platform/upstream/gstreamer.git
8 years agodvb/parsechannels: parse DVB-T hierarchy from v5 config file
Reynaldo H. Verdejo Pinochet [Sat, 8 Oct 2016 12:20:47 +0000 (05:20 -0700)]
dvb/parsechannels: parse DVB-T hierarchy from v5 config file

Avoids relying on the H/W default when the information
can be extracted from the configuration file.

8 years agodvb/parsechannels: parse DVB-T lp and hp code rates
Reynaldo H. Verdejo Pinochet [Sat, 8 Oct 2016 11:55:20 +0000 (04:55 -0700)]
dvb/parsechannels: parse DVB-T lp and hp code rates

8 years agompegts: Also clear packetizer on TIME DISCONT
Edward Hervey [Sat, 8 Oct 2016 12:08:06 +0000 (14:08 +0200)]
mpegts: Also clear packetizer on TIME DISCONT

When dealing with TIME-based input, the incoming stream could have
potentially changed completely.

In order to check whether it did or not, we need to re-check all sections
(PAT, PMT...). If it didn't, we will keep using the existing streams/pad,
and if it did we will act as if there was a program switch.

Fixes HLS streaming with decodebin3/playbin3

8 years agoadaptivedemux: Calculate values before queue2
Edward Hervey [Sun, 2 Oct 2016 07:34:51 +0000 (09:34 +0200)]
adaptivedemux: Calculate values before queue2

In order to calculate the *actual* bitrate for downloading a fragment
we need to take into account the time since we requested the fragment.

Without this, the bitrate calculations (previously reported by queue2)
would be biased since they wouldn't take into account the request latency
(that is the time between the moment we request a specific URI and the
moment we receive the first byte of that request).

Such examples were it would be biased would be high-bandwith but high-latency
networks. If you download 5MB in 500ms, but it takes 200ms to get the first
byte, queue2 would report 80Mbit/s (5Mb in 500ms) , but taking the request
into account it is only 57Mbit/s (5Mb in 700ms).

While this would not cause too much issues if the above fragment represented
a much longer duration (5s of content), it would cause issues with short
ones (say 1s, or when doing keyframe-only requests which are even shorter)
where the code would expect to be able to download up to 80Mbit/s ... whereas
if we take the request time into account it's much lower (and we would
therefore end up doing late requests).

Also calculate the request latency for debugging purposes and further
usage (it could allow us to figure out the maximum request rate for
example).

https://bugzilla.gnome.org/show_bug.cgi?id=733959

https://bugzilla.gnome.org/show_bug.cgi?id=772330

8 years agoopencv: Fix memory leak by gst_caps_to_string
Jimmy Ohn [Thu, 6 Oct 2016 06:54:00 +0000 (15:54 +0900)]
opencv: Fix memory leak by gst_caps_to_string

gst_caps_to_string function returned allocated memory.
So, It should be freed using g_free function.

https://bugzilla.gnome.org/show_bug.cgi?id=772499

8 years agoglwindow: don't use g_thread_join() to join the navigation thread
Matthew Waters [Wed, 5 Oct 2016 07:32:09 +0000 (18:32 +1100)]
glwindow: don't use g_thread_join() to join the navigation thread

Using g_thread_join() in _finalize() handlers may result in a deadlock
joining the current thread when the last reference is held by a signal
handler.

e.g.:

error 'Resource deadlock avoided' during 'pthread_join (pt->system_thread, NULL)'

The backtrace looks like this:
[...]
g_thread_join ()
gst_gl_window_finalize ()
gst_gl_window_x11_finalize ()
g_object_unref ()
g_value_unset ()
g_signal_emit_valist ()
g_signal_emit ()
gst_gl_window_send_mouse_event ()
gst_gl_window_mouse_event_cb ()
g_main_dispatch ()
[..]
g_main_loop_run ()
gst_gl_window_navigation_thread ()
g_thread_proxy ()
start_thread ()
clone ()

8 years agogluploadelement: fix leak of upload library object
Matthew Waters [Wed, 5 Oct 2016 07:28:48 +0000 (18:28 +1100)]
gluploadelement: fix leak of upload library object

When only linking the element, the upload object will be created from
_transform_caps() but will never be unreffed as the only case is in _stop().

Add an unref if non-NULL to a new finalize handler for this case.

8 years agovkdisplay: hold a weakref on the list of windows
Matthew Waters [Wed, 5 Oct 2016 01:11:19 +0000 (12:11 +1100)]
vkdisplay: hold a weakref on the list of windows

It's possible that the window may have been destroyed when a winsys
event comes in for it.

Fixes an assertion in make -C tests/check generic/states.check

8 years agovkdevice: only unref the gstcontext query if non-NULL
Matthew Waters [Wed, 5 Oct 2016 00:37:04 +0000 (11:37 +1100)]
vkdevice: only unref the gstcontext query if non-NULL

Fixes an assertion in make -C tests/check generic/states.check

8 years agotests/gl: make sure all GL commands are executed on the GL thread
Matthew Waters [Thu, 29 Sep 2016 01:45:57 +0000 (11:45 +1000)]
tests/gl: make sure all GL commands are executed on the GL thread

e.g. the final glGetError() must also be completed on the GL thread

8 years agogl/win32: remove egl implementation
Matthew Waters [Wed, 28 Sep 2016 08:04:21 +0000 (18:04 +1000)]
gl/win32: remove egl implementation

It hasn't been used in ages and is dead code.

8 years agotsdemux: Add conditions to identify ATSC EAC3
Wonchul Lee [Tue, 13 Sep 2016 13:07:05 +0000 (22:07 +0900)]
tsdemux: Add conditions to identify ATSC EAC3

https://bugzilla.gnome.org/show_bug.cgi?id=770528

8 years agodvb/parsechannels: make parsing errors delsys-neutral
Reynaldo H. Verdejo Pinochet [Mon, 26 Sep 2016 22:23:52 +0000 (15:23 -0700)]
dvb/parsechannels: make parsing errors delsys-neutral

DVB is a group of delivery standards but we support
several others. While at this, make message formating
more congruent.

8 years agomeson: Setup pre-commit hooks when configuring
Thibault Saunier [Fri, 30 Sep 2016 14:35:41 +0000 (11:35 -0300)]
meson: Setup pre-commit hooks when configuring

8 years agomeson: update version
Tim-Philipp Müller [Fri, 30 Sep 2016 10:41:34 +0000 (11:41 +0100)]
meson: update version

8 years agoplayer: Remove Return documentation for function returning void
Sebastian Dröge [Fri, 30 Sep 2016 10:42:06 +0000 (13:42 +0300)]
player: Remove Return documentation for function returning void

8 years agoRelease 1.9.90
Sebastian Dröge [Fri, 30 Sep 2016 10:02:55 +0000 (13:02 +0300)]
Release 1.9.90

8 years agoUpdate .po files
Sebastian Dröge [Fri, 30 Sep 2016 09:19:32 +0000 (12:19 +0300)]
Update .po files

8 years agopo: Update translations
Sebastian Dröge [Fri, 30 Sep 2016 08:44:47 +0000 (11:44 +0300)]
po: Update translations

8 years agowaylandsink: Actually use buffer pool config after setting it up
Arun Raghavan [Thu, 29 Sep 2016 16:12:24 +0000 (21:42 +0530)]
waylandsink: Actually use buffer pool config after setting it up

CID: 1373420

8 years agofix for https://bugzilla.gnome.org/show_bug.cgi?id=771871
Sergey Mamonov [Thu, 29 Sep 2016 10:36:11 +0000 (13:36 +0300)]
fix for https://bugzilla.gnome.org/show_bug.cgi?id=771871

8 years agoa2dpsink: Drop some dead-code
Arun Raghavan [Thu, 29 Sep 2016 16:15:50 +0000 (21:45 +0530)]
a2dpsink: Drop some dead-code

Left over from the last cleanup.

CID: 1373418

8 years agofdkaacenc: fix accessing freed memory
Vincent Penquerc'h [Thu, 29 Sep 2016 13:32:15 +0000 (14:32 +0100)]
fdkaacenc: fix accessing freed memory

The buffer data is not always copied in _Fill, and will be
read in _DecodeFrame. We unmap at the end of the function,
whether we get there via failure or early out, and keep a
ref to the buffer to ensure we can use it to unmap the
memory even after _finish_frame is called, as it unrefs
the buffer.

Note that there is an access beyond the allocated buffer,
which is only apparent when playing from souphttpsrc (ie,
not from filesrc). This appears to be a bug in the bit
reading code in libfdkaac AFAICT.

https://bugzilla.gnome.org/show_bug.cgi?id=772186

8 years agofdkaacdec: avoid memory corruption on decoding error
Vincent Penquerc'h [Thu, 29 Sep 2016 13:31:37 +0000 (14:31 +0100)]
fdkaacdec: avoid memory corruption on decoding error

The buffer size is expected to be in multiples of the sample size,
not in bytes.

https://bugzilla.gnome.org/show_bug.cgi?id=772186

8 years agofdkaacenc: fix buffer leak
Vincent Penquerc'h [Thu, 29 Sep 2016 13:29:46 +0000 (14:29 +0100)]
fdkaacenc: fix buffer leak

https://bugzilla.gnome.org/show_bug.cgi?id=772186

8 years agofdkaacenc: set framed=true on src caps
Vincent Penquerc'h [Wed, 28 Sep 2016 08:44:58 +0000 (09:44 +0100)]
fdkaacenc: set framed=true on src caps

This fixes muxing in MPEG TS.

https://bugzilla.gnome.org/show_bug.cgi?id=772108

8 years agogl/dispmanx: Implements set_render_rectangle to adjust the position of window
Gwang Yoon Hwang [Thu, 5 May 2016 06:53:57 +0000 (15:53 +0900)]
gl/dispmanx: Implements set_render_rectangle to adjust the position of window

We cannot set the x, y coordinate of the video frame at the dispmanx at
this point. We need to teach dispmanx backend to understand about
set_render_rectangle API to draw a video with other UI.

This patch keeps the current behavior which places video frame at the
center of the display if there is no set_render_rectangle call to the
dispmanx window.

https://bugzilla.gnome.org/show_bug.cgi?id=766018

8 years agogl: add necessary padding bytes to all public structs
Matthew Waters [Wed, 28 Sep 2016 05:20:25 +0000 (15:20 +1000)]
gl: add necessary padding bytes to all public structs

8 years agofdkaacenc: set transmux on the fdkaac lib
Vincent Penquerc'h [Tue, 27 Sep 2016 15:17:00 +0000 (16:17 +0100)]
fdkaacenc: set transmux on the fdkaac lib

Not doing so will fail to decode in a simple fdkaacenc ! fdkaacdec
pipeline, though would work if this goes through a file.

https://bugzilla.gnome.org/show_bug.cgi?id=772067

8 years agofdkaacdec: do not error out of out of sync return
Vincent Penquerc'h [Tue, 27 Sep 2016 15:15:44 +0000 (16:15 +0100)]
fdkaacdec: do not error out of out of sync return

The docs say we should continue feeding in data and decoding

https://bugzilla.gnome.org/show_bug.cgi?id=772067

8 years agofdkaac: fix error with AOT_MP2_AAC_LC removed from libfdkaac API
Vincent Penquerc'h [Tue, 27 Sep 2016 08:08:51 +0000 (09:08 +0100)]
fdkaac: fix error with AOT_MP2_AAC_LC removed from libfdkaac API

AOT_MP2_AAC_LC is a "pseudo AOT" which got removed after 0.1.4,
and maps to AOT_AAC_LC.

Remove mpegversion 2 from th caps to match.

https://bugzilla.gnome.org/show_bug.cgi?id=772067

8 years agomeson: bluez: fix build
Tim-Philipp Müller [Mon, 26 Sep 2016 15:30:02 +0000 (16:30 +0100)]
meson: bluez: fix build

8 years agodecklink: Update to SDK 10.8
Sebastian Dröge [Mon, 26 Sep 2016 13:38:04 +0000 (16:38 +0300)]
decklink: Update to SDK 10.8

OSX and Linux only, Windows needs someone on Windows to convert the .idl
files to C source code.

8 years agofdkaac: fix mixup setting interleaved output
Olivier Crête [Mon, 26 Sep 2016 10:41:42 +0000 (11:41 +0100)]
fdkaac: fix mixup setting interleaved output

https://bugzilla.gnome.org/show_bug.cgi?id=770455

8 years agoplayer: Only report the initial duration if the query was successful
Sebastian Dröge [Mon, 26 Sep 2016 10:26:36 +0000 (13:26 +0300)]
player: Only report the initial duration if the query was successful

8 years agoh265parse: Fix parsing of multiple NALs in the hvcC box
Kyungyong Kim [Mon, 26 Sep 2016 02:15:25 +0000 (11:15 +0900)]
h265parse: Fix parsing of multiple NALs in the hvcC box

https://bugzilla.gnome.org/show_bug.cgi?id=771970

8 years agobluez: dist new header file
Tim-Philipp Müller [Sun, 25 Sep 2016 16:38:15 +0000 (17:38 +0100)]
bluez: dist new header file

Fixes distcheck.

8 years agoa2dpsink: Don't try to set device caps on avdtpsink
Arun Raghavan [Sat, 24 Sep 2016 17:09:38 +0000 (22:39 +0530)]
a2dpsink: Don't try to set device caps on avdtpsink

We can't actually configure the transport, so we should only be working
with whatever we get.

8 years agoavdtpsink: Chain event handling up to basesink
Arun Raghavan [Fri, 23 Sep 2016 11:01:50 +0000 (16:31 +0530)]
avdtpsink: Chain event handling up to basesink

8 years agobluez: Clean up and make a2dpsink functional again
Arun Raghavan [Fri, 23 Sep 2016 10:34:43 +0000 (16:04 +0530)]
bluez: Clean up and make a2dpsink functional again

Was crashing, and had a lot of cruft (like a capsfilter and a fakesink)
that was no longer required.

8 years agoavdtpsrc: Add support for AVRCP metadata
Arun Raghavan [Thu, 15 Sep 2016 12:19:15 +0000 (17:49 +0530)]
avdtpsrc: Add support for AVRCP metadata

Metadata from AVRCP is emitted as tags, and the duration from AVRCP is
used in queries by avdtpsrc.

8 years agoopenjpegdec: Properly offset and shift for all formats when converting
Sebastian Dröge [Sat, 24 Sep 2016 14:49:10 +0000 (10:49 -0400)]
openjpegdec: Properly offset and shift for all formats when converting

And while at it, also clean up some code.

8 years agojpeg2000parse: Don't pass NULL strings to the colorspace / sampling parsing functions
Sebastian Dröge [Sat, 24 Sep 2016 14:05:16 +0000 (10:05 -0400)]
jpeg2000parse: Don't pass NULL strings to the colorspace / sampling parsing functions

They assert on NULL.

8 years agowaylandsink: Don't leak GValues in getcaps
Nicolas Dufresne [Fri, 23 Sep 2016 19:49:21 +0000 (15:49 -0400)]
waylandsink: Don't leak GValues in getcaps

8 years agowaylandsink: Update our window size on configure event
Nicolas Dufresne [Thu, 22 Sep 2016 22:55:23 +0000 (18:55 -0400)]
waylandsink: Update our window size on configure event

This is specific to when the waylandsink is not being embedded. In
this patch we pass the render lock to the window so it can safely
call gst_wl_window_set_render_rectangle() with the new size.

https://bugzilla.gnome.org/show_bug.cgi?id=722343

8 years agowaylandsink: Remove atomic operation
Nicolas Dufresne [Thu, 22 Sep 2016 22:53:22 +0000 (18:53 -0400)]
waylandsink: Remove atomic operation

We already take the render lock from the wlqueue thread in some other
place which indicates that there is no use of this atomic instead of
a proper locking mechanism.

8 years agowaylandsink: Do not pre-configure proposed pool
Nicolas Dufresne [Thu, 22 Sep 2016 19:37:41 +0000 (15:37 -0400)]
waylandsink: Do not pre-configure proposed pool

Upstream must configure the pool before using it. Pre-configuring
the proposed pool could hide bugs in upstream elements.

8 years agowaylandsink: Properly draw black border in absence of viewporter
Nicolas Dufresne [Thu, 22 Sep 2016 19:35:44 +0000 (15:35 -0400)]
waylandsink: Properly draw black border in absence of viewporter

When we don't have a viewporter (scaling support), we can't use the
1x1 scaleup image trick. Instead, we need to allocate a buffer with
the same size as the area that need to have black background.

8 years agodx9screencapsrc: throw error for invalid screen index
Florian Zwoch [Thu, 22 Sep 2016 08:55:03 +0000 (10:55 +0200)]
dx9screencapsrc: throw error for invalid screen index

Currently dx9screencapsrc prints a verbose warning in case the screen
index is out of range for the current number of detected monitors. This
value is then dropped.

However there is no initial indication (beside the console print) if it
worked or not. This may result in capturing an unwanted screen as it
would capture the last set index that was not rejected.

This patch sets the index regardless. Instead, the element throws an
error when it tries to run or getting caps for an invalid index.

https://bugzilla.gnome.org/show_bug.cgi?id=771817

8 years agogldisplay: also free the GWeakRef when removing dead contexts
Matthew Waters [Thu, 22 Sep 2016 08:19:36 +0000 (18:19 +1000)]
gldisplay: also free the GWeakRef when removing dead contexts

Otherwise we leak GWeakRef's.

Found with make -C tests/check libs/gstglcontext.valgrind

8 years agowayland/Makefile.am: add missing libs
Martin Kelly [Wed, 21 Sep 2016 22:51:52 +0000 (15:51 -0700)]
wayland/Makefile.am: add missing libs

libgstwayland is missing a dependency on $(GST_PLUGINS_BASE_LIBS), so add it.

https://bugzilla.gnome.org/show_bug.cgi?id=771794

8 years agowaylandsink: Destroy viewporter when done
Nicolas Dufresne [Wed, 21 Sep 2016 19:07:50 +0000 (15:07 -0400)]
waylandsink: Destroy viewporter when done

https://bugzilla.gnome.org/show_bug.cgi?id=738079

8 years agobuild: add missing $(GST_PLUGINS_BASE_LIBS)
Martin Kelly [Wed, 21 Sep 2016 18:33:18 +0000 (11:33 -0700)]
build: add missing $(GST_PLUGINS_BASE_LIBS)

The linker flags for several plugins (dashdemux, mssdemux, hlsdemux) are missing
$(GST_PLUGINS_BASE_LIBS), so add them so they can find -lgsttag and -lgstapp.

https://bugzilla.gnome.org/show_bug.cgi?id=771780

8 years agowaylandsink: Port to vmeta and GstVideoFrame
Nicolas Dufresne [Mon, 19 Sep 2016 16:11:59 +0000 (12:11 -0400)]
waylandsink: Port to vmeta and GstVideoFrame

This add support for non-standard strides to be used. Note that
some extra work is needed for multi-plane format which may have
a different GstMemory object per plane. This is not currently a
problem since SHM interface is limited to 1 memory.

8 years agowaylandsink: Don't propose multiple time the same buffer pool
Nicolas Dufresne [Mon, 19 Sep 2016 16:08:20 +0000 (12:08 -0400)]
waylandsink: Don't propose multiple time the same buffer pool

The buffer pool API does not allow multiple of owner. This otherwise
lead to error when renegotiation take place. Aso consider the
allocation query "need_pool" boolean.

8 years agowaylandsink: Promote debug trace into error trace
Nicolas Dufresne [Mon, 19 Sep 2016 16:06:17 +0000 (12:06 -0400)]
waylandsink: Promote debug trace into error trace

Otherwise those error may go unseen making debugging much
harder.

8 years agowaylandsink: Factor-out the pool creation
Nicolas Dufresne [Mon, 19 Sep 2016 16:04:52 +0000 (12:04 -0400)]
waylandsink: Factor-out the pool creation

8 years agowaylansink: Implement _show_frame()
Nicolas Dufresne [Fri, 16 Sep 2016 19:41:11 +0000 (15:41 -0400)]
waylansink: Implement _show_frame()

This fixes the property show-preroll-frame expose by the VideoSink base
class and removes redundant code.

8 years agortponviftimestamp: Update unit tests after changing default value of D-bit
Joakim Johansson [Wed, 21 Sep 2016 13:09:26 +0000 (15:09 +0200)]
rtponviftimestamp: Update unit tests after changing default value of D-bit

The D bit shall always be set to true for the first buffer.

https://bugzilla.gnome.org/show_bug.cgi?id=770221

8 years agodecklink: Print the Decklink API error return values in debug output in all places
Sebastian Dröge [Wed, 21 Sep 2016 13:31:23 +0000 (09:31 -0400)]
decklink: Print the Decklink API error return values in debug output in all places

8 years agowayland: fix distcheck some more
Tim-Philipp Müller [Tue, 20 Sep 2016 17:28:29 +0000 (18:28 +0100)]
wayland: fix distcheck some more

8 years agortponviftimestamp: Change default value of D-bit
Joakim Johansson [Tue, 13 Sep 2016 09:18:27 +0000 (11:18 +0200)]
rtponviftimestamp: Change default value of D-bit

The default value of D-bit is changed to TRUE so discontinuity
is set for initial request and seek request as well.

Only set the e_bit flag for the CUSTOM_DOWNSTREAM event if
a cached buffer exists.

https://bugzilla.gnome.org/show_bug.cgi?id=770221

8 years agoconfigure: fix --disable-external
Tim-Philipp Müller [Mon, 19 Sep 2016 09:18:22 +0000 (10:18 +0100)]
configure: fix --disable-external

conditional "USE_VULKAN" was never defined

8 years agoglcolorbalance: reconfigure on passthrough changes
Matthew Waters [Mon, 19 Sep 2016 06:54:43 +0000 (16:54 +1000)]
glcolorbalance: reconfigure on passthrough changes

Fixes an assertion when moving from passthrough to non-passthrough

Without an explicit reconfigure, glfiter won't have created the GL
resources such as the FBO, GL bufferpool, etc and basetransform will
allocate sysmem buffers instead.

8 years agoplayer: Remove some leftovers from external GstPlayer
Sebastian Dröge [Fri, 16 Sep 2016 11:43:50 +0000 (13:43 +0200)]
player: Remove some leftovers from external GstPlayer

8 years agowayland: fix distcheck when wayland-scanner is not installed
Tim-Philipp Müller [Sat, 17 Sep 2016 13:12:44 +0000 (14:12 +0100)]
wayland: fix distcheck when wayland-scanner is not installed

8 years agowaylandsink: Make wl_viewporter optional
Nicolas Dufresne [Thu, 15 Sep 2016 18:10:02 +0000 (14:10 -0400)]
waylandsink: Make wl_viewporter optional

This makes the viewporter interface optional. The end result is
obviously far from optimal, though it greatly helps testing on older
compostitors or gnome-wayland. We can make it strictly needed later when
this new interface get widely adopted.

8 years agowayland: Update from scaler to viewporter protocol
Scott D Phillips [Tue, 14 Jun 2016 23:34:35 +0000 (16:34 -0700)]
wayland: Update from scaler to viewporter protocol

Signed-off-by: Scott D Phillips <scott.d.phillips@intel.com>
https://bugzilla.gnome.org/show_bug.cgi?id=767671

8 years agoplayer: Fix unit test for minor API change
Sebastian Dröge [Thu, 15 Sep 2016 12:58:35 +0000 (14:58 +0200)]
player: Fix unit test for minor API change

8 years agoplayer: Remove boolean return from set_subtitle_uri()
Sebastian Dröge [Thu, 15 Sep 2016 11:15:22 +0000 (13:15 +0200)]
player: Remove boolean return from set_subtitle_uri()

It can't fail synchronously and is inconsistent with set_uri().

8 years agoplayer: Don't change state to PLAYING immediately if a seek is pending
Sebastian Dröge [Wed, 14 Sep 2016 16:07:55 +0000 (18:07 +0200)]
player: Don't change state to PLAYING immediately if a seek is pending

We first have to finish the seek (in PAUSED) and move to PLAYING once
the seek is actually finished (unless a new one is pending then).

8 years agoconfigure: Depend on gstreamer 1.9.2.1
Sebastian Dröge [Wed, 14 Sep 2016 09:30:48 +0000 (11:30 +0200)]
configure: Depend on gstreamer 1.9.2.1

8 years agodvbsuboverlay: map with READWRITE for overlay blend
Hyunjun Ko [Thu, 8 Sep 2016 02:57:52 +0000 (11:57 +0900)]
dvbsuboverlay: map with READWRITE for overlay blend

Just as in basetextoverlay [1], the frame to blend with the subtitles overlay
should be mapped with flags GST_MAP_READWRITE, because
gst_video_overlay_composition_blend() does both operations.

1. https://cgit.freedesktop.org/gstreamer/gst-plugins-base/commit/?id=1396f804

Signed-off-by: Víctor Manuel Jáquez Leal <vjaquez@igalia.com>
https://bugzilla.gnome.org/show_bug.cgi?id=771382

8 years agokmssink: update libdrm dependency to 2.4.55
Víctor Manuel Jáquez Leal [Tue, 13 Sep 2016 18:08:19 +0000 (20:08 +0200)]
kmssink: update libdrm dependency to 2.4.55

The latest libdrm feature supported by kmssink is the universal plane, which
was added in 2.4.55. Let's update the dependency to that version.

https://bugzilla.gnome.org/show_bug.cgi?id=770146

8 years agotsdemux: handle a NULL name in DVB short event descriptor
Graham Leggett [Sun, 11 Sep 2016 00:06:52 +0000 (00:06 +0000)]
tsdemux: handle a NULL name in DVB short event descriptor

Resolves the warning:

GStreamer-WARNING **: Trying to set NULL string on field 'title' on taglist.

https://bugzilla.gnome.org/show_bug.cgi?id=771198

8 years agoadaptivedemux: Also never have the SINK flag set on adaptivedemux
Sebastian Dröge [Mon, 12 Sep 2016 16:38:36 +0000 (18:38 +0200)]
adaptivedemux: Also never have the SINK flag set on adaptivedemux

8 years agoRevert "decklinkvideosink: Scale down scheduled frame times to milliseconds"
Sebastian Dröge [Mon, 12 Sep 2016 10:19:50 +0000 (12:19 +0200)]
Revert "decklinkvideosink: Scale down scheduled frame times to milliseconds"

This reverts commit 3b7e0d7de32f0728259d601daa13b9eeec3dd3ce.

It was a bug in the driver and is supposed to be fixed with 10.8 and newer.

8 years agoadaptivedemux: fix typo in new API
Thiago Santos [Sat, 10 Sep 2016 19:41:28 +0000 (16:41 -0300)]
adaptivedemux: fix typo in new API

Fixes supressed -> suppressed typo in previous commit

https://bugzilla.gnome.org/show_bug.cgi?id=770627

8 years agoadaptivedemux: prevent to propagate source flag to itself
Wonchul Lee [Mon, 5 Sep 2016 01:31:40 +0000 (10:31 +0900)]
adaptivedemux: prevent to propagate source flag to itself

https://bugzilla.gnome.org/show_bug.cgi?id=770627

8 years agoAutomatic update of common submodule
Jan Schmidt [Sat, 10 Sep 2016 10:51:27 +0000 (20:51 +1000)]
Automatic update of common submodule

From b18d820 to f980fd9

8 years agoAutomatic update of common submodule
Jan Schmidt [Fri, 9 Sep 2016 23:54:09 +0000 (09:54 +1000)]
Automatic update of common submodule

From 2c31690 to b18d820

8 years agomeson/gl: use separate deps for gl and glx
Matthew Waters [Fri, 9 Sep 2016 13:24:01 +0000 (23:24 +1000)]
meson/gl: use separate deps for gl and glx

e.g. passing with_gl_api=gles2 would still build the glx code but not be
linking against the libGL library which is where the glX* functions are
located and would result in a linker error.

Solved by checking for the libGL library if either opengl or glx may be
needed and then disabling the corresponding deps as requested.

8 years agomeson: add build files for vulkan
Matthew Waters [Fri, 9 Sep 2016 06:55:23 +0000 (16:55 +1000)]
meson: add build files for vulkan

8 years agomeson: add build files for the gtk plugin
Matthew Waters [Fri, 9 Sep 2016 06:36:03 +0000 (16:36 +1000)]
meson: add build files for the gtk plugin

8 years agomeson: add some starting build files for GL
Matthew Waters [Fri, 9 Sep 2016 01:08:53 +0000 (11:08 +1000)]
meson: add some starting build files for GL

Currently only works on linux with egl/glx + wayland/x11 but the general
principals have been layed out for adding the other GL platforms/winsys'.

8 years agokmssink: Scale up to the screen dimension
Nicolas Dufresne [Thu, 8 Sep 2016 15:23:57 +0000 (11:23 -0400)]
kmssink: Scale up to the screen dimension

In most display sink, the logic is to use as much as possible
of the given window. In this case, the window is the screen,
hence it's logical to scale up.

https://bugzilla.gnome.org/show_bug.cgi?id=767422

8 years agokmssink: Fix selection of source region
Nicolas Dufresne [Thu, 8 Sep 2016 15:21:09 +0000 (11:21 -0400)]
kmssink: Fix selection of source region

The source region was scaled for display before being passed
to drmModeSetPlane, which resulted in a portion of the video
being cropped. While when crop meta was present, the rectangle
was not centered since we where using unscaled width/height.

https://bugzilla.gnome.org/show_bug.cgi?id=767422

8 years agogstplayer: remove leftover update_interval api
Julien Isorce [Thu, 8 Sep 2016 07:40:59 +0000 (08:40 +0100)]
gstplayer: remove leftover update_interval api

old: gst_player_set_position_update_interval
new: gst_player_config_set_position_update_interval

https://bugzilla.gnome.org/show_bug.cgi?id=769144

8 years agovkupload: add video/x-raw to the VulkanBuffer uploader
Matthew Waters [Thu, 1 Sep 2016 09:46:31 +0000 (19:46 +1000)]
vkupload: add video/x-raw to the VulkanBuffer uploader

Allows upstream elements that use our memory to passthrough without copying.

8 years agovkutils: set the last element of the layer array to NULL
Matthew Waters [Thu, 1 Sep 2016 09:43:59 +0000 (19:43 +1000)]
vkutils: set the last element of the layer array to NULL

g_strv* functions require this to not read off the end of the array.

8 years agogl/examples: Really remove references to $(GST_PLUGINS_GL_*).
Alistair Buxton [Wed, 7 Sep 2016 19:11:55 +0000 (20:11 +0100)]
gl/examples: Really remove references to $(GST_PLUGINS_GL_*).

These seem to have been missed in the previous commit to this file.

https://bugzilla.gnome.org/show_bug.cgi?id=771023

8 years agogl: Remove duplicate GL_CFLAGS in Makefile.am.
Alistair Buxton [Wed, 7 Sep 2016 19:07:26 +0000 (20:07 +0100)]
gl: Remove duplicate GL_CFLAGS in Makefile.am.

https://bugzilla.gnome.org/show_bug.cgi?id=771021

8 years agokmssink: override stride if defined in driver
Víctor Manuel Jáquez Leal [Fri, 5 Aug 2016 16:17:32 +0000 (18:17 +0200)]
kmssink: override stride if defined in driver

Some kms drivers demands specific pitches over the ones calculated by
GstVideoInfo. For example, intel driver demands strides round up 64.

This patch queries the driver for the prefered pitch and overwrites it
in the pool's GstVideoInfo structure.

https://bugzilla.gnome.org/show_bug.cgi?id=768446

8 years agoaggregator: Use the event_full function for GstAggregatorPads
Thibault Saunier [Tue, 6 Sep 2016 19:05:53 +0000 (16:05 -0300)]
aggregator: Use the event_full function for GstAggregatorPads

Allowing us to tell GstPad why we are failing an event, which might
be because we are 'flushing' even if the sinkpad is not in flush state
at that point.

8 years agoqmlglsink: check qt_context_ first in GstQSGTexture::bind()
Haihua Hu [Tue, 6 Sep 2016 05:13:39 +0000 (13:13 +0800)]
qmlglsink: check qt_context_ first in GstQSGTexture::bind()

When start qmlglsink app, it will set NULL buffer to GstQSGTexture
in which case that qt_context_ will be a random value and cause
gst_gl_context_activate() fail.

https://bugzilla.gnome.org/show_bug.cgi?id=770925

8 years agobuild/vulkan: fix winsys detection based on the previously set variables
Matthew Waters [Tue, 6 Sep 2016 10:00:07 +0000 (20:00 +1000)]
build/vulkan: fix winsys detection based on the previously set variables

The X11/Wayland winsys checks weren't being used to enable the vulkan element.
Use them.

8 years agosrtpenc: prevent deadlock in create_session
Philippe Normand [Tue, 6 Sep 2016 09:02:06 +0000 (11:02 +0200)]
srtpenc: prevent deadlock in create_session

Temporarily release the object lock which is needed to post error
messages on the bus.

https://bugzilla.gnome.org/show_bug.cgi?id=770933

8 years agomeson: Fix building with meson 0.34 when using subprojects
Thibault Saunier [Tue, 6 Sep 2016 01:30:01 +0000 (22:30 -0300)]
meson: Fix building with meson 0.34 when using subprojects

When using subproject meson.source_root() returns the root
directory of the toplevel project which break build.

Upstream meson bug: https://github.com/mesonbuild/meson/issues/722

8 years agomeson: Bump version to 1.9.2
Thibault Saunier [Mon, 5 Sep 2016 15:22:17 +0000 (12:22 -0300)]
meson: Bump version to 1.9.2

8 years agoglupload: create the GstGLUpload object in ::transform_caps
Alessandro Decina [Mon, 5 Sep 2016 04:44:24 +0000 (14:44 +1000)]
glupload: create the GstGLUpload object in ::transform_caps

Previously it was created in the init function and destroyed in ::stop, which
lead to segfaults when reusing the element.

Now the upload object is created in ::transform_caps if it is NULL, which is the
earliest we need it. The other vfuncs already bail out if the upload object is
NULL, which means that negotiation wasn't done.