When we receive unimplemented/unhandled message types, we shouldn't
simply silently discard them, but print a log message for the
benefit of the user.
Change-Id: I65489578b1c214f193b1ce0e9ba59432dcd42a3e
The function handle_error asserts mgcp_ctx->conn to be non null,
but it does not access it otherwise.
remove unused variable conn
Change-Id: I09851c957395d1ddb2f9471b99ffc091bc250404
Even in the very early ST_CRCX_BTS phase, the error handler may
decide to go to ST_CALL in order to initate the termination of
a possibly half open connection.
Add ST_CALL to the out state list in ST_CRCX_BTS
Change-Id: Ic67aa7c67a4e98a38bff156be3ebf612012eb842
the switch statement in fsm_send_assignment_complete() has the
default case at the beginning.
Move the default case to the end to match common coding style
rules
Change-Id: I360842fe899b95972c44da3cb74a3dc51b379fdc
We recently started to use some symbols that were not available in
libosmo-mgcp-client-dev 1.0.0 or even 1.1.0. Let's depend on a newly
tagged version of libosmo-mgcp-client.
Change-Id: Ic5d3add1c69181aabbdb684a01a6ba7bcea1fe2c
So far, the config would log an error upon config parsing, and then continue to
use defaults, which is super easy to miss. On errors, return CMD_ERR_INCOMPLETE
to abort the program in a config parsing error.
Be fatal for non-existing addressbook entries and for using address book
entries from mismatching cs7 instances.
Though it is mixing in cosmetic changes, add "Error:" to the output and arrange
the erratic name to the end of the message, as is customary for error messages.
Related: libosmo-sccp I2f71b9c4dd30f919d2054da81283dd7035f44f60
Change-Id: Ia4e58902a2d3757b266cf35ac89f256cfb8f0eec
So far, if the user entered an invalid SCCP address in the config, the
osmo_bsc_sigtran_init() code simply replaced that with the default, i.e.
running with a completely different address than the user may intend.
Use the default SCCP addresses only when they are unset by the user.
Default MSC addr: set directly, do not detour via cs7 instance PC. The default
MSC SCCP addr is just a point code + SSN, deriving it from the cs7 instance
first is a confusing step. Just set the PC and SSN, and done.
Using default addresses does not constitute an "auto configuration": if we set
up a cs7 instance automatically, we do not want to have to create a second one
automatically, to prevent "auto-confusion", and want to bail instead. But for
each MSC on its own, using default SCCP addresses makes sense and is orthogonal
to automatic cs7 instance creation. Hence drop the auto config semantics from
the default SCCP address parts.
Always validate the SCCP addresses we will end up using, and bail immediately
if they are erratic. i.e. don't overwrite a non-empty invalid SCCP address with
defaults, but straight bail.
Beneficial side effects:
- Fix some grammar ultra confusion in log messages.
- Add context: log the MSC number the logging refers to.
- Drop code dup: since we're always logging the used SCCP addresses, might as
well log those once, unconditionally, in the end.
Change-Id: Iadbc2e9740457e1b389b7e7ad9c94274e7d8cb11
Use distinctive struct names: s/fsm_/fsm_bsc_reset/. They only exist
in the static context and it works fine, but the mad fsm-to-dot.py script
breaks with identical struct names. Can't hurt to have unique names.
Change-Id: I986377a74ccd83ca3b52e7f058bbc9115f05f741
Since Change-Id Ia2882b7ca31a3219c676986e85045fa08a425d7a, osmo-bsc
uses osmo-mgw and utilizes libosmo-mgcp-client to talk to it, so
let's make sure the Debian control file states that dependency.
Unfortuantely, this still won't make the osmo-bsc debian package
build again, as in fact the above commit uses symbols not even present
in 1.0.0 or 1.1.0 releases of libosmo-mgcp-client :( So we first
need a new release of that library, and we need to update the
configure.ac and debian/control version requirements in osmo-bsc
before this is fixed. This needs to be automatized in the future.
Change-Id: I41a0378d069f5383904cf92cc415c19beba26168
After the bssap test in Ie934c5d229140a89763bf2efff86d6a3766cd351, the
subsequent commit Ia2882b7ca31a3219c676986e85045fa08a425d7a was not tested
against the latest head, and its breaking bssap_test was not caught.
Fix current master of osmo-bsc's 'make check' target: add osmo_bsc_mgcp.c and
libosmo-mgcp-client dependencies to bssap_test linkage.
Change-Id: I28719d267452f66d65581c43433e24a9f46cf7dc
osmo-bsc currently negotiates the RTP stream directly with the
BTS and reports back the RTP IP/Port on the BTS. This works fine
for a single BTS, but for Handover the port/ip pointing to the
MSC side must not change, so an entity in between the BTSs and
the MSC is required.
Integrate the mgcp-client and use osmo-mgw to switch the RTP
streams.
Depends: osmo-mgw Ib5fcc72775bf72b489ff79ade36fb345d8d20736
Depends: osmo-mgw I44b338b09de45e1675cedf9737fa72dde72e979a
Depends: osmo-mgw I29c5e2fb972896faeb771ba040f015592487fcbe
Change-Id: Ia2882b7ca31a3219c676986e85045fa08a425d7a
3GPP TS § 08.08 defines various types of Cell Identifier List IEs, but we only
implement "entire BSS" and "one LAC". If the MSC sends a Cell Identifier List
that we don't implement, it is best for interoperability to page the entire BSS
and post a log message instead of rejecting the paging altogether. Apart from
resource management, it is not harmful to page more than the MSC requested; if
use of resources becomes an issue, the log message will guide towards the
solution of providing an actually implemented Cell Identifier List IE.
Upon IE length that is other than we expect, log the error, but also fall back
to paging the entire BSS. Overall message length correctness has been checked
earlier.
The particular case observed is that a Huwaei MSC sends a LAI for Cell
Identifier List (MCC+MNC in bcd, followed by a LAC), parsing of which we may
want to add later.
Improve logging: identify the subscriber that is being paged.
Coding style: use a switch() statement to clarify flow and provide a place to
add more implementations later.
Add regression test bssap_test.c: fabricates BSSAP Paging messages with the two
implemented Cell Identifier List IEs as well as the unimplemented LAI
identifier, verify the resulting paging LAC in wrapped function and stderr.
Change-Id: Ie934c5d229140a89763bf2efff86d6a3766cd351
To properly decide if a given OML link is degraded we have to use
BTS-specific information about MO state.
* move check function into BTS-specific part
* add generic wrapper
Related: OS#2486
Change-Id: Iddc7a4d20fbb95a6566eed1487a12733e5adb9e2
vty_install_default() and install_default() will soon be deprecated.
Depends: I5021c64a787b63314e0f2f1cba0b8fc7bff4f09b
Change-Id: If2edf59a687a78d6db6bc73117a27509374b0fc6
In Change-Id I469909ad7c597cde3d7a7d2ec86101a9f41d3aa6 we accidentially
also removed the libssl-dev dependency. osmo-bsc_nat still uses
RAND_getbytes directly, so we have to keep it for now, until we switch
to a future libosmocore-based mechanism that's in the works.
Change-Id: I3be26c566baf05278ba51b835a72e14ce6ecf3d0
See osmo-ci change I2409b2928b4d7ebbd6c005097d4ad7337307dd93 for rationale.
Depends: I2409b2928b4d7ebbd6c005097d4ad7337307dd93
Change-Id: I9b6afb59f0a8037d1510a7fddb63927f10d653e5
This fixes the following dpkg-shlibdeps warning:
Change-Id: Iea00c209652e8070a59942504bef660db0999e86
dpkg-shlibdeps: warning: package could avoid a useless dependency if debian/osmo-bsc-nat/usr/bin/osmo-bsc_nat was not linked against libosmoabis.so.6 (it uses none of the library's symbols)
We don't want to pacakge osmo-bsc_nat from osmo-bsc.git at this
point yet. It only suports SCCPlite, which is not yet fully supported
by osmo-bsc.
Rather, we continue to package osmo-bsc_nat from openbsc.git like we did
so far.
Also, the osmo-bsc_nat binary really doesn't belong into the osmo-bsc
package at all.
Change-Id: Icf0bf80d61141ec060b6d2efcf3e65e2ef1ac2d6
This fixes the following dpkg-shlibdeps warning:
Change-Id: I31af5fb8b52ef1fd5effb139d9cdea1ebe9a41b4
dpkg-shlibdeps: warning: package could avoid a useless dependency if debian/osmo-bsc/usr/bin/osmo-bsc_nat was not linked against libosmonetif.so.4 (it uses none of the library's symbols)
We are using quite a number of symbols that are definitely *not*
yet present in the respective library versions that we stated as
dependency. Rather than figuring this out individually, simply
require the latest releases.
Change-Id: Iecda06d206c24390bb10f3a8f8a70ef3036381e2
Those dependencies were introduced at a different time, when GPRS
related code was still in this repository
Change-Id: I469909ad7c597cde3d7a7d2ec86101a9f41d3aa6
We output a hexdump for each sccp message we receive, but not when
sending.
Also log the hexdump of sccp messages when sending.
Change-Id: Ibfe52a0b7dbca4c569c603a008d73d0d99d1c345
We already log the message type of sccp messages we receive, but
for transmitting the log output is missing.
Also log the message type for tramsitted sccp messages.
Change-Id: I6736f15ddc03e5f70c3504abffbae6cbf766c9d7
A FSM doesn't need "FSM" in its name, as it is obvious that it is a
FSM. Also, having two that are called RESET is confusing, so let's
try to come up with better names.
Also, after Change-Id I9ef59432f43a3cdb94e4cbb0c44ac3f9b2aac0f2 in
libosmocore, we now enforce that no FSM identifiers contain spaces
or other illegal characters.
Change-Id: I1b44d26cebc4a47094d7b8b3983e5737b88bf003
The library code for rate counter initialization, which is called
from the descendants of bsc_network_alloc() might already want to
log something (particularly after Change-Id
Ifc6ac824f5dae9a848bb4a5d067c64a69eb40b56 in libosmocore), so the
logging framework must be initialized before.
Change-Id: I1e893c97e023e63489fe8c46539b5e507d3cec8f
The test clearly fails unless bts->network is set correctly. Not sure
why this hasn't shown up before?
Change-Id: I47786ed06ff610213d7a0b56d0ebf1c537cd7568
unsigned long can be 32 bits on some arch/OS, while "current" field is
always 64 bit because it's a uint64_t.
Change-Id: Ibad1e4f09cf912cb654dbe3687a3f2182e2060f5
Related: OW#3893
Program terminated with signal SIGSEGV, Segmentation fault.
0 gsm_lchan_name (lchan=lchan@entry=0x0) at gsm_data_shared.c:342
(gdb) bt
0 gsm_lchan_name (lchan=lchan@entry=0x0) at gsm_data_shared.c:342
1 0x0805ab80 in lchan_release (lchan=0x0, sacch_deact=sacch_deact@entry=0, mode=mode@entry=RSL_REL_LOCAL_END)
at chan_alloc.c:410
2 0x0805c1dd in handle_ass_fail (msg=0x94142b8, conn=0x9251048) at bsc_api.c:459
3 dispatch_dtap (msg=0x94142b8, link_id=0 '\000', conn=0x9251048) at bsc_api.c:598
4 gsm0408_rcvmsg (msg=msg@entry=0x94142b8, link_id=0 '\000') at bsc_api.c:658
5 0x08058ca2 in abis_rsl_rx_rll (msg=0x94142b8) at abis_rsl.c:1686
6 abis_rsl_rcvmsg (msg=0x94142b8) at abis_rsl.c:2097
7 0xb7e8cf9a in handle_ts1_read (bfd=0x94e8e08) at input/ipaccess.c:271
8 ipaccess_fd_cb (bfd=0x94e8e08, what=1) at input/ipaccess.c:386
9 0xb7ee8434 in osmo_select_main (polling=polling@entry=0) at select.c:158
10 0x0804bd7c in main (argc=6, argv=0xbfc27144) at osmo_bsc_main.c:272
(gdb) print lchan
$2 = (const struct gsm_lchan *) 0x0
Possible scenario in which this crash can appear:
1- gsm0808_assign_req() calls handle_new_assignment() which sends an CHAN
ACTIVATE msg and arms T10 timer.
2- ACTIVATE ACK is received (handle_chan_ack), which calls
gsm48_send_rr_ass_cmd() which sends an ASSIGNMENT CMD, and doesn't
disable/modify T10 timer.
3- T10 timeout is triggered (assignment_t10_timeout()), which sets
conn->secondary_lchan = NULL
4- Immediately after, the ASSIGNMENT FAILURE message (which might have been
already queued) is processed in handle_ass_fail, and then the crash occurs.
This race condition is not an issue for handle_ass_compl() path because there's
this check there which would trigger most probably if secondary_lchan is NULL:
"if (conn->secondary_lchan != msg->lchan)"
Change-Id: I3798b36c628f75d4e8bc7b0996c27d695d53fbb1
Previously if we ran out of space while adding EARFCN, we simply return
which might result in malformed SI2q. Fix it by proper rollback of
entire EARFCN. While at it, let's be paranoid and introduce extra checks
against integer overflow in budget calculations.
Change-Id: I4b2aa3825e9affb6dfeadecdf24dd1a43a92b7b7
Related: OS#2357
Expose OML link uptime available via vts's "sh bts 0" command with the
new "bts.0.oml-uptime" ctrl command. To avoid code duplication, move
uptime computation into separate function and use it for both.
Change-Id: Iec405aa949d6a38a9c8e64cd7ee4b49fd416835d
Related: OS#2486
It wasn't used anyway because OsmoBTS relied on OpenBSC only. As of
ec33b0397f5d71248c5834513d4be7b9b0e46366 in OsmoBTS it does not need any
shared includes anymore.
Change-Id: Ia689c7f2163dd23e429ee9d17177345b5c9470c7
* fix insert routine to keep the list sorted by UARFCN
* fix rest octets generator to properly account for offset
* adjust test results accordingly
Change-Id: I443c5c5f937b490578354f3c8a0c5b92629f2794
Related: OS#2357
OML link state is available via vty ("sh bts 0" command) and
ctrl ("oml-connection-state" RO variable).
When showing OML link state, take into consideration RSL link state as
well: if OML is up but RSL is missing show it as degraded.
That's implemented via BTS model-specific functions (currently Sysmo- and
Nano- BTS only)
Change-Id: I5952fc59e4d82e0aa627ad91d20f964d9559a4c4
Related: OS#2486
* expand comments, fix typos
* constify parameter
* move try-add-adjust routine into separate function to facilitate
further modifications
* remove excessive checks and unnecessary return values
* move (UARFCN, Scrambling Code) tuple uniqueness check into separate
function and use it early
Change-Id: Ia72f848dec40723510ca56868e08081804227d47
Related: OS#2357
Currently, OSMO_ASSERT() is defined such that it ends in a semicolon, hence an
added ';' is redundant. However, the usual way this kind of macro should be
defined is
#define OSMO_ASSERT(x) do { ... } while(0)
so that the compiler requires a trailing semicolon.
To prepare for such a change possibly coming up in libosmocore, add ';' to all
OSMO_ASSERT() users.
Change-Id: If6dce81faee9177737a6e1b572a871aaf7e37138