From: Eric Dumazet <eric.dumazet@gmail.com>
To: Shawn Lu <shawn.lu@ericsson.com>
Cc: davem@davemloft.net, netdev@vger.kernel.org, xiaoclu@gmail.com
Subject: [PATCH net-next] tcp: md5: protects md5sig_info with RCU
Date: Wed, 01 Feb 2012 05:45:40 +0100 [thread overview]
Message-ID: <1328071540.2617.25.camel@edumazet-laptop> (raw)
In-Reply-To: <1328057424-11962-2-git-send-email-shawn.lu@ericsson.com>
This patch makes sure we use appropriate memory barriers before
publishing tp->md5sig_info, allowing tcp_md5_do_lookup() being used from
tcp_v4_send_reset() without holding socket lock (upcoming patch from
Shawn Lu)
Note we also need to respect rcu grace period before its freeing, since
we can free socket without this grace period thanks to
SLAB_DESTROY_BY_RCU
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Shawn Lu <shawn.lu@ericsson.com>
---
include/linux/tcp.h | 2 +-
include/net/tcp.h | 1 +
net/ipv4/tcp_ipv4.c | 32 ++++++++++++++++++++------------
net/ipv6/tcp_ipv6.c | 2 --
4 files changed, 22 insertions(+), 15 deletions(-)
diff --git a/include/linux/tcp.h b/include/linux/tcp.h
index c2025f1..115389e 100644
--- a/include/linux/tcp.h
+++ b/include/linux/tcp.h
@@ -463,7 +463,7 @@ struct tcp_sock {
const struct tcp_sock_af_ops *af_specific;
/* TCP MD5 Signature Option information */
- struct tcp_md5sig_info *md5sig_info;
+ struct tcp_md5sig_info __rcu *md5sig_info;
#endif
/* When the cookie options are generated and exchanged, then this
diff --git a/include/net/tcp.h b/include/net/tcp.h
index 10ae4c7..78880ba 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -1150,6 +1150,7 @@ struct tcp_md5sig_key {
/* - sock block */
struct tcp_md5sig_info {
struct hlist_head head;
+ struct rcu_head rcu;
};
/* - pseudo header */
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index da5d322..567cca9 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -879,14 +879,18 @@ struct tcp_md5sig_key *tcp_md5_do_lookup(struct sock *sk,
struct tcp_md5sig_key *key;
struct hlist_node *pos;
unsigned int size = sizeof(struct in_addr);
+ struct tcp_md5sig_info *md5sig;
- if (!tp->md5sig_info)
+ /* caller either holds rcu_read_lock() or socket lock */
+ md5sig = rcu_dereference_check(tp->md5sig_info,
+ sock_owned_by_user(sk));
+ if (!md5sig)
return NULL;
#if IS_ENABLED(CONFIG_IPV6)
if (family == AF_INET6)
size = sizeof(struct in6_addr);
#endif
- hlist_for_each_entry_rcu(key, pos, &tp->md5sig_info->head, node) {
+ hlist_for_each_entry_rcu(key, pos, &md5sig->head, node) {
if (key->family != family)
continue;
if (!memcmp(&key->addr, addr, size))
@@ -932,7 +936,8 @@ int tcp_md5_do_add(struct sock *sk, const union tcp_md5_addr *addr,
return 0;
}
- md5sig = tp->md5sig_info;
+ md5sig = rcu_dereference_protected(tp->md5sig_info,
+ sock_owned_by_user(sk));
if (!md5sig) {
md5sig = kmalloc(sizeof(*md5sig), gfp);
if (!md5sig)
@@ -940,7 +945,7 @@ int tcp_md5_do_add(struct sock *sk, const union tcp_md5_addr *addr,
sk_nocaps_add(sk, NETIF_F_GSO_MASK);
INIT_HLIST_HEAD(&md5sig->head);
- tp->md5sig_info = md5sig;
+ rcu_assign_pointer(tp->md5sig_info, md5sig);
}
key = sock_kmalloc(sk, sizeof(*key), gfp);
@@ -966,6 +971,7 @@ int tcp_md5_do_del(struct sock *sk, const union tcp_md5_addr *addr, int family)
{
struct tcp_sock *tp = tcp_sk(sk);
struct tcp_md5sig_key *key;
+ struct tcp_md5sig_info *md5sig;
key = tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&addr, AF_INET);
if (!key)
@@ -973,7 +979,9 @@ int tcp_md5_do_del(struct sock *sk, const union tcp_md5_addr *addr, int family)
hlist_del_rcu(&key->node);
atomic_sub(sizeof(*key), &sk->sk_omem_alloc);
kfree_rcu(key, rcu);
- if (hlist_empty(&tp->md5sig_info->head))
+ md5sig = rcu_dereference_protected(tp->md5sig_info,
+ sock_owned_by_user(sk));
+ if (hlist_empty(&md5sig->head))
tcp_free_md5sig_pool();
return 0;
}
@@ -984,10 +992,13 @@ void tcp_clear_md5_list(struct sock *sk)
struct tcp_sock *tp = tcp_sk(sk);
struct tcp_md5sig_key *key;
struct hlist_node *pos, *n;
+ struct tcp_md5sig_info *md5sig;
- if (!hlist_empty(&tp->md5sig_info->head))
+ md5sig = rcu_dereference_protected(tp->md5sig_info, 1);
+
+ if (!hlist_empty(&md5sig->head))
tcp_free_md5sig_pool();
- hlist_for_each_entry_safe(key, pos, n, &tp->md5sig_info->head, node) {
+ hlist_for_each_entry_safe(key, pos, n, &md5sig->head, node) {
hlist_del_rcu(&key->node);
atomic_sub(sizeof(*key), &sk->sk_omem_alloc);
kfree_rcu(key, rcu);
@@ -1009,12 +1020,9 @@ static int tcp_v4_parse_md5_keys(struct sock *sk, char __user *optval,
if (sin->sin_family != AF_INET)
return -EINVAL;
- if (!cmd.tcpm_key || !cmd.tcpm_keylen) {
- if (!tcp_sk(sk)->md5sig_info)
- return -ENOENT;
+ if (!cmd.tcpm_key || !cmd.tcpm_keylen)
return tcp_md5_do_del(sk, (union tcp_md5_addr *)&sin->sin_addr.s_addr,
AF_INET);
- }
if (cmd.tcpm_keylen > TCP_MD5SIG_MAXKEYLEN)
return -EINVAL;
@@ -1896,7 +1904,7 @@ void tcp_v4_destroy_sock(struct sock *sk)
/* Clean up the MD5 key list, if any */
if (tp->md5sig_info) {
tcp_clear_md5_list(sk);
- kfree(tp->md5sig_info);
+ kfree_rcu(tp->md5sig_info, rcu);
tp->md5sig_info = NULL;
}
#endif
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index bec41f9..c250181 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -571,8 +571,6 @@ static int tcp_v6_parse_md5_keys (struct sock *sk, char __user *optval,
return -EINVAL;
if (!cmd.tcpm_keylen) {
- if (!tcp_sk(sk)->md5sig_info)
- return -ENOENT;
if (ipv6_addr_v4mapped(&sin6->sin6_addr))
return tcp_md5_do_del(sk, (union tcp_md5_addr *)&sin6->sin6_addr.s6_addr32[3],
AF_INET);
next prev parent reply other threads:[~2012-02-01 4:45 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-02-01 0:50 (unknown), Shawn Lu
2012-02-01 0:50 ` [PATCH] tcp: md5: fix md5 RST when both sides have listener Shawn Lu
2012-02-01 4:08 ` Eric Dumazet
2012-02-01 4:45 ` Eric Dumazet [this message]
2012-02-01 7:18 ` [PATCH net-next] tcp: md5: protects md5sig_info with RCU David Miller
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=1328071540.2617.25.camel@edumazet-laptop \
--to=eric.dumazet@gmail.com \
--cc=davem@davemloft.net \
--cc=netdev@vger.kernel.org \
--cc=shawn.lu@ericsson.com \
--cc=xiaoclu@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
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