The abnormal case is that lchan_free ist getting called due
a RSL_MT_CHAN_REL_ACK in the RSL but the refcount of this
channel is not zero. This means that some "logical operation"
is still going on that needs to be cancelled.
Instead of always queuing up all operations in the
struct gsm_lchan use the signal framework to inform higher
layers about this abnormal case.
In gsm_04_08.c a signal handler is installed and in the
abnormal case the location updating request operation is
freed.
This was a stupid decision. We will have to assign the
network at some other place. The problem will be a problem
when we have two gsm_networks in one process and the same
subscriber is traveling in both networks.
The pointer was initially added to support to open a lchan
by using the gsm_subscriber structure. We would have restored
the struct gsm_bts from the db when loading the subscriber.
Instead we will dynamically resolve the gsm_bts and might end
up paging multiple bts in the same (stored/old) location area
until it answers.
It will define -DVTY_CRYPT_PW= if crypt is found. If -lcrypt is
needed, $(LIB_CRYPT) will be -lcrypt.
Compiles on OS X and Linux and reenables crypt.
Patch by: Lars Immisch
Tells mISDN not to release layer 2 on closing socket, when not
requested. If mISDN was told to release layer 2 once, it will continue
to release layer 2 on every shutdown of OpenBSC.
During bootstrap of BS11, the e1links are initialized. This must also be
done when BS11 is already bootstrapped (when restarting OpenBSC). It is
required to correctly multiplex the audio traffic between channels.
Without it, all time slots refer to card 0, slot 0, subslot 0, which
causes crashes when handling TRAU frames..
This patch will fix subscriber usage countinig.
It may happen, that the subscriber count will not be 0 after releasing
of a call. (This problem is solved with the application patch (27),
which will replace static call transaction by a dynamic transaction
list, and use subscriber for each transaction created.)
* add nm_state objects for RACK and CCLK
* add obj_class human-readable printing for all vendor obj_classes
* add human-readable printing of administrative state
* add enum for cell_alloc numbers
A new function trau_recv_lchan() is used to link a channel to a call reference
of a transaction. (Transactions are used in later patches.) TRAU frames will
then be forwarded to the application with the given call reference (in later
patches). Also the application can send TRAU frames by using trau_send_lchan().
A new list is introduced in trau_mux.c. (upqueue_entry) All subslots
that must be sent to application are listed here.
Received TRAU frames are written in the upqueue of application
interface, if a call reference is found in the upqueue-list. If an entry
is found the ss_entry list, the TRAU frames are bridged as before. The
frames have a message type (msg_type), a call reference (callref) and a
trau frame (data). The length of trau frame is defined by the content of
the c-bits inside the frame.
There is no support for ip.access yet, as they don't use the traditional
TRAU frame format. Harald must add this in order to use application interface
with ip-access. The bridging with ip-access works as before.
(Andreas Eversberg)
no IE type included in the message. These information elements are
mandatory, so their actual IE type is known. The improved parse_tlv()
function allows to parse zero, one, or two length-value elements.
(Andreas Eversberg)
is used for the E1's time slot. This timer replaces the "usleep()"
function, so the process will not block the execution of libbsc. The
timer is started after a frame is transmitted. If another frame is in
the transmit queue, the frame will only be queued until the timer times
out. If the timer is not running or times out, the frame is transmitted
and the timer is restarted.
The problem with partly provisioned TRX (locks show on LMT) is solved.
The adjustment for the inter frame delay of 50 miliseconds is for
further study.
(Andreas Eversberg)
select function ob libbsc. A "polling" flag is used to enable polling.
In this case select() will not sleep until file descriptor events occurr
or nearest timer expires. Also a return value will indicate if there was
an event that has been handled. If there was an event, the application
decides to poll again and don't wait.
In case for bsc_hack, the polling flag is not set. select will sleep as
usual.
(Andreas Eversberg)
The problem in the old logic was that we started talking to a given object
(e.g. radio carrier) one we received an administrative state change report. It
turns out we have to wait for the software activation report instead - and
everything suddenly works.
* keep track of site_id/bts_id in struct gsm_bts
* dynamically match incoming OML/RSL over TCP connections by BTS Unit ID
* introduce new debug category DINP (separate from DMI for hexdumps)
* remove ia_config() as it is no longer needed
*
* ensure that signalling links / E1 line information is correctly printed
* when bootstrapping RSL or OML, tell us for which BTS it is being doen
* separate bootstrap_bts() out from bootstrap_network()
* statically configure two ip.access BTS, one with unit id's 1800/0/0 and 1801/0/0
Patch by Andreas Eversberg to automatically release
layer2 on exit of the application. Made the naming of
the variables consistent (only release_l2 and not both
release_l2 and l2_release).
Let us establish the rule that there is one DB
per network. This will allow to assign a BTS of
the network to a subscriber that is loaded from
the db across restarts.
We directly assign the call->state and then check for something
that will never be true, and then immediately put the lchan and
schedule it's disconnect... and then directly after having closed
it down we send a message...
Change this to uncondtionally put down the lchan after having
changed the last(?) command.
Kill mi_set_pcap_fd from the header, introduce e1_set_pcap_fd
and call it from bsc_hack.c. Hook into abis_rsl_sendmsg and
_abis_nm_sendmsg for sending and e1inp_rx_ts for reading. It compiles
and should not cause a regression.
* gsm_util now uses caller-allocated data (rather than callee-allocated)
* correctly parse destination address
* parse (but not transcode) non-default encodings of SMS
* reject SMS to unknown destination number
* resolve target subscriber id and store incoming SMS (without header) in 'sms' table
What we're now missing is the sending part, i.e. a regular task iterating over
all pending SMS and trying to deliver them. Also, check for pending SMS once
we get a LOCATION UPDATE.
* initialize some data structures before using them in RSL
* DATA_REQ is a transparent message
* more elaborate DEBUGP statements here and there
* don't call 04.08 with zero-length RSL DATA INDICATION
* reject 04.08 CC HOLD and RETRIEVE, as we don't support them yet
- Fix the length (almost)
- Move the second part by 8 - b_off
e.g. with (ba) 0x62 0x61 the first bit of
0x61 goes to the highest on the first byte and
the second word contains 0x110000...
- The simple test case is almost passed... just a missing
character at the end.
- created for the creation of the SMS
- sent the time the message was submitted
- sender_id the subscriber for the return path/report
- receiver_id the subscriber
- header the (optional) SMS PDU header
- payload the payload (utf8), raw data
Apparently the BS-11 sends soem undocumented cause 0x18 as part of a CONN FAIL IND message
shortly after we establish the call. If we close the channel, the voice call
is aborted. If we ignore the message, everything just continues to work.
* add the concept of commands in addition to options
* add commands for querying essential configuration
* add command for explicit create/delete of TRX1
* software download is now an explicit operation
Holger Freyther
Harald Welte