netfilter: nf_conntrack: don't always initialize ct->proto

ct->proto is big(60 bytes) due to structure ip_ct_tcp, and we don't need
to initialize the whole for all the other protocols. This patch moves
proto to the end of structure nf_conn, and pushes the initialization down
to the individual protocols.

Signed-off-by: Changli Gao <xiaosuo@gmail.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>

diff --git a/include/net/netfilter/nf_conntrack.h b/include/net/netfilter/nf_conntrack.h
index caf17db..abfff1e 100644 (file)
--- a/include/net/netfilter/nf_conntrack.h
+++ b/include/net/netfilter/nf_conntrack.h
@@ -116,14 +116,14 @@ struct nf_conn {
        u_int32_t secmark;
 #endif
 
-       /* Storage reserved for other modules: */
-       union nf_conntrack_proto proto;
-
        /* Extensions */
        struct nf_ct_ext *ext;
 #ifdef CONFIG_NET_NS
        struct net *ct_net;
 #endif
+
+       /* Storage reserved for other modules, must be the last member */
+       union nf_conntrack_proto proto;
 };
 
 static inline struct nf_conn *

diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c
index 27a5ea6..0ba7d48 100644 (file)
--- a/net/netfilter/nf_conntrack_core.c
+++ b/net/netfilter/nf_conntrack_core.c
@@ -651,7 +651,8 @@ __nf_conntrack_alloc(struct net *net, u16 zone,
         * and ct->tuplehash[IP_CT_DIR_REPLY].hnnode.next unchanged.
         */
        memset(&ct->tuplehash[IP_CT_DIR_MAX], 0,
-              sizeof(*ct) - offsetof(struct nf_conn, tuplehash[IP_CT_DIR_MAX]));
+              offsetof(struct nf_conn, proto) -
+              offsetof(struct nf_conn, tuplehash[IP_CT_DIR_MAX]));
        spin_lock_init(&ct->lock);
        ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple = *orig;
        ct->tuplehash[IP_CT_DIR_ORIGINAL].hnnode.pprev = NULL;

diff --git a/net/netfilter/nf_conntrack_netlink.c b/net/netfilter/nf_conntrack_netlink.c
index b729ace..7f59be8 100644 (file)
--- a/net/netfilter/nf_conntrack_netlink.c
+++ b/net/netfilter/nf_conntrack_netlink.c
@@ -1375,6 +1375,7 @@ ctnetlink_create_conntrack(struct net *net, u16 zone,
        }
 #endif
 
+       memset(&ct->proto, 0, sizeof(ct->proto));
        if (cda[CTA_PROTOINFO]) {
                err = ctnetlink_change_protoinfo(ct, cda);
                if (err < 0)

diff --git a/net/netfilter/nf_conntrack_proto_dccp.c b/net/netfilter/nf_conntrack_proto_dccp.c
index 5292560..9ae57c5 100644 (file)
--- a/net/netfilter/nf_conntrack_proto_dccp.c
+++ b/net/netfilter/nf_conntrack_proto_dccp.c
@@ -452,6 +452,9 @@ static bool dccp_new(struct nf_conn *ct, const struct sk_buff *skb,
        ct->proto.dccp.role[IP_CT_DIR_ORIGINAL] = CT_DCCP_ROLE_CLIENT;
        ct->proto.dccp.role[IP_CT_DIR_REPLY] = CT_DCCP_ROLE_SERVER;
        ct->proto.dccp.state = CT_DCCP_NONE;
+       ct->proto.dccp.last_pkt = DCCP_PKT_REQUEST;
+       ct->proto.dccp.last_dir = IP_CT_DIR_ORIGINAL;
+       ct->proto.dccp.handshake_seq = 0;
        return true;
 
 out_invalid:

diff --git a/net/netfilter/nf_conntrack_proto_sctp.c b/net/netfilter/nf_conntrack_proto_sctp.c
index c6049c2..6f4ee70 100644 (file)
--- a/net/netfilter/nf_conntrack_proto_sctp.c
+++ b/net/netfilter/nf_conntrack_proto_sctp.c
@@ -413,6 +413,7 @@ static bool sctp_new(struct nf_conn *ct, const struct sk_buff *skb,
            test_bit(SCTP_CID_COOKIE_ACK, map))
                return false;
 
+       memset(&ct->proto.sctp, 0, sizeof(ct->proto.sctp));
        new_state = SCTP_CONNTRACK_MAX;
        for_each_sctp_chunk (skb, sch, _sch, offset, dataoff, count) {
                /* Don't need lock here: this conntrack not in circulation yet */

diff --git a/net/netfilter/nf_conntrack_proto_tcp.c b/net/netfilter/nf_conntrack_proto_tcp.c
index 3fb2b73..6f38d0e 100644 (file)
--- a/net/netfilter/nf_conntrack_proto_tcp.c
+++ b/net/netfilter/nf_conntrack_proto_tcp.c
@@ -1066,9 +1066,7 @@ static bool tcp_new(struct nf_conn *ct, const struct sk_buff *skb,
        BUG_ON(th == NULL);
 
        /* Don't need lock here: this conntrack not in circulation yet */
-       new_state
-               = tcp_conntracks[0][get_conntrack_index(th)]
-               [TCP_CONNTRACK_NONE];
+       new_state = tcp_conntracks[0][get_conntrack_index(th)][TCP_CONNTRACK_NONE];
 
        /* Invalid: delete conntrack */
        if (new_state >= TCP_CONNTRACK_MAX) {
@@ -1077,6 +1075,7 @@ static bool tcp_new(struct nf_conn *ct, const struct sk_buff *skb,
        }
 
        if (new_state == TCP_CONNTRACK_SYN_SENT) {
+               memset(&ct->proto.tcp, 0, sizeof(ct->proto.tcp));
                /* SYN packet */
                ct->proto.tcp.seen[0].td_end =
                        segment_seq_plus_len(ntohl(th->seq), skb->len,
@@ -1088,11 +1087,11 @@ static bool tcp_new(struct nf_conn *ct, const struct sk_buff *skb,
                        ct->proto.tcp.seen[0].td_end;
 
                tcp_options(skb, dataoff, th, &ct->proto.tcp.seen[0]);
-               ct->proto.tcp.seen[1].flags = 0;
        } else if (nf_ct_tcp_loose == 0) {
                /* Don't try to pick up connections. */
                return false;
        } else {
+               memset(&ct->proto.tcp, 0, sizeof(ct->proto.tcp));
                /*
                 * We are in the middle of a connection,
                 * its history is lost for us.
@@ -1107,7 +1106,6 @@ static bool tcp_new(struct nf_conn *ct, const struct sk_buff *skb,
                ct->proto.tcp.seen[0].td_maxend =
                        ct->proto.tcp.seen[0].td_end +
                        ct->proto.tcp.seen[0].td_maxwin;
-               ct->proto.tcp.seen[0].td_scale = 0;
 
                /* We assume SACK and liberal window checking to handle
                 * window scaling */
@@ -1116,13 +1114,7 @@ static bool tcp_new(struct nf_conn *ct, const struct sk_buff *skb,
                                              IP_CT_TCP_FLAG_BE_LIBERAL;
        }
 
-       ct->proto.tcp.seen[1].td_end = 0;
-       ct->proto.tcp.seen[1].td_maxend = 0;
-       ct->proto.tcp.seen[1].td_maxwin = 0;
-       ct->proto.tcp.seen[1].td_scale = 0;
-
        /* tcp_packet will set them */
-       ct->proto.tcp.state = TCP_CONNTRACK_NONE;
        ct->proto.tcp.last_index = TCP_NONE_SET;
 
        pr_debug("tcp_new: sender end=%u maxend=%u maxwin=%u scale=%i "