Attention is currently required from: pespin.

View Change

4 comments:

• File libosmo-trx/include/osmocom/trx/ep.h:

  • Patch Set #1, Line 27: struct osmo_trx_ep_cfg {

    Let's please have separate setter functions for all these, like osmo_stream, so we don't need to bre […]

    Done

• File libosmo-trx/src/trx_ep.c:

  • Patch Set #1, Line 60: uint8_t pdu_ver; /* TRXD PDU version in use */

    iirc PDU VER stuff is defined/envisioned only for TRX0, so this can be moved to struct osmo_trx_ep.

    No. The TRXD PDU version is actually negotiated for each endpoint separately. At least osmo-bts does negotiate the version for each transceiver connection individually, and osmo-trx handles this properly.

    Only the power on/off command propagates to other connections.

  • Patch Set #1, Line 70: struct osmo_trx_ep_chan *chans; /* array of cfg.num_chans channels */

    If the chans array is allocated at the end of the struct, it probably makes sense to use "struct osm […]

    See my comment below.

  • Patch Set #1, Line 302: ep->chans = talloc_zero_array(ep, struct osmo_trx_ep_chan, cfg->num_chans);

    Ah I see you are allocating the array here as a pointer to a seaprate memory region. […]

    I don't see the need for such a premature optimization, sorry. A `struct osmo_trx_ep` is normally allocated once and remains in the memory until the PHY connection is closed. It's not like we're allocating/freeing tons of endpoints per second.

To view, visit change 43111. To unsubscribe, or for help writing mail filters, visit settings.