ANNOUNCEMENT: Live Wireshark University & Allegro Packets online APAC Wireshark Training Session
April 17th, 2024 | 14:30-16:00 SGT (UTC+8) | Online

Wireshark-dev: Re: [Wireshark-dev] Informing user of incomplete dissection

From: Michael Mann <mmann78@xxxxxxxxxxxx>
Date: Fri, 4 Mar 2016 20:06:30 -0500
DISSECTOR_ASSERT is definitely wrong, that's intended for detecting coding screw ups not for anything the wire generates.
 
You should always put all bytes in some sort of field so that the user has access to them.  Fieldname could be a clue, like my_dissector.undissected_bytes.
 
Expert info is probably the best way to go because it should color some fields of your protocol differently to make the user aware of the incomplete dissection.  There's even a specific expert info type, PI_UNDECODED to note such an occurrence (typically PI_WARN)
 
 
 
-----Original Message-----
From: Mike Morrin <morrinmike@xxxxxxxxx>
To: Developer support list for Wireshark <wireshark-dev@xxxxxxxxxxxxx>
Sent: Fri, Mar 4, 2016 7:55 pm
Subject: [Wireshark-dev] Informing user of incomplete dissection

I have a dissector which does not yet dissect all of the possible information elements of a protocol.

What should the dissector do when it recognises that an IE cannot be handled?
  • Ignoring the un-dissected octets leaves the user with a false sense of believing the dissection was complete
  • DISSECTOR_ASSERT is much too invasive.
  • Expert info, would seem to be the best option, but I recall that it was intended to be used for providing information about the captured packets, rather than about dissector deficiencies.
What is the generally accepted solution?

This email has been sent from a virus-free computer protected by Avast.
www.avast.com
___________________________________________________________________________
Sent via: Wireshark-dev mailing list <wireshark-dev@xxxxxxxxxxxxx>
Archives: https://www.wireshark.org/lists/wireshark-dev
Unsubscribe: https://wireshark.org/mailman/options/wireshark-dev
mailto:wireshark-dev-request@xxxxxxxxxxxxx?subject=unsubscribe