3 years agochannel: Use frame deferral API for safe sleep.
Mark Michelson [Thu, 3 Nov 2016 21:46:41 +0000 (16:46 -0500)]
channel: Use frame deferral API for safe sleep.

This is another case where manual frame deferral can be replaced with
centralized routines instead.

Change-Id: I42cdf205f8f29a7977e599751a57efbaac07c30e
(cherry picked from commit d149c4b9e07eeb880d8428ad52c6fdb315cc15f5)

3 years agoautoservice: Use frame deferral API
Mark Michelson [Thu, 3 Nov 2016 21:46:03 +0000 (16:46 -0500)]
autoservice: Use frame deferral API

Rather than use manual frame deferral, just let the channel API do it
for us.


Change-Id: I688386f36e765dbc07be863943a43f26bd5eac49
(cherry picked from commit 8ba3e2fc27f9966b8c7ce75c1eca6208613a9315)

3 years agoAGI: Only defer frames when in an interception routine.
Mark Michelson [Thu, 3 Nov 2016 21:42:40 +0000 (16:42 -0500)]
AGI: Only defer frames when in an interception routine.

AGI recently was modified to defer important frames. This was because
when AGI was used in a connected line interception routine, the
resulting connected line frame would end up getting discarded by the

However, this caused bad behavior in other cases. Specifically, during a
transfer, if someone attempted to manually set the Caller ID on a
channel in an AGI, the deferred connected line frame would end up
overwriting what had been manually set in the AGI.

Since the initial issue was specific to interception routines, this
change removes the manual frame deferral from AGI and instead uses the
new frame deferral API in interception routines.

ASTERISK-26343 #close
Reported by Morton Tryfoss

Change-Id: Iab7d39436d0ee99bfe32ad55ef91e9bd88db4208

3 years agoAdd API for channel frame deferral.
Mark Michelson [Thu, 3 Nov 2016 21:36:13 +0000 (16:36 -0500)]
Add API for channel frame deferral.

There are several places in Asterisk that have duplicated logic
for deferring important frames until later.

This commit adds a couple of API calls to facilitate this automatically.

ast_channel_start_defer_frames(): Future reads of deferrable frames on
this channel will be deferred until later.

ast_channel_stop_defer_frames(): Any frames that have been deferred get
requeued onto the channel.


Change-Id: I3e1b87bc6796f222442fa6f7d1b6a4706fb33641

3 years agoMerge "res_http_websocket: Increase the buffer size for non-LOW_MEMORY systems"
Joshua Colp [Tue, 8 Nov 2016 10:59:53 +0000 (04:59 -0600)]
Merge "res_http_websocket: Increase the buffer size for non-LOW_MEMORY systems"

3 years agoMerge "res_stasis: Don't unsubscribe from a NULL bridge."
Joshua Colp [Tue, 8 Nov 2016 10:59:24 +0000 (04:59 -0600)]
Merge "res_stasis: Don't unsubscribe from a NULL bridge."

3 years agoMerge "chan_ooh323: reset rrq count on gk registration"
Joshua Colp [Tue, 8 Nov 2016 10:59:12 +0000 (04:59 -0600)]
Merge "chan_ooh323: reset rrq count on gk registration"

3 years agoMerge "chan_ooh323: Fixes to work right with Cisco devices"
Joshua Colp [Tue, 8 Nov 2016 10:58:04 +0000 (04:58 -0600)]
Merge "chan_ooh323: Fixes to work right with Cisco devices"

3 years agoMerge "stasis_recording/stored: remove calls to deprecated readdir_r function."
Joshua Colp [Tue, 8 Nov 2016 10:57:55 +0000 (04:57 -0600)]
Merge "stasis_recording/stored: remove calls to deprecated readdir_r function."

3 years agoMerge "res_stasis: Set a video source mode on Stasis created bridges"
Joshua Colp [Tue, 8 Nov 2016 01:07:55 +0000 (19:07 -0600)]
Merge "res_stasis: Set a video source mode on Stasis created bridges"

3 years agoMerge "main/bridge: Add some verbose logging for video source changes"
Joshua Colp [Mon, 7 Nov 2016 23:49:23 +0000 (17:49 -0600)]
Merge "main/bridge: Add some verbose logging for video source changes"

3 years agoMerge "main/bridge_channel: Fix channel reference leak on video source"
Joshua Colp [Mon, 7 Nov 2016 22:43:42 +0000 (16:43 -0600)]
Merge "main/bridge_channel: Fix channel reference leak on video source"

3 years agoMerge "bridges/bridge_softmix: Remove SSRC changes on join/leave; update video source"
Joshua Colp [Mon, 7 Nov 2016 22:01:03 +0000 (16:01 -0600)]
Merge "bridges/bridge_softmix: Remove SSRC changes on join/leave; update video source"

3 years agoMerge "pjproject_bundled: Fix issue with libasteriskpj needing libresample"
Joshua Colp [Mon, 7 Nov 2016 16:40:14 +0000 (10:40 -0600)]
Merge "pjproject_bundled:  Fix issue with libasteriskpj needing libresample"

3 years agores_stasis: Don't unsubscribe from a NULL bridge.
Joshua Colp [Wed, 2 Nov 2016 15:52:13 +0000 (15:52 +0000)]
res_stasis: Don't unsubscribe from a NULL bridge.

A NULL bridge has special meaning in res_stasis for
unsubscribing. It means that a subscription to ALL
bridges should be removed. This should not be done
as part of the normal subscription management in
the res_stasis channel loop.


Change-Id: I6d5bea8246dd13a22ef86b736aefbf2a39c15af0

3 years agochan_ooh323: Fixes to work right with Cisco devices
Alexander Anikin [Thu, 3 Nov 2016 12:42:20 +0000 (16:42 +0400)]
chan_ooh323: Fixes to work right with Cisco devices

Changed output packets queue processing algo to one read-one write
instead of all read-all send

Remove h.245 tunneling parameter from ReleaseComplete packet

ASTERISK-24400 #close
Reported by: Dmitry Melekhov
Tested by: Dmitry Melekhov

Change-Id: I0b31933b062a21011dbac9a82b8bcfe345f406f6

3 years agochan_ooh323: reset rrq count on gk registration
Alexander Anikin [Thu, 3 Nov 2016 18:10:53 +0000 (22:10 +0400)]
chan_ooh323: reset rrq count on gk registration

reset registration attempts count on success registration on gatekeeper

Change-Id: I5f47351852e0ca76c9ac78421659600e0f106336

3 years agoMerge "chan_ooh323: Fix infinite loop on read second part of H.225 packet"
Joshua Colp [Mon, 7 Nov 2016 14:07:10 +0000 (08:07 -0600)]
Merge "chan_ooh323: Fix infinite loop on read second part of H.225 packet"

3 years agoMerge "rtp_engine: Allow more than 32 dynamic payload types."
zuul [Mon, 7 Nov 2016 12:48:38 +0000 (06:48 -0600)]
Merge "rtp_engine: Allow more than 32 dynamic payload types."

3 years agores_http_websocket: Increase the buffer size for non-LOW_MEMORY systems
Matt Jordan [Fri, 4 Nov 2016 20:42:09 +0000 (15:42 -0500)]
res_http_websocket: Increase the buffer size for non-LOW_MEMORY systems

Not surprisingly, using Respoke (and possibly other systems) it is
possible to blow past the 16k limit for a WebSocket packet size. This
patch bumps it up to 32k, which, at least for Respoke, is sufficient.
For now.

Because 32k is laughable on a LOW_MEMORY system (as is 16k, for that
matter), this patch adds a LOW_MEMORY directive that sets the buffer to
8k for systems who have asked for their reduced memory availability to
be considered.

Change-Id: Id235902537091b58608196844dc4b045e383cd2e

3 years agores_stasis: Set a video source mode on Stasis created bridges
Matt Jordan [Fri, 4 Nov 2016 20:40:58 +0000 (15:40 -0500)]
res_stasis: Set a video source mode on Stasis created bridges

When a bridge is created via ARI (through res_stasis), no video source
mode is set by default. As a result, any endpoint sending video media
won't ever see any video reflected back to it.

This patch defaults a bridge to a 'follow the talker' video mode.
Further work can be done to add routes that allow for the video mode to
be controlled through the /bridges resource.

Change-Id: I7e9d530a5d7a97a4524a9ee4e468e1a6b3443866

3 years agomain/bridge_channel: Fix channel reference leak on video source
Matt Jordan [Fri, 4 Nov 2016 20:37:57 +0000 (15:37 -0500)]
main/bridge_channel: Fix channel reference leak on video source

When a channel is made the video source, the bridge holds a reference to
it. Whenever the video source changes, that reference is released.
However, a ref leak does occur if the channel leaves the bridge (such as
being hung up) while it is the video source, as the bridge never
releases the ref in such a case.

This patch adds a line to the bridge_channel_internal_join routine such
that, when a channel finishes its time in the bridge, it notifies the
bridge via ast_bridge_remove_video_src that if it is a video source its
reference should be released.

ASTERISK-26555 #close

Change-Id: I3a2f5238a9d2fc49c591f0e65199d782ab0be76a

3 years agomain/bridge: Add some verbose logging for video source changes
Matt Jordan [Fri, 4 Nov 2016 20:36:42 +0000 (15:36 -0500)]
main/bridge: Add some verbose logging for video source changes

It's actually quite useful to see the source of a video stream change.
This doesn't happen terribly often, even with talk detection - but when
it does, it's nice to know which channel is now providing your video

As a verbose 5 level message, it shouldn't be terribly spammy or costly
to have, and is 'lower level' then most other verbose messages that the
bridge system emits.


Change-Id: Ia1c20ecafa9670171fd38bddcf3beccae47fb15c

3 years agobridges/bridge_softmix: Remove SSRC changes on join/leave; update video source
Matt Jordan [Fri, 4 Nov 2016 20:33:35 +0000 (15:33 -0500)]
bridges/bridge_softmix: Remove SSRC changes on join/leave; update video source

WebRTC clients really, really want to know the SSRC of the media they're
getting. Changing the SSRC is generally not a good thing.

bridge_softmix, starting in Asterisk 12, started changing the SSRC of
parties as they joined or left the bridge. With most phones, this isn't
a problem: phones just play back the stream they're getting. With WebRTC
clients, however, the SSRC is tied to a media stream that may be
negotiated. When a new SSRC just shows up, the media can be dropped.

As it turns out, the SSRC change shouldn't even be necessary. From the
perspective of the client, it's still talking to Asterisk with the same
media stream: why indicate that the far party has suddenly changed to a
different source of media?

This patch opts to just remove the SSRC changes. With this patch, video
clients that join/leave a softmix bridge actually get the video stream
instead of freaking out.


Change-Id: I27fec098b32e7c8718b4b65f3fd5fa73527968bf

3 years agostasis_recording/stored: remove calls to deprecated readdir_r function.
Kevin Harwell [Fri, 28 Oct 2016 20:11:35 +0000 (15:11 -0500)]
stasis_recording/stored: remove calls to deprecated readdir_r function.

The readdir_r function has been deprecated and should no longer be used. This
patch removes the readdir_r dependency (replaced it with readdir) and also moves
the directory search code to a more centralized spot (file.c)

Also removed a strict dependency on the dirent structure's d_type field as it
is not portable. The code now checks to see if the value is available. If so,
it tries to use it, but defaults back to using the stats function if necessary.

Lastly, for most implementations of readdir it *should* be thread-safe to make
concurrent calls to it as long as different directory streams are specified.
glibc falls into this category. However, since it is possible that there exist
some implementations that are not safe, locking has been added for those other
than glibc.

ASTERISK-26509 #close

Change-Id: Id8f54689b1e2873e82a09d0d0d2faf41964e80ba

3 years agoRevert "chan_sip: Fix lastrtprx always updated"
Kevin Harwell [Fri, 4 Nov 2016 15:57:43 +0000 (10:57 -0500)]
Revert "chan_sip: Fix lastrtprx always updated"

This reverts commit 93332cb1d0eea18021ea6538237297e627d6e2fc.

Unfortunately, the aforementioned commit caused a regression (incoming calls
would eventually disconnect). Thus it is being removed.

ASTERISK-26523 #close

Change-Id: Ibf5586adc303073a8eac667a4cbfdb6be184a64d

3 years agochan_ooh323: Fix infinite loop on read second part of H.225 packet
Alexander Anikin [Thu, 3 Nov 2016 18:45:37 +0000 (22:45 +0400)]
chan_ooh323: Fix infinite loop on read second part of H.225 packet

Fix logic on read second part of H.225 packet. There was infinite loop on
wrong connections due to read before poll.

Change-Id: I42b4bf75c46e4a5c5df5c5ca1f0bd74b8944e7ff

3 years agopjproject_bundled: Fix issue with libasteriskpj needing libresample
George Joseph [Thu, 3 Nov 2016 16:55:06 +0000 (10:55 -0600)]
pjproject_bundled:  Fix issue with libasteriskpj needing libresample

libresample is only needed by pjproject if we're building pjsua, which
we only do if TEST_FRAMEWORK is selected.  It's required by pjsua to
process audio which is needed by some testsuite tests.  Unfortunately,
pjproject relies on a newer version of libresample than the version
that ships by most distros so we need to compile the version that's
bundled with pjproject.  Since we only need it for pjsua, we DON'T want
it's symbols exposed when we actually build asterisk.

There was a problem however... TEST_FRAMEWORK is only known AFTER we've
already run ./configure on both asterisk and pjproject but pjproject's
./configure needs to test it to know whether to set up to build
libresample or not.  The previous way of figuring this out was to
always tell ./configure "yes" but not actually build the library.  This
caused an issue where building libasteriskpj was being told to include
libresample but it wasn't actually there.

The solution is to still do a default pjproject configure during an
asterisk ./configure but if makeopts or menuselect.makeopts changes
subsequently, we now reconfigure pjproject, taking into account the
current state of TEST_FRAMEWORK.  Previously, if makeopts or
menuselect.makeopts changed, only a recompile of pjproject was done.

Change-Id: I9b5d84c61384a3ae07fe30e85c49698378cc4685

3 years agoMerge "chan_sip: add missing account code"
Joshua Colp [Thu, 3 Nov 2016 10:39:33 +0000 (05:39 -0500)]
Merge "chan_sip: add missing account code"

3 years agoMerge "app_dial: Fix incorrect device state when channel is picked up."
zuul [Wed, 2 Nov 2016 19:08:37 +0000 (14:08 -0500)]
Merge "app_dial: Fix incorrect device state when channel is picked up."

3 years agoMerge "chan_dahdi: remove by_name support"
zuul [Wed, 2 Nov 2016 15:51:59 +0000 (10:51 -0500)]
Merge "chan_dahdi: remove by_name support"

3 years agochan_sip: add missing account code
Sebastian Gutierrez [Wed, 2 Nov 2016 00:48:50 +0000 (21:48 -0300)]
chan_sip: add missing account code

Added missing account to AMI event of sip show peers

ASTERISK-26176 #close

Change-Id: Ieb6c2c80a838a1b59c82103eba4c63ba238dc482

3 years agoapp_dial: Fix incorrect device state when channel is picked up.
Joshua Colp [Wed, 2 Nov 2016 14:15:14 +0000 (14:15 +0000)]
app_dial: Fix incorrect device state when channel is picked up.

Given the scenario where multiple channels are dialed using Dial()
but the caller is picked up using PickupChan() all outgoing channels
except the channel specified to PickupChan() would be marked
as ringing until the call had been hung up.

When using the PickupChan application the channel executing the
application is swapped into place of another channel. As part
of this process the channel is answered. The Dial application
has explicit logic which checks if the channel is answered,
cancels all other outgoing channels, and bridges. This logic is
different than the normal logic that is executed when an outgoing
channel is answered. This different logic failed to publish dial
events stating that the other outgoing channels had been canceled.
As a result references to the outgoing channels were held onto by
the dial masquerade process until the call had been ended and
the channels had gone away. This would result in the channels
appearing in the "core show channels" list despite not being present
anymore and would also result in incorrect device state.

This change makes it so that this logic also publishes
dial events stating that the other outgoing channels have been


Change-Id: Iea7168e6e82f7d4609ec0366153804e4f55ea64f

3 years agortp_engine: Allow more than 32 dynamic payload types.
Alexander Traud [Tue, 13 Sep 2016 09:08:34 +0000 (11:08 +0200)]
rtp_engine: Allow more than 32 dynamic payload types.

Since adding all remaining rates of Signed Linear (ASTERISK-24274), SILK
(Gerrit 3136) and Codec 2 (ASTERISK-26217), no RTP Payload Type is left in the
dynamic range (96-127). RFC 3551 section 3 allows to reassign other ranges.
Consequently, when the dynamic range is exhausted, this change utilizes payload
types in the range between 35 and 63 giving room for another 29 payload types.

ASTERISK-26311 #close

Change-Id: I7bc96ab764bc30098a178b841cbf7146f9d64964

3 years agoMerge "res_pjsip_sdp_rtp: Limit number of formats to defined maximum."
Joshua Colp [Wed, 2 Nov 2016 13:14:09 +0000 (08:14 -0500)]
Merge "res_pjsip_sdp_rtp: Limit number of formats to defined maximum."

3 years agoMerge "define PATH_MAX for HURD"
Joshua Colp [Wed, 2 Nov 2016 10:26:14 +0000 (05:26 -0500)]
Merge "define PATH_MAX for HURD"

3 years agoMerge "netsock.c: fix includes for HURD"
Joshua Colp [Wed, 2 Nov 2016 10:24:21 +0000 (05:24 -0500)]
Merge "netsock.c: fix includes for HURD"

3 years agoMerge "bundled pjproject: Fix DNS write to freed memory."
zuul [Wed, 2 Nov 2016 06:21:45 +0000 (01:21 -0500)]
Merge "bundled pjproject: Fix DNS write to freed memory."

3 years agoMerge "pjproject_bundled: Fix compile of pjsua so it handles audio"
zuul [Wed, 2 Nov 2016 02:15:08 +0000 (21:15 -0500)]
Merge "pjproject_bundled:  Fix compile of pjsua so it handles audio"

3 years agoMerge "codecs.conf.sample: Add sample and option descriptions for codec_opus"
Joshua Colp [Tue, 1 Nov 2016 23:58:28 +0000 (18:58 -0500)]
Merge "codecs.conf.sample: Add sample and option descriptions for codec_opus"

3 years agoMerge "res_pjsip_outbound_publish: Fix crash when publishing device state."
Joshua Colp [Tue, 1 Nov 2016 23:32:54 +0000 (18:32 -0500)]
Merge "res_pjsip_outbound_publish: Fix crash when publishing device state."

3 years agoMerge "res/stasis: Add CLI commands for displaying/debugging ARI apps"
Joshua Colp [Tue, 1 Nov 2016 21:05:03 +0000 (16:05 -0500)]
Merge "res/stasis: Add CLI commands for displaying/debugging ARI apps"

3 years agobundled pjproject: Fix DNS write to freed memory.
Richard Mudgett [Tue, 1 Nov 2016 18:13:13 +0000 (13:13 -0500)]
bundled pjproject: Fix DNS write to freed memory.

PJPROJECT 2.5.5 introduced a race condition with the -r5349 IPv6 DNS

The patch below fixes a write to freed memory under cartain DNS lookup


Reported by:  Richard Mudgett

Change-Id: Ifdfae9ecf1e41b53080f33aab44ce1a220f349c5

3 years agoMerge "chan_sip: Incorrect display option Outbound reg. retry 403"
zuul [Tue, 1 Nov 2016 19:09:00 +0000 (14:09 -0500)]
Merge "chan_sip: Incorrect display option Outbound reg. retry 403"

3 years agores_pjsip_sdp_rtp: Limit number of formats to defined maximum.
Joshua Colp [Tue, 1 Nov 2016 11:56:24 +0000 (11:56 +0000)]
res_pjsip_sdp_rtp: Limit number of formats to defined maximum.

The res_pjsip_sdp_rtp module did not restrict the number of
formats added to a media stream in the SDP to the defined
limit. If allow=all was used with additional loaded codecs this
could result in the next media stream being overwritten some.

This change restricts the module to limit it to the defined
maximum and also increases the maximum in our bundled pjproject.

ASTERISK-26541 #close

Change-Id: I0dc5f59d3891246cafa2f3df5ec406f088559ee8

3 years agocodecs.conf.sample: Add sample and option descriptions for codec_opus
Kevin Harwell [Mon, 31 Oct 2016 22:35:47 +0000 (17:35 -0500)]
codecs.conf.sample: Add sample and option descriptions for codec_opus

codecs.conf.sample was missing codec opus's configuration options, descriptions,
and examples. This patch adds the configuration options and examples to
codecs.conf.sample that can be used with codec_opus.

ASTERISK-26538 #close

Change-Id: I1d89bb5e01d3e3b5bd78951b8dd0ff077a83dc8b

3 years agores/stasis: Add CLI commands for displaying/debugging ARI apps
Matt Jordan [Thu, 20 Oct 2016 12:27:21 +0000 (07:27 -0500)]
res/stasis: Add CLI commands for displaying/debugging ARI apps

This patch adds three new CLI commands:
 - ari show apps: list the registered ARI applications
 - ari show app: show detailed information about an ARI application
 - ari set debug: dump events being sent to an ARI application

Note that while these CLI commands live in the res_stasis module, we use
the 'ari' family for these commands. This was done as most users of
Asterisk aren't aware of the semantic differences between ARI and
res_stasis, and some 'ari' CLI commands already exist.

ASTERISK-26488 #close

Change-Id: I51ad6ff0cabee0d69db06858c13f18b1c513c9f5

3 years agochan_sip: Incorrect display option Outbound reg. retry 403
Grachev Sergey [Tue, 1 Nov 2016 13:32:35 +0000 (16:32 +0300)]
chan_sip: Incorrect display option Outbound reg. retry 403

If in sip.conf (general section) set option register_retry_403=no,
the command "sip show settings" return value:
Outbound reg. retry 403:0
If in sip.conf (general section) set option register_retry_403=yes,
the command "sip show settings" return value:
Outbound reg. retry 403:-1

* In static char "sip show settings" for "Outbound.reg. retry 403"
option use AST_CLI_YESNO

ASTERISK-26476 #close

Change-Id: I3c14272f05f1067bd2aeaa8b3ef9cf8fcb12dcf9

3 years agonetsock.c: fix includes for HURD
Tzafrir Cohen [Tue, 1 Nov 2016 09:18:49 +0000 (11:18 +0200)]
netsock.c: fix includes for HURD


Change-Id: I43bf94d2d36d3d8a8d0df40cd6c027d65a462814

3 years agodefine PATH_MAX for HURD
Tzafrir Cohen [Tue, 1 Nov 2016 09:00:21 +0000 (11:00 +0200)]
define PATH_MAX for HURD

PATH_MAX is not guaranteed to be defined. In parctice, all but the HURD
define it to a constant. It is indeed not safe to assume there won't be
longer paths and Asterisk generally does err safely on such cases.

So even for HURD we'll just pretend PATH_MAX is 4096.

ASTERISK-25070 #close

Change-Id: I53d10ba18c34c132bcb640a5fd8e0da1d9b22db3

3 years agopjproject_bundled: Fix compile of pjsua so it handles audio
George Joseph [Mon, 31 Oct 2016 21:12:57 +0000 (15:12 -0600)]
pjproject_bundled:  Fix compile of pjsua so it handles audio

In order for pjsua and its python binding to actually negotiate
audio for the testsuite tests, it needs g711 and resample.  The
pj* libraries themselves do not.  Unfortunately, pjproject relies
on a brand new libresample that most distros don't ship so we need
to use the libresample already bundled with pjproject.  Only the pjsua
executable and the python library are linked with it so it
shouldn't interfere with asterisk itself.

Also it was pointed out that apply_patches couldn't handle multiple
patches that depended on each other during the dry-run, so the
dry-run was removed.

Change-Id: I24f397462b486dcdde0dcafe40e6c55a6593f098

3 years agomanager: Add documentation for NewConnectedLine event.
Etienne Lessard [Mon, 31 Oct 2016 18:46:54 +0000 (14:46 -0400)]
manager: Add documentation for NewConnectedLine event.

The NewConnectedLine event has been added by commit fe7671f, but the
documentation was missing.

ASTERISK-26537 #close

Change-Id: I7fc331f18caa28492da9303e576f70884ca8c9e6

3 years agoMerge "bundled pjproject: Crashes while resolving DNS names."
zuul [Mon, 31 Oct 2016 17:44:45 +0000 (12:44 -0500)]
Merge "bundled pjproject: Crashes while resolving DNS names."

3 years agoMerge "astobj2: Declare private variable data_size for AO2_DEBUG only."
zuul [Mon, 31 Oct 2016 16:35:16 +0000 (11:35 -0500)]
Merge "astobj2: Declare private variable data_size for AO2_DEBUG only."

3 years agovector: Prevent NULL argument to memcpy.
Corey Farrell [Sun, 30 Oct 2016 18:33:12 +0000 (14:33 -0400)]
vector: Prevent NULL argument to memcpy.

Headers declare that memcpy does not accept NULL argument for the first
two parameters.  Add a conditional block to prevent memcpy and ast_free
from running on vectors with NULL element array.

ASTERISK-26526 #close

Change-Id: I988a476bb5fcfcbd3f6d6c6b3e7769e4f9629b71

3 years agoastobj2: Declare private variable data_size for AO2_DEBUG only.
Corey Farrell [Sat, 29 Oct 2016 15:19:53 +0000 (11:19 -0400)]
astobj2: Declare private variable data_size for AO2_DEBUG only.

Every ao2 object contains storage for a private variable data_size,
though the value is never read if AO2_DEBUG is disabled.  This change
makes the variable conditional, reducing memory usage.

ASTERISK-26524 #close

Change-Id: If859929e507676ebc58b0f84247a4231e11da07f

3 years agobundled pjproject: Crashes while resolving DNS names.
Richard Mudgett [Fri, 28 Oct 2016 19:55:08 +0000 (14:55 -0500)]
bundled pjproject: Crashes while resolving DNS names.

PJPROJECT 2.5.5 introduced a race condition with the -r5349 IPv6 DNS

The patches below fix the DNS lookup race condition crash caused by
attempting to send the same message twice for the single DNS lookup.


The patch below removes a cached DNS response from the hash table when
another thread is referencing the old entry.  The table still contained
the entry when it was destroyed which can result in inexplicable crashes.


ASTERISK-26344 #close
Reported by: Ian Gilmour

ASTERISK-26387 #close
Reported by: Harley Peters

Change-Id: I17fde80359e66f65a91341ceca58d914d0f61cc4

3 years agopjproject_bundled: Fix issue where "/version.mak" wasn't found
George Joseph [Fri, 28 Oct 2016 21:59:19 +0000 (15:59 -0600)]
pjproject_bundled:  Fix issue where "/version.mak" wasn't found

main/Makefile includes third-party/pjproject/build.mak but
doesn't set PJDIR beforehand so "include $(PJDIR)/version.mak"
evaluates to "/version.mak".  Fix is to set PJDIR in main/Makefile
before the include.

Change-Id: I0f7c67d60209049056fe9c4b041bf0463aa95604

3 years agoMerge "Fix shutdown crash caused by modules being left open."
zuul [Fri, 28 Oct 2016 21:21:50 +0000 (16:21 -0500)]
Merge "Fix shutdown crash caused by modules being left open."

3 years agores_pjsip_outbound_publish: Fix crash when publishing device state.
mkrokosz [Fri, 28 Oct 2016 18:30:02 +0000 (14:30 -0400)]
res_pjsip_outbound_publish: Fix crash when publishing device state.

While publishing device state between multiple instances of Asterisk,
a crash will sporadically occur under high CPS which looks to be a
race condition operating on the publisher queue.


Change-Id: I28da25d346deb358eff1d563485cabc433ce1ed6

3 years agoFix shutdown crash caused by modules being left open.
Corey Farrell [Fri, 28 Oct 2016 02:49:43 +0000 (22:49 -0400)]
Fix shutdown crash caused by modules being left open.

It is only safe to run ast_register_cleanup callbacks when all modules
have been unloaded.  Previously these callbacks were run during graceful
shutdown, making it possible to crash during shutdown.

ASTERISK-26513 #close

Change-Id: Ibfa635bb688d1227ec54aa211d90d6bd45052e21

3 years agoSAC documentation: don't specify transports for endpoints and registrations
Rusty Newton [Fri, 28 Oct 2016 14:50:32 +0000 (09:50 -0500)]
SAC documentation: don't specify transports for endpoints and registrations

Removing explicit transport definition for endpoints and registrations. It
isn't necessary and isn't generally advised.

ASTERISK-26514 #close

Change-Id: Ifdec5e631962438a4683600968dfa4bfd15909fb

3 years agoMerge "Remove ASTERISK_REGISTER_FILE."
zuul [Fri, 28 Oct 2016 03:23:00 +0000 (22:23 -0500)]

3 years agoMerge "res_pjsip_sdp_rtp: Fix address family of explicit media_address."
zuul [Fri, 28 Oct 2016 03:22:57 +0000 (22:22 -0500)]
Merge "res_pjsip_sdp_rtp: Fix address family of explicit media_address."

3 years agoMerge "pjsip: Fix a few media bugs with reinvites and asymmetric payloads."
Joshua Colp [Fri, 28 Oct 2016 00:37:47 +0000 (19:37 -0500)]
Merge "pjsip: Fix a few media bugs with reinvites and asymmetric payloads."

3 years agoMerge "app_voicemail: Clear voice mailbox in MailboxExists and MAILBOX_EXISTS."
zuul [Thu, 27 Oct 2016 21:48:07 +0000 (16:48 -0500)]
Merge "app_voicemail: Clear voice mailbox in MailboxExists and MAILBOX_EXISTS."

3 years agochan_dahdi: remove by_name support
Tzafrir Cohen [Tue, 18 Oct 2016 14:06:38 +0000 (17:06 +0300)]
chan_dahdi: remove by_name support

Support for referring to DAHDI channels by logical names was added in
(FIXME: when? Asterisk 11? 1.8?) and was intended to be part of support
of refering to channels by name.

While technically usable, it has never been properly supported in
dahdi-tools, as using it would require many changes at the Asterisk
level. Instead logical mapping was added at the kernel level.

Thus it seems that refering to DAHDI channels by name is not really used
by anyone, and therefore should probably be removed.

Change-Id: I7d50bbfd9d957586f5cd06570244ef87bd54b485

3 years agoMerge "res_pjsip_caller_id: Fix crash on session timers UPDATE on inbound calls."
zuul [Thu, 27 Oct 2016 20:15:35 +0000 (15:15 -0500)]
Merge "res_pjsip_caller_id: Fix crash on session timers UPDATE on inbound calls."

3 years agopjproject_bundled: Remove usage of tar's --strip-components option
George Joseph [Wed, 26 Oct 2016 23:48:24 +0000 (17:48 -0600)]
pjproject_bundled:  Remove usage of tar's --strip-components option

Older versions of tar don't support the --strip-components option so
instead of doing 'tar --strip-components=1 -C source', we now just
untar to the tarball's root directory (pjproject-<version>) and
rename that directory to 'source'.

Also fixed an issue where the pjproject source directory is a hard
coded absolute pathname.

ASTERISK-26510 #close
ASTERISK-22480 #close

Change-Id: I9ec92952507a91ff4e4d01e0149e09fd8e8f32b0

Corey Farrell [Thu, 27 Oct 2016 02:40:49 +0000 (22:40 -0400)]

ASTERISK_REGISTER_FILE no longer has any purpose so this commit removes
all traces of it.

Previously exported symbols removed:
* __ast_register_file
* __ast_unregister_file
* ast_complete_source_filename

This also removes the mtx_prof static variable that was declared when
MTX_PROFILE was enabled.  This variable was only used in lock.c so it
is now initialized in that file only.

ASTERISK-26480 #close

Change-Id: I1074af07d71f9e159c48ef36631aa432c86f9966

3 years agores_pjsip_caller_id: Fix crash on session timers UPDATE on inbound calls.
Joshua Colp [Thu, 27 Oct 2016 13:07:02 +0000 (13:07 +0000)]
res_pjsip_caller_id: Fix crash on session timers UPDATE on inbound calls.

The res_pjsip_caller_id module wrongly assumed that a
saved From header would always exist on sessions. This
is true until an inbound call is received and a session
timer causes an UPDATE to be sent. In this case there will
be no saved From header and a crash will occur. This change
makes it fall back to the From header of the outgoing request
if no saved From header is present.

ASTERISK-26307 #close

Change-Id: Iccc3bc8d243b5ede9b81abf960292930c908d4fa

3 years agoMerge "test_astobj2_thrash: Fix multithreaded issues"
zuul [Wed, 26 Oct 2016 22:20:11 +0000 (17:20 -0500)]
Merge "test_astobj2_thrash:  Fix multithreaded issues"

3 years agoMerge "cdr_radius,cel_radius: Fix old memleak in unload"
zuul [Wed, 26 Oct 2016 16:13:43 +0000 (11:13 -0500)]
Merge "cdr_radius,cel_radius: Fix old memleak in unload"

3 years agoapp_voicemail: Clear voice mailbox in MailboxExists and MAILBOX_EXISTS.
Joshua Colp [Wed, 26 Oct 2016 12:51:50 +0000 (12:51 +0000)]
app_voicemail: Clear voice mailbox in MailboxExists and MAILBOX_EXISTS.

When executing the MailboxExists dialplan application and
MAILBOX_EXISTS dialplan function the passed in temporary voice
mailbox was not cleared, causing it to try to free garbage.

ASTERISK-26503 #close

Change-Id: Ie21ccfa1b80b9c59318e596f6b8e17da2b5a7cb3

3 years agopjsip: Fix a few media bugs with reinvites and asymmetric payloads.
Joshua Colp [Sun, 23 Oct 2016 12:38:59 +0000 (12:38 +0000)]
pjsip: Fix a few media bugs with reinvites and asymmetric payloads.

When channel format changes occurred as a result of an RTP
re-negotiation the bridge was not informed this had happened.
As a result the bridge technology was not re-evaluated and the
channel may have been in a bridge technology that was incompatible
with its formats. The bridge is now unbridged and the technology
re-evaluated when this occurs.

The chan_pjsip module also allowed asymmetric codecs for sending
and receiving. This did not work with all devices and caused one
way audio problems. The default has been changed to NOT do this
but to match the sending codec to the receiving codec. For users
who want asymmetric codecs an option has been added, asymmetric_rtp_codec,
which will return chan_pjsip to the previous behavior.

The codecs returned by the chan_pjsip module when queried by
the bridge_native_rtp module were also not reflective of the
actual negotiated codecs. The nativeformats are now returned as
they reflect the actual negotiated codecs.

ASTERISK-26423 #close

Change-Id: I6ec88c6e3912f52c334f1a26983ccb8f267020dc

3 years agores_pjsip_sdp_rtp: Fix address family of explicit media_address.
Joshua Colp [Wed, 26 Oct 2016 11:32:04 +0000 (11:32 +0000)]
res_pjsip_sdp_rtp: Fix address family of explicit media_address.

When an explicit media_address is provided the address family
in the SDP needs to be set to reflect it.


Change-Id: Ib9350cc91c120eb2f96f0623d3907d12af67eb79

3 years agotest_astobj2_thrash: Fix multithreaded issues
George Joseph [Tue, 25 Oct 2016 16:20:16 +0000 (10:20 -0600)]
test_astobj2_thrash:  Fix multithreaded issues

The test uses 4 threads to grow, count, lookup and shrink 15K objects
in a container.  If there's only 1 execution engine available, the test
will complete in <50ms.  If each threads gets its own execution engine,
the test may timeout after 60 seconds because the count thread does a
locked ao2_callback on the whole container in a tight loop with only
a sched_yield to give up time.  The lock contention makes the test
execution times wildly variable and mostly timeout.  2 execution
engines are OK, 3 results in about 33% failure rate and >=4 causes
a 80% failure rate.

To fix, the sched_yield was changed to a usleep(500).

Also, the number of buckets specified for the container was an even
number so that was changed to the next prime number greater than
(MAX_HASH_ENTRIES / 100).  That's 151 currently.

Change-Id: I50cd2344161ea61bfe4b96d2a29a6ccf88385c77

3 years agochan_pjsip: segfault on already disconnected session
Alexei Gradinari [Tue, 18 Oct 2016 14:04:54 +0000 (10:04 -0400)]
chan_pjsip: segfault on already disconnected session

On heavy loaded system the TCP/TLS incoming calls could be
disconnected by pjproject while these calls are being
processed by asterisk.

This patch uses functions pjsip_inv_add_ref/pjsip_inv_dec_ref
to inform pjproject that an INVITE session is in use.

ASTERISK-26482 #close

Change-Id: Ia2e3e2f75358cdb530252a9ce158af3d5d9fdf33

3 years agocdr_radius,cel_radius: Fix old memleak in unload
Badalyan Vyacheslav [Mon, 10 Oct 2016 16:49:08 +0000 (12:49 -0400)]
cdr_radius,cel_radius: Fix old memleak in unload

- Call "rc_openlog" optional. If you do not call,
you will simply NULL instead of a name.

- On the one PID can be only one syslog channel.
And it can already be run in logger.c

- Calling rc_openlog we assigns a new name for
the channel syslog. This unexpected behavior for logger.c.

Most lesser evil, is to agree on a NULL name syslog
if the channel was not launched in logger.c.

It also solves the problem of memory leaks.

ASTERISK-26455 #close

Change-Id: Ic17c38de67583e971d78fe18807d1a9faf8f0afd

3 years agoMerge "pjsip: Support dual stack automatically."
Joshua Colp [Tue, 25 Oct 2016 10:29:45 +0000 (05:29 -0500)]
Merge "pjsip: Support dual stack automatically."

3 years agoMerge "pjproject_bundled: Fixed various build issues"
Joshua Colp [Tue, 25 Oct 2016 10:28:48 +0000 (05:28 -0500)]
Merge "pjproject_bundled:  Fixed various build issues"

3 years agoMerge "ARI: Add duplicate channel ID checking for channel creation."
Joshua Colp [Tue, 25 Oct 2016 01:01:47 +0000 (20:01 -0500)]
Merge "ARI: Add duplicate channel ID checking for channel creation."

3 years agoMerge "ARI: Detect duplicate channel IDs"
Joshua Colp [Tue, 25 Oct 2016 01:01:43 +0000 (20:01 -0500)]
Merge "ARI: Detect duplicate channel IDs"

3 years agopjproject_bundled: Fixed various build issues
George Joseph [Mon, 24 Oct 2016 15:55:23 +0000 (09:55 -0600)]
pjproject_bundled:  Fixed various build issues

* CFLAGS is now properly set when using older gcc.
* All third-party pjproject targets have been removed.  This fixes
  an issue with older libsrtp in some distros.
* Manually removing the source directory now causes a rebuild.
* EXTERNALS_CACHE_DIR is now properly checked.
* Whitespace fixes.

Change-Id: I98fec6847efc5602a9f41cb95096fd660a49fa60

3 years agotypo: s/paranthesis/parenthesis/ in a comment
Pascal Cadotte Michaud [Mon, 24 Oct 2016 19:13:43 +0000 (15:13 -0400)]
typo: s/paranthesis/parenthesis/ in a comment

Change-Id: I7c1f4eb051177ee22cbe97e063d4a3effe29be30

3 years agopjsip: Support dual stack automatically.
Joshua Colp [Mon, 19 Sep 2016 11:13:21 +0000 (11:13 +0000)]
pjsip: Support dual stack automatically.

This change adds support for dual stack automatically. No
configuration is required and the IP address and version
in the SIP messages and SDP will be automatically changed
based on the transport over which the message is being
sent. RTP usage has also been changed to listen on both
IPv4 and IPv6 simultaneously to allow media to flow, and
to allow ICE support on both simultaneously. This also
allows failover between IPv6 and IPv4 to work as expected.

ASTERISK-26309 #close

Change-Id: I235a421d8f9a326606d861b449fa6fe3a030572d

3 years agoARI: Add duplicate channel ID checking for channel creation.
Mark Michelson [Wed, 19 Oct 2016 17:05:28 +0000 (12:05 -0500)]
ARI: Add duplicate channel ID checking for channel creation.

This is similar to what is done for origination, but for the 14 and up
channel creation method. When attempting to create a channel, if a
channel ID is specified and a channel already exists with that ID, then
a 409 is returned.

Change-Id: I77f9253278c6947939c418073b6b31065489187c

3 years agoARI: Detect duplicate channel IDs
Mark Michelson [Mon, 17 Oct 2016 19:18:57 +0000 (14:18 -0500)]
ARI: Detect duplicate channel IDs

ARI and AMI allow for an explicit channel ID to be specified
when originating channels. Unfortunately, there is nothing in
place to prevent someone from using the same ID for multiple
channels. Further complicating things, adding ID validation to channel
allocation makes it impossible for ARI to discern why channel allocation
failed, resulting in a vague error code being returned.

The fix for this is to institute a new method for channel errors to be
discerned. The method mirrors errno, in that when an error occurs, the
caller can consult the channel errno value to determine what the error
was. This initial iteration of the feature only introduces "unknown" and
"channel ID exists" errors. However, it's possible to add more errors as

ARI uses this feature to determine why channel allocation failed and can
return a 409 error during origination to show that a channel with the
given ID already exists.


Change-Id: Ibba7ae68842dab6df0c2e9c45559208bc89d3d06

3 years agoFix issue with CLI not returning to prompt after running "features show"
snuffy [Wed, 19 Oct 2016 22:53:24 +0000 (09:53 +1100)]
Fix issue with CLI not returning to prompt after running "features show"

ASTERISK-26444 #close

Change-Id: I91d645b7e6e5dba35f8c410df2be77a8c0e3acb8

3 years agoMerge "utils.c: Fix ast_set_default_eid for multiple platforms"
zuul [Wed, 19 Oct 2016 22:35:52 +0000 (17:35 -0500)]
Merge "utils.c:  Fix ast_set_default_eid for multiple platforms"

3 years agoMerge "res_rtp_asterisk: Add ice_blacklist option"
zuul [Wed, 19 Oct 2016 19:58:23 +0000 (14:58 -0500)]
Merge "res_rtp_asterisk: Add ice_blacklist option"

3 years agoMerge "chan_sip: Support nat=auto_comedia or nat=force_rport,auto_comedia."
Joshua Colp [Wed, 19 Oct 2016 16:06:41 +0000 (11:06 -0500)]
Merge "chan_sip: Support nat=auto_comedia or nat=force_rport,auto_comedia."

3 years agores_rtp_asterisk: Add ice_blacklist option
Michael Walton [Tue, 4 Oct 2016 23:24:54 +0000 (12:24 +1300)]
res_rtp_asterisk: Add ice_blacklist option

Introduces ice_blacklist configuration in rtp.conf. Subnets listed in the
form ice_blacklist = <subnet spec>, e.g. ice_blacklist =, are excluded from ICE host, srflx and relay
discovery. This is useful for optimizing the ICE process where a system
has multiple host address ranges and/or physical interfaces and certain
of them are not expected to be used for RTP. Multiple ice_blacklist
configuration lines may be used. If left unconfigured, all discovered
host addresses are used, as per previous behavior.

Documention in rtp.conf.sample.

ASTERISK-26418 #close

Change-Id: Ibee88f80d7693874fda1cceaef94a03bd86012c9

3 years agoCDR: Alter destruction pattern for CDR chains.
Mark Michelson [Tue, 18 Oct 2016 21:30:17 +0000 (16:30 -0500)]
CDR: Alter destruction pattern for CDR chains.

CDRs form chains. When the root of the chain is destroyed, it then
unreferences the next CDR in the chain. That CDR is destroyed, and it
then unreferences the next CDR in the chain. This repeats until the end
of the chain is reached. While this typically does not cause any sort of
problems, it is possible in strange scenarios for the CDR chain to grow
way longer than expected. In such a scenario, the destruction pattern
can result in a stack overflow.

This patch fixes the problem by switching from a recursive pattern to an
iterative pattern for destruction. When the root CDR is destroyed, it is
responsible for iterating over the rest of the CDRs and unreferencing
each one. Other CDRs in the chain, since they are not the root, will
simply destroy themselves and be done. This causes the stack depth not
to increase.

ASTERISK-26421 #close
Reported by Andrew Nagy

Change-Id: I3ca90c2b8051f3b7ead2e0e43f60d2c18fb204b8

3 years agoMerge "menuselect: invalid test for GTK2"
zuul [Tue, 18 Oct 2016 19:57:08 +0000 (14:57 -0500)]
Merge "menuselect: invalid test for GTK2"

3 years agoMerge "cli: Auto-complete File not Module for core set debug."
Joshua Colp [Tue, 18 Oct 2016 18:22:02 +0000 (13:22 -0500)]
Merge "cli: Auto-complete File not Module for core set debug."

3 years agoari: Update model validator based on addition of asterisk_id.
Joshua Colp [Tue, 18 Oct 2016 16:51:20 +0000 (16:51 +0000)]
ari: Update model validator based on addition of asterisk_id.


Change-Id: I9c386f7a1c7d969161b28f189eb6298bbc5b7541

3 years agoMerge "Binaural synthesis (confbridge): On/off setting for binaural synthesis."
Joshua Colp [Tue, 18 Oct 2016 16:38:59 +0000 (11:38 -0500)]
Merge "Binaural synthesis (confbridge): On/off setting for binaural synthesis."

3 years agoMerge "chan_rtp: Set a sane default rtp engine for unicast."
Joshua Colp [Tue, 18 Oct 2016 16:38:13 +0000 (11:38 -0500)]
Merge "chan_rtp: Set a sane default rtp engine for unicast."

3 years agomenuselect: invalid test for GTK2
Tzafrir Cohen [Sun, 11 Sep 2016 15:13:00 +0000 (10:13 -0500)]
menuselect: invalid test for GTK2 was only checking for the existence of pkg-config
and not the gtk2 package itself.  Now it calls AST_PKG_CONFIG_CHECK
for gtk+-2.0.

ASTERISK-26356 #close

Change-Id: I93e9d0166341f0e7f84b52955bb6f81da42f2ef6