From: Akhmat Karakotov <hmukos@yandex-team.ru>
Date: Mon, 31 Jan 2022 16:31:21 +0300
Subject: txhash: Make rethinking txhash behavior configurable via sysctl
Patch-mainline: v5.18-rc1
Git-commit: e187013abeb4c2a7ec8a4bb978844c7e92a1a6ec
References: jsc#PED-1377
Add a per ns sysctl that controls the txhash rethink behavior:
net.core.txrehash. When enabled, the same behavior is retained,
when disabled, rethink is not performed. Sysctl is enabled by default.
Signed-off-by: Akhmat Karakotov <hmukos@yandex-team.ru>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Shung-Hsi Yu <shung-hsi.yu@suse.com>
---
include/net/netns/core.h | 1 +
include/net/sock.h | 34 +++++++++++++++++++++-------------
include/uapi/linux/socket.h | 3 +++
net/core/net_namespace.c | 2 ++
net/core/sysctl_net_core.c | 14 ++++++++++++--
5 files changed, 39 insertions(+), 15 deletions(-)
--- a/include/net/netns/core.h
+++ b/include/net/netns/core.h
@@ -10,6 +10,7 @@ struct netns_core {
struct ctl_table_header *sysctl_hdr;
int sysctl_somaxconn;
+ u8 sysctl_txrehash;
#ifdef CONFIG_PROC_FS
int __percpu *sock_inuse;
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -582,6 +582,18 @@ static inline bool sk_user_data_is_nocop
__tmp | SK_USER_DATA_NOCOPY); \
})
+static inline
+struct net *sock_net(const struct sock *sk)
+{
+ return read_pnet(&sk->sk_net);
+}
+
+static inline
+void sock_net_set(struct sock *sk, struct net *net)
+{
+ write_pnet(&sk->sk_net, net);
+}
+
/*
* SK_CAN_REUSE and SK_NO_REUSE on a socket mean that the socket is OK
* or not whether his port will be reused by someone else. SK_FORCE_REUSE
@@ -1980,10 +1992,18 @@ static inline void sk_set_txhash(struct
static inline bool sk_rethink_txhash(struct sock *sk)
{
- if (sk->sk_txhash) {
+ u8 rehash;
+
+ if (!sk->sk_txhash)
+ return false;
+
+ rehash = READ_ONCE(sock_net(sk)->core.sysctl_txrehash);
+
+ if (rehash) {
sk_set_txhash(sk);
return true;
}
+
return false;
}
@@ -2632,18 +2652,6 @@ static inline void sk_eat_skb(struct soc
__kfree_skb(skb);
}
-static inline
-struct net *sock_net(const struct sock *sk)
-{
- return read_pnet(&sk->sk_net);
-}
-
-static inline
-void sock_net_set(struct sock *sk, struct net *net)
-{
- write_pnet(&sk->sk_net, net);
-}
-
static inline bool
skb_sk_is_prefetched(struct sk_buff *skb)
{
--- a/include/uapi/linux/socket.h
+++ b/include/uapi/linux/socket.h
@@ -26,4 +26,7 @@ struct __kernel_sockaddr_storage {
};
};
+#define SOCK_TXREHASH_DISABLED 0
+#define SOCK_TXREHASH_ENABLED 1
+
#endif /* _UAPI_LINUX_SOCKET_H */
--- a/net/core/net_namespace.c
+++ b/net/core/net_namespace.c
@@ -367,6 +367,8 @@ out_undo:
static int __net_init net_defaults_init_net(struct net *net)
{
net->core.sysctl_somaxconn = SOMAXCONN;
+ net->core.sysctl_txrehash = SOCK_TXREHASH_ENABLED;
+
return 0;
}
--- a/net/core/sysctl_net_core.c
+++ b/net/core/sysctl_net_core.c
@@ -593,6 +593,15 @@ static struct ctl_table netns_core_table
.extra1 = SYSCTL_ZERO,
.proc_handler = proc_dointvec_minmax
},
+ {
+ .procname = "txrehash",
+ .data = &init_net.core.sysctl_txrehash,
+ .maxlen = sizeof(u8),
+ .mode = 0644,
+ .extra1 = SYSCTL_ZERO,
+ .extra2 = SYSCTL_ONE,
+ .proc_handler = proc_dou8vec_minmax,
+ },
{ }
};
@@ -611,7 +620,7 @@ __setup("fb_tunnels=", fb_tunnels_only_f
static __net_init int sysctl_core_net_init(struct net *net)
{
- struct ctl_table *tbl;
+ struct ctl_table *tbl, *tmp;
tbl = netns_core_table;
if (!net_eq(net, &init_net)) {
@@ -619,7 +628,8 @@ static __net_init int sysctl_core_net_in
if (tbl == NULL)
goto err_dup;
- tbl[0].data = &net->core.sysctl_somaxconn;
+ for (tmp = tbl; tmp->procname; tmp++)
+ tmp->data += (char *)net - (char *)&init_net;
/* Don't export any sysctls to unprivileged users */
if (net->user_ns != &init_user_ns) {