {
+               space += n->m_len;
+               if (n->m_next == 0)     /* keep pointer to last control buf */
+                       break;
+       }
+       if (space > sbspace(sb))
+               return (0);
+       if (asa->sa_len > MLEN)
+               return (0);
+       MGET(m, M_DONTWAIT, MT_SONAME);
+       if (m == 0)
+               return (0);
+       m->m_len = asa->sa_len;
+       bcopy((caddr_t)asa, mtod(m, caddr_t), asa->sa_len);

        if (n)
        if (n)

-               while (n->m_next)
+               n->m_next = m0;         /* concatenate data to control */
+       else
+               control = m0;
+       m->m_next = control;
+       for (n = m; n; n = n->m_next) {
+               eor |= n->m_flags & M_EOR;
+               sballoc(sb, n);
+       }
+       m->m_flags |= eor;
+       if (n = sb->sb_mb) {
+               while (n->m_nextpkt)
+                       n = n->m_nextpkt;
+               n->m_nextpkt = m;
+       } else
+               sb->sb_mb = m;
+       return (1);
+}
+
+sbappendcontrol(sb, m0, control)
+       struct sockbuf *sb;
+       struct mbuf *control, *m0;
+{
+       register struct mbuf *m, *n;
+       int space = 0, eor = 0;
+
+       if (control == 0)
+               panic("sbappendcontrol");
+       for (m = control; ; m = m->m_next) {
+               space += m->m_len;
+               if (m->m_next == 0)
+                       break;
+       }
+       n = m;                  /* save pointer to last control buffer */
+       for (m = m0; m; m = m->m_next)
+               space += m->m_len;
+       if (space > sbspace(sb))
+               return (0);
+       n->m_next = m0;                 /* concatenate data to control */
+       for (m = control; m; m = m->m_next) {
+               eor |= m->m_flags & M_EOR;
+               sballoc(sb, m);
+       }
+       control->m_flags |= eor;
+       if (n = sb->sb_mb) {
+               while (n->m_nextpkt)
+                       n = n->m_nextpkt;
+               n->m_nextpkt = control;
+       } else
+               sb->sb_mb = control;
+       return (1);
+}
+
+/*
+ * Compress mbuf chain m into the socket
+ * buffer sb following mbuf n.  If n
+ * is null, the buffer is presumed empty.
+ */
+sbcompress(sb, m, n0)
+       register struct sockbuf *sb;
+       register struct mbuf *m;
+       struct mbuf *n0;
+{
+       register struct mbuf *n = n0;
+       register int eor = 0;
+
+       if (n) {
+               if (n->m_flags & M_EOR)
+                       n = 0;
+               else while (n->m_next)

                        n = n->m_next;
                        n = n->m_next;

+       }

        while (m) {
        while (m) {

-               if (m->m_len == 0 && (int)m->m_act == 0) {
-                       m = m_free(m);
-                       continue;
+               eor |= m->m_flags & M_EOR;
+               if (m->m_len == 0) {
+                       if (eor == 0 || m->m_next || n) {
+                               m = m_free(m);
+                               continue;
+                       }

                }
                }

-               if (n && n->m_off <= MMAXOFF && m->m_off <= MMAXOFF &&
-                  (int)n->m_act == 0 && (int)m->m_act == 0 &&
-                  (n->m_off + n->m_len + m->m_len) <= MMAXOFF) {
+               if (n && (n->m_flags & M_EXT) == 0 &&
+                   (n->m_data + n->m_len + m->m_len) < &n->m_dat[MLEN] &&
+                   n->m_type == m->m_type) {

                        bcopy(mtod(m, caddr_t), mtod(n, caddr_t) + n->m_len,
                            (unsigned)m->m_len);
                        n->m_len += m->m_len;
                        bcopy(mtod(m, caddr_t), mtod(n, caddr_t) + n->m_len,
                            (unsigned)m->m_len);
                        n->m_len += m->m_len;


@@ -281,88 +629,66 @@ sbappend(sb, m)
                        m = m_free(m);
                        continue;
                }
                        m = m_free(m);
                        continue;
                }

-               sballoc(sb, m);
-               if (n == 0)
-                       sb->sb_mb = m;
-               else
+               if (n == 0) {
+                       if (n0)
+                               n0->m_nextpkt = m;
+                       else
+                               sb->sb_mb = m;
+                       n0 = m;
+               } else

                        n->m_next = m;
                        n->m_next = m;

+               sballoc(sb, m);

                n = m;
                n = m;

+               /*m->m_flags &= ~M_EOR;*/

                m = m->m_next;
                n->m_next = 0;
        }
                m = m->m_next;
                n->m_next = 0;
        }

-}
-
-/*
- * Append data and address.
- * Return 0 if no space in sockbuf or if
- * can't get mbuf to stuff address in.
- */
-sbappendaddr(sb, asa, m0)
-       struct sockbuf *sb;
-       struct sockaddr *asa;
-       struct mbuf *m0;
-{
-       struct sockaddr *msa;
-       register struct mbuf *m;
-       register int len = sizeof (struct sockaddr);
-
-       m = m0;
-       if (m == 0)
-               panic("sbappendaddr");
-       for (;;) {
-               len += m->m_len;
-               if (m->m_next == 0) {
-                       m->m_act = (struct mbuf *)1;
-                       break;
-               }
-               m = m->m_next;
+       if (eor) {
+               if (n0)
+                       n0->m_flags |= eor;
+               else
+                       panic("sbcompress");

        }
        }

-       if (len > sbspace(sb))
-               return (0);
-       m = m_get(M_DONTWAIT);
-       if (m == 0)
-               return (0);
-       m->m_off = MMINOFF;
-       m->m_len = sizeof (struct sockaddr);
-       msa = mtod(m, struct sockaddr *);
-       *msa = *asa;
-       m->m_act = (struct mbuf *)1;
-       sbappend(sb, m);
-       sbappend(sb, m0);
-       return (1);

 }
 
 /*
 }
 
 /*

- * Free all mbufs on a sockbuf mbuf chain.
- * Check that resource allocations return to 0.
+ * Free all mbufs in a sockbuf.
+ * Check that all resources are reclaimed.

  */
 sbflush(sb)
  */
 sbflush(sb)

-       struct sockbuf *sb;
+       register struct sockbuf *sb;

 {
 
        if (sb->sb_flags & SB_LOCK)
                panic("sbflush");
 {
 
        if (sb->sb_flags & SB_LOCK)
                panic("sbflush");

-       if (sb->sb_cc)
-               sbdrop(sb, sb->sb_cc);
-       if (sb->sb_cc || sb->sb_mbcnt || sb->sb_mb)
+       while (sb->sb_mbcnt)
+               sbdrop(sb, (int)sb->sb_cc);
+       if (sb->sb_cc || sb->sb_mb)

                panic("sbflush 2");
 }
 
 /*
                panic("sbflush 2");
 }
 
 /*

- * Drop data from (the front of) a sockbuf chain.
+ * Drop data from (the front of) a sockbuf.

  */
 sbdrop(sb, len)
        register struct sockbuf *sb;
        register int len;
 {
  */
 sbdrop(sb, len)
        register struct sockbuf *sb;
        register int len;
 {

-       register struct mbuf *m = sb->sb_mb, *mn;
+       register struct mbuf *m, *mn;
+       struct mbuf *next;

 
 

+       next = (m = sb->sb_mb) ? m->m_nextpkt : 0;

        while (len > 0) {
        while (len > 0) {

-               if (m == 0)
-                       panic("sbdrop");
+               if (m == 0) {
+                       if (next == 0)
+                               panic("sbdrop");
+                       m = next;
+                       next = m->m_nextpkt;
+                       continue;
+               }

                if (m->m_len > len) {
                        m->m_len -= len;
                if (m->m_len > len) {
                        m->m_len -= len;

-                       m->m_off += len;
+                       m->m_data += len;

                        sb->sb_cc -= len;
                        break;
                }
                        sb->sb_cc -= len;
                        break;
                }


@@ -371,5 +697,33 @@ sbdrop(sb, len)
                MFREE(m, mn);
                m = mn;
        }
                MFREE(m, mn);
                m = mn;
        }

-       sb->sb_mb = m;
+       while (m && m->m_len == 0) {
+               sbfree(sb, m);
+               MFREE(m, mn);
+               m = mn;
+       }
+       if (m) {
+               sb->sb_mb = m;
+               m->m_nextpkt = next;
+       } else
+               sb->sb_mb = next;
+}
+
+/*
+ * Drop a record off the front of a sockbuf
+ * and move the next record to the front.
+ */
+sbdroprecord(sb)
+       register struct sockbuf *sb;
+{
+       register struct mbuf *m, *mn;
+
+       m = sb->sb_mb;
+       if (m) {
+               sb->sb_mb = m->m_nextpkt;
+               do {
+                       sbfree(sb, m);
+                       MFREE(m, mn);
+               } while (m = mn);
+       }

 }
 }