Decompose the current single inpcbinfo lock into two locks:

- The existing ipi_lock continues to protect the global inpcb list and
  inpcb counter.  This lock is now relegated to a small number of
  allocation and free operations, and occasional operations that walk
  all connections (including, awkwardly, certain UDP multicast receive
  operations -- something to revisit).

- A new ipi_hash_lock protects the two inpcbinfo hash tables for
  looking up connections and bound sockets, manipulated using new
  INP_HASH_*() macros.  This lock, combined with inpcb locks, protects
  the 4-tuple address space.

Unlike the current ipi_lock, ipi_hash_lock follows the individual inpcb
connection locks, so may be acquired while manipulating a connection on
which a lock is already held, avoiding the need to acquire the inpcbinfo
lock preemptively when a binding change might later be required.  As a
result, however, lookup operations necessarily go through a reference
acquire while holding the lookup lock, later acquiring an inpcb lock --
if required.

A new function in_pcblookup() looks up connections, and accepts flags
indicating how to return the inpcb.  Due to lock order changes, callers
no longer need acquire locks before performing a lookup: the lookup
routine will acquire the ipi_hash_lock as needed.  In the future, it will
also be able to use alternative lookup and locking strategies
transparently to callers, such as pcbgroup lookup.  New lookup flags are,
supplementing the existing INPLOOKUP_WILDCARD flag:

  INPLOOKUP_RLOCKPCB - Acquire a read lock on the returned inpcb
  INPLOOKUP_WLOCKPCB - Acquire a write lock on the returned inpcb

Callers must pass exactly one of these flags (for the time being).

Some notes:

- All protocols are updated to work within the new regime; especially,
  TCP, UDPv4, and UDPv6.  pcbinfo ipi_lock acquisitions are largely
  eliminated, and global hash lock hold times are dramatically reduced
  compared to previous locking.
- The TCP syncache still relies on the pcbinfo lock, something that we
  may want to revisit.
- Support for reverting to the FreeBSD 7.x locking strategy in TCP input
  is no longer available -- hash lookup locks are now held only very
  briefly during inpcb lookup, rather than for potentially extended
  periods.  However, the pcbinfo ipi_lock will still be acquired if a
  connection state might change such that a connection is added or
  removed.
- Raw IP sockets continue to use the pcbinfo ipi_lock for protection,
  due to maintaining their own hash tables.
- The interface in6_pcblookup_hash_locked() is maintained, which allows
  callers to acquire hash locks and perform one or more lookups atomically
  with 4-tuple allocation: this is required only for TCPv6, as there is no
  in6_pcbconnect_setup(), which there should be.
- UDPv6 locking remains significantly more conservative than UDPv4
  locking, which relates to source address selection.  This needs
  attention, as it likely significantly reduces parallelism in this code
  for multithreaded socket use (such as in BIND).
- In the UDPv4 and UDPv6 multicast cases, we need to revisit locking
  somewhat, as they relied on ipi_lock to stablise 4-tuple matches, which
  is no longer sufficient.  A second check once the inpcb lock is held
  should do the trick, keeping the general case from requiring the inpcb
  lock for every inpcb visited.
- This work reminds us that we need to revisit locking of the v4/v6 flags,
  which may be accessed lock-free both before and after this change.
- Right now, a single lock name is used for the pcbhash lock -- this is
  undesirable, and probably another argument is required to take care of
  this (or a char array name field in the pcbinfo?).

This is not an MFC candidate for 8.x due to its impact on lookup and
locking semantics.  It's possible some of these issues could be worked
around with compatibility wrappers, if necessary.

Reviewed by:    bz
Sponsored by:   Juniper Networks, Inc.

12 files changed, 424 insertions, 369 deletions

diff --git a/sys/netinet/in_pcb.c b/sys/netinet/in_pcb.c
index 85e31dc..8ae4d6f 100644
--- a/sys/netinet/in_pcb.c
+++ b/sys/netinet/in_pcb.c
@@ -127,6 +127,10 @@ static VNET_DEFINE(int, ipport_tcplastcount);
 
 #define	V_ipport_tcplastcount		VNET(ipport_tcplastcount)
 
+static struct inpcb	*in_pcblookup_hash_locked(struct inpcbinfo *pcbinfo,
+			    struct in_addr faddr, u_int fport_arg,
+			    struct in_addr laddr, u_int lport_arg,
+			    int lookupflags, struct ifnet *ifp);
 static void	in_pcbremlists(struct inpcb *inp);
 
 #ifdef INET
@@ -212,11 +216,13 @@ in_pcbinfo_init(struct inpcbinfo *pcbinfo, const char *name,
 {
 
 	INP_INFO_LOCK_INIT(pcbinfo, name);
+	INP_HASH_LOCK_INIT(pcbinfo, "pcbinfohash");	/* XXXRW: argument? */
 #ifdef VIMAGE
 	pcbinfo->ipi_vnet = curvnet;
 #endif
 	pcbinfo->ipi_listhead = listhead;
 	LIST_INIT(pcbinfo->ipi_listhead);
+	pcbinfo->ipi_count = 0;
 	pcbinfo->ipi_hashbase = hashinit(hash_nelements, M_PCB,
 	    &pcbinfo->ipi_hashmask);
 	pcbinfo->ipi_porthashbase = hashinit(porthash_nelements, M_PCB,
@@ -234,10 +240,14 @@ void
 in_pcbinfo_destroy(struct inpcbinfo *pcbinfo)
 {
 
+	KASSERT(pcbinfo->ipi_count == 0,
+	    ("%s: ipi_count = %u", __func__, pcbinfo->ipi_count));
+
 	hashdestroy(pcbinfo->ipi_hashbase, M_PCB, pcbinfo->ipi_hashmask);
 	hashdestroy(pcbinfo->ipi_porthashbase, M_PCB,
 	    pcbinfo->ipi_porthashmask);
 	uma_zdestroy(pcbinfo->ipi_zone);
+	INP_HASH_LOCK_DESTROY(pcbinfo);
 	INP_INFO_LOCK_DESTROY(pcbinfo);
 }
 
@@ -309,8 +319,8 @@ in_pcbbind(struct inpcb *inp, struct sockaddr *nam, struct ucred *cred)
 {
 	int anonport, error;
 
-	INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo);
 	INP_WLOCK_ASSERT(inp);
+	INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo);
 
 	if (inp->inp_lport != 0 || inp->inp_laddr.s_addr != INADDR_ANY)
 		return (EINVAL);
@@ -351,8 +361,8 @@ in_pcb_lport(struct inpcb *inp, struct in_addr *laddrp, u_short *lportp,
 	 * Because no actual state changes occur here, a global write lock on
 	 * the pcbinfo isn't required.
 	 */
-	INP_INFO_LOCK_ASSERT(pcbinfo);
 	INP_LOCK_ASSERT(inp);
+	INP_HASH_LOCK_ASSERT(pcbinfo);
 
 	if (inp->inp_flags & INP_HIGHPORT) {
 		first = V_ipport_hifirstauto;	/* sysctl */
@@ -473,11 +483,10 @@ in_pcbbind_setup(struct inpcb *inp, struct sockaddr *nam, in_addr_t *laddrp,
 	int error;
 
 	/*
-	 * Because no actual state changes occur here, a global write lock on
-	 * the pcbinfo isn't required.
+	 * No state changes, so read locks are sufficient here.
 	 */
-	INP_INFO_LOCK_ASSERT(pcbinfo);
 	INP_LOCK_ASSERT(inp);
+	INP_HASH_LOCK_ASSERT(pcbinfo);
 
 	if (TAILQ_EMPTY(&V_in_ifaddrhead)) /* XXX broken! */
 		return (EADDRNOTAVAIL);
@@ -618,8 +627,8 @@ in_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct ucred *cred)
 	in_addr_t laddr, faddr;
 	int anonport, error;
 
-	INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo);
 	INP_WLOCK_ASSERT(inp);
+	INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo);
 
 	lport = inp->inp_lport;
 	laddr = inp->inp_laddr.s_addr;
@@ -907,8 +916,8 @@ in_pcbconnect_setup(struct inpcb *inp, struct sockaddr *nam,
 	 * Because a global state change doesn't actually occur here, a read
 	 * lock is sufficient.
 	 */
-	INP_INFO_LOCK_ASSERT(inp->inp_pcbinfo);
 	INP_LOCK_ASSERT(inp);
+	INP_HASH_LOCK_ASSERT(inp->inp_pcbinfo);
 
 	if (oinpp != NULL)
 		*oinpp = NULL;
@@ -983,8 +992,8 @@ in_pcbconnect_setup(struct inpcb *inp, struct sockaddr *nam,
 		if (error)
 			return (error);
 	}
-	oinp = in_pcblookup_hash(inp->inp_pcbinfo, faddr, fport, laddr, lport,
-	    0, NULL);
+	oinp = in_pcblookup_hash_locked(inp->inp_pcbinfo, faddr, fport,
+	    laddr, lport, 0, NULL);
 	if (oinp != NULL) {
 		if (oinpp != NULL)
 			*oinpp = oinp;
@@ -1007,8 +1016,8 @@ void
 in_pcbdisconnect(struct inpcb *inp)
 {
 
-	INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo);
 	INP_WLOCK_ASSERT(inp);
+	INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo);
 
 	inp->inp_faddr.s_addr = INADDR_ANY;
 	inp->inp_fport = 0;
@@ -1187,19 +1196,24 @@ void
 in_pcbdrop(struct inpcb *inp)
 {
 
-	INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo);
 	INP_WLOCK_ASSERT(inp);
 
+	/*
+	 * XXXRW: Possibly we should protect the setting of INP_DROPPED with
+	 * the hash lock...?
+	 */
 	inp->inp_flags |= INP_DROPPED;
 	if (inp->inp_flags & INP_INHASHLIST) {
 		struct inpcbport *phd = inp->inp_phd;
 
+		INP_HASH_WLOCK(inp->inp_pcbinfo);
 		LIST_REMOVE(inp, inp_hash);
 		LIST_REMOVE(inp, inp_portlist);
 		if (LIST_FIRST(&phd->phd_pcblist) == NULL) {
 			LIST_REMOVE(phd, phd_hash);
 			free(phd, M_PCB);
 		}
+		INP_HASH_WUNLOCK(inp->inp_pcbinfo);
 		inp->inp_flags &= ~INP_INHASHLIST;
 	}
 }
@@ -1328,7 +1342,8 @@ in_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp)
 }
 
 /*
- * Lookup a PCB based on the local address and port.
+ * Lookup a PCB based on the local address and port.  Caller must hold the
+ * hash lock.  No inpcb locks or references are acquired.
  */
 #define INP_LOOKUP_MAPPED_PCB_COST	3
 struct inpcb *
@@ -1346,7 +1361,7 @@ in_pcblookup_local(struct inpcbinfo *pcbinfo, struct in_addr laddr,
 	KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0,
 	    ("%s: invalid lookup flags %d", __func__, lookupflags));
 
-	INP_INFO_LOCK_ASSERT(pcbinfo);
+	INP_HASH_LOCK_ASSERT(pcbinfo);
 
 	if ((lookupflags & INPLOOKUP_WILDCARD) == 0) {
 		struct inpcbhead *head;
@@ -1450,10 +1465,12 @@ in_pcblookup_local(struct inpcbinfo *pcbinfo, struct in_addr laddr,
 #undef INP_LOOKUP_MAPPED_PCB_COST
 
 /*
- * Lookup PCB in hash list.
+ * Lookup PCB in hash list, using pcbinfo tables.  This variation assumes
+ * that the caller has locked the hash list, and will not perform any further
+ * locking or reference operations on either the hash list or the connection.
  */
-struct inpcb *
-in_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in_addr faddr,
+static struct inpcb *
+in_pcblookup_hash_locked(struct inpcbinfo *pcbinfo, struct in_addr faddr,
     u_int fport_arg, struct in_addr laddr, u_int lport_arg, int lookupflags,
     struct ifnet *ifp)
 {
@@ -1464,7 +1481,7 @@ in_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in_addr faddr,
 	KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0,
 	    ("%s: invalid lookup flags %d", __func__, lookupflags));
 
-	INP_INFO_LOCK_ASSERT(pcbinfo);
+	INP_HASH_LOCK_ASSERT(pcbinfo);
 
 	/*
 	 * First look for an exact match.
@@ -1574,6 +1591,56 @@ in_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in_addr faddr,
 
 	return (NULL);
 }
+
+/*
+ * Lookup PCB in hash list, using pcbinfo tables.  This variation locks the
+ * hash list lock, and will return the inpcb locked (i.e., requires
+ * INPLOOKUP_LOCKPCB).
+ */
+static struct inpcb *
+in_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in_addr faddr,
+    u_int fport, struct in_addr laddr, u_int lport, int lookupflags,
+    struct ifnet *ifp)
+{
+	struct inpcb *inp;
+
+	INP_HASH_RLOCK(pcbinfo);
+	inp = in_pcblookup_hash_locked(pcbinfo, faddr, fport, laddr, lport,
+	    (lookupflags & ~(INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)), ifp);
+	if (inp != NULL) {
+		in_pcbref(inp);
+		INP_HASH_RUNLOCK(pcbinfo);
+		if (lookupflags & INPLOOKUP_WLOCKPCB) {
+			INP_WLOCK(inp);
+			if (in_pcbrele_wlocked(inp))
+				return (NULL);
+		} else if (lookupflags & INPLOOKUP_RLOCKPCB) {
+			INP_RLOCK(inp);
+			if (in_pcbrele_rlocked(inp))
+				return (NULL);
+		} else
+			panic("%s: locking bug", __func__);
+	} else
+		INP_HASH_RUNLOCK(pcbinfo);
+	return (inp);
+}
+
+/*
+ * Public inpcb lookup routines, accepting a 4-tuple.
+ */
+struct inpcb *
+in_pcblookup(struct inpcbinfo *pcbinfo, struct in_addr faddr, u_int fport,
+    struct in_addr laddr, u_int lport, int lookupflags, struct ifnet *ifp)
+{
+
+	KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0,
+	    ("%s: invalid lookup flags %d", __func__, lookupflags));
+	KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0,
+	    ("%s: LOCKPCB not set", __func__));
+
+	return (in_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport,
+	    lookupflags, ifp));
+}
 #endif /* INET */
 
 /*
@@ -1588,8 +1655,9 @@ in_pcbinshash(struct inpcb *inp)
 	struct inpcbport *phd;
 	u_int32_t hashkey_faddr;
 
-	INP_INFO_WLOCK_ASSERT(pcbinfo);
 	INP_WLOCK_ASSERT(inp);
+	INP_HASH_WLOCK_ASSERT(pcbinfo);
+
 	KASSERT((inp->inp_flags & INP_INHASHLIST) == 0,
 	    ("in_pcbinshash: INP_INHASHLIST"));
 
@@ -1645,8 +1713,9 @@ in_pcbrehash(struct inpcb *inp)
 	struct inpcbhead *head;
 	u_int32_t hashkey_faddr;
 
-	INP_INFO_WLOCK_ASSERT(pcbinfo);
 	INP_WLOCK_ASSERT(inp);
+	INP_HASH_WLOCK_ASSERT(pcbinfo);
+
 	KASSERT(inp->inp_flags & INP_INHASHLIST,
 	    ("in_pcbrehash: !INP_INHASHLIST"));
 
@@ -1679,12 +1748,14 @@ in_pcbremlists(struct inpcb *inp)
 	if (inp->inp_flags & INP_INHASHLIST) {
 		struct inpcbport *phd = inp->inp_phd;
 
+		INP_HASH_WLOCK(pcbinfo);
 		LIST_REMOVE(inp, inp_hash);
 		LIST_REMOVE(inp, inp_portlist);
 		if (LIST_FIRST(&phd->phd_pcblist) == NULL) {
 			LIST_REMOVE(phd, phd_hash);
 			free(phd, M_PCB);
 		}
+		INP_HASH_WUNLOCK(pcbinfo);
 		inp->inp_flags &= ~INP_INHASHLIST;
 	}
 	LIST_REMOVE(inp, inp_list);
diff --git a/sys/netinet/in_pcb.h b/sys/netinet/in_pcb.h
index 14d4ea2..bde9e8d 100644
--- a/sys/netinet/in_pcb.h
+++ b/sys/netinet/in_pcb.h
@@ -268,22 +268,22 @@ struct inpcbport {
  * Global data structure for each high-level protocol (UDP, TCP, ...) in both
  * IPv4 and IPv6.  Holds inpcb lists and information for managing them.
  *
- * Each pcbinfo is protected by ipi_lock, covering mutable global fields (such
- * as the global pcb list) and hashed lookup tables.  The lock order is:
+ * Each pcbinfo is protected by two locks: ipi_lock and ipi_hash_lock,
+ * the former covering mutable global fields (such as the global pcb list),
+ * and the latter covering the hashed lookup tables.  The lock order is:
  *
- *    ipi_lock (before) inpcb locks
+ *    ipi_lock (before) inpcb locks (before) ipi_hash_lock
  *
  * Locking key:
  *
  * (c) Constant or nearly constant after initialisation
  * (g) Locked by ipi_lock
- * (h) Read using either ipi_lock or inpcb lock; write requires both.
+ * (h) Read using either ipi_hash_lock or inpcb lock; write requires both.
  * (x) Synchronisation properties poorly defined
  */
 struct inpcbinfo {
 	/*
-	 * Global lock protecting global inpcb list, inpcb count, hash tables,
-	 * etc.
+	 * Global lock protecting global inpcb list, inpcb count, etc.
 	 */
 	struct rwlock		 ipi_lock;
 
@@ -312,17 +312,22 @@ struct inpcbinfo {
 	struct	uma_zone	*ipi_zone;		/* (c) */
 
 	/*
+	 * Global lock protecting hash lookup tables.
+	 */
+	struct rwlock		 ipi_hash_lock;
+
+	/*
 	 * Global hash of inpcbs, hashed by local and foreign addresses and
 	 * port numbers.
 	 */
-	struct inpcbhead	*ipi_hashbase;		/* (g) */
-	u_long			 ipi_hashmask;		/* (g) */
+	struct inpcbhead	*ipi_hashbase;		/* (h) */
+	u_long			 ipi_hashmask;		/* (h) */
 
 	/*
 	 * Global hash of inpcbs, hashed by only local port number.
 	 */
-	struct inpcbporthead	*ipi_porthashbase;	/* (g) */
-	u_long			 ipi_porthashmask;	/* (g) */
+	struct inpcbporthead	*ipi_porthashbase;	/* (h) */
+	u_long			 ipi_porthashmask;	/* (h) */
 
 	/*
 	 * Pointer to network stack instance
@@ -406,6 +411,18 @@ void 	inp_4tuple_get(struct inpcb *inp, uint32_t *laddr, uint16_t *lp,
 #define INP_INFO_WLOCK_ASSERT(ipi)	rw_assert(&(ipi)->ipi_lock, RA_WLOCKED)
 #define INP_INFO_UNLOCK_ASSERT(ipi)	rw_assert(&(ipi)->ipi_lock, RA_UNLOCKED)
 
+#define	INP_HASH_LOCK_INIT(ipi, d) \
+	rw_init_flags(&(ipi)->ipi_hash_lock, (d), 0)
+#define	INP_HASH_LOCK_DESTROY(ipi)	rw_destroy(&(ipi)->ipi_hash_lock)
+#define	INP_HASH_RLOCK(ipi)		rw_rlock(&(ipi)->ipi_hash_lock)
+#define	INP_HASH_WLOCK(ipi)		rw_wlock(&(ipi)->ipi_hash_lock)
+#define	INP_HASH_RUNLOCK(ipi)		rw_runlock(&(ipi)->ipi_hash_lock)
+#define	INP_HASH_WUNLOCK(ipi)		rw_wunlock(&(ipi)->ipi_hash_lock)
+#define	INP_HASH_LOCK_ASSERT(ipi)	rw_assert(&(ipi)->ipi_hash_lock, \
+					    RA_LOCKED)
+#define	INP_HASH_WLOCK_ASSERT(ipi)	rw_assert(&(ipi)->ipi_hash_lock, \
+					    RA_WLOCKED)
+
 #define INP_PCBHASH(faddr, lport, fport, mask) \
 	(((faddr) ^ ((faddr) >> 16) ^ ntohs((lport) ^ (fport))) & (mask))
 #define INP_PCBPORTHASH(lport, mask) \
@@ -466,7 +483,16 @@ void 	inp_4tuple_get(struct inpcb *inp, uint32_t *laddr, uint16_t *lp,
 #define	INP_LLE_VALID		0x00000001 /* cached lle is valid */	
 #define	INP_RT_VALID		0x00000002 /* cached rtentry is valid */
 
-#define	INPLOOKUP_WILDCARD	1
+/*
+ * Flags passed to in_pcblookup*() functions.
+ */
+#define	INPLOOKUP_WILDCARD	0x00000001	/* Allow wildcard sockets. */
+#define	INPLOOKUP_RLOCKPCB	0x00000002	/* Return inpcb read-locked. */
+#define	INPLOOKUP_WLOCKPCB	0x00000004	/* Return inpcb write-locked. */
+
+#define	INPLOOKUP_MASK	(INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB | \
+			    INPLOOKUP_WLOCKPCB)
+
 #define	sotoinpcb(so)	((struct inpcb *)(so)->so_pcb)
 #define	sotoin6pcb(so)	sotoinpcb(so) /* for KAME src sync over BSD*'s */
 
@@ -527,7 +553,7 @@ struct inpcb *
 	in_pcblookup_local(struct inpcbinfo *,
 	    struct in_addr, u_short, int, struct ucred *);
 struct inpcb *
-	in_pcblookup_hash(struct inpcbinfo *, struct in_addr, u_int,
+	in_pcblookup(struct inpcbinfo *, struct in_addr, u_int,
 	    struct in_addr, u_int, int, struct ifnet *);
 void	in_pcbnotifyall(struct inpcbinfo *pcbinfo, struct in_addr,
 	    int, struct inpcb *(*)(struct inpcb *, int));
diff --git a/sys/netinet/ip_divert.c b/sys/netinet/ip_divert.c
index de88556..9f0e31c 100644
--- a/sys/netinet/ip_divert.c
+++ b/sys/netinet/ip_divert.c
@@ -659,9 +659,9 @@ div_pcblist(SYSCTL_HANDLER_ARGS)
 	INP_INFO_WLOCK(&V_divcbinfo);
 	for (i = 0; i < n; i++) {
 		inp = inp_list[i];
-		INP_WLOCK(inp);
-		if (!in_pcbrele(inp))
-			INP_WUNLOCK(inp);
+		INP_RLOCK(inp);
+		if (!in_pcbrele_rlocked(inp))
+			INP_RUNLOCK(inp);
 	}
 	INP_INFO_WUNLOCK(&V_divcbinfo);
 
diff --git a/sys/netinet/ipfw/ip_fw2.c b/sys/netinet/ipfw/ip_fw2.c
index 85ae1b2..6672084 100644
--- a/sys/netinet/ipfw/ip_fw2.c
+++ b/sys/netinet/ipfw/ip_fw2.c
@@ -657,7 +657,7 @@ check_uidgid(ipfw_insn_u32 *insn, int proto, struct ifnet *oif,
 	    (struct bsd_ucred *)uc, ugid_lookupp, ((struct mbuf *)inp)->m_skb);
 #else  /* FreeBSD */
 	struct inpcbinfo *pi;
-	int wildcard;
+	int lookupflags;
 	struct inpcb *pcb;
 	int match;
 
@@ -682,30 +682,31 @@ check_uidgid(ipfw_insn_u32 *insn, int proto, struct ifnet *oif,
 	if (*ugid_lookupp == -1)
 		return (0);
 	if (proto == IPPROTO_TCP) {
-		wildcard = 0;
+		lookupflags = 0;
 		pi = &V_tcbinfo;
 	} else if (proto == IPPROTO_UDP) {
-		wildcard = INPLOOKUP_WILDCARD;
+		lookupflags = INPLOOKUP_WILDCARD;
 		pi = &V_udbinfo;
 	} else
 		return 0;
+	lookupflags |= INPLOOKUP_RLOCKPCB;
 	match = 0;
 	if (*ugid_lookupp == 0) {
-		INP_INFO_RLOCK(pi);
 		pcb =  (oif) ?
-			in_pcblookup_hash(pi,
+			in_pcblookup(pi,
 				dst_ip, htons(dst_port),
 				src_ip, htons(src_port),
-				wildcard, oif) :
-			in_pcblookup_hash(pi,
+				lookupflags, oif) :
+			in_pcblookup(pi,
 				src_ip, htons(src_port),
 				dst_ip, htons(dst_port),
-				wildcard, NULL);
+				lookupflags, NULL);
 		if (pcb != NULL) {
+			INP_RLOCK_ASSERT(pcb);
 			*uc = crhold(pcb->inp_cred);
 			*ugid_lookupp = 1;
+			INP_RUNLOCK(pcb);
 		}
-		INP_INFO_RUNLOCK(pi);
 		if (*ugid_lookupp == 0) {
 			/*
 			 * We tried and failed, set the variable to -1
@@ -1827,21 +1828,32 @@ do {								\
 				else
 					break;
 
+				/*
+				 * XXXRW: so_user_cookie should almost
+				 * certainly be inp_user_cookie?
+				 */
+
 				/* For incomming packet, lookup up the 
 				inpcb using the src/dest ip/port tuple */
 				if (inp == NULL) {
-					INP_INFO_RLOCK(pi);
-					inp = in_pcblookup_hash(pi, 
+					inp = in_pcblookup(pi, 
 						src_ip, htons(src_port),
 						dst_ip, htons(dst_port),
-						0, NULL);
-					INP_INFO_RUNLOCK(pi);
-				}
-				
-				if (inp && inp->inp_socket) {
-					tablearg = inp->inp_socket->so_user_cookie;
-					if (tablearg)
-						match = 1;
+						INPLOOKUP_RLOCKPCB, NULL);
+					if (inp != NULL) {
+						tablearg =
+						    inp->inp_socket->so_user_cookie;
+						if (tablearg)
+							match = 1;
+						INP_RUNLOCK(inp);
+					}
+				} else {
+					if (inp->inp_socket) {
+						tablearg =
+						    inp->inp_socket->so_user_cookie;
+						if (tablearg)
+							match = 1;
+					}
 				}
 				break;
 			}
diff --git a/sys/netinet/raw_ip.c b/sys/netinet/raw_ip.c
index be099a8..635f08f 100644
--- a/sys/netinet/raw_ip.c
+++ b/sys/netinet/raw_ip.c
@@ -226,7 +226,7 @@ rip_append(struct inpcb *last, struct ip *ip, struct mbuf *n,
 {
 	int policyfail = 0;
 
-	INP_RLOCK_ASSERT(last);
+	INP_LOCK_ASSERT(last);
 
 #ifdef IPSEC
 	/* check AH/ESP integrity. */
@@ -834,16 +834,19 @@ rip_detach(struct socket *so)
 static void
 rip_dodisconnect(struct socket *so, struct inpcb *inp)
 {
+	struct inpcbinfo *pcbinfo;
 
-	INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo);
-	INP_WLOCK_ASSERT(inp);
-
+	pcbinfo = inp->inp_pcbinfo;
+	INP_INFO_WLOCK(pcbinfo);
+	INP_WLOCK(inp);
 	rip_delhash(inp);
 	inp->inp_faddr.s_addr = INADDR_ANY;
 	rip_inshash(inp);
 	SOCK_LOCK(so);
 	so->so_state &= ~SS_ISCONNECTED;
 	SOCK_UNLOCK(so);
+	INP_WUNLOCK(inp);
+	INP_INFO_WUNLOCK(pcbinfo);
 }
 
 static void
@@ -854,11 +857,7 @@ rip_abort(struct socket *so)
 	inp = sotoinpcb(so);
 	KASSERT(inp != NULL, ("rip_abort: inp == NULL"));
 
-	INP_INFO_WLOCK(&V_ripcbinfo);
-	INP_WLOCK(inp);
 	rip_dodisconnect(so, inp);
-	INP_WUNLOCK(inp);
-	INP_INFO_WUNLOCK(&V_ripcbinfo);
 }
 
 static void
@@ -869,11 +868,7 @@ rip_close(struct socket *so)
 	inp = sotoinpcb(so);
 	KASSERT(inp != NULL, ("rip_close: inp == NULL"));
 
-	INP_INFO_WLOCK(&V_ripcbinfo);
-	INP_WLOCK(inp);
 	rip_dodisconnect(so, inp);
-	INP_WUNLOCK(inp);
-	INP_INFO_WUNLOCK(&V_ripcbinfo);
 }
 
 static int
@@ -887,11 +882,7 @@ rip_disconnect(struct socket *so)
 	inp = sotoinpcb(so);
 	KASSERT(inp != NULL, ("rip_disconnect: inp == NULL"));
 
-	INP_INFO_WLOCK(&V_ripcbinfo);
-	INP_WLOCK(inp);
 	rip_dodisconnect(so, inp);
-	INP_WUNLOCK(inp);
-	INP_INFO_WUNLOCK(&V_ripcbinfo);
 	return (0);
 }
 
@@ -1077,9 +1068,9 @@ rip_pcblist(SYSCTL_HANDLER_ARGS)
 	INP_INFO_WLOCK(&V_ripcbinfo);
 	for (i = 0; i < n; i++) {
 		inp = inp_list[i];
-		INP_WLOCK(inp);
-		if (!in_pcbrele(inp))
-			INP_WUNLOCK(inp);
+		INP_RLOCK(inp);
+		if (!in_pcbrele_rlocked(inp))
+			INP_RUNLOCK(inp);
 	}
 	INP_INFO_WUNLOCK(&V_ripcbinfo);
 
diff --git a/sys/netinet/siftr.c b/sys/netinet/siftr.c
index 9d11262..6145a54 100644
--- a/sys/netinet/siftr.c
+++ b/sys/netinet/siftr.c
@@ -696,17 +696,16 @@ siftr_findinpcb(int ipver, struct ip *ip, struct mbuf *m, uint16_t sport,
 
 	/* We need the tcbinfo lock. */
 	INP_INFO_UNLOCK_ASSERT(&V_tcbinfo);
-	INP_INFO_RLOCK(&V_tcbinfo);
 
 	if (dir == PFIL_IN)
 		inp = (ipver == INP_IPV4 ?
-		    in_pcblookup_hash(&V_tcbinfo, ip->ip_src, sport, ip->ip_dst,
-		    dport, 0, m->m_pkthdr.rcvif)
+		    in_pcblookup(&V_tcbinfo, ip->ip_src, sport, ip->ip_dst,
+		    dport, INPLOOKUP_RLOCKPCB, m->m_pkthdr.rcvif)
 		    :
 #ifdef SIFTR_IPV6
-		    in6_pcblookup_hash(&V_tcbinfo,
+		    in6_pcblookup(&V_tcbinfo,
 		    &((struct ip6_hdr *)ip)->ip6_src, sport,
-		    &((struct ip6_hdr *)ip)->ip6_dst, dport, 0,
+		    &((struct ip6_hdr *)ip)->ip6_dst, dport, INPLOOKUP_RLOCKPCB,
 		    m->m_pkthdr.rcvif)
 #else
 		    NULL
@@ -715,13 +714,13 @@ siftr_findinpcb(int ipver, struct ip *ip, struct mbuf *m, uint16_t sport,
 
 	else
 		inp = (ipver == INP_IPV4 ?
-		    in_pcblookup_hash(&V_tcbinfo, ip->ip_dst, dport, ip->ip_src,
-		    sport, 0, m->m_pkthdr.rcvif)
+		    in_pcblookup(&V_tcbinfo, ip->ip_dst, dport, ip->ip_src,
+		    sport, INPLOOKUP_RLOCKPCB, m->m_pkthdr.rcvif)
 		    :
 #ifdef SIFTR_IPV6
-		    in6_pcblookup_hash(&V_tcbinfo,
+		    in6_pcblookup(&V_tcbinfo,
 		    &((struct ip6_hdr *)ip)->ip6_dst, dport,
-		    &((struct ip6_hdr *)ip)->ip6_src, sport, 0,
+		    &((struct ip6_hdr *)ip)->ip6_src, sport, INPLOOKUP_RLOCKPCB,
 		    m->m_pkthdr.rcvif)
 #else
 		    NULL
@@ -734,12 +733,7 @@ siftr_findinpcb(int ipver, struct ip *ip, struct mbuf *m, uint16_t sport,
 			ss->nskip_in_inpcb++;
 		else
 			ss->nskip_out_inpcb++;
-	} else {
-		/* Acquire the inpcb lock. */
-		INP_UNLOCK_ASSERT(inp);
-		INP_RLOCK(inp);
 	}
-	INP_INFO_RUNLOCK(&V_tcbinfo);
 
 	return (inp);
 }
diff --git a/sys/netinet/tcp_input.c b/sys/netinet/tcp_input.c
index 1a94d0a..080b4da 100644
--- a/sys/netinet/tcp_input.c
+++ b/sys/netinet/tcp_input.c
@@ -5,6 +5,7 @@
  *	Swinburne University of Technology, Melbourne, Australia.
  * Copyright (c) 2009-2010 Lawrence Stewart <lstewart@freebsd.org>
  * Copyright (c) 2010 The FreeBSD Foundation
+ * Copyright (c) 2010-2011 Juniper Networks, Inc.
  * All rights reserved.
  *
  * Portions of this software were developed at the Centre for Advanced Internet
@@ -16,6 +17,9 @@
  * Internet Architectures, Swinburne University of Technology, Melbourne,
  * Australia by David Hayes under sponsorship from the FreeBSD Foundation.
  *
+ * Portions of this software were developed by Robert N. M. Watson under
+ * contract to Juniper Networks, Inc.
+ *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions
  * are met:
@@ -197,10 +201,6 @@ SYSCTL_VNET_INT(_net_inet_tcp, OID_AUTO, recvbuf_max, CTLFLAG_RW,
     &VNET_NAME(tcp_autorcvbuf_max), 0,
     "Max size of automatic receive buffer");
 
-int	tcp_read_locking = 1;
-SYSCTL_INT(_net_inet_tcp, OID_AUTO, read_locking, CTLFLAG_RW,
-    &tcp_read_locking, 0, "Enable read locking strategy");
-
 VNET_DEFINE(struct inpcbhead, tcb);
 #define	tcb6	tcb  /* for KAME src sync over BSD*'s */
 VNET_DEFINE(struct inpcbinfo, tcbinfo);
@@ -591,8 +591,7 @@ tcp_input(struct mbuf *m, int off0)
 	char *s = NULL;			/* address and port logging */
 	int ti_locked;
 #define	TI_UNLOCKED	1
-#define	TI_RLOCKED	2
-#define	TI_WLOCKED	3
+#define	TI_WLOCKED	2
 
 #ifdef TCPDEBUG
 	/*
@@ -756,30 +755,25 @@ tcp_input(struct mbuf *m, int off0)
 	drop_hdrlen = off0 + off;
 
 	/*
-	 * Locate pcb for segment, which requires a lock on tcbinfo.
-	 * Optimisticaly acquire a global read lock rather than a write lock
-	 * unless header flags necessarily imply a state change.  There are
-	 * two cases where we might discover later we need a write lock
-	 * despite the flags: ACKs moving a connection out of the syncache,
-	 * and ACKs for a connection in TIMEWAIT.
+	 * Locate pcb for segment; if we're likely to add or remove a
+	 * connection then first acquire pcbinfo lock.  There are two cases
+	 * where we might discover later we need a write lock despite the
+	 * flags: ACKs moving a connection out of the syncache, and ACKs for
+	 * a connection in TIMEWAIT.
 	 */
-	if ((thflags & (TH_SYN | TH_FIN | TH_RST)) != 0 ||
-	    tcp_read_locking == 0) {
+	if ((thflags & (TH_SYN | TH_FIN | TH_RST)) != 0) {
 		INP_INFO_WLOCK(&V_tcbinfo);
 		ti_locked = TI_WLOCKED;
-	} else {
-		INP_INFO_RLOCK(&V_tcbinfo);
-		ti_locked = TI_RLOCKED;
-	}
+	} else
+		ti_locked = TI_UNLOCKED;
 
 findpcb:
 #ifdef INVARIANTS
-	if (ti_locked == TI_RLOCKED)
-		INP_INFO_RLOCK_ASSERT(&V_tcbinfo);
-	else if (ti_locked == TI_WLOCKED)
+	if (ti_locked == TI_WLOCKED) {
 		INP_INFO_WLOCK_ASSERT(&V_tcbinfo);
-	else
-		panic("%s: findpcb ti_locked %d\n", __func__, ti_locked);
+	} else {
+		INP_INFO_UNLOCK_ASSERT(&V_tcbinfo);
+	}
 #endif
 
 #ifdef INET
@@ -797,20 +791,18 @@ findpcb:
 		 * Transparently forwarded. Pretend to be the destination.
 		 * already got one like this?
 		 */
-		inp = in_pcblookup_hash(&V_tcbinfo,
-					ip->ip_src, th->th_sport,
-					ip->ip_dst, th->th_dport,
-					0, m->m_pkthdr.rcvif);
+		inp = in_pcblookup(&V_tcbinfo, ip->ip_src, th->th_sport,
+		    ip->ip_dst, th->th_dport, INPLOOKUP_WLOCKPCB,
+		    m->m_pkthdr.rcvif);
 		if (!inp) {
-			/* It's new.  Try to find the ambushing socket. */
-			inp = in_pcblookup_hash(&V_tcbinfo,
-						ip->ip_src, th->th_sport,
-						next_hop->sin_addr,
-						next_hop->sin_port ?
-						    ntohs(next_hop->sin_port) :
-						    th->th_dport,
-						INPLOOKUP_WILDCARD,
-						m->m_pkthdr.rcvif);
+			/*
+			 * It's new.  Try to find the ambushing socket.
+			 */
+			inp = in_pcblookup(&V_tcbinfo, ip->ip_src,
+			    th->th_sport, next_hop->sin_addr,
+			    next_hop->sin_port ? ntohs(next_hop->sin_port) :
+			    th->th_dport, INPLOOKUP_WILDCARD |
+			    INPLOOKUP_WLOCKPCB, m->m_pkthdr.rcvif);
 		}
 		/* Remove the tag from the packet.  We don't need it anymore. */
 		m_tag_delete(m, fwd_tag);
@@ -820,21 +812,19 @@ findpcb:
 	{
 #ifdef INET6
 		if (isipv6)
-			inp = in6_pcblookup_hash(&V_tcbinfo,
-						 &ip6->ip6_src, th->th_sport,
-						 &ip6->ip6_dst, th->th_dport,
-						 INPLOOKUP_WILDCARD,
-						 m->m_pkthdr.rcvif);
+			inp = in6_pcblookup(&V_tcbinfo, &ip6->ip6_src,
+			    th->th_sport, &ip6->ip6_dst, th->th_dport,
+			    INPLOOKUP_WILDCARD | INPLOOKUP_WLOCKPCB,
+			    m->m_pkthdr.rcvif);
 #endif
 #if defined(INET) && defined(INET6)
 		else
 #endif
 #ifdef INET
-			inp = in_pcblookup_hash(&V_tcbinfo,
-						ip->ip_src, th->th_sport,
-						ip->ip_dst, th->th_dport,
-						INPLOOKUP_WILDCARD,
-						m->m_pkthdr.rcvif);
+			inp = in_pcblookup(&V_tcbinfo, ip->ip_src,
+			    th->th_sport, ip->ip_dst, th->th_dport,
+			    INPLOOKUP_WILDCARD | INPLOOKUP_WLOCKPCB,
+			    m->m_pkthdr.rcvif);
 #endif
 	}
 
@@ -865,7 +855,7 @@ findpcb:
 		rstreason = BANDLIM_RST_CLOSEDPORT;
 		goto dropwithreset;
 	}
-	INP_WLOCK(inp);
+	INP_WLOCK_ASSERT(inp);
 	if (!(inp->inp_flags & INP_HW_FLOWID)
 	    && (m->m_flags & M_FLOWID)
 	    && ((inp->inp_socket == NULL)
@@ -906,28 +896,26 @@ findpcb:
 	 * legitimate new connection attempt the old INPCB gets removed and
 	 * we can try again to find a listening socket.
 	 *
-	 * At this point, due to earlier optimism, we may hold a read lock on
-	 * the inpcbinfo, rather than a write lock.  If so, we need to
-	 * upgrade, or if that fails, acquire a reference on the inpcb, drop
-	 * all locks, acquire a global write lock, and then re-acquire the
-	 * inpcb lock.  We may at that point discover that another thread has
-	 * tried to free the inpcb, in which case we need to loop back and
-	 * try to find a new inpcb to deliver to.
+	 * At this point, due to earlier optimism, we may hold only an inpcb
+	 * lock, and not the inpcbinfo write lock.  If so, we need to try to
+	 * acquire it, or if that fails, acquire a reference on the inpcb,
+	 * drop all locks, acquire a global write lock, and then re-acquire
+	 * the inpcb lock.  We may at that point discover that another thread
+	 * has tried to free the inpcb, in which case we need to loop back
+	 * and try to find a new inpcb to deliver to.
+	 *
+	 * XXXRW: It may be time to rethink timewait locking.
 	 */
 relocked:
 	if (inp->inp_flags & INP_TIMEWAIT) {
-		KASSERT(ti_locked == TI_RLOCKED || ti_locked == TI_WLOCKED,
-		    ("%s: INP_TIMEWAIT ti_locked %d", __func__, ti_locked));
-
-		if (ti_locked == TI_RLOCKED) {
-			if (INP_INFO_TRY_UPGRADE(&V_tcbinfo) == 0) {
+		if (ti_locked == TI_UNLOCKED) {
+			if (INP_INFO_TRY_WLOCK(&V_tcbinfo) == 0) {
 				in_pcbref(inp);
 				INP_WUNLOCK(inp);
-				INP_INFO_RUNLOCK(&V_tcbinfo);
 				INP_INFO_WLOCK(&V_tcbinfo);
 				ti_locked = TI_WLOCKED;
 				INP_WLOCK(inp);
-				if (in_pcbrele(inp)) {
+				if (in_pcbrele_wlocked(inp)) {
 					inp = NULL;
 					goto findpcb;
 				}
@@ -975,26 +963,24 @@ relocked:
 
 	/*
 	 * We've identified a valid inpcb, but it could be that we need an
-	 * inpcbinfo write lock and have only a read lock.  In this case,
-	 * attempt to upgrade/relock using the same strategy as the TIMEWAIT
-	 * case above.  If we relock, we have to jump back to 'relocked' as
-	 * the connection might now be in TIMEWAIT.
+	 * inpcbinfo write lock but don't hold it.  In this case, attempt to
+	 * acquire using the same strategy as the TIMEWAIT case above.  If we
+	 * relock, we have to jump back to 'relocked' as the connection might
+	 * now be in TIMEWAIT.
 	 */
-	if (tp->t_state != TCPS_ESTABLISHED ||
-	    (thflags & (TH_SYN | TH_FIN | TH_RST)) != 0 ||
-	    tcp_read_locking == 0) {
-		KASSERT(ti_locked == TI_RLOCKED || ti_locked == TI_WLOCKED,
-		    ("%s: upgrade check ti_locked %d", __func__, ti_locked));
-
-		if (ti_locked == TI_RLOCKED) {
-			if (INP_INFO_TRY_UPGRADE(&V_tcbinfo) == 0) {
+#ifdef INVARIANTS
+	if ((thflags & (TH_SYN | TH_FIN | TH_RST)) != 0)
+		INP_INFO_WLOCK_ASSERT(&V_tcbinfo);
+#endif
+	if (tp->t_state != TCPS_ESTABLISHED) {
+		if (ti_locked == TI_UNLOCKED) {
+			if (INP_INFO_TRY_WLOCK(&V_tcbinfo) == 0) {
 				in_pcbref(inp);
 				INP_WUNLOCK(inp);
-				INP_INFO_RUNLOCK(&V_tcbinfo);
 				INP_INFO_WLOCK(&V_tcbinfo);
 				ti_locked = TI_WLOCKED;
 				INP_WLOCK(inp);
-				if (in_pcbrele(inp)) {
+				if (in_pcbrele_wlocked(inp)) {
 					inp = NULL;
 					goto findpcb;
 				}
@@ -1027,13 +1013,16 @@ relocked:
 	/*
 	 * When the socket is accepting connections (the INPCB is in LISTEN
 	 * state) we look into the SYN cache if this is a new connection
-	 * attempt or the completion of a previous one.
+	 * attempt or the completion of a previous one.  Because listen
+	 * sockets are never in TCPS_ESTABLISHED, the V_tcbinfo lock will be
+	 * held in this case.
 	 */
 	if (so->so_options & SO_ACCEPTCONN) {
 		struct in_conninfo inc;
 
 		KASSERT(tp->t_state == TCPS_LISTEN, ("%s: so accepting but "
 		    "tp not listening", __func__));
+		INP_INFO_WLOCK_ASSERT(&V_tcbinfo);
 
 		bzero(&inc, sizeof(inc));
 #ifdef INET6
@@ -1371,13 +1360,17 @@ relocked:
 	return;
 
 dropwithreset:
-	if (ti_locked == TI_RLOCKED)
-		INP_INFO_RUNLOCK(&V_tcbinfo);
-	else if (ti_locked == TI_WLOCKED)
+	if (ti_locked == TI_WLOCKED) {
 		INP_INFO_WUNLOCK(&V_tcbinfo);
-	else
-		panic("%s: dropwithreset ti_locked %d", __func__, ti_locked);
-	ti_locked = TI_UNLOCKED;
+		ti_locked = TI_UNLOCKED;
+	}
+#ifdef INVARIANTS
+	else {
+		KASSERT(ti_locked == TI_UNLOCKED, ("%s: dropwithreset "
+		    "ti_locked: %d", __func__, ti_locked));
+		INP_INFO_UNLOCK_ASSERT(&V_tcbinfo);
+	}
+#endif
 
 	if (inp != NULL) {
 		tcp_dropwithreset(m, th, tp, tlen, rstreason);
@@ -1388,13 +1381,17 @@ dropwithreset:
 	goto drop;
 
 dropunlock:
-	if (ti_locked == TI_RLOCKED)
-		INP_INFO_RUNLOCK(&V_tcbinfo);
-	else if (ti_locked == TI_WLOCKED)
+	if (ti_locked == TI_WLOCKED) {
 		INP_INFO_WUNLOCK(&V_tcbinfo);
-	else
-		panic("%s: dropunlock ti_locked %d", __func__, ti_locked);
-	ti_locked = TI_UNLOCKED;
+		ti_locked = TI_UNLOCKED;
+	}
+#ifdef INVARIANTS
+	else {
+		KASSERT(ti_locked == TI_UNLOCKED, ("%s: dropunlock "
+		    "ti_locked: %d", __func__, ti_locked));
+		INP_INFO_UNLOCK_ASSERT(&V_tcbinfo);
+	}
+#endif
 
 	if (inp != NULL)
 		INP_WUNLOCK(inp);
@@ -1449,13 +1446,13 @@ tcp_do_segment(struct mbuf *m, struct tcphdr *th, struct socket *so,
 		INP_INFO_WLOCK_ASSERT(&V_tcbinfo);
 	} else {
 #ifdef INVARIANTS
-		if (ti_locked == TI_RLOCKED)
-			INP_INFO_RLOCK_ASSERT(&V_tcbinfo);
-		else if (ti_locked == TI_WLOCKED)
+		if (ti_locked == TI_WLOCKED)
 			INP_INFO_WLOCK_ASSERT(&V_tcbinfo);
-		else
-			panic("%s: ti_locked %d for EST", __func__,
-			    ti_locked);
+		else {
+			KASSERT(ti_locked == TI_UNLOCKED, ("%s: EST "
+			    "ti_locked: %d", __func__, ti_locked));
+			INP_INFO_UNLOCK_ASSERT(&V_tcbinfo);
+		}
 #endif
 	}
 	INP_WLOCK_ASSERT(tp->t_inpcb);
@@ -1601,13 +1598,8 @@ tcp_do_segment(struct mbuf *m, struct tcphdr *th, struct socket *so,
 				/*
 				 * This is a pure ack for outstanding data.
 				 */
-				if (ti_locked == TI_RLOCKED)
-					INP_INFO_RUNLOCK(&V_tcbinfo);
-				else if (ti_locked == TI_WLOCKED)
+				if (ti_locked == TI_WLOCKED)
 					INP_INFO_WUNLOCK(&V_tcbinfo);
-				else
-					panic("%s: ti_locked %d on pure ACK",
-					    __func__, ti_locked);
 				ti_locked = TI_UNLOCKED;
 
 				TCPSTAT_INC(tcps_predack);
@@ -1708,13 +1700,8 @@ tcp_do_segment(struct mbuf *m, struct tcphdr *th, struct socket *so,
 			 * nothing on the reassembly queue and we have enough
 			 * buffer space to take it.
 			 */
-			if (ti_locked == TI_RLOCKED)
-				INP_INFO_RUNLOCK(&V_tcbinfo);
-			else if (ti_locked == TI_WLOCKED)
+			if (ti_locked == TI_WLOCKED)
 				INP_INFO_WUNLOCK(&V_tcbinfo);
-			else
-				panic("%s: ti_locked %d on pure data "
-				    "segment", __func__, ti_locked);
 			ti_locked = TI_UNLOCKED;
 
 			/* Clean receiver SACK report if present */
@@ -2550,9 +2537,6 @@ tcp_do_segment(struct mbuf *m, struct tcphdr *th, struct socket *so,
 		}
 
 process_ACK:
-		INP_INFO_LOCK_ASSERT(&V_tcbinfo);
-		KASSERT(ti_locked == TI_RLOCKED || ti_locked == TI_WLOCKED,
-		    ("tcp_input: process_ACK ti_locked %d", ti_locked));
 		INP_WLOCK_ASSERT(tp->t_inpcb);
 
 		acked = BYTES_THIS_ACK(tp, th);
@@ -2716,9 +2700,6 @@ process_ACK:
 	}
 
 step6:
-	INP_INFO_LOCK_ASSERT(&V_tcbinfo);
-	KASSERT(ti_locked == TI_RLOCKED || ti_locked == TI_WLOCKED,
-	    ("tcp_do_segment: step6 ti_locked %d", ti_locked));
 	INP_WLOCK_ASSERT(tp->t_inpcb);
 
 	/*
@@ -2804,9 +2785,6 @@ step6:
 			tp->rcv_up = tp->rcv_nxt;
 	}
 dodata:							/* XXX */
-	INP_INFO_LOCK_ASSERT(&V_tcbinfo);
-	KASSERT(ti_locked == TI_RLOCKED || ti_locked == TI_WLOCKED,
-	    ("tcp_do_segment: dodata ti_locked %d", ti_locked));
 	INP_WLOCK_ASSERT(tp->t_inpcb);
 
 	/*
@@ -2938,13 +2916,8 @@ dodata:							/* XXX */
 			return;
 		}
 	}
-	if (ti_locked == TI_RLOCKED)
-		INP_INFO_RUNLOCK(&V_tcbinfo);
-	else if (ti_locked == TI_WLOCKED)
+	if (ti_locked == TI_WLOCKED)
 		INP_INFO_WUNLOCK(&V_tcbinfo);
-	else
-		panic("%s: dodata epilogue ti_locked %d", __func__,
-		    ti_locked);
 	ti_locked = TI_UNLOCKED;
 
 #ifdef TCPDEBUG
@@ -2973,9 +2946,6 @@ check_delack:
 	return;
 
 dropafterack:
-	KASSERT(ti_locked == TI_RLOCKED || ti_locked == TI_WLOCKED,
-	    ("tcp_do_segment: dropafterack ti_locked %d", ti_locked));
-
 	/*
 	 * Generate an ACK dropping incoming segment if it occupies
 	 * sequence space, where the ACK reflects our state.
@@ -3002,13 +2972,8 @@ dropafterack:
 		tcp_trace(TA_DROP, ostate, tp, (void *)tcp_saveipgen,
 			  &tcp_savetcp, 0);
 #endif
-	if (ti_locked == TI_RLOCKED)
-		INP_INFO_RUNLOCK(&V_tcbinfo);
-	else if (ti_locked == TI_WLOCKED)
+	if (ti_locked == TI_WLOCKED)
 		INP_INFO_WUNLOCK(&V_tcbinfo);
-	else
-		panic("%s: dropafterack epilogue ti_locked %d", __func__,
-		    ti_locked);
 	ti_locked = TI_UNLOCKED;
 
 	tp->t_flags |= TF_ACKNOW;
@@ -3018,12 +2983,8 @@ dropafterack:
 	return;
 
 dropwithreset:
-	if (ti_locked == TI_RLOCKED)
-		INP_INFO_RUNLOCK(&V_tcbinfo);
-	else if (ti_locked == TI_WLOCKED)
+	if (ti_locked == TI_WLOCKED)
 		INP_INFO_WUNLOCK(&V_tcbinfo);
-	else
-		panic("%s: dropwithreset ti_locked %d", __func__, ti_locked);
 	ti_locked = TI_UNLOCKED;
 
 	if (tp != NULL) {
@@ -3034,15 +2995,14 @@ dropwithreset:
 	return;
 
 drop:
-	if (ti_locked == TI_RLOCKED)
-		INP_INFO_RUNLOCK(&V_tcbinfo);
-	else if (ti_locked == TI_WLOCKED)
+	if (ti_locked == TI_WLOCKED) {
 		INP_INFO_WUNLOCK(&V_tcbinfo);
+		ti_locked = TI_UNLOCKED;
+	}
 #ifdef INVARIANTS
 	else
 		INP_INFO_UNLOCK_ASSERT(&V_tcbinfo);
 #endif
-	ti_locked = TI_UNLOCKED;
 
 	/*
 	 * Drop space held by incoming segment and return.
diff --git a/sys/netinet/tcp_subr.c b/sys/netinet/tcp_subr.c
index 2c013be..06854ec 100644
--- a/sys/netinet/tcp_subr.c
+++ b/sys/netinet/tcp_subr.c
@@ -1184,9 +1184,9 @@ tcp_pcblist(SYSCTL_HANDLER_ARGS)
 	INP_INFO_WLOCK(&V_tcbinfo);
 	for (i = 0; i < n; i++) {
 		inp = inp_list[i];
-		INP_WLOCK(inp);
-		if (!in_pcbrele(inp))
-			INP_WUNLOCK(inp);
+		INP_RLOCK(inp);
+		if (!in_pcbrele_rlocked(inp))
+			INP_RUNLOCK(inp);
 	}
 	INP_INFO_WUNLOCK(&V_tcbinfo);
 
@@ -1228,12 +1228,9 @@ tcp_getcred(SYSCTL_HANDLER_ARGS)
 	error = SYSCTL_IN(req, addrs, sizeof(addrs));
 	if (error)
 		return (error);
-	INP_INFO_RLOCK(&V_tcbinfo);
-	inp = in_pcblookup_hash(&V_tcbinfo, addrs[1].sin_addr,
-	    addrs[1].sin_port, addrs[0].sin_addr, addrs[0].sin_port, 0, NULL);
+	inp = in_pcblookup(&V_tcbinfo, addrs[1].sin_addr, addrs[1].sin_port,
+	    addrs[0].sin_addr, addrs[0].sin_port, INPLOOKUP_RLOCKPCB, NULL);
 	if (inp != NULL) {
-		INP_RLOCK(inp);
-		INP_INFO_RUNLOCK(&V_tcbinfo);
 		if (inp->inp_socket == NULL)
 			error = ENOENT;
 		if (error == 0)
@@ -1241,10 +1238,8 @@ tcp_getcred(SYSCTL_HANDLER_ARGS)
 		if (error == 0)
 			cru2x(inp->inp_cred, &xuc);
 		INP_RUNLOCK(inp);
-	} else {
-		INP_INFO_RUNLOCK(&V_tcbinfo);
+	} else
 		error = ENOENT;
-	}
 	if (error == 0)
 		error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred));
 	return (error);
@@ -1286,23 +1281,20 @@ tcp6_getcred(SYSCTL_HANDLER_ARGS)
 			return (EINVAL);
 	}
 
-	INP_INFO_RLOCK(&V_tcbinfo);
 #ifdef INET
 	if (mapped == 1)
-		inp = in_pcblookup_hash(&V_tcbinfo,
+		inp = in_pcblookup(&V_tcbinfo,
 			*(struct in_addr *)&addrs[1].sin6_addr.s6_addr[12],
 			addrs[1].sin6_port,
 			*(struct in_addr *)&addrs[0].sin6_addr.s6_addr[12],
-			addrs[0].sin6_port,
-			0, NULL);
+			addrs[0].sin6_port, INPLOOKUP_RLOCKPCB, NULL);
 	else
 #endif
-		inp = in6_pcblookup_hash(&V_tcbinfo,
+		inp = in6_pcblookup(&V_tcbinfo,
 			&addrs[1].sin6_addr, addrs[1].sin6_port,
-			&addrs[0].sin6_addr, addrs[0].sin6_port, 0, NULL);
+			&addrs[0].sin6_addr, addrs[0].sin6_port,
+			INPLOOKUP_RLOCKPCB, NULL);
 	if (inp != NULL) {
-		INP_RLOCK(inp);
-		INP_INFO_RUNLOCK(&V_tcbinfo);
 		if (inp->inp_socket == NULL)
 			error = ENOENT;
 		if (error == 0)
@@ -1310,10 +1302,8 @@ tcp6_getcred(SYSCTL_HANDLER_ARGS)
 		if (error == 0)
 			cru2x(inp->inp_cred, &xuc);
 		INP_RUNLOCK(inp);
-	} else {
-		INP_INFO_RUNLOCK(&V_tcbinfo);
+	} else
 		error = ENOENT;
-	}
 	if (error == 0)
 		error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred));
 	return (error);
@@ -1374,10 +1364,9 @@ tcp_ctlinput(int cmd, struct sockaddr *sa, void *vip)
 		th = (struct tcphdr *)((caddr_t)ip
 				       + (ip->ip_hl << 2));
 		INP_INFO_WLOCK(&V_tcbinfo);
-		inp = in_pcblookup_hash(&V_tcbinfo, faddr, th->th_dport,
-		    ip->ip_src, th->th_sport, 0, NULL);
+		inp = in_pcblookup(&V_tcbinfo, faddr, th->th_dport,
+		    ip->ip_src, th->th_sport, INPLOOKUP_WLOCKPCB, NULL);
 		if (inp != NULL)  {
-			INP_WLOCK(inp);
 			if (!(inp->inp_flags & INP_TIMEWAIT) &&
 			    !(inp->inp_flags & INP_DROPPED) &&
 			    !(inp->inp_socket == NULL)) {
@@ -2154,20 +2143,19 @@ sysctl_drop(SYSCTL_HANDLER_ARGS)
 	switch (addrs[0].ss_family) {
 #ifdef INET6
 	case AF_INET6:
-		inp = in6_pcblookup_hash(&V_tcbinfo, &fin6->sin6_addr,
-		    fin6->sin6_port, &lin6->sin6_addr, lin6->sin6_port, 0,
-		    NULL);
+		inp = in6_pcblookup(&V_tcbinfo, &fin6->sin6_addr,
+		    fin6->sin6_port, &lin6->sin6_addr, lin6->sin6_port,
+		    INPLOOKUP_WLOCKPCB, NULL);
 		break;
 #endif
 #ifdef INET
 	case AF_INET:
-		inp = in_pcblookup_hash(&V_tcbinfo, fin->sin_addr,
-		    fin->sin_port, lin->sin_addr, lin->sin_port, 0, NULL);
+		inp = in_pcblookup(&V_tcbinfo, fin->sin_addr, fin->sin_port,
+		    lin->sin_addr, lin->sin_port, INPLOOKUP_WLOCKPCB, NULL);
 		break;
 #endif
 	}
 	if (inp != NULL) {
-		INP_WLOCK(inp);
 		if (inp->inp_flags & INP_TIMEWAIT) {
 			/*
 			 * XXXRW: There currently exists a state where an
diff --git a/sys/netinet/tcp_syncache.c b/sys/netinet/tcp_syncache.c
index 8262f43..533de4b 100644
--- a/sys/netinet/tcp_syncache.c
+++ b/sys/netinet/tcp_syncache.c
@@ -661,6 +661,7 @@ syncache_socket(struct syncache *sc, struct socket *lso, struct mbuf *m)
 	inp = sotoinpcb(so);
 	inp->inp_inc.inc_fibnum = so->so_fibnum;
 	INP_WLOCK(inp);
+	INP_HASH_WLOCK(&V_tcbinfo);
 
 	/* Insert new socket into PCB hash list. */
 	inp->inp_inc.inc_flags = sc->sc_inc.inc_flags;
@@ -694,6 +695,7 @@ syncache_socket(struct syncache *sc, struct socket *lso, struct mbuf *m)
 			    s, __func__, error);
 			free(s, M_TCPLOG);
 		}
+		INP_HASH_WUNLOCK(&V_tcbinfo);
 		goto abort;
 	}
 #ifdef IPSEC
@@ -737,6 +739,7 @@ syncache_socket(struct syncache *sc, struct socket *lso, struct mbuf *m)
 				    s, __func__, error);
 				free(s, M_TCPLOG);
 			}
+			INP_HASH_WUNLOCK(&V_tcbinfo);
 			goto abort;
 		}
 		/* Override flowlabel from in6_pcbconnect. */
@@ -776,10 +779,12 @@ syncache_socket(struct syncache *sc, struct socket *lso, struct mbuf *m)
 				    s, __func__, error);
 				free(s, M_TCPLOG);
 			}
+			INP_HASH_WUNLOCK(&V_tcbinfo);
 			goto abort;
 		}
 	}
 #endif /* INET */
+	INP_HASH_WUNLOCK(&V_tcbinfo);
 	tp = intotcpcb(inp);
 	tp->t_state = TCPS_SYN_RECEIVED;
 	tp->iss = sc->sc_iss;
diff --git a/sys/netinet/tcp_timer.c b/sys/netinet/tcp_timer.c
index 5c2c5c2..73984c7 100644
--- a/sys/netinet/tcp_timer.c
+++ b/sys/netinet/tcp_timer.c
@@ -490,7 +490,7 @@ tcp_timer_rexmt(void * xtp)
 		INP_WUNLOCK(inp);
 		INP_INFO_WLOCK(&V_tcbinfo);
 		INP_WLOCK(inp);
-		if (in_pcbrele(inp)) {
+		if (in_pcbrele_wlocked(inp)) {
 			INP_INFO_WUNLOCK(&V_tcbinfo);
 			CURVNET_RESTORE();
 			return;
diff --git a/sys/netinet/tcp_usrreq.c b/sys/netinet/tcp_usrreq.c
index 318fe27..9bad970 100644
--- a/sys/netinet/tcp_usrreq.c
+++ b/sys/netinet/tcp_usrreq.c
@@ -2,8 +2,12 @@
  * Copyright (c) 1982, 1986, 1988, 1993
  *	The Regents of the University of California.
  * Copyright (c) 2006-2007 Robert N. M. Watson
+ * Copyright (c) 2010-2011 Juniper Networks, Inc.
  * All rights reserved.
  *
+ * Portions of this software were developed by Robert N. M. Watson under
+ * contract to Juniper Networks, Inc.
+ *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions
  * are met:
@@ -251,7 +255,6 @@ tcp_usr_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
 		return (EAFNOSUPPORT);
 
 	TCPDEBUG0;
-	INP_INFO_WLOCK(&V_tcbinfo);
 	inp = sotoinpcb(so);
 	KASSERT(inp != NULL, ("tcp_usr_bind: inp == NULL"));
 	INP_WLOCK(inp);
@@ -261,11 +264,12 @@ tcp_usr_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
 	}
 	tp = intotcpcb(inp);
 	TCPDEBUG1();
+	INP_HASH_WLOCK(&V_tcbinfo);
 	error = in_pcbbind(inp, nam, td->td_ucred);
+	INP_HASH_WUNLOCK(&V_tcbinfo);
 out:
 	TCPDEBUG2(PRU_BIND);
 	INP_WUNLOCK(inp);
-	INP_INFO_WUNLOCK(&V_tcbinfo);
 
 	return (error);
 }
@@ -292,7 +296,6 @@ tcp6_usr_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
 		return (EAFNOSUPPORT);
 
 	TCPDEBUG0;
-	INP_INFO_WLOCK(&V_tcbinfo);
 	inp = sotoinpcb(so);
 	KASSERT(inp != NULL, ("tcp6_usr_bind: inp == NULL"));
 	INP_WLOCK(inp);
@@ -302,6 +305,7 @@ tcp6_usr_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
 	}
 	tp = intotcpcb(inp);
 	TCPDEBUG1();
+	INP_HASH_WLOCK(&V_tcbinfo);
 	inp->inp_vflag &= ~INP_IPV4;
 	inp->inp_vflag |= INP_IPV6;
 #ifdef INET
@@ -316,15 +320,16 @@ tcp6_usr_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
 			inp->inp_vflag &= ~INP_IPV6;
 			error = in_pcbbind(inp, (struct sockaddr *)&sin,
 			    td->td_ucred);
+			INP_HASH_WUNLOCK(&V_tcbinfo);
 			goto out;
 		}
 	}
 #endif
 	error = in6_pcbbind(inp, nam, td->td_ucred);
+	INP_HASH_WUNLOCK(&V_tcbinfo);
 out:
 	TCPDEBUG2(PRU_BIND);
 	INP_WUNLOCK(inp);
-	INP_INFO_WUNLOCK(&V_tcbinfo);
 	return (error);
 }
 #endif /* INET6 */
@@ -341,7 +346,6 @@ tcp_usr_listen(struct socket *so, int backlog, struct thread *td)
 	struct tcpcb *tp = NULL;
 
 	TCPDEBUG0;
-	INP_INFO_WLOCK(&V_tcbinfo);
 	inp = sotoinpcb(so);
 	KASSERT(inp != NULL, ("tcp_usr_listen: inp == NULL"));
 	INP_WLOCK(inp);
@@ -353,8 +357,10 @@ tcp_usr_listen(struct socket *so, int backlog, struct thread *td)
 	TCPDEBUG1();
 	SOCK_LOCK(so);
 	error = solisten_proto_check(so);
+	INP_HASH_WLOCK(&V_tcbinfo);
 	if (error == 0 && inp->inp_lport == 0)
 		error = in_pcbbind(inp, (struct sockaddr *)0, td->td_ucred);
+	INP_HASH_WUNLOCK(&V_tcbinfo);
 	if (error == 0) {
 		tp->t_state = TCPS_LISTEN;
 		solisten_proto(so, backlog);
@@ -365,7 +371,6 @@ tcp_usr_listen(struct socket *so, int backlog, struct thread *td)
 out:
 	TCPDEBUG2(PRU_LISTEN);
 	INP_WUNLOCK(inp);
-	INP_INFO_WUNLOCK(&V_tcbinfo);
 	return (error);
 }
 #endif /* INET */
@@ -379,7 +384,6 @@ tcp6_usr_listen(struct socket *so, int backlog, struct thread *td)
 	struct tcpcb *tp = NULL;
 
 	TCPDEBUG0;
-	INP_INFO_WLOCK(&V_tcbinfo);
 	inp = sotoinpcb(so);
 	KASSERT(inp != NULL, ("tcp6_usr_listen: inp == NULL"));
 	INP_WLOCK(inp);
@@ -391,12 +395,14 @@ tcp6_usr_listen(struct socket *so, int backlog, struct thread *td)
 	TCPDEBUG1();
 	SOCK_LOCK(so);
 	error = solisten_proto_check(so);
+	INP_HASH_WLOCK(&V_tcbinfo);
 	if (error == 0 && inp->inp_lport == 0) {
 		inp->inp_vflag &= ~INP_IPV4;
 		if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0)
 			inp->inp_vflag |= INP_IPV4;
 		error = in6_pcbbind(inp, (struct sockaddr *)0, td->td_ucred);
 	}
+	INP_HASH_WUNLOCK(&V_tcbinfo);
 	if (error == 0) {
 		tp->t_state = TCPS_LISTEN;
 		solisten_proto(so, backlog);
@@ -406,7 +412,6 @@ tcp6_usr_listen(struct socket *so, int backlog, struct thread *td)
 out:
 	TCPDEBUG2(PRU_LISTEN);
 	INP_WUNLOCK(inp);
-	INP_INFO_WUNLOCK(&V_tcbinfo);
 	return (error);
 }
 #endif /* INET6 */
@@ -440,7 +445,6 @@ tcp_usr_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
 		return (error);
 
 	TCPDEBUG0;
-	INP_INFO_WLOCK(&V_tcbinfo);
 	inp = sotoinpcb(so);
 	KASSERT(inp != NULL, ("tcp_usr_connect: inp == NULL"));
 	INP_WLOCK(inp);
@@ -456,7 +460,6 @@ tcp_usr_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
 out:
 	TCPDEBUG2(PRU_CONNECT);
 	INP_WUNLOCK(inp);
-	INP_INFO_WUNLOCK(&V_tcbinfo);
 	return (error);
 }
 #endif /* INET */
@@ -482,7 +485,6 @@ tcp6_usr_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
 	    && IN6_IS_ADDR_MULTICAST(&sin6p->sin6_addr))
 		return (EAFNOSUPPORT);
 
-	INP_INFO_WLOCK(&V_tcbinfo);
 	inp = sotoinpcb(so);
 	KASSERT(inp != NULL, ("tcp6_usr_connect: inp == NULL"));
 	INP_WLOCK(inp);
@@ -493,6 +495,11 @@ tcp6_usr_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
 	tp = intotcpcb(inp);
 	TCPDEBUG1();
 #ifdef INET
+	/*
+	 * XXXRW: Some confusion: V4/V6 flags relate to binding, and
+	 * therefore probably require the hash lock, which isn't held here.
+	 * Is this a significant problem?
+	 */
 	if (IN6_IS_ADDR_V4MAPPED(&sin6p->sin6_addr)) {
 		struct sockaddr_in sin;
 
@@ -525,7 +532,6 @@ tcp6_usr_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
 out:
 	TCPDEBUG2(PRU_CONNECT);
 	INP_WUNLOCK(inp);
-	INP_INFO_WUNLOCK(&V_tcbinfo);
 	return (error);
 }
 #endif /* INET6 */
@@ -639,6 +645,7 @@ tcp6_usr_accept(struct socket *so, struct sockaddr **nam)
 
 	inp = sotoinpcb(so);
 	KASSERT(inp != NULL, ("tcp6_usr_accept: inp == NULL"));
+	INP_INFO_RLOCK(&V_tcbinfo);
 	INP_WLOCK(inp);
 	if (inp->inp_flags & (INP_TIMEWAIT | INP_DROPPED)) {
 		error = ECONNABORTED;
@@ -664,6 +671,7 @@ tcp6_usr_accept(struct socket *so, struct sockaddr **nam)
 out:
 	TCPDEBUG2(PRU_ACCEPT);
 	INP_WUNLOCK(inp);
+	INP_INFO_RUNLOCK(&V_tcbinfo);
 	if (error == 0) {
 		if (v4)
 			*nam = in6_v4mapsin6_sockaddr(port, &addr);
@@ -750,25 +758,17 @@ tcp_usr_send(struct socket *so, int flags, struct mbuf *m,
 	int error = 0;
 	struct inpcb *inp;
 	struct tcpcb *tp = NULL;
-	int headlocked = 0;
 #ifdef INET6
 	int isipv6;
 #endif
 	TCPDEBUG0;
 
 	/*
-	 * We require the pcbinfo lock in two cases:
-	 *
-	 * (1) An implied connect is taking place, which can result in
-	 *     binding IPs and ports and hence modification of the pcb hash
-	 *     chains.
-	 *
-	 * (2) PRUS_EOF is set, resulting in explicit close on the send.
+	 * We require the pcbinfo lock if we will close the socket as part of
+	 * this call.
 	 */
-	if ((nam != NULL) || (flags & PRUS_EOF)) {
+	if (flags & PRUS_EOF)
 		INP_INFO_WLOCK(&V_tcbinfo);
-		headlocked = 1;
-	}
 	inp = sotoinpcb(so);
 	KASSERT(inp != NULL, ("tcp_usr_send: inp == NULL"));
 	INP_WLOCK(inp);
@@ -805,7 +805,6 @@ tcp_usr_send(struct socket *so, int flags, struct mbuf *m,
 			 * initialize maxseg/maxopd using peer's cached
 			 * MSS.
 			 */
-			INP_INFO_WLOCK_ASSERT(&V_tcbinfo);
 #ifdef INET6
 			if (isipv6)
 				error = tcp6_connect(tp, nam, td);
@@ -830,10 +829,6 @@ tcp_usr_send(struct socket *so, int flags, struct mbuf *m,
 			socantsendmore(so);
 			tcp_usrclosed(tp);
 		}
-		if (headlocked) {
-			INP_INFO_WUNLOCK(&V_tcbinfo);
-			headlocked = 0;
-		}
 		if (!(inp->inp_flags & INP_DROPPED)) {
 			if (flags & PRUS_MORETOCOME)
 				tp->t_flags |= TF_MORETOCOME;
@@ -869,7 +864,6 @@ tcp_usr_send(struct socket *so, int flags, struct mbuf *m,
 			 * initialize maxseg/maxopd using peer's cached
 			 * MSS.
 			 */
-			INP_INFO_WLOCK_ASSERT(&V_tcbinfo);
 #ifdef INET6
 			if (isipv6)
 				error = tcp6_connect(tp, nam, td);
@@ -884,11 +878,6 @@ tcp_usr_send(struct socket *so, int flags, struct mbuf *m,
 				goto out;
 			tp->snd_wnd = TTCP_CLIENT_SND_WND;
 			tcp_mss(tp, -1);
-			INP_INFO_WUNLOCK(&V_tcbinfo);
-			headlocked = 0;
-		} else if (nam) {
-			INP_INFO_WUNLOCK(&V_tcbinfo);
-			headlocked = 0;
 		}
 		tp->snd_up = tp->snd_una + so->so_snd.sb_cc;
 		tp->t_flags |= TF_FORCEDATA;
@@ -899,7 +888,7 @@ out:
 	TCPDEBUG2((flags & PRUS_OOB) ? PRU_SENDOOB :
 		  ((flags & PRUS_EOF) ? PRU_SEND_EOF : PRU_SEND));
 	INP_WUNLOCK(inp);
-	if (headlocked)
+	if (flags & PRUS_EOF)
 		INP_INFO_WUNLOCK(&V_tcbinfo);
 	return (error);
 }
@@ -1087,13 +1076,13 @@ tcp_connect(struct tcpcb *tp, struct sockaddr *nam, struct thread *td)
 	u_short lport;
 	int error;
 
-	INP_INFO_WLOCK_ASSERT(&V_tcbinfo);
 	INP_WLOCK_ASSERT(inp);
+	INP_HASH_WLOCK(&V_tcbinfo);
 
 	if (inp->inp_lport == 0) {
 		error = in_pcbbind(inp, (struct sockaddr *)0, td->td_ucred);
 		if (error)
-			return error;
+			goto out;
 	}
 
 	/*
@@ -1106,11 +1095,14 @@ tcp_connect(struct tcpcb *tp, struct sockaddr *nam, struct thread *td)
 	error = in_pcbconnect_setup(inp, nam, &laddr.s_addr, &lport,
 	    &inp->inp_faddr.s_addr, &inp->inp_fport, &oinp, td->td_ucred);
 	if (error && oinp == NULL)
-		return error;
-	if (oinp)
-		return EADDRINUSE;
+		goto out;
+	if (oinp) {
+		error = EADDRINUSE;
+		goto out;
+	}
 	inp->inp_laddr = laddr;
 	in_pcbrehash(inp);
+	INP_HASH_WUNLOCK(&V_tcbinfo);
 
 	/*
 	 * Compute window scaling to request:
@@ -1129,6 +1121,10 @@ tcp_connect(struct tcpcb *tp, struct sockaddr *nam, struct thread *td)
 	tcp_sendseqinit(tp);
 
 	return 0;
+
+out:
+	INP_HASH_WUNLOCK(&V_tcbinfo);
+	return (error);
 }
 #endif /* INET */
 
@@ -1142,13 +1138,13 @@ tcp6_connect(struct tcpcb *tp, struct sockaddr *nam, struct thread *td)
 	struct in6_addr addr6;
 	int error;
 
-	INP_INFO_WLOCK_ASSERT(&V_tcbinfo);
 	INP_WLOCK_ASSERT(inp);
+	INP_HASH_WLOCK(&V_tcbinfo);
 
 	if (inp->inp_lport == 0) {
 		error = in6_pcbbind(inp, (struct sockaddr *)0, td->td_ucred);
 		if (error)
-			return error;
+			goto out;
 	}
 
 	/*
@@ -1156,18 +1152,23 @@ tcp6_connect(struct tcpcb *tp, struct sockaddr *nam, struct thread *td)
 	 * earlier incarnation of this same connection still in
 	 * TIME_WAIT state, creating an ADDRINUSE error.
 	 * in6_pcbladdr() also handles scope zone IDs.
+	 *
+	 * XXXRW: We wouldn't need to expose in6_pcblookup_hash_locked()
+	 * outside of in6_pcb.c if there were an in6_pcbconnect_setup().
 	 */
 	error = in6_pcbladdr(inp, nam, &addr6);
 	if (error)
 		return error;
-	oinp = in6_pcblookup_hash(inp->inp_pcbinfo,
+	oinp = in6_pcblookup_hash_locked(inp->inp_pcbinfo,
 				  &sin6->sin6_addr, sin6->sin6_port,
 				  IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)
 				  ? &addr6
 				  : &inp->in6p_laddr,
 				  inp->inp_lport,  0, NULL);
-	if (oinp)
-		return EADDRINUSE;
+	if (oinp) {
+		error = EADDRINUSE;
+		goto out;
+	}
 	if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
 		inp->in6p_laddr = addr6;
 	inp->in6p_faddr = sin6->sin6_addr;
@@ -1178,6 +1179,7 @@ tcp6_connect(struct tcpcb *tp, struct sockaddr *nam, struct thread *td)
 		inp->inp_flow |=
 		    (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK);
 	in_pcbrehash(inp);
+	INP_HASH_WUNLOCK(&V_tcbinfo);
 
 	/* Compute window scaling to request.  */
 	while (tp->request_r_scale < TCP_MAX_WINSHIFT &&
@@ -1192,6 +1194,10 @@ tcp6_connect(struct tcpcb *tp, struct sockaddr *nam, struct thread *td)
 	tcp_sendseqinit(tp);
 
 	return 0;
+
+out:
+	INP_HASH_WUNLOCK(&V_tcbinfo);
+	return error;
 }
 #endif /* INET6 */
 
diff --git a/sys/netinet/udp_usrreq.c b/sys/netinet/udp_usrreq.c
index c3503e6..5f2d7c3 100644
--- a/sys/netinet/udp_usrreq.c
+++ b/sys/netinet/udp_usrreq.c
@@ -2,8 +2,12 @@
  * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
  *	The Regents of the University of California.
  * Copyright (c) 2008 Robert N. M. Watson
+ * Copyright (c) 2010-2011 Juniper Networks, Inc.
  * All rights reserved.
  *
+ * Portions of this software were developed by Robert N. M. Watson under
+ * contract to Juniper Networks, Inc.
+ *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions
  * are met:
@@ -253,7 +257,7 @@ udp_append(struct inpcb *inp, struct ip *ip, struct mbuf *n, int off,
 #endif
 	struct udpcb *up;
 
-	INP_RLOCK_ASSERT(inp);
+	INP_LOCK_ASSERT(inp);
 
 	/*
 	 * Engage the tunneling protocol.
@@ -458,12 +462,12 @@ udp_input(struct mbuf *m, int off)
 	}
 #endif
 
-	INP_INFO_RLOCK(&V_udbinfo);
 	if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) ||
 	    in_broadcast(ip->ip_dst, ifp)) {
 		struct inpcb *last;
 		struct ip_moptions *imo;
 
+		INP_INFO_RLOCK(&V_udbinfo);
 		last = NULL;
 		LIST_FOREACH(inp, &V_udb, inp_list) {
 			if (inp->inp_lport != uh->uh_dport)
@@ -485,6 +489,13 @@ udp_input(struct mbuf *m, int off)
 			INP_RLOCK(inp);
 
 			/*
+			 * XXXRW: Because we weren't holding either the inpcb
+			 * or the hash lock when we checked for a match
+			 * before, we should probably recheck now that the
+			 * inpcb lock is held.
+			 */
+
+			/*
 			 * Handle socket delivery policy for any-source
 			 * and source-specific multicast. [RFC3678]
 			 */
@@ -542,7 +553,10 @@ udp_input(struct mbuf *m, int off)
 			 * or multicast datgram.)
 			 */
 			UDPSTAT_INC(udps_noportbcast);
-			goto badheadlocked;
+			if (inp)
+				INP_RUNLOCK(inp);
+			INP_INFO_RUNLOCK(&V_udbinfo);
+			goto badunlocked;
 		}
 		udp_append(last, ip, m, iphlen, &udp_in);
 		INP_RUNLOCK(last);
@@ -553,8 +567,9 @@ udp_input(struct mbuf *m, int off)
 	/*
 	 * Locate pcb for datagram.
 	 */
-	inp = in_pcblookup_hash(&V_udbinfo, ip->ip_src, uh->uh_sport,
-	    ip->ip_dst, uh->uh_dport, 1, ifp);
+	inp = in_pcblookup(&V_udbinfo, ip->ip_src, uh->uh_sport,
+	    ip->ip_dst, uh->uh_dport, INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB,
+	    ifp);
 	if (inp == NULL) {
 		if (udp_log_in_vain) {
 			char buf[4*sizeof "123"];
@@ -568,36 +583,31 @@ udp_input(struct mbuf *m, int off)
 		UDPSTAT_INC(udps_noport);
 		if (m->m_flags & (M_BCAST | M_MCAST)) {
 			UDPSTAT_INC(udps_noportbcast);
-			goto badheadlocked;
+			goto badunlocked;
 		}
 		if (V_udp_blackhole)
-			goto badheadlocked;
+			goto badunlocked;
 		if (badport_bandlim(BANDLIM_ICMP_UNREACH) < 0)
-			goto badheadlocked;
+			goto badunlocked;
 		*ip = save_ip;
 		ip->ip_len += iphlen;
 		icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0);
-		INP_INFO_RUNLOCK(&V_udbinfo);
 		return;
 	}
 
 	/*
 	 * Check the minimum TTL for socket.
 	 */
-	INP_RLOCK(inp);
-	INP_INFO_RUNLOCK(&V_udbinfo);
+	INP_RLOCK_ASSERT(inp);
 	if (inp->inp_ip_minttl && inp->inp_ip_minttl > ip->ip_ttl) {
 		INP_RUNLOCK(inp);
-		goto badunlocked;
+		m_freem(m);
+		return;
 	}
 	udp_append(inp, ip, m, iphlen, &udp_in);
 	INP_RUNLOCK(inp);
 	return;
 
-badheadlocked:
-	if (inp)
-		INP_RUNLOCK(inp);
-	INP_INFO_RUNLOCK(&V_udbinfo);
 badunlocked:
 	m_freem(m);
 }
@@ -656,17 +666,15 @@ udp_ctlinput(int cmd, struct sockaddr *sa, void *vip)
 		return;
 	if (ip != NULL) {
 		uh = (struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2));
-		INP_INFO_RLOCK(&V_udbinfo);
-		inp = in_pcblookup_hash(&V_udbinfo, faddr, uh->uh_dport,
-		    ip->ip_src, uh->uh_sport, 0, NULL);
+		inp = in_pcblookup(&V_udbinfo, faddr, uh->uh_dport,
+		    ip->ip_src, uh->uh_sport, INPLOOKUP_RLOCKPCB, NULL);
 		if (inp != NULL) {
-			INP_RLOCK(inp);
+			INP_RLOCK_ASSERT(inp);
 			if (inp->inp_socket != NULL) {
 				udp_notify(inp, inetctlerrmap[cmd]);
 			}
 			INP_RUNLOCK(inp);
 		}
-		INP_INFO_RUNLOCK(&V_udbinfo);
 	} else
 		in_pcbnotifyall(&V_udbinfo, faddr, inetctlerrmap[cmd],
 		    udp_notify);
@@ -756,9 +764,9 @@ udp_pcblist(SYSCTL_HANDLER_ARGS)
 	INP_INFO_WLOCK(&V_udbinfo);
 	for (i = 0; i < n; i++) {
 		inp = inp_list[i];
-		INP_WLOCK(inp);
-		if (!in_pcbrele(inp))
-			INP_WUNLOCK(inp);
+		INP_RLOCK(inp);
+		if (!in_pcbrele_rlocked(inp))
+			INP_RUNLOCK(inp);
 	}
 	INP_INFO_WUNLOCK(&V_udbinfo);
 
@@ -799,12 +807,11 @@ udp_getcred(SYSCTL_HANDLER_ARGS)
 	error = SYSCTL_IN(req, addrs, sizeof(addrs));
 	if (error)
 		return (error);
-	INP_INFO_RLOCK(&V_udbinfo);
-	inp = in_pcblookup_hash(&V_udbinfo, addrs[1].sin_addr, addrs[1].sin_port,
-				addrs[0].sin_addr, addrs[0].sin_port, 1, NULL);
+	inp = in_pcblookup(&V_udbinfo, addrs[1].sin_addr, addrs[1].sin_port,
+	    addrs[0].sin_addr, addrs[0].sin_port,
+	    INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB, NULL);
 	if (inp != NULL) {
-		INP_RLOCK(inp);
-		INP_INFO_RUNLOCK(&V_udbinfo);
+		INP_RLOCK_ASSERT(inp);
 		if (inp->inp_socket == NULL)
 			error = ENOENT;
 		if (error == 0)
@@ -812,10 +819,8 @@ udp_getcred(SYSCTL_HANDLER_ARGS)
 		if (error == 0)
 			cru2x(inp->inp_cred, &xuc);
 		INP_RUNLOCK(inp);
-	} else {
-		INP_INFO_RUNLOCK(&V_udbinfo);
+	} else
 		error = ENOENT;
-	}
 	if (error == 0)
 		error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred));
 	return (error);
@@ -924,6 +929,9 @@ udp_ctloutput(struct socket *so, struct sockopt *sopt)
 }
 
 #ifdef INET
+#define	UH_WLOCKED	2
+#define	UH_RLOCKED	1
+#define	UH_UNLOCKED	0
 static int
 udp_output(struct inpcb *inp, struct mbuf *m, struct sockaddr *addr,
     struct mbuf *control, struct thread *td)
@@ -1016,29 +1024,27 @@ udp_output(struct inpcb *inp, struct mbuf *m, struct sockaddr *addr,
 	 * conservative locks than required the second time around, so later
 	 * assertions have to accept that.  Further analysis of the number of
 	 * misses under contention is required.
+	 *
+	 * XXXRW: Check that hash locking update here is correct.
 	 */
 	sin = (struct sockaddr_in *)addr;
 	INP_RLOCK(inp);
 	if (sin != NULL &&
 	    (inp->inp_laddr.s_addr == INADDR_ANY && inp->inp_lport == 0)) {
 		INP_RUNLOCK(inp);
-		INP_INFO_WLOCK(&V_udbinfo);
 		INP_WLOCK(inp);
-		unlock_udbinfo = 2;
+		INP_HASH_WLOCK(&V_udbinfo);
+		unlock_udbinfo = UH_WLOCKED;
 	} else if ((sin != NULL && (
 	    (sin->sin_addr.s_addr == INADDR_ANY) ||
 	    (sin->sin_addr.s_addr == INADDR_BROADCAST) ||
 	    (inp->inp_laddr.s_addr == INADDR_ANY) ||
 	    (inp->inp_lport == 0))) ||
 	    (src.sin_family == AF_INET)) {
-		if (!INP_INFO_TRY_RLOCK(&V_udbinfo)) {
-			INP_RUNLOCK(inp);
-			INP_INFO_RLOCK(&V_udbinfo);
-			INP_RLOCK(inp);
-		}
-		unlock_udbinfo = 1;
+		INP_HASH_RLOCK(&V_udbinfo);
+		unlock_udbinfo = UH_RLOCKED;
 	} else
-		unlock_udbinfo = 0;
+		unlock_udbinfo = UH_UNLOCKED;
 
 	/*
 	 * If the IP_SENDSRCADDR control message was specified, override the
@@ -1048,7 +1054,7 @@ udp_output(struct inpcb *inp, struct mbuf *m, struct sockaddr *addr,
 	laddr = inp->inp_laddr;
 	lport = inp->inp_lport;
 	if (src.sin_family == AF_INET) {
-		INP_INFO_LOCK_ASSERT(&V_udbinfo);
+		INP_HASH_LOCK_ASSERT(&V_udbinfo);
 		if ((lport == 0) ||
 		    (laddr.s_addr == INADDR_ANY &&
 		     src.sin_addr.s_addr == INADDR_ANY)) {
@@ -1099,7 +1105,7 @@ udp_output(struct inpcb *inp, struct mbuf *m, struct sockaddr *addr,
 		    inp->inp_lport == 0 ||
 		    sin->sin_addr.s_addr == INADDR_ANY ||
 		    sin->sin_addr.s_addr == INADDR_BROADCAST) {
-			INP_INFO_LOCK_ASSERT(&V_udbinfo);
+			INP_HASH_LOCK_ASSERT(&V_udbinfo);
 			error = in_pcbconnect_setup(inp, addr, &laddr.s_addr,
 			    &lport, &faddr.s_addr, &fport, NULL,
 			    td->td_ucred);
@@ -1113,8 +1119,8 @@ udp_output(struct inpcb *inp, struct mbuf *m, struct sockaddr *addr,
 			/* Commit the local port if newly assigned. */
 			if (inp->inp_laddr.s_addr == INADDR_ANY &&
 			    inp->inp_lport == 0) {
-				INP_INFO_WLOCK_ASSERT(&V_udbinfo);
 				INP_WLOCK_ASSERT(inp);
+				INP_HASH_WLOCK_ASSERT(&V_udbinfo);
 				/*
 				 * Remember addr if jailed, to prevent
 				 * rebinding.
@@ -1209,25 +1215,25 @@ udp_output(struct inpcb *inp, struct mbuf *m, struct sockaddr *addr,
 	((struct ip *)ui)->ip_tos = inp->inp_ip_tos;	/* XXX */
 	UDPSTAT_INC(udps_opackets);
 
-	if (unlock_udbinfo == 2)
-		INP_INFO_WUNLOCK(&V_udbinfo);
-	else if (unlock_udbinfo == 1)
-		INP_INFO_RUNLOCK(&V_udbinfo);
+	if (unlock_udbinfo == UH_WLOCKED)
+		INP_HASH_WUNLOCK(&V_udbinfo);
+	else if (unlock_udbinfo == UH_RLOCKED)
+		INP_HASH_RUNLOCK(&V_udbinfo);
 	error = ip_output(m, inp->inp_options, NULL, ipflags,
 	    inp->inp_moptions, inp);
-	if (unlock_udbinfo == 2)
+	if (unlock_udbinfo == UH_WLOCKED)
 		INP_WUNLOCK(inp);
 	else
 		INP_RUNLOCK(inp);
 	return (error);
 
 release:
-	if (unlock_udbinfo == 2) {
+	if (unlock_udbinfo == UH_WLOCKED) {
+		INP_HASH_WUNLOCK(&V_udbinfo);
 		INP_WUNLOCK(inp);
-		INP_INFO_WUNLOCK(&V_udbinfo);
-	} else if (unlock_udbinfo == 1) {
+	} else if (unlock_udbinfo == UH_RLOCKED) {
+		INP_HASH_RUNLOCK(&V_udbinfo);
 		INP_RUNLOCK(inp);
-		INP_INFO_RUNLOCK(&V_udbinfo);
 	} else
 		INP_RUNLOCK(inp);
 	m_freem(m);
@@ -1376,15 +1382,15 @@ udp_abort(struct socket *so)
 
 	inp = sotoinpcb(so);
 	KASSERT(inp != NULL, ("udp_abort: inp == NULL"));
-	INP_INFO_WLOCK(&V_udbinfo);
 	INP_WLOCK(inp);
 	if (inp->inp_faddr.s_addr != INADDR_ANY) {
+		INP_HASH_WLOCK(&V_udbinfo);
 		in_pcbdisconnect(inp);
 		inp->inp_laddr.s_addr = INADDR_ANY;
+		INP_HASH_WUNLOCK(&V_udbinfo);
 		soisdisconnected(so);
 	}
 	INP_WUNLOCK(inp);
-	INP_INFO_WUNLOCK(&V_udbinfo);
 }
 
 static int
@@ -1453,11 +1459,11 @@ udp_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
 
 	inp = sotoinpcb(so);
 	KASSERT(inp != NULL, ("udp_bind: inp == NULL"));
-	INP_INFO_WLOCK(&V_udbinfo);
 	INP_WLOCK(inp);
+	INP_HASH_WLOCK(&V_udbinfo);
 	error = in_pcbbind(inp, nam, td->td_ucred);
+	INP_HASH_WUNLOCK(&V_udbinfo);
 	INP_WUNLOCK(inp);
-	INP_INFO_WUNLOCK(&V_udbinfo);
 	return (error);
 }
 
@@ -1468,15 +1474,15 @@ udp_close(struct socket *so)
 
 	inp = sotoinpcb(so);
 	KASSERT(inp != NULL, ("udp_close: inp == NULL"));
-	INP_INFO_WLOCK(&V_udbinfo);
 	INP_WLOCK(inp);
 	if (inp->inp_faddr.s_addr != INADDR_ANY) {
+		INP_HASH_WLOCK(&V_udbinfo);
 		in_pcbdisconnect(inp);
 		inp->inp_laddr.s_addr = INADDR_ANY;
+		INP_HASH_WUNLOCK(&V_udbinfo);
 		soisdisconnected(so);
 	}
 	INP_WUNLOCK(inp);
-	INP_INFO_WUNLOCK(&V_udbinfo);
 }
 
 static int
@@ -1488,25 +1494,23 @@ udp_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
 
 	inp = sotoinpcb(so);
 	KASSERT(inp != NULL, ("udp_connect: inp == NULL"));
-	INP_INFO_WLOCK(&V_udbinfo);
 	INP_WLOCK(inp);
 	if (inp->inp_faddr.s_addr != INADDR_ANY) {
 		INP_WUNLOCK(inp);
-		INP_INFO_WUNLOCK(&V_udbinfo);
 		return (EISCONN);
 	}
 	sin = (struct sockaddr_in *)nam;
 	error = prison_remote_ip4(td->td_ucred, &sin->sin_addr);
 	if (error != 0) {
 		INP_WUNLOCK(inp);
-		INP_INFO_WUNLOCK(&V_udbinfo);
 		return (error);
 	}
+	INP_HASH_WLOCK(&V_udbinfo);
 	error = in_pcbconnect(inp, nam, td->td_ucred);
+	INP_HASH_WUNLOCK(&V_udbinfo);
 	if (error == 0)
 		soisconnected(so);
 	INP_WUNLOCK(inp);
-	INP_INFO_WUNLOCK(&V_udbinfo);
 	return (error);
 }
 
@@ -1538,21 +1542,19 @@ udp_disconnect(struct socket *so)
 
 	inp = sotoinpcb(so);
 	KASSERT(inp != NULL, ("udp_disconnect: inp == NULL"));
-	INP_INFO_WLOCK(&V_udbinfo);
 	INP_WLOCK(inp);
 	if (inp->inp_faddr.s_addr == INADDR_ANY) {
 		INP_WUNLOCK(inp);
-		INP_INFO_WUNLOCK(&V_udbinfo);
 		return (ENOTCONN);
 	}
-
+	INP_HASH_WLOCK(&V_udbinfo);
 	in_pcbdisconnect(inp);
 	inp->inp_laddr.s_addr = INADDR_ANY;
+	INP_HASH_WUNLOCK(&V_udbinfo);
 	SOCK_LOCK(so);
 	so->so_state &= ~SS_ISCONNECTED;		/* XXX */
 	SOCK_UNLOCK(so);
 	INP_WUNLOCK(inp);
-	INP_INFO_WUNLOCK(&V_udbinfo);
 	return (0);
 }