The patterns provided by pbx_realtime were checked in the order in
which they were returned from the realtime backend. If there was
overlap between multiple patterns, the first one to correctly match was
chosen even though it may not have been the best match.
We now sort the patterns descending by their length and compare in that
order. There may be cases where this still results in a sub-optimal
match, but this patch should improve the overall behavior.
ASTERISK-18271 #close
Reported by: Charlie Smurthwaite
Sean Bright [Tue, 21 Feb 2017 21:09:47 +0000 (16:09 -0500)]
pbx_dundi: DUNDi weight parameter not processed correctly
The DUNDi weight field is not always converted from network byte order
to host byte order. This can result in incorrect weight values and
incorrect selection of DUNDi destinations.
ASTERISK-18731 #close
Reported by: Peter Racz
Patches:
dundi_weight.patch (license #6290) patch uploaded by Peter Racz
Joshua Colp [Tue, 21 Feb 2017 22:34:56 +0000 (16:34 -0600)]
Merge changes from topic 'ASTERISK-26580' into 13
* changes:
res_config_ldap: Don't try to delete non-existent attributes
res_config_ldap: Remove extraneous line numbers from log messages
res_config_ldap: Make memory allocation more consistent
res_config_ldap: Fix configuration inheritance from _general
ast_load_realtime_multientry() returns an ast_config structure whose
ast_categorys are keyed with the empty strings. Several modules were
giving semantic meaning to the category names causing problems at
runtime.
* app_directory: Treated the category name as the mailbox name, and
would fail to direct calls to the appropriate extension after an
entry was chosen.
* app_queue: Queues, queue members, and queue rules were all affected
and needed to be updated.
* pbx_realtime: Pattern matching would never succeed because the
extension entered by the user was always compared to the empty
string.
Sean Bright [Tue, 21 Feb 2017 14:56:54 +0000 (09:56 -0500)]
realtime: Centralize some common realtime backend code
All of the realtime backends create artificial ast_categorys to pass
back into the core as query results. These categories have no filename
or line number information associated with them and the backends differ
slightly on how they create them. So create a couple helper macros to
help make things more consistent.
Also updated the call sites to remove redundant error messages about
memory allocation failure.
Note that res_config_ldap sets the category filename to the 'table name'
but that is not read by anything in the core, so I've dropped it.
Joshua Colp [Thu, 16 Feb 2017 16:30:00 +0000 (16:30 +0000)]
build: Execute ldconfig to build cache.
On some platforms a multiarch approach is used for libraries.
The build system does not take this into account and still
places libraries into the lib directory if no --libdir is
specified to configure. On initial startup this results in
libasteriskssl.so not being found, as it is not in the multiarch
lib directory.
This change does the minimally invasive thing and executes
ldconfig so that the libraries in the lib directory are found
and their location cached. By doing so Asterisk starts up fine.
If DESTDIR is specified, however, the old logic is executed as
the install process may not have permission to alter the ldconfig
cache.
The inbound authentication object is supposed to be immutable when it is
stored in sorcery. However, the immutable property is violated if the
authentication object does not have a realm set.
The immutable contract violation has a different effect depending upon
what sorcery back end is used. If it is the config file back end you
would get the same object back until res_pjsip is reloaded. If it is the
real-time or AstDB back end you would get a new object on each query. If
it is cached you would get the same object back until it is refreshed from
the database.
Once an inbound authentication object has its realm set it may or may not
get updated again if the default_realm changes.
If the same authentication object is used for inbound and outbound
authentication then the immutable violation can make it very hard to
determine why the outbound authentication now fails. The only diagnostic
message is a complaint about no realms matching when it had worked
earlier. It fails because of the difference in behaviour for an empty
realm setting between inbound and outbound authentication objects.
* Fixed the sorcery object immutable violation by creating a new object
and setting the default_realm on it instead. The new object is a shallow
copy for speed.
* The auth_store thread storage no longer holds an auth ref. It
interferes with the shallow copy and never needed a ref anyway.
There was code attempting to update the artificial authentication object
whenever the default_realm changed. However, once the artificial
authentication object was created it would never get updated. The
artificial authentication object would require a system restart for a
change to the default_realm to take effect.
Using the same auth section for inbound and outbound authentication is not
recommended. There is a difference in meaning for an empty realm setting
between inbound and outbound authentication uses.
An empty inbound auth realm represents the global section's default_realm
value when the authentication object is used to challenge an incoming
request. An empty outgoing auth realm is treated as a don't care wildcard
when the authentication object is used to respond to an incoming
authentication challenge.
Richard Mudgett [Mon, 13 Feb 2017 23:11:06 +0000 (17:11 -0600)]
pjproject: Fixes to resolve DNS SRV crashes.
* Re #1945 (misc): Don't trigger SRV complete callback when there is a
parse error.
* srv_resolver.c: Don't try to send query if already considered resolved.
** In resolve_hostnames() don't try to resolve a query that is already
considered resolved.
** In resolve_hostnames() fix DNS typo in comments.
** In build_server_entries() move a common expression assigning to cnt
earlier.
* sip_transport.c: Fix tdata object name to actually contain the pointer.
It helps if the logs referencing a tdata object buffer actually have a
name that includes the correct pointer as part of the name. Also since
the tdata has its own pool it helps if any logs referencing the pool have
the same name as the tdata object. This change brings tdata logging in
line with how tsx objects are named.
Richard Mudgett [Sat, 4 Feb 2017 22:00:14 +0000 (16:00 -0600)]
res_pjsip: Record the serializer earlier on the tdata.
When PJPROJECT needs to do a DNS resolution and there is not a cached
entry available, the SIP request message goes out on the PJSIP monitor
thread instead of the original serializer thread. Thus when the response
comes back it does not get processed by the original sending serializer.
This patch records the serializer on tdata before passing a request
message to PJPROJECT where it can in Asterisk code. There are several
places in PJPROJECT for outbound registration and publishing support that
would need to record the serializer. Unfortunately, without replacing the
PJPROJECT DNS resolver as was done in v14 we cannot fix those without
modifying PJPROJECT.
Even if we backported the DNS resolver from v14, the outbound registration
refresh timer does not go out on a serializer thread but the PJSIP monitor
thread. Fortunately, Asterisk's outbound publish support doesn't use the
auto refresh timer that would also not go out under the serializer thread.
Sean Bright [Mon, 20 Feb 2017 11:53:47 +0000 (06:53 -0500)]
res_config_ldap: Don't try to delete non-existent attributes
OpenLDAP will raise an error when we try to delete an LDAP attribute
that doesn't exist. We need to filter out LDAP_MOD_DELETE requests
based on which attributes the current LDAP entry actually has. There
is of course a small window of opportunity for this to still fail,
but it is much less likely now.
Sean Bright [Mon, 20 Feb 2017 11:45:54 +0000 (06:45 -0500)]
res_config_ldap: Make memory allocation more consistent
The code in update_ldap() and update2_ldap() was using both Asterisk's
memory allocation routines as well as OpenLDAP's. I've changed it so
that everything that is passed to OpenLDAP's functions are allocated
with their routines.
Sean Bright [Mon, 20 Feb 2017 11:30:31 +0000 (06:30 -0500)]
res_config_ldap: Fix configuration inheritance from _general
The "_general" configuration section allows administrators to provide
both general configuration options (host, port, url, etc.) as well as a
global realtime-to-LDAP-attribute mapping that is a fallback if one of
the later sections do not override it. This neglected to exclude the
general configuration options from the mapping. As an example, during
my testing, chan_sip requested 'port' from realtime, and because I did
not have it defined, it pulled in the 'port' configuration option from
"_general." We now filter those out explicitly.
Sean Bright [Mon, 20 Feb 2017 11:27:38 +0000 (06:27 -0500)]
res_config_ldap: Fix erroneous LDAP_MOD_REPLACE in LDAP modify
We always treat the first change of our modification batch as a
replacement when it sometimes is actually a delete. So we have to pass
the correct arguments to the OpenLDAP library.
ASTERISK-26580 #close
Reported by: Nicholas John Koch
Patches:
res_config_ldap.c-11.24.1.patch (license #6833) patch uploaded
by Nicholas John Koch
Sean Bright [Wed, 15 Feb 2017 17:55:19 +0000 (12:55 -0500)]
res_config_sqlite3: Fix crash when loading with invalid config
When ast_config_load() fails with CONFIG_STATUS_FILEINVALID, it has
already destroyed the ast_config struct for us. Trying to do it again
results in a crash.
Sean Bright [Thu, 16 Feb 2017 14:38:06 +0000 (09:38 -0500)]
realtime: Fix LIKE escaping in SQL backends
The realtime framework allows for components to look up values using a
LIKE clause with similar syntax to SQL's. pbx_realtime uses this
functionality to search for pattern matching extensions that start with
an underscore (_).
When passing an underscore to SQL's LIKE clause, it will be interpreted
as a wildcard matching a single character and therefore needs to be
escaped. It is (for better or for worse) the responsibility of the
component that is querying realtime to escape it with a backslash before
passing it in. Some RDBMs support escape characters by default, but the
SQL92 standard explicitly says that there are no escape characters
unless they are specified with an ESCAPE clause, e.g.
SELECT * FROM table WHERE column LIKE '\_%' ESCAPE '\'
This patch instructs 3 backends - res_config_mysql, res_config_pgsql,
and res_config_sqlite3 - to use the ESCAPE clause where appropriate.
Looking through documentation and source tarballs, I was able to
determine that the ESCAPE clause is supported in:
MySQL 5.0.15 (released 2005-10-22 - earliest version available from
archives)
PostgreSQL 7.1 (released 2001-04-13)
SQLite 3.1.0 (released 2005-01-21)
The versions of the relevant libraries that we depend on to access MySQL
and PostgreSQL will not work on versions that old, and I've added an
explicit check in res_config_sqlite3 to only use the ESCAPE clause when
we have a sufficiently new version of SQLite3.
res_config_odbc already handles the escape characters appropriately, so
no changes were required there.
Joshua Colp [Thu, 16 Feb 2017 16:30:00 +0000 (16:30 +0000)]
build: Execute ldconfig to build cache.
On some platforms a multiarch approach is used for libraries.
The build system does not take this into account and still
places libraries into the lib directory if no --libdir is
specified to configure. On initial startup this results in
libasteriskssl.so not being found, as it is not in the multiarch
lib directory.
This change does the minimally invasive thing and executes
ldconfig so that the libraries in the lib directory are found
and their location cached. By doing so Asterisk starts up fine.
Sean Bright [Thu, 16 Feb 2017 11:46:04 +0000 (06:46 -0500)]
res_config_sqlite3: Properly create missing columns when necessary
There were two specific issues resolved here:
1) The code that iterated over the required fields
(via ast_realtime_require) was broken for the RQ_INTEGER1 field
type. Iteration would stop when the first RQ_INTEGER1 (0) field
was encountered.
2) sqlite3_changes() was used to try and count the number of rows
returned by a SELECT statement. sqlite3_changes() only counts
affected rows, so this was always returning the value from the
most recent data modification statement. We now separate read-only
queries from data modification queries and count rows appropriately
in both cases.
ASTERISK-23457 #close
Reported by: Scott Griepentrog
chan_unistim: fix char type to have consistent behavior on ARM
There is difference exists in behaviour of char type on x86 and ARM.
On x86 by default char variable type means signed char, but in ARM
unsigned char used. This make binary calculations and negative values
works wrong on ARM.
This patch change type of char variables used for store negative
values and binary calculations to signed char.
This patch fixes 2 original issues and more that those 2 exposed.
* When we send a NOTIFY, and the client either doesn't respond or
responds with a non OK, pjproject only calls our
pubsub_on_evsub_state callback, no others. Since
pubsub_on_evsub_state (which does the sub_tree cleanup) does not
expect to be called back without the other callbacks being called
first, it just returns leaving the sub_tree orphaned. Now
pubsub_on_evsub_state checks the event for PJSIP_EVENT_TSX_STATE
which is what pjproject will set to tell us that it was the
transaction that timed out or failed and not the subscription
itself timing our or being terminated by the client. If is
TSX_STATE, pubsub_on_evsub_state now does the proper cleanup
regardless of the state of the subscription.
* When a client renews a subscription, we don't update the
persisted subscription with the new expires timestamp. This causes
subscription_persistence_recreate to prune the subscription if/when
asterisk restarts. Now, pubsub_on_rx_refresh calls
subscription_persistence_update to apply the new expires timestamp.
This exposed other issues however...
* When creating a dialog from rdata (which sub_persistence_recreate
does from the packet buffer) there must NOT be a tag on the To
header (which there will be when a client refreshes a
subscription). If there is one, pjsip_dlg_create_uas will fail.
To address this, subscription_persistence_update now accepts a flag
that indicates that the original packet buffer must not be updated.
New subscribes don't set the flag and renews do. This makes sure
that when the rdata is recreated on asterisk startup, it's done
from the original subscribe packet which won't have the tag on To.
* When creating a dialog from rdata, we were setting the dialog's
remote (SUBSCRIBE) cseq to be the same as the local (NOTIFY) cseq.
When the client tried to resubscribe after a restart with the
correct cseq, we'd reject the request with an Invalid CSeq error.
* The acts of creating a dialog and evsub by themselves when
recreating a subscription does NOT restart pjproject's subscription
timer. The result was that even if we did correctly recreate the
subscription, we never removed it if the client happened to go away
or send a non-OK response to a NOTIFY. However, there is no
pjproject function exposed to just set the timer on an evsub that
wasn't created by an incoming subscribe request. To address this,
we create our own timer using ast_sip_schedule_task. This timer is
used only for re-establishing subscriptions after a restart.
An earlier approach was to add support for setting pjproject's
timer (via a pjproject patch) and while that patch is still included
here, we don't use that call at the moment.
While addressing these issues, additional debugging was added and
some existing messages made more useful. A few formatting changes
were also made to 'pjsip show scheduled tasks' to make displaying
the subscription timers a little more friendly.
Sean Bright [Wed, 15 Feb 2017 17:03:00 +0000 (12:03 -0500)]
res_rtp_asterisk: Use PJ_ICE_MAX_CAND instead of hard-coding 16
pjsip limits the total number of ICE candidates to PJ_ICE_MAX_CAND,
which is a compile-time constant. Instead of hard-coding 16 when we
enumerate local interfaces, use PJ_ICE_MAX_CAND so that we can
potentially collect more interfaces if the compile time options are
changed.
Tzafrir Cohen [Sat, 21 Jan 2017 05:59:15 +0000 (07:59 +0200)]
tcptls: use TLS_client_method with OpenSSL 1.1
OpenSSL 1.1 introduced TLS_client_method() and deprecated the previous
version-specific methods (such as TLSv1_client_method(). Other than
being simpler to use and more correct (gain support for TLS newer that
TLS1, in our case), the older ones produce a deprecation warning that
fails the build in dev-mode.
rrittgarn [Wed, 25 Jan 2017 22:25:21 +0000 (16:25 -0600)]
app_voicemail: VoiceMailPlayMsg did not play database stored messages
When attempting to use VoiceMailPlayMsg with a realtime data backend
the message is located, but never retrieved. This patch adds the
required RETRIEVE and DISPOSE calls that will fetch the message from
the database (and IMAP storage as well for that matter).
Sean Bright [Tue, 14 Feb 2017 14:12:31 +0000 (09:12 -0500)]
app_record: Add option to prevent silence from being truncated
When using Record() with the silence detection feature, the stream is
written out to the given file. However, if only 'silence' is detected,
this file is then truncated to the first second of the recording.
This patch adds the 'u' option to Record() to override that behavior.
ASTERISK-18286 #close
Reported by: var
Patches:
app_record-1.8.7.1.diff (license #6184) patch uploaded by var
Sean Bright [Sat, 11 Feb 2017 15:57:03 +0000 (10:57 -0500)]
cli: Fix various CLI documentation and completion issues
* app_minivm: Use built-in completion facilities to complete optional
arguments.
* app_voicemail: Use built-in completion facilities to complete
optional arguments.
* app_confbridge: Add missing colons after 'Usage' text.
* chan_alsa: Use built-in completion facilities to complete optional
arguments.
* chan_sip: Use built-in completion facilities to complete optional
arguments. Add completions for 'load' for 'sip show user', 'sip show
peer', and 'sip qualify peer.'
* chan_skinny: Correct and extend completions for 'skinny reset' and
'skinny show line.'
* func_odbc: Correct completions for 'odbc read' and 'odbc write'
* main/asterisk: Correct and extend completions for 'core show file
version.'
* main/astmm: Use built-in completion facilities to complete arguments
for 'memory' commands.
* main/bridge: Correct completions for 'bridge kick.'
* main/ccss: Use built-in completion facilities to complete arguments
for 'cc cancel' command.
* main/cli: Add 'all' completion for 'channel request hangup.' Correct
completions for 'core set debug channel.' Correct completions for 'core
show calls.'
* main/pbx_app: Remove redundant completions for 'core show
applications.'
* main/pbx_hangup_handler: Remove unused completions for 'core show
hanguphandlers all.'
* res_sorcery_memory_cache: Add completion for 'reload' argument of
'sorcery memory cache stale' and properly implement.
Norbert Varga [Fri, 13 Jan 2017 17:21:36 +0000 (18:21 +0100)]
chan_pjsip: Multidomain endpoint finding on call
When PJSIP tries to call an endpoint with a domain (e.g. 1000@test.com),
the user part is stripped down as it would be a trunk with a specified user,
and only the host part is called as a PJSIP endpoint and can't be found.
This is not correct in the case of a multidomain SIP account, so the stripping
after the @ sign is done only if the whole endpoint (in multidomain case
1000@test.com) can't be found.
Joshua Colp [Mon, 13 Feb 2017 11:05:51 +0000 (11:05 +0000)]
channel: Protect flags in ast_waitfor_nandfds operation.
The ast_waitfor_nandfds operation will manipulate the flags
of channels passed in. This was previously done without
the channel lock being held. This could result in incorrect
values existing for the flags if another thread manipulated
the flags at the same time.
This change locks the channel during flag manipulation.
Sean Bright [Thu, 9 Feb 2017 16:01:22 +0000 (11:01 -0500)]
manager: Restore Originate failure behavior from Asterisk 11
In Asterisk 11, if the 'Originate' AMI command failed to connect the provided
Channel while in extension mode, a 'failed' extension would be looked up and
run. This was, I believe, unintentionally removed in 51b6c49. This patch
restores that behavior.
This also adds an enum for the various 'synchronous' modes in an attempt to
make them meaningful.
Richard Mudgett [Wed, 8 Feb 2017 20:27:18 +0000 (14:27 -0600)]
core: Cleanup some channel snapshot staging anomalies.
We shouldn't unlock the channel after starting a snapshot staging because
another thread may interfere and do its own snapshot staging.
* app_dial.c:dial_exec_full() made hold the channel lock while setting up
the outgoing channel staging. Made hold the channel lock after the called
party answers while updating the caller channel staging.
* chan_sip.c:sip_new() completed the channel staging on off-nominal exit.
Also we need to use ast_hangup() instead of ast_channel_unref() at that
location.
* channel.c:__ast_channel_alloc_ap() added a comment about not needing to
complete the channel snapshot staging on off-nominal exit paths.
* rtp_engine.c:ast_rtp_instance_set_stats_vars() made hold the channel
locks while staging the channels for the stats channel variables.
The change in question was intended to prevent the need to reload in
order to update qualifies on contacts when an AOR changes. However, this
ended up causing a deadlock instead.
Joshua Colp [Tue, 7 Feb 2017 18:01:03 +0000 (18:01 +0000)]
srv: Fix crash when ast_srv_lookup is used and 0 records are returned.
When performing an SRV lookup using the ast_srv_lookup function it
did not properly handle the situation where 0 records are returned.
If this happened it would wrongly assume that at least one record
was present.
This change fixes the code so it will exit early if an error occurs
or if 0 records are returned.
ASTERISK-26772
patches:
srv_lookup.patch submitted by nappsoft (license 6822)
Joshua Colp [Mon, 6 Feb 2017 17:40:45 +0000 (17:40 +0000)]
res_stasis_device_state: Protect the adding/removing of subscriptions.
The adding and removing of device state subscriptions did not protect
fully against simultaneous manipulation. In particular the subscribe
case allowed a small window where two subscriptions could be added for
the same device state instead of just one.
This change makes the code hold the subscriptions lock for the entirety
of each operation to ensure that two are not occurring at the same time.
Mark Michelson [Wed, 1 Feb 2017 19:54:50 +0000 (13:54 -0600)]
Update qualifies when AOR configuration changes.
Prior to this change, qualifies would only update in the following
cases:
* A reload of res_pjsip.so was issued.
* A dynamic contact was re-registered after its AOR's qualify_frequency
had been changed
This does not work well if you are using realtime for your AORs. You can
update your database to have a new qualify_frequency, but the permanent
contacts on that AOR will not have their qualifies updated. And the
dynamic contacts on that AOR will not have their qualifies updated until
the next registration, which could be a long time.
This change seeks to fix this problem by making it so that whenever AOR
configuration is applied, the contacts pertaining to that AOR have their
qualifies updated.
Additions from this patch:
* AOR sorcery objects now have an apply handler that calls into a newly
added function in the OPTIONS code. This causes all contacts
associated with that AOR to re-schedule qualifies.
* When it is time to qualify a contact, the OPTIONS code checks to see
if the AOR can still be retrieved. If not, then qualification is
canceled on the contact.
Alterations from this patch:
* The registrar code no longer updates contact's qualify_frequence and
qualify_timeout. There is no point to this since those values already
get updated when the AOR changes.
* Reloading res_pjsip.so no longer calls the OPTIONS initialization
function. Reloading res_pjsip.so results in re-loading AORs, which
results in re-scheduling qualifies.
Richard Mudgett [Wed, 1 Feb 2017 00:28:15 +0000 (18:28 -0600)]
channel.c: Fix unbalanced read queue deadlocking local channels.
Using the timerfd timing module can cause channel freezing, lingering, or
deadlock issues. The problem is because this is the only timing module
that uses an associated alert-pipe. When the alert-pipe becomes
unbalanced with respect to the number of frames in the read queue bad
things can happen. If the alert-pipe has fewer alerts queued than the
read queue then nothing might wake up the thread to handle received frames
from the channel driver. For local channels this is the only way to wake
up the thread to handle received frames. Being unbalanced in the other
direction is less of an issue as it will cause unnecessary reads into the
channel driver.
ASTERISK-26716 is an example of this deadlock which was indirectly fixed
by the change that found the need for this patch.
* In channel.c:__ast_queue_frame(): Adding frame lists to the read queue
did not add the same number of alerts to the alert-pipe. Correspondingly,
when there is an exceptionally long queue event, any removed frames did
not also remove the corresponding number of alerts from the alert-pipe.
Richard Mudgett [Tue, 31 Jan 2017 22:38:49 +0000 (16:38 -0600)]
res_agi: Prevent an AGI from eating frames it should not. (Re-do)
A dialplan intercept routine is equivalent to an interrupt routine. As
such, the routine must be done quickly and you do not have access to the
media stream. These restrictions are necessary because the media stream
is the responsibility of some other code and interfering with or delaying
that processing is bad. A possible future dialplan processing
architecture change may allow the interception routine to run in a
different thread from the main thread handling the media and remove the
execution time restriction.
* Made res_agi.c:run_agi() running an AGI in an interception routine run
in DeadAGI mode. No touchy channel frames.
Richard Mudgett [Tue, 31 Jan 2017 22:32:18 +0000 (16:32 -0600)]
Frame deferral: Revert API refactoring.
There are several issues with deferring frames that are caused by the
refactoring.
1) The code deferring frames mishandles adding a deferred frame to the
deferred queue. As a result the deferred queue can only be one frame
long.
2) Deferrable frames can come directly from the channel driver as well as
the read queue. These frames need to be added to the deferred queue.
3) Whoever is deferring frames is really only doing the __ast_read() to
collect deferred frames and doesn't care about the returned frames except
to detect a hangup event. When frame deferral is completed we must make
the normal frame processing see the hangup as a frame anyway. As such,
there is no need to have varying hangup frame deferral methods. We also
need to be aware of the AST_SOFTHANGUP_ASYNCGOTO hangup that isn't real.
That fake hangup is to cause the PBX thread to break out of loops to go
execute a new dialplan location.
4) To properly deal with deferrable frames from the channel driver as
pointed out by (2) above, means that it is possible to process a dialplan
interception routine while frames are deferred because of the
AST_CONTROL_READ_ACTION control frame. Deferring frames is not
implemented as a re-entrant operation so you could have the unsupported
case of two sections of code thinking they have control of the media
stream.
A worse problem is because of the bad implementation of the AMI PlayDTMF
action. It can cause two threads to be deferring frames on the same
channel at the same time. (ASTERISK_25940)
* Rather than fix all these problems simply revert the API refactoring as
there is going to be only autoservice and safe_sleep deferring frames
anyway.
projects / thirdparty / asterisk.git / log
