Previously the expectations for number of CRCX and MDCX messages from
MGW was adjusted unconditionally for LCLS tests. However this is only
necessary for MGW-loop type of LCLS. Use explicit variable (with default
value preserving current behavior) to decide whether to apply this
adjustment or not. This simplifies support for other kinds of LCLS
loops.
Change-Id: I07b2c56991977b5e80c372a5b8338f348f14c076
Related: OS#3659
Move duplicated code into separate functions to simplify adding test
cases which do not use MGW MDCX methods for closing the loop.
Change-Id: Idf9fc8e639b0ece662ab26c481899ef39fb7edfe
Related: OS#3659
Add two helper functions which retry a VTY command until the
result matches a regular expression or a configurable timeout
expires.
Use these functions in BSC test's f_ts_dyn_mode_get, which has
seen sporadic failures due to a race condition during channel
reconfiguration, in order to hopefully close this race.
Change-Id: I308ddb06e440c165fe1e73fe2c1fb78be2e1d510
Related: OS#3690
Instead of vaguely allowing any release messages to be present or not, exactly
pinpoint for each test case the exact release messages expected during lchan
release.
Related: an osmo-bsc change broke sending of RR Release messages, which was
utterly ignored and hence not caught by ttcn tests. That must not happen again.
I am not actually sure that these expectations are 100% correct; if errors
become apparent, we shall change the expectations in ttcn3 and then fix
osmo-bsc according to that.
Adjust f_expect_chan_rel() and callers,
and the Assignment procedures (as_assignment and f_establish_fully).
The current state of the bsc tests should all pass with osmo-bsc
Id3301df059582da2377ef82feae554e94fa42035
Related: OS#3413
Change-Id: Ibc64058f1e214bea585f4e8dcb66f3df8ead3845
Allow osmo-bsc sending a Deact SACCH messages in most cases. Prepare the
ttcn3-bsc-tests to not break just because of those messages that will soon be
sent.
When releasing an lchan, it makes sense to Deactivate SACCH on it, if it was
ever active. So far osmo-bsc was fairly reluctant to send Deactivate SACCH, but
osmo-bsc Id3301df059582da2377ef82feae554e94fa42035 is about to change that.
In most test cases, Deact SACCH are still optional, but in one case, the
current missing Deact SACCH will introduce a test failure: in the 'interleave'
of BSC_Tests.TC_ho_out_fail_no_ho_detect.
As soon as abovementioned osmo-bsc patch is merged, the test will pass again.
Also, as soon as Ibc64058f1e214bea585f4e8dcb66f3df8ead3845 is merged here, the
bsc tests will properly ensure whether Deact SACCH is sent or not in all tests.
Change-Id: I27da24dbe3184fa7a076a35f6fa6af457c1db8d2
Receive RR Release messages if they happen during lchan release. Add RR Release
to the alt{}s in both f_expect_chan_rel() to cover a whole bunch of test cases,
and in f_tc_ho_out_fail_no_ho_detect() which has its own release expectations.
Before this, RR Release messages would typically be lost in the RSL.clear
recently removed by Ie1be30c3f109dda8c58c523df508211f8e20aad3.
However, I still expect tests to pass after this, since the current osmo-bsc
master has a bug that omits RR Release messages (since [1]).
By applying this patch, both the buggy osmo-bsc (omitting RR Release) and the
fix of that [2] should pass the BSC tests. So far by accepting whatever comes
along, and not complaining if it doesn't come along.
A subsequent patch will more precisely ensure that exactly the expected
messages will be sent by osmo-bsc (Ibc64058f1e214bea585f4e8dcb66f3df8ead3845).
[1] osmo-bsc I4fd582b41ba4599af704d670af83651d2450b1db
commit 8b818a01b00ea3daad4ad58c162ac52b4f08a5cb
"subscr conn: properly forget lchan before release"
[2] osmo-bsc I666b3b4f45706d898d664d380bd0fd2b018be358
"fix: send RR Release (e.g. after BSSMAP Clear Cmd"
Related: OS#3413
Change-Id: I4e6d266d091b140f56b28312cb3c5d67ffcc3a59
Instead of placing an own set of channel release expectations, just use the
common f_expect_chan_rel() that exists for exactly this purpose.
This will also be in line with upcoming changes to tighten checking of the
lchan release messages.
Related: OS#3413
Change-Id: Ib7dd886472337e2deb968e6f9de6cecdb7855319
When we're expecting release, it can be non-deterministic / timing dependent to
flush the RSL queue. Particularly the RR Release message is typically already
in the queue when f_expect_chan_rel() is called and would be lost.
It turns out that none of the current callers need the flush feature.
If a test needs it, we can add a separate f_rsl_flush() function and call that,
no need to clutter up the f_expect_chan_rel() argument list. I had such
function but found that no caller needs it, so dropped it.
Related: OS#3413
Change-Id: Ie1be30c3f109dda8c58c523df508211f8e20aad3
The receive template for the BSSMAP HANDOVER PERFORMED MESSAGE is
missing, lets add one.
Related OS#3645
Change-Id: I527913203b2d5bfa26c181c4bb79481a9cd283ae
* remove unused parameters
* add expected results
* use control for default test execution
* only tests which could (in theory) pass are left in control
Previously generic control had some tests which never called
"setverdict(pass)" which means they would always fail. Such tests might
be useful during development but should not be part of test set executed
by default.
Change-Id: I91910e18f108f8ee9a3b76ced16c420fdbdb665a
Related: OS#2890
Some test env may take more than 10 seconds to (re-)start a BTS, so
let's make it variable through this parameter.
For instance it was observed that running osmo-bts-sysmo through ssh
inside a sysmobts can sometimes take a good number of seconds (specially
because ssh connection can take a while to be established successfully).
Change-Id: Ieb046358d8266ac94bd7b9e916e85f84de3ad319
Drop 'return' statements after 'mtc.stop' as they cause following
warning from TTCN-3 compiler:
warning: Control never reaches this statement
Change-Id: I6210ecf5fcb39f751116ad63a69d2ae8651a60c5
Check that channel number IE is present in the received message. This
fixes following warning:
The value returned by function `@RSL_Types.f_rsl_find_ie' is not used
Change-Id: Icc6824d7acef809d76c70b5ff060a8261a82e2be
Related: OS#3659
The osmo-bsc.cfg should work on localhost. Having arbitrary IP address
in it seems confusing and might lead to hard-to-debug errors.
Change-Id: Ideb8a78348af435f9992c42bce0bb197d1e73e04
We cannot notify RSL Emulation layer about expecting a specific FN
(during ts_RSLDC_ChanRqd) because we only know the FN after sending the
RACH request, and since notification to RSL Emulation happens async, it
can happen (verified) that ChanRqd message from BTS arrives and is
handled before we register the RACH req into the ConnectionTable.
Change-Id: I438fd3ee82d88498d928dbcc89ce9bd80d37ab64
There is no need to pass session state from f_main_mo() to a
choosen EUSE handler (e.g. f_ss_echo), because a handler
itself is capable to extract the session state IE.
Change-Id: I1054baf3e7dafd05b797610b586e6202740f07b6
Unfortunately all component.stop can not be called from the non-mtc
component. f_shutdown is a wrapper that will try to shutdown a test in
the best way calling all component.stop if it is called from the mtc and
just stopping the mtc if called from any other component.
Change-Id: I9b71f7f7bd70d2da21fbad60c340d5bf8b3b9536
Some TC_no_out_fail_* testcases still used valueof() to get g_pars which
always set aoip to true.
Use f_gen_test_hdlr_pars() now so these tests can properly detect if
they are run with sccplite of aoip and can adjust their expectations
accordingly.
Change-Id: Ic164827d63c9f5452888951bba4c197c3fe6f57b
this allows us to match on specifically 3G MM AUTH REQ, whereas
so far we only had a generic template that matched the 2G and 3G
auth req.
Change-Id: I392d61d1348bee9c88abe4bb938e99b2c3702a94
Matching the CONFIG prompt using implicit '\w+(*)' pattern is
a bad idea, because it can actually match almost anything:
- 'OsmoBlaBla(config)# ',
- 'OsmoBlaBla(config) ',
- 'OsmoBlaBla> ',
- 'Mahlzeit'!
One problem is that the parentheses are interpreted as a matching
operator (which is used to group an expression), so they should
have been escaped by '\'. Another problem is that this pattern
is not complete, because '\# ' is missing. Let's fix this!
Change-Id: I8a0e3fcfb0c4e5854b7b1e39296052e679c63c73
Related: OS#3675
Otherwise RSL layer fails this way when this event is received:
RSL_Emulation.ttcn:429 Receive operation on port IPA_PT succeeded, message from TC_chan_act_a51-RSL-IPA(3): @IPA_Emulation.ASP_IPA_Event: { up_down := ASP_IPA_EVENT_DOWN (0) } id 9
- Function main_client was stopped. PTC terminates.
RSL_Emulation.ttcn:429 Message with id 9 was extracted from the queue of IPA_PT.
RSL_Emulation.ttcn:430 setverdict(fail): none -> fail reason: "Received unknown primitive from IPA", new component reason: "Received unknown primitive from IPA"
We now fail with a clearer message "Lost IPA connection!". These
failures seem to happen under high load when the BTS doesn't get a
steady clock from osmo-trx.
Change-Id: Idc6565c9de72d98015d56a41e5616c46051c8a8d
Replace all calls to setverdict(fail) with f_shutdown() since I'd rather
fail and stop early in case we encounter an error.
Only replace setverdict(pass) with f_shutdown() if it is followed by
mtc.stop
Remove internal function f_shutdown and use the one from Misc_Helpers
Change-Id: Ia8b01a1876e969d6f0760ea625e4df83af4f54ca
This function can now be called from anywhere to try and safely shutdown
a testcase. It is not optimal as we can't call "all component.stop" from
outside the mtc, but without any proper and orderly shutdown handling of
all our emulation components I believe this is the best we can do.
To use it:
import from Misc_Helpers all;
in your module and then call
Misc_Helpers.f_shutdown(__BFILE__, __LINE__);
You can also pass the function a verdict and a message and it will take care
of calling setverdict, but beware of the following:
While setverdict would accept any number of arguments as log message
and convert them to a log string f_shutdown expects one charstring.
It's possible to use the log2str function to use the log arguments in
setverdict for f_shutdown, for example
setverdict(fail, "Template didn't match: ", tmpl_foo);
would become
Misc_Helpers.f_shutdown(__BFILE__, __LINE__, fail, log2str("Template didn't match: ", tmpl_foo));
Change-Id: I84d1aa6732f6b748d2bfdeac8f6309023717f267
Absolute form of FAKE_RSSI command shall contain an optional
threshold value. Otherwise it's interpreted as relative form.
Change-Id: Ief89b2601277488bb1782b981aff1061ddaa6637
Temporary disable testing on FACCH/H because Calypso PHY is not stable.
Otherwise some tests fail half of the time due to this unstability.
Related: OS#3653
Change-Id: I078cdfbf8d27e543a723eab90f66b2fcda016b5d
The mncc guard timer in osmo-msc expires after 180 sec. This tests
terminates early and fails. Lets expand the timeout in order to give the
test a chance to pass.
Change-Id: I9f954a8807c0f0513422693ac24c647da0bfafd1
Related: OS#3599
Since OS#2988 was fixed, we should not overwrite nor ignore the
measurement results in f_build_meas_res_tmpl().
Change-Id: Ie902bfc7619181b528eafbce367c87e0b062243a
Add another IPA test to the BTS and BSC test suites.
This new test sends the header in one burst, followed by the
payload which is transmitted byte-per-byte.
The test uses an ID REQ message. If acting as a server, the test
can expect an ID RESP message. However, if acting as a client, the
server will close the connection when it receives the ID REQ.
The CTRL interface port on the BSC does not close the connection in
this case, so that particular port is skipped by the test for now.
Change-Id: If75cb90841bb25619b414f0cabe008a2428a9fdf
Related: OS#2010
Depends: I4804ccabd342b82d44e69dbc6eaaae220ec7d4e4
Run the chopped IPA ping test from the IPA_Testing module
as part of the BTS test suite. Contrary to the BSC version
of this test, this test listens for an IPA connection rather
than connecting to an IPA server. Make code in the IPA_Testing
module for accepting connections actually work.
Change-Id: I4804ccabd342b82d44e69dbc6eaaae220ec7d4e4
Related: OS#2010
With a related patch to libosmocore, this test will pass on
osmo-bsc's ctrl interface.
Change-Id: I20a4f251dbe3e75a5c53a0ff167998bbd1266f62
Depends: I9d7137c830981ccad03806b30b776e2b1f1b4699
Related: OS#2010
Use the output of 'git describe' to pin the commit checked out
for dependency titan.ProtocolModules.SGsAP_13.2.0_commit, i.e.
the tag which points to 81544cf008378058b95ff20d33f3770fadb7e235.
Using a plain commit hash will always re-trigger an update of
the working tree in this particular case because 'git describe'
won't ever return an error.
Change-Id: Ia994b5c4488fd9b0d410b992e515c6eec4dd23b1
RSPRO is the protocol used by osmo-remsim. It is embedded into an IPA
multiplex, and hence the TTCN-3 IPA code needs some extension to cover
support for it.
Change-Id: I536d6843b3e65b3ee35fbbcd6353e0fb0ce21c8e
Most differences between sccplite and AoIP are visible during the
assignment. The current implementation checks for the presence of a CIC
in the ASSIGNMENT REQUEST in order to detect if the communication should
be modeled by AoIP or sccplite. This method is error prone and does not
work very well in situations where only signalling is used, because
there in sccplite and AoIP no CIC or AoIP trasp. identifier is present,
so there is nothing to check on. To resolve this we need an explicit way
to tell the MSC_ConnectionHandler that it has to behave like an AoIP MSC
or like an sccplite MSC.
- Add an aoip flag to TestHdlrParams
- Make sure BSC_Tests.ttcn sets the AoIP depending on mp_bssap_cfg.transport
Change-Id: I800249e783deb018d99e81d814843e0574a5c69b
Related: OS#3639
There are two distinct types defined for a Mobile Identity LV IE.
One type definition lives in GSM_RR_Types and defines the "canonical"
IE form, with a full octet for the length.
Another one lives in RLCMAC_CSN1_Types which defines how a mobile
identity appears in paging requests. In this case, the length field
is only 4 bits in size. Rename this latter type from MobileIdentityLV to
MobileIdentityLV_Paging and add a comment to highlight this distinction.
TS 144 060 Table 11.2.10.2 explicitly states that only the value
part of this IE matches the definition of the canonical IE as
"defined in 3GPP TS 44.018" (actually, TS 44.018 further redirects
the reader to TS 124 008; see section 10.5.1.4 there).
As an aside, a third definition of the MobileIdentityLV type exists
in MobileL3_CommonIE_Types, which matches the "canonical" form.
Change-Id: I990316cd5ef5aaf079b03c344e3185ae6ab8ba6d
Related: OS#2404
We cannot test the control interface with an IPA ping message.
The control interface only supports osmocom-specific extensions,
and the IPA ping message is not part of those extensions.
Other tests will have to be devised for the control interface.
Change-Id: Iae0f16394c78196de621c14b34d1b1905c0cb7c8
Related: OS#2010
This new module allows us to test IPA code in libosmocore
and libosmo-netif. Currently only one test is implemented,
which sends a chopped IPA ping message and expects to receive
an IPA pong.
The system under test is any IPA speaker on any TCP port.
Any test suite may call parametrized functions to create
an IPA testing component and run a particular test.
So far, one such test has been added to the BSC_Tests suite.
Change-Id: I246a405414e36a44dc1e308692faab8bf04da0e6
Related: OS#2010
Since our BSC implementation supports AoIP, the COMPLETE LAYER 3
INFORMATION message must contan a Codec List (BSS Supported) information
element. The test currently just receive a L3 compl Template and then
continue. This is implemented without an altstep. So lets have an altstep
with timeout here and make sure that the codecList information element is
always included.
However, since AoIP was specified after sccp-lite, we need to make sure
that for sccp-lite configurations, the Codec List is not included.
- Check L3 compl message using an alt-step
- Make sure codecList is always included (for AoIP)
Change-Id: Ia16a454e78421430ec32cc37939d429970cb06ec
Related: OS#3548
Max
Daniel Willmann
Philipp Maier
Pau Espin
Harald Welte