[PATCH net-next 0/5] Remove expired routes with a separated list of routes.

[PATCH net-next 0/5] Remove expired routes with a separated list of routes.

[thinker.li]

From: Kui-Feng Lee <thinker.li@gmail.com>

This patchset is resent due to previous reverting. [1]

FIB6 GC walks trees of fib6_tables to remove expired routes. Walking a tree
can be expensive if the number of routes in a table is big, even if most of
them are permanent. Checking routes in a separated list of routes having
expiration will avoid this potential issue.

Background
==========

The size of a Linux IPv6 routing table can become a big problem if not
managed appropriately.  Now, Linux has a garbage collector to remove
expired routes periodically.  However, this may lead to a situation in
which the routing path is blocked for a long period due to an
excessive number of routes.

For example, years ago, there is a commit c7bb4b89033b ("ipv6: tcp:
drop silly ICMPv6 packet too big messages").  The root cause is that
malicious ICMPv6 packets were sent back for every small packet sent to
them. These packets add routes with an expiration time that prompts
the GC to periodically check all routes in the tables, including
permanent ones.

Why Route Expires
=================

Users can add IPv6 routes with an expiration time manually. However,
the Neighbor Discovery protocol may also generate routes that can
expire.  For example, Router Advertisement (RA) messages may create a
default route with an expiration time. [RFC 4861] For IPv4, it is not
possible to set an expiration time for a route, and there is no RA, so
there is no need to worry about such issues.

Create Routes with Expires
==========================

You can create routes with expires with the  command.

For example,

    ip -6 route add 2001:b000:591::3 via fe80::5054:ff:fe12:3457 \
        dev enp0s3 expires 30

The route that has been generated will be deleted automatically in 30
seconds.

GC of FIB6
==========

The function called fib6_run_gc() is responsible for performing
garbage collection (GC) for the Linux IPv6 stack. It checks for the
expiration of every route by traversing the trees of routing
tables. The time taken to traverse a routing table increases with its
size. Holding the routing table lock during traversal is particularly
undesirable. Therefore, it is preferable to keep the lock for the
shortest possible duration.

Solution
========

The cause of the issue is keeping the routing table locked during the
traversal of large trees. To solve this problem, we can create a separate
list of routes that have expiration. This will prevent GC from checking
permanent routes.

Result
======

We conducted a test to measure the execution times of fib6_gc_timer_cb()
and observed that it enhances the GC of FIB6. During the test, we added
permanent routes with the following numbers: 1000, 3000, 6000, and
9000. Additionally, we added a route with an expiration time.

Here are the average execution times for the kernel without the patch.
 - 120020 ns with 1000 permanent routes
 - 308920 ns with 3000 ...
 - 581470 ns with 6000 ...
 - 855310 ns with 9000 ...

The kernel with the patch consistently takes around 14000 ns to execute,
regardless of the number of permanent routes that are installed.

Major changes from the previous patchset [2]:

 - Split helpers.

   - fib6_set_expires() -> fib6_set_expires() and fib6_add_gc_list().

   - fib6_clean_expires() -> fib6_clean_expires() and
     fib6_remove_gc_list().

 - Fix rt6_add_dflt_router() to avoid racing of setting expires.

 - Remove unnecessary calling to fib6_clean_expires() in
   ip6_route_info_create().

 - Add test cases of toggling routes between permanent and temporary
   and handling routes from RA messages.

   - Clean up routes by deleting the existing device and adding a new
     one.

 - Fix a potential issue in modify_prefix_route().

---
[1] https://lore.kernel.org/all/20231219030243.25687-1-dsahern@kernel.org/
[2] https://lore.kernel.org/all/20230815180706.772638-1-thinker.li@gmail.com/

Kui-Feng Lee (5):
  net/ipv6: set expires in rt6_add_dflt_router().
  net/ipv6: Remove unnecessary clean.
  net/ipv6: Remove expired routes with a separated list of routes.
  net/ipv6: set expires in modify_prefix_route() if RTF_EXPIRES is set.
  selftests/net: Adding test cases of replacing routes and route
    advertisements.

 include/net/ip6_fib.h                    |  36 ++++-
 include/net/ip6_route.h                  |   3 +-
 net/ipv6/addrconf.c                      |  50 ++++++-
 net/ipv6/ip6_fib.c                       |  58 ++++++++-
 net/ipv6/ndisc.c                         |  14 +-
 net/ipv6/route.c                         |  20 ++-
 tools/testing/selftests/net/fib_tests.sh | 159 +++++++++++++++++++++--
 7 files changed, 308 insertions(+), 32 deletions(-)

-- 
2.34.1

[PATCH net-next 1/5] net/ipv6: set expires in rt6_add_dflt_router().

[thinker.li]

From: Kui-Feng Lee <thinker.li@gmail.com>

Pass the duration of a lifetime (in seconds) to the function
rt6_add_dflt_router() so that it can properly set the expiration time.

The function ndisc_router_discovery() is the only one that calls
rt6_add_dflt_router(), and it will later set the expiration time for the
route created by rt6_add_dflt_router(). However, there is a gap of time
between calling rt6_add_dflt_router() and setting the expiration time in
ndisc_router_discovery(). During this period, there is a possibility that a
new route may be removed from the routing table. By setting the correct
expiration time in rt6_add_dflt_router(), we can prevent this from
happening. The reason for setting RTF_EXPIRES in rt6_add_dflt_router() is
to start the Garbage Collection (GC) timer, as it only activates when a
route with RTF_EXPIRES is added to a table.

Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
---
 include/net/ip6_route.h | 3 ++-
 net/ipv6/ndisc.c        | 3 ++-
 net/ipv6/route.c        | 4 +++-
 3 files changed, 7 insertions(+), 3 deletions(-)

diff --git a/include/net/ip6_route.h b/include/net/ip6_route.h
index 28b065790261..52a51c69aa9d 100644
--- a/include/net/ip6_route.h
+++ b/include/net/ip6_route.h
@@ -170,7 +170,8 @@ struct fib6_info *rt6_get_dflt_router(struct net *net,
 struct fib6_info *rt6_add_dflt_router(struct net *net,
 				     const struct in6_addr *gwaddr,
 				     struct net_device *dev, unsigned int pref,
-				     u32 defrtr_usr_metric);
+				     u32 defrtr_usr_metric,
+				     int lifetime);
 
 void rt6_purge_dflt_routers(struct net *net);
 
diff --git a/net/ipv6/ndisc.c b/net/ipv6/ndisc.c
index a19999b30bc0..a68462668158 100644
--- a/net/ipv6/ndisc.c
+++ b/net/ipv6/ndisc.c
@@ -1382,7 +1382,8 @@ static enum skb_drop_reason ndisc_router_discovery(struct sk_buff *skb)
 			neigh_release(neigh);
 
 		rt = rt6_add_dflt_router(net, &ipv6_hdr(skb)->saddr,
-					 skb->dev, pref, defrtr_usr_metric);
+					 skb->dev, pref, defrtr_usr_metric,
+					 lifetime);
 		if (!rt) {
 			ND_PRINTK(0, err,
 				  "RA: %s failed to add default route\n",
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index 63b4c6056582..98abba8f15cd 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -4355,7 +4355,8 @@ struct fib6_info *rt6_add_dflt_router(struct net *net,
 				     const struct in6_addr *gwaddr,
 				     struct net_device *dev,
 				     unsigned int pref,
-				     u32 defrtr_usr_metric)
+				     u32 defrtr_usr_metric,
+				     int lifetime)
 {
 	struct fib6_config cfg = {
 		.fc_table	= l3mdev_fib_table(dev) ? : RT6_TABLE_DFLT,
@@ -4368,6 +4369,7 @@ struct fib6_info *rt6_add_dflt_router(struct net *net,
 		.fc_nlinfo.portid = 0,
 		.fc_nlinfo.nlh = NULL,
 		.fc_nlinfo.nl_net = net,
+		.fc_expires = jiffies_to_clock_t(lifetime * HZ),
 	};
 
 	cfg.fc_gateway = *gwaddr;
-- 
2.34.1

[PATCH net-next 2/5] net/ipv6: Remove unnecessary clean.

[thinker.li]

From: Kui-Feng Lee <thinker.li@gmail.com>

The route here is newly created. It is unnecessary to call
fib6_clean_expires() on it.

Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
---
 net/ipv6/route.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index 98abba8f15cd..dd6ff5b20918 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -3765,8 +3765,6 @@ static struct fib6_info *ip6_route_info_create(struct fib6_config *cfg,
 	if (cfg->fc_flags & RTF_EXPIRES)
 		fib6_set_expires(rt, jiffies +
 				clock_t_to_jiffies(cfg->fc_expires));
-	else
-		fib6_clean_expires(rt);
 
 	if (cfg->fc_protocol == RTPROT_UNSPEC)
 		cfg->fc_protocol = RTPROT_BOOT;
-- 
2.34.1

Re: [PATCH net-next 2/5] net/ipv6: Remove unnecessary clean.

[Hangbin Liu]

On Tue, Jan 30, 2024 at 10:40:38PM -0800, thinker.li@gmail.com wrote:
> From: Kui-Feng Lee <thinker.li@gmail.com>
> 
> The route here is newly created. It is unnecessary to call
> fib6_clean_expires() on it.
> 
> Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
> ---
>  net/ipv6/route.c | 2 --
>  1 file changed, 2 deletions(-)
> 
> diff --git a/net/ipv6/route.c b/net/ipv6/route.c
> index 98abba8f15cd..dd6ff5b20918 100644
> --- a/net/ipv6/route.c
> +++ b/net/ipv6/route.c
> @@ -3765,8 +3765,6 @@ static struct fib6_info *ip6_route_info_create(struct fib6_config *cfg,
>  	if (cfg->fc_flags & RTF_EXPIRES)
>  		fib6_set_expires(rt, jiffies +
>  				clock_t_to_jiffies(cfg->fc_expires));
> -	else
> -		fib6_clean_expires(rt);
>  
>  	if (cfg->fc_protocol == RTPROT_UNSPEC)
>  		cfg->fc_protocol = RTPROT_BOOT;
> -- 
> 2.34.1
> 

Reviewed-by: Hangbin Liu <liuhangbin@gmail.com>

[PATCH net-next 3/5] net/ipv6: Remove expired routes with a separated list of routes.

[thinker.li]

From: Kui-Feng Lee <thinker.li@gmail.com>

FIB6 GC walks trees of fib6_tables to remove expired routes. Walking a tree
can be expensive if the number of routes in a table is big, even if most of
them are permanent. Checking routes in a separated list of routes having
expiration will avoid this potential issue.

Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
---
 include/net/ip6_fib.h | 36 ++++++++++++++++++++++++++-
 net/ipv6/addrconf.c   | 50 +++++++++++++++++++++++++++++++------
 net/ipv6/ip6_fib.c    | 58 +++++++++++++++++++++++++++++++++++++++----
 net/ipv6/ndisc.c      | 11 +++++++-
 net/ipv6/route.c      | 14 +++++++++--
 5 files changed, 153 insertions(+), 16 deletions(-)

diff --git a/include/net/ip6_fib.h b/include/net/ip6_fib.h
index 360b12e61850..9cb616a4db9b 100644
--- a/include/net/ip6_fib.h
+++ b/include/net/ip6_fib.h
@@ -173,6 +173,9 @@ struct fib6_info {
 
 	refcount_t			fib6_ref;
 	unsigned long			expires;
+
+	struct hlist_node		gc_link;
+
 	struct dst_metrics		*fib6_metrics;
 #define fib6_pmtu		fib6_metrics->metrics[RTAX_MTU-1]
 
@@ -241,12 +244,18 @@ static inline bool fib6_requires_src(const struct fib6_info *rt)
 	return rt->fib6_src.plen > 0;
 }
 
+/* The callers should hold f6i->fib6_table->tb6_lock if a route has ever
+ * been added to a table before.
+ */
 static inline void fib6_clean_expires(struct fib6_info *f6i)
 {
 	f6i->fib6_flags &= ~RTF_EXPIRES;
 	f6i->expires = 0;
 }
 
+/* The callers should hold f6i->fib6_table->tb6_lock if a route has ever
+ * been added to a table before.
+ */
 static inline void fib6_set_expires(struct fib6_info *f6i,
 				    unsigned long expires)
 {
@@ -327,8 +336,10 @@ static inline bool fib6_info_hold_safe(struct fib6_info *f6i)
 
 static inline void fib6_info_release(struct fib6_info *f6i)
 {
-	if (f6i && refcount_dec_and_test(&f6i->fib6_ref))
+	if (f6i && refcount_dec_and_test(&f6i->fib6_ref)) {
+		DEBUG_NET_WARN_ON_ONCE(!hlist_unhashed(&f6i->gc_link));
 		call_rcu(&f6i->rcu, fib6_info_destroy_rcu);
+	}
 }
 
 enum fib6_walk_state {
@@ -382,6 +393,7 @@ struct fib6_table {
 	struct inet_peer_base	tb6_peers;
 	unsigned int		flags;
 	unsigned int		fib_seq;
+	struct hlist_head       tb6_gc_hlist;	/* GC candidates */
 #define RT6_TABLE_HAS_DFLT_ROUTER	BIT(0)
 };
 
@@ -498,6 +510,28 @@ void fib6_gc_cleanup(void);
 
 int fib6_init(void);
 
+/* Add the route to the gc list if it is not already there
+ *
+ * The callers should hold f6i->fib6_table->tb6_lock and make sure the
+ * route is on a table.
+ */
+static inline void fib6_add_gc_list(struct fib6_info *f6i)
+{
+
+	if (hlist_unhashed(&f6i->gc_link))
+		hlist_add_head(&f6i->gc_link, &f6i->fib6_table->tb6_gc_hlist);
+}
+
+/* Remove the route from the gc list if it is on the list.
+ *
+ * The callers should hold f6i->fib6_table->tb6_lock.
+ */
+static inline void fib6_remove_gc_list(struct fib6_info *f6i)
+{
+	if (!hlist_unhashed(&f6i->gc_link))
+		hlist_del_init(&f6i->gc_link);
+}
+
 struct ipv6_route_iter {
 	struct seq_net_private p;
 	struct fib6_walker w;
diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
index 733ace18806c..36bfa987c314 100644
--- a/net/ipv6/addrconf.c
+++ b/net/ipv6/addrconf.c
@@ -1255,6 +1255,7 @@ static void
 cleanup_prefix_route(struct inet6_ifaddr *ifp, unsigned long expires,
 		     bool del_rt, bool del_peer)
 {
+	struct fib6_table *table;
 	struct fib6_info *f6i;
 
 	f6i = addrconf_get_prefix_route(del_peer ? &ifp->peer_addr : &ifp->addr,
@@ -1264,8 +1265,18 @@ cleanup_prefix_route(struct inet6_ifaddr *ifp, unsigned long expires,
 		if (del_rt)
 			ip6_del_rt(dev_net(ifp->idev->dev), f6i, false);
 		else {
-			if (!(f6i->fib6_flags & RTF_EXPIRES))
+			if (!(f6i->fib6_flags & RTF_EXPIRES)) {
+				table = f6i->fib6_table;
+				spin_lock_bh(&table->tb6_lock);
 				fib6_set_expires(f6i, expires);
+				/* If fib6_node is null, the f6i is just
+				 * removed from the table.
+				 */
+				if (rcu_dereference_protected(f6i->fib6_node,
+							      lockdep_is_held(&table->tb6_lock)))
+					fib6_add_gc_list(f6i);
+				spin_unlock_bh(&table->tb6_lock);
+			}
 			fib6_info_release(f6i);
 		}
 	}
@@ -2706,6 +2717,7 @@ EXPORT_SYMBOL_GPL(addrconf_prefix_rcv_add_addr);
 void addrconf_prefix_rcv(struct net_device *dev, u8 *opt, int len, bool sllao)
 {
 	struct prefix_info *pinfo;
+	struct fib6_table *table;
 	__u32 valid_lft;
 	__u32 prefered_lft;
 	int addr_type, err;
@@ -2782,11 +2794,23 @@ void addrconf_prefix_rcv(struct net_device *dev, u8 *opt, int len, bool sllao)
 			if (valid_lft == 0) {
 				ip6_del_rt(net, rt, false);
 				rt = NULL;
-			} else if (addrconf_finite_timeout(rt_expires)) {
-				/* not infinity */
-				fib6_set_expires(rt, jiffies + rt_expires);
 			} else {
-				fib6_clean_expires(rt);
+				table = rt->fib6_table;
+				spin_lock_bh(&table->tb6_lock);
+				if (addrconf_finite_timeout(rt_expires)) {
+					/* not infinity */
+					fib6_set_expires(rt, jiffies + rt_expires);
+					/* If fib6_node is null, the f6i is
+					 * just removed from the table.
+					 */
+					if (rcu_dereference_protected(rt->fib6_node,
+								      lockdep_is_held(&table->tb6_lock)))
+						fib6_add_gc_list(rt);
+				} else {
+					fib6_clean_expires(rt);
+					fib6_remove_gc_list(rt);
+				}
+				spin_unlock_bh(&table->tb6_lock);
 			}
 		} else if (valid_lft) {
 			clock_t expires = 0;
@@ -4741,6 +4765,7 @@ static int modify_prefix_route(struct inet6_ifaddr *ifp,
 			       unsigned long expires, u32 flags,
 			       bool modify_peer)
 {
+	struct fib6_table *table;
 	struct fib6_info *f6i;
 	u32 prio;
 
@@ -4761,10 +4786,21 @@ static int modify_prefix_route(struct inet6_ifaddr *ifp,
 				      ifp->rt_priority, ifp->idev->dev,
 				      expires, flags, GFP_KERNEL);
 	} else {
-		if (!expires)
+		table = f6i->fib6_table;
+		spin_lock_bh(&table->tb6_lock);
+		if (!expires) {
 			fib6_clean_expires(f6i);
-		else
+			fib6_remove_gc_list(f6i);
+		} else {
 			fib6_set_expires(f6i, expires);
+			/* If fib6_node is null, the f6i is just removed
+			 * from the table.
+			 */
+			if (rcu_dereference_protected(f6i->fib6_node,
+						      lockdep_is_held(&table->tb6_lock)))
+				fib6_add_gc_list(f6i);
+		}
+		spin_unlock_bh(&table->tb6_lock);
 
 		fib6_info_release(f6i);
 	}
diff --git a/net/ipv6/ip6_fib.c b/net/ipv6/ip6_fib.c
index 38a0348b1d17..d53dc519d317 100644
--- a/net/ipv6/ip6_fib.c
+++ b/net/ipv6/ip6_fib.c
@@ -160,6 +160,8 @@ struct fib6_info *fib6_info_alloc(gfp_t gfp_flags, bool with_fib6_nh)
 	INIT_LIST_HEAD(&f6i->fib6_siblings);
 	refcount_set(&f6i->fib6_ref, 1);
 
+	INIT_HLIST_NODE(&f6i->gc_link);
+
 	return f6i;
 }
 
@@ -246,6 +248,7 @@ static struct fib6_table *fib6_alloc_table(struct net *net, u32 id)
 				   net->ipv6.fib6_null_entry);
 		table->tb6_root.fn_flags = RTN_ROOT | RTN_TL_ROOT | RTN_RTINFO;
 		inet_peer_base_init(&table->tb6_peers);
+		INIT_HLIST_HEAD(&table->tb6_gc_hlist);
 	}
 
 	return table;
@@ -1055,6 +1058,9 @@ static void fib6_purge_rt(struct fib6_info *rt, struct fib6_node *fn,
 				    lockdep_is_held(&table->tb6_lock));
 		}
 	}
+
+	fib6_clean_expires(rt);
+	fib6_remove_gc_list(rt);
 }
 
 /*
@@ -1115,10 +1121,13 @@ static int fib6_add_rt2node(struct fib6_node *fn, struct fib6_info *rt,
 					rt->fib6_nsiblings = 0;
 				if (!(iter->fib6_flags & RTF_EXPIRES))
 					return -EEXIST;
-				if (!(rt->fib6_flags & RTF_EXPIRES))
+				if (!(rt->fib6_flags & RTF_EXPIRES)) {
 					fib6_clean_expires(iter);
-				else
+					fib6_remove_gc_list(iter);
+				} else {
 					fib6_set_expires(iter, rt->expires);
+					fib6_add_gc_list(iter);
+				}
 
 				if (rt->fib6_pmtu)
 					fib6_metric_set(iter, RTAX_MTU,
@@ -1477,6 +1486,10 @@ int fib6_add(struct fib6_node *root, struct fib6_info *rt,
 		if (rt->nh)
 			list_add(&rt->nh_list, &rt->nh->f6i_list);
 		__fib6_update_sernum_upto_root(rt, fib6_new_sernum(info->nl_net));
+
+		if (rt->fib6_flags & RTF_EXPIRES)
+			fib6_add_gc_list(rt);
+
 		fib6_start_gc(info->nl_net, rt);
 	}
 
@@ -2280,9 +2293,8 @@ static void fib6_flush_trees(struct net *net)
  *	Garbage collection
  */
 
-static int fib6_age(struct fib6_info *rt, void *arg)
+static int fib6_age(struct fib6_info *rt, struct fib6_gc_args *gc_args)
 {
-	struct fib6_gc_args *gc_args = arg;
 	unsigned long now = jiffies;
 
 	/*
@@ -2307,6 +2319,40 @@ static int fib6_age(struct fib6_info *rt, void *arg)
 	return 0;
 }
 
+static void fib6_gc_table(struct net *net,
+			  struct fib6_table *tb6,
+			  struct fib6_gc_args *gc_args)
+{
+	struct fib6_info *rt;
+	struct hlist_node *n;
+	struct nl_info info = {
+		.nl_net = net,
+		.skip_notify = false,
+	};
+
+	hlist_for_each_entry_safe(rt, n, &tb6->tb6_gc_hlist, gc_link)
+		if (fib6_age(rt, gc_args) == -1)
+			fib6_del(rt, &info);
+}
+
+static void fib6_gc_all(struct net *net, struct fib6_gc_args *gc_args)
+{
+	struct fib6_table *table;
+	struct hlist_head *head;
+	unsigned int h;
+
+	rcu_read_lock();
+	for (h = 0; h < FIB6_TABLE_HASHSZ; h++) {
+		head = &net->ipv6.fib_table_hash[h];
+		hlist_for_each_entry_rcu(table, head, tb6_hlist) {
+			spin_lock_bh(&table->tb6_lock);
+			fib6_gc_table(net, table, gc_args);
+			spin_unlock_bh(&table->tb6_lock);
+		}
+	}
+	rcu_read_unlock();
+}
+
 void fib6_run_gc(unsigned long expires, struct net *net, bool force)
 {
 	struct fib6_gc_args gc_args;
@@ -2322,7 +2368,7 @@ void fib6_run_gc(unsigned long expires, struct net *net, bool force)
 			  net->ipv6.sysctl.ip6_rt_gc_interval;
 	gc_args.more = 0;
 
-	fib6_clean_all(net, fib6_age, &gc_args);
+	fib6_gc_all(net, &gc_args);
 	now = jiffies;
 	net->ipv6.ip6_rt_last_gc = now;
 
@@ -2382,6 +2428,7 @@ static int __net_init fib6_net_init(struct net *net)
 	net->ipv6.fib6_main_tbl->tb6_root.fn_flags =
 		RTN_ROOT | RTN_TL_ROOT | RTN_RTINFO;
 	inet_peer_base_init(&net->ipv6.fib6_main_tbl->tb6_peers);
+	INIT_HLIST_HEAD(&net->ipv6.fib6_main_tbl->tb6_gc_hlist);
 
 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
 	net->ipv6.fib6_local_tbl = kzalloc(sizeof(*net->ipv6.fib6_local_tbl),
@@ -2394,6 +2441,7 @@ static int __net_init fib6_net_init(struct net *net)
 	net->ipv6.fib6_local_tbl->tb6_root.fn_flags =
 		RTN_ROOT | RTN_TL_ROOT | RTN_RTINFO;
 	inet_peer_base_init(&net->ipv6.fib6_local_tbl->tb6_peers);
+	INIT_HLIST_HEAD(&net->ipv6.fib6_local_tbl->tb6_gc_hlist);
 #endif
 	fib6_tables_init(net);
 
diff --git a/net/ipv6/ndisc.c b/net/ipv6/ndisc.c
index a68462668158..5ca9fd4f7945 100644
--- a/net/ipv6/ndisc.c
+++ b/net/ipv6/ndisc.c
@@ -1410,8 +1410,17 @@ static enum skb_drop_reason ndisc_router_discovery(struct sk_buff *skb)
 		inet6_rt_notify(RTM_NEWROUTE, rt, &nlinfo, NLM_F_REPLACE);
 	}
 
-	if (rt)
+	if (rt) {
+		spin_lock_bh(&rt->fib6_table->tb6_lock);
 		fib6_set_expires(rt, jiffies + (HZ * lifetime));
+		/* If fib6_node is null, the f6i is just removed from the
+		 * table.
+		 */
+		if (rcu_dereference_protected(rt->fib6_node,
+					      lockdep_is_held(&rt->fib6_table->tb6_lock)))
+			fib6_add_gc_list(rt);
+		spin_unlock_bh(&rt->fib6_table->tb6_lock);
+	}
 	if (in6_dev->cnf.accept_ra_min_hop_limit < 256 &&
 	    ra_msg->icmph.icmp6_hop_limit) {
 		if (in6_dev->cnf.accept_ra_min_hop_limit <= ra_msg->icmph.icmp6_hop_limit) {
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index dd6ff5b20918..cfaf226ecf98 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -989,10 +989,20 @@ int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
 				 (rt->fib6_flags & ~RTF_PREF_MASK) | RTF_PREF(pref);
 
 	if (rt) {
-		if (!addrconf_finite_timeout(lifetime))
+		spin_lock_bh(&rt->fib6_table->tb6_lock);
+		if (!addrconf_finite_timeout(lifetime)) {
 			fib6_clean_expires(rt);
-		else
+			fib6_remove_gc_list(rt);
+		} else {
 			fib6_set_expires(rt, jiffies + HZ * lifetime);
+			/* If fib6_node is null, the f6i is just removed
+			 * from the table.
+			 */
+			if (rcu_dereference_protected(rt->fib6_node,
+						      lockdep_is_held(&rt->fib6_table->tb6_lock)))
+				fib6_add_gc_list(rt);
+		}
+		spin_unlock_bh(&rt->fib6_table->tb6_lock);
 
 		fib6_info_release(rt);
 	}
-- 
2.34.1

[PATCH net-next 4/5] net/ipv6: set expires in modify_prefix_route() if RTF_EXPIRES is set.

[thinker.li]

From: Kui-Feng Lee <thinker.li@gmail.com>

Make the decision to set or clean the expires of a route based on the
RTF_EXPIRES flag, rather than the value of the "expires" argument.

The function inet6_addr_modify() is the only caller of
modify_prefix_route(), and it passes the RTF_EXPIRES flag and an expiration
value. The RTF_EXPIRES flag is turned on or off based on the value of
valid_lft. The RTF_EXPIRES flag is turned on if valid_lft is a finite value
(not infinite, not 0xffffffff). Even if valid_lft is 0, the RTF_EXPIRES
flag remains on. The expiration value being passed is equal to the
valid_lft value if the flag is on. However, if the valid_lft value is
infinite, the expiration value becomes 0 and the RTF_EXPIRES flag is turned
off. Despite this, modify_prefix_route() decides to set the expiration
value if the received expiration value is not zero. This mixing of infinite
and zero cases creates an inconsistency.

Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
---
 net/ipv6/addrconf.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
index 36bfa987c314..2f6cf6314646 100644
--- a/net/ipv6/addrconf.c
+++ b/net/ipv6/addrconf.c
@@ -4788,7 +4788,7 @@ static int modify_prefix_route(struct inet6_ifaddr *ifp,
 	} else {
 		table = f6i->fib6_table;
 		spin_lock_bh(&table->tb6_lock);
-		if (!expires) {
+		if (!(flags & RTF_EXPIRES)) {
 			fib6_clean_expires(f6i);
 			fib6_remove_gc_list(f6i);
 		} else {
-- 
2.34.1

[PATCH net-next 5/5] selftests/net: Adding test cases of replacing routes and route advertisements.

[thinker.li]

From: Kui-Feng Lee <thinker.li@gmail.com>

Add tests of changing permanent routes to temporary routes and the reversed
case to make sure GC working correctly in these cases.  Add tests for the
temporary routes from RA.

The existing device will be deleted between tests to remove all routes
associated with it, so that the earlier tests don't mess up the later ones.

Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
---
 tools/testing/selftests/net/fib_tests.sh | 159 +++++++++++++++++++++--
 1 file changed, 148 insertions(+), 11 deletions(-)

diff --git a/tools/testing/selftests/net/fib_tests.sh b/tools/testing/selftests/net/fib_tests.sh
index b3ecccbbfcd2..f69b55304ebb 100755
--- a/tools/testing/selftests/net/fib_tests.sh
+++ b/tools/testing/selftests/net/fib_tests.sh
@@ -743,6 +743,16 @@ fib_notify_test()
 	cleanup &> /dev/null
 }
 
+# Create a new dummy_10 to remove all associated routes.
+reset_dummy_10()
+{
+	$IP link del dev dummy_10
+
+	$IP link add dummy_10 type dummy
+	$IP link set dev dummy_10 up
+	$IP -6 address add 2001:10::1/64 dev dummy_10
+}
+
 fib6_gc_test()
 {
 	setup
@@ -768,15 +778,19 @@ fib6_gc_test()
 	    $IP -6 route add 2001:20::$i \
 		via 2001:10::2 dev dummy_10 expires $EXPIRE
 	done
-	sleep $(($EXPIRE * 2))
-	N_EXP_SLEEP=$($IP -6 route list |grep expires|wc -l)
-	if [ $N_EXP_SLEEP -ne 0 ]; then
-	    echo "FAIL: expected 0 routes with expires, got $N_EXP_SLEEP"
+	sleep $(($EXPIRE * 2 + 1))
+	N_EXP=$($IP -6 route list |grep expires|wc -l)
+	if [ $N_EXP -ne 0 ]; then
+	    echo "FAIL: expected 0 routes with expires, got $N_EXP"
 	    ret=1
 	else
 	    ret=0
 	fi
 
+	log_test $ret 0 "ipv6 route garbage collection"
+
+	reset_dummy_10
+
 	# Permanent routes
 	for i in $(seq 1 5000); do
 	    $IP -6 route add 2001:30::$i \
@@ -788,19 +802,142 @@ fib6_gc_test()
 	    $IP -6 route add 2001:20::$i \
 		via 2001:10::2 dev dummy_10 expires $EXPIRE
 	done
-	sleep $(($EXPIRE * 2))
-	N_EXP_SLEEP=$($IP -6 route list |grep expires|wc -l)
-	if [ $N_EXP_SLEEP -ne 0 ]; then
-	    echo "FAIL: expected 0 routes with expires," \
-		 "got $N_EXP_SLEEP (5000 permanent routes)"
+	sleep $(($EXPIRE * 2 + 1))
+	N_EXP=$($IP -6 route list |grep expires|wc -l)
+	if [ $N_EXP -ne 0 ]; then
+	    echo "FAIL: expected 0 routes with expires, got $N_EXP"
 	    ret=1
 	else
 	    ret=0
 	fi
 
-	set +e
+	log_test $ret 0 "ipv6 route garbage collection (with permanent routes)"
 
-	log_test $ret 0 "ipv6 route garbage collection"
+	reset_dummy_10
+
+	# Permanent routes
+	for i in $(seq 1 100); do
+	    $IP -6 route add 2001:20::$i \
+		via 2001:10::2 dev dummy_10
+	done
+	# Replace with temporary routes
+	for i in $(seq 1 100); do
+	    # Expire route after $EXPIRE seconds
+	    $IP -6 route replace 2001:20::$i \
+		via 2001:10::2 dev dummy_10 expires $EXPIRE
+	done
+	N_EXP=$($IP -6 route list |grep expires|wc -l)
+	if [ $N_EXP -ne 100 ]; then
+	    log_test 1 0 "expected 100 routes with expires, got $N_EXP"
+	    set +e
+	    cleanup &> /dev/null
+	    return
+	fi
+	# Wait for GC
+	sleep $(($EXPIRE * 2 + 1))
+	N_EXP=$($IP -6 route list |grep expires|wc -l)
+	if [ $N_EXP -ne 0 ]; then
+	    echo "FAIL: expected 0 routes with expires, got $N_EXP"
+	    ret=1
+	else
+	    ret=0
+	fi
+
+	log_test $ret 0 "ipv6 route garbage collection (replace with expires)"
+
+	reset_dummy_10
+
+	# Temporary routes
+	for i in $(seq 1 100); do
+	    # Expire route after $EXPIRE seconds
+	    $IP -6 route add 2001:20::$i \
+		via 2001:10::2 dev dummy_10 expires $EXPIRE
+	done
+	# Replace with permanent routes
+	for i in $(seq 1 100); do
+	    $IP -6 route replace 2001:20::$i \
+		via 2001:10::2 dev dummy_10
+	done
+	N_EXP=$($IP -6 route list |grep expires|wc -l)
+	if [ $N_EXP -ne 0 ]; then
+	    log_test 1 0 "expected 0 routes with expires, got $N_EXP"
+	    set +e
+	    cleanup &> /dev/null
+	    return
+	fi
+
+	# Wait for GC
+	sleep $(($EXPIRE * 2 + 1))
+
+	N_PERM=$($IP -6 route list |grep -v expires|grep 2001:20::|wc -l)
+	if [ $N_PERM -ne 100 ]; then
+	    echo "FAIL: expected 100 permanent routes, got $N_PERM"
+	    ret=1
+	else
+	    ret=0
+	fi
+
+	log_test $ret 0 "ipv6 route garbage collection (replace with permanent)"
+
+	# ra6 is required for the next test. (ipv6toolkit)
+	if [ ! -x "$(command -v ra6)" ]; then
+	    echo "SKIP: ra6 not found."
+	    set +e
+	    cleanup &> /dev/null
+	    return
+	fi
+
+	# Delete dummy_10 and remove all routes
+	$IP link del dev dummy_10
+
+	# Create a pair of veth devices to send a RA message from one
+	# device to another.
+	$IP link add veth1 type veth peer name veth2
+	$IP link set dev veth1 up
+	$IP link set dev veth2 up
+	$IP -6 address add 2001:10::1/64 dev veth1 nodad
+	$IP -6 address add 2001:10::2/64 dev veth2 nodad
+
+	# Without stopping these two services, systemd may mess up the test
+	# by intercepting the RA message and adding routes.
+	if [ -x "$(command -v systemctl)" ]; then
+	    systemctl stop systemd-networkd.socket
+	    systemctl stop systemd-networkd.service
+	fi
+	# Make veth1 ready to receive RA messages.
+	$NS_EXEC sysctl -w net.ipv6.conf.veth1.accept_ra=2 &> /dev/null
+	$NS_EXEC sysctl -w net.ipv6.conf.veth1.accept_ra_rt_info_max_plen=127 &> /dev/null
+
+	# Send a RA message with a route from veth2 to veth1.
+	$NS_EXEC ra6 -i veth2 -d 2001:10::1 -R "2003:10::/64#1#$EXPIRE" -t $EXPIRE
+
+	# Wait for the RA message.
+	sleep 1
+
+	# There are 2 routes with expires. One is a default route and the
+	# other is the route to 2003:10::/64.
+	N_EXP=$($IP -6 route list |grep expires|wc -l)
+	if [ $N_EXP -ne 2 ]; then
+	    log_test 1 0 "expected 2 routes with expires, got $N_EXP"
+	    set +e
+	    cleanup &> /dev/null
+	    return
+	fi
+
+	# Wait for GC
+	sleep $(($EXPIRE * 2 + 1))
+
+	N_EXP=$($IP -6 route list |grep expires|wc -l)
+	if [ $N_EXP -ne 0 ]; then
+	    echo "FAIL: expected 0 routes with expires, got $N_EXP"
+	    ret=1
+	else
+	    ret=0
+	fi
+
+	log_test $ret 0 "ipv6 route garbage collection (RA message)"
+
+	set +e
 
 	cleanup &> /dev/null
 }
-- 
2.34.1

Re: [PATCH net-next 5/5] selftests/net: Adding test cases of replacing routes and route advertisements.

[Jakub Kicinski]

On Tue, 30 Jan 2024 22:40:41 -0800 thinker.li@gmail.com wrote:
> +	N_PERM=$($IP -6 route list |grep -v expires|grep 2001:20::|wc -l)
> +	if [ $N_PERM -ne 100 ]; then
> +	    echo "FAIL: expected 100 permanent routes, got $N_PERM"
> +	    ret=1
> +	else
> +	    ret=0
> +	fi

This fails on our slow VM:

#     TEST: expected 100 routes with expires, got 98                      [FAIL]

on a VM with kernel debug enable the test times out (it doesn't print
anything for 10min):

https://netdev-3.bots.linux.dev/vmksft-net-dbg/results/445222/6-fib-tests-sh/stdout
-- 
pw-bot: cr

Re: [PATCH net-next 5/5] selftests/net: Adding test cases of replacing routes and route advertisements.

[Kui-Feng Lee]

On 1/31/24 11:20, Jakub Kicinski wrote:
> On Tue, 30 Jan 2024 22:40:41 -0800 thinker.li@gmail.com wrote:
>> +	N_PERM=$($IP -6 route list |grep -v expires|grep 2001:20::|wc -l)
>> +	if [ $N_PERM -ne 100 ]; then
>> +	    echo "FAIL: expected 100 permanent routes, got $N_PERM"
>> +	    ret=1
>> +	else
>> +	    ret=0
>> +	fi
> 
> This fails on our slow VM:
> 
> #     TEST: expected 100 routes with expires, got 98                      [FAIL]
> 
> on a VM with kernel debug enable the test times out (it doesn't print
> anything for 10min):
> 
> https://netdev-3.bots.linux.dev/vmksft-net-dbg/results/445222/6-fib-tests-sh/stdout

I will reduce the number of routes to 100 instead of 5000 to see if it
helps.

Re: [PATCH net-next 5/5] selftests/net: Adding test cases of replacing routes and route advertisements.

[Hangbin Liu]

Hi,

On Tue, Jan 30, 2024 at 10:40:41PM -0800, thinker.li@gmail.com wrote:
> +# Create a new dummy_10 to remove all associated routes.
> +reset_dummy_10()
> +{
> +	$IP link del dev dummy_10
> +
> +	$IP link add dummy_10 type dummy
> +	$IP link set dev dummy_10 up
> +	$IP -6 address add 2001:10::1/64 dev dummy_10
> +}
> +
>  fib6_gc_test()
>  {
>  	setup
> @@ -768,15 +778,19 @@ fib6_gc_test()
>  	    $IP -6 route add 2001:20::$i \
>  		via 2001:10::2 dev dummy_10 expires $EXPIRE
>  	done
> -	sleep $(($EXPIRE * 2))
> -	N_EXP_SLEEP=$($IP -6 route list |grep expires|wc -l)
> -	if [ $N_EXP_SLEEP -ne 0 ]; then
> -	    echo "FAIL: expected 0 routes with expires, got $N_EXP_SLEEP"
> +	sleep $(($EXPIRE * 2 + 1))
> +	N_EXP=$($IP -6 route list |grep expires|wc -l)
> +	if [ $N_EXP -ne 0 ]; then
> +	    echo "FAIL: expected 0 routes with expires, got $N_EXP"
>  	    ret=1
>  	else
>  	    ret=0
>  	fi
>  
> +	log_test $ret 0 "ipv6 route garbage collection"
> +
> +	reset_dummy_10

Since you reset the dummy device and will not affect the later tests. Maybe
you can log the test directly, e.g.

	if [ "$($IP -6 route list |grep expires|wc -l)" -ne 0 ]; then
		log_test $ret 0 "ipv6 route garbage collection"
	fi

Or, if you want to keep ret and also report passed log, you can wrapper the
number checking like

check_exp_number()
{
	local exp=$1
	local n_exp=$($IP -6 route list |grep expires|wc -l)
	if [ "$n_exp" -ne "$exp" ]; then
		echo "FAIL: expected $exp routes with expires, got $n_exp"
		ret=1
	else
		ret=0
	fi
}

Then we can call it without repeating the if/else lines

	check_exp_number 0
	log_test $ret 0 "ipv6 route garbage collection"

Thanks
Hangbin

Re: [PATCH net-next 5/5] selftests/net: Adding test cases of replacing routes and route advertisements.

[Kui-Feng Lee]

On 2/1/24 00:46, Hangbin Liu wrote:
> Hi,
> 
> On Tue, Jan 30, 2024 at 10:40:41PM -0800, thinker.li@gmail.com wrote:
>> +# Create a new dummy_10 to remove all associated routes.
>> +reset_dummy_10()
>> +{
>> +	$IP link del dev dummy_10
>> +
>> +	$IP link add dummy_10 type dummy
>> +	$IP link set dev dummy_10 up
>> +	$IP -6 address add 2001:10::1/64 dev dummy_10
>> +}
>> +
>>   fib6_gc_test()
>>   {
>>   	setup
>> @@ -768,15 +778,19 @@ fib6_gc_test()
>>   	    $IP -6 route add 2001:20::$i \
>>   		via 2001:10::2 dev dummy_10 expires $EXPIRE
>>   	done
>> -	sleep $(($EXPIRE * 2))
>> -	N_EXP_SLEEP=$($IP -6 route list |grep expires|wc -l)
>> -	if [ $N_EXP_SLEEP -ne 0 ]; then
>> -	    echo "FAIL: expected 0 routes with expires, got $N_EXP_SLEEP"
>> +	sleep $(($EXPIRE * 2 + 1))
>> +	N_EXP=$($IP -6 route list |grep expires|wc -l)
>> +	if [ $N_EXP -ne 0 ]; then
>> +	    echo "FAIL: expected 0 routes with expires, got $N_EXP"
>>   	    ret=1
>>   	else
>>   	    ret=0
>>   	fi
>>   
>> +	log_test $ret 0 "ipv6 route garbage collection"
>> +
>> +	reset_dummy_10
> 
> Since you reset the dummy device and will not affect the later tests. Maybe
> you can log the test directly, e.g.
> 
> 	if [ "$($IP -6 route list |grep expires|wc -l)" -ne 0 ]; then
> 		log_test $ret 0 "ipv6 route garbage collection"
> 	fi
> 
> Or, if you want to keep ret and also report passed log, you can wrapper the
> number checking like
> 
> check_exp_number()
> {
> 	local exp=$1
> 	local n_exp=$($IP -6 route list |grep expires|wc -l)
> 	if [ "$n_exp" -ne "$exp" ]; then
> 		echo "FAIL: expected $exp routes with expires, got $n_exp"
> 		ret=1
> 	else
> 		ret=0
> 	fi
> }
> 
> Then we can call it without repeating the if/else lines
> 
> 	check_exp_number 0
> 	log_test $ret 0 "ipv6 route garbage collection"

If I read it correctly, the point here is too many boilerplate checks,
and you prefer to reduce them. Right?
No problem! I will do it.


> 
> Thanks
> Hangbin

Re: [PATCH net-next 5/5] selftests/net: Adding test cases of replacing routes and route advertisements.

[Hangbin Liu]

On Thu, Feb 01, 2024 at 09:14:17AM -0800, Kui-Feng Lee wrote:
> > > +	N_EXP=$($IP -6 route list |grep expires|wc -l)
> > > +	if [ $N_EXP -ne 0 ]; then
> > > +	    echo "FAIL: expected 0 routes with expires, got $N_EXP"
> > >   	    ret=1
> > >   	else
> > >   	    ret=0
> > >   	fi
> > > +	log_test $ret 0 "ipv6 route garbage collection"
> > > +
> > > +	reset_dummy_10
> > 
> > Since you reset the dummy device and will not affect the later tests. Maybe
> > you can log the test directly, e.g.
> > 
> > 	if [ "$($IP -6 route list |grep expires|wc -l)" -ne 0 ]; then
> > 		log_test $ret 0 "ipv6 route garbage collection"
> > 	fi
> > 
> > Or, if you want to keep ret and also report passed log, you can wrapper the
> > number checking like
> > 
> > check_exp_number()
> > {
> > 	local exp=$1
> > 	local n_exp=$($IP -6 route list |grep expires|wc -l)
> > 	if [ "$n_exp" -ne "$exp" ]; then
> > 		echo "FAIL: expected $exp routes with expires, got $n_exp"
> > 		ret=1
> > 	else
> > 		ret=0
> > 	fi
> > }
> > 
> > Then we can call it without repeating the if/else lines
> > 
> > 	check_exp_number 0
> > 	log_test $ret 0 "ipv6 route garbage collection"
> 
> If I read it correctly, the point here is too many boilerplate checks,
> and you prefer to reduce them. Right?
> No problem! I will do it.

Yes, thanks!

Hangbin