Comment # 15
on bug 9920
from Hadriel Kaplan
Patch uploaded:
https://code.wireshark.org/review/#/c/918/
Interestingly, solving issue (1) the way I did also resolved issue (2). Because
it's now ref-counted, so even though SDP gives RTP two different dyn_payloads
in the INVITE for the same conversation, RTP doesn't truly free the first one
because SDP still has a ref to it. So when SDP install the first one again in
the 200 OK, it's still valid.
There's still the issue of SDP inserting the rtpmap info into the wrong
channel, but that's unrelated and doesn't cause a crash.
Also, the uploaded patch resolves the memory leaks, but only insofar as they're
all destroyed at the end of the wmem file scope. SDP (and the other protocols)
still don't remove the RTP conversations when they could, when each call ends.
You are receiving this mail because:
- You are watching all bug changes.