From: Alexander Bluhm <bluhm@openbsd.org>
Subject: Re: shared netlock for soclose()
To: tech@openbsd.org
Date: Wed, 4 Jun 2025 17:09:25 +0200

On Tue, Jun 03, 2025 at 02:33:03PM +0200, Alexander Bluhm wrote:
> Hi,
> 
> Currently soclose() needs exclusive netlock.  I see deadlocks as
> tcp_newtcpcb() calls pool_get(&tcpcb_pool) with PR_WAITOK while
> holding shared netlock.  Running memory allocation in parallel
> although free needs an exclusive lock is a bad idea.
> 
> Solution is to run soclose() in parallel.
> 
> Diff consists of four parts:
> - always hold reference count of socket in inp_socket
> - inp_socket is not longer protected by special mutex
> - account when other CPU sets so_pcb to NULL
> - run soclose() and sofree() with shared netlock
> - tcp timeout reaper can go away
> 
> Please test.  I will split the diff for review.

Some parts have been commited.  New diff that applies to current.

bluhm

Index: kern/uipc_socket.c
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/kern/uipc_socket.c,v
diff -u -p -r1.378 uipc_socket.c
--- kern/uipc_socket.c	23 May 2025 23:41:46 -0000	1.378
+++ kern/uipc_socket.c	4 Jun 2025 14:18:03 -0000
@@ -213,10 +213,7 @@ socreate(int dom, struct socket **aso, i
 	if (error) {
 		so->so_state |= SS_NOFDREF;
 		/* sofree() calls sounlock(). */
-		soref(so);
-		sofree(so, 1);
-		sounlock_shared(so);
-		sorele(so);
+		sofree(so, 0);
 		return (error);
 	}
 	sounlock_shared(so);
@@ -304,7 +301,7 @@ sofree(struct socket *so, int keep_lock)
 
 	if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0) {
 		if (!keep_lock)
-			sounlock(so);
+			sounlock_shared(so);
 		return;
 	}
 	if (so->so_head) {
@@ -317,7 +314,7 @@ sofree(struct socket *so, int keep_lock)
 		 */
 		if (so->so_onq == &head->so_q) {
 			if (!keep_lock)
-				sounlock(so);
+				sounlock_shared(so);
 			return;
 		}
 
@@ -344,7 +341,7 @@ sofree(struct socket *so, int keep_lock)
 	}
 
 	if (!keep_lock)
-		sounlock(so);
+		sounlock_shared(so);
 	sorele(so);
 }
 
@@ -368,7 +365,7 @@ soclose(struct socket *so, int flags)
 	struct socket *so2;
 	int error = 0;
 
-	solock(so);
+	solock_shared(so);
 	/* Revoke async IO early. There is a final revocation in sofree(). */
 	sigio_free(&so->so_sigio);
 	if (so->so_state & SS_ISCONNECTED) {
@@ -430,7 +427,7 @@ discard:
 	if (so->so_sp) {
 		struct socket *soback;
 
-		sounlock(so);
+		sounlock_shared(so);
 		/*
 		 * Concurrent sounsplice() locks `sb_mtx' mutexes on
 		 * both `so_snd' and `so_rcv' before unsplice sockets.
@@ -477,7 +474,7 @@ notsplicedback:
 		task_del(sosplice_taskq, &so->so_sp->ssp_task);
 		taskq_barrier(sosplice_taskq);
 
-		solock(so);
+		solock_shared(so);
 	}
 #endif /* SOCKET_SPLICE */
 
Index: netinet/in_pcb.h
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/netinet/in_pcb.h,v
diff -u -p -r1.169 in_pcb.h
--- netinet/in_pcb.h	3 Jun 2025 16:51:26 -0000	1.169
+++ netinet/in_pcb.h	4 Jun 2025 14:18:03 -0000
@@ -138,7 +138,7 @@ struct inpcb {
 	u_int16_t inp_fport;		/* [t] foreign port */
 	u_int16_t inp_lport;		/* [t] local port */
 	struct	  socket *inp_socket;	/* [I] back pointer to socket */
-	caddr_t	  inp_ppcb;		/* pointer to per-protocol pcb */
+	caddr_t	  inp_ppcb;		/* [s] pointer to per-protocol pcb */
 	struct    route inp_route;	/* [s] cached route */
 	struct    refcnt inp_refcnt;	/* refcount PCB, delay memory free */
 	int	  inp_flags;		/* generic IP/datagram flags */
Index: netinet/ip_divert.c
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/netinet/ip_divert.c,v
diff -u -p -r1.104 ip_divert.c
--- netinet/ip_divert.c	4 Jun 2025 12:37:00 -0000	1.104
+++ netinet/ip_divert.c	4 Jun 2025 13:31:07 -0000
@@ -190,6 +190,7 @@ void
 divert_packet(struct mbuf *m, int dir, u_int16_t divert_port)
 {
 	struct inpcb *inp = NULL;
+	void *pcb;
 	struct socket *so;
 	struct sockaddr_in sin;
 
@@ -213,6 +214,12 @@ divert_packet(struct mbuf *m, int dir, u
 		divstat_inc(divs_noport);
 		goto bad;
 	}
+	pcb = READ_ONCE(inp->inp_socket->so_pcb);
+	if (pcb == NULL) {
+		divstat_inc(divs_closing);
+		goto bad;
+	}
+	KASSERT(pcb == inp);
 
 	memset(&sin, 0, sizeof(sin));
 	sin.sin_family = AF_INET;
Index: netinet/ip_divert.h
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/netinet/ip_divert.h,v
diff -u -p -r1.27 ip_divert.h
--- netinet/ip_divert.h	4 Jun 2025 12:37:00 -0000	1.27
+++ netinet/ip_divert.h	4 Jun 2025 13:31:07 -0000
@@ -22,6 +22,7 @@
 struct divstat {
 	u_long	divs_ipackets;	/* total input packets */
 	u_long	divs_noport;	/* no socket on port */
+	u_long	divs_closing;	/* inpcb exists, socket is closing */
 	u_long	divs_fullsock;	/* not delivered, input socket full */
 	u_long	divs_opackets;	/* total output packets */
 	u_long	divs_errors;	/* generic errors */
@@ -49,6 +50,7 @@ struct divstat {
 enum divstat_counters {
 	divs_ipackets,
 	divs_noport,
+	divs_closing,
 	divs_fullsock,
 	divs_opackets,
 	divs_errors,
Index: netinet/ip_var.h
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/netinet/ip_var.h,v
diff -u -p -r1.121 ip_var.h
--- netinet/ip_var.h	2 Mar 2025 21:28:32 -0000	1.121
+++ netinet/ip_var.h	4 Jun 2025 13:31:07 -0000
@@ -68,6 +68,7 @@ struct	ipstat {
 	u_long	ips_cantforward;	/* packets rcvd for unreachable dest */
 	u_long	ips_redirectsent;	/* packets forwarded on same net */
 	u_long	ips_noproto;		/* unknown or unsupported protocol */
+	u_long	ips_closing;		/* inpcb exists, socket is closing */
 	u_long	ips_delivered;		/* datagrams delivered to upper level*/
 	u_long	ips_localout;		/* total ip packets generated here */
 	u_long	ips_odropped;		/* lost output due to nobufs, etc. */
@@ -116,6 +117,7 @@ enum ipstat_counters {
 	ips_cantforward,	/* packets rcvd for unreachable dest */
 	ips_redirectsent,	/* packets forwarded on same net */
 	ips_noproto,		/* unknown or unsupported protocol */
+	ips_closing,		/* inpcb exists, socket is closing */
 	ips_delivered,		/* datagrams delivered to upper level*/
 	ips_localout,		/* total ip packets generated here */
 	ips_odropped,		/* lost output packets due to nobufs, etc. */
Index: netinet/raw_ip.c
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/netinet/raw_ip.c,v
diff -u -p -r1.166 raw_ip.c
--- netinet/raw_ip.c	11 Mar 2025 15:31:03 -0000	1.166
+++ netinet/raw_ip.c	4 Jun 2025 13:31:07 -0000
@@ -135,6 +135,7 @@ rip_input(struct mbuf **mp, int *offp, i
 	struct ip *ip = mtod(m, struct ip *);
 	struct inpcb_iterator iter = { .inp_table = NULL };
 	struct inpcb *inp, *last;
+	void *pcb;
 	struct in_addr *key;
 	struct sockaddr_in ripsrc;
 
@@ -169,6 +170,12 @@ rip_input(struct mbuf **mp, int *offp, i
 	while ((inp = in_pcb_iterator(&rawcbtable, inp, &iter)) != NULL) {
 		KASSERT(!ISSET(inp->inp_flags, INP_IPV6));
 
+		pcb = READ_ONCE(inp->inp_socket->so_pcb);
+		if (pcb == NULL) {
+			ipstat_inc(ips_closing);
+			continue;
+		}
+		KASSERT(pcb == inp);
 		/*
 		 * Packet must not be inserted after disconnected wakeup
 		 * call.  To avoid race, check again when holding receive
Index: netinet/tcp_input.c
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/netinet/tcp_input.c,v
diff -u -p -r1.450 tcp_input.c
--- netinet/tcp_input.c	3 Jun 2025 16:51:26 -0000	1.450
+++ netinet/tcp_input.c	4 Jun 2025 13:31:07 -0000
@@ -661,10 +661,9 @@ findpcb:
 		so = in_pcbsolock(inp);
 	}
 	if (so == NULL) {
-		tcpstat_inc(tcps_noport);
+		tcpstat_inc(tcps_closing);
 		goto dropwithreset_ratelim;
 	}
-
 	KASSERT(sotoinpcb(inp->inp_socket) == inp);
 	KASSERT(intotcpcb(inp) == NULL || intotcpcb(inp)->t_inpcb == inp);
 	soassertlocked(inp->inp_socket);
Index: netinet/tcp_subr.c
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/netinet/tcp_subr.c,v
diff -u -p -r1.211 tcp_subr.c
--- netinet/tcp_subr.c	3 Jun 2025 16:51:26 -0000	1.211
+++ netinet/tcp_subr.c	4 Jun 2025 14:18:03 -0000
@@ -440,8 +440,6 @@ tcp_newtcpcb(struct inpcb *inp, int wait
 	tp->t_inpcb = inp;
 	for (i = 0; i < TCPT_NTIMERS; i++)
 		TCP_TIMER_INIT(tp, i);
-	timeout_set_flags(&tp->t_timer_reaper, tcp_timer_reaper, tp,
-	    KCLOCK_NONE, TIMEOUT_PROC | TIMEOUT_MPSAFE);
 
 	tp->sack_enable = atomic_load_int(&tcp_do_sack);
 	tp->t_flags = atomic_load_int(&tcp_do_rfc1323) ?
@@ -528,9 +526,8 @@ tcp_close(struct tcpcb *tp)
 	}
 
 	m_free(tp->t_template);
-	/* Free tcpcb after all pending timers have been run. */
-	timeout_add(&tp->t_timer_reaper, 0);
 	inp->inp_ppcb = NULL;
+	pool_put(&tcpcb_pool, tp);
 	soisdisconnected(so);
 	in_pcbdetach(inp);
 	tcpstat_inc(tcps_closed);
Index: netinet/tcp_timer.c
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/netinet/tcp_timer.c,v
diff -u -p -r1.84 tcp_timer.c
--- netinet/tcp_timer.c	3 Jun 2025 16:51:26 -0000	1.84
+++ netinet/tcp_timer.c	4 Jun 2025 14:18:03 -0000
@@ -541,19 +541,3 @@ tcp_timer_2msl(void *arg)
  out:
 	tcp_timer_leave(inp, so);
 }
-
-void
-tcp_timer_reaper(void *arg)
-{
-	struct tcpcb *tp = arg;
-
-	/*
-	 * This timer is necessary to delay the pool_put() after all timers
-	 * have finished, even if they were sleeping to grab the net lock.
-	 * Putting the pool_put() in a timer is sufficient as all timers run
-	 * from the same timeout thread.  Note that neither softnet thread nor
-	 * user process may access the tcpcb after arming the reaper timer.
-	 * Freeing may run in parallel as it does not grab the net lock.
-	 */
-	pool_put(&tcpcb_pool, tp);
-}
Index: netinet/tcp_timer.h
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/netinet/tcp_timer.h,v
diff -u -p -r1.26 tcp_timer.h
--- netinet/tcp_timer.h	16 Jan 2025 11:59:20 -0000	1.26
+++ netinet/tcp_timer.h	4 Jun 2025 14:18:03 -0000
@@ -164,8 +164,5 @@ extern int tcp_keepidle_sec;	/* [a] copy
 extern int tcp_keepintvl_sec;	/* [a] copy of above in seconds for sysctl */
 extern int tcp_ttl;		/* time to live for TCP segs */
 extern const int tcp_backoff[];
-
-void	tcp_timer_reaper(void *);
-
 #endif /* _KERNEL */
 #endif /* _NETINET_TCP_TIMER_H_ */
Index: netinet/tcp_var.h
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/netinet/tcp_var.h,v
diff -u -p -r1.191 tcp_var.h
--- netinet/tcp_var.h	7 May 2025 14:10:19 -0000	1.191
+++ netinet/tcp_var.h	4 Jun 2025 14:18:03 -0000
@@ -70,7 +70,6 @@ struct tcpqent {
 struct tcpcb {
 	struct tcpqehead t_segq;		/* sequencing queue */
 	struct timeout t_timer[TCPT_NTIMERS];	/* tcp timers */
-	struct timeout t_timer_reaper;	/* reaper is special, no refcnt */
 	short	t_state;		/* state of this connection */
 	short	t_rxtshift;		/* log(2) of rexmt exp. backoff */
 	int	t_rxtcur;		/* current retransmit value */
@@ -393,6 +392,7 @@ struct	tcpstat {
 
 	u_int32_t tcps_pcbhashmiss;	/* input packets missing pcb hash */
 	u_int32_t tcps_noport;		/* no socket on port */
+	u_int32_t tcps_closing;		/* inpcb exists, socket is closing */
 	u_int32_t tcps_badsyn;		/* SYN packet with src==dst rcv'ed */
 	u_int32_t tcps_dropsyn;		/* SYN packet dropped */
 
@@ -583,6 +583,7 @@ enum tcpstat_counters {
 	tcps_preddat,
 	tcps_pcbhashmiss,
 	tcps_noport,
+	tcps_closing,
 	tcps_badsyn,
 	tcps_dropsyn,
 	tcps_rcvbadsig,
Index: netinet/udp_usrreq.c
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/netinet/udp_usrreq.c,v
diff -u -p -r1.341 udp_usrreq.c
--- netinet/udp_usrreq.c	3 Jun 2025 16:51:26 -0000	1.341
+++ netinet/udp_usrreq.c	4 Jun 2025 13:31:07 -0000
@@ -198,6 +198,7 @@ udp_input(struct mbuf **mp, int *offp, i
 	struct ip *ip = NULL;
 	struct udphdr *uh;
 	struct inpcb *inp = NULL;
+	void *pcb;
 	struct ip save_ip;
 	int len;
 	u_int16_t savesum;
@@ -419,6 +420,12 @@ udp_input(struct mbuf **mp, int *offp, i
 			else
 				KASSERT(!ISSET(inp->inp_flags, INP_IPV6));
 
+			pcb = READ_ONCE(inp->inp_socket->so_pcb);
+			if (pcb == NULL) {
+				udpstat_inc(udps_closing);
+				continue;
+			}
+			KASSERT(pcb == inp);
 			if (inp->inp_socket->so_rcv.sb_state & SS_CANTRCVMORE)
 				continue;
 			if (rtable_l2(inp->inp_rtableid) !=
@@ -596,7 +603,12 @@ udp_input(struct mbuf **mp, int *offp, i
 		return IPPROTO_DONE;
 	}
 
-	KASSERT(sotoinpcb(inp->inp_socket) == inp);
+	pcb = READ_ONCE(inp->inp_socket->so_pcb);
+	if (pcb == NULL) {
+		udpstat_inc(udps_closing);
+		goto bad;
+	}
+	KASSERT(pcb == inp);
 	soassertlocked_readonly(inp->inp_socket);
 
 #ifdef INET6
Index: netinet/udp_var.h
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/netinet/udp_var.h,v
diff -u -p -r1.53 udp_var.h
--- netinet/udp_var.h	2 Mar 2025 21:28:32 -0000	1.53
+++ netinet/udp_var.h	4 Jun 2025 13:31:07 -0000
@@ -61,6 +61,7 @@ struct	udpstat {
 	u_long	udps_badlen;		/* data length larger than packet */
 	u_long	udps_noport;		/* no socket on port */
 	u_long	udps_noportbcast;	/* of above, arrived as broadcast */
+	u_long	udps_closing;		/* inpcb exists, socket is closing */
 	u_long	udps_nosec;		/* dropped for lack of ipsec */
 	u_long	udps_fullsock;		/* not delivered, input socket full */
 	u_long	udps_pcbhashmiss;	/* input packets missing pcb hash */
@@ -104,6 +105,7 @@ enum udpstat_counters {
 	udps_badlen,		/* data length larger than packet */
 	udps_noport,		/* no socket on port */
 	udps_noportbcast,	/* of above, arrived as broadcast */
+	udps_closing,		/* inpcb exists, socket is closing */
 	udps_nosec,		/* dropped for lack of ipsec */
 	udps_fullsock,		/* not delivered, input socket full */
 	udps_pcbhashmiss,	/* input packets missing pcb hash */
Index: netinet6/ip6_divert.c
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/netinet6/ip6_divert.c,v
diff -u -p -r1.103 ip6_divert.c
--- netinet6/ip6_divert.c	4 Jun 2025 12:37:00 -0000	1.103
+++ netinet6/ip6_divert.c	4 Jun 2025 13:31:07 -0000
@@ -199,6 +199,7 @@ void
 divert6_packet(struct mbuf *m, int dir, u_int16_t divert_port)
 {
 	struct inpcb *inp = NULL;
+	void *pcb;
 	struct socket *so;
 	struct sockaddr_in6 sin6;
 
@@ -222,6 +223,12 @@ divert6_packet(struct mbuf *m, int dir, 
 		div6stat_inc(divs_noport);
 		goto bad;
 	}
+	pcb = READ_ONCE(inp->inp_socket->so_pcb);
+	if (pcb == NULL) {
+		div6stat_inc(divs_closing);
+		goto bad;
+	}
+	KASSERT(pcb == inp);
 
 	memset(&sin6, 0, sizeof(sin6));
 	sin6.sin6_family = AF_INET6;
Index: netinet6/raw_ip6.c
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/netinet6/raw_ip6.c,v
diff -u -p -r1.192 raw_ip6.c
--- netinet6/raw_ip6.c	27 May 2025 07:52:49 -0000	1.192
+++ netinet6/raw_ip6.c	4 Jun 2025 13:31:07 -0000
@@ -138,6 +138,7 @@ rip6_input(struct mbuf **mp, int *offp, 
 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
 	struct inpcb_iterator iter = { .inp_table = NULL };
 	struct inpcb *inp, *last;
+	void *pcb;
 	struct in6_addr *key;
 	struct sockaddr_in6 rip6src;
 	uint8_t type;
@@ -184,6 +185,12 @@ rip6_input(struct mbuf **mp, int *offp, 
 	while ((inp = in_pcb_iterator(&rawin6pcbtable, inp, &iter)) != NULL) {
 		KASSERT(ISSET(inp->inp_flags, INP_IPV6));
 
+		pcb = READ_ONCE(inp->inp_socket->so_pcb);
+		if (pcb == NULL) {
+			rip6stat_inc(rip6s_closing);
+			continue;
+		}
+		KASSERT(pcb == inp);
 		/*
 		 * Packet must not be inserted after disconnected wakeup
 		 * call.  To avoid race, check again when holding receive
Index: netinet6/raw_ip6.h
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/netinet6/raw_ip6.h,v
diff -u -p -r1.4 raw_ip6.h
--- netinet6/raw_ip6.h	9 Feb 2017 15:23:35 -0000	1.4
+++ netinet6/raw_ip6.h	4 Jun 2025 13:31:07 -0000
@@ -42,6 +42,7 @@ struct rip6stat {
 	u_int64_t rip6s_badsum;		/* of above, checksum error */
 	u_int64_t rip6s_nosock;		/* no matching socket */
 	u_int64_t rip6s_nosockmcast;	/* of above, arrived as multicast */
+	u_int64_t rip6s_closing;	/* inpcb exists, socket is closing */
 	u_int64_t rip6s_fullsock;	/* not delivered, input socket full */
 
 	u_int64_t rip6s_opackets;	/* total output packets */
@@ -68,6 +69,7 @@ enum rip6stat_counters {
 	rip6s_badsum,
 	rip6s_nosock,
 	rip6s_nosockmcast,
+	rip6s_closing,
 	rip6s_fullsock,
 	rip6s_opackets,
 	rip6s_ncounters,