Skip to contentSkip to conversation

simplify the code to dump the conntrack table

PATCH In review Pablo Neira Ayuso 2 participants 4 messages Sep 7, 2006 – Sep 14, 2006

Subsystems (2)
lore
Share & tools
lore

Pablo Neira Ayuso

author maintainer
#0 lore raw

PATCH 1/1 simplify the code to dump the conntrack table

Merge the bits to dump the conntrack table and the ones to dump and zero counters in a single piece of code. This patch does not change the default behaviour if accounting is not enabled.

[PATCH] Simplify the code to dump the conntrack table

Merge the bits to dump the conntrack table and the ones to dump and zero counters in a single piece of code. This patch does not change the default behaviour if accounting is not enabled.

net/ipv4/netfilter/ip_conntrack_netlink.c +8 -53
Index: net-2.6/net/ipv4/netfilter/ip_conntrack_netlink.c
===================================================================
@@ -436,6 +436,9 @@ restart:
 				cb->args[1] = (unsigned long)ct;
 				goto out;
 			}
+			if (NFNL_MSG_TYPE(cb->nlh->nlmsg_type) == 
+						IPCTNL_MSG_CT_GET_CTRZERO)
+				memset(&ct->counters, 0, sizeof(ct->counters));
 		}
 		if (cb->args[1]) {
 			cb->args[1] = 0;
@@ -451,46 +454,6 @@ out:
 	return skb->len;
 }
 
-#ifdef CONFIG_IP_NF_CT_ACCT
-static int
-ctnetlink_dump_table_w(struct sk_buff *skb, struct netlink_callback *cb)
-{
-	struct ip_conntrack *ct = NULL;
-	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=%u id=%u\n", __FUNCTION__, 
-			cb->args[0], *id);
-
-	write_lock_bh(&ip_conntrack_lock);
-	for (; cb->args[0] < ip_conntrack_htable_size; cb->args[0]++, *id = 0) {
-		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 (ctnetlink_fill_info(skb, NETLINK_CB(cb->skb).pid,
-		                        	cb->nlh->nlmsg_seq,
-						IPCTNL_MSG_CT_NEW,
-						1, ct) < 0)
-				goto out;
-			*id = ct->id;
-
-			memset(&ct->counters, 0, sizeof(ct->counters));
-		}
-	}
-out:	
-	write_unlock_bh(&ip_conntrack_lock);
-
-	DEBUGP("leaving, last bucket=%lu id=%u\n", cb->args[0], *id);
-
-	return skb->len;
-}
-#endif
-
 static const size_t cta_min_ip[CTA_IP_MAX] = {
 	[CTA_IP_V4_SRC-1]	= sizeof(u_int32_t),
 	[CTA_IP_V4_DST-1]	= sizeof(u_int32_t),
@@ -775,22 +738,14 @@ ctnetlink_get_conntrack(struct sock *ctn
 		if (msg->nfgen_family != AF_INET)
 			return -EAFNOSUPPORT;
 
-		if (NFNL_MSG_TYPE(nlh->nlmsg_type) ==
-					IPCTNL_MSG_CT_GET_CTRZERO) {
-#ifdef CONFIG_IP_NF_CT_ACCT
-			if ((*errp = netlink_dump_start(ctnl, skb, nlh,
-						ctnetlink_dump_table_w,
-						ctnetlink_done)) != 0)
-				return -EINVAL;
-#else
+#ifndef CONFIG_IP_NF_CT_ACCT
+		if (NFNL_MSG_TYPE(nlh->nlmsg_type) == IPCTNL_MSG_CT_GET_CTRZERO)
 			return -ENOTSUPP;
 #endif
-		} else {
-			if ((*errp = netlink_dump_start(ctnl, skb, nlh,
-		      		                        ctnetlink_dump_table,
-		                                	ctnetlink_done)) != 0)
+		if ((*errp = netlink_dump_start(ctnl, skb, nlh,
+	      		                        ctnetlink_dump_table,
+	                                	ctnetlink_done)) != 0)
 			return -EINVAL;
-		}
 
 		rlen = NLMSG_ALIGN(nlh->nlmsg_len);
 		if (rlen > skb->len)
net/netfilter/nf_conntrack_netlink.c +8 -57

86 lines not on this page · raw message

#1 lore raw
Quoted 3 lines · Pablo Neira Ayuso

Merge the bits to dump the conntrack table and the ones to dump and zero counters in a single piece of code. This patch does not change the default behaviour if accounting is not enabled.

Nice cleanup, thanks. And it saves me from removing the id based dumping, which I forgot to change for the counter case :)

BTW, what about my last question regarding your early_drop patch?

author maintainer
#2 lore raw
Quoted 7 lines · Patrick McHardy
Quoted 3 lines · Pablo Neira Ayuso

Merge the bits to dump the conntrack table and the ones to dump and zero counters in a single piece of code. This patch does not change the default behaviour if accounting is not enabled.

Nice cleanup, thanks. And it saves me from removing the id based dumping, which I forgot to change for the counter case :)

Thanks. So I guess that the next question is if there is any plan to remove the id ;)

Quoted 1 line

BTW, what about my last question regarding your early_drop patch?

Sorry, I forgot to do it, please mangle the patch and use '>' instead of '>=', otherwise the maximum number of conntracks will be ip_conntrack_max-1.

BTW, did you have the chance to have a look these patches?

http://lists.netfilter.org/pipermail/netfilter-devel/2006-August/025345.html
http://lists.netfilter.org/pipermail/netfilter-devel/2006-August/025353.html

@Jozsef: I forgot to include you in the CC, the second patch above touches some of your TCP tracking bits '[PATCH 2/3][CONNTRACK] Introduce the pickup facilities to take over TCP connections'

Signature · 2 lines
The dawn of the fourth age of Linux firewalling is coming; a time of
great struggle and heroic deeds -- J.Kadlecsik got inspired by J.Morris
#3 lore raw
Quoted 8 lines · Pablo Neira Ayuso
Quoted 6 lines · Patrick McHardy
Quoted 3 lines · Pablo Neira Ayuso

Merge the bits to dump the conntrack table and the ones to dump and zero counters in a single piece of code. This patch does not change the default behaviour if accounting is not enabled.

Quoted 6 lines
Quoted 2 lines

Nice cleanup, thanks. And it saves me from removing the id based dumping, which I forgot to change for the counter case :)

Thanks. So I guess that the next question is if there is any plan to remove the id ;)

I think that question has been discussed enough. Lets get rid of it. Small complaint about this patch btw, it didn't compile without conntrack accounting since the counters are only present when it is enabled.

Quoted 5 lines
Quoted 1 line

BTW, what about my last question regarding your early_drop patch?

Sorry, I forgot to do it, please mangle the patch and use '>' instead of '>=', otherwise the maximum number of conntracks will be ip_conntrack_max-1.

Done, thanks.

Quoted 3 lines

BTW, did you have the chance to have a look these patches?

http://lists.netfilter.org/pipermail/netfilter-devel/2006-August/025345.html

Not yet.

Quoted 1 line

http://lists.netfilter.org/pipermail/netfilter-devel/2006-August/025353.html

There were a couple of questions, still waiting for your response.