This line shows all BTS an their OML states in a single line.
Additionally the uptime or downtime is displayed, if there was a connect
or disconnect of the OML link.
Related: OS#6018
Change-Id: I003fd32e589ddf53b7dd42089f904cfb598e3625
According to Table 10.5.2.3.1 in TS 144.018, radio-link-timeout
values are between 4 to 64 in steps of 4.
Change-Id: I733591d5f72f2e4f822761ca9eda85de7a4c6c81
osmo-bsc would not start with a config written from the vty due
to incorrect identation on the pcu-socket parameters.
Change-Id: I36a66794e654989b4b8bf54bb3727ccbfc2131fa
Only check for intersecting full rate AMR codec, if the BTS has at least
one full rate or dynamic time slot configured.
Only check for intersecting half rate AMR codec, if the BTS has at least
one half rate or dynamic time slot configured.
Related: OS#5926
Change-Id: Ia4a8e7f22dc652655ee7c5458624df8ae136dd95
I believe the actual intention was to reset the 'lchan_nr' variable,
and not the 'chan_nr'. The 'lchan_nr' is used to compose the 'cbits':
cbits = 0x04;
cbits += lchan_nr;
If the value is 4, then the result is:
cbits = 0x04 + 4 = 0x08
which corresponds to SDCCH8 (not SDCCH4), and is clearly wrong.
Change-Id: Ic9c7c2e46e24dab0b721221e9adcbbae2ca56d23
Fixes: ec1b5a0e9 "gsm_ts2chan_nr(): add assertions for lchan_nr"
Fixes: CID#336586
A typical OS imposed limit is 1024 open FD, which is too low when there
are hundreds of BTS.
In systemd service file, set a super high limit of 65536.
In osmo-bsc's user manual, add section 'Configure limits' describing
this in detail.
Related: OS#6256
Change-Id: I26c4058484b11ff1d035a919bf88824c3af14e71
We already recover broken lchans where an ACTIV ACK or REL ACK arrives
late. Now add a recovery path for lchans that are broken because no
ACTIV ACK or REL ACK arrives at all.
Add a timeout of X28 = 30s to the lchan BORKEN state.
On timeout, attempt both a Channel Activation and a Channel Release. If
any of them is ACKed, we have successfully synced BTS and BSC's state.
After successful recovery, place the lchan back in the UNUSED state,
available for servicing subscribers.
If recovery is unsuccessful, just continue to attempt recovery every
further X28 seconds.
Patch-by: osmith, nhofmeyr
Related: osmo-ttcn3-hacks I9b4ddfc4a337808d9d5ec538c25fd390b1b2530f
Related: OS#5106
Related: SYS#6655
Change-Id: Ic4728b3efe843ea63e2a0b54b1ea8a925347484a
The assignment is repeated because the calling subscriber may not
receive the notification on the DCCH, during handover process. After the
assignment is complete, the calling subscriber will receive
notification.
This cannot be done automatically by the BTS, because the BTS has no
relation between the notifications and the channels.
The notification is required, so that the MS knows the channel to listen
to when leaving the uplink the first time. If no notification is
received, the MS will abort the call.
Change-Id: Ife568b8c2756be332c0b8de21111f66f6e537c4d
Title refers to the maximum length of the osmo_wqueue used for
the PCU socket connection.
Related: OS#5774
Change-Id: Ic5f19f4613bccaf582997a4d02b689adee083a0b
The lchan FSM timers were originally implemented to model earlier code
as closely as possible. Now it has come up that T3111 is used in the
wrong place:
3GPP TS 44.018 says:
T3111:
This timer is used to delay the channel deactivation after
disconnection of the main signalling link.
Its purpose is to let some time for possible repetition of the disconnection.
Its value is equal to the value of T3110.
Before this patch, we use it also to time the RF REL ACK message. That
is pretty bad, because T3111 is only 2 seconds by default, making RF
CHAN REL vulnerable for timeout. When a user increased T3111 to
alleviate the problem, the result is that each lchan also delays its
normal channel release procedure by the configured amount of time. Very
inelegant.
Instead, use the X6 timer for REL ACK, because X6 already times the CHAN
ACTIV ACK, which is semantically identical.
Compatibility / user impact: No negative impact expected.
We can assume that every user out there has X6 configured to work for
CHAN ACTIV ACK. From that logic, switching channel release ACK to the
same timer is guaranteed to be what the user intends. We could instruct
users in the release notes that they may now choose T3111 freely (as
short as 2 seconds) without jeopardising channel release anymore.
Related: SYS#6655
Change-Id: Ibd118fa23e5deb4381bc31b11a7b495f57901d6c
At the moment the PCU has no way of knowing with which BTS model it is
used with. However, some BTS models may require slightly different
behaviour by the PCU, depending on which BTS model is used. So, lets add
an additional bts_model field to struct gsm_pcu_if_info_ind in order to
convey the exact BTS model to the PCU.
Related: OS#6191
Depends: osmo-pcu.git I48eb75f65ab54fdec41ef913e24c1f18cd4a4047
Change-Id: I4b58912ad7be3070829614853901aa19108ba2c0
Not only l_bts must be declared outside the for-loop, but also last_i.
This is a fixup of I9dbbd066075f9ccb331616a2b59b46b1b44c8b4c.
Related: CID#330311
Change-Id: Ia10c5e68cb2940d9360d78f606af25bb207ee55f
l_bts must be declared outside the for-loop. If the loop is passed with
n_bts set the first time, l_bts is set. If the loop is passed with
n_bts set next time(s), l_bts is used to encode additional neighbor
cell infos.
Related: CID#330310 and CID#330311
Change-Id: I9dbbd066075f9ccb331616a2b59b46b1b44c8b4c
There are still some remains that are related to the old PCUIF v10
protocol version. Let's clean those up.
Related: OS#5927
Depends: osmo-pcu.git I68a3f59d5c960ae3a4fbd74f9d4a894295cb9ed8
Change-Id: Iebb3a634fee680bdc3636a61f3ccaa1e97e54a64
For each BTS, an SI 10 is generated with the informations about all
neighbor BTS that have the same group/broadcast call.
The SI 10 will only define neighbor cells within the same BSC, because
it does not know about neighbor cells within other BSCs.
When multiple channels are used for a group/broadcast call, the SI 10
is generated after all channels have been activated. Subsequent channel
activations result in an update of SI 10 on all channels.
Change-Id: Icd3101e6dd935a57f003253aaef400c2cf95a0c3
The error logging message within this function is moved to the user
neigh_list_get_arfcn().
In case of an error, which results in measurement report with cell
index that does not exist in the list of neigbor cells, the measurement
report is truncated to 0 neighbor cell measurements.
Change-Id: Ia8a1dca4837536129d17e7784b892bcb75b9ca4b
System Information 2 (bis/ter) uses BA_IND of 0. This refers to
"neigh_list". System information 5 (bis/ter) uses BA_IND of 1. This may
refer to "neigh_list" or optionally "si5_neigh_list", depending on the
VTY settings.
If BA_IND of 1 is received in measurement report and if the optional
"si5_neigh_list" is used, this list is chosen to decode the measurement
report.
Change-Id: Ie9123928fb3ae6f10921ecf01d1b50330661da38
Before this patch, neighbor cells with ARFCN 955 to 974 were ignored in
the GSM 900 band. This resulted an empty 'bit map 0' list in SI2/SI5
messages.
This patch includes R-GSM ARFCN in range 955 to 974. A different encoding
is chosen, if neigboring cells fall within this range.
Change-Id: I40d024290fa4be2ba8d3149ec841b182d0cc8c1f
The PCUIF flag PCU_IF_FLAG_SYSMO was originally used by osmo-bts-sysmo
to signal to the PCU that the direct PHY access for the sysmo-bts DSP
should be enabled. With time, support for other BTS models was added and
the flag became a synonym for "direct PHY access", so it makes sense to
rename it to "PCU_IF_FLAG_DIRECT_PHY"
Related: OS#6191
Depends: osmo-pcu.git I29b7b78a3a91d062b9ea3cd72623d30618cd3f0b
Change-Id: I23df067df99b76048667131905c4448d32d80640
Before this patch, the MGW was selected at startup, and the MGCP data
was always forwarded to that same MGW.
If several MGW were configured in the MGW pool, then osmo-bsc would
select any of those from the pool, and start configured the BTS-side
connection on an endpoint in that MGW. However, when the MSC submitted
the MGCP encapsulated in IPA to the BSC, the BSC would always forward
the MGCP message to that same MGW selected at startup.
As a result, multiple MGWs configured with osmo-bsc using SCCPlite was
broken.
This commit fixes support for multiple MGWs by looking up the already
selected MGW (to setup the BTS-side conn on the endpoint), based on the
CIC (MGCP Endpoint) which was provided by the MSC upon AssignReq.
Related: OS#6189
Depends: libosmocore.git Change-Id Iee361d740845257fa62c9093e30e8079fa933827
Depends: osmo-mgw.git Change-Id I18d7bdf650c0ec87ae16ed4944aed9f495400137
Change-Id: Ia106a21b7692eb5b2ac3b5ac2b358bedbc3b9da6
The first byte is the default version, the other bytes describe the
optional other versions supported by the MO. Print them all.
Change-Id: I01da4883cf59101ddaef575979519ac48fcf54b0
Even though the Abis/OML message flow looks the way it should look
on the wire, it does not actually reflect the sequence/flow of events
and actions in the NM FSMs. For example (extracted from a PCAP):
GPRS Cell(00,00,ff) State Changed Event Report
GPRS Cell(00,00,ff) Software Activate Request
GPRS Cell(00,00,ff) Software Activate Request ACK
GPRS Cell(00,00,ff) Activate Software
GPRS Cell(00,00,ff) Activate Software ACK
[a] GPRS Cell(00,00,ff) State Changed Event Report
[b] GPRS Cell(00,00,ff) Software Activated Report
[c] GPRS Cell(00,00,ff) Get Attributes
GPRS Cell(00,00,ff) Get Attributes Response
[d] GPRS Cell(00,00,ff) IPA Set Attributes
GPRS Cell(00,00,ff) IPA Set Attributes ACK
GPRS Cell(00,00,ff) Change Administrative State
GPRS Cell(00,00,ff) Change Administrative State ACK
GPRS Cell(00,00,ff) State Changed Event Report
GPRS Cell(00,00,ff) Opstart
GPRS Cell(00,00,ff) Opstart ACK
A follow-up patch [1] changes the logic generating message [d],
so that the IPA Object Version of the GPRS Cell MO is taken into
account when adding the attributes.
The problem is that both messages [c] and [d] are generated and
queued for transmission on the receipt of message [a], but *before*
message [b] has been processed. So the IPA Object Version is not
known and assumed to be 0 at that point in time.
This patch delays configure_loop() until message [b] is received.
So far only for nanoBTS and only for those MOs, for which Figure 2
in 3GPP TS 52.021 explicitly mentions that the SW downloading and
activation procedures may be required, plus for the ip.access
specific MOs which all seem to support the SW activation.
osmo-bts does send SW Activated Report only for a subset of MOs,
which does not include Baseband Transceiver, Radio Carrier, and
Radio Channel. 3GPP TS 52.021 is not clear on whether this
message shall be sent by all MOs either, so we consider it
optional and delay configure_loop() only for nanoBTS.
Change-Id: I3953a5e41eb27165f9ff203cac7447ee9d311abf
Related: [1] Ie0fb3eaf76e1f70e5a19bb088e1674b7e553d32a
3GPP TS 52.021 does not strictly mandate that the SW Activated Report
can only be received in state DISABLED/OFFLINE. The only requirement
is that the software load procedure (if needed) and activation is to
be performed in this state.
The successful outcome of software activation procedure is indicated
by the BTS using the above-mentioned SW Activated Report message,
which may be received in ST_OP_DISABLED_{DEPENDENCY,OFFLINE} too.
The MO state changes are triggered by the State Changed Event Report
messages, and happen asynchronously with the software activation.
This patch fixes the following warnings seen with a nanoBTS:
NM_BTS_OP(bts2){DISABLED_OFFLINE}: Event SW_ACT_REP not permitted
NM_GPRS_NSE_OP(nse2){DISABLED_OFFLINE}: Event SW_ACT_REP not permitted
NM_GPRS_CELL_OP(gprs-cell2){DISABLED_OFFLINE}: Event SW_ACT_REP not permitted
NM_GPRS_NSVC_OP(nsvc0){DISABLED_OFFLINE}: Event SW_ACT_REP not permitted
NM_GPRS_NSVC_OP(nsvc1){DISABLED_OFFLINE}: Event SW_ACT_REP not permitted
NM_BB_TRANSC_OP(bts2-trx0){DISABLED_OFFLINE}: Event SW_ACT_REP not permitted
NM_CHAN_OP(bts2-trx0-ts0){DISABLED_OFFLINE}: Event SW_ACT_REP not permitted
NM_CHAN_OP(bts2-trx0-ts1){DISABLED_OFFLINE}: Event SW_ACT_REP not permitted
NM_CHAN_OP(bts2-trx0-ts2){DISABLED_OFFLINE}: Event SW_ACT_REP not permitted
NM_CHAN_OP(bts2-trx0-ts3){DISABLED_OFFLINE}: Event SW_ACT_REP not permitted
NM_CHAN_OP(bts2-trx0-ts4){DISABLED_OFFLINE}: Event SW_ACT_REP not permitted
NM_CHAN_OP(bts2-trx0-ts5){DISABLED_OFFLINE}: Event SW_ACT_REP not permitted
NM_CHAN_OP(bts2-trx0-ts6){DISABLED_OFFLINE}: Event SW_ACT_REP not permitted
NM_CHAN_OP(bts2-trx0-ts7){DISABLED_OFFLINE}: Event SW_ACT_REP not permitted
NM_RCARRIER_OP(bts2-trx0){DISABLED_OFFLINE}: Event SW_ACT_REP not permitted
The following warning is still expected to show up though:
NM_BTS_SM_OP(bts_sm){ENABLED}: Event SW_ACT_REP not permitted
but is caused by a different problem, which is to be fixed later.
Change-Id: I00a423adcde5c34977f4c4dad920874687fa493c
These functions are called from a signal handler (SS_NM), and the
signal itself is sent from the generic OML logic whenever the
Software Activated Report is received from some BTS, which is not
necessarily a nanoBTS or osmo-bts.
It would be nice if we could check the BTS type once in the signal
handler, but the signal data is not always the same and depends on
the signal type, so unfortunately it's not possible.
Change-Id: I088ff75f2048e54e4bfd926a79c1dcf27b4fb3a4
Using bts->nr on the wire is wrong because:
* bts->nr is a BTS number in the BSC's config file,
* bts->bts_nr is a BTS number within the SITE-MANAGER MO.
The problem does not show up if there exists only one BTS node
in osmo-bsc.cfg. Otherwise, the Software Load and BTS Restart
procedures are broken for nanoBTS.
Change-Id: I99d9c72752e55c4553e2e9c60df5caa8343b7be0
Andreas Eversberg
Neels Hofmeyr
Harald Welte
Keith Whyte
Vadim Yanitskiy
Oliver Smith
Philipp Maier
Pau Espin