summaryrefslogtreecommitdiffstats
AgeCommit message (Collapse)AuthorFilesLines
2012-09-06ganc: use vTvLV_GAN TLV encoder/decoderlaforge/ganc-2012Harald Welte1-9/+2
This is required to correctly parse and generate messages with IEs of length > 0x7F, like long SMS messages. A recent copy of libosmocore is required for the core TLV parser support (2fe6847 or later).
2012-09-06ganc: Store more data (GERAN LAC, MAC addr, ESSID, ...) of peerHarald Welte3-5/+119
We store more of the data sent by the MS during REGISTER, and show that data during 'show gan-peer'. Also, there is now a table view (show gan-peer) as well as a detailed view (show gan-peer details)
2012-07-18ganc: TCP typo fixHermann Gausterer1-1/+1
Signed-off-by: Hermann Gausterer <git-openbsc-2012@mrq1.org>
2012-07-13ganc: fix free() order between osmo_conn and gan_peerHarald Welte1-5/+9
2012-07-13ganc: Make sure to close tcp socket when closing ganc / track REG stateHarald Welte2-5/+50
We now track if a peer is REGISTERED or not, and only page him if he is registered. Furthermore, we look for old stale peer records and remove them, if we get a REGISTER on a new TCP connection.
2012-07-13ganc vty: show keep-alive time-out for every peer in vtyHarald Welte1-4/+16
2012-07-13ganc: implement release timer for gan_peerHarald Welte2-8/+51
If we didn't see a GA-RC KEEP ALIVE for more than twice TU3906, then we assume the peer is dead and simply drop the tcp connection.
2012-07-13move BTS-specific timezone override into sub-structureHarald Welte3-16/+20
Group all three structure members involved in bts-specific timezone handling into a sub-structure.
2012-07-13handover: Not finding a HO record on CHAN ACT NACK is not an errorHarald Welte1-1/+1
... it might well a completely non-ho-related establishment
2012-07-13msc: Attempt to release the lchan immediately on IMSI DetachHolger Hans Peter Freyther1-6/+6
The Nokia E71 sends a "IMSI Detach" this msc code does not immediately send the "RR Channel Release", the E71 is impatient and sends a DISC, the "RELEASE INDICATION" is handled by starting the channel release procedure. OpenBSC sends a "RR Channel Release" which will never be answered, during the early release there is no timer and the lchan will be in "RELEASE REQUESTED" forever. This commit removes the anchor operation and checks if the channel can be released immediately. Regarding the channel release handling there is already a branch that needs to be tested.
2012-07-13GPRS LLC: fix segfault when transmitting downlink unit-data without mmctxHarald Welte1-5/+10
2012-07-13BSC: introduce new "sysmobts" BTS modelHarald Welte13-8/+101
so far, osmo-bts/sysmobts used to be entered as "sysmobts" type in the configuration file. However, there are some differences in the protocol/behaviour and we should reflect that by a new BTS plugin (with lots of code reuse from the nanobts driver).
2012-07-13pre-fix measurement reports with subscriber identityHarald Welte1-3/+7
... otherwise it's impossible to distinguish them from each other.
2012-07-13ganc: actually read + use settings from the config fileHarald Welte3-2/+28
2012-07-13ganc: add VTY for 'show gan-peer' and configuring virtual btsHarald Welte3-1/+317
2012-07-13ganc: Add a FIXME about our msgb memory leakHarald Welte1-0/+1
2012-07-13ganc: extend ganc_bts with description and SEGW / GANC IP/portHarald Welte1-0/+7
2012-07-13ganc: destroy peer after read fails or RELEASE has been sentHarald Welte1-0/+14
this is not quite as per spec. We probably need a timer of twice the KEEPALIVE timer of the MS to determine the phone is gone.
2012-07-12ganc: Add peer to global list of peers in networkDieter Spaar1-0/+5
TODO: remove peer from that list once tcp connection closes
2012-07-12ganc: use different host name for default UNCHarald Welte1-1/+1
2012-07-12ganc: implement sending of GA-CSR RELEASEDieter Spaar1-0/+17
2012-07-12ganc: announce < R99 MSC to avoid sequence number in msg_typeHarald Welte1-0/+4
>= R99 MSCs expect sequence numbers in the opper 2 bits of the message type of MS-originated >L3 messages. Our MSC code doesn't really deal with this yet, so don't announce R99 :)
2012-07-12ganc_data: add linked list to connect objectsHarald Welte2-1/+17
the peer needs a reference to a sccp connection, and the network needs to have a list of peers and a list of mscs.
2012-07-12experinmental and incomplete hack trying to build separate MSCHarald Welte8-226/+691
2012-07-12ganc: fix segfault during memset()Harald Welte1-1/+1
2012-07-04BSC VTY: Allow encryption to be set to A5/3Harald Welte1-2/+3
We are currently not checking if the BTS actually suports that cipher, and we particularly don't have any hack for ip.access which apparently seems to re-use the RSL algorithm identifier for A5/2.
2012-06-29ganc: change hard-coded MCC/MNC to 901-70Harald Welte1-2/+2
right now we use 901-70 based IMSIs for testing - but this of course needs to become vty configurable soon.
2012-06-29ganc: add (currently unused) functions for PAGING REQ and DL XFERHarald Welte1-0/+33
2012-06-29ganc: properly zero-initialize the GAN CCH description IEHarald Welte1-0/+3
As Dieter points out, the structure was not properly zero-initialized...
2012-06-28Use uint16_t for MCC and MNCHarald Welte1-2/+2
As Dieter correctly points out, MCC an MNC can very well be lager than 255 ;)
2012-06-28ganc: don't put IP addressess as string into a FQDN attributeHarald Welte1-3/+26
... there is a separate IP address IE that should be used for this. Otherwise, devices may treat 1.2.3.4 as a hostname and make forward dns lookups on it ;)
2012-06-28gsm_data_shared: add structure for decoded parameters of RLC configHarald Welte2-0/+69
this is primarily needed on the BTS side, but we should also use it on the BSC side and make the parameters VTY-configurable.
2012-06-28ipaccess: some more comments on timer valuesHarald Welte1-2/+5
2012-06-24ganc: handle classmark changeHarald Welte2-2/+29
2012-06-24ganc: fix TLV parsing and dump L3 content as hexHarald Welte2-42/+83
2012-06-24ganc: process some more messagesHarald Welte2-15/+82
2012-06-24ganc: update copyright headersHarald Welte3-0/+40
2012-06-24ganc: Correctly return DISCOVERY ACCEPT and REGISTRATION ACCEPTHarald Welte5-16/+197
2012-06-23ganc: fix code to do naything useful at allHarald Welte2-12/+46
2012-06-23osmo-ganc: actually register the accepted fd's with select loopHarald Welte1-0/+3
2012-06-23initial skeleton for accepting UMA/GAN connectionsHarald Welte8-1/+597
2012-06-17use new external libosmogb (part of libosmocore.git)0.12.0Harald Welte23-4453/+8
This removes the libgb (GPRS NS/BSSGP) code from the openbsc.git repository and uses the new version from libosmocore.git instead.
2012-06-17index on master: d24566a libgb: separate header files related to spec and ↵Harald Welte0-0/+0
implementation
2012-06-17libgb: separate header files related to spec and implementationHarald Welte7-219/+239
like in libosmogsm, we separate between header files that are just reflecting information in the respective specs, and header files that related to our specific implementation.
2012-06-17libgb: make sure all BSSGP functions have bssgp_ prefixHarald Welte9-35/+34
We change the minority of functions employing the gprs_bssgp_ prefix to match with the majority without gprs_ in front.
2012-06-17libgb/gprs: don't use log_info from libcommon anymoreHarald Welte4-5/+98
2012-06-17libgb: prefix all NS related functions with gprs_Harald Welte10-58/+95
2012-06-17libgb: don't call directly into GMM / LLC layerHarald Welte5-9/+143
Instead of direct function calls to individual functions, we now generate primitives (osmo_prim) and send them to one application-provided function "bssgp_prim_cb()"
2012-06-17libgb: remove dependency of BSSGP to include sgsn/gmm internal structsHarald Welte3-27/+55
2012-06-17update .gitignoreHarald Welte1-0/+6