Attention is currently required from: pespin, dexter. laforge has posted comments on this change. ( https://gerrit.osmocom.org/c/osmo-hnbgw/+/26792 )
Change subject: ranap_rab_ass: add decoder and rewrite functions for RAB-AssignmentRequest/Response ......................................................................
Patch Set 15:
(1 comment)
File src/osmo-hnbgw/ranap_rab_ass.c:
https://gerrit.osmocom.org/c/osmo-hnbgw/+/26792/comment/3974a6ae_911c1f01 PS14, Line 198: int ranap_rab_ass_req_extract_inet_addr(struct osmo_sockaddr *addr, uint8_t *rab_id, RANAP_RANAP_PDU_t *ranap_pdu)
Its actually two times at the moment for each messages at the moment. […]
As I stated, the best would likely be to use ranap_message which is an overall container for the RANAP message and all of its contents.
I would expect the ASN.1 decode is among the more expensive things that osmo-hnbgw does. However, performance is likely irrelevant in the kind of deployments we are looking at, where we have a few hundred hNB. Remember, it is control plane processing only.
To me it's more a question of code maintainability and KISS, performance is not super-critical.
We do have a relatively well-established way how to deal with ranap_message in the SGSN and MSC, so why should we deviate from this existing pattern here?