Comment # 24
on bug 8275
from Raúl Pérez Clavero
>
> Many of the packets in your sample capture (specifically of types request,
> subscribe, propose and cancel) show up as malformed with the current version
> of the patch. Is this correct (the packets are malformed) or is it a bug in
> the dissector?
Yes! It was a bug. When porting to wmem, I forgot to re-initialise a wmem_list.
The problem was only happening with reassembled packets and packets containing
more than one message. It was fixed.
>
> Also, once the dissector is finished, is it OK if we check it in without all
> the extra logging and testing code? It's not necessary for the actual
> dissection so I'd generally prefer not to have it included there. If you do
> want to include a test suite to run with the dissector it should probably go
> in the test/ instead so it is runnable through our automated test suite.
> This is probably not necessary though.
Absolutely. The logging was for me to help debugging. If you prefer not to
check it in, I have nothing to say.
>
> Two final notes on port range registration:
> - please use MAX_TCP_PORT instead of hard-coding 65536 when registering the
> range string
DONE!
> - I think you can use dissector_add_uint_range and
> dissector_delete_uint_range which will be simpler than calling range_foreach
> with your own callback functions?
>
Yes! Much simpler this way. Also done.
You are receiving this mail because:
- You are watching all bug changes.