X-Git-Url: https://git.saurik.com/apple/xnu.git/blobdiff_plain/6d2010ae8f7a6078e10b361c6962983bab233e0f..c18c124eaa464aaaa5549e99e5a70fc9cbb50944:/bsd/kern/kpi_socket.c

diff --git a/bsd/kern/kpi_socket.c b/bsd/kern/kpi_socket.c
index 70507beff..07c65bf39 100644
--- a/bsd/kern/kpi_socket.c
+++ b/bsd/kern/kpi_socket.c
@@ -1,8 +1,8 @@
 /*
- * Copyright (c) 2003-2011 Apple Inc. All rights reserved.
+ * Copyright (c) 2003-2012 Apple Inc. All rights reserved.
  *
  * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
- * 
+ *
  * 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
@@ -11,10 +11,10 @@
  * 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.
- * 
+ *
  * 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,
@@ -22,11 +22,11 @@
  * 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_OSREFERENCE_LICENSE_HEADER_END@
  */
 
-#define __KPI__
+#define	__KPI__
 #include <sys/systm.h>
 #include <sys/kernel.h>
 #include <sys/types.h>
@@ -42,83 +42,70 @@
 #include <sys/fcntl.h>
 #include <sys/filio.h>
 #include <sys/uio_internal.h>
-#include <kern/lock.h>
+#include <kern/locks.h>
 #include <netinet/in.h>
 #include <libkern/OSAtomic.h>
 
-extern int soclose_locked(struct socket *so);
-extern void soclose_wait_locked(struct socket *so);
-extern int so_isdstlocal(struct socket *so);
-
-errno_t sock_send_internal(
-	socket_t			sock,
-	const struct msghdr	*msg,
-	mbuf_t				data,
-	int					flags,
-	size_t				*sentlen);
-
-typedef	void	(*so_upcall)(struct socket *, caddr_t , int );
-
+static errno_t sock_send_internal(socket_t, const struct msghdr	*,
+    mbuf_t, int, size_t	*);
+static void sock_setupcalls_common(socket_t, sock_upcall, void *,
+    sock_upcall, void *);
 
 errno_t
-sock_accept(
-	socket_t		sock,
-	struct sockaddr	*from,
-	int				fromlen,
-	int				flags,
-	sock_upcall		callback,
-	void*			cookie,
-	socket_t		*new_sock)
+sock_accept(socket_t sock, struct sockaddr *from, int fromlen, int flags,
+    sock_upcall callback, void *cookie, socket_t *new_sock)
 {
 	struct sockaddr *sa;
 	struct socket *new_so;
 	lck_mtx_t *mutex_held;
-	int 	dosocklock;
+	int dosocklock;
 	errno_t	error = 0;
-	
-	if (sock == NULL || new_sock == NULL) return EINVAL;
+
+	if (sock == NULL || new_sock == NULL)
+		return (EINVAL);
+
 	socket_lock(sock, 1);
 	if ((sock->so_options & SO_ACCEPTCONN) == 0) {
 		socket_unlock(sock, 1);
-	       	return EINVAL;
+		return (EINVAL);
 	}
 	if ((flags & ~(MSG_DONTWAIT)) != 0) {
 		socket_unlock(sock, 1);
-		return ENOTSUP;
+		return (ENOTSUP);
 	}
 	if (((flags & MSG_DONTWAIT) != 0 || (sock->so_state & SS_NBIO) != 0) &&
-		sock->so_comp.tqh_first == NULL) {
+	    sock->so_comp.tqh_first == NULL) {
 		socket_unlock(sock, 1);
-		return EWOULDBLOCK;
+		return (EWOULDBLOCK);
 	}
 
 	if (sock->so_proto->pr_getlock != NULL)  {
 		mutex_held = (*sock->so_proto->pr_getlock)(sock, 0);
 		dosocklock = 1;
-	}
-	else {
+	} else {
 		mutex_held = sock->so_proto->pr_domain->dom_mtx;
 		dosocklock = 0;
 	}
-	
+
 	while (TAILQ_EMPTY(&sock->so_comp) && sock->so_error == 0) {
 		if (sock->so_state & SS_CANTRCVMORE) {
 			sock->so_error = ECONNABORTED;
 			break;
 		}
-		error = msleep((caddr_t)&sock->so_timeo, mutex_held, PSOCK | PCATCH, "sock_accept", NULL);
-		if (error) {
+		error = msleep((caddr_t)&sock->so_timeo, mutex_held,
+		    PSOCK | PCATCH, "sock_accept", NULL);
+		if (error != 0) {
 			socket_unlock(sock, 1);
 			return (error);
 		}
 	}
-	if (sock->so_error) {
+	if (sock->so_error != 0) {
 		error = sock->so_error;
 		sock->so_error = 0;
 		socket_unlock(sock, 1);
 		return (error);
 	}
-	
+
 	new_so = TAILQ_FIRST(&sock->so_comp);
 	TAILQ_REMOVE(&sock->so_comp, new_so, so_list);
 	sock->so_qlen--;
@@ -145,31 +132,28 @@ sock_accept(
 
 	if (dosocklock)	{
 		lck_mtx_assert(new_so->so_proto->pr_getlock(new_so, 0),
-					   LCK_MTX_ASSERT_NOTOWNED);
+		    LCK_MTX_ASSERT_NOTOWNED);
 		socket_lock(new_so, 1);
 	}
-	
+
 	new_so->so_state &= ~SS_COMP;
 	new_so->so_head = NULL;
 	(void) soacceptlock(new_so, &sa, 0);
-	
+
 	socket_unlock(sock, 1);	/* release the head */
 
-	if (callback) {
-		new_so->so_upcall = (so_upcall) callback;
-		new_so->so_upcallarg = cookie;
-		new_so->so_rcv.sb_flags |= SB_UPCALL;
-#if CONFIG_SOWUPCALL
-		new_so->so_snd.sb_flags |= SB_UPCALL;
-#endif
+	/* see comments in sock_setupcall() */
+	if (callback != NULL) {
+		sock_setupcalls_common(new_so, callback, cookie, NULL, NULL);
 	}
-	
-	if (sa && from)
-	{
-		if (fromlen > sa->sa_len) fromlen = sa->sa_len;
+
+	if (sa != NULL && from != NULL) {
+		if (fromlen > sa->sa_len)
+			fromlen = sa->sa_len;
 		memcpy(from, sa, fromlen);
 	}
-	if (sa) FREE(sa, M_SONAME);
+	if (sa != NULL)
+		FREE(sa, M_SONAME);
 
 	/*
 	 * If the socket has been marked as inactive by sosetdefunct(),
@@ -180,61 +164,57 @@ sock_accept(
 		    SHUTDOWN_SOCKET_LEVEL_DISCONNECT_INTERNAL);
 	}
 	*new_sock = new_so;
-	if (dosocklock)	
+	if (dosocklock)
 		socket_unlock(new_so, 1);
-	return error;
+	return (error);
 }
 
 errno_t
-sock_bind(
-	socket_t				sock,
-	const struct sockaddr	*to)
+sock_bind(socket_t sock, const struct sockaddr *to)
 {
-	int	error = 0;
+	int error = 0;
 	struct sockaddr *sa = NULL;
 	struct sockaddr_storage ss;
 	boolean_t want_free = TRUE;
 
-	if (sock == NULL || to == NULL) 
-		return EINVAL;
-	
-	if (to->sa_len > sizeof(ss)) {
+	if (sock == NULL || to == NULL)
+		return (EINVAL);
+
+	if (to->sa_len > sizeof (ss)) {
 		MALLOC(sa, struct sockaddr *, to->sa_len, M_SONAME, M_WAITOK);
 		if (sa == NULL)
-			return ENOBUFS;
+			return (ENOBUFS);
 	} else {
 		sa = (struct sockaddr *)&ss;
 		want_free = FALSE;
 	}
 	memcpy(sa, to, to->sa_len);
 
-	error = sobind(sock, sa);
-	
+	error = sobindlock(sock, sa, 1);	/* will lock socket */
+
 	if (sa != NULL && want_free == TRUE)
-		FREE(sa, M_SONAME);	
+		FREE(sa, M_SONAME);
 
-	return error;
+	return (error);
 }
 
 errno_t
-sock_connect(
-	socket_t				sock,
-	const struct sockaddr	*to,
-	int						flags)
+sock_connect(socket_t sock, const struct sockaddr *to, int flags)
 {
-	int	error = 0;
+	int error = 0;
 	lck_mtx_t *mutex_held;
 	struct sockaddr *sa = NULL;
 	struct sockaddr_storage ss;
 	boolean_t want_free = TRUE;
-	
-	if (sock == NULL || to == NULL) return EINVAL;
-	
-	if (to->sa_len > sizeof(ss)) {
+
+	if (sock == NULL || to == NULL)
+		return (EINVAL);
+
+	if (to->sa_len > sizeof (ss)) {
 		MALLOC(sa, struct sockaddr *, to->sa_len, M_SONAME,
-			(flags & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK);
+		    (flags & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK);
 		if (sa == NULL)
-			return ENOBUFS;
+			return (ENOBUFS);
 	} else {
 		sa = (struct sockaddr *)&ss;
 		want_free = FALSE;
@@ -244,37 +224,37 @@ sock_connect(
 	socket_lock(sock, 1);
 
 	if ((sock->so_state & SS_ISCONNECTING) &&
-		((sock->so_state & SS_NBIO) != 0 ||
-		 (flags & MSG_DONTWAIT) != 0)) {
+	    ((sock->so_state & SS_NBIO) != 0 || (flags & MSG_DONTWAIT) != 0)) {
 		error = EALREADY;
 		goto out;
 	}
 	error = soconnectlock(sock, sa, 0);
 	if (!error) {
 		if ((sock->so_state & SS_ISCONNECTING) &&
-			((sock->so_state & SS_NBIO) != 0 || (flags & MSG_DONTWAIT) != 0)) {
+		    ((sock->so_state & SS_NBIO) != 0 ||
+		    (flags & MSG_DONTWAIT) != 0)) {
 			error = EINPROGRESS;
 			goto out;
 		}
-		
-		if (sock->so_proto->pr_getlock != NULL)  
+
+		if (sock->so_proto->pr_getlock != NULL)
 			mutex_held = (*sock->so_proto->pr_getlock)(sock, 0);
-		else 
-               		mutex_held = sock->so_proto->pr_domain->dom_mtx;
+		else
+			mutex_held = sock->so_proto->pr_domain->dom_mtx;
 
-		while ((sock->so_state & SS_ISCONNECTING) && sock->so_error == 0) {
-			error = msleep((caddr_t)&sock->so_timeo, mutex_held, PSOCK | PCATCH,
-				"sock_connect", NULL);
-			if (error)
+		while ((sock->so_state & SS_ISCONNECTING) &&
+		    sock->so_error == 0) {
+			error = msleep((caddr_t)&sock->so_timeo,
+			    mutex_held, PSOCK | PCATCH, "sock_connect", NULL);
+			if (error != 0)
 				break;
 		}
-		
+
 		if (error == 0) {
 			error = sock->so_error;
 			sock->so_error = 0;
 		}
-	}
-	else {
+	} else {
 		sock->so_state &= ~SS_ISCONNECTING;
 	}
 out:
@@ -282,28 +262,25 @@ out:
 
 	if (sa != NULL && want_free == TRUE)
 		FREE(sa, M_SONAME);
-		
-	return error;
+
+	return (error);
 }
 
 errno_t
-sock_connectwait(
-	socket_t				sock,
-	const struct timeval	*tv)
+sock_connectwait(socket_t sock, const struct timeval *tv)
 {
-	lck_mtx_t * mutex_held;
+	lck_mtx_t *mutex_held;
 	errno_t	retval = 0;
 	struct timespec ts;
-	
+
 	socket_lock(sock, 1);
-	
-	// Check if we're already connected or if we've already errored out
-	if ((sock->so_state & SS_ISCONNECTING) == 0 || sock->so_error) {
-		if (sock->so_error) {
+
+	/* Check if we're already connected or if we've already errored out */
+	if ((sock->so_state & SS_ISCONNECTING) == 0 || sock->so_error != 0) {
+		if (sock->so_error != 0) {
 			retval = sock->so_error;
 			sock->so_error = 0;
-		}
-		else {
+		} else {
 			if ((sock->so_state & SS_ISCONNECTED) != 0)
 				retval = 0;
 			else
@@ -311,63 +288,61 @@ sock_connectwait(
 		}
 		goto done;
 	}
-	
-	// copied translation from timeval to hertz from SO_RCVTIMEO handling
+
+	/* copied translation from timeval to hertz from SO_RCVTIMEO handling */
 	if (tv->tv_sec < 0 || tv->tv_sec > SHRT_MAX / hz ||
-		tv->tv_usec < 0 || tv->tv_usec >= 1000000) {
+	    tv->tv_usec < 0 || tv->tv_usec >= 1000000) {
 		retval = EDOM;
 		goto done;
 	}
-	
+
 	ts.tv_sec = tv->tv_sec;
-	ts.tv_nsec = (tv->tv_usec * NSEC_PER_USEC);
-	if ( (ts.tv_sec + (ts.tv_nsec/NSEC_PER_SEC))/100  >  SHRT_MAX)  {
+	ts.tv_nsec = (tv->tv_usec * (integer_t)NSEC_PER_USEC);
+	if ((ts.tv_sec + (ts.tv_nsec/(long)NSEC_PER_SEC))/100  >  SHRT_MAX)  {
 		retval = EDOM;
 		goto done;
 	}
-	
-	if (sock->so_proto->pr_getlock != NULL)  
+
+	if (sock->so_proto->pr_getlock != NULL)
 		mutex_held = (*sock->so_proto->pr_getlock)(sock, 0);
-	else 
-          	mutex_held = sock->so_proto->pr_domain->dom_mtx;
+	else
+		mutex_held = sock->so_proto->pr_domain->dom_mtx;
+
+	msleep((caddr_t)&sock->so_timeo, mutex_held,
+	    PSOCK, "sock_connectwait", &ts);
 
-	msleep((caddr_t)&sock->so_timeo, mutex_held, PSOCK, "sock_connectwait", &ts);
-	
-	// Check if we're still waiting to connect
+	/* Check if we're still waiting to connect */
 	if ((sock->so_state & SS_ISCONNECTING) && sock->so_error == 0) {
 		retval = EINPROGRESS;
 		goto done;
 	}
-	
-	if (sock->so_error) {
+
+	if (sock->so_error != 0) {
 		retval = sock->so_error;
 		sock->so_error = 0;
 	}
-	
+
 done:
 	socket_unlock(sock, 1);
-	return retval;
+	return (retval);
 }
 
 errno_t
-sock_nointerrupt(
-	socket_t	sock,
-	int			on)
+sock_nointerrupt(socket_t sock, int on)
 {
 	socket_lock(sock, 1);
 
 	if (on) {
-		sock->so_rcv.sb_flags |= SB_NOINTR; // This isn't safe
-		sock->so_snd.sb_flags |= SB_NOINTR; // This isn't safe
-	}
-	else {
-		sock->so_rcv.sb_flags &= ~SB_NOINTR; // This isn't safe
-		sock->so_snd.sb_flags &= ~SB_NOINTR; // This isn't safe
+		sock->so_rcv.sb_flags |= SB_NOINTR;	/* This isn't safe */
+		sock->so_snd.sb_flags |= SB_NOINTR;	/* This isn't safe */
+	} else {
+		sock->so_rcv.sb_flags &= ~SB_NOINTR;	/* This isn't safe */
+		sock->so_snd.sb_flags &= ~SB_NOINTR;	/* This isn't safe */
 	}
 
 	socket_unlock(sock, 1);
 
-	return 0;
+	return (0);
 }
 
 errno_t
@@ -448,7 +423,7 @@ sock_getaddr(socket_t sock, struct sockaddr **psa, int peer)
 	socket_lock(sock, 1);
 	error = sogetaddr_locked(sock, psa, peer);
 	socket_unlock(sock, 1);
-	
+
 	return (error);
 }
 
@@ -460,65 +435,60 @@ sock_freeaddr(struct sockaddr *sa)
 }
 
 errno_t
-sock_getsockopt(
-	socket_t	sock,
-	int			level,
-	int			optname,
-	void		*optval,
-	int			*optlen)
+sock_getsockopt(socket_t sock, int level, int optname, void *optval,
+    int	*optlen)
 {
-	int				error = 0;
+	int error = 0;
 	struct sockopt	sopt;
-	
-	if (sock == NULL || optval == NULL || optlen == NULL) return EINVAL;
+
+	if (sock == NULL || optval == NULL || optlen == NULL)
+		return (EINVAL);
+
 	sopt.sopt_dir = SOPT_GET;
 	sopt.sopt_level = level;
 	sopt.sopt_name = optname;
-	sopt.sopt_val = CAST_USER_ADDR_T(optval); 
+	sopt.sopt_val = CAST_USER_ADDR_T(optval);
 	sopt.sopt_valsize = *optlen;
 	sopt.sopt_p = kernproc;
-	error = sogetopt(sock, &sopt); /* will lock socket */
-	if (error == 0) *optlen = sopt.sopt_valsize;
-	return error;
+	error = sogetoptlock(sock, &sopt, 1);	/* will lock socket */
+	if (error == 0)
+		*optlen = sopt.sopt_valsize;
+	return (error);
 }
 
 errno_t
-sock_ioctl(
-	socket_t	sock,
-	unsigned long request,
-	void *argp)
+sock_ioctl(socket_t sock, unsigned long request, void *argp)
 {
-	return soioctl(sock, request, argp, kernproc); /* will lock socket */
+	return (soioctl(sock, request, argp, kernproc)); /* will lock socket */
 }
 
 errno_t
-sock_setsockopt(
-	socket_t	sock,
-	int			level,
-	int			optname,
-	const void	*optval,
-	int			optlen)
+sock_setsockopt(socket_t sock, int level, int optname, const void *optval,
+    int	optlen)
 {
 	struct sockopt	sopt;
-	
-	if (sock == NULL || optval == NULL) return EINVAL;
+
+	if (sock == NULL || optval == NULL)
+		return (EINVAL);
+
 	sopt.sopt_dir = SOPT_SET;
 	sopt.sopt_level = level;
 	sopt.sopt_name = optname;
 	sopt.sopt_val = CAST_USER_ADDR_T(optval);
 	sopt.sopt_valsize = optlen;
 	sopt.sopt_p = kernproc;
-	return sosetopt(sock, &sopt); /* will lock socket */
+	return (sosetoptlock(sock, &sopt, 1)); /* will lock socket */
 }
 
 /*
- * This follows the recommended mappings between DSCP code points and WMM access classes
+ * This follows the recommended mappings between DSCP code points
+ * and WMM access classes.
  */
-static u_int8_t so_tc_from_dscp(u_int8_t dscp);
-static u_int8_t
+static u_int32_t so_tc_from_dscp(u_int8_t dscp);
+static u_int32_t
 so_tc_from_dscp(u_int8_t dscp)
 {
-	u_int8_t tc;
+	u_int32_t tc;
 
 	if (dscp >= 0x30 && dscp <= 0x3f)
 		tc = SO_TC_VO;
@@ -529,24 +499,23 @@ so_tc_from_dscp(u_int8_t dscp)
 	else
 		tc = SO_TC_BE;
 
-	return tc;
+	return (tc);
 }
 
 errno_t
-sock_settclassopt(
-	socket_t	sock,
-	const void	*optval,
-	size_t		optlen) {
-
+sock_settclassopt(socket_t sock, const void *optval, size_t optlen)
+{
 	errno_t error = 0;
 	struct sockopt sopt;
 	int sotc;
 
-	if (sock == NULL || optval == NULL || optlen != sizeof(int)) return EINVAL;
+	if (sock == NULL || optval == NULL || optlen != sizeof (int))
+		return (EINVAL);
 
 	socket_lock(sock, 1);
 	if (!(sock->so_state & SS_ISCONNECTED)) {
-		/* If the socket is not connected then we don't know 
+		/*
+		 * If the socket is not connected then we don't know
 		 * if the destination is on LAN  or not. Skip
 		 * setting traffic class in this case
 		 */
@@ -554,7 +523,8 @@ sock_settclassopt(
 		goto out;
 	}
 
-	if (sock->so_proto == NULL || sock->so_proto->pr_domain == NULL || sock->so_pcb == NULL) {
+	if (sock->so_proto == NULL || sock->so_proto->pr_domain == NULL ||
+	    sock->so_pcb == NULL) {
 		error = EINVAL;
 		goto out;
 	}
@@ -567,39 +537,38 @@ sock_settclassopt(
 
 	sopt.sopt_dir = SOPT_SET;
 	sopt.sopt_val = CAST_USER_ADDR_T(&sotc);
-	sopt.sopt_valsize = sizeof(sotc);
+	sopt.sopt_valsize = sizeof (sotc);
 	sopt.sopt_p = kernproc;
 	sopt.sopt_level = SOL_SOCKET;
 	sopt.sopt_name = SO_TRAFFIC_CLASS;
 
-	socket_unlock(sock, 0);
-	error = sosetopt(sock, &sopt);
-	socket_lock(sock, 0);
+	error = sosetoptlock(sock, &sopt, 0);	/* already locked */
 
 	if (error != 0) {
-		printf("sock_settclassopt: sosetopt SO_TRAFFIC_CLASS failed %d\n", error);
+		printf("%s: sosetopt SO_TRAFFIC_CLASS failed %d\n",
+		    __func__, error);
 		goto out;
 	}
 
-	/* Check if the destination address is LAN or link local address.
+	/*
+	 * Check if the destination address is LAN or link local address.
 	 * We do not want to set traffic class bits if the destination
-	 * is not local 
-	 */ 
-	if (!so_isdstlocal(sock)) {
+	 * is not local.
+	 */
+	if (!so_isdstlocal(sock))
 		goto out;
-	}
 
 	sopt.sopt_dir = SOPT_SET;
 	sopt.sopt_val = CAST_USER_ADDR_T(optval);
 	sopt.sopt_valsize = optlen;
 	sopt.sopt_p = kernproc;
 
-	switch (sock->so_proto->pr_domain->dom_family) {
-	case AF_INET:
+	switch (SOCK_DOM(sock)) {
+	case PF_INET:
 		sopt.sopt_level = IPPROTO_IP;
 		sopt.sopt_name = IP_TOS;
 		break;
-	case AF_INET6:
+	case PF_INET6:
 		sopt.sopt_level = IPPROTO_IPV6;
 		sopt.sopt_name = IPV6_TCLASS;
 		break;
@@ -607,140 +576,135 @@ sock_settclassopt(
 		error = EINVAL;
 		goto out;
 	}
-	
+
+	error = sosetoptlock(sock, &sopt, 0);	/* already locked */
 	socket_unlock(sock, 1);
-	return sosetopt(sock, &sopt);
+	return (error);
 out:
 	socket_unlock(sock, 1);
-	return error;
+	return (error);
 }
 
 errno_t
-sock_gettclassopt(
-	socket_t	sock,
-	void		*optval,
-	size_t		*optlen) {
+sock_gettclassopt(socket_t sock, void *optval, size_t *optlen)
+{
+	errno_t error = 0;
+	struct sockopt sopt;
 
-	errno_t		error = 0;
-	struct sockopt	sopt;
-	
-	if (sock == NULL || optval == NULL || optlen == NULL) return EINVAL;
+	if (sock == NULL || optval == NULL || optlen == NULL)
+		return (EINVAL);
 
 	sopt.sopt_dir = SOPT_GET;
-	sopt.sopt_val = CAST_USER_ADDR_T(optval); 
+	sopt.sopt_val = CAST_USER_ADDR_T(optval);
 	sopt.sopt_valsize = *optlen;
 	sopt.sopt_p = kernproc;
 
 	socket_lock(sock, 1);
 	if (sock->so_proto == NULL || sock->so_proto->pr_domain == NULL) {
 		socket_unlock(sock, 1);
-		return EINVAL;
+		return (EINVAL);
 	}
 
-	switch (sock->so_proto->pr_domain->dom_family) {
-	case AF_INET:
+	switch (SOCK_DOM(sock)) {
+	case PF_INET:
 		sopt.sopt_level = IPPROTO_IP;
 		sopt.sopt_name = IP_TOS;
 		break;
-	case AF_INET6:
+	case PF_INET6:
 		sopt.sopt_level = IPPROTO_IPV6;
 		sopt.sopt_name = IPV6_TCLASS;
 		break;
 	default:
 		socket_unlock(sock, 1);
-		return EINVAL;
+		return (EINVAL);
 
 	}
+	error = sogetoptlock(sock, &sopt, 0);	/* already locked */
 	socket_unlock(sock, 1);
-	error = sogetopt(sock, &sopt); /* will lock socket */
-	if (error == 0) *optlen = sopt.sopt_valsize;
-	return error;
+	if (error == 0)
+		*optlen = sopt.sopt_valsize;
+	return (error);
 }
 
 errno_t
-sock_listen(
-	socket_t	sock,
-	int			backlog)
+sock_listen(socket_t sock, int backlog)
 {
-	if (sock == NULL) return EINVAL;
-	return solisten(sock, backlog); /* will lock socket */
+	if (sock == NULL)
+		return (EINVAL);
+
+	return (solisten(sock, backlog)); /* will lock socket */
 }
 
-static errno_t
-sock_receive_internal(
-	socket_t		sock,
-	struct msghdr		*msg,
-	mbuf_t			*data,
-	int			flags,
-	size_t			*recvdlen)
+errno_t
+sock_receive_internal(socket_t sock, struct msghdr *msg, mbuf_t *data,
+    int flags, size_t *recvdlen)
 {
-	uio_t 		auio;
-	struct mbuf	*control = NULL;
-	int			error = 0;
-	int			length = 0;
-	struct sockaddr	*fromsa;
-	char		uio_buf[ UIO_SIZEOF((msg != NULL) ? msg->msg_iovlen : 0) ];
-	
-	if (sock == NULL) return EINVAL;
-	
-	auio = uio_createwithbuffer(((msg != NULL) ? msg->msg_iovlen : 0), 
-								  0, UIO_SYSSPACE, UIO_READ, 
-								  &uio_buf[0], sizeof(uio_buf));
-	if (msg && data == NULL) {
+	uio_t auio;
+	struct mbuf *control = NULL;
+	int error = 0;
+	int length = 0;
+	struct sockaddr	*fromsa = NULL;
+	char uio_buf[ UIO_SIZEOF((msg != NULL) ? msg->msg_iovlen : 0) ];
+
+	if (sock == NULL)
+		return (EINVAL);
+
+	auio = uio_createwithbuffer(((msg != NULL) ? msg->msg_iovlen : 0),
+	    0, UIO_SYSSPACE, UIO_READ, &uio_buf[0], sizeof (uio_buf));
+	if (msg != NULL && data == NULL) {
 		int i;
 		struct iovec *tempp = msg->msg_iov;
-		
+
 		for (i = 0; i < msg->msg_iovlen; i++) {
-			uio_addiov(auio, CAST_USER_ADDR_T((tempp + i)->iov_base), (tempp + i)->iov_len);
+			uio_addiov(auio,
+			    CAST_USER_ADDR_T((tempp + i)->iov_base),
+			    (tempp + i)->iov_len);
 		}
-		if (uio_resid(auio) < 0) return EINVAL;
-	}
-	else {
+		if (uio_resid(auio) < 0)
+			return (EINVAL);
+	} else if (recvdlen != NULL) {
 		uio_setresid(auio, (uio_resid(auio) + *recvdlen));
 	}
 	length = uio_resid(auio);
-	
-	if (recvdlen)
+
+	if (recvdlen != NULL)
 		*recvdlen = 0;
 
-	/* let pru_soreceive handle the socket locking */	
+	/* let pru_soreceive handle the socket locking */
 	error = sock->so_proto->pr_usrreqs->pru_soreceive(sock, &fromsa, auio,
 	    data, (msg && msg->msg_control) ? &control : NULL, &flags);
-	if (error) goto cleanup;
-	
-	if (recvdlen)
+	if (error != 0)
+		goto cleanup;
+
+	if (recvdlen != NULL)
 		*recvdlen = length - uio_resid(auio);
-	if (msg) {
+	if (msg != NULL) {
 		msg->msg_flags = flags;
-		
-		if (msg->msg_name)
-		{
+
+		if (msg->msg_name != NULL) {
 			int salen;
 			salen = msg->msg_namelen;
-			if (msg->msg_namelen > 0 && fromsa != 0)
-			{
+			if (msg->msg_namelen > 0 && fromsa != NULL) {
 				salen = MIN(salen, fromsa->sa_len);
 				memcpy(msg->msg_name, fromsa,
-					msg->msg_namelen > fromsa->sa_len ? fromsa->sa_len :  msg->msg_namelen);
+				    msg->msg_namelen > fromsa->sa_len ?
+				    fromsa->sa_len : msg->msg_namelen);
 			}
 		}
-		
-		if (msg->msg_control)
-		{
-			struct mbuf*	m = control;
-			u_char*			ctlbuf = msg->msg_control;
-			int				clen = msg->msg_controllen;
+
+		if (msg->msg_control != NULL) {
+			struct mbuf *m = control;
+			u_char *ctlbuf = msg->msg_control;
+			int clen = msg->msg_controllen;
+
 			msg->msg_controllen = 0;
-			
-			while (m && clen > 0)
-			{
+
+			while (m != NULL && clen > 0) {
 				unsigned int tocopy;
-				if (clen >= m->m_len)
-				{
+
+				if (clen >= m->m_len) {
 					tocopy = m->m_len;
-				}
-				else
-				{
+				} else {
 					msg->msg_flags |= MSG_CTRUNC;
 					tocopy = clen;
 				}
@@ -749,107 +713,109 @@ sock_receive_internal(
 				clen -= tocopy;
 				m = m->m_next;
 			}
-			msg->msg_controllen = (uintptr_t)ctlbuf - (uintptr_t)msg->msg_control;
+			msg->msg_controllen =
+			    (uintptr_t)ctlbuf - (uintptr_t)msg->msg_control;
 		}
 	}
 
 cleanup:
-	if (control) m_freem(control);
-	if (fromsa) FREE(fromsa, M_SONAME);
-	return error;
+	if (control != NULL)
+		m_freem(control);
+	if (fromsa != NULL)
+		FREE(fromsa, M_SONAME);
+	return (error);
 }
 
 errno_t
-sock_receive(
-	socket_t		sock,
-	struct msghdr	*msg,
-	int				flags,
-	size_t			*recvdlen)
+sock_receive(socket_t sock, struct msghdr *msg, int flags, size_t *recvdlen)
 {
-	if ((msg == NULL) ||
-		(msg->msg_iovlen < 1) ||
-		(msg->msg_iov[0].iov_len == 0) ||
-		(msg->msg_iov[0].iov_base == NULL))
-		return EINVAL;
-	return sock_receive_internal(sock, msg, NULL, flags, recvdlen);
+	if ((msg == NULL) || (msg->msg_iovlen < 1) ||
+	    (msg->msg_iov[0].iov_len == 0) ||
+	    (msg->msg_iov[0].iov_base == NULL))
+		return (EINVAL);
+
+	return (sock_receive_internal(sock, msg, NULL, flags, recvdlen));
 }
 
 errno_t
-sock_receivembuf(
-	socket_t		sock,
-	struct msghdr	*msg,
-	mbuf_t			*data,
-	int				flags,
-	size_t			*recvlen)
+sock_receivembuf(socket_t sock, struct msghdr *msg, mbuf_t *data, int flags,
+    size_t *recvlen)
 {
-	if (data == NULL || recvlen == 0 || *recvlen <= 0 || (msg &&
-		(msg->msg_iov != NULL || msg->msg_iovlen != 0)))
-		return EINVAL;
-	return sock_receive_internal(sock, msg, data, flags, recvlen);
+	if (data == NULL || recvlen == 0 || *recvlen <= 0 || (msg != NULL &&
+	    (msg->msg_iov != NULL || msg->msg_iovlen != 0)))
+		return (EINVAL);
+
+	return (sock_receive_internal(sock, msg, data, flags, recvlen));
 }
 
 errno_t
-sock_send_internal(
-	socket_t			sock,
-	const struct msghdr	*msg,
-	mbuf_t				data,
-	int					flags,
-	size_t				*sentlen)
+sock_send_internal(socket_t sock, const struct msghdr *msg, mbuf_t data,
+    int flags, size_t *sentlen)
 {
-	uio_t			auio = NULL;
-	struct mbuf		*control = NULL;
-	int				error = 0;
-	int				datalen = 0;
-	char			uio_buf[ UIO_SIZEOF((msg != NULL ? msg->msg_iovlen : 1)) ];
-	
+	uio_t auio = NULL;
+	struct mbuf *control = NULL;
+	int error = 0;
+	int datalen = 0;
+	char uio_buf[ UIO_SIZEOF((msg != NULL ? msg->msg_iovlen : 1)) ];
+
 	if (sock == NULL) {
 		error = EINVAL;
 		goto errorout;
 	}
-	
-	if (data == 0 && msg != NULL) {
+
+	if (data == NULL && msg != NULL) {
 		struct iovec *tempp = msg->msg_iov;
 
-		auio = uio_createwithbuffer(msg->msg_iovlen, 0, UIO_SYSSPACE, UIO_WRITE, 
-								  &uio_buf[0], sizeof(uio_buf));
-		if (tempp != NULL)
-		{
+		auio = uio_createwithbuffer(msg->msg_iovlen, 0,
+		    UIO_SYSSPACE, UIO_WRITE, &uio_buf[0], sizeof (uio_buf));
+		if (tempp != NULL) {
 			int i;
-			
+
 			for (i = 0; i < msg->msg_iovlen; i++) {
-				uio_addiov(auio, CAST_USER_ADDR_T((tempp + i)->iov_base), (tempp + i)->iov_len);
+				uio_addiov(auio,
+				    CAST_USER_ADDR_T((tempp + i)->iov_base),
+				    (tempp + i)->iov_len);
 			}
-			
+
 			if (uio_resid(auio) < 0) {
 				error = EINVAL;
 				goto errorout;
 			}
 		}
 	}
-	
-	if (sentlen)
+
+	if (sentlen != NULL)
 		*sentlen = 0;
-	
-	if (auio)
+
+	if (auio != NULL)
 		datalen = uio_resid(auio);
 	else
 		datalen = data->m_pkthdr.len;
-	
-	if (msg && msg->msg_control)
-	{
-		if ((size_t)msg->msg_controllen < sizeof(struct cmsghdr)) return EINVAL;
-		if ((size_t)msg->msg_controllen > MLEN) return EINVAL;
+
+	if (msg != NULL && msg->msg_control) {
+		if ((size_t)msg->msg_controllen < sizeof (struct cmsghdr)) {
+			error = EINVAL;
+			goto errorout;
+		}
+
+		if ((size_t)msg->msg_controllen > MLEN) {
+			error = EINVAL;
+			goto errorout;
+		}
+
 		control = m_get(M_NOWAIT, MT_CONTROL);
 		if (control == NULL) {
 			error = ENOMEM;
 			goto errorout;
 		}
-		memcpy(mtod(control, caddr_t), msg->msg_control, msg->msg_controllen);
+		memcpy(mtod(control, caddr_t), msg->msg_control,
+		    msg->msg_controllen);
 		control->m_len = msg->msg_controllen;
 	}
-	
+
 	error = sock->so_proto->pr_usrreqs->pru_sosend(sock, msg != NULL ?
-	    (struct sockaddr*)msg->msg_name : NULL, auio, data, control, flags);
+	    (struct sockaddr *)msg->msg_name : NULL, auio, data,
+	    control, flags);
 
 	/*
 	 * Residual data is possible in the case of IO vectors but not
@@ -868,8 +834,8 @@ sock_send_internal(
 		else
 			*sentlen = datalen;
 	}
-	
-	return error;
+
+	return (error);
 
 /*
  * In cases where we detect an error before returning, we need to
@@ -883,87 +849,84 @@ errorout:
 		m_freem(data);
 	if (sentlen)
 		*sentlen = 0;
-	return error;
+	return (error);
 }
 
 errno_t
-sock_send(
-	socket_t			sock,
-	const struct msghdr	*msg,
-	int					flags,
-	size_t				*sentlen)
+sock_send(socket_t sock, const struct msghdr *msg, int flags, size_t *sentlen)
 {
 	if (msg == NULL || msg->msg_iov == NULL || msg->msg_iovlen < 1)
-		return EINVAL;
-	return sock_send_internal(sock, msg, NULL, flags, sentlen);
+		return (EINVAL);
+
+	return (sock_send_internal(sock, msg, NULL, flags, sentlen));
 }
 
 errno_t
-sock_sendmbuf(
-	socket_t			sock,
-	const struct msghdr	*msg,
-	mbuf_t				data,
-	int					flags,
-	size_t				*sentlen)
+sock_sendmbuf(socket_t sock, const struct msghdr *msg, mbuf_t data,
+    int	flags, size_t *sentlen)
 {
-	if (data == NULL || (msg &&
-		(msg->msg_iov != NULL || msg->msg_iovlen != 0))) {
-		if (data)
+	if (data == NULL || (msg != NULL && (msg->msg_iov != NULL ||
+	    msg->msg_iovlen != 0))) {
+		if (data != NULL)
 			m_freem(data);
-		return EINVAL;
+		return (EINVAL);
 	}
-	return sock_send_internal(sock, msg, data, flags, sentlen);
+	return (sock_send_internal(sock, msg, data, flags, sentlen));
 }
 
 errno_t
-sock_shutdown(
-	socket_t	sock,
-	int			how)
+sock_shutdown(socket_t sock, int how)
 {
-	if (sock == NULL) return EINVAL;
-	return soshutdown(sock, how);
+	if (sock == NULL)
+		return (EINVAL);
+
+	return (soshutdown(sock, how));
 }
 
 
 errno_t
-sock_socket(
-	int				domain,
-	int				type,
-	int				protocol,
-	sock_upcall		callback,
-	void*			context,
-	socket_t		*new_so)
+sock_socket(int	domain, int type, int protocol, sock_upcall callback,
+    void *context, socket_t *new_so)
 {
-	int	error = 0;
-	if (new_so == NULL) return EINVAL;
+	int error = 0;
+
+	if (new_so == NULL)
+		return (EINVAL);
+
 	/* socreate will create an initial so_count */
 	error = socreate(domain, new_so, type, protocol);
-	if (error == 0 && callback)
-	{
-		(*new_so)->so_rcv.sb_flags |= SB_UPCALL;
-#if CONFIG_SOWUPCALL
-		(*new_so)->so_snd.sb_flags |= SB_UPCALL;
-#endif
-		(*new_so)->so_upcall = (so_upcall)callback;
-		(*new_so)->so_upcallarg = context;
+	if (error == 0) {
+		/* see comments in sock_setupcall() */
+		if (callback != NULL) {
+			sock_setupcalls_common(*new_so, callback, context,
+			    NULL, NULL);
+		}
+		/* 
+		 * last_pid and last_upid should be zero for sockets
+		 * created using sock_socket
+		 */
+		(*new_so)->last_pid = 0;
+		(*new_so)->last_upid = 0;
 	}
-	return error;
+	return (error);
 }
 
 void
-sock_close(
-	socket_t	sock)
+sock_close(socket_t sock)
 {
-	if (sock == NULL) return;
+	if (sock == NULL)
+		return;
+
 	soclose(sock);
 }
 
-/* Do we want this to be APPLE_PRIVATE API?: YES (LD 12/23/04)*/
+/* Do we want this to be APPLE_PRIVATE API?: YES (LD 12/23/04) */
 void
-sock_retain(
-	socket_t	sock)
+sock_retain(socket_t sock)
 {
-	if (sock == NULL) return;
+	if (sock == NULL)
+		return;
+
 	socket_lock(sock, 1);
 	sock->so_retaincnt++;
 	sock->so_usecount++;	/* add extra reference for holding the socket */
@@ -976,15 +939,17 @@ sock_release(socket_t sock)
 {
 	if (sock == NULL)
 		return;
-	socket_lock(sock, 1);
 
-	if (sock->so_flags & SOF_UPCALLINUSE)
+	socket_lock(sock, 1);
+	if (sock->so_upcallusecount > 0)
 		soclose_wait_locked(sock);
 
 	sock->so_retaincnt--;
-	if (sock->so_retaincnt < 0)
-		panic("sock_release: negative retain count for sock=%p "
-		    "cnt=%x\n", sock, sock->so_retaincnt);
+	if (sock->so_retaincnt < 0) {
+		panic("%s: negative retain count (%d) for sock=%p\n",
+		    __func__, sock->so_retaincnt, sock);
+		/* NOTREACHED */
+	}
 	if ((sock->so_retaincnt == 0) && (sock->so_usecount == 2)) {
 		/* close socket only if the FD is not holding it */
 		soclose_locked(sock);
@@ -996,62 +961,54 @@ sock_release(socket_t sock)
 }
 
 errno_t
-sock_setpriv(
-	socket_t	sock,
-	int			on)
+sock_setpriv(socket_t sock, int on)
 {
-	if (sock == NULL) return EINVAL;
+	if (sock == NULL)
+		return (EINVAL);
+
 	socket_lock(sock, 1);
 	if (on)
-	{
 		sock->so_state |= SS_PRIV;
-	}
 	else
-	{
 		sock->so_state &= ~SS_PRIV;
-	}
 	socket_unlock(sock, 1);
-	return 0;
+	return (0);
 }
 
 int
-sock_isconnected(
-	socket_t sock)
+sock_isconnected(socket_t sock)
 {
 	int retval;
+
 	socket_lock(sock, 1);
-	retval = (sock->so_state & SS_ISCONNECTED) != 0;
+	retval = ((sock->so_state & SS_ISCONNECTED) ? 1 : 0);
 	socket_unlock(sock, 1);
 	return (retval);
 }
 
 int
-sock_isnonblocking(
-	socket_t sock)
+sock_isnonblocking(socket_t sock)
 {
 	int retval;
+
 	socket_lock(sock, 1);
-	retval = (sock->so_state & SS_NBIO) != 0;
+	retval = ((sock->so_state & SS_NBIO) ? 1 : 0);
 	socket_unlock(sock, 1);
 	return (retval);
 }
 
 errno_t
-sock_gettype(
-	socket_t	sock,
-	int			*outDomain,
-	int			*outType,
-	int			*outProtocol)
+sock_gettype(socket_t sock, int *outDomain, int *outType, int *outProtocol)
 {
 	socket_lock(sock, 1);
-	if (outDomain)
-		*outDomain = sock->so_proto->pr_domain->dom_family;
-	if (outType)
+	if (outDomain != NULL)
+		*outDomain = SOCK_DOM(sock);
+	if (outType != NULL)
 		*outType = sock->so_type;
-	if (outProtocol)
-		*outProtocol = sock->so_proto->pr_protocol;
+	if (outProtocol != NULL)
+		*outProtocol = SOCK_PROTO(sock);
 	socket_unlock(sock, 1);
-	return 0;
+	return (0);
 }
 
 /*
@@ -1071,12 +1028,9 @@ sock_getlistener(socket_t sock)
 static inline void
 sock_set_tcp_stream_priority(socket_t sock)
 {
-	if ((sock->so_proto->pr_domain->dom_family == AF_INET || 
-		sock->so_proto->pr_domain->dom_family == AF_INET6) &&
-		sock->so_proto->pr_type == SOCK_STREAM) {
-
+	if ((SOCK_DOM(sock) == PF_INET || SOCK_DOM(sock) == PF_INET6) &&
+	    SOCK_TYPE(sock) == SOCK_STREAM) {
 		set_tcp_stream_priority(sock);
-
 	}
 }
 
@@ -1149,33 +1103,100 @@ socket_defunct(struct proc *p, socket_t so, int level)
 	return (retval);
 }
 
+static void
+sock_setupcalls_common(socket_t sock, sock_upcall rcallback, void *rcontext,
+    sock_upcall wcallback, void *wcontext)
+{
+	if (rcallback != NULL) {
+		sock->so_rcv.sb_flags |= SB_UPCALL;
+		sock->so_rcv.sb_upcall = rcallback;
+		sock->so_rcv.sb_upcallarg = rcontext;
+	} else {
+		sock->so_rcv.sb_flags &= ~SB_UPCALL;
+		sock->so_rcv.sb_upcall = NULL;
+		sock->so_rcv.sb_upcallarg = NULL;
+	}
+
+	if (wcallback != NULL) {
+		sock->so_snd.sb_flags |= SB_UPCALL;
+		sock->so_snd.sb_upcall = wcallback;
+		sock->so_snd.sb_upcallarg = wcontext;
+	} else {
+		sock->so_snd.sb_flags &= ~SB_UPCALL;
+		sock->so_snd.sb_upcall = NULL;
+		sock->so_snd.sb_upcallarg = NULL;
+	}
+}
+
 errno_t
-sock_setupcall(socket_t sock, sock_upcall callback, void* context)
+sock_setupcall(socket_t sock, sock_upcall callback, void *context)
 {
 	if (sock == NULL)
-		return EINVAL;
+		return (EINVAL);
 
 	/*
 	 * Note that we don't wait for any in progress upcall to complete.
+	 * On embedded, sock_setupcall() causes both read and write
+	 * callbacks to be set; on desktop, only read callback is set
+	 * to maintain legacy KPI behavior.
+	 *
+	 * The newer sock_setupcalls() KPI should be used instead to set
+	 * the read and write callbacks and their respective parameters.
 	 */
 	socket_lock(sock, 1);
+	sock_setupcalls_common(sock, callback, context, NULL, NULL);
+	socket_unlock(sock, 1);
 
-	sock->so_upcall = (so_upcall) callback;
-	sock->so_upcallarg = context;
-	if (callback) {
-		sock->so_rcv.sb_flags |= SB_UPCALL;
-#if CONFIG_SOWUPCALL
-		sock->so_snd.sb_flags |= SB_UPCALL;
-#endif /* CONFIG_SOWUPCALL */
+	return (0);
+}
+
+errno_t
+sock_setupcalls(socket_t sock, sock_upcall rcallback, void *rcontext,
+    sock_upcall wcallback, void *wcontext)
+{
+	if (sock == NULL)
+		return (EINVAL);
+
+	/*
+	 * Note that we don't wait for any in progress upcall to complete.
+	 */
+	socket_lock(sock, 1);
+	sock_setupcalls_common(sock, rcallback, rcontext, wcallback, wcontext);
+	socket_unlock(sock, 1);
+
+	return (0);
+}
+
+errno_t
+sock_catchevents(socket_t sock, sock_evupcall ecallback, void *econtext,
+    u_int32_t emask)
+{
+	if (sock == NULL)
+		return (EINVAL);
+
+	/*
+	 * Note that we don't wait for any in progress upcall to complete.
+	 */
+	socket_lock(sock, 1);
+	if (ecallback != NULL) {
+		sock->so_event = ecallback;
+		sock->so_eventarg = econtext;
+		sock->so_eventmask = emask;
 	} else {
-		sock->so_rcv.sb_flags &= ~SB_UPCALL;
-#if CONFIG_SOWUPCALL
-		sock->so_snd.sb_flags &= ~SB_UPCALL;
-#endif /* CONFIG_SOWUPCALL */
+		sock->so_event = sonullevent;
+		sock->so_eventarg = NULL;
+		sock->so_eventmask = 0;
 	}
-	
 	socket_unlock(sock, 1);
 
-	return 0;
+	return (0);
 }
 
+/*
+ * Returns true whether or not a socket belongs to the kernel.
+ */
+int
+sock_iskernel(socket_t so)
+{
+	return (so && so->last_pid == 0);
+}