The function setup_rtp_processing() in mgcp_protocol.c executes a
function pointer setup_rtp_processing_cb(). The function pointer
gets two struct mgcp_rtp_end pointers as parameter. To get those
parameters it has to dereference them from struct mgcp_conn_rtp
pointers. The variable conn_src is such a struct pointer and there
are conditions where this pointer may be NULL. The function at the
function pointer should get the conn pointers directly instead of
the dereferenced end (rtp) pointers. This also gives additional
flexibility to the implementation behind the function pointer,
which is not yet defined (the function pointer points always to
a stub function since we donot support transcoding yet.
- give conn pointers directly to setup_rtp_processing_cb() insed
of dereferencing conn_src->end
Change-Id: Id46e9bfba88613387026639eb4957221cce6820a
Closes OS#3406
When creating the mgcp statistics (DLCX) and also when printing
values in the VTY. The printf placeholder %lu is used. However,
this is not portable when the same code is compiled on a machine
with different integer size (e.g. armv7).
- Use PRIu64 when printing ->current value of the rate counters
Change-Id: Ifb8944cec83868845f74ad84551eb090f812daf8
In struct mgcp_rtp_end one finds unsigned int counters. Those should
be replaced with libosmocore rate counters
- replace packets_rx, octets_rx, packets_tx, octets_tx and
dropped_packets with libosmocore rate counters.
Change-Id: I47c5c9006df5044e59ddebb895e62adb849d72d5
Related: OS#2517
The codec negotiation via SDP is currently in a neglected state. Also
osmo-mgw does some kind of codec decision wile the SDP is parsed, the
result is information for one codec, even when there are multiple codecs
negotiated. This is problematic because we loose all information about
alternate codecs while we parse. This should be untangled and the
information should be presevered. Also we are not really capable
picking a default. Wehen we do not supply any codec information (not
even LCO), then we should pick a sane default codec.
- separate the codec decision from the sdp parser and concentrate
codec related code in a separate c file
- add support for multiple codecs in one SDP negotiation
- do not initalize "magic" codec defaults during conn allocation
- do not allow invalid payload types, especially not 255. When
someone tries to select an invalid payload type, do not fail
hard, just pick a sane default.
- handle the codec decision in protocol.c, pick a sane default
codec when no (valid) codec has been negotiated (no LCO, no SDP)
Change-Id: If730d022ba6bdb217ad4e20b3fbbd1114dbb4b8f
Closes: OS#2658
Related: OS#3114
Related: OS#2728
At the moment osmo-mgw will accept multiple lco options. (e.g.
p:10, a:PCMU, p:10) If LCO appear multiple times, than the first
appearance of will be parsed and used, all following appearances
will be ignored. However, having multiple appearances of LCO is
illegal and affected requests should be rejected. Also osmo-mgw
should reject illegal formatted LCO strings
- make sure that multiple appearances of LCOs will be rejected
- make sure that illegal formated LCOs are rejected
- add testcases with garbeled LCO and valid LCO examples
Change-Id: Iae2fddfa5f2bcfc952f8ab217b3056694e5f7812
Closes: OS#3119
In the current implementation the LCO parameters are reset. This means
that an MDCX without LCO will reset the LCO that have previously set
via CRCX. But according to RFC 3435 6.8 LocalConnectionOptions, the
LCO parameters should be preserved or left at their defaults if missing.
- Make sure LCO are retained if no LCO string is present.
- Also preserve the values of individual parameters if missing.
Change-Id: Ia0d73f61516618317dcd1d49384818fd8de27aa6
in mgcp_rtp_codec_init() tallo_free is called after codec->subtype_name
and codec->audio_name are set to NULL. So talloc_free() always sees
NULL-pointers and never frees anything. This may cause a memory leak.
- call talloc_free() first, then set pointers to NULL
Change-Id: I7373819c3689d34811846f6f48f27568297b26e4
"unable to create connection resource error" sounds a bit strange.
Lets just output "unable to create connection".
Change-Id: Ibef16b455f2e122c8e5ff95240c4d7a654c56a39
It is legal to create connection without setting the destination
ip and port (this usually done later through MDCX). However, if
some other connection tries to deliver an RTP packet through a
a half open connection, then the fact that no destination ip is
set is logged as error even if it is a pretty normal situation.
- Check if destination ip and port are set to zero. If yes, we
assume that the destination connection details are intentionally
not set yet. Only when one value is set and the other one not,
we log an error. Otherweise we log a message to debug.
Change-Id: If96e5a60b8ab92259d3bddaa143121893bb6c525
Related OS#3104
The function allocate_port tryes at least 200 different ports when
a new port is allocated. Since after every allocation the port
number is incremented the allocation should be able to allocate
a port with the first attempt. However, the number 200 is an
arbitrary number and it will not cover the whole port range in
most cases.
- Make sure that in the worst case at each port in the range
is tryed once, not only the next 200
Change-Id: Ic47f09869eaddd4aea817bb2517362883d65d029
Related: OS#2825
The alt_codec field is not used anywhere in the code
- remove unused alt_codec field
Change-Id: I5ff2899e3e61f33eb86f284b50ad8a94a949ed16
Related: OS#3114
The VTY command that sets the RTP port range does not check if the data
entered by the user actually makes sens. Also it allwos to configur a
range that starts at 0.
- Make sure 0 can not be used as start or end of the range
- make sure the end port number is always greater then the begin
port number
- Autocorrect uneaven port range beginnings to one port number before to
ensure the range starts at an even port number
- Autocorrect even port range ends to the next odd port number to
ensure the range ends at an odd port number.
Change-Id: Ib1312acba4f03f378594dbbeb4f31afd891d68d7
Related: OS#2825
This commit actually doesn't fix the entire code, since anyway osmux
conns are not supported and mgcp_conn_get_rtp() will return NULL.
However, it makes the code more logical and easier to understand once
somebody refactors the code to make it work again.
Change-Id: Ib57e12e5a36b5842c40673c236907bbcbfc390f3
The two counters: in_stream.err_ts_counter and out_stream.err_ts_counter
are still handcoded. To make them better accessible they should
be replaced with libosmocore rate counters.
- replace state.in_stream.err_ts_counter with libosmocore rate counter
- replace state.out_stream.err_ts_counter with libosmocore rate counter
Change-Id: I9fbd65bf2f4d1e015a05996db4c1f7ff20be2c95
Related: OS#2517
The struct state->out_stream.ssrc is initalized by first initalizing
state->in_stream and then copying state->in_stream over to
state->out_stream. This works as long as no pointers to other objects
are added to struct mgcp_rtp_stream_state but we may add pointers to
struct mgcp_rtp_stream_state in the future.
- Initalize out_stream and in_stream independently from each other
Change-Id: I5deb27e609448ee0b9f7034e644ae96f1e57887a
Related: OS#2517
The two counters: in_stream.err_ts_counter and out_stream.err_ts_counter
are still handcoded. To make them better accessible they should
be replaced with libosmocore rate counters.
- replace state.in_stream.err_ts_counter with libosmocore rate counter
- replace state.out_stream.err_ts_counter with libosmocore rate counter
Change-Id: I67aa7a8602f60366ef3ba2c5b1319b1b85719f64
Related: OS#2517
The function mgcp_conn_free() holds a few lines to de-initalize
members which are struct mgcp_conn_rtp specific. Since we already
have an mgcp_rtp_conn_init() that does the intialization, we should
have an mgcp_rtp_conn_cleanup() too.
- add function mgcp_rtp_conn_cleanup() and move rtp specific
code to that function.
Change-Id: Ib9bf6d2a3af4f1df1a4ab5ec789b39a2cee2532f
The function mgcp_rtp_codec_reset() is soley called from
mgcp_rtp_conn_init(), lets change the prefix here to _init too.
- rename mgcp_rtp_conn_reset() to mgcp_rtp_conn_init()
Change-Id: I246aabc896089c1f2b3d0409ec3422a85e43575c
The function mgcp_conn_alloc() calls mgcp_rtp_conn_init() to initalize
the RTP specific members (union u.rtp) but also touches u.rtp directly.
This should not be the case, only mgcp_rtp_conn_init() may touch the
union depending on which type of RTP connection is initialized
(currently there is only MGCP_CONN_TYPE_RTP).
- let mgcp_rtp_conn_init() set the backpointer to the generic
conn part.
Change-Id: I6f806f9bfa71b446c15bdc34ae59d2bc1cd10d7e
Note: This is merely a cosmetic change.
We do allocate connections dynamically and we initialize them
once by calling mgcp_rtp_conn_reset(). Calling this a reset
function implies that the reset happens multiple times while
the struct lives. This is not tha case, so lets change the
suffix to _init()
- rename mgcp_rtp_conn_reset() to mgcp_rtp_conn_init()
Change-Id: Ie48b575ff81c8f48afcc25f485967e011e90027b
Since we will support multiple different types of endpoints in the
future, all these endpoints will handle connections slightly different
and there will be possibly state that needs to be kept consistant
when a connection is deleted.
In mgcp_network.c where we implement the callback that is used to
create an rtp-bride-endpoint. In that callback we cache the pointer
of the connection we where we want to bride to (opposite connection).
When one of the connections is deleted using a DLCX operation, the
pointer is still there and the next incoming packet causes a use-
after-free segfault.
- introduce an endpoint specific callback function that is executed
before removing the connection.
- implement the endpoint specific callback for rtp bridge endpoints,
so that the use-after-free is prevented.
Change-Id: I921d9bbe58be1c3298e164a37f3c974880b3759f
When a wildcarded request is made with a DLCX or MDCX command
the MGW will search for a free endpoint and continues the command
execution with that endpoint.
- Catch the wildcarded request early on DLCX and MDCX and return
with an error code.
See also TTCN3 testcases:
MGCP_Test.TC_mdcx_wildcarded
MGCP_Test.TC_dlcx_wildcarded
Change-Id: Ia77d44a6a86083e62338e5845b553e5cf13ebd10
This is an internal library simmilar to 'libmsc' in osmo-msc, which
we don't expect to be used by other programs except osmo-mgw. Hence,
there's no need to install it as a shared library, which introduces
requirements about ABI/API stability and the like.
osmo-bsc_nat uses libosmo-legacy-mgcp, and once we should rewrite
osmo-bsc_nat, we might need some of the libosmo-mgcp related functions,
but at this point it's unclear what exactly would be needed and if
current libosmo-mgcp can provide that. As needed, we can introduce
a related shared library at that point.
Change-Id: Iba0a2c9c694e360356ac2ca584e97795281c6198
When a wildcarded CRCX is done flag "wildcarded_crcx" is set in the
endpoint struct. The flag tells other part of the code whether the
request was wildcarded or not since in some cases the behaviour
might be different for wildcarded requests. The implementation of
this mechanism is not entirely correct. The flag is set on wildcarded
requests but on non wildcarded requests it is not reset. Also the
name is misleading.
- rename wildcarded_crcx to wildcarded_req
- ensure the flag is refreshed with every new request
Change-Id: Ia5063ec65f5bc3a8a0943d1fd823aaeee20b8637
The connection mode setting (e.g. recvonly) is not checked on CRCX
and MDCX. This allows requests that set the connection mode to
sendrecv or sendonly without ever configuring the remote end of
the connection (half-open connection).
- reject sendrecv or sendonly on half open connections
See also TTCN3 Test:
MGCP_Test.TC_crcx_early_bidir_mode
Change-Id: I6ab01855d3b1faa36aebac357e7b97c563990678
Related: OS#2652
The final log lone in find_endpoint() lacks the \n causing a messed
up log output.
- Add missing \n
Change-Id: I97fca654b199dfb7aae2359322a56c6d0bae9599
When set_local_cx_options() returns an error code the MGCP command
execution is aborted and and the error code is returned, but on
CRCX the already seized eindpoint is not released.
- Do not generate the error response on the spot, jump to the
respective label and let the already existing error handling
do its work.
This patch is a follow-up page to:
Change-Id I02aaa3042f2a0e32eb4ec6b8753deab7082947a0
Change-Id: Iaef4ea6c6a2f24ac8b276966bda72d0b30f25cd5
Related: OS#2654
When an unsupported MGCP parameter (e.g. N) is used, then this
parameter is ignored and the command execution continues. However,
an MGCP command that contains an unsupported parameter should
be rejected.
- Make sure that MGCP commands DLCX, CRCX and MDCX are rejected,
when they contain unsupported parameters.
Change-Id: I8cd5987fc6befcd53a7c4916f77b1a24c904ba48
The short term of endpoint has always been "endp" througout the whole
project and not "ep".
- rename mcgp_ep.c to mgcp_endp.c
- rename mgcp_ep.h to mgcp_endp.h
Change-Id: Id52047bb2d0407655ac272c858ed3412b8ae9e6d
In order to allow clean prefixes for future endpoint related
functions the "rlease" should be moved to the end of the
function name.
- rename mgcp_release_endp to mgcp_endp_release
Change-Id: I22e938e702f57ad76d38c9f4a1370b110ac1ba11
The endpoint and the define that computes the endpoint number is
defined in mgcp_internal.h. Since we have a dedicated module for
endpoint related code it makes sense to move the endpoint related
parts there.
- move struct mgcp_endpoint to mgcp_ep.h
- move #define ENDPOINT_NUMBER(endp) to mgcp_ep.h
Change-Id: Ibae55e1859bd41e2d1918eda433418b4bf8365fe
When all endpoints are seized and a call agent tries to allocate
another one, then 500 is returned as response code, which means
that the endpoint was not found. This does not match.
- Return 403 which is defined as "Insufficient resources available
at this time"
Change-Id: Idf241b47e711e35e9f9b5a43f3cea5c0298ea30b
When a wildcarded CRCX is done, then the endpoint number is
returned as unsigned integer (%u). This results into problems
with endpoint numbers higher than 9.
- Return endpoint identifier with the endpoint number in
hexadecimal representation
Change-Id: I504f4658c193009347753b15256dbb46b32ad5a4
When the local connection options in an MDCX or CRCX request
are parsed, then the packetization interval is not checked.
- Check if the packetization is a multiple of 20ms
see also TTCN3 test: MGCP_Test.TC_crcx_unsupp_packet_intv
Change-Id: I02aaa3042f2a0e32eb4ec6b8753deab7082947a0
Related: OS#2654
Some of the line breaks lack the \r character, which leads to an
inconsistancy. While our software and even wireshark does ignore
the problem, other third party implementations might reject those
messages.
- Add the missing \r characters to make the message format
consistant.
Change-Id: I0cd80afae65accd3b4ddc5d82e5d30385879141c
The SDP parameter block must be detached from the regular parameters
using an additional line break (empty line). At the moment this works
because the empty OSMOX variable is added and by this also adds a
line break. It breaks as soon as OSMUX is used again.
- Make clear that no OSMUX variable is added when OSMUX is not in
use.
- Add the extra line break independently
Change-Id: I6261971040db527b96fe79676520ccd7794bd327
the virtual trunk is addressed without a prefix (just *@domain).
- reorganize find_endpoint() so that it accepts a prefix when
addressing the virtual trunk.
- do no longer accept wildcarded CRCX requests without prefix
(will not break anything, the feature of wildcarded CRCX is
not in use yet)
- keep the old prefix-less method but print a warning that it is
depreacted.
Change-Id: I2aac3ba0f1f3122dfbb3bf36f74198ecb2b21de5
The mgcp protocol in general allows wildcarded endpoints on CRCX.
osmo-mgw does not support this feature yet.
- when the endpoint name contains a wildcard character, search
a free endpoint automatically
- return the resulting endpoint name in the parameter section of
the mgcp response
- add parsing support for the returned endpoint names
- Be more concious about the parameters that are returned with
each response. Do not unnecessarily attach known parameters.
Return the connection ID only on CRCX commands. Only return
the endpoint ID on CRCX commands that are wildcarded.
Change-Id: Iebc95043569191b6f5fbc8fe266b13fcfcab2e48
related: OS#2631
Adding a NUL manually is a common idiom after calling strncpy() because
strncpy() does not always NUL-terminate the string. But snprintf() is
fine.
- remove NUL termination after snprintf in mgcp_send_reset_ep()
Change-Id: I5a1187b13b21b11674f13d3449c730616b0a4ddf
At the moment the MGW has a fixed domain name string that is not even
checked properly.
- Make domain name configurable, use the current "mgw" string as
defualt to maintain compatibility
- Check the domain name with each request. If the endpoint contains
an unexpected domain name, the request must be rejected.
Change-Id: Ia91ac428ba83ac1f9b52a0ec8dbf00ef7876da9e
there is a now obsolete constraint that endpoint numbers must
start at 1.
- remove the check to allow also endpoints starting at 0
Change-Id: Iec2f4e36e1ab01ff23875d99e4b0e04af7c1ad98
There's no need for us to use the sockets API directly: We have
pretty nice socket helper functions in libosmocore, let's make
use of them.
Change-Id: I39d47b8a27f683060a2facf2dbecff8d00c19ce9
Previously we
* did not distinguish between the cause of errors in mgcp_header_parse
* common errors were not handled in mgcp_handle_message() but in
individual per-verb functions
Let's centralize the handling of generating error responses and remove
that burden from the individual per-verb handler functions.
Change-Id: I463b27306e10ae3b021583ed102977e7299e5e66
Philipp Maier
Pau Espin
Alexander Couzens
Harald Welte