add support for optional tunables

CONTRIBUTING.md

@@ -169,6 +169,7 @@ struct bpftunable_desc {
         unsigned int id;
         enum bpftunable_type type;
         const char *name;
+        __u64 flags;
         __u8 num_values;
 };
 ```

include/bpftune/bpftune.h

@@ -85,11 +85,14 @@ struct bpftunable_scenario {
 
 #define BPFTUNE_MAX_VALUES	3
 
+#define BPFTUNABLE_NAMESPACED	0x1	/* settable in non-global namespace? */
+#define BPFTUNABLE_OPTIONAL	0x2	/* do not fail it tunable not found (e.g. ipv6 */
+
 struct bpftunable_desc {
 	unsigned int id;
 	enum bpftunable_type type;
 	const char *name;
-	bool namespaced;	/* settable in non-global namespace? */
+	__u64 flags;
 	__u8 num_values;
 };
 

src/libbpftune.c

@@ -940,6 +940,11 @@ int bpftuner_tunables_init(struct bpftuner *tuner, unsigned int num_descs,
 		num_values = bpftune_sysctl_read(0, descs[i].name,
 				tuner->tunables[i].current_values);
 		if (num_values < 0) {
+			if (descs[i].flags & BPFTUNABLE_OPTIONAL) {
+				bpftune_log(LOG_DEBUG, "error reading optional tunable '%s': %s\n",
+					    descs[i].name, strerror(-num_values));
+				continue;
+			}
 			bpftune_log(LOG_ERR, "error reading tunable '%s': %s\n",
 				    descs[i].name, strerror(-num_values));
 			return num_values;
@@ -1052,7 +1057,7 @@ int bpftuner_tunable_sysctl_write(struct bpftuner *tuner, unsigned int tunable,
 		}
 	}
 
-	if (t->desc.namespaced) {
+	if (t->desc.flags & BPFTUNABLE_NAMESPACED) {
 		fd = bpftuner_netns_fd_from_cookie(tuner, netns_cookie);
 		if (fd < 0) {
 			bpftune_log(LOG_DEBUG, "could not get netns fd for cookie %ld\n",

src/neigh_table_tuner.c

@@ -29,25 +29,30 @@ struct neigh_table_tuner_bpf *skel;
 
 static struct bpftunable_desc descs[] = {
 { NEIGH_TABLE_IPV4_GC_INTERVAL,		BPFTUNABLE_SYSCTL,
-  		"net.ipv4.neigh.default.gc_interval",	false,	1 },
+  		"net.ipv4.neigh.default.gc_interval",	0,	1 },
 { NEIGH_TABLE_IPV4_GC_STALE_TIME,	BPFTUNABLE_SYSCTL,
-		"net.ipv4.neigh.default.gc_stale_time",	true,	1, },
+		"net.ipv4.neigh.default.gc_stale_time",	0,	1, },
 { NEIGH_TABLE_IPV4_GC_THRESH1,		BPFTUNABLE_SYSCTL,
-		"net.ipv4.neigh.default.gc_thresh1",	false,	1, },
+		"net.ipv4.neigh.default.gc_thresh1",	0,	1, },
 { NEIGH_TABLE_IPV4_GC_THRESH2,		BPFTUNABLE_SYSCTL,
-		"net.ipv4.neigh.default.gc_thresh2",	false,	1, },
+		"net.ipv4.neigh.default.gc_thresh2",	0,	1, },
 { NEIGH_TABLE_IPV4_GC_THRESH3,		BPFTUNABLE_SYSCTL,
-		"net.ipv4.neigh.default.gc_thresh3",	false,	1, },
+		"net.ipv4.neigh.default.gc_thresh3",	0,	1, },
 { NEIGH_TABLE_IPV6_GC_INTERVAL,		BPFTUNABLE_SYSCTL,
-		"net.ipv6.neigh.default.gc_interval",   false,	1 },
+		"net.ipv6.neigh.default.gc_interval",   BPFTUNABLE_OPTIONAL,
+								1 },
 { NEIGH_TABLE_IPV6_GC_STALE_TIME,	BPFTUNABLE_SYSCTL,
-		"net.ipv6.neigh.default.gc_stale_time", true, 1, },
+		"net.ipv6.neigh.default.gc_stale_time", BPFTUNABLE_OPTIONAL,
+								1, },
 { NEIGH_TABLE_IPV6_GC_THRESH1,		BPFTUNABLE_SYSCTL,
-		"net.ipv6.neigh.default.gc_thresh1",    false, 1, },
+		"net.ipv6.neigh.default.gc_thresh1",    BPFTUNABLE_OPTIONAL,
+								1, },
 { NEIGH_TABLE_IPV6_GC_THRESH2,		BPFTUNABLE_SYSCTL,
-		"net.ipv6.neigh.default.gc_thresh2",    false, 1, },
+		"net.ipv6.neigh.default.gc_thresh2",    BPFTUNABLE_OPTIONAL,
+								1, },
 { NEIGH_TABLE_IPV6_GC_THRESH3,		BPFTUNABLE_SYSCTL,
-		"net.ipv6.neigh.default.gc_thresh3",    false, 1, },
+		"net.ipv6.neigh.default.gc_thresh3",    BPFTUNABLE_OPTIONAL,
+								1, },
 };
 
 static struct bpftunable_scenario scenarios[] = {

src/net_buffer_tuner.c

@@ -12,10 +12,10 @@ struct tcp_buffer_tuner_bpf *skel;
 
 static struct bpftunable_desc descs[] = {
 { NETDEV_MAX_BACKLOG,	BPFTUNABLE_SYSCTL, "net.core.netdev_max_backlog",
-								false, 1 },
+								0, 1 },
 { FLOW_LIMIT_CPU_BITMAP,
 			BPFTUNABLE_SYSCTL, "net.core.flow_limit_cpu_bitmap",
-								false, 1 },
+								0, 1 },
 };
 
 static struct bpftunable_scenario scenarios[] = {

src/netns_tuner.c

@@ -28,7 +28,7 @@ struct netns_tuner_bpf *skel;
 
 static struct bpftunable_desc descs[] = {
 {
- NETNS, BPFTUNABLE_OTHER, "Network namespace", true, 0 },
+ NETNS, BPFTUNABLE_OTHER, "Network namespace", BPFTUNABLE_NAMESPACED, 0 },
 };
 
 static struct bpftunable_scenario scenarios[] = {

src/route_table_tuner.c

@@ -27,7 +27,8 @@ struct route_table_tuner_bpf *skel;
 
 static struct bpftunable_desc descs[] = {
 { ROUTE_TABLE_IPV6_MAX_SIZE,		BPFTUNABLE_SYSCTL,
-		"net.ipv6.route.max_size",		true, 1 },
+		"net.ipv6.route.max_size",
+		BPFTUNABLE_NAMESPACED | BPFTUNABLE_OPTIONAL, 1 },
 };
 
 static struct bpftunable_scenario scenarios[] = {

src/tcp_cong_tuner.c

@@ -31,7 +31,7 @@
 
 static struct bpftunable_desc descs[] = {
 { 
- TCP_CONG, BPFTUNABLE_OTHER, "TCP congestion control", false, 0 },
+ TCP_CONG, BPFTUNABLE_OTHER, "TCP congestion control", 0, 0 },
 };
 
 static struct bpftunable_scenario scenarios[] = {

test/Makefile

@@ -24,7 +24,8 @@ TUNER_TESTS =	support_test log_test service_test inotify_test cap_test \
 		sysctl_test sysctl_legacy_test sysctl_netns_test \
 		netns_test netns_legacy_test \
 		backlog_test backlog_legacy_test \
-		neigh_table_test neigh_table_legacy_test \
+		neigh_table_test neigh_table_v4only_test \
+		neigh_table_legacy_test \
 		mem_pressure_test mem_pressure_legacy_test \
 		mem_exhaust_test mem_exhaust_legacy_test \
 		wmem_test wmem_legacy_test \

test/neigh_table_v4only_test.sh

@@ -0,0 +1,73 @@
+#!/usr/bin/bash
+#
+# SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note
+#
+# Copyright (c) 2023, Oracle and/or its affiliates.
+#
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU General Public
+# License v2 as published by the Free Software Foundation.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+# General Public License for more details.
+#
+# You should have received a copy of the GNU General Public
+# License along with this program; if not, write to the
+# Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+# Boston, MA 021110-1307, USA.
+#
+
+# run neigh table test when ipv6 is disabled
+
+. ./test_lib.sh
+
+LOGFILE=$TESTLOG_LAST
+
+SLEEPTIME=1
+
+for TUNER in neigh_table ; do
+
+ # gc_thresh3 is not namespaced...
+
+ for NS in global ; do
+  for TBL in arp_cache ; do
+
+   test_start "$0|neigh table test ($NS netns): does filling $TBL make it grow?"
+
+   test_setup "true"
+
+   sysctl -w net.ipv6.conf.all.disable_ipv6=1
+   if [[ $NS != "global" ]]; then
+     ip netns exec $NS sysctl -w net.ipv6.conf.all.disable_ipv6=1
+   fi
+   test_run_cmd_local "$BPFTUNE -s &" true
+
+   sleep $SETUPTIME
+
+   if [[ $NS != "global" ]]; then
+	PREFIX_CMD="ip netns exec $NETNS "
+        INTF=$VETH1
+   else
+	PREFIX_CMD=""
+	INTF=$VETH2
+   fi	
+   $PREFIX_CMD ip ntable change name $TBL dev $INTF thresh3 128
+
+   for ((i=3; i < 255; i++ ))
+   do
+      ipaddr="192.168.168.${i}"
+      ih=$(printf '%x' $i)
+      macaddr="de:ad:be:ef:de:${ih}"
+      $PREFIX_CMD ip neigh replace $ipaddr lladdr $macaddr dev $INTF
+   done
+   grep "updated gc_thresh3 for $TBL table" $LOGFILE
+   test_pass
+  done
+ done
+done
+
+test_cleanup
+
+test_exit

test/test_lib.sh

@@ -210,6 +210,7 @@ test_setup_local()
 	ip netns list 2>/dev/null | grep $NETNS
 	FOUND=$?
 	set -e
+	sysctl -qw net.ipv6.conf.all.disable_ipv6=0
 	if [[ $FOUND -ne 0 ]]; then
 		ip netns pids $NETNS 2>/dev/null| xargs -r kill
 		ip netns del $NETNS 2>/dev/null|true
@@ -278,6 +279,7 @@ test_cleanup_local()
 	ip --all netns del ${NETNS_PREFIX}\*
 	ip link del $VETH2 2>/dev/null
 	ip link del bpftunelocal 2>/dev/null
+	sysctl -w net.ipv6.conf.all.disable_ipv6=0
 	set -e
 
 	if [[ ! -f /usr/lib64/bpftune/tcp_buffer_tuner.so ]]; then