Currently the gprs_llc class handles both LLC queueing and the
partition into smaller pieces for RLC/MAC encapsulation. This hinders
the separation of TBF and MS related data, since LLC queueing belongs
to the MS related code while the RLC/MAC encoding/decoding belongs to
the TBF layer.
This commits takes the LLC queueing related methods and members and
puts them into a new class gprs_llc_queue. It puts the queueing
object into gprs_rlcmac_tbf and adds accessor functions. The
implementation in tbf.cpp and tbf_dl.cpp is adapted accordingly.
Ticket: #1674
Sponsored-by: On-Waves ehf
The acc_delay value is computed but not used, resulting in a warning.
This commit adds a logging message it case the function is executed
to make the warning disappear. It also adds a CPP warning to remind
of the incomplete implementation.
Addresses:
sysmo_l1_if.c:226:10: error: variable ‘acc_delay’ set but not used
[-Werror=unused-but-set-variable]
Sponsored-by: On-Waves ehf
Since the timing advance storage has been removed, the TA values are
lost, when the last TBF of a single MS is freed. The TA storage has
at least saved the last 30 TA values. So now in more cases a wrong TA
value can be transmitted to an MS. Note that this also could have
happened before the removal of the storage, especially if more than
30 MS were in use.
This commit changes the default value of ms_idle_sec to 60s which is
higher than default value for T3314 (44s, see TS 24.008, 11.2.2),
after which the SGSN will have to start paging anyway. In that case,
a RACH request will be triggered, that will trigger an update of the
TA value in the PCU.
Sponsored-by: On-Waves ehf
This commit sets the MS timeout when the MS object is created. The
value is taken from the ms_idle_sec field in gprs_rlcmac_bts which
can be changed by the VTY commands shown below.
The following VTY commands are added to the config-pcu node:
- ms-idle-time <1-7200> Set the timeout in seconds
- no ms-idle-time Disable the timeout
Another timer that is related is T3314 (Ready Timer, default 44s, GSM
24.008, 11.2.2) which requires the SGSN to use paging after its
expiry. Longer timeouts can help if adaptive coding scheme selection
is used (not yet implemented). On the other hand, measurement values
(TA, signal quality) can get invalid after some time, especially with
a moving MS. So a value slightly above T3314 is probably a good value
to start with.
Sponsored-by: On-Waves ehf
Currently the MS object is immediately idle when all TBFs are
detached and if no guard is being used. Since the plan is to use the
MS objects to pass information from one TBF to the next one even
across the gap of some seconds of inactivity, a mechanism is needed
to keep the MS objects around for some time.
This commit extends the GprsMs class by a timer that keeps the MS
objects in non-idle state for some time after all TBFs have been
detached. The set_timeout method must be used with a non-zero value
to activate this feature.
Sponsored-by: On-Waves ehf
Currently the TA storage stores up to 30 TLLI->TA mappings, if more
entries are created the oldest one is dropped. In theory this can
lead to missing TA information if many MS are present.
This commit removes the TimingAdvance class completely, since the TA
value is now stored in the GprsMs objects.
Note that the GprsMs objects are currently not kept after the TBFs
have detached from them, so the TA values are now kept for a shorter
time than before.
Ticket: #1674
Sponsored-by: On-Waves ehf
The TA value rather relates to an MS and not to a single TBF. So all
TBFs share the same TA value. Currently the TA value is stored per
TBF and eventually copied from an old TBF to a new one. It is in
general only passed with an RACH request when the TLLI and thus the
MS is not yet known.
This commit adds a TA member to the GprsMs class and uses that one
when the TBF is associated to an MS object. Since the TBF is not
always associated with an MS object (after RACH or when it has been
replaced by another TBF), the TA value is still stored in each TBF
and that value is used as long as no MS object is being associated.
Sponsored-by: On-Waves ehf
These tests cover the message exchange from receiving from the first
RACH request to the first data block when establishing an uplink TBF.
This will be used to check, whether TA and other values are passed to
an MS object correctly.
In addition, the RX RACH log message in rcv_rach is extended to
contain the single block fn.
Sponsored-by: On-Waves ehf
Currently a new TBF is chained to an existing older one, either of
the other direction (active or releasing) or of the same direction
(releasing). This does not work properly work if and uplink and a
downlink TBF are being established at the same time while an old TBF
is being released. In that case, one of them is thrown away and the
pending procedure is cancelled.
The chaining is no longer necessary since the GprsMs objects have
been introduced which keep track of the active TBFs.
This commit removes the TBF members m_new_tbf and m_old_tbf and the
related methods and code paths.
Note that a new TBF can replace an older TBF entry of the same
direction within an MS object when it is associated with an MS (e.g.
by TLLI or because it is assigned via another, already associated
TBF). In that case, the old TBF is no longer associated with an MS
object.
Ticket: #1674
Sponsored-by: On-Waves ehf
Currently the DL TBF's TLLI are not set. This will have to change in
the future, when the m_new_tbf chaining is replaced by the usage of MS
object to group TBFs.
This commit just calls update_ms() on newly created TBFs. This
changes many of the lines written to stderr due to the TLLI change,
but doesn't change other aspects or messages beside creating and
destroying MS objects.
Sponsored-by: On-Waves ehf
Currently the BTS::trigger_dl_ass() method assigns the IMSI to the MS
object. This should be (and is already) done earlier where the MS
object is retrieved/created.
This commit removes the corresponding code along with the 'imsi'
parameter from trigger_dl_ass.
Sponsored-by: On-Waves ehf
This commit extends get_ms() to really compare the IMSI if it has
been given. Matching by TLLI has a higher precedence than matching by
IMSI.
Ticket: #1674
Sponsored-by: On-Waves ehf
Currently the IMSI is stored in the TBFs. Since it directly refers to
an MS, it should rather be stored in an MS object.
This patch move the m_imsi field from gprs_rlcmac_tbf to GprsMs,
changes gprs_rlcmac_tbf::imsi() to get the IMSI from the associated
MS object, and adds getter and setter to GprsMs. Before changing the
IMSI of the associated MS object, assign_imsi() checks if there is
already another MS object with the same IMSI and eventually resets
the IMSI of that one. So using update_ms() and assign_imsi() ensures
that there are not two MS object entries is the storage with the
same TLLI or the same IMSI.
Ticket: #1674
Sponsored-by: On-Waves ehf
Currently the generated IMSI contains blanks due to a missing '0'
flag in the corresponding snprintf statement. In addition, the buffer
is effectively limited to 15 bytes which is shorter than the
generated character sequence, such that the last character of the
IMSI is removed.
This patch fixes both issues. Since snprintf itself will add a
terminating \0 character, the whole buffer can be passed to snprintf.
Sponsored-by: On-Waves ehf
Currently the m_tlli member in GprsMs is set by the constructor,
circumventing the TLLI confirmation mechanism.
This commit replaces the get_or_create_ms() method by a create_ms()
method which takes the TLLI and the direction (UL or DL) as
parameters to select either set_tlli() or confirm_tlli(). The MS
object is instantiated with TLLI = 0, and therefore GprsMs::tlli() is
extended to return the DL TLLI if both of the other TLLI are not set.
Note that create_ms() will not check whether an MS object with a
matching TLLI is already stored in the list, so it should only be
called after a corresponding get_ms() in general.
Sponsored-by: On-Waves ehf
Currently the TLLI is stored in each TBF. Since each MS is now
represented by a GprsMs object which takes care of TLLI updating,
and each TBF that has been associated with an TLLI also contains a
reference to a GprsMs object, per TBF TLLI handling is no longer
needed. Keeping all TBF m_tlli members up to date is complex and
doesn't currently work correctly in all circumstances.
This commit removes m_tlli and related members from the TBF class and
the tbf_by_tlli functions from the BTS class.
Ticket: #1674
Sponsored-by: On-Waves ehf
The write_queue is designed to have a maximum amount of pending
messages and will refuse to take new messages when it has been
reached. The caller can decide if it wants to flush the queue
and add the message again, create a log. But in all cases the
ownership of the msgb has not been transferred. Fix the potential
memory leak in the failure situation.
Debian is using the classic bfd linker and when passing
libosmogb as link dependency it always wants/needs to
resolve the bssgp_prim_cb symbol (which is to be provided
by the application).
Only keep the libosmocore dependency.
Fixes:
lib/libosmogb.so: undefined reference to `bssgp_prim_cb'
collect2: error: ld returned 1 exit status
Makefile:511: recipe for target 'llist/LListTest' failed
Currently the MS will be searched based on the TLLI in resue_tbf().
Since the MS object is already known in the TBF when the TLLI is set,
it can just be passed to the new TBF.
This commit removes the call to update_ms() and just adds
new_tbf->set_ms(ms()) which will also work as expected if ms() == NULL.
Sponsored-by: On-Waves ehf
The type of the TBF update_ms() is being called on does not always
reflect whether the TLLI has been signaled by the MS or the SGSN.
This commit adds an additional parameter to tell the method, in which
direction the TLLI has been passed.
Sponsored-by: On-Waves ehf
Since the synchronisation of the TBF's concerning the TLLIs has been
removed in 'Support new and old TLLI's', gprs_rlcmac_tbf::tlli() can
return the old TLLI which is probably different to ms()->tlli() in
that case. This can lead to a wrong TLLI being used in BSSGP messages.
This commit modifies the TBF's tlli() method to get the current TLLI
from the MS object.
Sponsored-by: On-Waves ehf
According to the specification (GSM 04.08/24.008, 4.7.1.5) after a
new P-TMSI has been assigned, the old P-TMSI must be kept basically
until it has been used by both sides. Since the TLLI will be derived
from the P-TMSI, the old TLLI must also be kept until the new TLLI
has been used by both MS and SGSN.
This commit modifies the TLLI handling of GprsMs accordingly.
set_tlli() is only used with TLLIs derived from MS messages,
confirm_tlli() is used with TLLIs derived from messages received from
the SGSN. tlli() returns the value set by the MS. check_tlli()
matches each of the TLLI used by either MS or SGSN as well as the old
TLLI until it has been confirmed.
Sponsored-by: On-Waves ehf
Use the MS storage to find a MS object for a given TLLI instead of
searching the TBF lists. The TBFs are then taken from the MS object,
if one has been found. If all TBF might be temporarily detached from
the MS object, a GprsMs::Guard is added to prevent the deletion of
the object, in case another TBF gets attached later on in the scope.
Ticket: #1674
Sponsored-by: On-Waves ehf
Currently the MS objects are contained in the TBF objects only. To
allow for an extended life time after the TBF objects have been freed
and to find them based on TLLI, a container for the MS objects is
needed.
This commit adds the container class and also adds the corresponding
m_list member to GprsMs. Further integration into the PCU code is not
yet done.
Ticket: #1674
Sponsored-by: On-Waves ehf
This commit adds the LListHead class which is a wrapper around the
linuxlist. It adds an additional member to refer to the container,
since the container_of macro doesn't work properly with C++ classes.
All functions and macros from linuxlist.h are support except for the
entry macros (e.g. llist_entry, llist_for_each_entry, ...). To access
the container (entry), an entry() method is provided instead:
llist_for_each(pos, &elems) {
pos->entry()->do_something();
}
Sponsored-by: On-Waves ehf
To show the talloc report more easily, this commit adds a call to
talloc_report_full if the TALLOC_REPORT_FULL environment variable is
set (the value is ignored). Since the stderr output is checked by the
test suite, this feature is not enabled by default.
Sponsored-by: On-Waves ehf
This commit adds MS object creation and cleanup to the TBF related
code. MS objects are created when a TBF that has been "anonymous" so
far gets associated with a TLLI. When a TBF is replaced by another,
the old TBF is detached and the new one is attached to the MS. When
all TBFs have been detached, the MS object gets deleted.
The TBF related code should not call attach_tbf/detach_tbf directly
but use set_ms instead to make sure, that the references are updated
properly. GprsMs::detach_tbf also calls set_ms(NULL) on the detached
TBF object.
The MS object is not really used yet, the focus is still on object
creation, TBF association, and cleanup.
Ticket: #1674
Sponsored-by: On-Waves ehf
Currently only TBF objects are used to handle the data flow between
the MS and the SGSN. MS specific data (e.g. pending LLC frames, TLLI)
is copied between successive TBFs. If all TBFs (uplink and downlink)
are idle for some time, all information about the MS is discarded in
the PCU. This makes the implementation of some features more
difficult, e.g. proper TLLI and timing advance handling,
connection based CS selection, and proper management of multiple TBF.
This commit adds the GprsMs class that is intended to hold
information directly related to the MS and to keep references to the
active TBFs.
The class is not yet integrated with the other PCU code. A GprsMs
object container and MS specific fields (TA, CS) will be added in
later commits.
Note that calling detach_tbf() can possibly delete the MS object
depending on the callback implementation.
Ticket: #1674
Sponsored-by: On-Waves ehf
Currently the tag value in FLOW CONTROL BVC messages is always 1.
This commit changes the implementation to increment that value with
each of the FLOW CONTROL BVC messages that is sent to the SGSN.
Sponsored-by: On-Waves ehf
The specification 28.018, allows to transmit the average LLC downlink
queueing delay in FLOW CONTROL BVC messages (BVC Measurement IE, see
GSM 28.018, 10.4.4 and 11.3.7).
This commit extends gprs_bssgp_pcu.cpp to compute the average delay
time between two subsequent FLOW CONTROL BVC messages. The average is
implemented as an arithmetic average without any weighting.
Ticket: OW#1432
Sponsored-by: On-Waves ehf
Currently the bucket size is by default being computed based on the
leak rate and the expected life time of LLC frames. The latter is
either taken from 'queue lifetime' (if given) or a fixed value is
used. Using 'queue lifetime' has the drawback that it sets a 'hard'
limit, since frames will be dropped if they stay in the queue
for a longer time.
This commit adds a VTY command to specifically set the time used for
the computation of the bucket size advertised to the SGSN. It does
not affect the PCUs queue handling in any way. If the bucket time is
not set (or if the 'no' command has been used), the old behaviour
(see above) is applied.
The following VTY commands are added (config-pcu node):
- flow-control bucket-time <1-65534> Sets the time in centisecs
- no flow-control bucket-time Don't use this time
Ticket: OW#1432
Sponsored-by: On-Waves ehf
Currently the PDCH assignment and coding scheme does not influence
the values transmitted by the FLOW-CONTROL-BVC messages.
This commit adds the computation of those values. If the leak rate is
not given explicitly, it is derived from the number of PDCH and the
allowed coding scheme. If the BVC bucket size is not given
explicitly, it is derived from the leak rate and the maximum buffer time.
The latter is taken from the 'queue lifetime' command (or 10s if this
has not been used). The MS default bucket size is set to 50% of the
BVC bucket size. The MS default rate computation assumes, that each MS
only supports up to 4 time slots for GPRS RX.
Sponsored-by: On-Waves ehf
Currently the FLOW-CONTROL_BVC message contains fixed values: The tag
is 1, the BVC bucket size is 6MB, the BVC bucket leak rate is
820kbit/s, the MS bucket size is 50kB, and the MS leak rate is
50kbit/s.
This commit makes the BVC parameters configurable and adds the
following VTY commands:
- flow-control force-bvc-bucket-size <1-6553500>
- no flow-control force-bvc-bucket-size
- flow-control force-bvc-leak-rate <1-6553500>
- no flow-control force-bvc-leak-rate
- flow-control force-ms-bucket-size <1-6553500>
- no flow-control force-ms-bucket-size
- flow-control force-ms-leak-rate <1-6553500>
- no flow-control force-ms-leak-rate
The 'no' variants restore the default behaviour.
Sponsored-by: On-Waves ehf
Currently this function which increments the corresponding counter is
just called in gprs_llc::clear(). It is not called on places where
LLC DISCARDED messages are sent.
This commit adds calls to llc_dropped_frame() at these places.
Sponsored-by: On-Waves ehf
Currently the ul_tbf is used in its own declaration as an argument to
tbf_alloc_ul_tbf, where dl_tbf can be used instead and makes more
sense.
This commit uses dl_tbf instead of ul_tbf as old_tbf.
Addresses:
tbf/TbfTest.cpp:71:14: warning: ‘ul_tbf’ may be used uninitialized in
this function
Sponsored-by: On-Waves ehf
Currently the global current_test is not initialised, which leads to
memory access failures. In addition, the variable is signed but used
in comparisons with unsigned values.
This commit changes the type to size_t and initialises the variable
to 0.
Sponsored-by: On-Waves ehf
A few prototypes are not part of any header files, leading to
corresponding compiler warnings.
This commit adds the missing prototypes to sysmo_l1_if.h.
Addresses:
sysmo_l1_if.c:347:2: warning: implicit declaration of function
‘l1if_transport_open’
sysmo_l1_if.c:364:3: warning: implicit declaration of function
‘l1if_transport_close’
sysmo_l1_fwd.c:89:3: warning: implicit declaration of function
‘l1if_handle_sysprim’
sysmo_l1_fwd.c:91:3: warning: implicit declaration of function
‘l1if_handle_l1prim’
sysmo_l1_hw.c:109:3: warning: implicit declaration of function
‘l1if_handle_sysprim’
sysmo_l1_hw.c:118:3: warning: implicit declaration of function
‘l1if_handle_l1prim’
Sponsored-by: On-Waves ehf
These fixes do not affect the semantics of the code. They either help
gcc by providing default values that won't be used ("may be
uninitialised"), remove unused variables, or change signed to
unsigned variables to avoid comparison warnings.
Addresses:
bts.cpp:494:32: warning: 'tbf' may be used uninitialized in this
function
emu/test_replay_gprs_attach.cpp:81:27: warning: comparison between
signed and unsigned integer expressions
emu/test_pdp_activation.cpp:95:23: warning: unused variable ‘budh’
emu/test_pdp_activation.cpp:97:6: warning: variable ‘rc’ set but
not used
emu/pcu_emu.cpp:109:26: warning: unused variable ‘bts’
alloc/AllocTest.cpp:74:27: warning: unused variable ‘tbf’
osmocom/core/utils.h:13:50: warning: comparison between signed and
unsigned integer expressions
types/TypesTest.cpp:319:7: warning: unused variable ‘count’
types/TypesTest.cpp:320:11: warning: unused variable ‘rbb’
alloc/AllocTest.cpp:74:27: warning: unused variable ‘tbf’
alloc/AllocTest.cpp:132:11: warning: unused variable ‘ts_no’
Currently an uint8_t csec variable is used to temporarily store
the number of centiseconds which can be set within the range
1 - 65535. This way values above 255 cannot be set properly. This
affects the VTY commands "queue lifetime", "queue hysteres", and
"queue idle-ack-delay".
This commit replaces the uint8_t by an uint16_t.
Sponsored-by: On-Waves ehf
This commit adds missing includes to pcu_vty.c and fixes the
llist_head type (missing struct keyword) used by tbf_print_vty_info.
Sponsored-by: On-Waves ehf
Currently the value -1 is assigned twice to m_last_dl_drained_fn
within append_data().
This commit removes the first of these assigments.
Sponsored-by: On-Waves ehf
Currently the BSSGP functions in libosmocore do not log correctly to
DBSSGP since the DBSSGP variable in common_vty.c is left
uninitialized.
This commit adds the call to bssgp_set_log_ss() to inform libosmocore
which sub system id it shall use for BSSGP.
Sponsored-by: On-Waves ehf
Currently the PCU silently discard LLC frames from the SGSN if a
DL TBF cannot be allocated.
This commit changes tbf_new_dl_assignment and reuse_tbf to send an
LLC discarded message to the SGSN in this case.
Ticket: #607
Sponsored-by: On-Waves ehf
This commit applies to following changes to pcu_vty.c:
- Use vty_install_default() instead of install_default()
- Add a blank after the '#' of the prompt
- Rename the 'pcu' configuration node to 'config-pcu'
Sponsored-by: On-Waves ehf
If the protocol layers above LLC (e.g. TCP) need an acknowledgement
to continue, it can take up to 400ms (single TS) until the MS is
polled for Ack/Nack which it can use to request an uplink TBF
quickly. The 400ms result from requesting an DL Ack/Nack every 20 RLC
blocks until all pending LLC frames have been sent.
Especially TCP's slow start mechanism can lead to a high delay at the
start of the connection, since the sender will eventually stop after
having sent the first packets (up to 4 (RFC2581) or 10 (RFC6928)).
This commit modifies append_data() to (re-)start
a timer every time it handles an LLC packet and to request an
Ack/Nack every time it expires. So if the server ceases to send IP
packets, the MS is polled in the assumption, that the server is
waiting for an ACK.
The following VTY commands are added (pcu node):
- queue idle-ack-delay <1-65535> timeout in centiseconds
- no queue idle-ack-delay disable this feature (default)
A sensible value is 10 (100ms) that at gave promising results when
testing locally.
Sponsored-by: On-Waves ehf
Currently single LLC blocks are discarded when the PDU lifetime
expires. If an IP packet has been fragmented either on the IP or on
the LLC layer and is therefore distributed over several LLC frames,
the kept fragments are transmitted and then discarded by the MS
because of the missing PDU. This can cause massive IP packet loss
when there are many fragmented packets (e.g. when trying 'ping
-s1800' or if the GGSN chops downlink IP packets into several SNDCP
packets).
On the other hand, discarding too many packets might disturb the
congestion handling of TCP. Dropping plain TCP ACKs might also hinder
flow control and congestion avoidance.
This commit adds a hysteresis algorithm to the LLC discard loop. If
an LLC message's age reaches the high water mark, further message's
with an age above the low water mark are discarded, too. This is
aborted, if a GMM, a non-UI, or a small message is detected. In
these cases, that message is kept.
The following VTY commands are added (pcu config node):
- queue hysteresis <1-65535> set the difference between high
(lifetime) and low watermark in
centiseconds
- no queue hysteresis disable this feature (default)
Since the SGSN will most probably send all fragments of a single
N-PDU without much delay between them, a value slightly above the
average transmission delay jitter between SGSN and PCU is probably a
sensible value to discard all fragments of a single IP packet.
This is an experimental feature that might be replaced by more
advanced means of active queue management in the future.
Sponsored-by: On-Waves ehf
If an MS wants to open a new UL TBF, it can either use (P)RACH or
request one in a Ack/Nack message for a DL TBF (PACCH). When a TBF
becomes idle (LCC queue is empty but the TBF is kept open), there
aren't any Ack/Nack requests that can be used by the MS to ask for an
UL TBF, therefore it has to use the RACH. This leads to many RACH
requests even for a single HTTP transaction, so it takes some time to
retrieve even a simple web page.
This commit modifies the scheduler to regularly send Ack/Nack
requests on idle DL TBFs. It does so by extending the priority based
scheduling algorithm to have 5 priority levels (highest priority
first):
- Control block is pending
- High age (100%) threshold reached (-> request Ack/Nack)
- Data is waiting or there are pending Nacks
- Low age (200ms) threshold reached (-> request Ack/Nack)
- Pending Nacks that have been resent already
- None of the above (-> send DL dummy control block)
The 'age' refers to the time since since the last control block has
been sent on the TBF. This high age threshold is set to
dl-tbf-idle-time or to 50% of T3190 (whichever is smaller), aiming
for at least a poll (and TBF shutdown) after the TBF has expired and
to safely prevent expiry of T3190. So if dl-tbf-idle-time > 200ms,
there will be a poll every 200ms and a final poll after
dl-tbf-idle-time. On high load, the interval between polls can get
higher, but the 'high age' poll should be in place.
This commit implements the scheduling with respect to GSM 44.060,
9.3.1a ("Delayed release of downlink TBF").
Ticket: #556
Sponsored-by: On-Waves ehf
Jacob Erlbeck