Closed lminiero closed 2 years ago
strange there was no test to catch this :(
strange there was no test to catch this :(
@pabuhler not sure, but my guess is this would have probably been tricky to catch using unit tests. We only noticed the issue when we were notified by someone using 2.4.1 about problems with srtp_unprotect
on incoming RTCP on the server, but there were no reported failures in srtp_protect
for outgoing RTP or RTCP for instance: when we worked on replicating the issue, we noticed media wouldn't work either, and that was because srtp_protect
would "succeed", but then the browser would not render anything, because of failures to unprotect on that end.
This PR fixes a typo introduced by #559, where for
SRTP_AES_GCM_256
the returned length isSRTP_AES_ICM_256_KEY_LEN_WSALT
instead ofSRTP_AES_GCM_256_KEY_LEN_WSALT
, which causes problems when SRTP sessions are created using AES-GCM-256.