Alexander Traud [Wed, 27 Jan 2021 17:42:06 +0000 (18:42 +0100)]
chan_sip: Set up calls without audio (text+video), again.
The previous commit 6d980de fixed this issue in the core of Asterisk.
With that, each channel technology can be used without audio
theoretically. Practically, the channel-technology driver chan_sip
turned out to have an invalid check preventing that. chan_sip tested
whether there is at least one audio format. However, chan_sip has to
test whether there is at least one format. More cannot be tested while
requesting chan_sip because only the [general] capabilities but not the
[peer] caps are known yet. And the [peer] caps might not be a subset or
show any intersection with the [general] caps. This change here fixes
this.
The original commit f04d5fb, thirteen years ago, contained a software
bug as it passed ANY audio capability to the channel-technology driver.
Instead, it should have passed NO audio format. Therefore, this
addressed issue here was not noticed in Asterisk 1.6.x and Asterisk 1.8.
Then, Asterisk 10 changed that from ANY to NO, but nobody reported since
then.
When a Transfer/REFER is executed, TRANSFERSTATUSPROTOCOL variable is
0 when no protocl specific error
SIP example of failure, 3xx-6xx for the SIP error code received
This allows applications to perform actions based on the failure
reason.
ASTERISK-29252 #close Reported-by: Dan Cropp
Change-Id: Ia6a94784b4925628af122409cdd733c9f29abfc4
roadkill [Fri, 22 Jan 2021 13:12:46 +0000 (14:12 +0100)]
res/res_pjsip.c: allow user=phone when number contain *#
if From number contain * or # asterisk will not add user=phone
Currently only number that uses AST_DIGIT_ANYNUM can have "user=phone" but the validation should use AST_DIGIT_ANY
this is a problem when you want to send call to ISUP
as they will disregard the From header and either replace From with anonymous or with p-asserted-identity
ASTERISK-29261
Reported by: Mark Petersen
Tested by: Mark Petersen
Alexander Traud [Thu, 21 Jan 2021 19:28:06 +0000 (20:28 +0100)]
chan_sip: SDP: Reject audio streams correctly.
This completes the fix for ASTERISK_24543. Only when the call is an
outgoing call, consult and append the configured format capabilities
(p->caps). When all audio formats got rejected the negotiated format
capabilities (p->jointcaps) contain no audio formats for incoming
calls. This is required when there are other accepted media streams.
Jaco Kroon [Fri, 8 Jan 2021 16:02:47 +0000 (18:02 +0200)]
AC_HEADER_STDC causes a compile failure with autoconf 2.70
From https://www.mail-archive.com/bug-autoconf@gnu.org/msg04408.html
> ... the long-obsolete AC_HEADER_STDC, previously used internally by
> AC_INCLUDES_DEFAULT, used AC_EGREP_HEADER. The AC_HEADER_STDC macro
> is now a no-op (and is not used at all within Autoconf anymore), so
> that change is likely what made the first use of AC_EGREP_HEADER the
> one inside the if condition, causing the observed results.
The implication is that the test does nothing anyway, and due to it
being a no-op from 2.70 onwards, results in the required not being set
to yes, resulting in ./configure to fail.
Sean Bright [Thu, 14 Jan 2021 14:47:21 +0000 (09:47 -0500)]
res_pjsip_pubsub: Fix truncation of persisted SUBSCRIBE packet
The last argument to ast_copy_string() is the buffer size, not the
number of characters, so we add 1 to avoid stamping out the final \n
in the persisted SUBSCRIBE message.
Ben Ford [Mon, 11 Jan 2021 20:25:09 +0000 (14:25 -0600)]
chan_pjsip.c: Add parameters to frame in indicate.
There are a couple of parameters (datalen and data) that do not get set
in chan_pjsip_indicate which could cause an Invalid message to pop up
for things such as fax. This patch adds them to the frame.
Robert Cripps [Tue, 22 Dec 2020 10:42:35 +0000 (11:42 +0100)]
res/res_pjsip_session.c: Check that media type matches in
function ast_sip_session_media_state_add.
Check ast_media_type matches when a ast_sip_session_media is found
otherwise when transitioning from say image to audio, the wrong
session is returned in the first if statement.
Jean Aunis [Wed, 30 Dec 2020 13:56:47 +0000 (14:56 +0100)]
Stasis/messaging: tech subscriptions conflict with endpoint subscriptions.
When both a tech subscription and an endpoint subscription exist for a given
endpoint, TextMessageReceived events are dispatched to the tech subscription
only.
Alexander Traud [Wed, 23 Dec 2020 14:44:45 +0000 (15:44 +0100)]
chan_sip: SDP: Sidestep stream parsing when its media is disabled.
Previously, chan_sip parsed all known media streams in an SDP offer
like video (and text) even when videosupport=no (and textsupport=no).
This wasted processor power. Furthermore, chan_sip accepted SDP offers,
including no audio but just video (or text) streams although
videosupport=no (or textsupport=no). Finally, chan_sip denied the whole
offer instead of individual streams when they had encryption (SDES-sRTP)
unexpectedly enabled.
Ivan Poddubnyi [Tue, 29 Dec 2020 18:16:00 +0000 (19:16 +0100)]
chan_pjsip: Assign SIPDOMAIN after creating a channel
session->channel doesn't exist until chan_pjsip creates it, so intead of
setting a channel variable every new incoming call sets one and the same
global variable.
This patch moves the code to chan_pjsip so that SIPDOMAIN is set on
a newly created channel, it also removes a misleading reference to
channel->session used to fetch call pickup configuraion.
Ivan Poddubnyi [Thu, 31 Dec 2020 11:53:34 +0000 (12:53 +0100)]
chan_pjsip: Stop queueing control frames twice on outgoing channels
The fix for ASTERISK-27902 made chan_pjsip process SIP responses twice.
This resulted in extra noise in logs (for example, "is making progress"
and "is ringing" get logged twice by app_dial), as well as in noise in
signalling: one incoming 183 Session Progress results in 2 outgoing 183-s.
This change splits the response handler into 2 functions:
- one for updating HANGUPCAUSE, which is still called twice,
- another that does the rest, which is called only once as before.
ASTERISK-28016 Reported-by: Alex Hermann
ASTERISK-28549 Reported-by: Gant Liu
ASTERISK-28185 Reported-by: Julien
Change-Id: I0a1874be5bb5ed12d572d17c7f80de6e5e542940
Jaco Kroon [Fri, 18 Dec 2020 19:06:20 +0000 (21:06 +0200)]
func_lock: fix multiple-channel-grant problems.
Under contention it becomes possible that multiple channels will be told
they successfully obtained the lock, which is a bug. Please refer
ASTERISK-29217
This introduces a couple of changes.
1. Replaces requesters ao2 container with simple counter (we don't
really care who is waiting for the lock, only how many). This is
updated undex ->mutex to prevent memory access races.
2. Correct semantics for ast_cond_timedwait() as described in
pthread_cond_broadcast(3P) is used (multiple threads can be released
on a single _signal()).
3. Module unload races are taken care of and memory properly cleaned
up.
Jaco Kroon [Wed, 23 Dec 2020 17:41:10 +0000 (19:41 +0200)]
pbx_lua: Add LUA_VERSIONS environment variable to ./configure.
On Gentoo it's possible to have multiple lua versions installed, all
with a path of /usr, so it's not possible to use the current --with-lua
option to determisticly pin to a specific version as is required by the
Gentoo PMS standards.
This environment variable allows to lock to specific versions,
unversioned check will be skipped if this variable is supplied.
Kevin Harwell [Wed, 23 Dec 2020 19:06:19 +0000 (13:06 -0600)]
app_mixmonitor: cleanup datastore when monitor thread fails to launch
launch_monitor_thread is responsible for creating and initializing
the mixmonitor, and dependent data structures. There was one off
nominal path after the datastore gets created that triggers when
the channel being monitored is hung up prior to monitor starting
itself.
If this happened the monitor thread would not "launch", and the
mixmonitor object and associated objects are freed, including the
underlying datastore data object. However, the datastore itself was
not removed from the channel, so when the channel eventually gets
destroyed it tries to access the previously freed datastore data
and crashes.
This patch removes and frees datastore object itself from the channel
before freeing the mixmonitor object thus ensuring the channel does
not call it when destroyed.
Dan Cropp [Mon, 7 Dec 2020 22:59:51 +0000 (16:59 -0600)]
chan_pjsip: Incorporate channel reference count into transfer_refer().
Add channel reference count for PJSIP REFER. The call could be terminated
prior to the result of the transfer. In that scenario, when the SUBSCRIBE/NOTIFY
occurred several minutes later, it would attempt to access a session which was
no longer valid. Terminate event subscription if pjsip_xfer_initiate() or
pjsip_xfer_send_request() fails in transfer_refer().
ASTERISK-29201 #close Reported-by: Dan Cropp
Change-Id: I3fd92fd14b4e3844d3d7b0f60fe417a4df5f2435
Kevin Harwell [Tue, 22 Dec 2020 23:40:38 +0000 (17:40 -0600)]
pbx_realtime: wrong type stored on publish of ast_channel_snapshot_type
A prior patch segmented channel snapshots, and changed the underlying
data object type associated with ast_channel_snapshot_type stasis
messages. Prior to Asterisk 18 it was a type ast_channel_snapshot, but
now it type ast_channel_snapshot_update.
When publishing ast_channel_snapshot_type in pbx_realtime the
ast_channel_snapshot was being passed in as the message data
object. When a handler, expecting a data object type of
ast_channel_snapshot_update, dereferenced this value a crash
would occur.
This patch makes it so pbx_realtime now uses the expected type, and
channel snapshot publish method when publishing.
Nick French [Sat, 19 Dec 2020 17:54:50 +0000 (11:54 -0600)]
res_pjsip: Prevent segfault in UDP registration with flow transports
Segfault occurs during outbound UDP registration when all
transport states are being iterated over. The transport object
in the transport is accessed, but flow transports have a NULL
transport object.
Alexander Traud [Tue, 1 Dec 2020 14:11:58 +0000 (15:11 +0100)]
codecs: Remove test-law.
This was dead code, test code introduced with Asterisk 13. This was
found while analyzing ASTERISK_28416 and ASTERISK_29185. This change
partly fixes, not closes those two issues.
Sean Bright [Fri, 11 Dec 2020 19:27:56 +0000 (14:27 -0500)]
app_chanspy: Spyee information missing in ChanSpyStop AMI Event
The documentation in the wiki says there should be spyee-channel
information elements in the ChanSpyStop AMI event.
https://wiki.asterisk.org/wiki/x/Xc5uAg
However, this is not the case in Asterisk <= 16.10.0 Version. We're
using these Spyee* arguments since Asterisk 11.x, so these arguments
vanished in Asterisk 12 or higher.
For maximum compatibility, we still send the ChanSpyStop event even if
we are not able to find any 'Spyee' information.
George Joseph [Thu, 10 Dec 2020 15:09:52 +0000 (08:09 -0700)]
logger.c: Automatically add a newline to formats that don't have one
Scope tracing allows you to not specify a format string or variable,
in which case it just prints the indent, file, function, and line
number. The trace output automatically adds a newline to the end
in this case. If you also have debugging turned on for the module,
a debug message is also printed but the standard log functionality
which prints it doesn't add the newline so you have messages
that don't break correctly.
* format_log_message_ap(), which is the common log
message formatter for all channels, now adds a
newline to the end of format strings that don't
already have a newline.
Pirmin Walthert [Tue, 8 Dec 2020 17:37:13 +0000 (18:37 +0100)]
res_pjsip_nat.c: Create deep copies of strings when appropriate
In rewrite_uri asterisk was not making deep copies of strings when
changing the uri. This was in some cases causing garbage in the route
header and in other cases even crashing asterisk when receiving a
message with a record-route header set. Thanks to Ralf Kubis for
pointing out why this happens. A similar problem was found in
res_pjsip_transport_websocket.c. Pjproject needs as well to be patched
to avoid garbage in CANCEL messages.
Joshua C. Colp [Thu, 29 Oct 2020 17:21:13 +0000 (14:21 -0300)]
pjsip: Match lifetime of INVITE session to our session.
In some circumstances it was possible for an INVITE
session to be destroyed while we were still using it.
This occurred due to the reference on the INVITE session
being released internally as a result of its state
changing to DISCONNECTED.
This change adds a reference to the INVITE session
which is released when our own session is destroyed,
ensuring that the INVITE session remains valid for
the lifetime of our session.
Sean Bright [Sat, 21 Nov 2020 17:51:48 +0000 (12:51 -0500)]
res_http_media_cache.c: Set reasonable number of redirects
By default libcurl does not follow redirects, so we explicitly enable
it by setting CURLOPT_FOLLOWLOCATION. Once that is enabled, libcurl
will follow up to CURLOPT_MAXREDIRS redirects, which by default is
configured to be unlimited.
This patch sets CURLOPT_MAXREDIRS to a more reasonable default (8). If
we determine at some point that this needs to be increased on
configurable it is a trivial change.
lvl [Thu, 29 Oct 2020 11:25:27 +0000 (12:25 +0100)]
Introduce astcachedir, to be used for temporary bucket files
As described in the issue, /tmp is not a suitable location for a
large amount of cached media files, since most distributions make
/tmp a RAM-based tmpfs mount with limited capacity.
I opted for a location that can be configured separately, as opposed
to using a subdirectory of spooldir, given the different storage
profile (transient files vs files that might stay there indefinitely).
This commit just makes the cache directory configurable, and changes
the default location from /tmp to /var/cache/asterisk.
res_pjsip: set Accept-Encoding to identity in OPTIONS response
RFC 3261 says that the Accept-Encoding header should be present
in an options response. Permitted values according to RFC 2616
are only compression algorithms like gzip or the default identity
encoding. Therefore "text/plain" is not a correct value here.
As long as the header is hard coded, it should be set to "identity".
Without this fix an Alcatel OmniPCX periodically logs warnings like
"[sip_acceptIncorrectHeader] Header Accept-Encoding is malformed"
on a SIP Trunk.
Alexander Traud [Wed, 4 Nov 2020 13:39:10 +0000 (14:39 +0100)]
chan_sip: Remove unused sip_socket->port.
12 years ago, with ASTERISK_12115 the last four get/uses of socket.port
vanished. However, the struct member itself and all seven set/uses
remained as dead code.
Boris P. Korzun [Fri, 13 Nov 2020 12:19:30 +0000 (15:19 +0300)]
bridge_basic: Fixed setup of recall channels
Fixed a bug (like a typo) in retransfer_enter() at main/bridge_basic.c:2641.
common_recall_channel_setup() setups common things on the recalled transfer
target, but used same target as source instead trasfered.
George Joseph [Wed, 11 Nov 2020 14:55:22 +0000 (07:55 -0700)]
app_queue: Fix deadlock between update and show queues
Operations that update queues when shared_lastcall is set lock the
queue in question, then have to lock the queues container to find the
other queues with the same member. On the other hand, __queues_show
(which is called by both the CLI and AMI) does the reverse. It locks
the queues container, then iterates over the queues locking each in
turn to display them. This creates a deadlock.
* Moved queue print logic from __queues_show to a separate function
that can be called for a single queue.
* Updated __queues_show so it doesn't need to lock or traverse
the queues container to show a single queue.
* Updated __queues_show to snap a copy of the queues container and iterate
over that instead of locking the queues container and iterating over
it while locked. This prevents us from having to hold both the
container lock and the queue locks at the same time. This also
allows us to sort the queue entries.
George Joseph [Mon, 2 Nov 2020 19:53:02 +0000 (12:53 -0700)]
res_pjsip_outbound_registration.c: Use our own scheduler and other stuff
* Instead of using the pjproject timer heap, we now use our own
pjsip_scheduler. This allows us to more easily debug and allows us to
see times in "pjsip show/list registrations" as well as being able to
see the registrations in "pjsip show scheduled_tasks".
* Added the last registration time, registration interval, and the next
registration time to the CLI output.
* Removed calls to pjsip_regc_info() except where absolutely necessary.
Most of the calls were just to get the server and client URIs for log
messages so we now just save them on the client_state object when we
create it.
* Added log messages where needed and updated most of the existong ones
to include the registration object name at the start of the message.
George Joseph [Mon, 2 Nov 2020 19:53:58 +0000 (12:53 -0700)]
pjsip_scheduler.c: Add type ONESHOT and enhance cli show command
* Added a ONESHOT type that never reschedules.
* Added "like" capability to "pjsip show scheduled_tasks" so you can do
the following:
CLI> pjsip show scheduled_tasks like outreg
PJSIP Scheduled Tasks:
Task Name Interval Times Run ...
============================================= ========= ========= ...
pjsip/outreg/testtrunk-reg-0-00000074 50.000 oneshot ...
pjsip/outreg/voipms-reg-0-00000073 110.000 oneshot ...
* Fixed incorrect display of "Next Start".
* Compacted the displays of times in the CLI.
* Added two new functions (ast_sip_sched_task_get_times2,
ast_sip_sched_task_get_times_by_name2) that retrieve the interval,
next start time, and next run time in addition to the times already
returned by ast_sip_sched_task_get_times().
sched: AST_SCHED_REPLACE_UNREF can lead to use after free of data
The data can be freed if the old object '_data' is the same object as
new 'data'. Because at first the object is unreferenced which can lead
to destroying it.
This could happened in res_pjsip_pubsub when the publication is updated
which could lead to segfault in function publish_expire.
Dovid Bender [Sun, 18 Oct 2020 18:40:10 +0000 (18:40 +0000)]
func_curl.c: Allow user to set what return codes constitute a failure.
Currently any response from res_curl where we get an answer from the
web server, regardless of what the response is (404, 403 etc.) Asterisk
currently treats it as a success. This patch allows you to set which
codes should be considered as a failure by Asterisk. If say we set
failurecodes=404,403 then when using curl in realtime if a server gives
a 404 error Asterisk will try to failover to the next option set in
extconfig.conf
Kevin Harwell [Wed, 4 Nov 2020 21:08:10 +0000 (15:08 -0600)]
AST-2020-001 - res_pjsip: Return dialog locked and referenced
pjproject returns the dialog locked and with a reference. However,
in Asterisk the method that handles this decrements the reference
and removes the lock prior to returning. This makes it possible,
under some circumstances, for another thread to free said dialog
before the thread that created it attempts to use it again. Of
course when the thread that created it tries to use a freed dialog
a crash can occur.
This patch makes it so Asterisk now returns the newly created
dialog both locked, and with an added reference. This allows the
caller to de-reference, and unlock the dialog when it is safe to
do so.
In the case of a new SIP Invite the lock, and reference are now
held for the entirety of the new invite handling process.
Otherwise it's possible for the dialog, or its dependent objects,
like the transaction, to disappear. For example if there is a TCP
transport error.
Ben Ford [Tue, 3 Nov 2020 16:38:34 +0000 (10:38 -0600)]
AST-2020-002 - res_pjsip: Stop sending INVITEs after challenge limit.
If Asterisk sends out and INVITE and receives a challenge with a
different nonce value each time, it will continually send out INVITEs,
even if the call is hung up. The endpoint must be configured for
outbound authentication in order for this to occur. A limit has been set
on outbound INVITEs so that, once reached, Asterisk will stop sending
INVITEs and the transaction will terminate.
Alexander Traud [Fri, 23 Oct 2020 15:26:15 +0000 (17:26 +0200)]
chan_sip: On authentication, pick MD5 for sure.
RFC 8760 added new digest-access-authentication schemes. Testing
revealed that chan_sip does not pick MD5 if several schemes are offered
by the User Agent Server (UAS). This change does not implement any of
the new schemes like SHA-256. This change makes sure, MD5 is picked so
UAS with SHA-2 enabled, like the service www.linphone.org/freesip, can
still be used. This should have worked since day one because SIP/2.0
already envisioned several schemes (see RFC 3261 and its augmented BNF
for 'algorithm' which includes 'token' as third alternative; note: if
'algorithm' was not present, MD5 is still assumed even in RFC 7616).
Walter Doekes [Thu, 4 Jun 2020 14:23:37 +0000 (16:23 +0200)]
main/say: Work around gcc 9 format-truncation false positive
Version: gcc (Ubuntu 9.3.0-10ubuntu2) 9.3.0
Warning:
say.c:2371:24: error: ‘%d’ directive output may be truncated writing
between 1 and 11 bytes into a region of size 10
[-Werror=format-truncation=]
2371 | snprintf(buf, 10, "%d", num);
say.c:2371:23: note: directive argument in the range [-2147483648, 9]
That's not possible though, as the if() starts out checking for (num < 0),
making this Warning a false positive.
(Also replaced some else<TAB>if with else<SP>if while in the vicinity.)
Kevin Harwell [Mon, 19 Oct 2020 20:31:14 +0000 (15:31 -0500)]
res_pjsip, res_pjsip_session: initialize local variables
This patch initializes a couple of local variables to some default values.
Interestingly, in the 'pj_status_t dlg_status' case the value not being
initialized caused memory to grow, and not be recovered, in the off nominal
path (at least on my machine).
Alexander Traud [Fri, 23 Oct 2020 14:55:18 +0000 (16:55 +0200)]
install_prereq: Add GMime 3.0.
Ubuntu 20.10 does not come with GMime 2.6. Ubuntu 16.04 LTS does not
come with GMime 3.0. aptitude ignores any missing package. Therefore,
it installs the correct package(s). However, in Ubuntu 18.04 LTS and
Ubuntu 20.04 LTS, both versions are installed alongside although only
one is really needed.
Alexander Traud [Fri, 23 Oct 2020 14:49:02 +0000 (16:49 +0200)]
BuildSystem: Enable Lua 5.4.
Note to maintainers: Lua 5.4, Lua 5.3, and Lua 5.2 have not been tested
at runtime with pbx_lua. Until then, use the lowest available version
of Lua, if you enabled the module pbx_lua at all.
Nick French [Tue, 13 Oct 2020 17:15:28 +0000 (12:15 -0500)]
res_pjsip_session: Restore calls to ast_sip_message_apply_transport()
Commit 44bb0858cb3ea6a8db8b8d1c7fedcfec341ddf66 ("debugging: Add enough
to choke a mule") accidentally removed calls to
ast_sip_message_apply_transport when it was attempting to just add
debugging code.
The kiss of death was saying that there were no functional changes in
the commit comment.
This makes outbound calls that use the 'flow' transport mechanism fail,
since this call is used to relay headers into the outbound INVITE
requests.
Joshua C. Colp [Tue, 6 Oct 2020 15:32:04 +0000 (12:32 -0300)]
res_pjsip: Adjust outgoing offer call pref.
This changes the outgoing offer call preference
default option to match the behavior of previous
versions of Asterisk.
The additional advanced codec negotiation options
have also been removed from the sample configuration
and marked as reserved for future functionality in
XML documentation.
The codec preference options have also been fixed to
enforce local codec configuration.
Kevin Harwell [Fri, 28 Aug 2020 21:32:37 +0000 (16:32 -0500)]
Logging: Add debug logging categories
Added debug logging categories that allow a user to output debug
information based on a specified category. This lets the user limit,
and filter debug output to data relevant to a particular context,
or topic. For instance the following categories are now available for
debug logging purposes:
These debug categories can be enable/disable via an Asterisk CLI command.
While this overrides, and outputs debug data, core system debugging is
not affected by this patch. Statements still output at their appropriate
debug level. As well backwards compatibility has been maintained with
past debug groups that could be enabled using the CLI (e.g. rtpdebug,
stundebug, etc.).
Sean Bright [Tue, 29 Sep 2020 18:04:48 +0000 (14:04 -0400)]
pbx.c: On error, ast_add_extension2_lockopt should always free 'data'
In the event that the desired extension already exists,
ast_add_extension2_lockopt() will free the 'data' it is passed before
returning an error, so we should not be freeing it ourselves.
Additionally, there were two places where ast_add_extension2_lockopt()
could return an error without also freeing the 'data' pointer, so we
add that.
George Joseph [Thu, 24 Sep 2020 18:46:15 +0000 (12:46 -0600)]
app_confbridge/bridge_softmix: Add ability to force estimated bitrate
app_confbridge now has the ability to set the estimated bitrate on an
SFU bridge. To use it, set a bridge profile's remb_behavior to "force"
and set remb_estimated_bitrate to a rate in bits per second. The
remb_estimated_bitrate parameter is ignored if remb_behavior is something
other than "force".
Stop advertising RFC2833 support on the rtp_engine when DTMF mode is
auto but no tel_event was found inside SDP file.
On an incoming call create_rtp will be called and when session->dtmf is
set to AST_SIP_DTMF_AUTO, the AST_RTP_PROPERTY_DTMF will be set without
looking at the SDP file.
Once get_codecs gets called we move the DTMF mode from RFC2833 to INBAND
but continued to advertise RFC2833 support.
This meant the native_rtp bridge would falsely consider the two channels
as compatible. In addition to changing the DTMF mode we now set or
remove the AST_RTP_PROPERTY_DTMF.
The property is checked in ast_rtp_dtmf_compatible and called by
native_rtp_bridge_compatible.
Arming response to both AST_SIP_SESSION_BEFORE_REDIRECTING and
AST_SIP_SESSION_BEFORE_MEDIA causes 302 to to be handled twice,
resulting in to 181 being generated.
Sean Bright [Fri, 18 Sep 2020 13:09:59 +0000 (09:09 -0400)]
func_curl.c: Prevent crash when using CURLOPT(httpheader)
Because we use shared thread-local cURL instances, we need to ensure
that the state of the cURL instance is correct before each invocation.
In the case of custom headers, we were not resetting cURL's internal
HTTP header pointer which could result in a crash if subsequent
requests do not configure custom headers.
Sean Bright [Fri, 18 Sep 2020 20:02:27 +0000 (16:02 -0400)]
res_musiconhold: Start playlist after initial announcement
Only track our sample offset if we are playing a non-announcement file,
otherwise we will skip that number of samples when we start playing the
first MoH file.
The ast_sip_dialog_get_session function returns the session
with reference count increased. This was not taken into
account and was causing sessions to remain around when they
should not be.
When a channel joins a bridge, we do topology change requests on all
existing channels to add the new participant to them. However the
announcer channel will return an error because it doesn't support
topology in the first place. Unfortunately, there doesn't seem to be a
reliable way to tell if the error is expected or not so the error is
ignored for all channels. If the request fails on a "real" channel,
that channel just won't get the new participant's video.
George Joseph [Fri, 11 Sep 2020 16:09:55 +0000 (10:09 -0600)]
res_pjsip_session: Fix issue with COLP and 491
The recent 491 changes introduced a check to determine if the active
and pending topologies were equal and to suppress the re-invite if they
were. When a re-invite is sent for a COLP-only change, the pending
topology is NULL so that check doesn't happen and the re-invite is
correctly sent. Of course, sending the re-invite sets the pending
topology. If a 491 is received, when we resend the re-invite, the
pending topology is set and since we didn't request a change to the
topology in the first place, pending and active topologies are equal so
the topologies-equal check causes the re-invite to be erroneously
suppressed.
This change checks if the topologies are equal before we run the media
state resolver (which recreates the pending topology) so that when we
do the final topologies-equal check we know if this was a topology
change request. If it wasn't a change request, we don't suppress
the re-invite even though the topologies are equal.
When both Asterisk and a UA send re-invites at the same time, both
send 491 "Transaction in progress" responses to each other and back
off a specified amount of time before retrying. When Asterisk
prepares to send its re-invite, it sets up the session's pending
media state with the new topology it wants, then sends the
re-invite. Unfortunately, when it received the re-invite from the
UA, it partially processed the media in the re-invite and reset
the pending media state before sending the 491 losing the state it
set in its own re-invite.
Asterisk also was not tracking re-invites received while an existing
re-invite was queued resulting in sending stale SDP with missing
or duplicated streams, or no re-invite at all because we erroneously
determined that a re-invite wasn't needed.
There was also an issue in bridge_softmix where we were using a stream
from the wrong topology to determine if a stream was added. This also
caused us to erroneously determine that a re-invite wasn't needed.
Regardless of how the delayed re-invite was triggered, we need to
reconcile the topology that was active at the time the delayed
request was queued, the pending topology of the queued request,
and the topology currently active on the session. To do this we
need a topology resolver AND we need to make stream named unique
so we can accurately tell what a stream has been added or removed
and if we can re-use a slot in the topology.
Summary of changes:
* bridge_softmix:
* We no longer reset the stream name to "removed" in
remove_all_original_streams(). That was causing multiple streams
to have the same name and wrecked the checks for duplicate streams.
* softmix_bridge_stream_sources_update() was checking the old_stream
to see if it had the softmix prefix and not considering the stream
as "new" if it did. If the stream in that slot has something in it
because another re-invite happened, then that slot in old might
have a softmix stream but the same stream in new might actually
be a new one. Now we check the new_stream's name instead of
the old_stream's.
* stream:
* Instead of using plain media type name ("audio", "video", etc) as
the default stream name, we now append the stream position to it
to make it unique. We need to do this so we can distinguish multiple
streams of the same type from each other.
* When we set a stream's state to REMOVED, we no longer reset its
name to "removed" or destroy its metadata. Again, we need to
do this so we can distinguish multiple streams of the same
type from each other.
* res_pjsip_session:
* Added resolve_refresh_media_states() that takes in 3 media states
and creates an up-to-date pending media state that includes the changes
that might have happened while a delayed session refresh was in the
delayed queue.
* Added is_media_state_valid() that checks the consistency of
a media state and returns a true/false value. A valid state has:
* The same number of stream entries as media session entries.
Some media session entries can be NULL however.
* No duplicate streams.
* A valid stream for each non-NULL media session.
* A stream that matches each media session's stream_num
and media type.
* Updated handle_incoming_sdp() to set the stream name to include the
stream position number in the name to make it unique.
* Updated the ast_sip_session_delayed_request structure to include both
the pending and active media states and updated the associated delay
functions to process them.
* Updated sip_session_refresh() to accept both the pending and active
media states that were in effect when the request was originally queued
and to pass them on should the request need to be delayed again.
* Updated sip_session_refresh() to call resolve_refresh_media_states()
and substitute its results for the pending state passed in.
* Updated sip_session_refresh() with additional debugging.
* Updated session_reinvite_on_rx_request() to simply return PJ_FALSE
to pjproject if a transaction is in progress. This stops us from
creating a partial pending media state that would be invalid later on.
* Updated reschedule_reinvite() to clone both the current pending and
active media states and pass them to delay_request() so the resolver
can tell what the original intention of the re-invite was.
Sungtae Kim [Mon, 31 Aug 2020 12:21:09 +0000 (14:21 +0200)]
realtime: Increased reg_server character size
Currently, the ps_contacts table's reg_server column in realtime database type is varchar(20).
This is fine for normal cases, but if the hostname is longer than 20, it returns error and then
failed to register the contact address of the peer.
Normally, 20 characters limitation for the hostname is fine, but with the cloud env.
So, increased the size to 255.
projects / thirdparty / asterisk.git / log
