Wireshark-dev: Re: [Wireshark-dev] Fwd: And again BER errors while decoding H248packets

From: "Graeme Lunt" <graeme.lunt@xxxxxxxxx>
Date: Thu, 5 Oct 2006 20:07:57 +0200
Hi,

> So, what's the conclusion? Is this packet correct or not?

Here is my opinion on the subject. 

I think the packet is correct - zero length encoding is perfectly fine.
I see zero length SETs every day.

[If you want an example - most X.400 systems use the default ia5 repertoire
for text bodyparts in the ASN.1 below - resulting in a zero length SET. I've
even seen implementations encode it in 6 bytes - which beats a "disgusting"
- but perfectly valid - indefinite length encoding.

IA5TextBodyPart ::= SEQUENCE {
  parameters  IA5TextParameters,
  data        IA5TextData
}

IA5TextParameters ::= SET {repertoire  [0]  Repertoire DEFAULT ia5
}
]

> I think that:
> 
> 1. If the packet is incorrect (in terms of BER-encoding), then
>    the "BER error" message should be put back, but the dissection
>    of the current sequence should not be aborted.

I don't think it is an error. 
(If there had been a constraint then it would have been an error - but
asn2wrs doesn't do any bounds checking at the moment.)
 
> 2. If we don't know whether it is correct or not, probably we should
>    give warning instead of error?

I think it is correct.

> 3. Anyway, those 'one item of zero length' lines in the decoded
>    packet tree should be changed to '0 items' or removed completely.

If I recall correctly, this issue was when displaying a zero length SEQUENCE
OF? 
In which case, I think it should be changed to '0 items' (and not removed)
to be clear and consistent.

Graeme