Re:Question about skbuff

[Catalin Marinas <c_marinas@yahoo.com>]

Hello,

On Sat, 11 Aug 2001, lam wai wm wrote:

>
> [...]
>
>     struct tcphdr * tcph = (void *) ip + ip->ihl *4;

I think is better to write

      struct tcphdr * tcph = (void *) ((char *) ip + ip->ihl * 4);

in order to have the correct sum. It is possible to work with your casts
but is safer to covert it to (char *).

>
> [...]
>
>     if (tcplen != tcph->doff * 4)
>       printk("TCP length not match\n");
>     else
>       printk("TCP length match..\n");

What does this part say? If it says "match", it means that the packet
contains no data and everything is OK.

> The result is :
> Source addr : 161.40.211.86:6190
> Dest addr : 161.40.211.86:1037
> Length : 52
> Skb Data addr : c7ca3d1c
> IP addr : c7ca3d1c
> IP->ihl : 5
> TCP addr : c7ca3d30
> Data :
> data addr : c7ca3d50
> tail : c7ca3d50
> end : c7ca3d50

Total Length : 52 = 0x34			// OK
Skb Data addr : 0xc7ca3d1c			// OK
IP addr : 0xc7ca3d1c				// OK
IP->ihl : 5					// OK
TCP addr : 0xc7ca3d30 = 0xc7ca3d1c + 5 * 4	// OK
tail : 0xc7ca3d50 = 0xc7ca3d1c + 52		// OK

data addr == tail : I think this is OK and maybe this is an ACK packet
with no data, only the header. Check the fields: syn, rst, ack in tcphdr
structure (http://www.rfc-editor.org/rfc/rfc793.txt).

I do not have too much experience with skbuffs but I think that nothing is
wrong with the data address being equal with tail address (only if this
happens for all the packets).

-- 
Catalin








-
Kernelnewbies: Help each other learn about the Linux kernel.
Archive:       http://mail.nl.linux.org/kernelnewbies/
IRC Channel:   irc.openprojects.net / #kernelnewbies
Web Page:      http://www.kernelnewbies.org/