Hadriel Kaplan
changed
bug 11413
Comment # 5
on bug 11413
from Hadriel Kaplan
(In reply to Daniel Starke from comment #4)
> Changing the code to the following does not change the problem for the given
> example trace.
Actually, it does change it - but it just hits a bug. :)
The change is that this dissector now correctly sets the desegment info in the
first pass; unfortunately wireshark's RTP dissector appears to have a bug
itself, not specific to Lua.
The RTP dissector doesn't store the frag information for more than one fragment
packet/pair, so in the second pass of the file when it's processing the second
packet it remembers that the first packet is a fragment, so it skips invoking
the dissector for the second packet; but it doesn't remember that the second
packet is also a fragment, so when it's processing the third packet it doesn't
concatenate the buffers from the first, second, and third packets into one big
buffer, but instead just invokes dissection with the buffer of the third packet
alone.
Looks like a simple fix (... famous last words). I'll upload it shortly.
You are receiving this mail because:
- You are watching all bug changes.