That field is not needed anymore, and it works only under the assumption
that only 1 poll request can be active at a time per TBF, which is not
true.
Change-Id: I9b8bed7741d385bab4cd8c64b841a78a02a05fe1
That field is not needed anymore, and it works only under the assumption
that only 1 poll request can be active at a time per TBF, which is not
true.
Change-Id: I63a34a702f028b871530fb7caeb13e8ea1cc78ac
Value 'false' is always passed by all callers of the function, so
there's no need to pass it. Furthermore, since it's false, there's no
need to access poll_fn since RRBP will always be invalid.
Change-Id: Ia48ce2a021865e76e813dedb22aca9c2522c5693
The poll_state logic was part of previous implementation (prior to pdch
ul controller) where the ssumption was that TBF could only had 1 POLL
request in transit, which is really not true. With current
infrastructure we don't need this state tracking at all.
Change-Id: Ie5b807ccd38aa736ae11b3310ca61ad0156ca4d4
This API is not really needed anymore, since anyway it works under the
assumption there can only be 1 POLL in transit per TBF, which isn't
necessarily true.
Change-Id: I875f51cade95faeb2d79dcebfead4c83e23a731b
This allows easily checking the initial reason to trigger the poll when
either it is received or times out.
Later on this reason can be transformed into an FSM event and sent to
the related FSM.
Related: OS#5020
Change-Id: Ie8fefd1f47ad674ce597a8065b15284088956bde
If Idle TBF timer (X2031) is set to 0, it means the TBF release is
immediately started once all queued data has been scheduled. In that
case, we must set FBI=1 (by setting cv=0) and move to FINISH state.
This used to work over the usual path where X2031 != 0, because release
start will alays happen at a later sched poll time where a dummy LLC
frame is sent and FBI set accordingly.
Change-Id: Ib20602936ae084c413f6bfe14eea33b602020be0
Old commit getting rid of LLC UI dummy and updating create_new_bsn()
function introduced a bug by not moving update of value m_last_dl_drained_fn
prior to a new break introduced.
As a result, the value is not updated in the case LLC queue becomes
drained but last few bytes are drained at exactly that moment.
Furthermore, then the IDLE tbf timer (X2031, keep_open())) returns always
true since according to it the drain never happened.
The impact of the bug is basically delaying a bit more than expected the
time the TBF stays in IDLE state with the TBF release process yet
to be started.
Related: OS#4849
Fixes: 7d0f9a0ec3
Change-Id: I7420aeffda3500bcdc990291e4a56511af433ff9
Under some circumstances, it could happen that a DL TBF is created as a
GPRS TBF due to not yet having enough information of the MS, and only
after the TBF is created the PCU gains that information and upgrades the
MS mode to "EGPRS". Hence, there's the possibility to run into a
situation where a GPRS TBF is attached to a EGPRS MS.
It may also happen sometimes that despite the TBF and the MS be EGPRS,
there's need to further limit the DL MCS to use, eg. MCS1-4 (GMSK).
As a result, when asking for the current DL (M)CS to use, we must tell
the MS which kind of limitations we want to apply. The later reasoning
was already implemented when GPRS+EGPRS multiplexing was added, but the
former was not being checked. Hence, by further spreading through the
call stack the "req_kind_mode" we match both cases.
Related: OS#4973
Change-Id: Ic0276ce045660713129f0c72f1158a3321c5977f
There's no real use of C++ in that file, and it causes problems when
using llist_head entry macros in future patches adding initial support
for multiple BTS in PCU object, so let's move it to plain C.
Change-Id: Ic771a89fd78b5e66151a5384f0ff6a8895589466
Previous work on BTS class started to get stuff out of the C++ struct
into a C struct (BTS -> struct gprs_glcmac_bts) so that some parts of
it were accessible from C code. Doing so, however, ended up being messy
too, since all code needs to be switching from one object to another,
which actually refer to the same logical component.
Let's instead rejoin the structures and make sure the struct is
accessible and usable from both C and C++ code by rewriting all methods
to be C compatible and converting 3 allocated suboject as pointers.
This way BTS can internally still use those C++ objects while providing
a clean APi to both C and C++ code.
Change-Id: I7d12c896c5ded659ca9d3bff4cf3a3fc857db9dd
As we integrate osmo-pcu more and more with libosmocore features, it
becomes really hard to use them since libosmocore relies heavily on C
specific compilation features, which are not available in old C++
compilers (such as designated initializers for complex types in FSMs).
GprsMs is right now a quite simple object since initial design of
osmo-pcu made it optional and most of the logic was placed and stored
duplicated in TBF objects. However, that's changing as we introduce more
features, with the GprsMS class getting more weight. Hence, let's move
it now to be a C struct in order to be able to easily use libosmocore
features there, such as FSMs.
Some helper classes which GprsMs uses are also mostly move to C since
they are mostly structs with methods, so there's no point in having
duplicated APIs for C++ and C for such simple cases.
For some more complex classes, like (ul_,dl_)tbf, C API bindings are
added where needed so that GprsMs can use functionalitites from that
class. Most of those APIs can be kept afterwards and drop the C++ ones
since they provide no benefit in general.
Change-Id: I0b50e3367aaad9dcada76da97b438e452c8b230c
According to:
* 3GPP TS 44.060 version 16.0.0 "9.3.1a Delayed release of downlink Temporary Block Flow"
* 3GPP TS 44.064 version 16.0.0 "6.4.2.2 Unconfirmed Information (UI) Dummy command"
LLC UI Dummy frames are to be used when there no more data to send, only
in order to delay the release of a TBF. Hence, while not incorrect per
se, makes no sense to send those LLC UI Dummy frames inserted into
rlcmac blocks which already contain other LLC frames, since the MS in
that case is already being kept active.
It only makes sense to send those LLC UI Dummy frames when we have
nothing else to send, that is, alone inside a RLCMAC block without other
LLC frames.
Related: OS#4849
Change-Id: Ifae1a7b2b3dfad8df19585063088ba0df2749c8f
In previous status, if USF for GPRS-only MS was selected, then EGPRS
TBFs were skipped and either a GPRS TBF was selected or a Dummy Block
was sent. That means the behavior was unfair towards EGPRS TBFs, because
sometimes they were skipped in favor of GPRS ones.
This patch imporves the situation in the above mentioned USF scenario, by
first, under specific conditions, allowing selection of an EGPRS TBF and
then forcing it to transmit in EGPRS-GMSK (MCS1-4) so that the
USF-targeted MS can still decode the USF, while at the same time
providing more fairness by allowing the EGPRS TBF to transmit data.
The specific conditions mentioned above are, mainly, related to the fact
that once a DL data block has been sent, and hence a BSN was assigned to
it, it cannot be retransmitted later using another MCS, since lower
MCS1-4 wouldn't be able to contain higher MCS RLC payload.
The set of conditions could be expanded in the future by also selecting
the EGPRS TBF if retransmition is required and the block to be
retransmitted was originally transmitted as MCS1-4.
Related: OS#4544
Change-Id: I9af23e175435fe9ae7b0e4119ad52fcd4707b9ca
For instance if PCU received DL data to be sent to an MS from an SGSN,
and the MS is not currently cached in the PCU (because there's no TBF
active for it), it will page it and transmit the DL data to it.
The SGSN is capable of sending (EGPRS) MS Class information in that same
DL data message, so it's the one responsible for providing that
information if not available at the PCU.
In the PCU if we don't have information about that MS and SGSN didn't
provide us information about it, we cannot assume the MS is going to be
EGPRS capable and even less expecting a specific EGPRS MS class.
So let's drop this code.
Related: OS#4544
Change-Id: Icce66cadb51af25ae0c3b3719940eccb548fe33b
If SGSN provides us with MS class information upon DL data, let's use it
and set it in an already existing MS object if not yet known.
Also remove all unneeded code passing ms_class to append_data() which
would simply try to (again) set the ms_class.
Change-Id: I4979c9344bffd3ba7657bbab94981d233eab801f
Avoid passing tons of params to internal helper function
tbf_nel_dl_assignment() in order to either fetch again the ms object or
create a new one. Let's instead create the ms function earlier if needed
and fill it with all the discovered information prior to calling the
helper function. This provides cleaner code and also better log output.
This way we also avoid trying to fill the MS twice and unneeded
getter+setter for TA.
tbf::imsi(): There' always an ms, so simply forward call to
ms()->imsi().
We can also get rid of assign_imsi, since the modified code is the only
place where it's used and there's already some code in place there to
update the MS. We instead merge it with set_imsi and keep the
duplication code to catch possible bugs from callers.
Move merge_and_clear_ms from tbf class to GprsMS, where it really
belongs.
Change-Id: Id18098bac3cff26fc4a8d2f419e21641a1f4c83b
This is another step forward towards a more clear data model where a TBF
always has a MS object (which may be lacking some information, and at a
later point when more information is found, it may actually be a
duplicated MS object and hence one duplicate removed and the TBF moved
to the object being kept).
This helps for instance in removing duplicated information stored in
the TBF which is really per MS, like ms_class, ta, etc. Since there's
always a MS object there's no need to keep a duplicate in both classes
in case there's no MS object.
It can already be seen looking at unit test logging that this kind of
data model already provides better information.
Some unit test parts were needed to adapt to the new model too.
Change-Id: I3cdf4d53e222777d5a2bf4c5aad3a7414105f14c
Return an interface to the window base class so that the tbf base class
can access the common window methods, such as set_ws(). It will be used
in next commit to get rid of duplicated function enable_egprs in both
dl_tbf and ul_tbf subclasses.
The user of the function can then decide to access more specific
functionaltiites of the window class by static casting it to the
specific direction (which is known by the caller since it operates on a
ul_tbf or a dl_tbf).
Change-Id: Ia2e1decf91be1184668e28297c2126affb9c7ae4
According to 3GPP TS 44.018 sec 10.5.2.40, Timing Advance value is 8 bit
and range is 0-63 (0-219 on GSM400). So there's no need for 16 bits to
store it. uint8_t is used in all other places in the code.
Change-Id: I38aa063ae30ca5680fef6252d2cef22cea98c123
Since commit 322456ed47 (and previous
one), it is expected that a tbf object ALWAYS has a MS object referend
to it, even if it's a temporary copy which will later be merged when
TLLI/IMSI is retrieved and it is found that several MS objects relate to
the same MS.
The purpose of set_tlli_from_ul was mainly to update TBF's ms() to
old_ms before going through usual tbf->update_ms() path. That's not
really needed since ms() is already always set and TBFs for old_ms are
already freed in update_ms() and children function.
Change-Id: Ie8795e7a02032336e53febb65c11f9150c36d2a0
We have same kind of object splitted into two layers, in coding_scheme
and gprs_coding_scheme. Let's merge them together and get rid of the
class, which is not really useful because it's only a set of functions
operating on one enum value.
This change also fixes gcc 10.1.0 error about memseting a complex type
in rlc.h init().
Change-Id: Ie9ce2144ba9e8dbba9704d4e0000a2929e3e41df
Newer gcc 10.1.0 is erroring due to memset being applied on a complex
type, so let's start by removing this only function outside of the
struct.
Change-Id: I20426557d9b3049ab275fadb92e10ea8a860a119
It's super annoying seeing lots of functions being called everywhere
only to find out they are only incrementing a counter. Let's drop all
those functions and increment the counter so people looking at code
doesn't see dozens of code paths evyerwhere.
Most of the commit was generated by following sh snippet:
"""
#!/bin/bash
define_pattern="^CREATE_COUNT_ADD_INLINE"
generic_func="do_rate_ctr_add"
grep -r -l "${define_pattern}" . | xargs cat | grep "${define_pattern}("| tr -d ",;" | tr "()" " " | awk '{ print $2 " " $3 }' >/tmp/hello
while read -r func_name ctr_name
do
#echo "$func_name -> $ctr_name";
files="$(grep -r -l "${func_name}(" .)"
for f in $files; do
echo "$f: $func_name -> $ctr_name";
sed -i "s#${func_name}(#${generic_func}(${ctr_name}, #g" $f
done;
done < /tmp/hello
grep -r -l "void ${generic_func}" | xargs sed -i "/void ${generic_func}(CTR/d"
grep -r -l "$define_pattern" | xargs sed -i "/$define_pattern/d"
"""
Change-Id: I966221d6f9fb9bb4f6068bf45ca2978008a0efed
It's super annoying seeing lots of functions being called everywhere
only to find out they are only incrementing a counter. Let's drop all
those functions and increment the counter so people looking at code
doesn't see dozens of code paths evyerwhere.
Most of the commit was generated by following sh snippet:
"""
#!/bin/bash
grep -r -l ^CREATE_COUNT_INLINE . | xargs cat | grep "^CREATE_COUNT_INLINE("| tr -d ",;" | tr "()" " " | awk '{ print $2 " " $3 }' >/tmp/hello
while read -r func_name ctr_name
do
#echo "$func_name -> $ctr_name"
files="$(grep -r -l "${func_name}()" .)"
for f in $files; do
echo "$f: $func_name -> $ctr_name";
sed -i "s#${func_name}()#do_rate_ctr_inc(${ctr_name})#g" $f
done;
done < /tmp/hello
grep -r -l "void do_rate_ctr_inc" | xargs sed -i "/void do_rate_ctr_inc(CTR/d"
grep -r -l "CREATE_COUNT_INLINE" | xargs sed -i "/^CREATE_COUNT_INLINE/d"
"""
Change-Id: I360e322a30edf639aefb3c0f0e4354d98c9035a3
We should really be using monotonic clock in all places that
gettimeofday is used right now. Since clock_gettime() uses timespec,
let's move all code to use timespecs instead to avoid having to convert
in several places between timespec and timeval.
Actually use osmo_clock_gettime() shim everywhere to be able to control
the time everywhere from unit tests.
Change-Id: Ie265d70f8ffa7dbf7efbef6030505d9fcb5dc338
As seen in OS#4420, setting the MetaInfo.recv_time outside of
llc_queue before calling llc_queue::enqueue() and later on using that
value in llc_queue itself at dequeue time is not a good idea, since it
can provoke errors if the recv_time was not set correctly.
For instance, LlcTest was not setting the value for recv_time on some
test, which ended up with a huge millisec value when substracting now()
from it:
"""
llc.cpp:215:29: runtime error: signed integer overflow: 1582738663 * 1000 cannot be represented in type 'long int'
"""
This issue only appeared when started building on a raspberrypi4.
Let's better set/store the MetaInfo.recv_time internally during
llc_queue::enqueue(). Then, enqueue() only needs the
MetaInfo.expire_time, so let's change its arg list to only receive that
to avoid confusions.
Take the chance to move the llc_queue APIs to use osmo_gettimeofday,
since we need to fake the time now that the API itself sets that time.
Also take the chance during this refactor to disallow passing null
pointer by default since no user needs that.
Finally, update the LlcTest accordingly with all API/behavior changes.
Related: OS#4420
Change-Id: Ief6b1464dc779ff22adc2b02da7a006cd772ebce
Pau Espin
Pau Espin Pedrol