ipv4: Allow amount of dirty memory from fib resizing to be controllable

fib_trie implementation calls synchronize_rcu when a certain amount of
pages are dirty from freed entries. The number of pages was determined
experimentally in 2009 (commit c3059477fce2d).

At the current setting, synchronize_rcu is called often -- 51 times in a
second in one test with an average of an 8 msec delay adding a fib entry.
The total impact is a lot of slow down modifying the fib. This is seen
in the output of 'time' - the difference between real time and sys+user.
For example, using 720,022 single path routes and 'ip -batch'[1]:

    $ time ./ip -batch ipv4/routes-1-hops
    real    0m14.214s
    user    0m2.513s
    sys     0m6.783s

So roughly 35% of the actual time to install the routes is from the ip
command getting scheduled out, most notably due to synchronize_rcu (this
is observed using 'perf sched timehist').

This patch makes the amount of dirty memory configurable between 64k where
the synchronize_rcu is called often (small, low end systems that are memory
sensitive) to 64M where synchronize_rcu is called rarely during a large
FIB change (for high end systems with lots of memory). The default is 512kB
which corresponds to the current setting of 128 pages with a 4kB page size.

As an example, at 16MB the worst interval shows 4 calls to synchronize_rcu
in a second blocking for up to 30 msec in a single instance, and a total
of almost 100 msec across the 4 calls in the second. The trade off is
allowing FIB entries to consume more memory in a given time window but
but with much better fib insertion rates (~30% increase in prefixes/sec).
With this patch and net.ipv4.fib_sync_mem set to 16MB, the same batch
file runs in:

    $ time ./ip -batch ipv4/routes-1-hops
    real    0m9.692s
    user    0m2.491s
    sys     0m6.769s

So the dead time is reduced to about 1/2 second or <5% of the real time.

[1] 'ip' modified to not request ACK messages which improves route
    insertion times by about 20%

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/Documentation/networking/ip-sysctl.txt b/Documentation/networking/ip-sysctl.txt
index bd029fc55ccb047766cac5f10b8f07acac0243c6..5eedc6941ce502b4dd2019341185c87922282040 100644 (file)
--- a/Documentation/networking/ip-sysctl.txt
+++ b/Documentation/networking/ip-sysctl.txt
@@ -81,6 +81,11 @@ fib_multipath_hash_policy - INTEGER
        0 - Layer 3
        1 - Layer 4
 
+fib_sync_mem - UNSIGNED INTEGER
+       Amount of dirty memory from fib entries that can be backlogged before
+       synchronize_rcu is forced.
+         Default: 512kB   Minimum: 64kB   Maximum: 64MB
+
 ip_forward_update_priority - INTEGER
        Whether to update SKB priority from "TOS" field in IPv4 header after it
        is forwarded. The new SKB priority is mapped from TOS field value

diff --git a/include/net/ip.h b/include/net/ip.h
index be3cad9c2e4c37b282e5c2d0e2ef5f05a79b7438..aa09ae5f01a58c5dad2aae2d697a15f3e606f7c8 100644 (file)
--- a/include/net/ip.h
+++ b/include/net/ip.h
@@ -38,6 +38,10 @@
 #define IPV4_MAX_PMTU          65535U          /* RFC 2675, Section 5.1 */
 #define IPV4_MIN_MTU           68                      /* RFC 791 */
 
+extern unsigned int sysctl_fib_sync_mem;
+extern unsigned int sysctl_fib_sync_mem_min;
+extern unsigned int sysctl_fib_sync_mem_max;
+
 struct sock;
 
 struct inet_skb_parm {

diff --git a/net/ipv4/fib_trie.c b/net/ipv4/fib_trie.c
index a573e37e0615b1b1b5093ae3ada97f07c44f27cb..1704f432de1fcef715ac5ad0ff7849da9e1418c5 100644 (file)
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -183,14 +183,16 @@ struct trie {
 };
 
 static struct key_vector *resize(struct trie *t, struct key_vector *tn);
-static size_t tnode_free_size;
+static unsigned int tnode_free_size;
 
 /*
- * synchronize_rcu after call_rcu for that many pages; it should be especially
- * useful before resizing the root node with PREEMPT_NONE configs; the value was
- * obtained experimentally, aiming to avoid visible slowdown.
+ * synchronize_rcu after call_rcu for outstanding dirty memory; it should be
+ * especially useful before resizing the root node with PREEMPT_NONE configs;
+ * the value was obtained experimentally, aiming to avoid visible slowdown.
  */
-static const int sync_pages = 128;
+unsigned int sysctl_fib_sync_mem = 512 * 1024;
+unsigned int sysctl_fib_sync_mem_min = 64 * 1024;
+unsigned int sysctl_fib_sync_mem_max = 64 * 1024 * 1024;
 
 static struct kmem_cache *fn_alias_kmem __ro_after_init;
 static struct kmem_cache *trie_leaf_kmem __ro_after_init;
@@ -504,7 +506,7 @@ static void tnode_free(struct key_vector *tn)
                tn = container_of(head, struct tnode, rcu)->kv;
        }
 
-       if (tnode_free_size >= PAGE_SIZE * sync_pages) {
+       if (tnode_free_size >= sysctl_fib_sync_mem) {
                tnode_free_size = 0;
                synchronize_rcu();
        }

diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index ba0fc4b1846561559ac995a444992f98a3187894..2316c08e9591736054bca54c91c304810ed24d89 100644 (file)
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -549,6 +549,15 @@ static struct ctl_table ipv4_table[] = {
                .mode           = 0644,
                .proc_handler   = proc_doulongvec_minmax,
        },
+       {
+               .procname       = "fib_sync_mem",
+               .data           = &sysctl_fib_sync_mem,
+               .maxlen         = sizeof(sysctl_fib_sync_mem),
+               .mode           = 0644,
+               .proc_handler   = proc_douintvec_minmax,
+               .extra1         = &sysctl_fib_sync_mem_min,
+               .extra2         = &sysctl_fib_sync_mem_max,
+       },
        { }
 };