X-Git-Url: https://git.saurik.com/apple/xnu.git/blobdiff_plain/0b4e3aa066abc0728aacb4bbeb86f53f9737156e..13fec9890cf095cc781fdf7b8917cb03bf32dd4c:/bsd/netinet/tcp_usrreq.c

diff --git a/bsd/netinet/tcp_usrreq.c b/bsd/netinet/tcp_usrreq.c
index 2dd2747fe..2a5ab6988 100644
--- a/bsd/netinet/tcp_usrreq.c
+++ b/bsd/netinet/tcp_usrreq.c
@@ -52,11 +52,9 @@
  * SUCH DAMAGE.
  *
  *	From: @(#)tcp_usrreq.c	8.2 (Berkeley) 1/3/94
+ * $FreeBSD: src/sys/netinet/tcp_usrreq.c,v 1.51.2.9 2001/08/22 00:59:12 silby Exp $
  */
 
-#if ISFB31
-#include "opt_tcpdebug.h"
-#endif
 
 #include <sys/param.h>
 #include <sys/systm.h>
@@ -75,14 +73,16 @@
 
 #include <netinet/in.h>
 #include <netinet/in_systm.h>
-#include <netinet/ip.h>
+#if INET6
+#include <netinet/ip6.h>
+#endif
 #include <netinet/in_pcb.h>
+#if INET6
+#include <netinet6/in6_pcb.h>
+#endif
 #include <netinet/in_var.h>
 #include <netinet/ip_var.h>
 #if INET6
-#include <netinet/ip6.h>
-#include <netinet6/ip6_var.h>
-#include <netinet6/in6_pcb.h>
 #include <netinet6/ip6_var.h>
 #endif
 #include <netinet/tcp.h>
@@ -104,22 +104,21 @@
  */
 extern	char *tcpstates[];	/* XXX ??? */
 
-static int	tcp_attach __P((struct socket *, struct proc *));
-static int	tcp_connect __P((struct tcpcb *, struct sockaddr *, 
-				 struct proc *));
+static int	tcp_attach(struct socket *, struct proc *);
+static int	tcp_connect(struct tcpcb *, struct sockaddr *, struct proc *);
 #if INET6
-static int	tcp6_connect __P((struct tcpcb *, struct sockaddr *,
-				 struct proc *));
+static int	tcp6_connect(struct tcpcb *, struct sockaddr *, struct proc *);
 #endif /* INET6 */
-static struct tcpcb *	tcp_disconnect __P((struct tcpcb *));
 static struct tcpcb *
-		tcp_usrclosed __P((struct tcpcb *));
+		tcp_disconnect(struct tcpcb *);
+static struct tcpcb *
+		tcp_usrclosed(struct tcpcb *);
 
 #if TCPDEBUG
-#define	TCPDEBUG0	int ostate
+#define	TCPDEBUG0	int ostate = 0
 #define	TCPDEBUG1()	ostate = tp ? tp->t_state : 0
 #define	TCPDEBUG2(req)	if (tp && (so->so_options & SO_DEBUG)) \
-				tcp_trace(TA_USER, ostate, tp, 0, req)
+				tcp_trace(TA_USER, ostate, tp, 0, 0, req)
 #else
 #define	TCPDEBUG0
 #define	TCPDEBUG1()
@@ -133,7 +132,6 @@ static struct tcpcb *
 static int
 tcp_usr_attach(struct socket *so, int proto, struct proc *p)
 {
-	int s = splnet();
 	int error;
 	struct inpcb *inp = sotoinpcb(so);
 	struct tcpcb *tp = 0;
@@ -154,7 +152,6 @@ tcp_usr_attach(struct socket *so, int proto, struct proc *p)
 	tp = sototcpcb(so);
 out:
 	TCPDEBUG2(PRU_ATTACH);
-	splx(s);
 	return error;
 }
 
@@ -168,16 +165,17 @@ out:
 static int
 tcp_usr_detach(struct socket *so)
 {
-	int s = splnet();
 	int error = 0;
 	struct inpcb *inp = sotoinpcb(so);
 	struct tcpcb *tp;
 	TCPDEBUG0;
 
-	if (inp == 0) {
-		splx(s);
+	if (inp == 0 || (inp->inp_state == INPCB_STATE_DEAD)) {
 		return EINVAL;	/* XXX */
 	}
+#if 1
+	lck_mtx_assert(((struct inpcb *)so->so_pcb)->inpcb_mtx, LCK_MTX_ASSERT_OWNED);
+#endif
 	tp = intotcpcb(inp);
 	/* In case we got disconnected from the peer */
         if (tp == 0) 
@@ -186,21 +184,19 @@ tcp_usr_detach(struct socket *so)
 	tp = tcp_disconnect(tp);
 out:
 	TCPDEBUG2(PRU_DETACH);
-	splx(s);
 	return error;
 }
 
 #define	COMMON_START()	TCPDEBUG0; \
 			do { \
-				     if (inp == 0) { \
-					     splx(s); \
+				     if (inp == 0 || (inp->inp_state == INPCB_STATE_DEAD)) { \
 					     return EINVAL; \
 				     } \
 				     tp = intotcpcb(inp); \
 				     TCPDEBUG1(); \
 		     } while(0)
 			     
-#define COMMON_END(req)	out: TCPDEBUG2(req); splx(s); return error; goto out
+#define COMMON_END(req)	out: TCPDEBUG2(req); return error; goto out
 
 
 /*
@@ -209,7 +205,6 @@ out:
 static int
 tcp_usr_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
 {
-	int s = splnet();
 	int error = 0;
 	struct inpcb *inp = sotoinpcb(so);
 	struct tcpcb *tp;
@@ -238,7 +233,6 @@ tcp_usr_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
 static int
 tcp6_usr_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
 {
-	int s = splnet();
 	int error = 0;
 	struct inpcb *inp = sotoinpcb(so);
 	struct tcpcb *tp;
@@ -258,8 +252,7 @@ tcp6_usr_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
 	}
 	inp->inp_vflag &= ~INP_IPV4;
 	inp->inp_vflag |= INP_IPV6;
-	if (ip6_mapped_addr_on && (inp->inp_flags & IN6P_BINDV6ONLY) == NULL) {
-
+	if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0) {
 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6p->sin6_addr))
 			inp->inp_vflag |= INP_IPV4;
 		else if (IN6_IS_ADDR_V4MAPPED(&sin6p->sin6_addr)) {
@@ -273,6 +266,8 @@ tcp6_usr_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
 		}
 	}
 	error = in6_pcbbind(inp, nam, p);
+	if (error)
+		goto out;
 	COMMON_END(PRU_BIND);
 }
 #endif /* INET6 */
@@ -283,7 +278,6 @@ tcp6_usr_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
 static int
 tcp_usr_listen(struct socket *so, struct proc *p)
 {
-	int s = splnet();
 	int error = 0;
 	struct inpcb *inp = sotoinpcb(so);
 	struct tcpcb *tp;
@@ -300,7 +294,6 @@ tcp_usr_listen(struct socket *so, struct proc *p)
 static int
 tcp6_usr_listen(struct socket *so, struct proc *p)
 {
-	int s = splnet();
 	int error = 0;
 	struct inpcb *inp = sotoinpcb(so);
 	struct tcpcb *tp;
@@ -308,8 +301,7 @@ tcp6_usr_listen(struct socket *so, struct proc *p)
 	COMMON_START();
 	if (inp->inp_lport == 0) {
 		inp->inp_vflag &= ~INP_IPV4;
-		if (ip6_mapped_addr_on &&
-		    (inp->inp_flags & IN6P_BINDV6ONLY) == NULL)
+		if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0)
 			inp->inp_vflag |= INP_IPV4;
 		error = in6_pcbbind(inp, (struct sockaddr *)0, p);
 	}
@@ -329,7 +321,6 @@ tcp6_usr_listen(struct socket *so, struct proc *p)
 static int
 tcp_usr_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
 {
-	int s = splnet();
 	int error = 0;
 	struct inpcb *inp = sotoinpcb(so);
 	struct tcpcb *tp;
@@ -347,6 +338,10 @@ tcp_usr_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
 		goto out;
 	}
 
+#ifndef __APPLE__
+	prison_remote_ip(p, 0, &sinp->sin_addr.s_addr);
+#endif
+
 	if ((error = tcp_connect(tp, nam, p)) != 0)
 		goto out;
 	error = tcp_output(tp);
@@ -357,7 +352,6 @@ tcp_usr_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
 static int
 tcp6_usr_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
 {
-	int s = splnet();
 	int error = 0;
 	struct inpcb *inp = sotoinpcb(so);
 	struct tcpcb *tp;
@@ -374,12 +368,13 @@ tcp6_usr_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
 		error = EAFNOSUPPORT;
 		goto out;
 	}
-	inp->inp_vflag &= ~INP_IPV4;
-	inp->inp_vflag |= INP_IPV6;
-	if (ip6_mapped_addr_on &&
-	    IN6_IS_ADDR_V4MAPPED(&sin6p->sin6_addr)) {
+
+	if (IN6_IS_ADDR_V4MAPPED(&sin6p->sin6_addr)) {
 		struct sockaddr_in sin;
 
+		if ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0)
+			return (EINVAL);
+
 		in6_sin6_2_sin(&sin, sin6p);
 		inp->inp_vflag |= INP_IPV4;
 		inp->inp_vflag &= ~INP_IPV6;
@@ -388,13 +383,13 @@ tcp6_usr_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
 		error = tcp_output(tp);
 		goto out;
 	}
+	inp->inp_vflag &= ~INP_IPV4;
+	inp->inp_vflag |= INP_IPV6;
 	if ((error = tcp6_connect(tp, nam, p)) != 0)
 		goto out;
 	error = tcp_output(tp);
 	if (error)
 		goto out;
-	if (ip6_mapped_addr_on)
-		inp->inp_vflag |= INP_IPV6;
 	COMMON_END(PRU_CONNECT);
 }
 #endif /* INET6 */
@@ -413,11 +408,13 @@ tcp6_usr_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
 static int
 tcp_usr_disconnect(struct socket *so)
 {
-	int s = splnet();
 	int error = 0;
 	struct inpcb *inp = sotoinpcb(so);
 	struct tcpcb *tp;
-
+	
+#if 1
+	lck_mtx_assert(((struct inpcb *)so->so_pcb)->inpcb_mtx, LCK_MTX_ASSERT_OWNED);
+#endif
 	COMMON_START();
         /* In case we got disconnected from the peer */
         if (tp == 0)
@@ -434,12 +431,20 @@ tcp_usr_disconnect(struct socket *so)
 static int
 tcp_usr_accept(struct socket *so, struct sockaddr **nam)
 {
-	int s = splnet();
 	int error = 0;
 	struct inpcb *inp = sotoinpcb(so);
-	struct tcpcb *tp;
+	struct tcpcb *tp = NULL;
+	TCPDEBUG0;
 
-	COMMON_START();
+	if (so->so_state & SS_ISDISCONNECTED) {
+		error = ECONNABORTED;
+		goto out;
+	}
+	if (inp == 0 || (inp->inp_state == INPCB_STATE_DEAD)) {
+		return (EINVAL);
+	}
+	tp = intotcpcb(inp);
+	TCPDEBUG1();
 	in_setpeeraddr(so, nam);
 	COMMON_END(PRU_ACCEPT);
 }
@@ -448,24 +453,30 @@ tcp_usr_accept(struct socket *so, struct sockaddr **nam)
 static int
 tcp6_usr_accept(struct socket *so, struct sockaddr **nam)
 {
-	int s = splnet();
 	int error = 0;
 	struct inpcb *inp = sotoinpcb(so);
-	struct tcpcb *tp;
+	struct tcpcb *tp = NULL;
+	TCPDEBUG0;
 
-	COMMON_START();
+	if (so->so_state & SS_ISDISCONNECTED) {
+		error = ECONNABORTED;
+		goto out;
+	}
+	if (inp == 0 || (inp->inp_state == INPCB_STATE_DEAD)) {
+		return (EINVAL);
+	}
+	tp = intotcpcb(inp);
+	TCPDEBUG1();
 	in6_mapped_peeraddr(so, nam);
 	COMMON_END(PRU_ACCEPT);
 }
 #endif /* INET6 */
-
 /*
  * Mark the connection as being incapable of further output.
  */
 static int
 tcp_usr_shutdown(struct socket *so)
 {
-	int s = splnet();
 	int error = 0;
 	struct inpcb *inp = sotoinpcb(so);
 	struct tcpcb *tp;
@@ -487,7 +498,6 @@ tcp_usr_shutdown(struct socket *so)
 static int
 tcp_usr_rcvd(struct socket *so, int flags)
 {
-	int s = splnet();
 	int error = 0;
 	struct inpcb *inp = sotoinpcb(so);
 	struct tcpcb *tp;
@@ -502,33 +512,54 @@ tcp_usr_rcvd(struct socket *so, int flags)
 
 /*
  * Do a send by putting data in output queue and updating urgent
- * marker if URG set.  Possibly send more data.
+ * marker if URG set.  Possibly send more data.  Unlike the other
+ * pru_*() routines, the mbuf chains are our responsibility.  We
+ * must either enqueue them or free them.  The other pru_* routines
+ * generally are caller-frees.
  */
 static int
-tcp_usr_send(struct socket *so, int flags, struct mbuf *m,
+tcp_usr_send(struct socket *so, int flags, struct mbuf *m, 
 	     struct sockaddr *nam, struct mbuf *control, struct proc *p)
 {
-	int s = splnet();
 	int error = 0;
 	struct inpcb *inp = sotoinpcb(so);
 	struct tcpcb *tp;
 #if INET6
 	int isipv6;
-#endif /* INET6 */
+#endif
+	TCPDEBUG0;
 
-	COMMON_START();
-	if (control && control->m_len) {
-		m_freem(control); /* XXX shouldn't caller do this??? */
+	if (inp == NULL || inp->inp_state == INPCB_STATE_DEAD) {
+		/*
+		 * OOPS! we lost a race, the TCP session got reset after
+		 * we checked SS_CANTSENDMORE, eg: while doing uiomove or a
+		 * network interrupt in the non-splnet() section of sosend().
+		 */
 		if (m)
 			m_freem(m);
-		error = EINVAL;
+		if (control)
+			m_freem(control);
+		error = ECONNRESET;	/* XXX EPIPE? */
+		tp = NULL;
+		TCPDEBUG1();
 		goto out;
 	}
-
 #if INET6
 	isipv6 = nam && nam->sa_family == AF_INET6;
 #endif /* INET6 */
-
+	tp = intotcpcb(inp);
+	TCPDEBUG1();
+	if (control) {
+		/* TCP doesn't do control messages (rights, creds, etc) */
+		if (control->m_len) {
+			m_freem(control);
+			if (m)
+				m_freem(m);
+			error = EINVAL;
+			goto out;
+		}
+		m_freem(control);	/* empty control, just free it */
+	}
 	if(!(flags & PRUS_OOB)) {
 		sbappend(&so->so_snd, m);
 		if (nam && tp->t_state < TCPS_SYN_SENT) {
@@ -547,7 +578,7 @@ tcp_usr_send(struct socket *so, int flags, struct mbuf *m,
 			if (error)
 				goto out;
 			tp->snd_wnd = TTCP_CLIENT_SND_WND;
-			tcp_mss(tp, -1, isipv6);
+			tcp_mss(tp, -1);
 		}
 
 		if (flags & PRUS_EOF) {
@@ -596,7 +627,7 @@ tcp_usr_send(struct socket *so, int flags, struct mbuf *m,
 			if (error)
 				goto out;
 			tp->snd_wnd = TTCP_CLIENT_SND_WND;
-			tcp_mss(tp, -1, isipv6);
+			tcp_mss(tp, -1);
 		}
 		tp->snd_up = tp->snd_una + so->so_snd.sb_cc;
 		tp->t_force = 1;
@@ -613,7 +644,6 @@ tcp_usr_send(struct socket *so, int flags, struct mbuf *m,
 static int
 tcp_usr_abort(struct socket *so)
 {
-	int s = splnet();
 	int error = 0;
 	struct inpcb *inp = sotoinpcb(so);
 	struct tcpcb *tp;
@@ -623,6 +653,7 @@ tcp_usr_abort(struct socket *so)
         if (tp == 0)
             goto out;
 	tp = tcp_drop(tp, ECONNABORTED);
+	so->so_usecount--;
 	COMMON_END(PRU_ABORT);
 }
 
@@ -632,7 +663,6 @@ tcp_usr_abort(struct socket *so)
 static int
 tcp_usr_rcvoob(struct socket *so, struct mbuf *m, int flags)
 {
-	int s = splnet();
 	int error = 0;
 	struct inpcb *inp = sotoinpcb(so);
 	struct tcpcb *tp;
@@ -662,7 +692,7 @@ struct pr_usrreqs tcp_usrreqs = {
 	tcp_usr_connect, pru_connect2_notsupp, in_control, tcp_usr_detach,
 	tcp_usr_disconnect, tcp_usr_listen, in_setpeeraddr, tcp_usr_rcvd,
 	tcp_usr_rcvoob, tcp_usr_send, pru_sense_null, tcp_usr_shutdown,
-	in_setsockaddr, sosend, soreceive, sopoll
+	in_setsockaddr, sosend, soreceive, pru_sopoll_notsupp
 };
 
 #if INET6
@@ -671,7 +701,7 @@ struct pr_usrreqs tcp6_usrreqs = {
 	tcp6_usr_connect, pru_connect2_notsupp, in6_control, tcp_usr_detach,
 	tcp_usr_disconnect, tcp6_usr_listen, in6_mapped_peeraddr, tcp_usr_rcvd,
 	tcp_usr_rcvoob, tcp_usr_send, pru_sense_null, tcp_usr_shutdown,
-	in6_mapped_sockaddr, sosend, soreceive, sopoll
+	in6_mapped_sockaddr, sosend, soreceive, pru_sopoll_notsupp
 };
 #endif /* INET6 */
 
@@ -714,35 +744,55 @@ tcp_connect(tp, nam, p)
 	error = in_pcbladdr(inp, nam, &ifaddr);
 	if (error)
 		return error;
+
+	tcp_unlock(inp->inp_socket, 0, 0);
 	oinp = in_pcblookup_hash(inp->inp_pcbinfo,
 	    sin->sin_addr, sin->sin_port,
 	    inp->inp_laddr.s_addr != INADDR_ANY ? inp->inp_laddr
 						: ifaddr->sin_addr,
 	    inp->inp_lport,  0, NULL);
+
+	tcp_lock(inp->inp_socket, 0, 0);
 	if (oinp) {
+		if (oinp != inp) /* 4143933: avoid deadlock if inp == oinp */
+			tcp_lock(oinp->inp_socket, 1, 0);
+		if (in_pcb_checkstate(oinp, WNT_RELEASE, 1) == WNT_STOPUSING) {
+			if (oinp != inp)
+				tcp_unlock(oinp->inp_socket, 1, 0);
+			goto skip_oinp;
+		}
+
 		if (oinp != inp && (otp = intotcpcb(oinp)) != NULL &&
 		otp->t_state == TCPS_TIME_WAIT &&
-		    otp->t_duration < TCPTV_MSL &&
+		    otp->t_starttime < tcp_msl &&
 		    (otp->t_flags & TF_RCVD_CC))
 			otp = tcp_close(otp);
-		else
+		else {
+			printf("tcp_connect: inp=%x err=EADDRINUSE\n", inp);
+			if (oinp != inp)
+				tcp_unlock(oinp->inp_socket, 1, 0);
 			return EADDRINUSE;
+		}
+		if (oinp != inp)
+			tcp_unlock(oinp->inp_socket, 1, 0);
 	}
+skip_oinp:
 	if ((inp->inp_laddr.s_addr == INADDR_ANY ? ifaddr->sin_addr.s_addr :
 		 inp->inp_laddr.s_addr) == sin->sin_addr.s_addr &&
 	    inp->inp_lport == sin->sin_port)
 			return EINVAL;
+	if (!lck_rw_try_lock_exclusive(inp->inp_pcbinfo->mtx)) {
+		/*lock inversion issue, mostly with udp multicast packets */
+		socket_unlock(inp->inp_socket, 0);
+		lck_rw_lock_exclusive(inp->inp_pcbinfo->mtx);
+		socket_lock(inp->inp_socket, 0);
+	}
 	if (inp->inp_laddr.s_addr == INADDR_ANY)
 		inp->inp_laddr = ifaddr->sin_addr;
 	inp->inp_faddr = sin->sin_addr;
 	inp->inp_fport = sin->sin_port;
 	in_pcbrehash(inp);
-
-	tp->t_template = tcp_template(tp);
-	if (tp->t_template == 0) {
-		in_pcbdisconnect(inp);
-		return ENOBUFS;
-	}
+	lck_rw_done(inp->inp_pcbinfo->mtx);
 
 	/* Compute window scaling to request.  */
 	while (tp->request_r_scale < TCP_MAX_WINSHIFT &&
@@ -753,13 +803,7 @@ tcp_connect(tp, nam, p)
 	tcpstat.tcps_connattempt++;
 	tp->t_state = TCPS_SYN_SENT;
 	tp->t_timer[TCPT_KEEP] = tcp_keepinit;
-#ifdef TCP_COMPAT_42
-	tp->iss = tcp_iss;
-	tcp_iss =+ TCP_ISSINCR/2;
-#else /* TCP_COMPAT_42 */
-	tp->iss = tcp_rndiss_next();
-#endif /* !TCP_COMPAT_42 */
-
+	tp->iss = tcp_new_isn(tp);
 	tcp_sendseqinit(tp);
 
 	/*
@@ -794,7 +838,7 @@ tcp6_connect(tp, nam, p)
 	struct socket *so = inp->inp_socket;
 	struct tcpcb *otp;
 	struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
-	struct in6_addr *addr6;
+	struct in6_addr addr6;
 	struct rmxp_tao *taop;
 	struct rmxp_tao tao_noncached;
 	int error;
@@ -813,38 +857,37 @@ tcp6_connect(tp, nam, p)
 	error = in6_pcbladdr(inp, nam, &addr6);
 	if (error)
 		return error;
+	tcp_unlock(inp->inp_socket, 0, 0);
 	oinp = in6_pcblookup_hash(inp->inp_pcbinfo,
 				  &sin6->sin6_addr, sin6->sin6_port,
 				  IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)
-				  ? addr6
+				  ? &addr6
 				  : &inp->in6p_laddr,
 				  inp->inp_lport,  0, NULL);
+	tcp_lock(inp->inp_socket, 0, 0);
 	if (oinp) {
 		if (oinp != inp && (otp = intotcpcb(oinp)) != NULL &&
 		    otp->t_state == TCPS_TIME_WAIT &&
-		    otp->t_duration < TCPTV_MSL &&
+		    otp->t_starttime < tcp_msl &&
 		    (otp->t_flags & TF_RCVD_CC))
 			otp = tcp_close(otp);
 		else
 			return EADDRINUSE;
 	}
+	if (!lck_rw_try_lock_exclusive(inp->inp_pcbinfo->mtx)) {
+		/*lock inversion issue, mostly with udp multicast packets */
+		socket_unlock(inp->inp_socket, 0);
+		lck_rw_lock_exclusive(inp->inp_pcbinfo->mtx);
+		socket_lock(inp->inp_socket, 0);
+	}
 	if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
-		inp->in6p_laddr = *addr6;
+		inp->in6p_laddr = addr6;
 	inp->in6p_faddr = sin6->sin6_addr;
 	inp->inp_fport = sin6->sin6_port;
-	/*
-	 * xxx kazu flowlabel is necessary for connect?
-	 * but if this line is missing, the garbage value remains.
-	 */
-	inp->in6p_flowinfo = sin6->sin6_flowinfo;
-
+	if ((sin6->sin6_flowinfo & IPV6_FLOWINFO_MASK) != NULL)
+		inp->in6p_flowinfo = sin6->sin6_flowinfo;
 	in_pcbrehash(inp);
-
-	tp->t_template = tcp_template(tp);
-	if (tp->t_template == 0) {
-		in6_pcbdisconnect(inp);
-		return ENOBUFS;
-	}
+	lck_rw_done(inp->inp_pcbinfo->mtx);
 
 	/* Compute window scaling to request.  */
 	while (tp->request_r_scale < TCP_MAX_WINSHIFT &&
@@ -855,11 +898,7 @@ tcp6_connect(tp, nam, p)
 	tcpstat.tcps_connattempt++;
 	tp->t_state = TCPS_SYN_SENT;
 	tp->t_timer[TCPT_KEEP] = tcp_keepinit;
-#ifdef TCP_COMPAT_42
-	tp->iss = tcp_iss; tcp_iss += TCP_ISSINCR/2;
-#else
-	tp->iss = tcp_rndiss_next();
-#endif /* TCP_COMPAT_42 */
+	tp->iss = tcp_new_isn(tp);
 	tcp_sendseqinit(tp);
 
 	/*
@@ -896,15 +935,13 @@ tcp_ctloutput(so, sopt)
 	struct socket *so;
 	struct sockopt *sopt;
 {
-	int	error, opt, optval, s;
+	int	error, opt, optval;
 	struct	inpcb *inp;
 	struct	tcpcb *tp;
 
 	error = 0;
-	s = splnet();		/* XXX */
 	inp = sotoinpcb(so);
 	if (inp == NULL) {
-		splx(s);
 		return (ECONNRESET);
 	}
 	if (sopt->sopt_level != IPPROTO_TCP) {
@@ -914,12 +951,10 @@ tcp_ctloutput(so, sopt)
 		else
 #endif /* INET6 */
 		error = ip_ctloutput(so, sopt);
-		splx(s);
 		return (error);
 	}
 	tp = intotcpcb(inp);
         if (tp == NULL) {
-                splx(s);
                 return (ECONNRESET);
         }
 
@@ -961,12 +996,24 @@ tcp_ctloutput(so, sopt)
 			if (error)
 				break;
 
-			if (optval > 0 && optval <= tp->t_maxseg)
+			if (optval > 0 && optval <= tp->t_maxseg &&
+			    optval + 40 >= tcp_minmss)
 				tp->t_maxseg = optval;
 			else
 				error = EINVAL;
 			break;
 
+                case TCP_KEEPALIVE:
+                        error = sooptcopyin(sopt, &optval, sizeof optval,
+                                            sizeof optval);
+                        if (error)
+                                break;
+                        if (optval < 0)
+                                error = EINVAL;
+			else
+				tp->t_keepidle = optval * PR_SLOWHZ;
+                        break;
+		
 		default:
 			error = ENOPROTOOPT;
 			break;
@@ -981,6 +1028,9 @@ tcp_ctloutput(so, sopt)
 		case TCP_MAXSEG:
 			optval = tp->t_maxseg;
 			break;
+		case TCP_KEEPALIVE:
+			optval = tp->t_keepidle / PR_SLOWHZ;
+			break;
 		case TCP_NOOPT:
 			optval = tp->t_flags & TF_NOOPT;
 			break;
@@ -995,7 +1045,6 @@ tcp_ctloutput(so, sopt)
 			error = sooptcopyout(sopt, &optval, sizeof optval);
 		break;
 	}
-	splx(s);
 	return (error);
 }
 
@@ -1005,12 +1054,17 @@ tcp_ctloutput(so, sopt)
  * be set by the route).
  */
 u_long	tcp_sendspace = 1024*16;
-SYSCTL_INT(_net_inet_tcp, TCPCTL_SENDSPACE, sendspace,
-	CTLFLAG_RW, &tcp_sendspace , 0, "");
+SYSCTL_INT(_net_inet_tcp, TCPCTL_SENDSPACE, sendspace, CTLFLAG_RW, 
+    &tcp_sendspace , 0, "Maximum outgoing TCP datagram size");
 u_long	tcp_recvspace = 1024*16;
-SYSCTL_INT(_net_inet_tcp, TCPCTL_RECVSPACE, recvspace,
-	CTLFLAG_RW, &tcp_recvspace , 0, "");
+SYSCTL_INT(_net_inet_tcp, TCPCTL_RECVSPACE, recvspace, CTLFLAG_RW, 
+    &tcp_recvspace , 0, "Maximum incoming TCP datagram size");
+
+__private_extern__ int	tcp_sockthreshold = 256;
+SYSCTL_INT(_net_inet_tcp, OID_AUTO, sockthreshold, CTLFLAG_RW, 
+    &tcp_sockthreshold , 0, "TCP Socket size increased if less than threshold");
 
+#define TCP_INCREASED_SPACE	65535	/* Automatically increase tcp send/rcv space to this value */
 /*
  * Attach TCP protocol to socket, allocating
  * internet protocol control block, tcp control block,
@@ -1025,30 +1079,31 @@ tcp_attach(so, p)
 	struct inpcb *inp;
 	int error;
 #if INET6
-	int isipv6 = INP_CHECK_SOCKAF(so, AF_INET) == NULL;
-#endif /* INET6 */
+	int isipv6 = INP_CHECK_SOCKAF(so, AF_INET6) != NULL;
+#endif
 
-	if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
-		error = soreserve(so, tcp_sendspace, tcp_recvspace);
-		if (error)
-			return (error);
-	}
 	error = in_pcballoc(so, &tcbinfo, p);
 	if (error)
 		return (error);
+
 	inp = sotoinpcb(so);
-#if IPSEC
-	error = ipsec_init_policy(so, &inp->inp_sp);
-	if (error) {
-#if INET6
-		if (isipv6)
-			in6_pcbdetach(inp);
-		else
-#endif /* INET6 */
-		in_pcbdetach(inp);
-		return (error);
+
+	if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
+		/*
+		 * The goal is to let clients have large send/rcv default windows (TCP_INCREASED_SPACE)
+		 * while not hogging mbuf space for servers. This is done by watching a threshold
+		 * of tcpcbs in use and bumping the default send and rcvspace only if under that threshold.
+		 * The theory being that busy servers have a lot more active tcpcbs and don't want the potential
+		 * memory penalty of having much larger sockbuffs. The sysctl allows to fine tune that threshold value.		 */
+
+		if (inp->inp_pcbinfo->ipi_count < tcp_sockthreshold)
+			error = soreserve(so, MAX(TCP_INCREASED_SPACE, tcp_sendspace), MAX(TCP_INCREASED_SPACE,tcp_recvspace));
+		else	
+			error = soreserve(so, tcp_sendspace, tcp_recvspace);
+		if (error)
+			return (error);
 	}
-#endif /*IPSEC*/
+
 #if INET6
 	if (isipv6) {
 		inp->inp_vflag |= INP_IPV6;