OpenBSC November 2011

openbsc@lists.osmocom.org

21 participants
25 discussions

Jenkins CI with GNU autotests is on
by Holger Hans Peter Freyther 14 Nov '11

14 Nov '11

Hi all, as can be seen here[1] I have merged the GNU autotest setup to libosmocore and the tests are now being executed during the normal build. I encourage everyone to write new tests for bugfixes in libosmocore but in other projects as well. How to write a test: - Create a new test executable or add a new function to one of the existing tests. Make repeatable output go to stdout, errors and things that can change to stderr. - Capture the stdout into the app_name.ok, add this file to the EXTRA_DIST of the Makefile.am - Append your test to the tests/testsuite.at file (copy and pasting one of the previous test cases) - test locally with make check - If you access local files during the test, add them to the EXTRA_DIST as well - Use make distcheck to verify everything is okay. regard holger [1] http://jenkins.osmocom.org/jenkins/job/libosmocore/label=linux_i386_debian_… (no permalink)

1 0

commitlog ML improvements
by Alexander Huemer 14 Nov '11

14 Nov '11

Hi all, I personally would like to keep track of the changes that happen in most of the osmocom git repositories. Generally there are currently two ways to do that. a) The atom feed that cgit offers, e.g. http://cgit.osmocom.org/cgit/osmocom-bb/atom/?h=jolly/testing cgit does not offer this feature in a way firefox would recognize, but it works. Atom feeds are quite convenient, but in the way this works ATM one has to opt-in explicitly to every branch one wants to "follow". Newly created branches are not advertised, ... b) The commitlog ML, which does not have any of the drawbacks of the cgit atom feed, but does not show the actual patch, in case there is one, only the revision hash. There are several thinkable ways to bring together the best of both systems, IMHO the easiest thing to do is modify the way the commitlog ML composes its emails. The example post-receive-email scripts of git (1.7.8rc1) contains the following comment: # hooks.showrev # The shell command used to format each revision in the email, with # "%s" replaced with the commit id. Defaults to "git rev-list -1 # --pretty %s", displaying the commit id, author, date and log # message. To list full patches separated by a blank line, you # could set this to "git show -C %s; echo". # To list a gitweb/cgit URL *and* a full patch for each change set, # use this: # "t=%s; printf 'http://.../?id=%%s' \$t; echo;echo; git show -C # \$t; echo" # Be careful if "..." contains things that will be expanded by shell # "eval" # or printf. The whole file is attached. As stated on IRC, I think pasting the patches in the emails would be most convenient, must takes up quite some space in mailboxes over time. Of course that problem would be addressable with rm and there is gmane/tin, ... In any case, an added cgit link does not harm and also provides an easy way to read the patches. Any comments and ideas how to do a similar RSS/atom feed welcome! Kind regards -Alex

2 3

slightly OT - new list for UMTS discussions
by David Burgess 14 Nov '11

14 Nov '11

All - I realize this is slightly off topic, but I have just set up a mailing list for discussion of the UMTS specifications. This list is not about building a specific implementation of UMTS (although it might lead to that) but it is instead a forum for discussion of the interpretation of the specifications. It is my impression so far that the 3GPP documents are constructed so as to meet some minimum legal requirement for the specification of the technology while still avoiding the conveyance of critical information actually required to implement the system. The likely purpose of this approach is to deter the implementation of 3G technology by anyone outside of the set of companies that staffed the specification committee. The eventual product of this list could be an alternative, wikipedia-style spec that conveys all of the information needed to actually build some minimum-but-fuctional 3G system. Anyone who is interested breaking down the door to this club, please subscribe here: https://lists.sourceforge.net/lists/listinfo/openbts-umts-discuss Thanks. -- David David A. Burgess Range Networks, Inc. 560 Brannan St. San Francisco, CA 94107 USA cell +1 707 208 2622

5 5

GSM network at 28C3
by Harald Welte 14 Nov '11

14 Nov '11

Hi all, I wanted to write this mail earlier, but somehow didn't get around to. MAny people have been assuming that there again will be a GSM network at the 28C3 conference in late december. However, given the large amount of work that was involved in the GSM network at the camp this summer, I have decided that I want to pull out a bit of this network operation. This means that I hope there will be some other project members that will step up to make it happen again. I've already applied for the regulatory license some 4 weeks ago (no response yet, but I doubt there will be any problems). So that should be OK (6 ARFCN, GSM 1800, 200mW each, indoor). The BTSs have always come from a quite large number of individuals and companies, and I'm sure we will be able to pool them together again. Whoever is able to provide a GSM1800 nanoBTS during the event should probably add themselves (and the number of BTS) to the following wiki page: http://openbsc.osmocom.org/trac/wiki/FieldTests/28c3 The BTSs will have to be available from at least 23rd of december to 31st december. Likely they'll be unmounted on 30th night, but nobody can guarantee that. What needs to be done: * BTS physical installation, patching + PoE This is typically done on dec 24 or 25, depends a bit when roh is doing the patching and when the house technicians lift up the big cross beam in the main hall * configuring osmo-nitb + LCR to connect with the POC, do some testing This again involves getting some physical E1 line from/to the PoC patched, and when the POC is available for configuring their side of the link * programming of SIM cards we have some 100 to 150 left-over 16in1 sim cards from the camp, but sysmocom will soon receive 1000 sysmoSIM' cards. Those cards store only one IMSI (not 16), but are otherwise much more flexible in terms of programming. You can create any file (DF/EF) and put any content inside, even files not specified in TS 11.11. The cards will need to be programmed (simple, quick). sysmocom will take care they are of that. * selling of SIM cards In the past, this has been done by the POC. However, there has been some discontempt with the fact that they get all the user questions and cannot do anything regarding GSM, and also they never got any share of the money (not that we made huge profits on it anyway). So I'd suggest to involve the POC more in the network, give them telnet access to the BSC so they can try to anlyze problems themselves, and maybe see if (at least in the beginning) one of the OpenBSC project members could spend some time at the POC desk to help with SIM card sales and user questions * putting together a publik wiki page (mostly copy+paste from last year) * Testing of newer BTSs. Both Fairwaves as well as sysmocom are develoing some new BTSs. We probably don't want them operational during all of the event, the network should run on nanoBTS. But when we feel like doing a test, we will shut down one nanoBTS and start up one of the new BTS models and check for bugs / performance. All in all, I will not be gone completely from the even GSM network but still be around and willing to help if some nasty problem needs to be debugged. And thanks at this point for the kind help of Peter Stuge and Khorben for helping with GSM @ the camp, as well as the POC for the SIM card sales in past years! [please make sure to keep the Cc list of this mail, as I don't think the POC guys are on the openbsc list...] Regards, Harald -- - Harald Welte <laforge(a)gnumonks.org> http://laforge.gnumonks.org/ ============================================================================ "Privacy in residential applications is a desirable marketing option." (ETSI EN 300 175-7 Ch. A6)

7 6

AW: OpenBSC Digest, Vol 35, Issue 3
by Daniel Brune privat 14 Nov '11

14 Nov '11

Hi Harald, I can also provide 3 fully functioning GSM1800 nanoBTS units, I just do not have the extension cables for multiTRX use, but I have adapters for external antennas if required. I would be able to assist for configuring, maintenance and testing the whole setup starting at 25th in Berlin as well. Best, Daniel

2 1

libosmocore and GNU autotest
by Holger Hans Peter Freyther 13 Nov '11

13 Nov '11

Hi Pablo, all I have pushed GNU autotest[1] integration of libosmocore into the zecke/gnu-autotest branch and invoking make check will execute the testsuite. The output looks like this: ## ------------------------------------- ## ## libosmocore 0.4.0.10-c015 test suite. ## ## ------------------------------------- ## Regression tests. 1: bits ok 2: msgfile ok 3: sms ok 4: smscb ok 5: timer FAILED (testsuite.at:38) 6: ussd FAILED (testsuite.at:44) GNU autotest will execute an external application and then can check the exit code, compare the stdout/stderr to a file. In this case the timer test fails as the test itself is randomized and does not always provide the same output. Pablo if your time permits it would be nice if you could: - Provide a cli option to make the test have less iterations (to make make check run faster) - Provide a cli option to produce a repeatable output (e.g. by omitting the expired output). What do you think? In some ways I think that executing the timer test as part of our regression tests makes sense but maybe specially on a loaded machine the test might be flaky... holger [1] http://www.gnu.org/s/hello/manual/autoconf/Using-Autotest.html#Using-Autote…

2 4

[PATCH 2/2] tests: timer: set maximum wait time to obtain test results
by Pablo Neira Ayuso 13 Nov '11

13 Nov '11

If the timer test takes more than 2 * (number of steps + 10), we abort the test. This calculation is based on the maximum timeout randomly set (10 seconds) plus the number of steps (some existing timers may be reset in each step). We double this to have some extra grace time to finish. --- tests/timer/timer_test.c | 19 +++++++++++++++++++ 1 files changed, 19 insertions(+), 0 deletions(-) diff --git a/tests/timer/timer_test.c b/tests/timer/timer_test.c index 72c07a9..3775151 100644 --- a/tests/timer/timer_test.c +++ b/tests/timer/timer_test.c @@ -24,6 +24,7 @@ #include <stdio.h> #include <stdlib.h> +#include <signal.h> #include <getopt.h> #include <osmocom/core/talloc.h> @@ -137,10 +138,22 @@ static void secondary_timer_fired(void *data) } } +static void alarm_handler(int signum) +{ + fprintf(stderr, "ERROR: We took too long to run the timer test, " + "something seems broken, aborting.\n"); + exit(EXIT_FAILURE); +} + int main(int argc, char *argv[]) { int c; + if (signal(SIGALRM, alarm_handler) == SIG_ERR) { + perror("cannot register signal handler"); + exit(EXIT_FAILURE); + } + while ((c = getopt_long(argc, argv, "s:", NULL, NULL)) != -1) { switch(c) { case 's': @@ -162,6 +175,12 @@ int main(int argc, char *argv[]) osmo_timer_schedule(&main_timer, 1, 0); + /* if the test takes too long, we may consider that the timer scheduler + * has hung. We set some maximum wait time which is the double of the + * maximum timeout randomly set (10 seconds, worst case) plus the + * number of steps (since some of them are reset each step). */ + alarm(2 * (10 + timer_nsteps)); + #ifdef HAVE_SYS_SELECT_H while (1) { osmo_select_main(0); -- 1.7.2.5 --fdj2RfSjLxBAspz7--

1 0

[PATCH 1/2] tests: timer: use stderr for non-repeatable output
by Pablo Neira Ayuso 13 Nov '11

13 Nov '11

This makes happy gnu-autotest for the timer test. We may still may fail if we run the test on a very heavy loaded system, but given the amount of timers that we using for the automatic test (only 32), this seems very unlikely to me. --- tests/timer/timer_test.c | 17 +++++++++-------- tests/timer/timer_test.ok | 21 ++------------------- 2 files changed, 11 insertions(+), 27 deletions(-) diff --git a/tests/timer/timer_test.c b/tests/timer/timer_test.c index a01a9e5..72c07a9 100644 --- a/tests/timer/timer_test.c +++ b/tests/timer/timer_test.c @@ -73,8 +73,8 @@ static void main_timer_fired(void *data) int i; if (*step == timer_nsteps) { - printf("Main timer has finished, please, wait a bit for the " - "final report.\n"); + fprintf(stderr, "Main timer has finished, please, " + "wait a bit for the final report.\n"); return; } /* add 2^step pair of timers per step. */ @@ -96,7 +96,7 @@ static void main_timer_fired(void *data) osmo_timer_schedule(&v->timer, seconds, 0); llist_add(&v->head, &timer_test_list); } - printf("added %d timers in step %u (expired=%u)\n", + fprintf(stderr, "added %d timers in step %u (expired=%u)\n", add_in_this_step, *step, expired_timers); total_timers += add_in_this_step; osmo_timer_schedule(&main_timer, TIME_BETWEEN_STEPS, 0); @@ -112,7 +112,8 @@ static void secondary_timer_fired(void *data) timersub(&current, &v->stop, &res); if (timercmp(&res, &precision, >)) { - printf("ERROR: timer %p has expired too late!\n", v->timer); + fprintf(stderr, "ERROR: timer %p has expired too late!\n", + v->timer); too_late++; } @@ -120,7 +121,7 @@ static void secondary_timer_fired(void *data) talloc_free(data); expired_timers++; if (expired_timers == total_timers) { - printf("test over: added=%u expired=%u too_late=%u \n", + fprintf(stdout, "test over: added=%u expired=%u too_late=%u \n", total_timers, expired_timers, too_late); exit(EXIT_SUCCESS); } @@ -155,8 +156,8 @@ int main(int argc, char *argv[]) } } - printf("Running timer test for %u steps, accepting imprecision " - "of %u.%.6u seconds\n", + fprintf(stdout, "Running timer test for %u steps, accepting " + "imprecision of %u.%.6u seconds\n", timer_nsteps, TIMER_PRES_SECS, TIMER_PRES_USECS); osmo_timer_schedule(&main_timer, 1, 0); @@ -166,6 +167,6 @@ int main(int argc, char *argv[]) osmo_select_main(0); } #else - printf("Select not supported on this platform!\n"); + fprintf(stdout, "Select not supported on this platform!\n"); #endif } diff --git a/tests/timer/timer_test.ok b/tests/timer/timer_test.ok index b1792f1..b4e0e11 100644 --- a/tests/timer/timer_test.ok +++ b/tests/timer/timer_test.ok @@ -1,19 +1,2 @@ -Running timer test for 16 steps, accepting imprecision of 0.010000 seconds -added 1 timers in step 0 (expired=0) -added 2 timers in step 1 (expired=0) -added 4 timers in step 2 (expired=0) -added 8 timers in step 3 (expired=0) -added 16 timers in step 4 (expired=1) -added 32 timers in step 5 (expired=4) -added 64 timers in step 6 (expired=33) -added 128 timers in step 7 (expired=87) -added 256 timers in step 8 (expired=183) -added 512 timers in step 9 (expired=466) -added 1024 timers in step 10 (expired=970) -added 2048 timers in step 11 (expired=1968) -added 4096 timers in step 12 (expired=3994) -added 8192 timers in step 13 (expired=8035) -added 16384 timers in step 14 (expired=16324) -added 32768 timers in step 15 (expired=32641) -Main timer has finished, please, wait a bit for the final report. -test over: added=65535 expired=65535 too_late=0 +Running timer test for 5 steps, accepting imprecision of 0.010000 seconds +test over: added=31 expired=31 too_late=0 -- 1.7.2.5 --fdj2RfSjLxBAspz7 Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="0002-tests-timer-set-maximum-wait-time-to-obtain-test-res.patch"

1 0

new timer code: semantics of osmo_timer_schedule for already scheduled timer
by Sylvain Munaut 13 Nov '11

13 Nov '11

Hi, While updating libosmocore in osmocom-bb, a problem arised: 100%CPU usage and hanging. Turns out 'mobile' sometimes call osmo_timer_schedule for an already scheduled timer. Obviously this used to work and now it doesn't. The question is : Was there a defined semantics for this case previously ? Or is it supposed to be unsupported ? What would you expect it to do ? Cheers, Sylvain

3 6

Re: new timer code: semantics of osmo_timer_schedule for already scheduled timer
by Pablo Neira Ayuso 13 Nov '11

13 Nov '11

On Sun, Nov 13, 2011 at 11:45:28AM +0100, Harald Welte wrote: > Hi Pablo and Sylvain, > > On Sun, Nov 13, 2011 at 01:28:30AM +0100, Pablo Neira Ayuso wrote: > > > Hm, I think I found one weird scenario which is not handled fine with > > your patch. [...] > > > > Yes, this scenario is strange, but I think we try to cover all possible > > situations. So I think my patch should be applied instead. > > Will the two of you please figure out what is the best option to > proceed? Revert one of Sylvains patches and apply Pablos patch? I think so. I can make it myself if you want.

1 0

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

OpenBSC November 2011