asterisk/asterisk.git
6 years agoRemove some unnecessary calls to ast_bridged_channel() in chan_unistim.c
Richard Mudgett [Wed, 1 May 2013 20:01:43 +0000 (20:01 +0000)]
Remove some unnecessary calls to ast_bridged_channel() in chan_unistim.c

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@387185 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoRemove some unnecessary calls to ast_bridged_channel() in chan_mgcp.c
Richard Mudgett [Wed, 1 May 2013 20:01:27 +0000 (20:01 +0000)]
Remove some unnecessary calls to ast_bridged_channel() in chan_mgcp.c

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@387184 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoRemove some unnecessary calls to ast_bridged_channel() in chan_skinny.c
Richard Mudgett [Wed, 1 May 2013 20:01:10 +0000 (20:01 +0000)]
Remove some unnecessary calls to ast_bridged_channel() in chan_skinny.c

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@387183 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoRemove some unnecessary calls to ast_bridged_channel() in chan_iax2.c
Richard Mudgett [Wed, 1 May 2013 20:00:53 +0000 (20:00 +0000)]
Remove some unnecessary calls to ast_bridged_channel() in chan_iax2.c

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@387182 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoRemove some unnecessary calls to ast_bridged_channel() in chan_dahdi.c/sig_analog.c
Richard Mudgett [Wed, 1 May 2013 20:00:31 +0000 (20:00 +0000)]
Remove some unnecessary calls to ast_bridged_channel() in chan_dahdi.c/sig_analog.c

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@387181 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoPrevent crash in 'sip show peers' when the number of peers on a system is large
Matthew Jordan [Wed, 1 May 2013 18:38:40 +0000 (18:38 +0000)]
Prevent crash in 'sip show peers' when the number of peers on a system is large

When you have lots of SIP peers (according to the issue reporter, around 3500),
the 'sip show peers' CLI command or AMI action can crash due to a poorly placed
string duplication that occurs on the stack. This patch refactors the command
to not allocate the string on the stack, and handles the formatting of a single
peer in a separate function call.

(closes issue ASTERISK-21466)
Reported by: Guillaume Knispel
patches:
  fix_sip_show_peers_stack_overflow_asterisk_11.3.0-v2.patch uploaded by gknispel (License 6492)
........

Merged revisions 387134 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@387135 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoMove some annoying chan_dahdi debug messages to level 5.
Richard Mudgett [Wed, 1 May 2013 17:15:26 +0000 (17:15 +0000)]
Move some annoying chan_dahdi debug messages to level 5.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@387108 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFix CDR not being created during an externally initiated blind transfer
Matthew Jordan [Tue, 30 Apr 2013 22:50:40 +0000 (22:50 +0000)]
Fix CDR not being created during an externally initiated blind transfer

Way back when in the dark days of Asterisk 1.8.9, blind transferring a call
in a context that included the 'h' extension would inadvertently execute the
hangup code logic on the transferred channel. This was a "bad thing". The fix
was to properly check for the softhangup flags on the channel and only execute
the 'h' extension logic (and, in later versions, hangup handler logic) if the
channel was well and truly dead (Jim).

Unfortunately, CDRs are fickle. Setting the softhangup flag when we detected
that the channel was leaving the bridge (but not to die) caused some crucial
snippet of CDR code, lying in ambush in the middle of the bridging code, to
not get executed. This had the effect of blowing away one of the CDRs that is
typically created during a blind transfer.

While we live and die by the adage "don't touch CDRs in release branches", this
was our bad. The attached patch restores the CDR behavior, and still manages to
not run the 'h' extension during a blind transfer (at least not when it's
supposed to).

Thanks to Steve Davies for diagnosing this and providing a fix.

Review: https://reviewboard.asterisk.org/r/2476

(closes issue ASTERISK-21394)
Reported by: Ishfaq Malik
Tested by: Ishfaq Malik, mjordan
patches:
  fix_missing_blindXfer_cdr2 uploaded by one47 (License 5012)
........

Merged revisions 387036 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 387038 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@387039 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoStasis Core: Refactor ACL Change events to go out over the stasis core msg bus
Jonathan Rose [Tue, 30 Apr 2013 22:37:24 +0000 (22:37 +0000)]
Stasis Core: Refactor ACL Change events to go out over the stasis core msg bus

(issue ASTERISK-21103)
Reported by: Matt Jordan
Review: https://reviewboard.asterisk.org/r/2481/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@387037 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoAdd forgotten event types to event_names array
Jonathan Rose [Tue, 30 Apr 2013 22:20:55 +0000 (22:20 +0000)]
Add forgotten event types to event_names array
........

Merged revisions 387030 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@387035 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFix a log message.
Jason Parker [Tue, 30 Apr 2013 18:12:36 +0000 (18:12 +0000)]
Fix a log message.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386990 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoUse the proper lower bound when doing saturation arithmetic.
Sean Bright [Tue, 30 Apr 2013 13:48:12 +0000 (13:48 +0000)]
Use the proper lower bound when doing saturation arithmetic.

16 bit signed integers have a range of [-32768, 32768).  The existing code
was using the interval (-32768, 32768) instead.  This patch fixes that.

Review: https://reviewboard.asterisk.org/r/2479/
........

Merged revisions 386929 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 386930 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386931 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoJust a couple of Stasis-HTTP nitpick fixes.
David M. Lee [Tue, 30 Apr 2013 13:37:09 +0000 (13:37 +0000)]
Just a couple of Stasis-HTTP nitpick fixes.

* Fixed crash when res_stasis_http is unloaded before the
  implementation modules.
* Cleaned up test initialization for test_stasis_http.so.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386928 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoModifying sounds/Makefile to pull down 1.4.24 core sounds
Rusty Newton [Mon, 29 Apr 2013 23:36:42 +0000 (23:36 +0000)]
Modifying sounds/Makefile to pull down 1.4.24 core sounds

1.4.24 core sounds includes a full set of Italian prompts for core sounds and a fix for the missing voicemail prompts in the Russian language.

(closes issue ASTERISK-19431)
(closes issue ASTERISK-19721)
........

Merged revisions 386877 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 386878 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386879 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoPlay periodic prompts for first call in a call queue
Olle Johansson [Mon, 29 Apr 2013 13:38:59 +0000 (13:38 +0000)]
Play periodic prompts for first call in a call queue

Review: https://reviewboard.asterisk.org/r/2263/
........

Merged revisions 386792 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 386794 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386841 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoChange pointer to existing wiki page instead of non-existing page
Olle Johansson [Mon, 29 Apr 2013 08:40:16 +0000 (08:40 +0000)]
Change pointer to existing wiki page instead of non-existing page

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386793 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFix spelling error in python doc
Kinsey Moore [Sun, 28 Apr 2013 03:32:35 +0000 (03:32 +0000)]
Fix spelling error in python doc

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386774 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoTweak res_sip priority so it gets loaded first before all other SIP stuff.
Joshua Colp [Sat, 27 Apr 2013 19:03:39 +0000 (19:03 +0000)]
Tweak res_sip priority so it gets loaded first before all other SIP stuff.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386760 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoUpdate res_config_sqlite to use the ast_variable lists.
Joshua Colp [Sat, 27 Apr 2013 16:17:01 +0000 (16:17 +0000)]
Update res_config_sqlite to use the ast_variable lists.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386746 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoAdd support for a realtime sorcery module.
Joshua Colp [Sat, 27 Apr 2013 12:01:29 +0000 (12:01 +0000)]
Add support for a realtime sorcery module.

This change does the following:

1. Adds the sorcery realtime module
2. Adds unit tests for the sorcery realtime module
3. Changes the realtime core to use an ast_variable list instead of variadic arguments
4. Changes all realtime drivers to accept an ast_variable list

Review: https://reviewboard.asterisk.org/r/2424/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386731 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoAdd missing module dependencies to various res_sip* modules
Matthew Jordan [Fri, 26 Apr 2013 21:52:06 +0000 (21:52 +0000)]
Add missing module dependencies to various res_sip* modules

This patch updates the various res_sip modules with their proper menuselect
options and proper dependencies, such that Asterisk still has a snowball's
chance in hell of compiling without pjproject.

Much thanks to snuffy(-home|-work) for making everyone's life
easier with this patch.

Review: https://reviewboard.asterisk.org/r/2472/

(closes issue ASTERISK-21669)
Reported by: snuffy
patches:
  xml-depends.diff uploaded by snuffy (license 5024)

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386686 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoClean up memory leak in config file on off nominal paths when glob is allowed
Matthew Jordan [Fri, 26 Apr 2013 21:34:16 +0000 (21:34 +0000)]
Clean up memory leak in config file on off nominal paths when glob is allowed

If a system allows for its usage, Asterisk will use glob to help parse
Asterisk .conf files. The config file loading routine was leaking the memory
allocated by the glob() routine when the config file was in an unmodified
or invalid state.

This patch properly calls globfree in those off nominal paths.

(closes issue ASTERISK-21412)
Reported by: Corey Farrell
patches:
  config_glob_leak.patch uploaded by Corey Farrell (license 5909)
........

Merged revisions 386672 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 386677 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386685 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoBy popular demand, putting the about-to-load-module printf back.
David M. Lee [Fri, 26 Apr 2013 21:31:39 +0000 (21:31 +0000)]
By popular demand, putting the about-to-load-module printf back.

But now it only prints during the initial startup, and prints at verbose 1
level.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386684 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoClean up resources in features on exit
Matthew Jordan [Fri, 26 Apr 2013 21:27:24 +0000 (21:27 +0000)]
Clean up resources in features on exit

This patch cleans up two things features:
* It properly unregisters the CLI commands that features registered
* It cancels and performs a pthread_join on the created parking thread. This
  not only properly joins a non-detached thread, but also prevents disposing
  of the parking lots prior to the parking thread completely exiting.

(closes issue ASTERISK-21407)
Reported by: Corey Farrell
patches:
  features_shutdown-r2.patch uploaded by Corey Farrell (License 5909)
........

Merged revisions 386641 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 386642 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386676 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoRemoving stray printf from r386540
David M. Lee [Fri, 26 Apr 2013 21:00:45 +0000 (21:00 +0000)]
Removing stray printf from r386540

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386640 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoAdd an \extref doxygen pointer for libuuid.
Mark Michelson [Fri, 26 Apr 2013 20:32:11 +0000 (20:32 +0000)]
Add an \extref doxygen pointer for libuuid.

Thanks to Olle Johansson for suggesting this.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386638 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoExample of how to use the Stasis message bus
David M. Lee [Fri, 26 Apr 2013 20:05:15 +0000 (20:05 +0000)]
Example of how to use the Stasis message bus

In order to get people familiar with the Stasis message bus, it would
be useful to have something of a tutorial. Since I'm not clever enough
to think of some cool integration we could do with Twitter, I settled
for something that might actually be useful.

This patch adds a res_statsd.so module, which implements a basic
statsd[1] client. Statsd is a very simple statistics gathering server,
which can publish its results to a backend graphing engine, like
Graphite[2]. There are several different Statsd server
implementations[3], so you can pick what works best for your
environment.

The actual example of how to use the Stasis message bus is in
res_chan_stats.so. This module demonstrates how to use subscriptions
and the message router by monitoring messages and posting channels
stats to the statsd server.

A wiki page walking through res_chan_stats.so is forthcoming.

 [1]: https://github.com/etsy/statsd/
 [2]: http://graphite.readthedocs.org/en/latest/
 [3]: http://joemiller.me/2011/09/21/list-of-statsd-server-implementations/

Review: https://reviewboard.asterisk.org/r/2460/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386624 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoIgnore *.[oi] files in res/res_sip
David M. Lee [Fri, 26 Apr 2013 20:03:32 +0000 (20:03 +0000)]
Ignore *.[oi] files in res/res_sip

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386623 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoDon't bind to anything in the sample configuration so we don't clash with chan_sip...
Joshua Colp [Thu, 25 Apr 2013 21:32:48 +0000 (21:32 +0000)]
Don't bind to anything in the sample configuration so we don't clash with chan_sip on a "make samples" right now.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386577 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoREmove automerge properties.
Mark Michelson [Thu, 25 Apr 2013 18:28:37 +0000 (18:28 +0000)]
REmove automerge properties.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386541 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoMerge the pimp_my_sip branch into trunk.
Mark Michelson [Thu, 25 Apr 2013 18:25:31 +0000 (18:25 +0000)]
Merge the pimp_my_sip branch into trunk.

The pimp_my_sip branch is being merged at this point because
it offers basic functionality, and from an API standpoint, things
are complete.

SIP work is *not* feature-complete; however, with the completion
of the SUBSCRIBE/NOTIFY API, all APIs (except a PUBLISH API) have
been created, and thus it is possible for developers to attempt
to create new SIP work.

API documentation can be found in the doxygen in the code, but
usability documentation is still lacking.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386540 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFix Displaying Symmetric RTP Global Setting
Michael L. Young [Thu, 25 Apr 2013 03:04:21 +0000 (03:04 +0000)]
Fix Displaying Symmetric RTP Global Setting

* Use comedia_string() to display correctly the symmetric rtp setting when
  running "sip show settings"
........

Merged revisions 386486 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386487 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoChange Case On Forcerport For Consistency
Michael L. Young [Thu, 25 Apr 2013 02:48:44 +0000 (02:48 +0000)]
Change Case On Forcerport For Consistency

* Change "ForcerPort" to "Forcerport" to match everywhere else it is displayed
........

Merged revisions 386483 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 386484 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386485 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoDocument JSON models in resource_*.h
David M. Lee [Wed, 24 Apr 2013 21:47:03 +0000 (21:47 +0000)]
Document JSON models in resource_*.h

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386462 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoOops. Mustache doesn't like dictionaries
David M. Lee [Wed, 24 Apr 2013 21:43:16 +0000 (21:43 +0000)]
Oops. Mustache doesn't like dictionaries

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386461 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoconfbridge: Make search the conference bridges container using OBJ_KEY.
Richard Mudgett [Tue, 23 Apr 2013 20:18:44 +0000 (20:18 +0000)]
confbridge: Make search the conference bridges container using OBJ_KEY.

* Make confbridge config parsing user profile, bridge profile, and menu
container hash/cmp functions correctly check the OBJ_POINTER, OBJ_KEY, and
OBJ_PARTIAL_KEY flags.

* Made confbridge load_module()/unload_module() free all resources on
failure conditions.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386375 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFix some bad whitespace
Kinsey Moore [Tue, 23 Apr 2013 18:57:00 +0000 (18:57 +0000)]
Fix some bad whitespace

This crept in with the RESTful HTTP interface merge.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386352 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFix crash when AMI redirect action redirects two channels out of a bridge.
Richard Mudgett [Mon, 22 Apr 2013 16:44:21 +0000 (16:44 +0000)]
Fix crash when AMI redirect action redirects two channels out of a bridge.

The two party bridging loops were changing the bridge peer pointers
without the channel locks held.  Thus when ast_channel_massquerade()
tested and used the pointer there is a small window of opportunity for the
pointers to become NULL even though the masquerade code has the channels
locked.

(closes issue ASTERISK-21356)
Reported by: William luke
Patches:
      jira_asterisk_21356_v11.patch (license #5621) patch uploaded by rmudgett
Tested by: William luke
........

Merged revisions 386256 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 386286 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386289 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoDoxygen - Markup Guidelines
Andrew Latham [Mon, 22 Apr 2013 16:22:00 +0000 (16:22 +0000)]
Doxygen - Markup Guidelines

Expand on a commit by OEJ to use the Coding-Guidelines

(issue ASTERISK-20259)

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386266 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoThis patch adds a RESTful HTTP interface to Asterisk.
David M. Lee [Mon, 22 Apr 2013 14:58:53 +0000 (14:58 +0000)]
This patch adds a RESTful HTTP interface to Asterisk.

The API itself is documented using Swagger, a lightweight mechanism for
documenting RESTful API's using JSON. This allows us to use swagger-ui
to provide executable documentation for the API, generate client
bindings in different languages, and generate a lot of the boilerplate
code for implementing the RESTful bindings. The API docs live in the
rest-api/ directory.

The RESTful bindings are generated from the Swagger API docs using a set
of Mustache templates.  The code generator is written in Python, and
uses Pystache. Pystache has no dependencies, and be installed easily
using pip. Code generation code lives in rest-api-templates/.

The generated code reduces a lot of boilerplate when it comes to
handling HTTP requests. It also helps us have greater consistency in the
REST API.

(closes issue ASTERISK-20891)
Review: https://reviewboard.asterisk.org/r/2376/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386232 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFix mistake in Doxygen.
Olle Johansson [Mon, 22 Apr 2013 12:45:26 +0000 (12:45 +0000)]
Fix mistake in Doxygen.

Doxygen is only *ONE* comment that applies to the NEXT piece of code.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386211 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agosla: remove redundant locking.
Russell Bryant [Mon, 22 Apr 2013 01:05:43 +0000 (01:05 +0000)]
sla: remove redundant locking.

sla.lock was already locked in the only place that sla_check_reload() was called.
Remove the redundant locking of sla.lock done in this function.  Less recursive
locking is A Good Thing.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386190 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoPrevent res_timing_pthread from blocking callers
Matthew Jordan [Fri, 19 Apr 2013 22:27:08 +0000 (22:27 +0000)]
Prevent res_timing_pthread from blocking callers

There were several reports of deadlock when using
res_timing_pthread. Backtraces indicated that one thread was blocked
waiting for the write to the pipe to complete and this thread held
the container lock for the timers.  Therefore any thread that wanted
to create a new timer or read an existing timer would block waiting
for either the timer lock or the container lock and deadlock ensued.

This patch changes the way the pipe is used to eliminate this source
of deadlocks:

1) The pipe is placed in non-blocking mode so that it would never
block even if the following changes someone fail...

2) Instead of writing bytes into the pipe for each "tick" that's
fired the pipe now has two states--signaled and unsignaled. If
signaled, the pipe is hot and any pollers of the read side
filedescriptor will be woken up. If unsigned the pipe is idle. This
eliminates even the chance of filling up the pipe and reduces the
potential overhead of calling unnecessary writes.

3) Since we're tracking the signaled / unsignaled state, we can
eliminate the exta poll system call for every firing because we know
that there is data to be read.

(closes issue ASTERISK-21389)
Reported by: Matt Jordan
Tested by: Shaun Ruffell, Matt Jordan, Tony Lewis
patches:
  0001-res_timing_pthread-Reduce-probability-of-deadlocking.patch uploaded by sruffell (License 5417)

(closes issue ASTERISK-19754)
Reported by: Nikola Ciprich

(closes issue ASTERISK-20577)
Reported by: Kien Kennedy

(closes issue ASTERISK-17436)
Reported by: Henry Fernandes

(closes issue ASTERISK-17467)
Reported by: isrl

(closes issue ASTERISK-17458)
Reported by: isrl

Review: https://reviewboard.asterisk.org/r/2441/
........

Merged revisions 386109 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 386159 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386160 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agocli.c: Properly initialize debug_modules and verbose_modules.
David M. Lee [Fri, 19 Apr 2013 05:20:02 +0000 (05:20 +0000)]
cli.c: Properly initialize debug_modules and verbose_modules.

This avoids some lock errors on the core set {debug,verbose} commands.
........

Merged revisions 386049 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 386051 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386054 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoAllow WebSocket connections on more URL's
David M. Lee [Thu, 18 Apr 2013 17:30:28 +0000 (17:30 +0000)]
Allow WebSocket connections on more URL's

This patch adds the concept of ast_websocket_server to
res_http_websocket, allowing WebSocket connections on URL's more more
than /ws.

The existing funcitons for managing the WebSocket subprotocols on /ws
still work, so this patch should be completely backward compatible.

(closes issue ASTERISK-21279)
Review: https://reviewboard.asterisk.org/r/2453/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386020 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFix lock errors on startup.
David M. Lee [Thu, 18 Apr 2013 17:26:29 +0000 (17:26 +0000)]
Fix lock errors on startup.

In messages.c, there are several places in the code where we create a
tmp_tech_holder and pass that into an ao2_find call. Unfortunately, we
weren't initializing the rwlock on the tmp_tech_holder, which the hash
function was locking. It's apparently harmless, but still not the best
code.

This patch extracts all that copy/pasted code into two functions,
msg_find_by_tech and msg_find_by_tech_name, which properly initialize
and destroy the rwlock on the tmp_tech_holder.

Review: https://reviewboard.asterisk.org/r/2454/
........

Merged revisions 386006 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@386019 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agores_xmpp and res_jabber need to search 'cachable' in the attrib section of the receiv...
Alec L Davis [Tue, 16 Apr 2013 23:44:18 +0000 (23:44 +0000)]
res_xmpp and res_jabber need to search 'cachable' in the attrib section of the received IE, not data.

(issue ASTERISK-20175)
(closes issue ASTERISK-21429)
(closes issue ASTERISK-21069)
(closes issue ASTERISK-21164)

Reported by: alecdavis
Tested by: alecdavis
alecdavis (license 585)

Review https://reviewboard.asterisk.org/r/2452/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385939 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoAllow res_corosync to build
Kinsey Moore [Tue, 16 Apr 2013 17:50:14 +0000 (17:50 +0000)]
Allow res_corosync to build

ast_enable_distributed_devstate is no longer applicable to how the
distributed device state system works and is no longer necessary.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385886 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoMove presence state distribution to Stasis-core
Kinsey Moore [Tue, 16 Apr 2013 15:48:16 +0000 (15:48 +0000)]
Move presence state distribution to Stasis-core

Convert presence state events to Stasis-core messages and remove
redundant serializers where possible.

Review: https://reviewboard.asterisk.org/r/2410/
(closes issue ASTERISK-21102)
Patch-by: Kinsey Moore <kmoore@digium.com>

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385862 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoMove device state distribution to Stasis-core
Kinsey Moore [Tue, 16 Apr 2013 15:33:59 +0000 (15:33 +0000)]
Move device state distribution to Stasis-core

In the move from Asterisk's event system to Stasis, this makes
distributed device state aggregation always-on, removes unnecessary
task processors where possible, and collapses aggregate and
non-aggregate states into a single cache for ease of retrieval. This
also removes an intermediary step in device state aggregation.

Review: https://reviewboard.asterisk.org/r/2389/
(closes issue ASTERISK-21101)
Patch-by: Kinsey Moore <kmoore@digium.com>

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385860 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFixed a typo
David M. Lee [Tue, 16 Apr 2013 14:09:25 +0000 (14:09 +0000)]
Fixed a typo

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385835 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoDon't unnecessarily rebuild things on every run of 'make'.
Jason Parker [Mon, 15 Apr 2013 17:26:49 +0000 (17:26 +0000)]
Don't unnecessarily rebuild things on every run of 'make'.

Review: https://reviewboard.asterisk.org/r/2449/
........

Merged revisions 385745 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 385768 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385782 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoAvoid unused variable warning when not in devmode
David M. Lee [Mon, 15 Apr 2013 16:47:25 +0000 (16:47 +0000)]
Avoid unused variable warning when not in devmode

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385743 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoMoved core logic from app_stasis to res_stasis
David M. Lee [Mon, 15 Apr 2013 16:43:47 +0000 (16:43 +0000)]
Moved core logic from app_stasis to res_stasis

After some discussion on asterisk-dev, it was decided that the bulk of
the logic in app_stasis actually belongs in a resource module instead
of the application module.

This patch does that, leaves the app specific stuff in app_stasis, and
fixes up everything else to be consistent with that change.

 * Renamed test_app_stasis to test_res_stasis
 * Renamed app_stasis.h to stasis_app.h
   * This is still stasis application support, even though it's no
     longer in an app_ module. The name should never have been tied to
     the type of module, anyways.
 * Now that json isn't a resource module anymore, moved the
   ast_channel_snapshot_to_json function to main/stasis_channels.c,
   where it makes more sense.

Review: https://reviewboard.asterisk.org/r/2430/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385742 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoDTMF events are now published on a channel's stasis_topic. AMI was
David M. Lee [Mon, 15 Apr 2013 16:22:03 +0000 (16:22 +0000)]
DTMF events are now published on a channel's stasis_topic. AMI was
refactored to use these events rather than producing the events directly
in channel.c. Finally, the code was added to app_stasis to produce
DTMF events on the WebSocket.

The AMI events are completely backward compatible, including sending
events on transmitted DTMF, and sending DTMF start events.

The Stasis-HTTP events are somewhat simplified. Since DTMF start and
DTMF send events are generally less useful, Stasis-HTTP will only send
events on received DTMF end.

(closes issue ASTERISK-21282)
(closes issue ASTERISK-21359)
Review: https://reviewboard.asterisk.org/r/2439

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385734 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFix the svn:keywords property on several files.
David M. Lee [Mon, 15 Apr 2013 16:10:10 +0000 (16:10 +0000)]
Fix the svn:keywords property on several files.

Normally I think keyword expansion is silly, but the one time it would have
been good, it didn't work because the property had quotes in it. This patch
fixes obviously busted svn:keywords properties.
........

Merged revisions 385683 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 385689 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385718 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoCalculate the timestamp for outbound RTP if we don't have timing information
Matthew Jordan [Sun, 14 Apr 2013 03:01:33 +0000 (03:01 +0000)]
Calculate the timestamp for outbound RTP if we don't have timing information

This patch calculates the timestamp for outbound RTP when we don't have timing
information. This uses the same approach in res_rtp_asterisk. Thanks to both
Pietro and Tzafrir for providing patches.

(closes issue ASTERISK-19883)
Reported by: Giacomo Trovato
Tested by: Pietro Bertera, Tzafrir Cohen
patches:
  rtp-timestamp-1.8.patch uploaded by tzafrir (License 5035)
  rtp-timestamp.patch uploaded by pbertera (License 5943)
........

Merged revisions 385636 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 385637 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385638 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoDon't attempt to create a voice frame on a read error
Matthew Jordan [Sun, 14 Apr 2013 02:35:04 +0000 (02:35 +0000)]
Don't attempt to create a voice frame on a read error

Prior to this patch, a read error in snd_pcm_readi would still be treated as a
nominal result when constructing a voice frame from the expected data. Since
the value returned is negative, as opposed to the number of samples read,
this could result in a crash. With this patch, we now return a null frame
when a read error is detected.

Note that the patch on ASTERISK-21329 was modified slightly for this commit,
in that we bail immediately on detecting the read error, rather than bypassing
the construction of the voice frame.

(closes issue ASTERISK-21329)
Reported by: Keiichiro Kawasaki
patches:
  chan_alsa.diff uploaded by kawasaki (License 6489)
........

Merged revisions 385633 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 385634 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385635 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFix Manager Segfault When app_queue Is Unloaded
Michael L. Young [Fri, 12 Apr 2013 22:38:56 +0000 (22:38 +0000)]
Fix Manager Segfault When app_queue Is Unloaded

When app_queue is unloaded, some manager commands are not being unregistered
which result in a segfault.  This patch corrects this.

(closes issue ASTERISK-21397)
Reported by: Peter Katzmann, Corey Farrell
Tested by: Corey Farrell
Patches:
    asterisk-21397-missing-unreg-manager-cmd_1.8.diff
                                                 Michael L. Young (license 5026)
    asterisk-21397-missing-unreg-manager-cmd_11.diff
                                                 Michael L. Young (license 5026)

Review: https://reviewboard.asterisk.org/r/2444/
........

Merged revisions 385593 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 385594 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385595 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoAllow codec_resample to be unloaded
Kinsey Moore [Fri, 12 Apr 2013 22:26:17 +0000 (22:26 +0000)]
Allow codec_resample to be unloaded

Ensure that trans_size is correct to prevent uninitialized entries from
preventing reload.

(closes issue ASTERISK-21401)
Reported by: Corey Farrell
Tested by: Corey Farrell
Patches:
    codec_resample-unload.patch uploaded by Corey Farrell
........

Merged revisions 385582 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385585 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFix app_voicemail Segfault And A Few Memory Leaks
Michael L. Young [Fri, 12 Apr 2013 22:22:58 +0000 (22:22 +0000)]
Fix app_voicemail Segfault And A Few Memory Leaks

The original report was that app_voicemail would crash.  This was caused by
ast_config_load() returning CONFIG_STATUS_FILEINVALID but no checks being
performed for that return status.  After adding the initial patch to fix this
issue, Jaco Kroon (jkroon) added some fixes to memory leaks he had discovered.

During review, Walter Doekes (wdoekes) suggested adding a helper function in
order to determine if we had a valid configuration or not.

This patch does the following:

* Creates a helper function to check if the configuration is valid

* Adds calls to the new helper function where appropiate

* Fixes memory leaks where the code returned without running
  ast_config_destroy() on the configuration that was loaded

(closes issue ASTERISK-21302)
Reported by: Jaco Kroon
Tested by: Jaco Kroon, Michael L. Young
Patches:
    asterisk-11.3.0-app_voicemail-ast_config-fixes.patch
                                                       Jaco Kroon (license 5671)
    asterisk-21302-valid_cfg_and_mem_leaks_v3-1.8.diff
                                                 Michael L. Young (license 5026)

Review: https://reviewboard.asterisk.org/r/2443/
........

Merged revisions 385551 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 385557 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385573 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFix documentation.
Jason Parker [Fri, 12 Apr 2013 21:48:10 +0000 (21:48 +0000)]
Fix documentation.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385548 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoExpose channel snapshot manager blob generation
Kinsey Moore [Fri, 12 Apr 2013 21:11:02 +0000 (21:11 +0000)]
Expose channel snapshot manager blob generation

These functions are already used in one branch (jrose's parking branch)
and will soon be used in other branches as well.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385522 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFix One-Way Audio With auto_* NAT Settings When SIP Calls Initiated By PBX
Michael L. Young [Fri, 12 Apr 2013 15:06:09 +0000 (15:06 +0000)]
Fix One-Way Audio With auto_* NAT Settings When SIP Calls Initiated By PBX

When we reload Asterisk or chan_sip, the flags force_rport and comedia that are
turned on and off when using the auto_force_rport and auto_comedia nat settings
go back to the default setting off.  These flags are turned on when needed or
off when not needed at the time that a peer registers, re-registers or initiates
a call.  This would apply even when only the default global setting
"nat=auto_force_rport" is being used, which in this case would only affect the
force_rport flag.

Everything is good except for the following:  The nat setting is set to
auto_force_rport and auto_comedia.  We reload Asterisk and the peer's
registration has not expired.  We load in the settings for the peer which turns
force_rport and comedia back to off.  Since the peer has not re-registered or
placed a call yet, those flags remain off.  We then initiate a call to the peer
from the PBX.  The force_rport and comedia flags stay off.  If NAT is involved,
we end up with one-way audio since we never checked to see if the peer is behind
NAT or not.

This patch does the following:

* Moves the checking of whether a peer is behind NAT into its own function

* Create a function to set the peer's NAT flags if they are using the auto_* NAT
  settings

* Adds calls in sip_request_call() to these new functions in order to setup the
  dialog according to the peer's settings

(closes issue ASTERISK-21374)
Reported by: Michael L. Young
Tested by: Michael L. Young
Patches:
    asterisk-21374-auto-nat-outgoing-fix_v2.diff Michael L. Young (license 5026)

Review: https://reviewboard.asterisk.org/r/2421/
........

Merged revisions 385473 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385474 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoIAX2 defer_full_frames fail to get sent
Alec L Davis [Fri, 12 Apr 2013 08:52:44 +0000 (08:52 +0000)]
IAX2 defer_full_frames fail to get sent

Ensure iax2_process_thread is signalled when a deferred frame is queued to it.

(closes issue ASTERISK-18827)
Reported by: alecdavis
Tested by: alecdavis
alecdavis (license 585)

Review https://reviewboard.asterisk.org/r/2426/
........

Merged revisions 385429 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 385430 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385431 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoIAX2, prevent network thread starting before all helper threads are ready
Alec L Davis [Fri, 12 Apr 2013 08:18:20 +0000 (08:18 +0000)]
IAX2, prevent network thread starting before all helper threads are ready

On startup, it's possible for a frame to arrive before the processing threads were ready.

In iax2_process_thread() the first pass through falls into ast_cond_wait, should a frame arrive
before we are at ast_cond_wait, the signal will be ignored.
The result iax2_process_thread stays at ast_cond_wait forever, with deferred frames being queued.

Fix: When creating initial idle iax2_process_threads, wait for init_cond to be signalled
after each thread is started.

(issue ASTERISK-18827)
Reported by: alecdavis
Tested by: alecdavis
alecdavis (license 585)

Review https://reviewboard.asterisk.org/r/2427/
........

Merged revisions 385402 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 385403 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385406 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoBlocked revisions 385356
Jason Parker [Thu, 11 Apr 2013 20:00:46 +0000 (20:00 +0000)]
Blocked revisions 385356

........
Add dependency on libuuid, for res_rtp_asterisk

pjproject is what actually requires libuuid.

(closes issue ASTERISK-21125)
reported by Private Name

(Ed. note: Really?  Private Name?  I am rolling my eyes so hard right now.)

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385357 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFix 'pri intense debug span' alias.
Richard Mudgett [Thu, 11 Apr 2013 16:53:21 +0000 (16:53 +0000)]
Fix 'pri intense debug span' alias.
........

Merged revisions 385313 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385314 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoEliminated dial_features_destroy() since it is equivalent to ast_free_ptr()
Richard Mudgett [Wed, 10 Apr 2013 23:08:02 +0000 (23:08 +0000)]
Eliminated dial_features_destroy() since it is equivalent to ast_free_ptr()

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385278 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years ago* Fix unlocked accesses to feature_list. The feature_list is now also
Richard Mudgett [Wed, 10 Apr 2013 23:03:30 +0000 (23:03 +0000)]
* Fix unlocked accesses to feature_list.  The feature_list is now also
protected by the features_lock.

* Made all calls to ast_find_call_feature() have the features_lock held.

* Fixed set_config_flags() to actually use find_group() to look for
feature groups in DYNAMIC_FEATURES.  The code originally assumed all
feature groups were listed in DYNAMIC_FEATURES.

* Make everyone use ast_rdlock_call_features(),
ast_unlock_call_features(), and new ast_wrlock_call_features() instead of
directly calling the rwlock API on features_lock.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385277 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFixed manager channelvars support.
David M. Lee [Wed, 10 Apr 2013 15:34:47 +0000 (15:34 +0000)]
Fixed manager channelvars support.

For the events that have been ported to Stasis, this was broken in
r384910, when a couple of lines of code was lost in a merge.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385236 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoUse LDAP memory management functions instead of Asterisk's
Matthew Jordan [Wed, 10 Apr 2013 14:26:22 +0000 (14:26 +0000)]
Use LDAP memory management functions instead of Asterisk's

When MALLOC_DEBUG is enabled with res_config_ldap, issues (munmap_chunk:
invalid pointer errors) can occur as the memory is being allocated with
Asterisk's wrappers around malloc/calloc/free/strdup, as opposed to the
LDAP library's wrappers.

This patch uses the LDAP library's wrappers where appropriate, so that
compiling with MALLOC_DEBUG doesn't cause more problems than it solves.

Note that the patch listed below was modified slightly for this commit
to account for some additional memory allocation/deallocations.

(closes issue ASTERISK-17386)
Reported by: John Covert
Tested by: Andrew Latham
patches:
  issue18789-1.8-r316873.patch uploaded by seanbright (License 5060)
........

Merged revisions 385190 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 385199 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385202 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFix crash in chan_sip when a core initiated op occurs at the same time as a BYE
Matthew Jordan [Wed, 10 Apr 2013 14:07:27 +0000 (14:07 +0000)]
Fix crash in chan_sip when a core initiated op occurs at the same time as a BYE

When a BYE request is processed in chan_sip, the current SIP dialog is detached
from its associated Asterisk channel structure. The tech_pvt pointer in the
channel object is set to NULL, and the dialog persists for an RFC mandated
period of time to handle re-transmits.

While this process occurs, the channel is locked (which is good).
Unfortunately, operations that are initiated externally have no way of knowing
that the channel they've just obtained (which is still valid) and that they are
attempting to lock is about to have its tech_pvt pointer removed. By the time
they obtain the channel lock and call the channel technology callback, the
tech_pvt is NULL.

This patch adds a few checks to some channel callbacks that make sure the
tech_pvt isn't NULL before using it. Prime offenders were the DTMF digit
callbacks, which would crash if AMI initiated a DTMF on the channel at the
same time as a BYE was received from the UA. This patch also adds checks on
sip_transfer (as AMI can also cause a callback into this function), as well
as sip_indicate (as lots of things can queue an indication onto a channel).

Review: https://reviewboard.asterisk.org/r/2434/

(closes issue ASTERISK-20225)
Reported by: Jeff Hoppe
........

Merged revisions 385170 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 385173 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385174 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoRename struct feature_ds to struct feature_datastore.
Richard Mudgett [Tue, 9 Apr 2013 19:58:35 +0000 (19:58 +0000)]
Rename struct feature_ds to struct feature_datastore.

Because "struct feature_ds *feature_ds" is not a good thing.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385142 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoBackported app_stasis fix from stasis-http branch.
David M. Lee [Tue, 9 Apr 2013 18:22:08 +0000 (18:22 +0000)]
Backported app_stasis fix from stasis-http branch.

The hash and compare functions for the control container was reusing
the wrong ones, causing some problems. I fixed it, but in the wrong
branch. Oh well, it happens.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385116 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoAdd inheritance support to FEATURE()/FEATUREMAP().
Russell Bryant [Tue, 9 Apr 2013 06:16:42 +0000 (06:16 +0000)]
Add inheritance support to FEATURE()/FEATUREMAP().

The settings saved on the channel for FEATURE()/FEATUREMAP() were only
for that channel.  This patch adds the ability to have these settings
inherited to child channels if you set FEATURE(inherit)=yes.

Closes issue ASTERISK-21306.

Review: https://reviewboard.asterisk.org/r/2415/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385088 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoModified the list of keys for the driver backends for sake of sample clarity
Rusty Newton [Mon, 8 Apr 2013 23:38:08 +0000 (23:38 +0000)]
Modified the list of keys for the driver backends for sake of sample clarity

Added a line showing the mapping of "mysql" to res_config_mysql available in add-ons. We used "mysql" as an example driver key in the sample, but didn't show what module it mapped too. Also added a subtitle above the list of keys for driver backends.
........

Merged revisions 385047 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 385048 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@385049 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoClean up Makefile "warning" clutter when makeopts doesn't exist.
Walter Doekes [Mon, 8 Apr 2013 18:24:50 +0000 (18:24 +0000)]
Clean up Makefile "warning" clutter when makeopts doesn't exist.

Review: https://reviewboard.asterisk.org/r/2304

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384989 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoDon't attempt a websocket protocol removal if res_http_websocket isn't there
Matthew Jordan [Mon, 8 Apr 2013 15:38:34 +0000 (15:38 +0000)]
Don't attempt a websocket protocol removal if res_http_websocket isn't there

This patch sets the protocols container provided by res_http_websocket to NULL
when the module gets unloaded and adds the necessary checks when adding/
removing a websocket protocol. This prevents some FRACKing on an invalid
pointer to the disposed container if a module that uses res_http_websocket is
unloaded after it.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384942 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoAdd multi-channel Stasis messages; refactor Dial AMI events to Stasis
Matthew Jordan [Mon, 8 Apr 2013 14:26:37 +0000 (14:26 +0000)]
Add multi-channel Stasis messages; refactor Dial AMI events to Stasis

This patch does the following:
 * A new Stasis payload has been defined for multi-channel messages. This
   payload can store multiple ast_channel_snapshot objects along with a single
   JSON blob. The payload object itself is opaque; the snapshots are stored
   in a container keyed by roles. APIs have been provided to query for and
   retrieve the snapshots from the payload object.
 * The Dial AMI events have been refactored onto Stasis. This includes dial
   messages in app_dial, as well as the core dialing framework. The AMI events
   have been modified to send out a DialBegin/DialEnd events, as opposed to
   the subevent type that was previously used.
 * Stasis messages, types, and other objects related to channels have been
   placed in their own file, stasis_channels. Unit tests for some of these
   objects/messages have also been written.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384910 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoStasis application WebSocket support
David M. Lee [Mon, 8 Apr 2013 13:27:45 +0000 (13:27 +0000)]
Stasis application WebSocket support

This is the API that binds the Stasis dialplan application to external
Stasis applications. It also adds the beginnings of WebSocket
application support.

This module registers a dialplan function named Stasis, which is used
to put a channel into the named Stasis app. As a channel enters and
leaves the Stasis diaplan application, the Stasis app receives a
'stasis-start' and 'stasis-end' events.

Stasis apps register themselves using the stasis_app_register and
stasis_app_unregister functions. Messages are sent to an application
using stasis_app_send.

Finally, Stasis apps control channels through the use of the
stasis_app_control object, and the family of stasis_app_control_*
functions.

Other changes along for the ride are:
 * An ast_frame_dtor function that's RAII_VAR safe
 * Some common JSON encoders for name/number, timeval, and
   context/extension/priority

Review: https://reviewboard.asterisk.org/r/2361/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384879 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoAdd a res_sorcery_astdb module which uses the astdb to persist objects.
Joshua Colp [Sat, 6 Apr 2013 16:00:20 +0000 (16:00 +0000)]
Add a res_sorcery_astdb module which uses the astdb to persist objects.

Review: https://reviewboard.asterisk.org/r/2420/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384857 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFix For Not Overriding The Default Settings In chan_sip
Michael L. Young [Fri, 5 Apr 2013 20:41:27 +0000 (20:41 +0000)]
Fix For Not Overriding The Default Settings In chan_sip

The initial report was that the "nat" setting in the [general] section was not
having any effect in overriding the default setting.  Upon confirming that this
was happening and looking into what was causing this, it was discovered that
other default settings would not be overriden as well.

This patch works similar to what occurs in build_peer().  We create a temporary
ast_flags structure and using a mask, we override the default settings with
whatever is set in the [general] section.

In the bug report, the reporter who helped to test this patch noted that the
directmedia settings were being overriden properly as well as the nat settings.

This issue is also present in Asterisk 1.8 and a separate patch will be applied
to it.

(issue ASTERISK-21225)
Reported by: Alexandre Vezina
Tested by: Alexandre Vezina, Michael L. Young
Patches:
  asterisk-21225-handle-options-default-prob_v4.diff
Michael L. Young (license 5026)

Review: https://reviewboard.asterisk.org/r/2385/
........

Merged revisions 384827 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384828 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoSeparate some event struct definitions from instantiation.
Richard Mudgett [Thu, 4 Apr 2013 18:15:34 +0000 (18:15 +0000)]
Separate some event struct definitions from instantiation.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384760 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agochan_dahdi: Change inband_on_proceeding option default to no/disabled.
Richard Mudgett [Wed, 3 Apr 2013 20:27:11 +0000 (20:27 +0000)]
chan_dahdi: Change inband_on_proceeding option default to no/disabled.

(issue ASTERISK-21151)

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384711 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agochan_dahdi: Add inband_on_proceeding compatibility option.
Richard Mudgett [Wed, 3 Apr 2013 20:20:09 +0000 (20:20 +0000)]
chan_dahdi: Add inband_on_proceeding compatibility option.

The new inband_on_proceeding option causes Asterisk to assume inband audio
may be present when a PROCEEDING message is received.

Q.931 Section 5.1.2 says the network cannot assume that the CPE side has
attached to the B channel at this time without explicitly sending the
progress indicator ie informing the CPE side to attach to the B channel
for audio.  However, some non-compliant ISDN switches send a PROCEEDING
without the progress indicator ie indicating inband audio is available and
assume that the CPE device has connected the media path for listening to
ringback and other messages.

ASTERISK-17834 which causes this issue was dealing with a non-compliant
network switch.

(closes issue ASTERISK-21151)
Reported by: Gianluca Merlo
Tested by: rmudgett
........

Merged revisions 384685 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 384689 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384696 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoUpdate documentation for CHANNEL function
Matthew Jordan [Wed, 3 Apr 2013 17:17:33 +0000 (17:17 +0000)]
Update documentation for CHANNEL function

Document that you can read/write the 'accountcode' and 'amaflags' on a channel.
........

Merged revisions 384640 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 384641 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384642 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoastobj2: Fix rbtree duplicate handling.
Richard Mudgett [Wed, 3 Apr 2013 16:01:51 +0000 (16:01 +0000)]
astobj2: Fix rbtree duplicate handling.

OBJ_PARTIAL_KEY searching a rbtree did not find all possible matches if
the container did not accept duplicates.

Added matching node bias to indicate which matching node is being searched
for: first, last, any.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384616 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFixed spurious rebuilds of func_version.
David M. Lee [Tue, 2 Apr 2013 17:35:45 +0000 (17:35 +0000)]
Fixed spurious rebuilds of func_version.

func_version.so was being rebuilt every time, because build.h was
changing every build, because of the cleantest dependency that was
added in r384410 to fix parallel make bugs.

Now build.h will only be created if it does not exist, which was the
original behavior of the Makefile.
........

Merged revisions 384544 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 384545 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384546 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoPass the object type name to the configuration framework.
Joshua Colp [Tue, 2 Apr 2013 12:18:50 +0000 (12:18 +0000)]
Pass the object type name to the configuration framework.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384518 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoMake things work again
Matthew Jordan [Tue, 2 Apr 2013 11:40:05 +0000 (11:40 +0000)]
Make things work again

Sorry folks. ',' are still greater than '|'.

Thanks for playing along :-)

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384514 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoinstall_prereq: Build jansson from source, when necessary
David M. Lee [Mon, 1 Apr 2013 20:10:47 +0000 (20:10 +0000)]
install_prereq: Build jansson from source, when necessary

When r383579 was committed, it made Jansson a required dependency.

While libjansson-dev and jansson-devel are available on recent
distros, some older (but still supported) distros don't have
it. There's a pull request[1] to get it into repoforge, but that still
doesn't help everyone. (And helps no one until the pull request is
merged and packages are built).

This patch adds Jansson install from source to the install_unpackaged()
function. There are a few gotcha's, which makes this change not
completely trivial.

 * Since Jansson may be installed by a package, don't install from
   source if a package installation can be found
   * libresample may also be installed via package, so I added a
     similar check to that.
 * Since Jansson installs into /usr/local, this patch also adds
   /usr/local/lib to /etc/ld.so.conf.d so that the library can be
   found.
   * The alternative was to install into /usr, but then it gets
     complicated having to deal with EL's /usr/lib{32,64} shenanigans.

 [1]: https://github.com/repoforge/rpms/pull/250

Review: https://reviewboard.asterisk.org/r/2414/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384488 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoMake appropriate items parse using '|' instead of ','
Matthew Jordan [Mon, 1 Apr 2013 14:44:30 +0000 (14:44 +0000)]
Make appropriate items parse using '|' instead of ','

This patch fixes a bug introduced in r76703, wherein Asterisk could only parse
arguments in the so-called 'recommended' way, e.g., NoOp(foo,bar). The proper
syntax of NoOp,foo|bar is now parsed correctly.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384452 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoRemove silly use of strncmp.
Joshua Colp [Mon, 1 Apr 2013 14:10:46 +0000 (14:10 +0000)]
Remove silly use of strncmp.
........

Merged revisions 384414 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384416 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agostasis: Fixed message ordering issues when forwarding
David M. Lee [Mon, 1 Apr 2013 13:37:51 +0000 (13:37 +0000)]
stasis: Fixed message ordering issues when forwarding

This patch fixes an issue of message ordering that occurs when
multiple topics are forwarded to an aggregator topic (such as
ast_channel_topic_all()).

It is (very reasonably) expected that the rules governing message
dispatch order still apply, so long as the messages start from the
same thread, and are received by the same subscription. Because the
existing code had an additional layer of dispatching via the Stasis
thread pool for forwards, those promises couldn't be kept.

Forwarding subscriptions no longer have their own mailbox, and now
dispatch directly from the forwarding topic's stasis_publish()
call. This means that the topic's lock is held for the duration of not
only a message's dispatch, but the dispatch of all the forwards. This
shouldn't be a problem right now, but if an aggregator topic had many
subscribers, it could become a problem. But I figure we can write more
clever code when the time comes, if necessary.

Review: https://reviewboard.asterisk.org/r/2419/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384413 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoFix parallel make problems.
David M. Lee [Mon, 1 Apr 2013 13:34:51 +0000 (13:34 +0000)]
Fix parallel make problems.

Occasionally, make -j would fail due to missing includes, or other
unusual errors.

This was due to the 'cleantest' target, which was designed to force a
make clean when some change in the code would cause the typical
depedency checking to fail. Several targets in the main Makefile did
not depend upon cleantest, hence would run in parallel to it. By
adding the dependency, make -j runs happily now.

Review: https://reviewboard.asterisk.org/r/2418/
........

Merged revisions 384410 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 384411 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384412 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoProperly format an intmax_t value
Matthew Jordan [Sat, 30 Mar 2013 05:15:42 +0000 (05:15 +0000)]
Properly format an intmax_t value

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384390 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoConvert TestEvent AMI events over to Stasis Core
Matthew Jordan [Sat, 30 Mar 2013 05:06:54 +0000 (05:06 +0000)]
Convert TestEvent AMI events over to Stasis Core

This patch migrates the TestEvent AMI events to first be dispatched over the
Stasis-Core message bus. This helps to preserve the ordering of the events
with other events in the AMI system, such as the various channel related
events.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384389 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoapp_voicemail: Add blank argument to externnotify if no context argument
Jonathan Rose [Fri, 29 Mar 2013 16:37:23 +0000 (16:37 +0000)]
app_voicemail: Add blank argument to externnotify if no context argument

At least one call to run_externnotify provides a NULL context parameter and
because the snprintf statement doesn't account for a NULL context parameter,
it simply writes '(null)' to the arguments string instead. This patch makes
it write two quotes back to back for that argument instead in the event of
a NULL context.

(closes issue ASTERISK-18207)
Reported by: Barry L. Kline
Patches:
modified from patch-20130306 uploaded by Karsten Wemheuer (License 5930)
........

Merged revisions 384325 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 384326 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384327 65c4cc65-6c06-0410-ace0-fbb531ad65f3

6 years agoAdd uuid wrapper API call ast_uuid_generate_str().
Richard Mudgett [Thu, 28 Mar 2013 23:59:20 +0000 (23:59 +0000)]
Add uuid wrapper API call ast_uuid_generate_str().

* Updated test_uuid.c to test the new API call.

* Made system use the new API call to eliminate "10's of lines" where
used.

* Fixed untested ast_strdup() return in stasis_subscribe() by eliminating
the need for it.  struct stasis_subscription now contains the uniqueid[]
string.

* Fixed some issues in exchangecal_write_event():
  Create uid with enough space for a UUID string to avoid a realloc.
  Fix off by one error if the calendar event provided a UUID string.
  There is no need to check for NULL before calling ast_free().

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@384302 65c4cc65-6c06-0410-ace0-fbb531ad65f3