OpenBSC October 2011

openbsc@lists.osmocom.org

23 participants
37 discussions

[PATCH] gsm: Verify the MNCC_VERSION of the BSC/MS and close the socket on mismatch
by Holger Hans Peter Freyther 21 Oct '11

21 Oct '11

The BSC/MS will send a Hello packet that includes the version number, make LCR verify this version number and close the socket in case it does not match a supported version. --- gsm.cpp | 12 ++++++++++++ mncc.h | 7 +++++++ 2 files changed, 19 insertions(+), 0 deletions(-) diff --git a/gsm.cpp b/gsm.cpp index e6806c3..227d618 100644 --- a/gsm.cpp +++ b/gsm.cpp @@ -1042,6 +1042,7 @@ static int mncc_fd_read(struct lcr_fd *lfd, void *inst, int idx) int rc; static char buf[sizeof(struct gsm_mncc)+1024]; struct gsm_mncc *mncc_prim = (struct gsm_mncc *) buf; + struct gsm_mncc_hello *hello = (struct gsm_mncc_hello *) buf; memset(buf, 0, sizeof(buf)); rc = recv(lfd->fd, buf, sizeof(buf), 0); @@ -1050,6 +1051,17 @@ static int mncc_fd_read(struct lcr_fd *lfd, void *inst, int idx) if (rc < 0) return rc; + /* TODO: size check? */ + switch (mncc_prim->msg_type) { + case MNCC_SOCKET_HELLO: + if (hello->version != MNCC_SOCK_VERSION) { + PERROR("MNCC version different. BSC version is %u\n", hello->version); + mncc_fd_close(lcr_gsm, lfd); + return 0; + } + break; + } + /* Hand the MNCC message into LCR */ switch (lcr_gsm->type) { #ifdef WITH_GSM_BS diff --git a/mncc.h b/mncc.h index e5f8216..fac7145 100644 --- a/mncc.h +++ b/mncc.h @@ -56,6 +56,8 @@ #define GSM_TCHF_FRAME 0x0300 #define GSM_TCHF_FRAME_EFR 0x0301 +#define MNCC_SOCKET_HELLO 0x0400 + #define GSM_MAX_FACILITY 128 #define GSM_MAX_SSVERSION 128 #define GSM_MAX_USERUSER 128 @@ -185,3 +187,8 @@ struct gsm_data_frame { unsigned char data[0]; }; +#define MNCC_SOCK_VERSION 1 +struct gsm_mncc_hello { + u_int32_t msg_type; + u_int32_t version; +}; -- 1.7.5.4 --------------020402070302030203020301--

1 0

R: Re: LCR + Asterisk + NanoBTS: Strange Voice Behaviour "Dropping incompatible voice frame on lcr/1"
by Luca Bongiorni 21 Oct '11

21 Oct '11

Hi Gregory, here the log from "asterisk -vvvvv": http://pastebin.com/3brJ9NxC > please send me a complete call trace. [verbose 3] Cheers, Luca

1 0

[RFT] Couple of changes to nanoBTS and nokia insites bootup sequences to be tested
by Sylvain Munaut 20 Oct '11

20 Oct '11

Hi, I've just pushed a couples of changes to a branch sylvain/pending that need to be tested. They alter the startup process for the nanoBTS (fixing the issues some people had with starting GPRS support, when the bts was not connecting to the SGSN). And also a fix for the Nokia InSite startup sequence where the reset command was not sent due to missing initialization of .do_reset. Since I know initialization sequence can be a rather fragile process, if people could test this didn't break anything for them, it'd be nice. Cheers, Sylvain

3 3

Re: [RFT] Couple of changes to nanoBTS and nokia bootup sequences to be tested
by Sylvain Munaut 20 Oct '11

20 Oct '11

Hi, > That’s all I get, and it never bootstraps. In libosmo-abis, in src/input/lapd.c line 90, try replacing 0,240000, by 0,480000, Also, make sure you plugged the power like 10 min before trying. Cheers, Sylvain

2 1

authentication and encryption???
by GSM Driller 20 Oct '11

20 Oct '11

hello list, I found some some where in openbts archives that authentication and encryption has been implemented in openbsc, kindly advise how can i add these features in openbts???, i studied libosmocore libraby, but unable to figure out, how it can be done, please advise!!!

2 2

Re: [RFT] Couple of changes to nanoBTS and nokia bootup sequences to be tested
by Dieter Spaar 19 Oct '11

19 Oct '11

Hello Sylvain, On Mon, 17 Oct 2011 16:11:25 +0200, "Sylvain Munaut" <246tnt(a)gmail.com> wrote: > > Actually to get it to start I have to: > - After powering it, wait for like 10 min. (If I try to start openbsc > any sooner, it will never bootstrap) If I remember, I never had to wait longer than half a minute after power on. But this was with the version from August, so some changes in the code might cause this behaviour. > But I could add this 'reset' process as an nokia specific option to > the vty (i.e. reset_on_start or something ). A configuration option is fine, I would set the default to "always reset" and only if someone know what he does, he can turn the reset off (I think I had strange effects while changing the configuration and not doing a reset, so this situation should be avoided). Best regards, Dieter -- Dieter Spaar, Germany spaar(a)mirider.augusta.de

2 1

Legal situation when using auth policy token
by Lennart Müller 19 Oct '11

19 Oct '11

Hi, we own a frequency usage licence from the Bundesnetzagentur and ran our network with auth policy token. So every new phone trying to attach to our network will receive a SMS with information how to register and a token. Thereafter, the phone is kicked out and will no longer be able to register again. Now there was a problem with some phones from Cupertino which, as I heard, registered to our network, received the SMS, were kicked out again but did not try to re-register with their home network. So some "nice" guys threaten to call the Bundesnetzagentur if we will not shut down the network immediately. As a quick solution, we changed the registration procedure from Token-Input to IMEI-Input. My question is: Are there really legal problems when using the "Auth Token" policy? Many Thanks, Lennart

4 6

[RFC] libosmocore: New libosmosdr sub library
by Sylvain Munaut 19 Oct '11

19 Oct '11

Hi, I've just pushed a libosmosdr sublibrary in a sylvain/sdr branch in the libosmocore git. Currently it contains mostly a complex vector 'class' along with various math functions used on them (convolution / correlation / ...). Some aspects the the API along with some of the underlying math are inspired by the code from openbts, but it is a complete rewrite in pure C with some aspects simplified, and some others extended. If anybody has comments before I merge it in master, they're welcome. The doxygen doc should be clear enough (and even includes fancy latex equation :P) Cheers, Sylvain

3 5

[PATCH 0/3] rb-tree based timer infrastructure for libosmocore
by pablo＠gnumonks.org 18 Oct '11

18 Oct '11

From: Pablo Neira Ayuso <pablo(a)gnumonks.org> Hi! This patch reworks the timer insfrastructure of libosmocore to make it more scalable. I remember some discussion with Harald while I was in Berlin that this can be of interest. It also reworks the timer torture test, the current list-based timer seems to have problem with it. You can find this changes in the pablo/timer branch. Pablo Neira Ayuso (3): add rb-tree implementation to libosmocore timer: add scalable RB-tree based timer infrastructure tests: add new torture test for timer infrastructure include/osmocom/core/Makefile.am | 2 +- include/osmocom/core/linuxrbtree.h | 160 +++++++++++++++ include/osmocom/core/timer.h | 6 +- src/Makefile.am | 2 +- src/rbtree.c | 389 ++++++++++++++++++++++++++++++++++++ src/timer.c | 176 +++++++++-------- tests/timer/timer_test.c | 141 ++++++++++--- 7 files changed, 757 insertions(+), 119 deletions(-) create mode 100644 include/osmocom/core/linuxrbtree.h create mode 100644 src/rbtree.c -- 1.7.2.5

2 5

[PATCH 1/1] openbsc: Fix Spelling error
by Caleb Pal 18 Oct '11

18 Oct '11

Fix to small spelling error. diff --git a/openbsc/src/gprs/gprs_gmm.c b/openbsc/src/gprs/gprs_gmm.c index cc9d297..ff2069a 100644 --- a/openbsc/src/gprs/gprs_gmm.c +++ b/openbsc/src/gprs/gprs_gmm.c @@ -1055,7 +1055,7 @@ static int gsm0408_rcv_gmm(struct sgsn_mm_ctx *mmctx, struct msgb *msg, break; case GSM48_MT_GMM_RA_UPD_COMPL: /* only in case SGSN offered new P-TMSI */ - DEBUGP(DMM, "-> ROUTEING AREA UPDATE COMPLETE\n"); + DEBUGP(DMM, "-> ROUTING AREA UPDATE COMPLETE\n"); mmctx_timer_stop(mmctx, 3350); mmctx->p_tmsi_old = 0; /* Unassign the old TLLI */

2 1

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

OpenBSC October 2011