Merge branch 'bind_addr_zero'

Kuniyuki Iwashima says:

====================
Improve bind(addr, 0) behaviour.

Currently we fail to bind sockets to ephemeral ports when all of the ports
are exhausted even if all sockets have SO_REUSEADDR enabled. In this case,
we still have a chance to connect to the different remote hosts.

These patches add net.ipv4.ip_autobind_reuse option and fix the behaviour
to fully utilize all space of the local (addr, port) tuples.

Changes in v5:
  - Add more description to documents.
  - Fix sysctl option to use proc_dointvec_minmax.
  - Remove the Fixes: tag and squash two commits.

Changes in v4:
  - Add net.ipv4.ip_autobind_reuse option to not change the current behaviour.
  - Modify .gitignore for test.
  https://lore.kernel.org/netdev/[email protected]/

Changes in v3:
  - Change the title and write more specific description of the 3rd patch.
  - Add a test in tools/testing/selftests/net/ as the 4th patch.
  https://lore.kernel.org/netdev/[email protected]/

Changes in v2:
  - Change the description of the 2nd patch ('localhost' -> 'address').
  - Correct the description and the if statement of the 3rd patch.
  https://lore.kernel.org/netdev/[email protected]/

v1 with tests:
  https://lore.kernel.org/netdev/[email protected]/
====================

Reviewed-by: Eric Dumazet <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

8 files changed, 243 insertions, 12 deletions

diff --git a/Documentation/networking/ip-sysctl.txt b/Documentation/networking/ip-sysctl.txt
index 5f53faff4e25..ee961d322d93 100644
--- a/Documentation/networking/ip-sysctl.txt
+++ b/Documentation/networking/ip-sysctl.txt
@@ -958,6 +958,15 @@ ip_nonlocal_bind - BOOLEAN
 	which can be quite useful - but may break some applications.
 	Default: 0
 
+ip_autobind_reuse - BOOLEAN
+	By default, bind() does not select the ports automatically even if
+	the new socket and all sockets bound to the port have SO_REUSEADDR.
+	ip_autobind_reuse allows bind() to reuse the port and this is useful
+	when you use bind()+connect(), but may break some applications.
+	The preferred solution is to use IP_BIND_ADDRESS_NO_PORT and this
+	option should only be set by experts.
+	Default: 0
+
 ip_dynaddr - BOOLEAN
 	If set non-zero, enables support for dynamic addresses.
 	If set to a non-zero value larger than 1, a kernel log
diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h
index 08b98414d94e..154b8f01499b 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -101,6 +101,7 @@ struct netns_ipv4 {
 	int sysctl_ip_fwd_use_pmtu;
 	int sysctl_ip_fwd_update_priority;
 	int sysctl_ip_nonlocal_bind;
+	int sysctl_ip_autobind_reuse;
 	/* Shall we try to damage output packets if routing dev changes? */
 	int sysctl_ip_dynaddr;
 	int sysctl_ip_early_demux;
diff --git a/net/ipv4/inet_connection_sock.c b/net/ipv4/inet_connection_sock.c
index a4db79b1b643..3b4f81790e3e 100644
--- a/net/ipv4/inet_connection_sock.c
+++ b/net/ipv4/inet_connection_sock.c
@@ -131,7 +131,7 @@ static int inet_csk_bind_conflict(const struct sock *sk,
 {
 	struct sock *sk2;
 	bool reuse = sk->sk_reuse;
-	bool reuseport = !!sk->sk_reuseport && reuseport_ok;
+	bool reuseport = !!sk->sk_reuseport;
 	kuid_t uid = sock_i_uid((struct sock *)sk);
 
 	/*
@@ -146,17 +146,21 @@ static int inet_csk_bind_conflict(const struct sock *sk,
 		    (!sk->sk_bound_dev_if ||
 		     !sk2->sk_bound_dev_if ||
 		     sk->sk_bound_dev_if == sk2->sk_bound_dev_if)) {
-			if ((!reuse || !sk2->sk_reuse ||
-			    sk2->sk_state == TCP_LISTEN) &&
-			    (!reuseport || !sk2->sk_reuseport ||
-			     rcu_access_pointer(sk->sk_reuseport_cb) ||
-			     (sk2->sk_state != TCP_TIME_WAIT &&
-			     !uid_eq(uid, sock_i_uid(sk2))))) {
-				if (inet_rcv_saddr_equal(sk, sk2, true))
-					break;
-			}
-			if (!relax && reuse && sk2->sk_reuse &&
+			if (reuse && sk2->sk_reuse &&
 			    sk2->sk_state != TCP_LISTEN) {
+				if ((!relax ||
+				     (!reuseport_ok &&
+				      reuseport && sk2->sk_reuseport &&
+				      !rcu_access_pointer(sk->sk_reuseport_cb) &&
+				      (sk2->sk_state == TCP_TIME_WAIT ||
+				       uid_eq(uid, sock_i_uid(sk2))))) &&
+				    inet_rcv_saddr_equal(sk, sk2, true))
+					break;
+			} else if (!reuseport_ok ||
+				   !reuseport || !sk2->sk_reuseport ||
+				   rcu_access_pointer(sk->sk_reuseport_cb) ||
+				   (sk2->sk_state != TCP_TIME_WAIT &&
+				    !uid_eq(uid, sock_i_uid(sk2)))) {
 				if (inet_rcv_saddr_equal(sk, sk2, true))
 					break;
 			}
@@ -176,12 +180,14 @@ inet_csk_find_open_port(struct sock *sk, struct inet_bind_bucket **tb_ret, int *
 	int port = 0;
 	struct inet_bind_hashbucket *head;
 	struct net *net = sock_net(sk);
+	bool relax = false;
 	int i, low, high, attempt_half;
 	struct inet_bind_bucket *tb;
 	u32 remaining, offset;
 	int l3mdev;
 
 	l3mdev = inet_sk_bound_l3mdev(sk);
+ports_exhausted:
 	attempt_half = (sk->sk_reuse == SK_CAN_REUSE) ? 1 : 0;
 other_half_scan:
 	inet_get_local_port_range(net, &low, &high);
@@ -219,7 +225,7 @@ other_parity_scan:
 		inet_bind_bucket_for_each(tb, &head->chain)
 			if (net_eq(ib_net(tb), net) && tb->l3mdev == l3mdev &&
 			    tb->port == port) {
-				if (!inet_csk_bind_conflict(sk, tb, false, false))
+				if (!inet_csk_bind_conflict(sk, tb, relax, false))
 					goto success;
 				goto next_port;
 			}
@@ -239,6 +245,12 @@ next_port:
 		attempt_half = 2;
 		goto other_half_scan;
 	}
+
+	if (net->ipv4.sysctl_ip_autobind_reuse && !relax) {
+		/* We still have a chance to connect to different destinations */
+		relax = true;
+		goto ports_exhausted;
+	}
 	return NULL;
 success:
 	*port_ret = port;
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index d9531b4b33f2..81b267e990a1 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -764,6 +764,15 @@ static struct ctl_table ipv4_net_table[] = {
 		.proc_handler	= proc_dointvec
 	},
 	{
+		.procname	= "ip_autobind_reuse",
+		.data		= &init_net.ipv4.sysctl_ip_autobind_reuse,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1         = SYSCTL_ZERO,
+		.extra2         = SYSCTL_ONE,
+	},
+	{
 		.procname	= "fwmark_reflect",
 		.data		= &init_net.ipv4.sysctl_fwmark_reflect,
 		.maxlen		= sizeof(int),
diff --git a/tools/testing/selftests/net/.gitignore b/tools/testing/selftests/net/.gitignore
index ecc52d4c034d..91f9aea853b1 100644
--- a/tools/testing/selftests/net/.gitignore
+++ b/tools/testing/selftests/net/.gitignore
@@ -23,3 +23,4 @@ so_txtime
 tcp_fastopen_backup_key
 nettest
 fin_ack_lat
+reuseaddr_ports_exhausted
\ No newline at end of file
diff --git a/tools/testing/selftests/net/Makefile b/tools/testing/selftests/net/Makefile
index 287ae916ec0b..48063fd69924 100644
--- a/tools/testing/selftests/net/Makefile
+++ b/tools/testing/selftests/net/Makefile
@@ -12,6 +12,7 @@ TEST_PROGS += udpgro_bench.sh udpgro.sh test_vxlan_under_vrf.sh reuseport_addr_a
 TEST_PROGS += test_vxlan_fdb_changelink.sh so_txtime.sh ipv6_flowlabel.sh
 TEST_PROGS += tcp_fastopen_backup_key.sh fcnal-test.sh l2tp.sh traceroute.sh
 TEST_PROGS += fin_ack_lat.sh
+TEST_PROGS += reuseaddr_ports_exhausted.sh
 TEST_PROGS_EXTENDED := in_netns.sh
 TEST_GEN_FILES =  socket nettest
 TEST_GEN_FILES += psock_fanout psock_tpacket msg_zerocopy reuseport_addr_any
@@ -22,6 +23,7 @@ TEST_GEN_FILES += tcp_fastopen_backup_key
 TEST_GEN_FILES += fin_ack_lat
 TEST_GEN_PROGS = reuseport_bpf reuseport_bpf_cpu reuseport_bpf_numa
 TEST_GEN_PROGS += reuseport_dualstack reuseaddr_conflict tls
+TEST_GEN_FILES += reuseaddr_ports_exhausted
 
 KSFT_KHDR_INSTALL := 1
 include ../lib.mk
diff --git a/tools/testing/selftests/net/reuseaddr_ports_exhausted.c b/tools/testing/selftests/net/reuseaddr_ports_exhausted.c
new file mode 100644
index 000000000000..7b01b7c2ec10
--- /dev/null
+++ b/tools/testing/selftests/net/reuseaddr_ports_exhausted.c
@@ -0,0 +1,162 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Check if we can fully utilize 4-tuples for connect().
+ *
+ * Rules to bind sockets to the same port when all ephemeral ports are
+ * exhausted.
+ *
+ *   1. if there are TCP_LISTEN sockets on the port, fail to bind.
+ *   2. if there are sockets without SO_REUSEADDR, fail to bind.
+ *   3. if SO_REUSEADDR is disabled, fail to bind.
+ *   4. if SO_REUSEADDR is enabled and SO_REUSEPORT is disabled,
+ *        succeed to bind.
+ *   5. if SO_REUSEADDR and SO_REUSEPORT are enabled and
+ *        there is no socket having the both options and the same EUID,
+ *        succeed to bind.
+ *   6. fail to bind.
+ *
+ * Author: Kuniyuki Iwashima <[email protected]>
+ */
+#include <arpa/inet.h>
+#include <netinet/in.h>
+#include <sys/socket.h>
+#include <sys/types.h>
+#include <unistd.h>
+#include "../kselftest_harness.h"
+
+struct reuse_opts {
+	int reuseaddr[2];
+	int reuseport[2];
+};
+
+struct reuse_opts unreusable_opts[12] = {
+	{0, 0, 0, 0},
+	{0, 0, 0, 1},
+	{0, 0, 1, 0},
+	{0, 0, 1, 1},
+	{0, 1, 0, 0},
+	{0, 1, 0, 1},
+	{0, 1, 1, 0},
+	{0, 1, 1, 1},
+	{1, 0, 0, 0},
+	{1, 0, 0, 1},
+	{1, 0, 1, 0},
+	{1, 0, 1, 1},
+};
+
+struct reuse_opts reusable_opts[4] = {
+	{1, 1, 0, 0},
+	{1, 1, 0, 1},
+	{1, 1, 1, 0},
+	{1, 1, 1, 1},
+};
+
+int bind_port(struct __test_metadata *_metadata, int reuseaddr, int reuseport)
+{
+	struct sockaddr_in local_addr;
+	int len = sizeof(local_addr);
+	int fd, ret;
+
+	fd = socket(AF_INET, SOCK_STREAM, 0);
+	ASSERT_NE(-1, fd) TH_LOG("failed to open socket.");
+
+	ret = setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, &reuseaddr, sizeof(int));
+	ASSERT_EQ(0, ret) TH_LOG("failed to setsockopt: SO_REUSEADDR.");
+
+	ret = setsockopt(fd, SOL_SOCKET, SO_REUSEPORT, &reuseport, sizeof(int));
+	ASSERT_EQ(0, ret) TH_LOG("failed to setsockopt: SO_REUSEPORT.");
+
+	local_addr.sin_family = AF_INET;
+	local_addr.sin_addr.s_addr = inet_addr("127.0.0.1");
+	local_addr.sin_port = 0;
+
+	if (bind(fd, (struct sockaddr *)&local_addr, len) == -1) {
+		close(fd);
+		return -1;
+	}
+
+	return fd;
+}
+
+TEST(reuseaddr_ports_exhausted_unreusable)
+{
+	struct reuse_opts *opts;
+	int i, j, fd[2];
+
+	for (i = 0; i < 12; i++) {
+		opts = &unreusable_opts[i];
+
+		for (j = 0; j < 2; j++)
+			fd[j] = bind_port(_metadata, opts->reuseaddr[j], opts->reuseport[j]);
+
+		ASSERT_NE(-1, fd[0]) TH_LOG("failed to bind.");
+		EXPECT_EQ(-1, fd[1]) TH_LOG("should fail to bind.");
+
+		for (j = 0; j < 2; j++)
+			if (fd[j] != -1)
+				close(fd[j]);
+	}
+}
+
+TEST(reuseaddr_ports_exhausted_reusable_same_euid)
+{
+	struct reuse_opts *opts;
+	int i, j, fd[2];
+
+	for (i = 0; i < 4; i++) {
+		opts = &reusable_opts[i];
+
+		for (j = 0; j < 2; j++)
+			fd[j] = bind_port(_metadata, opts->reuseaddr[j], opts->reuseport[j]);
+
+		ASSERT_NE(-1, fd[0]) TH_LOG("failed to bind.");
+
+		if (opts->reuseport[0] && opts->reuseport[1]) {
+			EXPECT_EQ(-1, fd[1]) TH_LOG("should fail to bind because both sockets succeed to be listened.");
+		} else {
+			EXPECT_NE(-1, fd[1]) TH_LOG("should succeed to bind to connect to different destinations.");
+		}
+
+		for (j = 0; j < 2; j++)
+			if (fd[j] != -1)
+				close(fd[j]);
+	}
+}
+
+TEST(reuseaddr_ports_exhausted_reusable_different_euid)
+{
+	struct reuse_opts *opts;
+	int i, j, ret, fd[2];
+	uid_t euid[2] = {10, 20};
+
+	for (i = 0; i < 4; i++) {
+		opts = &reusable_opts[i];
+
+		for (j = 0; j < 2; j++) {
+			ret = seteuid(euid[j]);
+			ASSERT_EQ(0, ret) TH_LOG("failed to seteuid: %d.", euid[j]);
+
+			fd[j] = bind_port(_metadata, opts->reuseaddr[j], opts->reuseport[j]);
+
+			ret = seteuid(0);
+			ASSERT_EQ(0, ret) TH_LOG("failed to seteuid: 0.");
+		}
+
+		ASSERT_NE(-1, fd[0]) TH_LOG("failed to bind.");
+		EXPECT_NE(-1, fd[1]) TH_LOG("should succeed to bind because one socket can be bound in each euid.");
+
+		if (fd[1] != -1) {
+			ret = listen(fd[0], 5);
+			ASSERT_EQ(0, ret) TH_LOG("failed to listen.");
+
+			ret = listen(fd[1], 5);
+			EXPECT_EQ(-1, ret) TH_LOG("should fail to listen because only one uid reserves the port in TCP_LISTEN.");
+		}
+
+		for (j = 0; j < 2; j++)
+			if (fd[j] != -1)
+				close(fd[j]);
+	}
+}
+
+TEST_HARNESS_MAIN
diff --git a/tools/testing/selftests/net/reuseaddr_ports_exhausted.sh b/tools/testing/selftests/net/reuseaddr_ports_exhausted.sh
new file mode 100755
index 000000000000..20e3a2913d06
--- /dev/null
+++ b/tools/testing/selftests/net/reuseaddr_ports_exhausted.sh
@@ -0,0 +1,35 @@
+#!/bin/bash
+# SPDX-License-Identifier: GPL-2.0
+#
+# Run tests when all ephemeral ports are exhausted.
+#
+# Author: Kuniyuki Iwashima <[email protected]>
+
+set +x
+set -e
+
+readonly NETNS="ns-$(mktemp -u XXXXXX)"
+
+setup() {
+	ip netns add "${NETNS}"
+	ip -netns "${NETNS}" link set lo up
+	ip netns exec "${NETNS}" \
+		sysctl -w net.ipv4.ip_local_port_range="32768 32768" \
+		> /dev/null 2>&1
+	ip netns exec "${NETNS}" \
+		sysctl -w net.ipv4.ip_autobind_reuse=1 > /dev/null 2>&1
+}
+
+cleanup() {
+	ip netns del "${NETNS}"
+}
+
+trap cleanup EXIT
+setup
+
+do_test() {
+	ip netns exec "${NETNS}" ./reuseaddr_ports_exhausted
+}
+
+do_test
+echo "tests done"