From: ebiederm@xmission•com (Eric W. Biederman)
To: David Miller <davem@davemloft•net>
Cc: edumazet@google•com, netdev@vger•kernel.org,
stephen@networkplumber•org, nicolas.dichtel@6wind•com,
roopa@cumulusnetworks•com, hannes@stressinduktion•org,
ddutt@cumulusnetworks•com, vipin@cumulusnetworks•com,
shmulik.ladkani@gmail•com, dsahern@gmail•com,
Julian Anastasov <ja@ssi•bg>
Subject: [PATCH net-next 8/8] tcp_metrics: Use a single hash table for all network namespaces.
Date: Wed, 11 Mar 2015 11:43:49 -0500 [thread overview]
Message-ID: <87h9treb7e.fsf_-_@x220.int.ebiederm.org> (raw)
In-Reply-To: <87vbi7fq92.fsf_-_@x220.int.ebiederm.org> (Eric W. Biederman's message of "Wed, 11 Mar 2015 11:33:29 -0500")
Now that all of the operations are safe on a single hash table accross
network namespaces, allocate a single global hash table and update the
code to use it.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission•com>
---
include/net/netns/ipv4.h | 2 --
net/ipv4/tcp_metrics.c | 66 ++++++++++++++++++++++--------------------------
2 files changed, 30 insertions(+), 38 deletions(-)
diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h
index 8f3a1a1a5a94..614a49be68a9 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -54,8 +54,6 @@ struct netns_ipv4 {
struct sock *mc_autojoin_sk;
struct inet_peer_base *peers;
- struct tcpm_hash_bucket *tcp_metrics_hash;
- unsigned int tcp_metrics_hash_log;
struct sock * __percpu *tcp_sk;
struct netns_frags frags;
#ifdef CONFIG_NETFILTER
diff --git a/net/ipv4/tcp_metrics.c b/net/ipv4/tcp_metrics.c
index baccb070427d..366728cbee4a 100644
--- a/net/ipv4/tcp_metrics.c
+++ b/net/ipv4/tcp_metrics.c
@@ -97,6 +97,9 @@ struct tcpm_hash_bucket {
struct tcp_metrics_block __rcu *chain;
};
+static struct tcpm_hash_bucket *tcp_metrics_hash __read_mostly;
+static unsigned int tcp_metrics_hash_log __read_mostly;
+
static DEFINE_SPINLOCK(tcp_metrics_lock);
static void tcpm_suck_dst(struct tcp_metrics_block *tm,
@@ -177,7 +180,7 @@ static struct tcp_metrics_block *tcpm_new(struct dst_entry *dst,
if (unlikely(reclaim)) {
struct tcp_metrics_block *oldest;
- oldest = rcu_dereference(net->ipv4.tcp_metrics_hash[hash].chain);
+ oldest = rcu_dereference(tcp_metrics_hash[hash].chain);
for (tm = rcu_dereference(oldest->tcpm_next); tm;
tm = rcu_dereference(tm->tcpm_next)) {
if (time_before(tm->tcpm_stamp, oldest->tcpm_stamp))
@@ -196,8 +199,8 @@ static struct tcp_metrics_block *tcpm_new(struct dst_entry *dst,
tcpm_suck_dst(tm, dst, true);
if (likely(!reclaim)) {
- tm->tcpm_next = net->ipv4.tcp_metrics_hash[hash].chain;
- rcu_assign_pointer(net->ipv4.tcp_metrics_hash[hash].chain, tm);
+ tm->tcpm_next = tcp_metrics_hash[hash].chain;
+ rcu_assign_pointer(tcp_metrics_hash[hash].chain, tm);
}
out_unlock:
@@ -221,7 +224,7 @@ static struct tcp_metrics_block *__tcp_get_metrics(const struct inetpeer_addr *s
struct tcp_metrics_block *tm;
int depth = 0;
- for (tm = rcu_dereference(net->ipv4.tcp_metrics_hash[hash].chain); tm;
+ for (tm = rcu_dereference(tcp_metrics_hash[hash].chain); tm;
tm = rcu_dereference(tm->tcpm_next)) {
if (addr_same(&tm->tcpm_saddr, saddr) &&
addr_same(&tm->tcpm_daddr, daddr) &&
@@ -261,9 +264,9 @@ static struct tcp_metrics_block *__tcp_get_metrics_req(struct request_sock *req,
net = dev_net(dst->dev);
hash ^= net_hash_mix(net);
- hash = hash_32(hash, net->ipv4.tcp_metrics_hash_log);
+ hash = hash_32(hash, tcp_metrics_hash_log);
- for (tm = rcu_dereference(net->ipv4.tcp_metrics_hash[hash].chain); tm;
+ for (tm = rcu_dereference(tcp_metrics_hash[hash].chain); tm;
tm = rcu_dereference(tm->tcpm_next)) {
if (addr_same(&tm->tcpm_saddr, &saddr) &&
addr_same(&tm->tcpm_daddr, &daddr) &&
@@ -310,9 +313,9 @@ static struct tcp_metrics_block *__tcp_get_metrics_tw(struct inet_timewait_sock
net = twsk_net(tw);
hash ^= net_hash_mix(net);
- hash = hash_32(hash, net->ipv4.tcp_metrics_hash_log);
+ hash = hash_32(hash, tcp_metrics_hash_log);
- for (tm = rcu_dereference(net->ipv4.tcp_metrics_hash[hash].chain); tm;
+ for (tm = rcu_dereference(tcp_metrics_hash[hash].chain); tm;
tm = rcu_dereference(tm->tcpm_next)) {
if (addr_same(&tm->tcpm_saddr, &saddr) &&
addr_same(&tm->tcpm_daddr, &daddr) &&
@@ -360,7 +363,7 @@ static struct tcp_metrics_block *tcp_get_metrics(struct sock *sk,
net = dev_net(dst->dev);
hash ^= net_hash_mix(net);
- hash = hash_32(hash, net->ipv4.tcp_metrics_hash_log);
+ hash = hash_32(hash, tcp_metrics_hash_log);
tm = __tcp_get_metrics(&saddr, &daddr, net, hash);
if (tm == TCP_METRICS_RECLAIM_PTR)
@@ -911,13 +914,13 @@ static int tcp_metrics_nl_dump(struct sk_buff *skb,
struct netlink_callback *cb)
{
struct net *net = sock_net(skb->sk);
- unsigned int max_rows = 1U << net->ipv4.tcp_metrics_hash_log;
+ unsigned int max_rows = 1U << tcp_metrics_hash_log;
unsigned int row, s_row = cb->args[0];
int s_col = cb->args[1], col = s_col;
for (row = s_row; row < max_rows; row++, s_col = 0) {
struct tcp_metrics_block *tm;
- struct tcpm_hash_bucket *hb = net->ipv4.tcp_metrics_hash + row;
+ struct tcpm_hash_bucket *hb = tcp_metrics_hash + row;
rcu_read_lock();
for (col = 0, tm = rcu_dereference(hb->chain); tm;
@@ -1010,10 +1013,10 @@ static int tcp_metrics_nl_cmd_get(struct sk_buff *skb, struct genl_info *info)
goto nla_put_failure;
hash ^= net_hash_mix(net);
- hash = hash_32(hash, net->ipv4.tcp_metrics_hash_log);
+ hash = hash_32(hash, tcp_metrics_hash_log);
ret = -ESRCH;
rcu_read_lock();
- for (tm = rcu_dereference(net->ipv4.tcp_metrics_hash[hash].chain); tm;
+ for (tm = rcu_dereference(tcp_metrics_hash[hash].chain); tm;
tm = rcu_dereference(tm->tcpm_next)) {
if (addr_same(&tm->tcpm_daddr, &daddr) &&
(!src || addr_same(&tm->tcpm_saddr, &saddr)) &&
@@ -1045,8 +1048,8 @@ out_free:
static void tcp_metrics_flush_all(struct net *net)
{
- unsigned int max_rows = 1U << net->ipv4.tcp_metrics_hash_log;
- struct tcpm_hash_bucket *hb = net->ipv4.tcp_metrics_hash;
+ unsigned int max_rows = 1U << tcp_metrics_hash_log;
+ struct tcpm_hash_bucket *hb = tcp_metrics_hash;
struct tcp_metrics_block *tm;
unsigned int row;
@@ -1090,8 +1093,8 @@ static int tcp_metrics_nl_cmd_del(struct sk_buff *skb, struct genl_info *info)
src = false;
hash ^= net_hash_mix(net);
- hash = hash_32(hash, net->ipv4.tcp_metrics_hash_log);
- hb = net->ipv4.tcp_metrics_hash + hash;
+ hash = hash_32(hash, tcp_metrics_hash_log);
+ hb = tcp_metrics_hash + hash;
pp = &hb->chain;
spin_lock_bh(&tcp_metrics_lock);
for (tm = deref_locked_genl(*pp); tm; tm = deref_locked_genl(*pp)) {
@@ -1147,6 +1150,9 @@ static int __net_init tcp_net_metrics_init(struct net *net)
size_t size;
unsigned int slots;
+ if (!net_eq(net, &init_net))
+ return 0;
+
slots = tcpmhash_entries;
if (!slots) {
if (totalram_pages >= 128 * 1024)
@@ -1155,14 +1161,14 @@ static int __net_init tcp_net_metrics_init(struct net *net)
slots = 8 * 1024;
}
- net->ipv4.tcp_metrics_hash_log = order_base_2(slots);
- size = sizeof(struct tcpm_hash_bucket) << net->ipv4.tcp_metrics_hash_log;
+ tcp_metrics_hash_log = order_base_2(slots);
+ size = sizeof(struct tcpm_hash_bucket) << tcp_metrics_hash_log;
- net->ipv4.tcp_metrics_hash = kzalloc(size, GFP_KERNEL | __GFP_NOWARN);
- if (!net->ipv4.tcp_metrics_hash)
- net->ipv4.tcp_metrics_hash = vzalloc(size);
+ tcp_metrics_hash = kzalloc(size, GFP_KERNEL | __GFP_NOWARN);
+ if (!tcp_metrics_hash)
+ tcp_metrics_hash = vzalloc(size);
- if (!net->ipv4.tcp_metrics_hash)
+ if (!tcp_metrics_hash)
return -ENOMEM;
return 0;
@@ -1170,19 +1176,7 @@ static int __net_init tcp_net_metrics_init(struct net *net)
static void __net_exit tcp_net_metrics_exit(struct net *net)
{
- unsigned int i;
-
- for (i = 0; i < (1U << net->ipv4.tcp_metrics_hash_log) ; i++) {
- struct tcp_metrics_block *tm, *next;
-
- tm = rcu_dereference_protected(net->ipv4.tcp_metrics_hash[i].chain, 1);
- while (tm) {
- next = rcu_dereference_protected(tm->tcpm_next, 1);
- kfree(tm);
- tm = next;
- }
- }
- kvfree(net->ipv4.tcp_metrics_hash);
+ tcp_metrics_flush_all(net);
}
static __net_initdata struct pernet_operations tcp_net_metrics_ops = {
--
2.2.1
next prev parent reply other threads:[~2015-03-11 16:47 UTC|newest]
Thread overview: 119+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-02-05 1:34 [RFC PATCH 00/29] net: VRF support David Ahern
2015-02-05 1:34 ` [RFC PATCH 01/29] net: Introduce net_ctx and macro for context comparison David Ahern
2015-02-05 1:34 ` [RFC PATCH 02/29] net: Flip net_device to use net_ctx David Ahern
2015-02-05 13:47 ` Nicolas Dichtel
2015-02-06 0:45 ` David Ahern
2015-02-05 1:34 ` [RFC PATCH 03/29] net: Flip sock_common to net_ctx David Ahern
2015-02-05 1:34 ` [RFC PATCH 04/29] net: Add net_ctx macros for skbuffs David Ahern
2015-02-05 1:34 ` [RFC PATCH 05/29] net: Flip seq_net_private to net_ctx David Ahern
2015-02-05 1:34 ` [RFC PATCH 06/29] net: Flip fib_rules and fib_rules_ops to use net_ctx David Ahern
2015-02-05 1:34 ` [RFC PATCH 07/29] net: Flip inet_bind_bucket to net_ctx David Ahern
2015-02-05 1:34 ` [RFC PATCH 08/29] net: Flip fib_info " David Ahern
2015-02-05 1:34 ` [RFC PATCH 09/29] net: Flip ip6_flowlabel " David Ahern
2015-02-05 1:34 ` [RFC PATCH 10/29] net: Flip neigh structs " David Ahern
2015-02-05 1:34 ` [RFC PATCH 11/29] net: Flip nl_info " David Ahern
2015-02-05 1:34 ` [RFC PATCH 12/29] net: Add device lookups by net_ctx David Ahern
2015-02-05 1:34 ` [RFC PATCH 13/29] net: Convert function arg from struct net to struct net_ctx David Ahern
2015-02-05 1:34 ` [RFC PATCH 14/29] net: vrf: Introduce vrf header file David Ahern
2015-02-05 13:44 ` Nicolas Dichtel
2015-02-06 0:52 ` David Ahern
2015-02-06 8:53 ` Nicolas Dichtel
2015-02-05 1:34 ` [RFC PATCH 15/29] net: vrf: Add vrf to net_ctx struct David Ahern
2015-02-05 1:34 ` [RFC PATCH 16/29] net: vrf: Set default vrf David Ahern
2015-02-05 1:34 ` [RFC PATCH 17/29] net: vrf: Add vrf context to task struct David Ahern
2015-02-05 1:34 ` [RFC PATCH 18/29] net: vrf: Plumbing for vrf context on a socket David Ahern
2015-02-05 13:44 ` Nicolas Dichtel
2015-02-06 1:18 ` David Ahern
2015-02-05 1:34 ` [RFC PATCH 19/29] net: vrf: Add vrf context to skb David Ahern
2015-02-05 13:45 ` Nicolas Dichtel
2015-02-06 1:21 ` David Ahern
2015-02-06 3:54 ` Eric W. Biederman
2015-02-06 6:00 ` David Ahern
2015-02-05 1:34 ` [RFC PATCH 20/29] net: vrf: Add vrf context to flow struct David Ahern
2015-02-05 1:34 ` [RFC PATCH 21/29] net: vrf: Add vrf context to genid's David Ahern
2015-02-05 1:34 ` [RFC PATCH 22/29] net: vrf: Set VRF id in various network structs David Ahern
2015-02-05 1:34 ` [RFC PATCH 23/29] net: vrf: Enable vrf checks David Ahern
2015-02-05 1:34 ` [RFC PATCH 24/29] net: vrf: Add support to get/set vrf context on a device David Ahern
2015-02-05 1:34 ` [RFC PATCH 25/29] net: vrf: Handle VRF any context David Ahern
2015-02-05 13:46 ` Nicolas Dichtel
2015-02-06 1:23 ` David Ahern
2015-02-05 1:34 ` [RFC PATCH 26/29] net: vrf: Change single_open_net to pass net_ctx David Ahern
2015-02-05 1:34 ` [RFC PATCH 27/29] net: vrf: Add vrf checks and context to ipv4 proc files David Ahern
2015-02-05 1:34 ` [RFC PATCH 28/29] iproute2: vrf: Add vrf subcommand David Ahern
2015-02-05 1:34 ` [RFC PATCH 29/29] iproute2: Add vrf option to ip link command David Ahern
2015-02-05 5:17 ` [RFC PATCH 00/29] net: VRF support roopa
2015-02-05 13:44 ` Nicolas Dichtel
2015-02-06 1:32 ` David Ahern
2015-02-06 8:53 ` Nicolas Dichtel
2015-02-05 23:12 ` roopa
2015-02-06 2:19 ` David Ahern
2015-02-09 16:38 ` roopa
2015-02-10 10:43 ` Derek Fawcus
2015-02-06 6:10 ` Shmulik Ladkani
2015-02-09 15:54 ` roopa
2015-02-11 7:42 ` Shmulik Ladkani
2015-02-06 1:33 ` Stephen Hemminger
2015-02-06 2:10 ` David Ahern
2015-02-06 4:14 ` Eric W. Biederman
2015-02-06 6:15 ` David Ahern
2015-02-06 15:08 ` Nicolas Dichtel
[not found] ` <87iofe7n1x.fsf@x220.int.ebiederm.org>
2015-02-09 20:48 ` Nicolas Dichtel
2015-02-11 4:14 ` David Ahern
2015-02-06 15:10 ` Nicolas Dichtel
2015-02-06 20:50 ` Eric W. Biederman
2015-02-09 0:36 ` David Ahern
2015-02-09 11:30 ` Derek Fawcus
[not found] ` <871tlxtbhd.fsf_-_@x220.int.ebiederm.org>
2015-02-11 2:55 ` network namespace bloat Eric Dumazet
2015-02-11 3:18 ` Eric W. Biederman
2015-02-19 19:49 ` David Miller
2015-03-09 18:22 ` [PATCH net-next 0/6] tcp_metrics: Network namespace bloat reduction Eric W. Biederman
2015-03-09 18:27 ` [PATCH net-next 1/6] tcp_metrics: panic when tcp_metrics can not be allocated Eric W. Biederman
2015-03-09 18:50 ` Sergei Shtylyov
2015-03-11 19:22 ` Sergei Shtylyov
2015-03-09 18:27 ` [PATCH net-next 2/6] tcp_metrics: Mix the network namespace into the hash function Eric W. Biederman
2015-03-09 18:29 ` [PATCH net-next 3/6] tcp_metrics: Add a field tcpm_net and verify it matches on lookup Eric W. Biederman
2015-03-09 20:25 ` Julian Anastasov
2015-03-10 6:59 ` Eric W. Biederman
2015-03-10 8:23 ` Julian Anastasov
2015-03-11 0:58 ` Eric W. Biederman
2015-03-10 16:36 ` David Miller
2015-03-10 17:06 ` Eric W. Biederman
2015-03-10 17:29 ` David Miller
2015-03-10 17:56 ` Eric W. Biederman
2015-03-09 18:30 ` [PATCH net-next 4/6] tcp_metrics: Remove the unused return code from tcp_metrics_flush_all Eric W. Biederman
2015-03-09 18:30 ` [PATCH net-next 5/6] tcp_metrics: Rewrite tcp_metrics_flush_all Eric W. Biederman
2015-03-09 18:31 ` [PATCH net-next 6/6] tcp_metrics: Use a single hash table for all network namespaces Eric W. Biederman
2015-03-09 18:43 ` Eric Dumazet
2015-03-09 18:47 ` Eric Dumazet
2015-03-09 19:35 ` Eric W. Biederman
2015-03-09 20:21 ` Eric Dumazet
2015-03-09 20:09 ` [PATCH net-next 0/6] tcp_metrics: Network namespace bloat reduction David Miller
2015-03-09 20:21 ` Eric W. Biederman
2015-03-11 16:33 ` [PATCH net-next 0/8] tcp_metrics: Network namespace bloat reduction v2 Eric W. Biederman
2015-03-11 16:35 ` [PATCH net-next 1/8] net: Kill hold_net release_net Eric W. Biederman
2015-03-11 16:55 ` Eric Dumazet
2015-03-11 17:34 ` Eric W. Biederman
2015-03-11 17:07 ` Eric Dumazet
2015-03-11 17:08 ` Eric Dumazet
2015-03-11 17:10 ` Eric Dumazet
2015-03-11 17:36 ` Eric W. Biederman
2015-03-11 16:36 ` [PATCH net-next 2/8] net: Introduce possible_net_t Eric W. Biederman
2015-03-11 16:38 ` [PATCH net-next 3/8] tcp_metrics: panic when tcp_metrics_init fails Eric W. Biederman
2015-03-11 16:38 ` [PATCH net-next 4/8] tcp_metrics: Mix the network namespace into the hash function Eric W. Biederman
2015-03-11 16:40 ` [PATCH net-next 5/8] tcp_metrics: Add a field tcpm_net and verify it matches on lookup Eric W. Biederman
2015-03-11 16:41 ` [PATCH net-next 6/8] tcp_metrics: Remove the unused return code from tcp_metrics_flush_all Eric W. Biederman
2015-03-11 16:43 ` [PATCH net-next 7/8] tcp_metrics: Rewrite tcp_metrics_flush_all Eric W. Biederman
2015-03-11 16:43 ` Eric W. Biederman [this message]
2015-03-13 5:04 ` [PATCH net-next 0/6] tcp_metrics: Network namespace bloat reduction v3 Eric W. Biederman
2015-03-13 5:04 ` [PATCH net-next 1/6] tcp_metrics: panic when tcp_metrics_init fails Eric W. Biederman
2015-03-13 5:05 ` [PATCH net-next 2/6] tcp_metrics: Mix the network namespace into the hash function Eric W. Biederman
2015-03-13 5:05 ` [PATCH net-next 3/6] tcp_metrics: Add a field tcpm_net and verify it matches on lookup Eric W. Biederman
2015-03-13 5:06 ` [PATCH net-next 4/6] tcp_metrics: Remove the unused return code from tcp_metrics_flush_all Eric W. Biederman
2015-03-13 5:07 ` [PATCH net-next 5/6] tcp_metrics: Rewrite tcp_metrics_flush_all Eric W. Biederman
2015-03-13 5:07 ` [PATCH net-next 6/6] tcp_metrics: Use a single hash table for all network namespaces Eric W. Biederman
2015-03-13 5:57 ` [PATCH net-next 0/6] tcp_metrics: Network namespace bloat reduction v3 David Miller
2015-02-11 17:09 ` network namespace bloat Nicolas Dichtel
2015-02-10 0:53 ` [RFC PATCH 00/29] net: VRF support Thomas Graf
2015-02-10 20:54 ` David Ahern
2016-05-25 16:04 ` Chenna
2016-05-25 19:04 ` David Ahern
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=87h9treb7e.fsf_-_@x220.int.ebiederm.org \
--to=ebiederm@xmission$(echo .)com \
--cc=davem@davemloft$(echo .)net \
--cc=ddutt@cumulusnetworks$(echo .)com \
--cc=dsahern@gmail$(echo .)com \
--cc=edumazet@google$(echo .)com \
--cc=hannes@stressinduktion$(echo .)org \
--cc=ja@ssi$(echo .)bg \
--cc=netdev@vger$(echo .)kernel.org \
--cc=nicolas.dichtel@6wind$(echo .)com \
--cc=roopa@cumulusnetworks$(echo .)com \
--cc=shmulik.ladkani@gmail$(echo .)com \
--cc=stephen@networkplumber$(echo .)org \
--cc=vipin@cumulusnetworks$(echo .)com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox