xnu-792.22.5.tar.gz

[apple/xnu.git] / bsd / netinet / tcp_output.c

diff --git a/bsd/netinet/tcp_output.c b/bsd/netinet/tcp_output.c
index 56043b0597ff688a4cd57d48dfdd3fc2d098e54e..b81dbd53a290a0cdb360e9d0f3f12bc68bb2e076 100644 (file)
--- a/bsd/netinet/tcp_output.c
+++ b/bsd/netinet/tcp_output.c
@@ -1,23 +1,29 @@
 /*
  * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
  *
 /*
  * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
  *

- * @APPLE_LICENSE_HEADER_START@
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_START@

  * 
  * 

- * The contents of this file constitute Original Code as defined in and
- * are subject to the Apple Public Source License Version 1.1 (the
- * "License").  You may not use this file except in compliance with the
- * License.  Please obtain a copy of the License at
- * http://www.apple.com/publicsource and read it before using this file.
+ * This file contains Original Code and/or Modifications of Original Code
+ * as defined in and that are subject to the Apple Public Source License
+ * Version 2.0 (the 'License'). You may not use this file except in
+ * compliance with the License. The rights granted to you under the License
+ * may not be used to create, or enable the creation or redistribution of,
+ * unlawful or unlicensed copies of an Apple operating system, or to
+ * circumvent, violate, or enable the circumvention or violation of, any
+ * terms of an Apple operating system software license agreement.

  * 
  * 

- * This Original Code and all software distributed under the License are
- * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
+ * Please obtain a copy of the License at
+ * http://www.opensource.apple.com/apsl/ and read it before using this file.
+ * 
+ * The Original Code and all software distributed under the License are
+ * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER

  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,

- * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT.  Please see the
- * License for the specific language governing rights and limitations
- * under the License.
+ * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
+ * Please see the License for the specific language governing rights and
+ * limitations under the License.

  * 
  * 

- * @APPLE_LICENSE_HEADER_END@
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_END@

  */
 /*
  * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
  */
 /*
  * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995


@@ -52,17 +58,16 @@
  * SUCH DAMAGE.
  *
  *     @(#)tcp_output.c        8.4 (Berkeley) 5/24/95
  * SUCH DAMAGE.
  *
  *     @(#)tcp_output.c        8.4 (Berkeley) 5/24/95

+ * $FreeBSD: src/sys/netinet/tcp_output.c,v 1.39.2.10 2001/07/07 04:30:38 silby Exp $

  */
 
  */
 

-#if ISFB31
-#include "opt_tcpdebug.h"
-#endif

 #define        _IP_VHL
 
 #define        _IP_VHL
 

-#include <stddef.h>

 
 #include <sys/param.h>
 #include <sys/systm.h>
 
 #include <sys/param.h>
 #include <sys/systm.h>

+#include <sys/kernel.h>
+#include <sys/sysctl.h>

 #include <sys/mbuf.h>
 #include <sys/domain.h>
 #include <sys/protosw.h>
 #include <sys/mbuf.h>
 #include <sys/domain.h>
 #include <sys/protosw.h>


@@ -74,13 +79,13 @@
 #include <netinet/in.h>
 #include <netinet/in_systm.h>
 #include <netinet/ip.h>
 #include <netinet/in.h>
 #include <netinet/in_systm.h>
 #include <netinet/ip.h>

+#include <netinet/in_pcb.h>

 #include <netinet/ip_var.h>
 #if INET6
 #include <netinet/ip_var.h>
 #if INET6

+#include <netinet6/in6_pcb.h>

 #include <netinet/ip6.h>
 #include <netinet/ip6.h>

-#include <netinet/ip_var.h>

 #include <netinet6/ip6_var.h>
 #endif
 #include <netinet6/ip6_var.h>
 #endif

-#include <netinet/in_pcb.h>

 #include <netinet/tcp.h>
 #define        TCPOUTFLAGS
 #include <netinet/tcp_fsm.h>
 #include <netinet/tcp.h>
 #define        TCPOUTFLAGS
 #include <netinet/tcp_fsm.h>


@@ -93,6 +98,9 @@
 #endif
 #include <sys/kdebug.h>
 
 #endif
 #include <sys/kdebug.h>
 

+#if IPSEC
+#include <netinet6/ipsec.h>
+#endif /*IPSEC*/

 
 #define DBG_LAYER_BEG          NETDBG_CODE(DBG_NETTCP, 1)
 #define DBG_LAYER_END          NETDBG_CODE(DBG_NETTCP, 3)
 
 #define DBG_LAYER_BEG          NETDBG_CODE(DBG_NETTCP, 1)
 #define DBG_LAYER_END          NETDBG_CODE(DBG_NETTCP, 3)


@@ -103,69 +111,266 @@
 extern struct mbuf *m_copypack();
 #endif
 
 extern struct mbuf *m_copypack();
 #endif
 

+static int path_mtu_discovery = 1;
+SYSCTL_INT(_net_inet_tcp, OID_AUTO, path_mtu_discovery, CTLFLAG_RW,
+       &path_mtu_discovery, 1, "Enable Path MTU Discovery");
+
+int ss_fltsz = 1;
+SYSCTL_INT(_net_inet_tcp, OID_AUTO, slowstart_flightsize, CTLFLAG_RW,
+       &ss_fltsz, 1, "Slow start flight size");
+
+int ss_fltsz_local = 4; /* starts with four segments max */
+SYSCTL_INT(_net_inet_tcp, OID_AUTO, local_slowstart_flightsize, CTLFLAG_RW,
+       &ss_fltsz_local, 1, "Slow start flight size for local networks");
+
+int     tcp_do_newreno = 0;
+SYSCTL_INT(_net_inet_tcp, OID_AUTO, newreno, CTLFLAG_RW, &tcp_do_newreno,
+        0, "Enable NewReno Algorithms");
+
+int    tcp_packet_chaining = 50;
+SYSCTL_INT(_net_inet_tcp, OID_AUTO, packetchain, CTLFLAG_RW, &tcp_packet_chaining,
+        0, "Enable TCP output packet chaining");
+
+struct mbuf *m_copym_with_hdrs(struct mbuf*, int, int, int, struct mbuf**, int*);
+static long packchain_newlist = 0;
+static long packchain_looped = 0;
+static long packchain_sent = 0;
+
+
+/* temporary: for testing */
+#if IPSEC
+extern int ipsec_bypass;
+#endif
+
+extern int slowlink_wsize;     /* window correction for slow links */
+extern u_long  route_generation;
+extern int fw_enable;          /* firewall is on: disable packet chaining */
+extern int ipsec_bypass;
+
+extern vm_size_t       so_cache_zone_element_size;
+
+static __inline__ u_int16_t
+get_socket_id(struct socket * s)
+{
+       u_int16_t               val;
+
+       if (so_cache_zone_element_size == 0) {
+               return (0);
+       }
+       val = (u_int16_t)(((u_int32_t)s) / so_cache_zone_element_size);
+       if (val == 0) {
+               val = 0xffff;
+       }
+       return (val);
+}

 
 /*
  * Tcp output routine: figure out what should be sent and send it.
  */
 int
 
 /*
  * Tcp output routine: figure out what should be sent and send it.
  */
 int

-tcp_output(tp)
-       register struct tcpcb *tp;
+tcp_output(struct tcpcb *tp)

 {
 {

-       register struct socket *so = tp->t_inpcb->inp_socket;
-       register long len, win;
+       struct socket *so = tp->t_inpcb->inp_socket;
+       long len, recwin, sendwin;

        int off, flags, error;
        register struct mbuf *m;
        struct ip *ip = NULL;
        int off, flags, error;
        register struct mbuf *m;
        struct ip *ip = NULL;

-       struct ipovly *ipov = NULL;
+       register struct ipovly *ipov = NULL;

 #if INET6
        struct ip6_hdr *ip6 = NULL;
 #endif /* INET6 */
 #if INET6
        struct ip6_hdr *ip6 = NULL;
 #endif /* INET6 */

-       struct tcphdr *th;
+       register struct tcphdr *th;

        u_char opt[TCP_MAXOLEN];
        unsigned ipoptlen, optlen, hdrlen;
        u_char opt[TCP_MAXOLEN];
        unsigned ipoptlen, optlen, hdrlen;

-       int idle, sendalot;
+       int idle, sendalot, howmuchsent = 0;
+       int i, sack_rxmit;
+       int sack_bytes_rxmt;
+       struct sackhole *p;
+
+       int maxburst = TCP_MAXBURST;

        struct rmxp_tao *taop;
        struct rmxp_tao tao_noncached;
        struct rmxp_tao *taop;
        struct rmxp_tao tao_noncached;

-#if INET6
-       int isipv6 = (tp->t_inpcb->inp_vflag & INP_IPV4) == 0;
-#endif
-       int    last_off;
+       int    last_off = 0;

        int    m_off;
        struct mbuf *m_last = 0;
        struct mbuf *m_head = 0;
        int    m_off;
        struct mbuf *m_last = 0;
        struct mbuf *m_head = 0;

+       struct mbuf *packetlist = 0;
+       struct mbuf *lastpacket = 0;
+#if INET6
+       int isipv6 = tp->t_inpcb->inp_vflag & INP_IPV6 ;
+#endif
+       short packchain_listadd = 0;
+       u_int16_t       socket_id = get_socket_id(so);

 
 
 
 

-       KERNEL_DEBUG(DBG_FNC_TCP_OUTPUT | DBG_FUNC_START, 0,0,0,0,0);
-
-       KERNEL_DEBUG(DBG_LAYER_BEG,
-                    ((tp->t_template->tt_dport << 16) | tp->t_template->tt_sport),
-                    (((tp->t_template->tt_src.s_addr & 0xffff) << 16) |
-                     (tp->t_template->tt_dst.s_addr & 0xffff)),
-                    0,0,0);

        /*
         * Determine length of data that should be transmitted,
         * and flags that will be used.
         * If there is some data or critical controls (SYN, RST)
         * to send, then transmit; otherwise, investigate further.
         */
        /*
         * Determine length of data that should be transmitted,
         * and flags that will be used.
         * If there is some data or critical controls (SYN, RST)
         * to send, then transmit; otherwise, investigate further.
         */

-       idle = (tp->snd_max == tp->snd_una);
-       if (idle && tp->t_idle >= tp->t_rxtcur)
+       idle = (tp->t_flags & TF_LASTIDLE) || (tp->snd_max == tp->snd_una);
+       if (idle && tp->t_rcvtime >= tp->t_rxtcur) {

                /*
                 * We have been idle for "a while" and no acks are
                 * expected to clock out any data we send --
                 * slow start to get ack "clock" running again.
                /*
                 * We have been idle for "a while" and no acks are
                 * expected to clock out any data we send --
                 * slow start to get ack "clock" running again.

+                *
+                * Set the slow-start flight size depending on whether
+                * this is a local network or not.

                 */
                 */

-               tp->snd_cwnd = tp->t_maxseg;
-    
-    /* Never send data that's already been acked */
-    if (SEQ_GT(tp->snd_una, tp->snd_nxt))
-        tp->snd_nxt = tp->snd_una;
+               if (
+#if INET6
+                   (isipv6 && in6_localaddr(&tp->t_inpcb->in6p_faddr)) ||
+                   (!isipv6 &&
+#endif
+                    in_localaddr(tp->t_inpcb->inp_faddr)
+#if INET6
+                    )
+#endif
+                   )
+                       tp->snd_cwnd = tp->t_maxseg * ss_fltsz_local;
+               else     
+                       tp->snd_cwnd = tp->t_maxseg * ss_fltsz;
+       }
+       tp->t_flags &= ~TF_LASTIDLE;
+       if (idle) {
+               if (tp->t_flags & TF_MORETOCOME) {
+                       tp->t_flags |= TF_LASTIDLE;
+                       idle = 0;
+               }
+       }

 again:
 again:

+       KERNEL_DEBUG(DBG_FNC_TCP_OUTPUT | DBG_FUNC_START, 0,0,0,0,0);
+
+#if INET6
+       if (isipv6) {
+       
+               KERNEL_DEBUG(DBG_LAYER_BEG,
+                    ((tp->t_inpcb->inp_fport << 16) | tp->t_inpcb->inp_lport),
+                    (((tp->t_inpcb->in6p_laddr.s6_addr16[0] & 0xffff) << 16) |
+                     (tp->t_inpcb->in6p_faddr.s6_addr16[0] & 0xffff)),
+                    sendalot,0,0);
+       }
+       else
+#endif
+
+       {
+               KERNEL_DEBUG(DBG_LAYER_BEG,
+                    ((tp->t_inpcb->inp_fport << 16) | tp->t_inpcb->inp_lport),
+                    (((tp->t_inpcb->inp_laddr.s_addr & 0xffff) << 16) |
+                     (tp->t_inpcb->inp_faddr.s_addr & 0xffff)),
+                    sendalot,0,0);
+       /*
+        * If the route generation id changed, we need to check that our
+        * local (source) IP address is still valid. If it isn't either
+        * return error or silently do nothing (assuming the address will
+        * come back before the TCP connection times out).
+        */
+
+      if ((tp->t_inpcb->inp_route.ro_rt != NULL &&
+           (tp->t_inpcb->inp_route.ro_rt->generation_id != route_generation)) || (tp->t_inpcb->inp_route.ro_rt == NULL)) {
+               /* check that the source address is still valid */
+               if (ifa_foraddr(tp->t_inpcb->inp_laddr.s_addr) == 0) {
+                       if (tp->t_state >= TCPS_CLOSE_WAIT) {
+                               tcp_close(tp);
+                               return(EADDRNOTAVAIL);
+                       }
+
+                       /* set Retransmit  timer if it wasn't set
+                        * reset Persist timer and shift register as the
+                        * adversed peer window may not be valid anymore
+                        */
+
+                        if (!tp->t_timer[TCPT_REXMT]) {
+                                tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
+                               if (tp->t_timer[TCPT_PERSIST]) {
+                                       tp->t_timer[TCPT_PERSIST] = 0;
+                                       tp->t_rxtshift = 0;
+                               }
+                       }
+
+                       if (packetlist) {
+                               error = ip_output_list(packetlist, packchain_listadd, tp->t_inpcb->inp_options, &tp->t_inpcb->inp_route,
+                                       (so->so_options & SO_DONTROUTE), 0);
+                               tp->t_lastchain = 0;
+                       }
+                       if (so->so_flags & SOF_NOADDRAVAIL)
+                               return(EADDRNOTAVAIL);
+                       else
+                               return(0); /* silently ignore and keep data in socket */
+               }
+        }
+       }
+
+       /*
+        * If we've recently taken a timeout, snd_max will be greater than
+        * snd_nxt.  There may be SACK information that allows us to avoid
+        * resending already delivered data.  Adjust snd_nxt accordingly.
+        */
+       if (tp->sack_enable && SEQ_LT(tp->snd_nxt, tp->snd_max))
+               tcp_sack_adjust(tp);

        sendalot = 0;
        off = tp->snd_nxt - tp->snd_una;
        sendalot = 0;
        off = tp->snd_nxt - tp->snd_una;

-       win = min(tp->snd_wnd, tp->snd_cwnd);
+       sendwin = min(tp->snd_wnd, tp->snd_cwnd);
+
+       if (tp->t_flags & TF_SLOWLINK && slowlink_wsize > 0)
+               sendwin = min(sendwin, slowlink_wsize);

 
        flags = tcp_outflags[tp->t_state];
 
        flags = tcp_outflags[tp->t_state];

+       /*
+        * Send any SACK-generated retransmissions.  If we're explicitly trying
+        * to send out new data (when sendalot is 1), bypass this function.
+        * If we retransmit in fast recovery mode, decrement snd_cwnd, since
+        * we're replacing a (future) new transmission with a retransmission
+        * now, and we previously incremented snd_cwnd in tcp_input().
+        */
+       /*
+        * Still in sack recovery , reset rxmit flag to zero.
+        */
+       sack_rxmit = 0;
+       sack_bytes_rxmt = 0;
+       len = 0;
+       p = NULL;
+       if (tp->sack_enable && IN_FASTRECOVERY(tp) &&
+           (p = tcp_sack_output(tp, &sack_bytes_rxmt))) {
+               long cwin;
+               
+               cwin = min(tp->snd_wnd, tp->snd_cwnd) - sack_bytes_rxmt;
+               if (cwin < 0)
+                       cwin = 0;
+               /* Do not retransmit SACK segments beyond snd_recover */
+               if (SEQ_GT(p->end, tp->snd_recover)) {
+                       /*
+                        * (At least) part of sack hole extends beyond
+                        * snd_recover. Check to see if we can rexmit data
+                        * for this hole.
+                        */
+                       if (SEQ_GEQ(p->rxmit, tp->snd_recover)) {
+                               /*
+                                * Can't rexmit any more data for this hole.
+                                * That data will be rexmitted in the next
+                                * sack recovery episode, when snd_recover
+                                * moves past p->rxmit.
+                                */
+                               p = NULL;
+                               goto after_sack_rexmit;
+                       } else
+                               /* Can rexmit part of the current hole */
+                               len = ((long)ulmin(cwin,
+                                                  tp->snd_recover - p->rxmit));
+               } else
+                       len = ((long)ulmin(cwin, p->end - p->rxmit));
+               off = p->rxmit - tp->snd_una;
+               if (len > 0) {
+                       sack_rxmit = 1;
+                       sendalot = 1;
+                       tcpstat.tcps_sack_rexmits++;
+                       tcpstat.tcps_sack_rexmit_bytes +=
+                           min(len, tp->t_maxseg);
+               }
+       }
+after_sack_rexmit:

        /*
         * Get standard flags, and add SYN or FIN if requested by 'hidden'
         * state flags.
        /*
         * Get standard flags, and add SYN or FIN if requested by 'hidden'
         * state flags.


@@ -182,7 +387,7 @@ again:
         * and go to transmit state.
         */
        if (tp->t_force) {
         * and go to transmit state.
         */
        if (tp->t_force) {

-               if (win == 0) {
+               if (sendwin == 0) {

                        /*
                         * If we still have some data to send, then
                         * clear the FIN bit.  Usually this would
                        /*
                         * If we still have some data to send, then
                         * clear the FIN bit.  Usually this would


@@ -201,18 +406,58 @@ again:
                         */
                        if (off < so->so_snd.sb_cc)
                                flags &= ~TH_FIN;
                         */
                        if (off < so->so_snd.sb_cc)
                                flags &= ~TH_FIN;

-                       win = 1;
+                       sendwin = 1;

                } else {
                        tp->t_timer[TCPT_PERSIST] = 0;
                        tp->t_rxtshift = 0;
                }
        }
 
                } else {
                        tp->t_timer[TCPT_PERSIST] = 0;
                        tp->t_rxtshift = 0;
                }
        }
 

-       len = (long)ulmin(so->so_snd.sb_cc, win) - off;
-
-       if ((taop = tcp_gettaocache(tp->t_inpcb)) == NULL) {
-               taop = &tao_noncached;
-               bzero(taop, sizeof(*taop));
+       /*
+        * If snd_nxt == snd_max and we have transmitted a FIN, the
+        * offset will be > 0 even if so_snd.sb_cc is 0, resulting in
+        * a negative length.  This can also occur when TCP opens up
+        * its congestion window while receiving additional duplicate
+        * acks after fast-retransmit because TCP will reset snd_nxt
+        * to snd_max after the fast-retransmit.
+        *
+        * In the normal retransmit-FIN-only case, however, snd_nxt will
+        * be set to snd_una, the offset will be 0, and the length may
+        * wind up 0.
+        *
+        * If sack_rxmit is true we are retransmitting from the scoreboard
+        * in which case len is already set.
+        */
+       if (sack_rxmit == 0) {
+               if (sack_bytes_rxmt == 0)
+                       len = ((long)ulmin(so->so_snd.sb_cc, sendwin) - off);
+               else {
+                       long cwin;
+
+                        /*
+                        * We are inside of a SACK recovery episode and are
+                        * sending new data, having retransmitted all the
+                        * data possible in the scoreboard.
+                        */
+                       len = ((long)ulmin(so->so_snd.sb_cc, tp->snd_wnd) 
+                              - off);
+                       /*
+                        * Don't remove this (len > 0) check !
+                        * We explicitly check for len > 0 here (although it 
+                        * isn't really necessary), to work around a gcc 
+                        * optimization issue - to force gcc to compute
+                        * len above. Without this check, the computation
+                        * of len is bungled by the optimizer.
+                        */
+                       if (len > 0) {
+                               cwin = tp->snd_cwnd - 
+                                       (tp->snd_nxt - tp->sack_newdata) -
+                                       sack_bytes_rxmt;
+                               if (cwin < 0)
+                                       cwin = 0;
+                               len = lmin(len, cwin);
+                       }
+               }

        }
 
        /*
        }
 
        /*


@@ -223,22 +468,23 @@ again:
        if ((flags & TH_SYN) && SEQ_GT(tp->snd_nxt, tp->snd_una)) {
                flags &= ~TH_SYN;
                off--, len++;
        if ((flags & TH_SYN) && SEQ_GT(tp->snd_nxt, tp->snd_una)) {
                flags &= ~TH_SYN;
                off--, len++;

-               if (len > 0 && tp->t_state == TCPS_SYN_SENT &&
-                   taop->tao_ccsent == 0) {
+               if (len > 0 && tp->t_state == TCPS_SYN_SENT) {
+                       if (packetlist) {
+                               error = ip_output_list(packetlist, packchain_listadd, tp->t_inpcb->inp_options, &tp->t_inpcb->inp_route,
+                               (so->so_options & SO_DONTROUTE), 0);
+                               tp->t_lastchain = 0;
+                       }

                  KERNEL_DEBUG(DBG_FNC_TCP_OUTPUT | DBG_FUNC_END, 0,0,0,0,0);
                  return 0;
                }
        }
 
        /*
                  KERNEL_DEBUG(DBG_FNC_TCP_OUTPUT | DBG_FUNC_END, 0,0,0,0,0);
                  return 0;
                }
        }
 
        /*

-        * Be careful not to send data and/or FIN on SYN segments
-        * in cases when no CC option will be sent.
+        * Be careful not to send data and/or FIN on SYN segments.

         * This measure is needed to prevent interoperability problems
         * with not fully conformant TCP implementations.
         */
         * This measure is needed to prevent interoperability problems
         * with not fully conformant TCP implementations.
         */

-       if ((flags & TH_SYN) &&
-           ((tp->t_flags & TF_NOOPT) || !(tp->t_flags & TF_REQ_CC) ||
-            ((flags & TH_ACK) && !(tp->t_flags & TF_RCVD_CC)))) {
+       if ((flags & TH_SYN) && (tp->t_flags & TF_NOOPT)) {

                len = 0;
                flags &= ~TH_FIN;
        }
                len = 0;
                flags &= ~TH_FIN;
        }


@@ -247,7 +493,7 @@ again:
                /*
                 * If FIN has been sent but not acked,
                 * but we haven't been called to retransmit,
                /*
                 * If FIN has been sent but not acked,
                 * but we haven't been called to retransmit,

-                * len will be -1.  Otherwise, window shrank
+                * len will be < 0.  Otherwise, window shrank

                 * after we sent into it.  If window shrank to 0,
                 * cancel pending retransmit, pull snd_nxt back
                 * to (closed) window, and set the persist timer
                 * after we sent into it.  If window shrank to 0,
                 * cancel pending retransmit, pull snd_nxt back
                 * to (closed) window, and set the persist timer


@@ -255,7 +501,7 @@ again:
                 * close completely, just wait for an ACK.
                 */
                len = 0;
                 * close completely, just wait for an ACK.
                 */
                len = 0;

-               if (win == 0) {
+               if (sendwin == 0) {

                        tp->t_timer[TCPT_REXMT] = 0;
                        tp->t_rxtshift = 0;
                        tp->snd_nxt = tp->snd_una;
                        tp->t_timer[TCPT_REXMT] = 0;
                        tp->t_rxtshift = 0;
                        tp->snd_nxt = tp->snd_una;


@@ -263,24 +509,41 @@ again:
                                tcp_setpersist(tp);
                }
        }
                                tcp_setpersist(tp);
                }
        }

+
+       /*
+        * len will be >= 0 after this point.  Truncate to the maximum
+        * segment length and ensure that FIN is removed if the length
+        * no longer contains the last data byte.
+        */

        if (len > tp->t_maxseg) {
                len = tp->t_maxseg;
        if (len > tp->t_maxseg) {
                len = tp->t_maxseg;

+               howmuchsent += len;

                sendalot = 1;
        }
                sendalot = 1;
        }

-       if (SEQ_LT(tp->snd_nxt + len, tp->snd_una + so->so_snd.sb_cc))
-               flags &= ~TH_FIN;
+       if (sack_rxmit) {
+               if (SEQ_LT(p->rxmit + len, tp->snd_una + so->so_snd.sb_cc))
+                       flags &= ~TH_FIN;
+       } else {
+               if (SEQ_LT(tp->snd_nxt + len, tp->snd_una + so->so_snd.sb_cc))
+                       flags &= ~TH_FIN;
+       }

 
 

-       win = sbspace(&so->so_rcv);
+       if (tp->t_flags & TF_SLOWLINK && slowlink_wsize > 0 )   /* Clips window size for slow links */
+               recwin = min(sbspace(&so->so_rcv), slowlink_wsize);
+       else
+               recwin = sbspace(&so->so_rcv);

 
        /*
 
        /*

-        * Sender silly window avoidance.  If connection is idle
-        * and can send all data, a maximum segment,
-        * at least a maximum default-size segment do it,
-        * or are forced, do it; otherwise don't bother.
-        * If peer's buffer is tiny, then send
-        * when window is at least half open.
-        * If retransmitting (possibly after persist timer forced us
-        * to send into a small window), then must resend.
+        * Sender silly window avoidance.   We transmit under the following
+        * conditions when len is non-zero:
+        *
+        *      - We have a full segment
+        *      - This is the last buffer in a write()/send() and we are
+        *        either idle or running NODELAY
+        *      - we've timed out (e.g. persist timer)
+        *      - we have more then 1/2 the maximum send window's worth of
+        *        data (receiver may be limited the window size)
+        *      - we need to retransmit

         */
        if (len) {
                if (len == tp->t_maxseg)
         */
        if (len) {
                if (len == tp->t_maxseg)


@@ -294,7 +557,9 @@ again:
                        goto send;
                if (len >= tp->max_sndwnd / 2 && tp->max_sndwnd > 0)
                        goto send;
                        goto send;
                if (len >= tp->max_sndwnd / 2 && tp->max_sndwnd > 0)
                        goto send;

-               if (SEQ_LT(tp->snd_nxt, tp->snd_max))
+               if (SEQ_LT(tp->snd_nxt, tp->snd_max))   /* retransmit case */
+                       goto send;
+               if (sack_rxmit)

                        goto send;
        }
 
                        goto send;
        }
 


@@ -304,14 +569,15 @@ again:
         * next expected input).  If the difference is at least two
         * max size segments, or at least 50% of the maximum possible
         * window, then want to send a window update to peer.
         * next expected input).  If the difference is at least two
         * max size segments, or at least 50% of the maximum possible
         * window, then want to send a window update to peer.

+        * Skip this if the connection is in T/TCP half-open state.

         */
         */

-       if (win > 0) {
+       if (recwin > 0 && !(tp->t_flags & TF_NEEDSYN)) {

                /*
                 * "adv" is the amount we can increase the window,
                 * taking into account that we are limited by
                 * TCP_MAXWIN << tp->rcv_scale.
                 */
                /*
                 * "adv" is the amount we can increase the window,
                 * taking into account that we are limited by
                 * TCP_MAXWIN << tp->rcv_scale.
                 */

-               long adv = min(win, (long)TCP_MAXWIN << tp->rcv_scale) -
+               long adv = min(recwin, (long)TCP_MAXWIN << tp->rcv_scale) -

                        (tp->rcv_adv - tp->rcv_nxt);
 
                if (adv >= (long) (2 * tp->t_maxseg))
                        (tp->rcv_adv - tp->rcv_nxt);
 
                if (adv >= (long) (2 * tp->t_maxseg))


@@ -321,7 +587,8 @@ again:
        }
 
        /*
        }
 
        /*

-        * Send if we owe peer an ACK.
+        * Send if we owe the peer an ACK, RST, SYN, or urgent data.  ACKNOW
+        * is also a catch-all for the retransmit timer timeout case.

         */
        if (tp->t_flags & TF_ACKNOW)
                goto send;
         */
        if (tp->t_flags & TF_ACKNOW)
                goto send;


@@ -332,13 +599,22 @@ again:
                goto send;
        /*
         * If our state indicates that FIN should be sent
                goto send;
        /*
         * If our state indicates that FIN should be sent

-        * and we have not yet done so, or we're retransmitting the FIN,
-        * then we need to send.
+        * and we have not yet done so, then we need to send.

         */
        if (flags & TH_FIN &&
            ((tp->t_flags & TF_SENTFIN) == 0 || tp->snd_nxt == tp->snd_una))
                goto send;
         */
        if (flags & TH_FIN &&
            ((tp->t_flags & TF_SENTFIN) == 0 || tp->snd_nxt == tp->snd_una))
                goto send;

-
+       /*
+        * In SACK, it is possible for tcp_output to fail to send a segment
+        * after the retransmission timer has been turned off.  Make sure
+        * that the retransmission timer is set.
+        */
+       if (tp->sack_enable && SEQ_GT(tp->snd_max, tp->snd_una) &&
+               tp->t_timer[TCPT_REXMT] == 0 &&
+           tp->t_timer[TCPT_PERSIST] == 0) {
+                       tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
+                       goto just_return;
+       } 

        /*
         * TCP window updates are not reliable, rather a polling protocol
         * using ``persist'' packets is used to insure receipt of window
        /*
         * TCP window updates are not reliable, rather a polling protocol
         * using ``persist'' packets is used to insure receipt of window


@@ -366,10 +642,15 @@ again:
                tp->t_rxtshift = 0;
                tcp_setpersist(tp);
        }
                tp->t_rxtshift = 0;
                tcp_setpersist(tp);
        }

-
+just_return:

        /*
        /*

-        * No reason to send a segment, just return.
+        * If there is no reason to send a segment, just return.
+        * but if there is some packets left in the packet list, send them now.

         */
         */

+       if (packetlist) {
+               error = ip_output_list(packetlist, packchain_listadd, tp->t_inpcb->inp_options, &tp->t_inpcb->inp_route,
+                       (so->so_options & SO_DONTROUTE), 0);
+       }

        KERNEL_DEBUG(DBG_FNC_TCP_OUTPUT | DBG_FUNC_END, 0,0,0,0,0);
        return (0);
 
        KERNEL_DEBUG(DBG_FNC_TCP_OUTPUT | DBG_FUNC_END, 0,0,0,0,0);
        return (0);
 


@@ -380,12 +661,12 @@ send:
         * NOTE: we assume that the IP/TCP header plus TCP options
         * always fit in a single mbuf, leaving room for a maximum
         * link header, i.e.
         * NOTE: we assume that the IP/TCP header plus TCP options
         * always fit in a single mbuf, leaving room for a maximum
         * link header, i.e.

-        *      max_linkhdr + sizeof (struct tcpiphdr) + optlen <= MHLEN
+        *      max_linkhdr + sizeof (struct tcpiphdr) + optlen <= MCLBYTES

         */
        optlen = 0;
 #if INET6
        if (isipv6)
         */
        optlen = 0;
 #if INET6
        if (isipv6)

-               hdrlen = sizeof (struct tcpip6hdr);
+               hdrlen = sizeof (struct ip6_hdr) + sizeof (struct tcphdr);

        else
 #endif
        hdrlen = sizeof (struct tcpiphdr);
        else
 #endif
        hdrlen = sizeof (struct tcpiphdr);


@@ -396,7 +677,7 @@ send:
 
                        opt[0] = TCPOPT_MAXSEG;
                        opt[1] = TCPOLEN_MAXSEG;
 
                        opt[0] = TCPOPT_MAXSEG;
                        opt[1] = TCPOLEN_MAXSEG;

-                       mss = htons((u_short) tcp_mssopt(tp, isipv6));
+                       mss = htons((u_short) tcp_mssopt(tp));

                        (void)memcpy(opt + 2, &mss, sizeof(mss));
                        optlen = TCPOLEN_MAXSEG;
 
                        (void)memcpy(opt + 2, &mss, sizeof(mss));
                        optlen = TCPOLEN_MAXSEG;
 


@@ -431,95 +712,107 @@ send:
                optlen += TCPOLEN_TSTAMP_APPA;
        }
 
                optlen += TCPOLEN_TSTAMP_APPA;
        }
 

-       /*
-        * Send `CC-family' options if our side wants to use them (TF_REQ_CC),
-        * options are allowed (!TF_NOOPT) and it's not a RST.
-        */
-       if ((tp->t_flags & (TF_REQ_CC|TF_NOOPT)) == TF_REQ_CC &&
-            (flags & TH_RST) == 0) {
-               switch (flags & (TH_SYN|TH_ACK)) {
-               /*
-                * This is a normal ACK, send CC if we received CC before
-                * from our peer.
+       if (tp->sack_enable && ((tp->t_flags & TF_NOOPT) == 0)) {
+               /* 
+                * Tack on the SACK permitted option *last*.
+                * And do padding of options after tacking this on.
+                * This is because of MSS, TS, WinScale and Signatures are
+                * all present, we have just 2 bytes left for the SACK
+                * permitted option, which is just enough.

                 */
                 */

-               case TH_ACK:
-                       if (!(tp->t_flags & TF_RCVD_CC))
-                               break;
-                       /*FALLTHROUGH*/
-

                /*
                /*

-                * We can only get here in T/TCP's SYN_SENT* state, when
-                * we're a sending a non-SYN segment without waiting for
-                * the ACK of our SYN.  A check above assures that we only
-                * do this if our peer understands T/TCP.
+                * If this is the first SYN of connection (not a SYN
+                * ACK), include SACK permitted option.  If this is a
+                * SYN ACK, include SACK permitted option if peer has
+                * already done so. This is only for active connect,
+                * since the syncache takes care of the passive connect.

                 */
                 */

-               case 0:
-                       opt[optlen++] = TCPOPT_NOP;
-                       opt[optlen++] = TCPOPT_NOP;
-                       opt[optlen++] = TCPOPT_CC;
-                       opt[optlen++] = TCPOLEN_CC;
-                       *(u_int32_t *)&opt[optlen] = htonl(tp->cc_send);
-
-                       optlen += 4;
-                       break;
+               if ((flags & TH_SYN) &&
+                   (!(flags & TH_ACK) || (tp->t_flags & TF_SACK_PERMIT))) {
+                       u_char *bp;
+                       bp = (u_char *)opt + optlen;
+
+                       *bp++ = TCPOPT_SACK_PERMITTED;
+                       *bp++ = TCPOLEN_SACK_PERMITTED;
+                       optlen += TCPOLEN_SACK_PERMITTED;
+               }

 
                /*
 
                /*

-                * This is our initial SYN, check whether we have to use
-                * CC or CC.new.
+                * Send SACKs if necessary.  This should be the last
+                * option processed.  Only as many SACKs are sent as
+                * are permitted by the maximum options size.
+                *
+                * In general, SACK blocks consume 8*n+2 bytes.
+                * So a full size SACK blocks option is 34 bytes
+                * (to generate 4 SACK blocks).  At a minimum,
+                * we need 10 bytes (to generate 1 SACK block).
+                * If TCP Timestamps (12 bytes) and TCP Signatures
+                * (18 bytes) are both present, we'll just have
+                * 10 bytes for SACK options 40 - (12 + 18).

                 */
                 */

-               case TH_SYN:
-                       opt[optlen++] = TCPOPT_NOP;
-                       opt[optlen++] = TCPOPT_NOP;
-                       opt[optlen++] = tp->t_flags & TF_SENDCCNEW ?
-                                               TCPOPT_CCNEW : TCPOPT_CC;
-                       opt[optlen++] = TCPOLEN_CC;
-                       *(u_int32_t *)&opt[optlen] = htonl(tp->cc_send);
-                       optlen += 4;
-                       break;
+               if (TCPS_HAVEESTABLISHED(tp->t_state) &&
+                   (tp->t_flags & TF_SACK_PERMIT) && tp->rcv_numsacks > 0 &&
+                   MAX_TCPOPTLEN - optlen - 2 >= TCPOLEN_SACK) {
+                       int nsack, sackoptlen, padlen;
+                       u_char *bp = (u_char *)opt + optlen;
+                       u_int32_t *lp;

 
 

-               /*
-                * This is a SYN,ACK; send CC and CC.echo if we received
-                * CC from our peer.
-                */
-               case (TH_SYN|TH_ACK):
-                       if (tp->t_flags & TF_RCVD_CC) {
-                               opt[optlen++] = TCPOPT_NOP;
-                               opt[optlen++] = TCPOPT_NOP;
-                               opt[optlen++] = TCPOPT_CC;
-                               opt[optlen++] = TCPOLEN_CC;
-                               *(u_int32_t *)&opt[optlen] =
-                                       htonl(tp->cc_send);
-                               optlen += 4;
-                               opt[optlen++] = TCPOPT_NOP;
-                               opt[optlen++] = TCPOPT_NOP;
-                               opt[optlen++] = TCPOPT_CCECHO;
-                               opt[optlen++] = TCPOLEN_CC;
-                               *(u_int32_t *)&opt[optlen] =
-                                       htonl(tp->cc_recv);
-                               optlen += 4;
+                       nsack = (MAX_TCPOPTLEN - optlen - 2) / TCPOLEN_SACK;
+                       nsack = min(nsack, tp->rcv_numsacks);
+                       sackoptlen = (2 + nsack * TCPOLEN_SACK);
+
+                       /*
+                        * First we need to pad options so that the
+                        * SACK blocks can start at a 4-byte boundary
+                        * (sack option and length are at a 2 byte offset).
+                        */
+                       padlen = (MAX_TCPOPTLEN - optlen - sackoptlen) % 4;
+                       optlen += padlen;
+                       while (padlen-- > 0)
+                               *bp++ = TCPOPT_NOP;
+
+                       tcpstat.tcps_sack_send_blocks++;
+                       *bp++ = TCPOPT_SACK;
+                       *bp++ = sackoptlen;
+                       lp = (u_int32_t *)bp;
+                       for (i = 0; i < nsack; i++) {
+                               struct sackblk sack = tp->sackblks[i];
+                               *lp++ = htonl(sack.start);
+                               *lp++ = htonl(sack.end);

                        }
                        }

-                       break;
+                       optlen += sackoptlen;

                }
                }

-       }
+       }
+
+       /* Pad TCP options to a 4 byte boundary */
+       if (optlen < MAX_TCPOPTLEN && (optlen % sizeof(u_int32_t))) {
+               int pad = sizeof(u_int32_t) - (optlen % sizeof(u_int32_t));
+               u_char *bp = (u_char *)opt + optlen;
+
+               optlen += pad;
+               while (pad) {
+                       *bp++ = TCPOPT_EOL;
+                       pad--;
+               }
+       }
+
+       hdrlen += optlen;

 
 

-       hdrlen += optlen;

 #if INET6
        if (isipv6)
                ipoptlen = ip6_optlen(tp->t_inpcb);
        else
 #endif
 #if INET6
        if (isipv6)
                ipoptlen = ip6_optlen(tp->t_inpcb);
        else
 #endif

-       if (tp->t_inpcb->inp_options) {
-               ipoptlen = tp->t_inpcb->inp_options->m_len -
+       {
+               if (tp->t_inpcb->inp_options) {
+                       ipoptlen = tp->t_inpcb->inp_options->m_len -

                                offsetof(struct ipoption, ipopt_list);
                                offsetof(struct ipoption, ipopt_list);

-       } else {
-               ipoptlen = 0;
+               } else
+                       ipoptlen = 0;

        }
 #if IPSEC
        }
 #if IPSEC

-#if INET6
-       ipoptlen += ipsec_hdrsiz_tcp(tp, isipv6);
-#else
-       ipoptlen += ipsec_hdrsiz_tcp(tp, 0);
-#endif
+       if (ipsec_bypass == 0)
+               ipoptlen += ipsec_hdrsiz_tcp(tp);

 #endif
 
        /*
 #endif
 
        /*


@@ -534,12 +827,18 @@ send:
                 */
                flags &= ~TH_FIN;
                len = tp->t_maxopd - optlen - ipoptlen;
                 */
                flags &= ~TH_FIN;
                len = tp->t_maxopd - optlen - ipoptlen;

+               howmuchsent += len;

                sendalot = 1;
        }
 
 /*#ifdef DIAGNOSTIC*/
                sendalot = 1;
        }
 
 /*#ifdef DIAGNOSTIC*/

+#if INET6
+       if (max_linkhdr + hdrlen > MCLBYTES)
+               panic("tcphdr too big");
+#else

        if (max_linkhdr + hdrlen > MHLEN)
                panic("tcphdr too big");
        if (max_linkhdr + hdrlen > MHLEN)
                panic("tcphdr too big");

+#endif

 /*#endif*/
 
        /*
 /*#endif*/
 
        /*


@@ -569,6 +868,18 @@ send:
                m->m_len += hdrlen;
                m->m_data -= hdrlen;
 #else
                m->m_len += hdrlen;
                m->m_data -= hdrlen;
 #else

+               /*
+                * try to use the new interface that allocates all 
+                * the necessary mbuf hdrs under 1 mbuf lock and 
+                * avoids rescanning the socket mbuf list if 
+                * certain conditions are met.  This routine can't
+                * be used in the following cases...
+                * 1) the protocol headers exceed the capacity of
+                * of a single mbuf header's data area (no cluster attached)
+                * 2) the length of the data being transmitted plus
+                * the protocol headers fits into a single mbuf header's
+                * data area (no cluster attached)
+                */

                m = NULL;
 #if INET6
                if (MHLEN < hdrlen + max_linkhdr) {
                m = NULL;
 #if INET6
                if (MHLEN < hdrlen + max_linkhdr) {


@@ -597,6 +908,12 @@ send:
                                m->m_data += max_linkhdr;
                                m->m_len = hdrlen;
                        }
                                m->m_data += max_linkhdr;
                                m->m_len = hdrlen;
                        }

+                       /* makes sure we still have data left to be sent at this point */
+                       if (so->so_snd.sb_mb == NULL || off == -1) {
+                               if (m != NULL)  m_freem(m);
+                               error = 0; /* should we return an error? */
+                               goto out;
+                       }

                        m_copydata(so->so_snd.sb_mb, off, (int) len,
                            mtod(m, caddr_t) + hdrlen);
                        m->m_len += len;
                        m_copydata(so->so_snd.sb_mb, off, (int) len,
                            mtod(m, caddr_t) + hdrlen);
                        m->m_len += len;


@@ -609,11 +926,31 @@ send:
                                        goto out;
                                }
                        } else {
                                        goto out;
                                }
                        } else {

+                               /*
+                                * determine whether the mbuf pointer and offset passed back by the 'last' call
+                                * to m_copym_with_hdrs are still valid... if the head of the socket chain has
+                                * changed (due to an incoming ACK for instance), or the offset into the chain we
+                                * just computed is different from the one last returned by m_copym_with_hdrs (perhaps
+                                * we're re-transmitting a packet sent earlier), than we can't pass the mbuf pointer and
+                                * offset into it as valid hints for m_copym_with_hdrs to use (if valid, these hints allow
+                                * m_copym_with_hdrs to avoid rescanning from the beginning of the socket buffer mbuf list.
+                                * setting the mbuf pointer to NULL is sufficient to disable the hint mechanism.
+                                */

                                if (m_head != so->so_snd.sb_mb || last_off != off)
                                        m_last = NULL;
                                last_off = off + len;
                                m_head = so->so_snd.sb_mb;
                                if (m_head != so->so_snd.sb_mb || last_off != off)
                                        m_last = NULL;
                                last_off = off + len;
                                m_head = so->so_snd.sb_mb;

-
+       
+                               /* makes sure we still have data left to be sent at this point */
+                               if (m_head == NULL) {
+                                       error = 0; /* should we return an error? */
+                                       goto out;
+                               }
+                               
+                               /*
+                                * m_copym_with_hdrs will always return the last mbuf pointer and the offset into it that
+                                * it acted on to fullfill the current request, whether a valid 'hint' was passed in or not
+                                */

                                if ((m = m_copym_with_hdrs(so->so_snd.sb_mb, off, (int) len, M_DONTWAIT, &m_last, &m_off)) == NULL) {
                                        error = ENOBUFS;
                                        goto out;
                                if ((m = m_copym_with_hdrs(so->so_snd.sb_mb, off, (int) len, M_DONTWAIT, &m_last, &m_off)) == NULL) {
                                        error = ENOBUFS;
                                        goto out;


@@ -647,35 +984,29 @@ send:
                        goto out;
                }
 #if INET6
                        goto out;
                }
 #if INET6

-               if (isipv6) {
+               if (isipv6 && (MHLEN < hdrlen + max_linkhdr) &&
+                   MHLEN >= hdrlen) {

                        MH_ALIGN(m, hdrlen);
                } else
 #endif
                m->m_data += max_linkhdr;
                m->m_len = hdrlen;
        }
                        MH_ALIGN(m, hdrlen);
                } else
 #endif
                m->m_data += max_linkhdr;
                m->m_len = hdrlen;
        }

-       m->m_pkthdr.rcvif = (struct ifnet *)0;
-       if (tp->t_template == 0)
-               panic("tcp_output");
+       m->m_pkthdr.rcvif = 0;

 #if INET6
        if (isipv6) {
                ip6 = mtod(m, struct ip6_hdr *);
                th = (struct tcphdr *)(ip6 + 1);
 #if INET6
        if (isipv6) {
                ip6 = mtod(m, struct ip6_hdr *);
                th = (struct tcphdr *)(ip6 + 1);

-               bcopy((caddr_t)&tp->t_template->tt_i6, (caddr_t)ip6,
-                     sizeof(struct ip6_hdr));
-               bcopy((caddr_t)&tp->t_template->tt_t, (caddr_t)th,
-                     sizeof(struct tcphdr));
-       } else {
+               tcp_fillheaders(tp, ip6, th);
+       } else

 #endif /* INET6 */
 #endif /* INET6 */

-       ip = mtod(m, struct ip *);
-       ipov = (struct ipovly *)ip;
-       th = (struct tcphdr *)(ip + 1);
-       bcopy((caddr_t)&tp->t_template->tt_i, (caddr_t)ip, sizeof(struct ip));
-       bcopy((caddr_t)&tp->t_template->tt_t, (caddr_t)th,
-             sizeof(struct tcphdr));
-#if INET6
+       {
+               ip = mtod(m, struct ip *);
+               ipov = (struct ipovly *)ip;
+               th = (struct tcphdr *)(ip + 1);
+               /* this picks up the pseudo header (w/o the length) */
+               tcp_fillheaders(tp, ip, th);

        }
        }

-#endif /* INET6 */

 
        /*
         * Fill in fields, remembering maximum advertised
 
        /*
         * Fill in fields, remembering maximum advertised


@@ -698,10 +1029,16 @@ send:
         * case, since we know we aren't doing a retransmission.
         * (retransmit and persist are mutually exclusive...)
         */
         * case, since we know we aren't doing a retransmission.
         * (retransmit and persist are mutually exclusive...)
         */

-       if (len || (flags & (TH_SYN|TH_FIN)) || tp->t_timer[TCPT_PERSIST])
-               th->th_seq = htonl(tp->snd_nxt);
-       else
-               th->th_seq = htonl(tp->snd_max);
+       if (sack_rxmit == 0) {
+               if (len || (flags & (TH_SYN|TH_FIN)) || tp->t_timer[TCPT_PERSIST])
+                       th->th_seq = htonl(tp->snd_nxt);
+               else
+                       th->th_seq = htonl(tp->snd_max);
+       } else {
+               th->th_seq = htonl(p->rxmit);
+               p->rxmit += len;
+               tp->sackhint.sack_bytes_rexmit += len;
+       }

        th->th_ack = htonl(tp->rcv_nxt);
        if (optlen) {
                bcopy(opt, th + 1, optlen);
        th->th_ack = htonl(tp->rcv_nxt);
        if (optlen) {
                bcopy(opt, th + 1, optlen);


@@ -712,13 +1049,33 @@ send:
         * Calculate receive window.  Don't shrink window,
         * but avoid silly window syndrome.
         */
         * Calculate receive window.  Don't shrink window,
         * but avoid silly window syndrome.
         */

-       if (win < (long)(so->so_rcv.sb_hiwat / 4) && win < (long)tp->t_maxseg)
-               win = 0;
-       if (win < (long)(tp->rcv_adv - tp->rcv_nxt))
-               win = (long)(tp->rcv_adv - tp->rcv_nxt);
-       if (win > (long)TCP_MAXWIN << tp->rcv_scale)
-               win = (long)TCP_MAXWIN << tp->rcv_scale;
-       th->th_win = htons((u_short) (win>>tp->rcv_scale));
+       if (recwin < (long)(so->so_rcv.sb_hiwat / 4) && recwin < (long)tp->t_maxseg)
+               recwin = 0;
+       if (recwin < (long)(tp->rcv_adv - tp->rcv_nxt))
+               recwin = (long)(tp->rcv_adv - tp->rcv_nxt);
+       if (tp->t_flags & TF_SLOWLINK && slowlink_wsize > 0) {
+               if (recwin > (long)slowlink_wsize) 
+                       recwin = slowlink_wsize;
+                       th->th_win = htons((u_short) (recwin>>tp->rcv_scale));
+       }
+       else {
+               if (recwin > (long)TCP_MAXWIN << tp->rcv_scale)
+                       recwin = (long)TCP_MAXWIN << tp->rcv_scale;
+               th->th_win = htons((u_short) (recwin>>tp->rcv_scale));
+       }
+
+       /*
+        * Adjust the RXWIN0SENT flag - indicate that we have advertised
+        * a 0 window.  This may cause the remote transmitter to stall.  This
+        * flag tells soreceive() to disable delayed acknowledgements when
+        * draining the buffer.  This can occur if the receiver is attempting
+        * to read more data then can be buffered prior to transmitting on
+        * the connection.
+        */
+       if (recwin == 0)
+               tp->t_flags |= TF_RXWIN0SENT;
+       else
+               tp->t_flags &= ~TF_RXWIN0SENT;

        if (SEQ_GT(tp->snd_up, tp->snd_nxt)) {
                th->th_urp = htons((u_short)(tp->snd_up - tp->snd_nxt));
                th->th_flags |= TH_URG;
        if (SEQ_GT(tp->snd_up, tp->snd_nxt)) {
                th->th_urp = htons((u_short)(tp->snd_up - tp->snd_nxt));
                th->th_flags |= TH_URG;


@@ -731,36 +1088,28 @@ send:
                 */
                tp->snd_up = tp->snd_una;               /* drag it along */
 
                 */
                tp->snd_up = tp->snd_una;               /* drag it along */
 

-

        /*
         * Put TCP length in extended header, and then
         * checksum extended header and data.
         */
        /*
         * Put TCP length in extended header, and then
         * checksum extended header and data.
         */

-       m->m_pkthdr.len = hdrlen + len;
+       m->m_pkthdr.len = hdrlen + len; /* in6_cksum() need this */

 #if INET6
 #if INET6

-       if (isipv6) {
-#if 0          /* ip6_plen will be filled in ip6_output. */
-               ip6->ip6_plen = htons((u_short)(sizeof(struct tcphdr) +
-                                               optlen + len));
-#endif
-
+       if (isipv6)
+               /*
+                * ip6_plen is not need to be filled now, and will be filled
+                * in ip6_output.
+                */

                th->th_sum = in6_cksum(m, IPPROTO_TCP, sizeof(struct ip6_hdr),
                                       sizeof(struct tcphdr) + optlen + len);
                th->th_sum = in6_cksum(m, IPPROTO_TCP, sizeof(struct ip6_hdr),
                                       sizeof(struct tcphdr) + optlen + len);

-       } else 
+       else

 #endif /* INET6 */
        {
 #endif /* INET6 */
        {

-
-       if (len + optlen)
-               ipov->ih_len = htons((u_short)(sizeof (struct tcphdr) +
-                                              optlen + len));
-         m->m_pkthdr.csum_flags = CSUM_TCP;
-                m->m_pkthdr.csum_data = offsetof(struct tcphdr, th_sum);
-                if (len + optlen) {       
-                        th->th_sum = in_addword(th->th_sum,
-                            htons((u_short)(optlen + len)));
-       }
-
-       }
+               m->m_pkthdr.csum_flags = CSUM_TCP;
+               m->m_pkthdr.csum_data = offsetof(struct tcphdr, th_sum);
+               if (len + optlen)
+                       th->th_sum = in_addword(th->th_sum, 
+                               htons((u_short)(optlen + len)));
+      }

 
        /*
         * In transmit state, time the transmission and arrange for
 
        /*
         * In transmit state, time the transmission and arrange for


@@ -780,6 +1129,8 @@ send:
                                tp->t_flags |= TF_SENTFIN;
                        }
                }
                                tp->t_flags |= TF_SENTFIN;
                        }
                }

+               if (sack_rxmit)
+                       goto timer;

                tp->snd_nxt += len;
                if (SEQ_GT(tp->snd_nxt, tp->snd_max)) {
                        tp->snd_max = tp->snd_nxt;
                tp->snd_nxt += len;
                if (SEQ_GT(tp->snd_nxt, tp->snd_max)) {
                        tp->snd_max = tp->snd_nxt;


@@ -787,8 +1138,8 @@ send:
                         * Time this transmission if not a retransmission and
                         * not currently timing anything.
                         */
                         * Time this transmission if not a retransmission and
                         * not currently timing anything.
                         */

-                       if (tp->t_rtt == 0) {
-                               tp->t_rtt = 1;
+                       if (tp->t_rtttime == 0) {
+                               tp->t_rtttime = 1;

                                tp->t_rtseq = startseq;
                                tcpstat.tcps_segstimed++;
                        }
                                tp->t_rtseq = startseq;
                                tcpstat.tcps_segstimed++;
                        }


@@ -802,39 +1153,39 @@ send:
                 * Initialize shift counter which is used for backoff
                 * of retransmit time.
                 */
                 * Initialize shift counter which is used for backoff
                 * of retransmit time.
                 */

+timer:

                if (tp->t_timer[TCPT_REXMT] == 0 &&
                if (tp->t_timer[TCPT_REXMT] == 0 &&

-                   tp->snd_nxt != tp->snd_una) {
-                       tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
+                   ((sack_rxmit && tp->snd_nxt != tp->snd_max) ||
+                       tp->snd_nxt != tp->snd_una)) {

                        if (tp->t_timer[TCPT_PERSIST]) {
                                tp->t_timer[TCPT_PERSIST] = 0;
                                tp->t_rxtshift = 0;
                        }
                        if (tp->t_timer[TCPT_PERSIST]) {
                                tp->t_timer[TCPT_PERSIST] = 0;
                                tp->t_rxtshift = 0;
                        }

+                       tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;

                }
                }

-       } else
-               if (SEQ_GT(tp->snd_nxt + len, tp->snd_max))
+       } else {
+               /*
+                * Persist case, update snd_max but since we are in
+                * persist mode (no window) we do not update snd_nxt.
+                */
+               int xlen = len;
+               if (flags & TH_SYN)
+                       ++xlen;
+               if (flags & TH_FIN) {
+                       ++xlen;
+                       tp->t_flags |= TF_SENTFIN;
+               }
+               if (SEQ_GT(tp->snd_nxt + xlen, tp->snd_max))

                        tp->snd_max = tp->snd_nxt + len;
                        tp->snd_max = tp->snd_nxt + len;

+       }

 
 #if TCPDEBUG
        /*
         * Trace.
         */
 
 #if TCPDEBUG
        /*
         * Trace.
         */

-       if (so->so_options & SO_DEBUG) {
-#if INET6
-               if (isipv6)
-                       ip6->ip6_vfc = IPV6_VERSION;
-               else
-                       ip->ip_vhl = IP_MAKE_VHL(IPVERSION,
-                                                IP_VHL_HL(ip->ip_vhl));
-#endif /* INET6 */
-               tcp_trace(TA_OUTPUT, tp->t_state, tp, 
-#if INET6
-                         isipv6 ? (void *)ip6 :
-#endif /* INET6 */
-                         ip,
-                         th, 0);
-
-       }
-#endif /* TCPDEBUG */
+       if (so->so_options & SO_DEBUG)
+               tcp_trace(TA_OUTPUT, tp->t_state, tp, mtod(m, void *), th, 0);
+#endif

 
        /*
         * Fill in IP length and desired time to live and
 
        /*
         * Fill in IP length and desired time to live and


@@ -842,9 +1193,13 @@ send:
         * to handle ttl and tos; we could keep them in
         * the template, but need a way to checksum without them.
         */
         * to handle ttl and tos; we could keep them in
         * the template, but need a way to checksum without them.
         */

+       /*
+        * m->m_pkthdr.len should have been set before cksum calcuration,
+        * because in6_cksum() need it.
+        */

 #if INET6
        if (isipv6) {
 #if INET6
        if (isipv6) {

-               /* 
+               /*

                 * we separately set hoplimit for every segment, since the
                 * user might want to change the value via setsockopt.
                 * Also, desired default hop limit might be changed via
                 * we separately set hoplimit for every segment, since the
                 * user might want to change the value via setsockopt.
                 * Also, desired default hop limit might be changed via


@@ -857,38 +1212,52 @@ send:
 
                /* TODO: IPv6 IP6TOS_ECT bit on */
 #if IPSEC
 
                /* TODO: IPv6 IP6TOS_ECT bit on */
 #if IPSEC

-               ipsec_setsocket(m, so);
+               if (ipsec_bypass == 0 && ipsec_setsocket(m, so) != 0) {
+                       m_freem(m);
+                       error = ENOBUFS;
+                       goto out;
+               }

 #endif /*IPSEC*/
 #endif /*IPSEC*/

+               m->m_pkthdr.socket_id = socket_id;

                error = ip6_output(m,
                            tp->t_inpcb->in6p_outputopts,
                            &tp->t_inpcb->in6p_route,
                error = ip6_output(m,
                            tp->t_inpcb->in6p_outputopts,
                            &tp->t_inpcb->in6p_route,

-                           (so->so_options & SO_DONTROUTE) /* | IP6_DONTFRAG */,
-                           NULL, NULL);
+                           (so->so_options & SO_DONTROUTE), NULL, NULL, 0);

        } else
 #endif /* INET6 */
        } else
 #endif /* INET6 */

-       {
-#if 1
+    {

        struct rtentry *rt;
        struct rtentry *rt;

-#endif

        ip->ip_len = m->m_pkthdr.len;
 #if INET6
        ip->ip_len = m->m_pkthdr.len;
 #if INET6

-       if (INP_CHECK_SOCKAF(so, AF_INET6))
-               ip->ip_ttl = in6_selecthlim(tp->t_inpcb,
-                                           tp->t_inpcb->in6p_route.ro_rt ?
-                                           tp->t_inpcb->in6p_route.ro_rt->rt_ifp
-                                           : NULL);
-       else
+       if (isipv6)
+               ip->ip_ttl = in6_selecthlim(tp->t_inpcb,
+                                           tp->t_inpcb->in6p_route.ro_rt ?
+                                           tp->t_inpcb->in6p_route.ro_rt->rt_ifp
+                                           : NULL);
+       else

 #endif /* INET6 */
        ip->ip_ttl = tp->t_inpcb->inp_ip_ttl;   /* XXX */
        ip->ip_tos = tp->t_inpcb->inp_ip_tos;   /* XXX */
 
 #endif /* INET6 */
        ip->ip_ttl = tp->t_inpcb->inp_ip_ttl;   /* XXX */
        ip->ip_tos = tp->t_inpcb->inp_ip_tos;   /* XXX */
 

-#define thtoti(x) \
-       ((struct tcpiphdr *)(((char *)(x)) - (sizeof (struct ip))))

 
 

-       KERNEL_DEBUG(DBG_LAYER_END, ((th->th_dport << 16) | th->th_sport),
-                  (((thtoti(th)->ti_src.s_addr & 0xffff) << 16) | (thtoti(th)->ti_dst.s_addr & 0xffff)),
-                   th->th_seq, th->th_ack, th->th_win);
-#if 1
+#if INET6
+       if (isipv6) {
+               KERNEL_DEBUG(DBG_LAYER_BEG,
+                    ((tp->t_inpcb->inp_fport << 16) | tp->t_inpcb->inp_lport),
+                    (((tp->t_inpcb->in6p_laddr.s6_addr16[0] & 0xffff) << 16) |
+                     (tp->t_inpcb->in6p_faddr.s6_addr16[0] & 0xffff)),
+                    0,0,0);
+       }
+        else 
+#endif
+       {
+               KERNEL_DEBUG(DBG_LAYER_BEG,
+                    ((tp->t_inpcb->inp_fport << 16) | tp->t_inpcb->inp_lport),
+                    (((tp->t_inpcb->inp_laddr.s_addr & 0xffff) << 16) |
+                     (tp->t_inpcb->inp_faddr.s_addr & 0xffff)),
+                    0,0,0);
+       }
+

        /*
         * See if we should do MTU discovery.  We do it only if the following
         * are true:
        /*
         * See if we should do MTU discovery.  We do it only if the following
         * are true:


@@ -896,31 +1265,91 @@ send:
         *      2) the MTU is not locked (if it is, then discovery has been
         *         disabled)
         */
         *      2) the MTU is not locked (if it is, then discovery has been
         *         disabled)
         */

-       if ((rt = tp->t_inpcb->inp_route.ro_rt)
+       if (path_mtu_discovery
+           && (rt = tp->t_inpcb->inp_route.ro_rt)

            && rt->rt_flags & RTF_UP
            && !(rt->rt_rmx.rmx_locks & RTV_MTU)) {
                ip->ip_off |= IP_DF;
        }
            && rt->rt_flags & RTF_UP
            && !(rt->rt_rmx.rmx_locks & RTV_MTU)) {
                ip->ip_off |= IP_DF;
        }

-#endif
-

 #if IPSEC
 #if IPSEC

-       ipsec_setsocket(m, so);
+       if (ipsec_bypass == 0)
+               ipsec_setsocket(m, so);

 #endif /*IPSEC*/
 
 #endif /*IPSEC*/
 

-       error = ip_output(m, tp->t_inpcb->inp_options, &tp->t_inpcb->inp_route,
-           so->so_options & SO_DONTROUTE, 0);
-    }
+       /*
+        * The socket is kept locked while sending out packets in ip_output, even if packet chaining is not active.
+        */
+
+       m->m_pkthdr.socket_id = socket_id;
+       if (packetlist) {
+               m->m_nextpkt = NULL;
+               lastpacket->m_nextpkt = m;
+               lastpacket = m;
+               packchain_listadd++;
+       }
+       else {
+               m->m_nextpkt = NULL;
+               packchain_newlist++;
+               packetlist = lastpacket = m;
+               packchain_listadd=0;
+       }
+
+       if ((ipsec_bypass == 0) || fw_enable || sendalot == 0 || (tp->t_state != TCPS_ESTABLISHED) || 
+                     (tp->snd_cwnd <= (tp->snd_wnd / 4)) || 
+                     (tp->t_flags & (TH_PUSH | TF_ACKNOW)) || tp->t_force != 0 ||
+                     packchain_listadd >= tcp_packet_chaining) {
+               lastpacket->m_nextpkt = 0;
+               error = ip_output_list(packetlist, packchain_listadd, tp->t_inpcb->inp_options, &tp->t_inpcb->inp_route,
+                       (so->so_options & SO_DONTROUTE), 0);
+               tp->t_lastchain = packchain_listadd;
+               packchain_sent++;
+               packetlist = NULL;
+               if (error == 0)
+                       howmuchsent = 0;
+       }
+       else {
+               error = 0;
+               packchain_looped++;
+               tcpstat.tcps_sndtotal++;
+               if (recwin > 0 && SEQ_GT(tp->rcv_nxt+recwin, tp->rcv_adv))
+                       tp->rcv_adv = tp->rcv_nxt + recwin;
+               tp->last_ack_sent = tp->rcv_nxt;
+               tp->t_flags &= ~(TF_ACKNOW|TF_DELACK);
+               goto again;
+       }
+   }

        if (error) {
        if (error) {

+
+               /*
+                * We know that the packet was lost, so back out the
+                * sequence number advance, if any.
+                */
+               if (tp->t_force == 0 || tp->t_timer[TCPT_PERSIST] == 0) {
+                       /*
+                        * No need to check for TH_FIN here because
+                        * the TF_SENTFIN flag handles that case.
+                        */
+                       if ((flags & TH_SYN) == 0) {
+                               if (sack_rxmit) {
+                                       p->rxmit -= howmuchsent;
+                                       tp->sackhint.sack_bytes_rexmit -= howmuchsent;
+                               } else
+                                       tp->snd_nxt -= howmuchsent;
+                       }
+               }
+               howmuchsent = 0;

 out:
                if (error == ENOBUFS) {
                         if (!tp->t_timer[TCPT_REXMT] &&
                              !tp->t_timer[TCPT_PERSIST])
                                 tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
                        tcp_quench(tp->t_inpcb, 0);
 out:
                if (error == ENOBUFS) {
                         if (!tp->t_timer[TCPT_REXMT] &&
                              !tp->t_timer[TCPT_PERSIST])
                                 tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
                        tcp_quench(tp->t_inpcb, 0);

+                       if (packetlist)
+                               m_freem_list(packetlist);
+                       tp->t_lastchain = 0;

                        KERNEL_DEBUG(DBG_FNC_TCP_OUTPUT | DBG_FUNC_END, 0,0,0,0,0);
                        return (0);
                }
                        KERNEL_DEBUG(DBG_FNC_TCP_OUTPUT | DBG_FUNC_END, 0,0,0,0,0);
                        return (0);
                }

-#if 1

                if (error == EMSGSIZE) {
                        /*
                         * ip_output() will have already fixed the route
                if (error == EMSGSIZE) {
                        /*
                         * ip_output() will have already fixed the route


@@ -929,19 +1358,28 @@ out:
                         * not do so here.
                         */
                        tcp_mtudisc(tp->t_inpcb, 0);
                         * not do so here.
                         */
                        tcp_mtudisc(tp->t_inpcb, 0);

+                       if (packetlist)
+                               m_freem_list(packetlist);
+                       tp->t_lastchain = 0;

                        KERNEL_DEBUG(DBG_FNC_TCP_OUTPUT | DBG_FUNC_END, 0,0,0,0,0);
                        return 0;
                }
                        KERNEL_DEBUG(DBG_FNC_TCP_OUTPUT | DBG_FUNC_END, 0,0,0,0,0);
                        return 0;
                }

-#endif

                if ((error == EHOSTUNREACH || error == ENETDOWN)
                    && TCPS_HAVERCVDSYN(tp->t_state)) {
                        tp->t_softerror = error;
                if ((error == EHOSTUNREACH || error == ENETDOWN)
                    && TCPS_HAVERCVDSYN(tp->t_state)) {
                        tp->t_softerror = error;

+                       if (packetlist)
+                               m_freem_list(packetlist);
+                       tp->t_lastchain = 0;

                        KERNEL_DEBUG(DBG_FNC_TCP_OUTPUT | DBG_FUNC_END, 0,0,0,0,0);
                        return (0);
                }
                        KERNEL_DEBUG(DBG_FNC_TCP_OUTPUT | DBG_FUNC_END, 0,0,0,0,0);
                        return (0);
                }

+               if (packetlist)
+                       m_freem_list(packetlist);
+               tp->t_lastchain = 0;

                KERNEL_DEBUG(DBG_FNC_TCP_OUTPUT | DBG_FUNC_END, 0,0,0,0,0);
                return (error);
        }
                KERNEL_DEBUG(DBG_FNC_TCP_OUTPUT | DBG_FUNC_END, 0,0,0,0,0);
                return (error);
        }

+sentit:

        tcpstat.tcps_sndtotal++;
 
        /*
        tcpstat.tcps_sndtotal++;
 
        /*


@@ -950,13 +1388,14 @@ out:
         * then remember the size of the advertised window.
         * Any pending ACK has now been sent.
         */
         * then remember the size of the advertised window.
         * Any pending ACK has now been sent.
         */

-       if (win > 0 && SEQ_GT(tp->rcv_nxt+win, tp->rcv_adv))
-               tp->rcv_adv = tp->rcv_nxt + win;
+       if (recwin > 0 && SEQ_GT(tp->rcv_nxt+recwin, tp->rcv_adv))
+               tp->rcv_adv = tp->rcv_nxt + recwin;

        tp->last_ack_sent = tp->rcv_nxt;
        tp->t_flags &= ~(TF_ACKNOW|TF_DELACK);
        tp->last_ack_sent = tp->rcv_nxt;
        tp->t_flags &= ~(TF_ACKNOW|TF_DELACK);

-       if (sendalot)
+
+       KERNEL_DEBUG(DBG_FNC_TCP_OUTPUT | DBG_FUNC_END,0,0,0,0,0);
+       if (sendalot && (!tcp_do_newreno || --maxburst))

                goto again;
                goto again;

-       KERNEL_DEBUG(DBG_FNC_TCP_OUTPUT | DBG_FUNC_END, 0,0,0,0,0);

        return (0);
 }
 
        return (0);
 }
 


@@ -964,10 +1403,10 @@ void
 tcp_setpersist(tp)
        register struct tcpcb *tp;
 {
 tcp_setpersist(tp)
        register struct tcpcb *tp;
 {

-       register int t = ((tp->t_srtt >> 2) + tp->t_rttvar) >> 1;
+       int t = ((tp->t_srtt >> 2) + tp->t_rttvar) >> 1;

 
        if (tp->t_timer[TCPT_REXMT])
 
        if (tp->t_timer[TCPT_REXMT])

-               panic("tcp_output REXMT");
+               panic("tcp_setpersist: retransmit pending");

        /*
         * Start/restart persistance timer.
         */
        /*
         * Start/restart persistance timer.
         */