Fix #4651: Improve SOMEIP.fragment()

[polybassa]

Usually, payloads are stored inside SOMEIP.data[0]. In the past I made this become a PacketListField, so that customizations for custom data interpretations can be added to the SOMEIP.get_payload_cls_by_srv_id().

However, this needs to be respected in the fragment function. This PR adds support for this custom handling. Also, there is still the possibility to use add_payload, which would create a SOMEIP/Raw packet. Therefore this changes will determine, if the payload is outside or inside the SOMEIP packet.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 82.12%. Comparing base (1dcad5d) to head (8f0978d).
Report is 9 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #4655      +/-   ##
==========================================
+ Coverage   81.53%   82.12%   +0.58%     
==========================================
  Files         359      359              
  Lines       86565    86583      +18     
==========================================
+ Hits        70582    71106     +524     
+ Misses      15983    15477     -506     

Files with missing lines	Coverage Δ
scapy/contrib/automotive/someip.py	95.17% <100.00%> (+1.11%)	⬆️

... and 27 files with indirect coverage changes

[gpotter2]

Aw, conflicts :/

[polybassa]

Aw, conflicts :/

fixed

[gpotter2]

Reading only this code and not the spec, I kinda feel like "data" could have been a Padding payload. It all in all seems a bit complicated.

If your fix fixes the issue, it's fine to merge it though.

[polybassa]

The current implementation "tries" to do two things which are somewhat in conflict with each other.

1. Multiple SOME/IP packets in one "frame" (Ether/IP/UDP/SOMEIP/SOMEIP/SOMEIP)
2. Custom payload classes which should support dynamic import of definitions

To support multiple packets in one frame (1), #4632 was merged. This uses bind_layers of SOMEIP

Custom payload classes are supported by an explicit data field for which the callback can be overloaded dynamically:
https://github.com/polybassa/scapy-1/blob/6ae85f977f86e332cf6630049372de3f06ebe364/scapy/contrib/automotive/someip.py#L129

By moving the data field in the padding payload, we would need to call bind_layers(SOMEIP, A) for each specific payload time, which then would need to bind to SOMEIP again, to have multiple SOMEIP packets supported bind_layers(A, SOMEIP)

I would say, the current solution is more "straight forward". Do you see an alternative way of doing this?

[gpotter2]

Fine by me !