asterisk/asterisk.git
8 months agoMerge "chan_pjsip: Transmit REFER waits for the REFER result setting TRANSFERSTATUS"
George Joseph [Tue, 23 Jul 2019 14:18:42 +0000 (09:18 -0500)]
Merge "chan_pjsip:  Transmit REFER waits for the REFER result setting TRANSFERSTATUS"

8 months agoCI: Add cleanWs to cleanup steps in jenkinsfiles
George Joseph [Fri, 19 Jul 2019 16:20:38 +0000 (10:20 -0600)]
CI: Add cleanWs to cleanup steps in jenkinsfiles

We're at the point where there are enough Jenkins jobs for
Asterisk branches than even cleaned checkouts of Asterisk
will add up to more disk space than is available on the
in-memory workspace mount.  Since we archive all relevent
artifacts anyway, there's no need to keep the workspace
around after the job finishes, whether it succeeds or fails.

Change-Id: I1cd3b73ebb045a987df0f62526d152a510210c39

8 months agoMerge "CI: Add install-headers to the install make targets"
George Joseph [Fri, 19 Jul 2019 16:04:50 +0000 (11:04 -0500)]
Merge "CI:  Add install-headers to the install make targets"

8 months agoMerge "README.md: Update year"
George Joseph [Fri, 19 Jul 2019 14:48:27 +0000 (09:48 -0500)]
Merge "README.md: Update year"

8 months agoMerge "sched: Don't allow ast_sched_del to deadlock ast_sched_runq from same thread"
George Joseph [Fri, 19 Jul 2019 13:46:21 +0000 (08:46 -0500)]
Merge "sched: Don't allow ast_sched_del to deadlock ast_sched_runq from same thread"

8 months agoCI: Add install-headers to the install make targets
George Joseph [Fri, 19 Jul 2019 13:38:39 +0000 (07:38 -0600)]
CI:  Add install-headers to the install make targets

The testsuite actually needs the headers installed to run
it's self_test.

Change-Id: Ice41d331131b876ad4a9c056085fe6aac34b32b2

8 months agoMerge "Build: Separate header install/uninstall"
George Joseph [Fri, 19 Jul 2019 12:54:28 +0000 (07:54 -0500)]
Merge "Build: Separate header install/uninstall"

8 months agoMerge "manager: Log AMI actions"
Joshua Colp [Fri, 19 Jul 2019 12:42:07 +0000 (07:42 -0500)]
Merge "manager: Log AMI actions"

8 months agosched: Don't allow ast_sched_del to deadlock ast_sched_runq from same thread
Walter Doekes [Wed, 17 Jul 2019 13:06:12 +0000 (15:06 +0200)]
sched: Don't allow ast_sched_del to deadlock ast_sched_runq from same thread

When fixing ASTERISK~24212, a change was done so a scheduled callback could not
be removed while it was running. The caller of ast_sched_del would have to wait.

However, when the caller of ast_sched_del is the callback itself (however wrong
this might be), this new check would cause a deadlock: it would wait forever
for itself.

This changeset introduces an additional check: if ast_sched_del is called
by the callback itself, it is immediately rejected (along with an ERROR log and
a backtrace). Additionally, the AST_SCHED_DEL_UNREF macro is adjusted so the
after-ast_sched_del-refcall function is only run if ast_sched_del returned
success.

This should fix the following spurious race condition found in chan_sip:
- thread 1: schedule sip_poke_peer_now (using AST_SCHED_REPLACE)
- thread 2: run sip_poke_peer_now
- thread 2: blank out sched-ID (too soon!)
- thread 1: set sched-ID (too late!)
- thread 2: try to delete the currently running sched-ID

After this fix, an ERROR would be logged, but no deadlocks (in do_monitor) nor
excess calls to sip_unref_peer(peer) (causing double frees of rtp_instances and
other madness) should occur.

(Thanks Richard Mudgett for reviewing/improving this "scary" change.)

Note that this change does not fix the observed race condition: unlocked
access to peer->pokeexpire (and potentially other scheduled items in chan_sip),
causing AST_SCHED_DEL_UNREF to look at a changing id. But it will make the
deadlock go away. And in the observed case, it will not have adverse affects
(like memory leaks) because the scheduled item is removed through a different
path.

ASTERISK-28282

Change-Id: Ic26777fa0732725e6ca7010df17af77a012aa856

8 months agoBuild: Separate header install/uninstall
George Joseph [Tue, 16 Jul 2019 12:55:49 +0000 (06:55 -0600)]
Build: Separate header install/uninstall

Asterisk headers are no longer installed and uninstalled
automatically when performing a "make install" or a
"make uninstall".  To install/uninstall the headers, use
"make install-headers" and "make uninstall-headers".
The headers also continue to be uninstalled when performing a
"make uninstall-all".

Also corrects an issue where /usr/include/asterisk.h was never
being removed at all.

Change-Id: Ia7399f3a0203a4825fc4a9f43b9034dae9a2b643

8 months agomanager: Log AMI actions
Kevin Harwell [Tue, 9 Jul 2019 19:42:51 +0000 (14:42 -0500)]
manager: Log AMI actions

When manager debugging is turned on, this patch makes it so incoming AMI actions
are now also logged.

Change-Id: I8047524510e7ac97d99482b2448f8e368f29cd47

8 months agores_rtp_asterisk: Move where DTLS MTU variable is defined.
Joshua Colp [Sun, 14 Jul 2019 18:26:41 +0000 (15:26 -0300)]
res_rtp_asterisk: Move where DTLS MTU variable is defined.

The DTLS MTU variable is not dependent on pjproject and should
not exist in its block.

Change-Id: I7e97d64dc192f2ac81bfe2b72b8229d321c7d026

8 months agoMerge "app_voicemail: Remove dependency on the stasis cache"
Kevin Harwell [Fri, 12 Jul 2019 14:21:15 +0000 (09:21 -0500)]
Merge "app_voicemail: Remove dependency on the stasis cache"

8 months agoMerge "MWI: Update modules that subscribe to MWI to use new API calls"
Kevin Harwell [Fri, 12 Jul 2019 14:19:18 +0000 (09:19 -0500)]
Merge "MWI: Update modules that subscribe to MWI to use new API calls"

8 months agoMerge "mwi: Update the MWI core to use stasis_state API"
Kevin Harwell [Fri, 12 Jul 2019 14:18:15 +0000 (09:18 -0500)]
Merge "mwi: Update the MWI core to use stasis_state API"

8 months agoMerge "stasis_state: Make unsubscribes NULL tolerant"
Kevin Harwell [Fri, 12 Jul 2019 14:17:55 +0000 (09:17 -0500)]
Merge "stasis_state: Make unsubscribes NULL tolerant"

8 months agoMerge "chan_sip: Handle invalid SDP answer to T.38 re-invite"
Friendly Automation [Thu, 11 Jul 2019 21:35:03 +0000 (16:35 -0500)]
Merge "chan_sip: Handle invalid SDP answer to T.38 re-invite"

8 months agores_pjsip_messaging: Check for body in in-dialog message
George Joseph [Wed, 12 Jun 2019 18:03:04 +0000 (12:03 -0600)]
res_pjsip_messaging:  Check for body in in-dialog message

We now check that a body exists and it has a length > 0 before
attempting to process it.

ASTERISK-28447
Reported-by: Gil Richard

Change-Id: Ic469544b22ab848734636588d4c93426cc6f4b1f

8 months agochan_sip: Handle invalid SDP answer to T.38 re-invite
Francesco Castellano [Fri, 28 Jun 2019 16:15:31 +0000 (18:15 +0200)]
chan_sip: Handle invalid SDP answer to T.38 re-invite

The chan_sip module performs a T.38 re-invite using a single media
stream of udptl, and expects the SDP answer to be the same.

If an SDP answer is received instead that contains an additional
media stream with no joint codec a crash will occur as the code
assumes that at least one joint codec will exist in this
scenario.

This change removes this assumption.

ASTERISK-28465

Change-Id: I8b02845b53344c6babe867a3f0a5231045c7ac87

8 months agoapp_voicemail: Remove dependency on the stasis cache
Kevin Harwell [Wed, 12 Jun 2019 18:49:30 +0000 (13:49 -0500)]
app_voicemail: Remove dependency on the stasis cache

app_voicemail utilized the stasis cache when polling mailboxes for MWI. This
caused a memory leak (items were not being appropriately removed from the
cache), and subsequent slowdown in system processing. This patch removes the
stasis cache dependency, thus alleviating the memory leak. It does this by
utilizing the new MWI API that better manages state lifetime.

ASTERISK-28443
ASTERISK-27121

Change-Id: Ie89fedaca81ea1fd03d150d9d3a1ef3d53740e46

8 months agoMWI: Update modules that subscribe to MWI to use new API calls
Kevin Harwell [Wed, 12 Jun 2019 18:11:42 +0000 (13:11 -0500)]
MWI: Update modules that subscribe to MWI to use new API calls

The MWI core recently got some new API calls that make tracking MWI state
lifetime more reliable. This patch updates those modules that subscribe to
specific MWI topics to use the new API. Specifically, these modules now
subscribe to both MWI topics and MWI state.

ASTERISK-28442

Change-Id: I32bef880b647246823dbccdf44a98d384fcabfbd

8 months agomwi: Update the MWI core to use stasis_state API
Kevin Harwell [Tue, 11 Jun 2019 19:12:12 +0000 (14:12 -0500)]
mwi: Update the MWI core to use stasis_state API

** Note **

This patch is meant to be the minimum needed in order for the MWI core to use
the now underlying stasis_state module. As such it does not completely remove
its reliance on the stasis_cache. Doing so has allowed current consumers to
not have to change, and update those code paths for this patch. When time
allows, subsequent patches can/will be made to those consumers to take advantage
of some of the new MWI API included here. Thus, eventually and ultimately
removing MWI dependency on the stasis_cache.

** End Note **

This patch makes it so the MWI core now takes advantage of the new stasis_state
API. Consumers of MWI should no longer need to depend upon stasis topic pooling,
and the stasis cache directly. Similar functionality and implementation details
have now been pushed into the stasis_state module. However, all MWI state should
be accessed via the MWI API itself.

As such a few new methods, and constructs have been added to the MWI core that
facilitate consumer publishing, subscribing, and iterating over MWI state data.

* ast_mwi_subscriber *

Created via ast_mwi_add_subscriber, a subscriber subscribes to a given mailbox
in order to receive updates about the given mailbox. Adding a subscriber will
create the underlying topic, and associated state data if those do not already
exist for it. The topic, and last known state data is guaranteed to exist for
the lifetime of the subscriber.

* ast_mwi_publisher *

Before publishing to a particular topic a publisher should be created. This can
be achieved by using ast_mwi_add_publisher. Publishing to a mailbox should then
be done using one of the MWI publish functions. This ensures the message is
published to the appropriate topic, and the last known state is maintained.

* ast_mwi_observer *

Add an observer in order to watch for particular MWI module related events. For
instance if a submodule needs to know when a subscription is added to any
mailbox an observer can be added to watch for that.

* other *

Urgent message count is now part of the published MWI state object. Also state
can be iterated over using defined callbacks.

ASTERISK-28442

Change-Id: I93f935f9090cd5ddff6d4bc80ff90703c05cf776

8 months agostasis_state: Make unsubscribes NULL tolerant
Kevin Harwell [Mon, 8 Jul 2019 23:10:07 +0000 (18:10 -0500)]
stasis_state: Make unsubscribes NULL tolerant

Regular stasis unsubscribes can handle NULL subscription objects. This patch
makes it so stasis state unsubscribes handles NULL's as well.

ASTERISK-28442

Change-Id: Ic3648e8df043a85b77cff085e9ff10356028e479

8 months agoREADME.md: Update year
Rodrigo Ramírez Norambuena [Fri, 5 Jul 2019 00:46:36 +0000 (20:46 -0400)]
README.md: Update year

Change-Id: I746fb94d112c7d797e206bca0fd1e13fcd26bae3

9 months agoMerge "stasis_state: Add new stasis_state module"
Friendly Automation [Tue, 2 Jul 2019 14:30:35 +0000 (09:30 -0500)]
Merge "stasis_state: Add new stasis_state module"

9 months agoMerge "chan_dahdi.c: crash in chan_dahdi"
Joshua Colp [Tue, 2 Jul 2019 13:25:41 +0000 (08:25 -0500)]
Merge "chan_dahdi.c: crash in chan_dahdi"

9 months agochan_dahdi.c: crash in chan_dahdi
Chris-Savinovich [Mon, 1 Jul 2019 21:57:25 +0000 (16:57 -0500)]
chan_dahdi.c: crash in chan_dahdi

Fixes a crash in chan_dahdi occurring on 32-bit systems. A previous
patch introduced a variable of type unassigned long long which is 64-bits.
Casting it as 'ast_json_int_t' along with JSON type 'I' makes it work
with 32-bit systems.

ASTERISK-28457

Change-Id: I9cef6b5f2d826fc5c93f2f6a1c997c4e3e6c93fe

9 months agores_pjsip_sdp_rtp: Remove unused variable
Kevin Harwell [Mon, 1 Jul 2019 15:49:56 +0000 (10:49 -0500)]
res_pjsip_sdp_rtp: Remove unused variable

The variable 'endpoint_caps' in function 'set_caps' is not used, so remove.

ASTERISK-28458

Change-Id: Ia8766d05a0738aecb29dd018302c2dafca5cab34

9 months agoMerge "app_voicemail.c: Build all three variants for app_voicemail at the same time"
George Joseph [Mon, 1 Jul 2019 15:20:43 +0000 (10:20 -0500)]
Merge "app_voicemail.c: Build all three variants for app_voicemail at the same time"

9 months agoMerge "tcptls.c: Add peer hostname and port to some error messages"
George Joseph [Mon, 1 Jul 2019 15:20:11 +0000 (10:20 -0500)]
Merge "tcptls.c:  Add peer hostname and port to some error messages"

9 months agoMerge "pjproject_bundled: Add peer information to most SSL/TLS errors"
Friendly Automation [Mon, 1 Jul 2019 15:05:26 +0000 (10:05 -0500)]
Merge "pjproject_bundled:  Add peer information to most SSL/TLS errors"

9 months agostasis_state: Add new stasis_state module
Kevin Harwell [Tue, 11 Jun 2019 17:30:27 +0000 (12:30 -0500)]
stasis_state: Add new stasis_state module

This new module describes an API that can be thought of as a combination of
stasis topic pools, and caching. Except, hopefully done in a more efficient
and less memory "leaky" manner.

The API defines methods, and data structures for managing, and tracking
published message state through stasis. By adding a subscriber or publisher,
consumers can more easily track the lifetime of the contained state. For
instance, when no more publishers and/or subscribers have need of the topic,
and associated state its data is removed from the managed container.

* stasis_state_manager *

The manager stores and well, manages state data. Each state is an association
of a unique stasis topic, and the last known published stasis message on that
topic. There is only ever one managed state object per topic. For each topic
all messages are forwarded to an "all" topic also maintained by the manager.

* stasis_state_subscriber *

Topic and state can be created, or referenced within the manager by adding a
stasis_state_subscriber. When adding a subscriber if no state currently exists
new managed state is immediately created. If managed state already exists then
a new subscriber is created referencing that state. The managed state is
guaranteed to live throughout the subscriber's lifetime. State is only removed
from the manager when no other entities require it.

* stasis_state_publisher *

Topic and state can be created, or referenced within the manager by also adding
a stasis_state_publisher. When adding a publisher if no state currently exists
new managed state is created. If managed state already exists then a new
publisher is created referencing that state. The managed state is guaranteed to
live throughout the publisher's lifetime. State is only removed from the
manager when no other entities require it.

* stasis_state_observer *

Some modules may wish to watch for, and react to managed state events. By
registering a state observer, and implementing handlers for the desired
callbacks those modules can do so.

* other *

Callbacks also exist that allow consumers to iterate over all, or some of the
managed state.

ASTERISK-28442

Change-Id: I7a4a06685a96e511da9f5bd23f9601642d7bd8e5

9 months agoapp_voicemail.c: Build all three variants for app_voicemail at the same time
Chris-Savinovich [Thu, 27 Jun 2019 18:50:57 +0000 (13:50 -0500)]
app_voicemail.c: Build all three variants for app_voicemail at the same time

Changes made to apps/Makefile to optionally build all three app_voicemail
variations at the same time: 1) file (default), 2) odbc, and 3) imap.
This functionality was requested by users. modules.conf.sample warns the
user to make sure only one voicemail is loaded at a time.

Change-Id: Iba3cd8ffb4b7e8b1c64a11dd383e1eafcd3ed0e7

9 months agoMerge "pjproject: Update to 2.9 release"
Kevin Harwell [Thu, 27 Jun 2019 21:52:59 +0000 (16:52 -0500)]
Merge "pjproject: Update to 2.9 release"

9 months agotcptls.c: Add peer hostname and port to some error messages
George Joseph [Thu, 27 Jun 2019 20:04:27 +0000 (14:04 -0600)]
tcptls.c:  Add peer hostname and port to some error messages

Where possble, hostname and port has been added to error
messages, mostly on the server side.

ASTERISK-26006
Reported by: Oleksandr Natalenko

Change-Id: Iff4f897277bc36ce8c5b493b71d0a4a7b74e62f0

9 months agopjproject_bundled: Add peer information to most SSL/TLS errors
George Joseph [Thu, 27 Jun 2019 17:46:44 +0000 (11:46 -0600)]
pjproject_bundled:  Add peer information to most SSL/TLS errors

Most SSL/TLS error messages coming from pjproject now have either
the peer address:port or peer hostname, depending on what was
available at the time and code location where the error was
generated.

ASTERISK-28444
Reported by: Bernhard Schmidt

Change-Id: I41770e8a1ea5e96f6e16b236692c4269ce1ba91e

9 months agoMerge "res/ari/resource_channels.c: Added hangup reason code for channels"
Friendly Automation [Thu, 27 Jun 2019 17:03:35 +0000 (12:03 -0500)]
Merge "res/ari/resource_channels.c: Added hangup reason code for channels"

9 months agoMerge "app_amd: issue with silence suppression fixed"
Kevin Harwell [Thu, 27 Jun 2019 16:33:22 +0000 (11:33 -0500)]
Merge "app_amd: issue with silence suppression fixed"

9 months agores/ari/resource_channels.c: Added hangup reason code for channels
sungtae kim [Mon, 15 Apr 2019 23:26:46 +0000 (01:26 +0200)]
res/ari/resource_channels.c: Added hangup reason code for channels

Currently, DELETE /ari/channels/<channelID> supports only few hangup reasons.
It's good enough for simple use, but when it needs to set the detail reason,
it comes challenges.
Added reason_code query parameter for that.

ASTERISK-28385

Change-Id: I1cf1d991ffd759d0591b347445a55f416ddc3ff2

9 months agoMerge "sig_pri: Address gcc9 issues"
George Joseph [Tue, 25 Jun 2019 14:55:17 +0000 (09:55 -0500)]
Merge "sig_pri:  Address gcc9 issues"

9 months agochan_pjsip: Transmit REFER waits for the REFER result setting TRANSFERSTATUS
Dan Cropp [Tue, 2 Apr 2019 19:42:44 +0000 (14:42 -0500)]
chan_pjsip:  Transmit REFER waits for the REFER result setting TRANSFERSTATUS

Previously, when a Transfer (REFER) was performed, chan_pjsip would set
the TRANSFERSTATUS to SUCCESS when the REFER was queued up.  This did not
reflect a successful/unsuccessful transfer the way chan_sip did.
Added a callback module to process the refer subscription information.

Now depends on res_pjsip_pubsub so call transfer progress can be monitored
and reported

ASTERISK-26968 #close
Reported-by: Dan Cropp

Change-Id: If6c27c757c66f71e8b75e3fe49da53ebe62395dc

9 months agoMerge "CI: New way to determnine libdir"
George Joseph [Tue, 25 Jun 2019 14:07:12 +0000 (09:07 -0500)]
Merge "CI:  New way to determnine libdir"

9 months agoMerge "res_fax: gateway sends T.38 request to both endpoints if V.21 detected"
George Joseph [Mon, 24 Jun 2019 20:16:36 +0000 (15:16 -0500)]
Merge "res_fax: gateway sends T.38 request to both endpoints if V.21 detected"

9 months agosig_pri: Address gcc9 issues
George Joseph [Mon, 24 Jun 2019 13:30:19 +0000 (07:30 -0600)]
sig_pri:  Address gcc9 issues

A few more format truncation issues addressed.

Change-Id: I047f373169caaca0eec4889d3c0e5e10f130017a

9 months agoMerge "translate.c do not log WARNING on empty audio frame"
George Joseph [Fri, 21 Jun 2019 18:41:29 +0000 (13:41 -0500)]
Merge "translate.c do not log WARNING on empty audio frame"

9 months agoMerge "app_confbridge: Attended transfer event fixup"
Friendly Automation [Fri, 21 Jun 2019 16:24:35 +0000 (11:24 -0500)]
Merge "app_confbridge:  Attended transfer event fixup"

9 months agoapp_amd: issue with silence suppression fixed
Nasir Iqbal [Tue, 21 May 2019 06:38:24 +0000 (11:38 +0500)]
app_amd: issue with silence suppression fixed

Now AMD algorithm will not ignore AST_FRAME_NULL, As I think using manual
wait time instead of `framelength` is enough to fix timeout / TOOLONG issue.

ASTERISK-28419 #close

Change-Id: I16ea2d6295bc99b975e8c092e5f9fbd9214debdb

9 months agores_fax: gateway sends T.38 request to both endpoints if V.21 detected
Alexei Gradinari [Wed, 29 May 2019 22:54:16 +0000 (18:54 -0400)]
res_fax: gateway sends T.38 request to both endpoints if V.21 detected

According T.38 Gateway 'Use case 3'
https://wiki.asterisk.org/wiki/display/AST/T.38+Gateway
T.38 Gateway should send T.38 negotiation request to called endpoint
if FAX preamble (using V.21 detector) generated by called endpoint.
But it does not, because fax_gateway_detect_v21 constructs T.38
negotiation request, but forwards it only to other channel,
not to the channel on which FAX preamble is detected.

Some SIP endpoints could be improperly configured to rely on the other side
to initiate T.38 re-INVITEs.

With this patch the T.38 Gateway tries to negotiate with both sides
by sending T.38 negotiation request to both endpoints supported T.38.

Change-Id: I73bb24799bfe1a48adae9c034a2edbae54cc2a39

9 months agoCI: New way to determnine libdir
George Joseph [Wed, 19 Jun 2019 16:58:39 +0000 (10:58 -0600)]
CI:  New way to determnine libdir

We were using the presence of /usr/lib64 to determine where
shared libraries should be installed.  This only existed on
Redhat based systems and was safe.  If it existed, use it,
otherwise use /usr/lib.

Unfortunately, Ubuntu 19 decided to create a /usr/lib64 BUT
NOT INCLUDE IT IN THE DEFAULT ld.so.conf.  So if anything is
installed there, it won't work.

The new method, just looks for $ID in /etc/os-release and if it's
centos or fedora, uses /usr/lib64 and if ubuntu, uses /usr/lib.

NOTE:  This applies only to the CI scripts.  Normal asterisk
build and install is not affected.

Change-Id: Iad66374b550fd89349bedbbf2b93f8edd195a7c3

9 months agotranslate.c do not log WARNING on empty audio frame
Alexei Gradinari [Fri, 14 Jun 2019 20:45:39 +0000 (16:45 -0400)]
translate.c do not log WARNING on empty audio frame

There is WARNING "no samples for ..." on each Playtones.
The function ast_playtones_start calls ast_activate_generator,
which calls ast_prod.
The function ast_prod calls ast_write with empty audio frame.
In this case it's spam log.

Change-Id: Id4ac309489d9ff281bad02abdef341cecdede660

9 months agochan_dahdi: Address gcc9 issues
George Joseph [Mon, 17 Jun 2019 17:11:49 +0000 (11:11 -0600)]
chan_dahdi:  Address gcc9 issues

Fixed format-truncation issues in chan_dahdi.c and
sig_analog.c.  Since they're related to fields provided
by dahdi-tools we can't change the buffer sizes so we're just
checking the return from snprintf and printing an errior if we
overflow.

Change-Id: Idc1f3c1565b88a7d145332a0196074b5832864e5

9 months agoapp_confbridge: Attended transfer event fixup
George Joseph [Mon, 10 Jun 2019 21:58:59 +0000 (15:58 -0600)]
app_confbridge:  Attended transfer event fixup

When a channel already in a conference bridge is attended transfered
to another extension, or when an existing call is attended
transferred into a conference bridge, we now generate ConfbridgeJoin
and ConfbridgeLeave events for the entering and departing channels.

Change-Id: Id7709cfbceb26fbcb828b2d0d2a6b2fbeaf028e1

9 months agopjproject: Update to 2.9 release
Sean Bright [Thu, 13 Jun 2019 15:11:48 +0000 (11:11 -0400)]
pjproject: Update to 2.9 release

Relies on https://github.com/asterisk/third-party/pull/4

Change-Id: Iec9cad42cb4ae109a86a3d4dae61e8bce4424ce3

9 months agores_rtp_asterisk: Add support for DTLS packet fragmentation.
Joshua Colp [Tue, 11 Jun 2019 12:26:42 +0000 (09:26 -0300)]
res_rtp_asterisk: Add support for DTLS packet fragmentation.

This change adds support for larger TLS certificates by allowing
OpenSSL to fragment the DTLS packets according to the configured
MTU. By default this is set to 1200.

This is accomplished by implementing our own BIO method that
supports MTU querying. The configured MTU is returned to OpenSSL
which fragments the packet accordingly. When a packet is to be
sent it is done directly out the RTP instance.

ASTERISK-28018

Change-Id: If2d5032019a28ffd48f43e9e93ed71dbdbf39c06

9 months agoMerge "app_attended_transfer: new application AttendedTransfer"
George Joseph [Wed, 12 Jun 2019 15:44:06 +0000 (10:44 -0500)]
Merge "app_attended_transfer: new application AttendedTransfer"

9 months agoMerge "app_blind_transfer: new application BlindTransfer"
Friendly Automation [Wed, 12 Jun 2019 14:31:36 +0000 (09:31 -0500)]
Merge "app_blind_transfer: new application BlindTransfer"

9 months agoMerge "chan_pjsip.c: Check for channel and session to not be NULL in hangup"
George Joseph [Wed, 12 Jun 2019 13:50:01 +0000 (08:50 -0500)]
Merge "chan_pjsip.c: Check for channel and session to not be NULL in hangup"

9 months agoapp_attended_transfer: new application AttendedTransfer
Alexei Gradinari [Tue, 21 May 2019 19:12:55 +0000 (15:12 -0400)]
app_attended_transfer: new application AttendedTransfer

AttendedTransfer queues up attended transfer to the given extension.

This application can be useful with Custom Dynamic Features.
For example to make attended transfer to a predefined number.

features.conf
;;;
[applicationmap]
my_atxfer => *7,self,GoSub,"my_atxfer,s,1",default
;;;

extensions.conf
;;;
[globals]
DYNAMIC_FEATURES=my_atxfer
TRANSFER_CONTEXT=my_transfer

[my_atxfer]
exten => s,1,AttendedTransfer(1234567890)
   same => n,Return()

[my_transfer]
include => default
;;;

This application also can be used to completly redefine Attended transfer
feature using dialplan. For example:

features.conf
;;;
[featuremap]
atxfer => *7

[applicationmap]
custom_atxfer => *2,self,GoSub,"custom_atxfer,s,1",default
;;;

extensions.conf
;;;
[globals]
DYNAMIC_FEATURES=custom_atxfer
TRANSFER_CONTEXT=my_transfer

[custom_atxfer]
exten => s,1,
   same => n,Playback(pbx-transfer)
   same => n,Read(dest,dial,10,i,3,3)
   same => n,AttendedTransfer(${dest})
   same => n,Return()

[my_transfer]
include => default
;;;

Change-Id: Ie5cfa455d0813cffd5c85a6fb117f07d8f0b903b

9 months agoMerge "cdr_pgsql: fix error in connection string"
Friendly Automation [Tue, 11 Jun 2019 13:03:09 +0000 (08:03 -0500)]
Merge "cdr_pgsql: fix error in connection string"

9 months agochan_pjsip.c: Check for channel and session to not be NULL in hangup
agupta [Thu, 6 Jun 2019 12:48:18 +0000 (18:18 +0530)]
chan_pjsip.c: Check for channel and session to not be NULL in hangup

We have seen some rare case of segmentation fault in hangup function
and we could notice that channel pointer was NULL.  Debug log shows
that there is a 200 OK answer and SIP timeout at the same time.  It
looks that while the SIP session was being destroyed due to timeout
call hangup due to answer event lead to race condition and channel
is being destroyed from two different places.  The check ensures we
check it not to be NULL before freeing it.

ASTERISK-25371

Change-Id: I19f6566830640625e08f7b87bfe15758ad33a778

9 months agoapp_blind_transfer: new application BlindTransfer
Alexei Gradinari [Tue, 21 May 2019 19:53:47 +0000 (15:53 -0400)]
app_blind_transfer: new application BlindTransfer

BlindTransfer redirects all channels currently bridged to the
caller channel to the specified destination.

This application can be useful with Custom Dynamic Features.
For example to make blind transfer to a predefined number.

features.conf
;;;
[applicationmap]
my_blindxfer => *6,self,GoSub,"my_blindxfer,s,1",default
;;;

extensions.conf
;;;
[globals]
DYNAMIC_FEATURES=my_blindxfer

[my_blindxfer]
exten => s,1,BlindTransfer(1234567890,default)
   same => n,Return()
;;;

This application also can be used to completly redefine Blind transfer
feature using dialplan. For example:

features.conf
;;;
[featuremap]
blindxfer =>

[applicationmap]
custom_blindxfer => ##,self,GoSub,"custom_blindxfer,s,1",default
;;;

extensions.conf
;;;
[globals]
DYNAMIC_FEATURES=custom_blindxfer

[custom_blindxfer]
exten => s,1,
   same => n,Playback(pbx-transfer)
   same => n,Read(dest,dial,10,i,3,3)
   same => n,BlindTransfer(${dest},default)
   same => n,Return()
;;;

Change-Id: I9d55e7f69ccfd4472dec00d62771d6de8803215a

9 months agopbx_dundi: added IPv4/IPv6 dual bind support to DUNDi
Kirsty Tyerman [Tue, 8 Jan 2019 06:14:07 +0000 (16:14 +1000)]
pbx_dundi: added IPv4/IPv6 dual bind support to DUNDi

ASTERISK-28234
Reported-by: Kirsty Tyerman

Change-Id: I5d6e6b52dbe51415046bb3953fd16f5b421bc2e1

9 months agocdr_pgsql: fix error in connection string
Chris-Savinovich [Tue, 4 Jun 2019 17:41:33 +0000 (12:41 -0500)]
cdr_pgsql: fix error in connection string

Fixes an error occurring in function pgsql_reconnect() caused when value of
hostname is blank. Which in turn will cause the connection string to look
like this: "host= port=xx", which creates a sintax error. This fix now checks
if the corresponding values for host, port, dbname, and user are blank. Note
that since this is a reconnect function the database library will replace any
missing value pairs with default ones.

ASTERISK-28435

Change-Id: I0a921f99bbd265768be08cd492f04b30855b8423

9 months agoMerge "res_fax: fix segfault on inactive "reserved" fax session"
Friendly Automation [Tue, 4 Jun 2019 10:07:14 +0000 (05:07 -0500)]
Merge "res_fax: fix segfault on inactive "reserved" fax session"

9 months agoMerge "app_readexten: new option 'p' to stop reading on '#' key"
Friendly Automation [Mon, 3 Jun 2019 15:05:11 +0000 (10:05 -0500)]
Merge "app_readexten: new option 'p' to stop reading on '#' key"

9 months agoMerge "pjsip: replace 180 by 183 if SDP negotiation has completed"
George Joseph [Mon, 3 Jun 2019 14:36:43 +0000 (09:36 -0500)]
Merge "pjsip: replace 180 by 183 if SDP negotiation has completed"

9 months agoMerge "res_fax: add channel name to CLI 'fax show session'"
Friendly Automation [Mon, 3 Jun 2019 14:29:12 +0000 (09:29 -0500)]
Merge "res_fax: add channel name to CLI 'fax show session'"

9 months agores_fax: fix segfault on inactive "reserved" fax session
Alexei Gradinari [Tue, 28 May 2019 20:35:17 +0000 (16:35 -0400)]
res_fax: fix segfault on inactive "reserved" fax session

The change #10017 "Handle fax gateway being started more than once"
introdiced a bug which leads to segfault in res_fax_spandsp.

The res_fax_spandsp module does not support reserving sessions, so
fax_session_reserve returns a fax session with state AST_FAX_STATE_INACTIVE.

The fax_gateway_start does not create a real fax session if the fax session
is already present and the state is not AST_FAX_STATE_RESERVED.
But the "reserved" session created for res_fax_spandsp has state
AST_FAX_STATE_INACTIVE, so fax_gateway_start not starting.

Then when fax_gateway_framehook is called and gateway T.38 state is
NEGOTIATED the call of gateway->s->tech->write(gateway->s, f) leads to
segfault, because session tech_pvt is not set, i.e. the tech session
was not initialized/started.

This patch adds check also on AST_FAX_STATE_INACTIVE to the "reserved"
session created for res_fax_spandsp will start.

This patch also adds extra check and log ERROR if tech_pvt is not set
before call tech->write.

ASTERISK-27981 #close

Change-Id: Ife3e65e5f18c902db2ff0538fccf7d28f88fa803

10 months agores_fax: add channel name to CLI 'fax show session'
Alexei Gradinari [Tue, 28 May 2019 22:15:40 +0000 (18:15 -0400)]
res_fax: add channel name to CLI 'fax show session'

This patch adds a channel name to output of CLI 'fax show session'
and also expands the channel name field up to 30 characters on
CLI 'fax show sessions'

Change-Id: Id059c43ff41811f5e76712b83fb63b8f246da953

10 months agobuild: Fix file format in CHANGES-staging.
Ben Ford [Fri, 24 May 2019 14:01:14 +0000 (09:01 -0500)]
build: Fix file format in CHANGES-staging.

One of the change files doesn't conform to the format that the release
scripts need in order to parse it.

Change-Id: Ie0b634cf27e4cbc671b9fe92993b6f2ecf60254c

10 months agochan_dahdi: add missing include.
Guido Falsi [Thu, 23 May 2019 14:44:07 +0000 (16:44 +0200)]
chan_dahdi: add missing include.

After some definitions have been moved to asterisk/mwi.h the files
channels/chan_dahdi.h channels/sig_pri.c are missing this new
include.

ASTERISK-28427 #close

Change-Id: Ia8cc595eeda653324643f40dcd9799d4c3f0ac91

10 months agoapp_readexten: new option 'p' to stop reading on '#' key
Alexei Gradinari [Fri, 17 May 2019 22:45:25 +0000 (18:45 -0400)]
app_readexten: new option 'p' to stop reading on '#' key

This patch adds the 'p' option.
The extension entered will be considered complete when a # is entered.

Change-Id: If77c40c9c8b525885730821e768f5dea71cf04c1

10 months agoMerge "res_rtp_asterisk: timestamp should be unsigned instead of signed int"
Friendly Automation [Thu, 23 May 2019 14:03:49 +0000 (09:03 -0500)]
Merge "res_rtp_asterisk: timestamp should be unsigned instead of signed int"

10 months agoMerge "pjproject-bundled: Add upstream timer fixes"
Friendly Automation [Wed, 22 May 2019 17:07:50 +0000 (12:07 -0500)]
Merge "pjproject-bundled:  Add upstream timer fixes"

10 months agoMerge "res_rtp_asterisk: Add ability to propose local address in ICE"
Friendly Automation [Wed, 22 May 2019 16:28:18 +0000 (11:28 -0500)]
Merge "res_rtp_asterisk:  Add ability to propose local address in ICE"

10 months agores_prometheus: Add metrics for PJSIP outbound registrations
Matt Jordan [Fri, 10 May 2019 14:36:01 +0000 (09:36 -0500)]
res_prometheus: Add metrics for PJSIP outbound registrations

When monitoring Asterisk instances, it's often useful to know when an
outbound registration fails, as this often maps to the notion of a trunk
and having a trunk fail is usually a "bad thing". As such, this patch
adds monitoring metrics that track the state of PJSIP outbound registrations.
It does this by looking for the Registry events coming across the Stasis
system topic, and publishing those as metrics to Prometheus. Note that
while this may support other outbound registration types (IAX2, SIP, etc.)
those haven't been tested. Your mileage may vary.

(And why are you still using IAX2 and SIP? It's 2019 folks. Get with the
program.)

This patch also adds Sorcery observers to handle modifications to the
underlying PJSIP outbound registration objects. This is useful when a
reload is triggered that modifies the properties of an outbound registration,
or when ARI push configuration is used and an object is updated or
deleted. Because we rely on properties of the registration object to
define the metric (label key/value pairs), we delete the relevant metric when
we notice that something has changed and wait for a new Stasis message to
arrive to re-create the metric.

ASTERISK-28403

Change-Id: If01420e38530fc20b6dd4aa15cd281d94cd2b87e

10 months agores_prometheus: Add CLI commands
Matt Jordan [Thu, 3 Jan 2019 16:28:28 +0000 (10:28 -0600)]
res_prometheus: Add CLI commands

This patch adds a few CLI commands to the res_prometheus module to aid
system administrators setting up and configuring the module. This includes:

* prometheus show status: Display basic statistics about the Prometheus
  module, including its essential configuration, when it was last scraped,
  and how long the scrape took. The last two bits of information are useful
  when Prometheus isn't generating metrics appropriately, as it will at
  least tell you if Asterisk has had its HTTP route hit by the remote
  server.

* prometheus show metrics: Dump the current metrics to the CLI. Useful for
  system administrators to see what metrics are currently available without
  having to cURL or go to Prometheus itself.

ASTERISK-28403

Change-Id: Ic09813e5e14b901571c5c96ebeae2a02566c5172

10 months agores_prometheus: Add Asterisk bridge metrics
Matt Jordan [Thu, 9 May 2019 14:41:49 +0000 (09:41 -0500)]
res_prometheus: Add Asterisk bridge metrics

This patch adds basic Asterisk bridge statistics to the res_prometheus
module. This includes:

* asterisk_bridges_count: The current number of bridges active on the
  system.

* asterisk_bridges_channels_count: The number of channels active in a
  bridge.

In all cases, enough information is provided with each bridge metric
to determine a unique instance of Asterisk that provided the data, along
with the technology, subclass, and creator of the bridge.

ASTERISK-28403

Change-Id: Ie27417dd72c5bc7624eb2a7a6a8829d7551788dc

10 months agores_prometheus: Add Asterisk endpoint metrics
Matt Jordan [Thu, 9 May 2019 14:41:02 +0000 (09:41 -0500)]
res_prometheus: Add Asterisk endpoint metrics

This patch adds basic Asterisk endpoint statistics to the res_prometheus
module. This includes:

* asterisk_endpoints_state: The current state (unknown, online, offline)
  for each defined endpoint.

* asterisk_endpoints_channels_count: The current number of channels
  associated with a given endpoint.

* asterisk_endpoints_count: The current number of defined endpoints.

In all cases, enough information is provided with each endpoint metric
to determine a unique instance of Asterisk that provided the data, as well
as the underlying technology and resource definition.

ASTERISK-28403

Change-Id: I46443963330c206a7d12722d08dcaabef672310e

10 months agores_rtp_asterisk: timestamp should be unsigned instead of signed int
Morten Tryfoss [Tue, 21 May 2019 16:29:05 +0000 (18:29 +0200)]
res_rtp_asterisk: timestamp should be unsigned instead of signed int

Using timestamp with signed int will cause timestamps exceeding max value
to be negative.
This causes the jitterbuffer to do passthrough of the packet.

ASTERISK-28421

Change-Id: I9dabd0718180f2978856c50f43aac4e52dc3cde9

10 months agores_prometheus: Add Asterisk channel metrics
Matt Jordan [Fri, 3 May 2019 00:45:27 +0000 (19:45 -0500)]
res_prometheus: Add Asterisk channel metrics

This patch adds basic Asterisk channel statistics to the res_prometheus
module. This includes:

* asterisk_calls_sum: A running sum of the total number of
  processed calls

* asterisk_calls_count: The current number of calls

* asterisk_channels_count: The current number of channels

* asterisk_channels_state: The state of any particular channel

* asterisk_channels_duration_seconds: How long a channel has existed,
  in seconds

In all cases, enough information is provided with each channel metric
to determine a unique instance of Asterisk that provided the data, as
well as the name, type, unique ID, and - if present - linked ID of each
channel.

ASTERISK-28403

Change-Id: I0db306ec94205d4f58d1e7fbabfe04b185869f59

10 months agopjproject/Makefile: Updates for Darwin compatible builds
Matt Jordan [Mon, 29 Apr 2019 15:10:35 +0000 (10:10 -0500)]
pjproject/Makefile: Updates for Darwin compatible builds

This patch fixes three compatibility issues for Darwin compatible builds:

(1) Use BSD compatible command line option for sed

For some versions of BSD sed, the -r command line option is unknown.
Both GNU and BSD sed support the -E command line option for enabling
extended regular expressions; as such, this patch replaces the -r
option with -E.

(2) Look for '_' in pjproject generated symbols

In Darwin comaptible systems, the symbols generated for pjproject may be
prefixed with an '_'. When exporting these to a symbol file, the invocation
to sed has to optionally look for a prefix of said '_' character.

(3) Use -all_load/-noall_load when linking

The flags -whole-archive/-no-whole-archive are not supported by the
linker, and must instead be replaced with -all_load/-noall_load.

Change-Id: I58121756de6a0560a6e49ca9d6bf9566a333cde3

10 months agoMerge "Add core Prometheus support to Asterisk"
Friendly Automation [Tue, 21 May 2019 15:11:04 +0000 (10:11 -0500)]
Merge "Add core Prometheus support to Asterisk"

10 months agoAdd core Prometheus support to Asterisk
Matt Jordan [Thu, 3 Jan 2019 16:28:28 +0000 (10:28 -0600)]
Add core Prometheus support to Asterisk

Prometheus is the defacto monitoring tool for containerized applications.
This patch adds native support to Asterisk for serving up Prometheus
compatible metrics, such that a Prometheus server can scrape an Asterisk
instance in the same fashion as it does other HTTP services.

The core module in this patch provides an API that future work can build
on top of. The API manages metrics in one of two ways:
(1) Registered metrics. In this particular case, the API assumes that
    the metric (either allocated on the stack or on the heap) will have
    its value updated by the module registering it at will, and not
    just when Prometheus scrapes Asterisk. When a scrape does occur,
    the metrics are locked so that the current value can be retrieved.
(2) Scrape callbacks. In this case, the API allows consumers to be
    called via a callback function when a Prometheus initiated scrape
    occurs. The consumers of the API are responsible for populating
    the response to Prometheus themselves, typically using stack
    allocated metrics that are then formatted properly into strings
    via this module's convenience functions.

These two mechanisms balance the different ways in which information is
generated within Asterisk: some information is generated in a fashion
that makes it appropriate to update the relevant metrics immediately;
some information is better to defer until a Prometheus server asks for
it.

Note that some care has been taken in how metrics are defined to
minimize the impact on performance. Prometheus's metric definition
and its support for nesting metrics based on labels - which are
effectively key/value pairs - can make storage and managing of metrics
somewhat tricky. While a naive approach, where we allow for any number
of labels and perform a lot of heap allocations to manage the information,
would absolutely have worked, this patch instead opts to try to place
as much information in length limited arrays, stack allocations, and
vectors to minimize the performance impacts of scrapes. The author of
this patch has worked on enough systems that were driven to their knees
by poor monitoring implementations to be a bit cautious.

Additionally, this patch only adds support for gauges and counters.
Additional work to add summaries, histograms, and other Prometheus
metric types may add value in the future. This would be of particular
interest if someone wanted to track SIP response types.

Finally, this patch includes unit tests for the core APIs.

ASTERISK-28403

Change-Id: I891433a272c92fd11c705a2c36d65479a415ec42

10 months agopjproject-bundled: Add upstream timer fixes
Joshua Colp [Mon, 20 May 2019 17:45:57 +0000 (14:45 -0300)]
pjproject-bundled:  Add upstream timer fixes

Fixed #2191:
  - Stricter double timer entry scheduling prevention.
  - Integrate group lock in SIP transport, e.g: for add/dec ref,
    for timer scheduling.

ASTERISK-28161
Reported-by: Ross Beer

Change-Id: I2e09aa66de0dda9414d8a8259a649c4d2d96a9f5

10 months agores_rtp_asterisk: Add ability to propose local address in ICE
George Joseph [Fri, 17 May 2019 23:44:37 +0000 (17:44 -0600)]
res_rtp_asterisk:  Add ability to propose local address in ICE

You can now add the "include_local_address" flag to an entry in
rtp.conf "[ice_host_candidates]" to include both the advertized
address and the local address in ICE negotiation:

[ice_host_candidates]
192.168.1.1 = 1.2.3.4,include_local_address

This causes both 192.168.1.1 and 1.2.3.4 to be advertized.

Change-Id: Ide492cd45ce84546175ca7d557de80d9770513db

10 months agopjsip: replace 180 by 183 if SDP negotiation has completed
Alexei Gradinari [Mon, 13 May 2019 20:37:50 +0000 (16:37 -0400)]
pjsip: replace 180 by 183 if SDP negotiation has completed

The caller endpoint hears dead silence if a callee replies 180 (without SDP)
and the caller already received 183 (with SDP).
It happens because Asterisk sends 180 (WITH SDP) to the caller,
there are not incoming RTP packets from the callee
and Asterisk does not generate inband ringing,
so there are not any outgoing RTP packets to the caller.

This patch replaces 180 by 183 if SDP negotiation has completed,
as if the caller endpoint is configured with "inband_progress=yes".

In this case Asterisk will generate inband ringing untill Asterisk receive
incoming RTP packets from the callee.

ASTERISK-27994 #close

Change-Id: I7450b751083ec30d68d6abffe922215a15ae5a73

10 months agoMerge "res_rtp_asterisk: Fix sequence number cycling and packet loss count."
Friendly Automation [Wed, 15 May 2019 22:45:56 +0000 (17:45 -0500)]
Merge "res_rtp_asterisk: Fix sequence number cycling and packet loss count."

10 months agoMerge "conversions.c: Add conversions for largest max sized integer"
Friendly Automation [Wed, 15 May 2019 12:04:37 +0000 (07:04 -0500)]
Merge "conversions.c: Add conversions for largest max sized integer"

10 months agoMerge "Fixes for GCC 9"
Friendly Automation [Wed, 15 May 2019 11:29:03 +0000 (06:29 -0500)]
Merge "Fixes for GCC 9"

10 months agoMerge "build: Pass --fno-partial-inlining to third-party when appropriate"
Friendly Automation [Wed, 15 May 2019 10:45:36 +0000 (05:45 -0500)]
Merge "build: Pass --fno-partial-inlining to third-party when appropriate"

10 months agoMerge "pjsip_options.c: Allow immediate qualifies for new contacts."
Friendly Automation [Mon, 13 May 2019 19:11:08 +0000 (14:11 -0500)]
Merge "pjsip_options.c: Allow immediate qualifies for new contacts."

10 months agoFixes for GCC 9
George Joseph [Fri, 10 May 2019 15:48:28 +0000 (09:48 -0600)]
Fixes for GCC 9

Various fixes for issues caught by gcc 9.  Mostly snprintf
trying to copy to a buffer potentially too small.

ASTERISK-28412

Change-Id: I9e85a60f3c81d46df16cfdd1c329ce63432cf32e

10 months agores_rtp_asterisk: Fix sequence number cycling and packet loss count.
Joshua Colp [Wed, 8 May 2019 15:41:43 +0000 (15:41 +0000)]
res_rtp_asterisk: Fix sequence number cycling and packet loss count.

This change fixes two bugs which both resulted in the packet loss
count exceeding 65,000.

The first issue is that the sequence number check to determine if
cycling had occurred was using the wrong variable resulting in the
check never seeing that cycling has occurred, throwing off the
packet loss calculation. It now uses the correct variable.

The second issue is that the packet loss calculation assumed that
the received number of packets in an interval could never exceed
the expected number. In practice this isn't true due to delayed
or retransmitted packets. The expected will now be updated to
the received number if the received exceeds it.

ASTERISK-28379

Change-Id: If888ebc194ab69ac3194113a808c414b014ce0f6

10 months agopjsip_options.c: Allow immediate qualifies for new contacts.
Ben Ford [Tue, 7 May 2019 16:08:33 +0000 (11:08 -0500)]
pjsip_options.c: Allow immediate qualifies for new contacts.

When multiple endpoints try to register close together using the same
AOR with qualify_frequency set, one contact would qualify immediately
while the other contacts would have to wait out the duration of the
timer before being able to qualify. Changing the conditional to check
the contact container count for a non-zero value allows all contacts to
qualify immediately.

Change-Id: I79478118ee7e0d6e76af7c354d66684220db9415

10 months agoconversions.c: Add conversions for largest max sized integer
Kevin Harwell [Mon, 6 May 2019 21:26:46 +0000 (16:26 -0500)]
conversions.c: Add conversions for largest max sized integer

Added a conversion for umax (largest maximum sized integer allowed). Adjusted
the other current conversion functions (uint and ulong) to be derivatives of
the umax conversion since they are simply subsets of umax.

Also made the negative check move the pointer on spaces since strtoumax does it
anyways.

Change-Id: I56c2ef2629d49b524c8df58af12951c181f81f08

10 months agostasis: Hangup channel for Local channel No such extension error
agupta [Fri, 3 May 2019 15:49:31 +0000 (21:19 +0530)]
stasis: Hangup channel for Local channel No such extension error

When we use early bridge with create and dial from stasis using Local channel
and the dialplan does not any entry the it is returned from core_local.c with
No such extension .

In such case asterisk locks up till the channel is not hangup with the error
Exceptionally long voice queue length

* Found that in such case app_control_dial fails on ast_call method and
  return -1
* Since it is called from stasis_app_send_command_async and return -1 does
  not cause resources to be freed and since no PBX exist it is not able to
  read from channel causing exceptionally long queue
* After putting this code found that the channel was releasing immediately
  and resources were freed.

ASTERISK-28399
Reported by: Abhay Gupta
Tested by: Abhay Gupta

Change-Id: I0a55c923fc6995559f808d63b9488762b4489318

10 months agobuild: Pass --fno-partial-inlining to third-party when appropriate
George Joseph [Fri, 3 May 2019 18:31:06 +0000 (12:31 -0600)]
build: Pass --fno-partial-inlining to third-party when appropriate

When the gcc version is >= 8.2.1, we were already setting the
--fno-partial-inlining flag for Asterisk source files to get around
a gcc bug but we weren't passing the flag down to the bundled
builds of pjproject and jansson.

ASTERISK-28392

Change-Id: I99ede9bc35408ecd096f7d5369e8192d3dc75704

10 months agoMerge "app_confbridge: Add "all" variants of REMB behavior."
Friendly Automation [Fri, 3 May 2019 15:54:13 +0000 (10:54 -0500)]
Merge "app_confbridge: Add "all" variants of REMB behavior."

10 months agoMerge "stasis: Only place stasis created and dialed channels into dial bridge."
Joshua Colp [Fri, 3 May 2019 15:50:38 +0000 (10:50 -0500)]
Merge "stasis: Only place stasis created and dialed channels into dial bridge."