5 years agores_pjproject: Add ability to map pjproject log levels to Asterisk log levels
George Joseph [Sun, 7 Feb 2016 23:34:20 +0000 (16:34 -0700)]
res_pjproject:  Add ability to map pjproject log levels to Asterisk log levels

Warnings and errors in the pjproject libraries are generally handled by
Asterisk.  In many cases, Asterisk wouldn't even consider them to be warnings
or errors so the messages emitted by pjproject directly are either superfluous
or misleading.  A good exampe of this are the level-0 errors pjproject emits
when it can't open a TCP/TLS socket to a client to send an OPTIONS.  We don't
consider a failure to qualify a UDP client an "ERROR", why should a TCP/TLS
client be treated any differently?

A config file for res_pjproject has bene added (pjproject.conf) and a new
log_mappings object allows mapping pjproject levels to Asterisk levels
(or nothing).  The defaults if no pjproject.conf file is found are the same
as those that were hard-coded into res_pjproject initially: 0,1 = LOG_ERROR,
2 = LOG_WARNING, 3,4,5 = LOG_DEBUG<level>

Change-Id: Iba7bb349c70397586889b8f45b8c3d6c6c8c3898

5 years agoMerge "app_queue: fix Calculate talktime when is first call answered"
Joshua Colp [Thu, 18 Feb 2016 19:12:03 +0000 (13:12 -0600)]
Merge "app_queue: fix Calculate talktime when is first call answered"

5 years agocel.c: Fix mismatch in ast_cel_track_event() return type.
Richard Mudgett [Wed, 17 Feb 2016 19:30:06 +0000 (13:30 -0600)]
cel.c: Fix mismatch in ast_cel_track_event() return type.

The return type of ast_cel_track_event() is not large enough to return all
64 potential bits of the event enable mask.  Fortunately, the defined CEL
events do not really need all 64 bits and the return value is only used to
determine if the requested CEL event is enabled.

* Made the ast_cel_track_event() return 0 or 1 only so the return value
can fit inside an int type instead of zero or a truncated 64 bit non-zero

Change-Id: I783d932320db11a95c7bf7636a72b6fe2566904c

5 years agoapp_queue: fix Calculate talktime when is first call answered
Rodrigo Ramírez Norambuena [Wed, 17 Feb 2016 05:37:43 +0000 (02:37 -0300)]
app_queue: fix Calculate talktime when is first call answered

Fix calculate of average time for talktime is wrong when is completed the
first call beacuse the time for talked would be that call.

ASTERISK-25800 #close

Change-Id: I94f79028935913cd9174b090b52bb300b91b9492

5 years agores_odbc: Fix for missing symbols
George Joseph [Tue, 16 Feb 2016 22:37:48 +0000 (15:37 -0700)]
res_odbc: Fix for missing symbols was missing a few symbols.
Changed to wildcards.

Change-Id: Ieadd76df24e43ea92577f651d478a0f7b742c30c

5 years agores_statsd: Fix for missing symbols
George Joseph [Tue, 16 Feb 2016 18:20:57 +0000 (11:20 -0700)]
res_statsd:  Fix for missing symbols was missing the _va variations of the log
functions causing Asterisk to crash in res_pjsip if OPTIONAL_API
wasn't enabled.

ASTERISK-25727 #close
Reported-by: Gergely Dömsödi

Change-Id: I395729f9f51bdd33c5ca757f5f96ebedad74077b

5 years agores_pjsip_caller_id: Fix segfault when replacing rpid or pai header
George Joseph [Mon, 15 Feb 2016 21:37:30 +0000 (14:37 -0700)]
res_pjsip_caller_id: Fix segfault when replacing rpid or pai header

If the PJSIP_HEADER dialplan function adds a PAI or RPID header and send_rpid
or send_pai is set, res_pjsip_caller_id attemps to retrieve, parse and modify
the header added by the dialplan function.  Since the header added by the
dialplan function is generic string, there are no virtual functions to parse
the uri and we get a segfault when we try.  Since the modify, was really only
an overwrite, we now just delete the old header if it was type PJSIP_H_OTHER
and recreate it.

This raises a question for another time though:  What should happen with
duplicate headers?  Right now res_pjsip_header_funcs doesn't check for dups
so if it's session supplement is loaded after res_pjsip_caller_id's (or any
other module that adds headers), there'll be dups in the message.

ASTERISK-25337 #close

Change-Id: I5e296b52d30f106b822c0eb27c4c2b0e0f71c7fa

5 years agoMerge "Fix creation race of contact_status structures."
zuul [Mon, 15 Feb 2016 21:39:47 +0000 (15:39 -0600)]
Merge "Fix creation race of contact_status structures."

5 years agoFix creation race of contact_status structures.
Mark Michelson [Mon, 15 Feb 2016 19:08:22 +0000 (13:08 -0600)]
Fix creation race of contact_status structures.

It is possible when processing a SIP REGISTER request to have two
threads end up creating contact_status structures in sorcery.
contact_status is created using a "find or create" function. If two
threads call into this at the same time, each thread will fail to find
an existing contact_status, and so both will end up creating a new
contact status.

During testing, we would see sporadic failures because the
PJSIP_CONTACT() dialplan function would operate on a different
contact_status than what had been updated by res_pjsip/pjsip_options.

The fix here is two-fold:
1) The "find or create" function for contact_status now has a lock
around the entire operation. This way, if two threads attempt the
operation simultaneously, the first to get there will create the object,
and the second will find the object created by the first thread.

2) res_sorcery_memory has had its create callback updated so that it
will not allow for objects with duplicate IDs to be created.

Change-Id: I55b1460ff1eb0af0a3697b82d7c2bac9f6af5b97

5 years agores_pjsip_pubsub: Move where the subscription is stored to after initialized.
Joshua Colp [Mon, 15 Feb 2016 18:52:22 +0000 (14:52 -0400)]
res_pjsip_pubsub: Move where the subscription is stored to after initialized.

A problem arose when testing the AMI subscription listing actions where it
was possible for a subscription that had not been fully initialized to be
listed. This was problematic as the underlying listing code would crash.

This change makes it so the subscription tree is fully set up before it is
added to the list of subscriptions. This ensures that when the listing actions
get the subscription it is valid.

ASTERISK-25738 #close

Change-Id: Iace2b13641c31bbcc0d43a39f99aba1f340c0f48

5 years agoMerge "res_pjsip: Refactor load_module/unload_module"
zuul [Fri, 12 Feb 2016 22:50:18 +0000 (16:50 -0600)]
Merge "res_pjsip:  Refactor load_module/unload_module"

5 years agoMerge "res_pjsip: Handle pjsip_dlg_create_uas deprecation"
zuul [Fri, 12 Feb 2016 22:50:13 +0000 (16:50 -0600)]
Merge "res_pjsip:  Handle pjsip_dlg_create_uas deprecation"

5 years agores_pjsip: Refactor load_module/unload_module
George Joseph [Tue, 9 Feb 2016 23:34:05 +0000 (16:34 -0700)]
res_pjsip:  Refactor load_module/unload_module

load_module was just too hairy with every step having to clean up all
previous steps on failure.

Some of the pjproject init calls have now been moved to a separate
load_pjsip function and the unload_pjsip function was enhanced to clean
up everything if an error happened at any stage of the load process.

In the process, a bunch of missing pj_shutdowns, serializer_pool_shutdowns
and ast_threadpool_shutdowns were also corrected.

Change-Id: I5eec711b437c35b56605ed99537ebbb30463b302

5 years agoMerge "Resources/res_phoneprov: fix memory leak and heap-use-after-free"
zuul [Thu, 11 Feb 2016 23:04:50 +0000 (17:04 -0600)]
Merge "Resources/res_phoneprov: fix memory leak and heap-use-after-free"

5 years agoResources/res_phoneprov: fix memory leak and heap-use-after-free
Badalyan Vyacheslav [Wed, 10 Feb 2016 04:42:11 +0000 (04:42 +0000)]
Resources/res_phoneprov: fix memory leak and heap-use-after-free

* heap-use-after-free happens when we free "cfg"
but then use "value" which refers to it

* A memory leak occurs because in some cases
it is not released "defaults"

ASTERISK-25721 #close
Reported by: Badalyan Vyacheslav
Tested by: Badalyan Vyacheslav

Change-Id: I3807d3f4726df6864430ec144cf6265d3f538469

5 years agofunc_iconv: Ensure output strings are properly terminated.
Sean Bright [Thu, 11 Feb 2016 17:21:42 +0000 (12:21 -0500)]
func_iconv: Ensure output strings are properly terminated.

ASTERISK-25272 #close
Reported by: Etienne Lessard
 AST-25272.patch submitted by Etienne Lessard (license #6394)

Change-Id: Id75ad202300960a1e91afe15e319d992936ecc17

5 years agoMerge "res_pjsip: Fix infinite recursion when loading transports from realtime"
Joshua Colp [Thu, 11 Feb 2016 12:10:06 +0000 (06:10 -0600)]
Merge "res_pjsip:  Fix infinite recursion when loading transports from realtime"

5 years agores_pjsip: Handle pjsip_dlg_create_uas deprecation
George Joseph [Wed, 10 Feb 2016 22:16:46 +0000 (15:16 -0700)]
res_pjsip:  Handle pjsip_dlg_create_uas deprecation

Pjproject has deprecated pjsip_dlg_create_uas in 2.5 and replaced it with
pjsip_dlg_create_uas_and_inc_lock which, as the name implies, automatically
increments the lock on the returned dialog.  To account for this,
now detects the presence of pjsip_dlg_create_uas_and_inc_lock and res_pjsip.c
has an #ifdef HAVE_PJSIP_DLG_CREATE_UAS_AND_INC_LOCK to decide whether to use
the original call or the new one.  If the new one was used, the ref count is
decremented before returning.

ASTERISK-25751 #close
Reported-by Josh Colp

Change-Id: I1be776b94761df03bd0693bc7795a75682615ca8

5 years agoMerge "res_config_pgsql: Show error message in reload if not connected."
Mark Michelson [Wed, 10 Feb 2016 21:35:01 +0000 (15:35 -0600)]
Merge "res_config_pgsql: Show error message in reload if not connected."

5 years agoMerge "Build: Fix menuselect USAN conflicts"
Joshua Colp [Wed, 10 Feb 2016 20:34:51 +0000 (14:34 -0600)]
Merge "Build: Fix menuselect USAN conflicts"

5 years agores_config_pgsql: Show error message in reload if not connected.
Rodrigo Ramírez Norambuena [Wed, 10 Feb 2016 02:13:07 +0000 (23:13 -0300)]
res_config_pgsql: Show error message in reload if not connected.

Change-Id: I9290115a1aaadb589eb1d02eaeb502eec01b31fa

5 years agoBuild: Added testing compiler to support the system sanitizes
Badalyan Vyacheslav [Wed, 10 Feb 2016 05:40:32 +0000 (05:40 +0000)]
Build: Added testing compiler to support the system sanitizes

In older versions of the compiler was not sanitizes.
Compilers other than GCC can not support the Usan and TSAN
or have other options for *FLAGS.

ASTERISK-25767 #close
Reported by: Badalyan Vyacheslav
Tested by: Badalyan Vyacheslav

Change-Id: Iefce6608221fa87884b82ae3cb5649b7b1804916

5 years agoBuild: Fix menuselect USAN conflicts
Badalyan Vyacheslav [Wed, 10 Feb 2016 02:57:38 +0000 (02:57 +0000)]
Build: Fix menuselect USAN conflicts

USAN can be used together with other sanitizers.

Reported by: Badalyan Vyacheslav
Tested by: Badalyan Vyacheslav

Change-Id: I3bffa350d70965c3026651dba3a12414d0aaa45f

5 years agoSimplify and fix conditional in FD_SET.
Corey Farrell [Tue, 9 Feb 2016 20:21:05 +0000 (15:21 -0500)]
Simplify and fix conditional in FD_SET.

FD_SET contains a conditional statement to protect against buffer
overruns.  The statement was overly complicated and prevented use
of the last array element of ast_fdset.  We now just verify the fd
is less than ast_FDMAX.

Change-Id: I41895c0b497b052aef5bf49d75c817c48b326f40

5 years agoMerge "res_config_pgsql: Add message on cli failed command status"
Joshua Colp [Tue, 9 Feb 2016 19:45:32 +0000 (13:45 -0600)]
Merge "res_config_pgsql: Add message on cli failed command status"

5 years agotests/test_sorcery_memory_cache_thrash: Improve termination process.
Joshua Colp [Tue, 9 Feb 2016 13:11:36 +0000 (09:11 -0400)]
tests/test_sorcery_memory_cache_thrash: Improve termination process.

When terminating the threads thrashing a sorcery memory cache each
would be told to stop and then we would wait on them. During at
least one thrashing test this was problematic due to the specific
usage pattern in use. It would take some time for termination of the
thread to occur.

This would occur due to contention between the threads retrieving
and the threads updating the cache. As the retrieving threads are
given priority it may be some time before the updating threads
are able to proceed.

This change makes it so all threads are told to stop and then each
are joined to ensure they stop. This way all the threads should
stop at around the same time instead of waiting for one to stop,
the next to stop, then the next, and so on. As a result of this
the execution time for each thrash test is much closer to their
expected value than previously seen as well.

Change-Id: I04a53470b0ea4170b8819180b0bd7475f3642827

5 years agores_pjsip: Fix infinite recursion when loading transports from realtime
George Joseph [Fri, 29 Jan 2016 23:56:42 +0000 (16:56 -0700)]
res_pjsip:  Fix infinite recursion when loading transports from realtime

Attempting to load a transport from realtime was forcing asterisk into an
infinite recursion loop.  The first thing transport_apply did was to do a
sorcery retrieve by id for an existing transport of the same name. For files,
this just returns the previous object from res_sorcery_config's internal
container, if any.  For realtime, the res_sourcery_realtime driver looks in the
database and finds the existing row but now it has to rehydrate it into a
sorcery object which means calling... transport_apply.  And so it goes.

The main issue with loading from realtime (apart from the loop) was that
transport stores structures and pointers directly in the ast_sip_transport
structure instead of the separate ast_transport_state structure.  This patch
separates those items into the ast_sip_transport_state structure.  The pattern
is roughly the same as res_pjsip_outbound_registration.

Although all current usages of ast_sip_transport and ast_sip_transport_state
were modified to use the new ast_sip_get_transport_state API, the original
items are left in ast_sip_transport and kept updated to maintain ABI
compatability for third-party modules.  They are marked as deprecated and
noted that they're now in ast_sip_transport_state.

ASTERISK-25606 #close
Reported-by: Martin Moučka

Change-Id: Ic7a836ea8e786e8def51fe3f8cce855ea54f5f19

5 years agoMerge "chan_misdn: Fix a few issues causing compile errors"
Joshua Colp [Mon, 8 Feb 2016 11:56:52 +0000 (05:56 -0600)]
Merge "chan_misdn: Fix a few issues causing compile errors"

5 years agores_config_pgsql: Add message on cli failed command status
Rodrigo Ramírez Norambuena [Sun, 7 Feb 2016 19:00:24 +0000 (16:00 -0300)]
res_config_pgsql: Add message on cli failed command status

In case failed of command "realtime show pgsql status" show a message the data
of connection to more clear information in error.

Change-Id: Ia8e9e2400466606e7118f52a46e05df0719b6a29

5 years agochan_misdn: Fix a few issues causing compile errors
George Joseph [Fri, 5 Feb 2016 16:29:00 +0000 (09:29 -0700)]
chan_misdn: Fix a few issues causing compile errors

Change-Id: I54b48c24d7ca88ed80496fdfd142d08772a7ab98

5 years agoapp_confbridge: Only use b_profile options from the conference.
Richard Mudgett [Mon, 25 Jan 2016 23:36:50 +0000 (17:36 -0600)]
app_confbridge: Only use b_profile options from the conference.

A user cannot set new bridge options after the conference is created by
the first user.  Attempting to do so is documented as undefined behavior.

This patch ensures that the bridge profile options used are from the
conference and not what a subsequent user may have tried to set.

Change-Id: I1b6383eba654679e5739d5a8de98199cf074a266

5 years agoMerge "pjsip/alembic: Add missing columns to system and registration"
Joshua Colp [Fri, 5 Feb 2016 17:50:35 +0000 (11:50 -0600)]
Merge "pjsip/alembic:  Add missing columns to system and registration"

5 years agoMerge "app_confbridge.c: Replace inlined code with existing function."
Joshua Colp [Fri, 5 Feb 2016 17:49:42 +0000 (11:49 -0600)]
Merge "app_confbridge.c: Replace inlined code with existing function."

5 years agoMerge topic 'ASTERISK-20987'
Joshua Colp [Fri, 5 Feb 2016 17:49:14 +0000 (11:49 -0600)]
Merge topic 'ASTERISK-20987'

* changes:
  app_confbridge: Add ability to get the muted conference state.
  app_confbridge.c: Update CONFBRIDGE and CONFBRIDGE_INFO documentation.
  app_confbridge: Make non-admin users join a muted conference muted.

5 years agoCheck for OpenSSL defines before trying to use them.
Mark Michelson [Thu, 4 Feb 2016 22:17:55 +0000 (16:17 -0600)]
Check for OpenSSL defines before trying to use them.

The SSL_OP_NO_TLSv1_1 and SSL_OP_NO_TLSv1_2 defines did not exist prior
to OpenSSL version 1.0.1. A recent commit attempts to, by default, set
these options, which can cause problems on systems with older OpenSSL

This commit adds a configure script check for those defines and will not
attempt to make use of those if they do not exist. We will print a
warning urging the user to upgrade their OpenSSL installation if those
defines are not present.

Change-Id: I6a2eb9a43fd0738b404d8f6f2cf4b5c22d9d752d

5 years agopjsip/alembic: Add missing columns to system and registration
George Joseph [Wed, 3 Feb 2016 20:25:23 +0000 (13:25 -0700)]
pjsip/alembic:  Add missing columns to system and registration

ps_systems needed disable_tcp_switch
ps_registrations needed line and endpoint

ASTERISK-25737 #close

Change-Id: Iaf9c2d69e62243d9fa53104c28c5339c47d4ac19

5 years agoMerge "logging: Remove/fix some message annoyances"
Mark Michelson [Thu, 4 Feb 2016 20:10:58 +0000 (14:10 -0600)]
Merge "logging: Remove/fix some message annoyances"

5 years agoMerge "res_stasis_device_state: Fix refcounting error."
Joshua Colp [Thu, 4 Feb 2016 18:35:37 +0000 (12:35 -0600)]
Merge "res_stasis_device_state: Fix refcounting error."

5 years agoMerge "app_queue: Add Lastpause field of queue member"
Joshua Colp [Thu, 4 Feb 2016 18:29:18 +0000 (12:29 -0600)]
Merge "app_queue: Add  Lastpause field of queue member"

5 years agoMerge "res_xmpp: Does not connect in component mode"
Joshua Colp [Thu, 4 Feb 2016 18:26:49 +0000 (12:26 -0600)]
Merge "res_xmpp: Does not connect in component mode"

5 years agores_stasis_device_state: Fix refcounting error.
Mark Michelson [Thu, 4 Feb 2016 17:39:10 +0000 (11:39 -0600)]
res_stasis_device_state: Fix refcounting error.

Device state subscription lifetimes were governed by when the
subscription was established and unsubscribed from. However, it is
possible that at the time of unsubscription, there could be device state
events still in flight. When those device state events occur, the device
state callback could attempt to dereference a freed pointer. Crash.

This change ensures that the lifetime of the device state subscription
does not end until the underlying stasis subscription has confirmed that
its final message has been sent.

Change-Id: I25a0f1472894c1a562252fb7129671478e25e9b2

5 years agores_rtp_asterisk: Allow ICE host candidates to be overriden
Sean Bright [Wed, 27 Jan 2016 16:44:10 +0000 (11:44 -0500)]
res_rtp_asterisk: Allow ICE host candidates to be overriden

During ICE negotiation the IPs of the local interfaces are sent to the remote
peer as host candidates. In many cases Asterisk is behind a static one-to-one
NAT, so these host addresses will be internal IP addresses.

To help in hiding the topology of the internal network, this patch adds the
ability to override the host candidates by matching them against a
user-defined list of replacements.

Change-Id: I1c9541af97b83a4c690c8150d19bf7202c8bff1f

5 years agoMerge "AST-2016-003 udptl.c: Fix uninitialized values."
Kevin Harwell [Wed, 3 Feb 2016 21:17:22 +0000 (15:17 -0600)]
Merge "AST-2016-003 udptl.c: Fix uninitialized values."

5 years agoMerge "AST-2016-002 chan_sip.c: Fix retransmission timeout integer overflow."
Kevin Harwell [Wed, 3 Feb 2016 21:14:53 +0000 (15:14 -0600)]
Merge "AST-2016-002 chan_sip.c: Fix retransmission timeout integer overflow."

5 years agoAST-2016-001 http: Provide greater control of TLS and set modern defaults.
Joshua Colp [Wed, 3 Feb 2016 18:05:20 +0000 (14:05 -0400)]
AST-2016-001 http: Provide greater control of TLS and set modern defaults.

This change exposes the configuration of various aspects of the TLS
support and sets the default to the modern standards.

The TLS cipher is now set to the best values according to the
Mozilla OpSec team, different TLS versions can now be disabled, and
the cipher order can be forced to be that of the server instead of
the client.

ASTERISK-24972 #close

Change-Id: I0a10f2883f7559af5e48dee0901251dbf30d45b8

5 years agoAST-2016-003 udptl.c: Fix uninitialized values.
Richard Mudgett [Mon, 7 Dec 2015 18:46:53 +0000 (12:46 -0600)]
AST-2016-003 udptl.c: Fix uninitialized values.

Sending UDPTL packets to Asterisk with the right amount of missing
sequence numbers and enough redundant 0-length IFP packets, can make
Asterisk crash.

ASTERISK-25603 #close
Reported by: Walter Doekes

ASTERISK-25742 #close
Reported by: Torrey Searle

Change-Id: I97df8375041be986f3f266ac1946a538023a5255

5 years agoAST-2016-002 chan_sip.c: Fix retransmission timeout integer overflow.
Richard Mudgett [Mon, 28 Sep 2015 22:07:42 +0000 (17:07 -0500)]
AST-2016-002 chan_sip.c: Fix retransmission timeout integer overflow.

Setting the sip.conf timert1 value to a value higher than 1245 can cause
an integer overflow and result in large retransmit timeout times.  These
large timeout times hold system file descriptors hostage and can cause the
system to run out of file descriptors.

NOTE: The default sip.conf timert1 value is 500 which does not expose the

* The overflow is now detected and the previous timeout time is

ASTERISK-25397 #close
Reported by: Alexander Traud

Change-Id: Ia7231f2f415af1cbf90b923e001b9219cff46290

5 years agologging: Remove/fix some message annoyances
George Joseph [Wed, 3 Feb 2016 20:07:07 +0000 (13:07 -0700)]
logging: Remove/fix some message annoyances

test_dlinklists doesn't need to NOTICE everyone that every macro worked.

res_phoneprov doesn't need to VERBOSE everyone that a phoneprov extension or
provider was registered.

res_odbc was missing a newline at the end of one message.

Change-Id: I6c06361518ef3711821795e535acd439782a995e

5 years agoMerge "res_sorcery_realtime: Fix regex regression."
Joshua Colp [Wed, 3 Feb 2016 16:14:45 +0000 (10:14 -0600)]
Merge "res_sorcery_realtime: Fix regex regression."

5 years agoMerge " REFACTOR Macro LENGTHEN_BUF"
Joshua Colp [Wed, 3 Feb 2016 12:20:18 +0000 (06:20 -0600)]

5 years agoMerge "app_queue: fix some tab format"
Joshua Colp [Wed, 3 Feb 2016 12:19:47 +0000 (06:19 -0600)]
Merge "app_queue: fix some tab format"

5 years agoMerge "README: Update year in copyright"
Joshua Colp [Wed, 3 Feb 2016 12:19:29 +0000 (06:19 -0600)]
Merge "README: Update year in copyright"

5 years agoMerge "app_queue: Fix preserved reason of pause when Asterisk is restared"
Joshua Colp [Wed, 3 Feb 2016 12:19:19 +0000 (06:19 -0600)]
Merge "app_queue: Fix preserved reason of pause when Asterisk is restared"

5 years agoMerge "app_queue.c: remove include for core_unreal.h not used in code."
Joshua Colp [Wed, 3 Feb 2016 12:18:58 +0000 (06:18 -0600)]
Merge "app_queue.c: remove include for core_unreal.h not used in code."

5 years agoMerge "chan_sip.c: AMI & CLI notify methods get different values of asterisk's own...
Mark Michelson [Tue, 2 Feb 2016 21:58:49 +0000 (15:58 -0600)]
Merge "chan_sip.c: AMI & CLI notify methods get different values of asterisk's own ip."

5 years agores_sorcery_realtime: Fix regex regression.
Mark Michelson [Tue, 2 Feb 2016 16:52:29 +0000 (10:52 -0600)]
res_sorcery_realtime: Fix regex regression.

A regression was introduced where searching for realtime PJSIP objects
by regex by starting the regex with a leading "^" would cause no items
to be returned.

This was due to a change which attempted to drop the requirement for a
leading "^" to be present due to how some CLI commands formulate their
regexes. However, the change, rather than simply eliminating the
requirement, caused any regexes that did begin with "^" to end up not
returning the expected results.

This change fixes the problem by inspecting the regex and formulating
the realtime query differently depending on if it begins with "^".

ASTERISK-25702 #close
Reported by Nic Colledge

    realtime_retrieve_regex.patch submitted by Alexei Gradinari License #5691

Change-Id: I055df608a6e6a10732044fa737a9fe8dca602693

5 years agores_xmpp: Does not connect in component mode
Karsten Wemheuer [Tue, 2 Feb 2016 10:05:15 +0000 (11:05 +0100)]
res_xmpp: Does not connect in component mode

The module res_xmpp does not accept usernames in the form used in component
mode (XEP-0114). In component mode there is no @something in the name.
In component mode the connection is now not dropped anymore.

If the xmpp server sends out a "stream" tag before handshake is finished,
the connection gets dropped in res_xmpp. Now this tag will be ignored and
the connection will be established.

After connecting there will be an exchange of presence states. This does
not work as expected in component mode. The responsible function
"xmpp_pak_presence" is left before the states get sent out. Sending
presence states in component mode is now moved to the top of the function.

ASTERISK-25735 #close

Change-Id: I70e036f931c3124ebb2ad1e56f93ed35cfdd9d5c

5 years agoMerge "res_odbc: Remove connection management"
Joshua Colp [Tue, 2 Feb 2016 12:46:41 +0000 (06:46 -0600)]
Merge "res_odbc: Remove connection management"

5 years agobuild_system: Fix some warnings highlighted by clang
George Joseph [Mon, 1 Feb 2016 19:04:06 +0000 (12:04 -0700)]
build_system:  Fix some warnings highlighted by clang

Fix some warnings found with clang.

Change-Id: I5195b6189b148c2ee3ed4a19d015a6d4ef3e77bd

5 years agopjsip/alembic: Fix definition of qualify_timeout
George Joseph [Mon, 1 Feb 2016 02:13:58 +0000 (19:13 -0700)]
pjsip/alembic: Fix definition of qualify_timeout

A recent commit set qualify_timeout to Decimal which isn't supported.
This path corrects it to Float.

Change-Id: I038f5274ba8cb60f8518a5845ce448d49306aadf

5 years agochan_sip.c: AMI & CLI notify methods get different values of asterisk's own ip.
StefanEng86 [Fri, 29 Jan 2016 13:39:06 +0000 (14:39 +0100)]
chan_sip.c: AMI & CLI notify methods get different values of asterisk's own ip.

When I ask asterisk to send a SIP NOTIFY message to a sip peer using either a)
AMI action: SIPnotify or b) cli command: sip notify <cmd> <peer>, I expect
asterisk to include the same value for its own ip in both cases a) and b),
but it seems a) produces a contact header like Contact:
<sip:asterisk@> whereas b) produces a contact header like
<sip:asterisk@>. is my udpbindaddr in sip.conf

My guess is that manager_sipnotify should call
ast_sip_ouraddrfor(&p->sa, &p->ourip, p) the same way sip_cli_notify does,
because after applying this patch, both cases a) and b) produce
the contact header that I expect: <sip:asterisk@>

Reported by: Stefan Engström
Tested by: Stefan Engström

Change-Id: I86af5e209db64aab82c25417de6c768fb645f476

5 years agoMerge "build_system: Prevent goals needing makeopts from running when it's missing"
Joshua Colp [Fri, 29 Jan 2016 14:06:22 +0000 (08:06 -0600)]
Merge "build_system: Prevent goals needing makeopts from running when it's missing"

5 years agoMerge "config: Allow options to register when documentation is unavailable."
Mark Michelson [Thu, 28 Jan 2016 21:56:30 +0000 (15:56 -0600)]
Merge "config: Allow options to register when documentation is unavailable."

5 years agoconfig_options.c: Fix warning message wording.
Richard Mudgett [Thu, 28 Jan 2016 18:44:43 +0000 (12:44 -0600)]
config_options.c: Fix warning message wording.

Change-Id: I915ea437936320393afde0e7552cf0a980a6b2e4

5 years agoapp_confbridge.c: Replace inlined code with existing function.
Richard Mudgett [Mon, 25 Jan 2016 23:34:20 +0000 (17:34 -0600)]
app_confbridge.c: Replace inlined code with existing function.

Change-Id: Ida5594e9f8d7c1fc18eeb733a11f8fb96326da51

5 years agoapp_confbridge: Add ability to get the muted conference state.
Richard Mudgett [Mon, 25 Jan 2016 22:05:09 +0000 (16:05 -0600)]
app_confbridge: Add ability to get the muted conference state.

* Added CONFBRIDGE_INFO(muted,) for querying the muted conference state.

* Added Muted header to AMI ConfbridgeListRooms action response list
events to indicate the muted conference state.

* Added Muted column to CLI "confbridge list" output to indicate the muted
conference state and made the locked column a yes/no value instead of a
locked/unlocked value.

Reported by: hristo

Change-Id: I4076bd8ea1c23a3afd4f5833e9291b49a0c448b1

5 years agoapp_confbridge.c: Update CONFBRIDGE and CONFBRIDGE_INFO documentation.
Richard Mudgett [Tue, 26 Jan 2016 23:59:28 +0000 (17:59 -0600)]
app_confbridge.c: Update CONFBRIDGE and CONFBRIDGE_INFO documentation.

Change-Id: Ic1f9e22ba1f2ff3b3f5cb017c5ddcd9bd48eccc7

5 years agoapp_confbridge: Make non-admin users join a muted conference muted.
Richard Mudgett [Mon, 25 Jan 2016 21:48:04 +0000 (15:48 -0600)]
app_confbridge: Make non-admin users join a muted conference muted.

ASTERISK-20987 #close
Reported by: hristo

Change-Id: Ic61a2b524ab3a4cfadf227fc6b3506527bc03f38

5 years agores_pjsip: Add res_pjproject dependency to samples
George Joseph [Wed, 27 Jan 2016 19:08:29 +0000 (12:08 -0700)]
res_pjsip:  Add res_pjproject dependency to samples

Since res_pjsip now depends on res_pjproject, this has been added to
basic-pbx modules.conf.

Change-Id: I42826597d5e10f08e518208860c44c96e52f1b2d

5 years agobuild_system: Prevent goals needing makeopts from running when it's missing
George Joseph [Wed, 27 Jan 2016 16:29:13 +0000 (09:29 -0700)]
build_system: Prevent goals needing makeopts from running when it's missing

The Makefile only optionally includes makeopts so when goals like uninstall that
dont depend on anything else are run after a distclean, rules like
'rm -f "$(DESTDIR)$(ASTMODDIR)/"*' get run as 'rm -f ""/*' which attempts
to remove everything in the root directory.

Although there's a rule defined for makeopts which prints a message and does
an 'exit 1', since '-include makepopts' was specified (with the -), the exit
was ignored letting the rest of the rules run.

This patch makes makeopts required unless the goal has the string 'clean' in it.

ASTERISK-25730 #close
Reported-by: George Joseph

Change-Id: I1bce59a7ea4f48e7a468e22b2abbb13c63417ac7

5 years agoconfig: Allow options to register when documentation is unavailable.
Joshua Colp [Mon, 25 Jan 2016 15:35:21 +0000 (11:35 -0400)]
config: Allow options to register when documentation is unavailable.

The config options framework is strict in that configuration options must
be documented unless XML documentation support is not available. In
practice this is useful as it ensures documentation exists however in
off-nominal cases this can cause strange problems.

If it is expected that a config option has a non-zero or non-empty
default value but the config option documentation is unavailable
this reasonable expectation will not be met. This can cause obscure
crashes and weirdness depending on how the code handles it.

This change tweaks the behavior to ensure that the config option
is still allowed to register, apply default values, and be set when
devmode is not enabled. If devmode is enabled then the option can
NOT be set.

This also does not remove the initial documentation error message that
is output on load when registering the configuration option.

ASTERISK-25725 #close

Change-Id: Iec42fca6b35f31326c33fcdc25473f6fd7bc8af8

5 years agoStasis: Use custom structure when setting variables.
Mark Michelson [Mon, 25 Jan 2016 16:23:18 +0000 (10:23 -0600)]
Stasis: Use custom structure when setting variables.

A recent change to queue channel variable setting to the Stasis control
queue caused a regression. When setting channel variables, it is
possible to give a NULL channel variable value in order to unset the
variable (i.e. remove it from the channel variable list). The change
introduced a call to ast_variable_new(), which is not tolerant of NULL
channel variable values.

This new change switches from using ast_variable to using a custom
channel variable struct that is lighter weight and NULL value-tolerant.

Change-Id: I784d7beaaa3c036ea936d103e7caf0bb1562162d

5 years agoMerge "res_pjsip_pubsub: Prevent crash from AMI command on freed subscription."
Matt Jordan [Tue, 26 Jan 2016 13:05:52 +0000 (07:05 -0600)]
Merge "res_pjsip_pubsub: Prevent crash from AMI command on freed subscription."

5 years agoMerge "sounds/Makefile: Incremented core and extra sounds versions to 1.5"
Matt Jordan [Tue, 26 Jan 2016 13:05:09 +0000 (07:05 -0600)]
Merge "sounds/Makefile: Incremented core and extra sounds versions to 1.5"

5 years agosounds/Makefile: Incremented core and extra sounds versions to 1.5
Rusty Newton [Mon, 25 Jan 2016 22:56:04 +0000 (16:56 -0600)]
sounds/Makefile: Incremented core and extra sounds versions to 1.5

Core and extra sounds 1.5 was recently released! The tarballs contain
change descriptions however I figure more people will see this one so
I'll try to be a bit detailed. Approximately 60 sounds were moved from Extra
to Core for en, en_GB, fr and added for languages that didn't already
have Extra sound sets (it,ja,ru).

In addition all of the English and Russian sounds have been completely

Sounds moved and added:

There were also a few random fixes here and there to file names for a few
of the languages.

ASTERISK-25068 #close

Change-Id: I2b594344ec585d7dfd922b40c1af43b1508828b3

5 years agores_pjsip_pubsub: Prevent crash from AMI command on freed subscription.
Mark Michelson [Mon, 25 Jan 2016 22:51:25 +0000 (16:51 -0600)]
res_pjsip_pubsub: Prevent crash from AMI command on freed subscription.

A test recently uncovered that running an ill-timed AMI command to show
inbound subscriptions could cause a crash since Asterisk will try to
operate on a freed subscription.

The fix for this is to remove the subscription tree from the list of
subscriptions at the time that we are sending our final NOTIFY request
out. This way, as the subscription is in the process of dying, it is
inaccessible from AMI.

Change-Id: Ic0239003d8d73e04c47c12dd2a7e23867e5b5b23

5 years agochan_sip: Fix buffer overrun in sip_sipredirect.
Corey Farrell [Mon, 25 Jan 2016 17:03:21 +0000 (12:03 -0500)]
chan_sip: Fix buffer overrun in sip_sipredirect.

sip_sipredirect uses sscanf to copy up to 256 characters to a stacked buffer
of 256 characters.  This patch reduces the copy to 255 characters to leave
room for the string null terminator.

ASTERISK-25722 #close

Change-Id: Id6c3a629a609e94153287512c59aa1923e8a03ab

5 years agoapp_queue: Add Lastpause field of queue member
Rodrigo Ramírez Norambuena [Sat, 23 Jan 2016 22:45:30 +0000 (19:45 -0300)]
app_queue: Add  Lastpause field of queue member

Add time when started a the last pause for a queue member for
QueueMemberStatus ami event.

Also show accumulate time in seconds when started a pause for a queue
member to CLI command 'queue show'.

ASTERISK-16394 #close

Change-Id: I4b12aa3b2efa8d02939db3e13712510b4879865c

5 years agoapp_queue: fix some tab format
Rodrigo Ramírez Norambuena [Sat, 23 Jan 2016 18:34:11 +0000 (15:34 -0300)]
app_queue: fix some tab format

Change-Id: I2734392b131f1fb0949515d538f83f30fbc15d8c

Rodrigo Ramírez Norambuena [Sat, 23 Jan 2016 17:41:38 +0000 (14:41 -0300)] REFACTOR Macro LENGTHEN_BUF

Remove repeated code on macro of assigned buffer to SQL vars.

Add table and connection name to log error message when is not possible
allocate memory.

Change-Id: I1fbf37d286a032d38fdda72a9f736356956c9ffe

5 years agoMerge "Stasis: Use control queue to prevent crash."
Joshua Colp [Sat, 23 Jan 2016 16:07:52 +0000 (10:07 -0600)]
Merge "Stasis: Use control queue to prevent crash."

5 years agoStasis: Fix potential memory leak of control data.
Mark Michelson [Fri, 22 Jan 2016 21:08:58 +0000 (15:08 -0600)]
Stasis: Fix potential memory leak of control data.

When queuing tasks onto the Stasis control queue, you can pass an
arbitrary data pointer and a function to free that data. All ARI
commands that use the Stasis control queue made the assumption that the
destructor function would be called in all paths, whether the task was
queued successfully or not. However, this was not correct. If a task was
queued onto a control structure that was already completed, the
allocated data would not be freed properly.

This patch corrects this by making sure that all return paths call the
data destructor.

Change-Id: Ibf06522094f8e5c4cce652537dc5d7222b1c4fcb

5 years agoStasis: Use control queue to prevent crash.
Mark Michelson [Thu, 21 Jan 2016 16:58:02 +0000 (10:58 -0600)]
Stasis: Use control queue to prevent crash.

A crash occurred when attempting to set a channel variable on a channel
that had already been hung up. This is because there is a small window
between when a control is grabbed and when the channel variable is set
that the channel can be hung up.

The fix here is to queue the setting of the channel variable onto the
control queue. This way, the manipulation of the channel happens in a
thread where it is safe to be done.

In this change, I also noticed that the setting of bridge roles on
channels was being done outside of the control queue, so I also changed
those operations to be done in the control queue.

ASTERISK-25709 #close
Reported by Mark Michelson

Change-Id: I2a0a4d51bce6fba6f1d9954e40935e42f366ea78

5 years agologger.c: Fix buffer overrun found by address sanitizer.
Richard Mudgett [Fri, 22 Jan 2016 17:48:24 +0000 (11:48 -0600)]
logger.c: Fix buffer overrun found by address sanitizer.

The null terminator of the tail struct member was not being allocated
when no logger.conf config file is installed.

ASTERISK-25714 #close
Reported by: Badalian Vyacheslav

Change-Id: I45770fdd08af39506a3bc33ba279c4f16e047a30

5 years agores_odbc: Remove connection management
Mark Michelson [Wed, 23 Dec 2015 21:07:05 +0000 (15:07 -0600)]
res_odbc: Remove connection management

Asterisk by default will create a single database connection and share
it among all threads that attempt to access the database. In previous
versions of Asterisk, this was tolerable, because the most used channel
driver, chan_sip, mostly accessed the database from a single thread.
With PJSIP, however, many threads may be attempting to perform database
operations, and there is the potential for many more database accesses,
meaning the concurrency is a horrible bottleneck if only one connection
is shared.

Asterisk has a connection pooling facility built into it, but the
implementation has flaws. For one, there is a strict limit on the number
of simultaneous connections that could be made to the database. Anything
beyond the maximum would result in a failed operation. Attempting to
predict what the maximum should be is nearly impossible even for someone
intimately familiar with Asterisk's threading model. In addition, use of
transactions in the dialplan can cause some severe bugs if connection
pooling is enabled.

This commit seeks to fix the concurrency problem by removing all
connection management code from Asterisk and leaving that to the
underlying unixODBC code instead. Now, Asterisk does not share a single
connection, nor does it try to maintain a connection pool. Instead, all
Asterisk ever does is request a connection from unixODBC and allow
unixODBC to either allocate those connections or retrieve them from a

Doing this has a bit of a ripple effect. For one, since connections are
not long-lived objects, several of the safeguards that previously
existed have been removed. We don't have to worry about trying to use a
connection that has gone stale. In every case, when we request a
connection, it has just been made and we don't need to perform any
sanity checks to be sure it's still active.

Another major player affected by this change is transactions.
Transactions and their respective connections were so tightly coupled
that it was almost pornographic. This code change moves
transaction-related code to its own file separate from the core ODBC
functionality. This way, the core of ODBC does not even have to know
that transactions exist.

In making this large change, I had to look at a lot of code and
understand it. When making this change, I discovered several places
where the behavior is definitely not ideal, but it seemed outside the
scope of this change to be fixing it. Instead, any place where I saw
some sort of room for improvement has had a XXX comment added explaining
what could be altered to improve it.

Change-Id: I37a84def5ea4ddf93868ce8105f39de078297fbf

5 years agoapp_queue.c: remove include for core_unreal.h not used in code.
Rodrigo Ramírez Norambuena [Fri, 22 Jan 2016 17:18:57 +0000 (14:18 -0300)]
app_queue.c: remove include for core_unreal.h not used in code.

Change-Id: Idc2ae8a6bd869a66544916906744a5678622262d

5 years agoMerge "Build System: Add support for checking alembic branches."
Matt Jordan [Fri, 22 Jan 2016 02:57:02 +0000 (20:57 -0600)]
Merge "Build System: Add support for checking alembic branches."

5 years agoMerge "res/res_pjsip/presence_xml.c: Add missing 2nd call presence state case."
Matt Jordan [Thu, 21 Jan 2016 23:25:26 +0000 (17:25 -0600)]
Merge "res/res_pjsip/presence_xml.c: Add missing 2nd call presence state case."

5 years agoMerge "main/asterisk.c: ast_el_read_char"
Matt Jordan [Thu, 21 Jan 2016 23:25:03 +0000 (17:25 -0600)]
Merge "main/asterisk.c: ast_el_read_char"

5 years agoBuild System: Add support for checking alembic branches.
Corey Farrell [Thu, 21 Jan 2016 22:40:47 +0000 (17:40 -0500)]
Build System: Add support for checking alembic branches.

* Add 'check-alembic' target to root Makefile.
* Create build_tools/make_check_alembic to do the actual checks.


Change-Id: Ibb3cae7d1202ac23dc70b0f3b5801571ad46b004

5 years agores/res_pjsip/presence_xml.c: Add missing 2nd call presence state case.
Richard Mudgett [Wed, 20 Jan 2016 00:20:59 +0000 (18:20 -0600)]
res/res_pjsip/presence_xml.c: Add missing 2nd call presence state case.

ASTERISK-25712 #close
Reported by: Richard Mudgett

Change-Id: I70634df24f8c6c3a2c66c45af61d021e4999253f

5 years agoMerge "chan_sip: option 'notifyringing' change and doc fix"
Mark Michelson [Thu, 21 Jan 2016 21:22:53 +0000 (15:22 -0600)]
Merge "chan_sip: option 'notifyringing' change and doc fix"

5 years agores_pjsip: Add CLI "pjsip dump endpt [details]"
Richard Mudgett [Wed, 13 Jan 2016 22:49:22 +0000 (16:49 -0600)]
res_pjsip: Add CLI "pjsip dump endpt [details]"

Dump the res_pjsip endpt internals.

In non-developer mode we will not document or make easily accessible the
"details" option even though it is still available.  The user has to know
it exists to use it.  Presumably they would also be aware of the potential
crash warning below.

Warning: PJPROJECT documents that the function used by this CLI command
may cause a crash when asking for details because it tries to access all
active memory pools.

Change-Id: If2d98a3641c9873364d1daaad971376311aef3cb

5 years agoMerge "taskprocessor.c: Increase CLI "core ping taskprocessor" timeout."
Mark Michelson [Wed, 20 Jan 2016 20:19:02 +0000 (14:19 -0600)]
Merge "taskprocessor.c: Increase CLI "core ping taskprocessor" timeout."

5 years agoMerge "taskprocessor.c: Fix some taskprocessor unrefs."
Mark Michelson [Wed, 20 Jan 2016 20:18:57 +0000 (14:18 -0600)]
Merge "taskprocessor.c: Fix some taskprocessor unrefs."

5 years agoMerge "res_pjproject: Add module providing pjproject logging and utils"
Joshua Colp [Wed, 20 Jan 2016 17:46:05 +0000 (11:46 -0600)]
Merge "res_pjproject:  Add module providing pjproject logging and utils"

5 years agomain/asterisk.c: ast_el_read_char
Diederik de Groot [Mon, 18 Jan 2016 09:49:48 +0000 (10:49 +0100)]
main/asterisk.c: ast_el_read_char

Make sure buf[res] is not accessed at res=-1 (buffer underrun).
Address Sanitizer will complain about this quite loudly.

ASTERISK-24801 #close

Change-Id: Ifcd7f691310815a31756b76067c56fba299d3ae9

5 years agores_pjproject: Add module providing pjproject logging and utils
George Joseph [Tue, 19 Jan 2016 01:27:57 +0000 (18:27 -0700)]
res_pjproject:  Add module providing pjproject logging and utils

res_pjsip_log_forwarder has been renamed to res_pjproject
and enhanced as follows:

As a follow-on to the recent 'Add CLI "pjsip show buildopts"' patch,
a new ast_pjproject_get_buildopt function has been added.  It
allows the caller to get the value of one of the buildopts.

The initial use case is retrieving the runtime value of
PJ_MAX_HOSTNAME to insure we don't send a hostname greater
than pjproject can handle.  Since it can differ between
the version of pjproject that Asterisk was compiled against
and the version of pjproject that Asterisk is running against,
we can't use the PJ_MAX_HOSTNAME macro directly in Asterisk
source code.

Change-Id: Iab6e82fec3d7cf00c1cf6185c42be3e7569dee1e

5 years agoMerge "funcs/func_cdr: Correctly report high precision values for duration and billsec"
Joshua Colp [Wed, 20 Jan 2016 16:33:27 +0000 (10:33 -0600)]
Merge "funcs/func_cdr: Correctly report high precision values for duration and billsec"

5 years agoMerge "pjsip_loging_refactor: Rename res_pjsip_log_forwarder to res_pjproject"
Joshua Colp [Wed, 20 Jan 2016 16:32:47 +0000 (10:32 -0600)]
Merge "pjsip_loging_refactor: Rename res_pjsip_log_forwarder to res_pjproject"