From 32b293a53deeb220769f9a29357cb151cfb8ee26 Mon Sep 17 00:00:00 2001
From: Josh Hunt <joshhunt00@gmail.com>
Date: Wed, 28 Dec 2011 13:23:07 +0000
Subject: IPv6: Avoid taking write lock for /proc/net/ipv6_route

During some debugging I needed to look into how /proc/net/ipv6_route
operated and in my digging I found its calling fib6_clean_all() which uses
"write_lock_bh(&table->tb6_lock)" before doing the walk of the table. I
found this on 2.6.32, but reading the code I believe the same basic idea
exists currently. Looking at the rtnetlink code they are only calling
"read_lock_bh(&table->tb6_lock);" via fib6_dump_table(). While I realize
reading from proc isn't the recommended way of fetching the ipv6 route
table; taking a write lock seems unnecessary and would probably cause
network performance issues.

To verify this I loaded up the ipv6 route table and then ran iperf in 3
cases:
  * doing nothing
  * reading ipv6 route table via proc
    (while :; do cat /proc/net/ipv6_route > /dev/null; done)
  * reading ipv6 route table via rtnetlink
    (while :; do ip -6 route show table all > /dev/null; done)

* Load the ipv6 route table up with:
  * for ((i = 0;i < 4000;i++)); do ip route add unreachable 2000::$i; done

* iperf commands:
  * client: iperf -i 1 -V -c <ipv6 addr>
  * server: iperf -V -s

* iperf results - 3 runs each (in Mbits/sec)
  * nothing: client: 927,927,927 server: 927,927,927
  * proc: client: 179,97,96,113 server: 142,112,133
  * iproute: client: 928,927,928 server: 927,927,927

lock_stat shows taking the write lock is causing the slowdown. Using this
info I decided to write a version of fib6_clean_all() which replaces
write_lock_bh(&table->tb6_lock) with read_lock_bh(&table->tb6_lock). With
this new function I see the same results as with my rtnetlink iperf test.

Signed-off-by: Josh Hunt <joshhunt00@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/include/net/ip6_fib.h b/include/net/ip6_fib.h
index 1e8a89f..b26bb81 100644
--- a/include/net/ip6_fib.h
+++ b/include/net/ip6_fib.h
@@ -199,6 +199,10 @@ struct fib6_node		*fib6_locate(struct fib6_node *root,
 					     const struct in6_addr *daddr, int dst_len,
 					     const struct in6_addr *saddr, int src_len);
 
+extern void			fib6_clean_all_ro(struct net *net,
+					       int (*func)(struct rt6_info *, void *arg),
+					       int prune, void *arg);
+
 extern void			fib6_clean_all(struct net *net,
 					       int (*func)(struct rt6_info *, void *arg),
 					       int prune, void *arg);
diff --git a/net/ipv6/ip6_fib.c b/net/ipv6/ip6_fib.c
index 246d8e4..b82bcde 100644
--- a/net/ipv6/ip6_fib.c
+++ b/net/ipv6/ip6_fib.c
@@ -1462,6 +1462,26 @@ static void fib6_clean_tree(struct net *net, struct fib6_node *root,
 	fib6_walk(&c.w);
 }
 
+void fib6_clean_all_ro(struct net *net, int (*func)(struct rt6_info *, void *arg),
+		    int prune, void *arg)
+{
+	struct fib6_table *table;
+	struct hlist_node *node;
+	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, node, head, tb6_hlist) {
+			read_lock_bh(&table->tb6_lock);
+			fib6_clean_tree(net, &table->tb6_root,
+					func, prune, arg);
+			read_unlock_bh(&table->tb6_lock);
+		}
+	}
+	rcu_read_unlock();
+}
 void fib6_clean_all(struct net *net, int (*func)(struct rt6_info *, void *arg),
 		    int prune, void *arg)
 {
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index 4a62c47..07361df 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -2680,7 +2680,7 @@ static int rt6_info_route(struct rt6_info *rt, void *p_arg)
 static int ipv6_route_show(struct seq_file *m, void *v)
 {
 	struct net *net = (struct net *)m->private;
-	fib6_clean_all(net, rt6_info_route, 0, m);
+	fib6_clean_all_ro(net, rt6_info_route, 0, m);
 	return 0;
 }
 
-- 
cgit v0.10.2