[PATCH] ip_queue and fragments

[PATCH] ip_queue and fragments

[Pablo Neira]

[-- Attachment #1: Type: text/plain, Size: 121 bytes --]

Hi,

I've tracing ip_queue source code and I think that it's not fragment 
aware. Am I missing anything?

regards,
Pablo

[-- Attachment #2: ip_queue_frags.patch --]
[-- Type: text/x-patch, Size: 537 bytes --]

diff -u -r1.1.1.1 ip_queue.c
--- a/net/ipv4/netfilter/ip_queue.c	11 May 2004 13:07:08 -0000	1.1.1.1
+++ b/net/ipv4/netfilter/ip_queue.c	4 Aug 2004 14:37:25 -0000
@@ -255,9 +255,10 @@
 				entry->skb->dev->hard_header_parse(entry->skb,
 				                                   pmsg->hw_addr);
 	}
-	
-	if (data_len)
-		memcpy(pmsg->payload, entry->skb->data, data_len);
+
+	if (data_len) 
+		if (skb_copy_bits(entry->skb, 0, pmsg->payload, data_len) != 0)
+			goto nlmsg_failure;
 		
 	nlh->nlmsg_len = skb->tail - old_tail;
 	return skb;

Re: [PATCH] ip_queue and fragments

[Patrick McHardy]

Pablo Neira wrote:

> Hi,
>
> I've tracing ip_queue source code and I think that it's not fragment 
> aware. Am I missing anything?

This check prevents copying data outside of skb->data:

        case IPQ_COPY_PACKET:
                if (copy_range == 0 || copy_range > entry->skb->len)
                        data_len = entry->skb->len;
                else
                        data_len = copy_range;

skb_copy_bits always stays inside limits if len is positive, so you can
do something like this:

if  (copy_range == 0)
  data_len = ~0UL;
else
  data_len = copy_range;

but you have to remove the jump to nlmsg_failure when skb_copy_bits fails.

Regards
Patrick

>------------------------------------------------------------------------
>
>diff -u -r1.1.1.1 ip_queue.c
>--- a/net/ipv4/netfilter/ip_queue.c	11 May 2004 13:07:08 -0000	1.1.1.1
>+++ b/net/ipv4/netfilter/ip_queue.c	4 Aug 2004 14:37:25 -0000
>@@ -255,9 +255,10 @@
> 				entry->skb->dev->hard_header_parse(entry->skb,
> 				                                   pmsg->hw_addr);
> 	}
>-	
>-	if (data_len)
>-		memcpy(pmsg->payload, entry->skb->data, data_len);
>+
>+	if (data_len) 
>+		if (skb_copy_bits(entry->skb, 0, pmsg->payload, data_len) != 0)
>+			goto nlmsg_failure;
> 		
> 	nlh->nlmsg_len = skb->tail - old_tail;
> 	return skb;
>  
>

Re: [PATCH] ip_queue and fragments

[Pablo Neira]

[-- Attachment #1: Type: text/plain, Size: 376 bytes --]

Hi Patrick,

Patrick McHardy wrote:

> skb_copy_bits always stays inside limits if len is positive, so you can
> do something like this:
>
> if  (copy_range == 0)
>  data_len = ~0UL;
> else
>  data_len = copy_range;
>
> but you have to remove the jump to nlmsg_failure when skb_copy_bits 
> fails.


thanks for your suggestions, I've applied them to my patch.

regards,
Pablo

[-- Attachment #2: ip_queue-fragments.patch --]
[-- Type: text/x-patch, Size: 604 bytes --]

diff -u -r1.1.1.1 ip_queue.c
--- a/net/ipv4/netfilter/ip_queue.c	4 Aug 2004 15:14:39 -0000	1.1.1.1
+++ b/net/ipv4/netfilter/ip_queue.c	6 Aug 2004 16:38:11 -0000
@@ -205,8 +205,8 @@
 		break;
 	
 	case IPQ_COPY_PACKET:
-		if (copy_range == 0 || copy_range > entry->skb->len)
-			data_len = entry->skb->len;
+		if (copy_range == 0)
+			data_len = ~0UL;
 		else
 			data_len = copy_range;
 		
@@ -257,7 +257,7 @@
 	}
 	
 	if (data_len)
-		memcpy(pmsg->payload, entry->skb->data, data_len);
+		skb_copy_bits(entry->skb, 0, pmsg->payload, data_len);
 		
 	nlh->nlmsg_len = skb->tail - old_tail;
 	return skb;

Re: [PATCH] ip_queue and fragments

[Patrick McHardy]

Pablo Neira wrote:

> thanks for your suggestions, I've applied them to my patch.

Actually your first patch was fine and my suggestion was stupid ;)
I'm going to put it in pom-ng later tonight.

Regards
Patrick