21 months agoMerge "bridge_softmix / app_confbridge: Add support for REMB combining."
George Joseph [Wed, 18 Apr 2018 20:37:45 +0000 (15:37 -0500)]
Merge "bridge_softmix / app_confbridge: Add support for REMB combining."

21 months agoMerge "utils/pval: Add -lBlocksRuntime for compiler clang conditionally."
Jenkins2 [Wed, 18 Apr 2018 20:24:24 +0000 (15:24 -0500)]
Merge "utils/pval: Add -lBlocksRuntime for compiler clang conditionally."

21 months agoMerge "utils: Add ast_assert_return"
Jenkins2 [Wed, 18 Apr 2018 19:43:56 +0000 (14:43 -0500)]
Merge "utils: Add ast_assert_return"

21 months agobridge_softmix / app_confbridge: Add support for REMB combining.
Joshua Colp [Wed, 4 Apr 2018 18:12:50 +0000 (15:12 -0300)]
bridge_softmix / app_confbridge: Add support for REMB combining.

This change adds the ability for multiple REMB reports in
bridge_softmix to be combined according to a configured
behavior into a single report. This single report is sent
back to the sender of video, which adjusts the encoding bitrate
to be at or below the bitrate of the report. The available
behaviors are: lowest, highest, and average. Lowest uses the
lowest received bitrate. Highest uses the highest received
bitrate. Average goes through the received bitrates adding
them to the previous average and creates a new average.

Other behaviors can be added in the future and the existing
average one may be adjusted, but this provides the foundation
to do so.

Support for configuring which behavior to use has been
added to app_confbridge.


Change-Id: I9eafe4e7c1f72d67074a8d6acb26bfcf19322b66

21 months agoMerge "res_musiconhold: Don't restart MOH from beginning after announcement."
Jenkins2 [Tue, 17 Apr 2018 17:21:37 +0000 (12:21 -0500)]
Merge "res_musiconhold: Don't restart MOH from beginning after announcement."

21 months agoutils: Add ast_assert_return
George Joseph [Fri, 13 Apr 2018 20:17:36 +0000 (14:17 -0600)]
utils: Add ast_assert_return

Similar to pjproject's PJ_ASSERT_RETURN macro, this one will do the

If the assert passes... NoOp

If the assert fails and AST_DEVMODE is defined, execute ast_assert()
then, if DO_CRASH isn't set, return from the calling function with
the supplied value.

If the assert fails and AST_DEVMODE is not defined, return from the
calling function with the supplied value.

The macro will execute a return without a value if one isn't suppled.

Change-Id: I0003844affeab550d5ff5bca7aa7cf8a559b873e

21 months agoutils/pval: Add -lBlocksRuntime for compiler clang conditionally.
Alexander Traud [Tue, 17 Apr 2018 12:06:55 +0000 (14:06 +0200)]
utils/pval: Add -lBlocksRuntime for compiler clang conditionally.


Change-Id: I930b364a33d54cc08dedfcd5bb45f7e83242f134

21 months agores_rtp_asterisk: Add support for receiving and handling NACK requests.
Ben Ford [Mon, 9 Apr 2018 22:09:03 +0000 (17:09 -0500)]
res_rtp_asterisk: Add support for receiving and handling NACK requests.

Adds the ability to receive and handle incoming NACK requests if
retransmissions are enabled. If retransmissions are enabled, a data
buffer is allocated that stores packets being sent. If a NACK request
is received, the packet requested for retransmission is sent if it is
still in the buffer. In the same request, if any of the following 16
packets are marked as not received, those will be sent as well if
available, as outlined in RFC4585.

Also changes RTCP RR and SR to use media source SSRC instead of packet
source SSRC when determining which instance to use for RTCP reports.

For more information, refer to the wiki page:

ASTERISK-27806 #close

Change-Id: I7f7f124af3b9d5d2fd9cffc6ba8cb48a6fff06ec

21 months agoMerge "res_pjsip.c: Split ast_sip_push_task_synchronous() to fit expectations."
George Joseph [Mon, 16 Apr 2018 16:12:30 +0000 (11:12 -0500)]
Merge "res_pjsip.c: Split ast_sip_push_task_synchronous() to fit expectations."

21 months agoMerge "pjsip_scheduler.c: Add ability to trace scheduled tasks."
Jenkins2 [Mon, 16 Apr 2018 12:11:26 +0000 (07:11 -0500)]
Merge "pjsip_scheduler.c: Add ability to trace scheduled tasks."

21 months agoMerge "pjsip_scheduler.c: Fix some corner cases."
Jenkins2 [Mon, 16 Apr 2018 11:49:14 +0000 (06:49 -0500)]
Merge "pjsip_scheduler.c: Fix some corner cases."

21 months agores_musiconhold: Don't restart MOH from beginning after announcement.
Ben Ford [Fri, 13 Apr 2018 19:32:48 +0000 (14:32 -0500)]
res_musiconhold: Don't restart MOH from beginning after announcement.

This reverts a problem introduced by the fix for ASTERISK_24329.
Now, when an announcement is played while waiting in a queue, music on
hold will not restart from the beginning of the sound file and will
instead pick up where it left off. However, the incorrect behavior in
ASTERISK_24329 is now present again; if an announcement X seconds
long is played when music on hold starts, music on hold will start X
seconds into the file.

ASTERISK-27774 #close
Reported by: lvl

Change-Id: I86b2885ee7063268f9b9747eddb788336ade989b

21 months agopjsip_scheduler.c: Add ability to trace scheduled tasks.
Richard Mudgett [Wed, 28 Mar 2018 20:13:13 +0000 (15:13 -0500)]
pjsip_scheduler.c: Add ability to trace scheduled tasks.

When a scheduled task is created you can pass in the
AST_SIP_SCHED_TASK_TRACK flag.  This new flag causes scheduling events to
be logged.

Change-Id: I91967eb3d5a220915ce86881a28af772f9a7f56b

21 months agores_pjsip.c: Split ast_sip_push_task_synchronous() to fit expectations.
Richard Mudgett [Tue, 27 Mar 2018 16:04:42 +0000 (11:04 -0500)]
res_pjsip.c: Split ast_sip_push_task_synchronous() to fit expectations.

ast_sip_push_task_synchronous() did not necessarily execute the passed in
task under the specified serializer.  If the current thread is any
registered pjsip thread then it would execute the task immediately instead
of under the specified serializer.  Reentrancy issues could result if the
task does not execute with the right serializer.

The original reason ast_sip_push_task_synchronous() checked to see if the
current thread was a registered pjsip thread was because of a deadlock
with masquerades and the channel technology's fixup callback
(ASTERISK_22936).  A subsequent masquerade deadlock fix (ASTERISK_24356)
involving call pickups avoided the original deadlock situation entirely.
The PJSIP channel technology's fixup callback no longer needed to call

However, there are a few places where this unexpected behavior is still
required to avoid deadlocks.  The pjsip monitor thread executes callbacks
that do calls to ast_sip_push_task_synchronous() that would deadlock if
the task were actually pushed to the specified serializer.  I ran into one
dealing with the pubsub subscriptions where an ao2 destructor called

* Split ast_sip_push_task_synchronous() into
ast_sip_push_task_wait_servant() and ast_sip_push_task_wait_serializer().
ast_sip_push_task_wait_servant() has the old behavior of
ast_sip_push_task_synchronous().  ast_sip_push_task_wait_serializer() has
the new behavior where the task is always executed by the specified
serializer or a picked serializer if one is not passed in.  Both functions
behave the same if the current thread is not a SIP servant.

* Redirected ast_sip_push_task_synchronous() to
ast_sip_push_task_wait_servant() to preserve API for released branches.


Change-Id: Id040fa42c0e5972f4c8deef380921461d213b9f3

21 months agopjsip_scheduler.c: Fix some corner cases.
Richard Mudgett [Thu, 22 Mar 2018 00:43:21 +0000 (19:43 -0500)]
pjsip_scheduler.c: Fix some corner cases.

* Fix the periodic interval wander because it may take significant time
between the sched thread queueing the task in the serializer and the
serializer actually executing the task.  The time it takes to actually
execute the task was already taken into account.

* Pass a schtd ref to the serializer when we queue a scheduled task on
the serializer.  We don't want it going away on us while it is in the
serializer queue.

* Skip the scheduled task if the task was canceled between queueing the
task to the serializer and the serializer actually executing the task.

* Reorder struct ast_sip_sched_task to avoid unnecessary padding.  Removed
task_id and added next_periodic.

* Hold a ref to the passed in serializer so the serializer cannot go away
on the scheduled task.


Change-Id: I6c8046b75f6953792c8c30e55b836a4291143f24

21 months agopjsip_scheduler.c: Sort "pjsip show scheduled_tasks" output.
Richard Mudgett [Fri, 23 Mar 2018 00:09:02 +0000 (19:09 -0500)]
pjsip_scheduler.c: Sort "pjsip show scheduled_tasks" output.

* A side benefit is that the scheduled tasks are not completely blocked
while the CLI command executes.

* Adjusted the "Task Name" column width to have more room for longer

Change-Id: Iec64aa463ee8b10eef90120e00c38b1fb444087e

21 months agoMerge "res_pjsip_notify.c: enable in-dialog NOTIFY"
Jenkins2 [Thu, 12 Apr 2018 20:03:54 +0000 (15:03 -0500)]
Merge "res_pjsip_notify.c: enable in-dialog NOTIFY"

21 months agoMerge "pjsip_scheduler.c: Fix ao2 usage errors."
Jenkins2 [Thu, 12 Apr 2018 15:25:18 +0000 (10:25 -0500)]
Merge "pjsip_scheduler.c: Fix ao2 usage errors."

21 months agoMerge "Build System: Strip '-std=c99' from CFLAGS provided by libraries."
Jenkins2 [Thu, 12 Apr 2018 14:52:02 +0000 (09:52 -0500)]
Merge "Build System: Strip '-std=c99' from CFLAGS provided by libraries."

21 months agocdr_mysql: Compile error because MYSQL_PORT definition is missing
Evandro Cesar Arruda [Mon, 2 Apr 2018 20:59:29 +0000 (17:59 -0300)]
cdr_mysql: Compile error because MYSQL_PORT definition is missing

If it is not defined, it will add MYSQL_PORT definition. After some
research on MySQL/MariaDB development tree, I couldn't find any reference
to MYSQL_PORT definition in include files.

ASTERISK-27782 #close

Change-Id: Ieee56c836fc2e8bd021c456145bba04c6068bb77

21 months agores_pjsip_session: Rewrite o= with external_media_address.
Chris-Savinovich [Tue, 10 Apr 2018 01:00:37 +0000 (21:00 -0400)]
res_pjsip_session: Rewrite o= with external_media_address.

It now appends the external IP address on the
o= line of the SDP packet.  The decision was made to write
the numeric IP address as opposed to the RFC that states
the FQDN should be used if and when available.  We believe
the usage of literal IP address will help avoid
potential problems.

ASTERISK-27614 #close

Change-Id: I84f3360f3606b8c4e8d161edb228799ec0b8a302

21 months agores_pjsip_notify.c: enable in-dialog NOTIFY
Nathan Bruning [Thu, 22 Feb 2018 18:18:48 +0000 (19:18 +0100)]
res_pjsip_notify.c: enable in-dialog NOTIFY

This patch adds support to send in-dialog SIP NOTIFY commands on
chan_pjsip channels, similar to the functionality recently added
for chan_sip (ASTERISK_27461).

This extends res_pjsip_notify to allow for in-dialog messages.


Change-Id: If7f3151a6d633e414d5dc319d5efc1443c43dd29

21 months agoMerge "res_pjsip_refer/chan_sip: Fix INVITE with replaces transfer to ConfBridge"
Jenkins2 [Wed, 11 Apr 2018 12:11:16 +0000 (07:11 -0500)]
Merge "res_pjsip_refer/chan_sip: Fix INVITE with replaces transfer to ConfBridge"

21 months agoMerge "chan_sip.c: Fix INVITE with replaces channel ref leak."
George Joseph [Tue, 10 Apr 2018 15:10:52 +0000 (10:10 -0500)]
Merge "chan_sip.c: Fix INVITE with replaces channel ref leak."

21 months agopjsip_scheduler.c: Fix ao2 usage errors.
Richard Mudgett [Thu, 22 Mar 2018 18:35:04 +0000 (13:35 -0500)]
pjsip_scheduler.c: Fix ao2 usage errors.

* Removed several invalid uses of OBJ_NOLOCK.  These uses resulted in the
'tasks' container being accessed without a lock in a multi-threaded
environment.  A recipe for crashes.

* Removed needlessly obtaining schtd object references.  If the caller
providing you a pointer to an object doesn't have a valid reference then
you cannot safely get one from it.

* Getting a ref to 'tasks' when you aren't copying the pointer into
another location is useless.  The 'tasks' container pointer is global.

* Removed many unnecessary uses of RAII_VAR.

* Make ast_sip_schedule_task() name parameter const.


Change-Id: I5c62488e651314e2a1dbc01f5b078a15512d73db

21 months agoMerge "pjsip / res_rtp_asterisk: Add support for sending REMB"
Jenkins2 [Mon, 9 Apr 2018 16:14:16 +0000 (11:14 -0500)]
Merge "pjsip / res_rtp_asterisk: Add support for sending REMB"

21 months agoMerge "res_rtp_asterisk: Fix minimum block word length for REMB."
Joshua Colp [Mon, 9 Apr 2018 15:58:00 +0000 (10:58 -0500)]
Merge "res_rtp_asterisk: Fix minimum block word length for REMB."

21 months agoMerge "app_confbridge / bridge_softmix: Add ability to configure REMB interval."
Joshua Colp [Mon, 9 Apr 2018 15:57:40 +0000 (10:57 -0500)]
Merge "app_confbridge / bridge_softmix: Add ability to configure REMB interval."

21 months agoMerge "Build System: Fixes for configure script."
Joshua Colp [Mon, 9 Apr 2018 15:32:49 +0000 (10:32 -0500)]
Merge "Build System: Fixes for configure script."

21 months agoMerge "app_originate: Add async option."
Joshua Colp [Mon, 9 Apr 2018 15:32:38 +0000 (10:32 -0500)]
Merge "app_originate: Add async option."

21 months agoMerge "res_rtp_asterisk: Queue video update on picture loss indication."
Jenkins2 [Mon, 9 Apr 2018 15:27:26 +0000 (10:27 -0500)]
Merge "res_rtp_asterisk: Queue video update on picture loss indication."

21 months agores_pjsip_refer/chan_sip: Fix INVITE with replaces transfer to ConfBridge
Richard Mudgett [Thu, 5 Apr 2018 23:33:40 +0000 (18:33 -0500)]
res_pjsip_refer/chan_sip: Fix INVITE with replaces transfer to ConfBridge

There is a problem when an INVITE-with-Replaces transfer targets a channel
in a ConfBridge.  The transfer will unconditionally swap out the
ConfBridge channel.  Unfortunately, the ConfBridge state will not be aware
of this change.  Unexpected behavior will happen as a result since
ConfBridge channels currently can only be replaced by a masquerade and not
normal bridge channel moves.

* We just need to pretend that the channel isn't in a bridge (like other
transfer methods already do) so the transfer channel will masquerade into
the ConfBridge channel.

Change-Id: I209beb0e748fa4f4b92a576f36afa8f495ba4c82

21 months agopjsip / res_rtp_asterisk: Add support for sending REMB
Joshua Colp [Wed, 28 Mar 2018 12:27:31 +0000 (12:27 +0000)]
pjsip / res_rtp_asterisk: Add support for sending REMB

This change allows chan_pjsip to be given an AST_FRAME_RTCP
containing REMB feedback and pass it to res_rtp_asterisk.
Once res_rtp_asterisk receives the frame a REMB RTCP feedback
packet is constructed with the appropriate contents and sent
to the remote endpoint.


Change-Id: Ic53f821c1560d8924907ad82c4d9c0bc322b38cd

21 months agoMerge "res_pjsip: Update authenticate_qualify documentation."
Jenkins2 [Fri, 6 Apr 2018 11:53:51 +0000 (06:53 -0500)]
Merge "res_pjsip: Update authenticate_qualify documentation."

21 months agores_rtp_asterisk: Fix minimum block word length for REMB.
Joshua Colp [Fri, 6 Apr 2018 01:02:27 +0000 (01:02 +0000)]
res_rtp_asterisk: Fix minimum block word length for REMB.

The minimum block word length is actually 4, not 5.

Change-Id: I878542218225aed72c72bdf1b856fc822cd2d649

21 months agores_rtp_asterisk: Queue video update on picture loss indication.
Joshua Colp [Thu, 5 Apr 2018 23:48:31 +0000 (23:48 +0000)]
res_rtp_asterisk: Queue video update on picture loss indication.

The previous payload specific feedback handling was very single
minded in that it just assumed everything should trigger a video
update. This was changed but the handling of picture loss indication
was not added. The result was that video may not flow. This change
adds it explicitly in.

Change-Id: I1894be02e39ee10a0af841b5a1dca5f0ec7d60b6

21 months agochan_sip.c: Fix INVITE with replaces channel ref leak.
Richard Mudgett [Thu, 5 Apr 2018 22:40:52 +0000 (17:40 -0500)]
chan_sip.c: Fix INVITE with replaces channel ref leak.

Given the below call scenario:
A -> Ast1 -> B
C <- Ast2 <- B

1) A calls B through Ast1
2) B calls C through Ast2
3) B transfers A to C

When party B transfers A to C, B sends a REFER to Ast1 causing Ast1 to
send an INVITE with replaces to Ast2.  Ast2 then leaks a channel ref of
the channel between Ast1 and Ast2.

Channel ref leaks are easily seen in the CLI "core show channels" output.
The leaked channels appear in the output but you can do nothing with them
and they never go away unless you restart Asterisk.

* Properly account for the channel refs when imparting a channel into a
bridge when handling an INVITE with replaces in handle_invite_replaces().
The ast_bridge_impart() function steals a channel ref but the code didn't
account for how many refs were held by the code at the time and which ref
was stolen.

* Eliminated RAII_VAR in handle_invite_replaces().


Change-Id: I7edbed774314b55acf0067b2762bfe984ecaa9a4

21 months agores_pjsip: Update authenticate_qualify documentation.
Richard Mudgett [Thu, 22 Mar 2018 00:40:46 +0000 (19:40 -0500)]
res_pjsip: Update authenticate_qualify documentation.

Change-Id: I3811de0014b1ffe96d4a3b49cddd5d4ca02ee5d4

21 months agoapp_agent_pool.c: Fix off nominal ref leak.
Richard Mudgett [Mon, 2 Apr 2018 21:49:38 +0000 (16:49 -0500)]
app_agent_pool.c: Fix off nominal ref leak.

Change-Id: Ib427ffc2c802620eaafb08b1c2a17dddd8fb8eb6

21 months agoBuild System: Strip '-std=c99' from CFLAGS provided by libraries.
Corey Farrell [Wed, 4 Apr 2018 15:02:06 +0000 (11:02 -0400)]
Build System: Strip '-std=c99' from CFLAGS provided by libraries.

Asterisk requires GNU C extensions.  On some systems certain libraries
may incorrectly push -std=c99 into CFLAGS, thus breaking the build.
This change causes that flag to be stripped so the Asterisk build is not
broken by those libraries.  This change is made for both pkgconfig and
tool based libraries.

ASTERISK-27629 #close

Change-Id: I13389613b194abbac77becf90cd950dc168704db

21 months agoBuild System: Fixes for configure script.
Corey Farrell [Tue, 3 Apr 2018 19:39:39 +0000 (15:39 -0400)]
Build System: Fixes for configure script.

* Replace all 'else if' statements with 'elif'.
* Use loop to detect versioned lua headers and libraries.

The loop for detecting lua fixes a bug where LUA_INCLUDE would be
appended with the directory of every lua version after the first one is

Change-Id: I3276f9aee955014108345be6092f51c932b43a0f

21 months agoapp_confbridge / bridge_softmix: Add ability to configure REMB interval.
Joshua Colp [Mon, 2 Apr 2018 13:53:17 +0000 (10:53 -0300)]
app_confbridge / bridge_softmix: Add ability to configure REMB interval.

This change adds a configuration option to app_confbridge which can be
used to set the interval at which we will send a combined REMB (remote
estimated maximum bitrate) frame to sources of video. The bridging API
has also been extended slightly to allow setting this so bridge_softmix
can use it.


Change-Id: I0e49eae60f369c86434414f3cb8278709c793c82

21 months agoMerge "install_prereq: Add Gentoo Linux."
Joshua Colp [Tue, 3 Apr 2018 12:32:23 +0000 (07:32 -0500)]
Merge "install_prereq: Add Gentoo Linux."

21 months agoMerge "install_prereq: Add Slackware (somehow)."
Jenkins2 [Tue, 3 Apr 2018 11:18:26 +0000 (06:18 -0500)]
Merge "install_prereq: Add Slackware (somehow)."

21 months agoMerge "res_pjsip: Correct usages of pjproject's timer heap"
Jenkins2 [Mon, 2 Apr 2018 18:06:44 +0000 (13:06 -0500)]
Merge "res_pjsip:  Correct usages of pjproject's timer heap"

21 months agoMerge "pjroject_bundled: Add already-destroyed check to tsx_timer_callback"
Jenkins2 [Mon, 2 Apr 2018 15:42:52 +0000 (10:42 -0500)]
Merge "pjroject_bundled:  Add already-destroyed check to tsx_timer_callback"

21 months agores_pjsip: Correct usages of pjproject's timer heap
George Joseph [Tue, 2 Jan 2018 13:54:54 +0000 (06:54 -0700)]
res_pjsip:  Correct usages of pjproject's timer heap

Fix some timer heap initializations and cancels to try and prevent
crashes and timer heap issues.

Change-Id: I64885d190fa22097d1b55987091375541e57a7ee

21 months agoMerge "main: Update copyright notice with year 2018"
George Joseph [Mon, 2 Apr 2018 15:14:15 +0000 (10:14 -0500)]
Merge "main: Update copyright notice with year 2018"

21 months agopjroject_bundled: Add already-destroyed check to tsx_timer_callback
George Joseph [Sun, 25 Mar 2018 18:35:12 +0000 (12:35 -0600)]
pjroject_bundled:  Add already-destroyed check to tsx_timer_callback

There have been cases that when the transaction timer callback is called
the tsx is already destroyed.  This causes a crash.  We now check the
tsx state and return if the tsx is already destroyed.

Change-Id: If93acd5e48d9ca5bb553f2405d5afc836842fe1c

21 months agopjproject_bundled: timer: Clean up usage of timer heap
George Joseph [Sun, 25 Mar 2018 18:25:30 +0000 (12:25 -0600)]
pjproject_bundled: timer: Clean up usage of timer heap

Added a new pj_timer_entry_reset function that resets a timer_entry
for re-use.

Changed direct settings of timer_entry fields to use
pj_timer_entry_init and pj_timer_entry_reset.

Fixed issues where timers were being rescheduled incorrectly.

Change-Id: I5b624bfbc5c1429117484b9b24567293002148e6

21 months agoMerge "BuildSystem: With external editline, do not require libs for internal editline."
Jenkins2 [Mon, 2 Apr 2018 13:36:01 +0000 (08:36 -0500)]
Merge "BuildSystem: With external editline, do not require libs for internal editline."

21 months agoMerge "core: Create main/options.c."
Jenkins2 [Mon, 2 Apr 2018 13:31:08 +0000 (08:31 -0500)]
Merge "core: Create main/options.c."

21 months agoMerge "pjproject_bundled: Add patch for pj_atomic crashes"
George Joseph [Mon, 2 Apr 2018 13:19:05 +0000 (08:19 -0500)]
Merge "pjproject_bundled:  Add patch for pj_atomic crashes"

22 months agoMerge "res_rtp_asterisk: Add support for raising additional RTCP messages."
Kevin Harwell [Thu, 29 Mar 2018 20:19:17 +0000 (15:19 -0500)]
Merge "res_rtp_asterisk: Add support for raising additional RTCP messages."

22 months agoMerge "main/indications: Use ast_cli_completion_add for all completions."
Kevin Harwell [Thu, 29 Mar 2018 20:06:45 +0000 (15:06 -0500)]
Merge "main/indications: Use ast_cli_completion_add for all completions."

22 months agoMerge "BuildSystem: pjsip_evsub_set_uas_timeout was not used (part 2)."
Jenkins2 [Thu, 29 Mar 2018 20:02:19 +0000 (15:02 -0500)]
Merge "BuildSystem: pjsip_evsub_set_uas_timeout was not used (part 2)."

22 months agoMerge "pjsip_transport_events.c: Fix crash using stale transport pointer."
Jenkins2 [Thu, 29 Mar 2018 19:39:26 +0000 (14:39 -0500)]
Merge "pjsip_transport_events.c: Fix crash using stale transport pointer."

22 months agoMerge "test_data_buffer.c: Add unit tests for data buffer API."
Jenkins2 [Thu, 29 Mar 2018 18:41:37 +0000 (13:41 -0500)]
Merge "test_data_buffer.c: Add unit tests for data buffer API."

22 months agoMerge "Add data buffer API to store packets."
Jenkins2 [Thu, 29 Mar 2018 18:38:02 +0000 (13:38 -0500)]
Merge "Add data buffer API to store packets."

22 months agoMerge "core: fix getopt(3) usage"
Jenkins2 [Thu, 29 Mar 2018 15:52:25 +0000 (10:52 -0500)]
Merge "core: fix getopt(3) usage"

22 months agopjsip_transport_events.c: Fix crash using stale transport pointer.
Ross Beer [Wed, 7 Mar 2018 12:15:05 +0000 (12:15 +0000)]
pjsip_transport_events.c: Fix crash using stale transport pointer.

Apparently it is possible for the transport to be destroyed without
triggering the transport callback logic.  As a result the transport gets
destroyed and we have a stale pointer in the active_transports container.

* Invoke the transport monitor callback checks when the transport is
destroyed in addition to when it is disconnected and shutdown.


Change-Id: Ia9b5469fea8f2b3f2d8476fae6b748a4d23e7261

22 months agotest_data_buffer.c: Add unit tests for data buffer API.
Ben Ford [Mon, 19 Mar 2018 14:36:44 +0000 (09:36 -0500)]
test_data_buffer.c: Add unit tests for data buffer API.

Added unit tests for the data buffer API. These tests include creating a
data buffer, putting payloads into the buffer, resizing the buffer, and
the nominal case for data buffer usage, which consists of adding
the max number of payloads to the buffer, checking to see if the correct
payloads are present, then adding more payloads and checking again to
see if the previous payloads were replaced or not.

For more information, refer to the wiki page:

Change-Id: Id5b599aa15a5e61d0ec080f97cd0c57bd07e6f8f

22 months agoAdd data buffer API to store packets.
Ben Ford [Fri, 23 Feb 2018 19:49:21 +0000 (13:49 -0600)]
Add data buffer API to store packets.

Adds a data buffer with a configurable size that can store different
kinds of packets (like RTP packets for retransmission). Given a number
it will store a data packet at that position relative to the others.
Given a number it will retrieve the given data packet if it is present.
This is purposely a storage of arbitrary things so it can be used not
just for RTP packets but also Asterisk frames in the future if needed.
The API does not internally use a lock, so it will be up to the user of
the API to properly protect the data buffer.

For more information, refer to the wiki page:

Change-Id: Iff13c5d4795d52356959fe2a57360cd57dfade07

22 months agopjproject_bundled: Add patch for pj_atomic crashes
George Joseph [Sun, 25 Mar 2018 18:12:39 +0000 (12:12 -0600)]
pjproject_bundled:  Add patch for pj_atomic crashes

There have been some crashes in the past where something attempts
to use a pj_atomic after it's already been destroyed.  This patch
tries to prevent it by making sure that pj_atomic_destroy sets
its mutex to NULL when it's done.  The pj_mutex functions already check
for a NULL mutex and just return PJ_EINVAL.

Teluu also added some checks to the win32 implementation as well.

Change-Id: Id25f70b79fdedf44ead6e6e1763a4417d3b3f825

22 months agores_rtp_asterisk: Add support for raising additional RTCP messages.
Joshua Colp [Wed, 21 Mar 2018 13:52:08 +0000 (13:52 +0000)]
res_rtp_asterisk: Add support for raising additional RTCP messages.

This change extends the existing AST_FRAME_RTCP frame type to be
able to contain additional RTCP message types, such as feedback
messages. The payload type is contained in the subclass which allows
knowing what is in the frame itself.

The RTCP feedback message type is now handled and REMB[1] messages
are raised with their containing information.

This also fixes a bug where all feedback messages were triggering
video updates instead of just FIR and FUR.

Finally RTCP frames are now passed up through the Asterisk core to
what is handling the channel, mapped appropriately in the case of
bridging, and written to an outgoing stream. Since RTCP frames are
on a per-stream basis this is only done on multistream capable



Change-Id: I680da0ad8d5059d5e9655d896fb9d92e9da8491e

22 months agomain: Update copyright notice with year 2018
Florian Floimair [Tue, 27 Mar 2018 13:27:09 +0000 (15:27 +0200)]
main: Update copyright notice with year 2018

Change-Id: I2d80bc5edf940fab914cba3d8a0fa0b5eb2a3148

22 months agoMerge "loader: Reserve space for additional pointers in ast_module_info."
Jenkins2 [Mon, 26 Mar 2018 16:44:52 +0000 (11:44 -0500)]
Merge "loader: Reserve space for additional pointers in ast_module_info."

22 months agocore: fix getopt(3) usage
Guido Falsi [Mon, 26 Mar 2018 12:42:52 +0000 (14:42 +0200)]
core: fix getopt(3) usage

Setting optind = 0 is forced to 1 in glibc implementation, but
causes option parsing to be flawed in other implementations, for
example on FreeBSD.

ASTERISK-27773 #close

Change-Id: Ia548e69f8302e9754dbbedb6bc451c0700c66f61

22 months agoinstall_prereq: Add Slackware (somehow).
Alexander Traud [Fri, 23 Mar 2018 18:15:09 +0000 (19:15 +0100)]
install_prereq: Add Slackware (somehow).


Change-Id: Ib87e0483c785542238cfe34c1e884d5a31edfaab

22 months agoinstall_prereq: Add Gentoo Linux.
Alexander Traud [Fri, 23 Mar 2018 14:13:44 +0000 (15:13 +0100)]
install_prereq: Add Gentoo Linux.


Change-Id: Ieb13293cd67481f3a33f58f6f7c8c3ee1e338e7a

22 months agomain/indications: Use ast_cli_completion_add for all completions.
Corey Farrell [Sat, 17 Mar 2018 06:02:06 +0000 (02:02 -0400)]
main/indications: Use ast_cli_completion_add for all completions.

Change-Id: I371be01f178fb542a9fbe8d97e7ae21aa4d82c36

22 months agoMerge "bridge_softmix: Clear "talking" when a channel is put on hold"
Jenkins2 [Thu, 22 Mar 2018 13:47:19 +0000 (08:47 -0500)]
Merge "bridge_softmix: Clear "talking" when a channel is put on hold"

22 months agoapp_originate: Add async option.
Russell Bryant [Wed, 21 Mar 2018 19:54:48 +0000 (15:54 -0400)]
app_originate: Add async option.

Add an option to make app_originate not wait for the created channel
to answer.

Change-Id: I7fc2facd77079abc6321f44e8bcd4e39298de2ae
Requested-by: Frederic Steinfels <>
Signed-off-by: Russell Bryant <>

22 months agoBuildSystem: pjsip_evsub_set_uas_timeout was not used (part 2).
Alexander Traud [Thu, 22 Mar 2018 12:27:52 +0000 (13:27 +0100)]
BuildSystem: pjsip_evsub_set_uas_timeout was not used (part 2).

The previous change was not complete.


Change-Id: I11082c14c0ef9c6af8c995084a6851337ea2a90f

22 months agoMerge "Revert "BuildSystem: In NetBSD, the Python Programming Language is python...
Jenkins2 [Thu, 22 Mar 2018 10:44:30 +0000 (05:44 -0500)]
Merge "Revert "BuildSystem: In NetBSD, the Python Programming Language is python-X.Y.""

22 months agoBuildSystem: With external editline, do not require libs for internal editline.
Alexander Traud [Thu, 22 Mar 2018 10:43:18 +0000 (11:43 +0100)]
BuildSystem: With external editline, do not require libs for internal editline.


Change-Id: Ib17a7415297a210cfcdbf149e4df9b6edadbfab6

22 months agocore: Create main/options.c.
Corey Farrell [Thu, 22 Mar 2018 03:00:56 +0000 (23:00 -0400)]
core: Create main/options.c.

This creates a separate source to 'own' symbols related to options.h and
paths.h.  This significantly reduces the number of exports created by
main/asterisk.o.  This change is required to eventually be able to
link unmodified Asterisk sources to utilities and/or stand-alone tests.


Change-Id: I5cf184f4757f9363b80c9e678bdc35c477122380

22 months agoRevert "BuildSystem: In NetBSD, the Python Programming Language is python-X.Y."
George Joseph [Thu, 22 Mar 2018 00:25:58 +0000 (18:25 -0600)]
Revert "BuildSystem: In NetBSD, the Python Programming Language is python-X.Y."

Something is causing a python2/python3 mismatch on Fedora27.

PYTHONDEV_CFLAGS='-I/usr/include/python3.6m '
PYTHONDEV_INCLUDE='-I/usr/include/python3.6m '
PYTHONDEV_LIB='-lpython3.6m '
PYTHONDEV_LIBS='-lpython3.6m '

This reverts commit be0e9920b64e3b07501b299d131309b58f9b0ddf.

Change-Id: I86dd102eb3ead199fe89178cdbadb36b4e2cfd1b

22 months agoMerge "core: Stop using AST_INLINE_API for allocator functions."
Jenkins2 [Wed, 21 Mar 2018 15:46:30 +0000 (10:46 -0500)]
Merge "core: Stop using AST_INLINE_API for allocator functions."

22 months agoMerge "rtp: Add REMB RTP property and set it on PJSIP video RTP."
Jenkins2 [Wed, 21 Mar 2018 15:38:32 +0000 (10:38 -0500)]
Merge "rtp: Add REMB RTP property and set it on PJSIP video RTP."

22 months agoMerge "chan_sip: Peers with distinct source ports don't match, regardless of transport."
Jenkins2 [Wed, 21 Mar 2018 14:42:54 +0000 (09:42 -0500)]
Merge "chan_sip: Peers with distinct source ports don't match, regardless of transport."

22 months agoMerge "func_channel: Delete dead CHANNEL_TRACE code"
Jenkins2 [Wed, 21 Mar 2018 11:26:09 +0000 (06:26 -0500)]
Merge "func_channel: Delete dead CHANNEL_TRACE code"

22 months agoloader: Reserve space for additional pointers in ast_module_info.
Corey Farrell [Thu, 8 Feb 2018 19:23:35 +0000 (14:23 -0500)]
loader: Reserve space for additional pointers in ast_module_info.

This creates 4 reserved pointers in case we need additional dependency
management fields.

Change-Id: If991ec99b779df1b2dfbd38ce1a0cd79f9e01821

22 months agobridge_softmix: Clear "talking" when a channel is put on hold
Kevin Harwell [Tue, 20 Mar 2018 20:28:12 +0000 (15:28 -0500)]
bridge_softmix: Clear "talking" when a channel is put on hold

This patch clears the talking flag from the channel (if already set), and
notifies listeners when that channel is put on hold. Note however, if the
endpoint continues to send audio frames and these are received by the bridge
then that channel will be put back into a "talking" state even though they
are on hold.

ASTERISK-27755 #close

Change-Id: I930e16c4662810f9f02043d69062f88173c5e2ef

22 months agoBuildSystem: For consistency, avoid extra libs to be empty.
Alexander Traud [Tue, 20 Mar 2018 16:53:19 +0000 (17:53 +0100)]
BuildSystem: For consistency, avoid extra libs to be empty.

AST_EXT_LIB_CHECK has several optional parameters. When an optional parameter
is left empty, [] is used to indicate this. However, this is done in the script
./configure only then, when a further parameter is not empty. For example, when
no extra libraries are needed to test the checked library, parameter 5 is not
mentioned. Except parameter 6 and higher are used, then parameter 5 must be

However, this general rule was broken
* four times for parameter 5 (extra libs) and
* three times for parameter 4 (header)
as found via the Regular Expression \[\]\). In case of parameter 5, all cases
were changed, because that happened for no reason. In case of parameter 4, an
[] improves readability actually. Therefore for parameter 4, the only case which
did not do it was changed. All this aims to create more consistency: Only do
something different if there is a reason to do so.

Change-Id: I037ef170cf1ad94497151a9ea5071a31c656cafe

22 months agoMerge "core: Remove additional symbols."
Joshua Colp [Tue, 20 Mar 2018 16:44:06 +0000 (11:44 -0500)]
Merge "core: Remove additional symbols."

22 months agoMerge "core: Remove dead symbols from"
Jenkins2 [Tue, 20 Mar 2018 16:40:39 +0000 (11:40 -0500)]
Merge "core: Remove dead symbols from"

22 months agoMerge "channel.c: Allow generic plc then channel formats are equal"
Jenkins2 [Tue, 20 Mar 2018 16:16:52 +0000 (11:16 -0500)]
Merge "channel.c:  Allow generic plc then channel formats are equal"

22 months agoMerge "BuildSystem: In NetBSD, the Python Programming Language is python-X.Y."
George Joseph [Tue, 20 Mar 2018 15:37:57 +0000 (10:37 -0500)]
Merge "BuildSystem: In NetBSD, the Python Programming Language is python-X.Y."

22 months agoMerge "BuildSystem: Instead of $PJPROJECT_LIBS with s, use $PJPROJECT_LIB everywhere."
Jenkins2 [Tue, 20 Mar 2018 15:24:11 +0000 (10:24 -0500)]
Merge "BuildSystem: Instead of $PJPROJECT_LIBS with s, use $PJPROJECT_LIB everywhere."

22 months agoMerge "main/sounds: Use ast_cli_completion_add."
Jenkins2 [Tue, 20 Mar 2018 15:09:26 +0000 (10:09 -0500)]
Merge "main/sounds: Use ast_cli_completion_add."

22 months agofunc_channel: Delete dead CHANNEL_TRACE code
Ivan Poddubny [Tue, 20 Mar 2018 14:58:38 +0000 (15:58 +0100)]
func_channel: Delete dead CHANNEL_TRACE code

The functions behind the flag and the flag itself were removed
from Asterisk 12 as incompatible with the new architecture.

Change-Id: I058493ef7a53ee290fd225bbcbb07bf46b623ccf

22 months agoMerge "named_acl: Use ast_cli_completion_add."
Joshua Colp [Tue, 20 Mar 2018 14:51:41 +0000 (09:51 -0500)]
Merge "named_acl: Use ast_cli_completion_add."

22 months agoMerge "manager: Use ast_cli_completion_add for completion generators."
Joshua Colp [Tue, 20 Mar 2018 14:36:56 +0000 (09:36 -0500)]
Merge "manager: Use ast_cli_completion_add for completion generators."

22 months agoMerge "main/test: Use ast_cli_completion_add."
Joshua Colp [Tue, 20 Mar 2018 14:18:57 +0000 (09:18 -0500)]
Merge "main/test: Use ast_cli_completion_add."

22 months agoMerge "core: Minor cleanup of ast_el_read_char."
Jenkins2 [Tue, 20 Mar 2018 13:47:15 +0000 (08:47 -0500)]
Merge "core: Minor cleanup of ast_el_read_char."

22 months agoMerge "aco: Use ast_cli_completion_add for 'config show help'."
Joshua Colp [Tue, 20 Mar 2018 13:31:49 +0000 (08:31 -0500)]
Merge "aco: Use ast_cli_completion_add for 'config show help'."

22 months agoMerge "main/config: Use ast_cli_completion_add for reload completion."
Jenkins2 [Tue, 20 Mar 2018 13:20:54 +0000 (08:20 -0500)]
Merge "main/config: Use ast_cli_completion_add for reload completion."

22 months agoMerge "main/translate: Use ast_cli_completion_add."
Jenkins2 [Tue, 20 Mar 2018 12:58:00 +0000 (07:58 -0500)]
Merge "main/translate: Use ast_cli_completion_add."

22 months agoMerge "main/taskprocessor: Use ast_cli_completion_add."
Joshua Colp [Tue, 20 Mar 2018 12:38:03 +0000 (07:38 -0500)]
Merge "main/taskprocessor: Use ast_cli_completion_add."