Merge branch 'net-remove-rtnl-from-fib_seq_sum'

kuba-moo · kuba-moo · commit 356d568fa9c0 · 2024-10-11T15:35:16.000-07:00
Eric Dumazet says: ==================== net: remove RTNL from fib_seq_sum() This series is inspired by a syzbot report showing rtnl contention and one thread blocked in: 7 locks held by syz-executor/10835: #0: ffff888033390420 (sb_writers#8){.+.+}-{0:0}, at: file_start_write include/linux/fs.h:2931 [inline] #0: ffff888033390420 (sb_writers#8){.+.+}-{0:0}, at: vfs_write+0x224/0xc90 fs/read_write.c:679 #1: ffff88806df6bc88 (&of->mutex){+.+.}-{3:3}, at: kernfs_fop_write_iter+0x1ea/0x500 fs/kernfs/file.c:325 #2: ffff888026fcf3c8 (kn->active#50){.+.+}-{0:0}, at: kernfs_fop_write_iter+0x20e/0x500 fs/kernfs/file.c:326 #3: ffffffff8f56f848 (nsim_bus_dev_list_lock){+.+.}-{3:3}, at: new_device_store+0x1b4/0x890 drivers/net/netdevsim/bus.c:166 #4: ffff88805e0140e8 (&dev->mutex){....}-{3:3}, at: device_lock include/linux/device.h:1014 [inline] #4: ffff88805e0140e8 (&dev->mutex){....}-{3:3}, at: __device_attach+0x8e/0x520 drivers/base/dd.c:1005 #5: ffff88805c5fb250 (&devlink->lock_key#55){+.+.}-{3:3}, at: nsim_drv_probe+0xcb/0xb80 drivers/net/netdevsim/dev.c:1534 #6: ffffffff8fcd1748 (rtnl_mutex){+.+.}-{3:3}, at: fib_seq_sum+0x31/0x290 net/core/fib_notifier.c:46 ==================== Link: https://patch.msgid.link/20241009184405.3752829-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
diff --git a/include/net/fib_notifier.h b/include/net/fib_notifier.h
@@ -28,7 +28,7 @@ enum fib_event_type {
 struct fib_notifier_ops {
 	int family;
 	struct list_head list;
-	unsigned int (*fib_seq_read)(struct net *net);
+	unsigned int (*fib_seq_read)(const struct net *net);
 	int (*fib_dump)(struct net *net, struct notifier_block *nb,
 			struct netlink_ext_ack *extack);
 	struct module *owner;
diff --git a/include/net/fib_rules.h b/include/net/fib_rules.h
@@ -176,7 +176,7 @@ int fib_default_rule_add(struct fib_rules_ops *, u32 pref, u32 table);
 bool fib_rule_matchall(const struct fib_rule *rule);
 int fib_rules_dump(struct net *net, struct notifier_block *nb, int family,
 		   struct netlink_ext_ack *extack);
-unsigned int fib_rules_seq_read(struct net *net, int family);
+unsigned int fib_rules_seq_read(const struct net *net, int family);
 
 int fib_nl_newrule(struct sk_buff *skb, struct nlmsghdr *nlh,
 		   struct netlink_ext_ack *extack);
diff --git a/include/net/ip6_fib.h b/include/net/ip6_fib.h
@@ -394,7 +394,7 @@ struct fib6_table {
 	struct fib6_node	tb6_root;
 	struct inet_peer_base	tb6_peers;
 	unsigned int		flags;
-	unsigned int		fib_seq;
+	unsigned int		fib_seq; /* writes protected by rtnl_mutex */
 	struct hlist_head       tb6_gc_hlist;	/* GC candidates */
 #define RT6_TABLE_HAS_DFLT_ROUTER	BIT(0)
 };
@@ -563,7 +563,7 @@ int call_fib6_notifiers(struct net *net, enum fib_event_type event_type,
 int __net_init fib6_notifier_init(struct net *net);
 void __net_exit fib6_notifier_exit(struct net *net);
 
-unsigned int fib6_tables_seq_read(struct net *net);
+unsigned int fib6_tables_seq_read(const struct net *net);
 int fib6_tables_dump(struct net *net, struct notifier_block *nb,
 		     struct netlink_ext_ack *extack);
 
@@ -632,7 +632,7 @@ void fib6_rules_cleanup(void);
 bool fib6_rule_default(const struct fib_rule *rule);
 int fib6_rules_dump(struct net *net, struct notifier_block *nb,
 		    struct netlink_ext_ack *extack);
-unsigned int fib6_rules_seq_read(struct net *net);
+unsigned int fib6_rules_seq_read(const struct net *net);
 
 static inline bool fib6_rules_early_flow_dissect(struct net *net,
 						 struct sk_buff *skb,
@@ -676,7 +676,7 @@ static inline int fib6_rules_dump(struct net *net, struct notifier_block *nb,
 {
 	return 0;
 }
-static inline unsigned int fib6_rules_seq_read(struct net *net)
+static inline unsigned int fib6_rules_seq_read(const struct net *net)
 {
 	return 0;
 }
diff --git a/include/net/ip_fib.h b/include/net/ip_fib.h
@@ -347,7 +347,7 @@ static inline int fib4_rules_dump(struct net *net, struct notifier_block *nb,
 	return 0;
 }
 
-static inline unsigned int fib4_rules_seq_read(struct net *net)
+static inline unsigned int fib4_rules_seq_read(const struct net *net)
 {
 	return 0;
 }
@@ -411,7 +411,7 @@ static inline bool fib4_has_custom_rules(const struct net *net)
 bool fib4_rule_default(const struct fib_rule *rule);
 int fib4_rules_dump(struct net *net, struct notifier_block *nb,
 		    struct netlink_ext_ack *extack);
-unsigned int fib4_rules_seq_read(struct net *net);
+unsigned int fib4_rules_seq_read(const struct net *net);
 
 static inline bool fib4_rules_early_flow_dissect(struct net *net,
 						 struct sk_buff *skb,
diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h
@@ -262,7 +262,7 @@ struct netns_ipv4 {
 #endif
 
 	struct fib_notifier_ops	*notifier_ops;
-	unsigned int	fib_seq;	/* protected by rtnl_mutex */
+	unsigned int	fib_seq;	/* writes protected by rtnl_mutex */
 
 	struct fib_notifier_ops	*ipmr_notifier_ops;
 	unsigned int	ipmr_seq;	/* protected by rtnl_mutex */
diff --git a/net/core/fib_notifier.c b/net/core/fib_notifier.c
@@ -43,7 +43,6 @@ static unsigned int fib_seq_sum(struct net *net)
 	struct fib_notifier_ops *ops;
 	unsigned int fib_seq = 0;
 
-	rtnl_lock();
 	rcu_read_lock();
 	list_for_each_entry_rcu(ops, &fn_net->fib_notifier_ops, list) {
 		if (!try_module_get(ops->owner))
@@ -52,7 +51,6 @@ static unsigned int fib_seq_sum(struct net *net)
 		module_put(ops->owner);
 	}
 	rcu_read_unlock();
-	rtnl_unlock();
 
 	return fib_seq;
 }
diff --git a/net/core/fib_rules.c b/net/core/fib_rules.c
@@ -101,7 +101,8 @@ static void notify_rule_change(int event, struct fib_rule *rule,
 			       struct fib_rules_ops *ops, struct nlmsghdr *nlh,
 			       u32 pid);
 
-static struct fib_rules_ops *lookup_rules_ops(struct net *net, int family)
+static struct fib_rules_ops *lookup_rules_ops(const struct net *net,
+					      int family)
 {
 	struct fib_rules_ops *ops;
 
@@ -370,7 +371,9 @@ static int call_fib_rule_notifiers(struct net *net,
 		.rule = rule,
 	};
 
-	ops->fib_rules_seq++;
+	ASSERT_RTNL();
+	/* Paired with READ_ONCE() in fib_rules_seq() */
+	WRITE_ONCE(ops->fib_rules_seq, ops->fib_rules_seq + 1);
 	return call_fib_notifiers(net, event_type, &info.info);
 }
 
@@ -397,17 +400,16 @@ int fib_rules_dump(struct net *net, struct notifier_block *nb, int family,
 }
 EXPORT_SYMBOL_GPL(fib_rules_dump);
 
-unsigned int fib_rules_seq_read(struct net *net, int family)
+unsigned int fib_rules_seq_read(const struct net *net, int family)
 {
 	unsigned int fib_rules_seq;
 	struct fib_rules_ops *ops;
 
-	ASSERT_RTNL();
-
 	ops = lookup_rules_ops(net, family);
 	if (!ops)
 		return 0;
-	fib_rules_seq = ops->fib_rules_seq;
+	/* Paired with WRITE_ONCE() in call_fib_rule_notifiers() */
+	fib_rules_seq = READ_ONCE(ops->fib_rules_seq);
 	rules_ops_put(ops);
 
 	return fib_rules_seq;
diff --git a/net/ipv4/fib_notifier.c b/net/ipv4/fib_notifier.c
@@ -22,15 +22,15 @@ int call_fib4_notifiers(struct net *net, enum fib_event_type event_type,
 	ASSERT_RTNL();
 
 	info->family = AF_INET;
-	net->ipv4.fib_seq++;
+	/* Paired with READ_ONCE() in fib4_seq_read() */
+	WRITE_ONCE(net->ipv4.fib_seq, net->ipv4.fib_seq + 1);
 	return call_fib_notifiers(net, event_type, info);
 }
 
-static unsigned int fib4_seq_read(struct net *net)
+static unsigned int fib4_seq_read(const struct net *net)
 {
-	ASSERT_RTNL();
-
-	return net->ipv4.fib_seq + fib4_rules_seq_read(net);
+	/* Paired with WRITE_ONCE() in call_fib4_notifiers() */
+	return READ_ONCE(net->ipv4.fib_seq) + fib4_rules_seq_read(net);
 }
 
 static int fib4_dump(struct net *net, struct notifier_block *nb,
diff --git a/net/ipv4/fib_rules.c b/net/ipv4/fib_rules.c
@@ -74,7 +74,7 @@ int fib4_rules_dump(struct net *net, struct notifier_block *nb,
 	return fib_rules_dump(net, nb, AF_INET, extack);
 }
 
-unsigned int fib4_rules_seq_read(struct net *net)
+unsigned int fib4_rules_seq_read(const struct net *net)
 {
 	return fib_rules_seq_read(net, AF_INET);
 }
diff --git a/net/ipv4/ipmr.c b/net/ipv4/ipmr.c
@@ -288,7 +288,7 @@ static int ipmr_rules_dump(struct net *net, struct notifier_block *nb,
 	return fib_rules_dump(net, nb, RTNL_FAMILY_IPMR, extack);
 }
 
-static unsigned int ipmr_rules_seq_read(struct net *net)
+static unsigned int ipmr_rules_seq_read(const struct net *net)
 {
 	return fib_rules_seq_read(net, RTNL_FAMILY_IPMR);
 }
@@ -346,7 +346,7 @@ static int ipmr_rules_dump(struct net *net, struct notifier_block *nb,
 	return 0;
 }
 
-static unsigned int ipmr_rules_seq_read(struct net *net)
+static unsigned int ipmr_rules_seq_read(const struct net *net)
 {
 	return 0;
 }
@@ -3035,11 +3035,9 @@ static const struct net_protocol pim_protocol = {
 };
 #endif
 
-static unsigned int ipmr_seq_read(struct net *net)
+static unsigned int ipmr_seq_read(const struct net *net)
 {
-	ASSERT_RTNL();
-
-	return net->ipv4.ipmr_seq + ipmr_rules_seq_read(net);
+	return READ_ONCE(net->ipv4.ipmr_seq) + ipmr_rules_seq_read(net);
 }
 
 static int ipmr_dump(struct net *net, struct notifier_block *nb,
diff --git a/net/ipv6/fib6_notifier.c b/net/ipv6/fib6_notifier.c
@@ -22,7 +22,7 @@ int call_fib6_notifiers(struct net *net, enum fib_event_type event_type,
 	return call_fib_notifiers(net, event_type, info);
 }
 
-static unsigned int fib6_seq_read(struct net *net)
+static unsigned int fib6_seq_read(const struct net *net)
 {
 	return fib6_tables_seq_read(net) + fib6_rules_seq_read(net);
 }
diff --git a/net/ipv6/fib6_rules.c b/net/ipv6/fib6_rules.c
@@ -56,7 +56,7 @@ int fib6_rules_dump(struct net *net, struct notifier_block *nb,
 	return fib_rules_dump(net, nb, AF_INET6, extack);
 }
 
-unsigned int fib6_rules_seq_read(struct net *net)
+unsigned int fib6_rules_seq_read(const struct net *net)
 {
 	return fib_rules_seq_read(net, AF_INET6);
 }
diff --git a/net/ipv6/ip6_fib.c b/net/ipv6/ip6_fib.c
@@ -345,17 +345,17 @@ static void __net_init fib6_tables_init(struct net *net)
 
 #endif
 
-unsigned int fib6_tables_seq_read(struct net *net)
+unsigned int fib6_tables_seq_read(const struct net *net)
 {
 	unsigned int h, fib_seq = 0;
 
 	rcu_read_lock();
 	for (h = 0; h < FIB6_TABLE_HASHSZ; h++) {
-		struct hlist_head *head = &net->ipv6.fib_table_hash[h];
-		struct fib6_table *tb;
+		const struct hlist_head *head = &net->ipv6.fib_table_hash[h];
+		const struct fib6_table *tb;
 
 		hlist_for_each_entry_rcu(tb, head, tb6_hlist)
-			fib_seq += tb->fib_seq;
+			fib_seq += READ_ONCE(tb->fib_seq);
 	}
 	rcu_read_unlock();
 
@@ -400,7 +400,7 @@ int call_fib6_entry_notifiers(struct net *net,
 		.rt = rt,
 	};
 
-	rt->fib6_table->fib_seq++;
+	WRITE_ONCE(rt->fib6_table->fib_seq, rt->fib6_table->fib_seq + 1);
 	return call_fib6_notifiers(net, event_type, &info.info);
 }
 
@@ -416,7 +416,7 @@ int call_fib6_multipath_entry_notifiers(struct net *net,
 		.nsiblings = nsiblings,
 	};
 
-	rt->fib6_table->fib_seq++;
+	WRITE_ONCE(rt->fib6_table->fib_seq, rt->fib6_table->fib_seq + 1);
 	return call_fib6_notifiers(net, event_type, &info.info);
 }
 
@@ -427,7 +427,7 @@ int call_fib6_entry_notifiers_replace(struct net *net, struct fib6_info *rt)
 		.nsiblings = rt->fib6_nsiblings,
 	};
 
-	rt->fib6_table->fib_seq++;
+	WRITE_ONCE(rt->fib6_table->fib_seq, rt->fib6_table->fib_seq + 1);
 	return call_fib6_notifiers(net, FIB_EVENT_ENTRY_REPLACE, &info.info);
 }
 
diff --git a/net/ipv6/ip6mr.c b/net/ipv6/ip6mr.c
@@ -276,7 +276,7 @@ static int ip6mr_rules_dump(struct net *net, struct notifier_block *nb,
 	return fib_rules_dump(net, nb, RTNL_FAMILY_IP6MR, extack);
 }
 
-static unsigned int ip6mr_rules_seq_read(struct net *net)
+static unsigned int ip6mr_rules_seq_read(const struct net *net)
 {
 	return fib_rules_seq_read(net, RTNL_FAMILY_IP6MR);
 }
@@ -335,7 +335,7 @@ static int ip6mr_rules_dump(struct net *net, struct notifier_block *nb,
 	return 0;
 }
 
-static unsigned int ip6mr_rules_seq_read(struct net *net)
+static unsigned int ip6mr_rules_seq_read(const struct net *net)
 {
 	return 0;
 }
@@ -1260,11 +1260,9 @@ static int ip6mr_device_event(struct notifier_block *this,
 	return NOTIFY_DONE;
 }
 
-static unsigned int ip6mr_seq_read(struct net *net)
+static unsigned int ip6mr_seq_read(const struct net *net)
 {
-	ASSERT_RTNL();
-
-	return net->ipv6.ipmr_seq + ip6mr_rules_seq_read(net);
+	return READ_ONCE(net->ipv6.ipmr_seq) + ip6mr_rules_seq_read(net);
 }
 
 static int ip6mr_dump(struct net *net, struct notifier_block *nb,

Original file line number	Diff line number	Diff line change
`@@ -347,7 +347,7 @@ static inline int fib4_rules_dump(struct net net, struct notifier_block nb,`
`347`	`347`	`return 0;`
`348`	`348`	`}`
`349`	`349`
`350`		`-static inline unsigned int fib4_rules_seq_read(struct net *net)`
	`350`	`+static inline unsigned int fib4_rules_seq_read(const struct net *net)`
`351`	`351`	`{`
`352`	`352`	`return 0;`
`353`	`353`	`}`
`@@ -411,7 +411,7 @@ static inline bool fib4_has_custom_rules(const struct net *net)`
`411`	`411`	`bool fib4_rule_default(const struct fib_rule *rule);`
`412`	`412`	`int fib4_rules_dump(struct net net, struct notifier_block nb,`
`413`	`413`	`struct netlink_ext_ack *extack);`
`414`		`-unsigned int fib4_rules_seq_read(struct net *net);`
	`414`	`+unsigned int fib4_rules_seq_read(const struct net *net);`
`415`	`415`
`416`	`416`	`static inline bool fib4_rules_early_flow_dissect(struct net *net,`
`417`	`417`	`struct sk_buff *skb,`
Original file line number	Diff line number	Diff line change
`@@ -43,7 +43,6 @@ static unsigned int fib_seq_sum(struct net *net)`
`43`	`43`	`struct fib_notifier_ops *ops;`
`44`	`44`	`unsigned int fib_seq = 0;`
`45`	`45`
`46`		`- rtnl_lock();`
`47`	`46`	`rcu_read_lock();`
`48`	`47`	`list_for_each_entry_rcu(ops, &fn_net->fib_notifier_ops, list) {`
`49`	`48`	`if (!try_module_get(ops->owner))`
`@@ -52,7 +51,6 @@ static unsigned int fib_seq_sum(struct net *net)`
`52`	`51`	`module_put(ops->owner);`
`53`	`52`	`}`
`54`	`53`	`rcu_read_unlock();`
`55`		`- rtnl_unlock();`
`56`	`54`
`57`	`55`	`return fib_seq;`
`58`	`56`	`}`
Original file line number	Diff line number	Diff line change
`@@ -22,15 +22,15 @@ int call_fib4_notifiers(struct net *net, enum fib_event_type event_type,`
`22`	`22`	`ASSERT_RTNL();`
`23`	`23`
`24`	`24`	`info->family = AF_INET;`
`25`		`- net->ipv4.fib_seq++;`
	`25`	`+ /* Paired with READ_ONCE() in fib4_seq_read() */`
	`26`	`+ WRITE_ONCE(net->ipv4.fib_seq, net->ipv4.fib_seq + 1);`
`26`	`27`	`return call_fib_notifiers(net, event_type, info);`
`27`	`28`	`}`
`28`	`29`
`29`		`-static unsigned int fib4_seq_read(struct net *net)`
	`30`	`+static unsigned int fib4_seq_read(const struct net *net)`
`30`	`31`	`{`
`31`		`- ASSERT_RTNL();`
`32`		`-`
`33`		`- return net->ipv4.fib_seq + fib4_rules_seq_read(net);`
	`32`	`+ /* Paired with WRITE_ONCE() in call_fib4_notifiers() */`
	`33`	`+ return READ_ONCE(net->ipv4.fib_seq) + fib4_rules_seq_read(net);`
`34`	`34`	`}`
`35`	`35`
`36`	`36`	`static int fib4_dump(struct net net, struct notifier_block nb,`
Original file line number	Diff line number	Diff line change
`@@ -74,7 +74,7 @@ int fib4_rules_dump(struct net net, struct notifier_block nb,`
`74`	`74`	`return fib_rules_dump(net, nb, AF_INET, extack);`
`75`	`75`	`}`
`76`	`76`
`77`		`-unsigned int fib4_rules_seq_read(struct net *net)`
	`77`	`+unsigned int fib4_rules_seq_read(const struct net *net)`
`78`	`78`	`{`
`79`	`79`	`return fib_rules_seq_read(net, AF_INET);`
`80`	`80`	`}`
Original file line number	Diff line number	Diff line change
`@@ -288,7 +288,7 @@ static int ipmr_rules_dump(struct net net, struct notifier_block nb,`
`288`	`288`	`return fib_rules_dump(net, nb, RTNL_FAMILY_IPMR, extack);`
`289`	`289`	`}`
`290`	`290`
`291`		`-static unsigned int ipmr_rules_seq_read(struct net *net)`
	`291`	`+static unsigned int ipmr_rules_seq_read(const struct net *net)`
`292`	`292`	`{`
`293`	`293`	`return fib_rules_seq_read(net, RTNL_FAMILY_IPMR);`
`294`	`294`	`}`
`@@ -346,7 +346,7 @@ static int ipmr_rules_dump(struct net net, struct notifier_block nb,`
`346`	`346`	`return 0;`
`347`	`347`	`}`
`348`	`348`
`349`		`-static unsigned int ipmr_rules_seq_read(struct net *net)`
	`349`	`+static unsigned int ipmr_rules_seq_read(const struct net *net)`
`350`	`350`	`{`
`351`	`351`	`return 0;`
`352`	`352`	`}`
`@@ -3035,11 +3035,9 @@ static const struct net_protocol pim_protocol = {`
`3035`	`3035`	`};`
`3036`	`3036`	`#endif`
`3037`	`3037`
`3038`		`-static unsigned int ipmr_seq_read(struct net *net)`
	`3038`	`+static unsigned int ipmr_seq_read(const struct net *net)`
`3039`	`3039`	`{`
`3040`		`- ASSERT_RTNL();`
`3041`		`-`
`3042`		`- return net->ipv4.ipmr_seq + ipmr_rules_seq_read(net);`
	`3040`	`+ return READ_ONCE(net->ipv4.ipmr_seq) + ipmr_rules_seq_read(net);`
`3043`	`3041`	`}`
`3044`	`3042`
`3045`	`3043`	`static int ipmr_dump(struct net net, struct notifier_block nb,`