platform/upstream/gstreamer.git
4 years agovideorate: Fix changing `rate` property during playback
Thibault Saunier [Sun, 31 May 2020 04:27:14 +0000 (00:27 -0400)]
videorate: Fix changing `rate` property during playback

We need to take into account the base_ts to compute next_ts and it needs
to be updated on rate change.

This introduces `pending_rate` so that change rate is properly handled
in the streaming thread in a safe way.

Added tests

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/679>

4 years agotests:validate: Run all test using a simple fakesink
Thibault Saunier [Thu, 28 May 2020 17:42:22 +0000 (13:42 -0400)]
tests:validate: Run all test using a simple fakesink

`fakevideosink` is in -bad and thus not available here.

Update the expectation files as video metas are not negotiated anymore.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/677>

4 years agolibs: gl: silence gir compiler
Víctor Manuel Jáquez Leal [Fri, 29 May 2020 09:39:57 +0000 (11:39 +0200)]
libs: gl: silence gir compiler

There were a couple complains of the gir compiler on these gstgl
files.

* Added namespace to public macros, even if they are helpers.
* Removed a misused private tag

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/678>

4 years agovideoscale: reorder code to avoid indent missmatches
Michael Gruner [Thu, 23 Jan 2020 18:38:44 +0000 (12:38 -0600)]
videoscale: reorder code to avoid indent missmatches

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/548>

4 years agovideoscale: transform size sensitive metas
Michael Gruner [Thu, 23 Jan 2020 17:02:52 +0000 (11:02 -0600)]
videoscale: transform size sensitive metas

Currently, videoscale just drops all metas that have other tags
besides video. However videoscale wont change the colorspace or
the orientation of the video so metas tagged as such may be
copied safely. Additionaly, given that videoscale will change
the frame size, we invoke the meta transform implementation
to give it the opportunity to scale accordingly.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/548>

4 years agotests: audiosink: Test class extension struct
Seungha Yang [Wed, 22 Jan 2020 13:30:11 +0000 (22:30 +0900)]
tests: audiosink: Test class extension struct

Test a vfunc which belongs to GstAudioSinkExtension struct.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/547>

4 years agoaudiosink: Keep baseclass extensible
Seungha Yang [Tue, 21 Jan 2020 15:14:14 +0000 (00:14 +0900)]
audiosink: Keep baseclass extensible

Add a structure for future extension.

Fixes: https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/issues/716
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/547>

4 years agodecodebin: only emit 'drained' signal when top chain is drained
uno20001 [Wed, 27 May 2020 13:11:17 +0000 (15:11 +0200)]
decodebin: only emit 'drained' signal when top chain is drained

Without this, decodebin emits 'drained' multiple times which then
causes (uri)playbin to emit 'about-to-finish' multiple times for
for file types.

Fixes #751

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/673>

4 years agoaudioresample: Add new test that checks for downstream renegotiation
Sebastian Dröge [Wed, 27 May 2020 16:59:56 +0000 (19:59 +0300)]
audioresample: Add new test that checks for downstream renegotiation

This test always consumes 48kHz and outputs different sample rates based
on downstream renegotiation. Previously this would produce completely
wrong timestamps and not output all samples.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/670>

4 years agoaudioresample: Fix up test_live_switch
Sebastian Dröge [Wed, 27 May 2020 16:08:45 +0000 (19:08 +0300)]
audioresample: Fix up test_live_switch

Actually check that we get back all samples, which we didn't before
because no draining was happening. Also remove commented out 0.10 code
and related comments.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/670>

4 years agoaudioresample: Drain resampler on discontinuities
Sebastian Dröge [Tue, 26 May 2020 19:51:06 +0000 (22:51 +0300)]
audioresample: Drain resampler on discontinuities

Otherwise we would lose the last few samples when resetting the
resampler.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/670>

4 years agoaudioresample: Drain resampler and reset timestamp tracking on stream-start event too
Sebastian Dröge [Tue, 26 May 2020 19:49:44 +0000 (22:49 +0300)]
audioresample: Drain resampler and reset timestamp tracking on stream-start event too

And also reset timestamp tracking on EOS events as more data might come
afterwards with a new stream-start event. This keeps the code the same.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/670>

4 years agoaudioresample: Drain the resampler and reset timestamp tracking on caps changes
Sebastian Dröge [Tue, 26 May 2020 19:45:41 +0000 (22:45 +0300)]
audioresample: Drain the resampler and reset timestamp tracking on caps changes

Especially when changing the sample rate our timestamp tracking will be
completely off, but even otherwise we would usually lose the last few
samples if we don't drain here as the resampler gets reset if anything
but the sample rate changes.

This is usually not a problem as the first buffer after a caps event
usually has the discont flag set, but can cause problems if
 - the caps event is followed by a segment event, which then causes
   draining according to the new sample rate
 - the caps were changed because of rengotiation due to a reconfigure
   event and there is not discontinuity from upstream

In both cases we would output buffers with completely wrong timestamps.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/670>

4 years agovideo-anc: init type before g_once_init_enter
Stéphane Cerveau [Fri, 8 May 2020 09:19:24 +0000 (11:19 +0200)]
video-anc: init type before g_once_init_enter

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/675>

4 years agovideo-hdr: fix typo
Stéphane Cerveau [Fri, 8 May 2020 09:03:52 +0000 (11:03 +0200)]
video-hdr: fix typo

fix typo in video_hdr_ascii_string_to_unsigned
method name.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/675>

4 years agoglimagesink: Correct signals gtkdoc
Niels De Graef [Tue, 26 May 2020 16:51:03 +0000 (18:51 +0200)]
glimagesink: Correct signals gtkdoc

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/669>

4 years agovideorate: Fix buffer selection logic in reverse playback
Thibault Saunier [Mon, 25 May 2020 20:59:53 +0000 (16:59 -0400)]
videorate: Fix buffer selection logic in reverse playback

Stop comparing all timestamps from buffers that are before the segment
with the segment.stop and compare with the actual end times.

Comparing to segment.stop for all the buffers that where before
the segment.stop was incorrect and leading to consuming wrong buffers
and not respecting segment.stop, this is now properly tested.

Expectations for `reverse.10_to_1fps.validatetest` have been fixed to
take that into account and comparing the checksums of the sinkpad and
srcpad expectations makes pretty clear how wrong that was.

(we can see in the expectations that videotestsrc outputs an extra
buffer with pts == segment.stop and this one is now properly dropped
by videorate as bec7f4ad5ed4bb1abfb92db946e654338766562b aimed at
doing)

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/668>

4 years agovideorate: Factor out a method for the`max-duplication-time` property
Thibault Saunier [Mon, 25 May 2020 20:29:15 +0000 (16:29 -0400)]
videorate: Factor out a method for the`max-duplication-time` property

Sensibly simplifying gst_video_rate_transform_ip

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/668>

4 years agovideorate: Use CLOCK_TIME_IS_VALID instead of checking CLOCK_TIME_NONE
Thibault Saunier [Mon, 25 May 2020 18:48:04 +0000 (14:48 -0400)]
videorate: Use CLOCK_TIME_IS_VALID instead of checking CLOCK_TIME_NONE

Making it more consistency with the rest of the code.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/668>

4 years agovideorate: Factor out a method to reset mode
Thibault Saunier [Mon, 25 May 2020 18:33:54 +0000 (14:33 -0400)]
videorate: Factor out a method to reset mode

Working on simplifying gst_video_rate_transform_ip

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/668>

4 years agovideorate: Do not push an extra buffer on EOS when we are done pushing already
Thibault Saunier [Mon, 25 May 2020 16:31:32 +0000 (12:31 -0400)]
videorate: Do not push an extra buffer on EOS when we are done pushing already

There is no reason that when we have already pushed all the buffers in
a segment we push a new one on EOS

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/668>

4 years agoaudiotestsrc: Fix the way we compute EOS in reverse playback
Thibault Saunier [Sat, 23 May 2020 03:24:55 +0000 (23:24 -0400)]
audiotestsrc: Fix the way we compute EOS in reverse playback

In reverse playback we were not taking into account the current buffer
samples to check if we had reached EOS which was leading to a buffer
with PTS = CLOCK_TIME_NONE containing too many frames followed by a
useless buffer with pts=0 duration=0, and a g_critical issue in
gst_object_sync_values.

Also add a validate based test case.
Without that patch this is how the expectation fails:

``` diff
--- log-asink-sink-expected       2020-05-22 23:22:42.654384579 -0400
+++ log-asink-sink-actual  2020-05-22 23:29:35.671586380 -0400
@@ -27,5 +27,6 @@
 buffer: pts=0:00:00.058820861, due=0:00:00.023219955, flags=discont
 buffer: pts=0:00:00.035600907, due=0:00:00.023219954, flags=discont
 buffer: pts=0:00:00.012380952, due=0:00:00.023219955, flags=discont
-buffer: pts=0:00:00.000000000, due=0:00:00.012380952, flags=discont
+buffer: due=0:00:00.012380953, flags=discont
+buffer: pts=0:00:00.000000000, flags=discont
 event eos: (no structure)
 ```

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/667>

4 years agogl: egl: fix gtk-doc doc start code
Guillaume Desmottes [Wed, 13 May 2020 13:28:33 +0000 (15:28 +0200)]
gl: egl: fix gtk-doc doc start code

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/664>

4 years agogl: workaround gir warning
Guillaume Desmottes [Wed, 13 May 2020 12:50:22 +0000 (14:50 +0200)]
gl: workaround gir warning

The gir generator wrongly assume that the vfunc
GstGLFilterClass.filter() and the method gst_gl_filter_filter_texture()
are related. As a result it complains about not matching argument names.

Workaround this by naming both of their arguments input and output.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/664>

4 years agogl: egl: add missing gir annotations
Guillaume Desmottes [Wed, 13 May 2020 08:08:25 +0000 (10:08 +0200)]
gl: egl: add missing gir annotations

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/664>

4 years agogl: x11: skip gst_gl_display_x11_new_with_display() from gir
Guillaume Desmottes [Wed, 13 May 2020 08:13:04 +0000 (10:13 +0200)]
gl: x11: skip gst_gl_display_x11_new_with_display() from gir

The X11 Display type is not usable in gir.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/664>

4 years agoriff-media: add H.265
Nicola Murino [Fri, 25 Jan 2019 21:48:17 +0000 (22:48 +0100)]
riff-media: add H.265

Closes #359

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/43>

4 years agoglcolorscale: fix documentation
Mathieu Duponchelle [Tue, 5 May 2020 15:39:04 +0000 (17:39 +0200)]
glcolorscale: fix documentation

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/653>

4 years agogl/display/egl: ensure debug category is initialized
Matthew Waters [Thu, 14 May 2020 04:00:22 +0000 (14:00 +1000)]
gl/display/egl: ensure debug category is initialized

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/662>

4 years agomeson: Pass native: false to add_languages()
Nirbheek Chauhan [Mon, 11 May 2020 23:28:24 +0000 (04:58 +0530)]
meson: Pass native: false to add_languages()

This is needed for cross-compiling without a build machine compiler
available. The option was added in 0.54, but we only need this in
Cerbero and it doesn't affect older versions so it should be ok.
Will only cause a spurious warning.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/660>

4 years agocheck: verify gst_gl_display_add_context()
Edward Hervey [Sat, 9 May 2020 05:31:04 +0000 (07:31 +0200)]
check: verify gst_gl_display_add_context()

As is done almost everywhere else. Doesn't cost anything.

CID #1462817

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/657>

4 years agotests/gl: add test for GL context removal
Matthew Waters [Fri, 8 May 2020 05:10:17 +0000 (15:10 +1000)]
tests/gl: add test for GL context removal

Tests functionality fixed by:

https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/654

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/656>

4 years agovideorate: Fix buffer timestamp underflow in reverse playback
Thibault Saunier [Mon, 27 Apr 2020 13:57:30 +0000 (09:57 -0400)]
videorate: Fix buffer timestamp underflow in reverse playback

And fix reverse playback buffer duration computation as in reverse
playback, buffer duration is prev_buffer.pts - buffer.pts not pts -
next_pts (buffers are displayed from buffer.pts + buffer.duration for
a duration of buffers.duration).

This is now tested with the `validate.test.clock_sync.videorate.*`
tests in the default integration testsuite where we check the exact
data flow and the synchronization on the clock behaviour with a
TestClock.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/646>

4 years agovideotestsrc: Fix buffer duration in reverse playback
Thibault Saunier [Fri, 24 Apr 2020 00:35:39 +0000 (20:35 -0400)]
videotestsrc: Fix buffer duration in reverse playback

In reverse playback, buffers have to be displayed at buffer.stop running
time, meaning:

    buffer.pts + buffer.duration = prev_buffer.pts
    =>
    buffer.duration = prev_buffer.pts - buffer.pts

We were setting buffer.duration = next_buffer.pts - buffer.pts which
is not correct.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/646>

4 years agogldisplay: Fix context leak when removing a context from the display
Sebastian Dröge [Wed, 6 May 2020 13:13:11 +0000 (16:13 +0300)]
gldisplay: Fix context leak when removing a context from the display

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/654>

4 years agogldisplay: Fix list iteration
Edward Hervey [Wed, 6 May 2020 07:26:35 +0000 (09:26 +0200)]
gldisplay: Fix list iteration

We were never moving past the first entry it seems...

CID #1461275

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/654>

4 years agotests: fix meson test env setup to make sure we use the right gst-plugin-scanner
Tim-Philipp Müller [Thu, 27 Feb 2020 00:05:52 +0000 (00:05 +0000)]
tests: fix meson test env setup to make sure we use the right gst-plugin-scanner

If core is built as a subproject (e.g. as in gst-build), make sure to use
the gst-plugin-scanner from the built subproject. Without this, gstreamer
might accidentally use the gst-plugin-scanner from the install prefix if
that exists, which in turn might drag in gst library versions we didn't
mean to drag in. Those gst library versions might then be older than
what our current build needs, and might cause our newly-built plugins
to get blacklisted in the test registry because they rely on a symbol
that the wrongly-pulled in gst lib doesn't have.

This should fix running of unit tests in gst-build when invoking
meson test or ninja test from outside the devenv for the case where
there is an older or different-version gst-plugin-scanner installed
in the install prefix.

In case no gst-plugin-scanner is installed in the install prefix, this
will fix "GStreamer-WARNING: External plugin loader failed. This most
likely means that the plugin loader helper binary was not found or
could not be run. You might need to set the GST_PLUGIN_SCANNER
environment variable if your setup is unusual." warnings when running
the unit tests.

In the case where we find GStreamer core via pkg-config we use
a newly-added pkg-config var "pluginscannerdir" to get the right
directory. This has the benefit of working transparently for both
installed and uninstalled pkg-config files/setups.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/582>

4 years agotests: don't look for plugins in -base installdir for tests
Tim-Philipp Müller [Wed, 26 Feb 2020 23:46:57 +0000 (23:46 +0000)]
tests: don't look for plugins in -base installdir for tests

-base plugins will always be found in the build directory, and
core plugins will be found either also via the build directory
(if both core and -base are a subproject) or by getting the
pluginsdir via pkg-config if core is installed.

The GST_PLUGIN_LOADING_WHITELIST env var will make sure we only
pick up plugins from core/base and base plugins only from the
builddir.

There is no reason to look for -base plugins in the install dir.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/582>

4 years agotcpserversrc: Add stats property
Vivia Nikolaidou [Thu, 12 Mar 2020 07:38:58 +0000 (09:38 +0200)]
tcpserversrc: Add stats property

Like in tcpclientsrc

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/651>

4 years agodiscoverer: Check sinkpad existence before retrieving caps
Sebastian Dröge [Wed, 6 May 2020 08:52:20 +0000 (11:52 +0300)]
discoverer: Check sinkpad existence before retrieving caps

Otherwise we would error out without releasing the caps first.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/655>

4 years agodiscoverer: Ensure that we have fixed, non-empty caps before passing passing to is_su...
Sebastian Dröge [Wed, 6 May 2020 08:44:39 +0000 (11:44 +0300)]
discoverer: Ensure that we have fixed, non-empty caps before passing passing to is_subtitle_caps()

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/655>

4 years agodiscoverer: Try to get negotiated caps first on pad-added and only then fall back...
Sebastian Dröge [Wed, 6 May 2020 08:42:05 +0000 (11:42 +0300)]
discoverer: Try to get negotiated caps first on pad-added and only then fall back to a caps query

The negotiated caps will be more accurate and are fixed caps in any
case.

Fixes https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/498

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/655>

4 years agogloverlaycompositor: Don't leak caps feature
Edward Hervey [Tue, 5 May 2020 15:17:57 +0000 (17:17 +0200)]
gloverlaycompositor: Don't leak caps feature

Only copy it if we really are going to modify and use it

CID #1439873

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/652>

4 years agovideoaggregator: Don't crash when setting pad properties after the aggregator was...
Sebastian Dröge [Fri, 1 May 2020 08:27:46 +0000 (11:27 +0300)]
videoaggregator: Don't crash when setting pad properties after the aggregator was finalized

The application might still have a strong reference to a pad and change
properties, which should work without crashing.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/649>

4 years agoAdd missing colons to Since markers in the docs
Sebastian Dröge [Fri, 1 May 2020 07:08:40 +0000 (10:08 +0300)]
Add missing colons to Since markers in the docs

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/648>

4 years agogl: Add since tags for new glbasesrc base class
Philippe Normand [Thu, 30 Apr 2020 16:59:45 +0000 (17:59 +0100)]
gl: Add since tags for new glbasesrc base class

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/647>

4 years agogst/gl: Add documentation for new scroll API
Philippe Normand [Thu, 30 Apr 2020 16:59:24 +0000 (17:59 +0100)]
gst/gl: Add documentation for new scroll API

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/647>

4 years agonavigation: Add since tag for new API
Philippe Normand [Thu, 30 Apr 2020 16:58:43 +0000 (17:58 +0100)]
navigation: Add since tag for new API

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/647>

4 years agotest: pbutils: Add check for high throughput scc.
He Junyan [Tue, 28 Apr 2020 13:27:36 +0000 (21:27 +0800)]
test: pbutils: Add check for high throughput scc.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/645>

4 years agolibs: pbutils: Add High throughput scc extensions profile idc(IDC 11)
He Junyan [Tue, 28 Apr 2020 13:14:07 +0000 (21:14 +0800)]
libs: pbutils: Add High throughput scc extensions profile idc(IDC 11)

It is compitable with scc and we can use scc's function to identify it.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/645>

4 years agolibs: pbutils: select extension profile by profile idc.
He Junyan [Tue, 28 Apr 2020 11:10:05 +0000 (19:10 +0800)]
libs: pbutils: select extension profile by profile idc.

the old manner does not consider the profile idc. The profile idc should
play an more important role in recognizing the profile than the other
information. And there is no need to mix profiles of different extensions
together to find the closest profile when the bits stream is not standard,
different extensions support different features and should not be mixed.

The correct way should be recognize the extension category by profile idc
firstly, and then find the closest profile.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/645>

4 years agolibs: pbutils: rename the GstH265FormatRangeExtensionProfile.
He Junyan [Tue, 28 Apr 2020 10:17:30 +0000 (18:17 +0800)]
libs: pbutils: rename the GstH265FormatRangeExtensionProfile.

GstH265FormatRangeExtensionProfile declares the common bits used
for not only format range extensions profiles, but also for several
different h265 extension profiles, such as high throughput, screen
content coding extensions, etc. And So the old name is not proper.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/645>

4 years agoopusdec: prevent 'channels=0' in caps
Guillaume Desmottes [Mon, 27 Apr 2020 07:40:23 +0000 (09:40 +0200)]
opusdec: prevent 'channels=0' in caps

gst_opus_channel_positions() should fail if caps contains "channels=0".
Prevent index underflow when indexing gst_opus_channel_positions[] (CID 1462590).

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/644>

4 years agoglbasesrc: avoid deadlock when querying for OpenGL context
Matthew Waters [Fri, 24 Apr 2020 12:40:10 +0000 (22:40 +1000)]
glbasesrc: avoid deadlock when querying for OpenGL context

Continuation of:
a4e49ba8c9657e2230aad32b6988bbda7a755440
https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/642

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/643>

4 years agogl: fix application context querying between elements
Matthew Waters [Fri, 24 Apr 2020 12:38:32 +0000 (22:38 +1000)]
gl: fix application context querying between elements

Fixes error introduced by:
a4e49ba8c9657e2230aad32b6988bbda7a755440
https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/642

Using the application GL context as the local GL context is not going to
work very well.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/643>

4 years agoopusdec: add 'stats' property
Guillaume Desmottes [Thu, 23 Apr 2020 08:17:31 +0000 (10:17 +0200)]
opusdec: add 'stats' property

Allow users to retrieve the number of samples, and their duration,
generated using PLC.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/641>

4 years agogl: avoid deadlock querying for OpenGL context
Matthew Waters [Fri, 24 Apr 2020 04:44:46 +0000 (14:44 +1000)]
gl: avoid deadlock querying for OpenGL context

If there are two elements and threads attempting to query each other for
an OpenGL context. The locking may result in a deadlock.

We need to unlock each element's context_lock when querying another
element for the OpenGL context in order to allow any other element to
take the lock when the other element is querying for an OpenGL context.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/642>

4 years agotests: add glviewconvert users integration unit test
Matthew Waters [Tue, 21 Apr 2020 10:09:41 +0000 (20:09 +1000)]
tests: add glviewconvert users integration unit test

Catch all smoke test for ensuring a basic pipeline can negotiate
successfully.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/638>

4 years agotextrender: Don't calculate caps on every buffer
Jan Schmidt [Tue, 21 Apr 2020 12:56:49 +0000 (22:56 +1000)]
textrender: Don't calculate caps on every buffer

Only renegotiate with downstream when the srcpad has a pending
reconfigure flag, instead of querying, fixating and sending caps
for every buffer.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/640>

4 years agovideo-anc: Register an init function for Bar meta
Sebastian Dröge [Sun, 19 Apr 2020 15:29:18 +0000 (18:29 +0300)]
video-anc: Register an init function for Bar meta

The init function is mandatory and we keep around uninitialized memory
otherwise.

Fixes https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/issues/747

4 years agoglvideomixer: fix black output after display changes
Matthew Waters [Fri, 10 Apr 2020 04:08:15 +0000 (14:08 +1000)]
glvideomixer: fix black output after display changes

Partial revert of 55e80b550e4ad02aaf44b30ec8b03da9ee485ac6

https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/issues/743

4 years agoglvideomixer: intersect with template caps when updating
Matthew Waters [Fri, 10 Apr 2020 04:05:08 +0000 (14:05 +1000)]
glvideomixer: intersect with template caps when updating

Downstream may provide extra things when we ask it for caps that we may
need to remove.

Fixes the following pipeline warning:

gltestsrc ! glvideomixerelement ! glimagesinkelement

** (gst-launch-1.0:908581): WARNING **: 13:53:28.518: glvideomixer0: update_src_caps returned caps which are not a real subset of the filter caps

4 years agotypefind: Consider MPEG-PS PSM to be a PES type
Jan Schmidt [Fri, 10 Apr 2020 12:45:26 +0000 (22:45 +1000)]
typefind: Consider MPEG-PS PSM to be a PES type

Include the Program Stream Map packet type 0xBC in the
set of packets we treat as PES. This fixes typefinding
on MPEG-PS streams with PSM, where the PSM would previously
be considered a loss-of-sync and cause the typefind
to require more data.

4 years agouridecodebin3: Activate suburi playback item
Philippe Normand [Mon, 12 Aug 2019 09:41:31 +0000 (10:41 +0100)]
uridecodebin3: Activate suburi playback item

The suburi playback item has to be activated after the main playback item so
that playsink can properly enable text rendering.

Fixes #451

4 years agotextrender: Fix AYUV output.
Jan Schmidt [Thu, 9 Apr 2020 18:24:03 +0000 (04:24 +1000)]
textrender: Fix AYUV output.

Fix the check for whether the element is operating in ARGB mode. It
was incorrectly checking if the output format has an alpha channel,
which is true for both ARGB and AYUV, leading to the element
incorrectly outputting ARGB values into AYUV caps.

4 years agolibs: gl: Fix a context leak when display_create_context failed
He Junyan [Wed, 8 Apr 2020 07:26:23 +0000 (15:26 +0800)]
libs: gl: Fix a context leak when display_create_context failed

4 years agodiscoverer: Enhance printed information when not verbose
Thibault Saunier [Fri, 3 Apr 2020 13:36:02 +0000 (10:36 -0300)]
discoverer: Enhance printed information when not verbose

Basically print information about media tracks making without tags
nor buffers in caps making still quite small but containing all
information infos.

Stop making 'Topology' section and just print the info

before:

```
Topology:
  container: MPEG-2 Transport Stream
    audio: AC-3 (ATSC A/52)
    video: H.264 (High Profile)

Properties:
  Duration: 0:00:05.512394259
  Seekable: yes
  Live: no
  Tags:
      audio codec: AC-3 (ATSC A/52)
      bitrate: 192000
      video codec: H.264 (High Profile)
      minimum bitrate: 12947
      maximum bitrate: 12947
```

After:

```
Properties:
  Duration: 0:00:05.512394259
  Seekable: yes
  Live: no
  container: MPEG-2 Transport Stream
    audio: AC-3 (ATSC A/52)
      Stream ID: b076403d73e0c5fc13985832e8d585945603993437ba14b0799f422f9495e8ef:1/00001100
      Language: <unknown>
      Channels: 2 (front-left, front-right)
      Sample rate: 48000
      Depth: 32
      Bitrate: 192000
      Max bitrate: 0
    video: H.264 (High Profile)
      Stream ID: b076403d73e0c5fc13985832e8d585945603993437ba14b0799f422f9495e8ef:1/00001011
      Width: 1920
      Height: 1080
      Depth: 24
      Frame rate: 30000/1001
      Pixel aspect ratio: 1/1
      Interlaced: true
      Bitrate: 10363396
      Max bitrate: 12947
```

4 years agotypefindfunctions: Fix otio typefinder to actually detect otio files
Sebastian Dröge [Mon, 6 Apr 2020 12:20:39 +0000 (15:20 +0300)]
typefindfunctions: Fix otio typefinder to actually detect otio files

The string "\"OTIO_SCHEMA\":" is 14 characters and not 15. Checking for
15 characters would also check for the final '\0', which does not exist
in any otio file as the string is the key of a JSON map.

4 years agotypefindfunctions: Fix otio typefinder detecting anything with curly braces at the...
Sebastian Dröge [Mon, 6 Apr 2020 12:14:41 +0000 (15:14 +0300)]
typefindfunctions: Fix otio typefinder detecting anything with curly braces at the start

memcmp() returns 0 (aka FALSE) on match and a difference otherwise.

Previously the typefinder was matching on anything but otio files that
happened to have some curly braces in the beginning of the file.

Fixes a false positive with a MOV file.

4 years agopbutils: Add EncodingProfile serialization support
Thibault Saunier [Fri, 20 Mar 2020 13:58:19 +0000 (10:58 -0300)]
pbutils: Add EncodingProfile serialization support

4 years agopbutils: Support for variable framerate in the encoding serialization
Thibault Saunier [Thu, 19 Mar 2020 20:02:57 +0000 (17:02 -0300)]
pbutils: Support for variable framerate in the encoding serialization

4 years agovideo-hdr: Fix a docs typo (x -> y)
Jan Schmidt [Thu, 2 Apr 2020 12:30:57 +0000 (23:30 +1100)]
video-hdr: Fix a docs typo (x -> y)

4 years agoaudiodecoder: Handle instant-rate-change event
Jan Schmidt [Fri, 30 Nov 2018 12:10:55 +0000 (23:10 +1100)]
audiodecoder: Handle instant-rate-change event

When receiving an instant-rate-change event, store the updated
seek flags and replace the flags in any input segments with them
to allow for instant switching between trickmodes and not.

4 years agovideodecoder: Handle instant-rate-change event
Jan Schmidt [Fri, 30 Nov 2018 11:56:11 +0000 (22:56 +1100)]
videodecoder: Handle instant-rate-change event

When receiving an instant-rate-change event, store the updated
seek flags and replace the flags in any input segments with them
to allow for instant switching between trickmodes and not.

4 years agoaudiobasesink: Handle an extra case of buffers being out of segment
Jan Schmidt [Fri, 28 Sep 2018 14:49:46 +0000 (00:49 +1000)]
audiobasesink: Handle an extra case of buffers being out of segment

It's possible that a buffer might be within the segment proper,
but not within the "valid" part we're playing, which is only
things after the 'offset' part of the segment. In that case,
the running-times of the buffer-start and buffer-stop will be
GST_CLOCK_TIME_NONE, and we'd better not schedule playback that
far in the future.

4 years agogst-play: Add -i parameter for instant rate changes
Jan Schmidt [Mon, 20 Aug 2018 16:21:23 +0000 (02:21 +1000)]
gst-play: Add -i parameter for instant rate changes

Add a command line flag that enables use of the
instant rate changes flag when doing rate changes.

4 years agotests/examples/seek/instant-rate-change: Add example app
Jan Schmidt [Fri, 17 Aug 2018 17:37:09 +0000 (03:37 +1000)]
tests/examples/seek/instant-rate-change: Add example app

Add an example app to exercise instant rate changes in a few
scenarios. Currently it deadlocks a lot sending rate changes to
paused pipelines.

4 years agoplayback-test: Add support for start_type=stop_type=NONE seeks
Sebastian Dröge [Mon, 18 Jun 2018 05:51:05 +0000 (08:51 +0300)]
playback-test: Add support for start_type=stop_type=NONE seeks

4 years agoplayback-test: Add support for sending instant-rate-change seeks
Sebastian Dröge [Fri, 25 May 2018 08:01:58 +0000 (11:01 +0300)]
playback-test: Add support for sending instant-rate-change seeks

4 years agovideoencoder: subclass 'set_format' should be optional
Stéphane Cerveau [Wed, 1 Apr 2020 15:47:04 +0000 (17:47 +0200)]
videoencoder: subclass 'set_format' should be optional

Aligns documentation with the code to make the subclass
'set_format' call optional.

4 years agovideo-hdr: Rework for GstVideoMasteringDisplayInfo and GstVideoContentLightLevel...
Seungha Yang [Tue, 10 Mar 2020 11:36:16 +0000 (20:36 +0900)]
video-hdr: Rework for GstVideoMasteringDisplayInfo and GstVideoContentLightLevel struct

This commit modifies GstVideoMasteringDisplayInfo and GstVideoContentLightLevel
structs so that each value is to be more like hdr_metadata_infoframe struct
of linux drm header and DXGI_HDR_METADATA_HDR10 struct of Windows.
So each value is no more fraction but normalized one as per CTA 861.G spec.
Also the unit of each value will be consistent with H.264, H.265
specifications, hdr_metadata_infoframe struct for linux and
DXGI_HDR_METADATA_HDR10 struct for Windows.

4 years agovideoaggregator: fix gir warning about invalid doc comment
Matthew Waters [Wed, 1 Apr 2020 09:05:06 +0000 (20:05 +1100)]
videoaggregator: fix gir warning about invalid doc comment

[38/1301] Generating GstVideo-1.0.gir with a custom command.
../subprojects/gst-plugins-base/gst-libs/gst/video/gstvideoaggregator.c:231: Error: GstVideo: identifier not found on the first line:
   *
    ^

4 years agogl: add support for Y412_LE / Y412_BE format
Haihao Xiang [Wed, 1 Apr 2020 05:55:50 +0000 (13:55 +0800)]
gl: add support for Y412_LE / Y412_BE format

Reuse Y410 code for Y412_LE / Y412_BE except gl format. The gl format is
RGBA16 for Y412_LE / Y412_BE.

Sample pipeline:
gst-launch-1.0 videotestsrc ! video/x-raw,format=Y412_LE ! glimagesink

4 years agogl: add support for Y212_LE / Y212_BE format
Haihao Xiang [Mon, 30 Mar 2020 06:37:26 +0000 (14:37 +0800)]
gl: add support for Y212_LE / Y212_BE format

Because the color value is stored in MSB, so we can reuse the
Y210 code for P012_LE / P012_BE

Sample pipeline:
gst-launch-1.0 videotestsrc ! video/x-raw,format=Y212_LE ! glimagesink

4 years agogstvideoaggregator: expose max-last-buffer-repeat property on pads
Mathieu Duponchelle [Fri, 27 Mar 2020 01:48:39 +0000 (02:48 +0100)]
gstvideoaggregator: expose max-last-buffer-repeat property on pads

This can be used to have compositor display either the background
or a stream on a lower zorder after a live input stream freezes
for a certain amount of time, for example because of network
issues.

4 years agosubparse: convert from pango-markup to utf8 ..
Mathieu Duponchelle [Fri, 20 Mar 2020 18:09:17 +0000 (19:09 +0100)]
subparse: convert from pango-markup to utf8 ..

when downstream requires it

4 years agovideo: add support for Y412 format
Haihao Xiang [Thu, 12 Mar 2020 02:27:34 +0000 (10:27 +0800)]
video: add support for Y412 format

Y412 is a packed 12 bits 4:4:4:4 format in the order U, Y, V, A, 2 bytes
per component with the color value stored in the 12 most significant
bits

Refer to https://github.com/torvalds/linux/blob/master/include/uapi/drm/drm_fourcc.h#L182
for the LE variant

4 years agovideo: add support for Y212 format
Haihao Xiang [Wed, 11 Mar 2020 02:58:00 +0000 (10:58 +0800)]
video: add support for Y212 format

Y212 is a packed 12 bits 4:2:2 format in the order Y, U, Y, V, 2 bytes
per component with the color value stored in the 12 most significant
bits

Refer to https://github.com/torvalds/linux/blob/master/include/uapi/drm/drm_fourcc.h#L174
for the LE variant

4 years agovideo-blend: Add support for blending on top of 16 bit per component formats
Sebastian Dröge [Thu, 12 Mar 2020 13:18:22 +0000 (15:18 +0200)]
video-blend: Add support for blending on top of 16 bit per component formats

With this we can support every currently supported video format for
blending.

4 years agoglcontext/cocoa: handle sending one message during shutdown
Matthew Waters [Thu, 26 Mar 2020 02:51:24 +0000 (13:51 +1100)]
glcontext/cocoa: handle sending one message during shutdown

gst_gl_window_quit() will attempt to send a message but will be called
from GstGLContext's finalize handler and so the weak ref that backs
gst_gl_window_get_context will return NULL as it has already been
cleared.  We need that context in send_message_async to decide whether
to run the provided callback immediately or queue in GCD

4 years agoglcontext: add context is valid precondition to get_thread()
Matthew Waters [Thu, 26 Mar 2020 02:50:52 +0000 (13:50 +1100)]
glcontext: add context is valid precondition to get_thread()

4 years agoglwindow: remove unused private alive variable
Matthew Waters [Thu, 26 Mar 2020 02:50:22 +0000 (13:50 +1100)]
glwindow: remove unused private alive variable

4 years agoglupload: fix segfault
Haihao Xiang [Thu, 26 Mar 2020 05:46:56 +0000 (13:46 +0800)]
glupload: fix segfault

Without this fix, it is possible that outbuf is not initialized, which
will result in segfault when call gst_buffer_replace (&outbuf, NULL). In
addition, the patch fixes potential memory leak in restart path.

The segfault can be reproduced by the pipeline below:

GST_GL_PLATFORM=egl \
gst-launch-1.0 videotestsrc ! msdkh265enc ! msdkh265dec ! \
'video/x-raw(memory:DMABuf)' ! glimagesink

4 years agogl: add support for P012_LE / P012_BE format
Haihao Xiang [Tue, 24 Mar 2020 04:39:50 +0000 (12:39 +0800)]
gl: add support for P012_LE / P012_BE format

Because the color value is stored in MSB, so we can reuse the
P016_LE / P016_BE code for P012_LE / P012_BE

4 years agoglupload: Fix fallback from direct dmabuf to dmabuf upload method
Chris Lord [Mon, 9 Mar 2020 10:21:53 +0000 (10:21 +0000)]
glupload: Fix fallback from direct dmabuf to dmabuf upload method

In the situation that the direct dmabuf path is chosen, but with an
unsupported texture format, this causes accept to fail rather than
continue and fail at the upload stage. It is also possibly necessary to
reconfigure after falling back from direct to non-direct dmabuf upload
paths.

4 years agovideorate: Signalling reconfigure to upstream whenever updating downstream caps
Seungha Yang [Mon, 23 Mar 2020 09:42:19 +0000 (18:42 +0900)]
videorate: Signalling reconfigure to upstream whenever updating downstream caps

Previously configured bufferpool can be expired/inactivate by the
updated caps. Therefore new reconfigure event should be signalled in order to
do allocation query dancing between upstream and downstream again.

Fixes: https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/issues/730
4 years agopkgconfig: fix missing gst-tag Requires
Loïc Minier [Sat, 14 Mar 2020 19:22:50 +0000 (20:22 +0100)]
pkgconfig: fix missing gst-tag Requires

Promote deps gstreamer-tag to Requires for pbutils and add missing
gstreamer-tag Requires for gst-audio. Uninstalled builds would fail
with:
stage/usr/local/lib/x86_64-linux-gnu/libgstpbutils-1.0.so: undefined reference to `gst_tag_list_to_vorbiscomment_buffer'
[...]
stage/usr/local/lib/x86_64-linux-gnu/libgstaudio-1.0.so: undefined reference to `gst_tag_register_musicbrainz_tags'

4 years agopkgconfig: add missing gst-rtp dep to gst-sdp
Loïc Minier [Sat, 14 Mar 2020 18:42:24 +0000 (19:42 +0100)]
pkgconfig: add missing gst-rtp dep to gst-sdp

Uninstalled builds using rtp would fail with:
stage/usr/local/lib/x86_64-linux-gnu/libgstsdp-1.0.so: undefined reference to `gst_rtp_payload_info_for_pt'

4 years agovideodecoder: fix documentation
Víctor Manuel Jáquez Leal [Sat, 21 Mar 2020 13:03:44 +0000 (14:03 +0100)]
videodecoder: fix documentation

Fix list indentation, othewise it is interpreted as verbatim.

4 years agonavigation: Use G_DEFINE_INTERFACE
Niels De Graef [Tue, 17 Mar 2020 17:40:26 +0000 (18:40 +0100)]
navigation: Use G_DEFINE_INTERFACE