[BACK]Return to uipc_socket2.c CVS log [TXT][DIR] Up to [cvs.NetBSD.org] / src / sys / kern

Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.

Diff for /src/sys/kern/uipc_socket2.c between version 1.85.8.1 and 1.85.8.2

version 1.85.8.1, 2007/08/02 02:42:40 version 1.85.8.2, 2007/08/02 02:42:41
Line 0 
Line 1 
   /*      $NetBSD$        */
   
   /*
    * Copyright (c) 1982, 1986, 1988, 1990, 1993
    *      The Regents of the University of California.  All rights reserved.
    *
    * Redistribution and use in source and binary forms, with or without
    * modification, are permitted provided that the following conditions
    * are met:
    * 1. Redistributions of source code must retain the above copyright
    *    notice, this list of conditions and the following disclaimer.
    * 2. Redistributions in binary form must reproduce the above copyright
    *    notice, this list of conditions and the following disclaimer in the
    *    documentation and/or other materials provided with the distribution.
    * 3. Neither the name of the University nor the names of its contributors
    *    may be used to endorse or promote products derived from this software
    *    without specific prior written permission.
    *
    * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
    * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
    * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
    * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
    * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
    * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
    * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
    * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
    * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
    * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
    * SUCH DAMAGE.
    *
    *      @(#)uipc_socket2.c      8.2 (Berkeley) 2/14/95
    */
   
   #include <sys/cdefs.h>
   __KERNEL_RCSID(0, "$NetBSD$");
   
   #include "opt_mbuftrace.h"
   #include "opt_sb_max.h"
   
   #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/poll.h>
   #include <sys/socket.h>
   #include <sys/socketvar.h>
   #include <sys/signalvar.h>
   #include <sys/kauth.h>
   
   /*
    * Primitive routines for operating on sockets and socket buffers
    */
   
   /* strings for sleep message: */
   const char      netcon[] = "netcon";
   const char      netcls[] = "netcls";
   const char      netio[] = "netio";
   const char      netlck[] = "netlck";
   
   u_long  sb_max = SB_MAX;        /* maximum socket buffer size */
   static u_long sb_max_adj;       /* adjusted sb_max */
   
   /*
    * Procedures to manipulate state flags of socket
    * 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,
    * and soisdisconnected() is called when the connection to the peer
    * is totally severed.  The semantics of these routines are such that
    * connectionless protocols can call soisconnected() and soisdisconnected()
    * only, bypassing the in-progress calls when setting up a ``connection''
    * takes no time.
    *
    * 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
    * cause software-interrupt process scheduling.
    */
   
   void
   soisconnecting(struct socket *so)
   {
   
           so->so_state &= ~(SS_ISCONNECTED|SS_ISDISCONNECTING);
           so->so_state |= SS_ISCONNECTING;
   }
   
   void
   soisconnected(struct socket *so)
   {
           struct socket   *head;
   
           head = so->so_head;
           so->so_state &= ~(SS_ISCONNECTING|SS_ISDISCONNECTING|SS_ISCONFIRMING);
           so->so_state |= SS_ISCONNECTED;
           if (head && soqremque(so, 0)) {
                   soqinsque(head, so, 1);
                   sorwakeup(head);
                   wakeup((void *)&head->so_timeo);
           } else {
                   wakeup((void *)&so->so_timeo);
                   sorwakeup(so);
                   sowwakeup(so);
           }
   }
   
   void
   soisdisconnecting(struct socket *so)
   {
   
           so->so_state &= ~SS_ISCONNECTING;
           so->so_state |= (SS_ISDISCONNECTING|SS_CANTRCVMORE|SS_CANTSENDMORE);
           wakeup((void *)&so->so_timeo);
           sowwakeup(so);
           sorwakeup(so);
   }
   
   void
   soisdisconnected(struct socket *so)
   {
   
           so->so_state &= ~(SS_ISCONNECTING|SS_ISCONNECTED|SS_ISDISCONNECTING);
           so->so_state |= (SS_CANTRCVMORE|SS_CANTSENDMORE|SS_ISDISCONNECTED);
           wakeup((void *)&so->so_timeo);
           sowwakeup(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, SS_ISCONFIRMING, or SS_ISCONNECTED.
    */
   struct socket *
   sonewconn(struct socket *head, int connstatus)
   {
           struct socket   *so;
           int             soqueue;
   
           soqueue = connstatus ? 1 : 0;
           if (head->so_qlen + head->so_q0len > 3 * head->so_qlimit / 2)
                   return ((struct socket *)0);
           so = pool_get(&socket_pool, PR_NOWAIT);
           if (so == NULL)
                   return (NULL);
           memset((void *)so, 0, 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;
           so->so_send = head->so_send;
           so->so_receive = head->so_receive;
           so->so_uidinfo = head->so_uidinfo;
   #ifdef MBUFTRACE
           so->so_mowner = head->so_mowner;
           so->so_rcv.sb_mowner = head->so_rcv.sb_mowner;
           so->so_snd.sb_mowner = head->so_snd.sb_mowner;
   #endif
           (void) soreserve(so, head->so_snd.sb_hiwat, head->so_rcv.sb_hiwat);
           so->so_snd.sb_lowat = head->so_snd.sb_lowat;
           so->so_rcv.sb_lowat = head->so_rcv.sb_lowat;
           so->so_rcv.sb_timeo = head->so_rcv.sb_timeo;
           so->so_snd.sb_timeo = head->so_snd.sb_timeo;
           so->so_rcv.sb_flags |= head->so_rcv.sb_flags & SB_AUTOSIZE;
           so->so_snd.sb_flags |= head->so_snd.sb_flags & SB_AUTOSIZE;
           soqinsque(head, so, soqueue);
           if ((*so->so_proto->pr_usrreq)(so, PRU_ATTACH,
               (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0,
               (struct lwp *)0)) {
                   (void) soqremque(so, soqueue);
                   pool_put(&socket_pool, so);
                   return (NULL);
           }
           if (connstatus) {
                   sorwakeup(head);
                   wakeup((void *)&head->so_timeo);
                   so->so_state |= connstatus;
           }
           return (so);
   }
   
   void
   soqinsque(struct socket *head, struct socket *so, int q)
   {
   
   #ifdef DIAGNOSTIC
           if (so->so_onq != NULL)
                   panic("soqinsque");
   #endif
   
           so->so_head = head;
           if (q == 0) {
                   head->so_q0len++;
                   so->so_onq = &head->so_q0;
           } else {
                   head->so_qlen++;
                   so->so_onq = &head->so_q;
           }
           TAILQ_INSERT_TAIL(so->so_onq, so, so_qe);
   }
   
   int
   soqremque(struct socket *so, int q)
   {
           struct socket   *head;
   
           head = so->so_head;
           if (q == 0) {
                   if (so->so_onq != &head->so_q0)
                           return (0);
                   head->so_q0len--;
           } else {
                   if (so->so_onq != &head->so_q)
                           return (0);
                   head->so_qlen--;
           }
           TAILQ_REMOVE(so->so_onq, so, so_qe);
           so->so_onq = NULL;
           so->so_head = NULL;
           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
    * informs the system that no more data is to be sent, by the protocol
    * code (in case PRU_SHUTDOWN).  Socantrcvmore indicates that no more data
    * will be received, and will normally be applied to the socket by a
    * protocol when it detects that the peer will send no more data.
    * Data queued for reading in the socket may yet be read.
    */
   
   void
   socantsendmore(struct socket *so)
   {
   
           so->so_state |= SS_CANTSENDMORE;
           sowwakeup(so);
   }
   
   void
   socantrcvmore(struct socket *so)
   {
   
           so->so_state |= SS_CANTRCVMORE;
           sorwakeup(so);
   }
   
   /*
    * Wait for data to arrive at/drain from a socket buffer.
    */
   int
   sbwait(struct sockbuf *sb)
   {
   
           sb->sb_flags |= SB_WAIT;
           return (tsleep((void *)&sb->sb_cc,
               (sb->sb_flags & SB_NOINTR) ? PSOCK : PSOCK | PCATCH, netio,
               sb->sb_timeo));
   }
   
   /*
    * Lock a sockbuf already known to be locked;
    * return any error returned from sleep (EINTR).
    */
   int
   sb_lock(struct sockbuf *sb)
   {
           int     error;
   
           while (sb->sb_flags & SB_LOCK) {
                   sb->sb_flags |= SB_WANT;
                   error = tsleep((void *)&sb->sb_flags,
                       (sb->sb_flags & SB_NOINTR) ?  PSOCK : PSOCK|PCATCH,
                       netlck, 0);
                   if (error)
                           return (error);
           }
           sb->sb_flags |= SB_LOCK;
           return (0);
   }
   
   /*
    * Wakeup processes waiting on a socket buffer.
    * Do asynchronous notification via SIGIO
    * if the socket buffer has the SB_ASYNC flag set.
    */
   void
   sowakeup(struct socket *so, struct sockbuf *sb, int code)
   {
           selnotify(&sb->sb_sel, 0);
           sb->sb_flags &= ~SB_SEL;
           if (sb->sb_flags & SB_WAIT) {
                   sb->sb_flags &= ~SB_WAIT;
                   wakeup((void *)&sb->sb_cc);
           }
           if (sb->sb_flags & SB_ASYNC) {
                   int band;
                   if (code == POLL_IN)
                           band = POLLIN|POLLRDNORM;
                   else
                           band = POLLOUT|POLLWRNORM;
                   fownsignal(so->so_pgid, SIGIO, code, band, so);
           }
           if (sb->sb_flags & SB_UPCALL)
                   (*so->so_upcall)(so, so->so_upcallarg, M_DONTWAIT);
   }
   
   /*
    * Socket buffer (struct sockbuf) utility routines.
    *
    * Each socket contains two socket buffers: one for sending data and
    * one for receiving data.  Each buffer contains a queue of mbufs,
    * information about the number of mbufs and amount of data in the
    * queue, and other fields allowing poll() statements and notification
    * on data availability to be implemented.
    *
    * 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:
    *
    * 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_CONTROL).
    * 3. If a name or rights record exists, then it must be followed by
    *    a data record, perhaps of zero length.
    *
    * 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.
    */
   
   int
   sb_max_set(u_long new_sbmax)
   {
           int s;
   
           if (new_sbmax < (16 * 1024))
                   return (EINVAL);
   
           s = splsoftnet();
           sb_max = new_sbmax;
           sb_max_adj = (u_quad_t)new_sbmax * MCLBYTES / (MSIZE + MCLBYTES);
           splx(s);
   
           return (0);
   }
   
   int
   soreserve(struct socket *so, u_long sndcc, u_long rcvcc)
   {
           /*
            * there's at least one application (a configure script of screen)
            * which expects a fifo is writable even if it has "some" bytes
            * in its buffer.
            * so we want to make sure (hiwat - lowat) >= (some bytes).
            *
            * PIPE_BUF here is an arbitrary value chosen as (some bytes) above.
            * we expect it's large enough for such applications.
            */
           u_long  lowat = MAX(sock_loan_thresh, MCLBYTES);
           u_long  hiwat = lowat + PIPE_BUF;
   
           if (sndcc < hiwat)
                   sndcc = hiwat;
           if (sbreserve(&so->so_snd, sndcc, so) == 0)
                   goto bad;
           if (sbreserve(&so->so_rcv, rcvcc, so) == 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 = lowat;
           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, so);
    bad:
           return (ENOBUFS);
   }
   
   /*
    * Allot mbufs to a sockbuf.
    * Attempt to scale mbmax so that mbcnt doesn't become limiting
    * if buffering efficiency is near the normal case.
    */
   int
   sbreserve(struct sockbuf *sb, u_long cc, struct socket *so)
   {
           struct lwp *l = curlwp; /* XXX */
           rlim_t maxcc;
           struct uidinfo *uidinfo;
   
           KDASSERT(sb_max_adj != 0);
           if (cc == 0 || cc > sb_max_adj)
                   return (0);
           if (so) {
                   if (l && kauth_cred_geteuid(l->l_cred) == so->so_uidinfo->ui_uid)
                           maxcc = l->l_proc->p_rlimit[RLIMIT_SBSIZE].rlim_cur;
                   else
                           maxcc = RLIM_INFINITY;
                   uidinfo = so->so_uidinfo;
           } else {
                   uidinfo = uid_find(0);  /* XXX: nothing better */
                   maxcc = RLIM_INFINITY;
           }
           if (!chgsbsize(uidinfo, &sb->sb_hiwat, cc, maxcc))
                   return 0;
           sb->sb_mbmax = min(cc * 2, sb_max);
           if (sb->sb_lowat > sb->sb_hiwat)
                   sb->sb_lowat = sb->sb_hiwat;
           return (1);
   }
   
   /*
    * Free mbufs held by a socket, and reserved mbuf space.
    */
   void
   sbrelease(struct sockbuf *sb, struct socket *so)
   {
   
           sbflush(sb);
           (void)chgsbsize(so->so_uidinfo, &sb->sb_hiwat, 0,
               RLIM_INFINITY);
           sb->sb_mbmax = 0;
   }
   
   /*
    * 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.
    */
   
   #ifdef SOCKBUF_DEBUG
   void
   sblastrecordchk(struct sockbuf *sb, const char *where)
   {
           struct mbuf *m = sb->sb_mb;
   
           while (m && m->m_nextpkt)
                   m = m->m_nextpkt;
   
           if (m != sb->sb_lastrecord) {
                   printf("sblastrecordchk: sb_mb %p sb_lastrecord %p last %p\n",
                       sb->sb_mb, sb->sb_lastrecord, m);
                   printf("packet chain:\n");
                   for (m = sb->sb_mb; m != NULL; m = m->m_nextpkt)
                           printf("\t%p\n", m);
                   panic("sblastrecordchk from %s", where);
           }
   }
   
   void
   sblastmbufchk(struct sockbuf *sb, const char *where)
   {
           struct mbuf *m = sb->sb_mb;
           struct mbuf *n;
   
           while (m && m->m_nextpkt)
                   m = m->m_nextpkt;
   
           while (m && m->m_next)
                   m = m->m_next;
   
           if (m != sb->sb_mbtail) {
                   printf("sblastmbufchk: sb_mb %p sb_mbtail %p last %p\n",
                       sb->sb_mb, sb->sb_mbtail, m);
                   printf("packet tree:\n");
                   for (m = sb->sb_mb; m != NULL; m = m->m_nextpkt) {
                           printf("\t");
                           for (n = m; n != NULL; n = n->m_next)
                                   printf("%p ", n);
                           printf("\n");
                   }
                   panic("sblastmbufchk from %s", where);
           }
   }
   #endif /* SOCKBUF_DEBUG */
   
   /*
    * Link a chain of records onto a socket buffer
    */
   #define SBLINKRECORDCHAIN(sb, m0, mlast)                                \
   do {                                                                    \
           if ((sb)->sb_lastrecord != NULL)                                \
                   (sb)->sb_lastrecord->m_nextpkt = (m0);                  \
           else                                                            \
                   (sb)->sb_mb = (m0);                                     \
           (sb)->sb_lastrecord = (mlast);                                  \
   } while (/*CONSTCOND*/0)
   
   
   #define SBLINKRECORD(sb, m0)                                            \
       SBLINKRECORDCHAIN(sb, m0, m0)
   
   /*
    * 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.
    */
   void
   sbappend(struct sockbuf *sb, struct mbuf *m)
   {
           struct mbuf     *n;
   
           if (m == 0)
                   return;
   
   #ifdef MBUFTRACE
           m_claimm(m, sb->sb_mowner);
   #endif
   
           SBLASTRECORDCHK(sb, "sbappend 1");
   
           if ((n = sb->sb_lastrecord) != NULL) {
                   /*
                    * XXX Would like to simply use sb_mbtail here, but
                    * XXX I need to verify that I won't miss an EOR that
                    * XXX way.
                    */
                   do {
                           if (n->m_flags & M_EOR) {
                                   sbappendrecord(sb, m); /* XXXXXX!!!! */
                                   return;
                           }
                   } while (n->m_next && (n = n->m_next));
           } else {
                   /*
                    * If this is the first record in the socket buffer, it's
                    * also the last record.
                    */
                   sb->sb_lastrecord = m;
           }
           sbcompress(sb, m, n);
           SBLASTRECORDCHK(sb, "sbappend 2");
   }
   
   /*
    * This version of sbappend() should only be used when the caller
    * absolutely knows that there will never be more than one record
    * in the socket buffer, that is, a stream protocol (such as TCP).
    */
   void
   sbappendstream(struct sockbuf *sb, struct mbuf *m)
   {
   
           KDASSERT(m->m_nextpkt == NULL);
           KASSERT(sb->sb_mb == sb->sb_lastrecord);
   
           SBLASTMBUFCHK(sb, __func__);
   
   #ifdef MBUFTRACE
           m_claimm(m, sb->sb_mowner);
   #endif
   
           sbcompress(sb, m, sb->sb_mbtail);
   
           sb->sb_lastrecord = sb->sb_mb;
           SBLASTRECORDCHK(sb, __func__);
   }
   
   #ifdef SOCKBUF_DEBUG
   void
   sbcheck(struct sockbuf *sb)
   {
           struct mbuf     *m;
           u_long          len, mbcnt;
   
           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 %lu != %lu || mbcnt %lu != %lu\n", len, sb->sb_cc,
                       mbcnt, sb->sb_mbcnt);
                   panic("sbcheck");
           }
   }
   #endif
   
   /*
    * As above, except the mbuf chain
    * begins a new record.
    */
   void
   sbappendrecord(struct sockbuf *sb, struct mbuf *m0)
   {
           struct mbuf     *m;
   
           if (m0 == 0)
                   return;
   
   #ifdef MBUFTRACE
           m_claimm(m0, sb->sb_mowner);
   #endif
           /*
            * Put the first mbuf on the queue.
            * Note this permits zero length records.
            */
           sballoc(sb, m0);
           SBLASTRECORDCHK(sb, "sbappendrecord 1");
           SBLINKRECORD(sb, m0);
           m = m0->m_next;
           m0->m_next = 0;
           if (m && (m0->m_flags & M_EOR)) {
                   m0->m_flags &= ~M_EOR;
                   m->m_flags |= M_EOR;
           }
           sbcompress(sb, m, m0);
           SBLASTRECORDCHK(sb, "sbappendrecord 2");
   }
   
   /*
    * As above except that OOB data
    * is inserted at the beginning of the sockbuf,
    * but after any other OOB data.
    */
   void
   sbinsertoob(struct sockbuf *sb, struct mbuf *m0)
   {
           struct mbuf     *m, **mp;
   
           if (m0 == 0)
                   return;
   
           SBLASTRECORDCHK(sb, "sbinsertoob 1");
   
           for (mp = &sb->sb_mb; (m = *mp) != NULL; mp = &((*mp)->m_nextpkt)) {
               again:
                   switch (m->m_type) {
   
                   case MT_OOBDATA:
                           continue;               /* WANT next train */
   
                   case MT_CONTROL:
                           if ((m = m->m_next) != NULL)
                                   goto again;     /* inspect THIS train further */
                   }
                   break;
           }
           /*
            * Put the first mbuf on the queue.
            * Note this permits zero length records.
            */
           sballoc(sb, m0);
           m0->m_nextpkt = *mp;
           if (*mp == NULL) {
                   /* m0 is actually the new tail */
                   sb->sb_lastrecord = m0;
           }
           *mp = m0;
           m = m0->m_next;
           m0->m_next = 0;
           if (m && (m0->m_flags & M_EOR)) {
                   m0->m_flags &= ~M_EOR;
                   m->m_flags |= M_EOR;
           }
           sbcompress(sb, m, m0);
           SBLASTRECORDCHK(sb, "sbinsertoob 2");
   }
   
   /*
    * 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.
    */
   int
   sbappendaddr(struct sockbuf *sb, const struct sockaddr *asa, struct mbuf *m0,
           struct mbuf *control)
   {
           struct mbuf     *m, *n, *nlast;
           int             space, len;
   
           space = asa->sa_len;
   
           if (m0 != NULL) {
                   if ((m0->m_flags & M_PKTHDR) == 0)
                           panic("sbappendaddr");
                   space += m0->m_pkthdr.len;
   #ifdef MBUFTRACE
                   m_claimm(m0, sb->sb_mowner);
   #endif
           }
           for (n = control; n; n = n->m_next) {
                   space += n->m_len;
                   MCLAIM(n, sb->sb_mowner);
                   if (n->m_next == 0)     /* keep pointer to last control buf */
                           break;
           }
           if (space > sbspace(sb))
                   return (0);
           MGET(m, M_DONTWAIT, MT_SONAME);
           if (m == 0)
                   return (0);
           MCLAIM(m, sb->sb_mowner);
           /*
            * XXX avoid 'comparison always true' warning which isn't easily
            * avoided.
            */
           len = asa->sa_len;
           if (len > MLEN) {
                   MEXTMALLOC(m, asa->sa_len, M_NOWAIT);
                   if ((m->m_flags & M_EXT) == 0) {
                           m_free(m);
                           return (0);
                   }
           }
           m->m_len = asa->sa_len;
           memcpy(mtod(m, void *), asa, asa->sa_len);
           if (n)
                   n->m_next = m0;         /* concatenate data to control */
           else
                   control = m0;
           m->m_next = control;
   
           SBLASTRECORDCHK(sb, "sbappendaddr 1");
   
           for (n = m; n->m_next != NULL; n = n->m_next)
                   sballoc(sb, n);
           sballoc(sb, n);
           nlast = n;
           SBLINKRECORD(sb, m);
   
           sb->sb_mbtail = nlast;
           SBLASTMBUFCHK(sb, "sbappendaddr");
   
           SBLASTRECORDCHK(sb, "sbappendaddr 2");
   
           return (1);
   }
   
   /*
    * Helper for sbappendchainaddr: prepend a struct sockaddr* to
    * an mbuf chain.
    */
   static inline struct mbuf *
   m_prepend_sockaddr(struct sockbuf *sb, struct mbuf *m0,
                      const struct sockaddr *asa)
   {
           struct mbuf *m;
           const int salen = asa->sa_len;
   
           /* only the first in each chain need be a pkthdr */
           MGETHDR(m, M_DONTWAIT, MT_SONAME);
           if (m == 0)
                   return (0);
           MCLAIM(m, sb->sb_mowner);
   #ifdef notyet
           if (salen > MHLEN) {
                   MEXTMALLOC(m, salen, M_NOWAIT);
                   if ((m->m_flags & M_EXT) == 0) {
                           m_free(m);
                           return (0);
                   }
           }
   #else
           KASSERT(salen <= MHLEN);
   #endif
           m->m_len = salen;
           memcpy(mtod(m, void *), asa, salen);
           m->m_next = m0;
           m->m_pkthdr.len = salen + m0->m_pkthdr.len;
   
           return m;
   }
   
   int
   sbappendaddrchain(struct sockbuf *sb, const struct sockaddr *asa,
                     struct mbuf *m0, int sbprio)
   {
           int space;
           struct mbuf *m, *n, *n0, *nlast;
           int error;
   
           /*
            * XXX sbprio reserved for encoding priority of this* request:
            *  SB_PRIO_NONE --> honour normal sb limits
            *  SB_PRIO_ONESHOT_OVERFLOW --> if socket has any space,
            *      take whole chain. Intended for large requests
            *      that should be delivered atomically (all, or none).
            * SB_PRIO_OVERDRAFT -- allow a small (2*MLEN) overflow
            *       over normal socket limits, for messages indicating
            *       buffer overflow in earlier normal/lower-priority messages
            * SB_PRIO_BESTEFFORT -->  ignore limits entirely.
            *       Intended for  kernel-generated messages only.
            *        Up to generator to avoid total mbuf resource exhaustion.
            */
           (void)sbprio;
   
           if (m0 && (m0->m_flags & M_PKTHDR) == 0)
                   panic("sbappendaddrchain");
   
           space = sbspace(sb);
   
   #ifdef notyet
           /*
            * Enforce SB_PRIO_* limits as described above.
            */
   #endif
   
           n0 = NULL;
           nlast = NULL;
           for (m = m0; m; m = m->m_nextpkt) {
                   struct mbuf *np;
   
   #ifdef MBUFTRACE
                   m_claimm(m, sb->sb_mowner);
   #endif
   
                   /* Prepend sockaddr to this record (m) of input chain m0 */
                   n = m_prepend_sockaddr(sb, m, asa);
                   if (n == NULL) {
                           error = ENOBUFS;
                           goto bad;
                   }
   
                   /* Append record (asa+m) to end of new chain n0 */
                   if (n0 == NULL) {
                           n0 = n;
                   } else {
                           nlast->m_nextpkt = n;
                   }
                   /* Keep track of last record on new chain */
                   nlast = n;
   
                   for (np = n; np; np = np->m_next)
                           sballoc(sb, np);
           }
   
           SBLASTRECORDCHK(sb, "sbappendaddrchain 1");
   
           /* Drop the entire chain of (asa+m) records onto the socket */
           SBLINKRECORDCHAIN(sb, n0, nlast);
   
           SBLASTRECORDCHK(sb, "sbappendaddrchain 2");
   
           for (m = nlast; m->m_next; m = m->m_next)
                   ;
           sb->sb_mbtail = m;
           SBLASTMBUFCHK(sb, "sbappendaddrchain");
   
           return (1);
   
   bad:
           /*
            * On error, free the prepended addreseses. For consistency
            * with sbappendaddr(), leave it to our caller to free
            * the input record chain passed to us as m0.
            */
           while ((n = n0) != NULL) {
                   struct mbuf *np;
   
                   /* Undo the sballoc() of this record */
                   for (np = n; np; np = np->m_next)
                           sbfree(sb, np);
   
                   n0 = n->m_nextpkt;      /* iterate at next prepended address */
                   MFREE(n, np);           /* free prepended address (not data) */
           }
           return 0;
   }
   
   
   int
   sbappendcontrol(struct sockbuf *sb, struct mbuf *m0, struct mbuf *control)
   {
           struct mbuf     *m, *mlast, *n;
           int             space;
   
           space = 0;
           if (control == 0)
                   panic("sbappendcontrol");
           for (m = control; ; m = m->m_next) {
                   space += m->m_len;
                   MCLAIM(m, sb->sb_mowner);
                   if (m->m_next == 0)
                           break;
           }
           n = m;                  /* save pointer to last control buffer */
           for (m = m0; m; m = m->m_next) {
                   MCLAIM(m, sb->sb_mowner);
                   space += m->m_len;
           }
           if (space > sbspace(sb))
                   return (0);
           n->m_next = m0;                 /* concatenate data to control */
   
           SBLASTRECORDCHK(sb, "sbappendcontrol 1");
   
           for (m = control; m->m_next != NULL; m = m->m_next)
                   sballoc(sb, m);
           sballoc(sb, m);
           mlast = m;
           SBLINKRECORD(sb, control);
   
           sb->sb_mbtail = mlast;
           SBLASTMBUFCHK(sb, "sbappendcontrol");
   
           SBLASTRECORDCHK(sb, "sbappendcontrol 2");
   
           return (1);
   }
   
   /*
    * Compress mbuf chain m into the socket
    * buffer sb following mbuf n.  If n
    * is null, the buffer is presumed empty.
    */
   void
   sbcompress(struct sockbuf *sb, struct mbuf *m, struct mbuf *n)
   {
           int             eor;
           struct mbuf     *o;
   
           eor = 0;
           while (m) {
                   eor |= m->m_flags & M_EOR;
                   if (m->m_len == 0 &&
                       (eor == 0 ||
                        (((o = m->m_next) || (o = n)) &&
                         o->m_type == m->m_type))) {
                           if (sb->sb_lastrecord == m)
                                   sb->sb_lastrecord = m->m_next;
                           m = m_free(m);
                           continue;
                   }
                   if (n && (n->m_flags & M_EOR) == 0 &&
                       /* M_TRAILINGSPACE() checks buffer writeability */
                       m->m_len <= MCLBYTES / 4 && /* XXX Don't copy too much */
                       m->m_len <= M_TRAILINGSPACE(n) &&
                       n->m_type == m->m_type) {
                           memcpy(mtod(n, char *) + n->m_len, mtod(m, void *),
                               (unsigned)m->m_len);
                           n->m_len += m->m_len;
                           sb->sb_cc += m->m_len;
                           m = m_free(m);
                           continue;
                   }
                   if (n)
                           n->m_next = m;
                   else
                           sb->sb_mb = m;
                   sb->sb_mbtail = m;
                   sballoc(sb, m);
                   n = m;
                   m->m_flags &= ~M_EOR;
                   m = m->m_next;
                   n->m_next = 0;
           }
           if (eor) {
                   if (n)
                           n->m_flags |= eor;
                   else
                           printf("semi-panic: sbcompress\n");
           }
           SBLASTMBUFCHK(sb, __func__);
   }
   
   /*
    * Free all mbufs in a sockbuf.
    * Check that all resources are reclaimed.
    */
   void
   sbflush(struct sockbuf *sb)
   {
   
           KASSERT((sb->sb_flags & SB_LOCK) == 0);
   
           while (sb->sb_mbcnt)
                   sbdrop(sb, (int)sb->sb_cc);
   
           KASSERT(sb->sb_cc == 0);
           KASSERT(sb->sb_mb == NULL);
           KASSERT(sb->sb_mbtail == NULL);
           KASSERT(sb->sb_lastrecord == NULL);
   }
   
   /*
    * Drop data from (the front of) a sockbuf.
    */
   void
   sbdrop(struct sockbuf *sb, int len)
   {
           struct mbuf     *m, *mn, *next;
   
           next = (m = sb->sb_mb) ? m->m_nextpkt : 0;
           while (len > 0) {
                   if (m == 0) {
                           if (next == 0)
                                   panic("sbdrop");
                           m = next;
                           next = m->m_nextpkt;
                           continue;
                   }
                   if (m->m_len > len) {
                           m->m_len -= len;
                           m->m_data += len;
                           sb->sb_cc -= len;
                           break;
                   }
                   len -= m->m_len;
                   sbfree(sb, m);
                   MFREE(m, mn);
                   m = mn;
           }
           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;
           /*
            * First part is an inline SB_EMPTY_FIXUP().  Second part
            * makes sure sb_lastrecord is up-to-date if we dropped
            * part of the last record.
            */
           m = sb->sb_mb;
           if (m == NULL) {
                   sb->sb_mbtail = NULL;
                   sb->sb_lastrecord = NULL;
           } else if (m->m_nextpkt == NULL)
                   sb->sb_lastrecord = m;
   }
   
   /*
    * Drop a record off the front of a sockbuf
    * and move the next record to the front.
    */
   void
   sbdroprecord(struct sockbuf *sb)
   {
           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) != NULL);
           }
           SB_EMPTY_FIXUP(sb);
   }
   
   /*
    * Create a "control" mbuf containing the specified data
    * with the specified type for presentation on a socket buffer.
    */
   struct mbuf *
   sbcreatecontrol(void *p, int size, int type, int level)
   {
           struct cmsghdr  *cp;
           struct mbuf     *m;
   
           if (CMSG_SPACE(size) > MCLBYTES) {
                   printf("sbcreatecontrol: message too large %d\n", size);
                   return NULL;
           }
   
           if ((m = m_get(M_DONTWAIT, MT_CONTROL)) == NULL)
                   return ((struct mbuf *) NULL);
           if (CMSG_SPACE(size) > MLEN) {
                   MCLGET(m, M_DONTWAIT);
                   if ((m->m_flags & M_EXT) == 0) {
                           m_free(m);
                           return NULL;
                   }
           }
           cp = mtod(m, struct cmsghdr *);
           memcpy(CMSG_DATA(cp), p, size);
           m->m_len = CMSG_SPACE(size);
           cp->cmsg_len = CMSG_LEN(size);
           cp->cmsg_level = level;
           cp->cmsg_type = type;
           return (m);
   }
Legend:
Removed from v.1.85.8.1  
changed lines
  Added in v.1.85.8.2
CVSweb <webmaster@jp.NetBSD.org>