[NETFILTER]: ctnetlink: change table dumping not to require an unique ID
authorPatrick McHardy <kaber@trash.net>
Tue, 30 May 2006 01:24:58 +0000 (18:24 -0700)
committerDavid S. Miller <davem@sunset.davemloft.net>
Sun, 18 Jun 2006 04:29:03 +0000 (21:29 -0700)
Instead of using the ID to find out where to continue dumping, take a
reference to the last entry dumped and try to continue there.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv4/netfilter/ip_conntrack_netlink.c
net/netfilter/nf_conntrack_netlink.c

index af152e3623dcbb71ea3f84481a4aae1b310cd13e..33891bb1fde438a97c5d85516b3c25bb826d1ba4 100644 (file)
@@ -399,38 +399,54 @@ nfattr_failure:
 static int ctnetlink_done(struct netlink_callback *cb)
 {
        DEBUGP("entered %s\n", __FUNCTION__);
+       if (cb->args[1])
+               ip_conntrack_put((struct ip_conntrack *)cb->args[1]);
        return 0;
 }
 
 static int
 ctnetlink_dump_table(struct sk_buff *skb, struct netlink_callback *cb)
 {
-       struct ip_conntrack *ct = NULL;
+       struct ip_conntrack *ct, *last;
        struct ip_conntrack_tuple_hash *h;
        struct list_head *i;
-       u_int32_t *id = (u_int32_t *) &cb->args[1];
 
        DEBUGP("entered %s, last bucket=%lu id=%u\n", __FUNCTION__, 
                        cb->args[0], *id);
 
        read_lock_bh(&ip_conntrack_lock);
-       for (; cb->args[0] < ip_conntrack_htable_size; cb->args[0]++, *id = 0) {
+       for (; cb->args[0] < ip_conntrack_htable_size; cb->args[0]++) {
+restart:
+               last = (struct ip_conntrack *)cb->args[1];
                list_for_each_prev(i, &ip_conntrack_hash[cb->args[0]]) {
                        h = (struct ip_conntrack_tuple_hash *) i;
                        if (DIRECTION(h) != IP_CT_DIR_ORIGINAL)
                                continue;
                        ct = tuplehash_to_ctrack(h);
-                       if (ct->id <= *id)
-                               continue;
+                       if (last != NULL) {
+                               if (ct == last) {
+                                       ip_conntrack_put(last);
+                                       cb->args[1] = 0;
+                                       last = NULL;
+                               } else
+                                       continue;
+                       }
                        if (ctnetlink_fill_info(skb, NETLINK_CB(cb->skb).pid,
                                                cb->nlh->nlmsg_seq,
                                                IPCTNL_MSG_CT_NEW,
-                                               1, ct) < 0)
+                                               1, ct) < 0) {
+                               nf_conntrack_get(&ct->ct_general);
+                               cb->args[1] = (unsigned long)ct;
                                goto out;
-                       *id = ct->id;
+                       }
+               }
+               if (last != NULL) {
+                       ip_conntrack_put(last);
+                       cb->args[1] = 0;
+                       goto restart;
                }
        }
-out:   
+out:
        read_unlock_bh(&ip_conntrack_lock);
 
        DEBUGP("leaving, last bucket=%lu id=%u\n", cb->args[0], *id);
index 8f27fe9446f27857a616a260389163b1f6d3c78c..b8c7c567c9df5a0c3a27e6cbf01680854a7b8914 100644 (file)
@@ -407,6 +407,8 @@ nfattr_failure:
 
 static int ctnetlink_done(struct netlink_callback *cb)
 {
+       if (cb->args[1])
+               nf_ct_put((struct nf_conn *)cb->args[1]);
        DEBUGP("entered %s\n", __FUNCTION__);
        return 0;
 }
@@ -416,10 +418,9 @@ static int ctnetlink_done(struct netlink_callback *cb)
 static int
 ctnetlink_dump_table(struct sk_buff *skb, struct netlink_callback *cb)
 {
-       struct nf_conn *ct = NULL;
+       struct nf_conn *ct, *last;
        struct nf_conntrack_tuple_hash *h;
        struct list_head *i;
-       u_int32_t *id = (u_int32_t *) &cb->args[1];
        struct nfgenmsg *nfmsg = NLMSG_DATA(cb->nlh);
        u_int8_t l3proto = nfmsg->nfgen_family;
 
@@ -427,7 +428,9 @@ ctnetlink_dump_table(struct sk_buff *skb, struct netlink_callback *cb)
                        cb->args[0], *id);
 
        read_lock_bh(&nf_conntrack_lock);
-       for (; cb->args[0] < nf_conntrack_htable_size; cb->args[0]++, *id = 0) {
+       for (; cb->args[0] < nf_conntrack_htable_size; cb->args[0]++) {
+restart:
+               last = (struct nf_conn *)cb->args[1];
                list_for_each_prev(i, &nf_conntrack_hash[cb->args[0]]) {
                        h = (struct nf_conntrack_tuple_hash *) i;
                        if (DIRECTION(h) != IP_CT_DIR_ORIGINAL)
@@ -438,17 +441,30 @@ ctnetlink_dump_table(struct sk_buff *skb, struct netlink_callback *cb)
                         * then dump everything. */
                        if (l3proto && L3PROTO(ct) != l3proto)
                                continue;
-                       if (ct->id <= *id)
-                               continue;
+                       if (last != NULL) {
+                               if (ct == last) {
+                                       nf_ct_put(last);
+                                       cb->args[1] = 0;
+                                       last = NULL;
+                               } else
+                                       continue;
+                       }
                        if (ctnetlink_fill_info(skb, NETLINK_CB(cb->skb).pid,
                                                cb->nlh->nlmsg_seq,
                                                IPCTNL_MSG_CT_NEW,
-                                               1, ct) < 0)
+                                               1, ct) < 0) {
+                               nf_conntrack_get(&ct->ct_general);
+                               cb->args[1] = (unsigned long)ct;
                                goto out;
-                       *id = ct->id;
+                       }
+               }
+               if (last != NULL) {
+                       nf_ct_put(last);
+                       cb->args[1] = 0;
+                       goto restart;
                }
        }
-out:   
+out:
        read_unlock_bh(&nf_conntrack_lock);
 
        DEBUGP("leaving, last bucket=%lu id=%u\n", cb->args[0], *id);