Title: SNDCP reassembly failure
Hi,
I have noticed that SNDCP re-assembly fails when the same N-PDU number is seen twice in a capture. The fragments of the second SNDCP frame are added wrongly to the first SNDCP frame, and therefore the second SNDCP frame is not re-assembled.
It is probably because msg_reassembled_table hashtable is not used in the code. So I have added the table and replaced fragment_add_seq() with fragment_add_seq_check() and it seems to work: details & proposed patch on bug report #1755.
Sample trace can be sent on request for validation.
Question: now if the very unlikely following sequence of packets exists in a trace:
……...
21:12:05.02 User1 sndcp NPDU 380 Segment 0
21:12:05.025 User2 sndcp NPDU 380 Segment 1
21:12:05.07 User1 sndcp NPDU 380 Segment 1
……...
I think the re assembly will also fail, won't it ? What would be the way to ensure that reassembly is done per user --> could we use a tap for this kind of issue ?
Thks and regards
Cyrille
