asterisk/asterisk.git
3 years agostasis/control: Fix possible deadlock with swap channel
George Joseph [Fri, 1 Sep 2017 10:17:02 +0000 (04:17 -0600)]
stasis/control:  Fix possible deadlock with swap channel

If an error occurs during a bridge impart it's possible that
the "bridge_after" callback might try to run before
control_swap_channel_in_bridge has been signalled to continue.
Since control_swap_channel_in_bridge is holding the control lock
and the callback needs it, a deadlock will occur.

* control_swap_channel_in_bridge now only holds the control
  lock while it's actually modifying the control structure and
  releases it while the bridge impart is running.
* bridge_after_cb is now tolerant of impart failures.

Change-Id: Ifd239aa93955b3eb475521f61e284fcb0da2c3b3

3 years agoMerge "alembic: Fix enum creation for dtls_fingerprint"
George Joseph [Wed, 6 Sep 2017 16:52:26 +0000 (11:52 -0500)]
Merge "alembic: Fix enum creation for dtls_fingerprint"

3 years agoMerge "alembic: fix erroneous commit for add_prune_on_boot"
Jenkins2 [Wed, 6 Sep 2017 15:55:35 +0000 (10:55 -0500)]
Merge "alembic: fix erroneous commit for add_prune_on_boot"

3 years agoMerge "res/res_pjsip: Standardize/fix localnet checks across pjsip."
Jenkins2 [Wed, 6 Sep 2017 15:17:06 +0000 (10:17 -0500)]
Merge "res/res_pjsip: Standardize/fix localnet checks across pjsip."

3 years agoMerge "formats: Restore previous fread() behavior"
Joshua Colp [Wed, 6 Sep 2017 14:25:40 +0000 (09:25 -0500)]
Merge "formats: Restore previous fread() behavior"

3 years agoalembic: Fix enum creation for dtls_fingerprint
George Joseph [Wed, 6 Sep 2017 12:54:00 +0000 (06:54 -0600)]
alembic: Fix enum creation for dtls_fingerprint

Change-Id: Ic061c5066a146616a68376881c7e4cf6d6e7e7db

3 years agoMerge "res_pjsip_t38: Make t38_reinvite_response_cb tolerant of NULL channel"
Jenkins2 [Wed, 6 Sep 2017 11:48:45 +0000 (06:48 -0500)]
Merge "res_pjsip_t38:  Make t38_reinvite_response_cb tolerant of NULL channel"

3 years agoMerge "res_calendar*, res_smdi: Move to "extended" support"
Jenkins2 [Wed, 6 Sep 2017 11:44:30 +0000 (06:44 -0500)]
Merge "res_calendar*, res_smdi: Move to "extended" support"

3 years agoalembic: fix erroneous commit for add_prune_on_boot
Florian Floimair [Tue, 5 Sep 2017 16:08:02 +0000 (18:08 +0200)]
alembic: fix erroneous commit for add_prune_on_boot

Added include for postgresql ENUM type and
redefined values in the same way as in the
other migration scripts.

ASTERISK-27254 #close

Change-Id: Id667304cdf3891b1c2f7d35fab3e2a84026159fa

3 years agochan_pjsip: Suppress frame warnings.
Ben Ford [Tue, 5 Sep 2017 14:35:12 +0000 (09:35 -0500)]
chan_pjsip: Suppress frame warnings.

When rtp_keepalive is on for a PJSIP endpoint dialing to another
Asterisk instance also using PJSIP, Asterisk will continue to print
warning messages about not being able to send frames of a certain
type. This suppresses that warning message.

Change-Id: I0332a05519d7bda9cacfa26d433909ff1909be67

3 years agoformats: Restore previous fread() behavior
Sean Bright [Tue, 5 Sep 2017 15:05:48 +0000 (11:05 -0400)]
formats: Restore previous fread() behavior

Some formats are able to handle short reads while others are not, so
restore the previous behavior for the format modules so that we don't
have spurious errors when playing back files.

ASTERISK-27232 #close
Reported by: Jens T.

Change-Id: Iab7f52b25a394f277566c8a2a4b15a692280a300

3 years agores/res_pjsip: Standardize/fix localnet checks across pjsip.
Walter Doekes [Tue, 5 Sep 2017 14:16:01 +0000 (16:16 +0200)]
res/res_pjsip: Standardize/fix localnet checks across pjsip.

In 2dee95cc (ASTERISK-27024) and 776ffd77 (ASTERISK-26879) there was
confusion about whether the transport_state->localnet ACL has ALLOW or
DENY semantics.

For the record: the localnet has DENY semantics, meaning that "not in
the list" means ALLOW, and the local nets are in the list.

Therefore, checks like this look wrong, but are right:

    /* See if where we are sending this request is local or not, and if
       not that we can get a Contact URI to modify */
    if (ast_apply_ha(transport_state->localnet, &addr) != AST_SENSE_ALLOW) {
        ast_debug(5, "Request is being sent to local address, "
                     "skipping NAT manipulation\n");

(In the list == localnet == DENY == skip NAT manipulation.)

And conversely, other checks that looked right, were wrong.

This change adds two macro's to reduce the confusion and uses those
instead:

    ast_sip_transport_is_nonlocal(transport_state, addr)
    ast_sip_transport_is_local(transport_state, addr)

ASTERISK-27248 #close

Change-Id: Ie7767519eb5a822c4848e531a53c0fd054fae934

3 years agoMerge "app_directory: Handle a NULL mailbox without crashing"
Joshua Colp [Tue, 5 Sep 2017 13:41:19 +0000 (08:41 -0500)]
Merge "app_directory: Handle a NULL mailbox without crashing"

3 years agores_calendar*, res_smdi: Move to "extended" support
George Joseph [Tue, 5 Sep 2017 12:50:36 +0000 (06:50 -0600)]
res_calendar*, res_smdi: Move to "extended" support

Change-Id: I31eee8be30c6b0fc3dadb31111dd47742da8892d

3 years agoMerge "chan_ooh323: Fix confusing indentation warning"
Joshua Colp [Tue, 5 Sep 2017 12:16:41 +0000 (07:16 -0500)]
Merge "chan_ooh323: Fix confusing indentation warning"

3 years agores_pjsip_t38: Make t38_reinvite_response_cb tolerant of NULL channel
George Joseph [Tue, 5 Sep 2017 10:23:04 +0000 (04:23 -0600)]
res_pjsip_t38:  Make t38_reinvite_response_cb tolerant of NULL channel

t38_reinvite_response_cb can get called by res_pjsip_session's
session_inv_on_tsx_state_changed in situations where session->channel
is NULL.  If it is, the ast_log warning segfaults because it tries
to get the channel name from a NULL channel.

* Check session->channel and print "unknown channel" when it's NULL.

ASTERISK-27236
Reported by: Ross Beer

Change-Id: I4326e288d36327f6c79ab52226d54905cdc87dc7

3 years agortp_engine: Prevent possible double free with DTLS config
Sean Bright [Fri, 1 Sep 2017 21:17:38 +0000 (17:17 -0400)]
rtp_engine: Prevent possible double free with DTLS config

ASTERISK-27225 #close
Reported by: Richard Kenner

Change-Id: I097b81734ef730f8603c0b972909d212a3a5cf89

3 years agochan_ooh323: Fix confusing indentation warning
Sean Bright [Fri, 1 Sep 2017 18:15:40 +0000 (14:15 -0400)]
chan_ooh323: Fix confusing indentation warning

ASTERISK-27177 #close
Reported by: Tzafrir Cohen

Change-Id: I40311c404edb2302a7543ad5ca7a06b2a38f2d97

3 years agoapp_directory: Handle a NULL mailbox without crashing
Sean Bright [Fri, 1 Sep 2017 14:51:06 +0000 (10:51 -0400)]
app_directory: Handle a NULL mailbox without crashing

ASTERISK-27241 #close
Reported by: David Moore

Change-Id: Ibbbca85517b04c315406ebfe3b6f7e0763daedc6

3 years agoMerge "chan_pjsip: Add tag info in CHANNEL function"
Jenkins2 [Thu, 31 Aug 2017 22:33:05 +0000 (17:33 -0500)]
Merge "chan_pjsip: Add tag info in CHANNEL function"

3 years agoMerge "res_rtp_asterisk: Allow remote SSRC to change on an RTP instance."
Joshua Colp [Thu, 31 Aug 2017 21:50:50 +0000 (16:50 -0500)]
Merge "res_rtp_asterisk: Allow remote SSRC to change on an RTP instance."

3 years agoMerge "res_rtp_asterisk: Only learn a new source in learn state."
Joshua Colp [Thu, 31 Aug 2017 13:34:48 +0000 (08:34 -0500)]
Merge "res_rtp_asterisk: Only learn a new source in learn state."

3 years agoMerge "pjsip_message_ip_updater: Fix issue handling "tel" URIs"
Jenkins2 [Thu, 31 Aug 2017 13:30:17 +0000 (08:30 -0500)]
Merge "pjsip_message_ip_updater:  Fix issue handling "tel" URIs"

3 years agopjsip_message_ip_updater: Fix issue handling "tel" URIs
George Joseph [Mon, 24 Jul 2017 15:48:14 +0000 (09:48 -0600)]
pjsip_message_ip_updater:  Fix issue handling "tel" URIs

sanitize_tdata was assuming all URIs were SIP URIs so when a non
SIP uri was in the From, To or Contact headers, the unconditional
cast of a non-pjsip_sip_uri structure to pjsip_sip_uri caused
a segfault when trying to access uri->other_param.

* Added PJSIP_URI_SCHEME_IS_SIP(uri) || PJSIP_URI_SCHEME_IS_SIPS(uri)
  checks before attempting to cast or use the returned uri.

ASTERISK-27152
Reported-by: Ross Beer

Change-Id: Id380df790e6622c8058a96035f8b8f4aa0b8551f

3 years agoAST-2017-006: Fix app_minivm application MinivmNotify command injection
Corey Farrell [Sun, 2 Jul 2017 00:24:27 +0000 (20:24 -0400)]
AST-2017-006: Fix app_minivm application MinivmNotify command injection

An admin can configure app_minivm with an externnotify program to be run
when a voicemail is received.  The app_minivm application MinivmNotify
uses ast_safe_system() for this purpose which is vulnerable to command
injection since the Caller-ID name and number values given to externnotify
can come from an external untrusted source.

* Add ast_safe_execvp() function.  This gives modules the ability to run
external commands with greater safety compared to ast_safe_system().
Specifically when some parameters are filled by untrusted sources the new
function does not allow malicious input to break argument encoding.  This
may be of particular concern where CALLERID(name) or CALLERID(num) may be
used as a parameter to a script run by ast_safe_system() which could
potentially allow arbitrary command execution.

* Changed app_minivm.c:run_externnotify() to use the new ast_safe_execvp()
instead of ast_safe_system() to avoid command injection.

* Document code injection potential from untrusted data sources for other
shell commands that are under user control.

ASTERISK-27103

Change-Id: I7552472247a84cde24e1358aaf64af160107aef1

3 years agores_rtp_asterisk: Only learn a new source in learn state.
Joshua Colp [Mon, 22 May 2017 15:36:38 +0000 (15:36 +0000)]
res_rtp_asterisk: Only learn a new source in learn state.

This change moves the logic which learns a new source address
for RTP so it only occurs in the learning state. The learning
state is entered on initial allocation of RTP or if we are
told that the remote address for the media has changed. While
in the learning state if we continue to receive media from
the original source we restart the learning process. It is
only once we receive a sufficient number of RTP packets from
the new source that we will switch to it. Once this is done
the closed state is entered where all packets that do not
originate from the expected source are dropped.

The learning process has also been improved to take into
account the time between received packets so a flood of them
while in the learning state does not cause media to be switched.

Finally RTCP now drops packets which are not for the learned
SSRC if strict RTP is enabled.

ASTERISK-27013

Change-Id: I56a96e993700906355e79bc880ad9d4ad3ab129c

3 years agores_rtp_asterisk: Allow remote SSRC to change on an RTP instance.
Joshua Colp [Wed, 30 Aug 2017 12:28:58 +0000 (12:28 +0000)]
res_rtp_asterisk: Allow remote SSRC to change on an RTP instance.

When SDP renegotiation occurs it is possible for an RTP
instance to be reused for a new stream, resulting in the remote
SSRC changing if it is part of a bundle group. This change
allows this and updates its mapping in the current bundle
group.

ASTERISK-27231

Change-Id: I6e3703974f236bc024c5dbe9bd43adae0c6fb490

3 years agoMerge "bridge_native_rtp.c: Fixup native_rtp_framehook()"
Jenkins2 [Wed, 30 Aug 2017 13:58:35 +0000 (08:58 -0500)]
Merge "bridge_native_rtp.c: Fixup native_rtp_framehook()"

3 years agochan_pjsip: Add tag info in CHANNEL function
Andre Nazario [Sat, 26 Aug 2017 02:06:10 +0000 (23:06 -0300)]
chan_pjsip: Add tag info in CHANNEL function

Create local_tag and remote_tag in CHANNEL info to get tag from From and
To headers of a SIP dialog.

ASTERISK-27220

Change-Id: I59b16c4b928896fcbde02ad88f0e98922b15d524

3 years agobridge_native_rtp.c: Fixup native_rtp_framehook()
Richard Mudgett [Tue, 29 Aug 2017 19:22:15 +0000 (14:22 -0500)]
bridge_native_rtp.c: Fixup native_rtp_framehook()

* Fix framehook to test frame type for control frame.
* Made framehook exit early if frame type is not a control frame.
* Eliminated RAII_VAR in framehook.
* Use switch instead of else-if ladder for control frame handling.

Change-Id: Ia555fc3600bd85470e3c0141147dbe3ad07c1d18

3 years agoconfbridge: Handle user hangup during name recording
Sean Bright [Tue, 29 Aug 2017 14:26:17 +0000 (10:26 -0400)]
confbridge: Handle user hangup during name recording

This prevents orphaned CBAnn channels from getting stuck in the bridge.

ASTERISK-26994 #close
Reported by: James Terhune

Change-Id: I5e43e832a9507ec3f2c59752cd900b41dab80457

3 years agoMerge "core: Reduce video update queueing."
Jenkins2 [Tue, 29 Aug 2017 11:13:09 +0000 (06:13 -0500)]
Merge "core: Reduce video update queueing."

3 years agoMerge "app_record: Resolve some absolute vs. relative filename bugs"
Jenkins2 [Tue, 29 Aug 2017 10:57:07 +0000 (05:57 -0500)]
Merge "app_record: Resolve some absolute vs. relative filename bugs"

3 years agoMerge "voicemail: Fix various abuses of mkstemp"
Jenkins2 [Tue, 29 Aug 2017 10:17:21 +0000 (05:17 -0500)]
Merge "voicemail: Fix various abuses of mkstemp"

3 years agocore: Reduce video update queueing.
Joshua Colp [Thu, 24 Aug 2017 16:45:08 +0000 (13:45 -0300)]
core: Reduce video update queueing.

A video update frame is used to indicate that a channel
with video negotiated should provide a full frame so the
decoder decoding the stream is able to do so. In situations
where a queue is used to store frames it makes no sense
for the queue to contain multiple video update frames. One
is sufficient to have a full frame be sent.

ASTERISK-27222

Change-Id: Id3f40a6f51b740ae4704003a1800185c0c658ee7

3 years agoMerge "res/res_pjsip_session: allow SDP answer to be regenerated"
Joshua Colp [Mon, 28 Aug 2017 12:34:47 +0000 (07:34 -0500)]
Merge "res/res_pjsip_session: allow SDP answer to be regenerated"

3 years agoMerge "alembic: Add dtls_fingerprint column in ps_endpoints table"
Jenkins2 [Mon, 28 Aug 2017 11:47:40 +0000 (06:47 -0500)]
Merge "alembic: Add dtls_fingerprint column in ps_endpoints table"

3 years agovoicemail: Fix various abuses of mkstemp
Sean Bright [Fri, 25 Aug 2017 18:44:35 +0000 (14:44 -0400)]
voicemail: Fix various abuses of mkstemp

mkstemp() returns a unique filename, but appending an extension to that
filename does not guarantee uniqueness. Instead, use mkdtemp() and we
can put whatever extension we want on the files that we create inside
the directory.

In the case of app_minivm, we also now properly clean up any temporary
files that we create.

ASTERISK-20858 #close
Reported by: Walter Doekes

Change-Id: I30ad04f0e115f0b11693ff678ba5184d8b938e43

3 years agoapp_record: Resolve some absolute vs. relative filename bugs
Sean Bright [Fri, 25 Aug 2017 17:20:16 +0000 (13:20 -0400)]
app_record: Resolve some absolute vs. relative filename bugs

If the Record() application is called with a relative filename that
includes directories, we were not properly creating the intermediate
directories and Record() would fail.

Secondarily, updated the documentation for RECORDED_FILE to mention
that it does not include a filename extension.

Finally, rewrote the '%d' functionality to be a bit more straight
forward and less noisy.

ASTERISK-16777 #close
Reported by: klaus3000

Change-Id: Ibc2640cba3a8c7f17d97b02f76b7608b1e7ffde2

3 years agoMerge "app_queue: Evaluate realtime queues when running dialplan functions"
Jenkins2 [Fri, 25 Aug 2017 14:32:28 +0000 (09:32 -0500)]
Merge "app_queue: Evaluate realtime queues when running dialplan functions"

3 years agoMerge "chan_pjsip.c: Fix topology refresh response code accuracy."
Joshua Colp [Fri, 25 Aug 2017 13:32:43 +0000 (08:32 -0500)]
Merge "chan_pjsip.c: Fix topology refresh response code accuracy."

3 years agoMerge "app_voicemail: Honor escape digits in "greeting only" mode"
Joshua Colp [Fri, 25 Aug 2017 13:28:11 +0000 (08:28 -0500)]
Merge "app_voicemail: Honor escape digits in "greeting only" mode"

3 years agoalembic: Add dtls_fingerprint column in ps_endpoints table
Florian Floimair [Wed, 23 Aug 2017 15:01:09 +0000 (17:01 +0200)]
alembic: Add dtls_fingerprint column in ps_endpoints table

The ps_endpoints table was missing the dtls_fingerprint column
introduced with commit adba2a8d7fd.

ASTERISK-27168 #close

Change-Id: I9cb5006f7f50718b5239919562773adabb334cfd

3 years agores/res_pjsip_session: allow SDP answer to be regenerated
Torrey Searle [Mon, 21 Aug 2017 09:28:52 +0000 (11:28 +0200)]
res/res_pjsip_session: allow SDP answer to be regenerated

If an SDP answer hasn't been sent yet, it's legal to change it.
This is required for PJSIP_DTMF_MODE to work correctly, and can
also have use in the future for updating codecs too.

ASTERISK-27209 #close

Change-Id: Idbbfb7cb3f72fbd96c94d10d93540f69bd51e7a1

3 years agoapp_queue: Evaluate realtime queues when running dialplan functions
Sean Bright [Thu, 24 Aug 2017 14:42:24 +0000 (10:42 -0400)]
app_queue: Evaluate realtime queues when running dialplan functions

ASTERISK-19103 #close
Reported by: Jim Van Meggelen

Change-Id: I4bd32a9d1fcebb8ac56bff0e084d4f53e31b692b

3 years agoapp_voicemail: Honor escape digits in "greeting only" mode
Sean Bright [Wed, 23 Aug 2017 14:19:35 +0000 (10:19 -0400)]
app_voicemail: Honor escape digits in "greeting only" mode

ASTERISK-21241 #close
Reported by: Eelco Brolman
Patches:
Patch uploaded by Eelco Brolman (License 6442)

Change-Id: Icbe39b5c82a49b46cf1d168dc17766f3d84f54fe

3 years agores_smdi: Clean up memory leak
Sean Bright [Thu, 24 Aug 2017 13:35:45 +0000 (09:35 -0400)]
res_smdi: Clean up memory leak

Change-Id: I1e33290929e1aa7c5b9cb513f8254f2884974de8

3 years agoMerge "res_pjsip_session.c: Fix crash when declining an active stream."
Joshua Colp [Wed, 23 Aug 2017 19:49:26 +0000 (14:49 -0500)]
Merge "res_pjsip_session.c: Fix crash when declining an active stream."

3 years agoMerge changes from topic 'ASTERISK-27212'
Jenkins2 [Wed, 23 Aug 2017 19:45:52 +0000 (14:45 -0500)]
Merge changes from topic 'ASTERISK-27212'

* changes:
  bridge_channel.c: Fix FRACK when mapping frames to the bridge.
  bridge: Fix softmix bridge deadlock.

3 years agoMerge "channel: Fix topology API locking."
Jenkins2 [Wed, 23 Aug 2017 19:17:11 +0000 (14:17 -0500)]
Merge "channel: Fix topology API locking."

3 years agoMerge "app_confbridge: Document sfu video_mode value."
Joshua Colp [Wed, 23 Aug 2017 18:05:35 +0000 (13:05 -0500)]
Merge "app_confbridge: Document sfu video_mode value."

3 years agoMerge "bridge_softmix.c: Restored softmix_bridge_leave() shortcut exit."
Jenkins2 [Wed, 23 Aug 2017 17:21:47 +0000 (12:21 -0500)]
Merge "bridge_softmix.c: Restored softmix_bridge_leave() shortcut exit."

3 years agoMerge "confbridge.h: Fix doxygen comments."
Jenkins2 [Wed, 23 Aug 2017 17:05:29 +0000 (12:05 -0500)]
Merge "confbridge.h: Fix doxygen comments."

3 years agoMerge "bridge_softmix.c: Remove always true test."
Jenkins2 [Wed, 23 Aug 2017 16:55:01 +0000 (11:55 -0500)]
Merge "bridge_softmix.c: Remove always true test."

3 years agoMerge "app_queue: Fix initial hold time queue statistic"
Jenkins2 [Wed, 23 Aug 2017 16:11:38 +0000 (11:11 -0500)]
Merge "app_queue: Fix initial hold time queue statistic"

3 years agores_pjsip_session.c: Fix crash when declining an active stream.
Richard Mudgett [Fri, 18 Aug 2017 22:37:12 +0000 (17:37 -0500)]
res_pjsip_session.c: Fix crash when declining an active stream.

If a previously active stream is declined we could crash because the
channel's thread is still using the stream while we are updating the
topology in the serializer thread.

* Defer removing any declined stream's handler until we have blocked the
channel's thread with the channel lock.

ASTERISK-27212

Change-Id: I50e1d3ef26f8e41948f4c411ee329aa3b960a420

3 years agobridge_channel.c: Fix FRACK when mapping frames to the bridge.
Richard Mudgett [Wed, 16 Aug 2017 22:50:18 +0000 (17:50 -0500)]
bridge_channel.c: Fix FRACK when mapping frames to the bridge.

* Add protection checks when mapping streams to the bridge.  The channel
and bridge may be in the process of updating the stream mapping when a
media frame comes in so we may not be able to map the frame at the time.

* We need to map the streams to the bridge's stream numbers right before
they are written into the bridge.  That way we don't have to keep
locking/unlocking the bridge and we won't have any synchronization
problems before the frames actually go into the bridge.

* Protect the deferred queue with the bridge_channel lock.

ASTERISK-27212

Change-Id: Id6860dd61b594b90c8395f6e2c0150219094c21a

3 years agochannel: Fix topology API locking.
Richard Mudgett [Fri, 11 Aug 2017 21:31:45 +0000 (16:31 -0500)]
channel: Fix topology API locking.

* ast_channel_request_stream_topology_change() must not be called with any
channel locks held.

* ast_channel_stream_topology_changed() must be called with only the
passed channel lock held.

ASTERISK-27212

Change-Id: I843de7956d9f1cc7cc02025aea3463d8fe19c691

3 years agobridge: Fix softmix bridge deadlock.
Richard Mudgett [Wed, 16 Aug 2017 20:22:04 +0000 (15:22 -0500)]
bridge: Fix softmix bridge deadlock.

* Fix deadlock in
bridge_softmix.c:softmix_bridge_stream_topology_changed() between
bridge_channel and channel locks.

* The new bridge technology topology change callbacks must be called with
the bridge locked.  The callback references the bridge channel list, the
bridge technology could change, and the bridge stream mapping is updated.

ASTERISK-27212

Change-Id: Ide4360ab853607e738ad471721af3f561ddd83be

3 years agochan_pjsip.c: Fix topology refresh response code accuracy.
Richard Mudgett [Mon, 14 Aug 2017 17:20:25 +0000 (12:20 -0500)]
chan_pjsip.c: Fix topology refresh response code accuracy.

There are other 1xx and 2xx codes than 100 and 200 respectively.

Change-Id: I680db0997343256add1478714f5bf5b5569aee17

3 years agobridge_softmix.c: Restored softmix_bridge_leave() shortcut exit.
Richard Mudgett [Fri, 11 Aug 2017 22:06:01 +0000 (17:06 -0500)]
bridge_softmix.c: Restored softmix_bridge_leave() shortcut exit.

Change-Id: I13026cd90954e0265eab94a0faf635a3e11f0e35

3 years agoapp_confbridge: Document sfu video_mode value.
Richard Mudgett [Thu, 17 Aug 2017 22:07:18 +0000 (17:07 -0500)]
app_confbridge: Document sfu video_mode value.

Change-Id: I26e17df2c93f3933b23f78070603adbcc84ba204

3 years agoconfbridge.h: Fix doxygen comments.
Richard Mudgett [Thu, 17 Aug 2017 22:06:21 +0000 (17:06 -0500)]
confbridge.h: Fix doxygen comments.

Change-Id: I16133166a85fdb557c66ffcbfe8128d0b4725b0e

3 years agobridge_softmix.c: Remove always true test.
Richard Mudgett [Fri, 11 Aug 2017 16:40:46 +0000 (11:40 -0500)]
bridge_softmix.c: Remove always true test.

Change-Id: I26238df2ff0d0f6dfe95c3aa35da588f1ee71727

3 years agoMerge "res_xmpp: fix inverted return code check in OAuth"
Jenkins2 [Tue, 22 Aug 2017 12:57:39 +0000 (07:57 -0500)]
Merge "res_xmpp: fix inverted return code check in OAuth"

3 years agoapp_queue: Fix initial hold time queue statistic
Sungtae Kim [Thu, 17 Aug 2017 21:46:49 +0000 (23:46 +0200)]
app_queue: Fix initial hold time queue statistic

Fixed to use correct initial value and fixed to use the
correct queue info to check the first value.

ASTERISK-27204

Change-Id: Ia9e36c828e566e1cc25c66f73307566e4acb8e73

3 years agoMerge "res_calendar_icalendar: Properly handle recurring events"
Joshua Colp [Tue, 22 Aug 2017 10:11:51 +0000 (05:11 -0500)]
Merge "res_calendar_icalendar: Properly handle recurring events"

3 years agores_xmpp: fix inverted return code check in OAuth
Michael Kuron [Sun, 20 Aug 2017 13:15:37 +0000 (15:15 +0200)]
res_xmpp: fix inverted return code check in OAuth

fetch_access_token calls func_curl via ast_func_read. The latter returns 0 upon
success and -1 if the function is not available.
This commit inverts the return code check so that an error is printed if the
module is not loaded and not if it is loaded.

ASTERISK-27207 #close

Change-Id: I9ef903f80702d1218e8701f65a4e5e918e6548fb

3 years agoMerge "Fix downloader not working with curl"
Jenkins2 [Fri, 18 Aug 2017 15:36:46 +0000 (10:36 -0500)]
Merge "Fix downloader not working with curl"

3 years agores_calendar_icalendar: Properly handle recurring events
Sean Bright [Thu, 17 Aug 2017 17:00:09 +0000 (13:00 -0400)]
res_calendar_icalendar: Properly handle recurring events

When looking for recurring events, use the correct end time based on the
configured 'timeframe.'

ASTERISK-27174 #close
Reported by: Mark Thompson

Change-Id: Id90c3cfc79d561a5521d79be176683e225f2edef

3 years agoFix downloader not working with curl
George Joseph [Wed, 16 Aug 2017 20:43:10 +0000 (14:43 -0600)]
Fix downloader not working with curl

The codec/dpma downloader wasn't handling curl correctly.  The logic
that transforms makeopts into a bash-sourceable file wasn't
handling the make 'or' command in DOWNLOAD_TIMEOUT so bash was
looking for an 'or' command.

That logic has been eliminated.  Instead of trying to transform
and source makeopts, the downloader now calls a make scriptlet
to print the value of a specific variable.  This way, make handles
the ors (or any other make construct that happens to creep into
that file).

ASTERISK-27202
Reported by: Sean McCord

Change-Id: Iadfb6693528e4d4da7b8bb201fa66da2c71c7f99

3 years agomanager: hook event is not being raised
Kevin Harwell [Tue, 15 Aug 2017 18:12:10 +0000 (13:12 -0500)]
manager: hook event is not being raised

When the iostream code went in it introduced a conditional that made it so the
hook event was not being raised even if a hook is present. This patch adds a
check to see if a hook is present in astman_append. If so then call into the
send_string function, which in turn raises the even for specified hook.

Also updated the ami hooks unit test, so the test could be automated.

ASTERISK-27200 #close

Change-Id: Iff37f02f9708195d8f23e68f959d6eab720e1e36

3 years agoMerge "configure: Check cache for valid pjproject tarball before downloading."
Jenkins2 [Wed, 16 Aug 2017 12:32:16 +0000 (07:32 -0500)]
Merge "configure: Check cache for valid pjproject tarball before downloading."

3 years agoconfigure: Check cache for valid pjproject tarball before downloading.
Richard Mudgett [Tue, 15 Aug 2017 20:15:58 +0000 (15:15 -0500)]
configure: Check cache for valid pjproject tarball before downloading.

On a fresh Asterisk source directory, the bundled pjproject tarball is
unconditionally downloaded even if the tarball is already in a specified
cache directory.

* Made check if the pjproject tarball is valid in the cache directory
before downloading the tarball on a fresh source directory.

Change-Id: Ic7ec842d3c97ecd8dafbad6f056b7fdbce41cae5

3 years agores_pjsip: Fix prune_on_boot to remove only contacts for the host.
Richard Mudgett [Tue, 15 Aug 2017 16:14:20 +0000 (11:14 -0500)]
res_pjsip: Fix prune_on_boot to remove only contacts for the host.

* Check that the contact's reg_server matches the host's name before
deleting any prune_on_boot contacts.  We don't want to delete reliable
transport contacts made with other servers if the ps_contacts database
table is shared with other servers.

Thanks to Ross Beer for pointing out that the original prune logic would
delete reliable transport contacts from other servers.

ASTERISK-27147

Change-Id: I8e439d0d1c266ffdfd7b73d1e5e466180a689bd0

3 years agores_xmpp: Google OAuth 2.0 protocol support for XMPP / Motif
Andrey Egorov [Fri, 4 Aug 2017 14:25:52 +0000 (17:25 +0300)]
res_xmpp: Google OAuth 2.0 protocol support for XMPP / Motif

Add ability to use tokens instead of passwords according to Google OAuth 2.0
protocol.

ASTERISK-27169
Reported by: Andrey Egorov
Tested by: Andrey Egorov

Change-Id: I07f7052a502457ab55010a4d3686653b60f4c8db

3 years agoMerge "STUN/netsock2: Fix some valgrind uninitialized memory findings."
Jenkins2 [Mon, 14 Aug 2017 18:45:25 +0000 (13:45 -0500)]
Merge "STUN/netsock2: Fix some valgrind uninitialized memory findings."

3 years agoMerge "res_pjsip_outbound_registration.c: Re-REGISTER on transport shutdown."
George Joseph [Mon, 14 Aug 2017 17:20:21 +0000 (12:20 -0500)]
Merge "res_pjsip_outbound_registration.c: Re-REGISTER on transport shutdown."

3 years agoMerge "res_pjsip: Remove ephemeral registered contacts on transport shutdown."
George Joseph [Mon, 14 Aug 2017 17:20:14 +0000 (12:20 -0500)]
Merge "res_pjsip: Remove ephemeral registered contacts on transport shutdown."

3 years agoMerge "res_pjsip: PJSIP Transport state monitor refactor."
George Joseph [Mon, 14 Aug 2017 17:19:53 +0000 (12:19 -0500)]
Merge "res_pjsip: PJSIP Transport state monitor refactor."

3 years agoMerge "res_pjsip_transport_management.c: Rename some variables."
Jenkins2 [Mon, 14 Aug 2017 14:28:41 +0000 (09:28 -0500)]
Merge "res_pjsip_transport_management.c: Rename some variables."

3 years agoSTUN/netsock2: Fix some valgrind uninitialized memory findings.
Richard Mudgett [Thu, 10 Aug 2017 19:18:01 +0000 (14:18 -0500)]
STUN/netsock2: Fix some valgrind uninitialized memory findings.

* netsock2.c: Test the addr->len member first as it may be the only member
initialized in the struct.

* stun.c:ast_stun_handle_packet(): The combinded[] local array could get
used uninitialized by ast_stun_request().  The uninitialized string gets
copied to another location and could overflow the destination memory
buffer.

These valgrind findings were found for ASTERISK_27150 but are not
necessarily a fix for the issue.

Change-Id: I55f8687ba4ffc0f69578fd850af006a56cbc9a57

3 years agores_pjsip_outbound_registration.c: Re-REGISTER on transport shutdown.
Richard Mudgett [Wed, 2 Aug 2017 23:44:12 +0000 (18:44 -0500)]
res_pjsip_outbound_registration.c: Re-REGISTER on transport shutdown.

The fix for the issue is broken up into three parts.

This is part three which handles the client side of REGISTER requests.
The registered contact may no longer be valid on the server when the
transport used is reliable and the connection is broken.

* Re-REGISTER our contact if the reliable transport is broken after
registration completes.  We attempt to re-REGISTER immediately to minimize
the time we are unreachable.  Time may have already passed between the
connection being broken and the loss being detected.

* Reorder sip_outbound_registration_state_alloc() so the STATSD_GUAGE's
are still correct if an allocation failure happens.

ASTERISK-27147

Change-Id: I3668405b1ee75dfefb07c0d637826176f741ce83

3 years agores_pjsip: Remove ephemeral registered contacts on transport shutdown.
Richard Mudgett [Mon, 31 Jul 2017 19:21:06 +0000 (14:21 -0500)]
res_pjsip: Remove ephemeral registered contacts on transport shutdown.

The fix for the issue is broken up into three parts.

This is part two which handles the server side of REGISTER requests when
rewrite_contact is enabled.  Any registered reliable transport contact
becomes invalid when the transport connection becomes disconnected.

* Monitor the rewrite_contact's reliable transport REGISTER contact for
shutdown.  If it is shutdown then the contact must be removed because it
is no longer valid.  Otherwise, when the client attempts to re-REGISTER it
may be blocked because the invalid contact is there.  Also if we try to
send a call to the endpoint using the invalid contact then the endpoint is
not likely to see the request.  The endpoint either won't be listening on
that port for new connections or a NAT/firewall will block it.

* Prune any rewrite_contact's registered reliable transport contacts on
boot.  The reliable transport no longer exists so the contact is invalid.

* Websockets always rewrite the REGISTER contact address and the transport
needs to be monitored for shutdown.

* Made the websocket transport set a unique name since that is what we use
as the ao2 container key.  Otherwise, we would not know which transport we
find when one of them shuts down.  The names are also used for PJPROJECT
debug logging.

* Made the websocket transport post the PJSIP_TP_STATE_CONNECTED state
event.  Now the global keep_alive_interval option, initially idle shutdown
timer, and the server REGISTER contact monitor can work on wetsocket
transports.

* Made the websocket transport set the PJSIP_TP_DIR_INCOMING direction.
Now initially idle websockets will automatically shutdown.

ASTERISK-27147

Change-Id: I397a5e7d18476830f7ffe1726adf9ee6c15964f4

3 years agores_pjsip: PJSIP Transport state monitor refactor.
Richard Mudgett [Fri, 28 Jul 2017 23:26:17 +0000 (18:26 -0500)]
res_pjsip: PJSIP Transport state monitor refactor.

The fix for the issue is broken up into three parts.

This is part one which refactors the transport state monitor code to allow
more modules to be able to monitor transports.

* Pull the management of PJPROJECT's transport state callback code from
res_pjsip_transport_management.c into res_pjsip.  Now other modules can
dynamically add and remove themselves from transport monitoring without
worrying about breaking PJPROJECT's callback chain.

* Add the ability for other modules to get a callback whenever a specific
transport is shutdown.

ASTERISK-27147

Change-Id: I7d9a31371eb1487c9b7050cf82a9af5180a57912

3 years agores_pjsip_transport_management.c: Rename some variables.
Richard Mudgett [Thu, 27 Jul 2017 20:36:20 +0000 (15:36 -0500)]
res_pjsip_transport_management.c: Rename some variables.

* Use monitored instead of the misleading keepalive name.

Change-Id: I9e5bcbb4ab2b82d49bcd0f06dfe85d15e0b552b6

3 years agoUPGRADE notes: Prepare for the eventual 16 branch.
Richard Mudgett [Wed, 9 Aug 2017 20:24:58 +0000 (15:24 -0500)]
UPGRADE notes: Prepare for the eventual 16 branch.

Change-Id: I4ca2f07ed62d77f1fdd10c3b216f6a28dd75720c

3 years agores_pjsip_messaging: IPv6 receive address needs brackets
Scott Griepentrog [Thu, 10 Aug 2017 14:09:29 +0000 (09:09 -0500)]
res_pjsip_messaging: IPv6 receive address needs brackets

When handling an incoming SIP MESSAGE, PJSIP
attaches the IP address that the message was
received from to the message in the variable
PJSIP_RECVADDR.  When the IP address is IPv6
the :PORT appended results in an unparseable
mess. By using an additional bit flag on the
pj_sockaddr_print call, the conventional use
of brackets around the address is achieved.

ASTERISK-27193 #close

Change-Id: I12342521f2ce87a5b6e4883d480a3fd957aa9fd9

3 years agoMerge "Make --with-pjproject-bundled the default for Asterisk 15"
Jenkins2 [Thu, 10 Aug 2017 12:25:26 +0000 (07:25 -0500)]
Merge "Make --with-pjproject-bundled the default for Asterisk 15"

3 years agoMerge "res_rtp_asterisk: Make P2P bridge Asymmetric codec aware"
Jenkins2 [Wed, 9 Aug 2017 20:39:34 +0000 (15:39 -0500)]
Merge "res_rtp_asterisk:  Make P2P bridge Asymmetric codec aware"

3 years agores_rtp_asterisk: enable rtcp & QOS stats on native bridge
Torrey Searle [Wed, 26 Jul 2017 14:17:02 +0000 (16:17 +0200)]
res_rtp_asterisk: enable rtcp & QOS stats on native bridge

Asterisk wasn't generating or forwarding RTCP packets when native
bridge was activated.  Also the stats weren't available via
CHANNEL(qos). Now the RTCP stats are always calculated.

ASTERISK-27158 #close

Change-Id: I46fb8f61c95e836b9d2dda6054b0cf205c16037b

3 years agores_rtp_asterisk: Make P2P bridge Asymmetric codec aware
Torrey Searle [Fri, 28 Jul 2017 12:53:44 +0000 (14:53 +0200)]
res_rtp_asterisk:  Make P2P bridge Asymmetric codec aware

Introduce a new property to rtp-engine to make it aware of
the desire for assymetric codecs or not.  If asymmetric codecs
is not allowed, the bridge will compare read/write formats
and shut down the p2p bridge if needed

ASTERISK-26745 #close

Change-Id: I0d9c83e5356df81661e58d40a8db565833501a6f

3 years agoMerge "res_pjsip_session/_sdp_rtp: Handling of 'msid' is incorrect"
Jenkins2 [Wed, 9 Aug 2017 13:15:24 +0000 (08:15 -0500)]
Merge "res_pjsip_session/_sdp_rtp: Handling of 'msid' is incorrect"

3 years agoMake --with-pjproject-bundled the default for Asterisk 15
George Joseph [Tue, 8 Aug 2017 18:33:50 +0000 (12:33 -0600)]
Make --with-pjproject-bundled the default for Asterisk 15

'--with-pjproject-bundled' is now the default when running
./configure. It can be disabled with '--without-pjproject-bundled'.

To make building without an internet connection easier, a new
./configure option '--with-download-cache' was added that sets
the cache for externals (like pjproject, the codecs and the DPMA),
AND the sounds files.  It can also be specified as an environment
variable named "AST_DOWNLOAD_CACHE".  The existing
'--with-sounds-cache' option / SOUNDS_CACHE_DIR env variable and
'--with-externals-cache' option / EXTERNALS_CACHE_DIR env variable
remain and if specified, will override '--with-downloads-cache'.

ASTERISK-27189

Change-Id: Ifa9783fddf44aafadb060c9feba713dfa81d38ce

3 years agores_pjsip_session: Release media resources on session end quicker.
Joshua Colp [Sat, 5 Aug 2017 11:36:49 +0000 (11:36 +0000)]
res_pjsip_session: Release media resources on session end quicker.

A change was made long ago where the session was kept around
until the underlying INVITE session had been destroyed. This
had the side effect of also keeping the underlying media resources
around for this time as well.

This change ensures that when we are told to terminate the
session we immediately release any media sessions associated
with it.

ASTERISK-27110

Change-Id: I643e431d5c3bf05cda220c1d39e824a505a29b82

3 years agoMerge "bridge: Fix stream topology/participant locking and video misrouting."
Jenkins2 [Mon, 7 Aug 2017 23:49:24 +0000 (18:49 -0500)]
Merge "bridge: Fix stream topology/participant locking and video misrouting."

3 years agoMerge "chan_sip: Access incoming REFER headers in dialplan"
Joshua Colp [Mon, 7 Aug 2017 14:51:57 +0000 (09:51 -0500)]
Merge "chan_sip: Access incoming REFER headers in dialplan"

3 years agoMerge "channel: Fix leak on successful call to chan->tech->requester."
Jenkins2 [Mon, 7 Aug 2017 14:30:21 +0000 (09:30 -0500)]
Merge "channel: Fix leak on successful call to chan->tech->requester."

3 years agoMerge "res_pjsip_nat.c: Remove unnecessary CMP_STOP."
Joshua Colp [Mon, 7 Aug 2017 13:31:50 +0000 (08:31 -0500)]
Merge "res_pjsip_nat.c: Remove unnecessary CMP_STOP."

3 years agoMerge "Support GMIME 3.0"
Jenkins2 [Mon, 7 Aug 2017 12:33:03 +0000 (07:33 -0500)]
Merge "Support GMIME 3.0"