6.6-stable review patch. If anyone has any objections, please let me know.
------------------
From: Eric Dumazet edumazet@google.com
[ Upstream commit 86e8921df05c6e9423ab74ab8d41022775d8b83a ]
TCPDIAG_GETSOCK and DCCPDIAG_GETSOCK diag are serialized on sock_diag_table_mutex.
This is to make sure inet_diag module is not unloaded while diag was ongoing.
It is time to get rid of this mutex and use RCU protection, allowing full parallelism.
Signed-off-by: Eric Dumazet edumazet@google.com Reviewed-by: Guillaume Nault gnault@redhat.com Reviewed-by: Kuniyuki Iwashima kuniyu@amazon.com Reviewed-by: Willem de Bruijn willemb@google.com Signed-off-by: Paolo Abeni pabeni@redhat.com Stable-dep-of: eb02688c5c45 ("ipv6: release nexthop on device removal") Signed-off-by: Sasha Levin sashal@kernel.org --- include/linux/sock_diag.h | 9 ++++++-- net/core/sock_diag.c | 43 +++++++++++++++++++++++---------------- net/ipv4/inet_diag.c | 9 ++++++-- 3 files changed, 40 insertions(+), 21 deletions(-)
diff --git a/include/linux/sock_diag.h b/include/linux/sock_diag.h index 7c07754d711b9..110978dc9af1b 100644 --- a/include/linux/sock_diag.h +++ b/include/linux/sock_diag.h @@ -23,8 +23,13 @@ struct sock_diag_handler { int sock_diag_register(const struct sock_diag_handler *h); void sock_diag_unregister(const struct sock_diag_handler *h);
-void sock_diag_register_inet_compat(int (*fn)(struct sk_buff *skb, struct nlmsghdr *nlh)); -void sock_diag_unregister_inet_compat(int (*fn)(struct sk_buff *skb, struct nlmsghdr *nlh)); +struct sock_diag_inet_compat { + struct module *owner; + int (*fn)(struct sk_buff *skb, struct nlmsghdr *nlh); +}; + +void sock_diag_register_inet_compat(const struct sock_diag_inet_compat *ptr); +void sock_diag_unregister_inet_compat(const struct sock_diag_inet_compat *ptr);
u64 __sock_gen_cookie(struct sock *sk);
diff --git a/net/core/sock_diag.c b/net/core/sock_diag.c index 72009e1f4380d..5c3666431df49 100644 --- a/net/core/sock_diag.c +++ b/net/core/sock_diag.c @@ -17,8 +17,9 @@ #include <linux/sock_diag.h>
static const struct sock_diag_handler __rcu *sock_diag_handlers[AF_MAX]; -static int (*inet_rcv_compat)(struct sk_buff *skb, struct nlmsghdr *nlh); -static DEFINE_MUTEX(sock_diag_table_mutex); + +static struct sock_diag_inet_compat __rcu *inet_rcv_compat; + static struct workqueue_struct *broadcast_wq;
DEFINE_COOKIE(sock_cookie); @@ -184,19 +185,20 @@ void sock_diag_broadcast_destroy(struct sock *sk) queue_work(broadcast_wq, &bsk->work); }
-void sock_diag_register_inet_compat(int (*fn)(struct sk_buff *skb, struct nlmsghdr *nlh)) +void sock_diag_register_inet_compat(const struct sock_diag_inet_compat *ptr) { - mutex_lock(&sock_diag_table_mutex); - inet_rcv_compat = fn; - mutex_unlock(&sock_diag_table_mutex); + xchg((__force const struct sock_diag_inet_compat **)&inet_rcv_compat, + ptr); } EXPORT_SYMBOL_GPL(sock_diag_register_inet_compat);
-void sock_diag_unregister_inet_compat(int (*fn)(struct sk_buff *skb, struct nlmsghdr *nlh)) +void sock_diag_unregister_inet_compat(const struct sock_diag_inet_compat *ptr) { - mutex_lock(&sock_diag_table_mutex); - inet_rcv_compat = NULL; - mutex_unlock(&sock_diag_table_mutex); + const struct sock_diag_inet_compat *old; + + old = xchg((__force const struct sock_diag_inet_compat **)&inet_rcv_compat, + NULL); + WARN_ON_ONCE(old != ptr); } EXPORT_SYMBOL_GPL(sock_diag_unregister_inet_compat);
@@ -259,20 +261,27 @@ static int __sock_diag_cmd(struct sk_buff *skb, struct nlmsghdr *nlh) static int sock_diag_rcv_msg(struct sk_buff *skb, struct nlmsghdr *nlh, struct netlink_ext_ack *extack) { + const struct sock_diag_inet_compat *ptr; int ret;
switch (nlh->nlmsg_type) { case TCPDIAG_GETSOCK: case DCCPDIAG_GETSOCK: - if (inet_rcv_compat == NULL) + + if (!rcu_access_pointer(inet_rcv_compat)) sock_load_diag_module(AF_INET, 0);
- mutex_lock(&sock_diag_table_mutex); - if (inet_rcv_compat != NULL) - ret = inet_rcv_compat(skb, nlh); - else - ret = -EOPNOTSUPP; - mutex_unlock(&sock_diag_table_mutex); + rcu_read_lock(); + ptr = rcu_dereference(inet_rcv_compat); + if (ptr && !try_module_get(ptr->owner)) + ptr = NULL; + rcu_read_unlock(); + + ret = -EOPNOTSUPP; + if (ptr) { + ret = ptr->fn(skb, nlh); + module_put(ptr->owner); + }
return ret; case SOCK_DIAG_BY_FAMILY: diff --git a/net/ipv4/inet_diag.c b/net/ipv4/inet_diag.c index c1ff0e426b677..5d09ab3ed735e 100644 --- a/net/ipv4/inet_diag.c +++ b/net/ipv4/inet_diag.c @@ -1445,6 +1445,11 @@ void inet_diag_unregister(const struct inet_diag_handler *h) } EXPORT_SYMBOL_GPL(inet_diag_unregister);
+static const struct sock_diag_inet_compat inet_diag_compat = { + .owner = THIS_MODULE, + .fn = inet_diag_rcv_msg_compat, +}; + static int __init inet_diag_init(void) { const int inet_diag_table_size = (IPPROTO_MAX * @@ -1463,7 +1468,7 @@ static int __init inet_diag_init(void) if (err) goto out_free_inet;
- sock_diag_register_inet_compat(inet_diag_rcv_msg_compat); + sock_diag_register_inet_compat(&inet_diag_compat); out: return err;
@@ -1478,7 +1483,7 @@ static void __exit inet_diag_exit(void) { sock_diag_unregister(&inet6_diag_handler); sock_diag_unregister(&inet_diag_handler); - sock_diag_unregister_inet_compat(inet_diag_rcv_msg_compat); + sock_diag_unregister_inet_compat(&inet_diag_compat); kfree(inet_diag_table); }