asterisk/asterisk.git
5 years agoastobj2: tweak ao2_replace to do nothing when it would be a NoOp
Corey Farrell [Fri, 11 Jul 2014 21:10:45 +0000 (21:10 +0000)]
astobj2: tweak ao2_replace to do nothing when it would be a NoOp

This change causes ao2_replace to do nothing when src == dst. This
avoids REF_DEBUG logging when we're not actually doing anything.

Review: https://reviewboard.asterisk.org/r/3743/
........

Merged revisions 418396 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@418397 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoconfig: inform config hook of change when writing file
Scott Griepentrog [Fri, 11 Jul 2014 16:42:49 +0000 (16:42 +0000)]
config: inform config hook of change when writing file

When updated configuration is written back to the conf
file - for example when a user changes their voicemail
pin, make sure that any config hook that wants to know
of changes is informed.

Review: https://reviewboard.asterisk.org/r/3708/
........

Merged revisions 418366 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 418369 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@418370 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoinclude/asterisk/xmpp.h: Convert indentation to tabs
Matthew Jordan [Thu, 10 Jul 2014 15:36:55 +0000 (15:36 +0000)]
include/asterisk/xmpp.h: Convert indentation to tabs

This is a whitespace only change.
........

Merged revisions 418323 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 418324 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@418325 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agochan_dahdi/sig_pri: Fix type mismatch in the idledial feature's channel creation.
Richard Mudgett [Thu, 10 Jul 2014 01:59:44 +0000 (01:59 +0000)]
chan_dahdi/sig_pri: Fix type mismatch in the idledial feature's channel creation.

Square pegs in round holes don't work very well.
........

Merged revisions 418261 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 418262 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 418263 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@418264 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoARI: Make mixing bridges propagate linkedids and accountcodes.
Richard Mudgett [Wed, 9 Jul 2014 16:34:51 +0000 (16:34 +0000)]
ARI: Make mixing bridges propagate linkedids and accountcodes.

* Create a Stasis bridge sub-class to propagate linkedids and
accountcodes.

* Fixed the basic bridge sub-class to update peeraccount codes when the
number of channels in the bridge drops back down to two parties.

* Refactored ast_bridge_channel_update_accountcodes() to handle channels
joining/leaving the bridge.

* Fixed the basic bridge sub-class to not call the base bridge class pull
method twice.

AFS-105 #close
ASTERISK-23852 #close
Reported by: Richard Mudgett

Review: https://reviewboard.asterisk.org/r/3720/
........

Merged revisions 418225 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@418226 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agomanager/ARI: Update version to 2.4.0/1.4.0; Update UPGRADE.txt
Matthew Jordan [Tue, 8 Jul 2014 14:48:30 +0000 (14:48 +0000)]
manager/ARI: Update version to 2.4.0/1.4.0; Update UPGRADE.txt
........

Merged revisions 418182 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@418183 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_rtp_asterisk: Fix undefined function when PJPROJECT is not installed
Matthew Jordan [Tue, 8 Jul 2014 14:38:42 +0000 (14:38 +0000)]
res_rtp_asterisk: Fix undefined function when PJPROJECT is not installed

The dtls_perform_handshake function was mistakenly placed under the guards for
USE_PJPROJECT. If PJPROJECT was not installed, the function would not be
defined, while other functions would attempt to still use it. This prevented
res_rtp_asterisk from being loaded.

ASTERISK-24001 #close
Reported by: Don Fanning
........

Merged revisions 418172 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@418174 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_dialog_info_body_generator: Add dialog-info+xml support for presence.
Joshua Colp [Mon, 7 Jul 2014 16:08:47 +0000 (16:08 +0000)]
res_pjsip_dialog_info_body_generator: Add dialog-info+xml support for presence.

This module implements dialog-info+xml for the purposes of presence. This means
that phones such as Grandstreams can now subscribe to receive presence information
for an extension.

ASTERISK-21443 #close
Reported by: Matt Jordan

Review: https://reviewboard.asterisk.org/r/3705/
........

Merged revisions 418116 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@418117 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoARI/res_stasis: Subscribe to both Local channel halves when originating to app
Matthew Jordan [Mon, 7 Jul 2014 02:15:00 +0000 (02:15 +0000)]
ARI/res_stasis: Subscribe to both Local channel halves when originating to app

This patch fixes two bugs:

1. When originating a channel into a Stasis application, we already create a
   subscription for the channel that is going into our Stasis app.
   Unfortunately, when you create a Local channel and pass it off to a Stasis
   app, you really aren't creating just one channel: you're creating two. This
   patch snags the second half of the Local channel pair (assuming it is a
   Local channel pair, but luckily core_local is kind about such assumptions)
   and subscribes to it as well.

2. Subscriptions are a bit sticky right now. If a subscription is made, the
   'interest' count gets bumped on the Stasis subscription - but unless
   something explicitly unsubscribes the channel, said subscription sticks
   around. This is not much of a problem is a user is creating the subscription
   - if they made it, they must want it. However, when we are creating
   implicit subscriptions, we need to make sure something clears them out.
   This patch takes a pessimistic approach: it watches the cache updates
   coming from Stasis and, if we notice that the cache just cleared out an
   object, we delete our subscription object. This keeps our ao2 container of
   Stasis forwards in an application from growing out of hand; it also is a
   bit more forgiving for end users who may not realize they were supposed to
   unsubscribe from that channel that just hung up.

Review: https://reviewboard.asterisk.org/r/3710/
#ASTERISK-23939 #close
........

Merged revisions 418089 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@418090 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoCEL: Fix incorrect/missing extra field information
Kinsey Moore [Mon, 7 Jul 2014 01:22:44 +0000 (01:22 +0000)]
CEL: Fix incorrect/missing extra field information

This corrects two issues with the extra field information in Asterisk
12+ in channel event logs.

It is possible to inject custom values into the dialstatus provided by
ast_channel_dial_type() Stasis messages that fall outside the
enumeration allowed for the DIALSTATUS channel variable. CEL now
filters for the allowed values and ignores other values.

The "hangupsource" extra field key is always blank if the far end
channel is a chan_pjsip channel. This is because the hangupsource is
never set for the pjsip channel driver. This change sets the
hangupsource whenever a hangup is queued for chan_pjsip channels.

This corrects an issue with the pjsip channel driver where the
hangupcause information was not being set properly.

Review: https://reviewboard.asterisk.org/r/3690/
........

Merged revisions 418071 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@418084 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoHTTP: Fix build for gcc 4.10
Kinsey Moore [Mon, 7 Jul 2014 01:10:55 +0000 (01:10 +0000)]
HTTP: Fix build for gcc 4.10
........

Merged revisions 418066 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@418067 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agomain/Makefile: fix compilation error of buildinfo occurring on 'make install'
Matthew Jordan [Fri, 4 Jul 2014 15:26:58 +0000 (15:26 +0000)]
main/Makefile: fix compilation error of buildinfo occurring on 'make install'

Egads. Another bad deletion of too much when attempting to remove h323 stuff.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@418050 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoconfigure: Remove last vestiges of h323; DO create menuselect-deps
Matthew Jordan [Fri, 4 Jul 2014 15:11:51 +0000 (15:11 +0000)]
configure: Remove last vestiges of h323; DO create menuselect-deps

The previous patch (r418034) fixed the 'glitch' that the channels/h323
Makefile no longer existed. Unfortunately, removing the entire line was a bit
of a blunder, as it meant that build_tools/menuselect-deps was never
generated. Hilarity ensued when actually trying to compile.

But hey! At least configure worked.

This patch fixes *that* glitch, and removes some more of the vestiges of h323.
(It had tendrils in the main Makefile? Crazy.)

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@418035 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoconfigure: Update script to pass if channels/h323/Makefile.in does not exist
Matthew Jordan [Fri, 4 Jul 2014 14:33:22 +0000 (14:33 +0000)]
configure: Update script to pass if channels/h323/Makefile.in does not exist

This simply removes that check from the configure script, as r418019 removed
chan_h323.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@418034 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoRemove many deprecated modules
Matthew Jordan [Fri, 4 Jul 2014 13:26:37 +0000 (13:26 +0000)]
Remove many deprecated modules

Billing records are fair,
To get paid is quite bright,
You should really use ODBC;
Good-bye cdr_sqlite.

Microsoft did once push H.323,
Hell, we all remember NetMeeting.
But try to compile chan_h323 now
And you will take quite a beating.

The XMPP and SIP war was fierce,
And in the distant fray
Was birthed res_jabber/chan_jingle;
But neither to stay.

For everyone did care and chase what Google professed.
"Free Internet Calling" was what devotees cried,
But Google did change the specs so often
That the developers were happy the day chan_gtalk died.

And then there was that odd application
Dedicated to the Polish tongue.
app_saycountpl was subsumed by Say;
One could say its bell was rung.

To read and parse a file from the dialplan
You could (I guess) use an application.
app_readfile did fill that purpose, but I think
A function is perhaps better in its creation.

Barging is rude, I'm not sure why we do it.
Inwardly, the caller will probably sigh.
But if you really must do it,
Don't use app_dahdibarge, use ChanSpy.

We all despise the sound of tinny robots
It makes our queues so cold.
To control such an abomination
It's better to not use Wait/SetMusicOnHold.

It's often nice to know properties of a channel
It makes our calls right
We have a nice function called CHANNEL
And so SIPCHANINFO is sent off into the night.

And now things get odd;
Apparently one could delimit with a colon
Properties from the SIPPEER function!
Commas are in; all others are done.

Finally, a word on pipes and commas.
We're sorry. We can't say it enough.
But those compatibility options in asterisk.conf;
To maintain them forever was just too tough.

This patch removes:

* cdr_sqlite
* chan_gtalk
* chan_jingle
* chan_h323
* res_jabber
* app_saycountpl
* app_readfile
* app_dahdibarge

It removes the following applications/functions:

* WaitMusicOnHold
* SetMusicOnHold
* SIPCHANINFO

It removes the colon delimiter from the SIPPEER function.

Finally, it also removes all compatibility options that were configurable from
asterisk.conf, as these all applied to compatibility with Asterisk 1.4 systems.

Review: https://reviewboard.asterisk.org/r/3698/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@418019 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agochan_dahdi: Add inband_on_setup_ack compatibility option.
Richard Mudgett [Thu, 3 Jul 2014 22:22:36 +0000 (22:22 +0000)]
chan_dahdi: Add inband_on_setup_ack compatibility option.

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

Q.931 Section 5.1.3 says that in scenarios with overlap dialing, when a
dialtone is sent from the network side, progress indicator 8 "Inband info
now available" MAY be sent to the CPE if no digits were received with the
SETUP.  It is thus implied that the ie is mandatory if digits came with
the SETUP and dialtone is needed.  This option should be enabled, when the
network sends dialtone and you want to hear it, but the network doesn't
send the progress indicator when needed.

NOTE: For Q.SIG setups this option should be enabled when outgoing overlap
dialing is also enabled because Q.SIG does not send the progress indicator
with the SETUP ACK.

The commit -r413714 (AST-1338) which causes this issue was dealing with a
SIP-to-ISDN interoperability issue.

This commit is a merge of the two patches indicated below.

ASTERISK-23897 #close
Reported by: Pavel Troller
Patches:
      pri-4.diff (license #6302) patch uploaded by Pavel Troller
      jira_asterisk_23897_v11.patch (license #5621) patch uploaded by rmudgett

Review: https://reviewboard.asterisk.org/r/3633/
........

Merged revisions 417956 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 417957 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 417958 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417976 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_ari: Fix some off-nominal paths just dropping the HTTP connection.
Richard Mudgett [Thu, 3 Jul 2014 19:06:12 +0000 (19:06 +0000)]
res_ari: Fix some off-nominal paths just dropping the HTTP connection.

* Removed some incorrect newlines on ast_http_error() messages in
manager.c.

* Removed an incorrect newline in res_ari_channels.c.

Addendum to ASTERISK-23552
........

Merged revisions 417932 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417933 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agochan_dahdi: Add AMI commands for controlling PRI debugging output
Jonathan Rose [Thu, 3 Jul 2014 17:34:32 +0000 (17:34 +0000)]
chan_dahdi: Add AMI commands for controlling PRI debugging output

Adds the following AMI commands:
PRIDebugSet - Set PRI debug levels for a specific span
PRIDebugFileSet - Set the file used for PRI debug message output
PRIDebugFileUnset - Disables file output for PRI debug messages

Review: https://reviewboard.asterisk.org/r/3681/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417916 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agopbx_config: Add manager actions to add/remove extensions
Jonathan Rose [Thu, 3 Jul 2014 17:20:00 +0000 (17:20 +0000)]
pbx_config: Add manager actions to add/remove extensions

Adds two new manager commands to pbx_config - DialplanExtensionAdd and
DialplanExtensionRemove which allow manager users to create and delete
extensions respectively.

Review: https://reviewboard.asterisk.org/r/3650/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417910 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoHTTP: Add persistent connection support.
Richard Mudgett [Thu, 3 Jul 2014 17:16:55 +0000 (17:16 +0000)]
HTTP: Add persistent connection support.

Persistent HTTP connection support is needed due to the increased usage of
the Asterisk core HTTP transport and the frequency at which REST API calls
are going to be issued.

* Add http.conf session_keep_alive option to enable persistent
connections.

* Parse and discard optional chunked body extension information and
trailing request headers.

* Increased the maximum application/json and
application/x-www-form-urlencoded body size allowed to 4k.  The previous
1k was kind of small.

* Removed a couple inlined versions of ast_http_manid_from_vars() by
calling the function.  manager.c:generic_http_callback() and
res_http_post.c:http_post_callback()

* Add missing va_end() in ast_ari_response_error().

* Eliminated unnecessary RAII_VAR() use in http.c:auth_create().

ASTERISK-23552 #close
Reported by: Scott Griepentrog

Review: https://reviewboard.asterisk.org/r/3691/
........

Merged revisions 417880 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417901 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agomain/tcptls: Add checks for OpenSSL Elliptic Curve support
Matthew Jordan [Thu, 3 Jul 2014 16:55:44 +0000 (16:55 +0000)]
main/tcptls: Add checks for OpenSSL Elliptic Curve support

The patch for ASTERISK-23905 that added PFS support in Asterisk depends on the
elliptic curve library support being present in OpenSSL. As it turns out, some
versions of OpenSSL don't have this library - notably the version running on
our build agents.

This patch fixes the build by providing a configure check for the specific
library calls that the PFS patch relies on.

Review: https://reviewboard.asterisk.org/r/3709/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417900 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoARI: Improvements to body parameters documentation
Sam Galarneau [Thu, 3 Jul 2014 16:14:39 +0000 (16:14 +0000)]
ARI: Improvements to body parameters documentation

The variables body parameter under the originate and originate with id
operations of the channel resource showed invalid JSON in its description.

The variables body parameter under the userEvent operation of the event
resource made no mention that the custom key/value pairs should be wrapped
in a variables key in order to be added to the custom user event.

ASTERISK-23975 #close

Review: https://reviewboard.asterisk.org/r/3692/
........

Merged revisions 417878 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417879 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoapi.wiki.mustache: Update wiki template to support body parameters
Sam Galarneau [Thu, 3 Jul 2014 15:35:22 +0000 (15:35 +0000)]
api.wiki.mustache: Update wiki template to support body parameters

This patch updates the api.wiki.mustache template and the swagger_model python
script to understand if an operation has a body parameter. If an operation
does have a body parameter, it will now be displayed in the corresponding
wiki entry.
........

Merged revisions 407389 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417877 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agodahdi_span_config_hook: automatically register new dahdi channels
Tzafrir Cohen [Thu, 3 Jul 2014 14:08:02 +0000 (14:08 +0000)]
dahdi_span_config_hook: automatically register new dahdi channels

Install a hook script for DAHDI to register new spans with Asterisk
automatically by running:

  asterisk -rx 'dahdi create channel FIRST LAST'

Review: https://reviewboard.asterisk.org/r/3157/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417863 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agomain/tcptls: Add support for Perfect Forward Secrecy
Matthew Jordan [Thu, 3 Jul 2014 12:10:17 +0000 (12:10 +0000)]
main/tcptls: Add support for Perfect Forward Secrecy

This patch enables Perfect Forward Secrecy (PFS) in Asterisk's core TLS API.
Modules that wish to enable PFS should consider the following:

- Ephemeral ECDH (ECDHE) is enabled by default. To disable it, do not
  specify a ECDHE cipher suite in a module's configuration, for example:
  tlscipher=AES128-SHA:DES-CBC3-SHA

- Ephemeral DH (DHE) is disabled by default. To enable it, add DH parameters
  into the private key file, i.e., tlsprivatekey. For an example, see the
  default dh2048.pem at
http://www.opensource.apple.com/source/OpenSSL098/OpenSSL098-35.1/src/apps/dh2048.pem?txt

- Because clients expect the server to prefer PFS, and because OpenSSL sorts
  its cipher suites by bit strength, (see "openssl ciphers -v DEFAULT")
  consider re-ordering your cipher suites in the conf file. For example:
tlscipher=AES128+kEECDH:AES128+kEDH:3DES+kEDH:AES128-SHA:DES-CBC3-SHA:-ADH:-AECDH
  will use PFS when offered by the client. Clients which do not offer PFS
  fall-back to AES-128 (or even 3DES as recommend by RFC 3261).

Review: https://reviewboard.asterisk.org/r/3647/

ASTERISK-23905 #close
Reported by: Alexander Traud
patches:
  tlsPFS_for_HEAD.patch uploaded by Alexander Traud (License 6520)
  tlsPFS.patch uploaded by Alexander Traud (License 6520)

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417803 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agomain/untils: Prevent potential infinite loop in ast_careful_fwrite
Matthew Jordan [Thu, 3 Jul 2014 11:27:25 +0000 (11:27 +0000)]
main/untils: Prevent potential infinite loop in ast_careful_fwrite

A loop in ast_careful_fwrite exists that will continually attempt to write to
a file stream, even in the presence of EAGAIN/EINTR errors. However, if a
connection that uses ast_careful_fwrite closes suddenly, ast_careful_fwrite's
call to fflush may return EAGAIN/EINTER along with EOF. A subsequent call to
fflush will return EOF but not clear errno, resulting in an infinite loop.

This patch clears errno after it is detected and handled the loop, such that
any subsequent call to fflush will not get erroneously stuck.

Review: https://reviewboard.asterisk.org/r/3704

#ASTERISK-23984 #close
Reported by: Steve Davies
patches:
  fflush_loop_fix uploaded by one47 (License 5012)
........

Merged revisions 417797 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 417798 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 417799 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417800 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoARI: Remove unnecessary \briefs from automatically generated documentation
Jonathan Rose [Wed, 2 Jul 2014 21:13:46 +0000 (21:13 +0000)]
ARI: Remove unnecessary \briefs from automatically generated documentation

Review: https://reviewboard.asterisk.org/r/3440/
........

Merged revisions 412653 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417770 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_rtp_asterisk: Don't leak memory or reset state if DTLS configuration is set multi...
Joshua Colp [Tue, 1 Jul 2014 14:42:46 +0000 (14:42 +0000)]
res_rtp_asterisk: Don't leak memory or reset state if DTLS configuration is set multiple times.
........

Merged revisions 417705 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417706 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoRecorded merge of revisions 417677 from http://svn.asterisk.org/svn/asterisk/branches/11
Joshua Colp [Mon, 30 Jun 2014 19:51:28 +0000 (19:51 +0000)]
Recorded merge of revisions 417677 from svn.asterisk.org/svn/asterisk/branches/11

........
res_rtp_asterisk: Add SHA-256 support for DTLS and perform DTLS negotiation on RTCP.

This change fixes up DTLS support in res_rtp_asterisk so it can accept and provide
a SHA-256 fingerprint, so it occurs on RTCP, and so it occurs after ICE negotiation
completes. Configuration options to chan_sip and chan_pjsip have also been added to
allow behavior to be tweaked (such as forcing the AVP type media transports in SDP).

ASTERISK-22961 #close
Reported by: Jay Jideliov

Review: https://reviewboard.asterisk.org/r/3679/
Review: https://reviewboard.asterisk.org/r/3686/
........

Merged revisions 417678 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417679 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoReverse logic during subscription persistence recreation.
Mark Michelson [Mon, 30 Jun 2014 18:39:56 +0000 (18:39 +0000)]
Reverse logic during subscription persistence recreation.

In the abstraction effort, this bit of logic got messed up. We
want to recreate the persistence if things go well, not if things
fail.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417663 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoapps/app_voicemail: Fix compilation error introduced in r417591
Matthew Jordan [Mon, 30 Jun 2014 13:02:43 +0000 (13:02 +0000)]
apps/app_voicemail: Fix compilation error introduced in r417591

Not sure why that change to ast_channel_alloc was made but ... okay.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417649 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoapp_voicemail, say: Add support for Japanese Language
Matthew Jordan [Mon, 30 Jun 2014 04:00:19 +0000 (04:00 +0000)]
app_voicemail, say: Add support for Japanese Language

This patch adds support for the Japanese language to both the say family of
applications, as well as for VoiceMail and VoiceMailMain. A new pack of
language sounds will be released at the same time as the next major version
of Asterisk to support the new language features.

The language features can be enabled using a language code of 'ja'.

Review: https://reviewboard.asterisk.org/r/3477

ASTERISK-23324 #close
Reported by: Kevin McCoy
patches:
  app_voicemail.c.20140226.jb.patch uploaded by Kevin McCoy (License 6586)
  say.c.20140226.jb.patch uploaded by Kevin McCoy (License 6586)

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417591 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agochan_sip: be more tolerant of whitespace between attributes in SDP fmtp line
Matthew Jordan [Mon, 30 Jun 2014 03:27:00 +0000 (03:27 +0000)]
chan_sip: be more tolerant of whitespace between attributes in SDP fmtp line

This patch is essentially a backport of a small portion of r397526 from
ASTERISK-21981. In that patch, pass through support and format attribute
negotiation was added for Opus. Part of that included being more tolerant to
whitespace in the fmtp line of an SDP; that part of the patch is being
applied here.

As the author of the backport pointed out, in SDP, the fmtp line is allowed to
include whitespace between attributes. RFC 3267 chapter 8.3 (from 2001)
includes an example for this. This was not removed in the updated RFC 4867 in
2007.

Review: https://reviewboard.asterisk.org/r/3658

#ASTERISK-23916 #close
Reported by: Alexander Traud
patches:
  sdpFMTPspace_Asterisk11.patch uploaded by Alexander Traud (License 6520)
........

Merged revisions 417587 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 417588 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 417589 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417590 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoevent.c: Fix type mismatch errors in ie_maps[].
Richard Mudgett [Fri, 27 Jun 2014 23:21:44 +0000 (23:21 +0000)]
event.c: Fix type mismatch errors in ie_maps[].

In v12+ the type values from the table are only used by the CEL unit
tests.  Since the unit tests were only comparing a generated expected
event with a real event to see if the ie contents matched and using the
same table IE_PLTYPE values to read the event contents, the type
mismatches were not detected.
........

Merged revisions 417565 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417571 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoEnsure REF_DEBUG records entrys for attempts to ao2_ref an invalid object
Corey Farrell [Fri, 27 Jun 2014 19:27:59 +0000 (19:27 +0000)]
Ensure REF_DEBUG records entrys for attempts to ao2_ref an invalid object

This change ensures that __ao2_ref_debug writes to ref_log when given a
non-NULL pointer to an invalid ao2 object.  This is to ensure that we
record any attempt manipulate references of already freed objects.

ASTERISK-23948 #close
Reported by: Corey Farrell
Review: https://reviewboard.asterisk.org/r/3677/
........

Merged revisions 417500 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 417505 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 417509 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417511 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agorefcounter.py: prevent use of excessive RAM with large refs logs
Corey Farrell [Fri, 27 Jun 2014 19:18:32 +0000 (19:18 +0000)]
refcounter.py: prevent use of excessive RAM with large refs logs

When processing a 212MB refs file, refcounter.py used over 3GB of RAM.
This change greatly reduces memory usage in two ways:

* Saving object history in whole lines instead of separated values.
* Not saving normal/skewed/leaked object lists unless they are requested.

ASTERISK-23921 #close
Reported by: Corey Farrell
Review: https://reviewboard.asterisk.org/r/3668/
........

Merged revisions 417480 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 417481 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 417483 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417485 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip: Add ActionID to events created as a result of PJSIP AMI actions
Matthew Jordan [Fri, 27 Jun 2014 13:50:02 +0000 (13:50 +0000)]
res_pjsip: Add ActionID to events created as a result of PJSIP AMI actions

A number of various PJSIP AMI actions were failing to parse out and place the
ActionID into their responses. This patch updates the various PJSIP actions
such that the passed in ActionID is emitted on any event list complete events,
as well as any intermediate events created as a result of the action.

#ASTERISK-23947 #close
Reported by: Mark Michelson

Review: https://reviewboard.asterisk.org/r/3675/
........

Merged revisions 417460 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417461 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoCEL: Update unit tests for bridge tech field
Kinsey Moore [Fri, 27 Jun 2014 02:04:50 +0000 (02:04 +0000)]
CEL: Update unit tests for bridge tech field

Update the CEL unit tests that handle BRIDGE_ENTER and BRIDGE_EXIT
events to expect the "bridge_technology" extra field key.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417447 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoCHANGES: Add missing changes
Kinsey Moore [Thu, 26 Jun 2014 19:15:10 +0000 (19:15 +0000)]
CHANGES: Add missing changes

Add missing CHANGES changes from r417361 and r417383.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417423 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_http_websocket: Export symbol for ast_websocket_set_timeout
Matthew Jordan [Thu, 26 Jun 2014 18:27:53 +0000 (18:27 +0000)]
res_http_websocket: Export symbol for ast_websocket_set_timeout

Thanks to Sean Bright for pointing out that this was missed in #asterisk-dev.
........

Merged revisions 417419 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 417420 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417421 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agochan_pjsip: Add a test event for fast picture updates
Matthew Jordan [Thu, 26 Jun 2014 17:17:25 +0000 (17:17 +0000)]
chan_pjsip: Add a test event for fast picture updates

This will drive the test on review r3419. Note that the patch for this was
done by Ben Ford, although it was slightly modified for this commit.

ASTERISK-23562
Reported by: Matt Jordan
........

Merged revisions 417399 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417400 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoCEL: Add bridge tech to relevant CEL records
Kinsey Moore [Thu, 26 Jun 2014 14:48:21 +0000 (14:48 +0000)]
CEL: Add bridge tech to relevant CEL records

Add the "bridge_technology" extra field key to BRIDGE_ENTER and
BRIDGE_EXIT CEL events to convey the bridge technology in use at the
time the record was generated.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417383 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoBridging: Allow channels to define bridging hooks
Kinsey Moore [Thu, 26 Jun 2014 12:43:47 +0000 (12:43 +0000)]
Bridging: Allow channels to define bridging hooks

This patch allows the current owner of a channel to define various
feature hooks to be made available once the channel has entered a
bridge. This includes any hooks that are setup on the
ast_bridge_features struct such as DTMF hooks, bridge event hooks
(join, leave, etc.), and interval hooks.

Review: https://reviewboard.asterisk.org/r/3649/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417361 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoapp_jack: Support audio with a sampling rate higher than 8kHz
Matthew Jordan [Thu, 26 Jun 2014 12:43:05 +0000 (12:43 +0000)]
app_jack: Support audio with a sampling rate higher than 8kHz

This patch enables the jack-audiohook to cope with dynamic sampling rates from
and to Asterisk. Information from the channel is taken to derive the channel's
sampling rate, suiting SLINxx format and frame->datalen.

There are stil a few limitations after this patch:
* Required information is taken from the channel during initialization as
  the audiohook does not provide this information.
  Audiohook.internal_sampl_rate(...) is set later, but no callback is available
  to inform app_jack.

* Frame.datalen is computed using "rate / 50" assuming a ptime of 20ms.
  There is no internal API available to determine datalen for a SLINxx.

* Ringbuffer size is now dynamic depending on the value of frame.datalen
  (see above) and the number of frames, which are in RINGBUFFER_FRAME_CAPACITY,
  that need to fit.

Review: https://reviewboard.asterisk.org/r/3618

Note that the patch being committed here is based on the patch posted on
ASTERISK-23836. However, Matthis Schmieder also provided a patch to enable
this functionality, and that patch is noted below.

ASTERISK-20696 #close
Reported by: Matthis Schmieder
patches:
  app_jack.patch uploaded by Matthis Schmieder (License 6445)

ASTERISK-23836 #close
Reported by: Dennis Guse
patches:
  patch-app_jack.c uploaded by Dennis Guse (License 6513)

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417360 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoudptl: Correct FEC to not consider negative sequence numbers as missing
Matthew Jordan [Thu, 26 Jun 2014 12:24:25 +0000 (12:24 +0000)]
udptl: Correct FEC to not consider negative sequence numbers as missing

When using FEC, with span=3 and entries=4 Asterisk will attempt to repair
the packet with sequence number 5, as it will see that packet -4 is
missing. The result is Asterisk sending garbage packets that can kill a
fax.

This patch adds a check to see if the sequence number is valid before
checking if the packet is missing.

Review: https://reviewboard.asterisk.org/r/3657/

#ASTERISK-23908 #close
Reported by: Torrey Searle
patches:
  udptl_fec.patch uploaded by Torrey Searle (License 5334)
........

Merged revisions 417318 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 417320 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 417324 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417327 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_http_websocket: Close websocket correctly and use careful fwrite
Matthew Jordan [Thu, 26 Jun 2014 12:21:14 +0000 (12:21 +0000)]
res_http_websocket: Close websocket correctly and use careful fwrite

When a client takes a long time to process information received from Asterisk,
a write operation using fwrite may fail to write all information. This causes
the underlying file stream to be in an unknown state, such that the socket
must be disconnected. Unfortunately, there are two problems with this in
Asterisk's existing websocket code:
1. Periodically, during the read loop, Asterisk must write to the connected
   websocket to respond to pings. As such, Asterisk maintains a reference to
   the session during the loop. When ast_http_websocket_write fails, it may
   cause the session to decrement its ref count, but this in and of itself
   does not break the read loop. The read loop's write, on the other hand,
   does not break the loop if it fails. This causes the socket to get in a
   'stuck' state, preventing the client from reconnecting to the server.
2. More importantly, however, is that the fwrite in ast_http_websocket_write
   fails with a large volume of data when the client takes awhile to process
   the information. When it does fail, it fails writing only a portion of
   the bytes. With some debugging, it was shown that this was failing in a
   similar fashion to ASTERISK-12767. Switching this over to ast_careful_fwrite
   with a long enough timeout solved the problem.

Note that this version of the patch, unlike r417310 in Asterisk 11, exposes
configuration options beyond just chan_sip's sip.conf. Configuration options
to configure the write timeout have also been added to pjsip.conf and ari.conf.

#ASTERISK-23917 #close
Reported by: Matt Jordan

Review: https://reviewboard.asterisk.org/r/3624/
........

Merged revisions 417310 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 417311 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417317 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agochan_sip: Fix handling of "From" headers longer than 256 characters
Corey Farrell [Thu, 26 Jun 2014 10:06:13 +0000 (10:06 +0000)]
chan_sip: Fix handling of "From" headers longer than 256 characters

From headers were processed using a 256 character buffer on the stack.
This change replaces that with a heap allocation by ast_strdup.

ASTERISK-23790 #close
Reported by: uniken1
Tested by: uniken1
Review: https://reviewboard.asterisk.org/r/3669/
Patches:
    chan_sip-large-from-header-1.8-r3.patch uploaded by wdoekes (license 5674)
........

Merged revisions 417248 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 417249 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 417250 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417251 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoAbstract PJSIP-specific elements from the pubsub API.
Mark Michelson [Wed, 25 Jun 2014 20:57:28 +0000 (20:57 +0000)]
Abstract PJSIP-specific elements from the pubsub API.

This helps to pave the way for RLS work that is to come.
Since this is a self-contained change and subscription
tests still pass, this work is being committed directly
to trunk instead of a working branch.

ASTERISK-23865 #close
Review: https://reviewboard.asterisk.org/r/3628

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417233 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoao2_container node object ignores REF_DEBUG in all places except one
Corey Farrell [Wed, 25 Jun 2014 18:57:04 +0000 (18:57 +0000)]
ao2_container node object ignores REF_DEBUG in all places except one

Almost every reference operation against container node's uses
__ao2_alloc or __ao2_ref, thereby preventing ref logging for
the nodes.  One node reference is released with ao2_t_ref, causing
refcounter.py to falsely report skews and leaks for many nodes.

ASTERISK-23922 #close
Reported by: Corey Farrell
Review: https://reviewboard.asterisk.org/r/3670/
........

Merged revisions 417212 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417213 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoSkinny: cleanup some log messages around sessions.
Damien Wedhorn [Wed, 25 Jun 2014 00:45:07 +0000 (00:45 +0000)]
Skinny: cleanup some log messages around sessions.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417193 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoMove eid functions to utils.c, mark netsock.h deprecated
Corey Farrell [Tue, 24 Jun 2014 02:50:15 +0000 (02:50 +0000)]
Move eid functions to utils.c, mark netsock.h deprecated

Move eid functions from netsock.c to utils.c.  These functions were
already published by utils.h.  Flag netsock.h as deprecated and switch
res_pjsip_session.h to use netsock2.h.  The only code that still uses
netsock.h is chan_iax2.

ASTERISK-23920 #close
Reported by: Corey Farrell
Review: https://reviewboard.asterisk.org/r/3661/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417167 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_rtp_asterisk: Return the length of data written when sending via ICE instead...
Joshua Colp [Mon, 23 Jun 2014 18:50:18 +0000 (18:50 +0000)]
res_rtp_asterisk: Return the length of data written when sending via ICE instead of 0.

ASTERISK-23834 #close
Reported by: Richard Kenner
........

Merged revisions 417141 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 417142 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417143 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agocore_unreal: Fix off by one buffer overwrite error.
Richard Mudgett [Mon, 23 Jun 2014 16:04:33 +0000 (16:04 +0000)]
core_unreal: Fix off by one buffer overwrite error.

Appending the ;2 to the user supplied ;1 uniqueid to create the ;2 version
if the user did not also supply an extra uniqueid for the ;2 channel
resulted in allocating a buffer that was one byte too small.

* Fix off by one error in ast_unreal_new_channels() when generating the ;2
uniqueid from the user suppled ;1 version.

* Pulled some long assignment lines from if tests to improve line break
readability in ast_unreal_new_channels().
........

Merged revisions 417119 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417120 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agosuspended destructions of pri spans on events
Tzafrir Cohen [Mon, 23 Jun 2014 07:44:19 +0000 (07:44 +0000)]
suspended destructions of pri spans on events

If a DAHDI span disappears, we wish for its representation in Asterisk
to be destroyed as well.

The information about the span's removal may come from several paths:

1. DAHDI sends DAHDI_EVENT_REMOVE on every channel.
2. An extra DAHDI_EVENT_REMOVED is sent on every subsequent call to
   DAHDI_GET_EVENT.
3. Every read (including the internal one by libpri on the D-channel)
   returns -ENODEV.

Asterisk responsds to DAHDI_EVENT_REMOVE on a channel by destroying it.

Destroying a channel requires holding the channel list lock (iflock).
Destroying a channel that is part of a span requires holding the span's
lock. Destroying a channel from a context that holds the span lock,
while at the same time another channel is destroyed directly, leads to a
deadlock. Solution: don't destroy span while holding the channels list
lock.

Thus changes in this patch:
* Deferring removal of PRI spans in response to events: doomed spans
  are collected on a list.
* Doomed spans are removed periodically by the monitor thread.
* ENODEV reads from the D-channel will warant the same deferred removal.

Review: https://reviewboard.asterisk.org/r/3548/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@417059 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoastobj2: Add an ao2_replace macro to astobj2.h
George Joseph [Sun, 22 Jun 2014 18:53:26 +0000 (18:53 +0000)]
astobj2: Add an ao2_replace macro to astobj2.h

This macro replaces one object reference with another cleaning up the original.

param dst Pointer to the object that will be cleaned up.
param src Pointer to the object replacing it.

src's ref count is bumped if it's non-NULL.
dst's ref count is decremented if it's non-NULL.
src is assigned to dst,

This patch was reviewed on IRC by coreyfarrell and mjordan.

Tested by: George Joseph
........

Merged revisions 416995 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416996 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agobuild: Allow autoconf/ast_ext_tool_check to handle cross-compiling better.
George Joseph [Fri, 20 Jun 2014 23:18:36 +0000 (23:18 +0000)]
build: Allow autoconf/ast_ext_tool_check to handle cross-compiling better.

ast_ext_tool_check.m4 isn't handling cases where a path to a package is
provided (E.G. --with-mysqlclient=/some/sysroot) and the package has a config
tool (E.G. mysql_config) and the package has its own subdirectories in include
or lib.  For example, mysql's libraries are in ${MYSQLCLIENT_DIR}/usr/lib/mysql
but ast_ext_tool_check sets MYSQLCLIENT_LIB to ${MYSQLCLIENT_DIR}/usr/lib.
libxml2 has the same problem with its includes.  They're in
${LIBXML2_DIR}/usr/include/libxml2 not directly in ${LIBXML2_DIR}/usr/include.
Both cause configure to fail and there are others in the same boat.

The problem is caused by logic in ast_ext_tool_check that overrides the result
of the config tool's --cflags and --libs options if package_DIR is set.

This patch prepends package_DIR (if specified) to the -L and -I results from
the package's config tool instead of overriding them.

A regenerated ./configure and include/asterisk/autoconfig.h.in are included
but can be regenerated by running ./bootstrap.sh at any time.

Tested by: George Joseph
Tested by: Matt Jordan
Review: https://reviewboard.asterisk.org/r/3550/
........

Merged revisions 416929 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 416930 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 416931 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416935 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agobuild: Allow autoconf/ast_ext_tool_check to handle cross-compiling better.
George Joseph [Fri, 20 Jun 2014 21:59:08 +0000 (21:59 +0000)]
build: Allow autoconf/ast_ext_tool_check to handle cross-compiling better.

ast_ext_tool_check.m4 isn't handling cases where a path to a package is
provided (E.G. --with-mysqlclient=/some/sysroot) and the package has a config
tool (E.G. mysql_config) and the package has its own subdirectories in include
or lib.  For example, mysql's libraries are in ${MYSQLCLIENT_DIR}/usr/lib/mysql
but ast_ext_tool_check sets MYSQLCLIENT_LIB to ${MYSQLCLIENT_DIR}/usr/lib.
libxml2 has the same problem with its includes.  They're in
${LIBXML2_DIR}/usr/include/libxml2 not directly in ${LIBXML2_DIR}/usr/include.
Both cause configure to fail and there are others in the same boat.

The problem is caused by logic in ast_ext_tool_check that overrides the result
of the config tool's --cflags and --libs options if package_DIR is set.

This patch prepends package_DIR (if specified) to the -L and -I results from
the package's config tool instead of overriding them.

Tested by: George Joseph
Tested by: Matt Jordan
Review: https://reviewboard.asterisk.org/r/3550/
........

Merged revisions 416870 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 416871 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416872 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_parking: Make manager commands register with module information
Jonathan Rose [Fri, 20 Jun 2014 20:57:46 +0000 (20:57 +0000)]
res_parking: Make manager commands register with module information

Previously module information was not included due to an oversight.

Review: https://reviewboard.asterisk.org/r/3626/
........

Merged revisions 416849 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416850 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoLogger: Add manager command 'LoggerRotate' to rotate logger
Jonathan Rose [Fri, 20 Jun 2014 20:29:45 +0000 (20:29 +0000)]
Logger: Add manager command 'LoggerRotate' to rotate logger

Part of a series of AMI command equivalents to existing CLI
commands

Review: https://reviewboard.asterisk.org/r/3651/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416848 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agovoicemail API callbacks: Extract the sayname API call to its own registerd callback.
Richard Mudgett [Fri, 20 Jun 2014 17:06:42 +0000 (17:06 +0000)]
voicemail API callbacks: Extract the sayname API call to its own registerd callback.

* Extract the sayname API call to its own registerd callback.  This allows
the app_directory and app_chanspy applications to say a mailbox owner's
name using an alternate provider when app_voicemail is not available
because you are using res_mwi_external.  app_directory still uses the
voicemail.conf file.

AFS-64 #close
Reported by: Mark Michelson

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416830 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoastobj2: Additional refactoring to push impl specific code down into the impls.
George Joseph [Fri, 20 Jun 2014 15:27:43 +0000 (15:27 +0000)]
astobj2: Additional refactoring to push impl specific code down into the impls.

Move some implementation specific code from astobj2_container.c into
astobj2_hash.c and astobj2_rbtree.c.  This completely removes the need for
astobj2_container to switch on RTTI and it no longer has any knowledge of
the implementation details.

Also adds AO2_DEBUG as a new compile option in menuselect which controls
astobj2 debugging independently of AST_DEVMODE and REF_DEBUG.

Tested by: George Joseph
Review: https://reviewboard.asterisk.org/r/3593/
........

Merged revisions 416806 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416807 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agopjsip cli: Change Identify to show CIDR notation instead of netmasks.
George Joseph [Thu, 19 Jun 2014 20:13:20 +0000 (20:13 +0000)]
pjsip cli: Change Identify to show CIDR notation instead of netmasks.

* Added ast_sockaddr_cidr_bits() to count the 1 bits in an ast_sockaddr.
* Added ast_ha_join_cidr() which uses ast_sockaddr_cidr_bits() for the netmask
  instead of ast_sockaddr_stringify_addr.
* Changed res_pjsip_endpoint_identifier_ip to call ast_ha_join_cidr() instead
  of ast_ha_join() for the CLI output.

This is a CLI change only.  AMI was not affected.

Tested by: George Joseph
Review: https://reviewboard.asterisk.org/r/3652/
........

Merged revisions 416737 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416738 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoFix build warnings with TEST_FRAMEWORK enabled
Kinsey Moore [Thu, 19 Jun 2014 19:40:45 +0000 (19:40 +0000)]
Fix build warnings with TEST_FRAMEWORK enabled
........

Merged revisions 416732 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 416733 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 416734 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416736 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoRemove the problematic and unneeded AST_MODFLAG_GLOBAL_SYMBOLS from pbx_lua.c
George Joseph [Thu, 19 Jun 2014 16:04:36 +0000 (16:04 +0000)]
Remove the problematic and unneeded AST_MODFLAG_GLOBAL_SYMBOLS from pbx_lua.c

AST_MODFLAG_GLOBAL_SYMBOLS was causing the module to be incorrectly loaded
before pbx_config.  pbx_config was therefore blowing away contexts that were
created by pbx_lua.  With AST_MODFLAG_DEFAULT the load order is now correct
and contexs are being properly merged.  AST_MODFLAG_GLOBAL_SYMBOLS was not
needed anyway since no other modules needed its global symbols that early.

ASTERISK-23818 #close
Reported by: Dennis Guse
Tested by: Dennis Guse
Tested by: George Joseph

Review: https://reviewboard.asterisk.org/r/3629/
........

Merged revisions 416668 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 416669 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416670 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoUpdate extensions.lua.sample with naming conflict guidance.
George Joseph [Wed, 18 Jun 2014 17:20:13 +0000 (17:20 +0000)]
Update extensions.lua.sample with naming conflict guidance.

The sample extensions.lua was causing pbx_lua to fail to load when parsing
'app.goto("default", "s", 1)' because in Lua 5.2, 'goto' is now a reserved
word.  This patch adds guidance to extensions.lua.sample and changed
'app.goto("default", "s", 1)' to 'app.['goto']("default", "s", 1)'.

ASTERISK-23844 #close
Reported by: rnewton
Tested by: gtjoseph
Review: https://reviewboard.asterisk.org/r/3627/

........

Merged revisions 416581 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 416582 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416589 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agostasis_channels: Update the stasis cache if manager variables are needed
Matthew Jordan [Wed, 18 Jun 2014 04:22:05 +0000 (04:22 +0000)]
stasis_channels: Update the stasis cache if manager variables are needed

In r416211, the publishing of variable changes was modified such that a
cached channel snapshot was used if manager variables were not requested
with each AMI event. This was done to reduce the amount of channel snapshots
created.

However, an assumption was made that generating a channel snapshot and
publishing the snapshot to the channel topic was sufficient to ensure that
the cache would be updated; this is not the case. The channel snapshot type
must be used to force a snapshot update.

This patch updates the publication of channel variables such that the cache
is updated prior to publication of the channel variable message if manager
variables are in use. This ensures that all AMI events receive the variable
update when they are supposed to.

Note that this issue was caught by the Asterisk Test Suite (go go testing)
........

Merged revisions 416557 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416561 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoAllow the PUSH and UNSHIFT functions to set inheritable channel variables.
Mark Michelson [Tue, 17 Jun 2014 18:45:11 +0000 (18:45 +0000)]
Allow the PUSH and UNSHIFT functions to set inheritable channel variables.
........

Merged revisions 416500 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 416501 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 416502 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416503 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoFix string growth algorithm for XML presence bodies.
Mark Michelson [Tue, 17 Jun 2014 16:35:04 +0000 (16:35 +0000)]
Fix string growth algorithm for XML presence bodies.

pjpidf_print() does not return < 0 if there is not enough
room for the document to be printed. Rather, it returns
39, the length of the XML prolog.

The algorithm also had a bug in that it would return if
it attempted to grow the string larger.
........

Merged revisions 416442 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416444 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoMoH: Don't restart stream on repeated start calls
Kinsey Moore [Tue, 17 Jun 2014 16:33:53 +0000 (16:33 +0000)]
MoH: Don't restart stream on repeated start calls

Currently, music on hold will stop and then start again from the
beginning if ast_moh_start() is called multiple times. This can happen
if a call is put on hold repeatedly (the channel receives multiple
HOLD control frames) and can be triggered from ARI by starting MoH on a
channel multiple times. This is fairly jarring/annoying to users.

This change prevents MoH from being restarted if the requested music
class is the same as the one currently playing.

This includes an extra check to prevent the errors previously
experienced in the testsuite and has 100+ test runs behind it.

Review: https://reviewboard.asterisk.org/r/3615/
........

Merged revisions 416439 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 416440 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 416441 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416443 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agochan_dahdi: Adds support for major update to libss7.
Richard Mudgett [Mon, 16 Jun 2014 18:27:51 +0000 (18:27 +0000)]
chan_dahdi: Adds support for major update to libss7.

* SS7 support now requires libss7 v2.0 or later.  The new libss7 is not
backwards compatible.

* Added SS7 support for connected line and redirecting.

* Most SS7 CLI commands are reworked as well as new SS7 commands added.
See online CLI help.

* Added several SS7 config option parameters described in
chan_dahdi.conf.sample.

* ISUP timer support reworked and now requires explicit configuration.
See ss7.timers.sample.

Special thanks to Kaloyan Kovachev for his support and persistence in
getting the original patch by adomjan updated and ready for release.

SS7-27 #close
Reported by: adomjan

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416416 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_http_websocket: read/write string fixup
Kevin Harwell [Mon, 16 Jun 2014 16:22:33 +0000 (16:22 +0000)]
res_http_websocket: read/write string fixup

There was a problem when reading a string from the websocket. It assumed the
received data had a null terminator and tried to write the data to an ast_str.
This of course could/would read past the end of the given buffer while
writing the data to the internal buffer of ast_str. Modified the the code to
correctly place a null terminator on the result string.

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416394 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoWe have faced situation when using CDR and CEL by sqlite3 modules. With system having...
Igor Goncharovskiy [Mon, 16 Jun 2014 09:04:05 +0000 (09:04 +0000)]
We have faced situation when using CDR and CEL by sqlite3 modules. With system having high load (~100 concurrent calls created by sipp) we found many cdr and cel records missed. There is special finction in sqlite3, that make able to fix this situation - sqlite3_wait_timeout, that also can replace awful code cdr_sqlite3 ad cel_sqlite3 modules. Also this function can be used for aastdb and res_config_sqlite3 to avoid missed writes to sqlite db.

#ASTERISK-23766 #close
Reported by: Igor Goncharovsky

Review: https://reviewboard.asterisk.org/r/3559/
........

Merged revisions 416336 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 416337 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 416338 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416339 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agochannels/chan_sip: Forbid remote bridging if T.38 is negotiated
Matthew Jordan [Mon, 16 Jun 2014 02:40:44 +0000 (02:40 +0000)]
channels/chan_sip: Forbid remote bridging if T.38 is negotiated

When a framehook is removed - such as the fax gateway framehook - the bridge
framework will re-evaluate the bridge mixing technologies to see if it can
improve the bridging. When this occurs, get_rtp_info will be called to
determine if local or remote bridging can be used. Using remote bridging
will cause a fax to fail, as direct media negotiation will cause some small
number of packets to not arrive at the remote endpoint.

This patch forces local native bridging if T.38 negotiation is in progress or
has been established.
........

Merged revisions 416318 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416319 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agochannel_internal_api: Publish a snapshot change when linkedids change
Matthew Jordan [Sun, 15 Jun 2014 22:12:49 +0000 (22:12 +0000)]
channel_internal_api: Publish a snapshot change when linkedids change

Snapshots are now not published *quite* as much as they used to. One instance
where they are not published any longer is during bridge enter and exit - the
state of the channel doesn't change, the bridge does. However, channels are
changed when a linkedid is propagated; previously, the channel's state would
be updated and published during the bridge enter event. Now this must be
explicitly done.
........

Merged revisions 416300 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416301 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agotest_stasis_endpoints: Remove expected channel snapshot
Matthew Jordan [Sun, 15 Jun 2014 21:42:45 +0000 (21:42 +0000)]
test_stasis_endpoints: Remove expected channel snapshot

We no longer publish a channel snapshot when it is associated with an endpoint;
after all, the channel itself hasn't changed - the endpoint state has changed.
This updates the channel_messages unit test accordingly.
........

Merged revisions 416298 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416299 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoMoH: Undo commit r416150 (1.8)
Matthew Jordan [Sun, 15 Jun 2014 21:24:04 +0000 (21:24 +0000)]
MoH: Undo commit r416150 (1.8)

This patch reverts r416150. When the comparison between mohclass->name and
state->class->name is made, you are not guaranteed that (a) state->class is
non-NULL or that state or state->class are in a safe state.

Crashes caught by the bridges/transfer_capabilities test.
........

Merged revisions 416251 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 416252 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 416255 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416267 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_manager_devicestate and res_manager_presencestate missing support level
Corey Farrell [Sat, 14 Jun 2014 19:26:16 +0000 (19:26 +0000)]
res_manager_devicestate and res_manager_presencestate missing support level

Add MODULEINFO comment block to define support level core for these new
modules.

Review: https://reviewboard.asterisk.org/r/3620/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416237 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agostasis: Reduce creation of channel snapshots to improve performance
Matthew Jordan [Fri, 13 Jun 2014 18:24:49 +0000 (18:24 +0000)]
stasis: Reduce creation of channel snapshots to improve performance

During some performance testing of Asterisk with AGI, ARI, and lots of Local
channels, we noticed that there's quite a hit in performance during channel
creation and releasing to the dialplan (ARI continue). After investigating
the performance spike that occurs during channel creation, we discovered
that we create a lot of channel snapshots that are technically unnecessary.
This includes creating snapshots during:
 * AGI execution
 * Returning objects for ARI commands
 * During some Local channel operations
 * During some dialling operations
 * During variable setting
 * During some bridging operations
And more.

This patch does the following:
 - It removes a number of fields from channel snapshots. These fields were
   rarely used, were expensive to have on the snapshot, and hurt performance.
   This included formats, translation paths, Log Call ID, callgroup, pickup
   group, and all channel variables. As a result, AMI Status,
   "core show channel", "core show channelvar", and "pjsip show channel" were
   modified to either hit the live channel or not show certain pieces of data.
   While this is unfortunate, the performance gain from this patch is worth
   the loss in behaviour.
 - It adds a mechanism to publish a cached snapshot + blob. A large number of
   publications were changed to use this, including:
   - During Dial begin
   - During Variable assignment (if no AMI variables are emitted - if AMI
     variables are set, we have to make snapshots when a variable is changed)
   - During channel pickup
   - When a channel is put on hold/unhold
   - When a DTMF digit is begun/ended
   - When creating a bridge snapshot
   - When an AOC event is raised
   - During Local channel optimization/Local bridging
   - When endpoint snapshots are generated
   - All AGI events
   - All ARI responses that return a channel
   - Events in the AgentPool, MeetMe, and some in Queue
 - Additionally, some extraneous channel snapshots were being made that were
   unnecessary. These were removed.
 - The result of ast_hashtab_hash_string is now cached in stasis_cache. This
   reduces a large number of calls to ast_hashtab_hash_string, which reduced
   the amount of time spent in this function in gprof by around 50%.

#ASTERISK-23811 #close
Reported by: Matt Jordan

Review: https://reviewboard.asterisk.org/r/3568/
........

Merged revisions 416211 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416216 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoMoH: Don't restart stream on repeated start calls
Kinsey Moore [Fri, 13 Jun 2014 13:11:38 +0000 (13:11 +0000)]
MoH: Don't restart stream on repeated start calls

Currently, music on hold will stop and then start again from the
beginning if ast_moh_start() is called multiple times. This can happen
if a call is put on hold repeatedly (the channel receives multiple
HOLD control frames) and can be triggered from ARI by starting MoH on a
channel multiple times. This is fairly jarring/annoying to users.

This change prevents MoH from being restarted if the requested music
class is the same as the one currently playing.

Review: https://reviewboard.asterisk.org/r/3615/
........

Merged revisions 416150 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 416151 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 416152 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416153 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoCEL: Expose parking retreiver in extra field
Kinsey Moore [Fri, 13 Jun 2014 12:56:06 +0000 (12:56 +0000)]
CEL: Expose parking retreiver in extra field

This exposes the retreiver of a parked call under the "retreiver" key
of the extra field when this information is available.

Review: https://reviewboard.asterisk.org/r/3608/
........

Merged revisions 416148 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416149 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoAST-2014-007: Fix of fix to allow AMI and SIP TCP to send messages.
Richard Mudgett [Fri, 13 Jun 2014 05:16:34 +0000 (05:16 +0000)]
AST-2014-007: Fix of fix to allow AMI and SIP TCP to send messages.

ASTERISK-23673 #close
Reported by: Richard Mudgett

Review: https://reviewboard.asterisk.org/r/3617/
........

Merged revisions 416066 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 416067 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 416070 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416071 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agomain/pbx - documentation - enhance 'core show hints' and 'core show hint' help text
Rusty Newton [Thu, 12 Jun 2014 21:27:02 +0000 (21:27 +0000)]
main/pbx - documentation - enhance 'core show hints' and 'core show hint' help text

Adds descriptive help text to 'core show hints' and 'core show hint'. The text describes the various columns for the sake of clarity. It takes into account recent changes to the content displayed by the commands https://reviewboard.asterisk.org/r/3604/ and https://reviewboard.asterisk.org/r/3611/.

ASTERISK-23764
Review: https://reviewboard.asterisk.org/r/3610/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416024 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoFix build in devmode for GCC 4.10
Kinsey Moore [Thu, 12 Jun 2014 20:17:37 +0000 (20:17 +0000)]
Fix build in devmode for GCC 4.10
........

Merged revisions 415980 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415982 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoAST-2014-007: Fix DOS by consuming the number of allowed HTTP connections.
Richard Mudgett [Thu, 12 Jun 2014 17:00:08 +0000 (17:00 +0000)]
AST-2014-007: Fix DOS by consuming the number of allowed HTTP connections.

Simply establishing a TCP connection and never sending anything to the
configured HTTP port in http.conf will tie up a HTTP connection.  Since
there is a maximum number of open HTTP sessions allowed at a time you can
block legitimate connections.

A similar problem exists if a HTTP request is started but never finished.

* Added http.conf session_inactivity timer option to close HTTP
connections that aren't doing anything.  Defaults to 30000 ms.

* Removed the undocumented manager.conf block-sockets option.  It
interferes with TCP/TLS inactivity timeouts.

* AMI and SIP TLS connections now have better authentication timeout
protection.  Though I didn't remove the bizzare TLS timeout polling code
from chan_sip.

* chan_sip can now handle SSL certificate renegotiations in the middle of
a session.  It couldn't do that before because the socket was non-blocking
and the SSL calls were not restarted as documented by the OpenSSL
documentation.

* Fixed an off nominal leak of the ssl struct in
handle_tcptls_connection() if the FILE stream failed to open and the SSL
certificate negotiations failed.

The patch creates a custom FILE stream handler to give the created FILE
streams inactivity timeout and timeout after a specific moment in time
capability.  This approach eliminates the need for code using the FILE
stream to be redesigned to deal with the timeouts.

This patch indirectly fixes most of ASTERISK-18345 by fixing the usage of
the SSL_read/SSL_write operations.

ASTERISK-23673 #close
Reported by: Richard Mudgett
........

Merged revisions 415841 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 415854 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 415896 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415907 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoBlocked revisions 415838
Jonathan Rose [Thu, 12 Jun 2014 15:52:20 +0000 (15:52 +0000)]
Blocked revisions 415838

........
Correct UPGRADE.txt notes in r415825

The change was marked against the wrong version of Asterisk. My apologies.
........

Merged revisions 415837 from http://svn.asterisk.org/svn/asterisk/branches/11

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415840 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoapp_queue: delayed state can cause early leavewhenempty ringing
Scott Griepentrog [Thu, 12 Jun 2014 15:50:48 +0000 (15:50 +0000)]
app_queue: delayed state can cause early leavewhenempty ringing

In app_queue, device state changes arrive in event messages and
update the queue member status value.  That value is checked in
get_member_status() to decide that the caller should leave when
there are no available members.  Although event messages can be
delayed by other activity, there is no adverse affect by lagged
status except in one specific case: there is only one available
member, it was just rung, and leavewhenempty is enabled set for
ringing members.  This change adds a direct check of the device
state only under this condition where the caller may be dropped
incorrectly, resolving this issue without affecting performance
of app_queue normally.

AST-1248 #close
Review: https://reviewboard.asterisk.org/r/3595/
Reported by: Thomas Arimont
........

Merged revisions 415833 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 415835 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 415836 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415839 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoMixMontior: Add class authorization requirements to MixMonitor AMI commands
Jonathan Rose [Thu, 12 Jun 2014 15:39:52 +0000 (15:39 +0000)]
MixMontior: Add class authorization requirements to MixMonitor AMI commands

MixMonitor AMI commands StartMixMonitor and StopMixMonitor lacked class
authorization. StopMixMonitor now requires that the manager user either have
the call or system class authorization. StartMixMonitor is a slightly larger
issue since it can execute shell commands if the right arguments are passed
into it, and we consider this a permission escalation. A security release
will be issued for problem this shortly.

ASTERISK-23609 #close
Reported by: Corey Farrell

........

Merged revisions 415825 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 415832 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415834 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_pubsub: unauthenticated remote crash in PJSIP pub/sub framework
Kevin Harwell [Thu, 12 Jun 2014 14:39:29 +0000 (14:39 +0000)]
res_pjsip_pubsub: unauthenticated remote crash in PJSIP pub/sub framework

A remotely exploitable crash vulnerability exists in the PJSIP channel driver's
pub/sub framework. If an attempt is made to unsubscribe when not currently
subscribed and the endpoint's "sub_min_expiry" is set to zero, Asterisk tries
to create an expiration timer with zero seconds, which is not allowed, so an
assertion raised.

The fix was to reject a subscription that is attempting to unsubscribe when not
being already subscribed.  Asterisk now checks for this situation appropriately
and responds with a 400 instead of crashing.

AST-2014-005

ASTERISK-23489 #close
........

Merged revisions 415812 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415813 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoFix potential deadlock situation in res_pjsip.
Mark Michelson [Thu, 12 Jun 2014 14:15:51 +0000 (14:15 +0000)]
Fix potential deadlock situation in res_pjsip.

SIP transaction timeouts are handled in the PJSIP monitor thread. When
this happens on a subscription, and the subscription is destroyed, the
subscription destruction is dispatched synchronously to the threadpool.
The issue is that the PJSIP dialog is locked by the monitor thread,
and then the dispatched task attempts to lock the dialog. This leads
to a deadlock that causes SIP traffic to no longer be accepted on the
Asterisk server.

The fix here is to treat the monitor thread as if it were a threadpool
thread when it attempts to dispatch synchronous tasks. This way, the
dispatched task turns into a simple function call within the same thread,
and the locking issue is averted.

AST-2014-008

ASTERISK-23802 #close
........

Merged revisions 415794 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415795 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agores_pjsip_pubsub: Persist subscriptions in sorcery so they are recreated on startup.
Joshua Colp [Thu, 12 Jun 2014 11:34:36 +0000 (11:34 +0000)]
res_pjsip_pubsub: Persist subscriptions in sorcery so they are recreated on startup.

This change makes res_pjsip_pubsub persist inbound subscriptions in sorcery. By default
this uses the local astdb but it can also be configured to store within an outside
database. When Asterisk is started these subscriptions are recreated if they have not
expired. Notifications are sent to the devices which have subscribed and they are none
the wiser that the system has restarted.

Review: https://reviewboard.asterisk.org/r/3598/
........

Merged revisions 415766 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415767 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agosafe_asterisk: Overwrite old safe_asterisk on make install.
Walter Doekes [Thu, 12 Jun 2014 07:52:59 +0000 (07:52 +0000)]
safe_asterisk: Overwrite old safe_asterisk on make install.

From now on, make install will overwrite safe_asterisk with the
latest version. You need to move any local modifications to files
inside /etc/asterisk/startup.d, if you have any.

See also commits r394939 and r397938.

ASTERISK-21965 #close
Patches:
  safe_asterisk.patch uploaded by jkister (License 6232, modified by me)
........

Merged revisions 415748 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415749 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoformat.c: Fix misuse of hash container function.
Richard Mudgett [Wed, 11 Jun 2014 23:01:19 +0000 (23:01 +0000)]
format.c: Fix misuse of hash container function.

The supplied hash function to a container must be idempotent given the
object's key value to figure out which container bucket the object belongs
in.  Returning a random number or the current container count is not
idempotent.  The "computed hash" value doesn't help find the object later
in those cases.

* Fixed the format_list container to actually be a list since that is how
the container is used.  Conceptually, if more than 283 formats were added
to the format_list then odd things may have happened before the fix.
........

Merged revisions 415728 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 415729 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415730 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoCLI: correct presence information on core show hints
Scott Griepentrog [Wed, 11 Jun 2014 20:22:23 +0000 (20:22 +0000)]
CLI: correct presence information on core show hints

Adds presence to core show hint and changes presence
string conversion to use the correct function.

ASTERISK-23858 #close
Review: https://reviewboard.asterisk.org/r/3611/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415715 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoCLI: add presence information to core show hints
Scott Griepentrog [Tue, 10 Jun 2014 22:31:04 +0000 (22:31 +0000)]
CLI: add presence information to core show hints

Adds presence state value to output of core show
hints.  Also reformats the output slightly so it
doesn't use as much space as it would otherwise.

Was:
                   1000@demo                : SIP/1000              State:Unavailable     Watchers  0

Now:
1000@demo           : SIP/1000              State:Unavailable     Presence:Idle            Watchers  0

AFS-53 #close
Review: https://reviewboard.asterisk.org/r/3604/

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415698 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoFix build in dev mode due to signed/unsigned mismatch
Kinsey Moore [Tue, 10 Jun 2014 18:32:12 +0000 (18:32 +0000)]
Fix build in dev mode due to signed/unsigned mismatch
........

Merged revisions 415678 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415679 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoPJSIP: PJSIPNotify - Strip content-length headers and add documentation
Jonathan Rose [Tue, 10 Jun 2014 16:06:12 +0000 (16:06 +0000)]
PJSIP: PJSIPNotify - Strip content-length headers and add documentation

Documentation for how to add custom headers/content to notifies created
with the PJSIPNotify manager action was a little sparse and it also
wasn't vetting application of Content-length headers like its chan_sip
equivalent was (so two Content-length headers could be applied... and
PJSIP determines the content length anyway, so it just opens people up
for error). This patch also flips the variable order so that the
variables are interpreted in the same order as they are put in the AMI
action.

Review: https://reviewboard.asterisk.org/r/3587/
........

Merged revisions 415658 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415659 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agochan_ooh323: fix loading module failure if there no accessible h323_log or ooh323...
Alexandr Anikin [Tue, 10 Jun 2014 09:28:55 +0000 (09:28 +0000)]
chan_ooh323: fix loading module failure if there no accessible h323_log or ooh323 config file

change return 1 to return AST_MODULE_LOAD_FAILURE on module load routine
few cosmetic changes

ASTERISK-23814 #close

(closes issue ASTERISK-23814)

Reported by: Igor Goncharovsky
Patches:
ASTERISK-23814-ast11.patch
........

Merged revisions 415599 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 415602 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415630 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agochan_pjsip: Fix bug where custom SIP headers could be duplicated on outgoing INVITEs.
Mark Michelson [Mon, 9 Jun 2014 20:21:42 +0000 (20:21 +0000)]
chan_pjsip: Fix bug where custom SIP headers could be duplicated on outgoing INVITEs.

When using PJSIP_HEADER() to add custom headers to outgoing INVITE requests, certain
situations could result in the headers being duplicated. For instance, if the request
were retransmitted, or if the INVITE were re-sent with authentication credentials,
the custom headers would be re-added to the request.

The fix here is to, after adding the custom headers to the outbound INVITE, remove
the datastore that holds the custom headers to add. This way, there is no risk in
accidentally adding them if the session supplement is called into a second or third
time.
........

Merged revisions 415579 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415580 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agosafe_asterisk: Cleanup additions to r415132.
Walter Doekes [Mon, 9 Jun 2014 12:12:25 +0000 (12:12 +0000)]
safe_asterisk: Cleanup additions to r415132.

* Replaced a stray echo that should've been a message call in
  safe_asterisk. This replaces a conditional log message by a slightly
  different message. Please update your log parsing scripts.
* Made the $NOTIFY mail Subject more verbose by adding the machine name
  and exitstatus.

(Note that a 'make install' still won't overwrite your old safe_asterisk
if it exists. See ASTERISK-21965.)

ASTERISK-23492 #close
........

Merged revisions 415521 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 415522 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 415523 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415524 65c4cc65-6c06-0410-ace0-fbb531ad65f3

5 years agoautoservice: stop thread on graceful shutdown
Corey Farrell [Mon, 9 Jun 2014 03:50:45 +0000 (03:50 +0000)]
autoservice: stop thread on graceful shutdown

This change adds thread shutdown to autoservice for graceful shutdowns only.
ast_register_cleanup is backported to 1.8 to allow this.  The logger callid
is also released on shutdown in 11+.

ASTERISK-23827 #close
Reported by: Corey Farrell
Review: https://reviewboard.asterisk.org/r/3594/
........

Merged revisions 415463 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........

Merged revisions 415464 from http://svn.asterisk.org/svn/asterisk/branches/11
........

Merged revisions 415465 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@415466 65c4cc65-6c06-0410-ace0-fbb531ad65f3