OpenBSC December 2018

openbsc@lists.osmocom.org

21 participants
33 discussions

by Gullik Webjorn

Hello, I have recently started to deploy OpenBSC . Initially I hope to contribute to manuals and howtos and with bug reports and testing. I have followed the suggested line, and installed everything from binarys, which puts for instance osmo-msc at 1.2.0. The platform is Orange Pi Zero, and UHD / B100, but will evolve to OPI Z + limesdr-mini. I have encountered the "swapped address" bug reported some 3-4 months ago, but is there in osmo-msc 1.2.0, i.e. resulting in one way audio, since the sdp is addressed to a byte reversed address. Apart from that problem I have some minor stability issues (more later ). so, my entry #1 Latest .deb packages have an issue with requesting the RTP stream for audio TO Mobile with a bogus IP address: 16:20:17.739790 IP (tos 0x0, ttl 64, id 50362, offset 0, flags [none], proto UDP (17), length 698) orangepizero.lan.5065 > slottet.lan.sip: [udp sum ok] SIP, length: 670 INVITE sip:1000@192.168.1.80:5060 SIP/2.0 Via: SIP/2.0/UDP 192.168.1.170:5065;rport;branch=z9hG4bKcgB6KF1r9Z8Zj Max-Forwards: 70 From: <sip:0757576000@192.168.1.170:5065>;tag=9U40t8B3a8p5r To: <sip:1000@192.168.1.80:5060> Call-ID: 590931be-7be8-1237-51bb-0242ed556d0c CSeq: 938485738 INVITE Contact: <sip:192.168.1.170:5065> User-Agent: sofia-sip/1.12.11devel Allow: INVITE, ACK, BYE, CANCEL, OPTIONS, PRACK, MESSAGE, SUBSCRIBE, NOTIFY, REFER, UPDATE Supported: timer, 100rel Content-Type: application/sdp Content-Length: 129 v=0 o=Osmocom 0 0 IN IP4 170.1.168.192 s=GSM Call c=IN IP4 170.1.168.192 t=0 0 m=audio 4036 RTP/AVP 0 a=rtpmap:0 GSM/8000 Asterisk on 192.168.1.80, osmo stack on 192.168.1.170, Best Regards, Gullik

6 years, 6 months

nano3g ip.access provisioning/configuration

by Mykola Shchetinin

Dear Osmocom community, Who deals with configuration of nano3g ip.access here? How do you configure it? We use the latest ip.access firmware that has a bunch of bugfixes and it doesn't support "older" telnet configuration methodology: https://osmocom.org/projects/cellular-infrastructure/wiki/Configuring_the_i… So, do you just use other/older firmware or do you have TR069 ACS set up? Kind regards, Mykola

6 years, 6 months

LAI_AND_LAC, curious

by Neels Hofmeyr

Re commit 6cb833608fa39943c1ce9fe046992922e09f4266 rename CELL_IDENT_LAI_AND_LAC to CELL_IDENT_LAI The name "LAI AND LAC" makes no sense because a LAC is part of a LAI. Keep the old name available for API backwards compatibility. Change-Id: I2749cf75b7b45de0cd43cf4c696a6b6984f5a065 Related: OS#3124 I remember giving similar code review during the gsm0808_cell_ident patches, and I think the conclusion was that even the specs speak of LAI_AND_LAC. Well, anyway, I'm glad that sanity made it to the CELL_IDENT API finally :) ~N

6 years, 6 months

Note about IuUP

by Mykola Shchetinin

Hello, It is about how I got a voice working between UE connected to Osmocom 3G (external call handling - Kamailio PBX) and VoIP softphone connected to the PBX. The part about is IuUP only. (Transcoding comes later) I have a traces of a call where a commercial NITB was used and I compared them to traces which I got from a call where Osmocom system was used. (Osmocom Core consisted of libosmocore of branch laforge/iu_up, osmo-mgw of branch neels/iuup and other stuff from Nightly Builds) My femtocell (nano3g ip.access) uses AMR 12.2k as a codec. The difference in IuUP <-> AMR conversion between two. When converting from IuUP to AMR: - Osmo-mgw only strips IuUP header from RTP payload; - Commercial NITB strips IuUP header and prepends AMR header to RTP payload; When converting from AMR to IuUP: - Osmo-mgw only prepends IuUP header to RTP payload; - Commercial NITB strips AMR header and prepends IuUP header to RTP payload; So that means basically that my femtocell sends AMR payload as IuUP payload without AMR header. And expects AMR payload inside IuUP message without AMR header. IuUP header is 4 bytes. AMR header for octet-aligned mode is 2 bytes. To test this I've modified the code of iuup_cn_node.c so that it does: - when converting from IuUP data (rx_data): after stripping IuUP header it prepends 0xf03c as a header (it is for AMR 12.2k). - when converting to IuUP data (osmo_iuup_cn_tx_payload): before prepending IuUP header it strips AMR header. The call performed had voice on both sides. It was performed between a regular phone connected to a femtocell and a VoIP softphone connected to Kamailio PBX. Kind regards, Mykola

6 years, 6 months

3G Working Setup

by Ron

Hi Community, Does anyone successfully run a 3G network using OSMOCOM elements? Currently we are trying to run the following OSMO elements using this URL "https://osmocom.org/projects/cellular-infrastructure/wiki/Osmocom_Network_I…. Instead of using a 3G femto cell as HnodeB, does anyone tried to use UMTRX, B210, and LimeSDR? Is there other software that will act as HnodeB between the SDR and OSMOHNBGW? TIA. Best Regard, Ron Menez ron.menez(a)entropysolution.com<mailto:ron.menez@entropysolution.com>

6 years, 6 months

nano3g ip.access femtocells and transcoding

by Mykola Shchetinin

Dear Osmocom community, Did somebody try to use nano3g ip.access femtocells and Osmocom system with external call handling? What I am trying to do is to make calls work between VoIP phones connected to PBX and UEs connected to Osmocom system. I've tried two PBXs: Kamailio with rtpengine and Asterisk. Signalling worked good between VoIP phone and UE with both PBXs; calls between two UEs registered on Osmocom system worked good also... But I fail to do transcoding (calls between VoIP phone and UE). The problem is that it is unknown which codec is used by nano3g ip.access femtocell. So, the question is: Do you know what codecs were used for speech in your setup with nano3g ip.access femtocells? Or do you know how to find out that information? (It was stated by ip.access that they use AMR but I failed to match AMR specs with RTP payload that was sent by femtocell...). I would be very grateful for any ideas. Kind regards, Mykola

6 years, 6 months

switched to gerrit +3, some ideas

by Neels Hofmeyr

There are obvious problems with the current +3 enforcing scheme on gerrit. a) if a patch has +2, the summary shows a tick mark, you don't see that another +1 is needed. b) if a patch has three +1s, the summary shows +1, you don't see that it can be merged. c) voting +1 for an own patch has an effect on the mergability. Ideas to improve: a) +2 hides missing +1 On IRC, we agreed that it is a good idea to require two reviews (by two people) -- three is too much. Currently that would be one +2 vote plus one +1 vote. (alternatively 3x +1, too). Idea: if we remove the "+2" ability, we avoid the problem that a +2 hides the missing +1. Everyone has only +1 votes, and we add up using that plugin. To require only two reviews, we merge once a sum of +2 is reached. b) 3x +1 == "+1" There's no solution short of fixing what gerrit shows in the summary. Everyone will still click on the patch and then see that it already has three votes. c) +1 to self The solution is that you don't vote for your own patches, unless it is really super trivial and not worth wasting others' review time on. Let's see for a few days how we fare with these quirks. ~N

6 years, 6 months

Re: Build failed in Jenkins: master-osmo-sgsn » --disable-iu,1,a3=default,a4=default,osmocom-master-debian9 #7511

by Neels Hofmeyr

Created issue OS#3736 for this. Let's collect failing build details there. ~N

6 years, 6 months

OT: owning a sysmoBTS we could use for 35c3?

by Neels Hofmeyr

Hi, if anyone owns a sysmoBTS and either comes to 35c3 or can easily pass it on: can we use your sysmoBTS for the GSM installation, please? They will get insured by the event, we just need the serial number(s). I might also ask you personally later this week / at sysmocom office. Thanks! ~N

6 years, 6 months

build slave segfaults

by Neels Hofmeyr

I see sporadic segfaults during jenkins builds every now and then... Let's keep an eye on it. This one built on build2-deb9build-ansible ~N On Wed, Dec 12, 2018 at 12:52:59AM +0000, jenkins(a)lists.osmocom.org wrote: > See <https://jenkins.osmocom.org/jenkins/job/master-osmo-msc/IU=--enable-iu,WITH…> > ... > CC auth_milenage.lo > /bin/bash: line 2: 6091 Segmentation fault (core dumped) /bin/bash ../../libtool --silent --tag=CC --mode=compile gcc -DHAVE_CONFIG_H -I. -I../.. -I../../include -I../../include -I/usr/include/p11-kit-1 -DBUILDING_LIBOSMOCORE -Wall -Wall -g -O2 -DBUILDING_LIBOSMOCORE -Wall -MT auth_milenage.lo -MD -MP -MF $depbase.Tpo -c -o auth_milenage.lo auth_milenage.c > make[3]: *** [auth_milenage.lo] Error 139 > Makefile:582: recipe for target 'auth_milenage.lo' failed > make[3]: *** Waiting for unfinished jobs.... > make[3]: Leaving directory '/build/deps/libosmocore/src/gsm'

6 years, 6 months

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

OpenBSC December 2018