In gsm04_08_clear_request(), in_release == 1 anyway and
msc_release_connection() would exit immediately without any effect. Don't
confuse the reader by passing release=1 arg.
Change-Id: I5bf9eb4889d32ad5e42ac7d096bf62fa3a493e20
Reviewed-on: https://gerrit.osmocom.org/93
Reviewed-by: Holger Freyther <holger@freyther.de>
Tested-by: Jenkins Builder
subscr_name() was called from several places:
* either without a check for subscr being NULL, which for example
was causing a segfault if we hand-over a channel before identifying the
subscriber
* or with an explicit NULL check and the ternary operator (?).
We now simplify the code by checking for the NULL Subscriber in subscr_name()
itself.
Change-Id: Ide09f4a515222eb2ec6c25e7a6a8c5f6cc2ffd4b
Reviewed-on: https://gerrit.osmocom.org/92
Tested-by: Jenkins Builder
Reviewed-by: Holger Freyther <holger@freyther.de>
In the past, normal migration was possible only if the actual
schema version differed from the version used in DB by 1. For
example, if DB uses an old version 3 and you need to use it
with the code written for version 5, the check_db_revision()
will convert it to 4 and DB will still use incompatible schema
version during Osmo-NITB running time. After next run it will
be converted to version 5.
This patch replaces a set of 'else-if' checks by a 'switch'
without 'break' statements between 'case' labels (waterfall).
It makes you able to migrate from current version to the
latest despite any difference between them.
Change-Id: Ia9c2aa86f96b88ad8a710d0a23879ce219bc82dc
Reviewed-on: https://gerrit.osmocom.org/62
Tested-by: Jenkins Builder
Reviewed-by: Holger Freyther <holger@freyther.de>
Rather than having a 'private' structure for kc, sres and rand, we
now finally (with 4 years delay) use osmo_auth_vector from libosmogsm,
which encapsulates authentication vectors that can be either GSM
triplets or UMTS quintuples or a combination of both.
gsm_auth_tuple becomes a wrapper around osmo_auth_vector, adding
use_count and key_seq to it.
key_seq is no longer initialized inside gprs_gsup_messages.c, as there
is no CKSN / key_seq inside the message anyway. If a usre of the code
needs key_seq, they need to manage it themselves.
In case both TCH/H and TCH/F or different codecs are configured and
internal MNCC handler is used we might end up in a situation where call
legs with incompatible channel types or codecs would be connected
resulting in a broken audio. Disconnect such calls with appropriate
error message.
Fixes: OS#1663
Extend the existing ctrl command to be able to specify the
algorithm and Ki. In contrast to the VTY no size check is
done. Together with the VTY this code only supports a small
part of what is supported by libosmocore.
The algorithm and ki are considered optional but if a valid
algorithm other than "none" is passed, a KI must be passed as
well.
Extend the test coverage by passing the potential values. It
is not verified that the KI/algorithm is stored.
We should not return a subscriber in case it was not written to
the database. Instead free the memory allocated and return NULL.
Callers in gsm_04_08.c are prepared to have the creation fail.
Related: OS Issue #1657
The issue of db_create_subscriber updating an already existing subscr
is that the same subscriber will then have two entries in the active
subscribers list. In general this will break assumptions that a subscr
can be compared by comparing the pointer.
In the case of the VTY this was not an issue as the created subscr
was immediately destroyed again but it is better to avoid this problem.
Change the VTY command to find the subscriber and then call sync to
have the updated time set. The side-effect is we will now have two
queries for the subscriber. Once through subscr_get_by_imsi and once
through db_create_subscriber.
Change the db_create_subscriber to fail if a subscriber already exists,
and add a testcase for this behavior and do not updated the 'updated'
timestamp of an already existing subscriber.
Add a testcase for this behavior.
Related: OS Issue #1657
Make sure a new auth tuple is initialized after
db_get_lastauthtuple_for_subscr() returns an error, i.e. if no tuple is present
for the subscriber yet.
Before this patch, the first key_seq depended on the typically uninitialized
value that was present in auth tuple's key_seq upon calling
auth_get_tuple_for_subscr().
The very first key_seq used for a new subscriber will now always be 0. Before,
it used to be mostly 1 ("(0 + 1) % 7"), but depended on whether the key_seq was
indeed initialized with 0, actually by random.
In auth_get_tuple_for_subscr(), add missing condition to match incoming
key_seq with stored key_seq, so that re-authentication is requested for
mismatching key_seqs.
Add test for this issue.
Instead of using hardcoded -1 for errors, include -1 in the enum auth_action
type; apply its use.
In the mm_auth test, the string output changes from '(internal error)' to
'AUTH_ERROR', since now the proper enum value is used in auth_action_names[].
Add basic MM Authentication test setup, with fake DB access and RAND_bytes().
So far implement simple tests for IO error during DB access and missing auth
entry.
To print the auth action during tests, add struct auth_action_names and
auth_action_str() inline function in auth.[hc].
In OpenBSC, we traditionally displayed a TMSI in its integer
representation, which is quite unusual in the telecom world. A TMSI is
normally printed as a series of 8 hex digits.
This patch aligns OpenBSC with the telecom industry standard.
Signed-off-by: Vadim Yanitskiy <axilirator@gmail.com>
libosmocore recently added inline functions to relieve callers from applying
bitmasks and bit shifts to access the transaction id of a GSM 04.08 header.
Apply these functions.
Replace hardcoded protocol discriminator and message type bitmasks with
function calls recently introduced in libosmocore.
Note that the release 98 bitmasks slightly differ from the release 99 bitmasks.
This patch uses the "default" gsm48_hdr_msg_type invocation, thus it depends on
libosmocore whether 98 or 99 bitmasks are used.
In some places, use of the bitmask was erratic. Fix these implicitly by
employing the bitmask functions:
* silent_call.c: silent_call_reroute(): add missing bitmask for MM.
* bsc_msg_filter.c: bsc_msg_filter_initial(): RR vs. MM messages.
* osmo_bsc_filter.c: bsc_find_msc() and bsc_scan_bts_msg(): RR vs. MM
messages.
* bsc_nat_rewrite.c: bsc_nat_rewrite_msg(): SMS vs. CC messages.
* bsc_ussd.c: no bitmask is applicable for the message types used here.
* gb_proxy.c: gbproxy_imsi_acquisition(): missing bit mask for pdisc.
In gprs_gb_parse.c: gprs_gb_parse_dtap(), add a log notice for unexpected
message types.
Make the SMPP bind address configurable (used to be harcoded as "0.0.0.0").
Add VTY command
smpp
local-tcp A.B.C.D <1-65535>
while keeping the old command 'local-tcp-port <1-65535>'. Both the old and the
new command immediately change the SMPP listening address and port.
Add a LOGL_NOTICE log when the SMPP listening address and/or port change.
However, to be useful, this patch has to go somewhat further: refactor the
initialization procedure, because it was impossible to run the VTY commands
without an already established connection.
The SMPP initialization procedure was weird. It would first open a connection
on the default port, and a subsequent VTY port reconfiguration while reading
the config file would try to re-establish a connection on a different port. If
that failed, smpp would switch back to the default port instead of failing the
program launch as the user would expect. If anything else ran on port 2775,
SMPP would thus refuse to launch despite the config file having a different
port: the first bind would always happen on 0.0.0.0:2775. Change that.
In the VTY commands, merely store address and port if no fd is established yet.
Introduce several SMPP initialization stages:
* allocate struct and initialize pointers,
* then read config file without immediately starting to listen,
* and once the main program is ready, start listening.
After that, the VTY command behaves as before: try to re-establish the old
connection if the newly supplied address and port don't work out. I'm not
actually sure why this switch-back behavior is needed, but fair enough.
In detail, replace the function
smpp_smsc_init()
with the various steps
smpp_smsc_alloc_init() -- prepare struct for VTY commands
smpp_smsc_conf() -- set addr an port only, for reading the config file
smpp_smsc_start() -- establish a first connection, for main()
smpp_smsc_restart() -- switch running connection, for telnet VTY
smpp_smsc_stop() -- tear down connection, used by _start() twice
And replace
smpp_openbsc_init()
smpp_openbsc_set_net()
with
smpp_openbsc_alloc_init()
smpp_openbsc_start()
I'd have picked function names like "_bind"/"_unbind", but in the SMPP protocol
there is also a bind/unbind process, so instead I chose the names "_start",
"_restart" and "_stop".
The smsc struct used to be talloc'd outside of smpp_smsc_init(). Since the smsc
code internally uses talloc anyway and employs the smsc struct as talloc
context, I decided to enforce talloc allocation within smpp_smsc_alloc_init().
Be stricter about osmo_signal_register_handler() return codes.
This patch lets the build script for openbsc find the libsmpp34 installation
with the help of pkg-config instead of assuming the header files are in
/usr/include.
memcpy has both the source and destination marked as non-null and
we were still passing NULL (with a zero size) to it. While this
makes sense it violates the constraints of the function. Add the
check to see if these values are NULL or not.
+db.c:583:2: runtime error: null pointer passed as argument 2, which is declared to never be null
+ #0 0x40d7f7 in get_equipment_by_subscr (/home/builder/jenkins/workspace/Osmocom_Sanitizer/source/openbsc/openbsc/tests/db/db_test+0x40d7f7)
+ #1 0x40f6d2 in db_get_subscriber (/home/builder/jenkins/workspace/Osmocom_Sanitizer/source/openbsc/openbsc/tests/db/db_test+0x40f6d2)
+ #2 0x40bfaa in sms_from_result_v3 (/home/builder/jenkins/workspace/Osmocom_Sanitizer/source/openbsc/openbsc/tests/db/db_test+0x40bfaa)
+ #3 0x40c847 in update_db_revision_3 (/home/builder/jenkins/workspace/Osmocom_Sanitizer/source/openbsc/openbsc/tests/db/db_test+0x40c847)
+ #4 0x40cbc3 in check_db_revision (/home/builder/jenkins/workspace/Osmocom_Sanitizer/source/openbsc/openbsc/tests/db/db_test+0x40cbc3)
+ #5 0x40cf85 in db_prepare (/home/builder/jenkins/workspace/Osmocom_Sanitizer/source/openbsc/openbsc/tests/db/db_test+0x40cf85)
+ #6 0x406f18 in main /home/builder/jenkins/workspace/Osmocom_Sanitizer/source/openbsc/openbsc/tests/db/db_test.c:179
+ #7 0x7fd625638a3f in __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x20a3f)
+ #8 0x405598 in _start (/home/builder/jenkins/workspace/Osmocom_Sanitizer/source/openbsc/openbsc/tests/db/db_test+0x405598)
+
+db.c:590:2: runtime error: null pointer passed as argument 2, which is declared to never be null
+ #0 0x40da23 in get_equipment_by_subscr (/home/builder/jenkins/workspace/Osmocom_Sanitizer/source/openbsc/openbsc/tests/db/db_test+0x40da23)
+ #1 0x40f6d2 in db_get_subscriber (/home/builder/jenkins/workspace/Osmocom_Sanitizer/source/openbsc/openbsc/tests/db/db_test+0x40f6d2)
+ #2 0x40bfaa in sms_from_result_v3 (/home/builder/jenkins/workspace/Osmocom_Sanitizer/source/openbsc/openbsc/tests/db/db_test+0x40bfaa)
+ #3 0x40c847 in update_db_revision_3 (/home/builder/jenkins/workspace/Osmocom_Sanitizer/source/openbsc/openbsc/tests/db/db_test+0x40c847)
+ #4 0x40cbc3 in check_db_revision (/home/builder/jenkins/workspace/Osmocom_Sanitizer/source/openbsc/openbsc/tests/db/db_test+0x40cbc3)
+ #5 0x40cf85 in db_prepare (/home/builder/jenkins/workspace/Osmocom_Sanitizer/source/openbsc/openbsc/tests/db/db_test+0x40cf85)
+ #6 0x406f18 in main /home/builder/jenkins/workspace/Osmocom_Sanitizer/source/openbsc/openbsc/tests/db/db_test.c:179
+ #7 0x7fd625638a3f in __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x20a3f)
+ #8 0x405598 in _start (/home/builder/jenkins/workspace/Osmocom_Sanitizer/source/openbsc/openbsc/tests/db/db_test+0x405598)
When a MNCC handler wants to issue the MNCC_BRIDGE primitive
overt the MNCC interface, this was not possible so far via the
MNCC socket. This primitive was so far only available from the
internal MNCC handler, more or less by accident I suppose. The reason
for this is in the way the array of two call references had been passed
into mncc_tx_to_cc().
In case of the RTP bridge mode we need to select the codec
ourselves. Rely on the same (incomplete) codec selection that
can be done using the mncc-int configuration node. This might
gain bearer capabilities support.
In case of a SDCCH a TCH/F will be attempted to be assigned.
This is an open issue for both modes and there should be a
preference for full or half-rate channels somewhere.
Implement sending MDCX on the newly allocated channel and send
the data to the same destination as the currently connected one.
This way the receiver can implement RTP RFC Appendix A.1 and
deal with the new source.
For the LCR rtp-bridge audio should directly flow to the
remote system. In contrast to the original patch audio
will now flow directly from the BTS to the remote system.
This assumes that BTS and the remote system are in the
same network segment and can directly communicate.
There are various limitations in the first iteration of
the implementation:
We could (and in the future) should delay the assignment
but currently we are forced to pick the channel and move
it to the audio state. In case we are located on a SDCCH
we always need to change but if we are on a TCH we could
send the ipa.CRCX and change the audio state a lot later.
The net effect is that the audio codec selection needs to
be done in the NITB code and not in the system connected
to it.
This only works with ip based systems. For E1 systems one
could still use the RTP socket or even try to move this
out of the process.
There is no code for handover handling and it relies on
the remote system dealing with the SSRC change of the
system.
This adds the protocol definition for the RTP bridge extension
of Andreas Eversberg and bumps the protocol version.
I added the missing mncc mappings from value to string.
[ 5cf8fb10ea3addcae74d37f4dbf1c1be664df53e protocol extension
5dac90de38990b188f499c602bf18a4f232070e8 payload extension]
I guess none of our users knows what a mi_type=0x02 is, but most would
know what an IMSI or a TMSI is. So let's use the newly introduced
gsm48_mi_type_name() function to fix this.
Traffic cannot sent to BTS, if there is (currently) no logical channel
associated with the transaction.
This happens, if TCH traffic is received from upper layer, but there is
no lchan available before completing immediate assignment, handover or
assignment process.
[hfreyther: The code has not been moved to tch_frame_down
but the issue looks similiar]
Using the talloc leak report we see that there are some msgb's
that are allocated for SMS but we don't have transactions or
SMS around. We need to improve the name of the messages to
uniquely dscribe where they are from but the obvious leak does
occur in this routine.
The no available transaction id is most likely the case where
we leak memory. This should not occur and shows another issue
with the smsqueue/smpp handling. It doesn't explain the subscr
reference count issue either.
Extract of the leak report:
GSM 04.11 contains 1160 bytes in 1 blocks (ref 0) 0x2517dc0
GSM 04.11 contains 1160 bytes in 1 blocks (ref 0) 0x24b56e0
GSM 04.11 contains 1160 bytes in 1 blocks (ref 0) 0x23e7930
In case the subscriber is currently busy we would omit the
subscr_put. This seems to be very hard to hit as the subscr
need to be active and at the same time be selected for the
purge operation.
Coordinate with the normal subscriber channel requests instead
of going to page ourselves. This might lead to getting a channel
that is of a different type though.
vty_interface_layer3.c:584:4: warning: format '%d' expects argument of type 'int', but argument 3 has type 'long unsigned int' [-Wformat=]
sizeof(subscr->extension)-1, VTY_NEWLINE);
Neels Hofmeyr
Harald Welte
Max
Daniel Willmann