Attention is currently required from: pespin.
dexter has posted comments on this change. ( https://gerrit.osmocom.org/c/osmo-mgw/+/32218 )
Change subject: mgcp_codec: fix codec decision ......................................................................
Patch Set 5:
(3 comments)
File src/libosmo-mgcp/mgcp_codec.c:
https://gerrit.osmocom.org/c/osmo-mgw/+/32218/comment/932b783e_30d440a2 PS2, Line 440: conn->end.codec = mgcp_codec_find_same(conn, found_same_codec_opposite);
Ah I see what you mean, you are looking the codec in each conn list. […]
The data is not duplicated. Its just that each endpoint holds two connections and each connections has a codec list. This is the list that had been issued by the call agent during CRCX/MDCX. We can not put both into one list since then we would lose the info which side supports what.
File src/libosmo-mgcp/mgcp_network.c:
https://gerrit.osmocom.org/c/osmo-mgw/+/32218/comment/f8e569a8_0e2f5fe9 PS4, Line 509: return;
so you decided to drop the return code while still logging an ERROR. […]
The caller is not doing anything meaningful with it but maybe it is better to drop the packet than sending it with a wrong PT.
https://gerrit.osmocom.org/c/osmo-mgw/+/32218/comment/5f1a8dad_f78b9f05 PS4, Line 1166: mgcp_patch_pt(conn_dst, msg);
all this code path failing looks buggy. […]
Technically nothing has changed but I can see mgcp_send() might need some refactoring.