* Fix deadlock in
bridge_softmix.c:softmix_bridge_stream_topology_changed() between
bridge_channel and channel locks.
* The new bridge technology topology change callbacks must be called with
the bridge locked. The callback references the bridge channel list, the
bridge technology could change, and the bridge stream mapping is updated.
ASTERISK-27212
Change-Id: Ide4360ab853607e738ad471721af3f561ddd83be
This change fixes a few locking issues and some video misrouting.
1. When accessing the stream topology of a channel the channel lock
must be held to guarantee the topology remains valid.
2. When a channel was joined to a bridge the bridge specific
implementation for stream mapping was not invoked, causing video
to be misrouted for a brief period of time.
ASTERISK-27182
Change-Id: I5d2f779248b84d41c5bb3896bf22ba324b336b03
This change does a few things to improve packet loss and renegotiation:
1. On outgoing RTP streams we will now properly reflect out of order
packets and packet loss in the sequence number. This allows the
remote jitterbuffer to better reorder things.
2. Video updates can now be discarded for a period of time
after one has been sent to prevent flooding of clients.
3. For declined and removed streams we will now release any
media session resources associated with them. This was not
previously done and caused an issue where old state was being
used for a new stream.
4. RTP bundling was not actually removing bundled RTP instances
from the parent. This has been resolved by removing based on
the RTP instance itself and not the SSRC.
5. The code did not properly handle explicitly unbundling an
RTP instance from its parent. This now works as expected.
ASTERISK-27143
Change-Id: Ibd91362f0e4990b6129638e712bc8adf0899fd45
When a participant leaves a bridge while operating in SFU mode
their respective stream on every other participant needs to be
removed. Leaving the stream out of the new topology results in
every stream after it being moved and reordered. This causes
problems with clients. Instead simply mark the stream as removed
which leaves it in place in the SDP and doesn't reorder or touch
any other streams.
ASTERISK-27136
Change-Id: I4b3f840adcdf69b83842b0d8a737665ba0ef9cb1
Streams are never truly removed in SDP, they still occupy
a location within the SDP. This location can be reused by
another stream if it so chooses.
This change takes advantage of this such that if a new stream
is needing to be added for a new participant any removed streams
are instead replaced first. This reduces the size of the SDP
and the number of streams.
ASTERISK-27134
Change-Id: I95cdcfd55cf47e02ea52abb5d94008db3fb68b1d
This change fixes a few things uncovered during SFU testing.
1. Unreal channels incorrectly forwarded video frames when
no video stream was present on them. This caused a crash when
they were read as the core requires a stream to exist for the
underlying media type. The Unreal channel will now ensure a
stream exists for the media type before forwarding the frame
and if no stream exists then the frame is dropped.
2. Mapping of frames during bridging from the stream number of
the underlying channel to the stream number of the bridge was
done in the wrong location. This resulted in the frame getting
dropped. This mapping now occurs on reading of the frame from
the channel.
3. Bridging was using the wrong ast_read function resulting in
it living in a non-multistream world.
4. In bridge_softmix when adding new streams to existing channels
the wrong stream topology was copied resulting in no streams
being added.
Change-Id: Ib7445722c3219951d6740802a0feddf2908c18c8
This sets up the "plumbing" in bridge_softmix to
be able to accommodate Asterisk asking as an SFU
(selective forwarding unit) for conferences.
The way this works is that whenever a channel enters or leaves a
conference, all participants in the bridge get sent a stream topology
change request. The topologies consist of the channels' original
topology, along with video destination streams corresponding to each
participants' source video streams. So for instance, if Alice, Bob, and
Carol are in the conference, and each supplies one video stream, then
the topologies for each would look like so:
Alice:
Audio,
Source video(Alice),
Destination Video(Bob),
Destination video (Carol)
Bob:
Audio,
Source video(Bob)
Destination Video(Alice),
Destination video (Carol)
Carol:
Audio,
Source video(Carol)
Destination Video(Alice),
Destination video (Bob)
This way, video that arrives from a source video stream can then be
copied out to the destination video streams on the other participants'
channels.
Once the bridge gets told that a topology on a channel has changed, the
bridge constructs a map in order to get the video frames routed to the
proper destination streams. This is done using the bridge channel's
stream_map.
This change is bare-bones with regards to SFU support. Some key features
are missing at this point:
* Stream limits. This commit makes no effort to limit the number of
streams on a specific channel. This means that if there were 50 video
callers in a conference, bridge_softmix will happily send out topology
change requests to every channel in the bridge, requesting 50+
streams.
* Configuration. The plumbing has been added to bridge_softmix, but
there has been nothing added as of yet to app_confbridge to enable SFU
video mode.
* Testing. Some functions included here have unit tests.
However, the functionality as a whole has only been verified by
hand-tracing the code.
* Selectivenss. For a "selective" forwarding unit, this does not
currently have any means of being selective.
* Features. Presumably, someone might wish to only receive video from
specific sources. There are no external-facing functions at the moment
that allow for users to select who they receive video from.
* Efficiency. The current scheme treats all video streams as being
unidirectional. We could be re-using a source video stream as a
desetnation, too. But to simplify things on this first round, I did it
this way.
Change-Id: I7c44a829cc63acf8b596a337b2dc3c13898a6c4d
Some codecs - codec_speex specifically - take voice frames and return
other types of frames, like CNG. If we subsequently treat those as
voice frames, we'll run into trouble when destroying the frame because
of the requirement that each voice frame have an associated format.
ASTERISK-26880 #close
Reported by: Kirsty Tyerman
Change-Id: I43f8450c48fb276ad8b99db8512be82949c1ca7c
Adds binaural synthesis to bridge_softmix (via convolution using libfftw3).
Binaural synthesis is conducted at 48kHz.
For a conference, only one spatial representation is rendered.
The default rendering is applied for mono-capable channels.
ASTERISK-26292
Change-Id: Iecdb381b6adc17c961049658678f6219adae1ddf
WebRTC clients really, really want to know the SSRC of the media they're
getting. Changing the SSRC is generally not a good thing.
bridge_softmix, starting in Asterisk 12, started changing the SSRC of
parties as they joined or left the bridge. With most phones, this isn't
a problem: phones just play back the stream they're getting. With WebRTC
clients, however, the SSRC is tied to a media stream that may be
negotiated. When a new SSRC just shows up, the media can be dropped.
As it turns out, the SSRC change shouldn't even be necessary. From the
perspective of the client, it's still talking to Asterisk with the same
media stream: why indicate that the far party has suddenly changed to a
different source of media?
This patch opts to just remove the SSRC changes. With this patch, video
clients that join/leave a softmix bridge actually get the video stream
instead of freaking out.
ASTERISK-26555
Change-Id: I27fec098b32e7c8718b4b65f3fd5fa73527968bf
ASTERISK_REGISTER_FILE no longer has any purpose so this commit removes
all traces of it.
Previously exported symbols removed:
* __ast_register_file
* __ast_unregister_file
* ast_complete_source_filename
This also removes the mtx_prof static variable that was declared when
MTX_PROFILE was enabled. This variable was only used in lock.c so it
is now initialized in that file only.
ASTERISK-26480 #close
Change-Id: I1074af07d71f9e159c48ef36631aa432c86f9966
softmix_bridge_join() failed because of an allocation failure. To address
this, the softmix bridge technology now checks if the channel failed to
join softmix successfully. In addition, the bridge now begins the process
of kicking the channel out of the bridge so we don't have channels
partially in the bridge for very long.
* Fix the test_channel_feature_hooks.c unit tests. The test channel must
have a valid codec to join the simple_bridge technology. This patch makes
joining a bridge more strict by not allowing partially joined channels to
remain in the bridge.
Change-Id: I97e2ade6a2bcd1214f24fb839fda948825b61a2b
Fix off nominal crash where we could not setup the channel to process
frames for the softmix bridge technology because of allocation failure.
Change-Id: Ic307a8386e46bf551e48fcd1eb97276714d56372
Git does not support the ability to replace a token with a version
string during check-in. While it does have support for replacing a
token on clone, this is somewhat sub-optimal: the token is replaced
with the object hash, which is not particularly easy for human
consumption. What's more, in practice, the source file version was often
not terribly useful. Generally, when triaging bugs, the overall version
of Asterisk is far more useful than an individual SVN version of a file. As a
result, this patch removes Asterisk's support for showing source file
versions.
Specifically, it does the following:
* Rename ASTERISK_FILE_VERSION macro to ASTERISK_REGISTER_FILE, and
remove passing the version in with the macro. Other facilities
than 'core show file version' make use of the file names, such as
setting a debug level only on a specific file. As such, the act of
registering source files with the Asterisk core still has use. The
macro rename now reflects the new macro purpose.
* main/asterisk:
- Refactor the file_version structure to reflect that it no longer
tracks a version field.
- Remove the "core show file version" CLI command. Without the file
version, it is no longer useful.
- Remove the ast_file_version_find function. The file version is no
longer tracked.
- Rename ast_register_file_version/ast_unregister_file_version to
ast_register_file/ast_unregister_file, respectively.
* main/manager: Remove value from the Version key of the ModuleCheck
Action. The actual key itself has not been removed, as doing so would
absolutely constitute a backwards incompatible change. However, since
the file version is no longer tracked, there is no need to attempt to
include it in the Version key.
* UPGRADE: Add notes for:
- Modification to the ModuleCheck AMI Action
- Removal of the "core show file version" CLI command
Change-Id: I6cf0ff280e1668bf4957dc21f32a5ff43444a40e
With this patch, chan_pjsip/res_pjsip now sets the native formats to the
codecs negotiated by a call.
* The changes in chan_pjsip.c and res_pjsip_sdp_rtp.c set the native
formats to include all the negotiated audio codecs instead of only the
initial preferred audio codec and later the currently received audio
codec.
* The audio frame handling in channel.c:ast_read() is more streamlined and
will automatically adjust to changes in received frame formats. The new
policy is to remove translation and pass the new frame format to the
receiver except if the translation was to a signed linear format. A more
long winded version is commented in ast_read() along with some caveats.
* The audio frame handling in channel.c:ast_write() is more streamlined
and will automatically adjust any needed translation to changes in the
frame formats sent. Frame formats sent can change for many reasons such
as a recording is being played back or the bridged peer changed the format
it sends. Since it is a normal expectation that sent formats can change,
the codec mismatch warning message is demoted to a debug message.
* Removed the short circuit check in
channel.c:ast_channel_make_compatible_helper(). Two party bridges need to
make channels compatible with each other. However, transfers and moving
channels among bridges can result in otherwise compatible channels having
sub-optimal translation paths if the make compatible check is short
circuited. A result of forcing the reevaluation of channel compatibility
is that the asterisk.conf:transcode_via_slin and codecs.conf:genericplc
options take effect consistently now. It is unfortunate that these two
options are enabled by default and negate some of the benefits to the
changes in channel.c:ast_read() by forcing translation through signed
linear on a two party bridge.
* Improved the softmix bridge technology to better control the translation
of frames to the bridge. All of the incoming translation is now normally
handled by ast_read() instead of splitting any translation steps between
ast_read() and the slin factory. If any frame comes in with an unexpected
format then the translation path in ast_read() is updated for the next
frame and the slin factory handles the current frame translation.
This is the final patch in a series of patches aimed at improving
translation path choices. The other patches are on the following reviews:
https://reviewboard.asterisk.org/r/4600/https://reviewboard.asterisk.org/r/4605/
ASTERISK-24841 #close
Reported by: Matt Jordan
Review: https://reviewboard.asterisk.org/r/4609/
........
Merged revisions 434671 from http://svn.asterisk.org/svn/asterisk/branches/13
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@434672 65c4cc65-6c06-0410-ace0-fbb531ad65f3
When a channel enters the bridging system it is first made compatible with
the bridge and then the bridge technology makes the channel compatible
with the technology. For all but the DAHDI native and softmix bridge
technologies the make channel compatible with the bridge step is an
effective noop because the other technologies allow all audio formats.
For the DAHDI native bridge technology it doesn't matter because it is not
an initial bridge technology and chan_dahdi allows only one native format
per channel. For the softmix bridge technology, it is a noop at best and
harmful at worst because the wrong translation path could be setup if the
channel's native formats allow more than one audio format.
This is an intermediate patch for a series of patches aimed at improving
translation path choices.
* Removed code dealing with the unnecessary step of making the channel
compatible with the bridge.
ASTERISK-24841
Reported by: Matt Jordan
Review: https://reviewboard.asterisk.org/r/4600/
........
Merged revisions 434424 from http://svn.asterisk.org/svn/asterisk/branches/13
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@434430 65c4cc65-6c06-0410-ace0-fbb531ad65f3
When more than one call using the same codec type enters into a softmix bridge
and no audio is present for a channel the bridge optimizes the out frame by
using the same one for all channels with the same codec type. Unfortunately,
when that number (channels with same codec type) dropped to <= 1 the codec
was not dereferenced. At least not until all parties left the bridge. Thus in
the case of G.729 the license was not released. This patch ensures that the
codec is dereferenced immediately when the optimization no longer applies.
ASTERISK-24797 #close
Reported by: Luke Hulsey
Review: https://reviewboard.asterisk.org/r/4429/
........
Merged revisions 432174 from http://svn.asterisk.org/svn/asterisk/branches/11
........
Merged revisions 432175 from http://svn.asterisk.org/svn/asterisk/branches/13
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@432176 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Playing back a file to a channel in an ARI bridge would attempt to wait until
the playback concluded before returning. The method used involved signaling the
waiting thread in the ARI custom playback function.
The problem with this is that there were some corner cases that were not accounted for:
* If a bridge channel could not be found, then we never would attempt the playback but
would still attempt to wait for the playback to complete.
* If the bridge playfile action failed to queue, we would still attempt to wait for the
playback to complete.
* If the bridge playfile action were queued but some circumstance caused the playback
not to occur (the bridge dies, the channel is removed from the bridge), then we would
never be notified.
The solution to this is to move the waiting logic into the bridge code. A new bridge
API function is added to queue a synchronous action on a bridge. The waiting thread
is notified when the queued frame has been freed, either due to an error occurring
or due to successful playback. As a failsafe, the waiting thread has a 10 minute
timeout just in case there is a frame leak somewhere.
Review: https://reviewboard.asterisk.org/r/3338
........
Merged revisions 410673 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@410684 65c4cc65-6c06-0410-ace0-fbb531ad65f3
The crash is caused by a race condition when switching between native RTP
and softmix bridging technologies. In this situation, the bridging
technology is switched from native RTP to softmix, and then back to native
RTP fast enough that the softmix private data gets destroyed before the
softmix mixing thread gets started.
Thanks to Kinsey Moore for the crash analysis.
* Fix race condition when starting the softmix mixing thread and switching
to another bridge technology.
(closes issue ASTERISK-22678)
Reported by: John Bigelow
Patches:
jira_asterisk_22678_v12.patch (license #5621) patch uploaded by rmudgett
Tested by: John Bigelow
........
Merged revisions 400849 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@400850 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Channel snapshots have string representations of the channel's native formats.
Prior to this change, the format strings were re-created on ever channel snapshot
creation. Since channel native formats rarely change, this was very wasteful.
Now, string representations of formats may optionally be stored on the ast_format_cap
for cases where string representations may be requested frequently. When formats
are altered, the string cache is marked as invalid. When strings are requested, the
cache validity is checked. If the cache is valid, then the cached strings are copied.
If the cache is invalid, then the string cache is rebuilt and copied, and the cache
is marked as being valid again.
Review: https://reviewboard.asterisk.org/r/2879
........
Merged revisions 400356 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@400363 65c4cc65-6c06-0410-ace0-fbb531ad65f3
This patch renames the bridging* files to bridge*. This may seem pedantic
and silly, but it fits better in line with current Asterisk naming conventions:
* channel is not "channeling"
* monitor is not "monitoring"
etc.
A bridge is an object. It is a first class citizen in Asterisk. "Bridging" is
the act of using a bridge on a set of channels - and the API that fulfills that
role is more than just the action.
(closes issue ASTERISK-22130)
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@395378 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Created a native_dahdi bridging technology for use with the new bridging
API.
The new bridging technology is part of the chan_dahdi channel driver
because it is very specific to that driver. Rather than include the new
code directly into chan_dahdi.c the new bridge technology is in its own
file and linked into chan_dahdi.so. A large part of this change is the
mechanical process of moving declarations around so chan_dahdi.c can be
split up into more files later.
* Changed the bridging core to pass NULL frames into the channel
technologies instead of discarding them. The channel technologies may
need the proding to determine if their configuration is still valid.
(closes issue ASTERISK-21886)
Reported by: Matt Jordan
Review: https://reviewboard.asterisk.org/r/2681/
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@395154 65c4cc65-6c06-0410-ace0-fbb531ad65f3
* Extract a useful routine from the softmix bridge technology for other
technologies. Make other technologies use it if they can.
* Made native and 1-1 bridges write to all parties if the bridge channel
writing the frame into the bridge is NULL. Softmix will also do the same
for frame types that make sense.
* Tweak the bridge write routine return value meaning and adjust the
bridge technologies to match.
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@392514 65c4cc65-6c06-0410-ace0-fbb531ad65f3
The bridge frame queue functions need to return an error status if the
frame failed to be queued because of an error condition. The main calls
that needed to return the status are:
ast_bridge_channel_queue_action_data() and
ast_bridge_channel_write_action_data(). The other return changes are
ripple effects.
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@392435 65c4cc65-6c06-0410-ace0-fbb531ad65f3
A three party bridge uses the softmix bridging technology. This
technology has a dedicated thread used to perform the analog mixing. When
one of these parties leaves the bridge, the bridge technology is changed
from the softmix technology to a two-party mixing technology. Changing
technologies is done by removing channels from the old technology and
adding them to the new technology. Since the remaining channels do not
leave the bridge, the softmix mixing thread could continue to process all
channels in the bridge. If the bridge code is not able to start
destruction of the softmix technology before the softmix mixing thread
wakes up, a crash happens.
* Added a stop technology callback that technologies can use to request
any helper threads to stop in preparation for being destroyed.
(closes issue AST-1156)
Reported by: John Bigelow
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@390975 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Breaks many things until they can be reworked. A partial list:
chan_agent
chan_dahdi, chan_misdn, chan_iax2 native bridging
app_queue
COLP updates
DTMF attended transfers
Protocol attended transfers
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@389378 65c4cc65-6c06-0410-ace0-fbb531ad65f3
* softmix_bridge_thread() was redundantly initializing an 8K buffer.
* Promoted a debug message to a warning in multiplexed_add_or_remove().
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@378858 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Currently, if an acknowledgement of a timer fails Asterisk will not realize
that a serious error occurred and will continue attempting to use the timer's
file descriptor. This can lead to situations where errors stream to the
CLI/log file. This consumes significant resources, masks the actual problem
that occurred (whatever caused the timer to fail in the first place), and
can leave channels in odd states.
This patch propagates the errors in the timing resource modules up through
the timer core, and makes users of these timers handle acknowledgement
failures. It also adds some defensive coding around the use of timers
to prevent using bad file descriptors in off nominal code paths.
Note that the patch created by the issue reporter was modified slightly for
this commit and backported to 1.8, as it was originally written for
Asterisk 10.
Review: https://reviewboard.asterisk.org/r/2178/
(issue ASTERISK-20032)
Reported by: Jeremiah Gowdy
patches:
jgowdy-timerfd-6-22-2012.diff uploaded by Jeremiah Gowdy (license 6358)
........
Merged revisions 375893 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........
Merged revisions 375894 from http://svn.asterisk.org/svn/asterisk/branches/10
........
Merged revisions 375895 from http://svn.asterisk.org/svn/asterisk/branches/11
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@375896 65c4cc65-6c06-0410-ace0-fbb531ad65f3
follow_talker mode originally echoed the same video stream
to all participants. As the primary talker switched around, the
video stream would result in the talker seeing themselves. Now
the primary talker sees the last person who was talking rather than
themselves.
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@327640 65c4cc65-6c06-0410-ace0-fbb531ad65f3