[Guy Harris <gharris@xxxxxxxxxxxx>]

On Thu, Jan 04, 2001 at 08:16:27PM +0100, Michael T�xen wrote:
> Having an SCTP packet with two M3UA packet a protocol tree like
> 
> IP
> SCTP
> M3UA
> M3UA
> 
> would suggest that M3UA is transported inside M3UA.

It *might* suggest that it's transported inside M3UA, or it might not. 

We already have protocols where you can have more than one PDU within a
frame; consider, for example, the OSI CLNP - I've seen one capture file
that has at least one CLNP PDU with two COTP PDUs in it.

We currently display that as

	Frame
	IEEE 802.3
	LLC
	CLNP
	COTP
	COTP

We do the same with the NetBIOS-over-TCP session service - there can be
more than one NBSS PDU within a TCP segment.  That could theoretically
occur with *any* protocol that runs atop TCP, as TCP segment boundaries
aren't visible to the protocol above TCP.

I'd be inclined to show something like

> The problem is, that SCTP supports the bundling of multiple M3UA packets in
> on SCTP packet. Furthermore in the SCTP packet there are some bytes belonging
> to SCTP, then the M3UA stuff of packet 1, then SCTP bytes and then the M3UA
> stuff of packet 2 and so on.
> So the byte sequence does not correspond to the linear structure of the 
> protocol stack.

as

	...

	SCTP
	M3UA
	SCTP
	M3UA