4 years agoMerge "rtp_engine: allocate RTP dynamic payloads per session"
zuul [Fri, 24 Mar 2017 21:22:55 +0000 (16:22 -0500)]
Merge "rtp_engine: allocate RTP dynamic payloads per session"

4 years agoMerge "pjproject_bundled: raise timeout value used when downloading"
zuul [Fri, 24 Mar 2017 20:42:48 +0000 (15:42 -0500)]
Merge "pjproject_bundled: raise timeout value used when downloading"

4 years agoMerge "res_xmpp: Correct implementation of JABBER_STATUS & JabberStatus"
zuul [Fri, 24 Mar 2017 17:25:07 +0000 (12:25 -0500)]
Merge "res_xmpp: Correct implementation of JABBER_STATUS & JabberStatus"

4 years agoMerge "res_xmpp: Include client name in connection related error messages"
zuul [Fri, 24 Mar 2017 16:55:38 +0000 (11:55 -0500)]
Merge "res_xmpp: Include client name in connection related error messages"

4 years agoMerge "res_xmpp: Don't crash when trying to send a message without a connection"
Joshua Colp [Fri, 24 Mar 2017 15:46:34 +0000 (10:46 -0500)]
Merge "res_xmpp: Don't crash when trying to send a message without a connection"

4 years agoMerge "res_xmpp: Correctly check return value of SSL_connect"
zuul [Fri, 24 Mar 2017 14:13:06 +0000 (09:13 -0500)]
Merge "res_xmpp: Correctly check return value of SSL_connect"

4 years agoMerge "res_xmpp: Try to provide useful errors messages from OpenSSL"
zuul [Fri, 24 Mar 2017 14:12:57 +0000 (09:12 -0500)]
Merge "res_xmpp: Try to provide useful errors messages from OpenSSL"

4 years agoMerge "audiohook.c: Lost RTP packets lead to out-of-sync MixMonitor."
Joshua Colp [Fri, 24 Mar 2017 12:25:07 +0000 (07:25 -0500)]
Merge "audiohook.c: Lost RTP packets lead to out-of-sync MixMonitor."

4 years agoAMI: Updated version
Kevin Harwell [Thu, 23 Mar 2017 19:01:40 +0000 (14:01 -0500)]
AMI: Updated version

Updated the AMI version for the following reason (see CHANGES for more details):

The 'PJSIPShowEndpoint' command's response event of 'IdentifyDetail' now
contains a new optional parameter, 'MatchHeader'.

Change-Id: Ie206913ef1dcfa6a2ebe3282da2387e52d6f05b9

4 years agopjproject_bundled: raise timeout value used when downloading
Kevin Harwell [Thu, 23 Mar 2017 17:07:09 +0000 (12:07 -0500)]
pjproject_bundled: raise timeout value used when downloading

After configuring Asterisk with '--with-pjproject-bundled' the configure/build
process attempts to download pjproject from its download site. Currently, a
timeout of 10 seconds is used that will stop the download process if pjproject
has not been fully downloaded in that time. For some systems this was not enough
time and the process was timing out too early.

This patch raises the download timeout value to '60'. Also, this patch fixes
another bug where the DOWNLOAD_TIMEOUT variable was not being properly exported
due to a naming error. DOWNLOAD_MAX_TIMEOUT is now properly renamed to

ASTERISK-26814 #close

Change-Id: Ia56e4e8a3d39db76bc8a1852b2cf07ec10b39842

4 years agores_xmpp: Correct implementation of JABBER_STATUS & JabberStatus
Sean Bright [Thu, 23 Mar 2017 01:33:02 +0000 (21:33 -0400)]
res_xmpp: Correct implementation of JABBER_STATUS & JabberStatus

The documentation for JABBER_STATUS (and the deprecated JabberStatus
app) indicate that a return value of 7 indicates that the specified
buddy was not in the roster. It also indicates that you can specify a
"bare" JID (one without a resource). Unfortunately the actual behavior
does not match the documented behavior.

Assuming that our roster includes the buddy online and available
"" and does *not* include the buddy
"", the JABBER_STATUS() function returns the
following before this patch:

| Buddy                        | Status     | Result                   |
|            |  Online    |  7 (Not in roster)       |
|      |  Online    |  1 (Online)              |
|    |  N/A       |  7 (Not in roster)       |
|          |  N/A       |  Error logged, no return |
|    |  N/A       |  Error logged, no return |

And after this patch:

| Buddy                        | Status     | Result                   |
|            |  Online    |  1 (Online)              |
|      |  Online    |  1 (Online)              |
|    |  N/A       |  6 (Offline)             |
|          |  N/A       |  7 (Not in roster)       |
|    |  N/A       |  7 (Not in roster)       |

This brings the behavior in line with the documentation.

ASTERISK-23510 #close
Reported by: Anthony Critelli

Change-Id: I9c3241035363ef4a6bdc21fabfd8ffcd9ec657bf

4 years agores_xmpp: Try to provide useful errors messages from OpenSSL
Sean Bright [Thu, 23 Mar 2017 14:45:35 +0000 (10:45 -0400)]
res_xmpp: Try to provide useful errors messages from OpenSSL

If any errors occur during the TLS connection setup, we currently dump a
fairly generic error message. So instead we try to pull in something
useful from OpenSSL to report instead.

Reported by: Matthias Urlichs

Change-Id: I288500991a9681f447d92913b11fedaf426087f4

4 years agores_xmpp: Correctly check return value of SSL_connect
Sean Bright [Thu, 23 Mar 2017 14:30:18 +0000 (10:30 -0400)]
res_xmpp: Correctly check return value of SSL_connect

SSL_connect returns non-zero for both success and some error conditions
so simply negating is inadequate.

Change-Id: Ifbf882896e598703b6c615407fa456d3199f95b1

4 years agores_xmpp: Don't crash when trying to send a message without a connection
Sean Bright [Wed, 22 Mar 2017 22:32:37 +0000 (18:32 -0400)]
res_xmpp: Don't crash when trying to send a message without a connection

If we never establish a connection to our Jabber server, iksemel never sets up
its internal transport pointer, so attempting to send a message dereferences a
NULL pointer and causes a crash.

ASTERISK-21855 #close
Reported by: Jeremy Kister

Change-Id: I204a568894e4a53ab929783ecc594a000f04d79c

4 years agores_xmpp: Include client name in connection related error messages
Sean Bright [Wed, 22 Mar 2017 20:40:29 +0000 (16:40 -0400)]
res_xmpp: Include client name in connection related error messages

ASTERISK-25622 #close
Reported by: Sean Darcy

Change-Id: I8472cb7bfb58d411a3cfbd482da98cae2d94d1e9

4 years agoMerge "res_pjsip_session: Enable RFC3578 overlap dialing support."
Joshua Colp [Wed, 22 Mar 2017 22:08:08 +0000 (17:08 -0500)]
Merge "res_pjsip_session: Enable RFC3578 overlap dialing support."

4 years agoMerge "CHANNEL(callid): Give dialplan access to the callid."
Joshua Colp [Wed, 22 Mar 2017 20:49:42 +0000 (15:49 -0500)]
Merge "CHANNEL(callid): Give dialplan access to the callid."

4 years agortp_engine: allocate RTP dynamic payloads per session
Kevin Harwell [Mon, 20 Mar 2017 18:27:31 +0000 (13:27 -0500)]
rtp_engine: allocate RTP dynamic payloads per session

Dynamic payload types were statically defined in Asterisk. This unfortunately
limited the number of dynamic payloads that could be registered. With this patch
dynamic payload type numbers are now assigned dynamically and per RTP instance.
However, in order to limit any issues where some clients expect the old
statically defined value this patch makes it so the value Asterisk used to pre-
designate is used for the dynamic assignment if available.

An option, "rtp_use_dynamic", has also been added (can be set in asterisk.conf)
that turns the new dynamic behavior on or off. When off it reverts back to using
statically defined payload values. This option defaults to "yes" in Asterisk 15.

ASTERISK-26515 #close
  ASTERISK-26515.diff submitted by jcolp (license 5000

Change-Id: I7653465c5ebeaf968f1a1cc8f3f4f5c4321da7fc

4 years agoMerge "res_pjsip_messaging: Check URI type before dereferencing"
zuul [Wed, 22 Mar 2017 17:36:43 +0000 (12:36 -0500)]
Merge "res_pjsip_messaging: Check URI type before dereferencing"

4 years agoMerge "Revert "app_queue: Handle the caller being redirected out of a queue bridge""
zuul [Wed, 22 Mar 2017 15:54:56 +0000 (10:54 -0500)]
Merge "Revert "app_queue: Handle the caller being redirected out of a queue bridge""

4 years agoMerge "app_queue: Member stuck as pending after forwarding previous call from queue"
zuul [Wed, 22 Mar 2017 14:50:22 +0000 (09:50 -0500)]
Merge "app_queue: Member stuck as pending after forwarding previous call from queue"

4 years agoMerge "pjsip: prevent memory corruption on creation of xml bodies"
zuul [Wed, 22 Mar 2017 13:32:06 +0000 (08:32 -0500)]
Merge "pjsip: prevent memory corruption on creation of xml bodies"

4 years agores_pjsip_session: Enable RFC3578 overlap dialing support.
Richard Begg [Tue, 14 Mar 2017 21:45:06 +0000 (08:45 +1100)]
res_pjsip_session: Enable RFC3578 overlap dialing support.

Support for RFC3578 overlap dialling (i.e. 484 Response to partially matched
destinations) as currently provided by chan_sip is missing from res_pjsip.
This patch adds a new endpoint attribute (allow_overlap) [defaults to yes]
which when set to yes enables 484 responses to partial destination
matches rather than the current 404.


Change-Id: Iea444da3ee7c7d4f1fde1d01d138a3d7b0fe40f6

4 years agoMerge "autochan/mixmonitor/chanspy: Fix unsafe channel locking and references."
zuul [Wed, 22 Mar 2017 02:51:49 +0000 (21:51 -0500)]
Merge "autochan/mixmonitor/chanspy: Fix unsafe channel locking and references."

4 years agoMerge "res_hep: Capture actual transport type in use"
zuul [Wed, 22 Mar 2017 00:47:16 +0000 (19:47 -0500)]
Merge "res_hep: Capture actual transport type in use"

4 years agores_hep: Capture actual transport type in use
Sean Bright [Tue, 21 Mar 2017 11:59:12 +0000 (07:59 -0400)]
res_hep: Capture actual transport type in use

Rather than hard-coding UDP, allow consumers of the HEP API to specify
which protocol is in use. Update the PJSIP provider to pass in the
current protocol type.

ASTERISK-26850 #close

Change-Id: I54bbb0a001cfe4c6a87ad4b6f2014af233349978

4 years agoRevert "app_queue: Handle the caller being redirected out of a queue bridge"
Sean Bright [Tue, 21 Mar 2017 14:57:46 +0000 (08:57 -0600)]
Revert "app_queue: Handle the caller being redirected out of a queue bridge"

This reverts commit 163e9e53dc7d84dd42721e733b7706c8147bdd27.

Change-Id: Ief28479c77a298879dfe2c56be7ee92dc465da4b

4 years agores_pjsip_messaging: Check URI type before dereferencing
Sean Bright [Tue, 21 Mar 2017 13:26:28 +0000 (09:26 -0400)]
res_pjsip_messaging: Check URI type before dereferencing

We aren't validating that the URI we just parsed is a SIP/SIPS one before
trying to access the user, host, and port members of a possibly uninitialized

Also update the MessageSend documentation to indicate what 'from' formats are

ASTERISK-26484 #close
Reported by: Vinod Dharashive

Change-Id: I476b5cc5f18a7713d0ee945374f2a1c164857d30

4 years agopjsip: prevent memory corruption on creation of xml bodies
Joshua Elson [Mon, 13 Mar 2017 20:21:23 +0000 (14:21 -0600)]
pjsip: prevent memory corruption on creation of xml bodies

ASTERISK-26776 #close

Change-Id: I884b6f4e8233a355d0be687ec78d41bc0e4d3fd2

4 years agobridge_softmix: Ignore non-voice frames from translator
Sean Bright [Mon, 20 Mar 2017 21:27:24 +0000 (17:27 -0400)]
bridge_softmix: Ignore non-voice frames from translator

Some codecs - codec_speex specifically - take voice frames and return
other types of frames, like CNG. If we subsequently treat those as
voice frames, we'll run into trouble when destroying the frame because
of the requirement that each voice frame have an associated format.

ASTERISK-26880 #close
Reported by: Kirsty Tyerman

Change-Id: I43f8450c48fb276ad8b99db8512be82949c1ca7c

4 years agoMerge "res/res_pjsip_session: Only check localnet if it is defined"
Joshua Colp [Mon, 20 Mar 2017 19:39:20 +0000 (14:39 -0500)]
Merge "res/res_pjsip_session: Only check localnet if it is defined"

4 years agoaudiohook.c: Lost RTP packets lead to out-of-sync MixMonitor.
Aaron An [Wed, 15 Mar 2017 04:49:12 +0000 (12:49 +0800)]
audiohook.c: Lost RTP packets lead to out-of-sync MixMonitor.

Fixed a bug in function "ast_audiohook_write_frame" that checked the
variable other_factory_samples and only flushed the factories, so they
would be in sync, when other_factory_samples > 0. When there is not any
rtp incoming the variable other_factory_samples will be 0, and although
the result of "our_factory_ms - other_factory_ms" may be very large,
this led to the record file not syncing.

ASTERISK-26875 #close
Reported-by: Aaron An
Tested-by: Aaron An

Change-Id: Ia4d890fb8fc1636a7188502bab35f555685aea22

4 years agoMerge "thread safety: Don't use getprotobyname()"
zuul [Mon, 20 Mar 2017 18:07:51 +0000 (13:07 -0500)]
Merge "thread safety: Don't use getprotobyname()"

4 years agothread safety: Don't use getprotobyname()
Sean Bright [Sat, 18 Mar 2017 17:30:32 +0000 (13:30 -0400)]
thread safety: Don't use getprotobyname()

POSIX does not require getprotobyname() to be thread safe and some
implementations use static memory which causes issues when multiple
threads are used.

Further, our usage of it today is just to ultimately get IPPROTO_TCP
for calls to setsockopt(). So instead we just use IPPROTO_TCP directly.

Change-Id: I2e14e58674808f7ce99b2f5e900d0f90d0d8da48

4 years agores_rtp_asterisk: Pass correct data length to ast_rtcp_interpret
Sean Bright [Sun, 19 Mar 2017 18:26:38 +0000 (14:26 -0400)]
res_rtp_asterisk: Pass correct data length to ast_rtcp_interpret

We are currently passing in the capacity of the read buffer instead of the
number of bytes that we actually read off the wire.

Change-Id: I60465049727d955c7f9a5e529e6f2aaff04cda36

4 years agoMerge "app_queue: Fix locking behavior in stasis message handlers"
Joshua Colp [Sat, 18 Mar 2017 11:53:08 +0000 (06:53 -0500)]
Merge "app_queue: Fix locking behavior in stasis message handlers"

4 years agoMerge "chan_sip: Add rtcp-mux support"
Joshua Colp [Sat, 18 Mar 2017 10:38:19 +0000 (05:38 -0500)]
Merge "chan_sip: Add rtcp-mux support"

4 years agoMerge "res_rtp_asterisk: Fix crash when RTCP is not present when DTLS is stopped."
Joshua Colp [Sat, 18 Mar 2017 10:37:29 +0000 (05:37 -0500)]
Merge "res_rtp_asterisk: Fix crash when RTCP is not present when DTLS is stopped."

4 years agoMerge "res_pjsip_asterisk.c: Fix compile error if libsrtp is not installed."
Joshua Colp [Sat, 18 Mar 2017 10:36:34 +0000 (05:36 -0500)]
Merge "res_pjsip_asterisk.c: Fix compile error if libsrtp is not installed."

4 years agoMerge "app_confbridge: Fix ConfbridgeTalking AMI event description."
Joshua Colp [Sat, 18 Mar 2017 00:49:21 +0000 (19:49 -0500)]
Merge "app_confbridge: Fix ConfbridgeTalking AMI event description."

4 years agoMerge "res_pjsip_sdp_rtp.c: Fix cut-n-paste error"
Joshua Colp [Fri, 17 Mar 2017 19:45:05 +0000 (14:45 -0500)]
Merge "res_pjsip_sdp_rtp.c: Fix cut-n-paste error"

4 years agoMerge "res_pjsip_sdp_rtp: RTP instance does not use same IP as explicit transport"
Joshua Colp [Fri, 17 Mar 2017 16:47:36 +0000 (11:47 -0500)]
Merge "res_pjsip_sdp_rtp: RTP instance does not use same IP as explicit transport"

4 years agoapp_queue: Member stuck as pending after forwarding previous call from queue
Robert Mordec [Tue, 14 Mar 2017 14:27:56 +0000 (15:27 +0100)]
app_queue: Member stuck as pending after forwarding previous call from queue

Queue member will get stuck in pending_members if queue calls a device
that is different from the one observed for state changes.

This patch removes members from pending_members as a result of channel stasis
events such as blind or attended transfers and hangup.

ASTERISK-26862 #close

Change-Id: I8bf6df487b9bb35726c08049ff25cdad5e357727

4 years agoCHANNEL(callid): Give dialplan access to the callid.
Richard Mudgett [Thu, 23 Feb 2017 05:26:13 +0000 (23:26 -0600)]
CHANNEL(callid): Give dialplan access to the callid.

* Added CHANNEL(callid) to retrieve the call identifier log tag associated
with the channel.  Dialplan now has access to the call log search key
associated with the channel so it can be saved in case there is a problem
with the call.


Change-Id: I2c97ebd928b6f3c5bc80c5729e4d3c07f453049f

4 years agoapp_queue: Fix locking behavior in stasis message handlers
Sean Bright [Thu, 16 Mar 2017 13:42:54 +0000 (09:42 -0400)]
app_queue: Fix locking behavior in stasis message handlers

The queue_stasis_data structure contains various mutable fields that require
appropriate locking. Specifically, the 'dying,' 'member_uniqueid,' and
'caller_uniqueid' fields need to be locked when read from or written to.

Change-Id: I246b7dbff8447acc957a1299f6ad0ebd0fd39088

4 years agochan_sip: Add rtcp-mux support
Sean Bright [Wed, 8 Mar 2017 01:28:18 +0000 (20:28 -0500)]
chan_sip: Add rtcp-mux support

ASTERISK-26846 #close

Change-Id: I541a1602ff55ab73684e9f8002edb9e0e745d639

4 years agoapp_confbridge: Fix ConfbridgeTalking AMI event description.
Richard Mudgett [Thu, 16 Mar 2017 21:50:17 +0000 (16:50 -0500)]
app_confbridge: Fix ConfbridgeTalking AMI event description.

Thanks to Chris Howard for pointing this out on the wiki.

Change-Id: I18e56de09a70e736b5d04719d45ef29cf0636705

4 years agores_pjsip_asterisk.c: Fix compile error if libsrtp is not installed.
Richard Mudgett [Thu, 16 Mar 2017 21:37:42 +0000 (16:37 -0500)]
res_pjsip_asterisk.c: Fix compile error if libsrtp is not installed.

struct ast_rtcp does not define the dtls member if SRTP is not enabled.


Change-Id: Id15ea212e04490e012f2cf4a56818b4dd948875e

4 years agoMerge "res_pjsip: Symmetric transports"
Joshua Colp [Thu, 16 Mar 2017 21:04:43 +0000 (16:04 -0500)]
Merge "res_pjsip:  Symmetric transports"

4 years agores_pjsip_sdp_rtp.c: Fix cut-n-paste error
Richard Mudgett [Thu, 16 Mar 2017 20:45:57 +0000 (15:45 -0500)]
res_pjsip_sdp_rtp.c: Fix cut-n-paste error

We were inadvertenly referencing the cos_video option to determine if we
should set the tos_audio and cos_audio value on the RTP instance.

Change-Id: Ia7964f486801d39dc6f5dae570baff079e1595b0

4 years agores/res_pjsip_session: Only check localnet if it is defined
Matt Jordan [Thu, 16 Mar 2017 15:39:00 +0000 (10:39 -0500)]
res/res_pjsip_session: Only check localnet if it is defined

If local_net is not defined on a transport, transport_state->localnet
will be NULL. ast_apply_ha will, be default, return AST_SENSE_ALLOW in
this case, causing the external_media_address, if set, to be skipped.

This patch causes us to only check if we are sending within a network if
local_net is defined.

ASTERISK-26879 #close

Change-Id: Ib661c31a954cabc9c99f1f25c9c9a5c5b82cbbfb

4 years agoMerge "RFC sdp: Initial SDP creation"
Joshua Colp [Thu, 16 Mar 2017 19:45:20 +0000 (14:45 -0500)]
Merge "RFC sdp: Initial SDP creation"

4 years agores_pjsip_sdp_rtp: RTP instance does not use same IP as explicit transport
Richard Begg [Tue, 14 Mar 2017 21:22:42 +0000 (08:22 +1100)]
res_pjsip_sdp_rtp: RTP instance does not use same IP as explicit transport

Currently a wildcard address is used for the local RTP socket, which
will not always result in the same address as used by the SIP socket
(e.g. if explicit transport addresses are configured).
Use the transport's host address when binding new local RTP sockets if


Change-Id: I098c29c9d1f79a4f970d72ba894874ac75954f1a

4 years agores_pjsip: Symmetric transports
George Joseph [Tue, 7 Mar 2017 14:33:26 +0000 (07:33 -0700)]
res_pjsip:  Symmetric transports

A new transport parameter 'symmetric_transport' has been added.

When a request from a dynamic contact comes in on a transport with
this option set to 'yes', the transport name will be saved and used
for subsequent outgoing requests like OPTIONS, NOTIFY and INVITE.
It's saved as a contact uri parameter named 'x-ast-txp' and will
display with the contact uri in CLI, AMI, and ARI output.  On the
outgoing request, if a transport wasn't explicitly set on the
endpoint AND the request URI is not a hostname, the saved transport
will be used and the 'x-ast-txp' parameter stripped from the
outgoing packet.

* config_transport was modified to accept and store the new parameter.

* config_transport/transport_apply was updated to store the transport
  name in the pjsip_transport->info field using the pjsip_transport->pool
  on UDP transports.

* A 'multihomed_on_rx_message' function was added to
  pjsip_message_ip_updater that, for incoming requests, retrieves the
  transport name from pjsip_transport->info and retrieves the transport.
  If transport->symmetric_transport is set, an 'x-ast-txp' uri parameter
  containing the transport name is added to the incoming Contact header.

* An 'ast_sip_get_transport_name' function was added to res_pjsip.
  It takes an ast_sip_endpoint and a pjsip_sip_uri and returns a
  transport name if endpoint->transport is set or if there's an
  'x-ast-txp' parameter on the uri and the uri host is an ipv4 or
  ipv6 address.  Otherwise it returns NULL.

* An 'ast_sip_dlg_set_transport' function was added to res_pjsip
  which takes an ast_sip_endpoint, a pjsip_dialog, and an optional
  pjsip_tpselector.  It calls ast_sip_get_transport_name() and if
  a non-NULL is returned, sets the selector and sets the transport
  on the dialog.  If a selector was passed in, it's updated.

* res_pjsip/ast_sip_create_dialog_uac and ast_sip_create_dialog_uas
  were modified to call ast_sip_dlg_set_transport() instead of their
  original logic.

* res_pjsip/create_out_of_dialog_request was modified to call
  ast_sip_get_transport_name() and pjsip_tx_data_set_transport()
  instead of its original logic.

* Existing transport logic was removed from endpt_send_request
  since that can only be called after a create_out_of_dialog_request.

* res_pjsip/ast_sip_create_rdata was converted to a wrapper around
  a new 'ast_sip_create_rdata_with_contact' function which allows
  a contact_uri to be specified in addition to the existing
  parameters.  (See below)

* res_pjsip_pubsub/internal_pjsip_evsub_send_request was eliminated
  since all it did was transport selection and that is now done in
  ast_sip_create_dialog_uac and ast_sip_create_dialog_uas.

* 'contact_uri' was added to subscription_persistence.  This was
  necessary because although the parsed rdata contact header has the
  x-ast-txp parameter added (if appropriate),
  subscription_persistence_update stores the raw packet which
  doesn't have it.  subscription_persistence_recreate was then
  updated to call ast_sip_create_rdata_with_contact with the
  persisted contact_uri so the recreated subscription has the
  correct transport info to send the NOTIFYs.

* res_pjsip_session/internal_pjsip_inv_send_msg was eliminated since
  all it did was transport selection and that is now done in

* pjsip_message_ip_updater/multihomed_on_tx_message was updated
  to remove all traces of the x-ast-txp parameter from the
  outgoing headers.

NOTE:  This change does NOT modify the behavior of permanent
contacts specified on an aor.  To do so would require that the
permanent contact's contact uri be updated with the x-ast-txp
parameter and the aor sorcery object updated.  If we need to
persue this, we need to think about cloning permanent contacts into
the same store as the dynamic ones on an aor load so they can be
updated without disturbing the originally configured value.

You CAN add the x-ast-txp parameter to a permanent contact's uri
but it would be much simpler to just set endpoint->transport.

Change-Id: I4ee1f51473da32ca54b877cd158523efcef9655f

4 years agores_rtp_asterisk: Fix crash when RTCP is not present when DTLS is stopped.
Joshua Colp [Thu, 16 Mar 2017 14:07:55 +0000 (14:07 +0000)]
res_rtp_asterisk: Fix crash when RTCP is not present when DTLS is stopped.

This change removes an assumption that when DTLS is stopped
an RTCP session will be present on the RTP session. This is not
always the case.


Change-Id: Ib9f7c09ce0b005efe362dbcc8795202b18f94611

4 years agoMerge "Add rtcp-mux support"
Joshua Colp [Thu, 16 Mar 2017 15:46:01 +0000 (10:46 -0500)]
Merge "Add rtcp-mux support"

4 years agoMerge "chan_iax2: Reload of iax peer results in loss of host address/port"
Joshua Colp [Thu, 16 Mar 2017 10:23:37 +0000 (05:23 -0500)]
Merge "chan_iax2: Reload of iax peer results in loss of host address/port"

4 years agoMerge "res/res_pjsip_refer: call xfer w/o extension"
zuul [Thu, 16 Mar 2017 04:03:52 +0000 (23:03 -0500)]
Merge "res/res_pjsip_refer: call xfer w/o extension"

4 years agoMerge "app_queue: Handle the caller being redirected out of a queue bridge"
zuul [Thu, 16 Mar 2017 01:30:55 +0000 (20:30 -0500)]
Merge "app_queue: Handle the caller being redirected out of a queue bridge"

4 years agoMerge "funcs/func_devstate: Remove new line in Device field of during module load"
zuul [Thu, 16 Mar 2017 01:13:17 +0000 (20:13 -0500)]
Merge "funcs/func_devstate: Remove new line in Device field of during module load"

4 years agoMerge "pbx.c: Fix crash from malformed exten pattern."
zuul [Thu, 16 Mar 2017 00:14:08 +0000 (19:14 -0500)]
Merge "pbx.c: Fix crash from malformed exten pattern."

4 years agoMerge "res_pjsip_endpoint_identifier_ip: Don't output error if no header_match."
zuul [Thu, 16 Mar 2017 00:01:40 +0000 (19:01 -0500)]
Merge "res_pjsip_endpoint_identifier_ip: Don't output error if no header_match."

4 years agoautochan/mixmonitor/chanspy: Fix unsafe channel locking and references.
Richard Mudgett [Wed, 15 Mar 2017 18:24:33 +0000 (13:24 -0500)]
autochan/mixmonitor/chanspy: Fix unsafe channel locking and references.

Dereferencing struct ast_autochan.chan without first calling
ast_autochan_channel_lock() is unsafe because the pointer could change at
any time due to a masquerade.  Unfortunately, ast_autochan_channel_lock()
itself uses struct ast_autochan.chan unsafely and can result in a deadlock
if the original channel happens to get destroyed after a masquerade in
addition to the pointer getting changed.

The problem is more likely to happen with v11 and earlier because
masquerades are used to optimize out local channels on those versions.
However, it could still happen on newer versions if the channel is
executing a dialplan application when the channel is transferred or
redirected.  In this situation a masquerade still must be used.

* Added a lock to struct ast_autochan to safely be able to use
ast_autochan.chan while trying to get the channel lock in
ast_autochan_channel_lock().  The locking order is the channel lock then
the autochan lock.  Locking in the other direction requires deadlock

* Fix unsafe ast_autochan.chan usages in app_mixmonitor.c.

* Fix unsafe ast_autochan.chan usages in app_chanspy.c.

* app_chanspy.c: Removed unused autochan parameter from next_channel().


Change-Id: Id29dd22bc0f369b44e23ca423d2f3657187cc592

4 years agoMerge "core: Add stream topology changing primitives with tests."
zuul [Wed, 15 Mar 2017 22:23:30 +0000 (17:23 -0500)]
Merge "core: Add stream topology changing primitives with tests."

4 years agoAdd rtcp-mux support
Mark Michelson [Tue, 7 Mar 2017 20:13:02 +0000 (14:13 -0600)]
Add rtcp-mux support

This commit adds support for RFC 5761: Multiplexing RTP Data and Control
Packets on a Single Port. Specifically, it enables the feature when
using chan_pjsip.

A new option, "rtcp_mux" has been added to endpoint configuration in
pjsip.conf. If set, then Asterisk will attempt to use rtcp-mux with
whatever it communicates with. Asterisk follows the rules set forth in
RFC 5761 with regards to falling back to standard RTCP behavior if the
far end does not indicate support for rtcp-mux.

The lion's share of the changes in this commit are in
res_rtp_asterisk.c. This is because it was pretty much hard wired to
have an RTP and an RTCP transport. The strategy used here is that when
rtcp-mux is enabled, the current RTCP transport and its trappings (such
as DTLS SSL session) are freed, and the RTCP session instead just
mooches off the RTP session. This leads to a lot of specialized if
statements throughout.

ASTERISK-26732 #close
Reported by Dan Jenkins

Change-Id: If46a93ba1282418d2803e3fd7869374da8b77ab5

4 years agoMerge "res_pjsip_endpoint_identifier_ip: Add an option to match requests by header"
Joshua Colp [Wed, 15 Mar 2017 19:49:13 +0000 (14:49 -0500)]
Merge "res_pjsip_endpoint_identifier_ip: Add an option to match requests by header"

4 years agoMerge "configure: Don't use the progress bar with curl when downloading to stdout"
Joshua Colp [Wed, 15 Mar 2017 18:01:16 +0000 (13:01 -0500)]
Merge "configure: Don't use the progress bar with curl when downloading to stdout"

4 years agores/res_pjsip_refer: call xfer w/o extension
Torrey Searle [Tue, 14 Mar 2017 13:49:54 +0000 (14:49 +0100)]
res/res_pjsip_refer: call xfer w/o extension

When transfering to a URI without an extension, ensure that the
s extension of the dialplan is entered

ASTERISK-26869 #close

Change-Id: I07403df66cf93f09e00a40ab5b41bfc6f72b1525

4 years agoapp_queue: Handle the caller being redirected out of a queue bridge
Sean Bright [Thu, 9 Mar 2017 17:05:12 +0000 (12:05 -0500)]
app_queue: Handle the caller being redirected out of a queue bridge

A caller can leave the Queue() application after being bridged with a
member in a few ways:

  * Caller or member hangup
  * Caller is transferred somewhere else (blind or atx)
  * Caller is externally redirected elsewhere

The first 2 scenarios are currently handled by subscribing to stasis
messages, but the 3rd is not explicitly covered. If a caller is
redirected away from the Queue() application, the member who was last
bridged with that caller will remain in an "In use" state until the
caller hangs up.

This patch adds handling of the caller leaving the queue via
redirection. We monitor the caller-member bridge, and if the caller is
the one that leaves, we treat it the same as we would a caller hangup.

ASTERISK-26400 #close
Reported by: Etienne Lessard

Change-Id: Iba160907770de5a6c9efeffc9df5a13e9ea75334

4 years agores_pjsip_endpoint_identifier_ip: Don't output error if no header_match.
Joshua Colp [Wed, 15 Mar 2017 13:44:43 +0000 (13:44 +0000)]
res_pjsip_endpoint_identifier_ip: Don't output error if no header_match.

This change ensures that if no header_match option is set on an
identify an error message is not output stating the option is set
to an invalid value.


Change-Id: I239bc6d2319dd3da24ba96a38d4d6e9b5526d62a

4 years agores_pjsip_endpoint_identifier_ip: Add an option to match requests by header
Matt Jordan [Tue, 14 Mar 2017 12:50:07 +0000 (07:50 -0500)]
res_pjsip_endpoint_identifier_ip: Add an option to match requests by header

This patch adds a new features to the endpoint identifier module,
'match_header'. When set, inbound requests are matched by a provided SIP
header: value pair. This option works in conjunction with the existing
'match' configuration option, such that if any 'match*' attribute
matches an inbound request, the request is associated with the specified

Since this module now identifies by more than just IP address,
appropriate renaming of the module and/or variables can be done in a
non-release branch.

ASTERISK-26863 #close

Change-Id: Icfc14835c962f92e35e67bbdb235cf0589de5453
(cherry picked from commit 30f52d79d7fc9ab0b628bef2b61ea515413795a2)

4 years agoMerge "res_pjsip_endpoint_identifier_ip: Clean up a spaces/tabs issue"
George Joseph [Wed, 15 Mar 2017 13:47:36 +0000 (08:47 -0500)]
Merge "res_pjsip_endpoint_identifier_ip: Clean up a spaces/tabs issue"

4 years agoMerge "configs/samples/hep.conf.sample: Clarify how the HEP stack works"
Joshua Colp [Wed, 15 Mar 2017 10:20:50 +0000 (05:20 -0500)]
Merge "configs/samples/hep.conf.sample: Clarify how the HEP stack works"

4 years agoMerge "main/stasis_cache: Demote the ERROR message when removing a nonexistent item"
Joshua Colp [Wed, 15 Mar 2017 10:19:33 +0000 (05:19 -0500)]
Merge "main/stasis_cache: Demote the ERROR message when removing a nonexistent item"

4 years agoMerge "res_pjsip_transport_websocket: Add support for IPv6."
zuul [Wed, 15 Mar 2017 02:22:26 +0000 (21:22 -0500)]
Merge "res_pjsip_transport_websocket: Add support for IPv6."

4 years agopbx.c: Fix crash from malformed exten pattern.
Richard Mudgett [Tue, 14 Mar 2017 21:16:23 +0000 (16:16 -0500)]
pbx.c: Fix crash from malformed exten pattern.

Forgetting to indicate an exten is a pattern can cause a crash if the
"pattern" has a character set range.  e.g., "9999[3-5]" The crash is due
to a buffer overwrite because the '-' exten eye-candy wasn't removed as
expected and overran the allocated space.

The buffer overwrite is fixed two ways in this patch.

1) Fix ext_strncpy() to distinguish between pattern and non-pattern
extens.  Now '-' characters are removed when they are eye-candy and not
when they are part of a pattern character set.  Since the function is
private to pbx.c, the return value now returns the number of bytes written
to the destination buffer instead of the strlen() of the final buffer so
the callers that care don't need to add one.

2) Fix callers to ext_strncpy() to supply the correct available buffer
size of the destination buffer.


Change-Id: I555d97411140e47e0522684062d174fbe32aa84a

4 years agochan_iax2: Reload of iax peer results in loss of host address/port
Richard Begg [Tue, 14 Mar 2017 21:51:41 +0000 (08:51 +1100)]
chan_iax2: Reload of iax peer results in loss of host address/port

When using a non-dynamic peer address, build_peer() invalidates the
peer address structure by setting the address family to unspecified.
However, if dnsmgr is enabled, the subsequent call to ast_dnsmgr_lookup()
will not amend the peer address if the cache is still valid, resulting
in peer connectivity failures.
To fix this, we call ast_dnsmgr_refresh() instead.


Change-Id: Id8a89a2f771ebbaf32255a35fe596a6dcb97a082

4 years agoconfigure: Don't use the progress bar with curl when downloading to stdout
Matt Jordan [Tue, 14 Mar 2017 20:12:28 +0000 (15:12 -0500)]
configure: Don't use the progress bar with curl when downloading to stdout

In some scenarios, such as when there may not be a terminal (such as
inside a Docker container), curl will apparently direct the progress bar
to stdout. This can cause extra data to be appended to a file curl'd
down to stdout, resulting in md5 verification failures.

This patch removes the progress bar, and tells curl to download the file

ASTERISK-26872 #close

Change-Id: Ie860b020f627d4372b3e7ce9453de5faafeebe6c

4 years agoMerge "chan_pjsip: Don't assume a session will have a channel."
zuul [Tue, 14 Mar 2017 19:07:51 +0000 (14:07 -0500)]
Merge "chan_pjsip: Don't assume a session will have a channel."

4 years agoRFC sdp: Initial SDP creation
George Joseph [Thu, 2 Mar 2017 23:11:06 +0000 (16:11 -0700)]
RFC sdp: Initial SDP creation

* Added additional fields to ast_sdp_options.
* Re-organized ast_sdp.
* Updated field names to correspond to RFC4566 terminology.
* Created allocs/frees for SDP children.
* Created getters/setters for SDP children where appropriate.
* Added ast_sdp_create_from_state.
* Refactored res_sdp_translator_pjmedia for changes.

Change-Id: Iefbd877af7f5a4d3c74deead1bff8802661b0d48

4 years agoMerge "chan_sip: Call not cancelled after receiving a 422 response"
Joshua Colp [Tue, 14 Mar 2017 16:47:30 +0000 (11:47 -0500)]
Merge "chan_sip: Call not cancelled after receiving a 422 response"

4 years agoconfigs/samples/hep.conf.sample: Clarify how the HEP stack works
Matt Jordan [Tue, 14 Mar 2017 14:55:06 +0000 (09:55 -0500)]
configs/samples/hep.conf.sample: Clarify how the HEP stack works

This patch updates the documenation in hep.conf.sample to better specify
how the various HEP modules interact.

ASTERISK-26717 #close

Change-Id: I337fb742a89e3ec5edc7fc7a7a0295218d841124

4 years agofuncs/func_devstate: Remove new line in Device field of during module load
Matt Jordan [Tue, 14 Mar 2017 14:59:48 +0000 (09:59 -0500)]
funcs/func_devstate: Remove new line in Device field of during module load

During module loading of func_devstate, Asterisk emits the current
device state of all Custom device states currently stored in the AstDB.
This was erroneously including a new line character ('\n') to the end of
the device state, causing two new lines to be emitted in
DeviceStateChange AMI events.

Note that this only happened for those device state changes that
occurred during startup. Regular device state changes for Custom device
states are handled elsewhere, and did not have the newline.

ASTERISK-26643 #close
Reported by: Roman Bedros
Tested by: Matt Jordan
  ami_devstate.diff uploaded by Roman Bedros (License 6842)

Change-Id: I1f4c02fc79c448d43bf725f5039c83d9611d7d93

4 years agomain/stasis_cache: Demote the ERROR message when removing a nonexistent item
Matt Jordan [Tue, 14 Mar 2017 14:37:34 +0000 (09:37 -0500)]
main/stasis_cache: Demote the ERROR message when removing a nonexistent item

This patch demotes the ERROR message that is displayed when a
nonexistent item is removed from the Stasis cache. The genesis of this
demotion is due to chan_sip's realtime peers and their interaction with
Asterisk's core ast_endpoint code, but ostensibly it could happen from
other channel drivers as well.

Since Mark Michelson already did an excellent job of explaining on this
issue, it is quoted here for posterity:

"Internally, when a realtime peer is retrieved, Asterisk creates an
ast_endpoint structure. When that peer is destroyed, the ast_endpoint is
destroyed as well. Part of the destruction of the ast_endpoint involves
clearing the Stasis cache of all information about that endpoint. The
problem here is that the act of creating the ast_endpoint is not enough
to actually put any information in the Stasis cache. Instead, something
has to happen, such as a state change, in order for the Stasis cache to
have any information about that endpoint. When a device registers,
chan_sip creates an ast_endpoint structure, processes the REGISTER, and
then destroys the ast_endpoint. When the ast_endpoint is destroyed,
there is nothing to destroy in the Stasis cache, so an error message is
emitted. When you use rtcachefriends, ast_endpoint structures persist
for the lifetime of the module and so you do not see this error

ASTERISK-25237 #close

Change-Id: I53cebc6b4a897a1ab9564182b75c177780feff70

4 years agores_pjsip_endpoint_identifier_ip: Clean up a spaces/tabs issue
Matt Jordan [Wed, 8 Mar 2017 18:39:20 +0000 (12:39 -0600)]
res_pjsip_endpoint_identifier_ip: Clean up a spaces/tabs issue

Tabs > spaces. Always.

Change-Id: I899ff662361c7ab0327173bd7851a67b53dd65f1

4 years agochan_pjsip: Don't assume a session will have a channel.
Joshua Colp [Sun, 12 Mar 2017 14:21:16 +0000 (14:21 +0000)]
chan_pjsip: Don't assume a session will have a channel.

When querying for PJSIP specific information using the dialplan
function CHANNEL() it is possible that the underlying session
will no longer have a channel associated with it. This is
most likely to occur when the RTCP HEP module attempts to get
the channel name. If this happens then a crash will occur.

This change just adds a check that the channel exists on the
session before querying it.


Change-Id: I113479cffff6ae64cf8ed089e9e1565223426f01

4 years agopjproject_bundled: Reduce the need for rebuilds
George Joseph [Sat, 11 Mar 2017 02:29:04 +0000 (19:29 -0700)]
pjproject_bundled:  Reduce the need for rebuilds

Bundled pjproject should now only rebuild if one of the menuselect
"Compiler Flags" options changes.

Change-Id: If114a2e16b9e77af371a600d6a5e197bbf28fe43

4 years agoMerge "pjsip/cli_commands: pjsip show channelstats shows wrong codec"
Joshua Colp [Fri, 10 Mar 2017 22:02:08 +0000 (16:02 -0600)]
Merge "pjsip/cli_commands: pjsip show channelstats shows wrong codec"

4 years agoMerge "res_musiconhold: moh general section is a class and issues warning"
zuul [Fri, 10 Mar 2017 00:32:03 +0000 (18:32 -0600)]
Merge "res_musiconhold: moh general section is a class and issues warning"

4 years agoMerge "media_cache: Prefer ast_file_is_readable() over access()"
Joshua Colp [Thu, 9 Mar 2017 22:10:25 +0000 (16:10 -0600)]
Merge "media_cache: Prefer ast_file_is_readable() over access()"

4 years agopjsip/cli_commands: pjsip show channelstats shows wrong codec
Daniel Journo [Sun, 5 Mar 2017 21:26:07 +0000 (21:26 +0000)]
pjsip/cli_commands: pjsip show channelstats shows wrong codec

* cli_commands.c Fixed CLI output

ASTERISK-26822 #close

Change-Id: I3889ef6a8f6738fc312fab42db5efacd6e452b01

4 years agoMerge "pbx_spool: Set AST_OUTGOING_ATTEMPT variable on channel"
Joshua Colp [Thu, 9 Mar 2017 20:29:02 +0000 (14:29 -0600)]
Merge "pbx_spool: Set AST_OUTGOING_ATTEMPT variable on channel"

4 years agores_musiconhold: moh general section is a class and issues warning
Daniel Journo [Wed, 8 Mar 2017 20:29:16 +0000 (20:29 +0000)]
res_musiconhold: moh general section is a class and issues warning

* res_musiconhold.c: Ensure the general section is not treated as
a moh class.

ASTERISK-26353 #close

Change-Id: Ia3dbd11ea2b43ab3e6c820a9827811dd24bea82d

4 years agomedia_cache: Prefer ast_file_is_readable() over access()
Sean Bright [Wed, 8 Mar 2017 23:08:52 +0000 (18:08 -0500)]
media_cache: Prefer ast_file_is_readable() over access()

Change-Id: Icc0dc6e61b2e68d5cdcb74b016b2726a388c7def

4 years agopbx_spool: Set AST_OUTGOING_ATTEMPT variable on channel
Sean Bright [Tue, 7 Mar 2017 12:25:25 +0000 (07:25 -0500)]
pbx_spool: Set AST_OUTGOING_ATTEMPT variable on channel

Set a variable on the channel that indicates which attempt number we
are currently performing to allow for attempt-specific behavior.

ASTERISK-26568 #close
Reported by: Roman Shubovich

Change-Id: Iacd7e8d43b0ed5b6cb021c62f41f1a1f5733dd89

4 years agores_pjsip_transport_websocket: Add support for IPv6.
Joshua Colp [Tue, 7 Mar 2017 13:37:52 +0000 (13:37 +0000)]
res_pjsip_transport_websocket: Add support for IPv6.

This change adds a PJSIP patch (which has been contributed upstream)
to allow the registration of IPv6 transport types.

Using this the res_pjsip_transport_websocket module now registers
an IPv6 Websocket transport and uses it for the corresponding


Change-Id: Id1f9126f995b31dc38db8fdb58afd289b4ad1647

4 years agoapp_voicemail: Cannot set fromstring on a per-mailbox basis
Daniel Journo [Wed, 8 Mar 2017 14:16:29 +0000 (14:16 +0000)]
app_voicemail: Cannot set fromstring on a per-mailbox basis

* apps/app_voicemail.c fromstring field added to mailbox which will
override the global fromstring if set.

ASTERISK-24562 #close

Change-Id: I5e90e3a1ec2b2d5340b49a0db825e4bbb158b2fe

4 years agoMerge "res_http_websocket: Fix faulty read logic."
zuul [Wed, 8 Mar 2017 16:05:19 +0000 (10:05 -0600)]
Merge "res_http_websocket: Fix faulty read logic."

4 years agoMerge "pbx_spool: Gracefully handle long lines in call files"
zuul [Tue, 7 Mar 2017 23:54:29 +0000 (17:54 -0600)]
Merge "pbx_spool: Gracefully handle long lines in call files"

4 years agores_http_websocket: Fix faulty read logic.
Mark Michelson [Tue, 7 Mar 2017 19:38:17 +0000 (13:38 -0600)]
res_http_websocket: Fix faulty read logic.

When doing some WebRTC testing, I found that the websocket would
disconnect whenever I attempted to place a call into Asterisk. After
looking into it, I pinpointed the problem to be due to the iostreams
change being merged in.

Under certain circumstances, a call to ast_iostream_read() can return a
negative value. However, in this circumstance, the websocket code was
treating this negative return as if it were a partial read from the
websocket. The expected length would get adjusted by this negative
value, resulting in the expected length being too large.

This patch simply adds an if check to be sure that we are only updating
the expected length of a read when the return from a read is positive.

ASTERISK-26842 #close
Reported by Mark Michelson

Change-Id: Ib4423239828a013d27d7bc477d317d2f02db61ab