[PATCH v3 net-next 13/15] ipv6: Defer fib6_purge_rt() in fib6_add_rt2node() to fib6_add().

[Kuniyuki Iwashima <kuniyu@amazon.com>]

The next patch adds per-nexthop spinlock which protects nh->f6i_list.

When rt->nh is not NULL, fib6_add_rt2node() will be called under the lock.
fib6_add_rt2node() could call fib6_purge_rt() for another route, which
could holds another nexthop lock.

Then, deadlock could happen between two nexthops.

Let's defer fib6_purge_rt() after fib6_add_rt2node().

Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
---
 include/net/ip6_fib.h |  1 +
 net/ipv6/ip6_fib.c    | 21 ++++++++++++++-------
 2 files changed, 15 insertions(+), 7 deletions(-)

diff --git a/include/net/ip6_fib.h b/include/net/ip6_fib.h
index 7c87873ae211..88b0dd4d8e09 100644
--- a/include/net/ip6_fib.h
+++ b/include/net/ip6_fib.h
@@ -198,6 +198,7 @@ struct fib6_info {
 					fib6_destroying:1,
 					unused:4;
 
+	struct list_head		purge_link;
 	struct rcu_head			rcu;
 	struct nexthop			*nh;
 	struct fib6_nh			fib6_nh[];
diff --git a/net/ipv6/ip6_fib.c b/net/ipv6/ip6_fib.c
index 79b672f3fc53..9e9db5470bbf 100644
--- a/net/ipv6/ip6_fib.c
+++ b/net/ipv6/ip6_fib.c
@@ -1083,8 +1083,8 @@ static void fib6_purge_rt(struct fib6_info *rt, struct fib6_node *fn,
  */
 
 static int fib6_add_rt2node(struct fib6_node *fn, struct fib6_info *rt,
-			    struct nl_info *info,
-			    struct netlink_ext_ack *extack)
+			    struct nl_info *info, struct netlink_ext_ack *extack,
+			    struct list_head *purge_list)
 {
 	struct fib6_info *leaf = rcu_dereference_protected(fn->leaf,
 				    lockdep_is_held(&rt->fib6_table->tb6_lock));
@@ -1308,10 +1308,9 @@ static int fib6_add_rt2node(struct fib6_node *fn, struct fib6_info *rt,
 		}
 		nsiblings = iter->fib6_nsiblings;
 		iter->fib6_node = NULL;
-		fib6_purge_rt(iter, fn, info->nl_net);
+		list_add(&iter->purge_link, purge_list);
 		if (rcu_access_pointer(fn->rr_ptr) == iter)
 			fn->rr_ptr = NULL;
-		fib6_info_release(iter);
 
 		if (nsiblings) {
 			/* Replacing an ECMP route, remove all siblings */
@@ -1324,10 +1323,9 @@ static int fib6_add_rt2node(struct fib6_node *fn, struct fib6_info *rt,
 				if (rt6_qualify_for_ecmp(iter)) {
 					*ins = iter->fib6_next;
 					iter->fib6_node = NULL;
-					fib6_purge_rt(iter, fn, info->nl_net);
+					list_add(&iter->purge_link, purge_list);
 					if (rcu_access_pointer(fn->rr_ptr) == iter)
 						fn->rr_ptr = NULL;
-					fib6_info_release(iter);
 					nsiblings--;
 					info->nl_net->ipv6.rt6_stats->fib_rt_entries--;
 				} else {
@@ -1397,6 +1395,7 @@ int fib6_add(struct fib6_node *root, struct fib6_info *rt,
 	     struct nl_info *info, struct netlink_ext_ack *extack)
 {
 	struct fib6_table *table = rt->fib6_table;
+	LIST_HEAD(purge_list);
 	struct fib6_node *fn;
 #ifdef CONFIG_IPV6_SUBTREES
 	struct fib6_node *pn = NULL;
@@ -1499,8 +1498,16 @@ int fib6_add(struct fib6_node *root, struct fib6_info *rt,
 	}
 #endif
 
-	err = fib6_add_rt2node(fn, rt, info, extack);
+	err = fib6_add_rt2node(fn, rt, info, extack, &purge_list);
 	if (!err) {
+		struct fib6_info *iter, *next;
+
+		list_for_each_entry_safe(iter, next, &purge_list, purge_link) {
+			list_del(&iter->purge_link);
+			fib6_purge_rt(iter, fn, info->nl_net);
+			fib6_info_release(iter);
+		}
+
 		if (rt->nh)
 			list_add(&rt->nh_list, &rt->nh->f6i_list);
 		__fib6_update_sernum_upto_root(rt, fib6_new_sernum(info->nl_net));
-- 
2.49.0