Age | Commit message (Collapse) | Author | Files | Lines |
|
Whenever we have FACCH stealing happening, the RTP clock didn't continue
to tick as we didn't report anything to RTP in those situations.
Change-Id: Iae832137b8ef8dc7c1e6d0746df69efd41aafcf0
Closes: OS#5974
|
|
ph_tch_req() is a recursive function and conditionally calls itself at
the very bottom. The recursive call happens iff all of the following
conditions are met:
* DTXd is enabled,
* AMR codec is in use,
* DTX DL AMR FSM state is recursive.
The problem is that ph_tch_req() may pull sizeof(*lsap) from the given
msgb twice: during the initial and the recursive calls. The second
attempt to pull sizeof(*lsap) causes the process to abort, because
the remaining room is less than it's attempting to pull.
AFAICT, doing msgb_pull() is not really necessary, given that
l1sap_tch_rts_ind() thankfully does set msg->l2h before pushing
the lsap header in front of the actual frame.
Update osmo-bts-sysmo and its copy-pasted siblings, which are likely
affected too, except osmo-bts-octphy which does not do the recursion.
Change-Id: Ib349b74a9e4bd48c902286f872d3b0e9a068256c
Related: OS#5925
|
|
Change-Id: Iec397f1ef4cb5ac1c48abf869f54e272ed0fbb55
Depends: libosmocore.git I6adda28698c0e479ef20f5d090c1f7f76a2ec97e
Depends: libosmocore.git Ia6b428e5b6aaecf151cbfa980b89eff6d0fe6006
Related: OS#1572
|
|
It makes no sense to push interference meas results for those TRX since
they are disabled, unused and hence won't be reported in RSL RF Res Ind.
Related: SYS#6370
Change-Id: Ie3fd80970585cb30808b0644568dbc8936a57721
|
|
The timeslot carrying BCCH/CCCH on C0 is a bit special in a way that
it's being activated implicitly - there is no explicit RSL CHANnel
ACTIVation for that. This is why we have TRX_CHAN_FLAG_AUTO_ACTIVE,
which marks sub-channels of BCCH/CCCH in the trx_chan_desc[].
The upcoming patch changes the burst buffer allocation strategy, so
that the memory is allocated on channel activation and then free()d
on channel release. This patch facilitates that by making the
activation/deactivation logic consistent for all sub-channels.
Change-Id: Id0c70d6a2b777a5117c85149bfa1b3a997299d1d
Related: OS#1572
|
|
The header file <osmocom/gprs/protocol/gsm_04_60.h> was recently
deprecated and moved to <osmocom/gsm/protocol/gsm_44_060.h>, so
it's now part of libosmogsn and depending libosmogb is not needed.
Change-Id: I823de7ebac2fe5dfddb88d533f1a9419f4a605db
Related: libosmocore.git I70cc21bf25a7081070738abacb409ed19094c3b2
Related: OS#5312
|
|
This change fixes a problem that prevents osmo-bts-virtual from
starting when dynamic (ipa/osmo) timeslots are in use.
Change-Id: I5db5b7dd6a8e84cf9a0d84f04a650c2ed8a4e368
|
|
Change-Id: Ia31e53552208a9d57e2fc5473440840d38945d00
|
|
Related: SYS#6370
Change-Id: I887e0cb03b2a5654accccf7a55fac51319981bfb
|
|
Change-Id: Ibcc3578cce23767445e53149796d0bd79f81ead8
|
|
Change-Id: If1de230f022c69f3dfce8a7c66d494a984e7f3f9
|
|
Change-Id: I8d9da8e8433feb1f022dc2f5199f9c15c01e9312
|
|
The code in that function is pretty rotten:
* osmo_fd_write_disable() is called for each message in the queue,
there's no need for that. Let's simply call it at the end if the queue
is empty.
* Asserting for obvious stuff like dequeue returning the first entry in
the list.
* Having error code path for empty message: That shouldn't happen, abort
immediately.
With all thse changes, the function is way simpler, easy to read and
more efficient.
Change-Id: I7ffff98cd8cdf2041bff486c3fde6f16365028d5
|
|
This option should be used for any executables which are used only
for testing, or for generating other files and are consequently never
installed. By specifying this option, we are telling Libtool that
the executable it links will only ever be executed from where it is
built in the build tree. Libtool is usually able to considerably
speed up the link process for such executables.
Change-Id: I06d3d5ab1dd21400a72f76eecc508886617ef4c6
|
|
These two messages indicate that no ACK/NACK message is going to be
sent to the BSC because activation/deactivation was requested by the
PCU. This is absolutely normal and requires no attention from the
user/operator, so better use LOGL_INFO.
Change-Id: I6eaf3a6c07fb30b31c045729c935c8ad6735e5c8
|
|
Change-Id: I69dc11c66a774ddb7364d6ecd9bdb2e94e6cad66
|
|
Change-Id: I692bbb5b7c06b5528a36c09a99a7580cbafa3cc0
|
|
Parse the RTP CSD Format and reply with NACK if the mode is not
RSL_IPAC_RTP_CSD_TRAU_BTS, which is the only one we plan to implement
for now.
Related: OS#4393
Change-Id: Ibfc7811387df5224682d7e6a313d38648d3d8c48
|
|
Add documentation for rsl_ipac_rtp_csd_format_d/_ir from
libosmocore.git, include/osmocom/gsm/protocol/gsm_08_58.h and
wireshark.git, epan/dissectors/packet-rsl.c.
Related: OS#4393
Change-Id: Ic8f40076698f2b341cc0096fd96ba2051a41f077
|
|
Forwarding of PDCH related data over multicast works for me.
Without these features osmo-bsc would reject the OML connection
if it's configured with [E]GPRS enabled.
Change-Id: I5e13c153805c56904e51d222007228e1c2872c88
Related: OS#5500
|
|
Change-Id: I8653d5521680bb6ac2b32371b5428263082bc7f3
|
|
osmo-pcu requires to get DATA.ind for all FN/TS it manages in order to
tick its internal FN clock and trigger timeouts. Without this, some
events are ticked in a delay fashion when osmo-pcu detects FN jumps.
Change-Id: I8f1856dd9061c1bfca8b15be30df7a51760231b0
|
|
rf_locked
Differentiate in each TRX between being provisioned (configuration available) and being provisioned *plus enabled*.
TRX0 waits for all other TRX to be ready before sending POWERON, since
all TRX need to have been minimally configured over TRXC before POWERON
is called. This "ready" state though, doesn't necessarily mean the
TRX!=0 are enabled (aka NM Enabled and rf_locked=0). With them being
configured it es enough to start the whole PHY.
With the old logic, given any TRX was rf_locked=1 at startup, the PHY
would not become UP because it the TRX_PROV FSM was waiting for OPSTART
to arrive on all TRX, which wouldn't happen on TRX that had rf_locked=1.
So in summary, the desired requirements to start the PHY are (in any
order):
1- Wait all TRX are configured
2- Wait for TRX0 OPSTART
Related: SYS#6370
Change-Id: Ie4836f5721ce8227a63c267730aeb17228994214
|
|
When we disable the old socket connection, let's use
osmo_fd_read_disable().
Change-Id: I6b6854e9881c79b5c4794bde4ba4f6841dd06386
|
|
Change-Id: If4f5a419b5af3f185219a879dcb2abb4eea45f1c
Fixes: f19f5331 "GSMTAP: allow configuring local address"
|
|
Change-Id: If3016bae9d03a9972c04b748fc9efab56a412e0e
Related: OS#1572
|
|
Even though it's unlikely, get_lchan_by_chan_nr() *may* return NULL.
Change-Id: I1a815c9675eebc16640b62308499dd784fc206bd
Related: OS#1572
|
|
Having a common body for PRIM_INFO_{ACTIVATE,DEACTIVATE,MODIFY},
but then branching using if-statements is a bit confusing.
Change-Id: I915c8a541249249e3c0b1f2eda4535e7c52db79f
Related: OS#1572
|
|
Change-Id: I8cb2979c62ebb05d06af49b40b145b7dee826cb1
|
|
Change-Id: I373b510efab4874f59f75385bd0e1229692299ec
|
|
The variable rc holds the return code of accept(), which returns a file
descriptor on success. So lets call the variable "fd" to make this
clear.
Change-Id: Ibc359d941786b1d1d52b356e239a76a090b52c1f
|
|
Change-Id: If047cbaf95b343ee115690bf7a724a8edc5df735
|
|
Change-Id: Iaa97dc36e92797db5f348e0bb79d2191ca0a58b5
|
|
Change-Id: Id53c1dd58b95c45f261fa9bfae16cc53b20edd18
|
|
A request to send an Immediate Assignment over the PCH (not AGCH)
is always coming from the PCU. It's used for DL TBF assignment.
Change-Id: If4b0aa01532ab65b96201ff8829e724c67df6993
|
|
Related: OS#5809
Change-Id: I1ae5a7169d864e5dbd052437addd167ffa2c510d
|
|
Premature activation of virtual scheduler in bts_model_phy_link_open()
leads to segfault in vbts_sched_start(). Fix this by moving scheduler
activation to bts_model_oml_estab()
Change-Id: I116c2548dd4d05df90c16e81fa2e85ed6027a2e1
|
|
Fix condition `block_nr == 4`, which was never reached, by effectively
replacing
- `4` by `msg->num_segs`, so we truly check if the current
block is the last of this message
- the index `block_nr` (which starts at zero) by `block_nr + 1`,
so it is comparable to `msg->num_segs`
Related: OS#5354
Change-Id: I3dc116d6c16c80b31712b2a969f0b2a6998b03f0
|
|
Change-Id: Iaa207d4996a00befd99194c65fc6e6cd9dbaa340
|
|
Related: OS#5783
Change-Id: I268ba716ded330a024a8f3c0d1bd2f28622cecab
|
|
Lets use ARRAY_SIZE instead of magic number (8) to determine the
number of timeslots to iterate.
Change-Id: I0e016e6fff420851dfabebef5d8d43c735b968b2
|
|
Change-Id: Ib33ffba348d6e9414eb904bfb7a2bd7ba2f55344
|
|
Looks like this is not needed, so make the jenkins build work without
installing dahdi-source.
Related: OS#5863
Change-Id: I61c483983a4793e0429bb37804dee0a128125daf
|
|
The VTY command is "local-port", but write-config would write "port"
instead, which would fail when re-reading the config file.
Realted: SYS#6237
Change-Id: Id08530b626b0e69c3b3bb9d8bb9e16080a73e74d
|
|
Related: OS#5809
Change-Id: I2da7d7bf2b07b6736ab09a11ec37afc12f5ec075
|
|
Change-Id: Ieca05004255c40287e6427560c2636b39529cf07
|
|
Both power_ramp_start() and power_ramp_force() are now small macros
around _power_ramp_start()
The new behavior is:
* ramp down power when stopping bts through Ctrl-C
* the other shutdown causes skip power ramping
This will cause the bts to reconnect faster when the oml link is
dropped and power ramping is enabled.
Change-Id: Ida1d7bbf094958b9142af306dbf84206729a609e
Related: SYS#6237
|
|
Fixes: OS#5832
Related: https://gitlab.com/gpsd/gpsd/-/blob/release-3.20/gps.h#L63
Change-Id: I6dc8ce303e5cb0fb412857a7f2c925e8cfe9b1e0
|
|
There is no reason why they should have different initial
states. Keep it consistent with the other MOs.
Change-Id: I9fd744ec79da9fc26d3ebe9857b2b0bbd5fcd1ff
|
|
This reverts commit c96d34f8281a307f9e4cf6fd00adc6eba377012c.
Reason for revert: This breaks ramping up power since the power ramp logic still assumes the output is full power.
Change-Id: I47a16a4b3ba02d74473569c0f4350a41fc12a464
|