vty_install_default() and install_default() will soon be deprecated.
Depends: I5021c64a787b63314e0f2f1cba0b8fc7bff4f09b
Change-Id: I5e8e9d7989ee20fe722bdd3109965d5bce9519fb
Previously clock source configured via vty was ignored. Check that it
wasn't configured and only apply default value in that case.
Change-Id: I8de74a12d2096971a8a83946dc6c7cdb9889acd7
Fixes: OS#1944
Make get_clk_cal() public and use it for ctrl commands handling. This
ensures that the calibration is properly selected depending on the clock
source.
Change-Id: I1c809d48f29ef8e8c50641f882d7a954dfcf88f9
Related: OS#1944
* copy-paste gsm_data_shared.* from OpenBSC master
* remove corresponding configure check and option
* remove .deb dependency
Actual refactoring with removal of unnecessary structures/parts, moving
common OpenBSC/OsmoBSC parts into libraries etc. are left for further
patches.
Current patch will make coexistence with *BSC easier and will simplify
our build infrastructure.
Change-Id: I9f004fb5c4c1db29d4792dfd281d388c7063da13
Related: OS#1923
Previously result of readv() was used unconditionally so when it failed
and returned negative value it was treated like very large positive
which lead to memory corruption. Fix this and add proper error log.
Change-Id: I956c8d551f45c9dd43b5e9de11dfe20dd8783647
Related: SYS#3865
When somebody kills the process, it's best to handle the signal
and to use the opportunity for some cleanup. We always did this
in the BTS on SIGINT, but never on SIGTERM. Let's change it.
Change-Id: I10009c08b7178988f646e2b6035197b9640ac9b5
So far, L1SAP code is hiding RSL_CHAN_OSMO_PDCH from the bts specific
code below L1SAP. This is some kind of a hack/workaround, making code
and debug output / logs more difficult to understand.
So let's teach the lower layer how to treat RSL_CHAN_OSMO_PDCH and
remove the "hiding" code from the common l1sap.c code.
Change-Id: Iaaa833febe45b82166d3901f10cc5466a7591c19
l1_if.c does not generate struct osmo_phsap_prim l1sap properly.
The frame number is not included in this struct. This renders the
logic that processes the reported measurements non functional,
since the logic (see measurement.c) is not able to detect the
end of the measurement period. This commit fixes the problem by
adding the missing frame number to the l1sap structure.
Change-Id: I09241d6cc2ff09e71a3d723d90e4468108a27ae1
We want to always call l1if_tch_rx and l1sap_up in order to avoid losing triggering
events on the upper layer.
With this change, the upper layer will increase correctly seq + ts for
RTP. It will then send an RTP packet with only the header and no payload, which is
not correct but at least we avoid drifting the RTP clock. Upcoming patch
in the series solves this issue.
This patch assumes that we are not lossing data events from the physical
layer and that we receive an event every 20ms, even if the MS is not
transmitting due to DTX.
Depends on libosmocore If4ae20c22b881e94585dad710f17b9e37f77bf82
Change-Id: If5df8940fab833eb4e3ed851880b66987d356031
This way the name of systemd service file will match the name of the
binary similar to lc15. Add aliases so the user can use both old and new
names regardless of which file is installed. Once the corresponding
changes to OE recipes are applied old file can be removed.
Based on work by Pau Espin Pedrol <pespin@sysmocom.de>
Change-Id: I08615eb625d488603aeb5962ad9f30869c0e77c5
Set (possibly incomplete) list of BTS model-specific features and report
them in response to attribute request via OML.
Change-Id: I5f8a6681c3562ec261441e84dde6e085b516d92f
Related: OS#1614
The higher layers relay on a normalized version of frame number
in order to detect the end of a measurement period.
The frame number on which the measurement reports are sent may
depend on the phy, so we need to normalize the frame number
before we report it to the higher layers.
Change-Id: I90bd01479e41f04a6b0aefe3845d071e3148d0c6
This is resubmission of 9eeb0b1a13 with
errorneous use of talloc_asprintf() removed which should fix OS#2316.
Change-Id: I02ae6fffdc808c1ea14185dbb4a720d27a62d4bb
Related: OS#1614
This reverts commit 9eeb0b1a13.
This commit caused osmo-gsm-tester test runs for the sysmoBTS to fail with
SIGABRT consistently. See below redmine issues.
In osmo-bts-sysmo/l1_if.c, it uses talloc_asprintf to write to the char
version[MAX_VERSION_LENGTH]; talloc_asprintf() however is intended to work on
string buffers allocated by talloc, and attempts to reallocate version[].
Furthermore, it is not clear why the patch passes a 'data' arg to
app_info_sys_compl_cb() that is not used.
Hence I will revert this instead of trying to fix. Please resolve issues and
re-submit.
Related: OS#2316 OS#1614
Change-Id: I2c9fd5e6739c1750365c0241476ce4b1aa2df3d4
The so-called "RTP mode" of the DSP contains a bug on all firmware
versions < 5.3.3 which causes the bit-order within each of the
non-aligned codec parameters to be wrong. Introduce a function
originally written by Sylvain Munaut during 32C3 in
http://git.osmocom.org/openbsc/commit/?h=sylvain/32c3_codec&id=5b4a16bbe93a7b1ace65cc23d6cce56ecf4f1275
to bring the bits into [the correct] order.
This has never been seen in a "pure sysmoBTS" setup, as all BTSs would
use the same (wrong) bit-ordering and thus interoperate. This patch now
checks for an affected DSP firmware version and then jumbles (old DSP
firmware version) or does nothing (new DSP firmware version).
Change-Id: Ia0eee4f514fb8fd81c052bb44a4facba898d6afb
Closes: SYS#2452
Make more use of TLVP_PRES_LEN() instead of plain TLVP_PRESENT() and
implicitly assuming a certain length of the information element.
What this obviously doesn't introduce is some kind of error
generation/reporting in case the minimum length is not fulfilled. An IE
that's too small is silently ignored by TLVP_PRES_LEN() and treated as
if the IE wouldn't exist in the first place.
Change-Id: If5c4eee65711c49bc8ba4675221b1d5fd16198e9
l1_if.c does not generate struct osmo_phsap_prim l1sap properly.
The (logical) frame number is not included in this struct. This
renders the logic that processes the reported measurements non
functional, since the logic (see measurement.c) is no longer
able to detect the end of the measurement period.
This commit fixes the problem by adding the missing frame number
to the l1sap structure.
Change-Id: I2bab40c30d727395eb3096026810917407419cd7
* move BTS model name resolution into separate function
* add convenience wrappers for BTS type and number fo TRX and use then
in L1 interface
Change-Id: I4649100df8f1b8e095f210fc294567ba014c0b6a
Related: OS#1614
Make it possible to store:
* Static vs. DHCP mode
* IPv4 address
* Netmask
* GW IPv4 address
* DNS IPv4 address
Add a simple CRC8 and pick 0xFF as initial value so an all
zero EEPROM will not generate a 0 CRC. The code tries to
differentiate exit code between unreadable EEPROM and CRC error.
This is a reference to see if we want to have store it in the
EEPROM or not.
Change-Id: Id8a37fe952ef2a8ef36778729f506f900accf8d1
In case a system has a high-gain external PA (like a 40dB PA) connected
externally, we cannot simply switch the transceiver to 0 dBm in
trx_init() only to then start the ramping at much lower levels once the
PHJ completes in trx_init_compl_cb(). The result would be a short
0 + 40 dBm spike followed by later ramping. We want to avoid that
spike, particularly its associated inrush current, so let's bring up the
board with smething very conservative like -50 dBm, and then ramp from
there.
Change-Id: I0ad91fce64f65e0213c9fcfde3390ace519055db
Fixes: SYS#3259
The simplistic approach of sysmobts_get_nominal_power() is insufficient
to cope for devices that have an internal PA. The Actual transceiver
board is driven to a certain level (0..23 dBm typically), and the
external PA must be handled independent of that. Increasing the return
value of sysmobts_get_nominal_power() would result in the sysmoBTS
mainboard attempting to reach a higher power, which is wrong.
This change affects sysmoBTS 1020 and 1100. It causes power-ramping to
be used by default. For 1002 and 2050, no behavior change is expected.
Change-Id: Ieff75d5becaa80a2097b6e744c75c2d16259c9a4
The vty write for phy/inst is broken, leading to a written-out config being
unparsable; fix all of these:
- in common/vty.c, actually call bts_model_config_write_phy_inst().
- in sysmo and lc15 write the phy instance elements in
bts_model_config_write_phy_inst() and not in bts_model_config_write_phy(),
which lead to writing the members above their parent 'instance'.
- sysmo, lc15 and oct omit the bts_model_config_write_phy_inst()
implementation. This did not cause a compilation problem because it
was in fact never called.
- sysmo writes 'clock-source None' when clk_src is zero, leading to unparsable
config (related: OS#1944). Instead omit the 'clock-source' when zero.
- osmo-bts-trx seems to be the only part that lacks nothing, yet it
also didn't work properly because bts_model_config_write_phy_inst()
was never called.
This problem existed since commit d784e50747
"Introduce new phy_link and phy_instance abstraction"
Change-Id: Icc54fa70045c8fa58e78cf9f788c21a437edfbd4
This reverts commit 1965b0d880, which was
a premature change. The 10dB gain are not the power of the TRX board,
but are due to an internal, factory mounted PA, i.e. the
trx_power_params.pa. This will be introduced after adding a set of
tx_power.c unit tests.
Change-Id: I524b63c51fb0fe1f90ced28486a8e712f2dc50ae
The sysmoBTS 1020 is a 2W BTS, hence its nominal transmit powre is
33dBm. We must handle this correctly based on the model-nr in the
EEPROM. As a result, proper power ramping will be made during start of
the unit by default.
Change-Id: I91a78dadfd7d2e1bc74c8086808c55effdcdd132
Fix SID_FIRST_INH detection during speech and when SID_FIRST is interrupted by FACCH.
Fix SID_UPDATE_INH detection during silence and when SID_UPDATE is interrupted by FACCH.
Add a delay for SID_FIRST to appear at the right time after FACCH.
Fix extra byte sent in downlink for SID_FIRST and SID_UPDATE.
Change-Id: Ia811305e15541f2376005df736bd610e8b0d2f69
Fix error during FACCH interruption of DTX for AMR HR.
Max's note: added fix for sysmobts.
Change-Id: Ib064952331b4f89676ee68a3d8078b1d9debe570
Related: OS#1801
* Unlike in AMR FR, in AMR HR incoming ONSET have to be treated
differently depending on whether we've recently sent SID UPDATE or
EMPTY frame. Split ST_SID_U FSM state into 2 states to accommodate for
that and make sure that additional states specific to AMR HR are not
used for AMR FR.
* Avoid sending E_VOICE and E_SID_U in corresponding states
as those do not initiate FSM state transitions anyway. This decrease
extra load from FSM signalling which otherwise would be triggered on
per-frame basis.
* Introduce separate signal for SID First P1 -> P2 transition to avoid
confusion with E_COMPL and E_SID_U initiated transitions from P1
state.
* Don't init DTX FSM for SDCCH channels.
Change-Id: I229ba39a38a223fada4881fc9aca35d3639371f8
Related: OS#1801
Do not assume that 1 == BS_AG_BLKS_RES but take that information from
SI3. Note: due to current implementation quirks we activate channels
before SI3 obtained, than we deactivate channels upon receiving SI3 and
activate them again. This might not be necessary once we migrate to
proper OML state machines.
This affects lc15 and sysmo hw.
Change-Id: I11377b12680ac3b2f77f80e742b6f0af63fc9c1e
Related: OS#1575
Remove lchan deactivation related code duplication to facilitate future
use for dynamic CCCH re-activation.
Change-Id: Id0d3b19dbfaa16d1734321a07a6eb0355bfd77c9
Add explicit state for recursion (sending the different payload data in
response to the RTS request for same FN) and corresponding
transition. Remove ST_FACCH_V as with new explicit recursion handling it
becomes unreacheable. This makes it easier to maintain
preemption (interruption of current procedure due to FACCH or
Inhibition). This also reduces the number of possible transitions out of
each state thus reducing graph's cyclomatic complexity.
Change-Id: If39b68083d23a4a35f468a5d75f54eb733ebfd14
Introduce dtx_dl_amr_enabled() function which checks that DTX is enabled
and FSM is allocated and use it for all corresponding checks.
Change-Id: Ifa68b641265ed14f242765c85e40da2d1021a541
* re-introduce ST_ONSET_F to guard from repetitive ONSET messages in case
multiple FACCH occur duriing DTX silence period.
* produce ONSET event after both SID FIRST and UPDATE in case of AMR FR.
* always dispatch E_SID_F (SID FIRST) signal if in talkspurt.
* allow E_SID_* right after ONSET (zero-length talkspurt).
* add missing E_ONSET signal description.
* fix FSM transitions for AMR HR *Inhibited and First P*.
* fix incorrect return from l1if_tch_encode() in ONSET FACCH with
incoming SID UPDATE
Change-Id: I0e9033c5f169da46aed9a0d1295faff489778dcf
Related: OS#1801
Handle ONSET cause by Voice and FACCH separately. In case of Voice we
have RTP payload which we have to cache and send later on in next
response to L1 RTS. FACCH have higher priority so it preempts both voice
and silence alike - hence we can send ONSET immediately but still have
to track previous state in order to get back to it gracefully.
This affects lc15 and sysmo hw as there's no FSM-based DTX
implementation for other models yet.
Note: this requires patch for OpenBSC which adds FACCH buffer to tch.dtx
struct.
Change-Id: Idba14dcd0cb12cd7aee86391fcc152c49fcd7052
Related: OS#1802
SDCCH occupy lchan 0..3 in combined configuration so for CCCH we've
always used lchan[4] - replace it with CCCH_LCHAN define and add
comment.
Change-Id: Ic5d742c292d638f119c6b4672120c1950adeb7f0
Use dedicated FSM to handle all DTX DL related events:
- add explicit checks if DTX DL is enabled (fixes regression for non-DTX
setup introduced in 654175f33b)
- fix handling of AMR CMI for SPEECH frames
- add FSM for DTX DL
- sync with corresponding changes in OpenBSC's
- handle FACCH-related DTX ONSET events
This affects both lc15 and sysmobts and requires corresponding change in
OpenBSC (Change-Id: Idac8609faf9b5ced818fde899ccfc6ed0c42e8fd).
Change-Id: I74a0b42cb34d525b8a70d264135e82994ca70d31
When ONSET event happens (either via Marker bit or due to unmarked
talkspurt) we should first send Onset event to L1 and than send voice
data in response to the same PH-RTS.ind.
Change-Id: I2a7b89430ca49eee4a350c5f980bd6bcbc386347
Move code from tch.c (lc15, sysmo) into generic function which:
- check if talkspurt is happening
- cache SID if necessary or invalidate cache
- fill in CMR & CMI prefix
This also fixes the problem when SID FIRST was cached without sending
just like SID UPDATE instead of being sent right away.
Change-Id: I6c7016a54749abadeef4fd4f5b6f750b256fb916
* consolidate AMR CMR and CMI handling in common/amr.c
* use it in save_last_sid()
* remove dead code
* properly compute RTP payload length for AMR
* use save_last_sid() for FR & HR as well
* invalidate cached SID if SPEECH frame is received
Fixes: OS #1800, #1801
Change-Id: I5a1c1ad0b0a295a50e67775a4db85f1d331755ed
Note: this also require changes to properly link against libosmocodec -
see 2bb65be159dfdabf664fec569b343320301701b0 in libosmocore.
Change-Id: I96594cf3aa1013d505bd20069d5bf261d9a2aefb
If Marker bit is set than it's a talkspurt which we have to explicitly
indicate to L1 by first sending ONSET message and than actual voice
data in a separate message.
This change affect sysmobts and LC15 hw.
Change-Id: I88c41568bcb0d82699f617adc4ad192603dd1bb6
Related: OS#1750
Some phone seems to not send SID_FIRST_P2 message which seems like a
different understanding of the DTX spec. L1 accommodates for that by
using P1 to supply data for SID. Hence we should use it to initiate DTX
and ignore P2 message in case of AMR HR.
Change-Id: Iaf993b89caa0ad49b97d1c745dcaf039f867f018
Previously receiving SID via RTP always caused it's transmission to L1
regardless of the time which might have resulted in excess traffic. Fix
this by only saving SID data and transmitting it later on as necessary
according to 3GPP TS 26.093 A.5.1.1.
Change-Id: Ifcdc5c60d0238b704a94f6778d4e00f2b087b090
Fixes: OS#1801
Previously SID retransmission was scheduled incorrectly based on GSM
frames instead of voice frames. Fix this by using GSM Fn only as elapsed
time estimation:
* move saved SID retransmission into generic function from lc15 and sysmo
specific code
* split retransmission time check into separate generic function
* compute estimation for elapsed time since last retransmission using
GSM Fn
Change-Id: Ib054b458a7345d9ba40dba53754ca59ab099c8e8
Fixes: OS#1799
Previously SID was saved explicitly by each BTS model (lc15, sysmo)
instead of relying on generic function. Fix it by using generic function
and propagating necessary parameters for it.
Change-Id: Ie545212cce5ed2b3ea3228597f18a473f5e1deb4
Fixes: OS#1800
The osmo-bts log used to say this a lot:
DL1C <0006> oml.c:758 (bts=0,trx=0,ts=0,ss=1) End of queue encountered. Now empty? 1
- Move this to DEBUG level instead of NOTICE.
- Tweak wording and logic so it says one of:
[...] End of SAPI cmd queue encountered. Queue is now empty.
[...] End of SAPI cmd queue encountered. More pending.
Change-Id: I5a46c90d016cee9b50f32db2af568765d3cb74cc
Based on the indication from L1, number of bits in RACH and burst
type is determined. Appropriate parameters are filled in osmo-bts-sysmo
These parameters are sent to osmo-pcu for processing of the RACH.
Change-Id: I93831ddfb3f31b637e6a576f23a9bb4557758582
Commit "sysmo,lc15: ts_connect_as(): log error also for pchan_as == TCH/F_PDCH"
introduced a check for TCH/F_PDCH intended only for TCH/F_TCH/H_PDCH. It looked
correct, but TCH/F_PDCH startup was designed differently:
For TCH/F_PDCH, the idea was to look it up in pchan_to_logChComb[] and obtain
the TCH/F channel combination, so that TCH/F_PDCH first initialize as TCH/F.
So pchan was in fact intended to be passed as TCH/F_PDCH.
For Osmocom TCH/F_TCH/H_PDCH, we've in the meantime added a ts_opstart()
function that makes this decision explicitly. So, instead of reverting the
erratic commit, add TCH/F_PDCH to ts_opstart(), for both sysmo and lc15.
In ts_opstart(), move to a switch statement to resolve the actual pchan to use
for ts_connect_as().
Drop TCH/F_PDCH and TCH/F_TCH/H_PDCH from pchan_to_logChComb[] and comment.
Change-Id: I376b92a06f4cbe943c1c913dea7487fac53a7d08
Though this patch theoretically makes a lot of sense, it is not entirely clear
why dyn TS are working without it. Committing due to common sense, not to fix
any actual breakage.
Change-Id: I6136cb42a4d627ebefc963eb5321fadfe63cca4b
Interface structure between osmo-bts and osmo-pcu is updated with the
parameters to differentiate the type of RACH and further support 11 bit
RACH. The function prototype and definitions are changed accordingly.
Interface version number is increased.
Change-Id: I4f4c501b2d86d77c78de32a84b1804172ffb6f4d
This reverts commit 53d792c3b0.
See http://osmocom.org/issues/1796
The commit caused this error for PDCH TS with SysmoBTS:
DL1P <0007> l1_if.c:164 Tx L1 prim MPH-ACTIVATE.req
DL1C <0006> oml.c:811 Error activating L1 SAPI PTCCH on TS 7: Invalid parameter
DL1C <0006> oml.c:1089 (bts=0,trx=0,ts=7,ss=0) act failed mark broken due status: -4
Plain PDCH TS show this error but continue to function despite the SAPI
activation failure.
As a side effect, the SAPI activation failure breaks dynamic TS. GPRS
initially works, but the broken status prevents transitions to TCH/* modes
in the BSC:
DRLL <0000> chan_alloc.c:355 Failed to allocate TCH_H channel
DRSL <0004> abis_rsl.c:1656 BTS 0 CHAN RQD: no resources for TCH_H 0x45
Since the commit only enabled PTCCH UL in sysmobts, no other BTS models are
affected. Notice that lc15 still has PTCCH UL disabled all the while, before
and after this commit and its revert.
Also note that PTCCH DL is and has been enabled without problems (see
further above in sapi_dir_pdtch_sapis[]). This is only about PTCCH UL.
Related: OS#1796
Change-Id: Ia59d95c75a8a5d3bd596f55cd1dc0906a6a95f7f
Fill in values for BER, BTO, Link quality in L1SAP and send them to
PCU. Note: this increases the version of BTS <-> PCU protocol. It also
requires corresponding changes in libosmocore.
All BTS models provide measurements data unless direct DSP access for
PCU is enabled. For BTS-specific notes see below.
Octphy: conversion from sSNRDb to Link Quality uses formulae which works
in practice instead of what's documented for sSNRDb value. Subject to
change in future revisions.
TRX: C / I link quality estimator is not computed.
Change-Id: Ic9693a044756fb1c7bd2ff3cfa0db042c3c4e01c
Related: OS#1616
Due to missing runtime cmd patching DSP tracing was not possible at phy
level of config file. No it is possible to specify it as follows:
...
phy 0
instance 0
dsp-trace-flag debug
dsp-trace-flag mph_cnf
...
Change-Id: Ibbbf81d2c4b5d3adbcbc1f08a844d262e603e571
Addition of GSM_PCHAN_TCH_F_TCH_H_PDCH should have added these assertions.
They clarify that only 'real' pchan types are expected in this function.
Change-Id: Ifd4b6820861ef716237e7eeed149b2add9e7d485
For chan_nr_by_sapi(), add GSM_PCHAN_TCH_F_TCH_H_PDCH to pick_pchan().
Add GSM_PCHAN_TCH_F_TCH_H_PDCH to pchan_to_logChComb[] for first pchan
initialization.
In ts_connect_as(), make sure that callers pass proper "real world" pchan
types, i.e. reject the "meta" GSM_PCHAN_TCH_F_TCH_H_PDCH pchan constant.
In ts_opstart(), connect as PCHAN_NONE since we will only know the desired
pchan when the first RSL chan activ is received.
Add GSM_PCHAN_TCH_F_TCH_H_PDCH to lchan_to_GsmL1_SubCh_t(), by using its
current "real" pchan mode.
Call cb_ts_[dis]connected() unconditionally, i.e. not only for TCH_F_PDCH when
a pending flag is set. The cb_ts_[dis]connected() will be a no-op if the pchan
type is not dynamic.
Change-Id: Iaffd2fc0aa9fc6c4a2acbc534ce6384392e0635b
ts_connect() is only called during OPSTART. Name it such because a special
case for OPSTART will be added there for dyn TS.
Change-Id: I09cc81e79661e8084aeebf8132435c207b2cf422
We're about to introduce a new kind of dynamic channel, which will also use
parts of the ip.access mode dyn PDCH code paths. Make sure the general parts
have general names and mark ip.access specific parts as such.
Rename dyn_pdch_ts_[dis]connected() to cb_ts_[dis]connected().
Rename dyn_pdch_complete to ipacc_dyn_pdch_complete().
From cb_ts_[dis]connected(), factor out the current code into static
functions ipacc_dyn_pdch_[dis]connected() -- this will make sense once the
new dynamic kind is added to cb_ts_[dis]connected().
Change-Id: I7da5b7cb7b48572671f50e0dec97d9eec3083df1
Correct the too short padding I introduced in the commit
a55b166c6c. The result needs to
be 121 and not 120. Add static asserts to make sure it does
not happen again.
Change-Id: I3da7f3b8d3c8e12deb8b805cd15ff52a103d4e56
Abstract code for checking/setting lchan's UL SID flag and RTP Marker
into generic function and use it for LC15 and sysmoBTS.
Change-Id: Ica5392e92bab29164711163e7b01adb174272883
Related: OS#1750
Implement bts_model_ts_disconnect() by sending an MphDisconnect message to L1.
Pass a disconnect callback to invoke dyn_pdch_ts_disconnected() in
common/rsl.c.
Implement bts_model_ts_connect() by calling ts_connect_as(). Pass a connect cb
to invoke dyn_pdch_connected() in common/rsl.c.
Change-Id: I61709fdf6b093689a6d3a046f67db6d02f1296ae
For upcoming dyn PDCH switching, I want to be able to set the pchan dynamically
upon ts_connect() and not continue with OPSTART ACK, but with the dyn PDCH.
Thus recoin ts_connect(ts) to ts_connect_as(ts, pchan, cb) and leave
ts_connect() as a thin wrapper to leave init code unchanged.
Change-Id: I09cc794cb424e17411e608c65f2b68e2f2544e07
Enhance bts_model_ API in preparation of dyn PDCH switching. These will be used
to re-connect a TCH/F_PDCH TS in a different mode: either as TCH/F or as PDCH.
All implementations so far return -ENOTSUP, and thus will cause a IPAC PDCH ACT
or DEACT *NACK* to be sent to the BSC as soon as these messages are handled.
Also add stubs in tests.
Change-Id: I21e60c028a1333431c3ed000f788b654d1170b0d
It is cosmetic since the 'default:' case already caught TCH_F_PDCH, but it's
good to mention all expected pchans explicitly.
Change-Id: I5aef84209e46c9288f6adf0730178fe08f26764f
Place a layer 3 handle into GSM L1 messages to better match up confirmations to
respective requests. This handle is a uint32_t transparently returned in the
confirmation messages, so a match-up is easy to add.
So far, a GSM L1 confirmation message received for a preceding L1 Request was
matched only by the prim_id. That meant that only one instance of the same
primitive could be waiting for a confirmation at any given time, or the
responses would get mixed up: the struct wait_l1_conf instances entered into
the fl1h->wlc_list queue would be returned to a possibly mismatching
confirmation handler. (Seen during testing of dyn pdch switching.)
Send the hLayer3 handle out via prim_init(), using new static functions to
produce handles on different scopes:
* l1p_handle_for_trx()
* l1p_handle_for_ts()
* l1p_handle_for_lchan()
(These could possibly move to a more general .h/.c file later.)
Remember the hLayer3 handle in
* struct wait_l1_conf.
Match the incoming confirmations' and stored hLayer3 handles up in, and remove
a now obsolete comment from:
* is_prim_compat()
Since the hLayer3 members are at different byte offsets in
GsmL1_Prim_t.u.*, use large switch statements to set/get the value:
* In prim_init(), extend existing switch statement to set in GsmL1_Prim_t.
* Add l1p_get_hLayer3() to retrieve from GsmL1_Prim_t (could possibly move to a
more general .h/.c file later).
Change-Id: Ie4533c6cbc160318917e7a672ab6f9a848f01d1b
We are using up to 48 (actually only 8) bytes to manage the boot
state of the device. Add it to the eeprom reservation. It turns out
the current padding was too large (37 + 84 don't end at 120).
Change-Id: I4c1de5925577f1d0b7b5cc08529642ffa333d7de
Store last SID received over RTP and repeat is if necessary (no new SID
or SPEECH frames) according to codec-specific scheduling rules.
Change-Id: I4d23846a27d3dbd2a6e75e481c1efcdb2a85f305
Related: OS#1563
Use libosmocodec function to parse RTP with AMR payload in sysmoBTS and
LC15. This replaces "manual" parsing of AMR frame with function covered
by test suite and makes adding DTXd support easier.
Related: OS#1563
Change-Id: I1464f9a12e3f92926d03d5dd5d18e8f0f7206dd9
Reviewed-on: https://gerrit.osmocom.org/204
Tested-by: Jenkins Builder
Reviewed-by: Harald Welte <laforge@gnumonks.org>
* set/clear DTXd activity indicator for measurement reporting
* set DTXd status based on information from RSL
Related: OS#1563
Change-Id: I148a75725c4e5089b6f2da6e9adcbe94170d3257
Depends-On: I4a033b03fcd0deb4db7a38273b5407511dbf1d6c
Reviewed-on: https://gerrit.osmocom.org/220
Tested-by: Jenkins Builder
Reviewed-by: Harald Welte <laforge@gnumonks.org>
Previously frame number was not saved in case of PRIM_TCH rendering many
debug statements with g_time useless.
Change-Id: Ib8d8c919862d0de8e2ebf7753c2592e0d91b09c5
Reviewed-on: https://gerrit.osmocom.org/195
Tested-by: Jenkins Builder
Reviewed-by: Harald Welte <laforge@gnumonks.org>
It remains up to the individual BTS hardware models to decide
whether or not to register those commands (depending on whether they
support the feature) via cfg_bts_auto_band_cmd / cfg_bts_no_auto_band_cmd
At the time the phy link / phy instance level VTY configuration
commands are parsed, we did not yet call l1if_open() and thus
pinst->u.{lc15,sysmobts}.hdl == NULL.
PHY or PHY instance specific configuration must thus be stored inside
the phy_link or phy_instance itself, and not inside the (not yet
existing) handle.
We solve this by moving around some parameters:
* clk_use_eeprom/clk_cal/clk_src/calib_path get replicated in
phy_instance
* min_qual_{rach,norm} are moved into the generic part (which means
that osmo-bts-octphy and osmo-bts-trx should also implement them)
Due to the changes introduced by the phy_link API, it's not easy to set
the default DSP trace flags via a command line argument anymore. We now
rather introduce a persitent VTY configuration command, by which the
default DSP tracing configuration can be set (for each PHY).
The persistent trace flags are stored in the phy_instance, while the
current operational run-time flags are in fl1h->phy_instance.
During the L1SAP related changes, somehow an old version of
check_for_ciph_cmd() was re-introduced, which didn't store the N(s) as
part of the lchan. To make things worse, the old code was still present
in the sysmobts specific part, but never executed.
by using a talloc pool, we avoid having to go back to the libc
malloc pool all the time. The msgb allocations and libortp allocations
happen quite frequently during processing and show up as one of the
high priority items in osmo-bts profiles on sysmoBTS with 14 concurrent
TCH/H calls (highest load scenario).
talloc still consumes significant CPU, this is mostly due to the
zero-initialization of all the associated buffers. Strictly speaking
we shouldn't need this, but any change there would require lots of
testing, as there might be hidden assumptions in the code?
In some percentage of cases, talloc still seems to fall back on malloc
for msgb allocations, which is currently a bit of a mystery. The pools
certainly are large enough, talloc_reprt() rarely reports more than a
few tens of kilobytes used by the msgb pool.
The way we recycle the msgb with a l1sap header when transforming a
PH-DATA.req L1SAP primitive into a PHY/L1 primitive was flawed in
several ways:
1) the way the L1SAP header was stored in the buffer didn't provide
sufficient tailroom for the L1 primitive
2) the alignment of the data in L1SAP is at a 32bit bounadry, but not
in the L1 primitive, causing unaligned accesses.
Instead of limiting the number of TRX at VTY to the actual number of
supported TRX, VTY allows to configure any possible number of TRX. If a
TRX is configured, which is not supported by BTS model, an error message is
returned, which states that the given TRX is not supported.
MS uplink power control is required in pretty much any BTS, and we
cannot assume that they PHY / L1 will always take care of it by
itself. So the correspondign code is moved to common/power_control.c
and called from the generic part of L1SAP.
The corresponding VTY paramter has been moved from the sysmobts-specific
trx VTY node to the common BTS VTY node.
There are three transitions:
1. LCHAN_CIPH_NONE -> LCHAN_CIPH_RX_REQ -> LCHAN_CIPH_RX_CONF
It is used to enable ciphering in RX (uplink) direction only.
2. LCHAN_CIPH_RX_CONF -> LCHAN_CIPH_RX_CONF_TX_REQ -> LCHAN_CIPH_RXTX_CONF
It is used to additionally enable ciphering in TX (downlink) direction.
3. LCHAN_CIPH_NONE -> LCHAN_CIPH_RXTX_REQ -> LCHAN_CIPH_RX_CONF_TX_REQ
-> LCHAN_CIPH_RXTX_CONF
It is used to enable ciphering in both TX and RX directions. This is used
when the channel is activated with encryption already enabled. (assignment
or handover)
In order to follow the order of these transitions, the RX direction must
always be set before the TX direction.
If no cipher key is set (A5/0), ciphering is set to ALG 0, but lchan cipher
state remains at LCHAN_CIPH_NONE.
This part moves control channel message primitives from osmo-bts-sysmo to
common part.
In order to control ciphering fo BTS model, CIPHER (MPH_INFO) messages are
used.
Max
Harald Welte
Pau Espin
Philipp Maier
Neels Hofmeyr