[L] Change in libosmo-sccp[master]: Implement M3UA-over-TCP (in addition to SCTP)

Attention is currently required from: fixeria, pespin. laforge has posted comments on this change. ( https://gerrit.osmocom.org/c/libosmo-sccp/+/35796?usp=email ) Change subject: Implement M3UA-over-TCP (in addition to SCTP) ...................................................................... Patch Set 1: (3 comments) Patchset: PS1: I agree that conceptually it is "mixing layers". However, from the user interface point of view, it is probably easier to have it exposed at the same level as m3ua. However, if this becomes a common pattern and thre would also be SUA-over-TCP or M2PA/M2UA/... over TCP, a more generic solution might be better. Let's check other SIGTRAN variants: * M2PA (RFC4165) reserves a TCP port but doesn't specify it: NO * M2UA (RFC3331) is specified both over SCTP and TCP: YES * SUA (RFC3868) does not mention TCP once: NO * IUA (RFC3057) is specified both over SCTP and TCP: YES * V5UA (RFC3807) is specified both over SCTP and TCP: YES So it seems that several other xUA versions also are specified to run on top of TCP, which (assuming we ever implement more of these) makes it a bit more clumsy to have everything duplicated at the level at whcih we're doing it now in this patch. I'm not saying I have a very clear opinion about it or a strict requirement - just sharing all of the facts/arguments I found. btw: it seems holger has created a SCTP<->TCP converter ages ago based on cellmgr_ng: https://github.com/moiji-mobile/stp-bridge File src/osmo_ss7_asp.c: https://gerrit.osmocom.org/c/libosmo-sccp/+/35796/comment/9df1dd50_b04707d0 PS1, Line 853: ss7_asp_m3ua_tcp_srv_conn_cb note that all of this de-segmentation works differently with osmo_io. There, the de-segmentation call-back is passed to Given that stream_{cli,srv}.c already fully supports osmo_io and provides osmo_stream_srv_set_segmentation_cb / osmo_stream_cli_set_segmentation_cb, I think this mechanism should be used. I guess the only reason it isn't yet used in libosmo-sigtran is that it still calls osmo_stream_srv_create instead of osmo_stream_srv_create2 There is a patch for making that switch in I7d02037990f4af405839309510dc6c04e36c3369 but it's held back by some incompatibility/inconsistency with non-blocking connect of SCTP client sockets when using the io_uring backend 😞 As @andreas@eversberg.eu is trying to fix that right now, I guess the half-life of this code is going to be very low. File src/osmo_ss7_vty.c: https://gerrit.osmocom.org/c/libosmo-sccp/+/35796/comment/2f8968db_89a6be03 PS1, Line 62: "MTP3 User Adaptation (SCTP)\n" \ Ack -- To view, visit https://gerrit.osmocom.org/c/libosmo-sccp/+/35796?usp=email To unsubscribe, or for help writing mail filters, visit https://gerrit.osmocom.org/settings Gerrit-Project: libosmo-sccp Gerrit-Branch: master Gerrit-Change-Id: I8c76d271472befacbeb998a93bbdc9e8660d9b5d Gerrit-Change-Number: 35796 Gerrit-PatchSet: 1 Gerrit-Owner: fixeria &lt;vyanitskiy(a)sysmocom.de&gt; Gerrit-Reviewer: Jenkins Builder Gerrit-Reviewer: laforge &lt;laforge(a)osmocom.org&gt; Gerrit-Reviewer: pespin &lt;pespin(a)sysmocom.de&gt; Gerrit-CC: jolly &lt;andreas(a)eversberg.eu&gt; Gerrit-Attention: pespin &lt;pespin(a)sysmocom.de&gt; Gerrit-Attention: fixeria &lt;vyanitskiy(a)sysmocom.de&gt; Gerrit-Comment-Date: Fri, 09 Feb 2024 08:44:50 +0000 Gerrit-HasComments: Yes Gerrit-Has-Labels: No Comment-In-Reply-To: pespin &lt;pespin(a)sysmocom.de&gt; Comment-In-Reply-To: fixeria &lt;vyanitskiy(a)sysmocom.de&gt; Gerrit-MessageType: comment