fsm4 branch and handover_test.c

historical

Hi Philipp,

I promised you to look at handover_test.c on the new fsm4 branch.

Looking at it now, I immediately ran into FSM events being dropped on the floor
because it was in the wrong state. When I initially fix that, the test runs
into the MGCP FSM not existing.

So at the end of the day, it is all related to the recently added FSMs and how
they interact, and it seems to me that you are more familiar with those than
me. In other words, I'm playing the ball back into your corner now:

Look at osmo-bsc branch neels/fsm4, here are the differences to pmaier/fsm4:

- right at the start, I add a patch to properly wrap abis_rsl_sendmsg(). That's
  just cosmetic, really. It was doing its job well, just I don't trust that
  implicit linking. It's at the start because I also submitted it to gerrit.
  5eaaf6120e21
  (I'd link to git.osmocom.org but after 10 minutes the branch still isn't
  synced there, strange.)

- I add the DMSC logging category to handover_test.c. That enables the FSM's
  error messages to be logged. This only makes sense as long as that FSM is
  actually logging on DMSC, which doesn't seem a good fit.
  8128a49a6c6ce4ae

- I kick the gscon into the ACTIVE state when a conn is created.  That might be
  taking a shortcut, since I'm not sure how e.g. the conn->user_plane.fi_bts
  should be populated...
  c2f90c36c445b2a38

So, e.g. when running './handover_test 1', my code state enabled the logging
for them and uncovered the errors:

DMSC handover_logic.c:135 SUBSCR_CONN[0x564f2a1d3b70]{INIT}: Event HO_START not permitted
and then
DMSC handover_logic.c:135 SUBSCR_CONN[0x564f2a1d3b70]{WAIT_CC}: Event HO_START not permitted

These two are fixed, and now I'm stuck at:

DMSC handover_logic.c:333 SUBSCR_CONN[0x560525a1cb70]{WAIT_HO_COMPL}: Received Event HO_COMPL
DMSC bsc_subscr_conn_fsm.c:710 SUBSCR_CONN[0x560525a1cb70]{WAIT_HO_COMPL}: state_chg to WAIT_MDCX_BTS_HO
Assert failed fi ../../../../src/osmo-mgw/src/libosmo-mgcp-client/mgcp_client_fsm.c:603

i.e. the handover_test.c is trying to indicate a 'Handover Complete' to the
libbsc machinery by calling handover_test.c:send_ho_complete(), and it runs
into a wall of the fake conn having conn->user_plane.fi_bts == NULL.

At this point I could start figuring out how to fake the fi_bts, but in fact I
assume you're more proficient in that area, so please take over from here.

Thanks!

~N

-- 
- Neels Hofmeyr <nhofmeyr at sysmocom.de>          http://www.sysmocom.de/
=======================================================================
* sysmocom - systems for mobile communications GmbH
* Alt-Moabit 93
* 10559 Berlin, Germany
* Sitz / Registered office: Berlin, HRB 134158 B
* Geschäftsführer / Managing Directors: Harald Welte
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://lists.osmocom.org/pipermail/openbsc/attachments/20180305/59ffec6a/attachment.bin>