[PATCH] Support of multiple RSL connections for ABIS/ipaccess (BTS side)

[Holger Hans Peter Freyther]

On Mon, Jan 20, 2014 at 08:04:22PM +0100, Andreas Eversberg wrote:

Dear Andreas,

my initial comment was that information should be at a single place.
The reason is that due bitrot/misc/etc. can change and one might think
one talks about the same thing when in fact one doesn't.

In the spirit of avoiding broken windows we have to avoid such
inconsistencies when they come up.


> in my last patch (for osmo-bts), the trx_id is set in each entry of
> an array of ipaccess_unit structure. the first entry is used for OML
> link, trx_id is not used there, so it is set to 0. the second (and
> subsequent) array entry is used for RSL. there the trx_id is set to
> the TRX number, starting with 0. (sysmobts only uses one TRX.)
> >>>-		e1inp_ipa_bts_rsl_connect(line, "127.0.0.1", IPA_TCP_PORT_RSL);
> >>>+		e1inp_ipa_bts_rsl_connect(line, "127.0.0.1", IPA_TCP_PORT_RSL, 0);
> >>>-	line = abis_open(bts, btsb->bsc_oml_host, "sysmoBTS");
> >>>+	line = abis_open(bts, btsb->bsc_oml_host, "sysmoBTS", 1);
> >Why do you start with trx_id == 1 in the BTS part but in the manual
> >case with 0?

> the abis_open() function gets the number of TRX to be used, not the
> trx_id. as shown above, the abis_open shall create the OML link and
> 1 RSL link for 1 TRX.
> the e1inp_ip_bts_rsl_connect() function, as shown above, gets the
> parameter 0, which means it should connect the first RSL link, which
> uses trx_id 0.


With your patch there is:

* trx->nr
* trx_id in the parameters 
* trx_id in the ipaccess_unit

So how can we reduce the amount of duplication? trx->nr will not go
away. Can we change e1inp_ipa_bts_rsl_connect to already work on a
sign_link (that normally would be created more late, probably not)?

any ideas?

	holger