commit 11052589cf5c0bab3b4884d423d5f60c38fcf25d upstream Commit e21145a9871a ("ipv4: namespacify ip_early_demux sysctl knob") made it possible to enable/disable early_demux on a per-netns basis. Then, we introduced two knobs, tcp_early_demux and udp_early_demux, to switch it for TCP/UDP in commit dddb64bcb346 ("net: Add sysctl to toggle early demux for tcp and udp"). However, the .proc_handler() was wrong and actually disabled us from changing the behaviour in each netns. We can execute early_demux if net.ipv4.ip_early_demux is on and each proto .early_demux() handler is not NULL. When we toggle (tcp|udp)_early_demux, the change itself is saved in each netns variable, but the .early_demux() handler is a global variable, so the handler is switched based on the init_net's sysctl variable. Thus, netns (tcp|udp)_early_demux knobs have nothing to do with the logic. Whether we CAN execute proto .early_demux() is always decided by init_net's sysctl knob, and whether we DO it or not is by each netns ip_early_demux knob. This patch namespacifies (tcp|udp)_early_demux again. For now, the users of the .early_demux() handler are TCP and UDP only, and they are called directly to avoid retpoline. So, we can remove the .early_demux() handler from inet6?_protos and need not dereference them in ip6?_rcv_finish_core(). If another proto needs .early_demux(), we can restore it at that time. Fixes: dddb64bcb346 ("net: Add sysctl to toggle early demux for tcp and udp") Signed-off-by: Kuniyuki Iwashima <kuniyu@xxxxxxxxxx> Link: https://lore.kernel.org/r/20220713175207.7727-1-kuniyu@xxxxxxxxxx Signed-off-by: Jakub Kicinski <kuba@xxxxxxxxxx> --- include/net/protocol.h | 4 --- include/net/tcp.h | 2 ++ include/net/udp.h | 1 + net/ipv4/af_inet.c | 14 ++------- net/ipv4/ip_input.c | 32 ++++++++++++++------- net/ipv4/sysctl_net_ipv4.c | 59 ++------------------------------------ net/ipv6/ip6_input.c | 23 +++++++++------ net/ipv6/tcp_ipv6.c | 9 ++---- net/ipv6/udp.c | 9 ++---- 9 files changed, 47 insertions(+), 106 deletions(-) diff --git a/include/net/protocol.h b/include/net/protocol.h index 4fc75f7ae23b..312a27393e0b 100644 --- a/include/net/protocol.h +++ b/include/net/protocol.h @@ -39,8 +39,6 @@ /* This is used to register protocols. */ struct net_protocol { - int (*early_demux)(struct sk_buff *skb); - int (*early_demux_handler)(struct sk_buff *skb); int (*handler)(struct sk_buff *skb); void (*err_handler)(struct sk_buff *skb, u32 info); unsigned int no_policy:1, @@ -54,8 +52,6 @@ struct net_protocol { #if IS_ENABLED(CONFIG_IPV6) struct inet6_protocol { - void (*early_demux)(struct sk_buff *skb); - void (*early_demux_handler)(struct sk_buff *skb); int (*handler)(struct sk_buff *skb); void (*err_handler)(struct sk_buff *skb, diff --git a/include/net/tcp.h b/include/net/tcp.h index 487b6c5f53f4..9a154fe06c60 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -901,6 +901,8 @@ static inline int tcp_v6_sdif(const struct sk_buff *skb) #endif return 0; } + +void tcp_v6_early_demux(struct sk_buff *skb); #endif static inline bool inet_exact_dif_match(struct net *net, struct sk_buff *skb) diff --git a/include/net/udp.h b/include/net/udp.h index 8482a990b0bb..618c83bea50d 100644 --- a/include/net/udp.h +++ b/include/net/udp.h @@ -173,6 +173,7 @@ typedef struct sock *(*udp_lookup_t)(struct sk_buff *skb, __be16 sport, struct sk_buff *udp_gro_receive(struct list_head *head, struct sk_buff *skb, struct udphdr *uh, udp_lookup_t lookup); int udp_gro_complete(struct sk_buff *skb, int nhoff, udp_lookup_t lookup); +void udp_v6_early_demux(struct sk_buff *skb); struct sk_buff *__udp_gso_segment(struct sk_buff *gso_skb, netdev_features_t features); diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c index 3cf9cd180898..4b26ae525d6d 100644 --- a/net/ipv4/af_inet.c +++ b/net/ipv4/af_inet.c @@ -1678,12 +1678,7 @@ static const struct net_protocol igmp_protocol = { }; #endif -/* thinking of making this const? Don't. - * early_demux can change based on sysctl. - */ -static struct net_protocol tcp_protocol = { - .early_demux = tcp_v4_early_demux, - .early_demux_handler = tcp_v4_early_demux, +static const struct net_protocol tcp_protocol = { .handler = tcp_v4_rcv, .err_handler = tcp_v4_err, .no_policy = 1, @@ -1691,12 +1686,7 @@ static struct net_protocol tcp_protocol = { .icmp_strict_tag_validation = 1, }; -/* thinking of making this const? Don't. - * early_demux can change based on sysctl. - */ -static struct net_protocol udp_protocol = { - .early_demux = udp_v4_early_demux, - .early_demux_handler = udp_v4_early_demux, +static const struct net_protocol udp_protocol = { .handler = udp_rcv, .err_handler = udp_err, .no_policy = 1, diff --git a/net/ipv4/ip_input.c b/net/ipv4/ip_input.c index c3a0683e83df..7ead5192b2a9 100644 --- a/net/ipv4/ip_input.c +++ b/net/ipv4/ip_input.c @@ -306,28 +306,38 @@ static inline bool ip_rcv_options(struct sk_buff *skb, struct net_device *dev) return true; } +int udp_v4_early_demux(struct sk_buff *); +int tcp_v4_early_demux(struct sk_buff *); static int ip_rcv_finish_core(struct net *net, struct sock *sk, struct sk_buff *skb, struct net_device *dev) { const struct iphdr *iph = ip_hdr(skb); - int (*edemux)(struct sk_buff *skb); struct rtable *rt; int err; - if (net->ipv4.sysctl_ip_early_demux && + if (READ_ONCE(net->ipv4.sysctl_ip_early_demux) && !skb_dst(skb) && !skb->sk && !ip_is_fragment(iph)) { - const struct net_protocol *ipprot; - int protocol = iph->protocol; + switch (iph->protocol) { + case IPPROTO_TCP: + if (READ_ONCE(net->ipv4.sysctl_tcp_early_demux)) { + tcp_v4_early_demux(skb); - ipprot = rcu_dereference(inet_protos[protocol]); - if (ipprot && (edemux = READ_ONCE(ipprot->early_demux))) { - err = edemux(skb); - if (unlikely(err)) - goto drop_error; - /* must reload iph, skb->head might have changed */ - iph = ip_hdr(skb); + /* must reload iph, skb->head might have changed */ + iph = ip_hdr(skb); + } + break; + case IPPROTO_UDP: + if (READ_ONCE(net->ipv4.sysctl_udp_early_demux)) { + err = udp_v4_early_demux(skb); + if (unlikely(err)) + goto drop_error; + + /* must reload iph, skb->head might have changed */ + iph = ip_hdr(skb); + } + break; } } diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c index ad132b6e8cfa..c97ba2a44b8b 100644 --- a/net/ipv4/sysctl_net_ipv4.c +++ b/net/ipv4/sysctl_net_ipv4.c @@ -332,61 +332,6 @@ static int proc_tcp_fastopen_key(struct ctl_table *table, int write, return ret; } -static void proc_configure_early_demux(int enabled, int protocol) -{ - struct net_protocol *ipprot; -#if IS_ENABLED(CONFIG_IPV6) - struct inet6_protocol *ip6prot; -#endif - - rcu_read_lock(); - - ipprot = rcu_dereference(inet_protos[protocol]); - if (ipprot) - ipprot->early_demux = enabled ? ipprot->early_demux_handler : - NULL; - -#if IS_ENABLED(CONFIG_IPV6) - ip6prot = rcu_dereference(inet6_protos[protocol]); - if (ip6prot) - ip6prot->early_demux = enabled ? ip6prot->early_demux_handler : - NULL; -#endif - rcu_read_unlock(); -} - -static int proc_tcp_early_demux(struct ctl_table *table, int write, - void __user *buffer, size_t *lenp, loff_t *ppos) -{ - int ret = 0; - - ret = proc_dointvec(table, write, buffer, lenp, ppos); - - if (write && !ret) { - int enabled = init_net.ipv4.sysctl_tcp_early_demux; - - proc_configure_early_demux(enabled, IPPROTO_TCP); - } - - return ret; -} - -static int proc_udp_early_demux(struct ctl_table *table, int write, - void __user *buffer, size_t *lenp, loff_t *ppos) -{ - int ret = 0; - - ret = proc_dointvec(table, write, buffer, lenp, ppos); - - if (write && !ret) { - int enabled = init_net.ipv4.sysctl_udp_early_demux; - - proc_configure_early_demux(enabled, IPPROTO_UDP); - } - - return ret; -} - static int proc_tfo_blackhole_detect_timeout(struct ctl_table *table, int write, void __user *buffer, @@ -638,14 +583,14 @@ static struct ctl_table ipv4_net_table[] = { .data = &init_net.ipv4.sysctl_udp_early_demux, .maxlen = sizeof(int), .mode = 0644, - .proc_handler = proc_udp_early_demux + .proc_handler = proc_douintvec_minmax, }, { .procname = "tcp_early_demux", .data = &init_net.ipv4.sysctl_tcp_early_demux, .maxlen = sizeof(int), .mode = 0644, - .proc_handler = proc_tcp_early_demux + .proc_handler = proc_douintvec_minmax, }, { .procname = "ip_default_ttl", diff --git a/net/ipv6/ip6_input.c b/net/ipv6/ip6_input.c index 57d84accbf1e..2bdb03a45baf 100644 --- a/net/ipv6/ip6_input.c +++ b/net/ipv6/ip6_input.c @@ -47,18 +47,25 @@ #include <net/inet_ecn.h> #include <net/dst_metadata.h> +void udp_v6_early_demux(struct sk_buff *); +void tcp_v6_early_demux(struct sk_buff *); static void ip6_rcv_finish_core(struct net *net, struct sock *sk, struct sk_buff *skb) { - void (*edemux)(struct sk_buff *skb); - - if (net->ipv4.sysctl_ip_early_demux && !skb_dst(skb) && skb->sk == NULL) { - const struct inet6_protocol *ipprot; - - ipprot = rcu_dereference(inet6_protos[ipv6_hdr(skb)->nexthdr]); - if (ipprot && (edemux = READ_ONCE(ipprot->early_demux))) - edemux(skb); + if (READ_ONCE(net->ipv4.sysctl_ip_early_demux) && + !skb_dst(skb) && !skb->sk) { + switch (ipv6_hdr(skb)->nexthdr) { + case IPPROTO_TCP: + if (READ_ONCE(net->ipv4.sysctl_tcp_early_demux)) + tcp_v6_early_demux(skb); + break; + case IPPROTO_UDP: + if (READ_ONCE(net->ipv4.sysctl_udp_early_demux)) + udp_v6_early_demux(skb); + break; + } } + if (!skb_valid_dst(skb)) ip6_route_input(skb); } diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c index 549ea5211bfa..7a5a7a4265cf 100644 --- a/net/ipv6/tcp_ipv6.c +++ b/net/ipv6/tcp_ipv6.c @@ -1673,7 +1673,7 @@ static int tcp_v6_rcv(struct sk_buff *skb) goto discard_it; } -static void tcp_v6_early_demux(struct sk_buff *skb) +void tcp_v6_early_demux(struct sk_buff *skb) { const struct ipv6hdr *hdr; const struct tcphdr *th; @@ -2028,12 +2028,7 @@ struct proto tcpv6_prot = { .diag_destroy = tcp_abort, }; -/* thinking of making this const? Don't. - * early_demux can change based on sysctl. - */ -static struct inet6_protocol tcpv6_protocol = { - .early_demux = tcp_v6_early_demux, - .early_demux_handler = tcp_v6_early_demux, +static const struct inet6_protocol tcpv6_protocol = { .handler = tcp_v6_rcv, .err_handler = tcp_v6_err, .flags = INET6_PROTO_NOPOLICY|INET6_PROTO_FINAL, diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c index fb71da6c6c74..9b0cae403027 100644 --- a/net/ipv6/udp.c +++ b/net/ipv6/udp.c @@ -912,7 +912,7 @@ static struct sock *__udp6_lib_demux_lookup(struct net *net, return NULL; } -static void udp_v6_early_demux(struct sk_buff *skb) +void udp_v6_early_demux(struct sk_buff *skb) { struct net *net = dev_net(skb->dev); const struct udphdr *uh; @@ -1533,12 +1533,7 @@ int compat_udpv6_getsockopt(struct sock *sk, int level, int optname, } #endif -/* thinking of making this const? Don't. - * early_demux can change based on sysctl. - */ -static struct inet6_protocol udpv6_protocol = { - .early_demux = udp_v6_early_demux, - .early_demux_handler = udp_v6_early_demux, +static const struct inet6_protocol udpv6_protocol = { .handler = udpv6_rcv, .err_handler = udpv6_err, .flags = INET6_PROTO_NOPOLICY|INET6_PROTO_FINAL, -- 2.30.2