It doesn't make much sense to pull whole libgtp to implement only the
user plan side of GTP. Let's drop it and add a minimal GTP-U
implementation here.
Related: SYS#5516
Change-Id: I53ad4915aaed3bc7574036e963be10514e370fe2
This is needed for instance:
* AUDIO: video calls
* GTP: secondary pdp contexts
For better abstraction, now both AUDIO and GTP conns use unique ID
namespaces, and one ID is used to identify each one. Each conn relate in
turn to a UE since a context_id is passed during connection
establishment.
Related: SYS#5516
Change-Id: Ib3f60d5ba21defe5259c25e2034fc2217c4d93df
This first implementation uses libgtp to manage the GTPv1-U socket and
pdp contexts, as a proof of concept to have something working for now.
At a later step, it is expected to drop libgtp and use some adhoc code
to handle GTP-U.
Related: SYS#5516
Change-Id: I5a6f5dfc4e508c92adb35210b4dc576d64353366
This first version can send and receive regular RTP, but does not talk IuUP.
Support for IuUP will be added in a future patch. This way we can
already test the whole LLSK<->RTP path.
Change-Id: I9909a7c054ddaabb1bb63d7d06331cc79f642b5d
The LLSK socket is an interface allowing thid-parties to implement and
hook their RLC/MAC/RRC stack to osmo-hnodbe, which takes care of
interconnection with the rest of the RAN and core network (Iuh, RTP,
GTP-U, etc.).
Related: SYS#5516
Change-Id: Icaabb2206d6f141d4fba47dedf71f8ec37e6257d
The shutdown_fsm doesn't do much right now, but it's included there in
order to allow easy clean up of resources upon Iuh reconnect in the
future, as well as reset the lower layers.
Change-Id: Id35d692218af366843186b0966b62f65bb9eb620
The Iuh code will be further extended next to properly support
reconnect, and hnb will also gain proper shutdown support soon.
Change-Id: I6e94210ab06a34b70c61bb074c58d7b0f4ee75de
Let's use already available functionality to handle the conn for us.
This is specially useful since we gain a connect_cb.
Change-Id: I9d884231577aa47af22cd6319130ca5c9e595d61
Pau Espin