Richard Mudgett [Tue, 27 Mar 2018 16:04:42 +0000 (11:04 -0500)]
res_pjsip.c: Split ast_sip_push_task_synchronous() to fit expectations.
ast_sip_push_task_synchronous() did not necessarily execute the passed in
task under the specified serializer. If the current thread is any
registered pjsip thread then it would execute the task immediately instead
of under the specified serializer. Reentrancy issues could result if the
task does not execute with the right serializer.
The original reason ast_sip_push_task_synchronous() checked to see if the
current thread was a registered pjsip thread was because of a deadlock
with masquerades and the channel technology's fixup callback
(ASTERISK_22936). A subsequent masquerade deadlock fix (ASTERISK_24356)
involving call pickups avoided the original deadlock situation entirely.
The PJSIP channel technology's fixup callback no longer needed to call
ast_sip_push_task_synchronous().
However, there are a few places where this unexpected behavior is still
required to avoid deadlocks. The pjsip monitor thread executes callbacks
that do calls to ast_sip_push_task_synchronous() that would deadlock if
the task were actually pushed to the specified serializer. I ran into one
dealing with the pubsub subscriptions where an ao2 destructor called
ast_sip_push_task_synchronous().
* Split ast_sip_push_task_synchronous() into
ast_sip_push_task_wait_servant() and ast_sip_push_task_wait_serializer().
ast_sip_push_task_wait_servant() has the old behavior of
ast_sip_push_task_synchronous(). ast_sip_push_task_wait_serializer() has
the new behavior where the task is always executed by the specified
serializer or a picked serializer if one is not passed in. Both functions
behave the same if the current thread is not a SIP servant.
* Redirected ast_sip_push_task_synchronous() to
ast_sip_push_task_wait_servant() to preserve API for released branches.
Richard Mudgett [Thu, 22 Mar 2018 00:43:21 +0000 (19:43 -0500)]
pjsip_scheduler.c: Fix some corner cases.
* Fix the periodic interval wander because it may take significant time
between the sched thread queueing the task in the serializer and the
serializer actually executing the task. The time it takes to actually
execute the task was already taken into account.
* Pass a schtd ref to the serializer when we queue a scheduled task on
the serializer. We don't want it going away on us while it is in the
serializer queue.
* Skip the scheduled task if the task was canceled between queueing the
task to the serializer and the serializer actually executing the task.
* Reorder struct ast_sip_sched_task to avoid unnecessary padding. Removed
task_id and added next_periodic.
* Hold a ref to the passed in serializer so the serializer cannot go away
on the scheduled task.
cdr_mysql: Compile error because MYSQL_PORT definition is missing
If it is not defined, it will add MYSQL_PORT definition. After some
research on MySQL/MariaDB development tree, I couldn't find any reference
to MYSQL_PORT definition in include files.
res_pjsip_session: Rewrite o= with external_media_address.
It now appends the external IP address on the
o= line of the SDP packet. The decision was made to write
the numeric IP address as opposed to the RFC that states
the FQDN should be used if and when available. We believe
the usage of literal IP address will help avoid
potential problems.
Nathan Bruning [Thu, 22 Feb 2018 18:18:48 +0000 (19:18 +0100)]
res_pjsip_notify.c: enable in-dialog NOTIFY
This patch adds support to send in-dialog SIP NOTIFY commands on
chan_pjsip channels, similar to the functionality recently added
for chan_sip (ASTERISK_27461).
This extends res_pjsip_notify to allow for in-dialog messages.
Richard Mudgett [Thu, 22 Mar 2018 18:35:04 +0000 (13:35 -0500)]
pjsip_scheduler.c: Fix ao2 usage errors.
* Removed several invalid uses of OBJ_NOLOCK. These uses resulted in the
'tasks' container being accessed without a lock in a multi-threaded
environment. A recipe for crashes.
* Removed needlessly obtaining schtd object references. If the caller
providing you a pointer to an object doesn't have a valid reference then
you cannot safely get one from it.
* Getting a ref to 'tasks' when you aren't copying the pointer into
another location is useless. The 'tasks' container pointer is global.
* Removed many unnecessary uses of RAII_VAR.
* Make ast_sip_schedule_task() name parameter const.
Richard Mudgett [Thu, 5 Apr 2018 23:33:40 +0000 (18:33 -0500)]
res_pjsip_refer/chan_sip: Fix INVITE with replaces transfer to ConfBridge
There is a problem when an INVITE-with-Replaces transfer targets a channel
in a ConfBridge. The transfer will unconditionally swap out the
ConfBridge channel. Unfortunately, the ConfBridge state will not be aware
of this change. Unexpected behavior will happen as a result since
ConfBridge channels currently can only be replaced by a masquerade and not
normal bridge channel moves.
* We just need to pretend that the channel isn't in a bridge (like other
transfer methods already do) so the transfer channel will masquerade into
the ConfBridge channel.
Richard Mudgett [Thu, 5 Apr 2018 22:40:52 +0000 (17:40 -0500)]
chan_sip.c: Fix INVITE with replaces channel ref leak.
Given the below call scenario:
A -> Ast1 -> B
C <- Ast2 <- B
1) A calls B through Ast1
2) B calls C through Ast2
3) B transfers A to C
When party B transfers A to C, B sends a REFER to Ast1 causing Ast1 to
send an INVITE with replaces to Ast2. Ast2 then leaks a channel ref of
the channel between Ast1 and Ast2.
Channel ref leaks are easily seen in the CLI "core show channels" output.
The leaked channels appear in the output but you can do nothing with them
and they never go away unless you restart Asterisk.
* Properly account for the channel refs when imparting a channel into a
bridge when handling an INVITE with replaces in handle_invite_replaces().
The ast_bridge_impart() function steals a channel ref but the code didn't
account for how many refs were held by the code at the time and which ref
was stolen.
* Eliminated RAII_VAR in handle_invite_replaces().
Build System: Strip '-std=c99' from CFLAGS provided by libraries.
Asterisk requires GNU C extensions. On some systems certain libraries
may incorrectly push -std=c99 into CFLAGS, thus breaking the build.
This change causes that flag to be stripped so the Asterisk build is not
broken by those libraries. This change is made for both pkgconfig and
tool based libraries.
George Joseph [Sun, 25 Mar 2018 18:35:12 +0000 (12:35 -0600)]
pjroject_bundled: Add already-destroyed check to tsx_timer_callback
There have been cases that when the transaction timer callback is called
the tsx is already destroyed. This causes a crash. We now check the
tsx state and return if the tsx is already destroyed.
George Joseph [Sun, 25 Mar 2018 18:12:39 +0000 (12:12 -0600)]
pjproject_bundled: Add patch for pj_atomic crashes
There have been some crashes in the past where something attempts
to use a pj_atomic after it's already been destroyed. This patch
tries to prevent it by making sure that pj_atomic_destroy sets
its mutex to NULL when it's done. The pj_mutex functions already check
for a NULL mutex and just return PJ_EINVAL.
Teluu also added some checks to the win32 implementation as well.
Corey Farrell [Wed, 28 Mar 2018 13:18:06 +0000 (09:18 -0400)]
core: Create main/options.c.
This creates a separate source to 'own' symbols related to options.h and
paths.h. This significantly reduces the number of exports created by
main/asterisk.o. This change is required to eventually be able to
link unmodified Asterisk sources to utilities and/or stand-alone tests.
Kevin Harwell [Tue, 20 Mar 2018 20:28:12 +0000 (15:28 -0500)]
bridge_softmix: Clear "talking" when a channel is put on hold
This patch clears the talking flag from the channel (if already set), and
notifies listeners when that channel is put on hold. Note however, if the
endpoint continues to send audio frames and these are received by the bridge
then that channel will be put back into a "talking" state even though they
are on hold.
Ross Beer [Wed, 7 Mar 2018 12:15:05 +0000 (12:15 +0000)]
pjsip_transport_events.c: Fix crash using stale transport pointer.
Apparently it is possible for the transport to be destroyed without
triggering the transport callback logic. As a result the transport gets
destroyed and we have a stale pointer in the active_transports container.
* Invoke the transport monitor callback checks when the transport is
destroyed in addition to when it is disconnected and shutdown.
Alexander Traud [Tue, 20 Mar 2018 16:55:56 +0000 (17:55 +0100)]
BuildSystem: For consistency, avoid extra libs to be empty.
AST_EXT_LIB_CHECK has several optional parameters. When an optional parameter
is left empty, [] is used to indicate this. However, this is done in the script
./configure only then, when a further parameter is not empty. For example, when
no extra libraries are needed to test the checked library, parameter 5 is not
mentioned. Except parameter 6 and higher are used, then parameter 5 must be
empty.
However, this general rule was broken
* three times for parameter 5 (extra libs) and
* three times for parameter 4 (header)
as found via the Regular Expression \[\]\). In case of parameter 5, all cases
were changed, because that happened for no reason. In case of parameter 4, an
[] improves readability actually. Therefore for parameter 4, the only case which
did not do it was changed. All this aims to create more consistency: Only do
something different if there is a reason to do so.
George Joseph [Fri, 16 Mar 2018 15:19:11 +0000 (09:19 -0600)]
channel.c: Allow generic plc then channel formats are equal
If the two formats on a channel are equal, we don't transcode and since
the generic plc needs slin to work, it doesn't get invoked.
* A new configuration option "genericplc_on_equal_codecs" was added
to the "plc" section of codecs.conf to allow generic packet loss
concealment even if no transcoding was originally needed.
Transcoding via SLIN is forced in this case.
Corey Farrell [Sat, 17 Mar 2018 21:41:13 +0000 (17:41 -0400)]
core: Minor cleanup of ast_el_read_char.
* Define CHAR_T_LIBEDIT and CHAR_TO_LIBEDIT based on
HAVE_LIBEDIT_IS_UNICODE. This avoids needing to repeatedly use
conditional blocks, eliminates having multiple function prototypes.
* Remove parenthesis from return values.
* Add missing code block brackets {}.
* Reduce use of 'else' conditional statements where possible.
Alexander Traud [Sat, 17 Mar 2018 15:54:09 +0000 (16:54 +0100)]
BuildSystem: Check for header file of OGG.
Asterisk uses various symbols of the shared library libogg within the module
format_ogg_vorbis. However, the source code of that module did not include the
header file of libogg explicitly but implicitly. Because that header was not
included before Asterisk 14, the script ./configure was told not to check for
it.
Anyway, even Asterisk 13 LTS uses symbols of libogg. Therefore, that header
should be included explicitly. Therefore, ./configure should check for that
header.
Alexander Traud [Fri, 9 Mar 2018 12:26:40 +0000 (13:26 +0100)]
BuildSystem: When no download utility is available, display the explanation.
./configure --with-pjproject-bundled
did not display an explanation, when no download utility like wget, curl, or
fetch was installed beforehand, although an explanation existed in code. This
happened because the code expected the variable DOWNLOAD_TO_STDOUT to be empty.
However, the script ./configure set that variable always.
Alexander Traud [Sat, 17 Mar 2018 10:00:06 +0000 (11:00 +0100)]
BuildSystem: Remove unused dependency on libltdl.
Asterisk does not need the development package of libltdl, because it does not
use any symbol of -lltdl directly. Instead, it uses the runtime package via the
shared library -lodbc. On the supported platforms, that shared library declares
its dependency on -lltdl correctly, otherwise AST_EXT_LIB_CHECK would have
failed.
Alexander Traud [Fri, 16 Mar 2018 14:53:22 +0000 (15:53 +0100)]
BuildSystem: Avoid an extra case for OpenBSD.
Nine years ago with Mantis 13639 (now ASTERISK-12841) an extra case for OpenBSD
was introduced: Vorbis required Ogg to be specified manually, because the shared
library libvorbis.so did not specify its required dependency on -logg itself.
Today with OpenBSD 6.2, all libvorbis*.so declare their dependencies correctly.
Therefore, an extra case is not required anymore.
Alexander Traud [Mon, 5 Mar 2018 16:17:23 +0000 (17:17 +0100)]
BuildSystem: Enable Advanced Linux Sound Architecture (ALSA) in NetBSD.
In the script ./configure, AST_EXT_LIB_CHECK checks for external libraries. Some
libraries do not specify all their dependencies and require additional shared
libraries. In AST_EXT_LIB_CHECK, this is the fifth parameter. However, if a
library is specified there, it must exist on the platform, because ./configure
tries to compile/link/execute a small app using those statements. For example,
the library libdl.so is Linux specific and does not exist on BSD-like platforms.
Furthermore, no supported platform/version was found, which still (ever?)
requires those additional libraries. Therefore, they were simply removed.
Finally, this change adds the error code ESTRPIPE to the channel driver
chan_alsa for those platforms which lack it, again for example NetBSD.
George Joseph [Fri, 16 Mar 2018 14:02:20 +0000 (08:02 -0600)]
app_voicemail: Fix json blob errors
When app_voicemail calls ast_test_suite_notify with the results of
a user keypress, it formats the keypress as '%c'. If the user hung up
or some other error occurrs, the result of the keypress is a non
printable character. This ultimately causes json_vpack_ex to think
it's being passed a non utf-8 string and return an error.
* Keypress results passed to ast_test_suite_notify are now checked with
isprint() and a '?' is substituted if the check fails.
projects / thirdparty / asterisk.git / log
