asterisk/asterisk.git
5 years agores_fax: Add T.38 negotiation timeout option
Kinsey Moore [Fri, 9 Jan 2015 14:53:09 +0000 (14:53 +0000)]
res_fax: Add T.38 negotiation timeout option

This change makes the T.38 negotiation timeout configurable via
't38timeout' in res_fax.conf or FAXOPT(t38timeout). It was previously
hard coded to be 5000 milliseconds.

This change also handles T.38 switch failures by aborting the fax since
in the case where this can happen, both sides have agreed to switch to
T.38 and Asterisk is unable to do so.

Review: https://reviewboard.asterisk.org/r/4320/
........

Merged revisions 430415 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 430416 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430417 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_pubsub: Fix persistent subscriptions not surviving graceful shutdown
George Joseph [Thu, 8 Jan 2015 21:41:02 +0000 (21:41 +0000)]
res_pjsip_pubsub: Fix persistent subscriptions not surviving graceful shutdown

If you do a 'core (shutdown|restart) graceful' persistent subscriptions won't
survive.  If you do a 'core (shutdown|restart) now' or asterisk terminates for
some reason, they do.  Here's why...

When asterisk shuts down gracefully, it sends a 'NOTIFY/terminated' to
subscribers for each subscription.  This not only tells the subscribers that the
dialog/state machine is done, it also frees the last reference to the
subscription tree which causes the persistent subscription to get deleted from
astdb.  When asterisk restarts, nothing's left.  Just preventing the delete from
astdb doesn't work because we already told the subscriber to terminate the
dialog so we can't restart it even if it was still in astdb.  Everything works
OK if asterisk terminates unexpectedly because we never send the 'terminated'
message so on restart, the subscription is still in astdb and the subscriber is
none the wiser.

This patch suppresses the sending of 'NOTIFY/terminated' on shutdown for
persistent connections.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4318/
........

Merged revisions 430397 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430398 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_outbound_registration: Fix reference leak.
George Joseph [Thu, 8 Jan 2015 21:38:26 +0000 (21:38 +0000)]
res_pjsip_outbound_registration: Fix reference leak.

Every time a registration started, sip_outbound_registration_response_cb bumps
the ref count on client_state then pushes a handle_registration_response task.
handle_registration_response never unreffed it though.  So every time a
registration goes out, the ref count goes up by one.

This patch adds the unreffs to handle_registration_response.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4303/
........

Merged revisions 430395 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430396 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_outbound_registration: Fix several reload issues
George Joseph [Thu, 8 Jan 2015 17:51:36 +0000 (17:51 +0000)]
res_pjsip_outbound_registration: Fix several reload issues

There are 2 issues with reloading registrations...

1.  The 'can_reuse_registration' test wasn't considering the intervals or
expiration in its determination of whether a registration changed or not so if
you changed any of the intervals or the expiration and reloaded, the object
would get reloaded but the actual timers wouldn't change.
can_reuse_registration now does a sorcery diff on the old and new objects
instead of discretely testing certain fields.  Now if you change expiration for
instance, and reload, the timer is updated and re-registration will occur on the
new value.

2.  If you mung up your password on an outbound registration you get a permanent
failure.  If you fix the password (on the outbound_auth object) and reload,
nothing tells outbound_registration to try again because the registration itself
didn't change.  This patch adds an observer on the "auth" object type and if any
auth changes, existing registration states are searched and those in a
REJECTED_PERMANENT state are retried.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4304/
........

Merged revisions 430373 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430374 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoARI: Allow usage of ASYNCGOTO with Stasis()
Kinsey Moore [Wed, 7 Jan 2015 21:26:48 +0000 (21:26 +0000)]
ARI: Allow usage of ASYNCGOTO with Stasis()

When the AMI Redirect action is used with a channel bridged inside
Stasis() and not running a pbx, the channel is hung up instead of
proceeding to the desired location in dialplan. This change allows
such channels to be Redirected properly by detecting the operation
used by Redirect (ASYNCGOTO) and using the code already established
for functionality of the ARI channel continue operation.

ASTERISK-24591 #close
Review: https://reviewboard.asterisk.org/r/4271/
........

Merged revisions 430355 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430356 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoAdd the ability to continue and originate using priority labels.
Mark Michelson [Wed, 7 Jan 2015 18:54:06 +0000 (18:54 +0000)]
Add the ability to continue and originate using priority labels.

With this patch, the following two ARI commands

POST /channels
POST /channels/{id}/continue

Accept a new parameter, label, that can be used to continue to or originate
to a priority label in the dialplan.

Because this is adding a new parameter to ARI commands, the API version of
ARI has been bumped from 1.6.0 to 1.7.0.

This patch comes courtesy of Nir Simionovich from Greenfield Tech. Thanks!

ASTERISK-24412 #close
Reported by Nir Simionovich

Review: https://reviewboard.asterisk.org/r/4285
........

Merged revisions 430337 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430338 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_exten_state: Change 'does not exist' warning to notice
George Joseph [Wed, 7 Jan 2015 18:17:42 +0000 (18:17 +0000)]
res_pjsip_exten_state: Change 'does not exist' warning to notice

The 'new_subscribe: Extension <> does not exist or has no associated hint'
is a config issue and doesn't need to clutter up logs with warnings.
Changed to notice.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4307/
........

Merged revisions 430319 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430321 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_mwi: Change "MWI Subscription failed" message from warning to notice
George Joseph [Wed, 7 Jan 2015 18:15:02 +0000 (18:15 +0000)]
res_pjsip_mwi: Change "MWI Subscription failed" message from warning to notice

The "MWI Subscription failed" message means the client is trying to subscribe
to a mailbox that doesn't exist.  There's no need to clutter up logs with
warnings for a client misconfiguration so I changed it to a notice.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4306/
........

Merged revisions 430317 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430318 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agofunc_config: Add ability to retrieve specific occurrence of a variable
George Joseph [Wed, 7 Jan 2015 17:54:13 +0000 (17:54 +0000)]
func_config: Add ability to retrieve specific occurrence of a variable

I guess nobody uses templates with AST_CONFIG because today if you have a
context that inherits from a template and you call AST_CONFIG on the context,
you'll get the value from the template even if you've overridden it in the
context.  This is because AST_CONFIG only gets the first occurrence which is
always from the template.

This patch adds an optional 'index' parameter to AST_CONFIG which lets you
specify the exact occurrence to retrieve, or '-1' to retrieve the last.
The default behavior is the current behavior.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4313/
........

Merged revisions 430315 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430316 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoFix ability to perform a remote attended transfer with PJSIP.
Mark Michelson [Wed, 7 Jan 2015 17:45:56 +0000 (17:45 +0000)]
Fix ability to perform a remote attended transfer with PJSIP.

This fix has two parts:

* Corrected an error message to properly state that external_replaces is an extension. The
  error message also prints what dialplan context the external_replaces extension was being
  looked for in.
* Corrected the printing of the Replaces: header in an INVITE request. We were duplicating
  "Replaces: " in the header.

ASTERISK-24376 #close
Reported by Matt Jordan

Review: https://reviewboard.asterisk.org/r/4296
........

Merged revisions 430313 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430314 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoconfig: Add option to NOT preserve effective context when changing a template
George Joseph [Wed, 7 Jan 2015 16:56:59 +0000 (16:56 +0000)]
config: Add option to NOT preserve effective context when changing a template

Let's say you have a template T with variable VAR1 = ON and you have a
context C(T) that doesn't specify VAR1.  If you read C, the effective value
of VAR1 is ON.  Now you change T VAR1 to OFF and call
ast_config_text_file_save.  The current behavior is that the file gets
re-written with T/VAR1=OFF but C/VAR1=ON is added.  Personally, I think this
is a bug. It's preserving the effective state of C even though I didn't
specify C/VAR1 in th first place.  I believe the behavior should be that if
I didn't specify C/VAR1 originally, then the effective value of C/VAR1 should
continue to follow the inherited state.  Now, if I DID explicitly specify
C/VAR1, the it should be preserved even if the template changes.

Even though I think the existing behavior is a bug, it's been that way forever
so I'm not changing it.  Instead, I've created ast_config_text_file_save2()
that takes a bitmask of flags, one of which is to preserve the effective context
(the current behavior).  The original ast_config_text_file_save calls *2 with
the preserve flag.  If you want the new behavior, call *2 directly without a
flag.

I've also updated Manager UpdateConfig with a new parameter
'PreserveEffectiveContext' whose default is 'yes'.  If you want the new behavior
with UpdateConfig, set 'PreserveEffectiveContext: no'.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4297/
........

Merged revisions 430295 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430296 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoFix dev-mode build on recent gcc
Kinsey Moore [Wed, 7 Jan 2015 03:01:39 +0000 (03:01 +0000)]
Fix dev-mode build on recent gcc
........

Merged revisions 430274 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430275 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoBlocked revisions 430252
Matthew Jordan [Tue, 6 Jan 2015 22:46:43 +0000 (22:46 +0000)]
Blocked revisions 430252

........
contrib/ast-db-manage: Correct down_revision path for user_eq_phone

When the user_eq_phone patch was backported to 13, it referenced the downward
revision that the PJSIP optimistic encryption option also references. This
creates a multi-path upgrade Exception when generating the SQL files.

This patch corrects this in the 13 branch. Note that trunk, which already
contained both of these features, is unaffected by this problem.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430254 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_mwi: Change warning to notice
George Joseph [Tue, 6 Jan 2015 17:53:42 +0000 (17:53 +0000)]
res_pjsip_mwi: Change warning to notice

When res_pjsip loads and an endpoint auto-subscribes a mailbox for mwi,
if a contact hasn't registered yet, res_pjsip_mwi spits out a warning.
This is a perfectly normal situation though and doesn't require something
as serious as a warning.  It's also self correcting. The device will start
getting mwi as soon as it registers.

This patch changes the warning to a notice.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4314/
........

Merged revisions 430227 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430228 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agobridge_native_rtp: Change local/remote message from debug/2 to verb/4
George Joseph [Tue, 6 Jan 2015 17:49:03 +0000 (17:49 +0000)]
bridge_native_rtp: Change local/remote message from debug/2 to verb/4

Change the "Locally bridged"/"Remotely bridged" messages from dbg/2 to verb/4.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4300/
........

Merged revisions 430225 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430226 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agooutbound_registration: Add 'pjsip send register' and update 'send unregister'
George Joseph [Tue, 6 Jan 2015 17:43:16 +0000 (17:43 +0000)]
outbound_registration: Add 'pjsip send register' and update 'send unregister'

The current behavior of 'pjsip send unregister' is to send the unregister
(REGISTER with 0 exp) but let the next scheduled register proceed normally.
I don't think that's a good idea.  If you unregister, it should stay
unregistered until you decide to start registrations again.  So this patch
just adds a cancel_registration call to the current unregister_task to
cancel the timer.

Of course, now you need  a way to start registration again so I've added
a 'pjsip send register' command that unregisters and cancels any existing
registration (the same as send unregister), then sends an immediate
registration and starts the timer back up again.

Both changes also ripple to AMI.  There's a new PJSIPRegister command.

There's no harm in calling either command repeatedly.  They don't care
about the actual state.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4301/
........

Merged revisions 430223 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430224 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agopjsip cli: Fix sorting of contacts for 'pjsip list contacts'
George Joseph [Tue, 6 Jan 2015 17:29:33 +0000 (17:29 +0000)]
pjsip cli: Fix sorting of contacts for 'pjsip list contacts'

For some reason I was using a hash container instead of a list to gather the
contacts for 'pjsip list/show contacts' so even though I had a sort function,
the output wasn't sorted.  This patch just changes the hash container to a
list container and the contacts now appear sorted in the CLI.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4305/
........

Merged revisions 430221 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430222 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agobridge: avoid leaking channel during blond transfer pt2
Scott Griepentrog [Mon, 5 Jan 2015 22:50:32 +0000 (22:50 +0000)]
bridge: avoid leaking channel during blond transfer pt2

A blond transfer to a failed destination, when followed
by a recall attempt, lead to a leak of the reference to
the destination channel.  In addition to correcting the
regression on the previous attempt (r429826) this fixes
the leak and two additional reference leaks on failures
of bridge_import.

ASTERISK-24513 #close
Review: https://reviewboard.asterisk.org/r/4302/
........

Merged revisions 430199 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 430200 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430201 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agopjsip: Document addition of 'PJSIP_AOR' and 'PJSIP_CONTACT' in CHANGES file.
Joshua Colp [Mon, 5 Jan 2015 17:57:43 +0000 (17:57 +0000)]
pjsip: Document addition of 'PJSIP_AOR' and 'PJSIP_CONTACT' in CHANGES file.
........

Merged revisions 430181 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430182 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agopjsip: Add 'PJSIP_AOR' and 'PJSIP_CONTACT' dialplan functions.
Joshua Colp [Mon, 5 Jan 2015 17:53:42 +0000 (17:53 +0000)]
pjsip: Add 'PJSIP_AOR' and 'PJSIP_CONTACT' dialplan functions.

The PJSIP_AOR dialplan function allows inspection of configured AORs including
what contacts are currently bound to them.

The PJSIP_CONTACT dialplan function allows inspection of contacts in existence.
These can include both externally added (by way of registration) or permanent
ones.

ASTERISK-24341
Reported by: xrobau

Review: https://reviewboard.asterisk.org/r/4308/
........

Merged revisions 430179 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430180 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agortp_engine: keep payload types in correct range
Scott Griepentrog [Wed, 31 Dec 2014 18:54:37 +0000 (18:54 +0000)]
rtp_engine: keep payload types in correct range

In r428708 additional codecs were added including
a payload type of 128 which is outside of nominal
range of 0-127.  This change moves changes 128 to
96 to avoid causing a pjsip assertion when making
a call to an endpoint configured with allow=all.

ASTERISK-24367 #close
Review: https://reviewboard.asterisk.org/r/4286/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430164 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoPJSIP: Update transport method documentation
Kinsey Moore [Mon, 29 Dec 2014 13:14:19 +0000 (13:14 +0000)]
PJSIP: Update transport method documentation

This updates the documentation for the 'method' configuration option to
be more verbose about the behaviors of values 'unspecified' and
'default'. They do exactly the same thing which is to select the
default as defined by PJSIP which is currently TLSv1.

Review: https://reviewboard.asterisk.org/r/4264/
........

Merged revisions 430145 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430146 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoapp_queue: Update sample conf documenation
Kevin Harwell [Wed, 24 Dec 2014 21:28:14 +0000 (21:28 +0000)]
app_queue: Update sample conf documenation

Updated the queues.conf.sample file to explicitly state which channel queue
variables are propagated to.

ASTERISK-24267
Reported by: Mitch Claborn
........

Merged revisions 430126 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 430127 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430128 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agomain/pbx.c: Fix double lock of contexts lock introduced by r429967
Matthew Jordan [Wed, 24 Dec 2014 16:59:42 +0000 (16:59 +0000)]
main/pbx.c: Fix double lock of contexts lock introduced by r429967

We only need to hold the context_merge_lock once. Locking it twice will make
many other parts of Asterisk very sad.

ASTERISK-24641 #close

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430111 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agopjsip_options: Fix continued qualifies after endpoint/aor deletion
George Joseph [Tue, 23 Dec 2014 23:19:30 +0000 (23:19 +0000)]
pjsip_options: Fix continued qualifies after endpoint/aor deletion

If you remove an endpoint/aor from pjsip.conf then do a core reload,
qualifies will continue even though the object are gone.  This happens
because nothing clears out the qualify tasks.

This patch unschedules all existing qualify tasks before scheduling
new ones on reload.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4290/
........

Merged revisions 430064 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430067 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agotest_astobj2: Fix warning for missing trailing slash in category
George Joseph [Tue, 23 Dec 2014 23:16:35 +0000 (23:16 +0000)]
test_astobj2: Fix warning for missing trailing slash in category

This patch adds a trailing slash to the category for this test.
No more warning.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4295/
........

Merged revisions 430059 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430060 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoDTMF atxfer: Setup recall channels as if the transferee initiated the call.
Richard Mudgett [Mon, 22 Dec 2014 21:20:11 +0000 (21:20 +0000)]
DTMF atxfer: Setup recall channels as if the transferee initiated the call.

After the initial DTMF atxfer call attempt to the transfer target fails to
answer during a blonde transfer, the recall callback channels do not get
setup with information from the initial transferrer channel.  As a result,
the recall callback to the transferrer does not have callid, channel
variables, datastores, accountcode, peeraccount, COLP, and CLID setup.  A
similar situation happens with the recall callback to the transfer target
but it is less visible.  The recall callback to the transfer target does
not have callid, channel variables, datastores, accountcode, peeraccount,
and COLP setup.

* Added missing information to the recall callback channels before
initiating the call.  callid, channel variables, datastores, accountcode,
peeraccount, COLP, and CLID

* Set callid of the transferrer channel on the DTMF atxfer controller
thread attended_transfer_monitor_thread().

* Added missing channel unlocks and props unref to off nominal paths in
attended_transfer_properties_alloc().

ASTERISK-23841 #close
Reported by: Richard Mudgett

Review: https://reviewboard.asterisk.org/r/4259/
........

Merged revisions 430034 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430041 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoFix compilation since the patch for ASTERISK-24363 went in.
Richard Mudgett [Mon, 22 Dec 2014 20:25:40 +0000 (20:25 +0000)]
Fix compilation since the patch for ASTERISK-24363 went in.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430028 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoqueue_log: Post QUEUESTART entry when Asterisk fully boots.
Richard Mudgett [Mon, 22 Dec 2014 20:08:35 +0000 (20:08 +0000)]
queue_log: Post QUEUESTART entry when Asterisk fully boots.

The QUEUESTART log entry has historically acted like a fully booted event
for the queue_log file.  When the QUEUESTART entry was posted to the log
was broken by the change made by ASTERISK-15863.

* Made post the QUEUESTART queue_log entry when Asterisk fully boots.
This restores the intent of that log entry and happens after realtime has
had a chance to load.

AST-1444 #close
Reported by: Denis Martinez

Review: https://reviewboard.asterisk.org/r/4282/
........

Merged revisions 430009 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 430010 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430011 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agochan_sip: Send CANCEL via original INVITE destination even after UPDATE request
Matthew Jordan [Mon, 22 Dec 2014 15:40:27 +0000 (15:40 +0000)]
chan_sip: Send CANCEL via original INVITE destination even after UPDATE request

Given the following scenario:
* Three SIP phones (A, B, C), all communicating via a proxy with Asterisk
* A call is established between A and B. B performs a SIP attended transfer of
  A to C. B sets the call on hold (A is hearing MOH) and dials the extension of
  C. While phone C is ringing, B transfers the call (that is, what we typically
  call a 'blond transfer').
* When the transfer completes, A hears the ringing of phone C, while B is idle.

In the SIP messaging for the above scenario, a REFER request is sent to
transfer the call. When "sendrpid=yes" is set in sip.conf, Asterisk may send an
UPDATE request to phone C to update party information. This update is sent
directly to phone C, not through the intervening proxy. This has the unfortunate
side effect of providing route information, which is then set on the sip_pvt
structure for C. If someone (e.g. B) is trying to get the call back (through a
directed pickup), Asterisk will send a CANCEL request to C. However, since we
have now updated the route set, the CANCEL request will be sent directly to C
and not through the proxy. The phone ignores this CANCEL according to RFC3261
(Section 9.1).

This patch updates reqprep such that the route is not updated if an UPDATE
request is being sent while the INVITE state is INV_PROCEEDING or
INV_EARLY_MEDIA. This ensures that a subsequent CANCEL request is still sent
to the correct location.

Review: https://reviewboard.asterisk.org/r/4279

ASTERISK-24628 #close
Reported by: Karsten Wemheuer
patches:
  issue.patch uploaded by Karsten Wemheuer (License 5930)
........

Merged revisions 429982 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 429983 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429984 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agopresencestate: Allow channel drivers to provide presence state information
Matthew Jordan [Mon, 22 Dec 2014 14:33:24 +0000 (14:33 +0000)]
presencestate: Allow channel drivers to provide presence state information

This patch adds the ability for channel drivers to supply presence information
in a similar manner to device state. The patch does not provide any channel
driver implementations, but it does provide the core infrastructure necessary
for channel drivers to provide such information.

The core handles multiple providers of presence state information. Ordering
of presence state is as follows:
 INVALID < NOT_SET < AVAILABLE < UNAVAILABLE < CHAT < AWAY < XA < DND

Each provider can trump the previous if it provides a presence state that
supercedes a previous one.

Review: https://reviewboard.asterisk.org/r/4050

ASTERISK-24363 #close
Reported by: Gareth Palmer
patches:
  chan_presencestate-428146.patch uploaded by Gareth Palmer (License 5169)

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429967 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoapp_confbridge: Fix build error caused by XML validation errors
Matthew Jordan [Mon, 22 Dec 2014 12:16:36 +0000 (12:16 +0000)]
app_confbridge: Fix build error caused by XML validation errors

Summaries can't contain XML nodes, as they are defined to contain only text
data.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429952 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoapp_confbridge: Add the ability to pass options/command to MixMonitor
Matthew Jordan [Mon, 22 Dec 2014 02:35:05 +0000 (02:35 +0000)]
app_confbridge: Add the ability to pass options/command to MixMonitor

This patch adds the ability to pass options and a command to MixMontor when
recording a conference using ConfBridge.

New options are -

* record_options: Options to MixMontor, eg: m(), W() etc.
* record_command: The command to execute when recording is over.
* record_file_timestamp: Append the start time to the file name.

These options can also be used with the CONFBRIDGE function, e.g.,
Set(CONFBRIDGE(bridge,record_command)=/path/to/command ^{MIXMONITOR_FILENAME}))

Review: https://reviewboard.asterisk.org/r/4023

ASTERISK-24351 #close
Reported by: Gareth Palmer
patches:
  record_command-428838.patch uploaded by Gareth Palmer (License 5169)

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429934 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_phoneprovi_provider: Fix reload
George Joseph [Mon, 22 Dec 2014 00:17:49 +0000 (00:17 +0000)]
res_pjsip_phoneprovi_provider: Fix reload

Reloading wasn't working correctly because on a reload, the sorcery apply
handler was never being called for unchanged users.  So, instead of using
an apply handler, I'm now iterating over all users.  Works much more reliably.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4288/
........

Merged revisions 429914 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429915 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoacl: Fix reloading of configuration if configuration file does not exist at startup.
Joshua Colp [Sat, 20 Dec 2014 20:57:47 +0000 (20:57 +0000)]
acl: Fix reloading of configuration if configuration file does not exist at startup.

The named ACL code incorrectly destroyed the config options information if loading
of the configuration file failed at startup. This would result in reloading
also failing even if a valid configuration file was put in place.

ASTERISK-23733 #close
Reported by: Richard Kenner
........

Merged revisions 429893 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 429894 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429895 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_http_websocket.c: Fix incorrect use of sizeof in ast_websocket_write().
Richard Mudgett [Fri, 19 Dec 2014 20:56:12 +0000 (20:56 +0000)]
res_http_websocket.c: Fix incorrect use of sizeof in ast_websocket_write().

This won't fix the reported issue but it is an incorrect use of sizeof.

ASTERISK-24566
Reported by:  Badalian Vyacheslav
........

Merged revisions 429867 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 429868 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429870 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agochan_dahdi: Don't ignore setvar when using configuration section scheme.
Richard Mudgett [Fri, 19 Dec 2014 17:34:33 +0000 (17:34 +0000)]
chan_dahdi: Don't ignore setvar when using configuration section scheme.

When the configuration section scheme of chan_dahdi.conf is used (keyword
dahdichan instead of channel) all setvar= options are completely ignored.
No variable defined this way appears in the created DAHDI channels.

* Move the clearing of setvar values to after the deferred processing of
dahdichan.

AST-1378 #close
Reported by: Guenther Kelleter
Patch by: Guenther Kelleter
........

Merged revisions 429825 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 429829 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429830 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agobridge: avoid leaking channel during blond transfer
Scott Griepentrog [Fri, 19 Dec 2014 17:27:25 +0000 (17:27 +0000)]
bridge: avoid leaking channel during blond transfer

After a blond transfer (start attended and hang up)
to a destination that also hangs up without answer,
the Local;1 channel was leaked and would show up on
core show channels.  This was happening because the
attended state blond_nonfinal_enter() resetting the
props->transfer_target to null while releasing it's
own reference, which would later prevent props from
releasing another reference during destruction. The
change made here is simply to not assign the target
to NULL.

ASTERISK-24513 #close
Reported by: Mark Michelson
Review: https://reviewboard.asterisk.org/r/4262/
........

Merged revisions 429826 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 429827 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429828 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agochan_dahdi.c, res_rtp_asterisk.c: Change some spammy debug messages to level 5.
Richard Mudgett [Thu, 18 Dec 2014 22:40:16 +0000 (22:40 +0000)]
chan_dahdi.c, res_rtp_asterisk.c: Change some spammy debug messages to level 5.

ASTERISK-24337 #close
Reported by: Rusty Newton
........

Merged revisions 429804 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 429805 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429806 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agochan_dahdi: Populate CALLERID(ani2) for incoming calls in featdmf signaling mode.
Richard Mudgett [Thu, 18 Dec 2014 20:09:21 +0000 (20:09 +0000)]
chan_dahdi: Populate CALLERID(ani2) for incoming calls in featdmf signaling mode.

For the featdmf signaling mode the incoming MF Caller-ID information is
formatted as follows: *${CALLERID(ani2)}${CALLERID(ani)}#*${EXTEN}#

Rather than discarding the ani2 digits, populate the CALLERID(ani2) value
with what is received instead.

AST-1368 #close
Reported by: Denis Martinez
Patches:
      extract_ani2_for_featdmf_v11.patch (license #5621) patch uploaded by Richard Mudgett
........

Merged revisions 429783 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 429784 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429785 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_sdp_rtp: wrong bridge chosen when the DTMF mode is not compatible
Kevin Harwell [Thu, 18 Dec 2014 15:55:03 +0000 (15:55 +0000)]
res_pjsip_sdp_rtp: wrong bridge chosen when the DTMF mode is not compatible

A native rtp bridge was being chosen (it shouldn't have been) when using two
pjsip channels with incompatible DTMF modes.  This patch sets the rtp instance
property, AST_RTP_PROPERTY_DTMF, for the appropriate DTMF mode(s) for pjsip.
It was not being set before, meaning all DTMF modes for pjsip were being treated
as compatible, thus native bridging would be chosen as the bridge type when it
shouldn't have been.

ASTERISK-24459 #close
Reported by: Yaniv Simhi
Review: https://reviewboard.asterisk.org/r/4265/
........

Merged revisions 429763 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429764 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoPrevent potential infinite outbound authentication loops in registration.
Mark Michelson [Thu, 18 Dec 2014 15:40:13 +0000 (15:40 +0000)]
Prevent potential infinite outbound authentication loops in registration.

Prior to this patch, Asterisk would always respond to 401 responses to
registration attempts by trying to provide a registration with authentication
credentials. Even if subsequent attempts were rejected with 401 responses,
Asterisk would continue this behavior. If authentication credentials were
incorrect, this could continue forever.

With this patch, we keep track of whether we have attempted authentication
on an outbound registration attempt. If we already have, we don not try
again until the next attempt. This prevents the infinite loop scenario.

Review: https://reviewboard.asterisk.org/r/4273
........

Merged revisions 429761 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429762 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoPrevent possible race condition on dual redirect of channels in the same bridge.
Mark Michelson [Thu, 18 Dec 2014 15:18:45 +0000 (15:18 +0000)]
Prevent possible race condition on dual redirect of channels in the same bridge.

The AST_FLAG_BRIDGE_DUAL_REDIRECT_WAIT flag was created to prevent bridges from
prematurely acting on orphaned channels in bridges. The problem with the AMI
redirect action was that it was setting this flag on channels based on the presence
of a PBX, not whether the channel was in a bridge. Whether a channel has a PBX
is irrelevant, so the condition has been altered to check if the channel is in a
bridge.

ASTERISK-24536 #close
Reported by Niklas Larsson

Review: https://reviewboard.asterisk.org/r/4268
........

Merged revisions 429741 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429745 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoEnsure the correct value is returned for CHANNEL(pjsip, secure)
Mark Michelson [Thu, 18 Dec 2014 14:50:06 +0000 (14:50 +0000)]
Ensure the correct value is returned for CHANNEL(pjsip, secure)

Prior to this patch, we were using the PJSIP dialog's secure flag
to determine if a secure transport was being used. Unfortunately,
the dialog's secure flag was only set if a SIPS URI were in use,
as required by RFC 3261 sections 12.1.1 and 12.1.2. What we're interested
in is not dialog security, but transport security. This code change
switches to a model where we use the dialog's target URI to determine
what transport would be used to communicate, and then check if that
transport is secure.

AST-1450 #close
Reported by John Bigelow

Review: https://reviewboard.asterisk.org/r/4277
........

Merged revisions 429739 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429740 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_config_wizard: fix unload SEGV
George Joseph [Thu, 18 Dec 2014 00:11:24 +0000 (00:11 +0000)]
res_pjsip_config_wizard: fix unload SEGV

If certain pjsip modules aren't loaded, the wizard causes a SEGV
when it unloads.  Added a check for the presense of the object
type wizard before trying to clean it up.

Tested-by: George Joseph
........

Merged revisions 429719 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429720 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_config_wizard: Change FILEUNCHANGED config_load2 flag determination
George Joseph [Wed, 17 Dec 2014 23:06:01 +0000 (23:06 +0000)]
res_pjsip_config_wizard: Change FILEUNCHANGED config_load2 flag determination

The module now applies the FILEUNCHANGED flag when both reloaded is
specified AND there's no last_config for the object type.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4276/
........

Merged revisions 429699 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429700 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoFix printf problems with high ascii characters after r413586 (1.8).
Walter Doekes [Wed, 17 Dec 2014 10:23:32 +0000 (10:23 +0000)]
Fix printf problems with high ascii characters after r413586 (1.8).

In r413586 (1.8) various casts were added to silence gcc 4.10 warnings.
Those fixes included things like:

    -out += sprintf(out, "%%%02X", (unsigned char) *ptr);
    +out += sprintf(out, "%%%02X", (unsigned) *ptr);

That works for low ascii characters, but for the high range that yields
e.g. FFFFFFC3 when C3 is expected.

This changeset:
- fixes those casts to use the 'hh' unsigned char modifier instead
- consistently uses %02x instead of %2.2x (or other non-standard usage)
- adds a few 'h' modifiers in various places
- fixes a 'replcaes' typo
- dev/urandon typo (in 13+ patch)

Review: https://reviewboard.asterisk.org/r/4263/

ASTERISK-24619 #close
Reported by: Stefan27 (on IRC)
........

Merged revisions 429673 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 429674 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 429675 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429683 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_config_wizard: fix test breakage
George Joseph [Tue, 16 Dec 2014 17:53:59 +0000 (17:53 +0000)]
res_pjsip_config_wizard: fix test breakage

Fix test breakage caused by not checking for res_pjsip before
calling ast_sip_get_sorcery.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4269/
........

Merged revisions 429653 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429654 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agochan_sip: Allow T.38 switch-over when SRTP is in use.
Joshua Colp [Tue, 16 Dec 2014 16:39:47 +0000 (16:39 +0000)]
chan_sip: Allow T.38 switch-over when SRTP is in use.

Previously when SRTP was enabled on a channel it was not possible
to switch to T.38 as no crypto attributes would be present.

This change makes it so it is now possible. If a T.38 re-invite
comes in SRTP is terminated since in practice you can't encrypt
a UDPTL stream. Now... if we were doing T.38 over RTP (which
does exist) then we'd have a chance but almost nobody does that so
here we are.

ASTERISK-24449 #close
Reported by: Andreas Steinmetz
patches:
 udptl-ignore-srtp-v2.patch submitted by Andreas Steinmetz (license 6523)
........

Merged revisions 429632 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 429633 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429634 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_t38: Fix T.38 failure when peer reinvites immediately.
Joshua Colp [Tue, 16 Dec 2014 15:44:43 +0000 (15:44 +0000)]
res_pjsip_t38: Fix T.38 failure when peer reinvites immediately.

If a remote endpoint reinvites to T.38 immediately the state machine
will go into a peer reinvite state. If a T.38 capable application
(such as ReceiveFax) queries it will receive this state. Normally
the application will then indicate so that the channel driver will
queue up the T.38 offer previously received. Once it receives this
offer the application will act normally and negotiate.

The res_pjsip_t38 module incorrectly partially squashed this indication.
This would cause the application to think the request had failed when
in reality it had actually worked.

This change makes it so that no T.38 control frames (or indications)
are squashed.
........

Merged revisions 429612 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429613 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_config_wizard: Allow streamlined config of common pjsip scenarios
George Joseph [Mon, 15 Dec 2014 17:08:24 +0000 (17:08 +0000)]
res_pjsip_config_wizard: Allow streamlined config of common pjsip scenarios

res_pjsip_config_wizard
------------------
 * This is a new module that adds streamlined configuration capability for
   chan_pjsip.  It's targetted at users who have lots of basic configuration
   scenarios like 'phone' or 'agent' or 'trunk'.  Additional information
   can be found in the sample configuration file at
   config/samples/pjsip_wizard.conf.sample.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4190/
........

Merged revisions 429592 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429593 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoActivate persistent subscriptions when they are recreated.
Mark Michelson [Mon, 15 Dec 2014 15:48:47 +0000 (15:48 +0000)]
Activate persistent subscriptions when they are recreated.

Prior to this change, recreating persistent subscriptions would
create the subscription but would not activate it. This led to subscriptions
being listed in the "NULL" state by diagnostics and not sending NOTIFYs
when expected.

Review: https://reviewboard.asterisk.org/r/4261
........

Merged revisions 429571 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429573 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoloader: Move definition of ast_module_reload from _private.h to module.h
George Joseph [Fri, 12 Dec 2014 23:57:50 +0000 (23:57 +0000)]
loader: Move definition of ast_module_reload from _private.h to module.h

No functionality change.  Just move the definition of ast_module_reload
from _private.h to module.h so it can be public.

Also removed the include of _private.h from manager.c since ast_module_load
was the only reason for including it.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4251/
........

Merged revisions 429542 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429543 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoDEBUG_THREADS: Fix regression and lock tracking initialization problems.
Richard Mudgett [Fri, 12 Dec 2014 23:49:36 +0000 (23:49 +0000)]
DEBUG_THREADS: Fix regression and lock tracking initialization problems.

This patch started with David Lee's patch at
https://reviewboard.asterisk.org/r/2826/ and includes a regression fix
introduced by the ASTERISK-22455 patch.

The initialization of a mutex's lock tracking structure was not protected
in a critical section.  This is fine for any mutex that is explicitly
initialized, but a static mutex may have its lock tracking double
initialized if multiple threads attempt the first lock simultaneously.

* Added a global mutex to properly serialize initialization of the lock
tracking structure.  The painful global lock can be mitigated by adding a
double checked lock flag as discussed on the original review request.

* Defer lock tracking initialization until first use.

* Don't be "helpful" and initialize an uninitialized lock when
DEBUG_THREADS is enabled.  Debug code is not supposed to fix or change
normal code behavior.  We don't need a lock initialization race that would
force a re-setup of lock tracking.  Lock tracking already handles
initialization on first use.

* Properly handle allocation failures of the lock tracking structure.

* No need to initialize tracking data in __ast_pthread_mutex_destroy()
just to turn around and destroy it.

The regression introduced by ASTERISK-22455 is the result of manipulating
a pthread_mutex_t struct outside of the pthread library code.  The
pthread_mutex_t struct seems to have a global linked list pointer member
that can get changed by other threads.  Therefore, saving and restoring
the contents of a pthread_mutex_t struct is a bad thing.

Thanks to Thomas Airmont for finding this obscure regression.

* Don't overwrite the struct ast_lock_track.reentr_mutex member to restore
tracking data in __ast_cond_wait() and __ast_cond_timedwait().  The
pthread_mutex_t struct must be treated as a read-only opaque variable.

Miscellaneous other items fixed by this patch:

* Match ast_suspend_lock_info() with ast_restore_lock_info() in
__ast_cond_timedwait().

* Made some uninitialized lock sanity checks return EINVAL and try a
DO_THREAD_CRASH.

* Fix bad canlog initialization expressions.

ASTERISK-24614 #close
Reported by: Thomas Airmont

Review: https://reviewboard.asterisk.org/r/4247/
Review: https://reviewboard.asterisk.org/r/2826/
........

Merged revisions 429539 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429541 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores/res_agi: Make Verbose message for 'stream file' match other playbacks
Matthew Jordan [Fri, 12 Dec 2014 22:54:02 +0000 (22:54 +0000)]
res/res_agi: Make Verbose message for 'stream file' match other playbacks

The Verbose message displayed when a file is played back via 'stream file'
was formatted differently than other playbacks:
* It didn't include the channel name
* It didn't include the channel language
It does, however, include the playback offset as well as any escape digits.
That information was kept; however, this patch updates the formatting to more
closely match the Verbose messages displayed when a file is played back by
'control stream file', Playback, ControlPlayback, or any other file playback
operation.
........

Merged revisions 429519 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429520 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agomedia: Fix crash when determining sample count of a frame during shutdown.
Joshua Colp [Fri, 12 Dec 2014 17:01:42 +0000 (17:01 +0000)]
media: Fix crash when determining sample count of a frame during shutdown.

When shutting down Asterisk the codecs are cleaned up. As a result anything
attempting to get a codec based on ID or details will find that no codec
exists. This currently occurs when determining the sample count of a frame.
This code did not take this situation into account.

This change fixes this by getting the codec directly from the format and
eliminates the lookup. This is both faster and also provides a guarantee
that the codec will exist and will be valid.

ASTERISK-24604 #close
Reported by: Matt Jordan

Review: https://reviewboard.asterisk.org/r/4260/
........

Merged revisions 429497 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429498 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agochan_pjsip: Race between channel answer and bridge setup when using direct media
Kevin Harwell [Fri, 12 Dec 2014 15:31:38 +0000 (15:31 +0000)]
chan_pjsip: Race between channel answer and bridge setup when using direct media

When direct media is enabled and a pjsip channel is answered a race would occur
between the handling of the answer and bridge setup. Sometimes the media
negotiation would take place after the native bridge was setup. This resulted
in a NULL media address, which in turn resulted in Asterisk using its address
as the remote media address when sending a reinvite.  This patch makes the
chan_pjsip answer handler synchronous thus alleviating the race condition (the
bridge won't start setting things up until after it returns).

ASTERISK-24563 #close
Reported by: Steve Pitts
Review: https://reviewboard.asterisk.org/r/4257/
........

Merged revisions 429477 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429478 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoFix crash for sorcery misconfigs
David M. Lee [Fri, 12 Dec 2014 15:03:16 +0000 (15:03 +0000)]
Fix crash for sorcery misconfigs

res_pjsip_outbound_publish was missing the CHECK_PJSIP_MODULE_LOADED()
call in load_module, and would crash with a segfault if res_pjsip
declined to load.

Review: https://reviewboard.asterisk.org/r/4258/
........

Merged revisions 429457 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429458 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoPJSIP: Allow use of 'inactive' streams for hold
Kinsey Moore [Fri, 12 Dec 2014 14:12:38 +0000 (14:12 +0000)]
PJSIP: Allow use of 'inactive' streams for hold

This allows use of the 'inactive' stream direction identifier to be
used for hold where 'sendonly' is normally used. Some Seimens phones
use 'inactive' and this change allows music on hold to operate
properly.

Review: https://reviewboard.asterisk.org/r/4252/
Reported by: Steve Pitts
........

Merged revisions 429432 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 429433 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429434 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoSorcery: Log when old config remains in use
Kinsey Moore [Fri, 12 Dec 2014 14:04:06 +0000 (14:04 +0000)]
Sorcery: Log when old config remains in use

This adds a log message notifying the user that a stale configuration
is in place upon reload when a config object fails to load. This
situation can end up causing confusion when the object failed to load
but exists from a previous config load especially when the old config
is significantly different from the new config.

Review: https://reviewboard.asterisk.org/r/4250/
Reported by: Thomas Thompson
........

Merged revisions 429429 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 429430 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429431 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_session: Delay sending BYE if a re-INVITE transaction is in progress.
Joshua Colp [Fri, 12 Dec 2014 13:06:24 +0000 (13:06 +0000)]
res_pjsip_session: Delay sending BYE if a re-INVITE transaction is in progress.

Given the scenario where a PJSIP channel is in a native RTP bridge with direct
media and the channel is then hung up the code will currently re-INVITE the channel
back to Asterisk and send a BYE at the same time. Many SIP implementations dislike
this greatly.

This change makes it so that if a re-INVITE transaction is in progress the BYE
is queued to occur after the completion of the transaction (be it through normal
means or a timeout).

Review: https://reviewboard.asterisk.org/r/4248/
........

Merged revisions 429409 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429410 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_session: Fix issue where a declined media stream in a re-INVITE would fail...
Joshua Colp [Fri, 12 Dec 2014 12:32:13 +0000 (12:32 +0000)]
res_pjsip_session: Fix issue where a declined media stream in a re-INVITE would fail SDP negotiation.

In the past the SDP negotiation within res_pjsip_session was made more tolerant of
certain situations. The only case where SDP negotiation will fail is when a major
error occurs during negotiation. Receiving an already declined media stream is
not considered a major error.

When producing the local SDP the logic took this into account so on the initial INVITE
the declined media stream did not cause an SDP negotiation failure. Unfortunately
the logic for handling media streams with a handler did not mirror this logic and
considered an already declined media stream an error and thus failed the SDP
negotiation.

This change makes the logic between both situations match so only under major
errors will the SDP negotiation fail.

ASTERISK-24607 #close
Reported by: Matt Jordan

Review: https://reviewboard.asterisk.org/r/4254/
........

Merged revisions 429407 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429408 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoARI/AMI: Include language in standard channel snapshot output
Kevin Harwell [Thu, 11 Dec 2014 20:32:21 +0000 (20:32 +0000)]
ARI/AMI: Include language in standard channel snapshot output

The CHANGES verbiage for the "language" addition had been put under the wrong
release. This moves it to be under 13.1 to 13.2 changes.

ASTERISK-24553
Reported by: Matt Jordan
........

Merged revisions 429387 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429388 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoStasis: Update unittest for channel snapshots
Kinsey Moore [Thu, 11 Dec 2014 13:53:39 +0000 (13:53 +0000)]
Stasis: Update unittest for channel snapshots

This adjusts the unit test for channel snapshots to take the new
language key into account.
........

Merged revisions 429352 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429353 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoARI/AMI: Include language in standard channel snapshot output
Kevin Harwell [Wed, 10 Dec 2014 15:43:48 +0000 (15:43 +0000)]
ARI/AMI: Include language in standard channel snapshot output

Adding information about including "language" in the standard channel snapshot
output to the CHANGES file. Note the actual source changes have already been
previously committed.

ASTERISK-24553
Reported by: Matt Jordan
........

Merged revisions 429325 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 429326 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429327 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_http_websocket: Fix crash due to double freeing memory when receiving a payload...
Joshua Colp [Wed, 10 Dec 2014 13:35:52 +0000 (13:35 +0000)]
res_http_websocket: Fix crash due to double freeing memory when receiving a payload length of zero.

Frames with a payload length of 0 were incorrectly handled in res_http_websocket.
Provided a frame with a payload had been received prior it was possible for a double
free to occur. The realloc operation would succeed (thus freeing the payload) but be
treated as an error. When the session was then torn down the payload would be
freed again causing a crash. The read function now takes this into account.

This change also fixes assumptions made by users of res_http_websocket. There is no
guarantee that a frame received from it will be NULL terminated.

ASTERISK-24472 #close
Reported by: Badalian Vyacheslav

Review: https://reviewboard.asterisk.org/r/4220/
Review: https://reviewboard.asterisk.org/r/4219/
........

Merged revisions 429270 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 429272 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 429273 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429274 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoPJSIP: Fix assert on initial mass qualify
Kinsey Moore [Wed, 10 Dec 2014 13:16:19 +0000 (13:16 +0000)]
PJSIP: Fix assert on initial mass qualify

This fixes the MWI test regressions caused by r429127 and ensures that
contacts have non-zero qualify_frequency before attempting scheduling.
........

Merged revisions 429245 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 429246 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429247 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agocore: avoid possible asterisk -r crash from long id
Scott Griepentrog [Tue, 9 Dec 2014 20:47:05 +0000 (20:47 +0000)]
core: avoid possible asterisk -r crash from long id

When connecting to the remote console, an id string
is first provided that consts of the hostname, pid,
and version.  This is parsed by the remote instance
using a buffer that may be too short, and can allow
a buffer overrun because it is not terminated. This
patch adds termination and a larger buffer.

Review: https://reviewboard.asterisk.org/r/4182/
........

Merged revisions 429223 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429224 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoARI/AMI: Include language in standard channel snapshot output
Kevin Harwell [Tue, 9 Dec 2014 20:20:27 +0000 (20:20 +0000)]
ARI/AMI: Include language in standard channel snapshot output

The channel "language" was already part of a channel snapshot, however is was
not sent out over AMI or ARI. This patch makes it so the channel "language" is
included in the appropriate AMI or ARI events.

ASTERISK-24553 #close
Reported by: Matt Jordan
Review: https://reviewboard.asterisk.org/r/4245/
........

Merged revisions 429204 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 429206 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429209 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoDirect Media calls within private network sometimes get one way audio
Kevin Harwell [Tue, 9 Dec 2014 20:03:22 +0000 (20:03 +0000)]
Direct Media calls within private network sometimes get one way audio

When endpoints with direct_media enabled, behind a firewall (Asterisk on a
separate network) and were bridged sometimes Asterisk would send the ip
address of the firewall in the sdp to one of the phones in the reinvite
resulting in one way audio. When sending the reinvite Asterisk will retrieve
the media address from the associated rtp instance, but if frames were being
read this can be overwritten with another address (in this case the
firewall's).  This patch ensures that Asterisk uses the original device
address when using direct media.

ASTERISK-24563
Reported by: Steve Pitts
Review: https://reviewboard.asterisk.org/r/4216/
........

Merged revisions 429195 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 429196 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429197 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_outbound_publish: stack overflow when using non-default sorcery wizard
Kevin Harwell [Tue, 9 Dec 2014 18:36:47 +0000 (18:36 +0000)]
res_pjsip_outbound_publish: stack overflow when using non-default sorcery wizard

When using a non-default sorcery wizard (in this instance realtime) for outbound
publishes Asterisk will crash after a stack overflow occurs due to the code
infinitely recursing.  The fix entails removing the outbound publish state
dependency from the outbound publish sorcery object and instead keeping an in
memory container that can be used to lookup the state when needed.

ASTERISK-24514 #close
Reported by: Mark Michelson
Review: https://reviewboard.asterisk.org/r/4178/
........

Merged revisions 429175 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429176 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoari: Add support for specifying an originator channel when originating.
Joshua Colp [Tue, 9 Dec 2014 15:45:19 +0000 (15:45 +0000)]
ari: Add support for specifying an originator channel when originating.

If an originator channel is specified when originating a channel the linked ID
of it will be applied to the newly originated outgoing channel. This allows
an association to be made between the two so it is known that the originator
has dialed the originated channel.

ASTERISK-24552 #close
Reported by: Matt Jordan

Review: https://reviewboard.asterisk.org/r/4243/
........

Merged revisions 429153 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429154 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoPJSIP: Stagger outbound qualifies
Kinsey Moore [Tue, 9 Dec 2014 14:01:43 +0000 (14:01 +0000)]
PJSIP: Stagger outbound qualifies

This change staggers initiation of outbound qualify (OPTIONS) attempts
to reduce instantaneous server load and prevent network congestion.

Review: https://reviewboard.asterisk.org/r/4246/
ASTERISK-24342 #close
Reported by: Richard Mudgett
........

Merged revisions 429127 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 429128 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429129 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoAMI/ARI: Update version to 2.6.0/1.6.0 respectively for new features
Matthew Jordan [Mon, 8 Dec 2014 16:54:43 +0000 (16:54 +0000)]
AMI/ARI: Update version to 2.6.0/1.6.0 respectively for new features

AMI/ARI are getting a few enhancements in the next release of Asterisk 13. Per
semantic versioning, that warrants a bump in the minor version number, as it
reflects a backwards compatible change. Hence, this commit.
........

Merged revisions 429091 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429092 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoFix a crash that would occur when receiving a 491 response to a reinvite.
Mark Michelson [Mon, 8 Dec 2014 16:43:00 +0000 (16:43 +0000)]
Fix a crash that would occur when receiving a 491 response to a reinvite.

The reviewboard description does a fine job of summarizing this, so here it is:

A reporter discovered that Asterisk would crash when attempting to retransmit
a reinvite that had previously received a 491 response. The crash occurred
because a pjsip_tx_data structure was being saved for reuse, but its reference
count was not being increased. The result was that the pjsip_tx_data was being
freed before we were actually done with it. When we attempted to re-use the
structure when re-sending the reinvite, Asterisk would crash.

The fix implemented here is not to try holding onto the pjsip_tx_data at all.
Instead, when we reschedule sending the reinvite, we create a brand new
pjsip_tx_data and send that instead. Because of this change, there is no need
for an ast_sip_session_delayed_request structure to have a pjsip_tx_data on
it any more. So any code referencing its use has been removed.

When this initial fix was introduced, I encountered a second crash when
processing a subsequent 200 OK on a rescheduled reinvite. The reason was
that when rescheduling the reinvite, we gave the wrong location for a
response callback. This has been fixed in this patch as well.

ASTERISK-24556 #close
Reported by Abhay Gupta

Review: https://reviewboard.asterisk.org/r/4233
........

Merged revisions 429089 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429090 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoAdd new AMI and ARI events for connected line changes on a channel.
Mark Michelson [Mon, 8 Dec 2014 16:24:36 +0000 (16:24 +0000)]
Add new AMI and ARI events for connected line changes on a channel.

The AMI event is called NewConnectedLine and the ARI event is called
ChannelConnectedLine.

ASTERISK-24554 #close
Reported by Matt Jordan

Review: https://reviewboard.asterisk.org/r/4231
........

Merged revisions 429064 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429084 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoStasis: Fix StasisStart/End order and missing events
Kinsey Moore [Mon, 8 Dec 2014 15:45:46 +0000 (15:45 +0000)]
Stasis: Fix StasisStart/End order and missing events

This corrects several bugs that currently exist in the stasis
application code.

* After a masquerade, the resulting channels have channel topics that
  do not match their uniqueids
** Masquerades now swap channel topics appropriately
* StasisStart and StasisEnd messages are leaked to observer
  applications due to being published on channel topics
** StasisStart and StasisEnd publishing is now properly restricted
   to controlling apps via app topics
* Race conditions exist where StasisStart and StasisEnd messages due to
  a masquerade may be received out of order due to being published on
  different topics
** These messages are now published directly on the app topic so this
   is now a non-issue
* StasisEnds are sometimes missing when sent due to masquerades and
  bridge swaps into and out of Stasis()
** This was due to StasisEnd processing adjusting message-sent flags
   after Stasis() had already exited and Stasis() had been re-entered
** This was corrected by adjusting these flags prior to sending the
   message while the initial Stasis() application was still shutting
   down

Review: https://reviewboard.asterisk.org/r/4213/
ASTERISK-24537 #close
Reported by: Matt DiMeo
........

Merged revisions 429061 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 429062 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429063 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores/res_monitor: Reset in/out sample counts on Monitor start
Matthew Jordan [Sat, 6 Dec 2014 18:16:49 +0000 (18:16 +0000)]
res/res_monitor: Reset in/out sample counts on Monitor start

When repeatedly starting/stopping a Monitor on a channel, the accumulated
in/out sample counts are never reset to 0. This can cause inadvertent jumps
in the recordings, as the code in the channel core will determine incorrectly
that a jump in the recorded file position should occur. Setting the sample
counts to 0 simply reflects the initial state a Monitor should be in when it
is started, as this is the initial count that would be on the channels at that
time.

ASTERISK-24573 #close
Reported by: Nuno Borges
patches:
  24573.patch uploaded by Nuno Borges (License 6116)
........

Merged revisions 429031 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 429032 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 429033 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429034 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoapps/app_meetme: Apply default values on initial load with no config file
Matthew Jordan [Sat, 6 Dec 2014 17:36:21 +0000 (17:36 +0000)]
apps/app_meetme: Apply default values on initial load with no config file

When the app_meetme module is loaded without its configuration file, the
module settings aren't initialized. In particular, this impacts the use
of logging realtime members. This patch guarantees that we always set the
default module settings on initial load.

Review: https://reviewboard.asterisk.org/r/4242/

ASTERISK-24572 #close
Reported by: Nuno Borges
patches:
  24572.patch uploaded by Nuno Borges (License 6116)
........

Merged revisions 429027 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 429028 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 429029 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429030 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agosorcery: Add additional observer capabilities.
George Joseph [Fri, 5 Dec 2014 17:08:27 +0000 (17:08 +0000)]
sorcery: Add additional observer capabilities.

Add new global, instance and wizard observers.
instance_created
wizard_registered
wizard_unregistered
instance_destroying
instance_loading
instance_loaded
wizard_mapped
object_type_registered
object_type_loading
object_type_loaded
wizard_loading
wizard_loaded

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4215/
........

Merged revisions 428999 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 429000 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@429001 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agomain/test: Fix compilation issue on 32-bit systems
Matthew Jordan [Thu, 4 Dec 2014 17:13:48 +0000 (17:13 +0000)]
main/test: Fix compilation issue on 32-bit systems

On a 32-bit system, a type of intmax_t will result in a compilation warning
when formatted as a 'long int'. Use the format specifier of %jd (which was
what was used originally in manager.c) to format the JSON extracted integer
on both 32-/64-bit systems.
........

Merged revisions 428972 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 428973 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428974 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agomain/test: Fix race condition between AMI topic and Test Suite topic
Matthew Jordan [Thu, 4 Dec 2014 15:48:26 +0000 (15:48 +0000)]
main/test: Fix race condition between AMI topic and Test Suite topic

This patch fixes a race condition between the raising of test AMI events (which
drive many tests in the Asterisk Test Suite) and other AMI events. Prior to
this patch, the Stasis messages published to the test topic were not forwarded
to the AMI topic. Instead, the code in manager had a dedicated handler for test
messages that was independent of the topics forwarded to the AMI topic. This
results in no synchronization between the test messages and the rest of the
Stasis messages published out over AMI. In some test with very tight timing
constraints, this can result in out of order messages and spurious test
failures. Properly forwarding the Test Suite topic to the AMI topic ensures
that the messages are synchronized properly.

This patch does that, and moves the message handling to the Stasis definition
of the Test Suite message in test.c as well.

Review: https://reviewboard.asterisk.org/r/4221/
........

Merged revisions 428945 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 428946 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428947 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agotests/test_cel: Add test_cel_attended_transfer_bridges_link to racey tests
Matthew Jordan [Wed, 3 Dec 2014 20:59:01 +0000 (20:59 +0000)]
tests/test_cel: Add test_cel_attended_transfer_bridges_link to racey tests

Despite failing less often, the ordering of the ATTENDEDTRANSFER event and the
BRIDGE_EXIT event for the Alice and David channels is not defined. This makes
the test still fail.
........

Merged revisions 428918 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 428919 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428920 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agotests/test_cel: Fix CEL unit test failures caused by attended transfer changes
Matthew Jordan [Wed, 3 Dec 2014 19:49:38 +0000 (19:49 +0000)]
tests/test_cel: Fix CEL unit test failures caused by attended transfer changes

When the publication of attended transfer messages were pushed to another
thread, some subtle race conditions were introduced with the CEL unit tests.
This patch fixes one of them, and pushes the other to ASTERISK-22367, which
already exists to fix another bouncy CEL unit test.

In particular, this patch fixes the test_cel_attended_transfer_bridges_link
test, and defers the test_cel_attended_transfer_bridges_swap test to the
aforementioned JIRA issue.

ASTERISK-22367
........

Merged revisions 428891 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 428892 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428893 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoapps/app_voicemail: Fix crash with IMAP when streams are opened simultaneously
Matthew Jordan [Wed, 3 Dec 2014 16:45:24 +0000 (16:45 +0000)]
apps/app_voicemail: Fix crash with IMAP when streams are opened simultaneously

The UW IMAP library is instrinsically not thread-safe, and relies upon higher
level applications to guarantee thread safety. For the most part, this is
provided by the vms object, which provides locking for individual streams.
Unfortunately, this is not sufficient for calls to mail_open which create the
IMAP stream. mail_open can, on some systems, call into a UW IMAP specific
function for determining the address of a system based on a hostname,
ip_nametoaddr.

In the ip6_unix implementation of this function, static variables are used
to hold parsing buffers. This can cause a crash if multiple threads attempt
to convert a hostname to an address at the same time. Locking on a single
mail stream is not sufficient to prevent simultaneous access to these static
variables.

In the IMAP library, this function can be called from the mail_open and
imap_status functions. As the imap_status function is not used by
app_voicemail, locking on access to mail_open is sufficient to prevent
any mangling of the buffers.

Review: https://reviewboard.asterisk.org/r/4188/

ASTERISK-24516 #close
Reported by: David Duncan Ross Palmer
Tested by: David Duncan Ross Palmer
patches:
  ASTERISK-24516.diff uploaded by David Duncan Ross Palmer (License 6660)
........

Merged revisions 428863 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 428864 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 428865 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428866 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoCHANGES: Add item for new 'pjsip show identif(y|ies) commands
George Joseph [Tue, 2 Dec 2014 21:54:05 +0000 (21:54 +0000)]
CHANGES: Add item for new 'pjsip show identif(y|ies) commands

Tested-by: George Joseph
........

Merged revisions 428836 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 428837 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428838 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agotests/test_stasis: Resolve compilation issues from Asterisk 12 merge
Matthew Jordan [Tue, 2 Dec 2014 19:04:02 +0000 (19:04 +0000)]
tests/test_stasis: Resolve compilation issues from Asterisk 12 merge

When merging the changes up stream in r428687, I missed the fact that the
signature for stasis_message_type_create was changed. This patch fixes
the compilation issues introduced by that merge.
........

Merged revisions 428815 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428816 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agopbx/pbx_loopback: Speed up switches by avoiding unneeded lookups
Matthew Jordan [Tue, 2 Dec 2014 17:10:57 +0000 (17:10 +0000)]
pbx/pbx_loopback: Speed up switches by avoiding unneeded lookups

This patch makes a small rearrangement to only do dialplan lookups during
loopback switches if the pattern matches. Prior to this patch, the dialplan
lookups were always performed, even when the result would be discarded.
Dialplan lookups can be very costly if remote switches - like DUNDi - are
present. In those cases extension matching is sped up considerably, making
the issue of lost digits more manageable.

As collateral damage, 6 trailing spaces were killed.

Review: https://reviewboard.asterisk.org/r/4211

ASTERISK-24577 #close
Reported by: Birger Harzenetter
patches:
  ast-loopback.patch uploaded by Birger Harzenetter (License 5870)
........

Merged revisions 428787 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 428788 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 428789 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428790 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agores_pjsip_refer: Fix issue where native bridge may not occur upon completion of a...
Joshua Colp [Tue, 2 Dec 2014 12:21:34 +0000 (12:21 +0000)]
res_pjsip_refer: Fix issue where native bridge may not occur upon completion of a transfer.

There are two methods within res_pjsip_refer for keeping track of the state of a transfer.
The first is a framehook which looks at frames passing by to determine the state. The second
subscribes to know when the channel joins a bridge. In the case when the channel joins the
bridge the framehook is *NOT* removed and this prevents the native RTP bridging technology
from getting used.

This change gets the channel and if it still exists remove the framehook.

Review: https://reviewboard.asterisk.org/r/4218/
........

Merged revisions 428760 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 428761 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428762 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoconfig: Create ast_variable_find_in_list()
George Joseph [Tue, 2 Dec 2014 00:38:56 +0000 (00:38 +0000)]
config: Create ast_variable_find_in_list()

Add
const char *ast_variable_find_in_list(const struct ast_variable *list,
   const char *variable);

ast_variable_find() requires a config category to search whereas
ast_variable_find_in_list() just needs the root list element which is
useful if you don't have a category.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4217/
........

Merged revisions 428733 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 428734 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428735 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agores_pjsip_endpoint_identifier_ip: Add 'show identify(ies)' cli commands
George Joseph [Tue, 2 Dec 2014 00:31:49 +0000 (00:31 +0000)]
res_pjsip_endpoint_identifier_ip: Add 'show identify(ies)' cli commands

While troubleshooting other things I realized there were no pjsip cli
commands for identify.  This patch adds them.  It also also fixes a
reference leak when a 'show endpoint' displayed identifies and properly
sets the return code if load_module can't allocate a cli formatter structure.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4212/
........

Merged revisions 428725 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 428731 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428732 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agortp_engine: Add support for transporting signed linear at 12kHz, 24kHz, 32kHz, 44kHz...
Joshua Colp [Mon, 1 Dec 2014 18:51:52 +0000 (18:51 +0000)]
rtp_engine: Add support for transporting signed linear at 12kHz, 24kHz, 32kHz, 44kHz, 48kHz, 96kHz, and 192kHz over RTP.

This change adds mappings in the RTP engine layer for the remaining signed linear formats.

ASTERISK-24274 #close
Reported by: Frankie Chin

Review: https://reviewboard.asterisk.org/r/4093/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428708 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agomain/stasis: Allow subscriptions to use a threadpool for message delivery
Matthew Jordan [Mon, 1 Dec 2014 17:59:21 +0000 (17:59 +0000)]
main/stasis: Allow subscriptions to use a threadpool for message delivery

Prior to this patch, all Stasis subscriptions would receive a dedicated
thread for servicing published messages. In contrast, prior to r400178
(see review https://reviewboard.asterisk.org/r/2881/), the subscriptions
shared a thread pool. It was discovered during some initial work on Stasis
that, for a low subscription count with high message throughput, the
threadpool was not as performant as simply having a dedicated thread per
subscriber.

For situations where a subscriber receives a substantial number of messages
and is always present, the model of having a dedicated thread per subscriber
makes sense. While we still have plenty of subscriptions that would follow
this model, e.g., AMI, CDRs, CEL, etc., there are plenty that also fall into
the following two categories:
* Large number of subscriptions, specifically those tied to endpoints/peers.
* Low number of messages. Some subscriptions exist specifically to coordinate
  a single message - the subscription is created, a message is published, the
  delivery is synchronized, and the subscription is destroyed.
In both of the latter two cases, creating a dedicated thread is wasteful (and
in the case of a large number of peers/endpoints, harmful). In those cases,
having shared delivery threads is far more performant.

This patch adds the ability of a subscriber to Stasis to choose whether or not
their messages are dispatched on a dedicated thread or on a threadpool. The
threadpool is configurable through stasis.conf.

Review: https://reviewboard.asterisk.org/r/4193

ASTERISK-24533 #close
Reported by: xrobau
Tested by: xrobau
........

Merged revisions 428681 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 428687 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428688 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoapp_record: Fix bug where using the 'k' option and hanging up would trim 1/4 of a...
Joshua Colp [Mon, 1 Dec 2014 13:41:43 +0000 (13:41 +0000)]
app_record: Fix bug where using the 'k' option and hanging up would trim 1/4 of a second of the recording.

The Record dialplan function trims 1/4 of a second from the end of recordings in case
they are terminated because of DTMF. When hanging up, however, you don't want this to happen.
This change makes it so on hangup this does not occur.

ASTERISK-24530 #close
Reported by: Ben Smithurst
patches:
 app_record_v2.diff submitted by Ben Smithurst (license 6529)

Review: https://reviewboard.asterisk.org/r/4201/
........

Merged revisions 428653 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 428654 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 428655 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428656 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agochannel: Extend size of buffer for codecs in "core show channeltype" CLI command.
Joshua Colp [Mon, 1 Dec 2014 13:08:23 +0000 (13:08 +0000)]
channel: Extend size of buffer for codecs in "core show channeltype" CLI command.

The static buffer for codecs when invoking the "core show channeltype" CLI command
did not have enough room for all codecs. This has been extended so it does.

ASTERISK-24542 #close
Reported by: snuffy
patches:
 channeltype-tech.diff submitted by snuffy (license 5024)

Review: https://reviewboard.asterisk.org/r/4204/
........

Merged revisions 428632 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428633 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agotest_channel_feature_hooks.c: Fix unit test for DTMF hooks.
Richard Mudgett [Mon, 24 Nov 2014 20:39:01 +0000 (20:39 +0000)]
test_channel_feature_hooks.c: Fix unit test for DTMF hooks.

Fix the failing /channels/features/test_features_channel_dtmf unit test.

DTMF emulation does not work without a stream of packets to prod the
emulation code.

Review: https://reviewboard.asterisk.org/r/4199/
........

Merged revisions 428604 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428605 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoDTMF hooks: Leaving channels need to push any collected digits into the bridge.
Richard Mudgett [Mon, 24 Nov 2014 20:32:29 +0000 (20:32 +0000)]
DTMF hooks: Leaving channels need to push any collected digits into the bridge.

Any partially collected DTMF digits for a DTMF hook need to be pushed into
the bridge when a channel leaves the bridging system as if there were a
timeout.

Review: https://reviewboard.asterisk.org/r/4199/
........

Merged revisions 428601 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 428602 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428603 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agomanager: Fix could not extend string messages.
Richard Mudgett [Fri, 21 Nov 2014 19:16:55 +0000 (19:16 +0000)]
manager: Fix could not extend string messages.

When shutting down Asterisk that has an active AMI connection, you get
several "failed to extend from %d to %d" messages because use of the
EVENT_FLAG_SHUTDOWN attempts to add all AMI permission strings to the
event.

* Created MAX_AUTH_PERM_STRING to use when creating stack based struct
ast_str variables used with the authority_to_str() and
user_authority_to_str() functions instead of a variety of magic numbers
that could be too small.

* Added a special check for EVENT_FLAG_SHUTDOWN to authority_to_str() so
it will not attempt to add all permission level strings.

Review: https://reviewboard.asterisk.org/r/4200/
........

Merged revisions 428570 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 428571 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 428572 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428573 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agosorcery: Make is_object_field_registered handle field names that are regexes.
George Joseph [Fri, 21 Nov 2014 17:49:39 +0000 (17:49 +0000)]
sorcery: Make is_object_field_registered handle field names that are regexes.

As a result of https://reviewboard.asterisk.org/r/3305, res_sorcery_realtime
was tossing database fields that didn't have an exact match to a sorcery
registered field.  This broke the ability to use regexes as field names which
manifested itself as a failure of res_pjsip_phoneprov_provider which uses
this capability.  It also broke handling of fields that start with '@' in
realtime but I don't think anyone noticed.

This patch does the following...
* Modifies ast_sorcery_fields_register to pre-compile the name regex.
* Modifies ast_sorcery_is_object_field_registered to test the regex if it
  exists instead of doing an exact strcmp.
* Modifies res_pjsip_phoneprov_provider with a few tweaks to get it to work
  with realtime.

Tested-by: George Joseph

Review: https://reviewboard.asterisk.org/r/4185/
........

Merged revisions 428543 from http://svn.asterisk.org/svn/asterisk/branches/12
........

Merged revisions 428544 from http://svn.asterisk.org/svn/asterisk/branches/13

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428545 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agosip.conf.sample - note that media_address does not change listen address, just the SDP
Olle Johansson [Fri, 21 Nov 2014 13:59:45 +0000 (13:59 +0000)]
sip.conf.sample - note that media_address does not change listen address, just the SDP

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@428526 65c4cc65-6c06-0410-ace0-fbb531ad65f3