Unlike xCCH, TCH/H channels are using block diagonal interleaving,
so every single burst carries 57 bits of one traffic frame, and 57
bits of another one. Moreover, unlike TCH/F where both traffic
and FACCH/F frames are interleaved over 8 bursts, a FACCH/H is
interleaved over 6 bursts, while a traffic frame is interleaved
over 4 bursts.
This is why a TCH/H burst transmission can't be initiated on
an arbitrary TDMA frame number. It shall be aligned as of
stated in GSM 05.02, clause 7, table 1.
This change introduces two basic functions:
- sched_tchh_block_map_fn - checks if a TCH/H block transmission
can be initiated / finished on a given frame number
and a given channel type;
- sched_tchh_block_dl_first_fn - calculates TDMA frame number of
the first burst using given frame number of the last burst;
and some auxiliary wrappers to simplify the usage of
sched_tchh_block_map_fn().
Change-Id: Iaf4cb33f1b79df23f8a90c8b14ebe0cd9907fbb9
The 'normal' math operations, such as addition and substraction,
are not applicable for TDMA frame numbers because they may result
in out-of-range values.
Having TDMA frame math helpers in a single place would allow
one to avoid possible out-of-range result mistakes.
Change-Id: Ibb66ba846cc3d6c2eaa88414569e5f3751128047
GSM48_CMODE_SIGN means 'signaling only', so we shall not send
bad frame indications in this state. Instead, it makes sense
to send dummy L2 frames like we do for xCCH channels.
Change-Id: Ie39d53522cafab265099076b3194fa96aff217ba
Despite the correct range of Timing Advance value is [0..63],
there is a special feature in OsmocomBB which allows one to
simulate the distance between both MS and a BTS by playing
with the signal delay.
This is why a signed 'int8_t' type is used in L1CTL protocol.
No need to limit the range, just forward it to TRX.
Change-Id: I06774b315b8451bf14083da6b2849d6e8594abc8
I am not sure we need the both control commands, as every burst
on DATA interface has a header that includes TX power.
Change-Id: Id14603e71df6dedb5a843bb3e20a320192dbca3d
Let's differentiate between 'expected' unimplemented messages
like L1CTL_NEIGH_PM_REQ and truly unknonw message types.
Change-Id: Id76993056fb514e6fb0242d505205316c61bb965
A BSC may allocate a dedicated channel on any ARFCN, not necessary
on the same one where a mobile station has requested this channel.
For some reason, the ARFCN info of L1CTL_DM_EST_REQ message was
not handled by trxcon. Let's fix this.
Related: OS#3526
Change-Id: I16ed5c64236c159bfa39002b05094c1f6c171f6b
In the most cases an ARFCN value is stored together with some
flags (e.g. DL/UL flag, DCS flag), so it should be taken into
account e.g. when printing. Let's use the proper naming.
Change-Id: I0b7634c80986dbff9d0da421c6a044cd36c9fd01
There is no need to check the range of timeslot number, which is
decoded from GSM 08.58 channel number (9.3.1) by applying 0x07
mask, because any result of this operation is always within
the correct range.
Change-Id: Ib84417099d303bd3ae3557f48a5c40b812c6cdfc
Use osmo_clock_gettime() to read the monotonic clock instead
of gettimeofday() which could drift backwards.
This requires switching the scheduler clock from struct timeval
to struct timespec. Expand some variables to 64 bits in order
to keep types used in calculations compatible.
The previous implementation unconditionally subtracted microsecond
values from different time measurements, causing overflow if the
current measurement was taken in less of a fraction of a second
than the past measurement. Use timespecsub() for the subtraction
instead which accounts for fractions of a second correctly.
Change-Id: Ic93f90685c6d6dc28dfc4ad48c998e0eac113cf8
Related: OS#3467
This field of the logical channel state structure was not used at
all as there is nothing related to A-bis / RSL in trxcon itself.
Change-Id: Iec1abf777a74cf57deadafa95e2337cba5d02842
When relying on GSM 04.08 channel mode (GSM48_CMODE_*), one should
distinguish between Bm (full rate) and Lm (half rate) channels.
This change prevents the scheduler from generating TCH/F BFI
instead of TCH/H BFI on the corresponding channels.
Change-Id: I4547aa7f6d38637692fef8a0122e85fb52039a46
Instead of passing the information about a logical channel, it
makes sense to pass the pointer to its state where everything
is stored. This approach would allow to avoid adding more
arguments every time, e.g. in case of AMR.
Change-Id: I91fe86fef43aac68776a58c9acc37ef2a9ee8042
Initially it was assumed that FACCH prioritization should be done
in the same way for both TCH/F and TCH/H. Moreover, it was not
possible to confirm this, because TCH/H was (and still) not
implemented yet. But according to the specs:
- unlike FACCH/F, FACCH/H transmissions shall be aligned
within a multiframe, i.e. can only be initiated on
particular frame numbers (see GSM 05.02, clause 7);
- unlike FACCH/F, a FACCH/H frame steals two TCH/F frames;
so the TCH/H (including FACCH/H) primitives should be handled
separately from the TCH/F (including FACCH/F) primitives.
Change-Id: I9b59f60e1cbac8fb8fd557b6c67b5e376c0a6bbb
The previous primitive dequeuing logic (especially for TCH/F
channels) was a bit complicated, and it could not be possible
to reuse the existing code parts in the upcoming implementation
of both TCH/H and FACCH/H channels without changing anything.
In particular, this change introduces two internal functions:
- prim_dequeue_one(), which merely dequeues a primitive
of a given channel type (e.g. TRXC_SDCCH4_0);
- prim_dequeue_tch(), which dequeues either a FACCH,
or a speech TCH primitive of a given channel
type (Lm or Bm).
So the logic of the TCH/F prim dequeuing function has become
cleaner, and the upcoming TCH/H prim dequeuing function, where
FACCH/H prioritization is more complex than FACCH/F, will
reuse the introduced functions.
Change-Id: Ib82ad2480ab1bc6b1df9576eb2bf5acbd398bf66
We used to trust (and still doing this) the messages coming from
L1CTL interface too much, and not to check the primitive length
before passing the payload to the libosmocoding API. As was
discovered and described in OS#3415, sending a L1CTL message
(either DATA_REQ, or TRAFFIC_REQ) with an incorrect length
(lower than expected) may cause heap overflow.
Let's explicitly check a primitive before encoding, and drop it
if its length doesn't match the expected value(s).
Change-Id: I258ee9f6d0124b183b1db23a73f1e523fcea89a8
Fixes: OS#3415
Currently Access Burst generated by trxcon
has 8 zero bits at the beginning. According to
the 3GPP 05.02 specification (Chapter 5.2.7
Access burst) custom 8-bit extended tail bits
sequence should be used:
(BN0, BN1, BN2 ... BN7) = (0,0,1,1,1,0,1,0)
After this fix trxcon sets correct 8-bit
sequence at the front of Access burst.
Change-Id: I1f624e783de6c585d2e292965c9e5810b0a4f27d
The main problem here is that the existing implementatin missing the L1
header in this message. A SACCH message doesn't have a 23byte LAPDm
message, but only a 21 byte LAPDm message prefixed by a 2-byte Layer1
header. So on the receiver in the BTS, right now the first two bytes of
the UL SACCH frame are misinterpreted as L1 header.
This it what causes RLL ERROR INDICATION on the Abis side, which is why
our BTS_Tests fail.
Change-Id: Id7776bf3604d0e8a32e04547e01b8bd377903272
Related: OS#3170
In order to be able to introspect not only the root application
context, but also all other contexts, e.g. allocated within
libosmocore or other libraries, let's enable tracking the
use of NULL contexts using the corresponding talloc API.
Change-Id: Id21cd5ee340def443f7a5d0b2b8f37f41188dd87
This is useless, and prevents us from finding potential memory
leaks at exit. Let's print talloc report instead of that.
Change-Id: Ibf04942070d654e97c3ed77d69ab19e44602758c
The osmo_init_logging() doesn't allow to specify a talloc context
for libosmocore logging subsystem, so this is why the new version
was introduced. Let's use it.
Change-Id: I06c4a1f7f839f774bc428e89cfac30132bae904d
the initial tch_mode is not always 0 (signalling) but can very well
be directly a codec mode, if the initial activation of the channel
is in speech mode as opposed to signalling
Change-Id: I96e4c89da1165e9c5287d863e0e65d811460c606
According to TS 144.006, section 5.2, the first octet containing
fill bits shall be set to the binary value "00101011" == 0x2b.
Change-Id: I8f0304bf84613a2dc07cb78aff0cb8bb4c5adf6c
We shall always send traffic frame indications, even if received
frame is incomplete or decoding was failed. This is required
for proper Measurement Reporting.
Change-Id: I99e134699796c7075299459e96b2f2d462636619
We shall always send data frame indications, even if received
frame is incomplete or decoding was failed. This is required
for proper Measurement Reporting.
Change-Id: I7beee7e797f488d04c3b59bee9501ce823717092
Since this change, each lchan handler shall manually indicate
a type of both message indications and confirmations.
Change-Id: I02e0b87d61c127d2f6f5b9532909af78332bf707
- change 'l1ctl_tx_data_ind' symbol to 'l1ctl_tx_dt_ind' in
order to indicate that it's used for both DATA and TRAFFIC;
- introduce a 'traffic' flag, which is used to define either
TRAFFIC or DATA indication type;
- pass L2 payload and its length separately from the
Downlink info header.
Change-Id: I9fe65ee9b2d772576b86b7bc85d53518530d1579
- change 'l1ctl_tx_data_conf' symbol to 'l1ctl_tx_dt_conf' in
order to indicate that it's used for both DATA and TRAFFIC;
- introduce a 'traffic' flag, which is used to define either
TRAFFIC or DATA confirmation type;
Change-Id: Iedd569086a264dc7d8740abea5c6e5ca21e299f6
Both functions are almost identical, and the only difference is
the message type they set. Let's combine them into a single
function and introduce a 'traffic' flag, which can be
used to define a message type.
Change-Id: I288f5d7b6cd242c4793973dcb3d2b1b6925d61a7
Each L1CTL message gets its own length pushed in front before
sending. This isn't specified in the 'l1ctl_proto.h', but
assumed in the code. Let's clarify this.
Change-Id: I118d00613aeaf5ff0bad1188fa5f7450d4ca8122
There are two types of L1CTL messages: received and to be
transmitted. Let's use proper names to indicate this.
Change-Id: I7c17687579282fa389bca35dc7edbc3582e55701
If at the moment of transmission there are no frames in TX buffer,
then either a dummy LAPDm frame (0x01, 0x03, 0x01, 0x2b ...) or a
silence frame (depending on a codec in use) shall be transmitted.
This is required for proper measurements on the BTS side.
Change-Id: Ie590990f2274ea476678f6b2079f90eeadab6501
This new flag is intended to indicate that continuous burst
transmission is assumed on particular logical channel. In other
words, if a logical channel has this flag, but there is nothing
to transmit in a TX buffer, then either a dummy LAPDm frame or
a silence frame shall be sent.
Change-Id: I25fcf9eeb787ffe5378d92532439e67d7d42fa65
The sched_frame_clck_cb() is responsible for UL burst transmission.
Iterating over each timeslot, it chooses a proper lchan handler
according to a current frame number and a multiframe layout in use,
takes a L2 UL frame from a TX buffer, and finally calls the chosen
handler in order to to encode and transmit a taken frame.
A handler should be called only for activated logical channels...
but for some long time, there was a bug, so each lchan was
processed, including inactive ones. It's time to fix this.
Change-Id: I33e3ecc14be3ae64dfd02789c7f0970c945582c9
The llist_for_each_entry_safe() should be used instead of the
llist_for_each_entry(), because it's safe against removal
of llist entry.
Found using Valgrind's memcheck tool.
Change-Id: I65234971ec152df038c5388da537a503060c215b
There's no need to express ToA value as a float. Let's turn it into
an int16_t with 1/256 symbol period accuracy throughout the code to
avoid both float arithmetic as well as loosing any precision.
Inspired by Idce4178e0b1f7e940ebc22b3e2f340fcd544d4ec.
Change-Id: I99c0f38db08a530d5846c474aba352aa0b68fe86
Vadim Yanitskiy
Harald Welte
Pau Espin