GitLab

Filters were reverted for the release so the binaries wouldn't include
libavfilter (which wasn't yet used).

This reverts commit 1e6a4716.

Change-Id: Ib9790a544b442066265b3fe1ea96313be35eff62
Reviewed-by: Sebastien Blin <sebastien.blin@savoirfairelinux.com>

Change-Id: I0c696b635bd7940d9d9803ecb577ef436943ded3
Reviewed-by: Sebastien Blin <sebastien.blin@savoirfairelinux.com>

As contributed by Uwe Kleine-König <uwe@kleine-koenig.org>:
 - https://bugs.debian.org/884249

The compat code needs atleast upnp 1.6.25

Change-Id: Ia5e50dc362a64b4acd252dac5137e2aba4795064

Change-Id: I04466980b941dd943637b1a0002997de7a77114f
Reviewed-by: Andreas Traczyk <andreas.traczyk@savoirfairelinux.com>

FFmpeg may check channels instead of channel_layout, so set both.

Change-Id: I25094b104529e7d5ce3476d580a43fc5a445f48d

Refactors the MediaEncoder class so it can be reused elsewhere in the
codebase.

Concretely, MediaEncoder now supports multiple streams and file formats.

Adds a unit test for MediaEncoder.

Change-Id: I27d6f2e8538809e01fef5c35978a0c35dcf8cae4
Reviewed-by: Sebastien Blin <sebastien.blin@savoirfairelinux.com>

Refactors out the common code from video and audio encoding/flushing
into a single method. This reuses code instead of duplicating it.

The method is public in the hopes of being able to reuse MediaEncoder
elsewhere in the codebase.

Allow caller to specify the stream index for the future case that the
AVFormatContext contains multiple streams (recording video and audio).

Change-Id: Ieae52bc453ef66d141a40819ca10fedbd38d9a86
Reviewed-by: Sebastien Blin <sebastien.blin@savoirfairelinux.com>

Change-Id: I7af53fd3b0f76a9652c8fb4d1faceffefe3a4d1f
Reviewed-by: Sebastien Blin <sebastien.blin@savoirfairelinux.com>

This reverts commit ea2933d4.

Change-Id: I624c07255f60186a0d1e6610a58956775f91b342

Change-Id: Ia84ed72a0ba48778fcdef404d87fd35f717f64ab

Adds dependency to libavfilter, which provides many useful utilities
such as hardware encoding, mixing audio or rotating video.

Access to libavfilter is done through the MediaFilter class, which takes
a filter graph string and AVFrames as input and outputs filtered
AVFrames.

Adds unit tests for MediaFilter.

Change-Id: I0bf3869a9f19ad9cbbe1fc5e2d37eb83ece66ccc
Reviewed-by: Sebastien Blin <sebastien.blin@savoirfairelinux.com>

Change-Id: I6ab8a0078f7b67b2342859e216ab4ff02d721736
Reviewed-by: Sebastien Blin <sebastien.blin@savoirfairelinux.com>

Allows to get PJSIP logs on Android, where stdout is not logged.

Change-Id: I6db7734dd608f3c8085bc7ed6cf0341ff3c9c787
Reviewed-by: Sebastien Blin <sebastien.blin@savoirfairelinux.com>

Currently, when a file is sent, the peer needs to accept the transfer.
So, connections with this peer will be opened until the peer accepts or
not the file. Since the multi-device support, it's a problem because
the peer has to accept/refuse the transfer on each devices.

To avoid to leave unused connections while we are using the turn,
we need to close these connections. This patch add a timeout to
outgoing file transfers. So, after 10 minutes with the status
awaiting peer, the connection is closed and the transfer canceled.

NOTE: the peer will still see the awaiting_host status even if the
connection is closed, but it's a known bug and it will need a major
change of the blocking startNewFile function in FtpServer.

Change-Id: Ic6e3a2bfebad92cc5c64f5ac2355c3c3765752c9
Reviewed-by: Philippe Gorley <philippe.gorley@savoirfairelinux.com>

With the current implementation, when a user send a file to a contact
who got several devices connected, a request will be sent to all of
these devices and the first which answer will get the transfer. So,
a user can't know on which device they will get the file.

With this patch, in the same configuration the peer will receives
the file on all of its devices and add the possibility to accept
or refuse the file. To avoid to see multiple file transfers because
a peer has several devices, I introduced a OptimisticMetaOutgoingInfo
which represents the best current state of the transfer (for example,
if one subtransfer is ongoing and one failed (cause refused) it will
show the outgoing one).

This is an ongoing work, in the near future we must:
+ Give the ability to clients to see the real status of each
subtransfer (with the ability to cancel/retry transfer by devices)
+ Add a timer to close awaiting transfers to avoid to let a unused
connection for too long.

Change-Id: I84eb243bff2bfdc087a83dd7eced45c361f27d16
Reviewed-by: Philippe Gorley <philippe.gorley@savoirfairelinux.com>

Second attempt to fix a heap-use-after-free found with ASAN.
First attempt is broken (commit b2e56f84)

Change-Id: I923963ec0a25d768c69c76d506d5bc5dedb6181f
Reviewed-by: Sebastien Blin <sebastien.blin@savoirfairelinux.com>

This reverts commit b2e56f84.

Change-Id: I832bcfe0e2994e82d33951a6499ba43345db1a69

Change-Id: I86d2bfcf88a2a2b0eac50e4bb2b1c74359f84f76

FFmpeg has the convenient macro av_err2str, but it cannot compile using
g++ (https://ffmpeg.org/pipermail/libav-user/2013-January/003440.html).
Adds an equivalent function for C++.

Change-Id: Ie9eb2c0dfe68c58fd6a5ff82a3729ff63b736a69

it points to an element of the servers_ map. When we call
servers_.erase(it), we free this element. Currently this is done
before calling connectedPeers_.erase(it->first.second) and
turnEndpoints_.erase(it->first.second) hence resulting in heap use
after free.

In this patch we change the order of operation so that
servers_.erase(it) is done at the end.

Change-Id: Ie1b87ebc83e39db189ed651690e9e1dce3496b41

Change-Id: Icce9d2e5a126421e853463a05538d4e75184ab75
Reviewed-by: Andreas Traczyk <andreas.traczyk@savoirfairelinux.com>

Change-Id: I03397f2a9c558b2969f4ce1d177b4905aff6b584

Before this patch a message got a failed status after 3 minutes
(3 retries x 1 min). Moreover, if no devices was found, the message
is keeped in the cache and wait for the next reboot. So it results in
extremely long sending status, which is not really useful for the user.

With this patch, the message fails to send after approximately one
minute, if the user is connected and we do not receive an answer
or if no device is found.

Change-Id: I41171be2ebfebc424f96b4a4d42c360ad35ede86

Do not delete an object of derived class type through a pointer to
its base class type that has a non-virtual destructor. Instead, the
base class should be defined with a virtual destructor. Deleting an
object through a pointer to a type without a virtual destructor
results in undefined behavior.

Change-Id: I5a9f1ade5d67056f9ebc2865bd3c1c17fe197fcf
Reviewed-by: Sebastien Blin <sebastien.blin@savoirfairelinux.com>

6f15399e broke the multi-transfer
with the same contact because it closes previous transfers. We need
to keep correct turn connections.

Change-Id: I0cb9942417b10cc1aaa7d6879613c685dc5de2ad
Reviewed-by: Philippe Gorley <philippe.gorley@savoirfairelinux.com>

When two persons are trying to call each others, we can assume they
want to talk together. So, the daemon should automatically answer.
It avoids some weird case where they both see the outgoing call view,
but cannot answer to the incoming call.

There is several scenarios.
1. Both are trying to make a video call, in this case, the daemon
with the incoming call from the higher ring_id should answer and drop
its outgoing call resulting in a video call.
2. Both are trying to make an audio call, same behavior as scenario 1,
resulting in an audio call.
3. One device is trying to make a video call, the other an audio call,
audio call is prioritary here, and the daemon with an incoming audio only
call should answer and drop the outgoing video call.

Change-Id: Id1e639448243c19bf492f2d057496cab6d05de8c
Reviewed-by: Philippe Gorley <philippe.gorley@savoirfairelinux.com>

Change-Id: Ic5ade415fda8f4adbf75ec55d92119f3611bab5a

For a lot of reasons, the connection to the turn server can be lost,
we should relaunch a connection between the daemon and the turn server
to avoid to block a future transfer.

Change-Id: I02fea8a220d7d42fcc563dc5c97436b4fd55c618

RFC 6156 propose to use REQUESTED-ADDRESS-FAMILY to be able to
communicate beetween a IPv6 network and a IPv4 network. Because
this parameter is negociated during the allocation we should use
two control channels to be able to communicate with all address
families.

Change-Id: Ie1659b20c4c7170697e4ad2949a519b0e87e35ae

This also fix a deadlock when calling this function inside a callback
called when no peer has been found.

Change-Id: I0a8af8fc7ccfa638fdae9da957275a578b43879e

Change-Id: Id151ff3a40d01871b8a59349465aa40e169822c0

Change-Id: I4512dad14927f01f56d6bf4472c3e504157f1b76

Prevent deadlock by stop waiting on the
condition variable after a second.

Change-Id: If707fac389615333d1074b0973d693a84c9a974e

A buffer can arrive before the ATTACH_INPUT or ATTACH_OUTPUT, we
should store this buffer and inject it as soon as a stream is attached
or we might have a blocked data transfer.

Change-Id: I110f7edc3d3e93fc4a9349c9e8a6d6aed699984b

Use turn server defined by the user instead of the hardcoded
turn.ring.cx

Change-Id: I47452dfc2a531af2f5ebda3e138d4b31e694631a
Reviewed-by: Philippe Gorley <philippe.gorley@savoirfairelinux.com>

Add libressl dependency to compile restbed with ssl support. Also,
this patch removes a unused file.
Bump asio to works with libressl instead of openssl.

Change-Id: I5a90dba97693ccda58899f69080e545bdea926d9
Signed-off-by: Sébastien Blin <sebastien.blin@savoirfairelinux.com>

nullptr access to detect if it an audio call only due to missing remote sdp.
In such case the incoming call is considered as audio only.

Change-Id: I70e5d45aaa047f4683f4277536eb851e5b222e6b

FFmpeg docs state that if avcodec_send_packet returns EAGAIN,
avcodec_receive_frame must be called; it is not an error.

Change-Id: I5e84a8a0103fe868056915b9c4ea2fc8d090f7ee

Change-Id: I6fcb702cedd6204b88ac9b6a6437a18a4728cc12
Reviewed-by: Sebastien Blin <sebastien.blin@savoirfairelinux.com>

* respect timeout in waitForRawData with UDP
* handshake until GnuTLS tells us it's time to stop or limit is reached

Change-Id: I349d90706e8e024d12c32f5d66d1c6850dd7c23c
Reviewed-By: Sebastien Blin <sebastien.blin@savoirfairelinux.com>