OsmoBTS fails to schedule SACCH filling like SI5
There are cases in which OsmoBTS is failing to schedule SACCH filling such as SYSTEM INFORAMTION 5.
This seems to be what's causing, among other things, #2963.
I've been instrumenting the code and found that the copying from BTS -> LCHAN is working as expected, and the lchan is started with valid information. Still,
lchan_sacch_get() returns NULL for all but its first invocation.
It seems there is a corner case when we have exactly only one SACCH filling. If two or more are set, the loop seems to work. If only one SACCH filling is specified, it's transmitted once and then never again.
This bug seems to go back all the way to 2012. It might be that OsmoBSC in earler incarnations always sent SI5+SI6 and therefore the problem only became visible when only SI5 but no other SI is scheduled on SACCH.